Disability and quality of life 20 years after traumatic brain injury

Abstract Objectives The study describes functional outcomes and health‐related quality of life (HRQL) in patients with traumatic brain injury (TBI) 20 years postinjury. Materials and Methods Forty‐four survivors who acquired moderate and severe TBI during 1995–1996 were followed 10 and 20 years postinjury. Outcomes were Glasgow Outcome Scale Extended (GOSE), Community Integration Questionnaire (CIQ), and SF‐36 questionnaire (SF‐36). Multiple regressions were performed to examine the relationship between follow‐up measurements, controlling for baseline demographics and injury severity. Results There were no significant differences in baseline age and civil status between moderate and severe TBI, but patients with severe injury had significantly lower employment rates (p = 0.05). Mean age at 20‐years follow‐up was 50.8 (SD 11.4) years, and 73% were males. Most patients showed good recovery (52%) or moderate disability (43%). Disability levels remained stable between and within severity groups from 10 to 20 years. Community integration including social integration improved from 10 to 20 years (p = 0.01 and p = 0.005, respectively). HRQL remained stable, except for subscales Bodily Pain and Role Emotional (p = 0.02 and p = 0.06). Depression at 10 years and females were associated with poorer mental health, while productive activity at 10 years indicated better physical and mental health at 20 years postinjury, respectively. Conclusions Functional limitations persist even decades after moderate and severe TBI, with poorer prognosis for females and persons who were depressed at the 10‐year follow‐up. Development and evaluation of targeted long‐term follow‐up programs and access to rehabilitation services for these groups should be highlighted. Improved community integration despite stable functional limitations draws attention to long‐term adaptation to adversity and illness.


| INTRODUC TI ON
A large proportion of patients with moderate and severe traumatic brain injury (TBI) sustain long-term physical, cognitive, and emotional impairments that have a deep impact on their functioning, reintegration to society and health-related quality of life (HRQL; Andelic et al., 2009;Forslund, Roe, Sigurdardottir, & Andelic, 2013;Hammond et al., 2004;Jacobsson, Westerberg, Soderberg, & Lexell, 2009;Ponsford, Draper, & Schonberger, 2008;Schulz-Heik et al., 2016;Sigurdardottir, Andelic, Roe, & Schanke, 2009;Wilson et al., 2017). TBI outcomes up to 10 years postinjury have been documented in several studies Dahm & Ponsford, 2015;Jourdan et al., 2016). The findings suggest that patients with moderate and severe TBI experience persisting functional limitations and decreased employment rates Forslund et al., 2017;Ponsford et al., 2008), reduced physical and mental health and elevated probabilities of social isolation Hawthorne, Gruen, & Kaye, 2009). Being older and female contribute to decreased physical and mental health Forslund et al., 2013), whereas being employed or involved in productive activities 10 years after injury contributes to better physical functioning, social and psychological well-being and HRQL Ponsford et al., 2008). Taken together, the studies indicate that persons with TBI may need professional assistance to improve functioning and quality of life, even a decade postinjury.
There is limited research worldwide on the very long-term outcomes after TBI (up to 20 years postinjury; Brown et al., 2011;Hoofien, Gilboa, Vakil, & Donovick, 2001;Nestvold & Stavem, 2009;Steadman-Pare, Colantonio, Ratcliff, Chase, & Vernich, 2001), and few studies have been conducted in Europe. Among these, McMillan, Teasdale, and Stewart (2012) found that half of the young people and adults that were assessed reported disability 12-14 years after head injury. Lexell, Wihlney, and Jacobsson (2016) assessed disability 6-15 years after injury and confirmed a strong association between disability and occupational status. Wood and Rutterford (Wood & Rutterford, 2006) assessed outcomes on average 17 years after severe injury and reported that although long-term psychosocial functioning remains weakened, community integration levels were just below those reported for nondisabled patients. Jacobsson, Westerberg, & Lexell (2010)  and found no association between HRQoL and injury data (Nestvold & Stavem, 2009).
It is challenging to establish which long-term services are needed to target functional domains that patients with TBI find particularly difficult (Hoofien et al., 2001), and not only HRQL as has been performed in previous studies. Multidimensional functional assessments may help patients and care providers define common goals, and coordinate service delivery. Such studies are important both from the epidemiological and rehabilitative perspectives (Hoofien et al., 2001). The present study aims to broaden the knowledge base on the functional outcomes and factors that contribute in a long-term perspective by following a patient cohort with moderate and severe TBI from 10 to 20 years after injury. The main objective was to describe disability levels and HRQL 20 years after injury.
Second, we studied changes in global functioning, community integration and HRQL from 10 to 20 years postinjury, and identified factors associated with physical and mental health at 20-year follow-up.
Finally, HRQL in the study population was compared to the general Norwegian population.

| MATERIAL S AND ME THODS
The study includes a 20-year follow-up of a 2-year TBI cohort injured in 1995/1996, consisting of 62 patients with moderate and severe TBI who were admitted to the university-affiliated Trauma Referral Center in Oslo, Norway.
The Glasgow Coma Scale score (GCS; Teasdale & Jennett, 1974) at the time of emergency admission to the hospital was used to classify injury severity (moderate injury 9-12 vs. severe 3-8). Baseline data, including socio-demographics and injury-related factors, were extracted from the hospital's medical records at the 10-year follow-up which was completed in 2005/2006, and reported previously ).
The 20-year follow-up was performed in 2015/2016. The

Regional Committee for Medical Research Ethics, South-East
Norway approved the study (2015/389). Written informed consent was obtained. Participants were contacted by mail and/or telephone, thereafter a face-to-face interview with physiatrists (NA and TH, n = 27) was scheduled. When a direct interview was not possible, patients or their relatives were contacted by telephone (EIH and NA, n = 17) to obtain information, including socio-demographics such as living situation, employment and outcome data, that is, functional status, community integration and HRQL.

| Outcome measures applied at 10-and 20-year follow-up
Functional status was measured by the Glasgow Outcome Scale Extended (GOSE; Wilson, Pettigrew, & Teasdale, 1998). GOSE is based on a structured interview and provides an ordinal classification of disability into eight categories ranging from death to vegetative state, lower and upper levels of severe and moderate disability and lower and upper levels of good recovery.

Community integration was assessed by the Community
Integration Questionnaire (CIQ; Willer, Ottenbacher, & Coad, 1994), a 15-item scale to assess possible restrictions and effective role performance within three domains: the home integration (score ranges 0-10), social integration (score ranges 0-12) and productive activities (score ranges 0-7). Subscale scores in sum provide a total CIQ score ranging from 0 to 29. Higher scores indicate greater integration and fewer restrictions.
Health-related quality of life (HRQL) was measured by the Medical Outcomes 36-Item Short Form Health Survey (SF-36; Ware, Gandek, & IQOLA Project Group, 1994). The SF-36 measures HRQL along eight subscales: physical function (PF), role limitations due to physical health (RP), bodily pain (BP), general health (GH), vitality (VT), social function (SF), role limitations due to emotional health (RE) and mental health (MH). In addition, a single item reports the changes in overall health over the past year. Raw scores were transformed into a scale score ranging from 0 to 100 (worst to best). The subscales were calculated into the Physical Component Summary

| Statistical analysis
Descriptive statistics were used to summarize the participants' demographic and injury characteristics at baseline, as well as the demographics, and measures of function and HRQL at the 10-and 20-years follow-up. Wilcoxon signed-rank test or paired t-tests were used to assess the statistical differences between the measurements. The dependent variables in the two regression models were PCS and MCS at 20 years. Considering the small sample size, a conservative approach was applied using stepwise regression.
With this method, the Statistical program selected which variables it would enter (stepping method criteria, probability of F, entry 0.05, removal 0.10) from a provided list of independent variables (age at injury and concurrent age, gender, GCS and associated injuries and 10-year functional level (GOSE, CIQ and BDI scores). The PCS, MCS, and CIQ were modeled as interval variables while the measures of GOSE (severe/moderate disability vs. good recovery) and Beck Depression Inventory (BDI, 0-12 vs. >12) were modeled as binary.
The results were presented as adjusted R 2 and B coefficients (95% CI). The strongest models out of three were presented in Table 2. Prior to regression, multicollinearity, and model assumptions were examined using the tolerance and variance inflation factor (VIF). Distribution of the residuals was examined for normality, and influential data points were examined using Cook's distance.
Bar charts were presented to compare the percentage differences in GOSE levels, and mean differences in the CIQ and SF-36 subscales between the 10-and 20-years follow-up. The mean measures of SF-36 subscales at 10-and 20-years follow-up were further compared with the general population in Norway (Loge & Kaasa, 1998). All analyses were performed using the Statistical Package for the Social Sciences (SPSS, version 24). Statistical significance was set to p < 0.05.

| RE SULTS
Since the 10-year follow-up, 4 of the 62 patients had died. A total of 44 of 58 surviving patients (71% of the original cohort) consented to participation. Table 1 shows the main demographics and injury characteristics at baseline, 10-and 20-year follow-ups. Mean current age was 50.8 years (SD 11.4,, and 73% were males. Patients with moderate and severe TBI were similar regarding age at the time of injury, gender, civil status, and education. At the 20-year follow-up, there were no differences between severity groups concerning age and civil status, whereas employment status differed significantly, as 75% of patients with moderate TBI were in full-time jobs in contrast to 37% of patients with severe injury (p = 0.05).

| Glasgow Outcome Scale Extended
In the total sample, the median GOSE score at 20-year follow-up was 6.0 (Interquartile range, IQR 2.5). Twenty-three patients (53%) had good recovery, while 19 (43%) presented with moderate, and two (4%) with severe disability, see Table 1. There were statistically significant differences in the GOSE median scores between injury severity groups (p = 0.003); 70% of patients with moderate TBI revealed good recovery, compared to 37.5% of patients with severe TBI, see Figure 1.
There were no significant differences effects of gender (p = 0.69), nor age (p = 0.38) in the total CIQ mean scores. There was a statistically significant difference in the total CIQ mean scores between patients with moderate and severe TBI (t = −1.94, p = 0.04), but not for the CIQ subscales (see Figure 2).
A statistically significant improvement was found on the total CIQ mean scores at the 20-years follow-up compared to 10 years (means 21.2, SD 3.4 vs. 19.4, SD 4.5, respectively, p = 0.01; see Figure 2). Of the CIQ subscales, the Social Integration showed statistically significant improvements from 10 to 20 years (means 8.8 SD 2.2 vs. 9.9 SD TA B L E 1 Demographics and injury characteristics at baseline, and functional status 10 and 20 years post-TBI 1.5, respectively, p = 0.005). Mean scores changes within the severity groups from 10-to 20-year follow-up showed statistically significant differences for total CIQ mean score (moderate TBI p = 0.02, severe TBI p = 0.08), Home Integration (moderate TBI p = 0.002) and Social Integration (severe TBI p = 0.004). Productive Activity mean scores approached significance in both severity groups (p = 0.08).

| Health-related quality of life
The mean score of the SF-36 domains is displayed in Figure 3.
There were no significant differences in the mean scores of any

| Factors associated with PCS and MCS 20-years after injury
The multiple regression models are presented in Table 2. The adjusted R 2 of PCS was 29%, meaning that the predictors (BDI and CIQ productivity score at 10-years) explained almost one-third of the PCS variance at 20-years. The B coefficient was negative for BDI meaning that depression at 10-years predicted poorer PCS 20-years after TBI. Furthermore, the B coefficient was positive for CIQ productivity, implying that productive activities at 10-years predicted better PCS at 20-year.
The adjusted R 2 of MCS was 45%, meaning that factors such as gender, BDI and CIQ productivity scores at 10-years explained almost half of the variance in MCS at 20-years. The B coefficients were negative for gender and BDI meaning that female gender and depression at 10-years predicted poorer MCS at 20-years, whereas the B coefficient was positive for CIQ productivity, meaning that productive activities at 10-years predicted better MCS at 20-years.

| D ISCUSS I ON
This study described disability levels and HRQL 20 years after moderate and severe TBI, changes in global functioning, community integration and HRQL from 10 to 20 years, and factors associated with physical and mental health.
In terms of demographic variables and TBI severity, differences were found only in employment status at 20-years post injury, where 75% of patients with moderate TBI were in full-time jobs compared to 37% with severe TBI. Employment rates remained stable in the moderate TBI group from baseline to 10-and 20-year follow-up. The total percentage of working patients in the severe TBI group remained unchanged from 10 to 20 years follow-up, and is similar to numbers reported by Wood (2008). However, the number of patients with severe TBI working full-time is 12% higher at the 20-year follow-up compared to the 10-year follow-up, suggesting a long-term potential for increased work-participation, particularly in those working part-time. Nonetheless, the percentage of patients with severe TBI receiving disability pension remained unchanged. As the patients with severe TBI included in this study are still in their productive ages, a loss of work productivity results in a substantial financial burden both at the individual and societal one-third of patients with severe TBI indicating a long-term differential effect of severe and moderate TBI which is in line with findings from previous research (Forslund et al., 2017;Ponsford et al., 2008).
The overall community integration at 20-year follow-up did not differ from the general population (Corrigan, 1994), suggesting successful long-term community reintegration (Wood & Rutterford, 2006). Age and gender had no impact on community reintegration.
However, injury severity influenced community integration, in line with previous studies Sandhaug, Andelic, Langhammer, & Mygland, 2015;Winkler, Unsworth, & Sloan, 2006 in line with the study by Brown et al. (2011) which indicated that adaptation to impairment-related limitations improves as the time since injury increases.
In the evaluation of HRQL at the 20-year follow-up, no statistically significant differences were found between SF-36 dimensions and age, gender or injury severity groups. The latter is in line with results from the previously mentioned studies by Jacobsson and Nestvold (Jacobsson et al., 2010;Nestvold & Stavem, 2009).
With respect to the 10-to 20-year follow-up, reduction in bodily pain from 10 to 20 years may suggest that the physical TBI-related consequences are weakened over time, as reported in other studies (Soberg, Bautz-Holter, Finset, Roise, & Andelic, 2015).
Furthermore, individuals with TBI may adapt over time and experience a large degree of psychological and emotional normalization and better HRQL in the long-term perspective (Bonanno, 2004;Brown et al., 2011). This study has several limitations that should be addressed. First, all study participants initially included were individuals with moderate and severe TBI, between 16 and 55 years old, and received care in the South-Eastern region of Norway. This may limit the generalizability to other populations. In addition, the statistical analyses were limited by a small sample size. Attrition to follow-up is a common problem for any long-term longitudinal study, and this may introduce selection bias, as the most improved patients often remain available (Gray et al., 2017). In this study, 70% of the original cohort consented for 20-year follow-up, which we consider acceptable. Only 4% had severe disability on the GOSE, and 53% showed good outcome. We can therefore not rule out that the sample may have been biased toward individuals who survived from moderate-to-severe TBI and are living in the community, and those who were able to self-report their outcomes.
In conclusion, the rate of disability reported in the study at health and well-being. Furthermore, access to comprehensive and coordinated rehabilitation services and vocational support in the very long-term post-TBI to optimize community integration and participation in productive activities, and to educate patients and their families to prevent or reduce late-developing problems is important. This is in line with a comprehensive review regarding brain injury rehabilitation, showing that comprehensive and holistic rehabilitation can improve community integration, functional independence, and productivity, even for patients who are many years post injury (Cicerone et al., 2011).

ACK N OWLED G M ENT
The authors want to thank all of the subjects for their participation.

CO N FLI C T O F I NTE R E S T
The authors report no conflict of interest.