http://informahealthcare.

com/bij
ISSN: 0269-9052 (print), 1362-301X (electronic)

Brain Inj, 2013; 27(9): 1000–1007

! 2013 Informa UK Ltd. DOI: 10.3109/02699052.2013.794971

ORIGINAL ARTICLE

Predictive factors for 1-year outcome of a cohort of patients with severe

traumatic brain injury (TBI): Results from the PariS-TBI study
C. Jourdan1,2,3, V. Bosserelle4,5, S. Azerad4,5, I. Ghout5, E. Bayen3,6,7, P. Aegerter2,5, J. J. Weiss4, J. Mateo8, T. Lescot9,
B. Vigué10, K. Tazarourte11, P. Pradat-Diehl3,6,7, P. Azouvi1,2,3, & the members of the steering committee of the
PariS-TBI study
1
Service de Médecine Physique et de Réadaptation, APHP Hôpital Raymond Poincaré, Garches, France, 2Université de Versailles – Saint-Quentin en
Brain Inj Downloaded from informahealthcare.com by SUNY Upstate Medical University on 08/20/14

Yvelines, Versailles, France, 3Unité ER 6 UPMC, Paris, France, 4Centre Ressources Francilien du Traumatisme Crânien (CRFTC), APHP Hôpital
Broussais, Paris, France, 5Unité de Recherche Clinique (URC), APHP Hôpital A Paré, Boulogne, France, 6Université Pierre et Marie Curie, Paris, France,
7
Service de Médecine Physique et Réadaptation, APHP Groupe Hospitalier Pitié-Salpêtrière, Paris, France, 8Département d’Anesthésiologie, Soins
intensifs & SAMU, APHP Hôpital Lariboisière, Paris, France, 9Département d’Anesthésiologie, Soins intensifs, APHP Groupe Hospitalier Pitié-
Salpêtrière, Paris, France, 10Département d’Anesthésiologie & Soins intensifs, APHP Hôpital Bicêtre, Le Kremlin Bicêtre, France, and 11SAMU 77,
Mobile Care Unit, Hôpital Marc Jacquet, Melun, France

Abstract Keywords
Objectives: To assess outcome and predicting factors 1 year after a severe traumatic brain injury Brain injury, craniocerebral trauma, disability,
(TBI). dysexecutive questionnaire, outcome
Methods: Multi-centre prospective inception cohort study of patients aged 15 or older with a
For personal use only.

severe TBI in the Parisian area, France. Data were collected prospectively starting the day of History
injury. One-year evaluation included the relatives-rating of the Dysexecutive Questionnaire
(DEX-R), the Glasgow Outcome Scale–Extended (GOSE) and employment. Univariate and Received 28 August 2012
multivariate tests were computed. Revised 26 March 2013
Results: Among 257 survivors, 134 were included (mean age 36 years, 84% men). Good recovery Accepted 7 April 2013
concerned 19%, moderate disability 43% and severe disability 38%. Among patients employed Published online 30 May 2013
pre-injury, 42% were working, 28% with no job change. DEX-R score was significantly
associated with length of education only. Among initial severity measures, only the IMPACT
prognostic score was significantly related to GOSE in univariate analyses, while measures
relating to early evolution were more significant predictors. In multivariate analyses,
independent predictors of GOSE were length of stay in intensive care (LOS), age and
education. Independent predictors of employment were LOS and age.
Conclusions: Age, education and injury severity are independent predictors of global disability
and return to work 1 year after a severe TBI.

Introduction Several large-scale prospective studies [3, 4] have

Traumatic brain injury (TBI) is a leading cause of death and
disability worldwide and its global incidence is rising [1]. TBI
epidemiology is in constant evolution, as the incidence of TBI
related to road traffic accidents decreases in high-income
countries, while falls in the ageing population are becoming
more prevalent [1]. Intensive care and monitoring of injury
have also evolved dramatically in the past years. Prospective
epidemiological data need thus to be continuously updated
and validated [2].
Correspondence: Professor Philippe Azouvi, Service de Médecine
Physique et de Réadaptation, Hôpital Raymond Poincaré, 104, bd
Raymond Poincaré, 92380, Garches, France. Tel: + (33)1 47107082.
Fax: + (33)1 47107726. Email: philippe.azouvi@rpc.aphp.fr
increased the knowledge on early evolution of patients with
TBI and produced validated predictive models based on 13
20
admission characteristics. These studies offer prognosis
scores to predict death or unfavourable outcome as defined
by the Glasgow Outcome Scale (GOS) [5], which is a five-
level rating scale of survival and global disability. These
scores have good accuracy for 6-month outcome, but their
value in predicting later outcome needs to be determined.
Predictors of late outcome are more diverse, as they
include trauma characteristics and severity [6], but also early
evolution parameters, motor and cognitive impairments [7],
socio-demographic characteristics [8] and environmental
factors [9]. Cohort studies yield conflicting results regarding
predictors of 1-year outcome [8, 10, 11], owing to the
multiplicity of outcome measures and prognosis factors [8],
which are often inter-related [12].
Improving the accuracy of late outcome prediction is
essential to inform patients and families about the
 DOI: 10.3109/02699052.2013.794971
consequences of TBI and to implement management
strategies, such as goals of rehabilitation and vocational
support. This knowledge is of particular importance for severe
TBI, which is responsible for the heaviest burden of death and
disability [13].
The aims of this prospective inception study were (1) to
assess 1-year outcome after a severe TBI in the Parisian area,
in terms of mental function, global disability and return to
work; and (2) to estimate the predictive values of various pre-
or post-injury factors, including recently validated prognosis
scores.
Methods
Design of the PariS-TBI (Severe Traumatic Brain Injury
in the Parisian area) study
Brain Inj Downloaded from informahealthcare.com by SUNY Upstate Medical University on 08/20/14
The present study was part of a larger regional prospective
inception cohort study called Severe Traumatic Brain Injury
in the Parisian area (PariS-TBI), which was undertaken in
2005 in Paris city and its suburbs (11.6 million inhabitants,
12 000 km2) [14, 15]. Consecutive patients were included by
all mobile emergency services of the area over a 22-month
period. Criteria for inclusion were patients aged 15 or more
with a severe TBI (lowest Glasgow Coma Scale (GCS) score
[16] 8 before hospital admission, in the absence of other
causes of coma). Data from intensive care units to home
discharge were collected prospectively in all participating
For personal use only.
centres.
A total of 504 patients were included in the PariS-TBI
cohort from July 2005 to April 2007. Subjects were mainly
men (77%), mean age was 42 years (SD ¼ 20; range ¼ 15–98).
Main causes of injury were road traffic accident (53%) and
falls (35%). Mean GCS score was 5 (SD ¼ 2).
Patients’ initial assessment
Pre-injury characteristics included gender, age, education
duration and professional status (seven categories: higher/
lower managers, white/blue collar workers, non-active, retired
and students). Pre-injury history of alcohol abuse was
recorded.
Initial brain injury severity was recorded using the last
GCS score assessed before arrival at the hospital, without any
previous sedation for most patients or after a transitory stop of
sedation. Presence of a non-reactive unilateral or bilateral
mydriasis on admission was recorded. Other measures of
injury severity included time to follow commands, length of
stay in the intensive care unit and an early evaluation of global
disability, by the Glasgow Outcome Scale (GOS) [5], scored
by intensive care unit practitioners upon discharge from the
intensive care unit.
One-year follow-up
Survivors and their relatives were contacted and interviewed
by telephone by a trained neuropsychologist, once 1 year had
passed since injury. Cognitive late outcome was assessed with
the Dysexecutive questionnaire, completed by a close relative
(DEX-R) [16]. It is a standardized questionnaire measuring
occurrence of cognitive, behavioural and emotional changes
in everyday life as a result of impairments of executive
One-year outcome after severe TBI 1001
functions in persons with brain injury. The overall score of the
DEX-R ranging from 0–80 represents the sum of ratings
across the 20 questions, with higher scores representing
greater problems with executive functioning. The reliability
of the questionnaire appears better when filled out by a
relative than by the patient [17]. This measure of cognitive
outcome was chosen as impairments of executive functions
are the major source of disability after a severe TBI [16].
The GOS-Extended (GOSE) [5] was used to assess global
disability. It is an 8-point scale, ranging from death (scoring
1) to Upper Good Recovery (scoring 8), based on a structured
interview covering seven main areas (consciousness, inde-
pendence in the home, independence outside the home, work,
social and leisure activities, family and friends, return to
normal life). Place of living (home or institution) was
recorded separately.
‘Return-to-work’ was defined as being employed in a
regular professional (paid or volunteer) activity. Type of
return-to-work was categorized into return to former full-time
job on one side and job change (adaptation of work time or of
occupation) on the other.
Statistical analyses
Patients’ characteristics were described by mean and standard
deviation (SD) or median (25–75th percentiles) for continu-
ous variables and counts and percentages for categorical
variables. To control for a potential bias, the patients lost to
follow-up were compared to included patients with respect to
all relevant variables using chi-square tests for categorical
variables and ANOVAs for quantitative variables.
The variable age was transformed into a three-class
ordinate variable (below 30, 30–45 and above 45 years old)
for further univariate and multivariate analyses, as graphical
analysis seemed to show a linear relationship between return-
to-work and these three classes of age. Statistical sensitivity
tests using age as a continuous variable showed similar results.
The individual prognosis for each patient was calculated
with the prognosis score developed by the International
Mission on Prognosis in Traumatic Brain Injury (IMPACT)
study group [3]. This score, validated on large-scale inter-
national cohort studies of patients with moderate-to-severe
TBI, uses age, motor score of the GCS and pupillary
reactivity to predict probability of unfavourable outcome 6
months post-injury (death, vegetative state or severe
disability).
To evaluate the univariate association between potential
prognostic factors and each of the three main outcome
measures (DEX-R total score, GOSE category and return to
work), ANOVAs analysis, Kendall’s correlation coefficients
test, chi square test and the Cochran-Armitage tendency test
were used as appropriate. Patients in vegetative state (n ¼ 2)
were excluded for the tests on the DEX-R score. Tests on
return-to-work were completed on the patients who were
professionally active before the injury (excluding students,
retired and non-active subjects). Similar univariate tests
assessed the association between the three outcome measures.
Multivariate logistic regression models were computed to
assess the independent predictive value of prognosis factors
on GOSE and return to work. Models were computed on
 1002 C. Jourdan et al. Brain Inj, 2013; 27(9): 1000–1007

complete data. Multicollinearity was evaluated by calculating
a variance inflation factor. The multivariate model for GOSE
category used the proportional odds logistic methodology
[18]. The equal slope proportional odds assumption of
proportional model was checked by using graphical analysis.
All factors found to be significant in univariate analyses were
included as explanatory factors, except time to follow
command and disability at intensive care discharge, which
would have induced multicollinearity, and an important
sample size reduction because of missing data. The discrim-
inative performances of the logistic models were measured
via the area under the Receiver Operating Characteristic
(ROC) curve, represented by the C-index for the binary model
and by the generalized C-index for the ordinal model [19].
Adjusted Odds Ratios (OR) and their 95% Confidence
Intervals (CI) were computed.
Brain Inj Downloaded from informahealthcare.com by SUNY Upstate Medical University on 08/20/14
measures (GCS score, IMPACT prognosis score, time to
follow command, length of stay in intensive care, disability at
intensive care discharge), gender, age or history of alcohol
abuse. Professional status showed a significant difference
(p50.05); patients were more often lost to follow-up if they
were non-active pre-injury (56%), as compared to other
professional categories (26–41%). Rates of lost to follow-up
were higher for violence-related traumas than for road traffic
accidents (p ¼ 0.03).
At the time of evaluation 124 patients (93%) were living at
home. Global outcome on the GOSE is shown in Table I.
Table I. Patients’ characteristics and global outcome (n ¼ 134).
Mean Count Missing
(SD) (%) data (%)

Statistical analyses were performed with the R 2.14.0 Age groups

(R Development Core Team, http://www.R-project.org) soft- 530 years 59 (44%) 0%
ware, using the rms library for ordinal logistic regression 30–45 years 41 (31%)
445 years 34 (25%)
(Frank E Harrell Jr, 2012-03-24). Gender (men) 112 (84%) 0%
Pre-injury alcohol abuse (yes) 21 (16%) 3%
Ethical concerns Years of education 11.3 (3.0) 3%
Professional status
Patients and families were informed about the purpose of the Higher managers 11 (8%) 0%
PariS-TBI study before the data were recorded. According to Lower managers 5 (4%)
French laws, the study was approved by the Consultative White collar workers 35 (26%)
Blue collar workers 30 (22%)
Committee for Treatment of Health Research Information and Retired 14 (10%)
written consent for participation was not necessary. Non-active 12 (9%)
For personal use only.

Students 27 (20%)
Results Initial Glasgow Coma Scale 5.7 (1.8) 1%
Initial prognostic score (IMPACT) 0.6 (0.2) 0%
Among the 257 acute care survivors, 134 (52%) were Time to follow command (days) 12.0 (11.2) 13%
Length of stay in intensive 26.9 (22.9) 0%
included for the 1-year follow-up assessment (Figure 1).
care (days)
The most common reasons for being lost-to-follow-up Disability at intensive care
(n ¼ 123) were administrative reasons (unknown discharge discharge
destination, homeless, move abroad or erroneous address), Vegetative State 2 (2%) 11%
Severe disability 56 (47%)
death or refusal to answer. Median time since injury at follow- Moderate disability 30 (25%)
up evaluation was 14.3 months (25–75th percentiles ¼ 13.1– Good recovery 31 (26%)
23.4 months). In the study sample, mean age at the time of One year GOSE score
injury was 36.0 (SD ¼ 16.3), 84% were men. Main trauma Vegetative state 2 (2%) 0%
Lower severe disability 13 (10%)
mechanisms were road traffic accidents for 98 patients (73%) Upper severe disability 37 (28%)
and falls for 29 (22%). Mean initial GCS score was 5.7 Lower moderate disability 37 (28%)
(SD ¼ 1.8). Other patients’ characteristics are summarized in Upper moderate disability 20 (15%)
Table I. Lower good recovery 21 (16%)
Upper good recovery 4 (3%)
Comparison of patients lost to follow-up to patients who
were evaluated reached no significance as to severity SD, Standard Deviation; GOSE, Glasgow Outcome Scale–Extended.

Figure 1. PariS-TBI study flow chart.

 DOI: 10.3109/02699052.2013.794971 One-year outcome after severe TBI 1003
Table II. Prognostic factors for cognitive outcome—DEX-R score Table III. Prognostic factors of global disability—GOSE score
(n ¼ 132).* (n ¼ 134).

Correlation Correlation
coefficienta Mean  SE p Value coefficienta -squareb Dfb p Value
Age groups 0.03 0.6 Age groups 0.23 0.002
Genderb Gender 1.00 1 0.3
Male 23.2  1.5 0.8 Pre-injury alcohol abuse 0.71 1 0.4
Female 22.3  3.8 Years of education 0.18 50.01
Pre-injury alcoholb Initial Glasgow Coma Scale 0.14 0.05
Yes 24.1  3.4 0.7 Initial prognostic score 0.17 50.01
No 22.6  1.6 (IMPACT)
Years of education 0.20 0.002 Time to follow command 0.21 0.002
Initial Glasgow Coma Scale 0.01 0.8 (days)
Initial prognostic score 0.03 0.7 Length of stay in intensive 0.31 50.00001
(IMPACT) care (days)
Time to follow command 0.11 0.1 Disability at intensive care 0.32 50.0001
(days) discharge
Length of stay in intensive 0.07 0.2
care (days) GOSE, Glasgow Outcome Scale–Extended; df, Degrees of freedom. Age
Brain Inj Downloaded from informahealthcare.com by SUNY Upstate Medical University on 08/20/14

Disability at intensive 0.10 0.2 groups ¼ younger than 30, 30–45, 45 or older. aKendall’s correlation.
b
care discharge results of Cochran-Armitage test.

DEX-R, proxy rating of the Dysexecutive questionnaire.

*Two patients in vegetative state were excluded from these analyses.
Age groups ¼ younger than 30, 30–45, 45 or older. aKendall’s
correlation. bANOVAs.
Mean DEX-R total score was 23.0  16.1. The proportion of
patients from the whole sample who were working or studying
was 40%. Among the 81 patients employed pre-injury
For personal use only.
45 (versus patients aged 30–45) had an OR of 0.30 (95%
CI ¼ 0.07–1.23). Gender, length of education, professional
category and pre-injury alcohol abuse were not significant
predictors for return-to-work.
The DEX-R score was significantly associated with the
GOSE score (Kendall’s tau ¼ 0.26, p value50.0001), but
not with return-to-work among patients working pre-injury
(p value ¼ 0.3). GOSE score was significantly associated with

(excluding students, retired and non-active subjects), 34

(42%) had returned to work: 23 (28% of patients employed
pre-injury) had resumed their former job at an identical level
and 11 (14%) had an adaptation either of work time or of
activity.
In univariate tests involving the DEX-R score (Table II),
neither age, gender nor any characteristic relating to injury
severity showed a significant interaction with the DEX-R.
Only higher years of education were significantly associated
with a better (lower) DEX-R score.
Results of univariate tests regarding GOSE score are
presented in Table III. Negative prognosis factors were older
age, shorter length of education and a more severe brain
injury. The significant relationship between GOSE and age
groups tended to be linear across the three age groups (see
Figure 2a). Mean GOSE difference was 0.47 between patients
younger than 30 and patients aged 30–45 (p ¼ 0.09) and was
0.48 between patients aged 30–45 and patients older than 45
(p ¼ 0.09). All variables relating to severity were associated
with outcome on the GOSE, but strength of association was
lower for initial severity markers (correlation coeffi-
cient ¼ 0.14 for GCS and 0.17 for IMPACT prognosis
score) than for early evolution markers (correlation coeffi-
cient40.30, p value50.0001 for length of stay in intensive
care and early disability).
Among patients employed pre-injury, those who had
returned to work were younger and had sustained shorter
lengths of coma and of stay in intensive care, as presented in
Table IV. The relationship between return-to-work and age
groups was globally significant (p ¼ 0.003) and showed a
linear trend across the three age groups (Figure 2b). Patients
aged 30–45 (versus younger than 30) had an OR of 0.44
(95% CI ¼ 0.16–1.19) for return-to-work, patients older than
return-to-work (p value50.0001).
Results of multivariate predictive models are summarized
in Table V. The model for the late GOSE had an area under
the ROC curve of predictions of 0.72. Older age, shorter
education duration and longer length of stay in intensive care
were significant independent predictors of poor outcome. The
predictive model for return-to-work, which included age and
length of stay in intensive care, had an area under the ROC
curve of 0.73.
Discussion
In this prospective late follow-up study of adult patients with
severe TBI, global disability remained high, with only 19% of
good recovery and, respectively, 43% and 38% of moderate
and severe disability over the whole sample. This emphasizes
the need for rehabilitation care and follow-up, contrasting
with previous findings on the same sample showing that only
45% were admitted to specialized inpatient rehabilitation [14].
Rate of 1-year return-to-work was 42% for patients employed
pre-injury. In prospective studies on mixed-severity TBI [20],
recent pooled estimates were 40.7% after 1 year, while rates
of employment following a severe TBI were usually lower
[21–23]. However, it is worth noting that only 28% of this
sample had resumed their former, full-time occupation, while
the remaining patients necessitated job adaptations. In
addition, employment 1 year post-TBI is known to be
unstable [24] and there is a need for long-term follow-up of
job stability.
There was no significant association between the DEX-R
score and any of the severity-related variables. The DEX can
distinguish persons with brain injury from controls [17, 25],
but there is little data on its relationships with early measures
 1004 C. Jourdan et al.
Brain Inj Downloaded from informahealthcare.com by SUNY Upstate Medical University on 08/20/14
Figure 2. (a) Association between age and Glasgow Outcome Scale–Extended score for the whole study sample. (b) Association between age and
return-to-work rates for pre-injury employed patients.
For personal use only.
of brain injury severity. Bennett et al. [26] found no
association between family ratings of DEX scores and
length of post-traumatic amnesia in patients who were still
receiving in-hospital rehabilitation. This finding is not really
surprising, as it has repeatedly been found that neuropsycho-
logical testing within patients with severe TBI is poorly
correlated with measures of initial injury severity [16].
Moreover, several authors reported weak or non-significant
correlations between proxy ratings of the DEX and executive
impairments in standard neuropsychological evaluation [25,
27–29], whereas ratings by professionals involved in the
rehabilitation process showed higher correlation with TBI
severity and executive testings [26]. DEX-R score seemed to
be related to global 1-year disability rather than TBI early
severity. Proxy ratings of the DEX could also be predomin-
antly related to difficulties in patient–proxies interaction, as
previous findings in this sample showed that DEX scores
were significantly and independently associated with proxies’
burden of care [15].
The association between DEX-R score and educational
level was expected, as education plays a role in most cognitive
function evaluations. Its effect was, however, small, explain-
ing only 5% of the variance of the DEX-R score. In the
general population, Gerstorf et al. [30] also found a relatively
small effect of educational level on the self-assessed DEX
score, while trait anxiety explained 30% of its variance. A
similar study in the TBI population would be interesting, in
order to quantify the impact of various factors on the DEX-R
score, including assessments of mood and affect.
Most severity measures significantly influenced 1-year
disability and return-to-work in this study, although the PariS-
TBI study had included patients with exclusively severe TBI
(initial GCS  8). Prognostic models based on admission data
Brain Inj, 2013; 27(9): 1000–1007
have been validated recently on large cohorts to predict either
early death or early death and severe disability after mixed-
severity TBI [3, 4]. The results suggest that these models
could also predict later disability in survivors from severe
TBI. Scores based on these models had a better prognostic
value than GCS alone, which did not reach statistical
significance. As the prognosis value of initial GCS for late
outcome appears inconsistent [6, 11], these models offer new
opportunities in the prediction of late outcome upon hospital
admission.
Baseline severity characteristics had, however, weaker
prognosis values in univariate and multivariate analyses than
subsequent evolution characteristics, such as time to follow
command and particularly length of stay in intensive care or
disability upon intensive care discharge. Similar findings have
been reported before [11, 21, 31–34], illustrating the need for
further research to include standardized variables obtained
during the clinical course in prognosis modelling [2].
The favourable role of education duration has been found
in previous studies on post-TBI return-to-work [12, 34, 35],
although inconsistently [7, 31, 32], and on home and social
integration [36]. In this study, education duration was
significantly associated with the GOSE score, independently
of other variables. These findings are in accordance with
previous reports on the role of socio-demographic variables
on outcomes which are not directly dependant from the
subject’s environment, such as early death and disability [37].
The lack of association between education and employ-
ment was surprising, considering previous literature results
[12, 34, 35] and considering the significant association with
global disability, which was closely related to vocational
outcome. One explanation could be the type of employment
that persons with a higher educational level are liable to
 DOI: 10.3109/02699052.2013.794971 One-year outcome after severe TBI 1005
Table IV. Prognostic factors for return-to-work (n ¼ 81 employed pre-injury).*

Unemployed (n ¼ 47) Employed (n ¼ 34)

Mean  SE Count (%) Mean  SE Count (%) p Value
c
Age groups
530 years 11 (39%) 17 (61%) 0.003
30–45 years 21 (60%) 14 (40%)
445 years 15 (83%) 3 (17%)
Genderb
Male 40 (57%) 30 (43%) 0.7
Female 7 (64%) 4 (36%)
Pre-injury alcoholb
Yes 7 (54%) 6 (46%) 0.7
No 39 (59%) 27 (41%)
Years of educationa 11.1  0.5 11.6  0.5 0.5
Professional statusa
Higher managers 9 (82%) 2 (18%) 0.2
Lower managers 2 (40%) 3 (60%)
White collar workers 22 (63%) 13 (37%)
Brain Inj Downloaded from informahealthcare.com by SUNY Upstate Medical University on 08/20/14

Blue collar workers 14 (47%) 16 (53%)

Initial Glasgow Coma Scalec 5.4  0.2 5.9  0.3 0.2
Initial prognostic score (IMPACT)a 0.5  0.03 0.4  0.03 0.06
Time to follow command (days)a 14.7  2.2 8.8  1.7 0.04
Length of stay in intensive care (days)a 31.7  3.8 16.9  2.8 0.005
Disability at intensive care dischargec
Vegetative State 2 (100%) 0 (0%) 0.1
Severe disability 19 (63%) 11 (37%)
Moderate disability 9 (50%) 9 (50%)
Good recovery 11 (48%) 12 (52%)

Age groups ¼ younger than 30, 30–45, 45 or older. SE, standard error. aResults of ANOVAs. bResults of chi square test. cResults of Cochran-
Armitage test.
For personal use only.

Table V. Multivariate predictive models.*

There was no significant association between pre-injury
OR (95% CI) p Value alcohol abuse and either GOSE score or return-to-work.
Previous studies reported a negative predictive value of pre-
Predictive factors for
injury substance abuse on TBI outcome [11, 39, 40]. These
GOSE score (n ¼ 128)
Age groups 0.62 (0.41–0.95) 0.03 negative results could be explained by the collection modality
Years of education 1.14 (1.02–1.28) 0.02 of this information, which did not use a standardized
Initial prognostic score 0.38 (0.06–2.47) 0.4 questionnaire, but a qualitative yes/no evaluation from
(IMPACT)
Length of stay in 0.96 (0.95–0.98) 50.0001
family interview and medical charts, which could be less
intensive care (days) reliable. It is, however, of concern that a previous study on the
Predictive factors for same sample found that alcohol abuse had a significant effect
return to work (n ¼ 81 on the decision to refer a patient to rehabilitation or not [14].
employed pre-injury)
Age groups 0.42 (0.21–0.85) 0.02 There was a highly significant unfavourable effect of older
Length of stay in 0.96 (0.93–1) 0.02 age on GOSE score. There was a trend towards a linear
intensive care (days) relationship between age groups and GOSE, but two-by-two
comparisons did not reach significance, sample size being
*Adjusted Odds Ratios (OR) and Confidence Intervals (CI) are given for
each additional class of age (younger than 30, 30–45, 45 or older) or for possibly insufficient. Most previous studies assessed the role
each additional unit of the explanatory variable. of age through dichotomization, with the worst outcome being
GOSE, Glasgow Outcome Scale–Extended. usually found after 40–60 years old [37]. However,
recent findings showed a quasi-linear relationship between
expect. A cognitively demanding professional activity could age and outcome on the GOS [37]. The present study used
prevent an effective return-to-work at the relatively early stage three age groups for clarity of graphical presentation of
of this study, considering the fact that all patients had results, but results were similar using age as a continuous
sustained severe TBIs. This hypothesis was consistent with variable.
the low return-to-work rate of higher managers in the cohort Previous studies reported inconsistent results regarding the
(18%, versus 53% for blue collar workers). The results effect of age on employment [6, 8]. Employment rates have
concerning pre-injury professional category were, however, been found lower after 40 [31, 41]. These results are in
non-significant and subject to caution, given the small size of accordance with these latter findings, with individuals aged
professional groups of patients. The relationship between over 45 having an adjusted OR of 0.42 for return-to-work,
professional status and return-to-work in patients with TBI is compared to the 30–45 group. As illustrated in Figure 1, there
challenging [38], owing to the multiplicity of individual was also a non-significant trend for a higher proportion of job
situations and the need to analyse large-size cohorts to yield adaptation in this older group. Furthermore, a similar effect
reliable, significant results. was shown for patients aged between 30–45, compared to
 1006 C. Jourdan et al.
younger subjects. These findings strongly suggest that age has
a major influence on post-TBI return-to-work.
The strengths of this study lie in its well-defined,
homogeneous population of patients with a severe TBI in a
unique geographical area and a limited time-period.
The evaluation was prospective over more than a year post-
TBI and the unique outcome assessor ensured the homogen-
eity of assessments over the sample. Integration of
recently recommended prognosis scores and statistical meth-
odology [42], including the use of multivariate analyses,
enabled one to assess the independent prognostic value of
several factors.
One limitation is the important proportion of lost to
follow-up patients. The TBI population is difficult to follow
and to contact after the injury [43]. The Parisian area has a
wide variety of places of care and drains population from
Brain Inj Downloaded from informahealthcare.com by SUNY Upstate Medical University on 08/20/14
various regions and countries, which were additional reasons
explaining the difficulties encountered in contacting the
whole sample after 1 year. Lost to follow-up patients did not
significantly differ from included patients in terms of TBI
severity, but presented more frequently than the included
patients some socio-demographic characteristics which could
have a deleterious influence on outcome [44]: assault-related
TBI or lack of pre-injury employment. Consequently, it is
probable that socially vulnerable patients were under-
represented in the present study, which could lead to a bias
towards a slightly over-optimistic rating of late outcome and
For personal use only.
towards an under-estimation of the role of social vulnerability
on outcome.
In conclusion, the 1-year follow-up of this prospective
inception cohort of patients with severe TBI showed 19% of
good recovery, 43% of moderate and 38% of severe disability.
Among patients employed pre-injury, 42% returned to work,
but only 28% without any job change. The proxy rating of the
DEX was not significantly associated with TBI severity and
was significantly related to education duration. GOSE score
was not significantly associated with initial GCS, but with the
IMPACT prognostic score. In multivariate analyses, inde-
pendent predictive factors for disability were length of stay in
intensive care, older age and length of education. Negative
predictive factors for return-to-work were length of stay in
intensive care and age. The negative impact of age on GOSE
score and employment seemed to be similar over all age
classes.
Acknowledgements
The authors thank all patients and family participants who
took the time to share their experience and all members of the
CRFTC Steering Committee (Centre Ressource Francilien des
Traumatisés Crâniens) for their valuable help. The authors
also thank Pr Pernot, Dr Dulou, Pr Tadie, Pr Truelle, Dr
Welschbillig and Dr Zouaoui for their participation in the
study.
Declaration of interest
This study was funded by a grant of the French Ministry of
Health (Programme Hospitalier de Recherche Clinique 2004,
AOM04084), sponsored by AP-HP (Département de la
Recherche Clinique et du Développement) and carried out
Brain Inj, 2013; 27(9): 1000–1007
with the support of Unité de Recherche Clinique Paris-Ouest.
The authors report no conflicts of interest.
References
1. Maas AI, Stocchetti N, Bullock R. Moderate and severe traumatic
brain injury in adults. Lancet Neurology 2008;7:728–741.
2. Lingsma HF, Roozenbeek B, Steyerberg EW,
Murray GD, Maas AIR. Early prognosis in traumatic brain
injury: From prophecies to predictions. Lancet Neurology 2010;9:
543–554.
3. Steyerberg EW, Mushkudiani N, Perel P, Butcher I, Lu J,
McHugh GS, Murray GD, Marmarou A, Roberts I, Habbema
JDF, et al. Predicting outcome after traumatic brain injury:
Development and international validation of prognostic scores
based on admission characteristics. PLoS Medicine 2008;5:e165;
discussion e165.
4. Perel P, Arango M, Clayton T, Edwards P, Komolafe E, Poccock S,
Roberts I, Shakur H, Steyerberg E, Yutthakasemsunt S. Predicting
outcome after traumatic brain injury: Practical prognostic models
based on large cohort of international patients. British Medical
Journal 2008;336:425–429.
5. Wilson JT, Pettigrew LE, Teasdale GM. Structured interviews for
the Glasgow Outcome Scale and the extended Glasgow Outcome
Scale: Guidelines for their use. Journal of Neurotrauma 1998;15:
573–585.
6. van Velzen JM, van Bennekom CAM, Edelaar MJA, Sluiter JK,
Frings-Dresen MHW. Prognostic factors of return to work after
acquired brain injury: A systematic review. Brain Injury 2009;23:
385–395.
7. Andelic N, Stevens LF, Sigurdardottir S, Arango-Lasprilla JC,
Roe C. Associations between disability and employment 1 year
after traumatic brain injury in a working age population. Brain
Injury 2012;26:261–269.
8. Ownsworth T, McKenna K. Investigation of factors related to
employment outcome following traumatic brain injury: A critical
review and conceptual model. Disability and Rehabilitation 2004;
26:765–783.
9. Levack WMM, Kayes NM, Fadyl JK. Experience of recovery and
outcome following traumatic brain injury: A metasynthesis of
qualitative research. Disability and Rehabilitation 2010;32:
986–999.
10. Crépeau F, Scherzer P. Predictors and indicators of work status after
traumatic brain injury: A meta-analysis. Neuropsychological
Rehabilitation 1993;3:5–35.
11. Willemse-van Son AHP, Ribbers GM, Verhagen AP,
Stam HJ. Prognostic factors of long-term functioning and prod-
uctivity after traumatic brain injury: A systematic review of
prospective cohort studies. Clinical Rehabilitation 2007;21:
1024–1037.
12. Schönberger M, Ponsford J, Olver J, Ponsford M, Wirtz M.
Prediction of functional and employment outcome 1 year after
traumatic brain injury: A structural equation modelling approach.
Journal of Neurology, Neurosurgery, and Psychiatry 2011;82:
936–941.
13. Tagliaferri F, Compagnone C, Korsic M, Servadei F, Kraus J.
A systematic review of brain injury epidemiology in Europe. Acta
Neurochirurgica (Wien) 2006;148:255–268; discussion 268.
14. Jourdan C, Bayen E, Bosserelle V, Azerad S, Genet F, Fermanian C,
Aegerter P, Pradat-Diehl P, Weiss J-J, Azouvi P, et al. Referral to
rehabilitation after severe traumatic brain injury: Results from the
PariS-TBI Study. Neurorehabilitation and neural repair
2013;27:35–44.
15. Bayen E, Pradat-Diehl P, Jourdan C, Ghout I, Bosserelle V,
Azerad S, Weiss J-J, Joël M-E, Aegerter P, Azouvi P. Predictors
of informal care burden 1 year after a severe traumatic brain
injury: Results from the PariS-TBI study. The Journal of Head
Trauma Rehabilitation [Internet] 2012. Available online at:
http://www.ncbi.nlm.nih.gov/pubmed/22691963, accessed 7 May
2013.
16. Azouvi P, Vallat-Azouvi C, Belmont A. Cognitive deficits after
traumatic coma. Progress in Brain Research 2009;177:89–110.
17. Burgess PW, Alderman N, Evans J, Emslie H, Wilson BA.
The ecological validity of tests of executive function. Journal of
the International Neuropsychological Society 1998;4:547–558.
 DOI: 10.3109/02699052.2013.794971
18. Ananth CV, Kleinbaum DG. Regression models for ordinal
responses: A review of methods and applications. International
Journal of Epidemiology 1997;26:1323–1333.
19. Harrell FE. Regression modeling strategies: With applications to
linear models, logistic regression, and survival analysis. New York:
Springer; 2001.
20. van Velzen JM, van Bennekom CAM, Edelaar MJA, Sluiter JK,
Frings-Dresen MHW. How many people return to work after
acquired brain injury?: A systematic review. Brain Injury 2009;23:
473–488.
21. Dikmen SS, Temkin NR, Machamer JE, Holubkov AL, Fraser RT,
Winn HR. Employment following traumatic head injuries. Archives
of Neurology 1994;51:177–186.
22. Ruff RM, Marshall LF, Crouch J, Klauber MR, Levin HS, Barth J,
Kreutzer J, Blunt BA, Foulkes MA, Eisenberg HM. Predictors of
outcome following severe head trauma: Follow-up data from the
Traumatic Coma Data Bank. Brain Injury 1993;7:101–111.
23. Bounds TA, Schopp L, Johnstone B, Unger C, Goldman H.
Gender differences in a sample of vocational rehabilitation clients
with TBI. NeuroRehabilitation 2003;18:189–196.
24. Kreutzer JS, Marwitz JH, Walker W, Sander A, Sherer M, Bogner J,
Brain Inj Downloaded from informahealthcare.com by SUNY Upstate Medical University on 08/20/14
One-year outcome after severe TBI 1007
value of acute injury characteristics related to complaints and return
to work. Journal of Neurology, Neurosurgery, and Psychiatry 1999;
66:207–213.
33. Temkin NR, Holubkov R, Machamer JE, Winn HR, Dikmen SS.
Classification and regression trees (CART) for prediction of
function at 1 year following head trauma. Journal of
Neurosurgery 1995;82:764–771.
34. Gollaher K, High W, Sherer M, Bergloff P, Boake C, Young ME,
Ivanhoe C. Prediction of employment outcome one to three years
following traumatic brain injury (TBI). Brain Injury 1998;12:
255–263.
35. Ketchum JM, Almaz Getachew M, Krch D, Baños JH, Kolakowsky-
Hayner SA, Lequerica A, Jamison L, Arango-Lasprilla JC. Early
predictors of employment outcomes 1 year post traumatic brain
injury in a population of Hispanic individuals. NeuroRehabilitation
2012;30:13–22.
36. Harrison-Felix C, Zafonte R, Mann N, Dijkers M, Englander J,
Kreutzer J. Brain injury as a result of violence: Preliminary findings
from the traumatic brain injury model systems. Archives of
Physical Medicine and Rehabilitation 1998;79:730–737.
37. Mushkudiani NA, Engel DC, Steyerberg EW, Butcher I, Lu J,

Fraser R, Bushnik T. Moderating factors in return to work and job
stability after traumatic brain injury. Jornal of Head Trauma
Rehabilitation 2003;18:128–138.
25. Boelen DHE, Spikman JM, Rietveld ACM, Fasotti L. Executive
dysfunction in chronic brain-injured patients: Assessment in
outpatient rehabilitation. Neuropsychological Rehabilitation 2009;
19:625–644.
26. Bennett PC, Ong B, Ponsford J. Measuring executive dysfunction in
an acute rehabilitation setting: Using the dysexecutive question-
naire (DEX). Journal of the International Neuropsychological
Society 2005;11:376–385.
27. Chan RCK, Yan C, Qing Y-H, Wang Y, Wang Y-N, Ma Z,
Hong X-H, Li Z-J, Gong Q-Y, Yu X. Subjective awareness of
For personal use only.
Marmarou A, Slieker F, McHugh GS, Murray GD, Maas AIR.
Prognostic value of demographic characteristics in traumatic brain
injury: Results from the IMPACT study. Journal of Neurotrauma
2007;24:259–269.
38. Walker WC, Marwitz JH, Kreutzer JS, Hart T, Novack TA.
Occupational categories and return to work after traumatic brain
injury: A multicenter study. Archives of Physical Medicine and
Rehabilitation 2006;87:1576–1582.
39. Fraser R, Machamer J, Temkin N, Dikmen S, Doctor J. Return to
work in traumatic brain injury (TBI): A perspective on capacity for
job complexity. Journal of Vocational Rehabilitation 2006;25:
141–148.
40. Sherer M, Bergloff P, High Jr W, Nick TG. Contribution of

everyday dysexecutive behavior precedes ‘‘objective’’ executive
problems in schizotypy: A replication and extension study.
Psychiatry Research 2011;185:340–346.
28. Chaytor N, Schmitter-Edgecombe M, Burr R. Improving the
ecological validity of executive functioning assessment. Archives
of Clinical Neuropsychology: The Official Journal of the National
Academy of Neuropsychologists 2006;21:217–227.
29. Wood RL, Liossi C. The ecological validity of executive tests in a
severely brain injured sample. Archives of Clinical
Neuropsychology: The Official Journal of the National Academy
of Neuropsychologists 2006;21:429–437.
30. Gerstorf D, Siedlecki KL, Tucker-Drob EM, Salthouse TA.
Executive dysfunctions across adulthood: Measurement properties
and correlates of the DEX self-report questionnaire.
Neuropsychology, Development, and Cognition. Section B,
Aging, Neuropsychology and Cognition 2008;15:424–445.
31. Ponsford JL, Olver JH, Curran C, Ng K. Prediction of employment
status 2 years after traumatic brain injury. Brain Injury 1995;9:
11–20.
32. van der Naalt J, van Zomeren AH, Sluiter WJ, Minderhoud JM.
One year outcome in mild to moderate head injury: The predictive
functional ratings to prediction of longterm employment outcome
after traumatic brain injury. Brain Injury 1999;13:973–981.
41. Keyser-Marcus LA, Bricout JC, Wehman P, Campbell LR, Cifu
DX, Englander J, High W, Zafonte RD. Acute predictors of return
to employment after traumatic brain injury: A longitudinal
follow-up. Archives of Physical Medicine and Rehabilitation
2002;83:635–641.
42. Maas AIR, Steyerberg EW, Marmarou A, McHugh GS, Lingsma
HF, Butcher I, Lu J, Weir J, Roozenbeek B, Murray GD. IMPACT
recommendations for improving the design and analysis of clinical
trials in moderate to severe traumatic brain injury.
Neurotherapeutics: The Journal of the American Society for
Experimental NeuroTherapeutics 2010;7:127–134.
43. Sample PL, Langlois JA. Linking people with traumatic brain
injury to services: Successes and challenges in Colorado.
The Journal of Head Trauma Rehabilitation 2005;20:270–278.
44. Gary KW, Arango-Lasprilla JC, Ketchum JM, Kreutzer JS,
Copolillo A, Novack TA, Jha A. Racial differences in employment
outcome after traumatic brain injury at 1, 2, and 5 years postinjury.
Archives of Physical Medicine & Rehabilitation 2009;90:
1699–1707.

Appendix: Site investigators of the PariS-TBI study

 Pre-hospital: Drs Frederic, Max, Ricard-Hibon, El Sayed, Cabaret, Le Quellec, Devaux, Sebbah, Cuvier, Lambros, Binda,
Rakotonirina, Hazan, Briole, Parpet, Faggianelli, Touitou, Nguyen, Letarnec, Soupizet, Chollet-Xemard, Adnet, Luis,
Lapostolle, Hennequin, Beruben, Telion, Kim An, Naon, Kierzek, Terraz, Ecollan, Latremoulle, Petit, Cabane, Lagrange, De
Stabenrath, Carli.
 Acute care: Drs Jost, Dolveck, Puybasset, Caille, Siyam, Mohebbi, Benayoun, Mantz, Bonneville, Paugam, Foucrier, Greef,
Trouiller, Alves, Pease, Restoux, Lakovlev, Mireau, Descorps-Declere, Marechal, Percheron, Gontier, Meyer, Fraisse, Vu Dinh,
Van De Louw, Imbert, Abadie, Hurel, Berbineau, Ho, Godier, Meleard, Loeb, Guinvarch, Frappier, Bellenfant, Tremey,
Chedevergne, Cerf, Labat, Yakhou, Heurtematte, Faucheux, At Mamar, Suen, Fernand, Bonnet, Grandclerc, Monier, Ract,
Engrand, Geerarerts, Laplace, Launy, Sitbon, Martin, Martinais, Payen, Rezlan, Rabuel, Losser, Cholet, Varjanian, Coulaud,
Abarrategui, Bekaert, Saidi, Debien, Abdennour, Langeron, Robin, Xin-Lu, Kahn, Breant, Denys, Roche, Hamada, Paules,
Bresson, Dakhlaoui, Repesse, Fangio, Richecoeur, Galliot, Boulet, Blanc, Pipien, Boufferrache, Pepion, Oswald, Merat, Bedos.
 Rehabilitation: Drs Canny Verrier, Thevenin Lemoine, Tiravy Silber, Witas, Rhein, Bonan, Bradai, Yelnik, Montagne,
Vivant, Darnault, De Crouy, Selma, Peskine, Genet, Lagniez Girardeau, Schnitzler, Gion, Memin, Gracies.