Long-term Outcomes and Quality of Life In Critically Ill Patients

Long-termOutcomesandQualityofLife

InCriticallyIllPatients

SandraOeyen

2

Everydayyoumaymakeprogress.Everystepmaybefruitful.

Yettherewillstretchoutbeforeyouaneverlengthening,everascending,

everimprovingpath.Youknowyouwillnevergettotheendofthejourney.

Butthis,sofarfromdiscouraging,onlyaddstothejoyandgloryoftheclimb.

WinstonChurchill

Cover:Sandra’squalityoflifebyAlineHartgers

©SandraOeyenGhentUniversityHospitalDepartmentofIntensiveCare1K12ICC.Heymanslaan109000Ghent,[email protected]

3

Long-termOutcomesandQualityofLife

InCriticallyIllPatients

SandraOeyen

Promotors:Prof.dr.JohanDecruyenaereandProf.dr.LievenAnnemans

ThesissubmittedtofulfilltherequirementsforthedegreeofDoctorinHealthSciences

Academicyear2017-2018

4

Promotors:Prof.dr.JohanDecruyenaere

FacultyofMedicineandHealthSciences,GhentUniversity,BelgiumDepartmentofIntensiveCare,GhentUniversityHospital,Ghent,Belgium

Prof.dr.LievenAnnemans

FacultyofMedicineandHealthSciences,GhentUniversity,BelgiumDepartmentofPublicHealth,GhentUniversity,Ghent,Belgium

Guidancecommittee:Prof.dr.DominiqueBenoit

FacultyofMedicineandHealthSciences,GhentUniversity,BelgiumDepartmentofIntensiveCare,GhentUniversityHospital,Ghent,Belgium

Examinationcommittee:ChairProf.dr.ir.P.VanEenoo

FacultyofMedicineandHealthSciences,GhentUniversity,Ghent,BelgiumDepartmentofClinicalChemistry,MicrobiologyandImmunology,GhentUniversity,Ghent,Belgium

SecretaryProf.dr.R.Peleman

FacultyofMedicineandHealthSciences,GhentUniversity,BelgiumDepartmentofInternalMedicine,GhentUniversityHospital,Ghent,Belgium

MembersProf.dr.M.Petrovic

FacultyofMedicineandHealthSciences,GhentUniversity,BelgiumDepartmentofGeriatrics,GhentUniversityHospital,Ghent,Belgium

Prof.dr.P.DepuydtFacultyofMedicineandHealthSciences,GhentUniversity,BelgiumDepartmentofIntensiveCare,GhentUniversityHospital,Ghent,Belgium

Prof.dr.K.ColpaertFacultyofMedicineandHealthSciences,GhentUniversity,BelgiumDepartmentofIntensiveCare,GhentUniversityHospital,Ghent,Belgium

Prof.dr.P.JorensFacultyofMedicineandHealthSciences,Antwerp,BelgiumDepartmentofIntensiveCare,AntwerpUniversityHospital,Antwerp,Belgium

Prof.dr.W.BrouwerErasmusSchoolofHealthPolicyandManagement,ErasmusUniversityRotterdam,Rotterdam,theNetherlandsDepartmentofHealthEconomics,ErasmusUniversityRotterdam,Rotterdam,theNetherlands

5

Dankwoord 9PartOne:Long-termoutcomesandqualityoflife:Introductionandresearchquestions 11I. Introduction 13

1.Movingtowardslong-termoutcomesandqualityoflife 132.Qualityoflife 15

2.1Definition 15 2.2Assessment 15 2.2.1TheEQ-5Dquestionnaire 19 2.2.2TheSF-36questionnaire 21

2.3QOLresearchinthecriticallyillpatient 25

3.CostsandOutcomeStudyintheICU(COSIstudy) 27 3.1Design,setting,patients,andQOLassessments 27

3.2Flowchart 29 3.3Datacollection 30

II. Researchquestions 31 1.Aimandoutline 31 2.Specificresearchquestions 31III. References 36

PartTwo:Systematicreviewandoriginalstudies 43I. 45Qualityoflifeafterintensivecare:Asystematicreviewoftheliterature Published as Oeyen SG, Vandijck DM, Benoit DD, Annemans L, Decruyenaere JM. Quality of life afterintensivecare:Asystematicreviewoftheliterature.CritCareMed2010;38:2386-2400II. 71Long-termoutcomesandqualityoflifeincriticallyillpatientswithhematologicalorsolidmalignanciesPublishedasOeyenSG,BenoitDD,AnnemansL,DepuydtPO,VanBelleSJ,TroisiRI,NoensLA,PattynP,Decruyenaere JM. Long-termoutcomesandqualityof life in critically ill patientswithhematological orsolidmalignancies:asinglecenterstudy.IntensiveCareMed2013;39:889-898III. 91Long-termqualityoflifeincriticallyillpatientswithacutekidneyinjurytreatedwithrenalreplacementtherapy:amatchedcohortstudy PublishedasOeyenS,DeCorteW,BenoitD,AnnemansL,DhondtA,VanholderR,DecruyenaereJ,HosteE.Long-termqualityoflifeincriticallyillpatientswithacutekidneyinjurytreatedwithrenalreplacementtherapy:amatchedcohortstudy.CritCare2015;19:289IV. 117Critically ill octogenarians and nonagenarians: Evaluation of long-term outcomes, posthospitaltrajectoriesandqualityoflifeoneyearandsevenyearsafterICUdischarge

6

Published as Oeyen S, Vermassen J, Piers R, Benoit D, Annemans L, Decruyenaere J: Critically illoctogenarians and nonagenarians: Evaluation of long-term outcomes, posthospital trajectories, andqualityoflifeoneyearandsevenyearsafterICUdischarge.MinervaAnestesiol2017;83:598-609V. 135Developmentofapredictionmodelforlong-termqualityoflifeincriticallyillpatientsPublishedasSandraOeyen,KarelVermeulen,DominiqueBenoit,LievenAnnemans,JohanDecruyenaere.Developmentofapredictionmodelforlong-termqualityoflifeincriticallyillpatients.JCritCare2018;43:133-138

PartThree:Overviewofthethesis 153I. Conciseoverviewofthestudyresults 155 1.Inclusions 155 2.Mortality 156

3.Qualityoflife 1564.Factorswithimpactonlong-termQOL 157

II. Generaldiscussion 159III. Conclusions 171IV. Futureperspectives 172 1.Researchlevel:anongoingbetterknowledgeoflong-termoutcomesandQOL172 1.1FurtherresearchbasedupontheCOSIcohort 172 1.2Globalresearch 172 2.ICUandhospitallevel:improvingoutcomesbypreventivemeasures 173 2.1TriageuponICUadmission 173 2.2Clinicalpatient-centeredoutcomepredictiontool 174 2.3Strategiestodecreaselong-termconsequencesofcriticalillness174 2.3.1IncreasingawarenessofPICSandPICS-F 174 2.3.2ImplementationoftheABCDEFGHbundle 176 2.3.3Attentionfortheenvironmentofcare 177 2.3.4ImplementationofICUstep-downunits:“IRC” 177

3.Post-hospitallevel:improvingoutcomesbyinterventionmeasures 178 3.1Post-dischargefollow-upprograms 178 3.2Peersupport 180

4.Health-economicslevel:resourceallocation 180V. Summary 182VI. Samenvatting 185VII. References 189

7

Addendum 197I. Listofabbreviations 199II. ConciseCurriculumVitae 201III. Additionalpublicationsrelatedtothesubjectofthethesis 210

8

aanmijnlievemanAlainaanonzekattenvoorjullieonvoorwaardelijkesteunaanmijnoudersaanAnnick,JeroenenAlineaanDaniella,Patsy,PatrickenJo“Lifeislikeridingabicycle.Tokeepyourbalanceyoumustkeepmoving.”

AlbertEinstein

9

Dankwoord

IkwilmijnpromotorProf.dr.JohanDecruyenaerealseerstebedankenvoorzijnsteunenadvies

bijhetmakenvanditwerk.ZonderhetbrainstormenmetJohanineencaféinBoston,terwijldebaseball

opgroteTV-schermeniedersaandachttrokbehalvehetonze,zoudezethesiserwellichthelemaalanders

hebbenuitgezien.Dankdat jemijhet vertrouwenhebt gegevenomvrij zelfstandig te kunnenwerken.

Maarookdankdatikbijjeterechtkonmetvragenentwijfels,zowelthesisalsniet-thesisgerelateerd.

Mijn andere promotor, Prof. dr. LievenAnnemans, alsook Prof. dr. DominiqueBenoit, Prof. dr.

EricHoste,dr.WouterDeCorteenProf.dr.RuthPiers,wilikbedankenvoorhuninputbijhetopzettenen

uitvoeren vande respectievelijke studies ivmhemato-oncopatiënten, dialysepatiënten, enouderenop

IntensieveZorg.DankaanProf.dr. PieterDepuydt voor zijn stilleenoprechte steun.De ledenvan de

examencommissie wil ik bedanken voor hun leeswerk en gewaardeerde feedback. Dank aan dr. Sofie

VanderhaeghenenBoVanDenBulckevoorhunmedewerkingaanhet“IntensiveCareRecovery”-project,

datnogsteedsvolopbezigisendatmezeernauwaanhethartligt.Hopelijkkanerverderopgebouwd

wordenindevolgendejaren.

Ik heb de eer gehad om tijdens de “COSI”-studie te mogen samenwerken met 4 fantastische

verpleegkundigen.Daniella, Patsy, Patricken Jo….dank julliewel voordeaangename samenwerkingen

voor het mee uitbouwen van een zeer mooie en nauwkeurige database die uiteindelijk tot deze

doctoraatsthesis hebben geleid! We hebben samen intens gewerkt aan deze studie en jullie

enthousiasme enmotivatie werkten steeds aanstekelijk. Vragenlijsten en informed consents afnemen,

patëntenopsporen,brievenopsturen,telefoneren,…allesgingmeteengroteglimlach.Hetwasheelfijn

omafentoede“COSI-bureau”eensbinnentevallen,nietenkelvooreenstudie-updatemaarookvoor

eengewonebabbel.DankaanProf.Dr. JohanDecruyenaerediehetmogelijkheeftgemaaktdatdeze4

verpleegkundigenmijkondenhelpenbijmijndatacollectie;alleenzouditonmogelijkgeweestzijn.Dank

aanChrisDanneelsdiehielpmethetuitbouwenvaneendatabaseenzorgdevoordenodige“COSI-back-

ups”.Chriszagsteedsstrengtoedatdegeïncludeerdepatiëntenniettelang“aangevinkt”blevenstaanin

IZIS zodat het systeem vlot operabel bleef. Hierdoor is onze dienst wellicht van een computer-crash

bespaard gebleven! Dank aan Lisa en Silvy, die me geholpen hebben met al de administratieve en

emotionelebeslommeringendieeendoctoraatsthesismetzichmeebrengtendiesteedseen luisterend

oorhadden.OokdankaanMarinavoordevelebabbelsindereceptie.

Eenspeciaaldankwoordwilikrichtentotdr.KarelVermeulen.AlsPhDindestatistiekisdeCOSI-

databaseonderworpenaanzijndeskundigheidmetalsdoeleenpredictiemodelvoorlevenskwalitieitop

langetermijnteontwikkelen.Zondermedischeachtergrondhadhij,naeenkorteuitlegvanmij,heelsnel

10

inzichtindealdannietklinischerelevantievansommigestatistischebevindingen.Datkannietiedereen.

Dankvoor jetoewijdingen jenauwgezetwerk.DankookaanProf.dr. JohanDecruyenaereommeedit

conceptuittewerkenenomdeaangenamesamenwerkingmetKarelmogelijktemaken.

Dank aan alle patiënten die steeds zeer bereidwillig hebbenmeegewerkt aan dit project. Hun

dankbaarheidenenthousiasmewaarmeeze,ookvelejarenlater,allevragenlijstenblevenbeantwoorden

zetaantotdenkenomonzeklassiekepost-intensievezorgteherzien.

Mijnouderswilikbedankendatzemedemogelijkheidgebodenhebbenom–reedszovelejaren

geleden-destudiesvangeneeskundeaantevatten,voorhunblijvendesteuntijdensaldiejaren,enook

nadien, en voor nog zoveel meer. Ook dank aan Annick, Jeroen, en Aline voor de vele leuke

familiemomenten,shoppingdagen,mails,enbabbelsaandetelefoon.DeoprichtingvanonzeWhatsApp-

groep“DeOeyens”iseensuperinitiatiefgeweest!!Dankaanmijnkattendiesteedshunonverstoorbare

zelvebleven.Vaakkwamenzenaastdecomputerliggenwaaraanikzattewerken,uiteraardjuistopde

papieren die ik nodig had. Onder tevreden gesnor duwden ze weleens op de toetsen van het

computerklavier.Zobenikenkelekereneenstuktekstkwijtgeraakt...Hoerade“undo”toets!!

En dan “last but absolutely not least” dank aan mijn lieve man Alain. Op onze vele mooie

fietstochten heb ik weleens aan de parallellen tussen het schrijven van een doctoraatsthesis en het

fietsengedacht.Teneerstemoetjevoorbeideneengoedebalansvinden.Eenbalanstussenvrijetijden

werk,eneenbalanstussenbewegingensnelheid.Zonderdiebalansdreigjeinbeidegevallenletterlijkof

figuurlijk omver te vallen. Sommige fietstochten zijn moeilijk, en dan is het afzien. Tegen de wind en

regen in of steil omhoog. Maar altijd was jij daar Alain, en met de uitspraak “Wanneer begint het

moeilijke deel?” kon je me zelfs door alle spierpijn heen steeds doen lachen. Ook op dagen dat het

mindergoedlukteomaanmijndoctoraattewerkenkoniksteedsopjesteunrekenen.Eveneentandje

lager zetten of een beetje harder trappen en alles ging weer vlotter. En in de voor mij vaak

angstaanjagendeafdalingreedjevoormijenriepdan“hiernueenbeetjeremmen”(alsofikdatnogniet

had gedaan!!) en “de bocht goed inschatten” en “voila, nu wat bijtrappen” en zo ook coachte je me

tijdens dit werk. Af en toe remmen (“tranquillo”) en soms een beetje bijtrappen. En elke bocht goed

blijveninschatten.

SandraOeyen,27juni2018

11

PartOne

Long-termoutcomesandQualityofLife:IntroductionandResearchQuestions

13

I.Introduction

1.Movingtowardslong-termoutcomesandqualityoflife

Theintensivecareunit(ICU),asadedicatedareainthehospital,emergedaround1960.Mainand

only focusat that timeand in the followingdecadeswas reducing ICU-mortality [1].Hospitalmortality

became themost importantparameter to reportoutcome, especiallywith the introductionof the first

generalseverityofillnessscore,theAcutePhysiologyandChronicHealthEvaluationII(APACHEIIscore)

[2]thatcouldestimatetheprobabilityofhospitalmortality.

Itwasbytheendofthe80sand90sthatcriticalcarephysiciansstartedtobeawareoftheneed

toevaluateotherendpointsbeyondshort-termmortality[3-5].Animportantdevelopmentinthefieldof

healthcareatthattimewastherecognitionofthecentralroleofthepatient’sviewregardingthequality

of medical care outcomes. A medical outcome became “the extent to which a change in a patient’s

functioningorwell-beingmeets thepatient’sneedsandexpectations” [6].Earlier, Lembckestated that

“thebestmeasureofqualityisnothowwellorhowfrequentlyamedicalserviceisgiven,buthowclosely

theresultapproachesthefundamentalobjectivesofprolonginglife,relievingdistress,restoringfunction

and preventing disability” [7]. European and American critical care societies were founded and held

roundtable conferences andworkshops concerning “outcomes research” and “surviving intensive care”

[8,9].

It was only since the 90s that clinical investigators began to use information about functional

status and well-being of patients. Earlier, data from patients regarding their experiences of disease,

treatment,andoutcomehadnotbeenroutinelycollected.Severaladvancesinthemethodsforassessing

patientperspectivesoccurredintheseyears.Someoftheseadvanceswereanimprovedunderstandingof

the major dimensions – physical, mental, and cognitive - of health and the validity of specific

measurementsinrelationtothesedimensions,ademonstrationoftheusefulnessofstandardizedhealth

surveys in clinical trials, and the development of general population health surveys. Techniques for

constructinghealthmeasuresandcontentofthesemeasuresimprovedovertime.

Some10yearsago,achapterintheyearbookoftheEuropeanSocietyofIntensiveCareMedicine

(ESICM)wasdedicatedtolong-termoutcomes[14],whichwastheproofthatmoreeffortshadbeenput

onmeasuringoutcomesotherthanonlysurvival.Gradually,thefocusonoutcomehadshiftedfromICUto

hospitalmortality,fromhospitalmortalitytopost-hospitalfunctionalityandwell-being,andtothe(very)

long-term-outcome.

Measuring and understanding the outcome of a treatment from the patients’ perspective

captures the essence of patient-centred care and incorporating this information in medical decision-

makingisessential.Althoughthischangeinoutcomeinterestseemsratherlateintime,it is logicalthat

formanyyears the traditional goalof critical caremedicinehasbeen todecrease short-termmortality

14

becausecriticalcaremedicineperdefinitiontreatsthemostcriticallyillpatientswithaninherenthighrisk

ofmortality.Themajorityofrandomizedcontrolledtrialsinthefieldofcriticalcaremedicinestillhaveas

primaryendpointshort-termmortalityandsomeverywellknownkey-studieshavefocusedonthis[10-

13]. While reducing short-term mortality is worthy, extremely important, and the core business of a

criticalcarephysician,thisgoalhoweverfailstoaddressthe issueofwhat itmeanstosurvive intensive

care[9].

The main reason for the increasing and still expanding interest in long-term outcomes is that

advancesindiagnostic,supportiveandtherapeuticoptionsmakethatmoreandmorepatientsnowadays

survivetheircriticalillness[15,16].Thereisalsoanincreasingacknowledgmentthattheepisodeofcritical

illnessisnotjusttheperiodoftimethepatientspendsintheICUbutistheperiodthatbeginswiththe

onsetofdeteriorationandendswhenthepatient’sriskoflatesequelaereturnstoabaselinelevelofrisk

of a similar patient who has not been critically ill [9]. Critical care can therefore be identified as one

importantpiece ina complex continuumof care. For this reason,wehave toquestionwhetherand to

whatextentcritical illnesswill affect the long-term (≥12monthsafter ICUdischarge) functionalityand

qualityoflife(QOL)insurvivors.

From: Angus DC, Carlet J, 2002 Brussels Roundtable Participants Surviving Intensive Care: A report from the 2002 BrusselsRoundtable.IntensiveCareMed2003;29:368-377

AsQOL incorporates a patient’ values andpreferences, it distinguishes itself fromother health

outcomemeasures[17].Hence,nexttosurvivalormortalityrate,indicesregardinglong-termmorbidity

andQOLafterICUdischargeshouldbetakenintoaccountaswelltofullyappreciatelong-termoutcomes

incriticallyillpatients.QOLconsiderationsmaybeparticularlyimportantinthecriticalcaresetting,where

interventionscansavelivesbutwherethefinaloutcomemaybevaluedasworsethandeath[18].

15

Abetterunderstandingofhowcriticalcareaffectsthelong-termhealthandQOLofitssurvivors

can help critical care physicianswhendeciding on allocation therapeutic efforts in the future, and can

helpinabetterandefficientadvancedcareplanningandcommunicationwithpatientandfamily.

2.Qualityoflife

2.1Definition

OneofthedifficultiesinQOLresearchisdefiningexactlywhatonemeansby“QOL”asthereisno

universallyacceptedorapplieddefinition.QOL,healthstatus,functionalstatus,functionality,andhealth-

relatedQOL(HRQOL)arealltermsthatareoftenusedinliterature,butwhichmayreflectquitedifferent

aspects of an individual's well-being. Differences in conceptualization of QOL may lead to different

measurementapproaches,whichmayleadtootherresults[18,19].

TheWorldHealthOrganizationdefines“QOL”as“anindividual'sperceptionoftheirpositioninlife

in the context of the culture and value systems in which they live and in relation to their goals,

expectations, standards and concerns. It is a broad ranging concept affected in a complexway by the

person'sphysicalhealth,psychologicalstate,personalbeliefs,socialrelationshipsandtheirrelationshipto

salient features of their environment” [20]. According toWikipedia, QOL is “the general well-being of

individuals and societies, outlining negative and positive features of life. It observes life satisfaction,

including everything from physical health, family, education, employment, wealth, religious beliefs,

financeandtheenvironment”[21].

Theoretically, it is important not tomix up the conceptofQOLwithHRQOL.An assessmentof

HRQOLiseffectivelyanevaluationofhowanindividual'swell-beingorQOLmaybeaffectedovertimeby

a disease, disability, or disorder. However, this distinction between QOL and HRQOL seems far too

theoretical since it is hardly imaginable that an individual’s well-being and perception of life, which is

definedasQOL,willnotbeinfluencedbyhealth,whichisdefinedasHRQOL.Inliterature,bothtermsare

oftenusedinterchangeably.Throughourresearchwewillalwaysrefertotheterm“QOL”,whichshould

theoreticallybe“HRQOL”.

2.2Qualityoflifeassessmentinthecriticallyillpatient

QOL measures will either be specific or generic. Specific QOL measures are designed to be

relevanttoaparticulardisease,toacertainpatientpopulation,toacertainfunction(forexamplesleep),

or to a specific condition (for example pain). As critically ill patients are a very heterogenic group of

patients,generic instrumentsthatcanbeusedacrossawiderangeofdiagnosticcategoriesareneeded

[22].Theymayhoweverbe lesssensitivetochanges incertainconditionsorsymptomsascomparedto

specific QOL instruments. Generic instruments should be reliable, valid, and contain a high

responsiveness.

16

Reliabilityistherepeatabilityofobservations(test-retest)wheninstrumentsareadministeredby

differentindividualsandatdifferentpointsintime.

Validityreferstoaninstrumentthatmeasureswhatitclaimstomeasure.ThewayaQOLmeasure

isvalidatedfallsgenerallyontooneofthreecategories:constructvalidity,contentvalidity,andcriterion

validity.

Constructvalidityisthedegreetowhichatestmeasureswhatitclaimstobemeasuring.Itisthe

overarchingconcernofvalidityresearch,subsumingallothertypesofvalidityevidence.Constructvalidity

examinesifthemeasurebehaveslikethetheorysaysthatthemeasureshouldbehave.Forexample,the

construct validity of a questionnaire can be checked to ensure that certain groups (older, lower social

classes, those with illnesses) will gain worse scores than other groups (younger, higher social classes,

thosewithoutillnesses).

Contentvalidityreferstochoiceof,andrelativeimportancegivento,itemsonaquestionnaire.It

is important that itemsappropriate to thephenomenonunder investigationarechosenand if theyare

weighted in some way, that the weights reflect the perceived level of difficulty or health problem.

ReferringtoQOLsurveys, it reflectshowwellaQOLquestionnaireeffectivelyandcomprehensivelycan

measurealldifferenthealthdomains.Contentvalidity isdifferentfromfacevalidity,whichrefersnotto

whatthetestactuallymeasures,butwhether itemsonaquestionnaireappearbothappropriatetothe

phenomenon being measured and to make sense, as well as being easily understood. Face validity

assesseswhetherthetest"looksvalid"totheexamineesthattakeit.Contentvalidityrequiresexpertsto

evaluate whether test items assess defined content and more rigorousstatistical teststhan does the

assessmentoffacevalidity.

Criterion validity refers to the ability of a QOL survey to be systematically related to the gold

standardsofoneormoreoutcomecriteria,whichisdifficultasgoldstandardsarehardtofindinthearea

ofQOLresearch.

Other formsof validity examine the extent towhich individual items in a domainmeasure the

sameunderlying(internalconsistency)ordifferentaspectsofQOL(factoranalysis)[19,22].

The sensitivity to change or “responsiveness” of an instrument is a very important criterion to

considerwhenselectingmeasures. It isessential thatevaluative instrumentsareable todetect change

andthelevelofthischangeovertime[22].

ExamplesofgenericQOLinstrumentsaretheNottinghamHealthProfile(NHP)[23],theSickness

Impact Profile (SIP) questionnaires [24, 25], theQuality ofWell-Being (QWB) Scale [26], the EuroQol-5

Dimensions(EQ-5D)[27,28],theRAND-36ItemHealthSurvey(RAND-36)[29],andtheMedicalOutcomes

Study 36-item Short Form Health Survey (SF-36®) [30-34]. All these instruments are commonly used

and/orcitedintheEnglishlanguageliterature.

17

TheNHPwasdevelopedtoreflect layratherthanprofessionalperceptionsofhealth.Itcontains

38 yes/no statements in 6 domains: mobility, pain, sleep, energy, emotional reactions, and social

isolation.Validityisgood,butitsreliabilityandresponsivenessincriticallyillpatientsarelesswell-known

[22]. The SIP survey was constructed as a measure of sickness in relation to impact on behavior. It

contains136itemsin12categories:work,recreation,emotionalbehavior,alertness,homemanagement,

sleep, body care, eating, ambulation, mobility, communication, and social interaction. Test-retest

reliability(r=0.92)andinternalconsistency(r-0.94)arehigh[24,25].TheQWBis,equaltotheEQ-5D,a

preference-basedmeasuredesignedtomeasureQOLoverthepreviousthreedaysinfourareas:physical

activities,socialactivities,mobility,andsymptom/problemcomplexes.Itconsistsof71itemsandtakes20

minutestocomplete.Thefourdomainscoresofthequestionnairearecombinedintoatotalscorethat

rangesfrom0to1,with1representingoptimumfunctionand0representingdeath[26].TheRAND-36is

avalidated,profile-basedQOLmeasurebasedontheSF-36.Questions intheRAND-36andintheSF-36

are similar and the correlation between themeasures is excellent (r=0.99) [29]. Scoring systems differ

slightly.

Therearenouniformly'worst'or'best'performinggenericinstruments.Thedecisiontouseone

overanother,touseacombinationof2ormore,ortouseagenericmeasurealongwithapreference-

basedmeasure will be driven by the purpose of themeasurement. The choice will also depend on a

varietyoffactorsincludingthecharacteristicsofthepopulation(age,healthstatus,language/culture)and

theenvironmentinwhichthemeasurementisundertaken(clinicaltrial,routinephysicianvisit)[35].

18

Examplesofgenericandspecificoutcomemeasurements

Genericinstruments

QOL NottinghamHealthProfile(NHP)

SicknessImpactProfile(SIP)

QualityofWell-Being(QWB)

EuroQoL-5D(EQ-5D)

RAND-36ItemHealthSurvey(RAND-36)

MedicalOutcomeStudyShortForm-36HealthSurvey(SF-36®andSF-36v2®)

Functionalstatus Katz’sActivitiesofDailyLiving(ADL)

KarnofskyIndex

BarthelIndex

Mentalstatus HospitalAnxietyandDepressionScale(HADS)

Specificinstruments

Diseasespecific NewYorkHeartAssociation(NYHA)FunctionalClass

AmericanThoracicSociety(ATS)RespiratoryQuestionnaire

GlasgowComaScore(GCS)

Patientgroupspecific ClinicalFrailtyScoreinolderpatients

Conditionspecific Numericratingscale(NRS)forpainassessment

Mini-MentalStateExamination(MMSE)forneuropsychologicalfunction

Function PittsburghSleepQualityIndex(PSQI)forassessmentofsleepquality

Wechose touse theEQ-5Dand theSF-36®questionnaires throughour research.Wepreferred

the combination of a respectively preference-based score with a single index value, reflecting the

preference of being in a health state and to be used in future economic evaluations, together with a

comprehensive short-form generic QOL measure with a better discriminative power [36]. They are

commonly used in critical care outcome research, arewell validated andhavepopulationnorms. Both

questionnaireswillnowbeexplainedmoreindetail.

19

2.2.1TheEQ-5Dquestionnaire

TheEQ-5Disastandardized,genericandpreference-basedmeasureofhealthstatedevelopedby

the EuroQol group (www.euroqol.org) [27, 28]. It is a simple and short questionnaire that is easily

understood and answered by patients. Furthermore, its usefulness and validity have been tested in

differentpatientgroupsandinthecriticallyillpatientpopulation[9,37-39].ItcanassessQOLinface-to-

face interviews, interviews by phone or by sending the questionnaire by regularmail. It consists of 3

parts:

Thefirstpartisasimpledescriptivepartwherehealthstatuscanbeassessedinfivedimensions:

mobility, self-care, usual activities, pain/discomfort, and anxiety/depression. Each dimension has three

levels:1=noproblems,2=moderateproblemsor3=severeproblems.Thedecisionperdimensionresultsin

a 1-digit number (1, 2 or 3) expressing the level selected for that dimension. The digits for the 5

dimensions can be combined in a 5-digit number describing the respondent’s health state. Therefore,

patientscanbeclassifiedinto1of243(35)possiblehealthstates.

TheEQ-5DMobilityIhavenoproblemsinwalkingabout � Ihavesomeproblemsinwalkingabout �Iamconfinedtobed �Self-CareIhavenoproblemswithself-care �Ihavesomeproblemswashingordressingmyself �Iamunabletowashordressmyself �Usualactivities(e.g.work,study,housework,familyorleisureactivities)Ihavenoproblemswithperformingmyusualactivities �Ihavesomeproblemswithperformingmyusualactivities �Iamunabletoperformmyusualactivities �Pain/DiscomfortIhavenopainordiscomfort �Ihavemoderatepainordiscomfort �Ihaveextremepainordiscomfort �Anxiety/DepressionIamnotanxiousordepressed �Iammoderatelyanxiousordepressed �Iamextremelyanxiousordepressed �

ThesecondpartistheEQ-visualanaloguescale(EQ-VAS),whichisa20-cmverticalhash-marked

scalewherepatientscanratetheirperceivedoverallhealthbetweentwoanchors“0”(worstimaginable

health state) and “100” (best imaginable health state). The EQ-VAS score is patient-based and canbe

usedas aquantitativemeasureofhealth status as judgedby the individual respondents.VAShas long

beenused in themeasurementofhealthstatusandQOL indiversepopulations [40,41]. It canalsobe

used tomeasure specific aspects ofQOL such aspain [42].MeasuringVAShas a good validity, andan

excellentreliability.TheEQ-VASscorehasagoodanchor-basedresponsiveness,meaningthatthescore

hastheabilitytodetectclinicallyimportantchangesovertimebetweenitstwo“0-100”anchors.Thelevel

20

of responsiveness calculatedbydistribution-basedmethods -using statistical analysis (i.e. standardized

responsemean,effectsize)tocalculatewhetherthemagnitudeofchangeinscoreovertimeshouldbe

consideredsignificant–ishowevermoderate,especiallyformentalhealth,meaningthatthereisabetter

distribution-basedresponsivenessforthephysicalcomparedtothementalhealthsubscales.VAScanbe

analternativetoamulti-itemmeasure,dependingontheresearchquestion[43].

InthethirdpartoftheEQ-5D,thehealthstatus–asassessedinthefirstpart–canbeconverted

by the researcher into a single index value,which indicates thepreferenceof being in a health status,

hence the name “utility index” (UI). This conversion is done by applying a formula that essentially

attaches values (=weights) to each of the levels in each dimension. The index can be calculated by

deducting the appropriate weights from 1, which is the value for full health (health state 11111).

ConvertingahealthstatetowardsaUIrequiresthusgeneralpopulation-basedvaluesets.Therationale

behind this is that the values are supposed to reflect the preferences of local taxpayers and potential

receiversofhealthcare.TheUIreflectsthereforetheopinionofthegeneralpopulation,whereastheEQ-

VASscoreispatient-basedandnotrepresentativeforthegeneralpopulation.

Generalpopulation-basedvaluesetshavebeenderivedforEQ-5Dinseveralcountriesusingthe

timetrade-off(TTO)valuationtechniqueortheEQ-5DVAS-technique.IntheTTOtechnique,respondents

fromthegeneralpopulationareasked,forexample,toimaginetheyliveinahealthstate(e.g.22222)for

10yearsand thenasked to specify theamountof time theyarewilling togiveup to live in fullhealth

instead(i.e.11111).Forexample,someonemightfind8yearsin11111equivalentto10yearsin22222.

TheVAStechniqueontheotherhand,askspeopletoindicatewhere,onaverticalthermometer-likescale

ranging frombest imaginablehealth (“100”) toworst imaginablehealth (“0”), they thinkahealthstate

shouldbepositioned.Althoughthereisstillanongoingdiscussionwhichofbothtechniquesispreferable,

there isnowmoreor lessaconsensusthat theTTO isamorereliablevaluationtechniquebut that the

VAStechniqueismorepracticalandfeasibleinuseandthereforeanacceptedtechniqueforpreference

valuemeasurement.

ForBelgium,722 value sets basedupon theEQ-VAS techniqueas valuationmethodwereused

[44].The indexvalueof theEQ-5D is thusapreference-basedmeasureofhealthstatus - reflecting the

preference to be in a certain health state - ensuring that consequences that are more preferred will

receive a greater weight in the analysis than less preferred ones. It makes the EQ-5D suitable for

quantifyinghealthoutcomes,which canbeuseful in clinical andeconomical evaluationsofhealth care

interventions.

The UI can range from -0.1584 (which is the index value for a health status indicating severe

problems on all 5 dimensions: the 5-digit number in part 1will be 33333) to 1.000 (which indicates a

healthstatuswithnoproblemsonthe5dimensions:the5-digitnumberinpart1willbe11111).Anindex

valueof0.0000equalsdead.In17ofthe243possiblehealthstatesthecorrespondingUIisbelowzero,so

21

itbecomesnegative.Thisindicatesahealthstatusthatisconsideredtobeworsethandead,soahealth

statusnoonepreferstobein.Inthatcase,thepatienthassevereproblemsinatleast3or4orinall5

dimensions,mainlyinthepain/discomfortandanxiety/depressiondimension.Comaalsocorrespondsto

aUIbelowzero[45].

TheEQ-5Dhasnowbeentranslatedintomorethan170languages–includingDutch-andisused

worldwidefreeofcharge.

However, ceiling effects, meaning that certain variations no longer could be captured, were

reported and a Task Force was established within the EuroQol Group [46] to investigate methods to

increasereliabilityandsensitivitywhilemaintainingthesamefeasibility.AnewversionoftheEQ-5Dwas

developedwhichincludedfivelevels(5L)ofseverity(noproblems,slightproblems,moderateproblems,

severeproblems,andextremeproblems)ineachoftheexistingfiveEQ-5Ddimensions.Itwascalledthe

“EQ-5D-5L”.TheexistingEQ-5Dwasrenamedthe“EQ-5D-3L”,referringtothe3levelsofseverityoneach

ofthe5dimensions.AsweusedtheEQ-5D-3Lthroughoutourresearch,westillwillusethename“EQ-

5D”forsimplicityreasons

2.2.2TheSF-36questionnaire

The SF-36® questionnaire is another example of a generic QOL-survey [30]. It is the most

commonly used QOL measure. The SF-36® was first published in 1992 and further developed and

validatedin1993and1994[31,32]. Itwasdevelopedasashort-formmeasureoffunctioningandwell-

being intheMedicalOutcomesStudy(MOS).TheMOSwasa4-year longitudinalobservationalstudyof

the variations in practice styles and of the health outcomes for chronically ill patients.Over 23000US

patientsparticipated in this study [47].TheMOSprovided theopportunity foral large-scale testof the

feasibilityof self-administeredpatientquestionnairesandgenerichealth scales.TheMOSsurveyswere

basedonamultidimensionalmodelofhealthandassessed40healthconceptsinacomprehensiveway.

TheSF-36®questionnairecontains11sectionsholdingatotalof36questionsoritemsmeasuring

QOL at 8multi-item health domains or scales. The 8 health domains representing in the SF-36®were

selected from the 40 health domains that were included in the MOS. Those 8 represent the health

domainsmostfrequentlymeasuredinhealthsurveysandthosebelievedtobemostaffectedbydisease

and health conditions. These 8 domains are: general perceptions of health, physical functioning, role

limitationsdue tophysical-, or emotional problems, social functioning, bodilypain, vitality, andmental

health. The36th item, health transition, provides information about perceived changes in health status

comparedtooneyearago.Twocomponentsummarymeasures,aphysicalandamental,arecalculated

summary measures where respectively the physical or the mental domains will account more in the

measureandwhererespectivelythementalandphysicalscalesweightnegatively.

22

AlthoughtheSF-36®provedtobeusefulformanypurposes,10yearsofexperiencerevealedthe

need and potential for improvements.At the endof the 90s and beginning of the years 2000, the SF-

36v2®wasdeveloped[33].Itisalsoa36-itemhealthsurveyyieldingthesame8healthdomainscalesand

the same 2 component summary measures. Compared to the SF-36® it has improved item wording

withoutambiguityorbias,improvedlay-outofquestions,andincreasedcomparabilityinrelationtoother

cultures.Responsechoicesfortherolelimitationduetophysicalhealthdomainandrolelimitationdueto

emotionalproblemsdomainwereincreasedanddecreasedforthementalhealthandvitalitydomains.All

these matters led to a survey which was easier to understand and which had a better validity and

reliability.

Althoughthe8healthdomainsoftheSF-36v2®areassessedin36questions,itisacomprehensive

and rather short QOLmeasure. Patients or other respondents are not tired of completing the survey,

whichiscertainlyanadvantageinthecriticallyillpopulation.

Eachofthe8healthdomainsoftheSF-36v2®hasarawscore,whichcanbeconvertedto0-100

scoresthroughasimplescoringalgorithm.Thehigherthescore,thebettertheconditiononthatdomain.

General population norms provide a basis for meaningful comparisons across the health scales. The

“physicalfunctioning”generalpopulationnormisbetween80and90whilethe“vitality”normisaround

60. Differences in norms for each health domain must be kept in mind which can make a correct

interpretationdifficult.

TheinterpretationofSF-36v2®resultshasbeengreatlysimplifiedwiththenorm-basedscoringof

itshealthdomainscalesandcomponentsummarymeasures.Thesenorm-basedscoresarebasedupon

the mean and standard deviations (0-100 scores) for each health domain of the US general healthy

populationin1998.Itisrecommendedthatusersbasetheirinterpretationsonnorm-basedscores,where

alldomainshavethesamemean(50)andthesamestandarddeviation(10).Norm-basedscoringdoesnot

only allow to comparewith a general healthy population (the 1998US general population) but it also

allowstocomparetheresultsofonedomainwithotherdomains,sincealldomainshavethesamemean

andstandarddeviation.

The first stepof transforming0-100scores tonorm-basedscoresconsistsof standardizingeach

SF-36v2®healthdomainscaleusingaz-scoretransformation.Az-scoreforeachdomainiscalculatedby

subtractingthe1998USgeneralpopulationmeanforthatrespectivedomainfromthe0-100score,and

thendividingthedifferencebythecorrespondingstandarddeviationofthe1998USgeneralpopulation

onthatdomain.Thenextstep is to transformthestandardz-scores tonorm-basedscoresbyaT-score

transformation(mean50;SD10).Thisisaccomplishedbymultiplyingeachz-scoreby10andthenadding

50 to the resultingproduct. The result is thenorm-based score for that respectivehealthdomain. The

transformationtowardsphysicalandmentalcomponentnorm-basedscoresgoesinananalogueway.

WeassessedSF-36v2®asnorm-basedscorestobeabletocomparethemdirectlywiththegeneral

23

healthypopulation,withagroup-levelrangeof47-53consideredasaverageornormal.Groupscoresless

than47indicateimpairedfunctioningwithinthathealthdomain;groupscoresgreaterthanorequalto53

shouldbeconsideredabovethenormativesample[33].Individualpatientdataareconsideredasaverage

ornormalwithinarangeof45-55.Scoreslessthan40orabove55indicateanimpairedorbetterhealth

conditionthanthatofthegeneralpopulation.Scoresbetween40-44shouldrequirefurtherinvestigation

todeterminethepresenceofimpairedfunctioningfortheindividualpatient.

Apartfromtheadvantageofnorm-basedscoringforinterpretationofthestudyresults,itisalso

importanttoexaminevisuallytheprofileofthedomainscores.Thisprofile,representingthescoresofan

individualpatientor themeansormediansof a groupprovidesabroadoverviewof thehealth status.

Therefore, the first scores in the profile should always be the physical andmental component scores.

These should be placed on the left side, emphasizing the importance of first considering the overall

results inthephysicalormentalhealthdomains.The8healthdomainsof theSF-36v2®shouldthenbe

placed from left to right in this specific order: physical functioning, role limitations due to physical

problems, bodily pain, general health, vitality, social functioning, role limitations due to emotional

problems,andmentalhealth.Hence,thehealthdomainsreflectingmainlyphysicalfunctioningareonthe

leftsideoftheprofile,whilehealthdomainsmainlyreflectingmentalhealthareontheright.

Averyquickinterpretationofahealthstatusatfirstsightisthuspossible.

AdaptedfromWareJEJr,KosinskiM,BjornerJB,Turner-BowkerDM,GandekB,MaruishME(2007).User’smanualfortheSF-36v2®Health Survey (2nd ed.). Lincoln, RI:QualityMetric Incorporated.Optum’s Table Abbreviated ItemContent for the SF-36v2®HealthSurveyHealthDomainScale,Figure7.1,page74.

24

Thereliability,validityandresponsivenessoftheSF-36v2®hasbeenconfirmedinthecriticallyill

population, and its use is validated in face-to-face interviews, interviews by phone, computer

administeredorbysendingthequestionnairebyregularmail[33,34].TheSF-36v2® iscurrentlyavailable

inmore than250 language translations, includingDutch.Itmayprovidemore informationandmaybe

more sensitive and discriminative than the EQ-5D [9, 18, 31-34, 37]. However, in the older patient

population,wherebrevityofQOLmeasuresispreferred,lowercompletionratesoftheSF-36v2®canbea

problem[38].

The SF-12v2® andSF-8™ health surveys are abbreviated versions of the SF-36v2® containing

respectively12and8questions.Theymeasurethesame8healthdomains,andeachsurveyprovidesalso

the physical and mental component summary scores. Their discriminative power is however less. Apreference-basedutilityindex,theSF-6Disalsoavailabletohelpunderstandeconomicbenefit.

The SF-36®, SF-36v2®, and their shorter versions, are registered trademarks of the Medical

OutcomesTrust andareusedunder license. The SF-36v2®Health Survey is copyrighted©1992, 1996,

2000,byMedicalOutcomesTrustandQualityMetric Incorporated.Permission to reproduceand touse

theSF-36v2®HealthSurveyforbothscholarlyandcommercialpurposescanbeobtainedbycompletinga

SurveyInformationRequestFormat:http://optum.com.WeusedtheSF-36v2®throughoutourresearch,

andwillrefertoitas“SF-36”forsimplicityreasons.

AbbreviatedquestionsfromtheSF-36

Question/section

Domain Abbreviatedcontent

1 Generalperceptionofhealth Isyourhealthexcellent,verygood,good,fair,poor

2 Healthtransition Howhealthisnowcomparedto1yearago3a Physicalfunctioning Vigorousactivities,suchasrunning,liftingheavyobjects,participatingin

strenuoussports3b Physicalfunctioning Moderateactivities,suchasmovingatable,pushingavacuum,bowling,

playinggolf3c Physicalfunctioning Liftingorcarryinggroceries3d Physicalfunctioning Climbingseveralflightsofstairs3e Physicalfunctioning Climbingoneflightofstairs3f Physicalfunctioning Bending,kneeling,orstooping3g Physicalfunctioning Walkingmorethanonekilometer3h Physicalfunctioning Walkingseveralhundredmeters3i Physicalfunctioning Walingonehundredmeters3j Physicalfunctioning Bathingordressingoneself4a Rolelimitationsdueto

physicalproblemsCutdowntheamountoftimespentonworkorotheractivitiesbecauseofphysicalhealth

4b Rolelimitationsduetophysicalproblems

Accomplishedlessthanyouwouldlikebecauseofphysicalhealth

4c Rolelimitationsduetophysicalproblems

Limitedinkindofworkorotheractivitiesbecauseofphysicalhealth

4d Rolelimitationsduetophysicalproblems

Haddifficultyperformingworkorotheractivitiesbecauseofphysicalhealth(Ittookextratime)

5a Rolelimitationsdueto Cutdowntheamountoftimespentonworkorotheractivitiesbecauseof

25

emotionalproblems emotionalproblems5b Rolelimitationsdueto

emotionalproblemsAccomplishedlessthanyouwouldlikebecauseofemotionalproblems

5c Rolelimitationsduetoemotionalproblems

Didworkorotheractivitieslesscarefullythanusualbecauseofemotionalproblems

6 Socialfunctioning Extenthealthproblemsinterferedwithnormalsocialactivities7 Bodilypain Intensityofbodilypain8 Bodilypain Extentpaininterferedwithnormalwork9a Vitality Feelfulloflife9b Mentalhealth Beenverynervous9c Mentalhealth Feltsodowninthedumpsthatnothingcouldcheerup9d Mentalhealth Feltcalmandpeaceful9e Vitality Havealotofenergy9f Mentalhealth Feltdownheartedanddepressed9g Vitality Feelwornout9h Mentalhealth Beenhappy9i Vitality Feeltired10 Socialfunctioning Frequencyhealthproblemsinterferedwithnormalsocialactivities11a Generalperceptionofhealth Seemtogetsickalittleeasierthanotherpeople11b Generalperceptionofhealth AshealthyasanybodyIknow11c Generalperceptionofhealth Expectmyhealthtogetworse11d Generalperceptionofhealth Healthisexcellent

AdaptedfromWareJEJr,KosinskiM,BjornerJB,Turner-BowkerDM,GandekB,MaruishME(2007).User’smanualfortheSF-36v2®Health Survey (2nd ed.). Lincoln, RI:QualityMetric Incorporated. Optum’s Table Abbreviated ItemContent for the SF-36v2®HealthSurveyHealthDomainScale,Table2.1,page15.

2.3Qualityofliferesearchinthecriticallyillpatient

QOLresearchstudiestheeffectsoftreatmentsonendpointsimportanttothepatient.Thegoalof

QOLresearchisnotonlytodiscriminatebetweenwhohasagoodorworseQOLatlong-termbutalsoto

evaluatehowQOLwillchangeovertime.AlthoughQOLhasnowbeenacceptedtobevaluableregarding

outcome,itisstillnotroutinelyincludedinstudies[48].Thishasmanyreasons.

Firstly,assessingQOLwithspecificquestionnaires ismore labour intensiveand timeconsuming

andwillalwaysbemoreambiguousfor interpretationthantheunequivocal“death”or“alive”outcome

binaryparameter,whichhastheadvantageofbeingunambiguousandveryeasytomeasure.Ascritical

carephysicians,wearenotveryfamiliarwithhandlingsuchapersonalandsubjectiveparameterasQOL.

QOLdependsonalotofdifferentissuesandwillnotonlydifferfrompatienttopatientbutalsofromthe

timepointofassessmentwithinthesamepatient.

Secondly, as QOL incorporates a patient’ personal values and preferences, QOL questionnaires

should ideallyonlybeansweredby thepatienthimself at everyQOLassessment timepoint.However,

manyICUpatientscannotcompleteQOLquestionnairesbecausetheyaretooill,tooweak,tooconfused,

orsedated.Askingthepatient tocompleteQOLsurveysafter the ICUadmissionholds theriskof recall

bias[17,49,50].Yet,proxiescancompleteQOLquestionnairesonbehalfofthepatient.Theycanprovide

a reasonably accurate estimate ofQOL of ICU patients, although they tend to underestimateQOL but

differencesareusuallysmallandnotclinicallyimportant[17,49-52].Theemotionaldimensionsseemto

26

beassessedlessaccuratelyandareoftenunderestimatedbyproxiescomparedtothephysicalones,that

are frequently overestimated,whichmeans that relatives tend to think that a patient has lessmental

powerandabetterphysicalhealththanthepatientactuallyhas[50-54].Nevertheless,QOLassessments

byproxiesatanytime,evenwithpossibleinherentsmallunder-oroverestimations,couldbeconsidered

asmoreimportantandmoreinformativethannoQOLassessmentatall.

Thirdly,whenQOLmeasuresareusedasdiscriminativeinstruments(whohasagoodandwhohas

apoorQOL?),possibleconfounders,whichcouldinfluenceQOL,shouldbeeliminated.Therefore,QOLin

ICUpatientscanbecomparedtoanage-andgender-matchedgeneralpopulation.Thestudyfindingscan

also be compared with an appropriate control group eliminating the influence of specific health

conditions.Moreimportant,long-termQOLshouldalsobecomparedwithQOLbeforeICUadmission,to

discriminate whether poor long-term QOL is a result of the severity of illness, or due to confounding

factors such as co-morbid disease, poor pre-admission QOL, age, gender, or acquired complications.

Baseline assessment of QOL (=QOL 2 weeks before ICU admission) is difficult but of great value to

examine the true burden of the critical illness. Evidence for poorer health status among patients

dischargedfromtheICUmaybemisleadingifthepriorhealthstatusoftheICUpatientisnottakeninto

account[49].Inthatcase,itwillbedifficulttomakehonestcomparisonsortodrawstrongconclusionsas

theimpactofthecriticalillnessmaybelargeandmaylastforalongtime.Wewillhoweverneverbeable

toseparatetheacuteillnessfromthepredispositiontotheacuteillness.

Fourthly, long-term outcomes and QOL research will always be observational. The prospective

observationalcohortstudyisthereforethemostpowerfulresearchdesigntomaximizetheimpactofthis

kind of research [55]. This should be coupled with the need to examine data longitudinally without

optimal time intervals for measurement of long-term QOL being known or defined [22, 39]. A very

complete picture of outcomes after critical care might require a long follow-up period, and one can

wonder when QOL measures will no longer give additional information. The shorter the follow-up

intervalsforQOLassessments,thelessinformativeresultswillbeandthehighertheriskof“assessment-

fatigue”inpatients.Thelongerthefollow-upperiodshowever,thehighertheriskthatmorepatientswill

belosttofollow-up,whichcouldleadtoimportantbiasofthestudyresults.Whileoptimaltimeintervals

for QOL assessments are not known, it is important to keep these intervals as strict and uniform as

possiblesoevolutionsinQOLovertimebetweendifferentpatientscanbeevaluatedinacorrectway.

Fifthly,toassessQOLovertime,itisnecessarilytotrackpatientsaftertheyaredischargedfrom

theICUandfromthehospital.Thiscanbedifficultandislabourintensive.ValidatedQOLquestionnaires

canaccessQOLbyface-to-faceinterview,byphone,orbyregularmail.Althoughahighresponserateto

QOL questionnaires is the aim of every QOL study, there will always be non-responders. If this is a

numerous group, it is important to describe these non-responders to find out if the reason for non-

respondingcanbeclarified[56].

27

Sixthly,evaluationsof long-termQOLalways implysurvivalbiasasQOLcanonlybeassessed in

survivors[57].Itshouldbeacknowledgedthatlong-termQOLmightbemodifiedbyeventshappeningto

thepatientafterhospitaldischarge.

Seventhly, the increasing interest in patients’ perceptions of health status has led to huge

variations in appliedmethodology tomeasure functional status andQOL,whichhampers theability to

compareresultsordrawstrongconclusionsoutofoutcomeresearch.AsQOLisasubjectiveparameterby

itself,itshouldthereforealwaysbepreferredtousestandardizedQOLquestionnaires,whichhaveawell-

knownvalidity,reliability,andareresponsivetorealchangesinhealth[22].Asalreadybeensaid,there

are no uniformly 'worst' or 'best' performing generic instruments and the decision to use one over

anotherortouseacombinationof2ormore,willdependuponthecharacteristicsofthepopulationand

the environment inwhich themeasurement is undertaken [35]. However, It should be encouraged to

uniformoutcome research so thatQOLevaluations couldbeeasier compared andplaced in a broader

perspective.

3.CostsandOutcomeStudyintheICU(COSIstudy)

3.1Design,setting,patients,andQOLassessments

With the knowledge that QOL is a subjective parameter, that we have to use standardized

questionnaires, preferentially completed by the patient, that baseline QOL should be assessed, that it

would be difficult to track patient at long-term, that there will be a survival bias and that outcome

researchimpliedanobservationalstudydesign,weperformedaprospectiveobservationalcohortstudy

inwhichwe,duringaone-yearperiod (March3rd2008 -March3rd2009) includedalladult (≥16years)

criticallyillpatientsconsecutivelyadmittedtothe14-bedmedicaland22-bedsurgicalICUandthe6-bed

burn unit of the Ghent University Hospital, Ghent, Belgium. Our main purpose was to gain data

concerning long-term outcomes and QOL in our own critically ill patient population. Within the total

patient cohort, we also predefined some subgroups namely patients admitted to the ICU due to

oncologicalorhematologicaldisease,patientswith livercirrhosisChild-PughBorC,patientsdeveloping

acutekidney injury (AKI)withneed for renal replacement therapy (RRT),patientswithaprolonged ICU

lengthofstay(LOS)(≥8days)orolder(≥80years)patients.

Incaseofmultiple ICUadmissionsduringthesamehospitalizationperiod,weonly includedthe

first admission.Wedid not include cardiac surgery patients as these patients represent a very specific

patientpopulation,whereICUadmissionisneededmainlyafterelectivecardiacsurgery,whereICUstays

areoftenshort,andwhereQOLat long-term is likelytobeverygood[58].Duetothehighturnover in

thesepatients, theywouldotherwisehavebecomethemajorpatientgroup inourstudycohort,which

couldhaveledtobiasinourmainstudyresults.

28

Thestudywasapprovedbythelocalethicalcommittee(EthischComitéGhentUniversityHospital;

project2007/423approved06December2007)(B67020072805),andconductedinaccordancewiththe

declarationofHelsinki.Asignedinformedconsentwasobtainedfromeveryincludedpatientorhislegal

representative.

QOLwasassessedusingtheEQ-5D(andcognitivefunctionassessment)andSF-36at3predefined

timepoints:baselineQOL,3monthsand1yearafter ICUdischarge.AcomputerfilewithICUdischarge

data for each included patient was created in order to respect in an accurateway the time points of

second (exactly 3 months after ICU discharge) and third (exactly 1 year after ICU discharge) QOL

assessment.

Following ICU admission and study inclusion, a face-to-face interview to assess baseline QOL

(defined as QOL 2 weeks before ICU admission) was done as soon as possible. This interview was

preferably taken from the patient, or, whenever impossible due to severity of illness, from the proxy.

Threemonthsand1yearafterICUdischarge,patientsorrelativesweresenttheEQ-5DandSF-36surveys

byregularmail,aftercheckingtheirlivingstatusandaddressthroughthehospitalcomputersystem.The

envelope contained the two questionnaires, and also a pre-addressed envelope with stamp and a

ballpointpen.At1year,questionsconcerning livingsituationof thepatient,memoriesof the ICUstay,

actual sleep disturbances, and if the patientwaswilling to be admitted to an ICUdepartment again if

needed, were added. If the questionnaires were not returnedwithin onemonth, patients or relatives

were contacted by phone to assess QOL. This was only done at the third time point. If there was no

contactbyphone,thefamilypractitionerwascontactedtoassessthelivingstatusofthepatient.

ItisimportanttonoticethatweanalyzedQOLinsurvivorsovertime.Therefore,thepopulationat

eachfollow-upintervalrepresentedadifferentsubsetoftheinitialcohort.

29

3.2Flowchartofincludedpatients,numberofQOLsurveysandoutcomesinthetotalcohort

(a)=151noinformedconsent:97refusals,40languageproblems,14socialreasons;(b)=21excluded:11refusals,3living abroad, 7 language problems; (c)= 17 excluded: 3 refusals, 7 living abroad, 7 language problems; (d)= 2excluded:1 refusal,1mentalproblem; (e)=13excluded:7 refusals,3 livingabroad,3 languageproblems; (f)=13excluded: 4 refusals, 1 living abroad, 8 language problems; ICU= intensive care unit; SICU= surgical ICU; MICU=medicalICU;CSICU=cardiacsurgeryICU;N=number;QOL=qualityoflife;IC=informedconsent

30

3.3DataCollection

Data collected within the first 24 hours of ICU admission included contact information of the

patient, proxy, andgeneralpractitioner, demographics, hospital daysprior to ICUadmission, livingand

workcircumstancesbeforeICUadmission,functionalityasmeasuredbytheKatzactivitiesofdaily living

(ADL) scale [59, 60], hospitalization in the last 6 months, comorbidity as measured by the Charlson

comorbidity index [61], main ICU admission reason and diagnosis, admission circumstances (planned-

unplanned/duringweekendor not), if the patient belonged to 1 ormore of the predefined subgroups

(sub)(oncological,hematological,livercirrhosisChild-PughBorC,patientsdevelopingAKIwithneedfor

RRT, patientswith a prolonged ICU-LOS (≥8 days) or older patients (≥ 80 years)), APACHE II score [2],

SequentialOrganFailureAssessment(SOFA)score[62],TherapeuticInterventionScoringSystem-28score

(TISS-28 score) [63], Nine Equivalent of Nursing Manpower Use score (NEMS-score) [64], do-not-

resuscitate (DNR) codes, need for invasivemechanical ventilation, vasopressors, RRT, medical imaging

(regardless of number and other than chest X-ray or ultrasound examinations), transfusionwith blood

products,surgery,ortracheotomy.Foreachincludedpatient,wealsocollectedallICUandhospitaldirect

costs.

During ICU stay SOFA, TISS-28 and NEMS-scores, DNR-codes, need for invasive mechanical

ventilation, vasopressors, RRT, medical imaging (regardless of number and other than chest X-ray or

ultrasoundexaminations),transfusion,surgery,ortracheotomywerecollectedonadailybase.

ICU-LOS, hospital-LOS, vital status at ICU and hospital discharge, and 1 year following ICU

dischargewerecollectedforeachpatient.Dependingonthesubstudy,wealsoassessedvitalstatusand

QOLatlongerterms.

31

II.Researchquestions

1.Aimandoutline

Drawing strong conclusions from a large case-mix of very heterogeneous medical, surgical, or

burnedcriticallyillpatientsisdifficult.PresentingQOLresultsforICUpatientsasawholemayobscurethe

factthatsometypesofpatientimprovewhilstothersremainstableordeteriorate[49].Amoreaccurate

pictureofICUoutcomesmightbeobtainedifthediagnosticcategoryistakenintoaccount,aspriorhealth

status,whichinfluencesQOL,hasbeenshowntovaryacrosssuchcategories[39,49,65].AssessingQOL

inmorespecificpatientgroupswillthereforeresultinmorerefineddata.Though,thenumberofpatients

inspecificdiagnosticpatientgroupswillbeinherentsmaller.

Nevertheless,accordingtoourmainstudygoal,wechosetoassesslong-termoutcomesandQOL

withinspecificpatientsubgroupsofourlargeCOSIstudycohortwherethereareoftendoubtsconsidering

effectivenessofcritical careorwhere thestartof specificexpensive treatmentsduring ICUstaycanbe

questioned, namely the oncological/hematological patients, the older patients (≥ 80 years), and the

patientswithneedforRRTduetoAKIdevelopedduringtheircriticalillness.

Asmore andmore critically ill patients – even in these specific andoften controversial patient

groups -nowadayssurvive theircritical illness; it is forcritical carephysiciansvery important tohavea

better understanding of how critical care affects the long-termhealth andQOL of its survivors. Better

knowledgeandinsightsoflong-termoutcomeswillhelpphysicianstoidentifywhowillbenefitthemost

fromICUadmissionwhendecidingonallocationtherapeuticeffortsinthefuture,andcanhelpinabetter

andefficientadvancedcareplanningandcommunicationwithpatientandfamily.

Thefocusofourresearchconcentratedthereforearound3major issues:1/reviewingliterature

concerning long-term QOL, reviewing applied methodology and quality of this published outcome

research,2/assessing long-termoutcomesandQOL in specificcritically illpatientwhere theadditional

benefitofcriticalcareisfrequentlyquestioned,and3/developingapredictionmodelforlong-termQOL

based upon readily available variables at the first day of ICU admission and so determining themost

important predictors for long-termQOL. Five specific research questions addressing these topicswere

formulated.

2.Specificresearchquestions

2.1.What is already known in the literature concerning long-term outcomes andQOL in critically ill

patients?Canweformulatemethodologicalrecommendationsforfurtherresearchonthistopic?

In this first study, it was our purpose to give a systematic review of the literature, published

betweenJanuary,1th1999andDecember,31th2009,ofQOLanditsinfluencingfactors,atleastoneyear

afterdischargefromtheICU,andofthemethodologyused.

32

AsearchthroughEMBASE-PubMed,MEDLINE(OVID),SCI/WebofScience,CochraneLibrary,and

GoogleScholarwasdoneonJanuary9,2010usingthemedicalsubjectheadings(MeSH)ortextkeywords

“qualityoflife”,or“long-termoutcome”crossreferencedwith“intensivecare”,“criticalcare”,“critically

ill patients”, “ICU patients”, “critical care patients”, “ICU stay”, or “ICU”. Limitations were applied

regarding language (only English language), time (articles published within the 10-years interval), age

(above 18 years), and humans. Only studies using SF-36, RAND-36, EQ-5D, and NHPwere considered.

Outcomes articles including exclusively cardiac or thoracic aortic surgery patients, methodological

articles, literaturereviews,case-reports,editorials,and letterswereexcluded.Studieswith lessthan50

patientswerealsonotincluded.

Foreacheligiblearticle,informationwasextractedonauthors,journal,yearofpublication,study

design,inclusionperiod,initialstudycohort,baselinevariablesandoutcome,numberofeligiblepatients

for long-termQOL assessment, instrument(s) andmethod(s) used for QOL assessment, response rate,

follow-up period, the use of other questionnaires or tests, the final conclusion concerning QOL, and

factorsdeterminingQOL.Studyqualitywasassessedusingfourcriteria:1)QOLassessmentpriortoICU

admission, 2) description of key inclusion or exclusion criteria, 3) description of non-responders and

comparison with those remaining in the study, and 4) adjustment for confounders such as age and

gender.We hopedwith this review to gain and give better insights into long-termQOL, and tomake

methodologicalrecommendationsforfurtherresearchonthistopic.

2.2. What is the long-term outcome and QOL of critically ill patients with a hematological or solid

malignancy?What istheevolutionofQOLovertimeinthesepatientscomparedtobaseline?Canwe

identifyprognosticindicatorsfortheevolutionofQOLafterICUdischarge?

The prognosis of patientswith a solid or hematologicalmalignancy has substantially improved

over thepastdecadesdue toadvances indiagnostics, antineoplastic therapyand supportive care [66].

Additionally,survivalofcancerpatientsdevelopingcriticalillness[66-68]hasincreasedaswell,including

thoserequiringmechanicalventilation [69]orRRT [70,71].Adiagnosisofcancershouldthereforenot

precludeICUadmission,asitistheseverityoftheacuteillnessthatwilldetermineshort-termmortality,

ratherthantheunderlyingcancercharacteristics[72-74].

In our review study, we demonstrated that major reductions in long-term QOL were seen in

critically ill patientswith severeacute respiratorydistress syndrome,prolongedmechanical ventilation,

and severe sepsis, all representing complications that affect cancer patients as much as non-cancer

patients [75]. Inaddition,poorperformance following ICUadmission incancerpatientsmay jeopardize

long-termoutcomebyinducingpostponementsorcancellationsofpotentiallycurativechemotherapy.So,

to fully appreciate outcomes of critically ill cancer patients, a better knowledge and insights regarding

long-termmorbidityandQOLafterICUdischargeisnecessarily.

33

Inordertoevaluatelong-termoutcomes,QOL,andevolutioninQOLofcriticallyillpatientswitha

hematologicalorsolidmalignancy,wefollowedtheCOSIstudydesignwithQOLassessmentsusingEQ-5D

andSF-36at the3different timepoints (baseline,3monthsandafter1 yearafter ICUdischarge), and

withadditionalquestionsafter1year.PrognosticindicatorsforapoorQOLat3monthsand1yearwere

formulated. Only patients of the COSI cohort with a solid or hematological malignancy as direct or

contributivecauseforICUadmissionwereincluded.Patientswithcompleteremissionfor>5yearswere

excluded.

2.3.Whatistheimpactofrenalreplacementtherapy(RRT)onlong-termoutcomeandQOLincritically

illpatientsdevelopingacutekidneyinjury(AKI)withneedforRRTduringICUstay?

Approximately 5-10%of critically ill patientswill developAKIwith need forRRT (AKI-RRT) [76].

ThesepatientsareamongstthemostseverelyillpatientsintheICU,asmaybeillustratedbythe50%in-

hospitalmortality[77-79].DecisionswhetherornottostartRRTarenoteasytomakeastheconsequence

towithholdthistherapywillleadinmanycasetothedeathofthepatient.AKI-RRTpatientswhosurvive

maydevelopchronickidneydisease,andexperiencedecreasedlong-termsurvival[79-82].Dataregarding

long-termQOL in AKI-RRT survivors show that these patients have a decreasedQOL compared to the

generalpopulationbutperceiveQOLasgood[83,84].However,thesedatawereretrospective[85-87],

evaluatedonly short-termQOL [83-90], lackedbaselineQOLassessment [83-86,88,91],ordatedback

morethanadecade[85,86,88,92].

TostudytheimpactofRRTonlong-termoutcomeandQOL,wethereforeperformedamatched

cohort study, according to the STROBE guidelines [93]. Included patientswereAKI-RRT patients of the

COSIcohort,aliveat1yearafterhospitaldischarge,whowereindividuallymatchedwith1-yearnon-AKI-

RRTsurvivorsfromthesamecohort.Equally,AKI-RRTpatientsaliveattimeofthestudy(average4years

later)wereindividuallymatchedwith4-yearnon-AKI-RRTsurvivors.Matchingwasbasedongender,age

(±5years),APACHEIIscore(±5),andadmissioncategory.Chronichemodialysispatientsandpatientswho

neededRRTbutwhodidnotreceiveRRTduetotherapeuticrestrictionswereexcluded.

2.4.Whatisthelong-termoutcomeandQOLofcriticallyillolderpatients(aged≥80years)?Whatisthe

evolution of QOL over time in these older patients compared to baseline? How do older survivors

perceivetheirlong-termQOL?Howaretheirpost-hospitaltrajectories?

Survival to older age has increased, which leads to more hospitalizations and more ICU

admissionsforolderpatients[57,94].Asprognosisofcriticallyillpatientsaged80ormoremaybepoor,

especially in those with severe comorbidity, or a greater illness severity, concerns may rise regarding

utilityor futilityofhigh-levelexpensive ICUtreatments forthesepatients [57,94-98].To identifywhich

34

critically illolderpatientwouldbenefit themost fromICUadmission, long-termoutcomesandQOLare

importantissuestobetakenintoaccount.

However, recent data regarding long-term QOL in critically ill older patients are still

limited[95-101].Studiesareeitherretrospective[95,102],evaluateonlyshort-termQOL[96,101,102],

lack baselineQOL evaluation [95, 96, 99], assessQOL after variable follow-up intervals [95], or define

olderpatientsaspatientsaged65yearsormoreorevenyounger[96,97,100,103].Inordertoevaluate

long-termoutcomes,QOL,andevolution inQOL inourcritically illolder(≥80years)patientpopulation,

we followed the COSI study design with QOL assessments using EQ-5D and SF-36 at 4 different time

points(baseline,3months,1yearand7yearsafterICUdischarge),andwithadditionalquestionsafter1

and7years.OnlypatientsoftheCOSIcohortwhowereatleast80yearsatICUadmissionwereincluded.

Older patients often perceive a worsening in long-term QOL but still evaluate their QOL as

acceptable [95-97,101-103]. It suggests thatQOLmighthaveanothermeaning forolderpatients,with

social andmental valuesbeing farmore important than limitedphysical functioningand thatage itself

influences QOL mainly due to increasing number of chronic conditions [96, 104]. We therefore also

determinedperceivedQOLperpatientbycomputingchangesbetweenthe3consecutivetimeintervals

(before ICU admission-3 months; 3 months-1 year; 1 year-7 years). These changes in QOL were

consideredclinically important ifpatients reportedanother level for thedifferentEQ-5Ddimensionsor

forthehealthtransitionoftheSF-36,oriftherewasaminimumdifferenceof7pointsintheEQ-VASor5

pointsinthenorm-basedphysicalandmentalcomponentmeasuresoftheSF-36[105].

Notonly long-termoutcomesandQOLareimportantissuestoconsiderwhendecidingtoadmit

olderpatientstotheICU,butcriticalcarephysiciansshouldconsiderthewholediseaseprocesstheolder

has toendure [106].Therefore,wealsoevaluatedposthospital trajectories incritically illolderpatients

whosurvivedtohospitaldischargetogainbetterinsightsinthefurthercourseofthediseaseandinthe

recoveryphase.

2.5. Can we predict long-term QOL based upon variables readily available at the first day of ICU

admission?

The true burden of a critical illness and its long-term consequences on physical, mental and

cognitive functioningmay be underestimated [107, 108], aswell as the possibility to return to former

daily lifeandQOL[109]. It isthe importanttaskofcriticalcarephysiciansto informcritically illpatients

andtheirfamily inareliablewayabouttheseoutcomes.However, forcriticalcarephysicianstoo, long-

termfunctionalityandQOLremaindifficulttopredict[110,111].

Accuratepredictionmodelscanguidephysiciansintheirhandling,communication,anddecision-

making. However, some existing prediction models are not applicable to a broad critically ill patient

population [112-117], are rather complex [118, 119], or not accurate enough [120]. Some focused on

35

long-term mortality [112, 121], or long-term functionality [113], but none of the existing prediction

modelsestimatedlong-termQOLingeneralcriticallyillpatients.

Therefore,itwasouraimtoretrospectivelydevelopaneasytouseandaccuratepredictionmodel

forthemeanQOLat1yearafterICUdischargeingeneralcriticallyillpatientsbasedupondataoftheCOSI

studyreadilyavailableatthefirstdayofICUadmission(=first24hoursofICUstay=D1).

ThehealthstatesassessedbythefirstpartoftheEQ-5D(the5-digitnumber)atbaselineandat1

year were converted into the corresponding UI at baseline (UIb) and UI at 1 year after ICU discharge

(UI1y) [45]. These were used as surrogate for QOL at that time point. VASb and VAS1y expressed

perceivedQOLatbaselineand1yearafterICUdischarge.UI1yandVAS1yfornon-survivorsweresetat

zero to avoid survival bias. QOL assessments 3 months after ICU discharge were not included in the

developmentoftheD1-modelduetotoomanymissingdataatthattimepoint.

For the development of theD1-predictionmodel, three differentmultivariate linear regression

models,respectivelyModelI,II,andIII,werefittedwithUI1yasprimaryoutcome.ModelIassessedthe

bivariate association between UIb and UI1y. Model II (“full” model) included all possible available D1

predictors in the linear regressionanalysis.Model III (“reduced”model) includedonlypredictors in the

linear regression, whichwere selected by the grouped lasso technique. This techniquewas applied to

identifytheoptimalnumberandmostimportantpredictorsforUI1yintheD1linearregressionmodelin

ordertosimplifythemodel,andtocopewiththecategoricalvariables[122,123]asitallowspredefined

groupsofcovariates,suchasallvariablesencodingacategoricalcovariate,tobeselectedintooroutofa

modeltogether.

Onlycompletecases,definedaspatientsincludedintheCOSIcohortwithoutmissingdata,were

included inthestatisticalanalysis.Themodelwiththebestpredictivecapability for themeanQOLat1

yearafterICUdischargewasselectedasD1-predictionmodel.

36

III.References

1. IbsenB.Theanaesthetist’sviewpointonthetreatmentofrespiratorycomplicationsinpoliomyelitisduringtheepidemicinCopenhagen,152.ProcRSocMed1954;47:72-742. KnausWA,DraperEA,WagnerDP,ZimmermanJE.APACHEII:aseverityofdiseaseclassificationsystem.CritCareMed1985;13:818-8293. ZarenB,HedstrandU.Qualityoflifeoflong-termsurvivorsofintensivecare.CritCareMed1987;15:743-7474. RidleySA,WallacePG.Qualityoflifeafterintensivecare.Anaesthesia1990;40:808-8135. SuterP,ArmaganidisA,BeaufilsF,BonfillX,BurchardiH,CookDetal.PredictingoutcomeinICUpatients.Consensusconferenceinintensivecaremedicine.IntensiveCareMed1994;20:390-3976. GeigleR,JonesSB.Outcomesmeasurement:areportfromthefront.Inquiry1990;27:7-137. Lembcke PA. Measuring the quality of medical care through vital statistics based on hospital service areas; I.Comparativestudyofappendectomyrates.AmJPublicHealthNationsHealth1952;42:276-868. RubenfeldGD,AngusDC,PinskyMR,CurtisJR,ConnorsAF,BernardGR,andthemembersoftheOutcomesResearchWorkshop.Outcomesresearchincriticalcare.AmJRespCritCareMed1999;160:358-3679. AngusDC, Carlet J, 2002Brussels Roundtable Participants Surviving Intensive Care: A report from the 2002BrusselsRoundtable.IntensiveCareMed2003;29:368-37710. Eisner MD, Thompson T, Hudson LD, Luce JM, Hayden D, Schoenfeld D, Matthay MA; Acute Respiratory DistressSyndromeNetwork.Efficacyoflowtidalvolumeventilationinpatientswithdifferentclinicalriskfactorsforacutelunginjuryandtheacuterespiratorystresssyndrome.AmJRespirCritCareMed2001;15:231-23611. VanDenBergheG,WoutersP,WeekersF,VerwaestC,BruyninckxF,SchetzM,VlasselaersD,FerdinandeP,LauwersP,BouillonR.Intensiveinsulintherapyincriticallyillpatients.NEnglJMed2001;8:1359-136712. RiversE,NguyenB,HavstadS,ResslerJ,MuzzinA,KnoblichB,PetersonE,TomlanovichM;EarlyGoal-DirectedTherapyCollaborativeGroup.Earlygoal-directedtherapyinthetreatmentofseveresepsisandsepticshock.NEnglJMed2001;8:1368-137713. AnnaneD, Sebille V, Charpentier C, Bollaert PE, Francois B, Korach JM, et al. Effect of treatmentwith low doses ofhydrocortisoneandfludrocortisoneonmortalityinpatientswithsepticshock.JAMA2002;288:862-87114. FlaattenHans.Long-termoutcomesafterintensivecare.25Yearsofprogressaninnovationinintensivecaremedicine.Ed.MedizinischWissenschaftlicheVerlagsgesellschaftBerlin2007;419-42715. Kaukonen KM, BaileyM, Suzuki S, Pilcher D, Bellomo R.Mortality related to severe sepsis and septic shock amongcriticallyillpatientsinAustraliaandNewZealand,2000-2012.JAMA2014;311:138-131616. AmatoMB,MeadeMO, Slutsky AS, Brochard L, Costa EL, Schoenfeld DA, Stewart TE, BrielM, Talmor D,Mercat A,RichardJC,CarvalhoCR,BrowerRG.Drivingpressureandsurvivalintheacuterespiratorydisresssyndrome.NEngJMed2015;372:747-75517. Granja C, Azevedo LF.When (quality) of life is at stake and intensive care Is needed: howmuch can we trust ourproxies?IntensiveCareMed2006;32:1681-168218. HeylandDK,GuyattG,CookDJ,MeadeM,JuniperE,CroninL,GafniA.Frequencyandmethodologicrigorofquality-of-lifeassessmentsinthecriticalcareliterature.CritCareMed1998;26:591-59819. HeylandDK,KutsogiannisDJ.Qualityof life followingcriticalcare:movingbeyondsurvival. IntensiveCareMed2000;26:1172-117520. WHOdefinitionQOL-http://www.who.int/healthinfo/survey/whoqol-qualityoflife/en/(lastassessedApril2018)21. WikipediadefinitionQOL-https://en.wikipedia.org/wiki/Quality_of_life(lastassessedApril2018)

37

22. BlackNA,JenkinsonC,HayesJA,YoungD,VellaK,RowanK,DalyK,RidleyS.Reviewofoutcomemeasuresusedinadultcriticalcare.CritCareMed2001;29:2119-212

23. Hunt SM,McKenna SP,McEwen J,Williams J, Papp E: The Nottingham Health Profile: subjective health status andmedicalconsultations.SocSciMed1981;15A:221-22924. BergnerM,BobbittRA,CarterWB,GilsonBS.TheSickness ImpactProfile:developmentandfinalrevisionofahealthstatemeasure.MedCare1981;19:787-80525. deBruinAF,deWitteLP,StevensF,Diederiks JP.Sickness ImpactProfile: thestateof theartofageneric functionalstatusmeasure.SocSciMed1992;35:1003-101426. KaplanRM,Bush JW,BerryCC.Healthstatus:Typesofvalidity foran IndexofWell-being.HealthServRes1976;11:478-50727. TheEuroQolGroup.EuroQol–anewfacilityforthemeasurementofhealth-relatedqualityoflife.HealthPolicy1990;16:199–208

28. BrooksR.EuroQol:thecurrentstateofplay.HealthPolicy1996;37:53-72

29. HaysRD,SherbourneCD,MazelR.TheRAND-36HealthSurvey1.0.HealthEcon1993;2:217-27730. Ware JE, Jr., Sherbourne CD. TheMOS 36-item short-formhealth survey (SF-36). I. Conceptual framework and itemselection.MedCare1992;30:473-48331. McHorneyCA,WareJEJr,RaczekAE.TheMOS36-itemshort-formhealthsurvey(SF-36). II.Psychometricandclinicaltestsofvalidityinmeasuringphysicalandmentalconstructs.MedCare1993;31:247-26332. McHorneyCA,WareJEJr,LuJF,SherbourneCD.TheMOS36-itemshort-formhealthsurvey(SF-36). III.Testsofdataquality,scalingassumptions,andreliabilityacrossdiversepatientgroups.MedCare1994;32:40-6633. Ware JE,KosinskiM,Bjorner JB, Turner-BowkerDM,GandekB,MaruishME.User’sManual for theSF-36v2®HealthSurvey.QualityMetricIncorporated,Lincoln,RhodeIsland;200734. Chrispin PS, Scotton H, Rogers J, Lloyd D, Ridley SA. Short-Form 36 in the intensive care unit: Assessment ofacceptability,reliabilityandvalidityofthequestionnaire.Anaesthesia1997;52:15-2335. Coons SJ, Rao S, Keininger DL, Hays RD. A comparative review of generic quality-of-life-instruments.Pharmacoeconomics2000;17:13-3536. KaarlolaA,PettiläV,KekkiP.Performanceoftwomeasuresofgeneralhealth-relatedqualityoflife,theEQ-5DandtheRAND-36amongcriticallyillpatients.IntensiveCareMed2004;30:2245-225237. Brazier J, Jones N, Kind P. Testing the validity of the Euroqol and comparing it with the SF-36 health surveyquestionnaire.QualLifeRes1993;2:169-18038. BrazierJE,WaltersSJ,Nicholl JP,KohlerB.UsingtheSF-36andEuroqolonanelderlypopulation.QualLifeRes1996;5:195-20439. GranjaC,Teixeira-PintoA,Costa-PereiraA.Qualityof lifeafter intensivecare–evaluationwithEQ-5Dquestionnaire.IntensiveCareMed2002;28:898-90740. PatrickD,BushJ,ChenM.Methodsformeasuringlevelsofwell-beingforahealthstatusindex.HealthServRes1973;8:228-24541. SelbyPJ,ChapmanJA,Etazadi-AmoliJ,DalleyD,BoydNF.Thedevelopmentofamethodforassessingthequalityoflifeincancerpatients.BrJCancer1984;50:13-2242. Hjermstad MJ, Favers PM, Haugen DF, Caraceni A, Hanks GW, Loge JH, et al. European Palliative Care ResearchCollaborative (EPCRC). Studies comparing Numerical Rating Scales, Verbal Rating Scales, and Visual Analogue Scales forassessmentofpainintensityinadults:asystematicliteraturereview.JPainSymptomManage2011;41:1073-109343. deBoerAG,vanLanschotJJ,StalmeierPF,vanSandickJW,HulscherJB,deHaesJC,SprangersMA.Isasingle-itemvisualanaloguescaleasvalid,reliableandresponsiveasmulti-itemsscalesinmeasuringqualityoflife?QualLifeRes2004;13:311-320

38

44. CleemputI.AsocialpreferencevaluationssetforEQ-5DhealthstatesinFlanders,Belgium.EurJHealthEcon2010;11:205-21345. LievenAnnemans.Gezondsheidseconomievoorniet-economen.1sted.Ghent:AcademiaPress;200746. HerdmanM,GudexC,LloydA,JanssenMF,KindP,ParkinD,BonselG,BadiaX.Developmentandpreliminarytestingofthenewfive-levelversionofEQ-5D(EQ-5D-5L).QualLifeRes2011;20:1727-173647. TarlovAR,Ware JE Jr,Greenfield S,Nelson EC, Perrin E, ZubkoffM. TheMedicalOutcomes Study.An applicationofmethodsformonitoringtheresultsofmedicalcare.JAMA1989;18;262:925-3048. TurnbullAE,RabieeA,DavisWE,FarhanNasserM,ReddyVennaV,LolithaR,HopkinsRO,etal.OutcomemeasurementinICUsurvivorshipresearchfrom1970to2013:Ascopingreviewof425publications.CritCareMed2016;44:1267-127749. BadiaX,Diaz-PrietoA,GorrizMT,et al.Using theEuroQol-5D tomeasure changes inqualityof life12monthsafterdischargefromanintensivecareunit.IntensiveCareMed2001;27:1901-190750. ScalesDC,TanseyCM,MatteA,HerridgeMS.Differenceinreportedpre-morbidhealth-relatedqualityolifebetweenARDSsurvivorsandtheirsubstitutedecisionmakers.IntensiveCareMed2006;32:1826-183151. Hofhuis J, Hautvast JL, Schrijvers AJ, Bakker J. Quality of life on admission to the intensive care: can we query therelatives?IntensiveCareMed2003;29:974-97952. DinglasVD,GiffordJM,HusainN,ColantuoniE,NeedhamDM.QualityoflifebeforeintensivecareusingEQ-5D:Patientversusproxyresponses.CritCareMed2013;41:9-1453. CapuzzoM,GrasselliC,CarrerS,GrittiG,AlvisiR.Qualityoflifebeforeintensivecareadmission:agreementbetweenpatientandrelativeassessment.IntensiveCareMed2000;26:1288-129554. Gifford JM, Husain N, Dinglas VD, Colantuoni E, Needham DM. Baseline quality of life before intensive care: acomparisonofpatientversusproxyresponses.CritCareMed2010;38:855-86055. DowdyDW,NeedhamDM,Mendez-Tellez PA, HerridgeMS, Pronovost PJ. Studying outcomes of intensive care unitsurvivors:theroleofthecohortstudy.IntensiveCareMed2005;31:914-92156. vonElmE,AltmanDG,EggerM,PoccockSJ,GøtzschePC,VandenbrouckeJP.STROBEinitiative:TheStrengtheningtheReportingofObservationalStudiesinEpidemiologystatement:Guidelinesforreportingobservationalstudies.JClinEpidemiol2008;61:344-34957. Conti M, Merlani P, Ricou B. Prognosis and quality of life of elderly patients after intensive care. Swiss MedWkly2012;142:w13671.58. Soliman IW,deLangeDW,PeelenLM,CremerOL,SlooterAJ,PasmaW,KeseciogluJ,vanDijkD.Single-center large-cohort study intoqualityof life inDutch intensivecareunit subgroups,1yearafteradmission,usingEuroQoLEQ-6D-3L. JCritCare2015;30:181-18659. KatzS,DownsTD,CashHR,GrotzRC.ProgressindevelopmentoftheindexofADL.Gerontologist1970,10:20-30.60. KatzS,FordAB,MoskowitzRW,JacksonBA,JaffeMW.Studiesofillnessintheaged:theindexofADL.Astandardizedmeasureofbiologicalandpsychologicalfunction.JAMA1963;185:914-91961. CharlsonME, Pompei P, Ales KL,Mackenzie CR. A newmethod of classifying prognostic comorbidity in longitudinalstudies:developmentandvalidation.JChronicDis1987;40:373-83.62. Vincent JL,Moreno R, Takala J,Willatts S, DeMendonça A, Bruining H, et al. The SOFA (Sepsis-related Organ FailureAssessment) score to describe organ dysfunction/failure. On behalf of the Working Group on Sepsis-related Problems of theEuropeanSocietyofIntensiveCareMedicine.IntensiveCareMed1996;22:707-10.63. Miranda DR et al. Simplified Therapeutic Intervention Scoring System: the TISS-28 items - results from amulticenterstudy.CritCareMed1996;24:64-73.64. ReisMirandaD,MorenoR, IapichinoG.Nineequivalentsofnursingmanpoweruse score (NEMS). IntensiveCareMed1997;23:760-5.

39

65. OrweliusL,NordlundA,NordlundP,SimonssonE,BäckmanC,SamuelssonA,SjöbergF.Pre-existingdisease:themostimportantfactorforhealthrelatedqualityoflifelong-termaftercriticalillness.CritCare2010;14(2):R6766. Azoulay E, Soares M, Darmon M, Benoit D, Pastores S, Afessa B. Intensive care of the cancer patient: recentachievementsandremainingchallenges.AnnIntensiveCare2011;1:567. SoaresM,CarusoP,SilvaE,TelesJM,LoboSM,FriedmanG,DalPizzolF,MelloPV,BozzaFA,SilvaUV,TorellyAP,KnibelMF, et al. Characteristics and outcomes of patients with cancer requiring admission to intensive care units: A prospectivemulticenterstudy.CritCareMed2010;38:9-1568. BenoitDD,DepuydtPO,VandewoudeKH,OffnerFC,BoterbergT,DeCockCA,NoensLA,JanssensAM,DecruyenaereJM.Outcome inseverely illpatientswithhematologicalmalignancieswhoreceived intravenouschemotherapy inthe intensivecareunit.IntensiveCareMed2006;32:93-9969. GristinaGR,AntonelliM,ContiG,CiarloneA,RoganteS,RossiC,BertoliniG.Noninvasiveversusinvasiveventilationforacuterespiratoryfailure inpatientswithhematologicalmalignancies;a5yearmulticenterobservationalsurvey.CritCareMed2011;39:2232-223970. BenoitDD,HosteEA,DepuydtPO,OffnerFC,LameireNH,VandewoudeKH,DhondtAW,NoensLA,DecruyenaereJM.Outcomeincritically illmedicalpatientstreatedwithrenalreplacementtherapy;comparisonbetweenpatientswithandthosewithouthaematologicalmalignancies.NephrolDialTransplant2005;20:552-55871. BenoitDD,HosteEA.Acutekidneyinjuryincriticallyillpatientswithcancer.CritCareClin2010;26:151-17972. MokartD,EtienneA,EsterniB,BrunJP,Chow-ChineL,SanniniA,FaucherM,BlacheJL.Criticallyillcancerpatientsintheintensivecareunit:short-termoutcomeand1-yearmortality.ActaAnaesthesiolScand2012;56:178-18973. RosolemMM,RabelloLS,LisboaT,CarusoP,CostaRT,LealJV,SalluhJI,SoaresM.Criticallyillpatientswithcancerandsepsis:Clinicalcourseandprognosticfactors.JCritCare2012;27:301-30774. Bird GT, Farquhar-Smith P, Wigmore T, Potter M, Gruber PC. Outcomes and prognostic factors in patients withhaematologicalmalignancyadmittedtoaspecialistcancerintensivecareunit:a5yrstudy.BrJAnaesth2012;108:452-45975. OeyenSG,VandijckDM,BenoitDD,Annemans L,Decruyenaere JM.Qualityof life after intensive care:A systematicreviewoftheliterature.CritCareMed2010;38:2386-240076. UchinoS,KellumJA,BellomoRetal.Acuterenalfailureincriticallyillpatients.JAMA.2005;294:813-81877. HosteEA,SchurgersM.Epidemiologyofacutekidneyinjury:howbigistheproblem?CritCareMed.2008;36Suppl4:146-15178. PalevskyPM,Zhang JH,O'ConnorTZ,ChertowGM,CrowleyST,ChoudhuryD,FinkelK,KellumJA,PaganiniE,ScheinRMHetal.Intensityofrenalsupportincriticallyillpatientswithacutekidneyinjury.NewEnglJMed.2008;359:7-2079. BellomoR, CassA,NortonR,GallagherM, Lo S, Su S, Cole L, Finfer S,McArthur C,McGuinness Set al. Intensity ofContinuousRenal-ReplacementTherapyinCriticallyIllPatients.NewEnglJMed.2009;361:1627-163880. Amdur RL, Chawla LS, Amodeo S, Kimmel PL, Palant CE. Outcomes following diagnosis of acute renal failure in U.S.veterans:focusonacutetubularnecrosis.KidneyInt.2009;76:1089-109781. Gammelager H, Christiansen CF, Johansen MB, Tonnesen E, Jespersen B, Sorensen HT. One-year mortality amongDanishintensivecarepatientswithacutekidneyinjury:acohortstudy.CritCare.2012;16:R12482. Chawla LS, Eggers PW, Star RA, Kimmel PL. Acute kidney injury and chronic kidney disease as interconnectedsyndromes.NEnglJMed.2014;371:58-6683. AhlstromA,TallgrenM,PeltonenS,RasanenP,PettilaV.Survivalandqualityof lifeofpatientsrequiringacuterenalreplacementtherapy.IntensiveCareMed.2005;31:1222-122884. DelannoyB, FloccardB, Thiolliere F, KaakiM,BadetM,Rosselli S, Ber CE, SaezA, FlandreauG,GuerinC. Six-monthoutcomeinacutekidneyinjuryrequiringrenalreplacementtherapyintheICU:amulticentreprospectivestudy.IntensiveCareMed.2009;35:1907-1915

40

85. MorgeraS,KraftAK,SiebertG,LuftFC,NeumayerHH.Long-termoutcomesinacuterenalfailurepatientstreatedwithcontinuousrenalreplacementtherapies.AmJKidneyDis.2002;40:275-27986. Korkeila M, Ruokonen E, Takala J. Costs of care, long-term prognosis and quality of life in patients requiring renalreplacementtherapyduringintensivecare.IntensiveCareMed.2000;26:1824-183187. Abelha FJ, BotelhoM, FernandesV, BarrosH.Outcome andquality of life of patientswith acute kidney injury aftermajorsurgery.Nefrologia2009;29:404-41488. MaynardSE,WhittleJ,ChelluriL,ArnoldR.Qualityoflifeanddialysisdecisionsincriticallyillpatientswithacuterenalfailure.IntensiveCareMed.2003;29:1589-159389. VaaraST,PettilaV,ReinikainenM,KaukonenKM,ConsortiumFIC.Population-basedincidence,mortalityandqualityoflife incritically illpatientstreatedwithrenalreplacementtherapy:anationwideretrospectivecohortstudy infinnish intensivecareunits.CritCare.2012;16:R1390. Hofhuis JG,vanStelHF,SchrijversAJ,Rommes JH,SpronkPE.Theeffectofacutekidney injuryon long-termhealth-relatedqualityoflife:aprospectivefollow-upstudy.CritCare.2013;17:R1791. NobleJS,SimpsonK,AllisonME.Long-termqualityoflifeandhospitalmortalityinpatientstreatedwithintermittentorcontinuoushemodialysisforacuterenalandrespiratoryfailure.RenFail.2006,28:323-33092. Hamel MB, Phillips RS, Davis RB, Desbiens N, Connors AF, Jr., Teno JM, Wenger N, Lynn J, Wu AW, Fulkerson W.Outcomesandcost-effectivenessofinitiatingdialysisandcontinuingaggressivecareinseriouslyillhospitalizedadults.SUPPORTInvestigators. Study toUnderstandPrognoses andPreferences forOutcomesandRisksof Treatments.Ann InternMed. 1997;127:195-20293. Vandenbroucke JP,vonElmE,AltmanDG,GøtzschePC,MulrowCD,PocockSJetal. Strengthening the reportingofobservationalstudiesinepidemiology.Epidemiology.2007;18:805-3594. NielssonMS,ChristiansenCF,JohansenMB,RasmussenBS,TønnesenE,NørgaardM.MortalityinelderlyICUpatients:acohortstudy.ActaAnaesthesiolScand2014;58:19-2695. AndersenFH,FlaattenH,KlepstadP,RomildU,KvåleR.Long-termsurvivalandqualityof lifeafter intensivecare forpatients80yearsofageorolder.AnnIntensiveCare2015;5:1396. DaubinC,ChevalierS,SéguinA,GaillardC,ValetteX,PrévostF,etal.Predictorsofmortalityandshort-termphysicalandcognitivedependenceincriticallyillpersons75yearsandolder:aprospectivecohortstudy.HealthQualLifeOutcomes2011;9:3597. SacanellaE,Pérez-CastejonJM,NicolasJM,MasanesF,NavarroM,CastroP,etal.Functionalstatusandqualityoflife12monthsafterdischargefromamedicalICUinhealthyelderlypatients:aprospectiveobservationalstudy.CritCare2011;15:R10598. Flaatten H, De Lange DW,Morandi A, Andersen FH, Artigas A, Bertolini G, Boumendil A, CecconiM, Christensen S,FaraldiL,FjølnerJ, JungC,MarshB,MorenoR,OeyenS,ÖhmanCA,PintoBB,SolimanIW,SzczeklikW,ValentinA,WatsonX,ZaferidisT,GuidetB;VIP1studygroup.The impactof frailtyon ICUand30-daymortalityand the levelofcare inveryelderlypatients(≥80years).IntensiveCareMed2017;Sept2017;43:1820-182899. HeylandDK,GarlandA,BagshawSM,CookD,RockwoodK,StelfoxHT,etal.Recoveryaftercritical illness inpatientsaged80yearsorolder:amulti-centerprospectiveobservationalcohortstudy.IntensiveCareMed2015;41:1911-1920100. BagshawSM,StelfoxHT,JohnsonJA,McDermidRC,RolfsonDB,TsuyukiRT,etal.Long-termassociationbetweenfrailtyandhealth-relatedqualityoflifeamongsurvivorsofcriticalillness:aprospectivemulticentercohortstudy.CritCareMed2015;43:973-982.101. Hofhuis JG, vanStelHF, SchrijversAJ,Rommes JH, SpronkPE.Changesofhealth-relatedqualityof life in critically illoctogenarians.Chest2011;140:1473-1483102. MerlaniP,ChenaudC,MariottiN,RicouB.Long-termoutcomeofelderlypatientsrequiringintensivecareadmissionforabdominalpathologies:survivalandqualityoflife.ActaAnaesthesiolScand2007;51:530-537103. KhouliH, AstuaA,DombrowskiW,Ahmad F,Homel P, Shapiro J, et al. Changes in health-relatedquality of life andfactorspredictinglong-termoutcomesinolderadultsadmittedtointensivecareunits.CritCareMed2011;39:731-737

41

104. FerranteLE,PisaniMA,MurphyTE,GahbauerEA,Leo-SummersLS,GillTM.Functionaltrajectoriesamongolderpersonsbeforeandaftercriticalillness.JAMAInternMed2015;175:523-529105. PickardAS,NearyMP,CellaD.EstimationofminimallyimportantdifferencesinEQ-5DutilityandVASscoresincancer.HealthQualLifeOutcomes2007;5:70106. RiouB,BoddaertJ.TheelderlypatientandtheICU:Wherearewegoing,whereshouldwego?CritCareMed2016;44:231-232107. NeedhamDM, Davidson J, Cohen H, Hopkins RO,Weinert C,Wunsch H, et al. Improving long-term outcomes afterdischargefromintensivecareunit:Reportfromastakeholders’conference.CritCareMed2012;40:502-509108. HashemMD,NallagangulaA,NalamalapuS,NunnaK,NausranU,RobinsonKA, et al. Patientoutcomesafter criticalillness:asystematicreviewofqualitativestudiesfollowinghospitaldischarge.CritCare2016;20:345109. Norman BC, Jackson JC, Graves JA, Girard TD, Pandharipande PP, Brummel NE, et al. Employment outcomes aftercriticalillness:AnanalysisofthebringingtolighttheriskfactorsandincidenceofneuropsychologicaldysfunctioninICUsurvivorscohort.CritCareMed2016,44:2003-2009110. LamasD.Chroniccriticalillness.NewEngJMed2014;370:175-177111. SimpkinAL,SchwartzsteinRM.Toleratinguncertainty–thenextmedicalrevolution?NewEnglJMed2016;375:1713-1715112. Carson SS, Kahn JM, Hough CL, Seeley EJ,White DB, Douglas IS, et al. Amulticentermortality predictionmodel forpatientsreceivingprolongedmechanicalventilation.CritCareMed2012;40:1171-1176113. HeylandDK,StelfoxHT,GarlandA,CookD,DodekP,Kutsogiannis J,etal.Predictingperformancestatus1yearaftercritical illness inpatients80yearsorolder:Developmentofamultivariableclinicalpredictionmodel.CritCareMed2016;44:1718-1726114. DecruyenaereA,DecruyenaereP,PeetersP,VermassenF,DhaeneT,Couckuyt I.Predictionofdelayedgraft functionafterkidneytransplantation:comparisonbetweenlogisticregressionandmachinelearningmethods.BMCMedInformDecisMak2015;15:83115. HarrisonDA,GriggsKA,PrabhuG,GomesM,LeckyFE,HutchinsonPJ,etal.ExternalvalidationandrecalibrationofriskpredictionmodelsforacutetraumaticbraininjuryamongcriticallyilladultpatientsintheUnitedKingdom.JNeurotrauma2015;32:1522-1537116. Peeters P, VanBiesenW,VeysN, LemahieuW,DeMoor B,DeMeester J. External validation of a risk stratificationmodeltoassistshareddecisionmakingforpatientsstartingrenalreplacementtherapy.BMCNephrol2016;17:41.117. WassenaarA, vandenBoogaardM, vanAchterbergT, SlooterAJ, KuiperMA,HoogendoornME,et al.MultinationaldevelopmentandvalidationofanearlypredictionmodelfordeliriuminICUpatients.IntensiveCareMed2015;41:1048-1056118. Zimmerman JE, Kramer AA, McNair DS, Malila FM. Acute Physiology and Chronic Health Evaluation (APACHE IV):hospitalmortalityassessmentfortoday’scriticallyillpatients.CritCareMed2006;34:1279-1310.119. MinneL,LudikhuizeJ,deJongeE,deRooijS,Abu-HannaA.Prognosticmodels forpredictingmortality inelderly ICUpatients:asystematicreview.IntensiveCareMed2011;37:1258-1268120. Veerbeeck JM,KwakkelG, vanWegenEH, Ket JCF,HeymansMW. Early predictionof outcomesof activities of dailylivingafterstroke:asystematicreview.Stroke2011;42:1482-1488121. BrinkmanS,Abu-HannaA,deJongeE,deKeizerNF.Predictionoflong-termmortalityinICUpatients:modelvalidationandassessing theeffectofusing in-hospitalversus long-termmortalityonbenchmarking. IntensiveCareMed2013;39:1925-1931122. TibshiraniR.Regressionshrinkageandselectionviathelasso.JRStatistSocB1996;58:267-288123. YuanM,LinY.Modelselectionandestimationinregressionwithgroupedvariables.JRStatistSocB2006;68:49-67

43

PartTwo

SystematicReview

And

OriginalStudies

45

I.Qualityoflifeafterintensivecare:A

systematicreviewoftheliterature

SandraGOeyen,MD1,DominiqueMVandijck,PhD2,DominiqueDBenoit,MD,PhD1,LievenAnnemans,

PhD2,JohanMDecruyenaere,MD,PhD1

1DepartmentofIntensiveCareMedicine,GhentUniversityHospital,Ghent,Belgium

2DepartmentofPublicHealth,GhentUniversity,Ghent,Belgium

PublishedinCriticalCareMedicine2010;38:2386-2400

46

ABSTRACT

Objectives:1)Toevaluatequalityoflife(QOL),atleast12-monthsafterdischargefromtheintensivecare

unit (ICU),ofadult critically ill patients,2) toevaluate themethodologyused toassess long-termQOL,

and3)togiveanoverviewoffactorsinfluencingQOL.

Datasources:EMBASE-PubMed,MEDLINE(OVID),SCI/WebofScience,CochraneLibrary,GoogleScholar,

andpersonalfiles.

Dataextraction:Dataextractionwascarriedoutindependentlyandcrosscheckedbytworeviewersusing

apredefineddataextractionform.Eligiblestudieswerepublishedbetween1999and2009,andassessed

QOL ≥ 12-months after ICU discharge by means of the Medical Outcomes Study 36-item Short Form

Health Survey (SF-36), RAND-36-item Health Survey, EuroQol-5D (EQ-5D), and/or Nottingham Health

Profile(NHP)inadultICUpatients.

Datasynthesis:53articles(10multicenters)wereincluded,withthemajorityperformedinEurope(68%).

TheSF-36wasusedin55%,andtheEQ-5D,NHP,RAND-36,oracombination,inrespectively21%,9%,8%

and8%.Aresponserateof≥80%wasattainedin26studies(49%).CriticallyillpatientshadalowerQOL

thananage-andgendermatchedpopulationbutQOLtendedtoimproveoveryears.Theworstreductions

inQOLwereseen in severeARDS,prolongedmechanicalventilation, severe trauma,andseveresepsis.

Studyqualitycriteria,definedasbaselineQOLassessment,nomajorexclusioncriteria,descriptionofnon-

responders, and a comparison with a reference population were only met in 4 studies (8%). Results

concerningtheinfluenceofseverityofillness,co-morbidity,pre-admissionQOL,age,gender,oracquired

complicationswereconflicting.

Conclusions: QOL differed upon diagnostic category, but overall, critically ill patients had a lowerQOL

thananage-andgendermatchedpopulation.Aminorityof studiesmet thepredefinedmethodological

quality criteria. Results concerning influence of the patients’ characteristics and illness upon long-term

QOLwereconflicting.

47

INTRODUCTION

Since intensive caremedicine per definition treats themost critically ill patients, who have an

inherenthighriskofmortality,itseemslogicalthatformanyyears,theprimaryoutcomeparameterhas

beensurvivalrate.Whilethisiswithoutanydoubtaveryimportantissue,survivalormortalityratehave

also the advantage of being unambiguous and very easy to measure. Advances in diagnostic and

therapeuticoptionsmakethatmoreandmorepatientssurvivecriticalillness.Whilestudiesinvestigating

survival rates of critically ill patients are widely performed, we also have to question whether critical

illnesshasany impactonan individuals (very) long-term(i.e.≥12monthsafter intensivecareunit (ICU)

discharge)healthstatusandqualityoflife(QOL).Therefore,nexttosurvivalormortalityrate,QOLhasto

beconsideredtobeofequalimportanceasoutcomeparameter.

AlthoughQOLhasbeenacceptedtobevaluableregardingoutcome,itisnotroutinelyincludedin

studies and research on this topic is still in its infancy. This has many reasons. Measuring QOL, with

specificquestionnaires,ismorelabourintensiveandtimeconsumingandwillalwaysbemoreambiguous

for interpretation than the “death” or “alive” outcome parameters. Optimal follow-up periods for

measuringQOLarenotdefined.BaselineassessmentofQOLisdifficultbutofgreatvaluetoexaminethe

burdenofthecriticalillness.

OnlyafewreviewsofQOLafterintensivecarehavebeenpublishedearlier(1-4).Therehasbeen

nosystematic reviewprovidingaccurateandrecentdataon theburdenofcritical illnessonapatients’

long-termQOL.Nevertheless,abetterunderstandingofhowintensivecareaffectshealthandwell-being

of its survivors will help physicians when deciding on allocating therapeutic efforts in the future.

Consequently,itisthepurposeofthispapertogiveasystematicreviewoftheliterature,publishedinthe

pastdecade,ofQOLand influencing factors,at leastoneyearafterdischarge fromthe ICU,andof the

methodology used. Finally, we hope to give better insights into long-term QOL, and to make

methodologicalrecommendationsforfurtherresearchonthistopic.

MATERIALSANDMETHODS

DataSources,SearchStrategy,StudySelectionandDataExtraction

A two-staged systematic review process of existing published original research articles was

conducted. First, two authors (SO, DV) independently searched EMBASE-PubMed, MEDLINE (OVID),

SCI/WebofScience,CochraneLibrary,andGoogleScholaronJanuary9,2010usingthemedicalsubject

headings (MeSH) or text keywords “quality of life”, or “long-term outcome” cross referenced with

“intensivecare”,“criticalcare”,“criticallyillpatients”,“ICUpatients”,“criticalcarepatients”,“ICUstay”,

or “ICU”. Limitations were applied regarding; language (English language), time (articles published

between January,1th1999andDecember,31th2009),age (above18years),andhumans.Personal files

that were known to the authors and reference lists of relevant articles were hand-searched as well.

48

Outcomes articles including exclusively cardiac or thoracic aortic surgery patients, methodological

articles, literaturereviews,case-reports,editorials,and letterswereexcluded.Studieswith lessthan50

patients were also not included. If it was unclear whether or not patients were admitted to the ICU,

articleswereexcludedaswell(5-7).

In stage two, all abstracts were evaluated independently by two authors (SO, DV) for the

following methodological criteria: 1) assessment of QOL by means of at least one of the following

instruments:MedicalOutcomesStudy36-itemShort FormHealthSurvey (SF-36),RAND-36-itemHealth

Survey, EuroQol-5D (EQ-5D), and/orNottinghamHealthProfile (NHP); and2) follow-upperiodof ≥12-

monthsfollowingdischargefromtheICU.Disagreementregardingeligibilitywasresolvedbyconsensus.

Subsequently,identifiedarticlesweredownloaded,andscreenedelectronically.Foreacheligible

article, using a predefined categorization system, information was extracted on respectively; authors,

journal, yearof publication, studydesign, inclusionperiod, initial study cohort, baseline variables (age)

and outcome (hospital mortality), number of eligible patients for long-term QOL assessment,

instrument(s)andmethod(s)usedforQOLassessment,responserate,follow-upperiod,theuseofother

questionnairesortests,thefinalconclusionconcerningQOL,andfactorsdeterminingQOL.Studyquality

wasassessedusingfourimportantcriteria,analogoustoDowdyetal.(1):1)QOLassessmentpriortoICU

admission, 2) description of key inclusion or exclusion criteria, 3) description of non-responders and

comparison with those remaining in the study, and 4) adjustment for confounders such as age and

gender.Abovementionedcriteriawerenotusedindecisionsregarding inclusionorexclusionofeligible

studies.Anydiscrepanciesbetweenbothreviewerswereresolvedbydiscussion.

QOLmeasurementinstruments

SF-36, RAND-36, EQ-5D, and NHPwere considered as they are generic instruments commonly

usedinintensivecareresearch(8);theyarewellvalidatedandhavepopulationnormsintheliterature(9-

16).

The SF-36 questionnaire contains 36 items measuring eight multi-item domains: physical and

social functioning,role limitationsduetophysicaloremotionalproblems,mentalhealth,vitality,bodily

pain,andgeneralperceptionofhealth(9-13).

Arising from SF-36, the RAND-36 questionnairewas developed.While the count system in the

latterdifferssomewhatcomparedtoSF-36,questionsandfinalresultsarealmostsimilar(14).

The EQ-5D is a short questionnaire consisting of three parts (15, 17-19). A descriptive system

measures health in five domains: mobility, self-care, usual activities, pain/discomfort, and

anxiety/depression.Eachdomainhasthreelevels:noproblems,moderate,orsevereproblems,andcan

thereforebeclassifiedintooneof243(35)possiblehealthstates.Eachofthesecanbeconvertedintoone

singlesummaryindex,whichcanbeusedinhealth-economystudies.Onavisualanaloguescale(EQ-VAS),

patients can rate their overall health between 0 and 100. Although the EQ-5D is a well-known and

49

validated instrument to measure QOL in general populations, it has been less well validated in the

critically illpopulation (17-19),and itmayprovide less informationandmaybe lessdiscriminative than

theSF-36(20).

TheNHPconsistof a twopartsquestionnaire (16). The firstone is composedof38 statements

related to six domains: physicalmobility, pain, sleep, energy, emotional reactions, and social isolation.

Thesecondpart lists7activitiesofdaily life:occupation,housework,socialactivity,home life,sex life,

hobbies and holidays. The NHP has already been used to evaluate QOL in the critically ill population,

especially in cardiac surgery patients (21). Nevertheless, internal consistency and sensitivity to change

werebetterfortheSF-36andRAND-36thanfortheNHP(22-24).

RESULTS

Atotalof53articleswerefinally includedinthereview.Thearticlesweregroupedaccordingto

diagnosticcategory.Studiesconcerningcriticallyillpatientsingeneralwereseparatedbaseduponfollow-

upperiod.Elevenarticlesconcerningacuterespiratorydistresssyndrome(ARDS)(25-35),3articlesabout

prolongedmechanicalventilation(36-38),8traumastudies(20,39-45),6concerningcardiacarrest (46-

51),6studiesaboutelderlypatients(52-57),2pancreatitisstudies(58,59),3sepsisstudies(60-62),and4

studieswithvarioustopics(63-66)wereincluded.Therewere4studiesconcerningoutcomeandQOLin

generalcriticallyillpatientsoneyearafterintensivecare(19,67-69)and6withlongerfollow-upperiods

(70-75).Table1givesanoverviewofthecharacteristicsofthesestudies.Allthestudieswereperformed

in large hospitals. Ten were multicenter studies (32-34, 40, 45, 48, 51, 57, 61, 66). Thirty six were

conductedinEurope(19,20,26-28,35,36,41,42,44-46,48,49,51-56,59,61-75),13intheUSA(25,29-

31,37-40,43,47,50,57,58)and4 inCanada (32-34,60).WithinEurope, themajorityof studieswere

doneinScandinaviancountries(42,44,45,51,54,59,61,64-67,72-75),Germany(26-28,35,49,71)and

theNetherlands(20,48,55,63).

Inclusionperiodsvariedbetweenlessthanoneyear(61,68,70,71)and10yearsormore(26-28,

35,47,50).Allbut3studiesconcerningcriticallyillpatientsingeneralhadaninclusionperiodofoneyear

(19,67,69,72-75).In3articles,theinclusionperiodwasnotfurtherspecified(32,33,40)(Table1).

Table2givesanoverviewofQOLassessmentafterICUdischarge.ThemostfrequentlyusedQOL-

instrumentwastheSF-36(55%),followedbytheEQ-5D(21%),theNHP(9%),andtheRAND-36(8%).Four

studies(8%)usedacombinationofQOLinstruments,eithertheSF-36withtheEQ-5D(19,53),theRAND-

36withtheEQ-5D(54),ortheNHPwiththePatrick’sPerceivedQualityofLifescore(PQL),anotherQOL

questionnaire(52).

Follow-up periods forQOL assessment varied between the included studies. Some had a strict

follow-upperiodofoneyear(29,30,37,40,41,56,67,68),whereasothershadlargerangeswithintheir

follow-op time (26-28, 35, 43-47, 50, 52, 55, 58-60), and in 1 study, although at least 12months, the

50

follow-upperiodforQOLevaluationwasnotclearlydefined(39).TwelvestudiesevaluatedQOLatvery

strict timepoints during the follow-upperiod (19, 31-34, 38, 51, 57, 66, 72, 74, 75).Median follow-up

periods of 5 years or more were found in 8 studies (26-28, 42, 48, 49, 71, 73). Particularly the

ScandinavianareaseemedtobeinterestedinresearchonQOLalongperiodafterICUdischarge(42,54,

59,64,65,72-75).

QOLwasassessedatfollow-upbyamailedsurveyin22studies(42%)(20,35,36,39,45,48,49,

53,54,57,59,61,67,68,63-66,72-75),byphonein14(26%)(19,25,32,33,40,41,44,52,55,56,58,60,

62,69),byface-to-faceinterviewsin12(23%)(26-31,34,43,46,47,50,51)orbyacombinationofthese

methodsin5studies(9%)(37,38,42,70,71).Togainthehighestresponseratepossible,manystudies

sentremindermailsorphonedinabsenceofanyresponsebymail(20,35,39,42,45,49,53,54,57,59,

65,67,68,72-74).Nevertheless,therewere3studies(6%)witharesponseratebelow50%(26,27,39),

24studies(45%)witharesponseratebetween50-79%(19,28,32,33,35,37,38,40,41,45,49,51,53,

57,58,61,62,64-66,69,73-75),and26studies(49%)hadaresponserateofatleast80%oftheeligible

patientpopulationforlong-termoutcomeandQOLassessment(20,25,29,30,31,34,36,42-44,46-48,

50,52,54-56,59,60,63,67,68,70-72).

Fourstudies(8%)metallofthe4predefinedstudyqualitycriteria;assessmentofQOLatbaseline,

nomajorexclusion criteriawithin the studypopulation,descriptionof thenon-respondergroupversus

the respondergroup,andcomparisonwithanage-andgendermatchednormalpopulation (19,37,53,

61)(Table3).ByomittingassessmentofbaselineQOLasqualitycriterion,thenumberofstudiesfulfilling

theother3qualitycriteriaincreasedto21(40%)(26-28,32,35,36,39,40,42,45,47,49,57,59,62,64-

66,69,72,74).Only9studies(17%)measuredQOLpriortoICU(19,37,38,44,52,53,61,68,70),andin

27articles(51%)(19,26-28,32,35,36,37,39,40,42,44,45,47,49,53,57,59,61,62,64-66,69,70,72,

74),adescriptionwasgivenofthenon-respondergroupandcomparedwithpatientswhorespondedto

theQOLsurvey.Allstudiesdefinedclearlywhichpatientswerein-orexcluded.

Table 4 summarizes the major finding concerning long-term QOL per article. Long-term QOL

varied between diagnostic categories. ARDS patients, patients after prolonged mechanical ventilation,

severetraumapatients,andsepsissurvivorsshowedsignificantimpairmentsinlong-termQOL(25-45,60-

62).While physical aspects improved slowly over the years, mental and emotional impairments were

stagnant or declined even further. On the other hand, survivors of cardiac arrest, severe pancreatitis,

oesophagectomy, and acute kidney injury had a goodQOLwhichwas comparablewith or evenbetter

thananage-andgendermatchedpopulation (46-51,58,59,63,64). In theelderly,QOLwassomewhat

decreased, especially in the physical domains, but elderly patients generally adapted well to these

limitationsandperceived theirQOLasgood (27-32).Oneyearafter ICU,critically illpatients ingeneral

hadalowerQOL,especiallyinphysicaldomains,thananage-andgendermatchedpopulation(19,67-69).

However,a slow improvement topre-morbidQOL levelscouldbe found.The increase inQOLcouldbe

51

furtherseenseveralyearsafterICU,whereQOLwasquitecomparablewiththatofthenormalpopulation

(70-75).

FactorsassociatedwithreductionsinQOLatleastoneyearafterICUdischargearealsodisplayed

in Table 4. In ARDS or patients with prolonged mechanical ventilation, the ARDS and its sequelae

influencedQOLbyimpairmentsinpulmonaryfunctions,cognitivedisorders,weakness,andposttraumatic

stressdisorders(25-35).Intraumapatients,theinjuryseverity,thedegreeofbraindamage,andfemale

gender dominated long-term QOL in a negative way (20, 41, 43, 44). However, in other studies the

severity of illness played a less important role (71, 74). In a mixed ICU-patient population, diagnostic

categorydeterminedQOL (67,68,70).Therewereconflicting results regarding the influenceofageon

long-termQOL (19, 37, 42, 57, 59, 63, 67, 70, 74). Two studies found that a poor pre-admission QOL

playedaroleinthereductioninQOLalongperiodafterICUdischarge(19,70).

DISCUSSION

Itwas thepurposeof this reviewtogiveanoverviewof the literatureofQOLat leastoneyear

afterdischargefromtheintensivecare,ofthefactorsthatdetermineQOL,andofthemethodologyused.

Because of differences in study design, patient population, QOL instruments, follow-up time, and

responserate,itisimpossibletomakeoneoverallconclusion.Thisreviewhashoweversomeimportant

findings.

First, long-term QOL depends largely upon diagnostic category. Patients with severe ARDS,

prolonged mechanical ventilation, severe trauma, and severe sepsis appeared to have the worst

reductions in QOL, which lasted a long time. While physical aspects improved slowly over the years,

mentalandemotionalimpairmentswerestagnantordeclinedevenfurther.Traumapatientswereusually

healthyandyoungbeforeICUadmission.TheirQOLoftendroppedsubstantiallyafterthetrauma,bothon

physical and psychosocial dimensions, and delusional memories and the inability to return to work

influencednegativelytheirperceivedQOL(20,41,45).Survivorsofcardiacarrest,elderly,patientswith

severe pancreatitis, after oesophagectomy, or patients with acute kidney injury had a good QOL or

perceiveditasevenbetterthanbeforeillness.Acceptanceofdisabilityis ingeneralhigheramongolder

patients,andevenbetteriftheyhaveagoodsocioeconomicstatus(52).AhighQOLdespitetheseverity

ofillnessorpersistingsymptoms,maybeexplainedbythefactthatpatientswhoareconfrontedwitha

life-threateningdiseaseare facedwith thenecessity toaccommodate to thedisease,whichmay lower

internalstandards(63).CriticallyillpatientsingeneralhadalowerQOLthananage-andgendermatched

populationoneyearafterICUdischarge,butaslowimprovementinQOLcouldbeseen,andseveralyears

afterICU,QOLwasquitecomparablewiththatofthenormalpopulation.

Thesecondfindingwasthatfactors,whichcouldbepresumedtoresultinapoorQOLafterICU,

suchasage,prolongedmechanicalventilation,oralongICUorhospitalstay,arenotperseindicatorsof

52

reductionsinQOLafterwards(25,27,44).Otherissuessuchascognitiveimpairments,sleepdisturbances,

posttraumaticstressdisorder, therehabilitationprocess,employmentstatus,andculturalandpayment

differences,caninfluenceQOLinalesstangiblewaythan,forexample,physicalimpairmentsaftermajor

trauma(26,27,35,49,52,66).

Third, therewere importantmethodological differences between the included studies. Four of

the53 includedstudiesmetallof4qualitycriteria. Onlyaminorityofstudieshadauniformfollow-up

timeormeasuredQOLprior to ICUadmission, and response rates toQOL surveyswere generally low,

whichresultedinalimitedinterpretationofstudyresults.

Theidealassessmentoflong-termQOLaftercriticalcareshouldusevalidatedQOLinstrumentsin

large cohorts without major exclusions, with an extensive but reasonably long and uniform follow-up

period, andwith comparisonwith pre-ICUbaseline evaluation (61). Future research on long termQOL

should focusonthat. In this review,onlystudieswhichusedat leastoneof4genericQOL instruments

(SF-36, EQ-5D, RAND-36, NHP) were included. Generic instruments apply for a broad spectrum of

populationsandarethereforelessresponsivetochangesinspecificconditionsascomparedwithspecific

QOL instruments (9).Although there is still no consensusaboutwhich tool shouldbeused tomeasure

QOL in critical care patients, SF-36 and EQ-5D are considered to be valid and reliable instruments for

criticallyillpatients(10).TheEQ-5DisvalidatedforEuropeanpopulations(76,77)butsomestillconsider

SF-36orRAND-36asthegeneric instrumentof firstchoice incritically illpatients (19,60,67). Itcanbe

recommendedtousebothEQ-5DandSF-36together(20).

Oneof thegoalsofQOLmeasures isdifferentiatingbetweenpeoplewithabetterandaworse

QOL,andmeasuringhowmuchQOLhaschangedovertime(9).ThischangeinQOLovertimeleadstoan

important and difficult issue in QOL studies. How long is “long” in long-term outcome and when will

functional outcome measures and questionnaires no longer give additional information? Follow-up

intervals forQOLwere verydifferent in the included studieswhichmade it difficult to concludewhich

timecourseshouldbeconsideredas thebest to interpret theoverall results,andas sufficient toallow

regainingthebestachievableQOL(71).Notonlybetweenstudiestherewerelargedifferencesintiming,

but also within the studies themselves the follow-up intervals differed a lot, which was correctly

consideredasa limitationof study results (26,27,35,36,45-47,50).A follow-upperiodofoneyear is

probablytooshortbecausephysical limitationsstill tendtodominateoveremotionalproblems(19,30,

31,35,37,41),andphysicalproblemswillnotalwaysberecovered(67).Oneyearmayalsobetooshort

to become accustomed tomore restrictions in daily live (72).When follow-upperiods extend tomore

thanoneyear,atendencytowardsmoreemotionalproblemswasfound.Itisgenerallyacceptedthatthe

realburdenofcritical illness isseenupto6monthsafter ICUdischarge(32,64),although it ispossible

thatstudiesusing6monthsasthefirsttimepoint fordatacollectionmissedanearlier fall inQOL(19).

Follow-upof1or2yearswillprobablycapturethemostanditmaybethelimitforimprovementinmost

53

QOLdimensionsasseenafterseveretrauma(44,68).Still,mentalhealthwillbeaffectedformanyyears

longer(35,70).

Themost important problemof long-term follow-up times is thatmorepatientswill be lost to

follow-up,whichcouldleadtoanimportantbiasinresults.Patientswhonotrespondcandosoforalot

ofdifferentreasons.TheycanconsiderQOLquestionnairestrivialiftheyrecoveredwell,theycansuffer

fromposttraumaticstressdisorderavoidingseekingmemoriesoftheirICUtreatment,theycanbetooill

tohavetheabilitytorespond,ortheymayhavediedbeforecompletingthesurvey(35,36,54).Assuch,

QOLrespondersmayrepresentasampleofhealthierpatients(47,58).Therefore,analyzesofresponders

versusnon-responders concerning severityof illness scores, co-morbidities,mortality,orage shouldbe

made (44). To avoid selection bias, every effort has to be made to target the highest response rate

possible.Inmanystudies,althoughtime-consumingandlabour-intensive,patients,whodidnotrespond

totheinitialmailedsurveyortoamailedreminder,werephoned,whichguaranteedhowevernotalways

ahighresponserate(35,39,73).Alosttofollow-upof20%isconsideredtobeacceptableforQOLstudies

(19)butonly49%ofthestudieshadaresponserateofatleast80%oftheeligiblepatientpopulationfor

long-termoutcomeandQOLassessment.Asaconsequence,thenumberofpatientswithareliableQOL

assessmentatleast1yearafterICUdischargewaslow.

WhenQOLmeasuresareusedasdiscriminative instruments,possibleconfounders,whichcould

influenceQOL, should be eliminated. Therefore,QOL in ICU patients can be compared to an age- and

gendermatched general population, which should be considered as the upper limits of what is

achievable (75). In most studies, QOL-responders were matched with a representative healthy

population.Thestudy findingscanalsobecomparedwithanappropriatecontrolgroupeliminatingthe

influenceofspecifichealthconditions(25,62).Moreimportant,long-termQOLshouldalsobecompared

withQOLbeforeICUadmission,todiscriminatewhetherpoorlong-termQOLisaresultoftheseverityof

illness, or due to confounding factors or ‘background variables’ such as co-morbid disease, poor pre-

admissionQOL,age,gender,oracquiredcomplications (44).Which factorwill influence themost long-

termQOLisaverydifficultquestion,andliteratureisdefinitivelynotconclusiveaboutthisissue(74).The

long-termeffectofa certain conditiononQOL is cohort-specificandmaybe the residuaofany severe

criticalillness(34).Itwillalsodependuponthefollow-upperiod,andthetoolsused,andwillprobablybe

amixtureofseverityofillness,priorhealthstatus,pre-morbidQOL,age,gender,anddiagnosticcategory.

PriorstudiesofQOLbeforeICUadmissionsupportthehypothesisthatpatients’premorbidQOL

has a large effect on QOL after critical illness (78, 79). It has been proved that pre-ICU QOL is low

comparedtothegeneralpopulationindicatingthatICUpatientsdifferfromtheaveragepopulationeven

before onset of critical illness (10, 44, 80). PoorQOLbefore critical illness is also correlatedwith poor

outcome (19,81,82,83,84). ImpairedQOLafter ICUmay thus reflect apoorbaseline situation rather

thanbeafunctionofintensivecare(19,67).MeasuringQOLatbaselineisdifficultandinthemajorityof

54

studies(83%)thiswasnotdone.Onethirdofthesestudiesconsideredthisasalimitation(20,25,31,36,

42,43,54-57,62,64,65,67). MostpatientswillnotbeabletocompletequestionnairesattimeofICU

admission andmany studies asked patients or proxies a long period afterwards how QOL was before

admission (20, 44, 52, 53, 62). Recall bias can influence results of theseQOL surveys. In retrospective

studiesrecallbiascanalsoaddsomeuncertaintytothestudyfindingsbecauseQOLassessmentisbased

upon patient’s recall of their memories from the ICU stay (45, 46). No baseline assessment of QOL

because itwouldhavebeenassessedretrospectivelycanbethereasonfornotmeasuringQOLprior to

ICUadmission(56).

SomeauthorsconsideredthatonlypatientscouldevaluatetheirownQOL(56)orconsidereditas

apotentialdangerforbiasifquestionnaireswerefilledinbyproxies(67).However,theSF-36andEQ-5D

questionnairecompletedbyproxiescanreliablyassesstheQOLofthecriticallyillpatientonadmissionat

the ICU (68, 81), although it is difficult to interview proxies when their relatives are critically ill (37).

ProxiestendtounderestimatetheQOLofthepatientbutdifferencesareusuallysmall(81).

Therearesomemethodological limitations in this review.First,only4genericQOL instruments

were included, which are, however, commonly used in critically ill patients (8). This allowed us to

compare among studies andmakemore comprehensive conclusions. Second, some studies had a low

number ofQOL responders and a non-uniform follow-up timewhich limits the interpretation of study

results.ThefindingsofthisreviewarealsolimitedbecauseofinfrequentcollectionofQOLatbaseline.

CONCLUSION

Future outcome evaluations should not be limited to “death” or “alive” but should also

incorporate QOL, even as this ismuchmore complicated to investigate. Long-termQOL in critically ill

patients depends largely upon diagnostic category, with the worst reductions found in patients who

survive severe ARDS, sepsis, trauma, and prolongedmechanical ventilation. For critically ill patients in

general,a lowerQOLcomparedtoanage-andgendermatchedhealthypopulationwasseen.However,

evidence forpoorerQOLafter ICU ismisleadingwhenthepriorhealthstateof thepatient isnot taken

intoaccount.BaselineQOLassessmentisnecessarywheninvestigatingtheinfluenceofthecriticalillness

andshouldbeassesseduponICUadmissiontoavoidrecallbias.Follow-upperiodsshouldbekeptstrictly

uniformalthoughthereisnoconsensusregardingthemostappropriatefollow-uptime.Measurestogain

thehighest response rate to avoid selectionbias shouldbe taken.Nevertheless, comparisonsbetween

respondersandnon-respondersshouldalwaysbemade.

55

Table1.Studycharacteristics

Reference Country Studydesign Inclusionperiod Patientcohort Eligiblepatientsforlong-term

QOLassessment,

N(%)*

ARDSDavidson,1999 USA prospective

matchedcontrolled

January1994-July1996

102sepsisortraumainducedARDSpatients

80(78%)

Schelling,2000 Germany follow-upcohort

January1985-January1995

192consecutiveARDSpatients 119(62%)

Rothenhäusler,2001

Germany exploratory January1985-January1995

192consecutiveARDSpatients 119(62%)

Kapfhammer,2004 Germany follow-upcohort

January1985-January1995

80long-termARDSsurvivors 80(100%)

Hopkins,1999 USA prospective February1994-July1998

106enrolledoutof274ARDSpatients

67(63%)

Orme,2003 USA prospective,cohortofaRCT

February1994-December1999

120ARDSpatientsenrolledinHTVvs.LTVstudy

74(62%)

Hopkins,2005 USA longitudinalprospective,cohortofaRCT

February1994-December1999

120ARDSpatientsenrolledinHTVvs.LTVstudy

74(62%)

Heyland,2005 Canada prospectiveobservationalmulticenter

NA 221ARDSpatientsenrolledinaphaseIIImulticenterRCT

103(47%)

Parker,2006 Canada prospectiveobservationalmulticenter

NA 221ARDSpatientsenrolledinaphaseIIImulticenterRCT

103(47%)

Herridge,2003 Canada longitudinalmulticenter

May1998–May2001

195adultARDSpatients 109(56%)

Deja,2006 Germany prospectivecontrolled

1991-2000 263patientswithsevereARDS 129(49%)

ProlongedmechanicalventilationCombes,2003 France prospective

cohortJanuary1995–June1999

347consecutivepatientsreceivingmechanicalventilationfor≥14d

99(29%)

Chelluri,2004 USA prospectiveobservational

June1997–July1999

817patientsreceivingmechanicalventilationfor≥48hrs

359(44%)

Cox,2009 USA prospectiveobservational

April2006-April2007

126consecutivepatientsreceivingmechanicalventilation≥21dorwithatracheotomyafter≥4dofmechanicalventilation

90(71%)

TraumaMiller,2000 USA retrospective January1991-

December1997115severelyinjuredpatientsspending≥3weeksintheICU

90(78%)

MacKenzie,2002 USA retrospective(hospitalstay),prospective(QOL)multicenter

NA sampleof1587patientsregisteredinthePennsylvaniaTraumaOutcomesStudy

1587(100%)

Dimopoulou,2004 Greece prospectivecohort

1999-2000 191consecutivemultipletraumapatientsrequiringmechanicalventilation

117(61%)

Sluys,2005 Sweden retrospective(patientcohort),prospective(QOL)

1996-1997 309traumapatients 246(80%)

Vles,2005 TheNetherlands

prospective January1996-January1999

295severelyinjuredpatients(ISS≥16)

196(66%)

Jackson,2007 USA retrospective 2003 97traumaICUsurvivorswithout 58(60%)

56

ICHUlvik,2008 Norway follow-up

cohort1998-2003 325traumapatients 228(70%)

Ringdal,2009 Sweden exploratorymulticenter

September2001-August2002

344adulttraumasurvivors 344(100%)

CardiacarrestSaner,2002 Switzerland retrospective

case-control1991-1996 439OOHCApatients

(of1307resuscitations)50(11%)

Bunch,2003 USA prospective(cardiacarrest,survival,QOL)

November1990-January2001

145OOHCApatients(of200resuscitations)

60(41%)

Kuilman,1999 TheNetherlands

retrospectivemulticenter

1988-1994 441OOHCApatients(of898resuscitations)

132(30%)

Graf,2008 Germany prospectivecohort

January1999-December2000

354consecutivepatientswithcardiacarrest

110(31%)

Mahapatra,2005 USA prospective(cardiacarrest,survival,QOL)

November1990-January2001

142OOHCApatients(of200resuscitations)

60(42%)

Lundgren-Nilsson,2005

Sweden longitudinalmulticenter

1996-1999 51cardiacarrestsurvivors 51(100%)

ElderlyMontuclard,2000 France prospective

cohortJanuary1993–August1998

75consecutivepatients>70yrswithICULOS≥30d

30(40%)

Merlani,2007 Switzerland retrospective January1999-December2000

141consecutivepatients≥70yrswithabdominalpathologies

52(37%)

Kaarlola,2006 Finland crosssectionalsurvey

1995-2000 882elderly(≥65yrs)1827controls(<65yrs)

354elderly(40%)

1074controls(59%)

deRooij,2008 TheNetherlands

retrospectivecohort

January1997-December2002

578consecutivepatients≥80yrs

231(40%)

Garrouste-Orgeas,2006

France prospectiveobservational

March2002-November2003

180patients≥80yrstriagedforICUadmission;48ICUadmissions

28(16%)(only9ICUpatients)

Kleinpell,2003 USA longitudinalprospectivemulticenter

periodof14months

883patients≥45yrs,ICU-LOS≥24hrs

284(32%)

PancreatitisSoran,2000 USA retrospective January1992-

December199652ICUpatientswithacutepancreatitis

39(75%)

Halonen,2003 Finland retrospective January1989-December1997

283consecutivepatientswithsevereacutepancreatitis

174(61%)

Sepsis

Heyland,2000 Canada cross-sectionalsurvey

1993-1998 78sepsispatients 30(38%)

Karlsson,2009 Finland prospectiveobservationalmulticenter

November2004-February2005

470severesepsispatients 278(59%)

Korosec,2006 Slovenia observational 2003 164patients(66sepsis,98trauma) 78patients(48%)

(21sepsis,57trauma)

MixedICUpatients1yearafterICU

Pettilä,2000 Finland prospectiveobservational

1995 591consecutiveICUpatients 354(60%)

Badia,2001 Spain prospectivecohort

October1994-June1995

523consecutivepatients(84T,239SS,57US,143M)

375(69T,198SS,23US,85M)

(72%)Cuthbertson,2005 United

Kingdomprospectivecohort

May2001-April2002

423consecutiveICUpatients 300(71%)

57

Stricker,2005 Switzerland prospectiveobservationalcase-control

September1998-August1999

173patientswithICU-LOS>7dvs1506withICU-LOS≤7d

116withanICU-LOS>7days(67%)

Long-termQOLGarciaLizana,2003 Belgium prospective

observationalJune25-September10,2000

202consecutiveadmittedpatients 118(58%)

Graf,2005 Germany prospectivecohort

November1997-February1998

303consecutivepatientswithICU-LOS>24hrs

190(63%)

Kaarlola,2003 Finland prospectiveobservational

1995 591consecutivepatients 169(29%)

Flaatten,2001 Norway retrospective(ICUstay),prospective(survival,QOL)

1987 219ICUpatients 88(40%)

Kvale,2003 Norway prospectivecohort

July1999-August2000

226patientswithICU-LOS>24hrsdischargedalive

226(100%)

Kvale,2002 Norway prospectiveandretrospectivecohort

1987comparedwith1997

219patientswithICU-LOS≥24hrsin1987,338in1997

88(40%)(1987)106

(31%)(1997)VariousdiseasesdeBoer,2000 The

Netherlandsprospectiveobservational

January1993-May1996

100consecutivepatientswhounderwentatranshiataloesophagectomy

35(35%)

Ahlström,2005 Finland crosssectionalcohort

1998-2002 703patientsreceivingRRTforAKI 229(33%)

Ylipalosaari,2007 Finland prospective May2002-June2003

272hospitalsurvivorswithICU-LOS>48hrs

187(69%)

Orwelius,2008 Sweden prospectivemulticentercohort

August2000-November2003

1625consecutiveadultpatientswithICU-LOS>24hrs

723(44%)

QOL=qualityoflife;N=number;ARDS=acuterespiratorydistresssyndrome;USA=UnitedStatesofAmerica;RCT=randomisedcontrolledtrial;HTV=hightidalvolume,LTV=lowtidalvolume;NA=notavailable;d=days;hrs=hours;ICU=intensivecareunit;ISS=injuryseverityscore;ICH=intracranialhemorrhage;OOHCA=outofhospitalcardiacarrest;yrs=years;LOS=lengthofstay;T=trauma,SS=scheduledsurgery;US=unscheduledsurgery;M=medical;vs=versus;RRT=renalreplacementtherapy;AKI=acutekidneyinjury;*Percentageofinitialpatientcohort

58

Table2.AssessmentofqualityoflifeafterICUReference QOL

assessmentinstrument

MethodofQOLassessment Responserate,%(NofQOLresponders)

Follow-upperiod

ARDSDavidson,1999 SF-36 telephone 96%(77) median23monthsSchelling,2000 SF-36 face-to-face 42%(50) median5.5years

(range1-10years)Rothenhäusler,2001 SF-36 face-to-face 39%(46) median6years(range1-12

years)Kapfhammer,2004 SF-36 face-to-face 58%(46) median8years(range3-13

years)Hopkins,1999 SF-36 face-to-face 82%(55) 1yearOrme,2003 SF-36 face-to-face 89%(66) 1yearHopkins,2005 SF-36 face-to-face 84%(62) 1and2yearsHeyland,2005 SF-36 telephone 71%(73) 3,6,12monthsParker,2006 SF-36 telephone 71%(73) 3,6,12monthsHerridge,2003 SF-36 face-to-face 80%(83)3months

82%(82)6months86%(83)at12months

3,6,12months

Deja,2006 SF-36 mail,telephoneifnoanswer

50%(65) 57±32months

ProlongedmechanicalventilationCombes,2003 NHP mail 88%(87) average3yearsChelluri,2004 SF-36 telephoneorface-to-face 64%(231)full

interview18%(65)mini-interview

1year

Cox,2009 EQ-5D telephoneorface-to-face 78%(70) 3,12monthsTraumaMiller,2000 RAND-36 mail,telephoneifno

answer39%(35) unclear,meanofseveralyears

MacKenzie,2002 SF-36 telephone 78%(1230) 1year(range10-14months)Dimopoulou,2004 NHP telephone 74%(87) 1yearSluys,2005 SF-36 mailortelephone,reminder

mail83%(205) 5years

Vles,2005 EQ-5D mail,telephoneifnoanswer

85%(166) mean41months

Jackson,2007 SF-36 face-to-face 100%(58) 12-24monthsUlvik,2008 EQ-5D telephone 92%(210) 2-7years(median4years)Ringdal,2009 SF-36 mail,onewrittenreminder,

thentelephone69%(239) 6-18months

CardiacarrestSaner,2002 NHP face-to-face 100%(50) mean31.7months

(range5-68months)Bunch,2003 SF-36 face-to-face 83%(50) 4.8±3.0yearsKuilman,1999 EQ-5D mail 83%(109) mean6.71yearsGraf,2008 SF-36 mailortelephoneifno

answer74%(81) 5years

Mahapatra,2005 SF-36 face-to-face 83%(50) 4.8±3.0yearsLundgren-Nilsson,2005

NHP face-to-face 51%(26)at1year 14 days, 45 days, 3months, 1year

ElderlyMontuclard,2000 PQL(1996)

NHP(1998)telephone 93%(28)(firststudy)

95%(21)(secondstudy)

557±117daysforthefirststudy,second2yearslater

Merlani,2007 ED-5D,SF-36 mail,telephoneifno/incompleteanswer

79%(41) 2years

Kaarlola,2006 EQ-5D,RAND-36

mail,remindermail 87%(307)elderly77%(828)controls

median3yearsforelderlymedian4yearsforcontrols

59

deRooij,2008 EQ-5D telephone 88%(204) 1to6years,median3.7yearsGarrouste-Orgeas,2006

NHP telephone 100%(28) 1year

Kleinpell,2003 SF-36 mail,remindermail,telephoneifnoanswer

70%(199) 1,3,6,12months

PancreatitisSoran,2000 SF-36 telephone 54%(21) median42months

(range17-69months)Halonen,2003 RAND-36 mail,remindermailor

telephone83%(145) median61months

(range19-127months)SepsisHeyland,2000 SF-36 telephone 100%(30)first

interview87%(26)secondinterview

16.6±10.6months

Karlsson,2009 EQ-5D mail 52%(252)QOLbefore58%(156)long-termQOL

median17months

Korosec,2006 EQ-5D telephone 50%(39) 2yearsMixedICUpatients1yearafterICUPettilä,2000 RAND-36 mail,remindermail 87%(307) 1yearBadia,2001 EQ-5D mail,telephoneorface-to-

faceinterviewifnoanswer89%(334) 1year

Cuthbertson,2005 SF-36,alsoEQ-5Dat12months

telephone 78%(233)3months67%(201)6months58%(173)12months

3,6,12months

Stricker,2005 SF-36 telephone 65%(75)

12-18months

Long-termQOLGarciaLizana,2003 EQ-5D mailortelephone 81%(96) 1,5yearsGraf,2005 SF-36 mailortelephone 91%(173) 5yearsKaarlola,2003 RAND-36 mail,remindermailifno

response84%(298)1year76%(192)6years

1yearand6years

Flaatten,2001 SF-36 mail,remindermailifnoresponse

58%(51) 12years

Kvale,2003 SF-36 mail,oneremindermail 56%(126)at6months79%(100)after2years

6monthsand2years

Kvale,2002 SF-36 mail 58%(51)in198762%(66)in1997

3yearsand13years

VariousdiseasesdeBoer,2000 SF-36 mail 100%(35) minimumof2yearsAhlström,2005 EQ-5D mail 67%(153) median2.4yearsYlipalosaari,2007 EQ-5D mail,telephoneifno

response76%(142) median22months

Orwelius,2008 SF-36 mail 69%(497)after12months

6and12months

ICU=intensivecareunit;QOL=qualityoflife;N=number;ARDS=acuterespiratorydistresssyndrome;SF-36=Short-Form36;NHP=NottinghamHealthProfile;EQ-5D=EuroQol-5D;PQL=Patrick’sPerceivedQualityofLife

60

Table3.StudyqualitycriteriaReference QOL

priortoICU

Keyinclusionorexclusioncriteria Descriptionofnon-responders

Age/gendermatchedgeneralpopulationtocompareQOL

ARDSDavidson,1999 no ARDSsurvivorswithsevereheadinjurieswere

excluded.no matchedwithsepsisand

traumapatientswithoutARDS

Schelling,2000 no Studypopulationwasafollow-upcohortof80long-termARDSsurvivorsandQOLrespondersinastudy3yearsbefore.

yes age-andgender-matchedcontrolgroupofnormalGermansubjects

Rothenhäusler,2001 no Onlylong-termARDSsurvivorswereincluded. yes age-andgender-matchedcontrolgroup

Kapfhammer,2004 no Onlylong-termARDSsurvivorswereincluded. yes standardvaluesoftheSF-36fromvolunteersoftheWestGermanpopulation

Hopkins,1999 no 168ARDSpatientswereexcludedforvariousreasons.

no normativepopulationdata

Orme,2003 no Onlylong-termARDSsurvivorswereincluded. no normativepopulationdata

Hopkins,2005 no Long-termARDSsurvivorswereincluded. no normativepopulationdata

Heyland,2005 no Long-termARDSsurvivorswereincluded. yes age-andgender-matchedpopulationderivedfromliterature

Parker,2006 no Long-termARDSsurvivorswereincluded. no no,primaryARDSpatientswerecomparedtosecondaryARDSpatients

Herridge,2003 no OnlysevereARDSpatientswereincluded.Immobilepatients,patientswithahistoryofpulmonaryresectionorwithaneurologicalorpsychiatricdiseasewereexcluded.

no thenormalCanadianpopulation

Deja,2006 no OnlysevereARDSpatientswereincluded. yes age-andgendermatchedhealthyGermancontrols

ProlongedmechanicalventilationCombes,2003 no Onlypatientswithprolongedmechanical

ventilation(≥14d)wereincluded.yes community-basedage-

andgendermatchedcontrols

Chelluri,2004 yes Patientswithprolongedmechanicalventilation(≥48hrs)wereincluded.

yes samplesoftheUSpopulation

Cox,2009 yes Patientswith≥21dmechanicalventilationorwithtracheotomyafter≥4dmechanicalventilationwereincluded.

no UKpopulationnormsforpersonsaged55-65years

TraumaMiller,2000 no Onlyseverelyinjuredpatientsspending≥3

weeksintheICUwereincluded.yes generalUSpopulation

MacKenzie,2002 no Blunttraumapatients(18-59yrs),withahospitalstayof≥72hrswereincluded.Drownings,electrocutions,burns,andhiporfemoralneckfractureswereexcluded.

yes age-andgendermatchedgeneralpopulation

Dimopoulou,2004 no Onlymechanicallyventilatedpolytraumapatientswereincluded.

no no

Sluys,2005 no BluntorpenetratingtraumapatientswithanISSof≥9wereincluded.Patientswithpsychiatricdisordersorcognitiveimpairmentswereexcluded.

yes aSwedishage-andgender-matchedreferencesample

Vles,2005 no OnlypatientswithISS≥16wereincluded. no Swedishreference

61

database,correctedforageandgender

Jackson,2007 no OnlytraumaICUsurvivors(ISS>25)withoutintracranialhemorrhagewereincluded.

no thegeneralUSpopulation

Ulvik,2008 yes Foreigntraumapatientswereexcludedduetodifficultieswithfollow-up.

yes no

Ringdal,2009 no Nonsurvivors,attemptedsuicide,notresidentinSweden,intellectualimpairment,andpatientswithunknownaddresswereexcluded.

yes ageandgendermatchedreferencesampledrawnfromtheSwedishSF-36normdatabase.

CardiacarrestSaner,2002 no Patientswithhypoxicbraindamage,drug

abusers,inhospitalresuscitation,non-Germanspeaking,and<20or>80yrswereexcluded.

no healthycontrolsofsimilarage,gender,andsocio-economicstatus

Bunch,2003 no OnlypatientswithanOOHCAwithVFwereincluded.

yes age-andgender-matchednormsfromasampleofthegeneralUSpopulation

Kuilman,1999 no Successfullyresuscitatedpatientswereincluded. no noGraf,2008 no PatientswhoreceivedCPRforanIHCAor

OOHCAwereincluded.yes thehealthyGerman

populationMahapatra,2005 no OnlypatientswithanOOHCAwithVFwere

included.no age-andgender-

matchednormsfromasampleofthegeneralUSpopulation

Lundgren-Nilsson,2005

no Onlycardiacarrestsurvivorswereincluded. no referenceSwedishpopulation

ElderlyMontuclard,2000 yes Consecutivepatients>70yrswithanICULOS≥

30dwereincluded.no thegeneralFrench

populationofmixedageand76-yrsoldSwedishurbancitizens

Merlani,2007 yes Patientsaged≥70yrswithabdominalpathologieswereincluded.

yes age-matchedpopulation

Kaarlola,2006 no Allconsecutivepatientsadmittedwithinthestudyperiodwereincluded.

no controlsandanage-andgender-matchedFinnishpopulation

deRooij,2008 no Consecutivepatientsaged≥80yrsadmittedwithinthestudyperiodwereincluded.

no age-matchedBritishnon-ICUgeneralpopulation


no In73%ofpatientsaged≥80yrsICUadmissionwasrefused.

no age-andgender-matchedgeneralFrenchpopulation

Kleinpell,2003 no Patients≥45yrswithICU-LOSof≥24hrswereincluded.

yes ageneralUSpopulation

PancreatitisSoran,2000 no Onlyacutepancreatitispatientswereincluded. no age-matchednormal

controlgroupHalonen,2003 no Patients(majorityneededICUadmission)with

acutepancreatitiswereincluded.yes age-andgender-

matchedFinnishpopulation

SepsisHeyland,2000 no Patientswithsepsiswereincluded.Patientswith

disabilitiesthatwouldprecludeatelephoneinterviewwereexcluded.

no generalUSpopulation

Karlsson,2009 yes AllseveresepsispatientsatadmissionorduringICUstaywereincluded.

yes age-andgenderadjustedFinnishreferencepopulation

Korosec,2006 no Onlysepsisandtraumapatientswereincluded. yes noMixedICUpatients1yearafterICUPettilä,2000 no nomajorexclusioncriteria no age-andgendermatched

62

generalFinnishpopulation

Badia,2001 yes nomajorexclusioncriteria no noCuthbertson,2005 yes PatientswhowerenotexpectedtosurviveICU

wereexcluded.yes age-andgendermatched

generalUKpopulationStricker,2005 no SurgicalandtraumapatientswithICU-LOS>7d

andwithICU-LOS≤7dwerematched.Burninjurieswereexcluded.

yes age-andgendermatchedsampleoftheGermanpopulation

Long-termQOLGarciaLizana,2003 yes ICU-admissionsforuncomplicatedelective

postoperativesurgerywereexcluded.yes no

Graf,2005 no PatientswithICU-LOS<24hrswereexcluded. no age-matchedgroupofhealthyGermans

Kaarlola,2003 no Patientswhorespondedtobothquestionnairesin1996and2001wereincluded.

yes age-andgendermatchedFinnishpopulation

Flaatten,2001 no Heartsurgeryandburnpatientswerenotincluded.

no age-andgendermatchedgeneralNorwegianpopulation

Kvale,2003 no Heartsurgeryandburnpatientswerenotincluded.

yes scoresafter6monthscomparedwithscoresafter2years

Kvale,2002 no Heartsurgeryandburnpatientswerenotincluded.

no age-andgendermatchedcontrolgroupsfromthegeneralNorwegianpopulation

VariousdiseasesdeBoer,2000 no Onlylong-termsurvivorswithouttumour

recurrencewereincluded.no age-matchedreference

populationAhlström,2005 no OnlyAKIpatientsneedingRRTwereincluded yes age-andgendermatched

populationYlipalosaari,2007 no OnlyhospitalsurvivorswithICU-LOS>48hrs

wereincluded.yes no

Orwelius,2008 no OnlyadultpatientswithICU-LOS>24hrsandalive6monthsafterdischargewereincluded.

yes randomsamplefromthemainintakeareaofthehospitalswasusedasareferencegroup

QOL=qualityof life; ICU= intensive careunit;ARDS=acute respiratorydistress syndrome; SF-36=Short-Form36; d=days; hrs=hours; US= United States of America; UK= United Kingdom; yrs=years; ISS= injury severityscore;OOHCA=outofhospitalcardiacarrest;VF=ventricularfibrillation;CPR=cardiopulmonaryresuscitation;IHCA=inhospitalcardiacarrest;LOS=lengthofstay;AKI:acutekidneyinjury;RRT=renalreplacementtherapy

63

Table4.Majorfindingsandfactorsinfluencinglong-termQOLReference Long-termQOL:Majorfinding QOL:Influencingfactors

ARDSDavidson,1999 ARDSsurvivorshadasignificantreductioninQOL.

Sepsis-inducedARDSpatientshadmoreseverereductionsinQOLthantrauma-inducedARDSpatients.

ARDSanditssequelaeNot:co-morbiddisease,severityoftraumaorillness,durationofmechanicalventilationorhospitalstay

Schelling,2000 Long-termARDSsurvivorshaveasignificantreducedQOL. multiplepulmonaryfunctionimpairments

Rothenhäusler,2001

Long-termQOLwasimpaired. cognitivedeficitsanddisability

Kapfhammer,2004

Long-termARDSsurvivorshadmajorimpairmentsinlong-termQOL. posttraumaticstressdisorder

Hopkins,1999 After1year,therewasimprovementforthephysicalbutnotfortheemotionaldomains.

cognitiveimpairments

Orme,2003 ARDSsurvivors,treatedwithhighorlowtidalvolumeventilation,hada reducedQOL,whichwas related tophysical rather thanemotionalconcerns.

pulmonaryfunctionimpairments

Hopkins,2005 ARDSsurvivorshaddecreasedQOL,withphysicalandemotionaldomainsimprovingat1year,butnoadditionalchangeordeclineat2years.

neurocognitiveimpairments,althoughthesemayrepresentmorbidityfromcriticalillnessratherthanbespecificforARDS

Heyland,2005 ARDSsurvivorshadasignificantlylowerQOLthanage-andgender-matchedcontrols.After1year,therewasanimprovementinthephysicaldomains,whilethementalscoresremainedunchanged.

pulmonaryfunctionimpairments,baselineco-morbidities

Parker,2006 PrimaryARDSpatientshadsignificantlybetterQOLscoresthanpatientswithsecondaryARDS.

primaryversussecondaryARDSNOT:ICULOS,hospitalLOS,durationofmechanicalventilation,co-morbidity,lungfunction

Herridge,2003 QOL improved over 1 year after ICU discharge but remained lowerthantheseofthecontrolpopulation.

functionaldisabilityduetomusclewasting,weakness,fatigue

Deja,2006 QOLinpatientswithARDSwassignificantlyreducedinalldimensions. posttraumaticstressdisorderProlongedmechanicalventilationCombes,2003 QOL was impaired but perceived as acceptable, with psychosocial

aspectsbeingbetterthanphysicalperformance.worseQOLseeninARDSsurvivors

Chelluri,2004 QOL was impaired mainly on the physical and social domains butcomparableonthementalhealthandemotionaldomains.

influenceofageandchronicillnesspredominatethelong-termoutcome

Cox,2009 OneyearafterICUdischarge,themajorityofpatientshadapoorQOL. NATraumaMiller,2000 QOLwaslow,especiallyinthephysicaldomains. NAMacKenzie,2002 Oneyearaftertrauma,QOLwaslow,exceptforvitalityandmental

health.NA

Dimopoulou,2004

QOLwasimpairedinphysicalfunctioning,workingability,andemotionalwell-being.

injuryseverity,degreeofbraintrauma

Sluys,2005 Fiveyearsaftertrauma,QOLwaslowinalldimensionsoftheSF-36. age,surgicalprocedures,ICU-andhospitalLOS,in-hospitalcomplications,inadequateinformation

Vles,2005 QOLwaslowandaquarterofthoseofworkingagewereunabletoreturntowork.

injuryseverity,femalegender

Jackson,2007 QOLwaslow. cognitiveimpairmentsUlvik,2008 Morethan2yearspost-injury,74%reportedimpairedQOL,mostly

duetopainanddiscomfort,butonlyaminorityhadsevereproblems.severityofillnessaninjury,timesincetrauma(pain),femalegender,degreeofbraintraumaNOT:age

Ringdal,2009 TraumapatientsscoredlowonallSF-36domains. delusionalmemories,co-morbidity

64

CardiacarrestSaner,2002 Long-termQOLremainedfulfillingwithonlyafewchangesinthe

psychosocialprofile.littleimpactofchangesinpsychosocialprofile

Bunch,2003 Exceptfromareductioninvitality,QOLwassimilartothatofthegeneralpopulation.

NA

Kuilman,1999 NodifferenceinQOLbetweenpatientsresuscitatedbyemergencypersonnel,physicians,orbystanders.

NA

Graf,2008 PatientswhosurvivewithoutsevereneurologicaldisabilitiesmayexpectagoodQOL.

NA

Mahapatra,2005 Long-termsurvivalandQOLareequallyfavourableinbothsexes. NALundgren-Nilsson,2005

QOLimprovedovertheyearwithvaluescomparabletothereferencepopulation.

cognitiveimpairments

ElderlyMontuclard,2000

After1year,perceivedQOLwasgood,especiallyemotionalandsocialfunctioning.

amoderatedisabilityinfluencedQOL

Merlani,2007 AhighmortalityandadecreaseinQOLwereobservedforelderlypatientswithabdominalpathologies.Thesepatientsadaptedwelltotheirphysicallimitations.

NA

Kaarlola,2006 AgingdecreasedQOLmostlyinthephysicaldomains,butelderlypatientshadbettervaluesformentalhealththantheyoungercontrols.

acceptanceofdisabilityisbetterwithagoodsocialnetwork

deRooij,2008 QOLwassignificantlylowerforusualactivities.MostpatientswerewillingtoreceiveICUtreatmentagainifnecessary.

NA


Afteroneyear,QOLwaspoorerthaninthegeneralpopulation.One-halfofthesurvivorsdidnotwantfurtherICUadmissionifnecessary.

NA

Kleinpell,2003 Inthemiddle-agedandelderlypatientgroup,SF-36scoresremainedbelowthegeneralpopulationnormsbutincreasedovertime.

severityofillnessratherthanage

PancreatitisSoran,2000 Long-termQOLisgoodandcomparablewithanage-matchedcontrol

population.NA

Halonen,2003 Long-termQOLisgoodandcomparablewithanage-matchedcontrolpopulation.

workingstatusbeforeacutepancreatitis,ageNOT:follow-uptime,cause,gender,ICUtreatment,ICU-LOS,MOF,operatingstatus

SepsisHeyland,2000 TheQOLofsepsissurvivorsislowerthanthatofthegeneral

populationandcomparabletoQOLofpatientswithchronicdiseaseorsurvivorsofacutelunginjury.

NA

Karlsson,2009 QOLinmostpatientswasalreadylowerbeforetheepisodeofseveresepsisthaninthegeneralpopulation,anditwasevenlowerafterthecriticalillness.

NA

Korosec,2006 SICU-patientswithsepsishaveahighermortalitythantraumapatients.However,QOLafter2yearsisreducedtothesamelevelinbothgroups.

anxietyanddepression(trauma)

MixedICUpatients1yearafterICUPettilä,2000 SurvivorshadalowerQOLthananage-andgender-matchedgeneral

population.However,patientsperceivedtheirQOLasbetterorsimilarasbeforetheirICUstay.

MOF,age,diagnosticcategory

Badia,2001 Traumapatientsexperiencedaworsening,unscheduledsurgeryandmedicalpatientsaslightdeterioration,andscheduledpatientsaconsiderableimprovementinQOL.

diagnosticcategory

Cuthbertson,2005

PhysicalQOLincreasedtopremorbidlevels1yearafterICUdischargebutphysicalscoresremainedbelowthepopulationnorms.Mentalscoresweresimilarorhigherthanpopulationnorms.Non-survivorshadalowerQOLthansurvivorsatalltimepoints.

poorbaselinesituationNOT:prolongedICU-LOS,age,surgicalormedicaladmissions

Stricker,2005 Whentakingintoaccountseverityofillness,QOL1yearafterICUdischargeiscomparablebetweenpatientswithshortandlongICUstay.QOLremainedlowerthaninageneralpopulation,mostlyin

NOT:prolongedICU-LOS

65

physicalaspects.Long-termQOLGarciaLizana,2003

38%felttheirQOLwasworse,37%feltittobesimilarand25%feltitwasbetterthanpriortotheirICUadmission.Psychologydomainswerethemostfrequentlyaffected.

previousQOL,prolongedhospitalstay,ICUreadmission,diagnosticcategory,APACHEIIscore,age,femalegender,organfailure

Graf,2005 After5years,mostpatientslivedindependentlyandhadagoodQOL. NOT:severityofillness,morbidity,resourceconsumption,age,gender

Kaarlola,2003 SixyearsafterICUdischarge,QOLwascomparablewiththatofthegeneralpopulation.QOLrevealedworsephysicalfunctioning,pain,andgeneralhealthbutimprovementinthepsychologicaldomains.

NA

Flaatten,2001 QOLwasacceptablebutitwasstilllowerthaninthegeneralpopulation.

NA

Kvale,2003 TherewasanincreaseinQOLfrom6monthsto2yearsinamixedICU-population.

ageminor:severityofillness,ICU-LOS

Kvale,2002 QOLwasstillreduced3and13yearsafterICU.QOLwasmorereducedin1997patients(3yearsfollow-up)thanin1987patients(13yearsfollow-up).

NA

VariousdiseasesdeBoer,2000 Althoughresidualsymptomsmaypersist,patientsreportedasimilar

orevenbetterQOL(emotionalwell-beinginparticular)thananage-matchedreferencegroup.

prolongedhospitalstay,age,fatigue,emotionalaspectsNOT:diseasespecificsymptoms

Ahlström,2005 Thelong-termoutcomeandQOLofpatientswithAKIwerepoorbutpatientsperceivedtheirQOLasgood.

NA

Ylipalosaari,2007

QOLwasequallyreducedinpatientswithorwithoutICU-acquiredinfection.

NOT:ICU-acquiredinfection

Orwelius,2008 QOLwasreducedduetophysicalproblems,bodilypain,generalhealth,vitality,andmentalhealth.

minor:sleepdisturbances

QOL=qualityoflife;ARDS=acuterespiratorydistresssyndrome;ICU=intensivecareunit;LOS=lengthofstay;NA=notavailable; SF-36=Short-Form 36; MOF= multiple organ failure; SICU= surgical intensive care unit; APACHE= acutephysiologyandchronichealthevaluation

66

REFERENCES

1. DowdyDW,EidMP,SedrakyanA,Mendez-TellezPA,PronovostPJ,HerridgeMS,NeedhamDM:Qualityof life inadultsurvivorsofcriticalillness:asystematicreviewoftheliterature.IntensiveCareMed2005;31:611-620

2. HennessyD,JuzwishinK,YergensD,NoseworthyT,DoigC:Outcomesofelderlysurvivorsofintensivecare.Areviewoftheliterature.Chest2005;127:1764-1774

3. ChaboyerW, ElliottD:Health-related quality of life of ICU survivors: reviewof the literature. Intensive Crit CareNurs

2000;16:88-974. BoumendilA,SommeD,Garrouste-OrgeasM,GuidetB:Shouldelderlypatientsbeadmittedtotheintensivecareunit?

IntensiveCareMed2007;33:1252-12625. InabaK,GoeckeM,SharkeyP,BrennemanF:Long-termoutcomesafterinjuryintheelderly.JTrauma2003;54:486-4916. MichaelsAJ,MichaelsCE,SmithJS,MoonCH,PetersonC,LongWB:Outcomefrominjury:generalhealth,workstatus,

andsatisfaction12monthsaftertrauma.JTrauma2000;48:841-8507. Polinder S, van Beeck EF, Essink-Bot ML, et al: Functional outcome at 2.5, 5, 9, and 24 months after injury in the

Netherlands.JTrauma2007;62:133-1418. GarrattA, Schmidt L,MackintoshA, FitzpatrickR:Qualityof lifemeasurement:bibliographic studyofpatientassessed

healthoutcomemeasures.BMJ2002;324:1417-14219. HeylandDK,GuyattG,CookDJ,MeadeM,JuniperE,CroninL,GafniA:Frequencyandmethodologicrigorofquality-of-

lifeassessmentsinthecriticalcareliterature.CritCareMed1998;26:591-598

10. Angus DC, Carlet J: Surviving Intensive Care: a report from the 2002 Brussels Roundtable. Intensive CareMed 2003;29:368-377

11. Ware JE, Jr., Sherbourne CD: The MOS 36-item short-form health survey (SF-36). I. Conceptual framework and item

selection.MedCare1992;30:473-48312. McHorneyCA,WareJE,RaczekAE:TheMOS36-itemshort-formhealthsurvey(SF-36).II.Psychometricandclinicaltests

ofvalidityinmeasuringphysicalandmentalconstructs.MedCare1993;31:247-26313. ChrispinPS,ScottonH,RogersJ,LloydD,RidleySA:ShortForm36intheintensivecareunit:assessmentofacceptability,

reliabilityandvalidityofthequestionnaire.Anaesthesia1997;52:15-2314. HaysRD,SherbourneCD,MazelRM:TheRAND36-ItemHealthSurvey1.0:HealthEcon1993;2:217-22715. The EuroQolGroup: EuroQol: a new facility for themeasurement of health-related quality of life.Health Policy 1990;

16:199-20816. HuntSM,McKennaSP,McEwenJ,WilliamsJ,PappE:TheNottinghamHealthProfile:subjectivehealthstatusandmedical

consultations.SocSciMedA1981;15:221-22917. BrazierJ,JonesN,KindP:TestingthevalidityoftheEuroqolandcomparingitwiththeSF-36healthsurveyquestionnaire.

QualLifeRes1993;2:169-18018. Brazier JE,Walters SJ,Nicholl JP, KohlerB:Using the SF-36 andEuroqol on anelderly population.Qual LifeRes 1996;

5:195-20419. CuthbertsonBH,ScottJ,StrachanM,KilonzoM,ValeL:Qualityoflifebeforeandafterintensivecare.Anaesthesia2005;

60:332-33920. VlesWJ,SteyerbergEW,Essink-BotML,vanBeeckEF,MeeuwisJD,LeenenLP:Prevalenceanddeterminantsofdisabilities

andreturntoworkaftermajortrauma.JTrauma2005;58:126-13521. NielsenD,SellgrenJ,RickstenSE:Qualityoflifeaftercardiacsurgerycomplicatedbymultipleorganfailure.CritCareMed

1997;25:52-57

67

22. VanderZeeKI,SandermanR,HeyinkJ:Acomparisonoftwomultidimensionalmeasuresofhealthstatus:theNottingham

HealthProfileandtheRAND-36ItemHealthSurvey1.0.QualLifeRes1996;5:165-17423. FalcozPE,ChocronS,MercierM,PuyraveauM,EtieventJP:ComparisonoftheNottinghamHealthProfileandthe36-item

healthsurveyquestionnairesincardiacsurgery.AnnThoracSurg2002;73:1222-122824. Black NA, Jenkinson C, Hayes JA, et al: Review of outcomemeasures used in adult critical care. Crit CareMed 2001;

29:2119-212425. Davidson TA, Caldwell ES, Curtis JR,Hudson LD, Steinberg KP: Reducedquality of life in survivors of acute respiratory

distresssyndromecomparedwithcriticallyillcontrolpatients.JAMA1999;281:354-36026. SchellingG,StollC,VogelmeierC,etal:Pulmonary functionandhealth-relatedqualityof life ina sampleof long-term

survivorsoftheacuterespiratorydsitresssyndrome.IntensiveCareMed2000;26:1304-131127. RothenhäuslerHB,EhrentrautS,StollC,SchellingG,KapfhammerHP:Therelationshipbetweencognitiveperformance

and employment and health status in long-term survivors of the acute respiratory distress syndrome: results of anexploratorystudy.GenHospPsychiatry2001;23:90-96

28. KapfhammerHP,RothenhäuslerHB,KrauseneckT,StollC,SchellingG:Posttraumaticstressdisorderandhealth-related

qualityoflifeinlong-termsurvivorsofacuterespiratorydistresssyndrome.AmJPsychiatry2004;161:45-5229. HopkinsRO,WeaverLK,PopeD,OrmeJFJr,BiglerED,Larson-LohrV:Neuropsychologicalsequelaeandimpairedhealth

statusinsurvivorsofsevereacuterespiratorydistresssyndrome.AmJRespirCritCareMed1999;160:50-5630 OrmeJJr,RomneyJS,HopkinsRO,etal:Pulmonaryfunctionandhealth-relatedqualityoflifeinsurvivorsofacute

respiratorydistresssyndrome.AmJRespirCritCareMed2003;167:690-69431. HopkinsRO,WeaverLK,CollingridgeD,ParkinsonRB,ChanKJ,OrmeJFJr:Two-yearcognitive,emotional,andqualityof

lifeoutcomesinacuterespiratorydistresssyndrome.AmJRespirCritCareMed2005;171:340-34732. Heyland DK, Groll D, Caeser M: Survivors of acute respiratory distress syndrome: Relationship between pulmonary

dysfunctionandlong-termhealth-relatedqualityoflife.CritCareMed2005;33:1549-155633. ParkerCM,HeylandDK,GrollD,CaeserM:Mechanismofinjuryinfluencesqualityoflifeinsurvivorsofacuterespiratory

distresssyndrome.IntensiveCareMed2006;32:1895-190034. HerridgeMS,CheungAM,TanseyCM,etal:One-yearoutcomesinsurvivorsoftheacuterespiratorydistresssyndrome.N

EnglJMed2003;348:683-69335. Deja M, Denke C, Weber-Carstens S, et al: Social support during intensive care unit stay might improve mental

impairmentandconsequentlyhealth-relatedqualityoflifeinsurvivorsofsevereacuterespiratorydistresssyndrome.CritCare2006;10:R147

36. Combes A, CostaMA, Trouillet JL, Baudot J,MokhtariM, Gibert C, Chastre J:Morbidity,mortality, and quality of life

outcomesofpatientsrequiring≥14daysofmechanicalventilation.CritCareMed2003;31:1373-138137. ChelluriL,ImKA,BelleSH,etal:Long-termmortalityandqualityoflifeafterprolongedmechanicalventilation.CritCare

Med2004;32:61-69 38. CoxCE,MartinuT,SathySJ,etal:Expectationsandoutcomesofprolongedmechanicalventilation.CritCareMed2009;

37:2888-289439. MillerRS,PattonM,GrahamRM,HollinsD:Outcomesoftraumapatientswhosurviveprolongedlengthsofstay inthe

intensivecareunit.JTrauma2000;48:229-23440. Mackenzie EJ, McCarthy ML, Ditunno JF, Forrester-Staz C, Gruen GS, Marion DW, Schwab WC: Using the SF-36 for

characterizingoutcomeaftermultipletraumainvolvingheadinjury.JTrauma2002;52:527-53441. Dimopoulou I,AnthiA,MastoraZ,etal:Health-relatedqualityof lifeanddisability insurvivorsofmultipletraumaone

yearafterintensivecareunitdischarge.AmJPhysMedRehabil2004;83:171-176

68

42. SluysK,HäggmarkT,IseliusL:Outcomeandqualityoflife5yearsaftermajortrauma.JTrauma2005;59:223-23243. JacksonJC,ObremskeyW,BauerR,etal:Long-termcognitive,emotional,andfunctionaloutcomesintraumaintensive

careunitsurvivorswithoutintracranialhemorrhage.JTrauma2007;62:80-8844. UlvikA,KvåleR,Wentzel-LarsenT,FlaattenH:Qualityoflife2-7yearsaftermajortrauma.ActaAnaesthesiolScand2008;

52:195-20145. RingdalM,PlosK, LundbergD, JohanssonL,Bergbom I:Outcomeafter injury:memories,health-relatedqualityof life,

anxiety,andsymptomsofdepressionafterintensivecare.JTrauma2009;66:1226-123346. SanerH,BornerRodriguezE,Kummer-BangerterA,SchüppelR,vonPlantaM:Qualityoflifeinlong-termsurvivorsofout-

of-hospitalcardiacarrest.Resuscitation2002;53:7-1347. Bunch TJ, White RD, Gersh BJ, et al: Long-term outcomes of out-of-hospital cardiac arrest after successful early

defibrillation.NEnglJMed2003;348:2626-263348. KuilmanM,BleekerJK,HartmanJA,SimoonsML:Long-termsurvivalafterout-of-hospitalcardiacarrest:an8yearfollow-

up.Resuscitation1999;41:25-3149. Graf J,MühlhoffC,DoigGS,etal:Healthcarecosts, long-termsurvival,andqualityof life following intensivecareunit

admissionaftercardiacarrest.CritCare2008;12:R9250. MahapatraS,BunchTJ,WhiteRD,HodgeDO,PackerDL:Sexdifferencesinoutcomeafterventricularfibrillationinout-of-

hospitalcardiacarrest.Resuscitation2005;65:197-20251. Lundgren-NilssonA,RosénH,HofgrenC,SunnerhagenKS:Thefirstyearaftersuccessfulcardiacresuscitation:function,

activity,participationandqualityoflife.Resuscitation2005;66:285-28952. MontuclardL,Garrouste-OrgeasM,TimsitJF,MissetB,DeJongheB,CarletJ:Outcome,functionalautonomy,andquality

oflifeofelderlypatientswithalong-termintensivecareunitstay.CritCareMed2000;28:389-339553. MerlaniP,ChenaudC,MariottiN,RicouB:Long-termoutcomeofelderlypatientsrequiringintensivecareadmissionsfor

abdominalpathologies:survivalandqualityoflife.ActaAnaesthesiolScand2007;51:530-53754. Kaarlola A, TallgrenM, Pettilä V: Long-term survival, quality of life, and quality-adjusted life-years among critically ill

elderlypatients.CritCareMed2006;2120-212655. deRooijSE,GoversAC,KorevaarJC,GiesbersAW,LeviM,deJongeE:Cognitive,functionalandquality-of-lifeoutcomesof

patientsaged80andolderwho survivedat least1 yearafterplannedorunplanned surgeryormedical intensive caretreatment.JAmGeriatrSoc2008;56:816-822

56. Garrouste-Orgeas M, Timsit JF, Montuclard L, et al: Decision-making process, outcome, and 1-year quality of life of

octogenariansreferredforintensivecareunitadmission.IntensiveCareMed2006;32:1045-105157. KleinpellRM:Exploringoutcomesaftercriticalillnessintheelderly.OutcomesManag2003;7:159-16958. Soran A, Chelluri L, Lee KK, Tisherman SA: Outcome and quality of life of patients with acute pancreatitis requiring

intensivecare.JSurgRes2000;91:89-8459. HalonenKI,PettiläV,LeppäniemiAK,KemppainenEA,PuolakkainenPA,HaapiainenRK:Long-termhealth-relatedquality

oflifeinsurvivorsofsevereacutepancreatitis.IntensiveCareMed2003;29:782-78660. HeylandDK,HopmanW, CooH, Tranmer J,McCollMA: Long-termhealth-related quality of life in survivors of sepsis.

ShortForm36:avalidandreliablemeasureofhealth-relatedqualityoflife.CritCareMed2000;28:3599-360561. KarlssonS,RuokonenE,VarpulaT,Ala-KokkoTI,PettiläV:Long-termoutcomeandquality-adjustedlifeyearsaftersevere

sepsis.CritCareMed2009;37:1268-127462. KorošecJagodičH,JagodičK,PodbregarM:Long-termoutcomeandqualityoflifeofpatientstreatedinsurgicalintensive

care:acomparisonbetweensepsisandtrauma.CritCare2006;10:R314

69

63. deBoerAG,GenovesiPI,SprangersMA,vanSandickJW,ObertopH,vanLanschotJJ:Qualityoflifeinlong-termsurvivorsaftercurativetranshiataloesophagectomyforoesophagealcarcinoma.BrJSurg2000;87:1716-1721

64. AhlströmA, TallgrenM, Peltonen S, Räsänen P, Pettilä V: Survival and quality of life of patients requiring acute renal

replacementtherapy.IntensiveCareMed2005;31:1222-122865. YlipalosaariP,Ala-KokkoTI,LaurilaJ,OhtonenP,SyrjäläH:Intensivecareunitacquiredinfectionhasnoimpactonlong-

termsurvivalorqualityoflife:aprospectivecohortstudy.CritCare2007;11:R3566. Orwelius L, Nordlund A, Nordlund P, Edéll-Gustafsson U, Sjöberg F: Prevalence of sleep disturbances and long-term

reducedhealth-relatedqualityoflifeaftercriticalcare:aprospectivemulticentercohortstudy.CritCare2008;12:R9767. PettiläV,KaarlolaA,MäkeläinenA:Health-relatedqualityof lifeofmultipleorgandysfunctionpatientsoneyearafter

intensivecare.IntensiveCareMed2000;26:1473-147968. Badia X, Diaz-Prieto A, GorrizMT, et al: Using the EuroQol-5D tomeasure changes in quality of life 12months after

dischargefromanintensivecareunit.IntensiveCareMed2001;27:1901-1907 69. Stricker KH, CavegnR, Takala J, RothenHU:Does ICU lengthof stay influencequality of life?ActaAnaesthesiol Scand

2005;49:975-98370. GarciaLizanaF,PeresBotaD,DeCubberM,VincentJL:Long-termoutcomesinICUpatients:Whataboutqualityoflife?

IntensiveCareMed2003;29:1286-129371. Graf J,WagnerJ,GrafC,KochKC, JanssensU:Five-yearsurvival,qualityof life,and individualcostsof303consecutive

medicalintensivecarepatients-acost-utilityanalysis.CritCareMed2005;33:547-55572. KaarlolaA,PettiläV,KekkiP:Qualityoflifesixyearsafterintensivecare.IntensiveCareMed2003;29:1294-129973. FlaattenH,KvåleR:Survivalandqualityoflife12yearsafterICU:AcomparisonwiththegeneralNorwegianpopulation.

IntensiveCareMed2001;27:1005-101174. KvåleR,FlaattenH:Changesinhealth-relatedqualityoflifefrom6monthsto2yearsafterdischargefromintensivecare.

HealthQualLifeOutcomes2003;1:275. KvåleR,FlaattenH:Changesinintensivecarefrom1987to1997-hasoutcomeimproved?Asinglecentrestudy.Intensive

CareMed2002;28:1110-111676. Badia X, Diaz-PrietoA, RueM, PatrickDL:Measuring health and health state preferences among critically ill patients.

IntensiveCareMed1996;22:1379-138477. Granja C, Teixeira-Pinto A, Costa-Pereira A: Quality of life after intensive care- evaluation with EQ-5D questionnaire.

IntensiveCareMed2002;28:898-90778. RidleySA,WallacePG:Qualityoflifeafterintensivecare.Anaesthesia1990;45:808-81379. WehlerM,MartusP,GeiseA,BostA,MuellerA,HahnEGetal:Changes inqualityof lifeaftermedical intensivecare.

IntensiveCareMed2001;27:154-15980. RidleySA,ChrispinPS,ScottonH,RogersJ,LloydD:Changesinqualityoflifeafterintensivecare:comparisonwithnormal

data.Anaesthesia1997;52:195-20281. Hofhuis JG, Spronk PE, van Stel HF, Schrijvers AJ, Bakker J: Quality of life before intensive care unit admission is a

predictorofsurvival.CritCare2007;11:R7882. Irribarren-DiarasarriS,Aizpuru-BarandiaranF,Munoz-MartinezT,Loma-OsorioA,Hernandez-LopezM,Ruiz-Zorrilla JM,

etal:Health-relatedqualityoflifeasaprognosticfactorofsurvivalincriticallyillpatients.IntensiveCareMed2009;35:833-839

83. AbelhaFJ,SantosCC,BarrosH:Qualityoflifebeforesurgicaladmission.BMCSurgery2007;7:2384. Rivera-Fernandez R, Sanchez-Cruz JJ, Abizanda-Campos R, Vazquez-Mata G: Quality of life before intensive care unit

admissionanditsinfluenceonresourceutilizationandmortalityrate.CritCareMed2001;29:1701-1709

71

II.Long-termoutcomesandqualityoflifein

criticallyillpatientswithhematologicalor

solidmalignancies

SGOeyen,MD1,DDBenoit,MD,PhD1,2,LAnnemans,PhD2,3,PODepuydt,MD,PhD1,2,SJVanBelle,MD,

PhD2,4,RITroisi,MD,PhD2,5,LANoens,MD,PhD2,6,PPattyn,MD,PhD2,7,JMDecruyenaere,MD,PhD1,2

1DepartmentofIntensiveCareMedicine,GhentUniversityHospital,Ghent,Belgium2FacultyofMedicineandHealthSciences,GhentUniversity,Ghent,Belgium3DepartmentofPublicHealth,GhentUniversity,Ghent,Belgium4DepartmentofMedicalOncology,GhentUniversityHospital,Ghent,Belgium5DepartmentofGeneralandHepato-BiliarySurgery,GhentUniversityHospital,Ghent,Belgium6DepartmentofHematology,GhentUniversityHospital,Ghent,Belgium7DepartmentofGastro-IntestinalSurgery,GhentUniversityHospital,Ghent,Belgium

PublishedinIntensiveCareMedicine2013;39:889-898

72

ABSTRACT

Purpose:Dataconcerning long-termoutcomesandqualityof life (QOL) incritically illcancerpatientsare

scarce. The aims of this studywere to assess long-term outcomes andQOL in critically ill patientswith

hematological (HM) or solid malignancies (SM) 3 months and 1 year after intensive care unit (ICU)

discharge,tocomparethesewithQOLbeforeICUadmission,andtoidentifyprognosticindicatorsoflong-

termQOL.

Methods:Duringa1yearprospectiveobservationalcohortanalysis,consecutivepatientswithHMorSM

admittedtothemedicalorsurgical ICUofauniversityhospitalwerescreenedfor inclusion.Cancerdata,

demographics, co-morbidity, severityof illness,organ failures,andoutcomeswerecollected.QOLbefore

ICUadmission,3months,and1yearafter ICUdischargewasassessedusingstandardizedquestionnaires

(EuroQoL-5D, Medical Outcomes Study 36-item Short Form Health Survey). Statistical significance was

attainedatP<0.05.

Results:483patients(85HM,398SM)(64%men)withamedianageof62yearswereincluded.Mortality

ratesofHMcomparedtoSMwererespectively:hospital(34%vs13%),3months(42%vs17%),and1year

(66%vs36%)(P<0.001).QOLdeclinedat3months,but improvedat1yearalthough it remainedunder

baselineQOL, particularly in HM. Older age (P=0.007), severe comorbidity (P=0.035), and HM (P=0.041)

wereindependentlyassociatedwithpoorerQOLat1year.

Conclusions:Long-termoutcomesandQOLwerepoor,particularlyinHM.Long-termexpectationsshould

playalargerroleduringmultidisciplinarytriagedecisionsuponreferraltotheICU.

73

INTRODUCTION

Theprognosisofpatientswithasolidorhematologicalmalignancyhassubstantiallyimprovedover

the past decades due to advances in diagnostics, antineoplastic therapy and supportive care [1, 2]. In

addition, survivalofcancerpatientsdevelopingcritical illness [1-7]has increasedaswell, including those

requiringmechanicalventilation[8,9]orrenalreplacementtherapy(RRT)[10-12].Asrecentstudieshave

shown that severity and cause of acute illness rather than the underlying cancer characteristics are

predictiveforshort-termmortality[13-18],adiagnosisofcancerassuchshouldnotprecludeadmissionto

theintensivecareunit(ICU).However,tofullyappreciateoutcomesofcriticallyillcancerpatients,indices

regardinglong-termmorbidityandqualityoflife(QOL)afterICUdischargeshouldbetakenintoaccountas

well.

Major reductions in long-term QOL were seen in cases of severe acute respiratory distress

syndrome, prolonged mechanical ventilation, and severe sepsis, representing complications that affect

cancer patients as much as non-cancer patients [19]. In addition, poor performance following ICU

admission in cancer patients may jeopardize long-term outcome by inducing postponements or

cancellationsofpotentiallycurativechemotherapy.

Thusfar,dataaboutQOLpostICUincancerpatients,thoughsorelyneededtoestimatelong-term

prognosis and to assist physicians in triage decisions, are virtually limited to patients with oesophageal

malignancy[20,21],ortoanolderreportconcerningcritically illhematologicalpatients[22]. Theaimof

thepresentstudywastoassess long-termoutcomesofcritically illpatientswithahematologicalorsolid

malignancy, tocompareQOLof thesepatients3monthsand1yearafter ICUdischargewithQOLbefore

ICU,andtoidentifyprognosticindicatorsoftheevolutionofQOLafterdischarge.

MATERIALSANDMETHODS

Design,Setting,andPatients

Thestudywasaprospectiveobservationalcohortanalysisperformedatthe14-bedmedical(MICU)

and 22-bed surgical ICU (SICU) of Ghent University Hospital, Belgium. FromMarch 3rd2008 -March 3rd

2009, all consecutive adult patients (≥ 16 years) with a solid or hematological malignancy as direct or

contributivecauseforICUadmissionwerescreenedforinclusion.Patientswithcompleteremissionfor>5

yearswereexcluded,aswerepatientswhounderwentcardiacsurgery.IncaseofmultipleICUadmissions,

onlythefirstwasconsidered.StudypatientswerepartofalargercohortofICUpatientsrecruitedtostudy

QOLandcost-effectivenessofintensivecare[23].

TheGhentUniversityHospitalICUisrunasa“closed”ICUwherepatientsaretreatedbyateamof

full-timecriticalcarephysicians.DecisionstoadmitapatienttotheICU,aswellastowithdraworwithhold

74

advanced life support are made by the critical care physician together with the referring physician,

consultingthewishesandexpectationsofthepatientandhisrepresentatives.

DataCollectionandDefinitions

Variables collected within the first 24 hours of ICU admission included age, gender, body mass

index (BMI), personal, proxy, and family practitioner contact data (address and phone number(s)), living

status, activity of daily living (ADL) (no limitations, moderate limitations, chair-bound, bedridden), co-

morbidityasmeasuredbytheCharlsonco-morbidity index(this indexwasalsocalculatedwithoutadding

cancer or hematological disease points in order to limit confounding in the multivariate analysis) [24],

hospitalization in the last 6 months before ICU admission, do-not-resuscitate (DNR) codes before ICU

admission,cancerstatus(controlledorremission,uncontrolledornewlydiagnosis,uncontrolledordisease

progression), weight loss (loss of > 10% of the usual body weight) and/or neutropenia (polynuclear

neutrophils < 500/mm3) at ICU admission, main reason for ICU admission, hospital days before ICU

admission,AcutePhysiologyandChronicHealthEvaluation(APACHEII)score[25],SequentialOrganFailure

Assessment(SOFA)score[26],needforinvasivemechanicalventilation,useofanyvasopressors,andneed

forRRT. During ICU stay, SOFA scores, need for invasivemechanical ventilation, vasopressors, RRT, and

DNR-codeswerecollectedonadailybase. ICU lengthofstay (LOS),hospitalLOS,vital statusat ICUand

hospitaldischarges,andvitalstatus3monthsand1yearfollowingICUdischargeswerecollectedforeach

patient.

Thestudywasapprovedbythelocalethicalcommittee.Asignedinformedconsentwasmandatory

foreveryincludedpatient.

Qualityoflife

QOLwasassessedbymeansoftheMedicalOutcomesStudy36-itemShortFormHealthSurvey(SF-

36)andtheEuroQoL-5D(EQ-5D).TheSF-36questionnaire[27,28]contains36itemsmeasuringeightmulti-

itemdomains:physical-(PF),andsocialfunctioning(SF),rolelimitationsduetophysical-(RP),oremotional

problems(RE),mentalhealth (MH),vitality(VT),bodilypain(BP),andgeneralperceptionofhealth(GH).

Two component scores, a physical (PCS) and a mental (MCS), are calculated summary scores where

respectively the physical or the mental domains will account more in the score.We assessed SF-36 as

norm-basedscorestobeabletocomparethemdirectlywiththegeneralhealthypopulation,withagroup-

level rangeof 47-53 consideredas averageornormal. The validity and reliabilityof the SF-36hasbeen

confirmed in the critically ill population, and its use is validated in face-to-face interviews, interview by

phoneorbysendingthequestionnairebyregularmail[29,30].

The EQ-5D is a questionnaire, which measures health in five domains: mobility, self-care, usual

activities, pain/discomfort, and anxiety/depression [31]. Each domain has three levels: no problems,

moderate problems or severe problems. Therefore, patients can be classified into 1 of 243 (35) possible

75

health states.Weconvertedeachhealth state intoautility index (range -0.1584 to1.000) indicating the

preferenceofbeinginahealthstatus.Onavisualanaloguescale,patientscanratetheirperceivedoverall

healthbetween0 and100. Though theEQ-5Dhasbeen lesswell validated in the critically ill population

[32],boththeEQ-5DandtheSF-36wereconsideredassuitable formeasuringQOL incriticalcareat the

BrusselsRoundtablemeeting[30].

QOL was assessed at 3 predefined time points: baseline QOL, 3 months and 1 year after ICU

discharge.Acomputerchartwith ICUdischargedata foreach includedpatientwaskept to respect inan

accurate way the time points of second (3 months) and third (1 year) QOL assessment. Following ICU

admission and study inclusion, a face-to-face interview to assess baselineQOL (defined asQOL 2weeks

beforeICUadmission)wasdoneassoonaspossible.Thisinterviewwaspreferablytakenfromthepatient,

or, whenever impossible due to severity of illness, from the proxy. Threemonths and 1 year after ICU

discharge,patientsorrelativesweresenttheEQ-5DandSF-36surveysbyregularmail;at1year,questions

concerning living situation of the patient, and if the patient was willing to be admitted to an ICU

department again if needed, were added. If the questionnaires were not returned within one month,

patientsorrelativeswerecontactedbyphonetoassessQOLafter1year.Iftherewasnocontactbyphone,

thefamilypractitionerwascontactedtoassessifthepatienthaddiedmeanwhile.

Statisticalanalysis

Valuesareexpressedasmedian(interquartilerange)(IQR)forcontinuousvariablesandasnumber

(%)forcategoricalvariableswhenappropriate.QOLbeforeICUadmissionandcharacteristicsbetweenboth

groups(hematologicalversussolidmalignancy)werecomparedbytheMann-WhitneyUtestforcontinuous

variables and by the Chi-square test for categorical variables. For long-term analysis ofQOL, differences

betweenQOL at baseline (only hospital survivors), at 3months and at 1 year after ICU dischargewere

assessedbyusingChi-square(EQ-5D)orFriedmantest(SF-36).

Linearregressionanalysis(entermethod)wasusedtoassessthemultivariaterelationshipbetween

patient characteristics and themeanutility index, as an indicator forQOL, at 3monthsandat1 year.A

significancelevelofP<0.2intheunivariateanalysiswasspecifiedforincludingvariablesinthemultivariate

model. Stepwise forward and backward elimination regression procedures were used. Variables that

remainedsignificantinthefinalmodelwereconsideredtobeindependentlyassociatedwithQOL3months

and1yearafterICUdischarge.Allstatisticalanalysesweretwo-tailedandcarriedoutwithSPSSv19(SPSS

Inc,Chicago,IL).Atwo-sidedP<0.05wasconsideredsignificant.

RESULTS

CharacteristicsandOutcomesoftheStudyPopulation

76

A total of 483 cancer patients fulfilled inclusion criteria (Figure 1). Forty-one (48%) of the

hematologicalmalignancies(N=85)werehigh-grade(25%non-Hodgkinlymphoma,18%acutemyelogenous

leukemia, 6% acute lymphoblastic leukemia) and 44 (52%) were low-grade (27%multiplemyeloma, 7%

chronic lymphocytic leukemia, 5% Hodgkin’s disease, 5% low-grade non-Hodgkin lymphoma, 4%

myelodysplastic syndrome,1%chronicmyelogenous leukemia,4%other).Within the solid tumorsgroup

(N=398), lower (26%)andhigher (25%)gastrointestinal tumorswere themostcommon followedby lung

(15%),urogenital(8.5%),brain(8%),headandneck(7%)breast(4%)andothertumors(4%).Almosthalfof

thesepatients(46%)hadmetastaticdisease.

Patientcharacteristics,reasonsforICUadmission,organfailureandoutcomesareshowninTable1.

Patients with hematological malignancies had more co-morbidity, had higher severity of illness at

admissionandrequiredmoreorgansupportthansolidtumorpatients;survivalrateswerealsosignificantly

loweratallmeasuredtimepoints.

Qualityoflife

The number of QOL surveyswas respectively 478 (admission), 392 (3months) and 331 (1 year)

whereas corresponding response rates were 99.0%, 75.8% and 99.4% respectively. Mortality increased

during the study course from16.4% (admission) to 21.7% (3months) and to 41.2%at 1 year (Figure 1).

Respectively79%,86%,and79%ofpatientsansweredthequestionnairesthemselvesatthedifferenttime

points(OnlineResource1).

QOLbeforeICUadmissionwasbetterinpatientswithsolidmalignancies,andinhospitalsurvivors

comparedtohospitalnonsurvivorswithineachmalignancygroup(datanotshown).

EQ-5DassessmentsthreemonthsafterICUdischargeshowedthatpatientswithhematologicaland

solidmalignancieshadmoredisabilitiesthanbeforeICUadmission(Figure2).QOLimprovedafter1year,

except for mobility (both malignancy groups) and for anxiety (solid tumors), but remained lower than

baseline.ChangesinQOLovertimeweresignificantinhematologicalpatientsforusualactivities(P<0.001),

and in patientswith solid tumors formobility (P=0.02), self-care (P=0.02), usual activities (P<0.001), and

pain (P<0.001). When comparing both groups, patients with hematological malignancies had more

problems at 3 months (mobility, P<0.001; self-care, P=0.004) and 1 year (mobility, P=0.004; self-care,

P=0.03;usualactivities,P=0.002)afterICUdischarge,exceptforusualactivitiesat3months.

EvolutionsinQOLassessedbytheSF-36areshowninFigure3.Forbothgroups,QOLdecreased3

monthsafter ICUdischargecomparedtobaseline, improvedafter1year,especially thementaldomains,

but remained under the baseline level. At anymoment, QOLwas lower in patientswith hematological

malignancies.Evolution inQOLforpatientswithsolidtumorswassignificant foralldomains(P<0.001for

respectively PCS, PF, RP, BP, VT, SF,MH;P=0.002 forGH;P=0.003 for RE;P=0.006 forMCS)while there

werenosignificantdifferencesinQOLovertimeforhematologicalpatients,exceptVT(P=0.03).

77

Long-term outcomes and utilities, based upon EQ-5Dmeasures, per type of cancer are given in

OnlineResource2.

Additionalquestionsafter1year

Among the one year survivors, patients with hematological malignancies were less likely to live

independentlywithoutadditionalhelp(62%versus79%;P=0.04)andmorewouldrefuseICUreadmission

again (10% versus 3%; P=0.04). 92% of all patients expressed a preference to be readmitted to an ICU

departmentincaseofdeterioration.

Independentpredictorsoflong-termQOL

Multivariate regression analysis showed that poor QOL 3 months after ICU discharge was

independently associated with female gender (P<0.001), higher comorbidity scores (P=0.001),

hematological malignancy (P=0.01), older age (P=0.03), and a higher mean SOFA score during ICU stay

(P=0.04)(OnlineResource3).OneyearafterICUdischarge,QOLwasstillnegativelyinfluencedbyolderage

(P=0.007), higher comorbidity scores (P=0.04), and hematological malignancy (P=0.04). These results

remainedconsistedregardlessofvariablesincludedinthemodel(datanotshown).Beingadmittedtothe

ICUforamedicalorsurgicalreason,orcancerstatushadnoinfluenceonlong-termQOL.

DISCUSSION

In this prospective study on cancer patients requiring ICU admission, in-hospital and 1-year

mortalitywas16%and41%,respectively.QOLmeasuredat3monthsand1yearafterICUdischargedidnot

returntobaselineandwasbelowtheaverageofthatofageneralhealthywesternpopulationatall time

points.

ICU and hospital mortality rates in our study reflect progress made in critical care of cancer

patients,showingfeasibilityofmajorsurgerybackedupbysafepostoperativeorgansupportinsolidtumor

patients,aswellasthepossibilitytoreverseacute,life-threateningcomplicationsinhematologicalpatients

[1-18].However,short-termmortalitymaynotfullyrepresenttheimpactofcriticalillnessandtheefficacy

ofcriticalcare.Whilethe20-30%declineinsurvivalbetweenhospitaldischargeand1yearmayhavebeen

duetotumorprogressionratherthantoadditionalcomplicationsgrafteduponpost-ICUfrailty,itservesto

remindthat1-yearsurvivalprovidesamorerealisticoutcomeestimateinthesepatients.

Themeasures of utility and QOLmay put the gains in survival into a larger perspective. QOL is

increasingly considered to represent a major measure of outcome, whilst being poorly studied in this

particularpatientpopulation.ThreemonthsafterICUdischarge,QOLwasworseoneverydomainoftheSF-

36 and more patients reported problems on the different domains of the EQ-5D, particularly in usual

activitiesandpain.After1year,QOLimproved,especiallyonthementaldomainsbutstillremainedunder

baseline level. The divergence between mental and physical performance probably reflects a gradual

78

process in which patients adapt to a diminished performance status and come to accept their physical

limitations.Thisiswellillustratedbythefactthatthevastmajorityofourpatientswhowerealiveafter1

yearansweredpositivetothequestionwhethertheywouldchoosetobereadmittedtoanICUincaseof

deterioration.

Evidently, malignancy represents a highly diverse spectrum of disease and cancer patients are

heterogeneousinperformancestatusandco-morbidity.Assuch,outcomeshouldbedifferentiatedamong

subgroups. We found important differences between solid tumor patients and hematological patients

relative toco-morbidity, reason for ICUadmission,andseverityof illness.These translated intodifferent

survivalratesandQOLinsurvivors,withhematologicalpatientshavingworseonQOLoneverymomentof

thestudyperiod,andexperiencingnosignificant improvementsbeyond1year.Somesmallerdifferences

couldbediscernedadditionallybetweendifferentcategoriesofmalignancy.Patientswithgastro-intestinal

tumors had highest survival and highest utility after 1 year, a findingwhich is in accordancewith other

studies [21]. In the subgroupswith the highestmortality at one year, namely high-grade hematological

malignancy and head and neck cancer, a remarkable recovery in QOL was seen within the survivors,

howeverprobablyduetosurvivorbias.Thebestlong-termsurvivalwasseeninpatientswithlungcancer,

althoughincontrast,long-termQOLwasratherpoor.

Prognosticationat the individual level incritically ill cancerpatients isextremelydifficultbecause

manyfactorsrelatedtotheunderlyingcancer,theacuteseverityofillness,andprojectionsonfutureanti-

cancertreatmenthavetobetakenintoaccount.Agoodcollaborationwithopencommunicationregarding

theseissuesisthereforemandatorybetweenallpartieswithdifferentexpertiseinvolvedintheICUtriage

decisionmakingprocess[1,5,7,12].Formanyyears,criticalcarephysicianshavebeenreluctanttoadmit

cancerpatientstotheICU[1-3,5,12],mainlybecauseofthehighmortalityatshort-termreportedinolder

seriesbutalsobecauseofthetoooptimisticlong-termsurvivalexpectationsofreferringphysicians[33,34]

andthepoorcommunicationregardingtheseexpectationstothepatient’sand/ortherelatives[33,35,36].

Futurestudiesshouldtrytofocusonthecomplexdynamic interplayofshort-and long-termexpectations

and evolutions in QOL while taking multidisciplinary triage decisions. Evidently, even the most detailed

long-termoutcomeandQOLdatacannotreplaceclinicalevaluationoftheindividualpatientoroverrulea

patient’spersonalview,thoughtheycertainlyassistintakinganinformeddecision.

Thestrengthofthisstudyliesintheaccurateandprospectivelycollecteddata.QOLwasassessed

with validated questionnaires at baseline, which is rarely done in QOL studies but allows for the only

reliable evaluation of evolution inQOL over timewithout recall bias [19]. Very strict time intervals of 3

monthsand1yearafterICUdischargetoassessQOLagain,wererespectedinallpatients.Responserate

wasveryhighandonly2patientswere lost-to-follow-up.Ontheotherhand,some limitationsshouldbe

mentioned. First, single centre data from a university hospital may not reflect general practice, as the

79

volumeofICUadmissionhasbeenshowntorelatetooutcomeinthesepatients[37,38].Second,medical

decisionsleadingtoICUreferralmayhaveselectedforpatientswithbetterprospects;indifferentadmission

of any cancer patient for advanced life support conceivablywill result in aworse long-termQOL. Third,

thereispotentiallylackofstatisticalpowertodetectdifferencesamongtheQOLdomainsinhematological

patients.Fourth,althoughwetriedtoadjustforimportantdifferencesbetweensurgical(scheduledsurgery

59%,emergencysurgery9%)andmedicalpatients (31%) in themultivariate linear regression,wedonot

knowwhether thiswas sufficient tounweave thecomplex interplaybetweenunderlyingmalignancyand

admissiontype.Fifth,wedidnotcollectinformationaboutoncologicalstatusandanticancertherapyafter

ICUdischarge,whichcouldhaveinfluencedlong-termQOL.

CONCLUSIONS

Our study showed that despite substantial immediate survival of cancer patients following ICU

admission, outcome at longer term was more limited, especially with regard to QOL. Long-term

expectationsofmortalityandQOLshouldbe taken intoaccountwhendecidingwhetherornota cancer

patientshouldbeconsideredforreferraltotheICU.

ACKNOWLEDGEMENTS

TheauthorswishtothankthestudynursesPatrickDeBaets,PatsyPriem,JoVandenbossche,and

Daniella Van der Jeught for their tremendous help, motivation, and enthusiasm concerning inclusions,

interviewingpatients,andcallingpatientsorrelatives.TheythankDominiqueVandijck,whodidagreatjob

in helping with the start-up and preparation of the study, inclusions of patients, calling relatives, and

supervisingdatacollection,whileworkingonhisPhD.TheauthorsalsothankChrisDanneelsforhishelpin

settingupthedatabase.

80

Table1.Patientcharacteristics,organfailuresandoutcomes

Allpatients(N=483)

Solidtumor(N=398)

Hematologicalmalignancy(N=85)

P

Characteristics

age,yrs,(median,IQR) 62(54-70) 62(54-69) 60(48-71) 0.31

malegender,N(%) 310(64) 261(84) 49(58) 0.17BMI,kg/m2(median,IQR) 25(22-28) 25(22-27) 25(22-27) 0.87hospitaldayspriortoICU,days(median,IQR)

1(1-3) 1(1-1) 2(0-8) 0.02

Comorbiditylivesathomebeforeadmission,N(%)

478(99) 393(99) 85(100) 0.30

ADL,N(%) nolimitations 297(61) 271(68) 26(31) <0.001

moderatelimitations 157(33) 108(27) 49(58) <0.001chair-bound 13(3) 8(2) 5(6) 0.05bedridden 16(3) 11(3) 5(6) 0.15

hospitalisationinlast6monthsbeforeICU,N(%)

313(65) 254(64) 59(69) 0.33

Charlsoncomorbidityindex(median,IQR)

4(2-8) 6(2-8) 3(2-4) <0.001

Charlsonrecoded(median,IQR)

0(0-1) 0(0-1) 1(0-2) 0.004

Cancerstatus,N(%)controlled/remission 65(13) 36(9) 29(34) <0.001uncontrolled,newlydiagnosis 247(51) 221(56) 26(31) <0.001uncontrolled,recurrence/progression 171(35) 141(35) 30(35) 0.98neutropeniaatICUadmission 32(7) 3(1) 29(34) <0.001weightloss 65(13) 54(14) 11(13) 0.88Typeofadmission,N(%)medical 152(31) 75(19) 77(90) <0.001scheduledsurgery 287(59) 283(71) 4(5) <0.001emergencysurgery 44(9) 40(10) 4(5) 0.12MainreasonforICUadmission,N(%)postoperativecare 331(69) 324(81) 7(8) <0.001respiratoryfailure 63(13) 25(6) 38(45) <0.001septicshock 18(4) 10(3) 8(9) 0.002neurologicaldisorder 12(2) 7(2) 5(6) 0.03metabolicdisorder 11(2) 9(2) 2(2) 0.96MOF 11(2) 2(1) 9(11) <0.001GIhemorrhage 9(2) 9(2) 0(0) 0.16surveillance 7(1) 3(1) 4(5) 0.006cardiovascularcomplications 5(1) 4(1) 1(1) 0.89renalfailure 5(1) 5(1) 0(0) 0.30other 11(2) 0(0) 11(13) <0.001SeverityofillnessatICUadmission(day1)

APACHEIIscore(median,IQR) 15(11-20) 13(11-18) 21(17-29) <0.001SOFAscore(median,IQR) 3(2-5) 3(2-5) 6(3-9) <0.001

OrganfailureduringICUstaymechanicalventilation,N(%) 144(30) 114(29) 30(35) 0.22vasopressors,N(%) 103(21) 71(8) 32(38) <0.001RRT,N(%) 26(5) 14(4) 12(14) <0.001meanSOFAscore(median,IQR) 3(2-5) 3(2-4) 6(4-8) <0.001OutcomesICULOS,days(median,IQR) 3(2-4) 2(2-4) 4(2-9) <0.001readmissions,N(%) 43(9) 32(8) 11(13) 0.15

81

ICUmortality,N(%) 38(8) 20(5) 18(21) <0.001

hospitalLOS,days(median,IQR) 15(10-27) 14(10-24) 25(11-49) 0.001

hospitalmortality,N(%) 79(16) 50(13) 29(34) <0.001

DNRdecisions,N(%) 53(11) 28(7) 25(29) <0.001

3monthsmortality,N(%) 105(22) 69(17) 36(42) <0.001

1yearmortality,N(%) 199(41) 143(36) 56(66) <0.001

N=number;yrs=years;IQR=interquartilerange(25%-75%);BMI=bodymassindex;ICU=intensivecareunit;ADL=activityofdailyliving;Charlsonrecoded=Charlsonco-morbidityindexminuspointsforsolidorhematologicalmalignancy;MOF=multipleorganfailure;GI=gastro-intestinal;APACHE=AcutePhysiologyandChronicHealthEvaluation;SOFA=SequentialOrganFailureAssessment;RRT=renalreplacementtherapy;LOS=lengthofstay;DNR=do-not-resuscitate

82

Figure1.Flowchartofthepatientcohortoverthe1yearstudyperiod,numberofsurveys,andresponserates2414patientsscreenedforinclusionover1year

151refusedtoparticipate(6.3%)244readmissions(10.1%)3patients<16years(0.1%)63patientsaftercardiacsurgery(2.6%)1470nocancerrelatedproblems(60.9%)

483patientsincluded(398solidtumorand85hematologicalmalignancy)Admission 483478surveysbyface-to-faceinterview79nonsurvivors 5refusedtoanswerquestionnaires(16.4%) 99.0%responserate8diedmeanwhile87nonsurvivors 483 4excluded(3livingabroad,1refusedtoanswerquestionnaires)3months 392surveys(373regularmail,19face-to-faceinterviewinhospital)

105nonsurvivors 18responsesofdecease(21.7%) 279completedquestionnaires 75.8%responserate41diedmeanwhile 146nonsurvivors 483 6excluded(4livingabroad,2refusedtoanswerquestionnaires)1year 331surveysbyregularmail199nonsurvivors 53responsesofdecease(41.2%) 276completedquestionnaires(73.9%mail;26.1%phone)

2losttofollow-up(0.6%)99.4%responserate

83

Figure2.EQ-5D:PercentageofpatientswithsomeorextremeproblemsbeforeICU(hospitalsurvivorsonly),3monthsand1yearafterICUdischargeSolidmalignancies

Hematologicalmalignancies

TheX-axis represents thedifferentdimensionsof theEQ-5D.TheY-axis represents thepercentages (%)ofpatientswithsomeorsevereproblemsinarespectivedimension.Chi-squaretestwasusedtocalculateP-valuesperdomainoverthe3differenttimepoints(P<0.05wasconsideredsignificant).Foreachdomain,P-valuesoverthedifferenttimepoints are shown between brackets. (*) = P<0.001; ICU=intensive care unit; discomf= discomfort; anx= anxiety;depress=depressionSolid malignancies: Total numbers of patients at the different time points were respectively: 344 (before ICUadmission,hospitalsurvivorsonly);240(3monthsafterICUdischarge);246(1yearafterICUdischarge)Hematologicalmalignancies:Totalnumbersofpatientsatthedifferenttimepointswererespectively:56(beforeICUadmission,hospitalsurvivorsonly);39(3monthsafterICUdischarge);29(1yearafterICUdischarge)

0102030405060708090100

mobility(0.02) self-care(0.02) ususalacçviçes(*) pain/discomf(*) anx/depress(0.19)

0102030405060708090100

mobility(0.18) self-care(0.47) ususalacçviçes(*) pain/discomf(0.27) anx/depress(0.21)

beforeICU aéer3months aéer1year

84

Figure3.SF-36beforeICU(hospitalsurvivorsonly),3months,and1yearafterICUdischarge:norm-basedscoresSolidmalignancies

Hematologicalmalignancies

The X-axis represents the different domains of the SF-36. The Y-axis represents the norm-based scores (medianvalues) per respective domain. Friedman testwas used to calculate P-values per domain over the 3 different timepoints (P<0.05 was considered significant). For each domain, P-values over the different time points are shownbetweenbrackets;(*)=P<0.001;ICU=intensivecareunit;PCS=physicalcomponentscore;MCS=mentalcomponentscore; PF=physical functioning; RP= role physical; BP= bodily pain; GH= general health; VT= vitality; SF= socialfunctioning;RE=roleemotional;MH=mentalhealthSolid malignancies: Total numbers of patients at the different time points were respectively: 346 (before ICUadmission,hospitalsurvivorsonly);239(3monthsafterICUdischarge);245(1yearafterICUdischarge).Hematologicalmalignancies:Totalnumbersofpatientsatthedifferenttimepointswererespectively:56(beforeICUadmission,hospitalsurvivorsonly);39(3monthsafterICUdischarge);29(1yearafterICUdischarge).

0

10

20

30

40

50

60

70

PCS(*) MCS(0.006) PF(*) RP(*) BP(*) GH(0.002) VT(*) SF(*) RE(0.003) MH(*)

0

10

20

30

40

50

60

70

PCS(0.16) MCS(0.28) PF(0.28) RP(0.32) BP(0.55) GH(0.25) VT(0.03) SF(0.08) RE(0.26) MH(0.31)

beforeICU aéer3months aéer1year

85

Onlineresource1.PersonswhocompletedtheQOLquestionnaires,N(%)

All Solidtumorpatients Hematologicalpatients

Base-line

N=478

3monthsafterICU

dis-chargeN=279

1yearafterICUdischargeN=276

P(*)Base-line

N=394

3monthsafterICUdischargeN=240


P(*)

Base-lineN=84

3monthsafterICUdischargeN=39


P(*)

Patient 378(79) 240(86) 218(79) 0.04 325

(82) 210(88) 197(80) 0.07 53(63) 30(77) 21(72) 0.27

Husband/wife 53(11) 26(9) 39(14) 0.19 37(9) 19(8) 33(13) 0.11 16(19) 7(18) 6(21) 0.96

Son/daughter 32(7) 9(3) 10(4) 0.05 24(6) 7(3) 10(4) 0.16 8(10) 2(5) 0(0) 0.20

Father/mother 6(1) 0(0) 2(1) 0.16 2(1) 0(0) 1(1) 0.56 4(5) 0(0) 1(3) 0.40

Otherfamily,friends 9(2) 4(2) 7(3) 0.64 6(2) 4(2) 6(2) 0.69 3(4) 0(0) 1(3) 0.50

QOL=qualityoflife;N=number;ICU=intensivecareunit;P(*):P-valueoverthe3differenttimepoints

86

OnlineResource2.Outcomeandutilityindex(baseduponEQ-5D)pertypeofcancer

Typeofcancer N Hospital

mortality(%)

Mortality3monthsafterICUdischarge(%)

Mortality1yearafterICUdischarge(%)

Utilityindexatbaseline(ICUsurvivors)(*)

Utilityindex3monthsafterICUdischarge(*)

Utilityindex1yearafterICUdischarge(*)

LowerGI

102 8.8 10.8 31.4 0.76(0.53-1.00) 0.73(0.63-1.00) 0.75(0.65-1.00)

UpperGI

99 12.1 15.2 33.3 0.74(0.42-1.00) 0.71(0.57-0.80) 0.73(0.63-0.95)

Lung

73 9.6 15.1 24.7 0.74(0.43-1.00) 0.70(0.56-0.76) 0.71(0.56-0.76)

Urogenital

34 8.8 26.5 41.2 0.74(0.37-0.77) 0.73(0.55-0.77) 0.66(0.49-0.82)

Brain

31 16.1 22.6 41.9 0.77(0.76-1.00) 0.69(0.57-1.00) 0.73(0.56-1.00)

Headandneck

26 30.8 38.5 65.4 0.77(0.51-1.00) 0.55(0.33-0.91) 0.79(0.60-1.00)

Breast

16 18.8 18.8 37.5 0.66(0.20-1.00) 0.56(0.19-0.74) 0.70(0.63-1.00)

OthersolidT

17 17.6 17.6 58.8 0.74(0.32-0.83) 0.69(0.52-0.94) 0.69(0.56-0.77)

HighgradeHM

41 41.5 46.3 68.3 0.71(0.29-0.95) 0.66(0.39-0.74) 0.66(0.64-0.82)

LowgradeHM 44 27.3 38.6 63.6 0.66(0.29-0.76) 0.33(0.19-0.68) 0.60(0.14-0.77)

(*)Utilityindexisexpressedasmedian(interquartilerange);N=number;ICU=intensivecareunit;GI=gastro-intestinal;T=tumors;HM=hematologicalmalignancy

87

Onlineresource3.Univariateandmultivariateanalysesoffactorsassociatedwithlong-termQOLFactorsassociatedwithutility(asindicatorforQOL)3monthsafterICUdischarge Univariate Multivariate

R2=0.151Variable R2 Β(SE)

95%CI P 95%CI P

age(peryear)

0.008 -0.002(0.001) -0.004to0.001 0.14 -0.005to0.000 0.03

femalegender 0.026 -0.10(0.04) -0.16to-0.03 0.007 -0.19to-0.05 <0.001Charlsonrecoded 0.043 -0.04(0.01) -0.06to-0.02 0.001 -0.06to-0.02 0.001hospitaldayspriorICU

0.034 -0.01(0.003) -0.02to-0.004 0.002

cancerstatus controlled

disease- - - reference

uncontrollednewdiagnosis

0.031 0.16(0.05) 0.05to0.26 0.003

uncontrolledrecurr/progr

0.031 0.12(0.06) 0.009to0.23 0.03

STvsHM 0.058 -0.20(0.05) -0.29to-0.10 <0.001 -0.23to-0.03 0.01surgical/medical 0.052 -0.16(0.04) -0.23to-0.08 <0.001 emergencysurgeryversusother

0.009 -0.11(0.07) -0.25to0.03 0.11

APACHEII 0.071 -0.12(0.003) -0.017to-0.007 <0.001 SOFAday1 0.031 -0.02(0.006) -0.03to-0.006 0.003 SOFAmean 0.045 -0.03(0.008) -0.04to-0.01 <0.001 -0.03to-0.001 0.04Factorsassociatedwithutility(asindicatorforQOL)1yearafterICUdischarge Univariate Multivariate

R2=0.057Variable R2 Β(SE)

95%CI P 95%CI P

age(peryear)

0.021 -0.003(0.001) -0.005to0.000 0.02 -0.005to-0.001 0.007

femalegender 0.002 0.026(0.033) -0.04to-0.09 0.43 Charlsonrecoded 0.022 -0.03(0.13) -0.06o-0.006 0.02 -0.05to-0.002 0.04

cancerstatus controlleddiseae - - - reference uncontrollednew

diagnosis0.036 0.15(0.05) 0.05to0.25 0.004

uncontrolledrecurr/progr

0.036 0.17(0.05) 0.06to0.27 0.002

STvsHM 0.013 -0.09(0.05) -0.19to0.05 0.06 -0.20to-0.004 0.04surgical/medical 0.011 -0.07(0.04) -0.16o0.008 0.08 SOFAmean 0.001 -0.004(0.008) -0.02to0.01 0.60 QOL=qualityof life; ICU= intensive careunit;R2= (Pearsoncorrelation coefficient)2; SE= standarderror;CI= confidence interval;Charlsonrecoded=Charlsonco-morbidity indexminuspoints forsolidorhematologicalmalignancy;ST=solid tumor;vs=versus;HM=hematologicalmalignancy;recurr=recurrence;progr=progression;APACHE=AcutePhysiologyandChronicHealthEvaluation;SOFA=SequentialOrganFailureAssessment

88

REFERENCES

1. AzoulayE,SoaresM,DarmonM,BenoitD,PastoresS,AfessaB(2011)Intensivecareofthecancerpatient:recentachievementsandremainingchallenges.AnnIntensiveCare1:52. SoaresM,CarusoP,SilvaE,TelesJM,LoboSM,FriedmanG,DalPizzolF,MelloPV,BozzaFA,SilvaUV,TorellyAP,KnibelMF,etal(2010)Characteristicsandoutcomesofpatientswithcancerrequiringadmissiontointensivecareunits:Aprospectivemulticenterstudy.CritCareMed38:9-153. Benoit DD, Vandewoude KH, Decruyenaere JM, Hoste EA, Colardyn F (2003) Outcome and early prognosticindicatorsinpatientswithahematologicmalignancyadmittedtotheintensivecareunitforalife-threateningcomplication.CritCareMed31:104-1124. PèneF,PercheronS,LemialeV,ViallonV,ClaessensYE,MarquéS,CharpentierJ,AngusDC,CariouA,ChicheJ-D,Mira JP (2008) Temporal changes in management and outcome of septic shock in patients with malignancies in theintensivecareunit.CritCareMed36:690-6965. Vandijck DM, Benoit DD, Depuydt PO, Offner FC, Blot SI, Van Tilborgh AK, Nollet J, Steel E, Noens LA,DecruyenaereJM(2008)Impactofrecentintravenouschemotherapyonoutcomeinseveresepsisandsepticshockpatientswithhematologicalmalignancies.IntensiveCareMed34:847-8556. DarmonM,ThieryG,CiroldiM,deMirandaS,GalicierL,RaffouxE, LeGall JR,SchlemmerB,AzoulayE (2005).Intensivecareinpatientswithnewlydiagnosedmalignanciesandaneedforcancerchemotherapy.CritCareMed33:2488-2937. Benoit DD, Depuydt PO, Vandewoude KH, Offner FC, Boterberg T, De Cock CA, Noens LA, Janssens AM,Decruyenaere JM (2006) Outcome in severely ill patients with hematological malignancies who received intravenouschemotherapyintheintensivecareunit.IntensiveCareMed32:93-998. Soares M, Salluh JI, Spector N, Rocco JR (2005) Characteristics and outcomes of cancer patients requiringmechanicalventilatorsupportfor>24hrs.CritCareMed33:520-5269. GristinaGR,AntonelliM,ContiG,CiarloneA,RoganteS,RossiC,BertoliniG(2011)Noninvasiveversus invasiveventilation for acute respiratory failure in patientswith hematologicalmalignancies; a 5 yearmulticenter observationalsurvey.CritCareMed39:2232-223910. BenoitDD,HosteEA,DepuydtPO,OffnerFC,LameireNH,VandewoudeKH,DhondtAW,NoensLA,DecruyenaereJM(2005)Outcomeincriticallyillmedicalpatientstreatedwithrenalreplacementtherapy;comparisonbetweenpatientswithandthosewithouthaematologicalmalignancies.NephrolDialTransplant20:552-55811. SoaresM,SalluhJI,CarvalhoMS,DarmonM,RoccoJR,SpectorN(2006).Prognosisofcritically illpatientswithcancerandacuterenaldysfunction.JClinOncol24:4003-401012. BenoitDD,HosteEA(2010)Acutekidneyinjuryincriticallyillpatientswithcancer.CritCareClin26:151-17913. SoaresM,SalluhJI,TorresVB,LealJV,SpectorN(2008)Short-andlong-termoutcomesofcriticallyillpatientswithcancerandprolongedICUlengthofstay.Chest134:520-52614. MerzTM,SchärP,BühlmannM,Takala J,RothenHU (2008)Resourceuseandoutcome incritically illpatientswithhematologicalmalignancy:aretrospectivecohortstudy.CritCare12:R7515. McGrathS,ChatterjeeF,WhiteleyC,OstermannM(2010)ICUand6-monthoutcomeofoncologypatientsintheintensivecareunit.QJMed103:397-40316. MokartD,EtienneA,EsterniB,BrunJP,Chow-ChineL,SanniniA,FaucherM,BlacheJL(2012)Criticallyillcancerpatientsintheintensivecareunit:short-termoutcomeand1-yearmortality.ActaAnaesthesiolScand56:178-18917. RosolemMM,RabelloLS,LisboaT,CarusoP,CostaRT,LealJV,SalluhJI,SoaresM(2012)Criticallyillpatientswithcancerandsepsis:Clinicalcourseandprognosticfactors.JCritCare27:301-30718. BirdGT,Farquhar-SmithP,WigmoreT,PotterM,GruberPC(2012)Outcomesandprognosticfactorsinpatientswithhaematologicalmalignancyadmittedtoaspecialistcancerintensivecareunit:a5yrstudy.BrJAnaesth108:452-459

89

19. OeyenSG,VandijckDM,BenoitDD,AnnemansL,DecruyenaereJM(2010)Qualityoflifeafterintensivecare:asystematicreviewoftheliterature.CritCareMed38:2386-240020. CenseHA,HulscherJB,deBoerAG,DongelmansDA,TilanusHW,ObertopH,SprangersMA,vanLanschotJJ(2006)Effectsofprolonged intensivecareunit stayonqualityof lifeand long-termsurvivalafter transthoracicesophageal resection.CritCareMed34:354-36221. deBoerAG,GenovesiPI,SprangersMA,VanSandickJW,ObertopH,VanLanschotJJ(2000)Qualityof lifeinlong-termsurvivorsaftercurativetranshiataloesophagectomyforoesophagealcarcinoma.BrJSurg87:1716-172122. YauE,RohatinerAZ, ListerTA,HindsCJ (1991)Long termprognosisandqualityof life following intensivecare for life-threateningcomplicationsofhaematologicalmalignancy.BrJCancer64:938-94223. OeyenS,BenoitD,AnnemansL,DecruyenaereJ(2010)Qualityoflifebefore,3months,and1yearafterICUdischarge.CritCareMed38:P587(supplDecember)24. CharlsonME,PompeiP,AlesKL,MackenzieCR(1987)Anewmethodofclassifyingprognosticcomorbidityinlongitudinalstudies:developmentandvalidation.JChronDis40:373-38325. KnausWA,DraperEA,WagnerDP,ZimmermanJE(1985)Prognosisinacuteorgan-systemfailure.AnnSurg202:685-69326. VincentJL,MorenoR,TakalaJ,WillattsS,DeMendonçaA,BruiningH,ReinhartCK,SuterPM,ThijsLG(1996)TheSOFA(Sepsis-relatedOrganFailureAssessment)scoretodescribeorgandysfunction/failure.OnbehalfoftheWorkingGrouponSepsis-relatedProblemsoftheEuropeanSocietyofIntensiveCareMedicine.IntensiveCareMed22:707-71027. WareJEJr,SherbourneCD(1992)TheMOS-36itemshort-formhealthsurvey(SF-36).I.Conceptualframeworkanditemselection.MedCare30:473-48328. McHornyCA,WareJEJr,RaczekAE(1993)TheMOS36-itemshort-formhealthsurvey(SF-36):II.Psychometricandclinicaltestsofvalidityinmeasuringphysicalandmentalhealthconstructs.MedCare31:247-26329. Chrispin PS, Scotton H, Rogers J, Lloyd D, Ridley SA (1997) Short-Form 36 in the intensive care unit: Assessment ofacceptability,reliabilityandvalidityofthequestionnaire.Anaesthesia52:15-2330. AngusDC,CarletJ,2002BrusselsRoundtableParticipants(2003)SurvivingIntensiveCare:Areportfromthe2002BrusselsRoundtable.IntensiveCareMed29:368-37731. TheEuroQolGroup(1990)EuroQOL-Anewfacilityforthemeasurementofhealth-relatedqualityoflife.HealthPolicy16:199-20832. Brazier J, Jones N, Kind P (1993) Testing the validity of the Euroqol and comparing it with the SF-36 health surveyquestionnaire.QualLifeRes2:169-18033. LamontEB,ChristakisA (2001)Prognosticdisclosure topatientswithcancernear theendof life.Ann InternMed134:1096-110534. LambBW,SevdalisN,MostafidH,VincentC,GreenJS(2011)Quality improvement inmultidisciplinarycancerteams:Aninvestigationofteamworkandclinicaldecision-makingandcross-validationofassessments.AnnSurgOncol18:3535-354335. Audry S, Abel J, Blazeby JM, Falk S, Campbell R (2008) What oncologists tell patients about survivals benefits ofchemotherapyandimplicationsforinformedconsent:qualitativestudy.BMJ337:a75236. WeeksJC,CatalanoPJ,CroninA,FinkelmanMD,MackJW,KeatingNL,SchragD(2012)Patients'ExpectationsaboutEffectsofChemotherapyforAdvancedCancer.NEnglJMed367:1616-162537. Lecuyer L, Chevret S, Guidet B, Aegerter P, Martel P, Schlemmer B, Azoulay E (2008) Case volume and mortality inhaematologicalpatientswithacuterespiratoryfailure.EurRespirJ32:748-75438. Zuber B, Tran TC, Aegerter P, Grimaldi D, Charpentier J, Guidet B,Mira JP, Pene F (2012) Impact of case volume onsurvivalofsepticshockinpatientswithmalignancies.CritCareMed40:55-62

91

III.Long-termqualityoflifeincriticallyill

patientswithacutekidneyinjurytreated

withrenalreplacementtherapy:Amatched

cohortstudy

SandraOeyen,MD1,2,a,WouterDeCorte,MD1,3,a,DominiqueBenoit,MD,PhD1,2,LievenAnnemans,PhD1,4,

AnnemiekeDhondt,MD,PhD1,5,RaymondVanholder,MD,PhD,ProfessorEmeritus1,5,JohanDecruyenaere,

MD,PhD1,2,EricHoste,MD,PhD1,2

aSandraOeyenandWouterDeCorteequallycontributedtothisstudyandarejointfirstauthors.1FacultyofMedicineandHealthSciences,GhentUniversity,Ghent,Belgium2DepartmentofIntensiveCare,GhentUniversityHospital,Ghent,Belgium3DepartmentofAnaesthesiaandIntensiveCareMedicine,AZGroeningeHospital,Courtray,Belgium4DepartmentofPublicHealth,GhentUniversity,Ghent,Belgium5DepartmentofNephrology,GhentUniversityHospital,Ghent,Belgium

PublishedinCriticalCare2015;19:289

92

ABSTRACT

Introduction:Acutekidneyinjury(AKI)isacommoncomplicationinintensivecareunit(ICU)patientsand

associatedwith increasedmorbidity andmortality.We compared long-term outcome and quality of life

(QOL) in ICUpatientswithAKI treatedwith renal replacement therapy (RRT)withmatchednonAKI-RRT

patients.

Methods:During1yearadultICUpatientsconsecutivelywereincludedinaprospectivecohortstudy.AKI-

RRTpatientsaliveat1yearand4yearswerematchedwithnonAKI-RRTsurvivorsfromthesamecohortin

a1:2 (1year)and1:1 (4years) ratioongender,age,APACHE II score,andadmissioncategory.QOLwas

assessedbytheEuroQoL-5DandtheShortForm-36surveybeforeICUadmissionandat3months,1and4

yearsafterICUdischarge.

Results:Of1953patients,121(6.2%)hadAKI-RRT.AKI-RRThospitalsurvivors(44.6%;N=54)hada1-year

and 4-year survival rate of 87.0% (N=47) and 64.8% (N=35) respectively. Forty-seven 1-year AKI-RRT

patientswerematchedwith 94 1-year non AKI-RRT patients. Of 35 4-years survivors 3 refused further

cooperation, 3 were lost-to-follow-up, and 1 had no control. Finally, 28 4-years AKI-RRT patients were

matchedwith 28 non AKI-RRT patients. During ICU stay, 1-year and 4-years AKI-RRT patients hadmore

organdysfunctioncomparedtotheirrespectivematches(SOFAscores7vs.5,P<0.001;7vs.4,P<0.001).

Long-termQOLwashowevercomparablebetweenbothgroupsbutlowerthaninthegeneralpopulation.

QOLdecreasedat3months,improvedafter1and4yearsbutremainedunderbaselinelevel.Respectively

1 and 4 years after ICUdischarge, 19.1% and 28.6%ofAKI-RRT survivors remainedRRTdependent, and

81.8%and71%ofthemwerewillingtoundergoICUadmissionagainifneeded.

Conclusion:Inlong-termcriticallyillAKI-RRTsurvivors,QOLwascomparabletomatchedlong-termcritically

ill non AKI-RRT survivors, but lower than in the general population. The majority of AKI-RRT patients

wantedtobereadmittedtotheICUwhenneeded,despiteahigherseverityofillnesscomparedtomatched

nonAKI-RRTpatients,anddespitethefactthatonequarterhadpersistentdialysisdependency.

93

INTRODUCTION

Acutekidneyinjurytreatedwithrenalreplacementtherapy(AKI-RRT)affectsapproximately5-10%

ofintensivecareunit(ICU)patients[1].ThesepatientsareamongstthemostseverelyillpatientsintheICU,

asmay be illustrated by the 50% in-hospitalmortality [2-4]. AKI-RRT patientswho survivemay develop

chronickidneydisease,includingendstagerenaldisease,andexperiencedecreasedlong-termsurvival[4-

8]. Therefore, to fullyappreciateoutcomesof critically illAKI-RRT survivors, indices regarding long-term

morbidityandqualityoflife(QOL)shouldbetakenintoaccountaswell[9,10].

Major reductions in long-term QOL in critically ill patients are seen in severe acute respiratory

distresssyndrome,prolongedmechanicalventilation,severesepsis,andaftermajortrauma,allconditions

frequentlyassociatedwithAKI-RRT[11].DataregardingQOLinAKI-RRTpatientsshowthatthesepatients

haveadecreasedQOLcomparedtothegeneralpopulationbutperceiveQOLasgood[12,13].However,

these studieswere either retrospective [14-17], evaluatedQOL after a short term [12-15, 17-21], lacked

baselineQOLassessment[12-15,18,22],ordatedbackmorethanadecade[14-16,18,23].Itisalsounclear

whether impairment in long-term QOL is the consequences of critical illness, AKI-RRT, pre-existing co-

morbidities,oracombinationofthese.

The aimof thepresent studywas to assess long-termoutcomes andQOLof critically ill AKI-RRT

patientsatbaseline,andat3months,1yearand4yearsafterICUdischargeandtocompareQOLwitha

cohortofmatchednonAKI-RRTpatients[24].

METHODS

Design,Patients,andSetting

Thecohortdescribedinthisstudyisasubgroupofaprospectiveobservationalcohort.Duringone

year (March2008-March2009),all consecutivelyadmittedadultpatientsat the14-bedmedical (MICU),

the22-bed surgical ICU (SICU), and the6-bedburnunit of theGhentUniversityHospital, Belgium,were

screenedtostudyQOLandcost-effectivenessofintensivecare[25].Exclusioncriteriawereage<16yand

admissiontotheICUaftercardiacsurgery.IncaseofmultipleICUadmissions,onlythefirstwasconsidered.

In this study, only AKI-RRT patients of the larger cohort were included. Chronic hemodialysis

patients were excluded. The attending critical care physician and consulting nephrologist assessed

indicationforRRTandmodality.

TostudytheimpactofRRTonlong-termoutcomeandQOL,weperformedamatchedcohortstudy,

accordingtotheSTROBEguidelines[26].IncludedAKI-RRTpatientsaliveat1yearafterhospitaldischarge

weredefinedasexposedpatientsandindividuallymatchedwith1-yearnonAKI-RRTsurvivors(definedas

non-exposed patients) from the same cohort. Being a patient in the non AKI-RRT group did not imply

normalkidneyfunction;itimpliednotreatmentwithRRT.Tocorrectforpossiblebias,weexcludedpatients

94

who needed RRT butwho did not receive RRT due to therapeutic restrictions. Equally, AKI-RRT patients

alive at time of this study (average 4 years later) were individually matched with 4-years non AKI-RRT

survivors.Theexposed:non-exposedratiowasaimedat1:2 to reduceriskof selectionbias.Whenthere

weremorethan2non-exposedpatientsforanexposedpatient,onlythenon-exposedpatientwiththebest

overall match was selected. If an exposed patient could only be properly matched to 1 non-exposed

patient,weacceptedmatching ina1:1 ratio for the respectivecohort inorder toavoidan imbalanceof

characteristics and to retain thebest possiblematching.Matchingwasbasedon gender, age (±5 years),

AcutePhysiologyandChronicHealthEvaluation(APACHEII)score(±5),andadmissioncategory.


Variables collected within the first 24 hours of ICU admission included age, gender, body mass

index, personal, proxy, and family practitioner contact data, living situation, activity of daily living, co-

morbidityasmeasuredbytheCharlsonco-morbidityindex[27],hospitalizationinthelast6months,main

reason for ICU admission, APACHE II score [28], SequentialOrgan Failure Assessment (SOFA) score [29],

needformechanicalventilation,useofanyvasopressors,andneedforRRT.DuringICUstaySOFAscores,

needformechanicalventilation,vasopressors,RRT,anddo-not-resuscitatecodeswerecollectedonadaily

base.ICUlengthofstay(LOS),hospitalLOS,vitalstatusatICUandhospitaldischarge,andat3months,1

yearand4yearsfollowingICUdischargewerecollectedforeachpatient.

Valuesof serumcreatinineofAKI-RRTpatientswereextracted fromtheSTARRTdatabase,which

includes all relevant renal and RRT data of ICU patientswith AKI–RRT treated in our hospital, and from

laboratorydataincontrolpatients.Theestimatedglomerularfiltrationrate(eGFR)wascalculatedusingthe

Chronic Kidney Disease Epidemiology Collaboration formula [30]. Renal recovery was defined as

independencefromRRT.

Thestudywasapprovedbythelocalethicalcommittee(EthischComitéGhentUniversityHospital;

amendment project 2007/423 approved February 19th, 2013) (B67020072805), and conducted in

accordancewiththedeclarationofHelsinki.Asignedinformedconsentwasobtainedfromeveryincluded

patient.

Qualityoflife

QOLwasassessedbymeansoftheMedicalOutcomesStudy36-itemShortFormHealthSurvey(SF-

36v2®) and the EuroQoL-5D (EQ-5D). The SF-36 questionnaire contains 36 items measuring 8 health

domains: physical- (PF), and social functioning (SF), role limitations due to physical- (RP), or emotional

problems (RE),mental health (MH), vitality (VT), bodily pain (BP), and general perceptionof health (GH)

[31].Twocomponentscores,aphysical (PCS)andamental (MCS),arecalculatedsummaryscoreswhere

respectively the physical domains (PF, RP, BP, GH) or themental domains (VT, SF, RE,MH)will account

moreinthescore.WeassessedSF-36asnorm-basedscorestobeabletocomparethemdirectlywiththe

95

generalhealthypopulation,withagroup-levelrangeof47-53consideredasaverageornormal[31].Group

scoreslessthan47indicateimpairedfunctioningwithinthathealthdomain;groupscoresgreaterthanor

equalto53shouldbeconsideredaverageorabovethenormativesample.

The36thitem,healthtransition,providesinformationaboutperceivedchangesinhealthstatus.The

validity and reliabilityof theSF-36hasbeen confirmed in critically ill patients, and itsuse is validated in

face-to-faceinterviews,interviewbyphoneorbysendingthequestionnairebyregularmail[32].

TheEQ-5DisagenericQOLquestionnairethatmeasureshealth infivedimensions:mobility,self-

care,usual activities, pain/discomfort, andanxiety/depression [33]. Eachdimensionhas three levels: no

problems,moderateproblemsorsevereproblems.Onavisualanaloguescale(VAS),patientscanratetheir

perceivedoverallhealthbetween0and100.TheEQ-5DissuitableformeasuringQOLincriticalcare[34,

35].

QOLwasassessedatdifferenttimepoints:baselineQOLandatstrictly3monthsand1yearafter

ICUdischarge.QOLwasalsoassessedinAugust2013,amedianof4.1years(3.9years–4.3years)afterICU

discharge. Following ICU admission and study inclusion, a face-to-face interview to assess baselineQOL

(definedasQOL2weeksbeforeICUadmission)wasdoneassoonaspossible.Thisinterviewwaspreferably

takenfromthepatient,orwhenimpossible, fromtheproxy.Threemonths,1year,and4yearsafter ICU

discharge, patientswere sent the EQ-5D and SF-36 surveys by regularmail; at 1 and 4 years, questions

concerning living situation, memories, sleep quality, and willingness to be readmitted to an ICU

department,wereadded.Ifthequestionnaireswerenotreturnedwithinonemonth,patientsorrelatives

werecontactedbyphonetoassessQOLafter1yearandafter4years.Eventually,thefamilypractitioner

wascontacted.

Statisticalanalysis

Dataareexpressedasmedian (interquartile range) (IQR) forcontinuousvariablesandasnumber

(%)forcategoricalvariables.QOLatthedifferenttimepointsandcharacteristicsbetweenbothgroups(AKI-

RRTversusnonAKI-RRTpatients)werecomparedbytheMann-WhitneyUtestforcontinuousvariablesand

bytheChi-squaretestforcategoricalvariables.Forlong-termanalysisofQOL,differencesbetweenQOLat

baseline(onlyhospitalsurvivors),at3months,at1and4yearsafterICUdischargewereassessedbyChi-

square (EQ-5D)orFriedmantest (SF-36). P-valueswere two-sidedandstatistical significancewas setat

0.05.AllstatisticalanalysesweredoneusingIBMSPSSStatisticssoftwareversion21.

RESULTS

Characteristicsofthestudypopulation

Duringthe1-yearstudyperiod1953patientswereincluded(Figure1).Onehundredforty-

sevenpatients (7.5%)developedAKIwithneed forRRT.Of these, 121patients (6.2%) receivedRRT. ICU

96

(46.3%), hospital (55.4%), 3months (57.9%), 1-year (61.1%) and4-years (71.1%)mortality rates in these

patientswerehigh.Twenty-sixAKIpatients(1.3%)didnotreceiveRRTduetotherapeuticrestrictionsand

wereexcludedforfurtheranalysis.

AKI-RRT hospital survivors (44.6%) had a 1-year and 4-years survival rate of 87.0% and 64.8%

respectively.Forty-seven1-yearAKI-RRTsurvivorswere individuallymatchedwith941-yearnonAKI-RRT

survivors(2matchesforallAKI-RRTpatients).Of354-yearssurvivors3refusedfurthercooperation,3were

lost-to-follow-up,and1hadadoublematch. In13of the28 included4-yearsAKI-RRTsurvivorsonlyone

goodmatchcouldbewithhold, somatchingoccurred ina1:1 ratio. Finally,284-yearsAKI-RRT survivors

were individuallymatchedwith 28 non AKI-RRT patients. AKI-RRT and non AKI-RRT patients had similar

gender,age,APACHEIIscore,andadmissioncategoryat1yearand4years(Table1).

During ICU stay, 1-year and 4-years AKI-RRT patients had higher SOFA scores compared to their

respectivematches,andmoreneededmechanicalventilationorvasopressorsforalongertime(Table1).

Renalcharacteristicsandrenaloutcomes

One year AKI-RRT patients had higher baseline serum creatinine concentrations and lower eGFR

comparedtotheirmatches.Thesemeasurementsdidnotsignificantlydifferbetween4-yearsAKI-RRTand

nonAKI-RRTpatients(Table1).

Respectively 12 1-year (25.5%) and 10 4-years AKI-RRT patients (35.7%)were RRT dependent at

hospitaldischarge.Nine(19.1%)ofthe1-yearand8(28.6%)ofthe4-yearsAKI-RRTpatientsremainedRRT

dependentovertime.

Qualityoflife

AnoverviewofthepersonswhoratedQOL,howQOLwasassessedandthenumberofcompleted

QOLsurveysisgiveninTable2.MostpatientsratedtheirownQOLatthedifferenttimepoints,exceptat

baselinein1-yearAKI-RRTpatients.

Significantdifferences inQOLbetweenAKI-RRTandnonAKI-RRT survivorsateachdifferent time

point were small. Figure 2 and Figure 3 show that the 1-year AKI-RRT versus (vs) 1-year non AKI-RRT

patientshadcomparablebaselineQOL.The1-yearAKI-RRTpatientswerepooreremotionallyat3-months

(RE28.7vs38.4;P=0.035),buthadabettermentalscore(MCS53.3vs47.8;P=0.039)andlessbodilypain

(BP46.5vs41.6;P=0.041)at1year(Figure3).Figure4and5showthatthe4-yearsAKI-RRTvs4-yearsnon

AKI-RRTpatientswereemotionallybetteratbaseline (RE55.9vs40.3;P=0.030) (Figure5),buthadmore

problems with usual activities (81.0% vs 47.8%; P=0.023), pain (71.4% vs 26.1%; P=0.003) and anxiety

(61.9%vs17.4%;P=0.002)at3months(Figure4).QOLafter1and4yearsshowednodifferences(Figure4

andFigure5).

ComparingQOLwithineachgroupbetweenthedifferenttimepointsrevealedthatQOLparticularly

decreasedafter3months.

97

EvolutioninQOLovertime:1year-cohort

All 1-year AKI-RRT patients reportedmore problems on the EQ-5D after 3months compared to

baseline.After1year,theyexperiencedfewerproblemsbutstillmorethanbeforeICUadmission.TheEQ-

5Dshowedthesameevolutionfor1-yearnonAKI-RRTpatients(AdditionalFile1A/1B).

TheSF-36showedsignificantevolutionsinQOLovertimefor1-yearAKI-RRTpatients innearlyall

dimensions.QOLdecreasedafter3months,improvedafter1yearbutwithoutreturntothebaselinelevel.

QOLalsoremainedundertheleveloftheaveragepopulation.Thesamepattern,althoughlesspronounced,

wasseenin1-yearnonAKI-RRTpatients(AdditionalFile2A/2B).

For1-yearAKI-RRTpatientsmedianVASscoresrangedfrom70(baseline),to60(3months)and70

(1 year) (P=0.048). In non AKI-RRT patients the VAS remained the same, respectively 68, 65 and 65 at

baseline,3monthsand1yearafterICUdischarge(P=0.917).

EvolutioninQOLovertime:4years-cohort

ChangesinQOLovertimeassessedbytheEQ-5DweresignificantinAKI-RRTpatientsformobility

(P=0.040),usualactivities(P<0.001),andanxiety(P=0.040)(AdditionalFile1C)andin4-yearsnonAKI-RRT

patients formobility (P=0.017), andusual activities (P=0.014)withmostproblemsat 3months after ICU

discharge followed by an improvement in QOL after 1 year (Additional File 1D). QOL never returned to

baselinelevel.

The SF-36 showed that in both groups, QOL decreased after 3 months compared to baseline

(Additional File 2C/2D). For the 4-years AKI-RRT patients, QOL improved after 1 year, especially in the

mentaldomains.At4years,QOLsignificantlydecreasedmainlyphysicallybut improvedor remainedthe

sameinthementalcomponents(AdditionalFile2C).Changesinlong-termQOLinthe4-yearsnonAKI-RRT

patientswerelesspronounced(AdditionalFile2D).

The 4-years AKI-RRT patients showed a decrease in VAS after 3months (63), and improvements

after1(70)and4years(68)butwithoutregainofthebaselinelevel(70)(P=0.044).The4-yearsnonAKI-

RRTpatientshadthesameevolutionbutwithoutsignificance(P=0.327).

Additionalfile3andadditionalfile4illustratemoreindetailthevariabilityinEQ-5DandSF-36over

time.

Overall, long-termQOL remainedunder thebaseline level forAKI-RRTandnonAKI-RRTpatients,

andundertheQOLoftheaveragepopulation.

Additionalquestionsafter1yearand4years

Oneand4yearsafterICUdischarge,mostsurvivorslivedindependently,andonlyaminoritystayed

inaspecialcarefacility(Table1).Therewerenomajorsleepingproblems.Oneyearand4yearsafterICU

discharge,AKI-RRTpatientshadmorebadmemoriesthannonAKI-RRTpatients (17.4%vs4.3%,P=0.010;

21.4% vs 3.8%, P=0.055). 81.8% of the 1-year AKI-RRT patients preferred to be readmitted to an ICU

98

department in case of deterioration versus 83.0% of their 1-year matches (P=0.867). This number

decreased to 71.4% for the 4-years AKI-RRT patients versus 84.6% for the 4-years non AKI-RRT patients

(P=0.244).

DISCUSSION

Inthisprospectivesinglecentermatchedcohortstudyconcerninglong-termoutcomesandQOLof

AKI-RRTpatients,wefoundhighmortalityratesandlowerQOLlevelscomparedtothegeneralpopulation.

Similar to others, we found high hospital mortality (55%) in this cohort of critically ill AKI-RRT

patients,withonlymoderate increaseofmortalityat longer followup (58%at3months,61%at1year,

71%at4years)[4,14,15,20,36].

At hospital discharge and at long-term, a quarter of AKI-RRT hospital survivors were RRT

dependent.Thesefindingsaresimilartothosereportedinliterature[37].

Long-termsurvivaldatawouldbemeaninglesswithoutconsideringQOL.Remarkably,therewasno

difference inQOLatdifferent timepointsbetweenAKI-RRTpatients andmatchednonAKI-RRTpatients,

although changes inQOLover timewere less pronounced in the latter group.QOLdecreased 3months

afterICUdischargecomparedtobaseline,improvedafter1year,andstayedthesameorimprovedslightly

after4years,butstillremainedunderbaselinelevel.

Thefactthatlong-termQOLhadthesameevolutionovertimeinAKI-RRTandnonAKI-RRTpatients

was quite surprising suggesting that the AKI-RRT component during critical illness did not have an

importantimpactonlong-termQOL.Othersreportedverysimilarfindings,however,thesestudiesreported

onlyonQOLafter6months,andin1studynotallAKIpatientsreceivedRRT,andsomepatientsreceived

RRTwithoutAKI[20,21].

The fact thatAKI-RRTpatientsweremoreseverely illduring their ICUstaycomparedtomatched

patientshadnoinfluenceonQOLovertheyears.This is inaccordancewiththefindingsofOrweliusetal

[38].Inamulticenterstudytheyfoundthat6monthsafterICUdischarge,perceivedQOLinsepsispatients

did not differ from ICU survivors with other diagnoses, even though these sepsis patients were more

severelyill,andhadalongerICUstay.AnotherstudybyOrweliussuggestedthatlong-termQOLwasmainly

affectedbyco-morbidity [39]. Inour studyAKI-RRTandnonAKI-RRTpatientshadaverycomparableco-

morbidityandmedicalhistory,whichmayexplain thecomparable long-termQOLbetweengroups inour

study.

QOL was perceived as acceptable and both AKI-RRT and non AKI-RRT patients reported low

dependenceindailylifelateron.ThenumberofAKI-RRTandnonAKI-RRTpatientswhoagreedtoundergo

life-sustaining interventions again in case of deterioration remained high. However, QOL was lower

99

comparedtothatoftheaveragepopulationinbothgroupsspecificallyinthemorephysicaldomains.Thisis

inaccordancewiththefindingsofothers[12-16,20,21].

Ourstudyhasseveralstrengths.First,thematchedcohortdesigndemonstratestherealimpactof

AKI-RRT upon long-term QOL. This has not been evaluated thus far. Second, QOL was assessed with

validatedquestionnairesatbaseline,whichallowsfortheonlyreliableevaluationofQOLovertimewithout

recall or selectionbias [11, 40]. Third, the additional questions andVAS score allowedevaluationof the

patients’ perception of the ICU admission and the consequences of severe illness. Finally, most studies

report QOL in AKI survivors as a short-term endpoint, while this study provides also data for a longer

follow-up period. Strict time intervals of 3months and 1 year after ICU dischargewere respected in all

patients.For long-termassessmentofQOL,anarbitrary timepointwaschosen (August2013)whichwas

between47-52monthsafterICUdischargeforallpatients.Responseratewasveryhighandonly3patients

werelost-to-follow-up.

Somelimitationsshouldalsobementioned.First,singlecenterdatafromauniversityhospitalmay

notreflectgeneralpracticeandmaylimitexternalvalidityofthedata.Second,although1-yearand4-years

AKI-RRT patients were matched to non AKI-RRT patients based on 4 criteria, we cannot exclude that

matchedpatientshadadifferentprofilecomparedtoAKI-RRTpatients.Third,thestudycohortisrelatively

smallandmaylackofstatisticalpowertodetectdifferencesamongtheQOLdomainsinourstudypatients.

Fourth, medical decisions leading to ICU referral may have selected for patients with better prospects.

Fifth, long-termQOLmay also bemodified by events happening to the patient after hospital discharge.

Thesewerenotrecordedinthepresentstudy.

CONCLUSIONS

We found high mortality rates in AKI-RRT patients. However, in long-term critically ill AKI-RRT

survivors,QOLwascomparabletomatchedlong-termcriticallyillsurvivorswithoutAKI-RRT,butlowerthan

in the general population. Themajority of AKI-RRT patients wanted to be readmitted to the ICU when

needed,despiteahigher severityof illness compared tomatchednonAKI-RRTpatients, anddespite the

factthatonequarterhadpersistentdialysisdependency.

100

KEYMESSAGES

Long-termcriticallyillAKI-RRTsurvivorshavecomparableQOLthanmatchedlong-termcriticallyillsurvivorswithoutRRT.

QOLinlong-termAKI-RRTsurvivorsislowerthaninthegeneralpopulation.

AKI-RRTpatientsaremoreseverelyillduringtheirICUstaycomparedtomatchednonAKI-RRTpatients.

Themajorityoflong-termAKI-RRTsurvivorsprefertobereadmittedtotheICUdepartmentincaseofdeterioration.

Onequarteroflong-termAKI-RRTsurvivorshavepersistentdialysisdependency.

ACKNOWLEDGEMENTS



interviewingpatients,andcallingpatientsorrelatives.TheauthorsalsothankChrisDanneelsforhishelpin


101

Table1.PatientcharacteristicsatICUadmission,organfailureduringICUadmission,andoutcomes

1-yearAKI-RRTpatients

(N=47)

1-yearnonAKI-RRTpatients(N=94)

P 4-yearAKI-RRTpatients(N=28)

4-yearnonAKI-RRTpatients(N=28)

P

age,yrs,(median,IQR) 57(45-69) 57(48-70) 0.897 54(45-66) 53(45-68) 0.718malegender,N(%) 31(66.0) 62(66.0) 0.999 16(57.1) 16(57.1) 0.999BMI,kg/m2(median,IQR)

26.2(22.8-29.7) 25.9(22.0-29.4) 0.444 27.3(22.9-31.6) 24.5(22.9-27.8) 0.092

serumcreatininebaseline(mg/dL)(median,IQR)*

1.14(0.94-1.51) 0.82(0.66-1.04) 0.001 0.97(0.80-1.26) 0.78(0.65-1.11) 0.062

eGFRbaseline(mL/minper1.73m²)(median,IQR)*

86(71-100) 100(83-116) 0.007 99(85-109) 102(87-116) 0.629

livesathomebeforeadmission,N(%)

45(95.7) 90(95.75) 0.999 26(92.9) 27(96.4)0.553

ADL,N(%) nolimitations 25(53.2) 47(50.0) 0.721 18(63.4) 21(75.0) 0.383

moderatelimitations 19(40.4) 42(44.7) 0.631 7(25.0) 7(25.0) 0.999chair-bound 0(0) 3(3.2) 0.216 0(0) 0(0) NAbedridden 3(6.4) 2(2.1) 0.198 3(10.7) 0(0) <0.001

hospitalizationinlast6monthsbeforeICU,N(%)

20(42.6) 46(48.9) 0.474 10(35.7) 14(50.0) 0.280


1(0-3) 2(0-3) 0.115 0(0-2) 2(0-3) 0.110

Typeofadmission,N(%)medical 32(68.1) 67(71.3) 0.696 18(64.3) 18(64.3) 0.999scheduledsurgery 1(2.1) 4(4.3) 0.519 0(0) 4(14.3) 0.038emergencysurgery 10(21.3) 18(19.1) 0.765 7(25.0) 3(10.7) 0.163trauma 3(6.4) 4(4.3) 0.376 2(7.1) 2(7.1) 0.999burns 1(1) 0.614 1(3.6) 1(3.6) 0.999SeverityofillnessatICUadmission(first24hours)APACHEIIscore(median,IQR)

26(21-31) 24(20-30) 0.251 23(20-28) 22(18-25) 0.362

SOFAscore(median,IQR) 9(5-11) 7(5-10) 0.047 7(4-12) 6(4-9) 0.139Mechanicalventilation,N(%)

29(61.7) 49(52.1) 0.281 21(75.0) 13(46.4) 0.029

Vasopressors,N(%) 21(44.7) 37(39.4) 0.545 11(39.3) 9(32.1) 0.577RRT,N(%) 11(23.4) 0(0) <0.001 6(21.4) 0(0) 0.010OrganfailureduringICUstayMechanicalventilation,N(%)

39(83.0) 50(53.2) <0.001 24(85.7) 13(46.4) 0.002

Lengthofmechanicalventilation,days(median,IQR)

16(3-27) 1(0-3) <0.001 18(4-31) 0(0-7) <0.001

Vasopressors,N(%) 36(76.6) 42(44.7) <0.001 21(75.0) 10(35.7) 0.003Lengthofvasopressortherapy,days(median,IQR)

5(1-8) 0(0-3) <0.001 3(0-10) 0(0-3) 0.002

RRT,N(%) 47(100) 0(0) <0.001 28(100.0) 0(0) <0.001MeanSOFAscore(median,IQR)

7(6-9) 5(4-7) <0.001 7(5-10) 4(4-7) <0.001

OutcomesICULOS,days(median,IQR)

22(11-42) 5(3-9) <0.001 24(13-49) 7(3-10) <0.001

Readmissions,N(%) 8(17.0) 12(12.8) 0.495 3(10.7) 4(14.3) 0.686HospitalLOS,days(median,IQR)

70(30-100) 21(13-44) <0.001 62(20-130) 19(10-46) 0.003

DNRdecisions,N(%) 4(8.5) 3(3.2) 0.170 2(7.1) 1(3.6) 0.312

102

Long-termmortality,N(%)

12(25.5) 20(21.3) 0.570 NA NA -

NeedforRRTathospitaldischarge,N(%)

12(25.5) NA - 10(35.7) NA -

NeedforRRTat3months,N(%)

9(19.1) NA - 8(28.6) NA -

NeedforRRTat1year,N(%)

9(19.1) NA - 8(28.6) NA -

NeedforRRTat4years,N(%)

NA NA - 8(28.6) NA -

Livingsituationafter1year,N(%) 46answers 93answers 27answers 26answers

independentwithoutadditionalhelp

25(54.3) 47(50.5) 0.672 16(59.3) 14(53.8) 0.691

independentwithsomehelp

12(26.1) 22(23.7) 0.754 6(22.2) 6(23.1) 0.941

togetherwithrelatives(othersthanspouse)

6(13.0) 14(15.1) 0.751 3(11.1) 4(15.4) 0.646

specialcarefacility 3(6.5) 5(5.4) 0.786 2(7.4) 1(3.8) 0.575other 0(0) 5(5.4) 0.109 0(0) 1(3.8) 0.304

Livingsituationafter4years,N(%) NA NA 27answers 26answers

independentwithoutadditionalhelp

NA NA - 18(66.7) 14(53.8) 0.340

independentwithsomehelp

NA NA - 5(18.5) 6(23.1) 0.682

togetherwithrelatives(othersthanspouse)

NA NA - 2(7.4) 5(19.2) 0.204

specialcarefacility NA NA - 2(7.4) 1(3.8) 0.575other NA NA - 0(0) 0(0) 0.999

AKI=acutekidney injury;RRT=renalreplacementtherapy;yrs=years; IQR=interquartilerange(25%-75%);N=number;BMI=bodymassindex;eGFR=estimatedglomerularfiltrationrate;ICU=intensivecareunit;ADL=activityofdailyliving;NA=notapplicable;ICU=intensive care unit; APACHE=Acute Physiology and ChronicHealth Evaluation; SOFA= SequentialOrgan FailureAssessment; LOS=lengthofstay;DNR=do-not-resuscitate;NA=notapplicable*Serumcreatinineatbaselinewasdefinedasserumcreatinine6monthsbeforeICUadmission.Valuesweremissingin27ofthe1-yearAKI-RRTpatients,in14ofthe941-yearnonAKI-RRTpatients,in21ofthe4-yearsAKI-RRTpatients,andin4the4-yearsnonAKI-RRTpatients

103

Table2.PersonswhoratedQOL,assessmentofQOL,numberofcompletedQOLsurveys

(a) All QOL surveys completed by face-to-face interviews; (b) All QOL surveys completed by regular mail; (c) 46 QOL surveyscompleted; 32 by regularmail (69.6%) and 14 by phone interview (30.4%); (d) 94 QOL surveys completed; 67 by regularmail(71.3%)and27byphoneinterview(28.7%);(e)27QOLsurveyscompleted;18byregularmail(66.7%)and9byphoneinterview(33.3%);(f)26QOLsurveyscompleted;19byregularmail(73.1%)and7byphoneinterview(26.9%);(g)28QOLsurveyscompleted;14byregularmail(50.0%)and14byphoneinterview(50.0%);(h)28QOLsurveyscompleted;20byregularmail(71.4%)and8byphoneinterview(28.6%);QOL=qualityoflife;N=number;AKI=acutekidneyinjury;RRT=renalreplacementtherapy

104

Figure1.Patientcohort

N=number;AKI=acutekidneyinjury;RRT=renalreplacementtherapy;ICU=intensivecareunit

105

Figure2.EQ-5Dassessmentsinthe1-yearcohort:Percentagesofpatientswithsomeorsevereproblemsperdimensionatthe3differenttimepoints

TheX-axis represents thedifferentdimensionsof theEQ-5D.TheY-axis represents thepercentages (%)ofpatientswithsomeorsevereproblemsinarespectivedimension.OnlysignificantP-values(Chi-Squaretest)areshownabovetherespectivedimensions.QOL=qualityoflife;AKI=acutekidneyinjury;RRT=renalreplacementtherapy

106

Figure3.SF-36assessmentsinthe1-yearcohort:Norm-basedmedianscoresperdomainatthe3different

timepoints

TheX-axis represents thedifferentdomainsof theSF-36.TheY-axis represents thenorm-basedmedianscores inarespectivedomainof theSF-36.Anorm-basedmedianscorebetween47-53 inagroupofpatients isconsideredasnormaloraverage.Norm-basedmedianscoresbelow47indicateimpairedfunctioningorbelowaverage;norm-basedmedian scores above 53 indicate better functioning or above average. Only significant P-values (Mann-Whitney Uanalysis)areshownabovetherespectivedomains.QOL=qualityoflife;AKI=acutekidneyinjury;RRT=renalreplacementtherapyPCS=physicalcomponentscore;MCS=mental component score; PF= physical functioning; RP= role physical; BP = bodily pain; GH= general health; VT=vitality;SF=socialfunctioning;RE=roleemotional;MH=mentalhealth

107

Figure4.EQ-5Dassessmentsinthe4-yearscohort:Percentagesofpatientswithsomeorsevereproblemsperdimensionatthe4differenttimepoints

TheX-axis represents thedifferentdimensionsof theEQ-5D.TheY-axis represents thepercentages (%)ofpatientswithsomeorsevereproblemsinarespectivedimension.OnlysignificantP-values(ChiSquaretest)areshownabovetherespectivedimensions.QOL=qualityoflife;AKI=acutekidneyinjury;RRT=renalreplacementtherapy

108

Figure5.SF-36assessmentsinthe4-yearscohort:Norm-basedmedianscoresperdomainatthe4differenttimepoints

TheX-axis represents thedifferentdomainsof theSF-36.TheY-axis represents thenorm-basedmedianscores inarespectivedomainof theSF-36.Anorm-basedmedianscorebetween47-53 inagroupofpatients isconsideredasnormaloraverage.Norm-basedmedianscoresbelow47indicateimpairedfunctioningorbelowaverage;norm-basedmedian scores above 53 indicate better functioning or above average. Only significant P-values (Mann-Whitney Uanalysis)areshownabovetherespectivedomains.QOL=qualityoflife;AKI=acutekidneyinjury;RRT=renalreplacementtherapyPCS=physicalcomponentscore;MCS=mental component score; PF= physical functioning; RP= role physical; BP = bodily pain; GH= general health; VT=vitality;SF=socialfunctioning;RE=roleemotional;MH=mentalhealth

109

AdditionalFile1.EQ-5Dassessmentsovertime

EvolutionsinEQ-5Dassessmentsaredescribedthroughfiguresinthe1-yearcohort(47AKI-RRT(1A)and94nonAKI-RRTpatients(1B))andinthe4-yearscohort(28AKI-RRT(1C)patientsand28nonAKI-RRTpatients(1D)).PercentagesofpatientswithsomeorsevereproblemsinthedifferentdimensionsoftheEQ-5Daregivenoverthedifferenttimepoints:baseline,3monthsand1year(1-yearcohort)andbaseline,3months,1yearand4years(4-yearscohort).TheX-axisrepresentsthedifferentdimensionsoftheEQ-5D.TheY-axisrepresentsthepercentages(%)ofpatientswithsomeorsevereproblemsinarespectivedimension.OnlysignificantP-values(ChiSquaretest)areshownabovetherespectivedimensions.QOL=qualityoflife;AKI=acutekidneyinjury;RRT=renalreplacementtherapy

110

AdditionalFile2.SF-36assessmentsovertime

EvolutionsinSF-36assessmentsaredescribedthroughfiguresinthe1-yearcohort(47AKI-RRT(2A)and94nonAKI-RRTpatients(2B))andinthe4-yearscohort(28AKI-RRT(2C)patientsand28nonAKI-RRTpatients(2D)).Norm-basedmedianscoresinthedifferentdomainsoftheSF-36aregivenoverthedifferenttimepoints:baseline,3monthsand1year(1-yearcohort)andbaseline,3months,1yearand4years(4-yearscohort).TheX-axis represents thedifferentdomainsof theSF-36.TheY-axis represents thenorm-basedmedianscores inarespectivedomainof theSF-36.Anorm-basedmedianscorebetween47-53 inagroupofpatients isconsideredasnormaloraverage.Norm-basedmedianscoresbelow47indicateimpairedfunctioningorbelowaverage;norm-basedmedian scoresabove53 indicatebetter functioningoraboveaverage.Only significantP-values (Friedman test) areshownabovetherespectivedomains.QOL=qualityoflife;AKI=acutekidneyinjury;RRT=renalreplacementtherapyPCS=physicalcomponentscore;MCS=mental component score; PF= physical functioning; RP= role physical; BP = bodily pain; GH= general health; VT=vitality;SF=socialfunctioning;RE=roleemotional;MH=mentalhealth

111

AdditionalFile3.VariabilityinEQ-5D471-yearAKI-RRTpatients Baseline 3months 1year P %(95%CI) Mobility 39.1(26.4-53.5) 63.6(46.6-77.8) 60.9(46.5-73.6) 0.045 Self-care 23.9(13.9-37.9) 42.4(27.2-59.2) 37.0(24.5-51.4) 0.190 Ususalactivities 37.0(24.5-51.4) 81.8(65.6-91.4) 60.9(46.5-73.6) <0.001 Pain/discomfort 45.7(32.2-59.8) 75.8(59.0-87.2) 54.3(40.2-67.8) 0.013 Anxiety/depression 30.4(19.1-44.8) 60.6(43.7-75.3) 30.4(19.1-44.8) 0.009 941-yearnonAKI-RRTpatients Baseline 3months 1year P %(95%CI) Mobility 37.2(28.1-47.3) 54.9(43.4-66.0) 55.4(45.3-65.2) 0.021 Self-care 24.5(16.9-34.0) 40.8(30.2-52.5) 38.0(28.8-48.3) 0.050 Ususalactivities 46.8(37.0-56.8) 81.7(71.2-89.0) 66.3(56.2-75.1) <0.001 Pain/discomfort 51.1(41.1-60.9) 70.4(59.0-79.8) 63.0(52.8-72.2) 0.035 Anxiety/depression 40.4(31.1-50.5) 39.4(28.9-51.1) 41.3(31.8-51.5) 0.971 284-yearsAKI-RRTpatients Baseline 3months 1year 4years P%(95%CI) Mobility 25.9(13.2-44.7) 61.9(40.9-79.2) 59.3(40.7-75.5) 50.0(32.6-67.4) 0.040Self-care 14.8(5.9-32.5) 47.6(28.3-67.6) 33.3(18.6-52.2) 25.9(13.2-44.7) 0.090Ususalactivities 25.9(13.2-44.7) 81.0(60.0-92.3) 55.6(37.3-72.4) 70.4(51.5-84.1) <0.001Pain/discomfort 48.1(30.7-66.0) 71.4(50.0-86.2) 59.3(40.7-75.5) 55.6(37.3-72.4) 0.439Anxiety/depression 29.6(15.9-48.5) 61.9(40.9-79.2) 25.9(13.2-44.7) 29.6(15.9-48.5) 0.040 284-yearsnonAKI-RRTpatients Baseline 3months 1year 4years P%(95%CI) Mobility 18.5(8.2-36.7) 39.1(22.2-59.2) 41.7(24.5-61.2) 60.7(42.4-76.4) 0.017Self-care 11.1(3.9-28.1) 21.7(9.7-41.9) 25.0(12.0-44.9) 28.6(15.3-47.1) 0.436Ususalactivities 29.6(15.9-48.5) 47.8(29.2-67.0) 70.8(50.8-85.1) 64.3(45.8-79.3) 0.014Pain/discomfort 37.0(21.5-55.8) 26.1(12.5-46.5) 45.8(27.9-64.9) 53.6(35.8-70.5) 0.227Anxiety/depression 51.9(34.0-69.3) 17.4(7.0-37.1) 25.0(12.0-44.9) 32.1(17.9-50.7) 0.054Percentagesand95%confidence intervalsofpatientswithsomeorsevereproblemsontherespectivedimensionsof theEQ-5Dovertimearegiven.AKI=acutekidneyinjury;RRT=renalreplacementtherapy;CI=confidenceinterval

(*)TheconfidenceintervalwascalculatedaccordingtoDGAltman,DMachin,TNBryant,MGardner(2000).Statisticswithconfidence:Confidenceintervalsandstatisticalguidelines.BMJBooks

112

AdditionalFile4.VariabilityinSF-36471-yearAKI-RRTpatients Baseline 3months 1year P Median(IQR) PCS 41.7(28.5-54.2) 30.7(25.1-40.4) 38.3(27.7-47.4) 0.003 MCS 53.8(38.9-61.6) 39.5(29.3-47.2) 53.3(39.2-58.6) 0.014 Physicalfunctioning 44.4(29.1-53.4) 27.6(19.2-39.1) 40.2(26.5-46.5) <0.001 Rolephysical 34.8(22.6-56.9) 27.5(17.7-29.9) 34.8(25.0-45.8) <0.001 Bodilypain 62.1(37.2-62.1) 39.7(29.2-50.9) 46.5(37.2-62.1) 0.015 Generalhealth 40.1(30.5-48.2) 36.3(31.1-41.0) 41.0(30.5-50.6) 0.078 Vitality 55.2(42.7-61.5) 45.8(39.6-50.5) 50.5(41.9-59.1) 0.041 Socialfunctioning 51.4(35.0-56.8) 35.0(24.1-40.5) 45.9(29.6-56.8) 0.005 Roleemotional 55.9(40.3-55.9) 28.7(20.9-38.4) 48.1(32.6-55.9) <0.001 Mentalhealth 50.0(33.1-61.3) 41.6(30.3-50.0) 50.0(40.2-58.4) 0.022 941-yearnonAKI-RRTpatients Baseline 3months 1year P Median(IQR) PCS 39.4(29.1-49.6) 31.3(26.3-43.2) 36.6(26.0-46.4) 0.007 MCS 48.0(37.5-55.7) 47.3(31.6-54.9) 47.8(34.8-54.0) 0.759 Physicalfunctioning 40.2(23.4-53.4) 31.8(21.3-44.4) 33.9(22.3-48.6) 0.001 Rolephysical 34.8(22.6-56.9) 27.5(17.7-37.3) 32.4(23.2-42.2) 0.059 Bodilypain 46.5(33.3-62.1) 39.5(29.2-50.5) 41.6(29.2-55.4) 0.008 Generalhealth 37.7(30.5-50.6) 40.1(31.1-45.8) 37.7(30.5-45.8) 0.871 Vitality 49.0(36.5-58.3) 49.0(39.6-55.2) 49.0(36.5-58.3) 0.896 Socialfunctioning 48.7(35.0-56.8) 35.0(24.1-45.9) 35.0(24.1-51.4) <0.001 Roleemotional 55.9(31.6-55.9) 38.4(20.9-55.9) 44.2(24.8-55.9) 0.410 Mentalhealth 47.2(33.1-58.4) 50.0(34.5-55.7) 47.2(34.5-55.6) 0.562 284-yearsAKI-RRTpatients Baseline 3months 1year 4years PMedian(IQR) PCS 46.1(38.7-53.7) 33.2(26.0-40.4) 39.8(31.6-46.7) 38.1(31.6-47.1) 0.007MCS 57.6(42.8-62.3) 39.5(29.3-47.1) 53.5(40.9-61.6) 53.9(42.4-60.3) 0.010Physicalfunctioning 48.6(36.5-57.0) 27.6(18.1-43.4) 42.3(29.7-48.6) 33.9(29.7-40.2) <0.001Rolephysical 42.2(27.5-56.9) 27.5(17.7-31.8) 34.8(27.5-47.1) 45.9(27.5-56.9) <0.001Bodilypain 51.1(38.2-62.1) 41.8(30.1-50.9) 51.1(41.8-62.1) 50.7(34.4-62.1) 0.178Generalhealth 42.9(30.3-47.9) 36.3(32.9-42.9) 43.4(36.3-50.6) 38.2(32.9-48.0) 0.093Vitality 55.2(43.5-64.6) 45.8(42.7-50.5) 52.1(45.8-61.5) 49.0(45.8-58.3) 0.037Socialfunctioning 56.8(40.5-56.8) 35.0(26.9-40.5) 51.4(35.0-56.8) 45.9(35.0-56.8) 0.101Roleemotional 55.9(50.0-55.9) 24.8(9.2-38.4) 48.1(32.6-55.9) 55.9(20.9-55.9) 0.001Mentalhealth 55.6(33.1-64.1) 41.6(33.1-51.4) 50.0(41.6-61.3) 52.8(41.6-58.5) 0.188 284-yearsnonAKI-RRTpatients Baseline 3months 1year 4years PMedian(IQR) PCS 48.4(36.3-57.0) 37.1(26.1-45.5) 40.8(27.9-46.5) 41.0(32.1-52.6) 0.358MCS 48.6(34.3-57.6) 48.9(37.2-54.8) 49.7(40.6-54.7) 47.0(37.4-55.5) 0.913Physicalfunctioning 52.8(40.2-54.9) 39.1(19.2-44.4) 38.1(22.3-48.6) 38.1(25.5-48.6) <0.001Rolephysical 52.0(17.7-56.9) 27.5(25.0-39.7) 32.4(25.0-39.7) 39.7(25.0-47.1) 0.158Bodilypain 50.3(41.2-62.1) 46.1(37.2-55.4) 46.1(36.1-62.1) 46.1(37.2-62.1) 0.489Generalhealth 41.0(35.3-55.3) 40.1(29.8-49.4) 41.0(35.3-48.8) 41.0(34.7-53.5) 0.577Vitality 52.1(42.7-58.3) 49.0(39.6-58.3) 52.1(39.6-58.3) 49.0(42.7-55.2) 0.403Socialfunctioning 56.8(35.0-56.8) 40.5(24.1-51.4) 35.0(22.8-52.8) 45.9(24.1-56.8) 0.058Roleemotional 40.3(20.9-55.9) 40.3(28.7-55.9) 40.3(24.8-55.9) 44.2(24.8-55.9) 0.071Mentalhealth 52.8(35.9-58.4) 50.0(37.3-58.5) 50.0(37.3-58.5) 50.0(41.6-52.8) 0.962Mediannorm-basedscoreswithinterquartilerangesonthedifferentdomainsoftheSF-36overtimearegiven.AKI=acutekidney injury;RRT= renal replacement therapy; IQR= interquartile range (25%-75%);PCS=physical component score;MCS=mentalcomponentscore

113

REFERENCES1. UchinoS,KellumJA,BellomoRetal.Acuterenalfailureincriticallyillpatients.JAMA.2005;294:813-82. HosteEA,SchurgersM.Epidemiologyofacutekidney injury:howbig is theproblem?CritCareMed.2008;36Suppl4:146-513. PalevskyPM,ZhangJH,O'ConnorTZ,ChertowGM,CrowleyST,ChoudhuryD,FinkelK,KellumJA,PaganiniE,ScheinRMHetal.Intensityofrenalsupportincriticallyillpatientswithacutekidneyinjury.NewEnglJMed.2008;359:7-204. Bellomo R, Cass A, Norton R, GallagherM, Lo S, Su S, Cole L, Finfer S,McArthur C,McGuinness S et al. Intensity ofContinuousRenal-ReplacementTherapyinCriticallyIllPatients.NewEnglJMed.2009;361:1627-385. Amdur RL, Chawla LS, Amodeo S, Kimmel PL, Palant CE. Outcomes following diagnosis of acute renal failure in U.S.veterans:focusonacutetubularnecrosis.KidneyInt.2009;76:1089-976. IshaniA,XueJL,HimmelfarbJ,EggersPW,KimmelPL,MolitorisBA,CollinsAJ.AcutekidneyinjuryincreasesriskofESRDamongelderly.JAmSocNephrol.2009;20:223-87. GammelagerH,ChristiansenCF,JohansenMB,TonnesenE,JespersenB,SorensenHT.One-yearmortalityamongDanishintensivecarepatientswithacutekidneyinjury:acohortstudy.CritCare.2012;16:R1248. ChawlaLS,EggersPW,StarRA,KimmelPL.Acutekidneyinjuryandchronickidneydiseaseasinterconnectedsyndromes.NEnglJMed.2014;371:58-669. Mehlhron J, Freytag A, Schmidt K, Brunkhorst FM, Graf J, Troitzsch U et al. Rehabilitation interventions forpostintesivecareunitsyndrome:asystematicreview.CritCareMed.2014;42:1263-7110. OeyenS,VandijckD,BenoitD,Decruyenaere J;Annemans L,HosteE. Long-termoutcomeafter acute kidney injury incritically-illpatients.ActaClinBelg.2007;62Suppl2:337-4011. OeyenSG,VandijckDM,BenoitDD,AnnemansL,DecruynaereJM.Qualityoflifeafterintensivecare:asystematicreviewoftheliterature.CritCareMed.2010;38:2386-40012. AhlstromA, TallgrenM, Peltonen S, Rasanen P, Pettila V. Survival and quality of life of patients requiring acute renalreplacementtherapy.IntensiveCareMed.2005;31:1222-813. Delannoy B, Floccard B, Thiolliere F, KaakiM, BadetM, Rosselli S, Ber CE, Saez A, Flandreau G, Guerin C. Six-monthoutcome in acute kidney injury requiring renal replacement therapy in the ICU: amulticentre prospective study. IntensiveCareMed.2009;35:1907-1514. MorgeraS,KraftAK,SiebertG,LuftFC,NeumayerHH.Long-termoutcomes inacuterenal failurepatientstreatedwithcontinuousrenalreplacementtherapies.AmJKidneyDis.2002;40:275-915. Korkeila M, Ruokonen E, Takala J. Costs of care, long-term prognosis and quality of life in patients requiring renalreplacementtherapyduringintensivecare.IntensiveCareMed.2000;26:1824-3116. Gopal I, Bhonagiri S, Ronco C, Bellomo R. Out of hospital outcome and quality of life in survivors of combined acutemultipleorganandrenalfailuretreatedwithcontinuousvenovenoushemofiltration/hemodiafiltration.IntensiveCareMed.1997;23:766-7217. AbelhaFJ,BotelhoM,FernandesV,BarrosH.Outcomeandqualityoflifeofpatientswithacutekidneyinjuryaftermajorsurgery.Nefrologia2009;29:404-1418. MaynardSE,Whittle J,ChelluriL,ArnoldR.Qualityof lifeanddialysisdecisions incritically illpatientswithacuterenalfailure.IntensiveCareMed.2003;29:1589-9319. Morsch C, Thome FS, Balbinotto A, Guimaraes JF, Barros EG. Health-related quality of life and dialysis dependence incriticallyillpatientsurvivorsofacutekidneyinjury.RenFail.2011;33:949-56

114

20. VaaraST,PettilaV,ReinikainenM,KaukonenKM,ConsortiumFIC.Population-based incidence,mortalityandqualityoflifeincriticallyillpatientstreatedwithrenalreplacementtherapy:anationwideretrospectivecohortstudyinfinnishintensivecareunits.CritCare.2012;16:R1321. Hofhuis JG, van Stel HF, Schrijvers AJ, Rommes JH, Spronk PE. The effect of acute kidney injury on long-term health-relatedqualityoflife:aprospectivefollow-upstudy.CritCare.2013;17:R1722. NobleJS,SimpsonK,AllisonME.Long-termqualityoflifeandhospitalmortalityinpatientstreatedwithintermittentorcontinuoushemodialysisforacuterenalandrespiratoryfailure.RenFail.2006,28:323-3023. HamelMB,PhillipsRS,DavisRB,DesbiensN,ConnorsAF,Jr.,TenoJM,WengerN,LynnJ,WuAW,FulkersonW.Outcomesandcost-effectivenessofinitiatingdialysisandcontinuingaggressivecareinseriouslyillhospitalizedadults.SUPPORTInvestigators.StudytoUnderstandPrognosesandPreferencesforOutcomesandRisksofTreatments.AnnInternMed.1997;127:195-20224. WouterDeCorte,SandraOeyen,LievenAnnemans,DominiqueBenoit,AnnemiekeDhondt,RaymondVanholder,JohanDecruyenaere, Eric Hoste. Long-term outcome and quality of life in ICU patients with acute kidney injury treated with renalreplacementtherapy:acasecontrolstudy.IntensiveCareMed.2014;40Suppl1:1325. OeyenS,BenoitD,AnnemansL,DecruyenaereJ.Qualityoflifebefore,3months,and1yearafterICUdischarge.CritCareMed.2010;38SupplDecember:P58726. Vandenbroucke JP, von Elm E, Altman DG, Gøtzsche PC,Mulrow CD, Pocock SJ et al. Strengthening the reporting ofobservationalstudiesinepidemiology.Epidemiology.2007;18:805-3527. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinalstudies:developmentandvalidation.JChronicDis.(1987);40:373-83.28. KnausWA,DraperEA,WagnerDP,ZimmermanJE:APACHEII.Aseverityofdiseaseclassificationsystem.CritCareMed.1985;13:818-2929. VincentJL,MorenoR,TakalaJ,WillattsS,DeMendoncaA,BruiningH,ReinhartCK,SuterPM,ThijsLG.TheSOFA(Sepsis-relatedOrganFailureAssessment)scoretodescribeorgandysfunction/failure.OnbehalfoftheWorkingGrouponSepsis-RelatedProblemsoftheEuropeanSocietyofIntensiveCareMedicine.IntensiveCareMed.1996;22:707-1030. StevensLA,SchmidCH,GreeneT,ZhangYL,BeckGJ,FroissartMetal.ComparativeperformanceoftheCKDEpidemiologyCollaboration (CKI-EPI)and themodificationofdiet in renaldisease (MDRD)studyequations forestimatingGFR levelsabove60mL/min/1.73m2.AmJKidneyDis.2010;56:486-9531. JohnEWare,MarkKosinski,JakobBBjorner,DianeMTurner-Bowker,BarbaraGandek,MarkEMaruish.User’sManualfortheSF-36v2®HealthSurvey.QualityMetricIncorporated,Lincoln,RhodeIsland;2007.32. ChrispinPS,ScottonH,RogersJ,LloydD,RidleySA.ShortForm36intheintensivecareunit:assessmentofacceptability,reliabilityandvalidityofthequestionnaire.Anaesthesia.1997;52:15-2333. EuroQolGroup.EuroQol--anewfacilityforthemeasurementofhealth-relatedqualityoflife.HealthPolicy.1990;16:199-20834. BrazierJ,JonesN,KindP.TestingthevalidityoftheEuroqolandcomparingItwiththeSf-36HealthSurveyQuestionnaire.QualLifeRes.1993;2:169-8035. AngusDC,CarletJ,BrusselsRoundtableP.Survivingintensivecare:areportfromthe2002BrusselsRoundtable.IntensiveCareMed.2003;29:368-7736. PalevskyPM.Renalsupportinacutekidneyinjury--howmuchisenough?NewEnglJMed.2009;361:1699-70137. Wald R,McArthur E, Adhikari NK, Bagshaw SM, Burns KE, Garg AX et al. Changing incidence and outcomes followingdialysis-requiringacutekidneyinjuryamongcriticallyilladults:Apopulation-basedcohortstudy.AmJKidneyDis.2015;65:870-738. OrweliusL,LoboC,TeixeiraPintoA,CarneiroA,Costa-PereiraA,GranjaC.Sepsispatientsdonotdifferinhealth-relatedqualityoflifecomparedwithotherICUpatients.ActaAnaesthesiolScand.2013;57:1201-5

115

39. OrweliusL,NordlundA,NordlundP,SimonssonE,BackmanC,SamuelssonA,SjobergF.Pre-existingdisease: themostimportant factor forhealthrelatedqualityof life long-termaftercritical illness:aprospective, longitudinal,multicentretrial.CritCare.2010;14:R6740. JoyceVR,SmithMW,JohansenKL,UnruhML,SirokaAM,O'ConnorTZ,PalevskyPM,VeteranAffairs/NationalInstitutesofHealthAcuteRenalFailureTrialN.Health-relatedqualityof lifeasapredictorofmortalityamongsurvivorsofAKI.ClinJAmSocNephrol.2012;7:1063-70

117

IV.CriticallyIllOctogenariansand

Nonagenarians:EvaluationofLong-Term

Outcomes,Post-HospitalTrajectoriesand

QualityofLifeOneYearandSevenYears

afterICUDischarge

SandraOeyen*,MD1,2,JorisVermassen,MD1,2,RuthPiers,MD,PhD2,3,DominiqueBenoit,MD,PhD1,2,

LievenAnnemans,PhD4,JohanDecruyenaere,MD,PhD1,2

1DepartmentofIntensiveCareMedicine,GhentUniversityHospital,Ghent,Belgium2FacultyofMedicineandHealthSciences,GhentUniversity,Ghent,Belgium3DepartmentofGeriatrics,GhentUniversityHospital,Ghent,Belgium4I-CHERFacultyofMedicineandHealthSciences,GhentUniversity,Ghent,Belgium

PublishedinMinervaAnestesiologica2017,83:598-609

118

ABSTRACT

Background: To investigate long-term outcomes, post-hospital trajectories, and quality of life (QOL) in

patients≥80yearsadmittedtotheintensivecareunit(ICU)ofatertiarycarehospital.

Methods:A1-yearprospectiveobservationalcohortanalysiswasperformed.Allconsecutivepatients≥80

years admitted to the ICU were screened for inclusion. Demographics, comorbidity, organ failures, and

outcomeswere analyzed.QOLbefore admission, 3months, 1 year, and 7 years after ICUdischargewas

assessedusingEuroQoL-5D(EQ-5D)andMedicalOutcomesStudy36-itemShortFormHealthSurvey(SF-36)

questionnaires.StatisticalsignificancewasattainedatP<0.05.

Results:131patientswithamedianageof83years(IQR81-85),aCharlsoncomorbidityindexof2(IQR0-

4),aSOFAscoreof4(3-8)uponICUadmissionandanAPACHEIIscoreof20(IQR15-24)wereincluded.ICU,

hospital,3months,1-year,and7-yearsmortality rateswere17%,29%,39%,50%,and84%respectively.

QOL decreased significantly over time.Most elderly consideredQOL as acceptable and perceived only a

worseninginphysicalfunctioningandself-careatlong-term.Ofthe1-yearand7-yearssurvivors,21%and

39%(P=0.122) lived innursinghomes,and81%and72%(P=0.423)preferredtobereadmittedtoan ICU

departmentifnecessarily.

Conclusions:Mostcriticallyilllong-termelderlysurvivorslivedathome,perceivedtheirQOLasacceptable,

andwantedtobereadmittedtotheICUifnecessarily.Inolderpatients,agealoneisapoorindicatorofthe

possiblevaluetobegainedfromanICUadmission.

119

INTRODUCTION

Survivaltoolderagehas increased,which leadstomorehospitalizationsandmore intensivecare

unit (ICU) admissions for older patients.1-3 Concerns may rise regarding utility or futility of high-level

expensive ICU treatments for these patients. Prognosis of critically ill patients aged 80 ormoremay be

poor, especially in those admitted from a chronic care facility, orwith severe comorbidity, or a greater

illnessseverity.1-6

ToidentifywhowouldbenefitfromICUtreatment,long-termqualityoflife(QOL)shouldbetaken

intoaccountaswell.7Majorreductionsinlong-termQOLincriticallyillpatientswereseeninsevereacute

respiratory distress syndrome, prolonged mechanical ventilation, and severe sepsis, representing

complicationsthataffectelderlypatientsasmuchasyoungerpatients.8

Recentdataregardinglong-termQOLincriticallyillelderlypatientsareincreasingbutstilllimited.4,

5-7, 9-17 They show that elderly have a comparable or slightly decreased QOL compared to the general

population but perceive QOL as good.4, 5, 6, 11, 12, 15, 16 However, these studies were either based on a

retrospectivecohort,4,12,15evaluatedQOLafterashortterm,5,11,15lackedbaselineQOLassessment,4,5,9,10,

12-14,16 assessedQOLafter variable follow-up intervals,4,12,13 includedonly elderlywith an ICU stayof 24

hoursormore,9-11,17ordefinedelderlyaspatientsaged65yearsormore5,6,16,17orevenyounger.10Most

studies identified independent predicting factors for outcome 5,13 but lacked any information about the

post-hospitalcoursesofsurvivors.

Theaimofthepresentstudywastoevaluatelong-termoutcomesofelderlypatientsaged80years

ormoreadmitted to the ICU, toassesspost-hospital trajectories,and tocomparebaselineQOLof these

patientswithQOL3months,1yearand7yearsafterICUdischarge.

MATERIALSANDMETHODS

Design,setting,andpatients

The study was a prospective observational cohort analysis performed at the 14-bed medical

(MICU),22-bedsurgicalICU(SICU),and6-bedburnunitoftheGhentUniversityHospitalinBelgium.From

March 3rd2008 -March 3rd 2009, all consecutive patients ≥ 80 yearswere screened for inclusion. Study

patientsconsistedofapredefinedsubgroupofa largerobservationalcohortstudyconcerningQOL inan

ICUpopulation.18IncaseofreadmissionormultipleICUadmissions,onlythefirstwasconsidered.Elderly

patientsadmittedatthecardiacsurgicalunitaftercardiacsurgerywerenotincluded.

TheGhentUniversityHospital ICUsareclosedICUswherepatientsaretreatedbyfull-timecritical

care physicians. Decisions concerning admission, withdrawing or withholding advanced life support are

made by the critical care physician together with the referring physician, consulting the wishes and

expectationsofthepatientandrepresentatives.

120

The study was approved by the Ethics Committee of the Ghent University Hospital (project

2007/423; amendment 0095/2015) and conducted in accordancewith theHelsinki declaration. A signed

informedconsentwasobtainedfromeveryincludedpatientorhislegalrepresentative.


Data collected within the first 24 hours of ICU admission included demographics, contact

information of the patient, proxy, and general practitioner, hospital days prior to ICU admission, living

circumstancesbeforeICUadmission,functionalityaccordingtoactivityofdailyliving(ADL),19hospitalization

in the last6months,comorbidityasmeasuredby theCharlsoncomorbidity index,20main reason for ICU

admission,AcutePhysiology andChronicHealth Evaluation (APACHE II) score,21 SequentialOrgan Failure

Assessment(SOFA)score,22needforinvasivemechanicalventilation,useofanyvasopressors,orneedfor

renal replacementtherapy (RRT). During ICUstay,SOFAscoresneedfor invasivemechanicalventilation,

vasopressors,RRT, tracheotomy,anddo-not-resuscitate (DNR) codeswere collectedonadailybase. ICU

lengthofstay(LOS),hospitalLOS,vitalstatusatICUandhospitaldischarge,andat3months,1yearand7

yearsfollowingICUdischargewerecollectedforeachpatient.

Qualityoflife

QOLwas assessed bymeans of theMedical Outcomes Study 36-item Short FormHealth Survey

version 2 (SF-36v2®)23 and the EuroQoL-5D (EQ-5D).24 Both questionnaires were validated and found

suitableformeasuringQOLinthecriticallyillpopulation.25,26Anextensiveexplanationofthesesurveyscan

befoundinpreviouspublicationsofourgroup.27,28

Qualityoflife:evolutionovertime

QOLwasassessedat4differenttimepoints:baselineQOLandstrictlyat3monthsand1yearafter

ICUdischarge.QOLwasalsoassessedbetween18-24February2015,amedianof6.6years(interquartile

range(IQR)6.0years–6.8years)-roundedto7years-afterICUdischarge.FollowingICUadmissionand

study inclusion, a face-to-face interview to assess baseline QOL (defined as QOL 2 weeks before ICU

admission) was done as soon as possible. This interview was preferably taken from the patient, or if

deemedimpossible,fromtheproxy.Threemonths,1year,and7yearsafterICUdischarge,patientswere

senttheEQ-5DandSF-36surveysbyregularmail;at1and7years,questionsconcerning livingsituation,

memories,sleepquality,andwillingnesstobereadmittedtoanICUdepartment,wereadded.After7years,

patients were also questioned about their social network, medical follow-up, financial situation, and

happiness.Ifthequestionnaireswerenotreturnedwithinonemonth,patientsorrelativeswerecontacted

byphonetoassessQOLafter1yearandafter7years.Eventually,thegeneralpractitionerwascontacted

concerningvitalstatusofthepatient.

Qualityoflife:changesperpatientpertimeinterval

121

Changes in QOL per patient between the 3 consecutive time intervals (before ICU admission-3

months;3months-1year;1year-7years)wereassessedforeachdimensionoftheEQ-5Dandeachdomain

of theSF-36.Thesechangescouldonlybeassessed if thepatientanswered theQOLsurveyonboth the

start and end of the respective time interval. Changes in QOL were considered clinically important if

patientsreportedanotherlevelforthedifferentEQ-5Ddimensionsorforthehealthtransition(HT)ofthe

SF-36,oriftherewasaminimumdifferenceof7pointsintheEQ-visualanaloguescale(VAS)or5pointsin

the norm-based physical (PCS) andmental (MCS) component scores of the SF-36. Otherwise, QOLwas

consideredthesamebetweenthedifferenttimeintervals.29

Post-hospitaltrajectories

Post-hospitaltrajectorieswereassessedforeachsurvivingpatientbytheelectronicpatientrecord,

whichiskeptinthehospitalcomputersystem.Withinthissystem,thepatient’srecordsandconsultations

in other hospitals can also be assessed so a complete trajectory of the patient after the initial hospital

admissioncanbemade.

Statisticalanalysis

Valuesareexpressedasmedian (IQR) forcontinuousvariablesandasnumber (%) forcategorical

variables.QOLbeforeICUadmissionandcharacteristicsbetweenhospitalsurvivorsandnon-survivorswere

compared by the Mann-Whitney U test for continuous variables and by Chi-square test for categorical

variables.Chi-square(EQ-5D)orFriedmantest(SF-36)assesseddifferencesbetweenQOLatbaseline(only

hospitalsurvivors),at3months,at1yearand7yearsafterICUdischarge.Allstatisticalanalysesweredone

usingIBMSPSSStatisticssoftwareversion22.Atwo-sidedP<0.05wasconsideredsignificant.

RESULTS

CharacteristicsandOutcomesoftheStudyPopulation

Patientcharacteristics,organfailuresandoutcomesareshowninTablesIandII.131patients(60%

males) with median age of 83 years (IQR 81-85) and Charlson comorbidity index of 2 (IQR 0-4) were

included. ICU admission reasons weremedical (55%), emergency surgery (23%), elective surgery (12%),

trauma(9%),andburns(1%).APACHEIIandSOFAscoresuponICUadmissionwere20(IQR15-24)and4(3-

8)respectively.Hospitalnon-survivorshadhigherseverityofillnessatadmissionandrequiredmoreorgan

supportthanhospitalsurvivorsalthoughtherewerenodifferences incomorbidity,baselinefunctionality,

orICUadmissionreason.Therapeuticlimitationsweresetin34patients(26%)after2days(IQR1-5)atthe

ICU. ICU, hospital, 3 months, 1-year and 7-years mortality rates were 17%, 29%, 39%, 50%, and 84%

respectively.

Qualityoflife:evolutionovertime

122

ThenumberofQOLsurveysateachtimepoint,responserate,andpatientsdyingduringthestudy

course are shown in Figure 1.Most patients answered the questionnaires themselves, respectively 60%

beforeICU,60%at3monthsand57%1year(P=0.94)afterICUdischarge.After7years,QOLsurveyswere

completedbynextofkin(44%),bythepatientthemself(28%),orbyotherfamily(28%).MedianageatQOL

evaluationafter7yearswas89years(IQR88-90years).

There were no differences in QOL before ICU admission between hospital survivors and non-

survivors(datanotshown).

EQ-5D assessments over time showed that the number of patients with disabilities increased

almost at each of the consecutive time points, which was significant for mobility (P=0.018), self-care

(P=0.011),usualactivities(P=0.007),andanxiety/depression(P=0.035)(Figure2.I).

SF-36measurementsdemonstratedthatQOLdecreased3monthsafterICUdischargecomparedto

baseline,improvedafter1year,especiallymentally,butworsenedagainafter7years(Figure2.II).Thiswas

significant for physical functioning (P=0.001), general health (P=0.009), and social functioning (P=0.001).

Long-termQOLremainedunderbaselinelevelandunderQOLofthegeneralpopulation.

VASinelderlydidnotsignificantlychangeovertime(respectively70,60,65,and63atbaseline,3

months,1year,and7yearsafterICUdischarge(P=0.464)).

Qualityoflife:Perceptionofchangesperpatientpertimeinterval

ForallEQ-5Ddimensions,mostpatientsperceivednochange inQOLper time interval (Figure3).

After7years,significantmoreelderlyexperiencedaworsening inmobility (P=0.025),self-care (P=0.044),

andVAS(P=0.030).

PerceptionofchangesinPCS,MCS,andHTareshowninFigure4.After3months,themajorityof

patientsperceiveddeterioration in PCS,MCS, andHT,which changed into aperceptionof no changeor

evenbetterafter1yearandagainaperceptionofworseninginmostpatientsafter7years.

Post-hospitaltrajectoriesandadditionalquestionsafter1and7years

Post-hospitaltrajectories(TableIII)showednobigdifferencesbetweensurvivorsandnon-survivors

per respective time interval. In the first 3 months after hospital discharge, more non-survivors were

discharged to other hospitals (30.8% vs 5.0%; P=0.002) andmore had therapeutic limitations (53.8% vs

11.3%;P<0.001),whichincreasedfurtherintheyearafterhospitaldischarge(64.3%vs9.2%;P<0.001).Few

patients had a living will, whichwas drawn up belatedly. The number of new hospital admissions was

similarbetweensurvivorsandnon-survivorspertimeinterval.

Amongthe1-yearand7-yearssurvivorsrespectively,37%and11%(P=0.036) lived independently

athome,26%and28%(P=0.867)hadadditionalhomehelp,13%and22%(P=0.330) livedwith relatives,

and21%and39%(P=0.122)livedinaspecialcarefacility.Themajorityofpatientshadgood(48%and28%;

P=0.134)ornomemories(38%and67%;P=0.031)oftheirICUstay.Increasedsleepingdisturbanceswere

123

rare(11%and17%;P=0.528).81%and72%(P=0.423)ofthelong-termsurvivorsexpressedapreferenceto

bereadmittedtoanICUdepartmentincaseofdeterioration.

All but 1 of the 7-years survivors reported a very good familial and social network, a good

paramedical andmedical follow-up, experienced no financial problems, andwere happy to be still alive

despitetheiradvancedage.

DISCUSSION

TheICU(17%),hospital(29%)andlong-term(50%at1year,84%at7years)mortalityratesfoundin

our studycanbecompared toother studies1,4,5,6,9,15,16,30,31 althoughmortality ratesmaybedifficult to

compare because of differences in patient selection, in the applied definition of elderly patients,

differencesinpre-ICUtriagedecisions,andintimeline.3,9,32Ahighnumberoftherapeuticlimitations(26%)

weresetshortlyafterICUadmission.

Objectively seen, the long-term QOL in elderly in our study was low compared to a general

population,particularlyinself-care,usualactivitiesandthephysicaldomains,withanincreasingnumberof

patientsexperiencingmoreproblemsovertime.Itseemshoweverwithinnormalevolutionaryexpectations

thatthemorephysicalcomponentsofQOLwilldeterioratewithadvancedage,evenwhetherornotelderly

have been admitted to an ICU department before.3 More important is to assess their perceptions and

changesinQOL.

ElderlyperceivedsomeworseninginQOLatlong-termbutstillevaluatedtheirQOLasacceptable.

ThisisinaccordancewithQOLandADLmeasurementsfoundinotherstudies.4-6,11-17ItsuggeststhatQOL

mighthaveanothermeaning forolderpatients,with social andmental valuesbeing farmore important

than limited physical functioning and that age itself influencesQOLmainly due to increasing number of

chronicconditions.5,13,14,31Therefore,QOLcanbehelpful indecision-makingconcerningICUadmissionof

elderly patients but its rolemay be limited at the same time.QOL interpretation in elderly is therefore

difficultandintensivistsshouldnotusetheirownframeofvaluesandreferencesinmakingjudgments.

The elderly in our study also expressed preferences for a longer life, even with reduced QOL,

probablyduetochangesinindividual’sexpectations,values,andsteadyacceptanceofdisability,especially

when theyhadagoodsocialnetwork.3,7,11,15,16Thismayexplainwhy1and7yearsafter ICUdischarge,

81% and 72% of the elderly patients in our study wanted ICU admission again if needed, which is in

accordancewithpercentages found in literature.6,14,15Thesenumbersmayseemsurprisingasphysicians

oftenincorrectlyassumethatelderlypatientsdonotwantlife-extendingcare.12

Still, it remains essential to identify these elderly patients who are most likely to benefit from

critical care, not only to prevent suffering fromunnecessary treatments but also to optimise the use of

resources.33,34 Reaching this balance is difficult andwouldbeeasierwith reliableprognostication,which

124

unfortunatelyhasbeenproventoremainchallengingatthemoment.Neithertriagescores,norhighquality

prognosticmodelscancurrentlybeconsideredassufficientlyvalidtobeapplicableinclinicalpracticeinthe

elderly.35, 36 Decisions based upon chronological age and comorbidities may also not be appropriate, as

thesemaynot capture sufficientlyall characteristicsofelderlypatients.37Recent literaturehighlights the

importanceofknowledgeoffrailtyandbaselinefunctionalityinprognosticationandappropriatedecision-

makingforelderlycritically illpatientsaspatientswhoare less frailaremore likely tosurviveandregain

good physical functioning.3, 6, 9, 10, 31, 33, 38 Biological age and frailty also proved to bemore important in

determiningoutcomesinelderlycomparedtoseverityofillnessscores.38

Intensivecareshouldthereforeonlybeindicatedwhenthecriticalconditionhasthepotentialtobe

reversible,whenbenefitsoutweighburdensandwhentheoutcomeisacceptableforthepatient.39Helpful

guidelines for decision-making concerning ICU admission or refusal are published in the SIAARTI

recommendations.40

Importantly, indecidingtoadmitelderlytotheICU, intensivistsshouldconsiderthewholehealth

process rather than focusing on the ICU period alone.37 Therefore, we also evaluated post-hospital

trajectories in elderly hospital survivors. Overall, there were no big differences after hospital discharge

between survivors and non-survivors per respective time interval. The majority of the 1- and 7-years

survivors lived at home - with or without additional help - which is an important patient-centered

outcome.3Agoodfamilial,paramedicalandmedicalnetworkwithoutfinancialproblemsaddedtoperceive

QOLasacceptable.Over time,morepatientshad therapeutic limitationsbut fewhada livingwill,which

was drawn up belatedly. Factors associated with admission in nursing homes were mainly cognitive

impairmentsandhighdependencyindailyactivities.

To thebestofourknowledge, this is the first study thatevaluated long-termoutcomesandQOL

with validated questionnaires in patients aged 80 years ormore at baseline, and 3months, 1 year and

almost7yearsafter ICUdischarge.Responserateswerehighandonlyonepatientwas lostto-follow-up.

Consequently,theimpactofanICUadmissionuponlong-termphysical,mentalandcognitivefunctioningin

theelderlycouldbeassessed,whichisrarelypossible.7,17Wehopeourstudyprovidesbetterinsightsinthe

long-termQOLandtrajectoriesofelderlyICUsurvivorsandcanhelpinbetterdecision-makingandadvance

careplanninginthisgrowingpatientcohort.

However,somelimitationshavetobementioned.First,thiswasasinglecenterstudyperformedin

a large university hospital. Study resultsmight not be applicable to other centers. Second, the inclusion

period of 12monthswas short and consequently, although all eligible patientswere included, the total

number of study patientswas low andmay lack statistical power to detect differences inQOL. Still,we

believethatourstudygivesagoodoverviewofthelong-termoutcomesincriticallyillpatientsaged80or

more.Third,mostpatientsdidnotrespondtotheQOLsurveysthemselvesforlong-termQOLassessments

125

after7years. AlthoughQOLmaybepreferentiallyevaluated fromthepatient,webelievethat forsome

elderlypatientsproxiesmayprovidethemostreliableinformation.Fourth,wedonothavedataonmedical

decision-making leading to ICU referral. Consequently, the included patients, of whom only a minority

chair-boundorbedriddenatbaseline,mightalreadyrepresentaselectionoffitterelderlypatientswitha

possible inherent better prognosis andQOL. This limitation is hardly avoidable and can alsobe found in

otherstudiesonthistopic.2,4,5,11,12Fifth,evaluationsof long-termQOLalways implysurvivalbiasasQOL

canonlybeassessed insurvivors.3Weacknowledgethat long-termQOLmayalsobemodifiedbyevents

happeningtothepatientafterhospitaldischarge.Sixth,wedidnotassessdegreeoffrailtyduringfollow-

up,aswedidnothavedataofbaselinefrailty.Nevertheless,wecanrelyuponverydetaileddatafromour

QOLsurveysandadditionalquestions.

CONCLUSION

Mostcritically ill long-termelderlysurvivors livedathome,perceivedonlydecline inmobilityand

self-care, considered theirQOLasacceptable,andwanted tobe readmitted to the ICU ifnecessarily. In

olderpatients,knowledgeofbaselineconditionismoreimportantthanageinestimatingthepossiblevalue

ofanICUadmission.

KEYMESSAGES

Themajorityofcriticallyilllong-termelderlysurvivorsperceivesonlychangesinmobilityandself-careovertimeandevaluatesQOLasacceptable.

Themajority of critically ill long-termelderly survivors prefer to be readmitted to an ICUdepartment incaseofdeterioration.

Intensivecareforveryelderlypeopleshouldonlybeindicatedwhenthecriticalconditionhasthepotentialtobereversible,whenbenefitsoutweighburdensandwhentheoutcomeisacceptableforthepatient.

AgealoneisapoorindicatorofthevaluetobegainedfromanICUadmissionincriticallyillelderlypatients.

ACKNOWLEDGEMENTS



interviewingpatients,andcallingpatientsorrelatives.TheauthorsalsothankChrisDanneelsforhishelpin


126

Table1.Patientcharacteristicsandcomorbidities

Allpatients(N=131)

Hospitalsurvivors(N=93)

Hospitalnon-survivors(N=38)

P

age,yrs(median,IQR) 83(81-85) 83(81-85) 83(81-86) 0.70

agebetween80-84years,N(%) 91(69.5) 66(71.0) 25(65.8) 0.56



malegender,N(%) 78(59.5) 57(61.3) 21(55.3) 0.52

BMI,kg/m2(median,IQR) 25.3(22.6-27.4) 25.2(23.1-27.3) 25.4(21.2-27.7) 0.97

hospitaldayspriortoICU(median,IQR) 1(0-3) 1(0-3) 0(0-4) 0.80

hospitalizationinlast6months,N(%) 45(34.3) 28(30.1) 17(44.7) 0.11

livingstatusbeforeadmission,N(%)

athome 122(93.1) 86(92.5) 36(94.7) 0.64

chroniccarefacility 8(6.1) 6(6.5) 2(5.3) 0.80

other 1(0.8) 1(1.1) 0(0) 0.52

ADL,N(%)

nolimitations 52(39.7) 39(41.9) 13(34.2) 0.41

moderatelimitations 67(51.1) 45(48.4) 22(57.9) 0.32

chair-bound 10(7.6) 7(7.5) 3(7.9) 0.94

bedridden 2(1.0) 2(2.2) 0(0) 0.36


2(0-4) 1(0-3) 2(1-3) 0.93

specificcomorbidity,N(%)

cardiovascular 79(60.3) 56(60.2) 23(60.5) 0.97

neurological 34(26.0) 24(25.8) 10(26.3) 0.95

solidtumor 34(26.0) 25(26.9) 9(23.7) 0.70

respiratory 31(23.7) 21(22.6) 10(26.3) 0.65

gastrointestinal 21(16.0) 14(15.1) 7(18.4) 0.63

renal 19(14.5) 16(17.2) 3(7.9) 0.17

Immunocompromised 7(5.3) 5(5.4) 2(5.3) 0.97

metastaticcancer 7(5.3) 4(4.3) 3(7.9) 0.41

hematologicalcancer 6(4.6) 5(5.4) 1(2.6) 0.50

N=number;yrs=years;IQR=interquartilerange(25%-75%);BMI=bodymassindex;ICU=intensivecareunit;ADL=activityofdailyliving

127

Table2.ICUadmissionreasons,organfailuresandoutcomes

Allpatients(N=131)

Hospitalsurvivors(N=93)

Hospitalnon-survivors(N=38)

P

MainreasonforICUadmission,N(%)

Medical 72(55.0) 52(55.9) 20(52.6) 0.73

Emergencysurgery 30(22.9) 18(19.4) 12(31.6) 0.13

Scheduledsurgery 15(11.5) 12(12.9) 3(7.9) 0.41

Trauma 12(9.2) 10(10.8) 2(5.2) 0.32

Burns 2(1.5) 1(1.1) 1(2.6) 0.51

SeverityofillnessatICUadmission(first24hours)

APACHEIIscore(median,IQR) 20(15-24) 18(14-23) 24(19-29) <0.001

SOFAscore(median,IQR) 4(3-8) 4(2-6) 8(4-10) <0.001

mechanicalventilation,N(%) 46(35.1) 24(25.8) 22(57.9) <0.001

vasopressors,N(%) 35(26.7) 18(19.4) 17(44.7) 0.002

RRT,N(%) 5(3.8) 3(3.2) 2(5.3) 0.58

OrganfailureduringICUstay

meanSOFAscore(median,IQR) 4(3-6) 5(3-7) 7(4-11) <0.001

mechanicalventilation,N(%) 56(42.7) 29(31.2) 27(71.1) <0.001

vasopressors,N(%) 43(32.8) 23(24.7) 20(52.6) 0.002

RRT,N(%) 7(5.3) 3(3.2) 4(10.5) 0.09

Outcomes

ICUreadmissions,N(%) 10(7.6) 4(4.3) 6(15.8) 0.02

ICULOS,days(median,IQR) 3(2-5) 3(2-5) 3(2-5) 0.33

hospitalLOS,days(median,IQR) 17(9-38) 22(11-47) 10(3-21) <0.001

DNRdecisions,N(%) 34(25.9) 11(11.8) 23(60.5) <0.001

ICUmortality,N(%) 22(16.8) 0(0) 22(57.9) <0.001

hospitalmortality,N(%) 38(29.0) 0(0) 38(100) <0.001

3-monthsmortality,N(%) 51(38.9) 13(14.0) NA -

1-yearmortality,N(%) 65(49.6) 27(29.7) NA -

7-yearsmortality,N(%) 110(84.0) 82(77.4) NA -

N=number;ICU=intensivecareunit;IQR=interquartilerange(25%-75%);APACHEII=AcutePhysiologyandChronicHealthEvaluation;SOFA=SequentialOrganFailureAssessment;RRT=renalreplacementtherapy;LOS=lengthofstay;DNR=do-not-resuscitate;NA=notapplicable

128

Table3.Trajectoriespertimeinterval

hospitaldischargeto3months

P 3monthsto1yearafterhospitaldischarge

P* 1yearto7yearsafterhospitaldischarge

P**

survivorsN=80

3monthsnon-survivorsN=13

survivorsN=65

1-yearnon-survivorsN=14

survivorsN=20

7-yearsnon-survivorsN=45

1livingabroad

dischargelocationfromtheinitialhospitaladmission,N(%)

home 57(71.3) 8(61.5) 0.479 44(67.7) 12(85.7) 0.178 13(65.0) 31(68.9) 0.757otherhospital 4(5.0) 4(30.8) 0.002 3(4.6) 1(7.1) 0.696 7(35.0) 3(6.7) 0.003specialcarefacility19(23.8) 1(7.7) 0.191 18(27.7) 1(7.1) 0.103 0(0) 11(24.4) 0.015patientswiththerapeuticlimitations,N(%) 9(11.3) 7(53.8) <0.001 6(9.2) 9(64.3) <0.001 5(25.0) 21(46.7) 0.100newhospitaladmission,N(%)none 52(65.0) 5(38.5) 0.068 39(60.0) 7(50.0) 0.491 4(20.0) 17(37.8) 0.1571 20(25.0) 6(46.2) 0.115 17(26.2) 4(28.6) 0.854 6(30.0) 9(20.0) 0.3772 4(5.0) 1(7.8) 0.690 5(7.7) 1(7.1) 0.944 1(5.0) 8(17.8) 0.169>2 4(5.0) 1(7.8) 0.690 4(6.2) 2(14.3) 0.298 9(45.0) 11(24.4) 0.097patientswithlastwill,N(%)

0(0) 0(0) NA 0(0) 4(28.6) <0.001 0(0) 5(11.1) 0.121

ICUadmissiontodeath,days(median,IQR) NA 43

(29-78)- NA 248

(150-327)

- NA 1196(689-1737)

-

hospitaldischargetodeath,days(median,IQR) NA 20

(8-40)- NA 196

(110-319)

- NA 1130(646-1712)

-

placeswherepatientsdied,N(%)tertiaryhospital,ICU NA 1(7.8) - NA 0(0) - NA 1(2.2) -tertiaryhospital,ward

NA 3(23.1) - NA 4(28.6) - NA 6(13.3) -

otherhospital NA 4(30.8) - NA 2(14.3) - NA 4(8.9) -athome NA 4(30.8) - NA 3(21.4) - NA 9(20.0) -specialcarefacility NA 1(7.8) - NA 1(7.1) - NA 17(37.8) -unknown NA 0(0) - NA 4(28.6) NA 8(17.8) -N=number; IQR=interquartilerange; ICU=intensivecareunit;NA=notapplicable;P= levelofsignificancebetweensurvivorsandnon-survivorsintheafterhospitaldischarge-3monthsafterICUdischargetimerange;P*=levelofsignificancebetweensurvivorsandnon-survivors in the3months-1yearafter ICUdischarge time range;P**= levelof significancebetweensurvivorsandnon-survivorsinthe1year-7yearsafterICUdischargetimerange

129

Figure1.Patientcohort

N=number;ICU=intensivecareunit;QOL=qualityoflife

130

Figure2.QOLassessmentsovertime

EQ-5D assessments over time: Percentage of patients with moderate or severe problems per dimension at the 4different time points. The X-axis represents the different dimensions of the EQ-5D. The Y-axis represents thepercentages(%)ofpatientswithmoderateorsevereproblemsinarespectivedimension.SignificantP-values(P<0.05)(Chi-Squaretest)areshownabovetherespectivedimensions.SF-36 assessments over time: Norm-based median scores per domain at the 4 different time points. The X-axisrepresents thedifferentdomainsof theSF-36.TheY-axis represents thenorm-basedmedianscores ina respectivedomainof theSF-36.Anorm-basedmedianscorebetween47-53 inagroupofpatients is consideredasnormaloraverage.Norm-basedmedianscoresbelow47 indicate impaired functioningorbelowaverage;norm-basedmedianscores above 53 indicate better functioning or above average; the higher the score, the better the condition.SignificantP-values(P<0.05)(Mann-WhitneyUanalysis)areshownabovetherespectivedomains.PCS= physical component score;MCS=mental component score; PF= physical functioning; RP= role physical; BP =bodily pain;GH= general health; VT= vitality; SF= social functioning; RE= role emotional;MH=mental health; ICU=intensivecareunit;*=hospitalsurvivorsonly

131

Figure3.PerceptionsofchangesinQOLperpatientpertimeinterval

TheX-axisrepresentsthedifferenttimeintervalswiththenumberofpatientswhorespondedonbothstartandendoftherespectivetimeinterval.TheY-axisrepresentsthepercentages(%)ofpatientswhoperceivedthechangeinQOLasthesame(blue),worse(red),orbetter (green)per respective time intervalandper respectivedimension (EQ-5D)ordomain (SF-36).Onlysignificantp-levelsofdifferencesinpercentageofpatientswhoperceivethechangeasthesame,worseorbetterovertimeareshown.QOL=qualityoflife;N=number;VAS=visualanaloguescale;base-3m=changeinQOLperpatientbetweenQOLbeforeICUadmissionand3monthsafterICUdischarge;3m-1yr=changeinQOLperpatientbetween3monthsand1yearafterICUdischarge;1yr-7yrs=changeinQOLperpatientbetween1yearand7yearsafterICUdischarge

132

REFERENCES

1. BagshawSM,WebbSA,DelaneyA,GeorgeC,PilcherD,HartGK,etal.Veryoldpatientsadmitted to intensivecare inAustraliaandNewZealand:amulti-centrecohortanalysis.CritCare2009;13:R45.2. NielssonMS,ChristiansenCF,JohansenMB,RasmussenBS,TønnesenE,NørgaardM.MortalityinelderlyICUpatients:acohortstudy.ActaAnaesthesiolScand2014;58:19-26.3. Conti M, Merlani P, Ricou B. Prognosis and quality of life of elderly patients after intensive care. Swiss Med Wkly2012;142:w13671.4. Andersen FH, FlaattenH, Klepstad P, RomildU, Kvåle R. Long-term survival and quality of life after intensive care forpatients80yearsofageorolder.AnnIntensiveCare2015;5:13.5. DaubinC,ChevalierS,SéguinA,GaillardC,ValetteX,PrévostF,etal.Predictorsofmortalityandshort-termphysicalandcognitivedependenceincriticallyillpersons75yearsandolder:aprospectivecohortstudy.HealthQualLifeOutcomes2011;9:35.6. SacanellaE,Pérez-CastejonJM,NicolasJM,MasanesF,NavarroM,CastroP,etal.Functionalstatusandqualityoflife12monthsafterdischargefromamedicalICUinhealthyelderlypatients:aprospectiveobservationalstudy.CritCare2011;15:R105.7. HennessyD,JuzwishinK,YergensD,NoseworthyT,DoigC.Outcomesofelderlysurvivorsof intensivecare:areviewoftheliterature.Chest2005;127:1764-74.8. Oeyen SG, Vandijck DM, Benoit DD, Annemans L, Decruyenaere JM. Quality of life after intensive care: a systematicreviewoftheliterature.CritCareMed2010;38:2386-400.9. HeylandDK,GarlandA,BagshawSM,CookD,RockwoodK,StelfoxHT,etal.Recoveryaftercriticalillnessinpatientsaged80yearsorolder:amulti-centerprospectiveobservationalcohortstudy.IntensiveCareMed2015;41:1911-20.10. BagshawSM,StelfoxHT,JohnsonJA,McDermidRC,RolfsonDB,TsuyukiRT,etal.Long-termassociationbetweenfrailtyandhealth-relatedqualityoflifeamongsurvivorsofcriticalillness:aprospectivemulticentercohortstudy.CritCareMed2015;43:973-82.11. Hofhuis JG, van Stel HF, Schrijvers AJ, Rommes JH, Spronk PE. Changes of health-related quality of life in critically illoctogenarians.Chest2011;140:1473-83.12. deRooijSE,GoversAC,KorevaarJC,GiesbersAW,LeviM,deJongeE.Cognitive,functional,andquality-of-lifeoutcomesofpatientsaged80andolderwhosurvivedatleast1yearafterplannedorunplannedsurgeryormedicalintensivecaretreatment.JAmGeriatrSoc2008;56:816-22.13. RochA,WiralusS,PaulyV,Forel JM,GuervillyC,GainnierM,etal.Long-termoutcome inmedicalpatientsaged80oroverfollowingadmissiontoanintensivecareunit.CritCare2011;15:R36.14. TabahA,PhilippartF,TimsitJF,WillemsV,FrançaisA,LeplègeA,etal.Qualityoflifeinpatientsaged80oroverafterICUdischarge.CritCare2010;14:R2.15. MerlaniP,ChenaudC,MariottiN,RicouB.Long-termoutcomeofelderlypatientsrequiringintensivecareadmissionforabdominalpathologies:survivalandqualityoflife.ActaAnaesthesiolScand2007;51:530-7.16. Kaarlola A, TallgrenM, Pettilä V. Long-term survival, quality of life, and quality-adjusted life-years among critically illelderlypatients.CritCareMed2006;34:2120-6.17. KhouliH,AstuaA,DombrowskiW,AhmadF,HomelP,ShapiroJ,etal.Changesinhealth-relatedqualityoflifeandfactorspredictinglong-termoutcomesinolderadultsadmittedtointensivecareunits.CritCareMed2011;39:731-7.18. OeyenS,BenoitD,AnnemansL,DecruyenaereJ.Qualityoflifebefore,3monthsand1yearafterICUdischarge.CritCareMed2010;38:P587(supplDecember).19. KatzS,DownsTD,CashHR,GrotzRC.ProgressindevelopmentoftheindexofADL.Gerontologist1970;10:20-30.20. Charlson ME, Pompei P, Ales KL, Mackenzie CR. A new method of classifying prognostic comorbidity in longitudinalstudies:developmentandvalidation.JChronDis1987;40:373-83.

133

21. KnausWA,DraperEA,WagnerDP,ZimmermanJE.APACHEII:aseverityofdiseaseclassificationsystem.CritCareMed1985;13:818-29.22. Vincent JL, Moreno R, Takala J, Willatts S, De Mendonça A, Bruining H, et al. The SOFA (Sepsis-related Organ FailureAssessment) score to describe organ dysfunction/failure. On behalf of the Working Group on Sepsis-related Problems of theEuropeanSocietyofIntensiveCareMedicine.IntensiveCareMed1996;22:707-10.23. Ware JE, KosinskiM, Bjorner JB, Turner-Bowker DM, Gandek B,MaruishME. User’sManual for the SF-36v2® HealthSurvey.QualityMetricIncorporated,Lincoln,RhodeIsland;2007.24. EuroQolGroup.EuroQol--anewfacilityforthemeasurementofhealth-relatedqualityoflife.HealthPolicy1990;16:199-208.25. ChrispinPS,ScottonH,RogersJ,LloydD,RidleySA.Short-Form36intheintensivecareunit:assessmentofacceptability,reliabilityandvalidityofthequestionnaire.Anaesthesia1997;52:15-23.26. Angus DC, Carlet J, 2002 Brussels Roundtable Participants. Surviving intensive Care: A report from the 2002 BrusselsRoundtable.IntensiveCareMed2003;29:368-77.27. OeyenSG,BenoitDD,AnnemansL,DepuydtPO,VanBelleSJ,TroisiRI,etal. Long-termoutcomeandqualityof life incriticallyillpatientswithhematologicalorsolidmalignancies:asinglecenterstudy.IntensiveCareMed2013;39:889-98.28. OeyenS,DeCorteW,BenoitD,AnnemansL,DhondtA,VanholderR,etal.Long-termqualityoflifeincriticallyillpatientswithacutekidneyinjurytreatedwithrenalreplacementtherapy:amatchedcohortstudy.CritCare2015;19:289.29. PickardAS,NearyMP,CellaD.Estimationofminimally importantdifferences inEQ-5DutilityandVASscores incancer.HealthQualLifeOutcomes2007;5:70.30. RellosK,FalagasM,VardakasK,SermaidesG,MichalopoulosA.Outcomeofcriticallyilloldest-oldpatients(aged90andolder)admittedtotheintensivecareunit.JAmGeriatrSoc2006;54:110-4.31. FerranteLE,PisaniMA,MurphyTE,GahbauerEA,Leo-SummersLS,GillTM.Functionaltrajectoriesamongolderpersonsbeforeandaftercriticalillness.JAMAInternMed2015;175:523-9.32. FlaattenH,Garrouste-OrgeasM.TheveryoldICUpatient:anever-endingstory.IntensiveCareMed2015;41:1996-8.33. PiersRD,AzoulayE,RicouB,DekeyserGF,DecruyneaereJ,MaxA,etal.PerceptionsofappropriatenessofcareamongEuropeanandIsraeliintensivecareunitnursesandphysicians.JAMA2011;306:2694-703.34. Pelavski AD, De Miguel M, Rochera MI, Lacasta A, Roca M. Immediate postoperative and mid-term survival innonagenariansundergoingnon-traumaticemergencysurgery.MinervaAnestesiol2014;80:796-804.35. LaMantiaMA,StewartPW,Platts-MillsTF,BieseKJ,ForbachC,ZamoraE,etal.Predictivevalueofinitialtriagevitalsignsforcriticallyillolderadults.WestJEmergMed2013;14:453-60.36. Minne L, Ludikhuize J, de Jonge E, de Rooij S, Abu-HannaA. Prognosticmodels for predictingmortality in elderly ICUpatients:asystematicreview.IntensiveCareMed2011;37:1258-68.37. Riou B, Boddaert J. The elderly patient and the ICU: Where are we going, where should we go? Crit Care Med2016;44:231-2.38. LeMaguet P, Roquilly A, Lasocki S, Asehnoune K, Carise E, SaintMartinM, et al. Prevalence and impact of frailty onmortalityinelderlyICUpatients:aprospective,multicentre,observationalstudy.IntensiveCareMed2014;40:674-82.39. ZamperettiN,PiccinniP.Endoflifeintheintensivecareunit.MinervaAnestesiol2010;76:541-7.40. End of life care and the intensivist: SIAARTI recommendations on the management of the dying patient. MinervaAnestesiol2006;72:927-47.

135

V.Developmentofapredictionmodelfor

long-termqualityoflifeincriticallyill

patientsSandraOeyen,MD1,2,KarelVermeulen,PhD3,DominiqueBenoit1,2,MD,PhD1,2,LievenAnnemans,PhD4,JohanDecruyenaere,MD,PhD1,2

1FacultyofMedicineandHealthSciences,GhentUniversity,Ghent,Belgium2DepartmentofIntensiveCare,GhentUniversityHospital,Ghent,Belgium3FacultyofBioscienceEngineering,DepartmentofMathematicalModelling,StatisticsandBioinformatics,Ghent

University,Ghent,Belgium4FacultyofMedicineandHealthSciences,DepartmentofPublicHealth,GhentUniversity,DePintelaan185,9000

Ghent,Belgium

PublishedinJCritCare,2018;43:133-138

136

ABSTRACT

Purpose:Wedevelopedapredictionmodel forqualityof life (QOL)1yearafter intensivecareunit (ICU)

dischargebasedupondataavailableatthefirstICUdaytoimprovedecision-making.

Methods:Thedatabaseofa1-yearprospectivestudyconcerninglong-termoutcomeandQOL(assessedby

EuroQol-5D) in critically ill adult patients consecutively admitted to the ICU of a university hospital was

used. Caseswithmissing datawere excluded.Utility indices at baseline (UIb) and at 1 year (UI1y)were

surrogates forQOL.For1-yearnon-survivorsUI1ywassetat zero.Thegrouped lasso techniqueselected

themostimportantvariablesinthepredictionmodel.R2andadjustedR2werecalculated.

Results: 1831of1953cases (93.8%)werecomplete.UI1ydependedsignificantlyon:UIb (P<0.001); solid

tumor (P<0.001); age (P<0.001); activity of daily living (P<0.001); imaging (P<0.001); APACHE II-score

(P=0.001);≥80years (P=0.001);mechanical ventilation (P=0.006);hematologicalpatient (P=0.007); SOFA-

score (P=0.008); tracheotomy (P=0.018); admission diagnosis (surgical P<0.001 (versus medical); and

comorbidity (P=0.049). Only baseline health status and surgical patients were positively associatedwith

UI1y.R2was0.3875andadjustedR20.3807.

Conclusion: Althoughonly40%of variability in long-termQOL couldbeexplained, thispredictionmodel

canbehelpfulindecision-making.

137

INTRODUCTION

Uncertainty about outcomes in critically ill patients admitted to the intensive care unit (ICU) is

heavytobearforpatientsandfamily.Ingeneral,patientsandfamilyonlyassociateoutcomewithsurvival

andoften,unrealisticexpectationsatlong-termarehopedfor[1].Thetrueburdenofdiseaseanditslong-

termconsequencesonphysical,mentalandcognitivefunctioningmaybeunderestimated[2,3],aswellas

thepossibilitytoreturntoformerdailylifeandoverallqualityoflife(QOL)[4].

It is the important task of critical care physicians to informpatients and family in a reliableway

about these outcomes. However, for critical care physicians too, uncertainty concerning long-term

functionalityandQOL isdifficult tohandle [5].Major reductions in long-termQOLwereseen incasesof

severeacute respiratorydistress syndrome,prolongedmechanical ventilation, trauma,and severe sepsis

[6]. Still, long-term QOL remains difficult to predict for the individual patient and patients and families

frequentlyarenotwellbriefedaboutexpectedlong-termsurvivalandfunctionalitydespiteexplicitwishes

tohavethisinformation[7].

Accuratepredictionmodels canguidephysicians in their handling, communication, anddecision-

making.Predictionmodelsincriticalcaredoexistbuttheirroleindecision-makingishoweverlimited[8].

Severityofillnessandorganfailurescoresmainlyfocusonestimationofshort-termmortalityrisk[9-15].

Some prediction models may focus on very specific patient populations or problems and are not

generalizabletoabroadpatientapplicationincriticalcare[7,16-22].Somemodelsarerathercomplex[10,

23], not accurate enough [24], or ignore that better future treatments may improve prognosis [19].

Although somepredictionmodels focusedon long-termmortality [7, 25], short-term [24] and long-term

functionaloutcome[16],noneofthemodelsestimatedlong-termQOLingeneralcriticallyillpatients.

Therefore,itwasouraimtodevelopaneasytouseandaccuratepredictionmodelforthemean

QOLat1yearafterICUdischargeingeneralcriticallyillpatientsbasedupondatareadilyavailableatthe

firstICUday(D1)(D1=first24hoursofICUadmission).

MATERIALSANDMETHODS

Designandsetting

TheD1-predictionmodelwas retrospectivelydevelopedbasedupondataofa1-yearprospective

cohortstudy.Thisstudyfocusedonlong-termoutcomeandQOLincriticallyilladult(≥16years)patients

consecutivelyadmittedtothe22-bedsurgicalICU,the14-bedmedicalICU,andthe6-bedburnunitofthe

GhentUniversityHospital,a tertiarycare facility inBelgium[26]. Incaseofmultiple ICUadmissions,only

the firstwasconsidered.Patientsadmittedto the10-bedcardiacsurgicalunitaftercardiacsurgerywere

notincludedinthestudycohort.

138

TheGhentUniversityHospitalICUisaclosedICUwherepatientsaretreatedbyateamoffull-time

criticalcarephysicians,nursesandphysiotherapists.

The original observational study was approved by the local ethical committee (Ethisch Comité

GhentUniversityHospital;project2007/423approved06December2007)(B67020072805),andconducted

in accordance with the declaration of Helsinki. A signed informed consent was obtained from every

includedpatientorhislegalrepresentative.


Data collected within the first 24 hours of ICU admission included contact information of the

patient, proxy, and general practitioner, demographics, hospital days prior to ICU admission, living and

work circumstances before ICU admission, functionality asmeasuredby the Katz activities of daily living

(ADL)scale[27],hospitalizationinthelast6months,comorbidityasmeasuredbytheCharlsoncomorbidity

index [28],main ICU admission diagnosis (surgical,medical, burns, or trauma), admission circumstances

(planned-unplanned/during weekend or not), if the patient belonged to 1 or more of the predefined

subgroups (sub) (oncological, hematological, liver cirrhosis Child-Pugh B or C, or elderly (≥ 80 years)

patient),AcutePhysiology andChronicHealth Evaluation (APACHE II) score [9], SequentialOrgan Failure

Assessment(SOFA)score[13],TherapeuticInterventionScoringSystem-28score(TISS-28score)[29],Nine

Equivalent ofNursingManpowerUse score (NEMS-score) [30], do-not-resuscitate (DNR) codes, need for

invasive mechanical ventilation, vasopressors, renal replacement therapy (RRT), medical imaging

(regardlessofnumberortype),transfusionwithbloodproducts,surgery,ortracheotomy.

During ICU stay SOFA, TISS-28 and NEMS-scores, DNR-codes, need for invasive mechanical

ventilation,vasopressors,RRT,medical imaging,transfusion,surgery,ortracheotomywerecollectedona

daily base. ICU length of stay (LOS), hospital LOS, vital status at ICU and hospital discharge, and 1 year

followingICUdischargewerecollectedforeachpatient.

Qualityoflifeassessments

QOLwasassessedbymeansof theEuroQoL-5D (EQ-5D) [31].Thisquestionnaire is validatedand

foundsuitableformeasuringQOLinthecriticallyillpopulation[32].Itmeasureshealthinfivedimensions:

mobility, self-care, usual activities, pain/discomfort, and anxiety/depression. Each dimension has three

levels:noproblems,moderateproblemsorsevereproblems.Therefore,patientscanbeclassifiedinto1of

243(35)possiblehealthstates.

Weconvertedeachhealthstateintothecorrespondingutilityindex(UI),indicatingthepreference

ofbeinginahealthstatus[33].UIcanrangefrom-0.1584(severeproblemsonalldimensions)to1.000(no

problems on all dimensions). UI=0.0000 equals dead. In 17 of the 243 possible health states the

correspondingUIgoesbelowzero, indicatingahealthstateassumedtobeworsethandead.Thepatient

139

willthenhavesevereproblemsinatleast3or4orinall5dimensions,mainlyinthepain/discomfortand

anxiety/depressiondimensions.

Another part of the EQ-5D is the visual analogue scale (VAS), where patients can rate their

perceivedoverallhealthbetween0and100.

QOLwasassessedatbaseline(definedasQOL2weeksbeforeICUadmission)andatstrictly1year

afterICUdischarge.FollowingICUadmissionandstudyinclusion,aface-to-faceinterviewtoassessbaseline

QOLwas done as soon as possible. This interviewwas preferably taken from the patient, or if deemed

impossible, from theproxy.Oneyearafter ICUdischarge,patientswere sent theEQ-5Dby regularmail.

Patients or relatives were contacted by phone to assess the 1-year QOL if the questionnaire was not

returnedwithinonemonth.Eventually, thegeneralpractitionerwascontactedconcerningsurvivalstatus

ofthepatient.

UIatbaseline(UIb)andUIat1yearafterICUdischarge(UI1y)wereusedassurrogateforQOLat

thattimepoint.VASbandVAS1yexpressedperceivedQOLatbaselineand1yearafterICUdischarge.UI1y

andVAS1yfornon-survivorsweresetatzerotoavoidsurvivalbias.

Statisticalanalysis

For the development of the D1-prediction model, three different multivariate linear regression

models, respectivelyModel I, II, and III,were fittedwithUI1yasprimaryoutcome.Model Iassessed the

bivariate association between UIb and UI1y. Model II (“full” model) included all possible available D1

predictors in the linear regression analysis.Model III (“reduced”model) included only predictors in the

linearregression,whichwereselectedbythegroupedlassotechnique.

Lasso (least absolute shrinkage and selection operator) is a regression analysis method that

performs both variable selection and regularization in order to enhance the prediction accuracy and

interpretabilityofthestatisticalmodelitproduces.Thegroupedlassotechniqueallowspredefinedgroups

ofcovariates,suchasallvariablesencodingacategoricalcovariate,tobeselected intooroutofamodel

together. This techniquewas applied to identify the optimal number andmost important predictors for

UI1y in theD1 linear regressionmodel in order to simplify themodel, and to copewith the categorical

variables[34,35].

Onlycompletecases(=patientswithoutmissingdata)wereincludedinthestatisticalanalysis.The

numberofincludedcasesvariedrelativetotheconsideredmodel.

Foreachrespectivemodel,theR2(=proportionofexplainedvariance),adjustedR2(=proportionof

explained variance, taking into account the number of variables), and the root of the cross-validated

prediction error were calculated. By using (10-fold) cross-validation, the root of the cross-validated

predictionerrorgivesanhonestreflectionofthepredictivecapabilityoftheconsideredmodelbysplitting

140

thedata intoatrainingsetandtestset10timesenablingpredictionofthetestdatabasedonsolelythe

trainingdata.

The F-test compared the fit of the reducedModel IIIwith the fullModel II. Descriptive statistics

weredonewithIBMSPSSStatisticssoftwareversion23.Linearregressionanalysisforthedevelopmentof

theD1-modelwasdonewiththeR3.2.2softwarepackage[36].Thegroupedlassotechniquewasexecuted

usingthe“grpreg”routineavailableinthe“grpreg”package[37].

RESULTS

A total of 1953 patients (847 surgical, 895medical, 48 burn, 163 trauma) were included in the

originalobservationalstudy.Respectively1867(95.6%),1809(92.6%),and1831(93.8%)ofthe1953cases

werecompleteandincludedfordevelopmentofrespectivelymodelsI,II,andIII.Demographics,admission

characteristics,organfailuresandoutcomesforallcasesandforthesubsetsofcompletecasespermodel

are described in Table 1. Results were very similar between the different models, which is a strong

indication that there were no systematic differences in the subsets of included cases per model.

MissingnessofvariablesisdescribedinTable2.

Development of the D1-prediction model was based upon all 32 variables (10 continuous, 16

binary,6categorical)readilyavailableatD1ofICUadmission(Table3).

ForeachrespectivemodeltheR2,adjustedR2,andtherootofthecross-validatedpredictionerror

werecalculated(Table4).

ModelIrevealedapositiveassociationbetweenUIbandUI1y.UIbcouldexplain20%ofvariability

inUI1y(Table4).

Model II (“full” model) held all possible 32 D1-predictors (Table 3). The multivariate linear

regression analysis (data not shown) revealed the following significant D1-predictors (significance level

0.10) for UI1y (in order of decreasing importance): UIb, main ICU diagnosis, sub oncological, ADL, age,

APACHE II, D1.medical imaging, sub elderly, sub hematological, D1.surgery, origin of ICU admission,

D1.SOFA,D1.MV,D1.tracheotomy,originofhospitaladmissionandCharlsonco-morbidity index.UIbwas

positivelyassociatedwithUI1y.Themodelcouldexplain40%ofthevariability inUI1y (Table4).Variable

selectionwasdifficultbecauseofthemanycorrelationsbetweenthedifferentcovariates(datanotshown).

The grouped lasso technique revealed 17 possible D1-predictors to be included in Model III

(“reduced”model) (Figure 1). We excluded oneD1-predictor (D1.NEMS) because of lack of significance

(coefficientestimate0.00006,standarderror=0.0018,p=0.973)andfinally,16selectedD1-predictorswere

included inModel III.Multivariate regressionanalysis is shown inTable5.Finally,D1-predictionofmean

UI1ybaseduponModelIIIcanbeobtainedby:

141

MeanUI1y=0.56+0.0009*VASb+0.3017*UIb–0.1190*suboncological–0.1077*subhematological–

0.1035*subelderly-0.0023*age–0.0931*ADL2–0.1794*ADL3–0.1186*ADL4–0.0067*Charlson–

0.0047*APACHEII+0.1102*mainICUdiagnosis2+0.0346*mainICUdiagnosis3–0.0151*mainICU

diagnosis4–0.0092*D1.SOFA–0.0728*D1.DNR–0.0530*D1.mechanicalventilation–

0.0329*D1.vasopressors–0.0689*D1.medicalimaging–0.1238*D1.tracheotomy.

Only UIb, VASb, and surgical or burn patients (versus medical patients) were positively associated with

UI1y.

ExplanationofvariabilityinUI1yandcross-validatedpredictionerrorofModelIIIwerecomparable

orevenbetter than theseofModel II (Table4). Byusing cross-validation, the latterprovidesanhonest

reflection of the uncertainty for making new predictions using the corresponding model. The F-test

revealednosignificantbetterfitforthefullModelIIcomparedtothereducedModelIII(p=0.432).

DISCUSSION

Wefitted3different linear regressionmodels todevelopaneasy touseandaccurateprediction

model for themeanQOL at 1 year after ICU discharge in general critically ill patients based upon data

readilyavailableatthefirstICUday.ModelI,whichpositivelyrelatedUIbandUI1ycouldonlyexplain20%

ofthevariability inUI1y.Bothmodels II,whichheldall32possibleD1-predictors,and III,withareduced

amountofthemost importantandpowerfulD1-predictors,explained40%ofvariability inmeanUI1y.As

this latterD1-predictionmodelwas less complex, had a better performance and fit, and could easily be

implemented in an electronic patient data file, we preferred this “reduced” D1-prediction model for

predictionofUI1y.

Forcenturies,humanshavetriedtopredictthefuture. Inmedicine,thedatarichenvironmentof

criticalcarehasledthewayinoutcomepredictionbecauseofitsusefulnessinimprovingdecision-making

under uncertainty, especially when the stakes are so high. However, ICU risk predicting systems lack

patient-centerednessandoftenfailtopredictlong-termmortalityandlong-termfunctionaloutcomes[38].

Even until recent, estimation of long-term QOL was considered too challenging to be reliably used in

medicaldecision-makingasQOLwasthoughttobetoopersonalandtoosubjective[39].

Apredictionmodel for long-termQOLbasedupon readily availabledata inanearly stageof ICU

admission could thereforehelp critical carephysicians to identify thosepatientswhowill return to their

baselinefunctionality,orthosewhowillneedalongrevalidation.Itcouldalsohelptoinformpatientsand

familiesinareliableway,totriagepatientsforICUadmission,toguideintreatmentdecisions,anditcould

eventually help to transform future healthcare by making better prospects of recovery and better

allocationofresources[40,41].

Still, predictionmodels have not gainedmuch acceptance in clinical practice,mainly because of

142

complexalgorithmsthathamperimplementationindailypractice,andbecauseofconcernsofbeingwrong

[24]. Our reduced D1-prediction model could explain 40% of variability of UI1y. This is acceptable but

nevertheless,ahigheraccuracywouldbebetter.Still,model III,as it isbaseduponreadilyavailabledata

within the first 24hoursof ICUadmission, andas it is easy tousewithinanelectronicpatientdata file,

couldbeconsideredasahelpfultool foramoresystematicapproachof integrationofallD1-variablesof

theindividualcriticallyillpatient.

Althoughitisnotdefinedtowhatlevelmodelpredictionscouldbehelpfulandbeyondthescopeof

our study, it certainly might facilitate decisions, which otherwise should have been taken based upon

subjective evaluation alone [42]. The D1-prediction model will never replace clinician’s judgments, but

rather informand reinforce these judgments, as recommendations for further carehighly correlatewith

physician’sestimationsofagoodlong-termQOL[7,8,16,43].Furtherresearchshouldfocusonrefiningof

thisQOLpredictionmodel.

WithinourQOLpredictionmodel,wewereabletoidentify16D1-variablesthathadgreat impact

onlong-termoutcome.BaselineQOLandfunctionalityappearedtobestrongpositivepredictorsforlong-

termQOL. This is in accordancewith the findings of Veerbeeck [24] and Heyland [16]who respectively

demonstratedthatagoodbaselineneurologicalstatusinstrokepatientsandgoodbaselinefunctionalityin

elderlypatientshadagreatimpactonlong-termADLandfunctionality.

WealsofoundthatthepredictedUI1yforsurgicalpatientswassignificantlyhigherversusmedical

patients(p<0.001).Thiswasincontrasttoburnpatient(p=0.484)ortrauma(0.618)patients,forwhomwe

couldnotdemonstrateanysignificantdifferenceinUI1yversusmedicalpatients.

The study has several strengths. First, to the best of our knowledge, this is the first simple D1-

predictionmodelwhichhasanacceptableaccuracyandwhichfocusonlong-termQOLingeneralcritically

illpatients.Second, it isoriginalanddealswithavery important issuenowadays incriticalcare. Itmight

have several consequences on resources allocation and anticipates a clear discussion with patients and

family members regarding prognosis and preparation for outcomes. Third, the prediction model was

developed upon prospectively accurately collected data. Fourth, there was no selection bias in the

database, because of the consecutive and prospective enrollment of patients and the high long-term

follow-uprateformortalityandQOL.Fifth,theD1-modelisnottoocomplexandcanaidindecision-making

early in ICU stay. Sixth, the database held data concerning baseline condition and QOL, which is of

importance in outcome studies and in developing objective prediction models, but still is exceptionally

assessed[6].ThehighimpactofUIbonUI1yillustratestherequirementofknowledgeofbaselinecondition

tomakeanypredictiononoutcomeatlong-term.Seventh,weusedagroupedlassotechnique,whichisan

objective selectionandshrinkageestimationmethod for linear regressionmodels [34,35].Wepreferred

this technique above the widely used stepwise selection method – where prediction accuracy only

143

improveswhencovariateshaveastrongrelationshipwiththeoutcome-toselecttheoptimalnumberand

mostimportantpredictorsforUI1yintheD1linearregressionmodelinordertosimplifythemodel,andto

copewiththecategoricalvariables.

Ourstudyalsohassomelimitations.First,theD1-modelwasdevelopedbaseduponasingle-center

dataset. Second, the model was not externally validated, nor was it validated into clinical practice.

Implementation studies are needed to investigate the added value of our model in decision-making

comparedtoclinicalexpertisealone[24].Third,themodelcouldonlyexplain40%ofvariabilityofUI1y.This

couldbeconsideredasnotaccuratelyenough.However,atthismoment,itshouldbeseenasauniquehelp

ininformingpatientsandfamilies,indecision-makingandinadvancedcareplanning.

CONCLUSION

WedevelopedaneasytousepredictionmodelforthemeanQOLat1yearafter ICUdischargein

generalcriticallyillpatientsbasedupondatareadilyavailableatthefirstICUday.Althoughonly40%ofthe

variability in long-termQOL could be explained, this predictionmodel can be a helpful tool in decision-

making,ingoodandinformativecommunicationtowardspatientsandfamilies,inresourceallocation,and

inadvancedcareplanning. Further research shouldnow focusonprospectiveandmulticentervalidation

andrefiningofthisQOLpredictionmodel.

ACKNOWLEDGEMENTS



callingandinterviewingpatients,whowereincludedintheoriginaldatabase.TheauthorsalsothankChris

Danneelsforhishelpinsettinguptheoriginaldatabase.

144

Table1.Demographics,ICUadmission,D1characteristics,andoutcomes*

1953cases ModelI ModelII ModelIII

Completecasesincluded,N(%) 1953(100%) 1867(95.6%) 1809(92.6) 1831(93.8)BaselineCharacteristicsMalegender,N(%) 1211(62.0) 1152(61.7) 1120(61.9) 1133(61.9)Age(years) 57.2±16.8 57.6±16.7 57.5±16.6 57.5±16.7BMI(kg/m2) 25.6±5.4 25.6±5.3 25.6±5.3 25.6±5.3Charlsonco-morbidityindex 2.5±2.7 2.5±2.7 2.5±2.7 2.5±2.7Previoushospitalizationinpast6months,N(%)

843(43.2) 813(43.5) 784(43.3) 794(43.4)

LivingathomebeforeICUadmission,N(%) 1891(96.8) 1808(96.8) 1754(97.0) 1773(96.8)ADLatbaseline,N(%)NolimitationsModeratelimitationsChairboundBedridden

1162(59.5) 1099(58.9) 1080(59.7) 1089(59.7)625(32.0) 609(32.6) 576(31.8) 587(32.1)96(4.9) 94(5.0) 91(5.0) 92(5.0)70(3.6) 65(3.5) 62(3.4) 63(3.4)

UIb 0.62±0.33(a) 0.62±0.33 0.63±0.33 0.62±0.33

VASb 65.6±20.0(b) 65.7±19.9 65.8±19.9 65.7±19.9ICUadmissioncharacteristicsICUadmissionduringweekend,N(%)

564(28.9) 535(28.7) 512(28.3) 522(28.5)

ICUadmissionunplanned,N(%)

1430(73.2) 1364(73.1) 1318(72.9) 1333(72.8)

HospitaldayspriorICUadmission(days) 3.1±14.0 2.9±11.7 2.7±9.8 2.7±9.8ICU-D1characteristicsAPACHEII 16.9±8.2(c) 17.0±8.2 16.9±8.1 16.9±8.1SOFAscore 4.6±3.8 4.6±3.8 4.6±3.7 4.6±3.8Needformechanicalventilation,N(%) 606(31.0) 572(30.6) 557(30.8) 564(30.8)Needforvasopressortherapy,N(%) 390(20.0) 371(19.9) 361(20.0) 364(19.9)NeedforRRT,N(%) 43(2.2) 43(2.3) 39(2.2) 40(2.2)Needfortracheotomy,N(%) 35(1.8) 35(1.9) 34(1.9) 35(1.9)

OutcomesICU-LOS(days) 6.5±10.5 6.5±10.3 6.5±10.4 6.5±10.3

ICUmortality,N(%) 168(8.6) 160(8.6) 151(8.3) 152(8.3)Hospital-LOS(days) 29.3±42.4 29.0±40.7 28.7±40.4 28.6±40.3Hospitalmortality,(%) 285(14.6) 275(14.7) 259(14.3) 262(14.3)UI1y* 0.46±0.38(d) 0.46±0.38 0.47±0.38 0.46±0.381-yearmortality,N(%) 515(26.4) 504(27.0) 477(26.4) 483(26.4)D1=first24hoursofICUadmission;±=meanandstandarddeviation;ICU=intensivecareunit;N=number;BMI=bodymassindex;ADL=activitiesofdailyliving;UIb=utilityindexatbaseline;VASb=visualanaloguescaleatbaseline;APACHEII=AcutePhysiologyandChronicHealthEvaluationscore;SOFA=sequentialorganfailureassessment;RRT=renalreplacementtherapy;LOS=lengthofstay;UI1y=utilityindexat1yearafterICUdischarge;*=basedupon1953casesindatabaseunlessindicatedotherwise;(a)=28/1953missingdata(1.43%);(b)=39/1953missingdata(2.00%);(c)=5/1953missingdata(0.26%);(d)=72/1953missingdata(3.7%),*UI1yfornon-survivors=0

145

Table2.Descriptionofmissingness

Variable Numbermissing(N)(total1953cases)

Proportionmissing(%)

Numberofcaseswithatleast1variablemissing

144 7.37

UI1y 72 3.69

VASb 39 2.00

UIb 28 1.43

Suboncological 20 1.02

Subhematological 1 0.05

BMI 27 1.38

APACHEII 5 0.26

Baselinejob 24 1.23

D1.TISS-28score 1 0.05

D1.NEMS-score 1 0.05

D1.medicalimaging 1 0.05

D1.transfusion 1 0.05

N=number;UI1y=utilityindexat1yearafterICUdischarge;VASb=visualanaloguescaleatbaseline;UIb=utilityindexatbaseline;sub=predefinedsubgroupofaspecificpatientpopulation;BMI=bodymassindex;APACHEII=AcutePhysiologyandChronicHealthEvaluationscore;D1=describesvariableatD1(D1=first24hoursofICUadmission);TISS-28score=TherapeuticInterventionScoringSystem28-score;NEMS-score=NineEquivalentofNursingManpowerUsescore

146

Table3.All32possibleD1-variablestopredictUI1y

Variable Description

10continuousvariables

UIb,VASb,age,BMI,Charlsonco-morbidityindex,

hospitaldayspriorICUadmission,APACHEII,D1.SOFA,D1.TISS-28,D1.NEMS

16binaryvariables(only1dummypossibleforeachbinaryvariableintheD1-model:0/1*)

suboncological,subhematological,subcirrhosis,subelderly(≥80years),

gender, previous hospitalization in the past 6 months, admission during weekend,

admissionunplanned,D1.DNR,D1.MV,D1.VP,D1.RRT,D1.surgery,D1.medical imaging,

D1.tracheotomy,D1.transfusion

6categoricalvariables(morethan1dummyforeachcategoricalvariableintheD1-model)

livingsituationatbaseline(reference=1/athomewith2dummies:2/specialcarefacility;

3/other);ADL(reference=1/nolimitationswith3dummies:2/moderatelimitations,

3/chairbound,4/bedridden);originofhospitaladmission(reference=1/homewith5

dummies:2/emergencydepartment,3/otherhospital,4/psychiatricinstitution,5/special

carefacility,6/other);originofICUadmission(reference=1/emergencydepartmentwith

8dummies:2/hospitalward,3/high-careunit,4/coronarycareunit,5/operationtheatre,

6/catheterizationroom,7/recoveryroom,8/otherhospital,9/other);baselinework

(reference=1/studentwith5dummies:2/atwork,3/unemployed,4/housekeeping,

5/invalidity,6/retired);mainICUdiagnosis(reference=1/medicalwith3dummies:

2/surgical,3/burns,4/trauma)

D1=first24hoursof ICUadmission;UI1y=utility indexat1yearafter ICUdischarge;UIb=utility indexatbaseline;VASb=visualanalogue scale at baseline; BMI= body mass index; APACHE II= Acute Physiology and Chronic Health Evaluation score; D1.=describes variable at D1; SOFA= Sequential Organ Failure Assessment (SOFA) score; TISS-28= Therapeutic Intervention ScoringSystem28score;NEMS-score=NineEquivalentofNursingManpowerUsescore; sub=predefinedsubgroupofa specificpatientpopulation;DNR=do-not-resuscitatescore;MV=mechanicalventilation;VP=vasopressors;RRT=renalreplacementtherapy;ADL=activitiesofdailyliving;ICU=intensivecareunit;*0/1=eitherthevariableispresent(1)ornot(0)

147

Table4.Fittingofthe3differentD1-predictionmodelstopredictUI1y

Model Description NumberofD1-variablesIncluded

(N)

Numberofcompletecases

included(of1953cases)(N)

(%)*

R2 AdjustedR2 Rootofcross-validated

predictionerror

I Bivariateassociation

betweenUIb-UI1y

1 1867(95.6%) 0.2050 0.2050 NA

II Fullmodel 32 1809(92.6%) 0.3980 0.3800 0.3068

III Reducedmodel 16 1831(93.8%) 0.3875 0.3807 0.3026

D1= first 24 hours of ICU admission; UI1y= utility index at 1 year after ICU discharge; R2= proportion of explained variance;adjusted R2= proportion of explained variance, taking into account the number of variables; N= number; UIb= utility index atbaseline;NA=notapplicable;*=caseswithpartialinformation(=missingofatleast1variableinatleast1case)wereexcludedforthedevelopmentoftherespectivemodel

148

Table5.ModelIII:Multivariateregressionanalysis

D1variables Estimate SE

t-value

p-value

95%CI

VASb 0.0009 0.0004 1.956 0.051 -0.000to0.002UIb 0.3017 0.0325 9.277 <0.001 0.238to0.365Suboncological -0.1190 0.0232 -5.120 <0.001 -0.165to-0.073Subhematological -0.1077 0.0402 -2.679 0.007 -0.187to-0.029Subelderly(≥80yrs) -0.1035 0.0318 -3.259 0.001 -0.166to-0.041Age -0.0023 0.0005 -4.330 <0.001 -0.003to-0.001ADL,Reference=nolimitations

moderatelimitationschairboundbedridden

-0.0931-0.1794-0.1186

0.01980.03840.0456

-4.712-4.675-2.601

<0.001<0.0010.009

-0.132to-0.054-0.255to-0.104-0.021to-0.029

Charlsonco-morbidityindex -0.0067 0.0034 -1.969 0.049 -0.013to-0.000APACHEII -0.0047 0.0014 -3.289 0.001 -0.007to-0.002MainICUdiagnosis,Reference=medical

surgicalburns

trauma

0.11020.0346-0.0151

0.01720.04950.0302

6.4230.700-0.499

<0.0010.4840.618

0.076to0.144-0.063to0.132-0.074to0.044

D1.SOFA -0.0092 0.0035 -2.656 0.008 -0.016to-0.002D1.DNR -0.0728 0.0480 -1.517 0.129 -0.167to0.021D1.mechanicalventilation -0.0530 0.0192 -2.761 0.006 -0.091to-0.015D1.vasopressors -0.0329 0.0258 -1.273 0.203 -0.084to0.018D1.medicalimaging -0.0689 0.0191 -3.603 <0.001 -0.106to-0.031D1.tracheotomy -0.1238 0.0525 -2.360 0.018 -0.227to-0.021D1=first24hoursofICUadmission;SE=standarderror;CI=confidenceinterval;VASb=visualanaloguescaleatbaseline;UIb=utilityindexatbaseline;sub=predefinedsubgroupofaspecificpatientpopulation;ADL=activitiesofdailyliving;APACHEII=AcutePhysiologyandChronicHealthEvaluationscore;ICU=intensivecareunit;D1=describesvariableatD1;SOFA=SequentialOrganFailureAssessment(SOFA)score;DNR=do-not-resuscitatescore

149

Figure1.GroupedlassotechniquetoselectD1-variablesinModelIII

Description:X-axis(above):all32D1-variables;X-axis(under):logarithmofpenaltyparameterλY-axis:cross-validatedpredictionerror(reddots)witherror-bar(±standarderrorofthecross-validatedpredictionerror)Cross-section of X-axes and Y-axis (light grey dotted line) revealed that the lowest value of the cross-validated prediction error was reached when 24 of all 32 D1-variables were selected in the predictionmodel. Subsequently, the one-standard-error rule was applied in order to select theλ-valuewhere thecorresponding cross-validated prediction error is within 1 standard error of the optimal (lowest) cross-validatedpredictionerror.Thiswasdone toavoid toomanyD1-variables in thepredictionmodel.Cross-sectionofX-axesandY-axis (bluedotted line)afterapplyingof theone-standard-errorrulerevealedthattheoptimalnumberofD1-variablesinthepredictionmodelwas17outofall32D1-variables.D1=first24hoursofICUadmission;log=logarithm;λ=penaltyparameter

150

REFERENCES

1. LamasD.Chroniccriticalillness.NewEngJMed2014;370:175-72. NeedhamDM,DavidsonJ,CohenH,HopkinsRO,WeinertC,WunschH,etal.Improvinglong-termoutcomesafterdischargefromintensivecareunit:Reportfromastakeholders’conference.CritCareMed2012;40:502-93. HashemMD, Nallagangula A, Nalamalapu S, Nunna K, Nausran U, Robinson KA, et al. Patient outcomes aftercriticalillness:asystematicreviewofqualitativestudiesfollowinghospitaldischarge.CritCare2016;20:3454. NormanBC,JacksonJC,GravesJA,GirardTD,PandharipandePP,BrummelNE,etal.Employmentoutcomesaftercritical illness:Ananalysisof thebringing to light therisk factorsand incidenceofneuropsychologicaldysfunction in ICUsurvivorscohort.CritCareMed2016,44:2003-95. SimpkinAL,SchwartzsteinRM.Toleratinguncertainty–thenextmedicalrevolution?NewEnglJMed2016;375:1713-56. Oeyen SG, Vandijck DM, Benoit DD, Annemans L, Decruyenaere JM. Quality of life after intensive care: asystematicreviewoftheliterature.CritCareMed2010;38:2386-4007. CarsonSS,KahnJM,HoughCL,SeeleyEJ,WhiteDB,DouglasIS,etal.Amulticentermortalitypredictionmodelforpatientsreceivingprolongedmechanicalventilation.CritCareMed2012;40:1171-68. SinuffT,AdhikariNK,CookDJ,SchünemannHJ,GriffithLE,RockerG,etal.Mortalitypredictionsintheintensivecareunit:comparingphysicianswithscoringsystems.CritCareMed2006;34:878-859. KnausWA,DraperEA,WagnerDP,ZimmermanJE.APACHEII:aseverityofdiseaseclassificationsystem.CritCareMed1985;13:818-2910. ZimmermanJE,KramerAA,McNairDS,MalilaFM.AcutePhysiologyandChronicHealthEvaluation(APACHEIV):hospitalmortalityassessmentfortoday’scriticallyillpatients.CritCareMed2006;34:1279-31011. LeGallJR,LemeshowS,SaulnierF.ANewSimplifiedAcutePhysiologyScore(SAPSII)basedonaEuropean/NorthAmericanmulticenterstudy.JAMA1993;270:2957-6312. MorenoRP,MetnizPG,AlmeidaE,JordanB,BauerP,CamposRA,etal.SAPS3–Fromevaluationofthepatienttoevaluationofhe intensive careunit.Part2:Developmentofaprognosticmodel forhospitalmortalityat ICUadmission.IntensiveCareMed2005;31:1345-5513. VincentJL,MorenoR,TakalaJ,WillattsS,DeMendonçaA,BruiningH,etal.TheSOFA(Sepsis-relatedOrganFailureAssessment)scoretodescribeorgandysfunction/failure.OnbehalfoftheWorkingGrouponSepsis-relatedProblemsoftheEuropeanSocietyofIntensiveCareMedicine.IntensiveCareMed1996;22:707-1014. Ferreira FL, Bota DP, Bross A,Mélot C, Vincent JL. Serial evaluation of the SOFA score to predict outcome incriticallyillpatients.JAMA2001;286:1754-815. JainA,PaltaS,SaroaR,PaltaA,SamaS,GombarS.Sequentialorganfailureassessmentscoringandpredictionofpatient’soutcomeinIntensiveCareUnitofatertiaryhospital.JAnaesthesiolClinPharmacol2016;32:364-816. HeylandDK,StelfoxHT,GarlandA,CookD,DodekP,KutsogiannisJ,etal.Predictingperformancestatus1yearaftercritical illness inpatients80yearsorolder:Developmentofamultivariableclinicalpredictionmodel.CritCareMed2016;44:1718-2617. Decruyenaere A, Decruyenaere P, Peeters P, Vermassen F, Dhaene T, Couckuyt I. Prediction of delayed graftfunctionafter kidney transplantation: comparisonbetween logistic regressionandmachine learningmethods.BMCMedInformDecisMak2015;15:8318. HarrisonDA,GriggsKA,PrabhuG,GomesM,LeckyFE,HutchinsonPJ,etal.Externalvalidationandrecalibrationof risk prediction models for acute traumatic brain injury among critically ill adult patients in the United Kingdom. JNeurotrauma2015;32:1522-3719. denBoerS,deKeizerNF,deJongeE.Performanceofprognosticmodelsincriticallyillcancerpatients.CritCare2005;9:R458

151

20. PeetersP,VanBiesenW,VeysN,LemahieuW,DeMoorB,DeMeesterJ.Externalvalidationofariskstratificationmodeltoassistshareddecisionmakingforpatientsstartingrenalreplacementtherapy.BMCNephrol2016;17:4121. Wassenaar A, van den Boogaard M, van Achterberg T, Slooter AJ, Kuiper MA, Hoogendoorn ME, et al.Multinational development and validationof an early predictionmodel for delirium in ICUpatients. IntensiveCareMed2015;41:1048-5622. ReidJM,GubitzGJ,DaiD,KyddD,EskesG,ReidyYetal.PredictingfunctionaloutcomeafterstrokebymodellingbaselineclinicalandCTvariables.AgeAgeing2010;39:360-623. MinneL,LudikhuizeJ,deJongeE,deRooijS,Abu-HannaA.PrognosticmodelsforpredictingmortalityinelderlyICUpatients:asystematicreview.IntensiveCareMed2011;37:1258-6824. Veerbeeck JM,KwakkelG, vanWegenEH,Ket JCF,HeymansMW.Earlypredictionofoutcomesof activitiesofdailylivingafterstroke:asystematicreview.Stroke2011;42:1482-825. Brinkman S, Abu-Hanna A, de Jonge E, de Keizer NF. Prediction of long-termmortality in ICU patients:modelvalidationandassessing theeffectofusing in-hospital versus long-termmortalityonbenchmarking. IntensiveCareMed2013;39:1925-3126. OeyenS,BenoitD,AnnemansL,DecruyenaereJ.Qualityoflifebefore,3months,and1yearafterICUdischarge.CritCareMed2010;38SupplDecember:18227. KatzS,DownsTD,CashHR,GrotzRC.ProgressindevelopmentoftheindexofADL.Gerontologist1970,10:20-3028. Charlson ME, Pompei P, Ales KL, Mackenzie CR. A new method of classifying prognostic comorbidity inlongitudinalstudies:developmentandvalidation.JChronicDis1987;40:373-8329. MirandaDRetal.SimplifiedTherapeuticInterventionScoringSystem:theTISS-28items-resultsfromamulticenterstudy.CritCareMed1996;24:64-7330. ReisMirandaD,MorenoR, IapichinoG.Nineequivalentsofnursingmanpowerusescore (NEMS). IntensiveCareMed1997;23:760-531. EuroQolGroup.EuroQol--anewfacilityforthemeasurementofhealth-relatedqualityoflife.HealthPolicy1990;16:199-20832. Angus DC, Carlet J, 2002 Brussels Roundtable Participants. Surviving intensive Care: A report from the 2002BrusselsRoundtable.IntensiveCareMed2003;29:368-7733. LievenAnnemans.Gezondsheidseconomievoorniet-economen.1sted.Ghent:AcademiaPress;200734. TibshiraniR.Regressionshrinkageandselectionviathelasso.JRStatistSocB1996;58:267-8835. YuanM,LinY.Modelselectionandestimationinregressionwithgroupedvariables.JRStatistSocB2006;68:49-6736. RFoundationforStatisticalComputing.R:a languageandenvironmentforstatisticalcomputing(version3.2.2),http://www.R-project.org/;2017[accessed13.07.17]37. Grpreg-package.https://cran.r-project.org/web/packages/grpreg/grpreg.pdf;2017[accessed13.07.17]38. KahnJM.Predictingoutcomeincriticalcare:past,present,andfuture.CurrOpinCritCare2014;20:542-339. FrickS,UehlingerDE,ZuercherZenklusenRM.Medicalfutility:Predictingoutcomeofintensivecareunitpatientsbynursesanddoctors–aprospectivecomparativestudy.CritCareMed2003;31:456-6140. Timmers TK, Verhofstad MH, Moons KGM, Leenen LP. Intensive care performance: How should we monitorperformanceinthefuture?WorldJCritCareMed2014;3:74-941. BlackN.Patientreportedoutcomemeasurescouldhelptransformhealthcare.BMJ2013;346:167

152

42. CoslovskyM, TakalaJ, ExadaktylosAE,Martinolli L,Merz TM.A clinical predictionmodel to identify patients at

highriskofdeathintheemergencydepartment.IntensiveCareMed2015,41:1029-36

43. PutmanMS,TakHJ,CurlinFA,YoonJD.Qualityoflifeandrecommendationsforfurthercare.CritCareMed2016;

44:1996-2002

153

PartThree

OverviewoftheThesis

155

I.Conciseoverviewofthestudyresults

1.Inclusions

Inour reviewstudy,we includeda totalof53articles.Therewere4studiesconcerningoutcome

andQOLingeneralcriticallyillpatientsoneyearafterintensivecareand6withlongerfollow-upperiods.

The other articles were grouped according to diagnostic category: acute respiratory distress syndrome

(ARDS) (N=11), prolonged mechanical ventilation (N=3), trauma (8), cardiac arrest (N=6), older patients

(N=6),pancreatitis(N=2),sepsis(N=3),andstudieswithvarioustopics(N=4).Hugevariationswerefoundin

usedQOL instruments, in timingandmethod for long-termQOLassessments,and in final response rate.

Only4ofallthe53includedstudies(8%)metallofthe4predefinedstudyqualitycriteria;assessmentof

QOLatbaseline,nomajorexclusioncriteria,descriptionofthenon-respondergroupversustheresponder

group, and comparison with an age-and gendermatched normal population. All studies defined clearly

whichpatientswerein-orexcludedbutonly9studies(17%)measuredQOLpriortoICU.In our second study, 483 cancer patients (398 oncological and 85 hematological patients) were

included.Patientswithhematologicalmalignancieshadsignificanthigherco-morbidities,significanthigher

severityofillnessatadmission,requiredsignificantmoreorgansupportduringICUstayandhadsignificant

longer ICU and hospital stays although their disease status was significant more under control or in

remissioncomparedtosolidtumorpatients.

Inourthirdstudy,wefoundthat147patients(7.5%) inthetotalCOSIcohortdevelopedAKIwith

needforRRT.Ofthese,26AKIpatients(1.3%)didnotreceiveRRTduetotherapeuticrestrictionsandwere

excluded for further analysis; the other 121 patients (6.2%) received RRT. Forty-seven 1-year AKI-RRT

survivorswereindividuallymatchedwith941-yearnon-AKI-RRTsurvivors,and284-yearAKI-RRTsurvivors

were individually matched with 28 non-AKI-RRT patients. During ICU stay, 1-year and 4-year AKI-RRT

patientsweremoreseverelyillcomparedtotheirrespectivematches.

In our fourth study concerning patients aged 80ormore,we included131patients (60%males)

withmedianageof83years (IQR81-85)andCharlsoncomorbidity indexof2 (IQR0-4).Reasons for ICU

admission were mainly medical (55%) or postoperative after emergency surgery (23%). Fewer older

patientswere admitted after elective surgery (12%), trauma (9%), or burns (1%) Therapeutic limitations

weresetin34patients(26%)after2days(IQR1-5)attheICU.

Inourfifthstudy,theCOSIdatabasewasusedfordevelopmentofapredictionmodelforthemean

QOLat1yearafter ICUdischarge ingeneralcritically illpatientsbasedupondatareadilyavailableatthe

firstICUday.Respectively1867(95.6%),1809(92.6%),and1831(93.8%)ofthe1953caseswerecomplete

and included for development of respectively models I, II, and III. We fitted these 3 different linear

regression models and compared their performance towards prediction accuracy and usability. We

preferredthereducedModelIII,whichheldonly16ofthemostimportantandpowerfulD1-variables,for

156

predictionofUI1y.

2.Mortality

Wemeasuredmortalityatbaseline(ICUandhospitalmortality),andat3monthsand1yearafter

ICU discharge. In the COSI cohort, in the AKI-RRT patients, and in the patients aged ≥80 years, we also

assessedmortalityatlonger-term,respectivelyatabout9years(*),4years(**)and7years(***)afterICU

discharge.Highmortalityrateswerefoundinallthecriticallyillpatientswestudied.

Cohort

Mortality

AllCOSIpatients

Oncologicalpatients

Hematologicalpatients

AKI-RRTpatients

Olderpatients

N=1953 N=398 N=85 N=121 N=131

ICU(%) 9 5 21 46 17

hospital(%) 14 13 34 55 29

3months(%) 17 17 42 58 39

1year(%) 26 36 66 61 50

long-term(%) 48(*) - - 71(**) 84(***)

3.Qualityoflife

Inourreviewarticle,wefoundthatoneyearafterICU,criticallyillpatientsingeneralhadalower

QOL, especially in physical domains, than an age-and gender matched population. However, a slow

improvementtopre-morbidQOLlevelscouldbefound.ParticularlyARDSpatients,patientsafterprolonged

mechanical ventilation, severe trauma patients, and sepsis survivors showed significant impairments in

long-termQOL.Whilephysicalaspectsimprovedslowlyovertheyears,mentalandemotionalimpairments

werestagnantordeclinedevenfurther.Inolderpatients,QOLwassomewhatdecreased,especiallyinthe

physicaldomains,butthesepatientsgenerallyadaptedwelltotheselimitationsandperceivedtheirQOLas

good.

Inoursecondstudy,QOLassessmentsshowedthatforbothoncologicalandhematologicalpatient

groups long-termQOLwas lower than that of a general population.QOL decreased 3months after ICU

dischargecomparedtobaseline,improvedafter1year,especiallythementaldomains,butremainedunder

thebaselinelevel.Atanymoment,QOLwasespeciallylowerinpatientswithhematologicalmalignancies.

157

Among the one-year survivors, patients with hematological malignancies were also less likely to live

independentlywithoutadditionalhelpandmorewouldrefuseICUreadmissionagain.

In our third study, we found that differences in QOL between AKI-RRT and their non-AKI-RRT

matchesateachdifferenttimepointwereverysmall.EvolutioninQOLovertimeforthe1-yearand4-year

AKI-RRT patients showed that most problems in QOL were seen at 3 months after ICU discharge,

particularlyintheAKI-RRTgroup.QOLimprovedafter1year,especiallyinthementaldomains,butwithout

return to the baseline level. At 4 years, QOL significantly decreased mainly physically but improved or

remainedthesameinthementalcomponents.Thesamepattern,althoughlesspronounced,wasseenin

the1-yearand4-yearnon-AKI-RRTpatients.Overall,long-termQOLremainedunderthebaselinelevelfor

AKI-RRTandnon-AKI-RRTpatients,andundertheQOLof theaveragepopulationspecifically in themore

physical domains. QOL was however perceived as acceptable. Both AKI-RRT and non-AKI-RRT patients

reportedlowdependenceindailylifelateron.ThemajorityofAKI-RRTpatientswantedtobereadmittedto

theICUwhenneeded,despitethefactthatonequarterhadpersistentdialysisdependency.

Inourfourthstudy,wesawthatthenumberofolderpatientswithproblemsinmobility,self-care,

usualactivities,andanxiety/depressionsignificantlyincreasedateachoftheconsecutivetimepoints.QOL

decreased3monthsafterICUdischargecomparedtobaseline,improvedafter1year,especiallymentally,

butworsened again after 7 years. Long-termQOL remained under baseline level and underQOL of the

generalpopulation.PerceiveddeteriorationinQOLwasseenafter3months,whichhoweverchangedinto

aperceptionofnochangeorevenbetterafter1yearandagainaperceptionofworseninginmostpatients

after7years,mainlyinthedimensionsofmobilityandself-care.Allbut1ofthe7-yearssurvivorsreported

a very good familial and social network, a good paramedical and medical follow-up, experienced no

financialproblems,andwerehappytobestillalivedespitetheiradvancedage.Amongthe1-yearand7-

yearssurvivorsrespectively,37%and11%livedindependentlyathome,26%and28%hadadditionalhome

help,13%and22%livedwithrelatives,and21%and39%livedinaspecialcarefacility.Themajorityofthe

long-term older survivors expressed a preference to be readmitted to an ICU department in case of

deterioration.

4.Factorswithimpactonlong-termqualityoflife

Althoughitwasnotthemaintargetthroughourresearch,wealsotriedtodeterminefactorswith

impactonlong-termQOL.Inourreviewarticle,resultsconcerninginfluenceofthepatients’characteristics

and illnessupon long-termQOLwereconflicting. Itwasdifficult towithholdcertain factors impactingon

long-term QOL due to different study designs, methodologies, patient populations, applied QOL

instruments, follow-upperiods,andresponserates throughthe includedarticles.Wefoundthat inARDS

patientsorpatientswithprolongedmechanicalventilation, theARDSand itssequelae influencedQOLby

158

impairmentsinpulmonaryfunctions,cognitivedisorders,weakness,andposttraumaticstressdisorders.In

traumapatients,theinjuryseverity,thedegreeofbraindamage,andfemalegenderdominatedlong-term

QOL in a negative way. However, in other studies the severity of illness played a less important role.

Medicalornon-scheduledsurgicalpatients,olderage,andapoorpre-admissionQOLhadalsoanegative

impactonlong-termQOL.

Inourstudyconcerningoncologicalandhematologicalpatients,wespecificallysearchedforfactors

with impactonQOL.Beingadmittedto the ICUforamedicalorsurgical reason,orcancerstatushadno

influence on long-termQOL.Multivariate regression analysis showed however that poor QOL 3months

after ICU discharge was independently associated with female gender (p<0.001), higher comorbidity

(p=0.001),hematologicalmalignancy(p=0.010),olderage(p=0.030),andahighermeanSOFAscoreduring

ICUstay (p=0.040).QOL1yearafter ICUdischargewasstillnegatively influencedbyolderage (p=0.007),

highercomorbidity(p=0.035),andhematologicalmalignancy(p=0.041).

These factors alsoplayedan important role inourD1-predictionmodel formeanQOLat1 year.

BaselineQOLandbaselineVASappeared tobe stronglypositively relatedwith long-termQOL.Variables

negatively related with mean QOL at 1 year were an oncological or hematological disease, older age,

limitations in ADL, higher APACHE II score, organ failure with need for mechanical ventilation or

vasopressors, and a high comorbidity. We also found that the predicted UI1y for surgical patients was

significantlyhigherversusmedicalpatients,whichwasincontrasttoburnortraumapatients,forwhomwe

couldnotdemonstrateanysignificantdifferenceinUI1yversusmedicalpatients.

159

II.Generaldiscussion

Thefocusofourresearchconcentratedaround3majorissuesresultinginasystematicreviewand

4original studies:1/ reviewing literatureconcerning long-termQOL, reviewingappliedmethodologyand

qualityofthispublishedoutcomeresearch,2/assessinglong-termoutcomesandQOLinspecificcriticallyill

patientpopulations(oncological-hematological,AKI-RRTandolderpatients(≥80years)),and3/developing

a prediction model for long-term QOL based upon readily available variables at the first day of ICU

admissionandsodeterminingthemostimportantpredictorsforlong-termQOL.

At first,weevaluatedwhatwasalreadyknownconcerning long-termQOL incritically illpatients.

We found huge variations in appliedmethodology resulting in a rather poor overall quality of outcome

research,whichhamperedtheabilitytocompareresultsordrawstrongconclusionsoutofthisresearch.

Thisproblemwasalreadyunderlinedsomedecadesago[1,2].Recently,manyprofessionalandscientific

organizationshaveprioritizedoutcomeresearchonsurvivorsofcriticalillnessafterhospitaldischargeand

peer-reviewedpublicationsreportingonthesepatientoutcomesgrewfrom3in1970uptonearly500just

now[3].However,thereisstillnoconsensusonthemostimportantoutcomes,measurementinstruments

forassessments,andtimingoftheseassessments[4].

So, within critical care medicine thus far, there has been little critical evaluation of outcome

measuresused in clinicaloutcome research.Thispartly reflects the largenumberofmeasures thathave

been used in critical care research in the past and partly the poor quality of this research. Our

recommendation, therefore, is that the research community should agree on a limited list ofmeasures

fromwhich to select for any given project and a common time point for follow-up. Thiswould at least

enableaconsiderablebodyofexperienceandknowledgetobebuiltuparounda fewmeasures [4,5]. It

would also allow investigators to make comparisons between studies, facilitate overviews of published

resultsandenablephysicianstodrawconclusionsoutofthegrowingnumberofstudiesinthisfield[3,4,

6].

Lately, more attention has been paid to this problem and there are some projects within

internationalsocietiesfocusingontheneedforstandardizeddefinitionsofappropriateandvalidoutcome

measures, standardized timing of outcome assessments, minimizing loss to follow-up, and appropriate

statisticalmethods[6].

AsQOL isapatient-centeredandsubjectiveoutcomeparameterby itself,webelievethat

theuseofvalidated tools toassessQOL isanabsolute“must”. Incritical careoutcomes researchmainly

genericQOLmeasuresarebeingused.Inourreviewarticle,wechosetoincludeonlystudiesassessingQOL

bySF-36,RAND-36,EQ-5D,andNHPbecausethesearegenericinstrumentscommonlyusedincriticalcare

research; they are validatedandhavepopulationnorms in the literature.Although thesequestionnaires

160

haveawell-knownvalidity, reliability, andare responsive to changes inhealth [5], theyhave substantial

gapsintheircoverageofthesurvivors’QOL[7].ForexampletheEQ-5DandSF-36v2®,whicharethemost

commonly used QOL measures, and which we used throughout our research, do not assess memory,

concentration,theabilitytocompletetasks,multi-tasking,problemsolving,ordecision-making[8,9].The

dimensionof“usualactivities“of theEQ-5D isverybroaddefinedandmighteventually includecognitive

problemsalthoughthismightnotbeclearlyinterpretedbypatients.However,cognitivefunctionstogether

withphysicalandmentalfunctionsarethethreemainplayersindetermininglong-termoutcomes[10].

As we considered evaluation and evolution of cognition in the critically ill patient to be very

important,weaddedanextra6thdimension“cognition”tothefirstpartoftheEQ-5D,whichhas,equalto

theEQ-5D,3 levelsofproblems.This sixthdimension ishowevernot incorporated intocalculationofUI.

This“EQ-6D” is in factanextendedformoftheEQ-5DandwasdevelopedwithinthescopeoftheDutch

DisabilityWeights Study, which was carried out to obtain disease-specific preference weights for many

diseases[11].TheexpertgroupofthestudyproposedtoextendtheEQ-5Dwithacognitivedimensionto

capturecognitivedysfunction.TheEQ-6Dconstructvaliditywasexaminedwithgoodresults[12].TheEQ-

6Dishoweverfarlesswellknownandconsequently,itsuseisratherlimited.TheEQ-6Disparticularlyused

in theNetherlands for outcome research in a Dutch patient population [12-16]. During analyzing of our

studyresults,wethereforepreferredtheuseofEQ-5DandSF-36v2®,whicharebothcommonlyusedand

verywell-knownstandardizedQOLquestionnairesinoutcomeresearch.Weconsideredtheextraquestion

regardingcognitionasabonustogainmorecompleteinformationaboutthehealthstatusofthepatient.

Both questionnaires also do not address sexual functioning, social support, family and marital

functioning,placeofresidence,livingsituation,finances,problemstoreturntowork,sleepquality,health

distress,andmanyotherissuessuchaschangesinappearance,problemswithclothingduetoweightloss,

relationship toothers, etc.All thesephysical andpsychophysiological symptoms couldheavily impacton

QOL [7]. To overcome somewhat these shortcomings, we added in our research 4 additional short

questionsatlong-term(regardinglivingsituation,memoriesoftheICUstay,sleepqualityandpreferences

tobereadmittedtoanICU), inanattempttoovercomepartlyandeasilythesegaps.Weareunawareof

measures to specifically assess cognitive function except for the Informant Questionnaire on Cognitive

Decline(IQCODE).Thisquestionnairehastobeansweredbythenextofkinandassessesactualcognitive

functioning of the older patient comparedwith cognition 10 years ago. It is a very frequently used and

validatedquestionnaireingeriatricsbutinthegeneralcriticallyillsettingithasnotbeenusedbefore[17].

It is difficult to select the most appropriate survey(s), both in number and in content. All have

shortcomingsand it is important to selectdependingon the researchquestion, the researchpopulation,

and timing of the survey. The advantage of the EQ-5D is that it is a very short survey, which has

nevertheless the possibility to gain a lot of information. However, due to its shortness, it is less

161

discriminative than the SF-36v2®, which is very well validated in critically ill patients, and may be

considered as the first choice for QOL assessment in this patient group. Therefore, we believe that the

combination of SF-36v2®with EQ-5D yields themost to assess baseline QOL and QOL shortly after ICU

discharge:a lotofdiscriminativeQOL informationcombinedwithapreference-basedQOLmeasurewith

thepossibilityofanindexvaluetobeusedinhealtheconomicsstudies.

TimingofQOLassessmentswillalsoinfluencethechoiceandnumberofmeasures.Atbaseline,too

many questionnaires will tire the critically ill patient or the family, and will increase the probability for

incompletesurveysanddecreasetheprobabilityforfurtherparticipationinthestudy.Atlonger-term,after

aperiodofsomerecovery,itwillbeeasierformostpatientstocompletequestionnaires.Theseissuesmust

bebalancedtoensurethatsufficientandmeaningfuldataarecollectedatappropriatetimepoints,without

overburdening patients, family or researchers. A clear explanation of why, when, and how QOL

assessmentswill bemadeandwhatwill happen to thedatapatientsprovide,will help in keeping study

participantsmotivated.

Patients are often unable tomake a clear distinction between normal disease-specific processes

andconsequencesofbeinginacriticalcaredepartment[7].Therefore,tohaveamorecompletepictureof

outcomes and QOL at long-term, when the critical illness has been past for a while, we can now

recommendaddingadditionalvalidatedquestionnairestothegenericQOLquestionnairessuchasthePost-

traumatic Stress Syndrome 14-questions inventory (PTSS-14), the Hospital Anxiety and Depression Scale

(HADS),andtheMontrealCognitiveAssessmenttest(MoCA).ThePTSS-14isa14-itemscreeningtoolthat

has been validated in ICU patients [18, 19] and has a high sensitivity (86%) and specificity (97%) for

diagnosisofpost-traumaticstressdisorder(PTSD).ThePTSS-14isshort(5to10minutestocomplete),can

beeasilyusedinanoutpatientsettingoroverthetelephoneanddoesnotovertirepatients.TheHADSisa

reliable and valid instrument for detecting the presence and for measuring severity of

depressionandanxietyinthesettingofahospitalmedicaloutpatientclinic,inpsychiatriccases,inprimary

carepatientsandinthegeneralpopulation[20,21].TheMoCAtestisavalidatedone-page30-pointtest,

which can be administered in approximately 10 minutes. It assesses several cognitive domains such as

short-term memory, visual-spatial abilities, executive functions, attention, concentration and working

memory,language,fluencyandorientationtotimeandplace[22].

Whencombiningallthesemeasures,itshouldbepossibletoassessamorecompletepictureofthe

physical,mentalandcognitivefunctioningofthecriticallyillsurvivorandtomakeabetteradvancedcare

plan.

WhereQOLisasubjectiveoutcomeparameter,whichcanbedifficult,time-consumingandlabor-

intensivetoassess,death is,onthecontrary,aneasytodetermineandunequivocalendpoint.Thereare

severalpointsintimeatwhichtomeasureit:ICU,orhospitalmortality,timeuntildeath,ordeathatafixed

162

timepoint.Wemeasuredmortalityatbaseline (ICUandhospitalmortality), andat3monthsand1year

afterICUdischarge.IntheolderandAKI-RRTgroup,wealsoassessedmortalityatlonger-term.Wefound

highmortalityratesinallgroupsofcriticallyillpatientswestudied,especiallyinthefirst3monthssinceICU

admission,withonlymoderateincreaseofmortalityatlongerfollowup.Thesemortalityratesarehowever

comparablewiththenumbersfoundinliterature[23-31].Practicepatternssuchasadmissionpolicybefore

ICU, therapeutic restrictions during ICU, discharge policy and destination, and case-mix of patientsmay

haveimpactontheinterpretationofthesemortalityrates.Asatertiarycarefacility,thechanceofreceiving

complex and high-risk patients transferred from other hospitals is high, which can attribute to the high

mortality rates. Although there is an actual trend for a significant decrease in short-and long-term

mortality,itisalsoknownthatICUsurvivorshaveanongoingincreasedriskofmortalitymuchbeyondICU

discharge,whencomparedtoamatchedgeneralpopulation [32-35]. Ingeneral, ICUpatients reacha life

expectancysimilartothatofthegeneralpopulation2yearsafterICUadmission[34,35].

Themeasuresoflong-termQOLmayputsurvivingacriticalillnessintoalargerperspective.When

makingaglobalconclusionconcerninglong-termQOL,wefoundthatcriticallyillpatientshadalowerlong-

termQOLthanageneralpopulation,butaslowimprovementinQOLcouldbeseen,althoughitremained

under baseline level. Several years after ICU, QOL was quite comparable with that of the normal

population. In our review study, we found that patients with severe ARDS, prolonged mechanical

ventilation,severetrauma,andseveresepsisappearedtohavetheworstreductionsinQOL,whichlasted

alsoforalongtime.Theimpactofdiagnosticcategoryuponlong-termQOLwasalsopartlyreflectedinour

predictionmodel. Wesawthatthepredictedlong-termQOLforsurgicalpatientswassignificantlyhigher

comparedtomedicalpatients.

Being a hematological, oncological, AKI-RRT, or older patient certainly impacted on outcome.

Evidently,cancerpatients,AKI-RRTpatientsorolderpatientsadmittedtothe ICUrepresentednotonlya

highlydiversespectrumofdiseasesbutalsopatientswithaveryheterogeneousperformancestatusand

co-morbidityatbaseline.Assuch,outcomesshouldbedifferentiatedamongthesesubgroups.

Wefoundimportantdifferencesbetweensolidtumorpatientsandhematologicalpatientsrelative

to co-morbidity, reason for ICUadmission, and severityof illness. These translated into adifferent long-

termQOL in survivors, with hematological patients having aworse QOL on everymoment of the study

period, and experiencing no significant improvements beyond 1 year. Recent outcome studies in the

criticallyillcancerpatientstillfocusonmortality[24,36,37].OtherQOLstudiesinthegroupofcriticallyill

cancerpatients,beyondours,arevery scarcewhich is ratherbizarregiven thegrowingnumberof these

patients being admitted to the ICU combinedwith increasing short-term survival rates although overall

mortality remains high [23, 24, 37-39]. QOL assessments seem therefore of particular interest to

differentiateifthedyingprocessisbeingprolongedorifwecanguaranteeaqualityandmeaningfullifeat

163

longer-term[40,41].Azoulayetal.foundinahugestudyconcerningoutcomesincriticallyillhematological

patientsthatQOL,assessedbySF-36surveys,wasnotsignificantlydifferentfromage-andgendermatched

cancerpatientsnotadmittedtotheICU.OnlyaminorityofpatientsperceivedalterationsinQOL3months

afterICUdischarge.Inthisstudyhowever,onlyshort-termQOLwasassessedwithoutbaselineevaluation

andwitharesponserateofonly69%[37].Anotherrecentstudyfoundquitethesame[38].Thissuggests

that the critical illnessdoesnot impact thatmuchon long-termQOLand consequently, shouldnotbe a

reason not to transfer these patients to the ICU [23]. Very recently, the study by Normilio-Silva et al.

confirmedourdata[39].Theydemonstratedinamixedcriticallyillcancerpopulation-withpredominantly

oncological patients - that QOL in patients with a good baseline status decreased directly after ICU

admission, and then gradually increased but never returned to baseline level. However, patientswith a

poorbaselineconditionandQOLsteadilyimprovedover18monthsreachingamoderateQOL.

WealsocomparedQOLofAKI-RRTpatientswiththatofmatchednon-AKI-RRTpatients,andfound

verysimilarmeasurementsinbothgroups.This impliedthattheRRTcomponentduringcritical illnessdid

nothaveanimportantimpactonlong-termQOL.QOLwashoweverlowerthaninthegeneralpopulation.

IncontrasttotheextensiveliteratureonepidemiologyandRRTmodalities,thereisstillapaucityof

literatureonQOLandlong-termoutcomeincriticallyillpatientswhosurviveanepisodeofAKI-RRT[42-44].

However, these patients are some of the most severely ill patients in the ICU were prognosis, survival

estimation, and starting or withholding RRT is frequently a matter of difficult clinical decision-making,

taking also into account the high costs of RRT [45, 46]. Recently, some studies concerning QOL were

published, but only a minority reported on long-term QOL [25, 27, 47-51]. Consistent with outcome

research ingeneral, interpretationofstudyresultswaschallengingduetoheterogeneityofstudydesign,

QOLassessmenttools,case-mixofpatients,RRTmodalitiesanddurationoffollow-up.Nevertheless,overall

QOL data in these studies were very similar to ours with a QOL of AKI or AKI-RRT survivors that was

comparable with QOL of matched non-AKI or non-AKI-RRT patients but lower than in the general

population.QOLwasseldomassessedatbaselinebutoftenalreadyimpairedatthatmoment,consistentto

ourfindings[25,48,49].

A recentstudyshowedthatalthoughdevelopmentofAKIwasnotan independent risk factor for

increased 3-yearmortality in 30-day AKI-survivors, an episode of AKI-RRTmight portend long-term risks

such as evolution to chronic kidney disease (CKD), accelerated progression to end-stage kidney disease

(ESKD),chronicRRTdependencyormajorcardiovascularevents,whichallmayimpactheavilyonlong-term

outcomeandQOL[43,44,52].Wefoundthat19%ofthe1-yearAKI-RRTand29%ofthe4-yearAKI-RRT

survivors remained RRT dependent, which is an adverse outcome strongly associated with an ongoing

increasedriskofdeath[42].RatesofRRTdependencyafteranepisodeofAKI-RRTdifferamongpopulations

andcanvarybetween0%-40%[53].Patient-relatedfactorssuchasageandcomorbiditymayberiskfactors

164

fornon-recoveryofkidneyfunction,butalsotheseverityoftheAKIandoftheacute illness[53,54].The

impactofRRTmodalityonrenalrecoveryatlong-termremainscontroversialandwasnotoneofourstudy

endpoints [53].Whether there is a role for robustpathways tomonitor and screenAKI-RRT survivors to

improvetheselong-termoutcomeshasnotbeenformallystudiedalthoughpotentialfollow-upschemesdo

exist[43,53,55].

Determiningpatientswhoshouldbenefit themost from ICUadmissionbecomesmoreandmore

complicatedandthis isparticularly the fact inpatientsaged80yearsorolder.The long-termQOL in the

criticallyillolderpatientsinourstudywaslowcomparedtoageneralpopulation,particularlyinself-care,

usual activities and the physical domains, with an increasing number of patients experiencing more

problemsover time. This is in accordancewith data found in other recent studies concerning long-term

QOLinthe(very)oldpatient[56-60].TheseolderpatientshoweverrecognizedlittlechangesinQOLover

timeexceptformobilityandself-careatlong-term.Wefound,similaraswhatisdescribedinliterature,that

olderpatientsadaptedwelltotheiradvancedageandperceivedtheiroverallQOLasacceptable[58-64].It

suggests thatQOLmighthaveanothermeaning foroldpatients,with social andmental valuesbeing far

more important than limited physical functioning and that age itself influences QOL mainly due to

increasingnumberofchronicconditions[28,59,62,64,65].

A difference has indeed to be made between QOL measurements and perception of QOL as

experiencedbythepatienthimself,assessedbytheVAS.Oncologicalpatientshadabetterperceptionof

their QOL compared to hematological patients, but for both groups QOL was still acceptable. AKI-RRT

patientsperceivedQOLasgoodandbothAKI-RRTandnon-AKI-RRTpatientsreported lowdependence in

daily life lateron,whichwasalsofound inotherstudies[51].Thisperceptionofa fairQOLwasalsowell

illustratedbythefactthatthevastmajorityofallourincludedpatientswhowerealiveafter1yearoreven

longeransweredpositivetothequestionwhethertheywouldchoosetobereadmittedtoanICUincaseof

deterioration.AgoodperceptionofQOLdespitepersisting symptomsmaybeexplainedby the fact that

patientswhoareconfrontedwithalife-threateningdiseasearefacedwiththenecessitytoaccommodate

to the disease, which may lower internal standards. The divergence between mental and physical

performanceprobably reflects thisgradualprocess inwhichpatientsadapt toadiminishedperformance

status and come to accept their physical limitations. Acceptance of disability is in general higher among

olderpatients,andevenbetteriftheyhaveagoodsocioeconomicstatus[66].Indeed,theolderpatientsin

ourstudyexpressedpreferencesforalongerlife,evenwithreducedQOL,especiallywhentheyhadagood

socialnetwork.

Wenotonlydifferentiatedbetweenpatientswith abetter andaworseQOL,but alsomeasured

howQOLchangedovertimewithinacertainpatientgroup.Generally,QOLdecreased3monthsafterICU

discharge compared to baseline, improved after 1 year or longer, especially the mental domains, but

165

remainedunderthebaselinelevel.ThischangeinQOLovertimeleadstoanimportantanddifficultissuein

QOL studies. How long is “long” in long-term outcome and when will outcome measures and

questionnaires no longer give additional information? In all our included patients, mainly the physical

components deterioratedover time.While physical aspects improved slowly over the years,mental and

emotional impairmentswere rather stagnant. Our follow-up period of one yearwas probably too short

becausephysicallimitationsstilltendedtodominateoveremotionalproblemsandphysicalproblemswere

notalwaysrecovered.Oneyearmayalsobetooshorttobecomeaccustomedtomorerestrictionsindaily

live [38, 67]. The absence of any correlation between the physical and mental problems through our

researchisremarkable.ThismayhoweverbeexplainedbythefactthatICUsurvivorscanaccommodateto

the critical illness and its consequences leading to acceptance and adjustments to the disease [68].

Althoughwedonotdoubttheseobservations, itshouldbeunderlinedthatmentalorcognitiveproblems

bearahigherrisktoberemainedunrecognized.

The most important problem of long-term follow-up times is that more patients will be lost to

follow-up,whichcouldleadtoanimportantbiasinresults.Patientswhonotrespondcandosoforalotof

differentreasons.TheycanconsiderQOLquestionnairestrivialiftheyrecoveredwell,theycansufferfrom

posttraumaticstressdisorderavoidingseekingmemoriesoftheirICUtreatment,theycanbetooilltohave

the ability to respond, or they may have died before completing the survey [69-71]. As such, QOL

respondersmay representa sampleofhealthierpatients. Selectionbiasmayalsobe inducedbefore ICU

admission.PatientswhoarereferredtotheICUmightalreadyrepresentaselectionoffitterpatientswitha

possible inherent better prognosis and QOL. This was probably seen in our study concerning long-term

outcome in older patients ofwhom only aminoritywas chair-bound or bedridden at baseline.We also

cannotrulethisoutinthestudyevaluatingoncologicalandhematologicalpatients.Thislimitationishardly

avoidableandcanalsobefoundinotherstudiesconcerningolderorcancerpatients[28,41,61-63,72].

Anyway,toavoidselectionbias in long-termQOLdata,everyefforthastobemadetotargetthe

highest possible response rate. Otherwise, analyses of responders versus non-responders concerning

severityofillnessscores,co-morbidities,mortality,orageshouldbemade[73].Alosttofollow-upof20%

isconsideredtobeacceptableforQOLstudies[74],butmorethanhalfofthestudiesinourreviewarticle

didnotattaintothis.ToassessQOL1yearafterICUdischargeintheCOSIstudy,andalsoatlongertermin

theAKI-RRTandelderlystudy,wephonedallpatientswhodidnotrespondtotheinitialmailedsurveyafter

one month, although it was time-consuming and labour-intensive. This finally resulted in a very high

responserate(97.7%)andavery lownumberofpatients-only18outof1953patients inthetotalCOSI

cohort–,whichwerelosttofollow-up.Becauseoftheconsecutiveandprospectiveenrollmentofpatients

intheCOSIstudyandthehighlong-termfollow-uprateformortalityandQOL,wetriedtoreduceanyform

ofselectionbiastoanabsoluteminimum.

166

Besidestheriskofselectionbias,survivalbiasremainsaprobleminoutcomeresearch.Correction

for patients,who died during the observational period,was not necessarily in our study concerning the

impact of RRT on long-term outcome sincewe only included long-term survivors in the analysis. In the

studiesregardingcancerandolderpatients,itislikelythatonlythe“best”orthe“fittest”patientssurvived

long-term.Wecannotchangethis fact.AsQOLat long-termcanonlybemeasured insurvivors, inwhom

you may assume that overall QOL will be better that in nonsurvivors, QOL at long-term may be

overestimated. To correct for patients who died during the total observational period, QOL may be

indicated as “zero” for the nonsurvivors in the study cohort, which however will underestimate the

observedQOLofthesurvivorsinthecohort.WhendevelopingtheD1-predictionmodelwegaveQOLat1

yeara“zero”inputfor1-yearnonsurvivors.Thisallowedforcomparisonsbetweenthesamepatientcohort

atbaselineandat1yearafterICUdischargeandavoidedthatlong-termpredictionofQOLonlywouldbe

developedupondataofsurvivors.

Although survivors of critical illness share the common experience of coming extremely close to

deathastheysurvivealife-threateningillness,theycandifferfromoneanotherinmanywayssuchastheir

health statusbefore the illness, the specific eventordisease triggering the illness, their reactions to the

illness, and their capacity to recover. Another problem in interpretation and comparison of long-term

outcomeincriticallyillpatientsisthattheperiodofcriticalillnessisonlyasmallpartofthewholeillness

episodeandtherefore, thewholeprocessof illnessandcareshould in factbescrutinized: ICUadmission

policy, level of care during the ICU stay, end-of-life (EOL) decision, ICUdischarge policy, further hospital

stay,andpost-hospitalaftercare.Possibleconfounders,whichcouldinfluenceQOL,shouldbeeliminated.

Therefore, QOL in ICU patients can be compared to an age- and gendermatched general

population,whichshouldbeconsideredastheupperlimitsofwhatisachievable.Inallouroriginalstudies,

we thereforeused thenorm-based scoresof the SF-36v2®,which allowed fordirect comparisonswith a

generalhealthypopulation.Moreimportant,long-termQOLshouldalsobecomparedwithQOLbeforeICU

admission, to discriminate whether poor long-term QOL is a result of the severity of illness, or due to

confounding factors such as co-morbid disease, poor pre-admission QOL, age, gender, or acquired

complications.

Ourresearchwasobservational,solookingforcausesorexplanationsforlong-termQOLisdifficult.

However,wetriedtodeterminethemostimportantpredictors,besidesdiagnosticcategory,forlong-term

QOL.AlthoughbaselineQOLcanbeviewedasaanimportantpredictoroflong-termQOL,only17%ofthe

included studies in our review article measured QOL prior to ICU. In more recent outcome research,

measurementofbaselineQOL isstill rarelydone[25,39,48,49,58,75].PriorstudiesofQOLbefore ICU

admission support the hypothesis that patients’ premorbid QOL has a large effect on QOL after critical

illness[39,76].Ithasbeenprovedthatpre-ICUQOLislowcomparedtothegeneralpopulationindicating

167

that ICUpatients differ from the average population even before onset of critical illness [73]. PoorQOL

beforecritical illness isalsocorrelatedwithpooroutcome [74,76-78]. ImpairedQOLafter ICUmay thus

reflectapoorbaselinesituationratherthanbeafunctionofintensivecare[74,76].Wefoundaveryhigh

impactofbaselineQOLon long-termQOL inourD1-predictionmodel.This illustratestherequirementof

knowledgeofbaselineconditiontomakeanypredictiononoutcomeatlong-term.

Inourreviewarticle,itwasdifficulttowithholdcertainfactorsimpactingonlong-termQOLdueto

different studydesigns,methodologies,patientpopulations,appliedQOL instruments, follow-upperiods,

and response rates through the included articles. We however found that factors, which could be

presumedtoresultinapoorQOLafterICU,suchasalongICUorhospitalstay,arenotperseindicatorsof

reductions in QOL afterwards [73]. Other issues such as cognitive impairments, sleep disturbances,

posttraumatic stress disorder, the rehabilitation process, employment status, and cultural and payment

differences,can influenceQOLina lesstangiblewaythan,forexample,physical impairmentsaftermajor

trauma[69,79,80].WematchedAKI-RRTsurvivorswithnon-AKI-RRTsurvivors, toevaluate theeffectof

RTTonlong-termQOL.Thefactor“RRT”seemedsurprisinglynottohaveaverybig impactonlong-term-

QOL. However, long-term QOL was impaired, mainly driven by poor physical functioning. The great

comorbidburdeninthesesurvivorscombinedwithanalreadyimpairedbaselineQOLmayalsocontribute

tothefinallong-termQOL.Inourstudyconcerningoncologicalandhematologicalpatients,wespecifically

searchedforfactorswithimpactonQOL.MultivariateregressionanalysisshowedthatpoorQOL3months

after ICUdischargewas independentlyassociatedwith femalegender,highercomorbidity,hematological

malignancy, older age, andmore organ failure during ICU stay.One year after ICU discharge, older age,

highercomorbidity,andhematologicalmalignancystillnegativelyinfluencedQOL.

These factors alsoplayedan important role inourD1-predictionmodel formeanQOLat1 year.

Withinthispredictionmodel,wewereabletoidentify16D1-variablesthathadgreatimpactonlong-term

outcome.Asalreadymentioned,baselineQOLappearedtobestrongpositivepredictorforlong-termQOL.

Thisunderlinestheimportanceofknowledgeofthisbaselinecondition.Itisalsoisinaccordancewiththe

findingsofVeerbeeck[81]andHeyland[82]whorespectivelydemonstratedthatagoodbaselinestatusin

stroke patients and in older patients had a great impact on long-term functionality. Normilio-Silva also

confirmedthatbaselineQOLandfunctionalitywerethevariablesthatbestdiscriminatedQOLat18months

[39].VariablesinourD1-predictionmodelthathadanegativeinfluenceonQOLat1yearwereolderage,

limitations in functionality, a higher comorbidity, amore severe critical illness, amedical reason as ICU

maindiagnosisandmoreorganfailure.

This is similar to what is found in literature and In general, we may conclude that the most

important determinants of long-termQOL are baseline QOL, co-morbidity, age, functionality, and social

interplays[76-78,83-85].Inalargemulticenterlongitudinalstudyevaluatinglong-termQOL,Orweliusetal

168

found that comorbidity was a very important factor that influenced long-term QOL [83]. In another

multicenter study, theyalso found that ICU-related factorsor the severityof the critical illnesshad little

effectonthereportedlong-termQOL[86].Theysawthat6monthsafterICUdischarge,perceivedQOLin

sepsispatientsdidnotdiffer from ICUsurvivorswithotherdiagnoses,even though thesesepsispatients

weremore severely ill, and had a longer ICU stay. Indeed, our AKI-RRT and theirmatched non-AKI-RRT

patientshadaverycomparableco-morbidityandmedicalhistory,whichmayexplainthefactthattheRRT

componentduringICUstayhadnoeffectonlong-termQOL,whichwasverysimilarbetweenbothgroupsin

ourstudy.AKI-RRTpatientswerealsomoreseverelyillduringtheirICUstaycomparedtomatchedpatients

but thishadno influenceonQOLover theyears. This ishowevernot inaccordancewithour findings in

cancerpatients,wherehematologicalpatientshadahigherseverityofillnessduringICUstayandalower

long-termQOLcomparedtooncologicalpatients.

InourD1-predictionmodel,wefoundthatcomorbiditycertainlyimpactedonlong-termQOLbutto

alesserextentthanbaselineQOL,age,functionality,andseverityof illnessororganfailure. Inastudyby

Luna et al. the presence of comorbidities was associated with poorer outcome in patients with a

community-acquiredpneumonia[87].However,whentherewasnooronlyonecomorbidity,thefactitself

of being 80 years or older increasedmortality. Althoughwe clearly demonstrated the impact of age on

long-term outcome in older patients, in cancer patients and in our D1-predictionmodel, age remains a

difficult parameter to handle in outcome research. Using QOL instruments that are not specific to a

particularagegroupenablescomparisonstobemadewithotheragegroups, i.e.youngerormiddle-aged

groups. However, the questionnaire items of QOL instruments tend to be phrased predominantly in

relationtophysicalfunctionandthusmayinadvertentlydiscriminateagainstolderpersons,whosephysical

functionislikelytobenotasgoodasthatofyoungerpeople.ParticularissuesintheassessmentofQOLin

older patient populations include the persistent finding of a poor relationship between QOL and

disability/diseaseseverity,andtheimportanceofvalidproxyratingsforthoseunabletomakedecisionsor

communicate for themselves. It is important, therefore, that assessment of QOL incorporates issues of

importance to individual older people by broadening the scope of the measurement instruments, thus

representingmore validly theQOL statusofolderpatient groups. Therefore,QOLmeasurements canbe

helpful in decision-making concerning ICU admissionof older patients but its rolemaybe limited at the

same time. Biological age as comorbid burden is therefore more important than chronological age in

outcomeresearch.Biologicalagedoesnotnecessarilyparallelchronologicalageandit ismoredifficultto

estimate [31]. This conceptof “frailty”asmarkerofbiological ageandpredictorofoutcome is relatively

newincriticalcaremedicine. Itreflectsadecline inreserveandfunction inawiderangeofphysiological

systems and accordingly,may represent amore robust predictor of vulnerability and recoverability than

chronologicalagealone.

169

TheClinicalFrailtyScore(CFS)willgiveamorecompletepictureofthegeneralhealthstatusofthe

(older) critically ill patient [56, 88, 89]. Although we did not measure the CFS in our studies, recent

literaturehighlightstheimportanceofknowledgeofthisCFSinprognosticationandappropriatedecision-

makingforoldercriticallyillpatientsaspatientswhoarelessfrailaremorelikelytosurviveandregaingood

physicalfunctioning[28,31,56,57,65,89-91].Althoughfrailtyisfrequentlyassociatedwithadvancedage,

notallolderpatientsarefrail.Youngerpatientscanalsobefrailastheaccumulationofhealthimpairments

drivingthedevelopmentoffrailtymayoccurduringthetotaladultlifespan[92,93].Inanyagegroup,this

CFSisthereforeagoodparametertooutweighthebalancebetweentheburdenofICUmanagementand

thegoaltorestoreanacceptableQOLthatismeaningfulbasedonlifeexpectancy[94,95].Wetherefore

recommendassessingCFSforanycriticalillpatientatICUadmission.

Theoverallfunctionalityorperformancestatusofacriticallyillpatient,whichissomewhatinline

withtheCFS,israthereasytodetermine.WemeasureditthroughtheADLwith4differentcategories(no

limitations,moderate limitations,chairboundorbedridden)andfound it tobeoneofthe importantD1-

variables for prediction of long-term QOL, although only aminority of our included patients was chair-

boundorbedriddenatbaseline.Poor functionalityoftenreflects irreversible factorssuchasolderageor

severecomorbiditiesandstrongassociationsbetweenfunctionalityandQOLwerefound[39].Recently,its

keyroleinoutcomeincritically illpatientswasalsodemonstratedinanotherstudy[84].Theyfoundthat

poor functionality was associatedwith highermortality, irrespective of othermarkers of chronic health

status such as age or comorbidity, and concluded that assessment of functionality was necessarily to

captureamorecompletepictureofapatient’shealthstatus.

Another important variable with impact on long-term outcome and QOL, although difficult to

measure,istheroleofsocial interplaysandintegration.Wesawinourreviewarticlethatpatientswitha

good familial surrounding had a better long-term QOL. This was confirmed in a controlled multicenter

prospectiveexplorative studywhere the levelof social integration,measuredby theAVSI (Availabilityof

SocialIntegration)instrument,significantlyaffectedlong-termQOLinformerICUpatients,eventoalarger

extentthanage[85].Althoughwedidnotmeasuresocialrelationshipswithavalidatedinstrument,wealso

demonstratedinourelderlystudythatagoodfamilial,paramedicalandmedicalnetwork,andnofinancial

problemsaddedtoperceiveQOLasacceptable.

As already highlighted, all these important determinants of long-term QOL - baseline QOL, co-

morbidity,age,andfunctionality,asdemonstratedinliteratureandinourstudies-werealsocapturedin

ourD1-predictionmodel,withtheexceptionofthevariable“socialintegration”becausewedidnothada

quantitativemeasurement of it. Severity of illness and the level of organ failure at the first day of ICU

admission appeared to have also an important impact on long-term QOL. This illustrates the complex

interplayofpre-ICUhealthstate,andacuteandpersistingillnessindetermininglong-termQOL[96].

170

Althoughonly40%ofthevariabilityinlong-termQOLcouldbeexplained,thispredictionmodelcan

beahelpful tool toguidecriticalcarephysicians indecision-making,communication, resourceallocation,

andadvancedcareplanning.Althoughitisnotdefinedtowhatlevelmodelpredictionscouldbehelpfuland

beyond the scope of our study, it certainlymight facilitate decisions,which otherwisewould have been

takenbaseduponsubjectiveevaluationalone.Decision-makingcanbedifficultparticularly inthespecific

patient subgroups we studied, namely critically ill cancer patients, AKI-RRT patients and older patients,

where there are often doubts considering effectiveness of critical care or where the start of specific

expensivetreatmentsduringICUstaycanbequestioned.

Specificpredictionmodelsforcriticallyillolderpatientsdoexistbuttheyfocusedonmortality[97,

98]orfunctionality[82]inthisspecificpatientgroup.Onestudyspecificallystudiedthepredictivevalueof

earlydevelopmentofAKIonsurvivaland long-termQOL[99].Ourpredictionmodel isunique in its form

because it has the advantage that it canbe applied in any critically ill patient butmeanwhile alsohas a

patient-centeredoutcomeapproachasitpredictsmeanlong-termQOLoftheindividualpatientinsteadof

short-termmortalityestimatedbytheclassicalseverityofillnessscores[100].

Thereforewemaystatethatitrespondstothecriteriaofmodernandpatient-centeredoutcomeprediction

research[101].

Still, our D1-prediction model will never replace clinician’s judgments, but rather inform and

reinforce these judgments, as recommendations for further care highly correlate with physician’s

estimationsofagoodlong-termQOL[102].Arecentstudydemonstratedthatprognosesmadebycritical

care physicians at ICU discharge incorrectly predicted long-term survival and QOL in one-third of ICU

survivors.Inaccurateprognosesweregenerallytheresultofoveroptimisticexpectationsofoutcome[103].

Theneedforanobjectivepredictiontool toaiddecision-making inthecomplexenvironmentofacritical

caredepartmentseemsthereforeobvious.

171

III.Conclusionsofthethesis

Ourstudyresultsmighthelpingainingbetterknowledgeaboutlong-termQOLandinthedesignof

futurestudiesonlong-termQOL.Whilethefocusincriticalcaremedicineisstillon“survival”,webelieve

that long-term QOL must become as important in outcome target. With more and more studies now

focusing on long-term QOL, it will certainly influence our decision-making process, although to which

extendwillbequitehardtomeasure.Despitethefactthattheinterestandthenumberofstudiesreporting

onpost-dischargeoutcomesofICUsurvivorshasnowincreasedsubstantially,theabilitytocompareresults

or draw strong conclusions remains impeded by the use ofmany different outcomemeasurements and

varioustimingsofassessments.

Nevertheless,wenowknowthattheburdenofcriticalillnessinICUsurvivorsisasubstantialandan

underrecognizedproblem.IntheyearsafterICUdischarge,critically illsurvivorspresentexcessmortality

andprolongedphysical,mentalandcognitivemorbiditywithdifferentdegreesofseverity.Consequently,

the overall well-being of the individual patient at long-term must be taken into account when taking

decisions during the ICU stay. Critical care physicians shouldnot only use their own frameof ideals and

standards tomake these decisions but respect the patient’s preferences and values. Consequently, the

degreeanddurationofadvanced life-supportingmeasures shouldbe inbalancewith theexpected long-

termsurvivalandQOLinthecriticallyillpatient.

With the growing andbetter knowledgeof theseproblems that ICU survivors and their relatives

mayexperienceafterICUdischarge,ithasbecomeclearthatawarenessoftheseconsequencesarecrucial

if we want to improve long-term outcomes and QOL. As we now have the tools to recognize and

understandthesesequelae,itenablestheintroductionofbetterpreventivemeasuresandmorestructured

andestablishedpost-hospitalinterventions.AlthoughthebenefitofICUfollow-upconsultationsorspecific

rehabilitationprogramsneedstobeprovenyet,itisofimportancethatbothpreventionandintervention

measures should be patient-tailored to guarantee the best possible results. This leads us to the next

chapter:futureperspectivesinoutcomeresearch.

172

IV.Futureperspectives

1.Researchlevel:anongoingbetterknowledgeoflong-termoutcomesandQOL

CommitmentofcriticalcarephysicianstowardscriticallyillpatientsshouldnotendatICUdischarge

butinsteadprolongsmuchbeyond.Thefocusoffurtherresearchseemsthereforeratherstraightforward.

1.1FurtherresearchbasedupontheCOSIcohort

Many critically ill patient populations are of interest for outcome research. Long-termoutcomes

andQOLinCOSIpatientswithaprolongedICU-LOS(atleast8days)havebeenassessedanddataneedto

be further analyzed. Thesepatients are especially at risk to developmajor dysfunctions on the physical,

mentalandcognitive level.Theirprolonged ICU-staycanbeexplainedbyseveral reasons: thecomplexity

andseverityoftheiracutecritical illness,combinedwitha longtimeperiodneededtorecoverandtobe

abletobedischargedtoageneralhospitalwardforfurtherrehabilitation.TheirprolongedICU-stayisalso

partofthedecisiontonotwithdrawornotwithholdtherapyandtogivethesepatientsachancetosurvive

without medical obstinacy or futility. This implies a good baseline condition in these patients, which

howeverwillmakethefinal long-termoutcomeinmanycasesconfronting.Preventiveandinterventional

measuresforgoodrecoveryareimportantinthesepatients.

OurD1-predictionmodelshouldbeexternallyvalidatedandshouldfinallybeevaluatedinaclinical

prospective impact study comparingpredictionsmadeby theD1-model and real life long-termoutcome

data[104].Auser-friendlyelectronicformatcouldeventuallybeimplementedbedsideforconvenientdata

processingandtransmission[101,105].

Wealsodevelopedalongitudinalpredictionmodeltakingintoaccountthefactor“time”(datanot

publishedyet).ItisamorecomplexmodelthantheD1-modelbutwiththerefinedpossibilityoflong-term

QOLpredictionperconsecutivedayoftheICUstay.Althoughtimehadaweakeffectonpredictionoflong-

term QOL, taking into account the “time” variable increased the predictive power of the model as it

consideratetheday-to-dayevolution-improvementordeterioration–oftheindividualpatientduringICU

treatment.Thislongitudinalmodelalsoshouldbeexternallyvalidatedinthefuture.

1.2Globalresearch

Anongoingbetterknowledgeandbroaderpictureof long-termoutcomesandQOL incritically ill

patients remains important.Bettermethodology, amoreuniformoutcome researchwithmoreuniform,

standardizedandvalidatedQOL instruments,with a reasonable longanduniform follow-upperiod (long

enough tohaveany ideaabout the long-termQOLbutwitha lownumberofpatients lost to followup),

with assessment of baseline (pre-ICU) QOL (to compare with long-term QOL) and with a focus on very

specifiedcriticallyillpatientgroupstogainthebestandmostinformationarehoweverneeded[3,6].The

173

critically care society recentlyhaspaidmoreattention to this and recently somearticles concerning this

werepublished[23,31,43,106].Futurestudiesshouldtryto focusonthecomplexdynamic interplayof

short-andlong-termexpectationsandevolutionsinQOLwhiletakingmultidisciplinarydecisions.Evidently,

even the most detailed long-term outcome and QOL data cannot replace clinical evaluation of the

individualpatientoroverruleapatient’spersonalview,thoughtheycertainlyassistintakinganinformed

decision. Future research inQOL should ideally incorporate the perspective of the individual in order to

enablevalidconclusionstobederivedbasedoncontentthat isrelevanttothe individualbeingassessed,

thusinformingmanagementdecisions,policyandpracticemoremeaningfully[107].

2.ICUandhospitallevel:improvingoutcomesbypreventivemeasures

2.1.TriageuponICUadmission

Throughourresearchandthroughtherecentlyexpandingcurrentliterature,wenowhaveabetter

understanding of long-term outcomes and QOL in critically ill patients. Long-term QOL is impaired

compared tobaseline and lower thanQOL in the general population. It is thereforeessential to identify

these patients who are most likely to benefit from critical care, not only to prevent suffering from

unnecessary treatments but also to optimise the use of resources. Reaching this balance is difficult and

wouldbeeasierwithreliableprognostication,whichunfortunatelyhasbeenproventoremainchallenging

at themoment. The classical ICU scoring systems frequently take age and comorbidity into account but

theyarenotadaptedtothespecificcharacteristicsoftheindividualpatientandtheyarenotdesignedfor

triage [30, 31]. Routine knowledge and implementation of bedsideQOL instruments, such as the EQ-5D

withimmediateandautomaticcalculationofUI-toacknowledgebaselinesituation-willalreadygivesome

informationconcerningoutcomeandfuturelong-termQOL,incombinationwithcomorbidity,functionality,

age,socialenvironment,andfrailtyassessment.Itshouldbecomeanautomatismtoassessallthesefactors

before or at ICU admission, next to medical history, use of medication, and a clinical examination, to

estimatepatients’prognosesatlonger-term.EvensmallchangesinQOLmaybeofimportancetopatients

andQOLdatashouldthereforebeusedtoinformpatients.

Additionally,whendecidingtoreferoradmitapatienttotheICU,prognosticationsattheindividual

level in critically ill patients should consider the whole health process rather than focusing on the ICU

period alone. This remains however extremely difficult because many factors related to the underlying

disease,theacuteseverityofillness,andprojectionsonfuturetreatmenthavetobetakenintoaccount.

Anequally importantpartofawell-consideredICUadmissionpolicy isknowledgeofthepatient’s

wishes, preferences and thoughts before or at ICU admission. This is what personalized medicine

differentiatesfromprecisionmedicine.Ittakesintoaccountthepatient’spersonality,preferences,values,

goals,healthbeliefs,socialnetworks,financialresources,andlifecircumstances–“thepersonomics”ofthe

174

patient[108].Ithasbeendemonstrated,particularlyintheolderpopulation,thatphysiciansfrequentlydo

not seek or are unaware of patient’s preferences regarding ICU admission or level of ICU treatment,

althoughtheirdecisionsandactionsduringICUstayarefrequentlybaseduponapatient’swishesandmay

evenchangewhenknowingthesechoices[109-112].Goodinsightsofapatient’swishesmaynotonlyassist

clinicians inprovidingbetterandmorepatient-centeredcare, itmay finallyhelp to transformhealthcare

[113]. Survivalperseshouldnotbeouronlyaim,rathersurvivalwithagoodQOL,oral leastaQOLthat

matchesapatient’spreference[114].

2.2Clinicalpatient-centeredoutcomepredictiontool

When referring a patient to the ICU, QOL is frequently of secondary importance when medical

outcomes–particularsurvival–canbesignificantlyaffectedbycriticalcaretreatment.Cliniciansoftendo

nothavesufficientconfidenceinQOLdatatoincorporatetheminclinicaldecision-making,becauseoflack

ofknowledge,experienceandunderstandingofthemeasurementsandscores.However,estimationofthe

benefitsofanICUadmissionshouldbeconsiderednotonlyintermsofsurvivalbutalsotakingintoaccount

therestorationofanacceptableQOL.Apredictionmodelforlong-termQOLbaseduponreadilyavailable

data could therefore help critical care physicians to triage patients for ICU admission, to identify those

patientswhowillreturntotheirbaselinefunctionality,orthosewhowillneedalongrevalidation.Itcould

also help to informpatients and families in a reliableway, to guide in treatment decisions, and it could

eventually help to transform future healthcare by making better prospects of recovery and better

allocationof resources.Although it isnotdefined towhat levelpredictionmodels couldbehelpful, they

certainly might facilitate decisions, which otherwise should have been taken based upon subjective

evaluation alone. Our developed D1-predictionmodel will therefore rather inform and reinforce clinical

judgments, as recommendations for further care highly correlatewith physician’s estimations of a good

long-termQOL [102, 115]. As highlighted earlier, further research should focus on prospective external

multicenter validation of our D1-prediction model and our longitudinal prediction model for long-term

QOL.

2.3Strategiestodecreaselong-termconsequencesofcriticalillness

2.3.1IncreasingawarenessofPICSandPICS-F

Asshown in literatureand inourstudies,manycritically illpatientswill suffer from long-

termconsequencesoftheiracuteillnessonthephysical,mentalandorcognitivelevel.“Post-IntensiveCare

Syndrome” (PICS) was agreed as the recommended term to describe these new or worsening physical,

mentalorcognitiveproblemsarisingafteracriticalillnessandpersistingbeyondacutecarehospitalization.

The term could be applied to either a survivor or family member (PICS-F) [10, 116, 117]. Although the

175

criticalcarecommunity isbecoming increasinglyawareofPICSorPICS-F,patients, families,andthepost-

hospital carecommunityneedmore information.This is important sinceawarenesscandecrease fearof

the unknown, decrease feelings of being unique, alone, abandoned, or of something else being terribly

wrongwiththem,andalertthemmeanwhiletothepossibleneedforfollow-upassessmentsandprevent

unrealisticexpectationsandfrustrations[118]. Aclear informationbrochurededicatedtoPICSshouldbe

availableontheICUandshouldbeprovidedtopatientsand/orfamiliesadriskforPICSorPICS-F.

From:NeedhamDM,DavidsonJ,CohenH,HopkinsRO,WeinertC,WunschH,etal. Improving long-termoutcomesafterdischargefromintensivecareunit:reportfromastakeholders'conference.CritCareMed2012;40:502-509

In order to increase awareness of PICS, the American Society of Critical Care Medicine (SCCM)

establishedaWikipediasection,videosonYouTubewithpatientsandfamiliesdescribingtheirexperiences,

and a “PICS” pamphlet on their website. Dedicated websites with specific information on the ICU

environmentandonpost-ICUandpost-hospitalcaremaybeasourceof feedback, information,comfort,

less stress, and continued follow-up for patients, families, outpatient clinicians or general practitioners

[119]. Although such websites already exist in other countries (www.fcic.nl; www.aftertheicu.org;

www.intensiva.it; www.opeenicliggen.nl), it would be an opportunity for our ICU department and our

hospitaltodevelopasimilarwebsitebutwiththeadditionaluniquepossibilityofapersonallogintoreceive

-asapatientorasafamilymember-veryspecificpatient-or family-centered informationtailoredupon

thecriticalillnessandhealthstateoftheindividualpatient.Itwouldalsobeanopportunitytoreceive,as

criticalcarephysician,datafromthepatientorfamilyconcerningphysical,mentalandcognitivefunctioning

forfurtherresearchandtogiveadviseuponthemostappropriateaftercareforthatmoment.

176

2.3.2ImplementationoftheABCDEFGHbundle

PatientrisksforPICSincludeimmobility,durationofsedationandmechanicalventilation,lengthof

ICU stay, delirium, sepsis,ARDS,hypoglycemia, andhypoxia [118]. An importantpreventivemeasure to

reduce the prevalenceof these risk factors is the implementationof themultifaceted “ABCDEFGH” care

bundle,whichstandsforAirwayandAwakeningmanagement,spontaneousBreathingtrials,Coordination

ofCareandCommunication,Deliriumassessmentandtreatment,Earlymobilization,Family involvement,

Good handoff communication, andHandoutmaterial for PICS and PICS-F [118]. Each component of this

bundleaddressesaspecificpractice in the ICU independentlyassociatedwith improvedpatient-centered

outcomes.Theeffectivenessandsafetyofthebundlewasdemonstratedinabefore-and-afterstudy[120],

andthebundlealsofacilitatedthe implementationofthePain-Agitation-DeliriumguidelinesoftheSCCM

[121]. Higher bundle compliancewas associatedwith improved survival, and less delirium and sedation

afteradjustmentforage,severityofillnessandpresenceofmechanicalventilation[121].

Although itspromising results, aworldwide survey showed thatonly57%of all respondentshad

implementedthisbundlewithhighvariationsacrossimplementationoftheindividualcomponents.Useof

sedation and pain scales scored the best, moderate adherence scores were seen for awakening trials,

spontaneousbreathingtrails,andearlymobilization.Lowadherencewasfoundindeliriumassessment,and

aminorityreportedtheirunit tobe24/7openfor family,ortohaveadedicatedpsychologist tosupport

families[122].

ThisreflectsacompellingneedforgreateruseandimplementationoftheABCDEFGHcarebundle

to reduceorpreventPICS in the future.Only adecadeago, themajorityof ICUs - includingours -were

closedtofamilymembers–withexceptionof2veryshortvisitmomentsaday-practicingheavysedation

andventilation,andpatientimmobilization.Now,theABCDEFGHcarebundlereflectsashiftawayfromthis

approach to a “less ismore” culture in the ICUwith less sedated or awake patients,who are breathing

spontaneouslyasquickandasmuchaspossibleandwhoaremobilizedearlyandmoreactively,toreduce

thedeconditioninganddysfunctionsooftenseeninICUsurvivors[123].Theattentionliesalsoinamore

multidisciplinaryapproachwithanimportantroleforphysiotherapistsandpsychologists.Thiscultureshift

needstimetoexpandandtobecomestandardofcare,whichisnormalforeverychangeinpractice.Inour

ICU, the implementation of the bundle goes further, and compared to some years ago, progression has

certainlybeenmadeonalldifferentcomponents.

However,familyinvolvementinroundsorincareisstillrarelydone.AnopenICUvisitationpolicyis

uncommon, also in our ICU,where pure architecturally it is almost impossible for families to stay 24/7.

Although we are now more flexible regarding visiting hours and visiting possibilities, there is need to

improveorchangeourinteractionswithfamilymembersinthefuture[124].

177

2.3.3Attentionfortheenvironmentofcare

Moreattentionshouldbepaidtoprovideamorehealingandcompassionateenvironmentthatcan

decreaseanxietyanddeliriumandpromotesleepincriticallyillpatients.Itisthereforeimportanttoattend

toroomtemperatureandlighting,decreasenoiseandfalsealarms,tomakesurethepatientcanusetheir

glassesorhearingaids ifnecessary,andpromotefeasibilityoffamilypresenceandfamilyparticipationin

care.FutureICUdepartmentsandfuturehospitalsshouldbedesignedandshouldbebuilttakingthisinto

account.

2.3.4ImplementationofICUstep-downunits:“TheIntensiveCareRecoveryCenter”(IRC)

Most ICU patients, once their acute medical problems are resolved, will be discharged to the

generalward.However,manyofthesepatientswillstillbeveryweakandthestepfromtheintensivecare

unitand intensivemonitoringat the ICUtothegeneralwardwillbe (too)big.Prematuredischarge from

theICUisassociatedwithhigherriskofdeath[125].Thecomplexityandmagnitudeofthephysical,mental

andcognitive rehabilitation in combinationwith further recovery fromelaborateorgan-relatedproblems

mayexceedthecapacityofthewardwherethenursetopatientratioisfarbelowthatoftheICU.

Earlierdischarge fromthe ICUforpatientsneedingmorecarethancouldbeprovidedongeneral

wardsmaybefacilitatedbyaspecificallydesignedICUstepdown-unit.Patientsexpressedapreferenceto

namethisICU-stepdownunitthe“IntensiveCareRecoveryCentre”(IRC),combiningboththeaspectsofan

ongoingneedforcareandneedforrecovery.TheIRCshouldbeadepartment,onlydedicatedforformer

ICU patients and parallel to the intensive care unit, that has the potential for intensive physical

rehabilitation,whichshouldbedonebycriticalcarephysiotherapistsandspecificrehabilitationspecialists

inclosecollaborationwithcriticalcarephysicians.Mentalandcognitiverecoveryshouldbeequallytreated

withintensivetrainingandcareofpsychologistsandoccupationaltherapists.TheIRCshouldalsohaveno

familyvisitingrestrictionsandfacilitatepresenceandaidofclosefamilymembers.

Wepreferaparallelmodelofsuchan IRCbecauseweseemanyadvantages:excellenttreatment

continuity in the transfer from ICU to IRCwithnoor very little lossof information, a very short transfer

distance between ICU and IRC, simplified patient allocation, a common use of intensive care technical

devices (if needed), a common administration, and high flexibility in the exchange of medical and

paramedicalpersonnelbetween ICUand IRC.An integrationmodel–where IRCpatients stayat the ICU

department – has asmost important disadvantage that IRC-patients are obligated to rehabilitate in the

turbulentenvironmentofanICU,whichisnotdesignedforthatpurpose,andwhereitwillbelessfeasible

for family tohave thepossibility tobepresent 24/7.An independentmodel (standaloneunit) couldbe

usefulasaspecializedtreatmentunitforspecificpatients,suchasacoronarycareorastrokeunit,butnot

asrecoveryunitforsuchaheterogeneousandweakpatientgroupasformercomplexICUpatients[126].

178

DefiningwhichpatientsshouldbetransferredfromtheICUtotheIRCcanonlybedoneinavery

general way, as conditions will be very patient-specific. Overall, these patients should have an ongoing

needofcareandmonitoringbutatanother leveland inanotherwaythan ICUpatients.Thisalsowillbe

reflectedina,comparedtotheICU,lowernurse-to-patientratio(andlowercosts),from1:2to1:3or1:4

depending on complexity of the patients and time of the day. IRC-patientswill no longer need invasive

mechanical ventilationor vasopressorsbut theywill need intensive revalidation, beforedischarge to the

general ward can be considered. A dedicated team of critical care physicians, critical care nurses,

physiotherapists, occupational therapists, psychologists and rehabilitation physicians should have the

leading of this IRC. Defining the need for the number of beds in such an IRC is difficult, as there is no

reliable information of an upper limit for bed numbers in such a step-down unit. We propose for our

hospital,astertiarycarecenter,atleast10to12beds;largerunitswillbemoredifficulttomanage[126].

3.Post-hospitallevel:improvingoutcomesbyinterventionmeasures

3.1Post-dischargefollow-upprograms

An intervention measure to treat patients with PICS is the implementation of an ICU follow-up

clinic. Post-hospital follow-up clinics or consultations will give us a better understanding of specific

problemsinphysical,mentalorcognitivefunctioning.Theinformationgainedthroughtheseconsultations

canbeusedtoimprovecriticalcareitselfandcanbeitselfaqualityserviceforpatientsandtheirrelatives

[4].Still,thesefollow-upconsultationsareyetnotcommonplaceincriticalcare.Traditionallyseen,critical

care is not a medical subspecialty that has a well-established patient follow-up program and follow-up

consultationsarenotcommon.Ultimately,manycriticallyillpatientswillgettheirmedicalfollow-upbyan

organ specialist or by their general practitioner. Bothmay have a limited knowledge ofwhat happened

during ICUstayand therefore,bothmayhavedifficulties tohavegood insights into thepost-ICUrelated

problems of the patient. The critical care physician together with an ICU psychologist, a rehabilitation

specialist, and dedicated ICU nurse may be in a better position to understand the consequences that

patients suffer from after having survived their critical illness. They also may better understand which

interventions may improve outcome. Continuity of care through the continuum of care is therefore a

challenge.

Consequently, ICU-aftercareneeds abetter andmore structuredorganization. In theUK, around

30% of ICU departments run a follow-up clinic [127]. Although it seems as though post-discharge

rehabilitationwith specific programs and follow-up clinics would be a logical way to address PICS, until

now, ICU follow-up clinics or randomized controlled trials concerning specialized rehabilitationprograms

versus standard care, still not have proven their benefit [118, 128-131]. The effect of ICU-follow-up

179

consultations improved when the ICU diary, kept by relatives and/or members of the ICU team, was

discussed[132].

At the moment, there are no gold standards for post-ICU follow-up programs but a pragmatic

modelintheScandinaviancountriesandclearrecommendationsintheNetherlandshavebeenformulated

[133, 134]. Nevertheless, a recent electronic survey of ICU-aftercare in Denmark demonstrated an

abundantheterogeneity of criteria and interventions [135]. So,manyquestions still arise.Whowill fund

thisfollow-up?WhichpatientsshouldbetargetsforICU-follow-upclinics,thesickestofthesickorjustany

ICUsurvivor?ItiscommontothinkofARDSpatients,sepsispatients,patientswithprolongedmechanical

ventilation, or patientswith a prolonged ICU stay.What kindof post-ICU interventiondo these patients

need? They will certainly need physical, functional and cognitive rehabilitation but they will also need

education,informationandcarecoordinationfortransitiontoprimarycareinthefuture.Whatshouldbe

offered:a rehabilitationpackage,post-hospitalvisitsanddialogue,or smartphonesappswithadvices for

self-rehabilitation?Where?Whatisthebesttiming?Atthismoment,theoptimaltimetostartwiththese

follow-upconsultationsafterICU-discharge,thebesttimeintervalbetweenvisitsorthebestplaceforthese

follow-upvisitsarestillunknown.

Althoughthere isnoprovenbenefitof ICU-follow-upconsultationsat themoment, intuitivelywe

might assume that theymay be important for both patients and relatives. Itmight be possible thatwe

cannotmeasure the possible positive effects of post-ICU follow-up through easymeasurable biomedical

tests.Walsh et al. found a higher patient satisfactionwithmany aspects of recovery in the intervention

group where patients received more physical and nutritional rehabilitation and more information

comparedtothestandardgroup[131].Overall,whereextendedICUfollow-upexisted,patientsreported

greatsatisfactionwiththeservice[127,136].

Aslong-termoutcomesandQOLcanbeverydifferentfrompatienttopatient,somustbeanykind

ofrevalidationtoo.PatientswithPICSareaveryheterogeneousgroupofformercriticallyillpatientswho

willrehabilitateinadifferentwayandwhereonepatientwillrespondbettertoacertaintherapythanthe

other. So an individually based rehabilitation program should therefore possibly be preferred above the

“one size fits all” approach, which will make the whole discussion concerning post-ICU follow-up

interventionsevenmoredifficult.

Tryingto implementpost-ICUfollow-upconsultationswithoutevidenceandwithmanybarriers is

hard.Commonbarriersforimplementationofpost-ICUfollow-uparelowevidence,noconceptofproof,no

funding, no staff, noplace, too complicated, no clinical benefit, noquick fix, andnot scalable.However,

based upon a small pilot study we performed regarding feasibility of establishing post-ICU follow-up

consultations3monthsafterICUdischarge(unpublisheddata),Istronglybeliefitcouldhelpsomepatients,

althoughitmightindeedbestronglyindividuallybased.Ourstudysamplewassmall,butallthe43patients

180

we sawappreciated these follow-up consultations.Most patientswere accompaniedby a close relative,

eitherahusbandorwife,oroneoftheirchildren.Duringtheconsultation,theformerICU-patientshadto

completetheEQ-5D,theHADS,thePTSS-14,andtheMoCAtestsinaface-to-faceinterview[8,18-22].Next

totheseassessments,wealsoaskedabouttheirlivingcircumstances,returntoworkplans,weightlossand

gain, perceived changes in taste, problemswith talking, swallowing, eatingor sleeping, sexualproblems,

driving a car possibilities, financial problems, healthcare utilization, current use of medication, and

appreciationofthefollow-upconsultation.Allpatientsandtheir familywerehappytocomebackandto

tell their story about their experienceswhile beingon the ICU andpost-ICU andpost-hospital. They felt

respected and appreciated the follow-up initiative a lot. We have to acknowledge that we only saw a

selectionof “thebest”post-ICUpatients as theoneswho still neededmore careand inpatient recovery

were admitted to special care facilities andwere unable to attend the consultation 3months after ICU

discharge.

Sofurtherresearchisabsoluteneededtoprovideaclear“proorcon”basedevidenceforpost-ICU

follow-up consultations. In my opinion, these post-ICU follow-up consultations should become an

integratedpartinthegeneralstrategyofpatientwell-beingandrecovery.Forpatientsforwhomitmaybe

lessconvenienttovisitthepost-ICUfollow-upclinicduetolongtraveldistanceortransportationproblems,

a telephone follow-up or a dedicated, individualized and well-developed website could be of help, as

highlightedearlier.Suchawebsitecouldalsobeinformativeandofhelpformanyothers,suchasgeneral

practitioners,physiotherapists,revalidationphysicians,etc.

3.2Peersupport

At thismoment,our ICU-psychologists startedanew initiativewhere ICU-survivorscanmeet ICU

physicians,physiotherapists,psychologists,andotherformerICU-patientsintheveryinformalenvironment

of a pub and talk about their experiences during and after ICU. These “drop-in” meetings started in

November2017and futuremeetings in 2018havebeenplanned.As survivors and their caregivershave

first-hand experience of the challenges that survivors face, they arewell suited to educate and prepare

peer survivors for certain aspects of the recovery process [137]. They can also be an inspiration for

professionalcaregiversintheirunderstandingandimprovingoftherehabilitationafterICU.

BiggereventsandgroupsforspecificformerICU-patientssuchasTransplantoux,forpatientswho

receivedasolidorgantransplant,alreadyhaveproventheirsuccess[138].

4.Health-economicslevel:resourceallocation

Thecostsofintensivecarearehighandconsumealargefractionofavailableresourcesforhealth

care.AsignificantamountofresourcesintheICUaredevotedtopatientswithapoorprognosis,andmany

181

of them will ultimately die or survive with a poor QOL. Given this, there is an increasing pressure to

examine,evaluateandjustifyutilizationofcriticalcareresources.Furtherresearchandinsightsintopatient

preferences and long-term outcomes combined with cost-effectiveness and cost-utility studies are

necessarilytoimproveallocationofscarceresources[139].

Cost-utilitystudiesanalyzecostsperqualityadjustedlifeyears(QALYs)andallowforcomparisons

betweencertaintherapies. QALYsaremeasuresthatcombinedurationandqualityof life,thuscapturing

boththeeffectsoftherapyorinterventionsandconsequencesofadisease.Theyarecalculatedbasedupon

thepatient’sestimatedsurvivaltimewhileweighingeachlifeyearbyaQOLindexvalue,forexampletheUI

oftheEQ-5D.Abetterunderstandingoflong-termQOLwillalsoleadtoabetterestimationofQALYs,but

still,decisionstooptimizeresourceallocationwillremaindifficultincriticallyillpatients.

Sohowwillthefutureofhealthcareexpenditureincriticalcaremedicinelookalike?Governments

will make further choices to minimize expensive care based upon quality improvement programs.

Techniques and treatments will focus on reducing the need for inpatient hospital care and promote

outpatient treatment, eventually leading to hospitalswith relativelymore ICU beds [140]. This does not

imply that intelligentallocationof resources incritical caremedicinewillno longerbenecessary.On the

contrary,thecrucialquestionwillstillbehowtoselectthesepatientswhowillbenefitthemostfromICU

treatmenttoaimforacost-effectiveuseof ICUbeds.Preventionofhighcostsforpatientswitha limited

life expectation and poor long-term outcomes will be the main tool for optimizing the use of scarce

resources. A better knowledge of long-term outcomes, more transparency and insights into costs and

benefitsofcertainmedicaltreatments,combinedwithagoodwellthoughtoutadmissionpolicy,andwell-

consideredEOL-decisions, in respectwith the individualpatient’s valuesandpreferences,might improve

cost-efficiencyinthefuture.

182

V.Summary

Ourresearchconcentratedaround3majorissues:1/reviewingliteratureandappliedmethodology

concerning long-termQOLandoutcomes research,2/ assessing long-termoutcomesandQOL in specific

critically illpatientpopulationswhere thereareoftendoubts concerningeffectivenessof critical care,or

where the start of specific treatments during ICU stay can be questioned (namely the oncological-

hematological,AKI-RRTandolder(≥80years)patients),and3/developingapredictionmodelforlong-term

QOLbaseduponreadilyavailablevariablesatthefirstdayofICUadmissionandsodeterminingthemost

importantpredictorsforlong-termQOL.

Inourreviewarticle,wefoundthatatleastoneyearafterICU,criticallyillpatientshadalowerQOL

thananage-andgendermatchedgeneralpopulation.Itwasdifficulttowithholdcertainfactorswithimpact

on long-termQOLdue tohugevariations inmethodologyand studydesign,patientpopulations, applied

QOL instruments, follow-up periods, and response rates through the included articles. Recently, more

attention has been paid to this problem and international societies now focus on the need for more

standardizationinoutcomesresearch.

ItisdifficulttoselectthemostappropriateQOLsurvey(s),bothinnumber,incontentandintiming.

As QOL is a patient-centered and subjective outcome parameter by itself, we believe that the use of

validatedtoolstoassessQOLisanabsolutemust.Throughourresearch,wechosetoassessQOLbytheEQ-

5D and SF-36 because these are generic instruments commonly used in critical care; they have a well-

known validity, reliability, and are responsive to changes in health. As they do not assess memory,

concentration, the ability to complete tasks, problem solving, or decision-making, we added a sixth

dimension“cognition”totheEQ-5D.

Weassessedbaselinemortalityrates(ICUandhospitalmortality)andbaselineQOL(definedasQOL

2weeksbefore ICUadmission),andat3monthsand1yearafter ICUdischarge. In thestudyconcerning

AKI-RRTandolderpatients,wealsoassessed livingstatusandQOLatrespectively4yearsandat7years

after ICU discharge. We found high mortality rates in all groups of critically ill patients we studied,

especially in the first 3months since ICU admission,with onlymoderate increase ofmortality at longer

follow-up.Thesemortalityrateswerehowevercomparablewiththenumbersfoundinliterature.

Themeasuresof long-termQOLputsurvivingacritical illness intoa largerperspective.Wefound

thatthecriticallyillpatientsinourresearchhadalowerlong-termQOL,mainlyinthephysicaldimensions,

thanageneralpopulation,butaslowimprovementinQOLovertimecouldbeseen,althoughitremained

underbaselinelevel.Inourreviewstudy,wefoundthatpatientswithsevereARDS,prolongedmechanical

ventilation,severetrauma,andseveresepsisappearedtohavetheworstreductionsinQOL,whichlasteda

long time. The impact of diagnostic category upon long-term QOL was also reflected in our prediction

model;withsurgicalpatientshavingasignificantlybetterpredictedlong-termQOLthanmedicalpatients.

183

Evidently,theincludedcancer,AKI-RRTandolderpatientsrepresentednotonlyahighlydiversespectrum

of diseases but also patientswith a very heterogeneous performance status and co-morbidity. As such,

outcomesshouldbedifferentiatedamongsubgroups.

We found important differences between solid tumor patients and hematological patients with

hematological patients having aworseQOL on everymoment of the study period, and experiencing no

significant improvements beyond 1 year. Differences in QOL between AKI-RRT and their non-AKI-RRT

matchesateachdifferenttimepointwereverysmall.ThisimpliedthattheRRTcomponentduringcritical

illnessdidnothavean important impacton long-termQOL.Overall, long-termQOL remainedunder the

baseline level for AKI-RRT and non-AKI-RRT patients, and under the QOL of the average population,

specifically in the more physical domains. Determining patients who should benefit the most from ICU

admissionbecomesmoreandmorecomplicated,andthisisparticularlythefactinpatientsaged80years

orolder.Thelong-termQOLinthecriticallyillolderpatientsinourstudywaslowcomparedtoageneral

population,particularlyinself-care,usualactivitiesandthephysicaldomains,withanincreasingnumberof

patientsexperiencingmoreproblemsovertime.Theseolderpatientshoweverrecognizedlittlechangesin

QOLover timeexcept formobility and self-care.Older patients adaptedwell to their advanced age and

perceived their overall QOL as acceptable. It suggests that QOL might have another meaning for older

patients,withsocialandmentalvaluesbeingfarmoreimportantthanlimitedphysicalfunctioning.

AdifferencehastobemadebetweenQOLmeasurementsandperceptionofQOLasexperiencedby

the patient himself, assessed by the VAS. Oncological patients had a better perception of their QOL

compared to hematological patients, but for both groups QOL was still acceptable. AKI-RRT patients

perceivedQOLasgoodandbothAKI-RRTandnon-AKI-RRTpatientsreportedlowdependenceindailylife

later on. This perception of a fair QOL was also illustrated by the fact that the vastmajority of all our

includedpatientswhowerealiveafter1yearorevenlongerwantedtobereadmittedtoanICUincaseof

deterioration.

Ourresearchwasobservational,solookingforcausesorexplanationsforlong-termQOLisdifficult.

However,we developed a predictionmodel for long-termQOL and hence, tried to determine themost

importantvariablesforpredictingthisoutcome.WefoundaverystrongpositiverelationofbaselineQOL

withlong-termQOLinourD1-predictionmodel.Thisillustratestherequirementofknowledgeofbaseline

conditiontomakeanypredictiononoutcomeatlong-term.VariablesnegativelyrelatedwithmeanQOLat

1yearwereanoncologicalorhematologicaldisease,olderage,limitationsinfunctionality,ahigherseverity

of illness, organ failure with need for mechanical ventilation or vasopressors, and a high comorbidity.

Althoughonly40%ofthevariabilityinlong-termQOLcouldbeexplainedbyourpredictionmodel,itmight

certainly facilitate decisions,which otherwise should have been taken based upon subjective evaluation

alone.

184

Based upon literature and based upon our research, wemay conclude that themost important

determinants of long-termQOL are baselineQOL, co-morbidity, age, functionality, and social interplays.

Althoughwe clearly demonstrated the impact of age on long-termoutcome in older patients, in cancer

patientsandinourD1-predictionmodel,ageremainsadifficultparametertohandleinoutcomeresearch.

In fact, biological age as comorbid burden is more important than chronological age. The concept of

“frailty” as marker of biological age reflects a decline in reserve and function and accordingly, may

represent a more robust predictor of vulnerability and recoverability than chronological age alone.

Althoughfrailty isfrequentlyassociatedwithadvancedage,youngerpatientscanalsobefrail. Inanyage

group,assessingfrailtyisthereforeagoodparametertooutweighthebalancebetweentheburdenofICU

managementandthegoaltorestoreanacceptableQOLthatismeaningfulbasedonlifeexpectancy.

Throughourresearchandthroughtherecentlyexpandingcurrentliterature,wenowhaveabetter

understandingoflong-termoutcomesandQOLincriticallyillpatients.Thereisnodoubtthatcriticalillness

affects long-term outcomes in the physical, mental, and cognitive dimensions, a syndrome which was

recentlydefinedas“PICS”.ImplementationoftheABCDEFGHbundleduringICUstaycouldbethefirststep

topreventpatientsfromdevelopingPICS.ThisbundleimpliesashiftincultureattheICU,withlesssedated

patients,whoarebreathingspontaneouslyasquickandasmuchaspossibleandwhoaremobilizedearly

andmoreactively. Theattentionwill lie in amoremultidisciplinary approachwithan important role for

physiotherapists,psychologists,andmorefamilyinvolvement.Thiscultureshiftneedstimetoexpandand

tobecomestandardofcare.

Although the critical care community is now becoming increasingly aware of PICS, patients,

families,andthepost-hospitalcarecommunityneedmoreinformation.Thisisimportantsinceawareness

can decrease fear of the unknown, and alert them meanwhile to the possible need for follow-up

assessmentsandpreventunrealisticexpectationsandfrustrations.Dedicatedwebsitesorappswithspecific

informationontheICUenvironmentandonpost-ICUandpost-hospitalcaremaybeasourceoffeedback,

information, comfort, and continued follow-up for patients, families, outpatient clinicians or general

practitioners.Itwouldalsobeanopportunitytoreceive,ascriticalcarephysician,datafromthepatientor

familyconcerningpost-hospitalphysical,mentalandcognitivefunctioningforfurtherresearchandtogive

adviseuponthemostappropriateaftercareforthatmoment.

ICU step-down units to facilitate the step towards the general ward and post-ICU follow-up

consultationsmaybefutureinitiativestofurtherimprovelong-termoutcomes,QOLandcost-effectivecare

incriticallypatients.

185

VI.Samenvatting

Het onderzoek in deze doctoraatsthesis concentreerde zich op 3 belangrijke domeinen: 1/ het

bestuderen van de literatuur aangaande levenskwaliteit (QOL) op lange termijn en van de toegepaste

methodologie,2/hetanalyserenvanlange-termijngevolgenenQOLbijdiekritiekziekepatiëntenwaarer

vaaktwijfelszijnoverdeeffectiviteitvanIntensievezorg(IZ),ofwaardestartvanbepaaldebehandelingen

tijdensdeIZ-opnameinvraagwordtgesteld(metnamedeoncologische-hematologische,deAKI-RRT,ende

oudere (≥80 jaar) patiënten), en3/ het ontwikkelen vaneenpredictiemodel voorQOLop lange termijn,

gebaseerdopgegevensdiebeschikbaarzijnopdeeerstedagvanIZ-opname.

Inonsoverzichtsartikel vondenwedat,minstens1 jaarnaontslag van IZ, kritiek ziekepatiënten

een verminderde QOL hadden ten opzichte van een algemene populatie met vergelijkbaar geslacht en

leeftijd.Hetwasmoeilijkombepaalde factoren teweerhoudendieeen impacthaddenop lange termijn

QOL door grote variaties in methodologie en studie design, patiëntenpopulaties, gebruikte

meetinstrumenten voor QOL, opvolgperiodes, en responspercentage binnen de geïncludeerde artikels.

Recentisermeeraandachtvoorditprobleemeninternationaleverenigingenconcentrerenzichopdenood

voorbeterestandaardisatiebinnenoutcomeonderzoek.

Het ismoeilijkomdemeestgeschiktevragenlijstenteselecterenvoorhetmetenvanQOL,zowel

naar inhoudalsnaartiming.GezienQOLeenpatiënt-gerichteensubjectieveparameter isopzich,vinden

wedat het gebruik van gevalideerde vragenlijstenomQOLna te gaan, een echte “must” is. Binnenons

onderzoekkozenwij voordeEQ-5DendeSF-36vragenlijstenomdathet algemenevragenlijsten zijndie

vaak gebruikt worden binnen kritiek zieke patiënten. Ze hebben een goed gekende validiteit en

betrouwbaarheid en zijn gevoelig voor veranderingen in de gezondheidstoestand van de patiënt. Deze

vragenlijstenevaluerenechterniethetgeheugen,concentratievermogen,ofmogelijkhedenomopdrachten

uittevoeren,problemenoptelossen,ofombeslissingentenemen.Daaromvoegdenwezelfeen6evraag

overcognitieaandeEQ-5Dtoe.

In ons onderzoek werd basis-mortaliteit (mortaliteitspercentage op IZ en in het ziekenhuis) en

basis-QOL (gedefinieerd als QOL 2 weken voor IZ-opname) nagegaan, alsook 3 maanden en 1 jaar na

ontslagvanIZ.IndestudiesaangaandeAKI-RRTpatiëntenenouderepatiëntenwerdenmortaliteitenQOL

ook nagegaan na respectievelijk 4 en 7 jaar na ontslag van de IZ-afdeling. We vonden hoge

mortaltiteitspercentagesinallegroepenvanIZ-patiëntenbinnenonzestudies,voornamelijkindeeerste3

maandensindsIZ-opname,metenkeleenmatigetoenameoplangeretermijn.Dezemortaliteitscijferszijn

vergelijkbaarmetdiegenediebeschrevenwordenindeliteratuur.

QOLop lange termijnplaatst het overleven vaneen kritieke ziekte in een anderperspectief.De

kritiek zieke patiënten binnen ons onderzoek hadden een lagereQOL op lange termijn, voornamelijk op

186

fysiek vlak, in vergelijking met de algemene bevolking. Een trage verbetering van QOL kon worden

waargenomen,maardezebleefwelonderhetniveauvandebasis-QOL.Inonsoverzichtsartikelzagenwe

datvoornamelijkpatiëntenmeteenernstigARDS,langdurigemechanischeventilatie,naeenzwaartrauma

ofnaernstige sepsisdemeestuitgesprokenverminderinghadden inQOL.Dezedaling inQOLhield lang

aan. De impact van een bepaalde diagnose op lange-termijn QOL werd ook teruggevonden in ons

predictiemodelwaar chirurgische patiënten een significant betere voorspelde lange-termijnQOLhadden

dan medische patiënten. Het is dan ook logisch dat onze geïncludeerde kanker-, AKI-RRT, en oudere

patiënten niet enkel een zeer divers spectrum van zieke patiënten vertegenwoordigden, maar ook

patiënten waren met een verschillende functionaliteit en comorbiditeit bij aanvang van de IZ-opname.

Bijgevolgmoetoutcomegedifferentieerdwordentussendezeverschillendepatiëntengroepen.

Er waren belangrijke verschillen tussen oncologische en hematologische patiënten, waarbij de

hematologischepatiëntenopelkogenblik vande studieeen slechtereQOLhaddendandeoncologische

patiënten,enwaarbijergeensignificanteverbeteringenwarenbinnenhetjaar.DeverschillentussenAKI-

RRTennietAKI-RRTpatiëntenwarenopelkgemetentijdstipergklein.Ditzoukunnenbetekenendatde

factor“dialyse”weinigimpacthadoplange-termijnQOL.Inhetalgemeenwasdelange-termijnQOLvoor

AKI-RRTénnietAKI-RRTpatiëntenonderhetbasisniveauvanQOLen lagerdandeQOLvandealgemene

bevolking, voornamelijk op fysiek vlak. Bepalen welke patiënten het meest voordeel halen uit een IZ-

opnameisergcomplex,endatisvoorzekerzovoordegroepvanouderepatiënten.Delange-termijnQOL

indezegroepvanpatiëntenwas laag invergelijkingmeteenalgemenepopulatie.Voornamelijkopfysiek

vlak, zelf-zorg en dagdagelijkse activiteitenwerden ermeer enmeer problemenwaargenomenover het

verloopvantijd.Tochervaardenouderepatiënten,behalveinmobiliteitenzelf-zorg,weinigveranderingin

QOL.Ouderepatiëntenpastenzich inhetalgemeenvrijgoedaanaanhungevorderdeleeftijdenvonden

hunQOLaanvaardbaar.DitsuggereertdatQOLvoorouderepatiënteneenanderebetekenisheeft,waarbij

een goede sociale omgeving en een goede mentale functionaliteit van veel groter belang zijn dan een

verminderdemobiliteitofzelf-zorg.

DaaromishetbelangrijkeenverschiltemakentussendegemetenQOLendeQOLzoalsdieervaren

wordtdoorpatiënten.DezeperceptievanQOLkanwordennagegaanviadeVAS.Oncologischepatiënten

haddeneenbeterperceptievanhunQOLdanhematologischepatiëntenmaarvoorbeidegroepenwasde

lange-termijnQOLaanvaardbaar.OokvoorAKI-RRTennietAKI-RRTpatiëntenwasde lange-termijnQOL

erg aanvaardbaar en beide patiëntengroepen hadden van een vrij onafhankelijk leven op lange termijn.

Deze perceptie van accepteerbareQOL op lange termijnwerd ook bevestigd door het feit dat de grote

meerderheidvanalonzegeIncludeerdestudiepatiëntenopnieuwwenstenopgenomentewordenopeen

IZ-afdelingindienditnodigzouzijn.

187

Ons onderzoekwas observationeel dus oorzaken of verklaringen vinden voor lange-termijn QOL

wasmoeilijk. Desondanks kondenwe door het ontwikkelen van een predictiemodel voor QOL op lange

termijnwelenkele factoren selecterendiebelangrijkblekenvoor lange-termijnQOL.Wevondenbinnen

ons D1-predictiemodel een zeer sterke positieve relatie tussen basis-QOL en lange-termijn QOL. Dit

illustreert het belang van het kennen van deze basisconditie om enige inschatting te kunnenmaken op

langeretermijn.Variabelendienegatiefgerelateerdwarenaanlange-termijn-QOLwarendeaanwezigheid

van een oncologische of hematologische aandoening, oudere leeftijd, verminderde functionaliteit, een

grotereernstvanziek-zijn,orgaanfalenmetnoodaanmechanischeventilatieen/ofvasopressoren,eneen

groterecomorbiditeit.Ondankshetfeitdatweslechts40%vandevariabiliteitinlange-termijnQOLkonden

verklarendooronspredicitiemodel, zalditmodelons tochkunnenhelpenmethetnemenvanbepaalde

beslissingendieanderslouteropsubjectievebasiszoudengenomenzijn.

Gebaseerd op ons predictiemodel en op de literatuur, kunnen we besluiten dat basis-QOL,

comorbiditetit,leeftijd,functionaliteitensociaalmilieudemeestbelangrijkefactorenzijndieinvloedzullen

hebben op lange-termijn QOL. Ondanks het feit dat we de invloed van leeftijd op lange-termijn QOL

duidelijk konden aantonen bij ouderen, bij kankerpatiënten en in ons predictiemodel, blijft leeftijd een

moeilijke parameter in outcome onderzoek. Eigenlijk is biologische leeftijd van groter belang dan

kalenderleeftijd. Het concept van “frail-zijn” als merker van deze biologische leeftijd kenmerkt een

vermindering in fysiologische reserve en functie en zal een betere voorspellende waarde hebben voor

kwetsbaarheidenmatevanrevalideerbaarheiddankalenderleeftijdalleen.Dezematevanfrail-zijnwordt

vaakgeassocieerdmethogereleeftijdmaarookjongerenkunnenevengoedfrailzijn.Daaromzal,inwelke

leeftijdsgroepdanook,hetbepalenvanditfrail-zijneengoedeparameterzijnomdeimpactvanhetkritiek

ziek-zijnaftewegentenopzichtevanmogelijkhedentotherstelnaareenaanvaardbareQOL.

Dooronsonderzoekendoorderecenttoegenomenliteratuurhebbenwenueenbeter inzicht in

lange-termijnoutcomeenQOLinkritiekziekepatiënten.Erisgeentwijfelmeerdatditkritiekziek-zijnde

lange-termijn outcome zal beïnvloeden op fysiek, mentaal en cognitief vlak; een syndroom dat recent

“PICS”werd genoemd. Het implementeren van de “ABCDEFGH” zorgbundel kan een eerste stap zijn ter

preventie van PICS bij patiënten opgenomen op IZ. Deze bundel veronderstelt wel een zekere

cultuursveranderingopIZ,waarbijpatiëntenminderénminderlanggesedeerdzullenzijn,meerensneller

spontaanzullenademenenmeerenactieverzullengemobiliseerdworden.Erzalmeeraandachtzijnvoor

eenmultidisciplinaire samenwerkingwaarbij kinesisten, psychologen en familieleden een belangrijke rol

zullenspelen.Sowiesozalhettijdvergenvooraleerdezezorgbundelalsalgemenenormwordterkend.

OndankshetfeitdatbinnendeIZ-wereldermeerenmeererkenningenherkenningisvanPICS,is

hetnoodzakelijkompatiënten,familieleden,enpost-hospitaalzorgverlenershierovergoedteinformeren.

Dezeinformatieisbelangrijkomangstvoorhetonbekendetevoorkomen,ominzichttegevenindenood

188

voorverderopvolgingenomonrealistischeverwachtingenenfrustatiestebeperken.Speciaalontworpen

websitesen/ofappsmetgoede informatieover IZ,depost-IZperiodeendepost-hospitaalzorgverlening

kunneneenbron zijn van feedback,uitleg, comfort, and continueopvolging vanpatiënten, familieleden,

poliklinieken of huisartsen. Het zou tevens een opportuniteit zijn om als IZ-arts vervolg-data op fysiek,

mentaalencognitiefvlakvandepatientof familie tekrijgen;datadiebelangrijkkunnenzijnvoorverder

onderzoek en die omgekeerd ook een zeer gerichte en persoonlijke nazorg naar de patient en familie

mogelijkmaken.

Step-downeenhedennaeenIZ-opname,omdeovergangnaardealgemeneafdelingmakkelijkerte

maken,enpost-IZopvolgconsultatieskunneninitiatievenzijnindetoekomstdieeenverdereverbetering

van lange-termijnoutcome,QOLeneenkosten-effectievezorg inkritiekziekepatiëntenmogelijkkunnen

maken.

189

VII.References

1. Heyland DK, Guyatt G, Cook DJ,MeadeM, Juniper E, Cronin L, Gafni A. Frequency andmethodologic rigor ofquality-of-lifeassessmentsinthecriticallyill.CritCareMed1998;26:591-5982. Flaatten Hans. Long-term outcomes after intensive care. 25 Years of progress an innovation in intensive caremedicine.Ed.MedizinischWissenschaftlicheVerlagsgesellschaftBerlin2007;419-4273. Turnbull AE, Rabiee A, Davis WE, Farhan Nasser M, Reddy Venna V, Lolitha R, Hopkins RO, et al. OutcomemeasurementinICUsurvivorshipresearchfrom1970to2013:Ascopingreviewof425publications.CritCareMed2016;44:1267-12774. Granja C, Amaro A, Dias C, Costa-Pereira A. Outcome of ICU survivors: a comprehensive review. The role ofpatient-reportedoutcomestudies.ActaAanesth.Scand2012;56:1092-1103

5. BlackNA,JenkinsonC,HayesJA,YoungD,VellaK,RowanK,DalyK,RidleyS.Reviewofoutcomemeasuresusedinadultcriticalcare.CritCareMed2001;29:2119-2124

6. NeedhamDM.Understandingandimprovingclinicaltrialoutcomemeasuresinacuterespiratoryfailure.AmJRespirCritCareMed2014;189:875-8777. LimWC,BlackN,LampingD, RowanK,MaysN. Conceptualizing andmeasuringhealth-relatedquality of life incriticalcare.JCritCare2016;31:183-1938. TheEuroQolGroup.EuroQol–anewfacilityforthemeasurementofhealth-relatedqualityoflife.HealthPolicy1990;16:199–208

9. Ware JE, KosinskiM, Bjorner JB, Turner-Bowker DM, Gandek B,MaruishME. User’sManual for the SF-36v2®HealthSurvey.QualityMetricIncorporated,Lincoln,RhodeIsland;200710. NeedhamDM,DavidsonJ,CohenH,HopkinsRO,WeinertC,WunschH,etal.Improvinglong-termoutcomesafterdischargefromintensivecareunit:reportfromastakeholders'conference.CritCareMed2012;40:502-50911. StouthardME, Essink-BotML, Bonsel GJ. Disabilityweights for diseases. Amodified protocol and results for aWesternEuropeanregion.EurJPublicHealth2000;10:24-3012. HoeymansN, van LindertH,WestertGP. Thehealth statusof theDutchpopulationas assessedby theEQ-6D.QualLifeRes2005;14:655-66313. Soliman IW,deLangeDW,PeelenLM,CremerOL, SlooterAJ,PasmaW,Kesecioglu J, vanDijkD.Single-centerlarge-cohortstudyintoqualityoflifeinDutchintensivecareunitsubgroups,1yearafteradmission,usingEuroQoLEQ-6D-3L.JCritCare2015;30:181-18614. WoltersAE, vanDijkD, PasmaW,CremerOL, LooijeMF, de LangeDW,VeldhuijzenDS, SlooterAJ. Long-termoutcomeofdeliriumduringintensivecareunitstayinsurvivorsofcriticalillness:aprospectivecohortstudy.CritCare2014;18:R12515. TimmersTK,vanHerwaardenJA,deBorstGJ,MollFL,LeenenLP.Long-termsurvivalandqualityoflifeafteropenabdominalaneurysmrepair.WordJSurg2013;37:2957-296416. TimmersTK,VerhofstadMH,MoonsKG,vanBeeckEF,LeenenLP.Long-termqualityoflifeaftersurgicalintensivecareadmission.ArchSurg2011;146:412-41817. Jorm AF,Jacomb PA. The Informant Questionnaire on Cognitive Decline in the Elderly (IQCODE): socio-demographiccorrelates,reliability,validityandsomenorms.PsycholMed1989;19:1015-1022

18. Ballenger JC, Davidson JR, Lecrubier Y: Consensus statement on posttraumatic stress disorder from theInternationalConsensusGrouponDepressionandAnxiety.JClinPsychiatry2000;61(Suppl5):60-66

190

19. TwiggE,HumphrisG,JonesC,BramwellR,GriffithsRD:Useofascreeningquestionnaireforpost-traumaticstressdisorder(PTSD)onasampleofUKICUpatients.ActaAnaesthesiolScand2008;52:202-208

20. ZigmondAS,SnaithRP.TheHospitalAnxietyAndDepressionScale.ActaPsychiatrScand1983;67:361-370

21. BjellandI,DahlAA,HaugTTandNeckelmannD.ThevalidityoftheHospitalAnxietyandDepressionScale;anupdatedreview.JPsychiatRes2002;52:69-77

22. NasreddineZS, PhillipsNA,BédirianV, CharbonneauS, WhiteheadV,Collin I, Cummings JL, ChertkowH. TheMontreal CognitiveAssessment,MoCA: a brief screening tool formild cognitive impairment.J AmGeriatr Soc 2005; 53:695–69923. AzoulayE, Schellongowski P, Darmon M, Bauer PR, Benoit D, Depuydt P, et al. The Intensive Care Medicineresearchagendaoncriticallyilloncologyandhematologypatients.IntensiveCareMed2017;43:1366-138224. OstermannM, Ferrando-Vivas P, Gore C, Power S, HarrisonD. Characteristics and outcome of cancer patientsadmittedtotheICUinEngland,Wales,andNorthernIrelandandnationaltrendsbetween1997and2013.CritCareMed2017;45:1668-167625. VaaraST,PettiläV,ReinikainenM,KaukonenKM;FinnishIntensiveCareConsortium.Population-basedincidence,mortality and quality of life in critically ill patients treated with renal replacement therapy: a nationwide retrospectivecohortstudyinFinnishintensivecareunits.CritCare2012;16:R1326. CocaSG,YusufB,ShlipakMG,GargAX,ParikhCR.Long-termriskofmortalityandotheradverseoutcomesafteracutekidneyinjury:asystematicreviewandmeta-analysis.AmJKidneyDis2009;53:961-97327. GallagherM,CassA,BellomoR,FinferS,GattasD,LeeJ,etal;POST-RENALStudyInvestigatorsandtheANZICSClinicalTrialsGroup.Long-termsurvivalanddialysisdependencyfollowingacutekidneyinjuryinintensivecare:extendedfollow-upofarandomizedcontrolledtrial.PLoSMed2014;11:e100160128. ContiM,MerlaniP,RicouB.Prognosisandqualityoflifeofelderlypatientsafterintensivecare.SwissMedWkly2012;142:w1367129. FlaattenH,Garrouste-OrgeasM.TheveryoldICUpatient:aneverendingstory.IntensiveCareMed.2015;41:1996-199830. LeblancG,BoumendilA,GuidetB.Tenthingstoknowaboutcriticallyillelderlypatients.IntensiveCareMed2017;43:217-21931. FlaattenH,deLangeDW,ArtigasA,BinD,MorenoR,ChristensenS,JoyntGM,BagshawSM,SprungCL,BenoitD,SoaresM,GuidetB.Thestatusof intensivecaremedicineresearchanda futureagendaforveryoldpatients in the ICU.IntensiveCareMed2017;43:1319-132832. KarakusA,HaasLEM,BrinkmanS,deLangeDW,deKeizerNF.Trendsinshort-termand1-yearmortality inveryelderly intensive care patients in the Netherlands: a retrospective study from 2008 to 2014. Intensive CareMed 2017;43:1476-148433. KaukonenKM,BaileyM,SuzukiS,PilcherD,BellomoR.MortalityrelatedtoseveresepsisandsepticshockamongcriticallyillpatientsinAustraliaandNewZealand,2000-2012.JAMA2014;311:1308-131634. WintersBD,EberleinM,LeungJ,NeedhamD,PronovostPJ,SevranskyJE.Long-termmortalityandqualityoflifeinsepsis:asystematicreview.CritCareMed2010;38:1276-128335. WunschH,GuerraC,BarnatoAE,AngusDC,LiG,Linde-ZwirbleWT.Three-yearoutcomesformedicarebeneficiarieswhosurviveintensivecare.JAMA2010;303:849-85636. vanVlietM,VerburgIW,vandenBoogaardM,deKeizerNF,PeekN,BlijlevensNM,PickkersP.Trensinadmissionprevalence, illness severity and survival of haematological patients treated in Dutch intensive care units. Intensive CareMed2014;40:1275-1284

191

37. Azoulay E,Mokart D, Pène F, Lambert J, Kouatchet A,Mayaux J, et al. Outcomes of critically ill patients withhematologicmalignancies:prospectivemulticenterdatafromFranceandBelgium–agroupederechercherespiratoireenréanimationonco-hématologiquestudy.JClinOncol2013;31:2810-280838. van VlietM,van den BoogaardM,Donnelly JP, Evers AW, BlijlevensNM, Pickkers P. Long-term health relatedqualityoflifefollowingintensivecareduringtreatmentforhaematologicalmalignancies.PLoSOne2014;9:e8777939. Normilio-SilvaK,deFigueiredoAC,Pedroso-de-LimaAC,Tunes-da-SilvaG,NunesdaSilvaA,DelgadoDiasLevitesA,etal. Long-termsurival,qualityof life,andquality-adjustedsurvival incritically illpatientswithcancer.CritCareMed2016;44:1327-133740. Azoulay E, SoaresM, DarmonM, Benoit D, Pastores S, Afessa B. Intensive care of the cancer patient: recentachievementsandremainingchallenges.AnnIntensiveCare2011;1:541. AzoulayE,PèneF,DarmonM,LenglinéE,BenoitD,SoaresM,VincentF,BruneelF,PerezP,LemialeV,MokartD.Groupe de Recherche Respiratoire en Réanimation Onco-Hématologique (Grrr-OH). Managing critically ill hematologypatients:Timetothinkdifferently.BloodRev2015;29:359-36742. Hoste EA,Bagshaw SM, BellomoR, Cely CM, Colman R, CruzDN, et al. Epidemiology of acute kidney injury incriticallyillpatients:themultinationalAKI-EPIstudy.IntensiveCareMed2015;41:1411-142343. PickkersP,OstermannM,JoannidisM,ZarbockA,HosteE,BellomoR,etal.Theintensivecaremedicineagendaonacutekidneyinjury.IntensiveCareMed2017;43:1198-120944. BagshawSM,DarmonM,OstermannM,FinkelsteinFO,WaldR,TolwaniAJ,etal.Currentstateoftheartforrenalreplacementtherapyincriticallyillpatientswithacutekidneyinjury.IntensiveCareMed2017;43:841-85445. DeSmedtDM,ElseviersMM,LinsRL,AnnemansL.Economicevaluationofdifferenttreatmentmodalitiesinacutekidneyinjury.NephrolDialTransplant.2012;27:4095-410146. Ethgen0, SchneiderAG,BagshawSM,BellomoR,Kellum JA. Economicsofdialysisdependence following renalreplacementtherapyforcriticallyillacutekidneyinjurypatients.NephrolDialTransplant.2015;30:54-6147. MorschC,ThoméFS,BalbinottoA,GuimaraesJF,BarrosEG.Health-relatedqualityoflifeanddialysisdependenceincriticallyillpatientsurvivorsofacutekidneyinjury.RenFail2011;33:949-95648. Hofhuis JG, van Stel HF, Schrijvers AJ, Rommes JH, Spronk PE. The effect of acute kidney injury on long-termhealth-relatedqualityoflife:aprospectivefollow-upstudy.CritCare2013;17:R1749. Nisula S, Vaara ST, Kaukonen KM, Reinikainen M, Koivisto SP, Inkinen O, Poukkanen M, Tiainen P, Pettilä V,KorhonenAM;FINNAKI-QOLStudyGroup.Six-monthsurvivalandqualityoflifeofintensivecarepatientswithacutekidneyinjury.CritCare2013;17:R25050. WangAY,BellomoR,CassA,FinferS,GattasD,MyburghJ,etal;POST-RENALStudyInvestigatorsandtheANZICSClinicalTrialsGroup.Health-relatedqualityoflifeinsurvivorsofacutekidneyinjury:TheProlongedOutcomesStudyoftheRandomized Evaluation of Normal versusAugmented Level Replacement Therapy study outcomes. Nephrology 2015;20:492-498

51. VilleneuvePM,ClarkEG,SikoraL,SoodMM,BagshawSM.Health-relatedqualityoflifeamongsurvivorsofacuteinjuryintheintensivecareunit:asystematicreview.IntensiveCareMed2016;42:137-14652. MildhH,PettiläV,KorhonenAM,KarlssonS,Ala-KokkoT,ReinikainenM,VaaraST;FINNAKIStudyGroup.Three-yearmortalityin30-daysurvivorsofcriticalcarewithacutekidneyinjury:datafromtheprospectiveobservationalFINNAKIstudy.AnnIntensiveCare2016;6:11853. Forni LG,DarmonM,OstermannM,Oudemans-van StraatenHM, Pettilä V, Prowle JR, SchetzM, JoannidisM.Renalrecoveryafteracutekidneyinjury.IntensiveCareMed2017;43:855-86654. CommereucM,GuérotE,Charles-NelsonA,ConstanA,KatsahianS,SchortgenF.ICUpatientsrequiringrenalreplacementtherapyInitiation:Fewersurvivorsandmoredialysisdependentsfrom80yearsold.CritCareMed2017;45:e772-e781

192

55. Stoumpos S, Mark PB, McQuarrie EP, Traynor JP, Geddes CC. Continued monitoring of acute kidney injury

survivorsmightnotbenecessaryinthoseregaininganestimatedglomerularfiltrationrate>60 mL/minat1year.Nephrol

DialTransplant20171;32:81-88

56. BagshawSM,StelfoxHT,JohnsonJA,McDermidRC,RolfsonDB,TsuyukiRT,etal.Long-termassociationbetween

frailtyandhealth-relatedqualityoflifeamongsurvivorsofcriticalillness:aprospectivemulticentercohortstudy.CritCare

Med2015;43:973-982

57. Heyland DK, Garland A, Bagshaw SM, Cook D, Rockwood K, Stelfox HT, et al. Recovery after critical illness in

patientsaged80yearsorolder:amulti-centerprospectiveobservationalcohortstudy.IntensiveCareMed2015;41:1911-

1920

58. KhouliH,AstuaA,DombrowskiW,AhmadF,HomelP,ShapiroJ,etal.Changesinhealth-relatedqualityoflifeand

factorspredictinglong-termoutcomesinolderadultsadmittedtointensivecareunits.CritCareMed2011;39:731-737

59. RochA,WiralusS,PaulyV,ForelJM,GuervillyC,GainnierM,etal.Long-termoutcomeinmedicalpatientsaged

80oroverfollowingadmissiontoanintensivecareunit.CritCare2011;15:R36

60. SacanellaE,Pérez-CastejonJM,NicolasJM,MasanesF,NavarroM,CastroP,etal.Functionalstatusandqualityof

life12monthsafterdischargefromamedicalICUinhealthyelderlypatients:aprospectiveobservationalstudy.CritCare

2011;15:R105

61. AndersenFH,FlaattenH,KlepstadP,RomildU,KvåleR.Long-termsurvivalandqualityoflifeafterintensivecare

forpatients80yearsofageorolder.AnnIntensiveCare2015;5:13

62. Daubin C, Chevalier S, Séguin A, Gaillard C, Valette X, Prévost F, et al. Predictors ofmortality and short-term

physicalandcognitivedependenceincriticallyillpersons75yearsandolder:aprospectivecohortstudy.HealthQualLife

Outcomes2011;9:35

63. HofhuisJG,vanStelHF,SchrijversAJ,RommesJH,SpronkPE.Changesofhealth-relatedqualityoflifeincritically

illoctogenarians.Chest2011;140:1473-1483

64. TabahA,PhilippartF,TimsitJF,WillemsV,FrançaisA,LeplègeA,etal.Qualityoflifeinpatientsaged80orover

afterICUdischarge.CritCare2010;14:R2

65. FerranteLE,PisaniMA,MurphyTE,GahbauerEA,Leo-SummersLS,GillTM.Functionaltrajectoriesamongolder

personsbeforeandaftercriticalillness.JAMAInternMed2015;175:523-529

66. MontuclardL,Garrouste-OrgeasM,TimsitJF,MissetB,DeJongheB,CarletJ:Outcome,functionalautonomy,and

qualityoflifeofelderlypatientswithalong-termintensivecareunitstay.CritCareMed2000;28:389-3395

67. KaarlolaA,PettiläV,KekkiP:Qualityoflifesixyearsafterintensivecare.IntensiveCareMed2003;29:1294-1299

68. Deimling GT, Kahana B, Bowman KF, SchaeferML. Cancer survivorship and psychological distress in later life.

Psycho-Oncology2002;11:479-494

69. DejaM,DenkeC,Weber-CarstensS,etal: Social supportduring intensivecareunit staymight improvemental

impairmentandconsequentlyhealth-relatedqualityoflifeinsurvivorsofsevereacuterespiratorydistresssyndrome.Crit

Care2006;10:R147

70. CombesA,CostaMA,TrouilletJL,BaudotJ,MokhtariM,GibertC,ChastreJ:Morbidity,mortality,andqualityof

lifeoutcomesofpatientsrequiring≥14daysofmechanicalventilation.CritCareMed2003;31:1373-1381

71. KaarlolaA,TallgrenM,PettiläV:Long-termsurvival,qualityoflife,andquality-adjustedlife-yearsamongcritically

illelderlypatients.CritCareMed2006;2120-2126

72. NielssonMS, Christiansen CF, JohansenMB, Rasmussen BS, Tønnesen E, NørgaardM.Mortality in elderly ICU

patients:acohortstudy.ActaAnaesthesiolScand2014;58:19-26

73. Ulvik A, Kvåle R,Wentzel-Larsen T, FlaattenH:Quality of life 2-7 years aftermajor trauma. Acta Anaesthesiol

Scand2008;52:195-201

193

74. CuthbertsonBH,ScottJ,StrachanM,KilonzoM,ValeL:Qualityoflifebeforeandafterintensivecare.Anaesthesia2005;60:332-33975. JoyceVR, SmithMW, Johansen KL,UnruhML, SirokaAM,O'Connor TZ, Palevsky PM, VeteranAffairs/NationalInstitutesofHealthAcuteRenalFailureTrialN.Health-relatedqualityoflifeasapredictorofmortalityamongsurvivorsofAKI.ClinJAmSocNephrol.2012;7:1063-107076. Irribarren-Diarasarri S, Aizpuru-Barandiaran F, Munoz-Martinez T, Loma-Osorio A, Hernandez-Lopez M, Ruiz-ZorrillaJM,etal:Health-relatedqualityoflifeasaprognosticfactorofsurvivalincriticallyillpatients.IntensiveCareMed2009;35:833-83977. HofhuisJG,SpronkPE,vanStelHF,SchrijversAJ,BakkerJ:Qualityoflifebeforeintensivecareunitadmissionisapredictorofsurvival.CritCare2007;11:R7878. AbelhaFJ,SantosCC,BarrosH:Qualityoflifebeforesurgicaladmission.BMCSurgery2007;7:2379. GrafJ,MühlhoffC,DoigGS,etal:Healthcarecosts,long-termsurvival,andqualityoflifefollowingintensivecareunitadmissionaftercardiacarrest.CritCare2008;12:R9280. Orwelius L,NordlundA,NordlundP, Edéll-GustafssonU, Sjöberg F: Prevalenceof sleepdisturbances and long-termreducedhealth-relatedqualityoflifeaftercriticalcare:aprospectivemulticentercohortstudy.CritCare2008;12:R9781. Veerbeeck JM,KwakkelG, vanWegenEH,Ket JCF,HeymansMW.Earlypredictionofoutcomesof activitiesofdailylivingafterstroke:asystematicreview.Stroke2011;42:1482-148882. HeylandDK,StelfoxHT,GarlandA,CookD,DodekP,KutsogiannisJ,JiangX,TurgeonAF,DayAG;CanadianCriticalCareTrialsGroupand theCanadianResearchers at theEndof LifeNetwork.PredictingPerformanceStatus1YearAfterCriticalIllnessinPatients80YearsorOlder:DevelopmentofaMultivariableClinicalPredictionModel.CritCareMed2016;44:1718-172683. OrweliusL,NordlundA,NordlundP,SimonssonE,BackmanC,SamuelssonA,SjobergF.Pre-existingdisease:themostimportantfactorforhealthrelatedqualityoflifelongtermaftercriticalillness:aprospective,longitudinal,multicentertrial.CritCare2010;14:R6784. ZampieriFG,BozzaFA,MoralezGM,MazzaDD,ScottiAV,SantinoMS,etal.Theeffectsofperformancestatusoneweekbeforehospitaladmissionontheoutcomesofcriticallyillpatients.IntensiveCareMed2017;43:39-4785. OrweliusL,BackmanC,FredriksonM,SimonssonE,NordlundP,SamuelssonA,SjobergF.Social integration:animportantfactorforhealth-relatedqualityoflifeaftercriticalillness.IntensiveCareMed2011;5:831-83886. OrweliusL,LoboC,TeixeiraPintoA,CarneiroA,Costa-PereiraA,GranjaC.Sepsispatientsdonotdifferinhealth-relatedqualityoflifecomparedwithotherICUpatients.ActaAnaesthesiolScand.2013;57:1201-120587. LunaCM,PalmaI,NiedermanMS,MembrianiE,GioviniV,WiemkenTL,PeyraniP,RamirezJ.TheImpactofageandcomorbiditiesonthemortalityofpatientsofdifferentagegroupsadmittedwithcommunity-acquiredpneumonia.AnnAmThoracSoc2016;13:1519-152688. RockwoodK,SongX,MacKnightC,BergmanH,HoganDB,McDowell I,MitnitskiA.Aglobalclinicalmeasureoffitnessandfrailtyinelderlypeople.CMAJ.2005;173:489-49589. FlaattenH,DeLangeDW,MorandiA,AndersenFH,ArtigasA,BertoliniG,etal;VIP1studygroup.TheimpactoffrailtyonICUand30-daymortalityandthelevelofcareinveryelderlypatients(≥80years).IntensiveCareMed2017;43:1820-182890. PiersRD,AzoulayE,RicouB,DekeyserGF,DecruyneaereJ,MaxA,etal.PerceptionsofappropriatenessofcareamongEuropeanandIsraeliintensivecareunitnursesandphysicians.JAMA2011;306:2694-2670391. LeMaguetP,RoquillyA,LasockiS,AsehnouneK,CariseE,SaintMartinM,etal.PrevalenceandimpactoffrailtyonmortalityinelderlyICUpatients:aprospective,multicentre,observationalstudy.IntensiveCareMed2014;40:674-68292. BagshawM,MajumdarSR,RolfsonDB,IbrahimQ,McDermidRC,StelfoxHT.Aprospectivemulticentercohortstudyoffrailtyinyoungercriticallyillpatients.CritCare2016;20:175

194

93. Montgomery C, Bagshaw SM. Frailty in the age of VIPs (very old intensive care patients). Intensive CareMed2017;43:1887-188894. Muscedere J,WatersB,VaramballyA,BagshawSM,Boyd JG,MasloveD, Sibley S, RockwoodK. The impactoffrailtyonintensivecareunitoutcomes:asystematicreviewandmeta-analysis.IntensiveCareMed2017;43:1105-112295. BrummelNE,BellSP,GirardTD,PandharipandePP,JacksonJC,MorandiA,etal.Frailtyandsubsequentdisabilityandmortalityamongpatientswithcriticalillness.AmJRespirCritCareMed.2017;196:64-7296. LatronicoN,HerridgeM,HopkinsRO,AngusD,HartN,HermansG,etal.TheICMresearchagendaonintensivecareunit-acquiredweakness.IntensiveCareMed2017;43:1270-128197. MinneL,LudikhuizeJ,deJongeE,deRooijS,Abu-HannaA.PrognosticmodelsforpredictingmortalityinelderlyICUpatients:asystematicreview.IntensiveCareMed2011;37:1258-126898. BallIM,BagshawSM,BurnsKE,CookDJ,DayAG,DodekPM,etal.Aclinicalpredictiontoolforhospitalmortalityincriticallyillelderlypatients.JCritCare2016;35:206-21299. Soliman IW, Frencken JF, Peelen LM, Slooter AJ, Cremer OL, van Delden JJ, van Dijk D, de Lange DW. Thepredictivevalueofearlyacutekidneyinjuryforlong-termsurvivalandqualityoflifeofcriticallyillpatients.CritCare2016;20:242100. KahnJM.Predictingoutcomeincriticalcare:past,presentandfuture.CurrOpinCritCare2014;20:542-543101. WyshamNG,AbernethyAP,CoxCE.Settingthevision:appliedpatient-reportedoutcomesandsmart,connecteddigitalhealthcaresystemstoimprovepatient-centeredoutcomespredictionincriticalillness.CurrOpinCritCare2014;20:566-572102. PutmanMS,TakHJ,CurlinFA,YoonJD.Qualityoflifeandrecommendationsforfurthercare.CritCareMed2016;44:1996-2002103. Soliman IW, Cremer OL, de Lange DW, Slooter AJC, van Delden JHJM, van Dijk D, Peelen LM. The ability ofintensivecareunitphysicianstoestimatelong-termprognosisinsurvivorsofcriticalillness.JCritCare2018;43:148-155104. Labarère J,Renaud B,FineMJ. How to derive and validate clinical predictionmodels for use in intensive caremedicine.IntensiveCareMed2014;40:513-527105. CeliLA,CseteM,StoneD.Optimaldatasystems:thefutureofclinicalpredictionsanddecisionsupport.CurrOpinCritCare2014;20:573-580106. AzoulayE,VincentJL,AngusDC,ArabiYM,BrochardL,BrettSJ,etalRecoveryaftercriticalillness:puttingthepuzzletogether-aconsensusof29.CritCare2017;21:296107. HickeyA,BarkerM,McGeeH,O’BoyleC.Measuringhealth-relatedqualityoflifeinolderpatientpopulations:areviewofcurrentapproaches.Pharmacoeconomics2005;23:971-993108. ZiegelsteinRC.Personomics:TheMissingLinkintheEvolutionfromPrecisionMedicinetoPersonalizedMedicine.JPersMed2017;7:11109. LeGuenJ,BoumendilA,GuidetB,CorvolA,Saint-JeanO,SommeD.Areelderlypatients'opinionssoughtbeforeadmissiontoanintensivecareunit?ResultsoftheICE-CUBstudy.AgeAgeing.2016;45:303-309110. Philippart F,VesinA, Bruel C, Kpodji A,Durand-GasselinB,GarçonP, et al. The ETHICA study (part I): elderly'sthoughtsaboutintensivecareunitadmissionforlife-sustainingtreatments.IntensiveCareMed2013;39:1565-1573111. Garrouste-OrgeasM,TabahA,VesinA,PhilippartF,KpodjiA,BruelC,GrégoireC,MaxA,TimsitJF,MissetB.TheETHICA study (part II): simulation studyofdeterminantsandvariabilityof ICUphysiciandecisions inpatientsaged80orover.IntensiveCareMed2013;39:1574-1583112. Heyland DK,Dodek P, Mehta S, Cook D, Garland A, Stelfox HT, et al. Canadian Critical Care Trials Group andCanadianResearchersatEndofLifeNetwork (CARENET).Admissionof theveryelderly to the intensivecareunit: familymembers'perspectivesonclinicaldecision-makingfromamulticentercohortstudy.PalliatMed2015;29:324-335

195

113. BlackN.Patientreportedoutcomemeasurescouldhelptransformhealthcare.BMJ2013;346:f167114. VincentJL,CreteurJ.Isthiscriticallyillpatientgoingtosurvive?IntensiveCareMed2016;42:426-428115. CollinsGS,deGrootJA,DuttonS,OmarO,ShanyindeM,TajarA,VoyseyM,WhartonR,YuLM,MoonsKG,AltmanDG.Externalvalidationofmultivariablepredictionmodels:asystematicreviewofmethodologicalconductandreporting.BMCMedResMethodol2014;14:40116. ElliottD,DavidsonJE,HarveyMA,Bemis-DoughertyA,HopkinsRO,IwashynaTJ,etal.Exploringthescopeofpost-intensive care syndrome therapy and care: engagement of non-critical care providers and survivors in a secondstakeholdersmeeting.CritCareMed2014;42:2518-2526117. Davidson JE,Jones C, Bienvenu OJ. Family response to critical illness: postintensive care syndrome-family. CritCareMed2012;40:618-624118. HarveyMA,DavidsonJE.PostintensiveCareSyndrome:RightCare,RightNow…andLater.CritCareMed2016;44:381-385119. Mistraletti G, Umbrello M, Mantovani ES, Moroni B, Formenti P, Spanu P, et al; http://www.intensiva.itInvestigators.Afamily informationbrochureanddedicatedwebsitetoimprovetheICUexperienceforpatients'relatives:anItalianmulticenterbefore-and-afterstudy.IntensiveCareMed2017;43:69-79120. BalasMC, Vasilevskis EE, Olsen KM, Schmid KK, Shostrom V, CohenMZ, et al. Effectiveness and safety of theawakening and breathing coordination, deliriummonitoring/management, and early exercise/mobility bundle. Crit CareMed2014;42:1024-1036121. Barnes-DalyMA, Phillips G, Ely EW. Improving Hospital Survival and Reducing Brain Dysfunction at SevenCalifornia Community Hospitals: Implementing PADGuidelines Via the ABCDEF Bundle in 6,064 Patients. Crit CareMed2017;45:171-178122. MorandiA,PivaS,ElyEW,MyatraSN,SalluhJIF,AmareD,etal.WorldwideSurveyofthe"AssessingPain,BothSpontaneousAwakeningandBreathingTrials,ChoiceofDrugs,DeliriumMonitoring/Management,EarlyExercise/Mobility,andFamilyEmpowerment"(ABCDEF)Bundle.CritCareMed2017;45:e1111-e1122123. KressJP,HallJB.ICU-acquiredweaknessandrecoveryfromcriticalillness.NEnglJMed2014;370:1626-1635124. DavidsonJE,AslaksonRA,LongAC,PuntilloKA,KrossEK,HartJ,etal.GuidelinesforFamily-CenteredCareintheNeonatal,Pediatric,andAdultICU.CritCareMed2017;45:103-128125. CapuzzoM,VoltaC,TassinatiT,MorenoR,ValentinA,GuidetB,etal;WorkingGrouponHealthEconomicsoftheEuropeanSocietyofIntensiveCareMedicine.HospitalmortalityofadultsadmittedtoIntensiveCareUnitsinhospitalswithandwithoutIntermediateCareUnits:amulticentreEuropeancohortstudy.CritCare2014;18:551126. WaydhasC,HertingE,KlugeS,MarkewitzA,MarxG,MuhlE,etal.Intermediatecareunits:Recommendationsonfacilities and structure.Med Klin Intensivmed Notfmed; 2017: Nov 7. doi: 10.1007/s00063-017-0369-7. [Epub ahead ofprint]127. Griffiths JA, Barber VS, Cuthbertson BH, Young JD. A national survey of intensive care follow-up clinics.Anaesthesia2006;61:950-955128. Cuthbertson BH, Rattray J, Campbell MK, Gager M, Roughton S, Smith A, et al; PRaCTICaL study group. ThePRaCTICaLstudyofnurseled,intensivecarefollow-upprogrammesforimprovinglongtermoutcomesfromcriticalillness:apragmaticrandomisedcontrolledtrial.BMJ.2009;339:b3723129. JensenJF,EgerodI,BestleMH,ChristensenDF,ElklitA,HansenRL,KnudsenH,GrodeLB,OvergaardD.Arecoveryprogramtoimprovequalityoflife,senseofcoherenceandpsychologicalhealthinICUsurvivors:amulticenterrandomizedcontrolledtrial,theRAPITstudy.IntensiveCareMed2016;42:1733-1743130. Elliott D,McKinley S, Alison J, Aitken LM, KingM, LeslieGD, Kenny P, Taylor P, Foley R, Burmeister E. Health-relatedqualityof life andphysical recovery after a critical illness: amulti-centre randomised controlled trial of ahome-basedphysicalrehabilitationprogram.CritCare2011;15:R142

196

131. WalshTS,SalisburyLG,Merriweather JL,Boyd JA,GriffithDM,HubyG,etal;RECOVER Investigators. IncreasedHospital-Based Physical Rehabilitation and Information Provision After Intensive Care Unit Discharge: The RECOVERRandomizedClinicalTrial.JAMAInternMed2015;175:901-910132. JensenJF,ThomsenT,OvergaardD,BestleMH,ChristensenD,EgerodI.Impactoffollow-upconsultationsforICUsurvivorsonpost-ICUsyndrome:asystematicreviewandmeta-analysis.IntensiveCareMed2015;41:763-775133. Egerod I,RisomSS,ThomsenT, Storli SL,EskerudRS,HolmeAN,SamuelsonKA. ICU-recovery inScandinavia:acomparativestudyofintensivecarefollow-upinDenmark,NorwayandSweden.IntensiveCritCareNurs2013;29:103-111134. Van Der Schaaf M, Bakhshi-Raiez F, Van Der Steen M, Dongelmans DA, De Keizer NF. Recommendations forintensivecare follow-upclinics; report froma surveyandconferenceofDutch intensivecares.MinervaAnestesiol2015;81:135-144135. KjerCKW,EstrupS,PoulsenLM,MathiesenO.Follow-upafterintensivecaretreatment:aquestionnairesurveyofintensivecareaftercareinDenmark.ActaAnaesthesiolScand.2017;61:925-934136. GlimeliusPeterssonC,BergbomI,BrodersenK,RingdalM.Patients'participationinandevaluationofafollow-upprogramfollowingintensivecare.ActaAnaesthesiolScand2011;55:827-834137. MikkelsenME,JacksonJC,HopkinsRO,ThompsonC,AndrewsA,NetzerG,etal.PeersupportasanovelstrategytomitigatePost-IntensiveCareSyndrome.AACNAdvCritCare2016;27:221-229138. Transplantouxorganisation:https://www.transplantoux.be/(lastassessedApril2018)139. PinskyMR,ValentinA,RubenfeldG.IntensiveCareMedicinein2050:cost-effectivenessanalysis. IntensiveCareMed2017;43:1039-1040140. GuidetB,vanderVoortPHJ,CsomosA.Intensivecarein2050:healthcareexpenditure.IntensiveCareMed2017;43:1141-1143

197

Addendum

199

I.Listofabbreviations

• ABCDEFGHbundle Airwayandawakeningmanagement,spontaneousbreathingtrials,

coordinationofcareandcommunication,deliriumassessmentand

treatment,earlymobilization,familyinvolvement,goodhandoff

communication,andhandoutmaterialforPICSandPICS-F

• ADL activitiesofdailyliving

• AKI acutekidneyinjury

• APACHEII AcutePhysiologyandChronicHealthEvaluationII

• CFS ClinicalFrailtyScore

• CKD chronickidneydisease

• COSI CostsandOutcomeStudyintheICU

• D1 day1=first24hoursofICUadmission

• DNR do-not-resuscitate

• EOL end-of-life

• EQ-5D EuroQol-5Dimensions

• ESICM EuropeanSocietyofIntensiveCareMedicine

• ESKD end-stagekidneydisease

• HADS HospitalAnxietyandDepressionScale

• HRQOL health-relatedqualityoflife

• ICU intensivecareunit

• ICU-AW intensivecareunit-acquiredweakness

• IRC intensivecarerecoverycenter

• IZ IntensieveZorg

• LOS lengthofstay

• MoCA MontrealCognitiveAssessmenttest

• MOS MedicalOutcomesStudy

• NEMS NineEquivalentofNursingManpowerUsescore

• NHP NottinghamHealthProfile

• PICS post-intensivecaresyndrome

• PICS-F post-intensivecaresyndrome-family

• PTSD post-traumaticstressdisorder

• PTSS-14 Post-traumaticStressSyndrome14-questionsinventory

• QOL qualityoflife

200

• QWB QualityofWell-Being

• RAND-36 RAND-36-itemHealthSurvey

• RRT renalreplacementtherapy

• SCCM SocietyofCriticalCareMedicine

• SF-36 MedicalOutcomesStudy36-itemShortFormHealthSurvey

• SIP SicknessImpactProfile

• SOFA SequentialOrganFailureAssessment

• TISS-28 TherapeuticInterventionScoringSystem-28

• UI utilityindex

• UIb utilityindexatbaseline(=2weeksbeforeICUadmission)

• UI1y utilityindex1yearafterICUdischarge

• VAS visualanaloguescale

• VASb visualanaloguescaleatbaseline(=2weeksbeforeICUadmission)

• VAS1y visualanaloguescale1yearafterICUdischarge

201

II.ConciseCurriculumVitae

PERSONALIAName: OEYENSandraGermaineRaymondaBorn: Antwerp,Belgium,January15th1970Civilstate: MarriedwithAlainSmetsHomeaddress: Beekstraat116,9800Astene,BelgiumWorkaddress: GhentUniversityHospital DepartmentofIntensiveCare1K12IC C.Heymanslaan10,9000Ghent,BelgiumPositiontitle: MD StaffmemberoftheDepartmentofIntensiveCare GhentUniversityHospital Ghent,BelgiumTelephone: +3292822500 home +3293326316 work +32478467555 mobileE-mail: [email protected] 1-35677-26-100DEGREESANDEDUCATION

Institutionandlocation Degree Year Fieldofstudy

KoninklijkAtheneumMalle,Malle,Belgium Diplomaofsecondaryschool 1982-1988 Latijn-Wetenschappen

GhentUniversity,Ghent,Belgium MD,withgreatdistinction 1988-1995 Medicine

GhentUniversity,Ghent,Belgium Certificate 1997 Advanced

AnesthesiologyGhentUniversity,Ghent,

Belgium Anesthesiologist 1995-2000 Anesthesiology

GhentUniversity,Ghent,Belgium Criticalcarephysician 2000-2001 CriticalCareMedicine

GhentUniversity,Ghent,Belgium Certificate 2000 EmergencyMedicine

202

POSTGRADUATECOURSES

Institutionandlocation Course Year Fieldofstudy

SocietyofMedicalDecisionMaking,Atlanta,USA

Causalinferenceandcausaldiagramsinmedicaldecision

making2004 Statistics

HospitalErasme,Brussels,Belgium

Cardiovascularandrespiratoryphysiology 2004 Criticalcare

SocietyofMedicalDecisionMaking,Boston,USA

Changingphysicianbehaviour 2006 Evidencebased

medicine

VlerickSchoolGent-Leuven Financialmanagementinhospitals 2007 Economics

GhentUniversity,Ghent,Belgium Statistics 2007 Statistics

GhentUniversity,Ghent,Belgium

StatisticalanalysiswithPASW18 2010 Statistics

GhentUniversity,Ghent,Belgium

Multivariateanalysisandlogisticregression 2012 Statistics

Medicalevaluationtechnologyassessment,

Ghent,Belgium

Economicevaluationsinhealthscience 2014 Health-economics

GhentUniversity,Ghent,Belgium Trainthetrainer 2016 Management

EXPERIENCEINCLINICALTRIALS

• Experienceassub-investigatorinseveralmulticenterandinternationalstrials(phaseII-IV)inthefieldofsepsis,ARDSandinfectiology

• PrincipalinvestigatoroftheLIPOSTMstudy(GSK)(severesepsistrial)2005-2006• PrincipalinvestigatoroftheACCESSstudy(severesepsistrial)2009-2010• PrincipalinvestigatoroftheOasisstudy(severesepsistrial)2011-2012• CountryCoordinatorforBelgiumfortheEloisestudy(2013),endorsedbyESICM(principal

investigatorMauriziaCapuzzo) • Country Coordinator for Belgium for the VIP1 study (2016), endorsed by ESICM (principal

investigatorHansFlaatten)• Country Coordinator for Belgium for the VIP2 study (2018), endorsed by ESICM (principal

investigatorHansFlaatten)PROFESSIONALMEMBERSHIP

• EuropeanSocietyofIntensiveCareMedicine

203

EDUCATIONALTASKS

• Teachingpathophysiologyinthe3rdyearMedicine2002-2011• “HemodynamicmonitoringandshockintheICU”;Continuingeducationofphysicianandnursing

staff• “Long-termoutcomes”;Continuingeducationofphysicianandnursingstaff• “VasopressorsintheICU”;Teachinginthe7thyearMedicine:2007-ongoing• “Long-termoutcomes”;Teachinginthe7thyearMedicine:2007-ongoing• “Outcomes,qualityoflife,scoringsystems”;TeachingintheInteruniversitypostgraduatecourse

criticalcaremedicine:2013-ongoing• Reviewerfunctionindifferentcriticalcarejournals:CriticalCareMedicine,IntensiveCareMedicine,

CriticalCare,JournalofCriticalCare,BritishMedicalJournalA1PUBLICATIONS

• Adherence to and efficacy and safety of an insulin protocol in the critically ill: A prospectiveobservationalstudy.OeyenSG,HosteEA,RoosensCD,DecruyenaereJM,BlotSI.AJCC2007;16:599-608

• Long-termoutcomeafteracutekidneyinjuryincriticallyillpatients.

OeyenS,VandijckD,BenoitD,DecruyenaereJ,AnnemansL,HosteE.ActaClinBelg.2007;62(Suppl2):337-340

• Acutekidneyinjury,lengthofstay,andcostsinpatientshospitalizedintheintensivecareunit.

DMVandijck,SOeyen.JMDecruyenaere,LAnnemans,EAHoste.ActaClinBelg.2007;62(Suppl2):341-345

• Daily cost of antimicrobial therapy in patients with intensive care unit-acquired, laboratory-

confirmedbloodstreaminfection.VandijckDM,DepaemelaereM,LabeauSO,DepuydtPO,AnnemansL,BuyleFM,OeyenS,ColpaertKE,PelemanRP,BlotSI,DecruyenaereJM.IntJAntimicrobAgents2008;31:161-165

• Hyperglycemia upon Onset of ICU-acquired Bloodstream Infection is Associated with Adverse

OutcomeinaMixedICUPopulation.VandijckDM,OeyenS,BuyleFM,ClausBO,BlotSI,DecruyenaereJM.AnaesthIntensiveCare2008;36:25-29.

• A50-yearoldmanwithseverehypercalcemia:acasereport.

KVandenHauwe,SGOeyen,BFScrijvers,JMDecruyenaere,WABuylaert.ActaClinBelg2009;64:442-446

• Qualityoflifeafterintensivecare:Asystematicreviewoftheliterature.

OeyenSG,VandijckDM,BenoitDD,AnnemansL,DecruyenaereJM.CritCareMed2010;38:2386-2400

204

• Long-term outcomes and quality of life in critically ill patients with hematological or solidmalignancies:asinglecenterstudy.Oeyen SG, Benoit DD, Annemans L, Depuydt PO, Van Belle SJ, Troisi RI, Noens LA, Pattyn P,DecruyenaereJM.IntensiveCareMed2013;39:889-898

• Effect of eritoran, an antagonist ofMD2-TLR4, onmortality in patientswith severe sepsis: the

Accessrandomizedtrial.OpalSM,LaterrePF,FrancoisB,LaRosaSP,AngusDC,MiraJP,WitteboleX,DugernierT,PerrotinD,TidswellM, Jauregui L,KrellK,Pachl J,TakahashiT,PeckelsenC,CordascoE,ChangCS,OeyenS,AikawaN,MaruyamaT,ScheinR,KalilAC,VanNuffelenM,LynnM,RossignolDP,GogateJ,RobertsMB,WheelerJL,VincentJL;ACCESSStudyGroup.JAMA2013;309:1154-1162

• Lowserumcreatinekinaseisassociatedwithworseoutcomeincriticallyillpatients.

VanDeMoortelL,SpeeckaertM,FiersT,OeyenS,DecruyenaereJ,DelangheJJCritCare2014;29(5):786-790

• Hospital mortality of adults admitted to Intensive Care Units in hospitals with and without

IntermediateCareUnits:amulticentreEuropeancohortstudy.CapuzzoM,VoltaC,TassinatiT,MorenoR,ValentinA,GuidetB,etalCritCare2014;18:551

• Oraltalactoferrininseveresepsisstudyinvestigators.

VincentJL,MarshallJC,DellingerRP,SimonsonSG,GuntupalliK,LevyMM,SingerM,MalikR.CritCareMed2015;43:1832-1838

• Influenceofsmartreal-timeelectronicalertingonglucosecontrolincriticallyillpatients.

ColpaertK,OeyenS,SijnaveB,PelemanR,BenoitD,DecruynaereJ.JCritCare2015;30:216

• Long-term quality of life in critically ill patients with acute kidney injury treated with renal

replacementtherapy:amatchedcohortstudy.OeyenS,DeCorteW,BenoitD,AnnemansL,DhondtA,VanholderR,DecruynaereJ,HosteE.CritCare2015;19:289

• Long-termoutcomeandhealth-relatedqualityoflifeindifficult-to-weanpatientswithorwithout

ventilatordependencyatICUdischarge:aretrospectivecohortstudy.DepuydtP,OeyenS,DeSmetS,DeRaedtS,BenoitD,DecruyenaereJ,DeromE.BMCPulmMed2016;27:133

• Critically ill octogenarians andnonagenarians: Evaluationof long-termoutcomes, post-hospital

trajectories,andqualityoflifeoneyearandsevenyearsafterICUdischarge.OeyenS,VermassenJ,PiersR,BenoitD,AnnemansL,DecruyenaereJ.MinervaAnestesiol2017,83:598-609

• TheimpactoffrailtyonICUand30-daymortalityandthelevelofcareinveryelderlypatients(≥

80years).FlaattenH,DeLangeDW,MorandiA,AndersenFH,ArtigasA,BertoliniG,BoumendilA,CecconiM,Christensen S, Faraldi L, Fjølner J, Jung C, Marsh B, Moreno R, Oeyen S, Öhman CA, Pinto BB,SolimanIW,SzczeklikW,ValentinA,WatsonX,ZaferidisT,GuidetB;VIP1studygroup.

205

IntensiveCareMed2017;43:1820-1828

• Developmentofapredictionmodelforlong-termqualityoflifeincriticallyillpatients.SandraOeyen,KarelVermeulen,DominiqueBenoit,LievenAnnemans,JohanDecruyenaere.JCritCare2018;43:133-138

• Withholdingorwithdrawingof life-sustainingtherapy inolderadultpatients (≥80years)admitted to the intensive care unit. B Guidet, H Flaatten, A Boumendil, A Morandi, FHAndersen, A Artigas, G Bertolini, M Cecconi, S Christensen, L Faraldi, J Fjølner, C Jung, BMarsh,RMoreno, SOeyen,CAÖhman,BBPinto18; IWSoliman,WSzczeklik,AValentin, XWatson,TZafeiridis,DWDeLange;OnbehalfoftheVIP1studygroup.IntensiveCareMed;2018May17.doi:10.1007/s00134-018-5196-7[Epubaheadofprint]

• Influence of neutropenia onmortality of critically ill cancer patients: Results of ameta-

analysisonindividualdata.Georges Quentin, Azoulay Elie, Mokart Djamel, Soares Marcio, Jeon Kyeongman, SandraOeyen,etal.AcceptedforpublicationinCriticalCare

• Development of a simplified geriatric score predictingmortality in elderly patients (≥ 80years)whoareacutelyadmittedtotheIntensiveCareUnitsinEurope.DWDeLange,SBrinkman,HFlaatten,ABoumendil,AMorandi,FHAndersen,AArtigas,GBertolini,MCecconi,SChristensen,LFaraldi,JFjølner,CJung,BMarsh,RMoreno,SOeyen,CAÖhman,etal;OnbehalfoftheVIP1studygroup.Submitted

• Hugevariationinobtainingethicalpermissionforanon-interventionalobservationalstudy

inEurope.DeLangeD,GuidetB,AndersenFH,ArtigasA,BertoliniG,MorenoR,ChristensenS,CecconiM,Agvald-OhmanC,GradisekP,JungC,MarshBJ,OeyenS,etal.Submitted

EDITORIALS

• Admissionhyperglycemiaandoutcome:Theongoingstory.OeyenS.CritCareMed2005;33(12):2848-2849

• Aboutprotocolsandguidelines:It’stimetoworkinharmony!

OeyenS.CritCareMed2007;35(1):292-293

• Freshfrozenplasmatransfusioninthecriticallyill:Yes,noormaybe?OeyenS.CritCareMed2007;35(7):1777-1778

• Closingthegapbetweenknowledgeandbehavior:Missionimpossible?

OeyenS.CritCareMed2007;35(9):2219-2220

• Doyou(still)believeintightbloodglucosecontrol?OeyenS.CritCareMed2008;36(12):3277-3278

206

OTHERPUBLICATIONS

• Cost-effectivenessincriticalcare.VandijckD,AnnemansL,OeyenS,BlotSI,DecruyenaereJM

ICUManagement2007;7:6-8

• Commenton“Health-relatedqualityoflifeasaprognosticfactorforsurvivalincriticallyillpatients”.DMVandijck,SOeyen,LAnnemans,JMDecruyenaere.

IntensiveCareMed2009;35:1308

BOOKCHAPTER

• QualityoflifeafterICUClinicalevidenceinIntensiveCarebyTheESICMSystematicreviewgroup:pp236-240

ABSTRACTS

• Efficacy and side effects of a single dose of trometamol or bicarbonate as a buffer inpatientswithmildacidosis.ColpaertK,HosteE,NolletJ,OeyenS,DepuydtP,DeWaeleJ,DecruyenaereJ,MonsieursK,

OsipowskaE,ColardynF.

IntensiveCareMed2001;27(Suppl.2)

• A 10-years analysis of adult acute liver failure with request for a high-urgent livertransplant.OeyenS,HosteE,DanneelsC,MaeneL,TroisiR,DecruynaereJ,deHemptinneB,ColardynF.


• HeparinmonitoringintheICU:thevalueoftheactivatedclottingtime.DeWaeleJ,VanCauwenbergheS,HosteE,OeyenS,BenoitD,DepuydtP,ColardynF.


• Efficacyandsideeffectsoftheintravenouscool-linecatheter.ColpaertK,OeyenS,DeWaeleJ,HosteE,DecruyenaereJ,ColardynF.


• Impactofbloodstreaminfectionontheoutcomeofpatientswithacuterenalfailure.HosteE,BlotS,DeWaeleJ,ColpaertK,OeyenS,DecruyenaereJ,ColardynF.


• Targettingandmaintainingatightbloodglucoserangeinthecriticallyill:feasibleornot?OeyenS,PoelaertJ,VandewoudeK,DecruyenaereJ.


• Calculationofthetotalcostofownershipofanintensivecareinformationsystem.JDecruyenaere,CDanneels,SOeyen,KColpaert,GVerwaeren,DMyny.

207

CritCareMed2004;32(12,Suppl.)

• AcuterespiratoryeffectsoftheuprightpositioninARDSpatients.DeWaeleJ,ColpaertK,OeyenS,DecruyenaereJ,PoelaertJ,RoosensC.

ActaAnaesthesiol.Belg.2004;55:269

• Salinevolumeintransvesicalintra-abdominalpressuremeasurement:enoughisenough. DeWaele J, Pletinckx P, Decruyenaere J, Colpaert K, Oeyen S, Nollet J, Roosens C, Blot S,

HosteE.


• BloodstreaminfectionsfromabdominaloriginintheICU.DeWaeleJ,HosteE,VandewoudeK,DecruyenaereJ,ColpaertK,OeyenS,NolletJ,Roosens

C,BlotS.


• Acute kidney injury defined by the Rifle classification: which baseline serum creatininelevel?De Laet I, DeWaele JJ, Blot SI, Decruyenaere J, Oeyen S, Colpaert K, Nollet J, Roosens C,

HosteEA.


• Oliguria during a 2-hour period (U2): a beautiful day for the detection of acute kidneyinjury?De Laet I, DeWaele JJ, Blot SI, Decruyenaere J, Oeyen S, Colpaert K, Nollet J, Roosens C,

HosteEA.IntensiveCareMed2006;32(Suppl.1)

• Reliabilityoftransvesicalintra-abdominalpressuremeasurementusingminimalinstillationvolumes.DeLaetIE,HosteEA,OeyenS,ColpaertK,NolletJ,RoosensC,DecruyenaereJ,DeWaeleJJ.


• Adrenalfunctioninpatientsatriskforintra-abdominalhypertension.DeLaetIE,HosteE,OeyenS,NolletJ,ColpaertK,RoosensC,DecruyenaereJ,DeWaeleJJ.


• Hyperglycemiaupononsetofnosocomialbloodstreaminfectionadverselyaffectsoutcomeinamixedintensivecareunitpopulation.VandijckD,OeyenS,BuyleF,ClausB,BlotS,DecruyenaereJ.

CritCare2007;11(Suppl.2)

• Dailycostofantimicrobialtherapyincriticallyillpatientswithnosocomialsepsis.Vandijck DM, Blot SI, Depaemelaere M, Oeyen S, Colpaert KE, Annemans L, Peleman RP,

BuyleFM,LabeauSO,DecruyenaereJM.


• Candidemiainthecriticallyill:Aneconomicanalysisofdailyantimicrobialtherapyrelatedcosts.VandijckDM,BlotSI,DepaemelaereM,OeyenS,ColpaertKE,AnnemansL,VandewoudeKH,

PelemanRP,BuyleFM,LabeauSO,DecruyenaereJM.


208

• Acutekidneyinjuryinlivertransplantpatients.AkbasT,DeWaeleJJ,RoosensC,OeyenS,ColpaertK,NolletJ,DecruyeanereJ,HosteE.


• CompliancewithrestrictedmeropenemprescriptioninasurgicalICU.RavytsM,BlotS,DepuydtP,HosteE,ColpaertK,OeyenS;RoosensC,VogelaersD,DeWaele

JJ.IntensiveCareMed2007;33(Suppl.1)

• Antibiotic treatment for intra-abdominal infections in the ICU: is XXL necessary for allpatients?HLebbinck,EHoste,SBlot,KColpaert,SOeyen,CRoosens,DVogelaers,JDecruyenaere,J

DeWaele.


• Temocillin:avalidoptionfordirectedtherapyinICUpatients?D Njdekembo Shango, P Depuydt, S Blot, K Colpaert, E Hoste, S Oeyen, C Roosens, D

Vogelaers,JDecruyenaere,JDeWaele.


• Characteristicsandoutcomesofinterhospitaltransferpatientsadmittedtoatertiarycareintensivecareunit.SGOeyen,PMCoucke,DMVandijck,PLafaire,JMDecruyenaere.


• Outcomeandresourceutilizationfollowinginterhospitaltransferofcriticallyillpatients.DVandijck,SOeyen,PCoucke,PLafaire,DBenoit,JDecruyenaere.

CriticalCare2009;13(Suppl.1)

• PatientsadmittedtotheICUaftercardiopulmonaryresuscitation:ananalysisofoutcome,qualityoflifeandcost-effectiveness.OeyenS,VandijckD,VandenbosscheJ,BenoitD,AnnemansL,ColardynF,DecruyenaereJ.

ValueinHealth2009;12:A329

• Agreementbetweenpatientandproxyassessmentofhealth-relatedqualityoflifebeforeintensivecareunitadmission.VandijckD,OeyenS,CostersS,AnnemansL,DecruyenaereJ.

ValueinHealth2009;12:A397

• QualityoflifebeforeICU,3monthsand1yearafterICUdischarge.SandraOeyen,DominiqueBenoit,LievenAnnemans,JohanDecruyenaere.

CritCareMed2010;38(12,Suppl)

• The patient with decompensated liver cirrhosis admitted to the ICU: Evaluation ofoutcomes,qualityoflife,costsandcost-effectiveness.SandraOeyen,DominiqueBenoit,LievenAnnemans,JohanDecruyenaere.


• Thepatientwithhematologicalmalignancyadmittedtothe ICU:Evaluationofoutcomes,qualityoflife,costsandcost-effectiveness.SandraOeyen,DominiqueBenoit,LievenAnnemans,JohanDecruyenaere.

CritCareMed2012;40(12,Suppl

209

• The (very)oldpatient admitted to the ICU: Evaluationofoutcomes,qualityof life, costsandcost-effectiveness.SandraOeyen,DominiqueBenoit,LievenAnnemans,JohanDecruyenaere.


• Long-termoutcomeandqualityoflifeinICUpatientswithacutekidneyinjurytreatedwithrenalreplacementtherapy:Acasecontrolstudy.DeCorteW,OeyenS,AnnemansL,BenoitD,ADhondt,RVanholder,DecruyenaereJ,Hoste

E.IntensiveCareMed2014;40(Suppl1)

• Developmentofapredictionmodelforlong-termqualityoflifeincriticallyillpatients.SandraOeyen,KarelVermeulen,DominiqueBenoit,LievenAnnemans,JohanDecruyenaere.


• Longitudinalpredictionmodelforlong-termqualityoflifeincriticallyillpatients.SandraOeyen,KarelVermeulen,DominiqueBenoit,LievenAnnemans,JohanDecruyenaere.


• Influenceofneutropeniaonmortalityofcriticallyillcancerpatients:resultsofasystematicreviewonindividualdata.QuentinGeorges,ElieAzoulay,DjamelMokart,MSoares,KyeongmanJeon,SandraOeyenet

al.AnnalsofIntensiveCare2017;7(Suppl1)

• Outcomeandqualityoflifeinpatientswithaprolongedintensivecareunitstay.KSteenhaut,SOeyen,DBenoit,JDecruyenaere,EHoste.

Posterpresentationat the38th International Symposiumon IntensiveCare andEmergency

Medicine(March2018,Brussels);theabstractwillbepublishedinaSupplementaleditionof

CriticalCare2018

210

III.Additionalpublicationsrelatedtothesubjectofthethesis

Long-termoutcomeandhealth-relatedqualityoflifeindifficult-to-weanpatientswithorwithoutventilatordependencyatICUdischarge:aretrospectivecohortstudy.PublishedasDepuydtP,OeyenS,DeSmetS,DeRaedtS,BenoitD,DecruyenaereJ,DeromE.Long-term outcome and health-related quality of life in difficult-to-wean patients with or withoutventilatordependencyatICUdischarge:aretrospectivecohortstudy.BMCPulmMed2016;27:133

TheimpactoffrailtyonICUand30-daymortalityandthelevelofcareinveryelderlypatients(≥80years).PublishedasFlaattenH,DeLangeDW,MorandiA,AndersenFH,ArtigasA,BertoliniG,BoumendilA,CecconiM,ChristensenS,FaraldiL,FjølnerJ,JungC,MarshB,MorenoR,OeyenS,ÖhmanCA,PintoBB, Soliman IW, Szczeklik W, Valentin A, Watson X, Zaferidis T, Guidet B; VIP1 study group. Theimpact of frailty on ICU and 30-daymortality and the level of care in very elderly patients (≥ 80years).IntensiveCareMed2017;43:1820-1828

Withholdingorwithdrawingoflife-sustainingtherapyinolderadultpatients(≥80years)admittedtotheintensivecareunit.Guidet B, Flaatten H, Boumendil A, Morandi A, Andersen FH, Artigas A, Bertolini G, Cecconi M,ChristensenS,FaraldiL,FjølnerJ,JungC,MarshB,MorenoR,OeyenS,ÖhmanCA,PintoBB,SolimanIW,SzczeklikW,ValentinA,WatsonX,ZafeiridisT,DeLangeDW;VIP1studygroup. IntensiveCareMed;2018May17.doi:10.1007/s00134-018-5196-7.[Epubaheadofprint]

211

It’snotaboutthedestination,

itisabouttheride.

Long-term Outcomes and Quality of Life In Critically Ill Patients

Documents

Transcript of Long-term Outcomes and Quality of Life In Critically Ill Patients