Reference Values of the Quality of Life after Brain Injury (QOLIBRI) from a General Population Sample in Italy

Traumatic brain injury (TBI) may affect the lives of the individuals concerned and their relatives negatively in many dimensions. Health-related quality of life (HRQoL) is a comprehensive and complex concept that can assess one’s satisfaction with a broad range of areas of life and health. The Quality of Life after Traumatic Brain Injury (QOLIBRI) questionnaire is a TBI-specific measure for HRQoL which is used in research and health services worldwide. When evaluating self-reported HRQoL after TBI, reference values from a general population are helpful to perform clinically relevant evaluations and decisions about the condition of an affected person by comparing the patient scores with reference values. Despite the widespread use of the QOLIBRI, reference values have until now only been available for the Netherlands and the United Kingdom. The aim of this study was to validate the QOLIBRI for the general population in Italy and to provide reference values. An adapted form of the QOLIBRI was administered to 3298 Italians from a healthy general population using an online survey. Their scores were compared with those of 298 individuals post-TBI recruited within the international longitudinal observational cohort CENTER-TBI study in Italian hospitals, who completed the original questionnaire. The psychometric characteristics and the measurement invariance of the QOLIBRI were assessed. A regression analysis was performed to identify predictors relevant for HRQoL in the general population. Reference values were provided using percentiles. Measurement invariance analysis showed that the QOLIBRI captures the same HRQoL constructs in an Italian general population and Italian TBI sample from the observational Center-TBI study. Higher age, higher education and the absence of a chronic health condition were associated with higher QOLIBRI scores, suggesting better HRQoL. Reference values were provided for a general Italian population adjusted for age, sex, education and presence of chronic health conditions. We recommend using these for a better interpretation of the QOLIBRI score in clinical practice and research in Italy.


Introduction
Traumatic brain injury (TBI) is an important cause of burden of disease worldwide, as more than 50 million people acquire it every year [1]. In a study published in 2018, Dewan et al. [2] estimated that approximately 69 million people worldwide experience TBI each year. In Italy, the incidence of TBI varies from 212.4 [3] to 848 [4,5] cases per 100,000, depending on the study, placing it among the countries with the highest TBI rates in Europe.

Instruments 2.3.1. Quality of Life after Traumatic Brain Injury (QOLIBRI)
The QOLIBRI is the first instrument specifically developed for individuals after TBI to assess their disease-specific HRQoL. It comprises 37 items associated with four scales (Cognition, Self, Daily Life and Autonomy, and Social Relationships) with items measuring satisfaction with various aspects of HRQoL (part A) and two scales (Emotions and Physical Problems) measuring issues that individuals after TBI feel bothered by (part B). Responses to the Part A items are coded on a 5-point Likert scale with 1 corresponding to not at all satisfied and 5 to very satisfied. Responses to the items in Part B are reversely scored to correspond with the items of the Part A. Here, 1 indicates very (bothered) and 5 means not at all bothered. Like other instruments measuring quality of life, when scoring the QOLIBRI scale, means are converted to a 0 to 100 rating scale by subtracting 1 from the mean score and then multiplying it by 25, with a value of 0 indicating the worst possible HRQoL and a value of 100 the best possible HRQoL.
For the general population sample, three items of the original QOLIBRI had to be reworded to remove the reference to a TBI. The fifth item from the scale "Self", "How satisfied are you with what you have achieved since your brain injury?", was changed to "How satisfied are you with what you have achieved recently?". The second item from the scale "Physical", "How bothered are you by effects of any other injuries you sustained at the same time as your brain injury?", was changed to "How bothered are you by the effects of any injuries you sustained?". The last item, also assigned to the scale "Physical", "Overall, how bothered are you by the effects of your brain injury?", was changed to "Overall, how bothered are you by the effects of any health problems?".

Sociodemographic and Health Status Data
The sociodemographic and health status data for both samples contained information on sex, age and the highest level of education achieved. In addition, the presence of chronic health conditions (CHC) was recorded for the general population sample, where multiple answers were possible. The question was: "Do you have any of the following chronic health complaints?" Subjects were asked to tick a box for the response options (multiple answers were possible) listed in Table A1.
Additionally, the Glasgow Coma Scale (GCS) was used in the TBI sample to rate TBI severity [36]. A score of 13 to 15 points indicates mild TBI, 9 to 12 moderate TBI, and 3 to 8 severe TBI. The Glasgow Outcome Scale Extended (GOSE), ranging from 1 (death) to 8 (upper good recovery), was used as a measure of recovery status after TBI [37].

General Population Sample
Participants from the general population sample were recruited by a market research agency (Dynata, Shelton, CT, USA) between 29 June and 31 July 2017. To obtain a representative sample, participants were invited until the required quotas for age, sex and level of education had been achieved. Due to the self-reported nature of the data collection, the sex of participants was collected as gender (male, female). Since gender/sex corresponds to the biological categories of males and females, the word "sex" will be used for consistency and to avoid any confusion. Comparison of the quotas with demographic information obtained from the Organization for Economic Cooperation and Development databank (OECD) [38] and Eurostat database [39] revealed a widely comparable distribution of the groups. Within this online survey based on self-report, the data were collected in Italy, the Netherlands, and the UK. The reference values of the QOLIBRI for the Netherlands and the UK have already been published [40].
In order to increase the representativeness of the sample, Dynata deployed a variety of methods to engage people with diverse motivations to take part in research and to reach participants with different socioeconomic statuses. To avoid self-selection bias, specific details of the project were not visible at the time of the invitation. The project details were only disclosed later on. Participants who answered the survey in less than five minutes were automatically excluded from the analysis. Additionally, participants with contradictory response patterns were excluded. For the QOLIBRI, the following answers were excluded as they were contradictory: If someone chose responses at either the left or right extremes of the Likert scale, that meant that they were not satisfied at all, but also not bothered at all. All collected data were anonymized. The nonresponse rate of the survey was 14.1%. Figure 1 shows the general Italian population sample attrition.

TBI Sample
Participants in the TBI sample were a part of the CENTER-TBI study (EC grant 602150), which collected data from 4509 patients in 18 countries [35]. The following inclusion criteria had to be fulfilled: a clinical diagnosis of TBI, presentation in the hospital fewer than 24 h after injury, and an indication for computed tomography (CT). Data were collected between 9 December 2014 and 17 December 2017 via face-to-face visits, in hospital visits, via telephone interviews, or a combination of telephone interview and e-mail.

TBI Sample
Participants in the TBI sample were a part of the CENTER-TBI study (EC grant 602150), which collected data from 4509 patients in 18 countries [35]. The following inclusion criteria had to be fulfilled: a clinical diagnosis of TBI, presentation in the hospital fewer than 24 h after injury, and an indication for computed tomography (CT). Data were collected between 9 December 2014 and 17 December 2017 via face-to-face visits, in hospital visits, via telephone interviews, or a combination of telephone interview and e-mail. Data on sex, age, time since injury and education was collected at study enrollment based on medical records and self-report. The information on age at study enrollment reflects the age at injury. The QOLIBRI data used was obtained around three months post-injury (i.e., minus two to plus five weeks). Figure 2 shows the TBI sample attrition. No participants with contradictory response patterns were identified. Therefore, all were included in the analyses.

TBI Sample
Participants in the TBI sample were a part of the CENTER-TBI study (EC grant 602150), which collected data from 4509 patients in 18 countries [35]. The following inclusion criteria had to be fulfilled: a clinical diagnosis of TBI, presentation in the hospital fewer than 24 h after injury, and an indication for computed tomography (CT). Data were collected between 9 December 2014 and 17 December 2017 via face-to-face visits, in hospital visits, via telephone interviews, or a combination of telephone interview and e-mail. Data on sex, age, time since injury and education was collected at study enrollment based on medical records and self-report. The information on age at study enrollment reflects the age at injury. The QOLIBRI data used was obtained around three months post-injury (i.e., minus two to plus five weeks). Figure 2 shows the TBI sample attrition. No participants with contradictory response patterns were identified. Therefore, all were included in the analyses.

Statistical Analyses
The following section describes the statistical analyses in detail. All the analyses were carried out using R version 4.0.3 [41] employing the packages lavaan [42] and semTools [43] for the calculation of Confirmatory Factor Analysis (CFA) and MI, respectively. The significance level was set at 5%.

Statistical Analyses
The following section describes the statistical analyses in detail. All the analyses were carried out using R version 4.0.3 [41] employing the packages lavaan [42] and semTools [43] for the calculation of Confirmatory Factor Analysis (CFA) and MI, respectively. The significance level was set at 5%.

Item and Scale Characteristics of QOLIBRI in General Population
Firstly, the item characteristics of the reworded QOLIBRI were examined. This included means, standard deviations, skewness, and a check of the floor and ceiling effects. Skewness was characterized as symmetric for values from −0.5 to 0.5, moderately skewed from ±0.5 to ±1, and highly skewed for values above ±1 [44]. On the scale level, the internal consistency of items was calculated using Cronbach's alpha. Then, the correlation between scales and the range of correlations between items and their home scales were checked. In order to evaluate the ceiling effects, a cut-off value of 40% was chosen for the highest category "very". This is twice as high as the 20% that could be expected by chance with five categories. For the floor effects, we controlled by combining the response categories "not at all" and "slightly", with a cut-off of 10%. The recommendations of the World Health Organization Quality of Life (WHOQOL) Group [45] were followed to exclude items with a Corrected Item-Total Correlation (CITC) higher than 0.4. However, no items had to be excluded.

Construct Validity of QOLIBRI in General Population
We used confirmatory factor analysis (CFA) to verify whether the six-factor structure of the original questionnaire could be replicated for the adapted QOLIBRI applied in the general reference population sample. For this purpose, we first estimated three models: a one-factor model, a two-factor model, and the original six-factor model. The one-factor model assumed a general factor HRQoL that is associated equally with all 37 QOLIBRI items. The two-factor model assumed two intercorrelated factors, where one factor included items from the QOLIBRI that represented satisfaction with certain aspects of an individual's life (Part A) and the second factor reflected feeling bothered with some aspects of one's life (Part B). The six-factor model which was described above in detail comprised six factors (Cognition, Self, Daily Life and Autonomy, Social Relationships, Emotions and Physical Problems). Finally, the models were compared using chi-square difference tests.

Measurement Invariance between Samples
The examination of the MI included analyses of individual responses from both samples. Due to the limited sample size in the TBI sample, we had to dichotomize the response categories of the QOLIBRI, with the response categories "not at all" and "slightly" forming the lower category and the response categories "moderately", "quite" and "very" the higher category. We therefore followed the approach of Wu and Estabrook [46] when testing MI for dichotomized response categories. We estimated increasingly constrained models and compared the model fit among these. We first estimated the baseline model, which is mostly equivalent to configural MI and freely estimates all four parameters (thresholds, loadings, intercepts and residuals). Here, the requirement of configural MI is satisfied when the same number of factors and the same pattern of loadings are equal for both groups. We then estimated the second model, where three parameters are restricted and the thresholds are freely estimated, which corresponds to partial MI. Finally, in the last model, all four parameters were restricted, which is equivalent to full MI.

Regression Analysis
Research suggests that age [47], gender [48], education [49] and the presence of chronic health conditions (CHC) [50,51] have an impact on HRQoL. Therefore, to generate reference values that represent HRQoL for meaningful subgroups, we investigated the influence of these factors on HRQoL as measured by the QOLIBRI total score using multiple linear regression. Available information from the general population sample on these variables and their interactions was included in the regression model. Age was binned into six ordered age categories (18 to 24; 25 to 34; 35 to 44; 45 to 54; 55 to 64; older than 65 years). Bearing in mind that the age-in the form of 10-year age bins-had a significant influence on the total score of the short form of the QOLIBRI, its overall scale-QOLIBRI-OS, in the Italian population [52], the same age bins were used here. Sex was categorized as female and male. Education was assessed as the highest level of education and categorized as one of the following three: low (primary school), middle (diploma, secondary school, high school, or post-high school), or high (college or university). Participants were categorized in terms of CHCs either being present (when they reported at least one CHC) or being absent. The dependent variable was the participants' QOLIBRI total score.

Reference Values from the General Population Sample
Based on the results of a linear regression analysis, tables were presented with population reference values in form of percentiles (2.5%, 5%, 16%, 30%, 40%, 50%, 60%, 70%, 85%, 95% and 97.25%). Values below the 16th percentile and above the 85th percentile (both rounded up to the next integer) represent low and excellent HRQoL, respectively. These can be used to evaluate whether an individual's QOLIBRI total score is below, equal to, or above the value of the respective reference group.

Results
The sociodemographic characteristics of the general population sample are presented in Table 1. Both sexes were represented equally. The mean age of this sample was 45.27 (SD = 14.85) years. Slightly more than a half of the participants (53.97%) reported no CHCs. Detailed information on specific CHCs per age group can be found in Appendix A, Table A1.

Item and Scale Characteristics of QOLIBRI in the General Population
Item characteristics including mean value, skewness, and floor and ceiling effects are presented in Table 3. On average, individuals were rather satisfied with their HRQoL (M = 3.62 [3.11-4.02]). The lowest satisfaction scores related to questions on anger or aggression (M = 3.11, SD = 1.26) and the highest satisfaction scores were reported in connection with the ability to find one's way around (M = 4.02, SD = 0.98), the ability to get out and about (M = 4.02, SD = 1.02), and the ability to carry out domestic activities (M = 4.02, SD = 0.97). With skewness values from 0 to ±0.99, the item distribution can be considered as moderately skewed. None of the satisfaction items from the Part A exceeded the cut-off value for ceiling effects. The reversed scales "Emotions" and "Physical Problems" in Part B, containing bothered items, showed higher values, indicating that individuals from the general population sample were mostly not bothered by problems present in the TBI population. The scales "Cognition" and "Physical Problems" were below the cut-off value of 10%, indicating that the healthy population sample had very few problems in these domains.  Table 4 provides Cronbach's alpha characterizing the internal consistency of the six QOLILBRI scales. Coefficients ranged from 0.87 to 0.92 indicating good to excellent internal consistency of the QOLIBRI scales [53]. Based on corrected item-total correlations (CITC) and the cut-off of 0.40, all items were considered consistent. The subscales were moderately to highly intercorrelated (r between 0.35 and 0.77). The highest correlation was found between the subscales "Daily Life and Autonomy" and "Self" (r = 0.83), while the lowest correlation was between the scales "Emotions" and "Cognition" (r = 0.35).

Construct Validity of the QOLIBRI in the General Population
In order to evaluate the latent factor structure of the adapted QOLIBRI, CFAs were carried out, comparing the one, two and six factorial models. Table 5 summarizes the goodness of fit indices for these models, showing the best fit for the six factorial model with χ2(614) = 7473, p < 0.001, CFI = 0.994, and RMSEA = 0.058, 90% CI (0.057; 0.059) [54].

Measurement Invariance
The results of the MI analyses indicated no significant difference between the configural and partial invariance models (Table 6), thus partial invariance can be assumed. However, a comparison of the partial and full invariance models revealed statistically significant differences, indicating that thresholds differed between these models. Further analysis has been undertaken to assess the practical significance of these differences. Examining the thresholds in the partial invariance model showed that these values differed between the general population sample and the TBI sample (Table A2), indicating that the response behavior was not identical in both groups. However, these threshold differences did not exceed 5%. Therefore, the difference between partial and full measurement invariance can be interpreted as being non-significant, resulting in full measurement invariance between the TBI and general population sample. Thus, when comparing QOLIBRI scores between general population and TBI population samples, the differences in scores can be attributed to real differences in HRQoL. Identification constraints for the invariance models: Configural: item intercepts = 0, residual variances = 1, latent factor means = 0, latent factor variances = 1; Partial: item intercepts = 0, residual variances = 1. Only in the reference group latent factor means = 0 and variances = 1; Full: item intercepts = 0, residual variances = 1. Only in the reference group factor means = 0, factor variances = 1.

Linear Regression Analysis
Regression analysis revealed a significant impact of age, CHCs and education (Table 7). Individuals in all other age groups displayed significantly higher QOLIBRI scores than individuals aged 18 to 24 years. The presence of a CHC significantly influenced HRQoL, since healthy individuals had higher QOLIBRI scores than individuals with at least one chronic health condition. Individuals with a high, but not those with a medium level of education had significantly higher QOLIBRI scores than individuals with lower education. The effect of sex or any other interaction did not significantly contribute to explaining the QOLIBRI scores.
The following example will try to illustrate how to use these values. After a TBI, a 50-year-old woman with diabetes presented with a QOLIBRI total score of 65. The appropriate reference values are those of females with at least one CHC in the age group of 45 to 54 years (Table 8). Table 8 shows that about 65% of individuals in her age group reported the same or a lower level of HRQoL. Her value lies in the range of one standard deviation above the median and can thus be considered as being average. Based on the 16%-percentile cut-off value, HRQoL is interpreted as being below average for female individuals of 50 years with one CHC when the QOLIBRI total score is lower than 42.

Discussion
The aim of this study was to provide reference values for the QOLIBRI derived from a general Italian population sample. For that purpose, some conditions had to be fulfilled. First, CFA was used to verify that the adjusted QOLIBRI had the assumed six-factorial structure (Cognition, Self, Daily Life and Autonomy, Social Relationships, Emotions, Physical Problems) like the original QOLIBRI version for adults after TBI. This requirement was met and the results were almost consistent with an earlier study [40] that applied the adapted QOLIBRI questionnaire to general population samples from the Netherlands and the UK.
Gorbunova et al. [40] showed that in the Dutch population, the interaction between gender and CHCs was also significant in the regression analysis. This was not the case in the Italian or in the United Kingdom populations. Concerning the QOLIBRI total score without further stratification, a value below 50 obtained from general Italian sample indicates impaired HRQoL. The values obtained from the English and Dutch general population samples were lower (i.e., 44 for the UK) and higher (i.e., 55 for the Netherlands), respectively [40]. Since no differences can be observed in terms of the distribution of the sociodemographic or health-related factors, these findings can be explained by the differences in HRQoL across the countries [58][59][60][61]. For example, Alonso et al. [61] found that participants from the Netherlands (M = 55.2) reported the highest generic mental HRQoL score as measured using the Short Form-36 (SF-36) mental component summary score, compared with other countries (e.g., Italy: M = 50.3). In addition, the European Study of Epidemiology of Mental Disorders within six countries found that the proportion of respondents reporting problems on any of the EuroQol-5 Dimensions (EQ-5D) [62] was significantly higher in France and lower in Spain and Italy [58]. Taken together, all these differences emphasize the importance of country-specific reference values, which is also the case for TBI-specific HRQoL assessments.
The MI analyses indicated that the same construct was measured in the general Italian reference sample and in the TBI population. Although the full MI model differed from the partial MI model in terms of model fit, analyses of threshold fluctuations indicated that thresholds did not differ more than 5% and were thus negligible. The same conclusion could also be drawn for the QOLIBRI in the Dutch and UK samples [40,[58][59][60][61].
Our results showed that younger age, presence of CHCs, and lower level of education are associated with worse HRQoL measured using the QOLIBRI. Wu et al. [52] found similar results for the use of the short version of the QOLIBRI, the QOLIBRI-OS, in an Italian general population sample providing reference values. The use of the same age bins in calculating the regression analyses presented, as well as in stratifying the reference values, ensures that the reference values of the two instruments are comparable in the future. Regarding age differences, a study examining HRQoL after heart failure found that older patients' HRQoL exceeded expectations for their age, whereas younger individuals complained of loss of activities or roles and rated their HRQoL as being correspondingly worse. The authors suggested that better HRQoL in older compared with younger patients was due to the older patients' ability to reconceptualize their expectations in relation to their health problems. Duke et al. [63] also demonstrated that older people who had adapted their activities to the chronic illness in question had better mental health, suggesting that it is not just the presence of health problems or young age that determines good quality of life.
In addition, it should be noted that sex did not play a role either in the study by Wu et al. [52] nor in the present study. However, the literature on TBI regarding sex or gender differences is inconsistent [57,[64][65][66], while there is strong evidence that gender represents an influential factor in TBI [67]. Previous research shows that sex differences were found to possibly affect sustaining a TBI [68], to impact post-concussion symptoms [56,69], depression [70], anxiety [70], as well as recovery after TBI [71,72]. A recent study by Mikolic et al. (2021), examining differences between men and women in treatment and outcome after TBI, finds that after mild TBI women reported lower generic and diseasespecific HRQoL than men. Despite controversial research findings, gender/sex seems to be important for outcome assessment after TBI. Therefore, we have also added a stratification of reference values by sex in addition to the stratification by age, presence of CHC and education.
In contrast to our TBI sample showing a negative association between the HRQoL and age (r = −0.18), as well as to prior research that has found a decrease in HRQoL in older subjects with a TBI history [47], the general population sample investigated in the present study displayed higher HRQoL with increasing age. This is in line with findings from a non-TBI Taiwanese sample, which showed a positive effect of age on mental HRQoL and negative influence on physical HRQoL measured using the generic Short Form 12 (SF-12) [73]. In our sample, we used the QOLIBRI total score, which incorporates both mental and physical aspects of HRQoL. Further research should investigate the differential effects of age on individual QOLIBRI dimensions.
It is reasonable to assume that chronic health problems have an influence on HRQoL [74]. Our results showed that individuals with CHCs exhibited lower QOLIBRI total scores than individuals without CHCs. These results are consistent with previous research which indicates an inverse relationship between CHCs and HRQoL [75].
In addition, level of education was also associated with better HRQoL. Individuals with a higher education level reported higher QOLIBRI total scores in comparison to individuals with low education levels. These findings are in line with prior research showing an association between higher education levels and better HRQoL in non-TBI [76] and TBI [77,78] populations. The relationship between education and HRQoL can likely be explained by the opportunities higher education and better socioeconomic status provide, furthering, for example, self-determination through better income and better access to health services [79][80][81].

Strengths and Limitations
The most important strength of our study is the number of survey participants, which allowed reference values to be calculated stratified by several sub-groups. For example, we were able to provide reference tables for the individuals with and without CHCs and integrating the education levels. The interpretation of HRQoL for Italian individuals after TBI has thereby been improved. Furthermore, reference values based on percentiles are a common approach in clinical practice, facilitating the interpretation and communication of the QOLIBRI scores. The comparison with a (healthy) general population improves the comprehensibility of the test results for the patients.
This study also has several limitations that may require discussion. The first limitation concerns the recruitment of the general population sample. Recruitment was carried out via online platforms and strived for maximum representativeness. However, the online nature of the recruitment only captures certain population groups, such as only those who have Internet access, which may have led to selection biases [82]. In addition, we do not have information about those who declined the survey invitation, which is one of the main issue of online surveys [83]. Possible carelessness in answering online surveys [84] as well as the lack of opportunity to verify the authenticity of the data are notable limitations [82]. Moreover, the severity of the CHCs, as well as their duration, were not recorded because the analyses of these characteristics were beyond the scope of this study. Future studies may investigate the influence of these factors on disease-specific HRQoL.
With respect to the TBI sample, it should be noted that its relatively small size made a dichotomization of the QOLIBRI's response categories necessary, which always results in a loss of information [85,86]. Furthermore, the vast majority of the TBI sample (71%) consisted of mild TBI, which could have led to response categories not being exhausted (e.g., not at all satisfied or very bothered), requiring the modification of the number of response categories for MI analysis. However, this limitation only concerns the comparison of the QOLIBRI between the general and the TBI sample. To fill this gap, future research should investigate potential differences between Italian TBI and general population samples employing larger TBI samples. With regard to injury severity in the TBI sample, it should be noted that 13.7% of subjects had missing information, which is common in clinical trials. These missing data were not imputed since this information has not been used in the further analyses. The 13.7% of missing values for education were either due to the fact that the level of education was unknown or not reported. Since we did not include any of the above variables in determining the reference values and used them only for the descriptive statistics of the TBI sample, the missing values had no further impact on our results.
The QOLIBRI is an internationally widely used instrument, which has been translated into 26 languages. The reference values for the Italian population presented here may help to consider cultural differences in HRQoL. In addition to the total score, reference values on the subscale level allow the HRQoL domains to be evaluated more precisely. However, to date, there are reference values only for two further countries (i.e., the Netherlands and the UK). Therefore, further studies are required that investigate country-specific reference values for the QOLIBRI in the general population to enable multinational studies on TBI supporting the understanding of the clinical meaning of HRQoL after TBI.

Conclusions
This study contributes to TBI outcome research by providing reference values for the TBI-specific instrument QOLIBRI for an Italian general population stratified by age, education, gender, and the presence of CHCs. Researchers and clinicians are now able to employ reference values for individuals from Italy which could help them to better interpret HRQoL after TBI in individuals and to adjust their treatment accordingly, which in turn could help to improve the quality of life of the individuals concerned.

Institutional Review Board Statement:
The CENTER-TBI study (EC grant 602150) has been conducted in accordance with all relevant laws of the EU if directly applicable or of direct effect and all relevant laws of the country where the recruiting sites were located, including, but not limited to, the relevant privacy and data protection laws and regulations (the "Privacy Law"), the relevant laws and regulations on the use of human materials, and all relevant guidance relating to clinical studies from time to time in force, including, but not limited to, the ICH Harmonized Tripartite Guideline for Good Clinical Practice (CPMP/ICH/135/95) ("ICH GCP") and the World Medical Association Declaration of Helsinki entitled "Ethical Principles for Medical Research Involving Human Subjects". Informed consent was obtained for all patients recruited in the Core Dataset of CENTER-TBI and documented in the e-CRF. Ethical approval was obtained for each recruiting site. The list of sites, Ethical Committees, approval numbers and approval dates can be found on the project's website https://www.center-tbi.eu/project/ethical-approval (accessed on 15 July 2022).
Data Availability Statement: All relevant data are available upon request from CENTER-TBI, and the authors are not legally allowed to share it publicly. The authors confirm that they received no special access privileges to the data. CENTER-TBI is committed to data sharing and in particular to responsible further use of the data. Hereto, we have a data sharing statement in place: https://www. center-tbi.eu/data/sharing (accessed on 1 July 2022). The CENTER-TBI Management Committee, in collaboration with the General Assembly, established the Data Sharing policy, and Publication and Authorship Guidelines to assure correct and appropriate use of the data as the dataset is hugely complex and requires help of experts from the Data Curation Team or Bio-Statistical Team for correct use. This means that we encourage researchers to contact the CENTER-TBI team for any research plans and the Data Curation Team for any help in appropriate use of the data, including sharing of scripts. Requests for data access can be submitted online: https://www.center-tbi.eu/data (accessed on 1 July 2022). The complete Manual for data access is also available online: https: //www.center-tbi.eu/files/SOP-Manual-DAPR-2402020.pdf (accessed on 1 July 2022).

Acknowledgments:
The authors would like to cordially thank all patients, study participants and CENTER-TBI investigators.