RAATVDH

qsstoc

tcp(tubmw3

Pminmcafa

aopdaimac

A

ASH

©P

CA

RD

IOV

ASC

ULA

R

elation of the Sequential Organ Failuressessment Score to Morbidity and Mortalityfter Cardiac Surgery

ommi Pätilä, MD, Sinikka Kukkonen, MD, PhD, Antti Vento, MD, PhD,ille Pettilä, MD, PhD, and Raili Suojaranta-Ylinen, MD, PhD

epartments of Cardiothoracic Surgery and Anesthesiology and Intensive Care Medicine, Helsinki University Meilahti Hospital,

elsinki, Finland

mclraTmwp

dpae

Background. Organ dysfunction evaluation using Se-uential Organ Failure Assessment (SOFA) has beenhown to predict mortality and morbidity in adult cardiacurgical patients with prolonged recovery. The purpose ofhis study was to evaluate the utility of SOFA in predictionf mortality and morbidity in a cohort of heterogeneousonsecutive adult cardiac surgical patients.

Methods. A prospective study of 857 consecutive pa-ients entering in a single cardiac postoperative intensiveare unit was assigned during the year 2004. The Euro-ean System for Cardiac Operative Risk Evaluation

EuroSCORE) of each patient was assessed preopera-ively. SOFA was calculated daily until intensive carenit discharge or for a maximum of 7 days. SOFA changeetween the first and the third postoperative day, maxi-um SOFA during the first 3 days, and maximal SOFAere calculated. Length of intensive care unit stay and

0-day mortality were assessed.

ti

o1dum(we

ascsspor

ItS

urgery, University of Helsinki Meilahti Hospital, PO Box 340, FIN-00029US, Helsinki, Finland; e-mail: tommi.patila@hus.fi.

2006 by The Society of Thoracic Surgeonsublished by Elsevier Inc

Results. Maximum SOFA during the first 3 days andaximal SOFA-predicted 30-day mortality (area under the

urve, 0.763 and 0.779, respectively) also correlated with theength of intensive care unit stay (p < 0.001 and p < 0.001,espectively). The EuroSCORE predicted both mortalitynd intensive care unit stay (p < 0.0001 and p < 0.0001).he correlation coefficient between the EuroSCORE andaximum SOFA during the first 3 days or maximal SOFAas low (r � 0.34 and 0.33, respectively, p < 0.0001 and� 0.0001).Conclusions. The SOFA score is an independent pre-

ictor of mortality and length of stay in cardiac surgicalatients. The SOFA score is associated with mortalitynd morbidity even when assessed in the early postop-rative period after adult cardiac surgery.

(Ann Thorac Surg 2006;82:2072–9)

© 2006 by The Society of Thoracic Surgeons

ostoperative scoring of cardiac surgical patients inthe intensive care unit (ICU) is necessary for the

easurement of morbidity. The measurement of morbid-ty is more sensitive than mortality in the evaluation ofew treatment modalities, cost/benefit, or the quality andanagement of the ICU. It also holds possibilities for the

omparison of results of institutions or surgeons. Risk-djusted morbidity differences between surgeons’ per-ormances might differ much more than mortality ratesnd also have economic aspects.When preoperative risk algorithms are widely used

nd accepted in the practice of cardiac surgery, post-perative scoring would give another time point foratient evaluation, if assessed as an aid for clinicalecision-making. After the operation, certain factorsffecting the patient might be downgraded and newssues raised. The question to be asked at this time

ight be whether to continue ICU care. Many of thevailable ICU outcome prediction models ignorehanges in patient status [1–3], which is the strength of

ccepted for publication June 2, 2006.

ddress correspondence to Dr Pätilä, Department of Cardiothoracic

he Sequential Organ Failure Assessment (SOFA) scor-ng system.

The SOFA system was created in a consensus meetingf the European Society of Intensive Care Medicine in994 and further revised in 1996. The SOFA is a six-organysfunction/failure score measuring multiple organ fail-re daily. Each organ is graded from 0 (normal) to 4 (theost abnormal), providing a daily score of 0 to 24 points

Table 1). The objective in the development of the SOFAas to create a simple, reliable, and continuous score

asily obtained in every institution [4].Although SOFA was developed primarily to describe

nd quantify organ function, it has been demonstrated ineveral studies to predict mortality and morbidity ofritically ill patients [5–9]. Reports on the SOFA scoringystem in cardiac surgical patient population have beenparse. The study by Ceriani and colleagues [10] includedatients with an ICU stay of 4 days or more. SOFA scoresf adult cardiac surgical population with uneventfulecovery have not been published [10–12].

A lack of well validated, widely accepted postoperativeCU outcome prediction score for cardiac surgical pa-ients is evident. Our aim was to examine the utility of the

OFA scoring system in the postoperative evaluation of

0003-4975/06/$32.00doi:10.1016/j.athoracsur.2006.06.025

cp(cbi[tdi

M

SBpgMcdaFfuti

DEeweoaacs

apl

hcfCpsto

m�vPi

T

T

R

C

L

C

C

R

a

d e;

2073Ann Thorac Surg PATILA ET AL2006;82:2072–9 SOFA SCORE IN CARDIAC SURGERY PATIENTS

CA

RD

IOV

ASC

ULA

R

ardiac surgical patients. We calculated associated Euro-ean System for Cardiac Operative Risk Evaluation

EuroSCORE) values in the same population to enableomparison with the SOFA scores. The EuroSCORE haseen well documented to predict morbidity and mortal-

ty after cardiac surgery in Europe and in North America13–17]. We also evaluated the association of postopera-ive SOFA with preoperative EuroSCORE and sought toetermine whether this later assessment would have

mpact in the prediction of morbidity and mortality.

aterial and Methods

tudy Populationetween January 1, 2004 and December 31, 2004, 902atients underwent cardiac surgery or surgery of thereat intrathoracic vessels in the University of Helsinkieilahti Hospital. The study included all of the 855

ardiac surgical patients admitted in the 14-bed postcar-iac surgery ICU. Excluded were transplantation patientsnd patients transferred into other ICUs after operation.rom the postcardiac surgery ICU, patients were trans-erred directly to the surgical ward; no intermediate carenit was available. Approval for the study was given by

he local ethics committee, which waived the need fornformed consent.

ata Collection and MeasurementsuroSCORE data of each patient were collected preop-ratively. Risk factors included in the EuroSCORE modelere age, female gender, chronic pulmonary disease,

xtracardiac arteriopathy, neurologic dysfunction, previ-us cardiac surgery, increased serum creatinine level,ctive endocarditis, critical preoperative state, unstablengina, decreased left ventricular function, recent myo-ardial infarction, pulmonary hypertension, emergent

able 1. The Sequential Organ Failure Assessment (SOFA) Sc

0 1

espirationPao2/Fio2 �400 �400

oagulationPlatelets � 103/�L �150 �150

iver Bilirubinmg/dL (�mol/L)

�1.2 (20–32) 1.2–1.9 (20–32)

ardiovascularb

Hypotension No hypotension MAP �70 mm H

entral nervous systemGlasgow Coma Scale 15 13–14

enal creatininemg/dL (�mol/L) orurine output mL/d

�1.2 (�110) 1.2–1.9 (110–170

Values are with respiratory support. b Adrenergic agents administe

op � dopamine; epi � epinefrine; MAP � mean arterial pressur

urgery, cardiac operation other than isolated coronaryAg

rtery bypass grafting, surgery on the thoracic aorta, andostinfarct septal rupture [13]. Patients’ demographic,

aboratory, and clinical data were collected in the ICU.The SOFA score (Table 1) was calculated every 24

ours until discharge or for a maximum of 7 days. In thealculation of the score, the worst values for a given dayor each variable were included. The assumed Glasgowoma Score values were used in sedated patients untilroven otherwise. A value of 3 was used for the renalcore in patients with continuous venovenous hemofil-ration started in an indication of relatively low urineutput and massive fluid load.The maximum SOFA (maxSOFA) score was deter-ined as the highest SOFA value during the ICU stay,-SOFA31 was calculated by subtracting lowest SOFAalue from highest SOFA value during the first 3 days.atients discharged from the ICU before day 3 were

ncluded in the �-SOFA31 calculations. MaxSOFA3d was

able 2. Surgical Procedures

Procedure N

CABG 407CABG (off-pump) 117AVR 81MV procedure 49Composite graft 35Rec thoracic aorta 27CABG � AVR 70CABG � MV procedure 19Double valve procedure 13Tricuspidal valve procedure 7Others 30Total 855

SOFA Score

2 3 4

�300 �200a �100a

�100 �50 �202.0–5.9 (33–101) 6.0–11.9 (102–204) �12.0 (�204)

Dop �5, or dobany dosec

Dop �5, epi �0.1,or norepi �0.1c

Dop �15, epi �0.1,or norepi �0.1c

10–12 6–9 �62.0–3.4 (171–299) 3.5–4.9 (300–440)

or �500�5 (�440) �200

r at least 1 h. c Dosages are in �g/(kg · min).

norepi � norepinefrine.

ore

g

)

red fo

VR � aortic valve replacement; CABG � Coronary bypass sur-ery; MV � mitral valve.

md

aoid

f

st

SDIt(d

Fm(wcfa(

T

I

PAGElm�

m36

a

D

I CardiA

2074 PATILA ET AL Ann Thorac SurgSOFA SCORE IN CARDIAC SURGERY PATIENTS 2006;82:2072–9

CA

RD

IOV

ASC

ULA

R

easured as the highest SOFA score during the first 3ays.The Central Statistical office of Finland provided the 30-

nd 60-day mortality. Accuracy of the Central Statisticalffice of Finland concerning mortality is excellent. Allndividuals permanently living in Finland at the time ofeath are compiled in these statistics.The SOFA and EuroSCORE were separately calculated

or risk of death and compared. EuroSCORE was as-

ig 1. The maximum Sequential Organ Failure Assess-ent score during first 3 days of intensive care unit

MaxSOFA3d) stay correlated with mortality alongith the highest SOFA during first week of intensive

are unit stay (maxSOFA). Logistic European Systemor Cardiac Operative Risk Evaluation (EuroSCORE)nd EuroSCORE also correlated with mortality.CI � confidence interval.)

able 3. Patients Divided in Groups According to Length of I

CU StayGroup 1 Gro

1 day 2–4

atient number 391 2ge 64 (19–88) 68 (22ender M/F 296/95 203uroSCORE 3 (0–13) 6 (0–

og Eur 4.2 (0.9–59.7) 8.8 (0.9ax SOFA 6 (1–11) 7 (1–SOFA13 1 (0–3) �3 (�ax SOFA3 6 (1–11) 7 (1–

0-day mortality 20-day mortality 2

p � 0.05 compared with group 1. b p � 0.05 compared with group 2

ata are numbers (range) or ratios.

CU � intensive care unit; EuroSCORE � European System forssessment.

essed from a larger group of patients: 5 patients died inhe operating room.

tatistical Methodsata were analyzed with SPSS 12.0 (SPSS Inc, Chicago,

L). Predictive power regarding 30-day and 60-day mor-ality was assessed by receiver operating characteristicROC) curve analysis. SOFA scores and EuroSCOREsuring ICU stay were evaluated using nonparametric

ive Care Unit Stay

Group 3 Group 45–7 days �8 days

101 6870.0 (34–87)a 67 (33–87)

73/28 49/196 (0–17)a,b 8 (0–19)a,b,c

)a 6.2 (0.9–63.3)a,b 13.7 (0.9–81.4)a,b,c

9 (6–15)a,b 11 (6–20)a,b,c

�1 (�10–5) 1 (�7–13)c

9 (5–15)a,b 11 (6–20)a,b,c

3 7a

4 13a

p � 0.05 compared with group 3.

ac Operative Risk Evaluation; SOFA � Sequential Organ Failure

ntens

up 2days

95–85)a

/9217)a

–68.513)a

9–4)a

13)a

55

. c

KWacSt

R

Duwcsttaal

51p

aT2mt

n9SgElFt0Swac

FpmsCl

FsStit((d

2075Ann Thorac Surg PATILA ET AL2006;82:2072–9 SOFA SCORE IN CARDIAC SURGERY PATIENTS

CA

RD

IOV

ASC

ULA

R

ruskal-Wallis test with several groups and the Mann-hitney U test between two groups. ICU mortalities

mong different groups according to ICU stay wereompared using the Fisher test. Correlation betweenOFA scores and the EuroSCORE values was assessed by

he Spearman nonparametric test.

esults

uring the study period, complete data from the consec-tive 855 patients entering the postcardiac surgery ICUere gathered for the evaluated variables. Surgical pro-

edures are presented in Table 2. Of these patients, 444tayed 2 days or longer and 84 stayed 7 days or longer inhe ICU. All patients needed respiratory support whenhey arrived at the ICU, and the infusions given weredrenalin in 73%, noradrenalin in 80%, dopamine in 2%,nd dobutamine in 0.6%. Prevalence of intraaortic bal-oon pump use was 2.8%.

The EuroSCORE (mean SD) of these patients was.3 3.5, and the mean logistic EuroSCORE was 8.0 1.1%. Five patients died in the operating room, 12

ig 2. (A) Intensive care unit stay (continuous line) was longer inatients with a higher maximum Sequential Organ Failure Assess-ent score during first 3 days (maxSOFA3d) and (B) higher SOFA

cores during first week (maxSOFA). Higher European System forardiac Operative Risk Evaluation (EuroSCORE) values (dotted

ine) were obtained from the patients who stayed longer in the ICU.

atients died during their ICU stay, and 13 patients died (

fter ICU discharge and within 60 days postoperatively.he 30-day mortality of all the operated on patients was.4%. Readmission rate to ICU was 2.6%. Table 3 sum-arizes the EuroSCORE and SOFA scores according to

he length of stay in the ICU.The predictive power of SOFA variables in mortality,

amely maxSOFA3d (area under the curve [AUC], 0.76;5% confidence interval [CI], 0.62 to 0.91) and max-OFA (AUC, 0.78; 95% CI, 0.64 to 0.92), was reasonablyood by ROC analysis (Fig 1) and comparable with theuroSCORE (AUC, 0.87; 95% CI, 0.79 to 0.94) and the

ogistic EuroSCORE (AUC, 0.84; 95% CI, 0.76 to 0.92,ig 1). The correlation of �-SOFA31 with 30-day mor-ality was only fair (AUC, 0.66; 95% CI, 055 to 0.77; p �.006). Lengths of ICU stay according to different max-OFA scores are presented in Figure 2. Mortality increasedith increasing SOFA scores, higher SOFA scores were

ssociated with EuroSCORE, and there was also signifi-ant association with higher EuroSCORE and higher

ig 3. (A) The continuous line describes the 30-day mortality (leftcale) of the patients divided in groups according to maximumequential Organ Failure Assessment scores during first 3 days in

he intensive care unit (ICU) (maxSOFA3d). The number of patientsn each group can be seen in the upper row. The dotted line showshe Higher European System for Cardiac Operative Risk EvaluationEuroSCORE) values of the patients in the same groups (right scale).B) Similar parallel curves can be seen when patient groups wereivided according to maximum SOFA score during the first week

MaxSOFA).

lbw�SwStr

C

Itpdch

ttfSgpeossmsTSspp

timbsiatso

atotpsv

atpt

cptodc

aclctbpgsa

iAtpfou

mpdabrphig1nadoai

loipaegcSpvc

2076 PATILA ET AL Ann Thorac SurgSOFA SCORE IN CARDIAC SURGERY PATIENTS 2006;82:2072–9

CA

RD

IOV

ASC

ULA

R

ogistic EuroSCORE, respectively (Fig 3). Correlationsetween EuroSCORE and maxSOFA and maxSOFA3dere low (p � 0.0001; r � 0.33; 95% CI, 0.27 to 0.39; and p

0.0001; r � 0.34; 95% CI, 0.27 to 0.40) respectively,pearman correlation test). A similarly low correlationas observed between logistic EuroSCORE and max-OFA and maxSOFA3d (p � 0.0001, r � 0.34, 95% CI, 0.2

o 0.40 and p � 0.0001, r � 0.35, 95% CI, 0.29 to 0.41,espectively, Spearman correlation test).

omment

n this study we demonstrate a significant correlation ofhe SOFA score in the mortality of the cardiac surgeryatients in postoperative period. The highest SOFA scoreuring the first 3 postoperative days and a greater in-rease in SOFA score during first 3 postoperative daysad a significant correlation in the 30-day mortality.In adult cardiac surgical patients with ICU stay of more

han 4 days, Ceriani and colleagues [10] demonstratedhat the SOFA score identified patients at increased riskor postoperative mortality. They also postulated, thatOFA could be used without any specific adaptations torade postoperative morbidity. Our study also includedatients with uneventful recovery and minimum postop-rative ICU stay. We especially assessed SOFA variablesf the first postoperative days to gauge the patients withtraightforward recovery without incidents. We demon-trated an association of SOFA score in the mortality andorbidity during the first 3 days of postoperative ICU

tay, even when uncomplicated patients were included.hus, our study further supports the applicability ofOFA in adult cardiac surgical patients. However, de-pite the SOFA calculations providing information oferoperative and postoperative variables, the predictiveower of the EuroSCORE for mortality was better.Numerous preoperative risk-assessment scoring sys-

ems have been published. Preoperative evaluation givesnformation about the patient, mainly demographics, co-

orbidities, and heart disease. We used the EuroSCOREecause it is widely accepted in Europe and has beenelected as an official preoperative scoring system in ournstitution. It has also been validated in North Americand Japan [16]. The EuroSCORE has been evaluated inhe prediction of morbidity in many studies. Also in ourtudy, the EuroSCORE was associated with a long post-perative stay.Preoperative risk stratifications ignore the individual

natomic aspects of the patient’s heart disease and alsohe performance of the operating surgeon, whereas post-perative patient scoring gives information of the situa-ion after the operation. In addition to the preoperativeatient status assessment, postoperative organ functioncoring support the existing data with another point ofiew.The rather low correlation coefficient between SOFA

nd the EuroSCORE is because they both predict mor-ality and morbidity well, but reveal different high-riskatients. A patient with a low preoperative risk according

o the EuroSCORE might have a stormy postoperative p

ourse that can be assessed by SOFA. This supports theotential of a look at the evolution of patient risk during

he treatment process. Furthermore, additive assessmentf these scores might be advantageous. The low inci-ence of index events in our material prevented furtheralculations.

The EuroSCORE has been proven to predict morbiditynd mortality, and our study supports this observationompared with morbidity assessed by SOFA and also byength of ICU stay. The combination of preoperativeardiac lesions, perioperative cardiac events, complica-ions, comorbidities, and the use of cardiopulmonaryypass (CPB) may predispose cardiac surgical patients toostoperative organ dysfunction resembling that of aeneral ICU population. The compatibility of the SOFAcoring system has been evaluated in both prospectivend retrospective studies [4–9].When organ function after cardiac surgery is exam-

ned, the cardiovascular evaluation is of great interest.ssessment of perfusion and imaging of ventricular wall

hickening and movement with controlled end-diastolicressure would give the most information about heart

unction [18], but the routine performance of these kindsf measurements would be laborious, expensive, andnnecessary in most cases.In the SOFA scoring system, cardiovascular measure-ent is performed on treatment criteria, and the inotro-

ic agents and vasopressors needed for appropriate car-iac output and blood pressure are gauged. Extensivend liberal use of inotropic agents and vasopressorsecause of myocardial stunning and the inflammatoryesponse caused by CPB might lead to high scores in thisarticular organ system in all the patients in the firstours after ICU admission. It has been demonstrated that

f ejection fraction decreased to below 0.46 in cardiosur-ical patients, the need for inotropic agents was 71% to00%. Even patients with a proper ejection fraction mayeed inotropic support because of local wall motionbnormalities and an increase in left ventricular end-iastolic pressure. Transient myocardial ischemia sec-ndary to cardiac surgery and cardioplegic arrest may bessociated with myocardial stunning and the need fornotropic and mechanical support [19–21].

Cardiovascular scoring simply by medication is thusess sensitive in detecting the differences among patientsn the first postoperative day, nevertheless sensitivity

ncreases over time. On the other hand, institutionalreferences in the use of inotropic and vasoactive drugsffect the scoring. In our institution, we use mostlypinephrine and less dopamine and dobutamine, whichive higher scoring in SOFA. However, patient-relatedhanges in vasoactive treatment reflected by change inOFA score during the first ICU days may be highlyredictive for later death because extended need forasoactive treatment is associated with unfavorable out-ome after cardiac surgery [11].

Postoperative cardiac surgical patients have certain

eculiarities in other organ systems that may modify

Sppccp[

sitebth

fqttaawab

eblio

flpSefmi

opsafllsacASuuCoamaf

Msp

slwwpTmttsTn

aSdmIsomorlemgsdc

Ssgspts[

R

2077Ann Thorac Surg PATILA ET AL2006;82:2072–9 SOFA SCORE IN CARDIAC SURGERY PATIENTS

CA

RD

IOV

ASC

ULA

R

OFA scoring, especially in the early postoperativeeriod. CPB and general anesthesia causes ventilation/erfusion mismatch and intrapulmonary shunting,ompromising oxygenation. The underlying causes in-lude atelectasis, hemodilution, and noncardiogeniculmonary edema from the endothelial dysfunction

22].Hepatic dysfunction is fairly common after open heart

urgery, and transient hyperbilirubinemia will developn about one third of the patients. Most of the increase ofotal bilirubin caused by CPB occurs on the first postop-rative day. This comes from an increase in unconjugatedilirubin by increased hemolysis peroperatively [23]. In

he case of hepatic failure, it takes a few days foryperbilirubinemia to develop.In critically ill patients, discontinuation of sedatives

or neurologic evaluation was not practiced until ade-uate hemodynamic stability was reached. Thus, pa-

ients with stroke might receive too low SOFA scores inhe first few days. However, critically ill patients withn unstable postoperative course after cardiac surgeryre almost without exception in severe fluid excesshen it is uncommon to reach normal mental status

ccording to the Glasgow Coma Scale even with intactrain [24].CPB has been shown to lower platelet count in the

arly postoperative period, thus affecting the SOFA scorey itself. Depression of hematopoietic system shows

owering of the platelet count during days, when thembalance between normal depletion and the formationf new platelets becomes evident [25].Continuous hemofiltration as a renal support for

uid removal is indicated in some cardiac surgicalatients despite normal renal function measured byOFA. These treatments are initiated individually byvaluating the whole status of the patient not byollowing rigid criteria. In these cases, the treatment

ight change the renal SOFA score without alterationsn the organ function.

In a recent study, Hekmat and colleagues [26] devel-ped a specific postoperative score for cardiac surgicalatients, the Cardiac Surgery Score (CASUS). It hasimilarities with the SOFA score. Same measured vari-bles include arterial partial pressure of oxygen/raction of inspired oxygen ratio, serum creatinineevel, serum bilirubin level, platelet count, and neuro-ogic state, although point limits differ and neurologiccore is simplified from Glasgow Coma Scale. CASUSlso measures the cardiovascular score, although ac-ording to the multiple organ dysfunction score [27].dditional variables in CASUS compared with theOFA are blood lactate level, intraaortic balloon pumpsage, ventricular assist device existence, and contin-ous venovenous hemofiltration or dialysis. In theASUS model inotropic agents were excluded basedn different therapy protocols between ICUs and vari-tion in patient volume status. This kind of minorodification of SOFA scoring might be advantageous,

lthough not proven. For a clinical use, the organ

unction score system should be as simple as possible.

ulticenter studies would be needed for a scoringystem to identify the most important postoperativearameters for cardiac surgical patients.Our purpose was to analyze a cohort of major cardiac

urgery patients. Because of the rather small cohort andow mortality, no restrictions according to type of surgeryere made. Different surgical procedures are associatedith different pathophysiologic heart diseases, differenterfusion times, and different amount of surgical trauma.hese factors lead to variability in the morbidity andortality of the population. Restriction of analysis in one

ype of surgery might have given more detailed informa-ion of the function of SOFA. The rather small sampleize of 855 patients is a significant limitation in our study.he single institution assessment also limits the useful-ess of our results.Our study shows the feasibility of SOFA scoring in the

ssessment of a cardiac surgical patient population.OFA correlates well in the morbidity and mortality,espite being rather simple. This kind of daily scoringight be advantageous for clinical use for evaluating the

CU population at certain time for optimizing the con-umption of ICU resources and to characterize severityf the patient material as an outcome measure instead ofortality in clinical trials. Moreover, the description of

rgan dysfunction is simple enough to be repeatedly andeliably measured in every institution. Values based onaboratory, oxygenation, and vasoactive treatment can beasily gathered automatically if a patient data manage-ent system is available. The SOFA score would be too

ross for individual decision-making, but SOFA scoreequences might be helpful in identifying certain organysfunction development for a patient recovering afterardiac surgery.

There are evident problems associated with theOFA score in the assessment of postoperative cardiacurgical patients. The SOFA score system has a veryood association to the mortality and morbidity in ourtudy. Concomitant evaluation of the EuroSCORE furtherroves the usefulness of postoperative scoring system in

he evaluation of cardiac surgical patients. For optimalcoring system, multicenter studies should be executed.28].

eferences

1. Knaus WA, Zimmerman JE, Wagner DP, Draper EA, Law-rence DE. APACHE-acute physiology and chronic healthevaluation: a physiologically based classification system. CritCare Med 1981;9:591–7.

2. Le Gall JR, Lemeshow S, Saulnier F. A new Simplified AcutePhysiology Score (SAPS II) based on a European/NorthAmerican multicenter study. JAMA 1993;270:2957–63.

3. Lemeshow S, Teres D, Avrunin JS, Gage RW. Refiningintensive care unit outcome prediction by using changingprobabilities of mortality. Crit Care Med 1988;16:470–7.

4. Vincent JL, Moreno R, Takala J, et al. The SOFA (Sepsis-related Organ Failure Assessment) score to describe organdysfunction/failure. On behalf of the Working Group onSepsis-Related Problems of the European Society of Inten-

sive Care Medicine. Intensive Care Med 1996;22:707–10.

1

1

1

1

1

1

1

1

1

1

2

2

2

2

2

2

2

2

2

I

Tmspvsucrs

Sspwoblw

2078 PATILA ET AL Ann Thorac SurgSOFA SCORE IN CARDIAC SURGERY PATIENTS 2006;82:2072–9

©P

CA

RD

IOV

ASC

ULA

R

5. Vincent JL, de Mendonca A, Cantraine F, et al. Use of theSOFA score to assess the incidence of organ dysfunction/failure in intensive care units: results of a multicenter,prospective study. Working group on “sepsis-related prob-lems” of the European Society of Intensive Care Medicine.Crit Care Med 1998;26:1793–800.

6. Pettila V, Pettila M, Sarna S, Voutilainen P, Takkunen O.Comparison of multiple organ dysfunction scores in theprediction of hospital mortality in the critically ill. Crit CareMed 2002;30:1705–11.

7. Ferreira FL, Bota DP, Bross A, Melot C, Vincent JL. Serialevaluation of the SOFA score to predict outcome in criticallyill patients. JAMA 2001;286:1754–8.

8. Timsit JF, Fosse JP, Troche G, et al. Calibration and discrim-ination by daily Logistic Organ Dysfunction scoring compar-atively with daily Sequential Organ Failure Assessmentscoring for predicting hospital mortality in critically ill pa-tients. Crit Care Med 2002;30:2003–13.

9. Antonelli M, Moreno R, Vincent JL, et al. Application ofSOFA score to trauma patients. Sequential Organ FailureAssessment. Intensive Care Med 1999;25:389–94.

0. Ceriani R, Mazzoni M, Bortone F, Gandini S, Solinas C,Susini G. Application of the sequential organ failure assess-ment score to cardiac surgical patients. Chest 2003;123:1229–39.

1. Janssens U, Graf C, Graf J, et al. Evaluation of the SOFAscore: a single center experience of a medical intensive careunit in 303 consecutive patients with predominately cardio-vascular disorders. Intensive Care Med 2000;26:1037–45

2. Shime N, Kageyama K, Ashida H, Tanaka Y. Application ofmodified sequential organ failure assessment score in chil-dren after cardiac surgery. J Cardiothorac Vasc Anesth2001;15:463–8.

3. Nashef SA, Roques F, Michel P, et al. European system forcardiac operative risk evaluation (EuroSCORE). Eur J Car-diothorac Surg 1999;16:9–13.

4. Toumpoulis IK, Anagnostopoulos CE, Swistel DG, DeRoseJJ. Does EuroSCORE predict length of stay and specificpostoperative complications after cardiac surgery? Eur J Car-diothorac Surg 2005;27:128–33.

5. Pitkänen O, Niskanen M, Rehnberg S, Hippeläinen M,Hynynen M. Intra-institutional prediction of outcome aftercardiac surgery: comparison between a locally derivedmodel and the EuroSCORE. Eur J Cardiothorac Surg 2000;18:

703–10.

ith SOFA scores. The SOFA score’s simplicity and

acPsp

ppaoitp

mdscs

2006 by The Society of Thoracic Surgeonsublished by Elsevier Inc

6. Nashef SA, Roques F, Hammill BG, et al. Validation ofEuropean System for Cardiac Operative Risk Evaluation(EuroSCORE) in North American cardiac surgery. Eur J Car-diothorac Surg 2002;22:101–5.

7. Nilsson J, Algotsson L, Höglund P, Lührs C, Brandt J. Earlymortality in coronary bypass surgery: the EuroSCORE ver-sus The Society of Thoracic Surgeons risk algorithm. AnnThorac Surg 2004;77:1235–9.

8. Reichert CL, Visser CA, Koolen JJ, et al. Transesophagealechocardiography in hypotensive patients after cardiac op-erations. Comparison with hemodynamic parameters. J Tho-rac Cardiovasc Surg 1991;8104:321–6.

9. Hardy JF, Belisle S. Inotropic support of the heart that fails tosuccessfully wean from cardiopulmonary bypass: the Mon-treal Heart Institute experience. J Cardiothorac Vasc Anesth1993;7:33–9.

0. Bernard F, Denault A, Babin D, et al. Diastolic dysfunction ispredictive of difficult weaning from cardiopulmonary by-pass. Anesth Analg 2001;92:291–8.

1. Bolli R. Mechanism of myocardial “stunning.” Circulation1990;82:723–38.

2. MA Matthay and JP Wiener-Kronish. Respiratory manage-ment after cardiac surgery. Chest 1989;95:257–8.

3. Hachenberg T, Tenling A, Nystrom SO, Tyden H, Hedensti-erna G. Ventilation-perfusion inequality in patients under-going cardiac surgery. Anesthesiology 1994;80:509–19.

4. Wang MJ, Chao A, Huang CH, et al. Hyperbilirubinemiaafter cardiac operation. Incidence, risk factors, and clinicalsignificance. J Thorac Cardiovasc Surg 1994;108:429–36.

5. Andersson RE, Li TQ, Hindmarsh T, Settergren G, Vaage J.Increased intracellular brain water in coronary artery bypassgrafting is avoided by off-pump surgery. J Cardiothorac VascAnaest 1999;13:698–02.

6. Bevan DH. Cardiac bypass haemostasis: putting bloodthroughout the mill. Br J Haematol 1999;104:208–19.

7. Hekmat K, Kroener A, Stuetzer H, et al. Daily assessment oforgan dysfunction and survival in intensive care unit cardiacsurgical patients. Ann Thorac Surg 2005;79:1555–62.

8. Marshall JC, Cook DJ, Christou NV, Bernard GR, Sprung CL,Sibbald WJ. Multiple organ dysfunction score: a reliabledescriptor of a complex clinical outcome. Crit Care Med

1995;23:1638–52.

NVITED COMMENTARY

he currently available perioperative mortality andorbidity risk assessment tools, such as Parsonnet’s

core and Euroscore, are exclusively based on patients’reoperative comorbidities. In contrast, there exists noalidated model to identify and assess postcardiacurgery patients’ mortality risk at the intensive carenit (ICU) bedside. Such a tool would be useful forlinicians so as to estimate each patient’s day-to-dayisk status and divert ICU resources and attention toicker patients.

In this study, Pätilä and colleagues [1] suggest that theequential Organ Failure Assessment (SOFA) scoringystem be used as a mortality risk estimation tool inostoperative cardiac surgery patients. Their reasoningas that SOFA scores are a straightforward, reliable wayf measuring the degree of multiorgan failure on a dailyasis. To corroborate their assertion, the authors calcu-

ated concomitant Euroscore values and compared them

bility of continuous assessment make it a preferredhoice over other ICU prediction scores such as the Acutehysiology And Chronic Health Evaluation (APACHE)core, which provides only an admission risk factorrofile.Although SOFA scoring was not designed to be ap-

lied to cardiac surgery patients, such use has beenreviously reported in the literature. This study furtherdds to the limited evidence available. Based on theirbserved correlation between SOFA scores and mortal-

ty, the authors conclude that their study further supportshe applicability of SOFA scoring in adult cardiac surgeryatients.However,SOFA scoring is probably not the risk assess-ent tool we seek. It is too crude a device, not actually

esigned de novo to be applied to postoperative cardiacurgery patients. It is limited in accurately evaluatingardiovascular function, renal function, and neurologic

tatus in those patients. It is not therefore surprising that

0003-4975/06/$32.00doi:10.1016/j.athoracsur.2006.06.033