Prognostic factors of long-term functioning and productivity after traumatic brain injury: a...

http://cre.sagepub.com

Clinical Rehabilitation

DOI: 10.1177/0269215507077603 2007; 21; 1024 Clin Rehabil

Agnes H.P. Willemse-van Son, Gerard M. Ribbers, Arianne P. Verhagen and Henk J. Stam systematic review of prospective cohort studies

Prognostic factors of long-term functioning and productivity after traumatic brain injury: a

http://cre.sagepub.com/cgi/content/abstract/21/11/1024 The online version of this article can be found at:

Published by:

http://www.sagepublications.com

can be found at:Clinical Rehabilitation Additional services and information for

http://cre.sagepub.com/cgi/alerts Email Alerts:

http://cre.sagepub.com/subscriptions Subscriptions:

http://www.sagepub.com/journalsReprints.navReprints:

http://www.sagepub.com/journalsPermissions.navPermissions:

http://cre.sagepub.com/cgi/content/refs/21/11/1024SAGE Journals Online and HighWire Press platforms):

(this article cites 47 articles hosted on the Citations

© 2007 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution. at Ebsco Host temp on November 6, 2007 http://cre.sagepub.comDownloaded from

http://cre.sagepub.com/cgi/alerts

http://cre.sagepub.com/subscriptions

http://www.sagepub.com/journalsReprints.nav

http://www.sagepub.com/journalsPermissions.nav

http://cre.sagepub.com/cgi/content/refs/21/11/1024


Clinical Rehabilitation 2007; 21: 1024–1037

Prognostic factors of long-term functioning andproductivity after traumatic brain injury: a systematicreview of prospective cohort studiesAgnes HP Willemse-van Son Department of Rehabilitation Medicine, Erasmus MC, Gerard M Ribbers Rijndam RehabilitationCenter and Department of Rehabilitation Medicine, Erasmus MC, Arianne P Verhagen Department of General Practice, ErasmusMC and Henk J Stam Department of Rehabilitation Medicine, Erasmus MC, Rotterdam, The Netherlands

Received 29th September 2006; returned for revisions 30th December 2006; revised manuscript accepted 30th January 2007.

Objective: To systematically review prospective cohort studies that investigated

prognostic factors associated with long-term activity limitations or participation

restrictions and productivity after a traumatic brain injury.

Data sources: PubMed and Psychinfo were searched from 1995 to April 2005, and

references were checked.

Review methods: Publications were selected if the study assessed prognostic

factors for activity limitations or participation restrictions at least one year post injury;

outcome was measured with another or additional measure besides the Glasgow

Outcome Scale; the design was a prospective cohort study of adult traumatic brain

injury patients; the article was a full-text article written in English, French, German or

Dutch. Two reviewers independently assessed methodological quality. A study was

considered as ‘high quality’ if it satisfied at least half of the maximum available quality

score.

Results: Thirty-five articles reporting on 14 cohorts were included. Due to hetero-

geneity in prognostic factors and outcome measures, a best-evidence synthesis was

performed. All cohorts were of high quality. Strong evidence for predicting disability

was found for older age, pre-injury unemployment, pre-injury substance abuse, and

more disability at rehabilitation discharge. Strong prognostic factors for being

non-productive were pre-injury unemployment, longer post-traumatic amnesia, more

disability at rehabilitation admission, and pre-injury substance abuse.

Conclusion: Older age, pre-injury unemployment, pre-injury substance abuse and

more disability at rehabilitation discharge are important predictors of long-term

disability. Pre-injury unemployment, longer post-traumatic amnesia, more disability at

rehabilitation admission and pre-injury substance abuse are important predictors of

being non-productive.

Introduction

Traumatic brain injury affects approximately 1.4million people in the USA each year and about235 000 are hospitalized.1 In Europe approxi-mately 1.6 million traumatic brain injury patients

Address for correspondence: Agnes Willemse-van Son,Erasmus MC, Department of Rehabilitation Medicine,PO Box 2040, 3000 CA Rotterdam, The Netherlands.e-mail: [email protected]

� SAGE Publications 2007Los Angeles, London, New Delhi and Singapore 10.1177/0269215507077603



are admitted to hospital on a yearly basis.Although reported incidence rates vary per coun-try, the incidence rate for the European popula-tion is 235/100 000 with 66 000 deaths per year.2,3

Direct health costs related to traumatic braininjury in Europe are estimated at 2.9 billioneuros. Non-medical costs (e.g. due to loss ofproductivity and intangible costs due to reducedquality of life) are not taken into account in thisestimation.4

The outcome after traumatic brain injury canvary from complete recovery to death, withmany patients having long-term disabilities.Especially after severe injury, serious cognitive,behavioural, emotional and sensorimotor impair-ments can occur.5 These impairments can havemajor consequences for activity patterns, socialparticipation and quality of life issues.

Reliable guidelines for prediction of long-termoutcome and optimal clinical management inpatients at risk for activity limitations or partici-pation restrictions are lacking. Prediction modelsfor patients at risk of developing these long-termrestrictions are essential to optimize the use oflimited health care and social resources forpatients and their relatives.

Although several studies have investigated thelong-term prognosis of traumatic brain injury, toour knowledge no systematic reviews have beenpublished on prognostic factors of long-term out-come. Therefore, this study aims to summarize theliterature on prognostic factors associated withactivity limitations, participation restrictions andproductivity at least one year post-injury. Wesystematically investigated the influence of socio-demographic factors, pre-morbid comorbidity,injury characteristics, neuropsychological factors,treatment factors, and post-acute functioning andtheir relation to long-term outcome andproductivity.

Method

Search strategyWe searched PubMed and Psychinfo from 1995

to April 2005. Additionally, references of identi-fied publications were checked.

The search strategy was developed and testedfor PubMed and adapted for Psychinfo.

To describe the population the MeSH term‘craniocerebral trauma’ was used. To describethe design the following key terms were used: ‘pre-dictive value of tests’ (MeSH term) and prognos*and predict*. To select the adult populationthe MeSH terms ‘adult’ and ‘middle aged’ wereused. The search strategy is available fromthe last author.

One reviewer (AWvS) conducted the search.Two reviewers (AWvS and GMR) independentlyscreened titles and abstracts to identify relevantarticles. Full papers were retrieved when abstractswere absent or provided insufficient informationto enable selection.

Selection criteriaAn article was included if all following criteria

were met: (1) the study investigated factors asso-ciated with functional outcome after traumaticbrain injury; (2) traumatic brain injury wasdefined as ‘an alteration in brain function as aresult of an acute external violent force to thehead’; (3) outcome was described as activitylimitations or participation restrictions, as definedin the International Classification of Functioning,Disability and Health6; (4) another or an addi-tional measure besides the Glasgow OutcomeScale7 or its extended version8 was used to mea-sure outcome; (5) time post injury was at least oneyear; (6) the study population consisted of trau-matic brain injury patients or a separatelyanalysed subgroup of traumatic brain injurypatients; (7) the majority (at least 80%) of thepatients in the studies were 18–65 years old;(8) the article was written in English, French,German or Dutch; (9) the article was a full-textarticle; (10) the study design was a prospectivecohort study.

A study was excluded if: (1) the study popula-tion suffered from additional serious neurological,oncological or systemic impairments; (2) the studypopulation included animals.

Two reviewers (AWvS and GMR) assessed allcriteria independently in the full-text articles.In case of disagreement, consensus was sought.If disagreements were not resolved a third reviewer(APV) made the final decision.

Long-term functional prognosis after traumatic brain injury 1025



Methodological qualityWe assessed the methodological quality of the

cohorts with a modified version of an establishedcriteria list for prospective cohort studies.9

The criteria list was modified in concordancewith the framework for assessing validity inprognostic studies.10

The criteria list consisted of 16 items(Table 1), with each having a ‘yes/no/don’tknow option’. The item was scored positive(yes), if it fulfilled the criterion. If a criterionwas not fulfilled, the item was scored negative(no). If there was insufficient information, theitem was scored unclear (don’t know). Thetotal sum of positive items was calculated asthe quality score (maximum 16 points).A study that scored at least 8 points was con-sidered as high quality. We calculated one qual-ity score for each cohort, based on theinformation of all publications of that cohort.

Two reviewers independently scored the quality(AWvS and GMR). In case of disagreement,consensus was sought. If consensus could not bereached, a third reviewer (APV) made the

final decision. Interobserver agreement wasderived with kappa statistics because of dichoto-mous values.

Data extractionData on study cohort, inclusion and exclusion

criteria, number of participants, time post injury,loss to follow-up, outcome measurements, prog-nostic factors, and results on associations wereextracted, using a standardized form.

One reviewer (AWvS) extracted the data andone reviewer (GMR) checked an unselectedsubsample.

AnalysisA best-evidence synthesis was performed, in

which four levels of evidence9 (Table 2) weredefined to determine the strength of associationof prognostic factors with disability andbeing non-productive at least one year postonset. Disability was defined as allmeasures that described activity limitations or

Table 1 Criteria list for the quality assessment of studies on prognosis of patients withtraumatic brain injury

Criteria Score

Study populationa) Inception cohort þ/�/?b) Description of source population þ/�/?c) Description of relevant inclusion and exclusion criteria þ/�/?

Follow-upd) Time since injury/follow-up at least 12 months þ/�/?e) Drop-outs/loss to follow-up < 20% þ/�/?f) Information completers versus loss to follow-up/drop-outs þ/�/?g) Prospective data collection þ/�/?

Treatmenth) Treatment in cohort is fully described/standardized þ/�/?

Prognostic factorsi) Clinically relevant potential prognostic factors þ/�/?j) Standardized or valid measurements þ/�/?k) Data presentation of most important prognostic factors þ/�/?

Outcomel) Clinically relevant outcome measures þ/�/?m) Standardized or valid measurements þ/�/?n) Data presentation of most important outcome measures þ/�/?

Analysiso) Appropriate univariate crude estimates þ/�/?p) Appropriate multivariate analysis techniques þ/�/?

1026 AHP Willemse-van Son et al.



participation restrictions. Although being non-productive was a component of disability, we eval-uated it separately. Being non-productive wasdefined as ‘all measures that describe not returningto work, not returning to school, unemploymentor otherwise not being productive’.

Significant relative risk ratios (RRs), odds ratios(ORs), or significant associations (P<0.05) thatwere provided by the studies, were used to deter-mine the levels of evidence. If a multivariateanalysis was performed in the studies, then theseresults were used to establish levels of evidence.Otherwise, presented results from univariateanalysis were used.

Results

Selection of studiesIn total, 501 non-duplicate citations were found

from which 183 full-text articles were retrieved.Agreement was reached in 83% (151 of 183papers), and consensus was sought and found for32 articles. Finally the two reviewers selected35 articles (Figure 1). Six cohorts published morethan one paper on the same cohort, resulting in atotal of 14 cohorts.

Methodological qualityThe overall interobserver agreement of the

methodological quality assessment was K¼ 0.46,representing moderate agreement. Disagreementoccurred mainly because of reading errors anddifference in interpretation of the criteria list andwas easily resolved. For 11 items disagreementpersisted and a third reviewer (APV) made thefinal decision. The final results of the methodolo-gical assessment are presented in Table 3.

The cohorts were ranked by their quality score,in which a higher score indicated a higher quality.The names were abstracted from the city, region ordatabase where the cohort was recruited. Allcohorts scored at least 8 points and were all con-sidered high quality. Methodological quality ofindividual papers within the cohorts varied dueto differences in the presentations of methodsand analysis.

The most important methodological shortcom-ings concerned the following items: no descriptionof source population, no description of relevantinclusion and exclusion criteria, unclear informa-tion on percentage of drop-outs or loss to follow-up of more than 20%, no information ofcompleters versus loss to follow-up, and nodescription of what treatment was given, or therewas no treatment.

Study characteristicsTable 4 presents the main characteristics of the

cohorts ordered into source populations andranked by cohort and quality scores. The samplesize ranged from n¼ 2111 to n¼ 277112 per indivi-dual study. Seven cohorts enrolled over 100 casesand two cohorts (Colorado, Traumatic BrainInjury Model Systems) enrolled over 1000 cases.In one cohort (Traumatic Brain Injury ModelSystems) the sample sizes in the individual studiesranged from 59 to 2363 cases.13–26 The longestfollow-up period was up to 10 years (TraumaticBrain Injury Model Systems). The loss to follow-up after one year between the individual studiesranged from 4% (Groningen) to 58% (TraumaticBrain Injury Model Systems). Three cohorts(Groningen, Traumatic Brain Injury ModelSystems, Seattle) reported a loss to follow-up ofbelow 20%.

Table 2 Levels of evidence for prognostic factors

Levels of evidence

Strong Consistent (�80%) findings in at least two high-quality cohortsModerate One high-quality cohort and consistent (�80%) findings in one or more low-quality cohortsLimited Findings of one cohort or consistent (�80%) findings in one or more low-quality cohortsInconclusive Inconsistent findings irrespective of study quality




Table 5 presents a summary of the results fromthe multivariate analysis performed by the indivi-dual studies. Results are presented for the longestfollow-up period only. Over 100 prognostic factors

were examined: sociodemographic and injury-related factors, neuropsychological tests, pre-morbid comorbidity, and post-acute determinants.Twenty different outcome measures were used on

PubMed: 476 articlesPsycinfo: 25 additional articles

318 articles excluded based on title and abstract

157 articles excluded based on final selection

Reference checking: 9 additional articles

Total: 501 non-duplicate articles

Total : 183 full articles retrieved

Total: 26 articles included

35 articles on 14 cohorts

Figure 1 Flowchart showing the selection of the studies.

Table 3 Results of the methodological assessment

Cohort name a b c d e f g h i j k l m n o p Qualityscore

Range ofindividual papers

Traumatic Brain InjuryModel Systems13–26

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 16 9–14

Colorado12,27 1 1 1 1 0 1 1 1 1 1 1 1 1 1 1 1 15 14–15Groningen32–34 1 1 1 1 1 0 1 0 1 1 1 1 1 1 1 1 14 11–14Houston35 1 0 0 1 1 1 1 1 1 1 1 1 1 1 1 1 14 –Melbourne36 1 1 0 1 0 1 1 1 1 1 1 1 1 1 1 1 14 –Toronto37 1 1 1 1 0 1 1 0 1 1 1 1 1 1 1 1 14 –Seattle38,39 1 1 1 1 0 1 1 1 1 1 1 1 1 1 0 0 13 12, 12Westmead40 1 1 0 1 0 1 1 0 1 1 1 1 1 1 1 1 13 –Rome41 1 0 1 1 0 0 1 0 1 1 1 1 1 1 1 1 12 –Columbus42–44 1 0 0 1 0 1 1 0 1 1 1 1 1 1 1 1 12 11–12Alabama 45 1 1 1 1 0 0 1 0 1 1 1 1 1 1 0 1 12Veruno11,46,47 1 1 0 1 0 0 1 0 1 1 1 1 1 1 0 1 11 9–11Montreal48 1 0 0 1 0 0 1 0 1 1 1 1 1 0 1 1 10 –Magdeburg49 1 0 0 1 0 0 1 1 1 1 0 1 1 1 1 0 10 –




Table 4 Study characteristics

Cohort recruitment Case definition and numberenrolled in cohort

Follow-up period % lost to follow-up

Acute care cohorts

Colorado, USA12,27 ICD-9 codes 800.0–801.9, 803.0–804.9,850.0–854.1 (n¼2,77112; n¼ 1059complete cases27)

1 year12; 12–18 months27 42.6%12; 41.9%27

Groningen,The Netherlands32–34

PTA 1 day–60 days32/GCS 9–14 and PTA1 hour–28 days33,34 (n¼ 6032; n¼ 7033;n¼ 6734)

1 year and 2–5 years32;1 year33,34

5% (1 year); 15%(2–5 years)32;4%33

Toronto, Canada37 Responsive within one month (n¼ 92) 1 and 4 years 37% (1 year); 50%(4 years)

Seattle, USA38,39 Loss of consciousness or PTA >1 hour, orother objective evidence (n¼ 44838;n¼ 466 complete cases39)

1 year 9%39

Alabama, USA45 Acute care length �3 days (n¼ 116 com-plete cases)

1 and 2 years 37% (2 years)

Magdeburg, Germany49 GCS 3–12, and intracranial injury on com-puted tomography (n¼ 34 completecases)

3–8 years Unclear

Acute care and inpatient rehabilitation cohorts

Traumatic BrainInjury ModelSystems Database,USA13–26

Receipt of acute care and inpatient rehabi-litation in Model Systems (n¼ 301complete cases)15; (n¼ 292 completecases)14; (n¼ 71 complete cases)16;(n¼ 2362)13; (n¼538)17; (n¼ 293complete cases)18; (n¼ 132 completecases)20; (n¼99 complete cases)19;(n¼ 59 complete cases)21; (n¼ 423complete cases)22; (n¼ 1839)23;(n¼ 637)24; (n¼ 633)25; (n¼586)26

1 and 5 years14,15; 4months–10 years16; 1year13,20–24,26; 1, 2, 3, 4,and 5 years17; 1–4years18; 1–3 years19; 1, 2and 3 or 4 years25

1 year: 58%14,15;54.4%13; 16.2%17;39%22;48%26 5 years:54%15, 46%14;77.7%17

Melbourne, Australia36 Working at time of injury (n¼ 517,subsample used n¼ 74 complete cases)

2 years 50.9%

Columbus,USA42–44

Admitted to brain injury unit (n¼ 351)42;(n¼ 340)43; (n¼ 357)44

1 year42,43; 1 and 2 years44 1 year: 42%42,43;22%44; 2 years:26.6%44

Inpatient rehabilitation cohorts

Westmead, Australia40 Admitted to brain injury rehabilitation(n¼ 110)

6 months and 2 years 50% (2 years)

Rome, Italy41 GCS �8, spontaneous eye opening aftermore than 1 week, PTA �4weeks (n¼ 37)

3, 6, and 12 months Unclear

Veruno, Italy11,46,47 Severe injury (n¼ 143 complete cases)46;(n¼ 21)11; (n¼ 27)47

> 1 year46; 6 months and 1year11; 1 year47

Unclear

Montreal48 No vegetative state (n¼35 completecases)

1.5 years Unclear

Outpatient rehabilitation cohort

Houston, USA35 Followed-up >3 months post discharge(n¼ 76 complete cases)

22.5 months Unclear

ICD-9, International Classification of Disease Ninth Revision; GCS, Glasgow Coma Scale; PTA, post-traumatic amnesia.




Table 5 Outcome measures used and prognostic factors related to them

Cohort Outcome measure Significant prognostic factors and non-significant factorsa

Acute care cohortsColorado12 Worse cognitive alertness

behaviourModerate injuries, higher injury severity score, age <65,female, member of ethnic minority (R 2¼3.9%)

Worse handicap and socialintegration

Violence-related injury, severe injury, moderate injuries,higher injury severity score, age �65, females, members ofethnic minority, not working at time of injury, single, receivedgovernment funding (R 2¼22.5%)

Colorado27 More independence Admission in a long-term care facility (nursing home), severeinjuries, moderate injuries, age �65 years, female, membersof minorities, not working at injury, government fundingNot significant: Inpatient rehabilitation, outpatient services

Worse handicap and socialintegration

Receiving inpatient rehabilitation, admission in a long-termcare facility (nursing home), receiving outpatient services,severe injuries, moderate injuries, female, members of mino-rities, not working at injury, government fundingNot significant: Age

Worse cognitivesymptoms

Receiving inpatient rehabilitation, admission in a long-termcare facility (nursing home), receiving outpatient services,severe injuries, female, age �65 years, members of minori-ties, not working at injury, government funding.Not significant: Moderate injuries

Worse health status Receiving inpatient rehabilitation, admission in a long-termcare facility (nursing home), receiving outpatient services,severe injuries, female, members of minorities, not workingat injury, government funding

Groningen33 Higher% return to work Fewer symptoms 3 months, shorter post-traumatic amnesia,fewer cognitive problems 1 year (R 2¼ 42%)Not significant: Age, sex, level of education

Toronto37 More psychosocial distress Female, lower GCS, longer length of coma, longer length ofstay, more days to recall (R2¼ 29%)

Higher% return toproductivity

Lower GCS, longer length of coma, more days to recall(R 2¼ 34%)

Seattle38 Higher% return to work Shorter time to follow commands, stable pre-injury employ-ment, younger age, working after 1 month (regression treeanalysis)

Alabama45 Improved quality of life 1–2years

Path model (R 2¼ 49%): direct pathways: higher employmentstability, better FIM at 24 months, less impairments, non-white race. Indirect pathways: presence of self blame,strong family support, ability to pay for services, better FIMat 12 months, less rehabilitation

Acute care and inpatient rehabilitation cohortsTBIMS14 Improvement and deterioration of independence 1–5 years

Expression and compre-hension improvement

Lower year 1 FIM-cognitive, lower Wechsler AdultIntelligence Scale-Block Design

Social interactionimprovement

Lower Year 1 FIM-cognitive

(continued )




Table 5 Continued


TBIMS16 Positive driving status Better neuropsychological functioning, no awareness problems,better ability to drive safely (perceived by patient and significantother) (R 2¼ 53%)

Estimated miles drivenpost injury

Years post injury, DRS at discharge, neuropsychological functioning,significant others’ perceived safe driving ability, patients’ perceivedsafe driving ability (R 2¼ 33%)

Driving safety Years post injury, DRS at discharge, neuropsychological functioning,significant others’ perceived safe driving ability, patients’ perceivedsafe driving ability (R 2¼ 30%)

TBIMS18 Productivity Patients that completed at least 1 neuropsychological test at reha-bilitation and not injured in assault were more productive thenpatients that did not compete 1 neuropsychological test not testedin rehabilitation and injured in assault

TBIMS19 Higher% employment More years of education, pre-injury productive, lower (better) DRSscore at admission rehabilitation, lower (better) DRS score at dis-charge rehabilitation (R 2¼ 26%)

TBIMS21 Higher community integra-tion self report

Younger age, higher REY Auditory Verbal Learning Test, worse TMT-A, better TMT-B (R 2¼ 38%)

Higher community integra-tion significant other report

Younger age, higher REY Auditory Verbal Learning Test, worse TMT-A, better TMT-B (R 2¼ 29%)

Higher home integration Younger age, higher REY Auditory Verbal Learning Test, worse TMT-A, better TMT-B (R 2¼ 20%)

Social integration Younger age, higher REY Auditory Verbal Learning Test, worse TMT-A, better TMT-B (R 2¼ 21%)

TBIMS22 Higher FIM motor Higher FIM motor at discharge rehabilitation (R2¼ 49%)Not significant: Aetiology, age, GCS score, length of unconscious-ness, gender, marital status, rehabilitation discharge FIM cognitive,race, education level, productive activity, acute payer, rehabilitationpayer, blood alcohol level, alcohol use (1 year).

Higher FIM cognitive Higher FIM motor and cognitive at discharge rehabilitation(R2¼ 32%)Not significant: Aetiology, age, GCS score, length of unconscious-ness, gender, marital status, race, education level, productive activ-ity, acute payer, rehabilitation payer, blood alcohol level, alcohol use(1 year)

Higher home integration Higher level of education (R 2¼ 12%)Not significant: Aetiology, age, GCS score, length of unconscious-ness, gender, marital status, rehabilitation discharge FIM motor,rehabilitation discharge FIM cognitive, race, productive activity,acute payer, rehabilitation payer, blood alcohol level, alcohol use (1year)

Higher social integration Younger age, white race, higher education level (R 2¼ 17%)Not significant: Aetiology, GCS score, length of unconsciousness,gender, marital status, rehabilitation discharge FIM motor, rehabilita-tion discharge FIM cognitive, productive activity, acute payer, reha-bilitation payer, blood alcohol level, alcohol use (1 year)

Higher productivity Higher productivity before injury (R 2¼ 23%)Not significant: Aetiology, age, GCS score, length of unconscious-ness, gender, marital status, FIM motor, rehabilitation discharge FIMcognitive, race, education level, acute payer, rehabilitation payer,blood alcohol level, alcohol use (1 year)

Higher communityintegration

Younger age, higher education level, white race, higher productivitybefore injury (R 2¼ 22%)

(continued )




Table 5 Continued


Not significant: Aetiology, GCS score, length of unconsciousness, gender, maritalstatus, rehabilitation discharge FIM motor, rehabilitation discharge FIM cognitive,acute payer, rehabilitation payer, blood alcohol level, alcohol use (1 year)

TBIMS25 More productivity Younger age, length of coma, length of acute care stay, length of rehabilitation stay,rehabilitation admission DRS (R 2¼ 34%)

Social integration Not significant: Age, length of coma, length of acute care stay, length of rehabilitationstay, rehabilitation admission DRS

Home integration Not significant: Age, length of coma, length of acute care stay, length of rehabilitationstay, rehabilitation admission DRS

TBIMS26 Better FIM motor Not injured by falls, younger age, gender, type of funding, better FIM motor atrehabilitation discharge, better FIM cognitive at rehabilitation discharge (R 2¼ 39%)

Better FIM cognitive Not injured by falls, younger age, higher FIM cognitive at rehabilitation discharge,higher Rancho Los Amigos Levels of Cognitive Functioning Scale at discharge reha-bilitation, marital status at follow-up (R 2¼ 23%)Not significant: GCS score

Higher homeintegration

Younger age, type of funding, better Rancho Los Amigos Levels of CognitiveFunctioning Scale at rehabilitation discharge, better DRS at rehabilitation discharge,living alone (R 2¼ 17%)

Higher socialintegration

White race, shorter post-traumatic amnesia, younger age, better DRS at rehabilitationdischarge, higher family income, no drug use (R 2¼ 24%)Not significant: Aetiology

Higher productivity White race, no violence related injury, younger age, better DRS at rehabilitation dis-charge, higher family income, alcohol use (R 2¼ 25%)

Higher communityintegration

White race, younger age, better DRS at rehabilitation discharge, higher family income,alcohol use, no drug abuse (R 2¼ 28%)Not significant: post-traumatic amnesia

Melbourne36 Employment status Age, initial GCS score, total DRS score at admission rehabilitationColumbus42 Higher life satisfaction Pre-injury employment, higher FIM motor score at discharge rehabilitation, no sub-

stance abuse history (R 2¼ 14%)

Higher productivity Younger age, pre-injury employment, higher FIM cognitive at discharge rehabilitation,no history of substance abuse (R 2¼ 16%)

Columbus44 Higher life satisfaction Not depressed at 2 years, socially integrated at 2 years, employed at 2 years, nohistory of substance abuse (R 2¼ 30%)

Improvement of satis-faction with life 1–2years post-injury

Not married at 2 years, not depressed at 2 years (R 2¼8%)

Inpatient rehabilitation cohorts

Westmead40 Employment Younger age (R 2¼ 19%), pre-injury employed (R 2¼ 18%), employed at 6 months(R 2¼ 24%), less psychological distress at 6 months (R 2¼ 26%)

Veruno46 Functionalindependence

Epilepsy patients had lower Functional independence scores than no epilepsypatients, controlled for GCS at time of injury, duration of coma, severity of computedtomography scan, age

Disability Epilepsy patients had higher disability scores than no epilepsy patients, controlled forGCS at time of injury, duration of coma, severity of computed tomography scan, age

Outpatient rehabilitation cohort

Houston35 Higher% employment Pre-injury substance use (OR¼ 0.12, 0.02–0.62)

aNon-significant are in italics.OR, odds ratio (OR >1 has a protective effect); TBIMS, Traumatic Brain Injury Model Systems database; FIM, FunctionalIndependence Measure; GCS, Glasgow Coma Scale; DRS, Disability Rating Scale; TMT-A, Trail Making Test part A;TMT-B, Trail Making Test part B.




handicap, disability, psychosocial distress, socialintegration, return to work, quality of life andindependence. The heterogeneity of the prognosticfactors and outcome measures precluded statisticalpooling and necessitated a qualitative summary ofthe results.

Overall levels of evidenceWe established the levels of evidence for deter-

minants of disability (Table 6) with the results ofthe multivariate analysis that was performed in theindividual studies. To establish the levels ofevidence for being non-productive (Table 7), weused the results of the univariate or multivariateanalysis that was performed by the studies.Frequently it was unclear which prognostic factors

were actually tested and often only significantresults were presented. Therefore, it was notalways clear whether factors absent in the resultswere found to be not significant, or whether theywere not tested. Therefore we used only negativeresults when they were reported as non-significantin the paper.

In the 35 selected articles, 51 determinants ofdisability (multivariate results) and 82determinants of productivity (univariate andmultivariate results) were registered. Onlyprognostic factors examined in more than onecohort are presented. Several publicationsof one cohort reported contradictoryresults.12,19,21,22,26,27 Differences in size andinclusion criteria within the individual papersmight explain these contradictory results.

Table 6 Levels of evidence for prognostic factors and their associations with disability

Prognostic factor Cohorts assessed Positive findings Negative findings Level of evidence

Older age 7 7/7 (100%) 0/7 (0%) StrongFemale gender 3 2/3 (66.7%) 1/3 (33.3%) InconclusiveUnemployment at time of injury 4 4/4 (100%) 0/4 (0%) StrongLower (worse) Glasgow Coma Scale score 3 2/3 (66.7%) 1/3 (33.3%) InconclusivePre-injury substance abuse 3 3/3 (100%) 0/3 (0%) StrongHigher (worse) Disability RatingScale score at discharge rehabilitation

2 2/2 (100%) 0/2 (0%) Strong

Positive findings were considered ORs >2 or <0.5, or significant associations (P< 0.05).

Table 7 Levels of evidence for prognostic factors and their associations with being non-productive

Prognostic factor Cohorts assessed Positive findings Negative findings Level of evidence

Older age 7 4/7 (57.1%) 3/7 (42.9%) InconclusiveFemale gender 4 0/4 (0%) 4/4 (100%) Strong noUnemployment at time of injury 5 4/5 (80%) 1/5 (20%) StrongLower education level 2 0/2 (0%) 2/2 (100%) Strong noFewer years of education 4 2/4 (50%) 2/4 (50%) InconclusiveLower (worse) Glasgow Coma Scale score 6 2/6 (33.3%) 4/6 (66.7%) InconclusiveLonger post-traumatic amnesia 3 3/3 (100%) 0/3 (0%) StrongViolence-related aetiology 3 1/3 (33.3%) 2/3 (66.7%) InconclusiveLonger loss of consciousness/coma 3 2/3 (66.7%) 1/3 (33.3%) InconclusivePre-injury substance abuse 2 2/2 (100%) 0/2 (0%) StrongLonger length of stay in acute care/hospital 2 1/2 (50%) 1/2 (50%) InconclusiveLower Functional Independence Measurecognitive score at discharge rehabilitation

2 1/2 (50%) 1/2 (50%) Inconclusive

Higher (worse) Disability Rating Scale scoreat admission rehabilitation

2 2/2 (100%) 0/2 (0%) Strong

Positive findings were considered RRs or ORs >2 or <0.5, or significant associations (P< 0.05).




If there was a clear majority in favour of oragainst a determinant, the results were reportedin concordance with the majority. If evidencewas conflicting within one cohort, it was leftout of the analysis.

DisabilityOlder age, pre-injury unemployment, pre-injury

substance abuse, and more severe disability atrehabilitation discharge (measured with theDisability Rating Scale) were strong predictorsfor disability. For female gender and lowerGlasgow Coma Scale scores the evidence wasinconclusive. Limited evidence was found for:non-white race, violence-related aetiology, longerlength of coma, and fewer years of education.Only one cohort found significant results forthese predictors.

ProductivityPre-injury unemployment, longer post-

traumatic amnesia, more severe disability at reha-bilitation admission (measured with DisabilityRating Scale), and pre-injury substance abusewere strong predictors of being non-productive.Female gender and a lower education level werenot predictors. Inconclusive evidence was foundfor older age, fewer years of education, lowerGlasgow Coma Scale scores (more severe injury),violence-related aetiology, longer loss of uncon-sciousness, longer length of stay in acute hospital,and lower cognitive functioning at rehabilitationdischarge (measured with FunctionalIndependence Measure28–30).

Using the results of multivariate analysis forbeing non-productive did not alter the conclusionsconsiderably, except for lower GlasgowComa Scale scores. The Glasgow Coma Scalescore became a predictor. The prognostic valueof fewer years of education and longerpost-traumatic amnesia changed from inconclu-sive to limited. There was limited evidence thatviolence-related aetiology was not associatedwith being non-productive. Length of stay in anacute hospital was not tested with a multivariateanalysis in the studies.

Discussion

This systematic review has summarized theresults of 35 papers on 14 cohorts concerningthe prognostic value of various factors onlong-term activity limitations and participationrestrictions after traumatic brain injury.All cohorts were of good quality. Due toheterogeneity of prognostic factors andoutcome measures, we performed a qualitativeanalysis.

Older age, pre-injury unemployment, substanceabuse and more severe disability at rehabilitationdischarge were strong predictors for ongoingdisability. Inconclusive evidence was found forfemale gender, and lower Glasgow Coma Scalescores.

Pre-injury unemployment, longer post-traumatic amnesia, substance abuse and moredisability at rehabilitation admission werestrong predictors for being non-productive.Female gender and lower education level werenot predictors for being non-productive.Inconclusive evidence was found for older age,fewer years of education, lower Glasgow ComaScale scores, violence-related aetiology, longerloss of consciousness, longer length of stay inacute hospital, and less independence atrehabilitation discharge.

Although these profiles are best evidence forprognosis of restrictions in activities and parti-cipation one year after traumatic brain injury,they seem to be of limited value. Most factorsare not modifiable by prevention or treatment.Hence, it is not possible to change the courseand prevent a future worse outcome. Further,the identified prognostic factors are of limitedvalue in planning adequate and cost-effectivelong-term care. Which patients are at risk oflosing their jobs or of developing maritalor parental difficulties? Who is at risk of devel-oping substance abuse or will depend heavilyon the caregiver? These are examples ofunanswered questions that are of paramountimportance in clinical practice. As long asthese issues are not dealt with, traumaticbrain injury patients as well as their caregiversshould be followed-up over long periodsof time.




Limitations of the reviewFor pragmatic reasons, we searched

PubMed and Psychinfo from 1995 to April 2005.It is therefore possible that relevantpublications before that time were not includedin this review.

In the literature, different definitions fortraumatic brain injury and methods fordiagnosis are used. We chose to definetraumatic brain injury as ‘an alteration inbrain function as a result of an acute externalviolent force to the head’. We chose this ratherbroad definition of traumatic brain injuryin order to include as many as possible relevantstudies.

Publication bias might have occurred in thissystematic review. Six cohorts published morethan one paper, which might have resulted in pub-lication bias. Studies with significant results aremore easily published than studies without signifi-cant results, and were therefore easier to find.31

Further, studies published in additional languageswere not included in the review.

Cohorts with multiple publications may havereceived higher methodological quality scoresbecause the information was retrieved from allavailable articles originating from the samecohort. Lacking information in one article couldbe completed with information from the other arti-cles, resulting in a higher quality score for cohortswith more than one paper. Nevertheless, this strat-egy was used to prevent exclusion ofvaluable information on additional determinantsdue to incomplete descriptions in an individualpaper.

Levels of evidence for determinants of beingnon-productive were established with results ofunivariate or multivariate analysis performed inthe individual studies. The use of results fromunivariate analysis might have biasedconclusions because these results were notadjusted for confounding. This pragmatic choicewas made because it provided additionalinformation on some determinants forproductivity. Using only results of multivariateanalysis on productivity did not alter theconclusions, except for lower Glasgow ComaScale scores. Further investigation on this factoris needed in the future.

Recommendations for future researchIn the reviewed studies, activity limitations and

participation restrictions were assessed with manydifferent outcome measures. Therefore compari-sons between the studies were complicated.International consensus in the assessment ofactivity limitations and participation restrictionswould improve comparability of studies.

In general, it is important that additional, largeprospective cohort studies be performed tostrengthen the conclusions on prognostic factors.If possible, these studies should also includemodifiable prognostic factors, in order to developfuture clinical interventions.

AcknowledgementsThis project has been performed as part of the

‘Long term prognosis of functional outcome inneurological disorders’, supervised by the depart-ment of Rehabilitation Medicine of the VUMedical Center, Amsterdam and supported bythe Netherlands Organisation for HealthResearch and Development (project: 1435.0020).On behalf of the FuPro-II study group:GJ Lankhorst, J Dekker, AJ Dallmeijer,MJ IJzerman, H Beckerman, V de Groot:VU University Medical Center Amsterdam (pro-ject coordination); E Lindeman, G Kwakkel, IGLvan de Port: University Medical Center Utrecht/Rehabilitation Center De Hoogstraat, Utrecht; HJStam, AHP Willemse-van Son: Erasmus MedicalCenter, Rotterdam; GM Ribbers: Rehabilitation

Clinical messages

� Older patients with pre-injury unemploy-ment, pre-injury substance abuse or moredisability at rehabilitation discharge shouldbe considered at risk for long-term disability.

� Patients with pre-injury unemployment,longer post-traumatic amnesia, pre-injurysubstance abuse or more disability at rehabi-litation admission, should be considered atrisk for being non-productive.




Center Rijndam, Rotterdam; GAM van den Bos:Academic Medical Center, University ofAmsterdam, Amsterdam.

References

1 Langlois JA, Rutland-Brown W, Thomas KE.Traumatic brain injury in the United States: emer-gency department visits, hospitalizations, and deaths.Centers for Disease Control and Prevention,National Center for Injury Prevention and Control,2004.

2 Andlin-Sobocki P, Jonsson B, Wittchen HU,Olesen J. Cost of disorders of the brain in Europe.Eur J Neurol 2005; 12(suppl 1): 1–27.

3 A systematic review of brain injury epidemiology inEurope. Acta Neurochir (Wien) 2006; 148: 255–68;discussion 68.

4 Berg J, Tagliaferri F, Servadei F. Cost of trauma inEurope. Eur J Neurol 2005; 12(suppl 1): 85–90.

5 Thurman DJ, Alverson C, Dunn KA, Guerrero J,Sniezek JE. Traumatic brain injury in the UnitedStates: a public health perspective. J Head TraumaRehabil 1999; 14: 602–15.

6 World Health Organization. InternationalClassification of Functioning, Disability and Health.WHO, 2002.

7 Jennett B, Bond M. Assessment of outcome aftersevere brain damage. Lancet 1975; 1: 480–84.

8 Jennett B, Snoek J, Bond MR, Brooks N.Disability after severe head injury: observations onthe use of the Glasgow Outcome Scale. J NeurolNeurosurg Psychiatry 1981; 44: 285–93.

9 Scholten-Peeters GG, Verhagen AP,Bekkering GE et al. Prognostic factors of whiplash-associated disorders: a systematic review ofprospective cohort studies. Pain 2003; 104: 303–22.

10 Altman DG. Systematic reviews of evaluations ofprognostic variables. BMJ 2001; 323: 224–28.

11 Mazzini L, Zaccala M, Gareri F, Giordano A,Angelino E. Long-latency auditory-evokedpotentials in severe traumatic brain injury. ArchPhys Med Rehabil 2001; 82: 57–65.

12 Gerhart KA, Mellick DC, Weintraub AH.Violence-related traumatic brain injury: a popula-tion-based study. J Trauma 2003; 55: 1045–53.

13 Duong TT, Englander J, Wright J et al.Relationship between strength, balance, and swal-lowing deficits and outcome after traumatic braininjury: a multicenter analysis. Arch Phys MedRehabil 2004; 85: 1291–97.

14 Hammond FM, Hart T, Bushnik T, Corrigan JD,Sasser H. Change and predictors of change in

communication, cognition, and social functionbetween 1 and 5 years after traumatic brain injury.J Head Trauma Rehabil 2004; 19: 314–28.

15 Hammond FM, Grattan KD, Sasser H et al. Fiveyears after traumatic brain injury: a study of indi-vidual outcomes and predictors of change in func-tion. NeuroRehabilitation 2004; 19: 25–35.

16 Coleman RD, Rapport LJ, Ergh TC et al.Predictors of driving outcome after traumatic braininjury. Arch Phys Med Rehabil 2002; 83: 1415–22.

17 Keyser-Marcus LA, Bricout JC, Wehman P et al.Acute predictors of return to employment aftertraumatic brain injury: a longitudinal follow-up.Arch Phys Med Rehabil 2002; 83: 635–41.

18 Boake C, Millis SR, High Jr WM et al. Using earlyneuropsychologic testing to predict long-termproductivity outcome from traumatic brain injury.Arch Phys Med Rehabil 2001; 82: 761–68.

19 Gollaher K, High W, Sherer M et al. Prediction ofemployment outcome one to three years followingtraumatic brain injury (TBI). Brain Inj 1998; 12:255–63.

20 Cifu DX, Keyser-Marcus L, Lopez E et al. Acutepredictors of successful return to work 1 year aftertraumatic brain injury: a multicenter analysis. ArchPhys Med Rehabil 1997; 78: 125–31.

21 Ross SR, Millis SR, Rosenthal M.Neuropsychological prediction of psychosocialoutcome after traumatic brain injury. ApplNeuropsychol 1997; 4: 165–70.

22 Harrison-Felix C, Zafonte R, Mann N et al. Braininjury as a result of violence: preliminary findingsfrom the traumatic brain injury model systems.Arch Phys Med Rehabil 1998; 79: 730–37.

23 Englander J, Cifu DX, Wright JM, Black K. Theassociation of early computed tomography scanfindings and ambulation, self-care, and supervisionneeds at rehabilitation discharge and at 1 year aftertraumatic brain injury. Arch Phys Med Rehabil2003; 84: 214–20.

24 Englander J, Cifu D, Wrigley M et al. The impactof acute complications, fractures, and motordeficits on functional outcome and length of stayafter traumatic brain injury: a multicenter analysis.J Head Trauma Rehabil 1996; 11: 15–26.

25 Sanders B, Kreutzer JS, Rosenthal P,Delmonico RL, Young ME. A multicenter longi-tudinal investigation of return to work andcommunity integration following traumatic braininjury. J Head Trauma Rehabil 1996; 11: 70–84.

26 Rosenthal P, Dijkers M, Harrison-Felix C et al.Impact of minority status on functionaloutcome and community integration followingtraumatic brain injury. J Head Trauma Rehabil1996; 11: 40–57.




27 Mellick D, Gerhart KA, Whiteneck GG.Understanding outcomes based on the post-acutehospitalization pathways followed by persons withtraumatic brain injury. Brain Inj 2003; 17: 55–71.

28 Keith RA, Granger CV, Hamilton BB, Sherwin FS.The functional independence measure: a new toolfor rehabilitation. Adv Clin Rehabil 1987; 1: 6–18.

29 Linacre JM, Heinemann AW, Wright BD,Granger CV, Hamilton BB. The structure andstability of the functional independence measure.Arch Phys Med Rehabil 1994; 75: 127–32.

30 Stineman MG, Shea JA, Jette A et al. Thefunctional independence measure: tests of scalingassumptions, structure, and reliability across 20diverse impairment categories. Arch Phys MedRehabil 1996; 77: 1101–108.

31 Easterbrook PJ, Berlin JA, Gopalan R,Matthews DR. Publication bias in clinical research.Lancet 1991; 337: 867–72.

32 Spikman JM, Timmerman ME, Zomeren van AH,Deelman BG. Recovery versus retest effects inattention after closed head injury. J Clin ExpNeuropsychol 1999; 21: 585–605.

33 van der Naalt J, van Zomeren AH, Sluiter WJ,Minderhoud JM. One year outcome in mild tomoderate head injury: the predictive value of acuteinjury characteristics related to complaints andreturn to work. J Neurol Neurosurg Psychiatry1999; 66: 207–13.

34 van der Naalt J, van Zomeren AH, Sluiter WJ,Minderhoud JM. Acute behaviouraldisturbances related to imaging studiesand outcome in mild-to-moderate head injury.Brain Inj 2000; 14: 781–88.

35 Sherer M, Bergloff P, High Jr W, Nick TG.Contribution of functional ratings to prediction oflongterm employment outcome after traumaticbrain injury. Brain Inj 1999; 13: 973–81.

36 Ponsford JL, Olver JH, Curran C, Ng K.Prediction of employment status 2 years aftertraumatic brain injury. Brain Inj 1995; 9: 11–20.

37 Dawson DR, Levine B, Schwartz ML, Stuss DT.Acute predictors of real-world outcomes followingtraumatic brain injury: a prospective study. BrainInj 2004; 18: 221–38.

38 Temkin NR, Holubkov R, Machamer JE,Winn HR, Dikmen SS. Classification and regres-sion trees (CART) for prediction of function at

1 year following head trauma. J Neurosurg 1995;82: 764–71.

39 Dikmen SS, Ross BL, Machamer JE, Temkin NR.One year psychosocial outcome in head injury.J Int Neuropsychol Soc 1995; 1: 67–77.

40 Felmingham KL, Baguley IJ, Crooks J. A com-parison of acute and postdischarge predictors ofemployment 2 years after traumatic brain injury.Arch Phys Med Rehabil 2001; 82: 435–39.

41 Pierallini A, Pantano P, Fantozzi LM et al.Correlation between MRI findings and long-termoutcome in patients with severe brain trauma.Neuroradiology 2000; 42: 860–67.

42 Bogner JA, Corrigan JD, Mysiw WJ, Clinchot D,Fugate L. A comparison of substance abuse andviolence in the prediction of long-term rehabilita-tion outcomes after traumatic brain injury. ArchPhys Med Rehabil 2001; 82: 571–77.

43 Bogner JA, Corrigan JD, Fugate L, Mysiw WJ,Clinchot D. Role of agitation in prediction ofoutcomes after traumatic brain injury. Am J PhysMed Rehabil 2001; 80: 636–44.

44 Corrigan JD, Bogner JA, Mysiw WJ, Clinchot D,Fugate L. Life satisfaction after traumatic braininjury. J Head Trauma Rehabil 2001; 16: 543–55.

45 Webb CR, Wrigley M, Yoels W, Fine PR.Explaining quality of life for persons withtraumatic brain injuries 2 years after injury. ArchPhys Med Rehabil 1995; 76: 1113–19.

46 Mazzini L, Cossa FM, Angelino E et al.Posttraumatic epilepsy: neuroradiologic and neu-ropsychological assessment of long-term outcome.Epilepsia 2003; 44: 569–74.

47 Mazzini L, Pisano F, Zaccala M et al.Somatosensory and motor evoked potentials atdifferent stages of recovery from severetraumatic brain injury. Arch Phys Med Rehabil1999; 80: 33–39.

48 Stemmer B, Gahl B, Lacher S, Schoenle PW.Predicting vocational and independence statusfrom early assessment of motor, cognitive, andsocial abilities in traumatic brain injury patients.Brain Cogn 2000; 44: 25–30.

49 Woischneck D, Firsching R, Ruckert N et al.Clinical predictors of the psychosocial long-termoutcome after brain injury. Neurol Res 1997; 19:305–10.




Prognostic factors of long-term functioning and productivity after traumatic brain injury: a...

Documents

Transcript of Prognostic factors of long-term functioning and productivity after traumatic brain injury: a...