Quality of life up to 10 years after traumatic brain injury: a cross-sectional analysis

For this cross-sectional study, 439 consecutively admitted TBI patients were identified in the electronic database of the Schoen Rehabilitation Center, Bad Aibling, Germany, who underwent neurorehabilitation due to a mild, moderate or severe TBI between January 1, 2005 through June 30, 2015. Before admission to the Schoen Rehabilitation Center, these patients received critical care in a hospital of the Southern Upper-Bavaria Trauma Network, thereby assuring a comparable and high standard of critical and surgical care. To gain knowledge on how patients cope with the long-term consequences after TBI, these patients were invited in writing to participate in this cross-sectional analysis on November 4, 2015 (Fig. 1 and supplementary Fig. S1). The letter contained a cover letter with a request to answer the QOLIBRI questionnaire, the QOLIBRI questionnaire as well as a prepaid return envelope. Thus, HRQoL was assessed up to 10 years following TBI using the health-related and disease-specific QOLIBRI questionnaire. According to local legislation and the ethics committee of the Ludwig-Maximilians University, Munich, Germany, ethical approval was not required for this study as this cross-sectional analysis was performed as catamnesis and all data from medical records were anonymously used. In detail, this catamnesis was performed to gain knowledge how former patients cope with the long-term consequences after neurorehabilitation due to a mild, moderate or severe TBI, which does not require an ethical approval. In accordance to this approach, a second contact of patients was not allowed and not performed.

Patients equal or older than 18 years at survey who experienced an open or closed, mild, moderate, or severe TBI were eligible; patients with either intracranial bleeding without a TBI or a chronic subdural hematoma were excluded.

The QOLIBRI is a form-based questionnaire designed to measure HRQoL in patients following brain injury. It has been internationally validated in a variety of languages, including German [28, 29]. The QOLIBRI total score ranges from 0 to 100, representing the lowest and highest HRQoL, respectively. It comprises the following six scales with 37 items: cognition (7 items), self (7 items), daily life & autonomy (7 items), social relationships (6 items), emotions (5 items), and physical problems (5 items). Two major key aspects in life—satisfaction and bothered items, i.e. restrictions—are assessed by merging items 1 through 4 (cognition, self, daily life & autonomy, and social relationships) and items 5 and 6 (emotions and physical problems) [31], with an additive maximum score of 400 (merged items 1–4 with each having a maximum score of 100) and 200 (merged items 5–6 with each having a maximum score of 100), respectively. A QOLIBRI total score of 60 or higher represents good HRQoL; a score below 60 indicates unsatisfactory HRQoL with an increased risk of having either an affective or anxiety disorder, and a score below 40 indicates an increased risk of having both disorders [29]. In the case of severe cognitive and/or motor impairment, the patient’s caregiver helped the patient complete the questionnaire. HRQoL was then quantified and the results were correlated with the following demographic and basic characteristics, namely i) TBI severity (initial GCS), ii) TBI etiology (traffic accident/fall), iii) the patient’s age at the time of TBI, iv) the patient’s age at the time of the survey, v) the time elapsed between TBI and completion of the questionnaire, and vi) the patient’s sex (male/female).

To rule out non-responder bias, the following data are given for the main unit (n = 439), the non-responders (n = 251) and the QOLIBRI cohort (n = 135) and were statistically compared between the non-responders and the QOLIBRI cohort: i) TBI severity, ii) TBI etiology, iii) age at the time of TBI, iv) age at the time of the survey, v) time elapsed since TBI, vi) sex distribution of the patient cohort, vii) whether decompressive craniectomy, viii) ICP monitoring/ a shunt device, and ix) a tracheostomy was performed as well as x) time to onset of neurorehabilitation, xi) duration of neurorehabilitation, xii) functional status at admission to, and xiii) at discharge from neurorehabilitation. The relative frequency (%) of each TBI severity level was quantified using the categories: mild (TBI I°: GCS 13–15), moderate (TBI II°: GCS 9–12), or severe (TBI III°: GCS 3–8) injury, based on the initially documented score on the Glasgow Coma Scale (GCS) [36], which was extracted from the medical referral letter that was sent to the Schoen Rehabilitation Center. The relative frequency (%) of each TBI etiology was determined for the following three categories: traffic accidents, falls, and other.

In this analysis of chronic TBI patients, we addressed the common lack of GCS documentation regarding TBI severity as followed [25, 37]. 62% of patients (n = 84) were initially classified for their TBI severity (GCS) and were included in our correlation analysis (model 1 and 2). For the descriptive analysis of demographic data, we included all 135 patients. For the overall data interpretation, we assumed that most probably 58 to 70% underwent either a primary severe TBI or a secondary deterioration resulting in severe secondary brain injury, indicated by the portion of decompressive craniectomy and ICP monitoring during the acute treatment (Table 1).

Each QOLIBRI questionnaire returned was checked for completeness, and a self- or caregiver-assessed rating was recorded. Each QOLIBRI responder was then assigned an interim ID number. The QOLIBRI scores were added to the 13 demographic results listed above, which were obtained from the medical records using the interim ID numbers. Finally, all ID numbers were removed, and the entire data set was anonymized.

Statistical analysis was performed using R (R Core Team, Vienna, Austria). Group differences were calculated between the QOLIBRI cohort (n = 135) and the non-responders (n = 251) using the Mann-Whitney-U test for numeric or the Fisher test for categorical variables. All QOLIBRI data were tested for a Gaussian distribution using a QQ plot. A multivariate regression model (model 1) was used to assess the parameters that influence HRQoL measured using the QOLIBRI total score. HRQoL was modeled as a dependent variable and as a function of the following potential predictors (as independent variables): i) TBI severity (GCS), ii) TBI etiology, iii) age at the time of TBI (calculated by subtracting the age at the time of the survey from the time elapsed since TBI), iv) age at the time of the survey, v) time elapsed since TBI, and vi) sex distribution. A second multivariate regression model (model 2) was used to analyze the putative effect of TBI severity (GCS) on the QOLIBRI total score, on the two major aspects in life, namely satisfaction and restrictions, and on each of the six scales. Patients with missing data of TBI severity (n = 51) were excluded for regression analysis (model 1 and 2), thus these regression models included 84 out of 135 patients. Data are reported as a relative frequency (%), the mean ± SEM (demographics), the mean ± SD (QOLIBRI results), or the median with interquartile range (IQR_25–75). Differences were considered significant at p <  0.05. Effect size was assessed using the determination coefficients R² and adjusted R². Four of the original 139 returned questionnaires were incomplete and were therefore excluded from the analysis.

One hundred thirty-five out of 439 consecutive admitted TBI patients of the electronic database of the Schoen Rehabilitation Center, completed the analysis, from now on termed as the “QOLIBRI cohort” (Fig. 1). In detail, 251 patients did not respond to our written invitation to participate and at least 38 out of 439 potentially eligible cTBI patients had already deceased, as indicated by their relative’s answer. One hundred fifty out of 401 (37%) former patients responded to our inquiry. Eleven of the responding patients did not return the QOLIBRI questionnaire, and an additional 4 responders did not complete the questionnaire. Thus, 135 out of 150 (90%) chronic cTBI (cTBI) patients of our survey cohort provided complete information regarding their HRQoL measured using the QOLIBRI questionnaire (Fig. 1).

The QOLIBRI cohort (n = 135) and the non-responders (n = 251) did not differ with respect to TBI severity (p = 0.08), TBI etiology (p = 0.22), age at TBI (p = 0.08), age at survey (p = 0.13), elapsed time since TBI (p = 0.14), sex distribution (p = 0.63), decompressive craniectomy (p = 0.42), ICP monitoring/ shunt device (p = 0.38), tracheostomy (p = 0.23), time to onset of neurorehabilitation (p = 0.76), duration of neurorehabilitation (p = 0.08), and functional status at admission (p = 0.09). However, patients of the QOLIBRI cohort gained better functional status at discharge from neurorehabilitation with a mean (± SEM) modified Rankin Scale (mRS) of 2.3 ± 0.1 in comparison to the non-responders (n = 251) with a mean (± SEM) mRS of 3.1 ± 0.1 (p < 0.001). Thus, patients of the QOLIBRI cohort gained on average good mobility and independence in activities of daily living though unable to carry out all previous activities compared to the non-responders, who remained on average moderate disabled though able to walk unassisted (Table 1). Among the 135 patients in the QOLIBRI cohort, 13, 13, and 36% had mild, moderate, or severe TBI, respectively; TBI severity was not classified for the remaining 38% of patients. With respect to TBI etiology, 49% of cases were due to a traffic accident, 44% were due to a fall, and the remaining 7% were due to another cause. The mean (± SEM) age at the time of the survey was 53.1 ± 1.6 years, and 76% of the participants were male. 27% of patients underwent decompressive craniectomy in the acute phase, representing a life-threatening increased intracranial pressure (ICP) or the need for a neuroprotective intervention. 57% of patients required either ICP monitoring during the acute treatment—using external ventricular drain or intraparenchymal probe—or a permanent ventriculo-peritoneal shunt system, all interventions indicating either primary or secondary severe brain damage.

Initial TBI severity was weakly correlated with HRQoL (p = 0.04; adjusted R² = 0.02) (Fig. 2). The median QOLIBRI total score was 74.4 (IQR_25–75: 53.5–85.2), 70.2 (IQR_25–75: 51.2–81.8), and 68.5 (IQR_25–75: 46.2–78.1) in the patients with mild, moderate, and severe TBI, respectively. In contrast, HRQoL was not correlated with TBI etiology, age at TBI, age at survey, time elapsed since TBI, or sex distribution (Supplementary Table S1 and Fig. S2). The model’s effect size was low, with an R² value of 0.09 and an adjusted R² of 0.02, indicating a weak correlation, revealing that each 1-point increase in the GCS is associated with a 1.59-point increase in the QOLIBRI total score.

Our analysis also revealed that TBI severity influenced the patient’s level of satisfaction (p = 0.03; adjusted R² = 0.1), one of the QOLIBRI key aspects influenced primarily by the factors of daily life & autonomy (p = 0.03; adjusted R² = 0.09) and cognition (p = 0.049; adjusted R² = 0.05) (Figs. 2b, Supplementary Table S2). In contrast, TBI severity had no effect on the second QOLIBRI key aspect, restrictions (p = 0.31; adjusted R² = 0.08), which was obtained by merging the scales emotions (p = 0.29; adjusted R² = 0.04) and physical problems (p = 0.39; adjusted R² = 0.04).

Seventeen patients (13%) of the QOLIBRI cohort were included during the first year after their TBI. These 13% of patients (n = 17) reported insufficient HRQoL with a median QOLIBRI total score of 54 (IQR_25–75: 39.2–75.8) (Fig. 3). Two to 5 years after TBI, 33% of patients (n = 45) reported good HRQoL with a median QOLIBRI total score of 77 (IQR_25–75: 57.8–85.2). Finally, 6 to 10 years after TBI, 54% of patients (n = 73) reported good HRQoL with a median QOLIBRI total score of 65.5 (IQR_25–75: 48.8–80.5) (Fig. 3).

Among the 135 participants, 76% (n = 102) were male and 24% (n = 33) were female. Sex was not correlated with HRQoL following TBI (p = 0.91; adjusted R² = 0.02). The median QOLIBRI total scores among the male and female patients were 69.9 (IQR_25–75: 53.5–84.0) and 61 (IQR_25–75: 48.8–75.0), respectively (Supplementary Fig. S2 D).

Among our cohort of cTBI patients, 64% indicated good HRQoL, reflected by a mean (± SD) QOLIBRI total score of 65.5 ± 22.6, a value well in line with the general population and the QOLIBRI validation cohort (Supplementary Table S3) [38]. The remaining 36% of patients had a QOLIBRI total score below 60, indicating an increased risk of a depressive and/or anxiety disorder. In detail, 20% of patients had either an increased risk of a depressive or anxiety disorder, and 16% of patients for both disorders (Fig. 4). Furthermore, the majority of patients reported on average good HRQoL in all six QOLIBRI scales with the following mean (± SD) scores: cognition (62.4 ± 27.2), self (61.1 ± 25.6), daily life & autonomy (63.5 ± 31.0), relationships (69.3 ± 23.7), emotions (75.0 ± 23.9), and physical problems (66.1 ± 24.1), values again well in line with the previously published QOLIBRI validation cohort of 795 chronic TBI patients [38], indicating that the results of the current study are robust and in line with already published results (Supplementary Table S3 and Fig. S3).

The response and participation rate were 37% (150 of 401 alive patients) and 90% (135 out of 150 participant of our survey cohort), respectively. These values are relatively high, given (1) the long-follow up period of up to 10 years after TBI, (2) the high age of participating patients, (3) the severity of TBI, and (4) the fact that a second contact by telephone was not permitted by German data protection laws. This response rate is consistent with a large German multicenter epidemiological survey of 4307 pediatric and adult TBI patients evaluated 1 year after TBI, which yielded a primary response rate of 40% [37].

Most demographic data and basic characteristics did not differ between the QOLIBRI cohort and the non-responders. However, patients of the QOLIBRI cohort gained better functional status at discharge from neurorehabilitation with a mean modified Rankin Scale of 2, indicating slight disability with good mobility and independence in activities of daily living though unable to carry out all previous activities, while the non-responders remained moderate disabled—indicating that a non-responder bias cannot fully be excluded, but most parameters were comparable to the entire cohort of 439 TBI patients, and thus the QOLIBRI cohort is most likely representative for this larger group of chronic TBI patients.

All participants underwent neurorehabilitation at Schoen Rehabilitation Center in Bad Aibling, Germany—one of the largest neurorehabilitation units in Europe—and were primarily referred by certified TBI centers within the Southern Upper-Bavaria Trauma Network, thereby providing the highest standard of medical care to patients in the acute phase of TBI. Despite the common lack of GCS documentation in 38% of medical records [25, 37], it is reasonable to assume that most patients experienced a severe brain injury, namely 36% due to the initially documented GCS and up to 34% due to secondary brain injury. Specifically, 27% of patients underwent decompressive craniectomy and 57% received either intracranial pressure (ICP) monitoring to detect and guide therapy of intracranial hypertension and brain edema or a permanent ventriculo-peritoneal shunt system, procedures indicating severe brain damage. Based on a close examination of the relative proportions of patients who underwent decompressive craniectomy and/or received ICP monitoring and/or a shunt device, 58 to 70% were either primarily or secondarily severe brain injured, thus our cohort is comparable to the QOLIBRI validation cohort [38]. Other factors regarding our cohort, including TBI etiology and sex distribution, were consistent with the literature [23, 28, 39].

On average, our cohort was 20 years older than the cohort used to validate the QOLIBRI questionnaire, but this age difference did not change HRQoL and supports our finding that age is not a contributor to HRQoL in cTBI patients [38]. In contrast, the Dutch study found younger age as an independent predictor of lower HRQoL after DAI, while others report older age as an independent risk factor for a decreased HRQoL or likewise no age effects [21, 40, 41]. In detail, Scholten et al. analyzed HRQoL using the SF-36 instrument at 6 and 12 months after predominantly mild and after moderate to severe TBI with a mean age of 44 years (range 27–57) compared to 53 (range 18–85) years at survey in our QOLIBRI cohort [40]. This study emphasizes—well in line with our findings—that HRQoL increases over time until 12 months after TBI and this finding was most evident in the primarily mildly injured patients. Furthermore, severer injuries, less functional recovery, older age and female sex negatively correlated with better HRQoL but were potentially influenced by the 56 and 84% of patients lost to follow up at 6 and 12 months, respectively. Besides the initial TBI severity, i.e. patients of our QOLIBRI cohort reported better HRQoL after milder brain injuries—those findings are contradictory to the QOLIBRI cohort. Van Eijck and colleagues analyzed 86 chronic TBI patients aged 16–87 years on average up to 57 months (range 14–100 months post-TBI) after DAI due to a TBI with younger patients reporting less good HRQoL. This age-related finding which contrasts our current results, might be explained by the cognitive impairment after DAI [21]. Nevertheless, the QOLIBRI validation cohort included 795 chronic TBI patients—aged 17 to 68 years—3 months and up to 18 years after TBI, thus highly comparable with our QOLIBRI cohort, with only very weak correlations (r ≤ 0.11) for age effects, education, time since injury, and severity of injury (GCS) with HRQoL [30]. However, evidence of HRQoL in the elderly after TBI is scarce and might rather be associated with the preinjury and psychosocial capabilities than with the injury-related factors [41]. Taking together, age might be relevant for HRQoL in a variety of subgroups including TBI injury patterns, timepoint of follow up and potentially sample sizes and certainly need further attention in future long-term HRQoL outcome studies.

During the first year after TBI, HRQoL was reduced in our cohort indicating an increased risk of psychiatric sequelae and maladaptation to the post-TBI changes and this relevant finding is in line with the literature [24, 33, 40, 42]. Beyond one year post-TBI, HRQoL was even slightly better in our cohort than in the general German population and comparable to previous findings [22, 37, 43]. We suggest that future studies should investigate if good individual adaptation and resilience to the TBI-related changes are associated with better HRQoL outcome and the absence of psychiatric sequelae. Thus, these two factors—adaptation and resilience—should be prospectively focused and might be contributors to good HRQoL and long-term outcome after TBI.

Psychiatric disorders such as anxiety and depression are common in the general population with a 12-months prevalence of 18 and 9.5%, respectively [44]. Following TBI, psychiatric sequelae are relevant and frequencies vary in the literature ranging from 18 to 83% due to methodological issues such as diagnostic criteria, injury severity and elapsed time since the brain impact [45]. Especially, anxiety and affective disorders are most relevant after TBI with data ranges up to 70% for anxiety and between 25 and 77% for depressive disorders [9, 10, 46]. Approximately 40% of TBI patients even suffer from more than two psychiatric sequelae [10, 46].

In our chronic TBI cohort, one third of patients suffered from insufficient HRQoL associated with an increased risk of psychiatric sequelae, namely anxiety and/or depressive disorders. This finding is in line with results of a prospective study analyzing 817 TBI patients of which a total of 31% reported psychiatric disorders at 12 months after TBI [13]. The latter finding suggests an increased risk for mental health changes such as depressive and/or anxiety disorders after TBI with the need for routine diagnostic of these posttraumatic sequelae that hamper patient’s quality of life. Furthermore, depressive symptoms might increase over time and psychiatric sequelae are a major reason for rehospitalization after TBI [22, 47]. Whether these psychiatric sequelae are associated with the functional decline described in one third of chronic TBI patients is still not elucidated. However, the link between the posttraumatic heterogeneous psychiatric sequelae and patient’s outcome becomes more and more obvious and need our attention throughout neurorehabilitation.

This study has several limitations that warrant discussion. First, the cross-sectional study design with the sample size of 135—albeit representative for the entire cohort of 439—chronic TBI patients allows descriptive conclusions. But, a recent systematic review and meta-analysis on post-TBI HRQoL revealed that the majority of the included 49 studies between 1991 to 2013 analyzed less than 100 post-TBI patients [33]. Two further TBI studies analyzed HRQoL of 60 and 51 patients 10 years after TBI, respectively [22, 25]. Thus, the sample size of 135 chronic TBI patients in our study seems to be appropriate to highlight the need for psychiatric assessment on a regular base after TBI, especially when regarding the age range up to the 85-years-old, the long time period of up to 10 years after TBI as well as the given evidence, so far. Second, premorbid psychiatric sequelae, comorbidities, education, employment, living environment, injury patterns, and pharmacotherapy were not available. Third, neuroimaging data were not included to further characterize injury patterns as i) patients were referred from different trauma centers to neurorehabilitation, i.e. neuroimaging was per se less comparable or not available, ii) TBI patients—except for clinical deterioration or scheduling the bone flap replacement following decompressive craniectomy—do not get a routine neuroimaging while in neurorehabilitation, iii) state-of-the-art imaging for DAI (diffusion tensor imaging (DTI), tractography and susceptibility weighted imaging using a gradient recall echo (GRE-SWI)) was not available at the time when most of our patients were injured, i.e. more than 10 years ago, and iv) a multilevel diagnostic approach including neuroimaging and fluid biomarkers is recommended, but has not been implemented in the clinical routine, and thus were not available for our cohort [48]. But, DAI does not seem to influence HRQoL up to 5 years after TBI, therefore the injury pattern itself might be a less relevant factor for long-term outcome [49]. Fourth, functional status and comorbidities were not assessed in this survey-like cross-sectional study as written self-rating of functional status is most probably a less valid approach to get sustainable data. Fifth, caregiver’s quality of life and external assessments to elucidate a more objective perspective of the patients’ outcome was not done, but seem less relevant as indicated in the literature so far [50, 51]. Sixth, although a total score below 60 on the QOLIBRI questionnaire indicates an increased risk of psychiatric sequelae, a precise cut-off score has not been established; accordingly, our results on psychiatric disorders must be interpreted with care, but might help to implement cut-off scores and highlight the need for further evidence. Seventh, the initial GCS was not documented in 38% of cases in our cohort—a well-known finding in previous studies [25, 37]. Finally, the results’ generalizability might be limited due to the following issues: i) the age span of 18- to 85-year-olds, ii) the German population, iii) the heterogeneity of TBI in our cohort, and iv) treatment regimens including neurorehabilitation [21, 22]. Nevertheless, age does not seem to be relevant for long-term outcome regarding HRQoL as the presented results are comparable to the large QOLIBRI validation study, that included 20 years younger patients on average [38]. Future long-term outcome studies with larger sample sizes should better stratify for TBI severity subgroups beyond GCS as targeted by the large TRACK-TBI and CENTER-TBI studies. The acute TBI treatment in certified hospitals of the Southern Upper-Bavaria Trauma Network and the subsequent neurorehabilitation in one neurorehabilitation center—using standardized rehabilitation protocols—most probably represent the highest standard of medical care, and therefore results are at least generalizable within industrialized countries.