The sample of respondents was recruited by a non-probabilistic convenience sampling process. ABI participants had to meet the following inclusion criteria: (a) have an ABI; (b) are at least 16 years old; (c) are users of ABI-specific rehabilitation center; and (d) sign the informed consent to participate in the study. The ABI sample was composed of 421 adults from 17 rehabilitation centers providing health and social services throughout Spain. Just over half of the sample were men (60%), aged from 17 to 91 years (M = 53.12; SD = 14.87). Half of the sample were married (49.5%) and the employment situation prior to the injury shows that almost the entire sample (76.1%) was active (i.e., working and/or studying). However, it should be noted that after the injury, practically the whole sample (94.5%) was inactive. The main causes of ABI were CVA (56.3%) and TBI (21.9%), and the average time since the injury was 8.12 years (M = 6; SD = 7.3; range 1-57).

Although the ABI participants’ QoL was assessed, the respondents were 155 people, given that a report of others was used. To be an informant, they had to meet two requirements: (a) knowing the person for at least three months; and (b) being able to observe the person in different environments for substantial periods of time. Among respondents, 97 (62.6%) were health professionals and 58 (37.4%) were relatives. The relatives were mostly partners (43.1%) or parents (27.6%). Professionals were mostly neuropsychologists (24.7%), occupational therapists (22.7%), or physiotherapists (11.3%). The professionals knew the assessed person for periods between three months and 19 years (M=2.90; SD = 3.04 years).

The assessment of QoL was carried out through the administration of the CAVIDACE scale (Fernández, Verdugo, Gómez, Aguayo, & Arias, 2018; Verdugo, Gómez, Fernández, Aguayo, & Arias, 2018), which is a disease-specific instrument aimed to assess the QoL-related personal outcomes of adults with ABI. The scale has shown excellent evidences of validity and reliability (Fernández et al., 2019), and it has a suitable internal consistency (alpha ordinal values ranged from .81 to .93) and high inter-rater reliability (ICC = .97). The analysis of construct validity supported the eight correlated factors model (TLI = .87; CFI = .89; RMSEA = .06; SRMR = .07). The CAVIDACE scale comprises 64 items assessing different aspects of a person’s life, eight items per domain randomly distributed around the eight core domains proposed by Schalock and Verdugo (2002): Emotional well-being (EW), Interpersonal relationships (IR), Material well-being (MW), Personal development (PD), Physical well-being (PW), Self-determination (SD), Social inclusion (SI), and Rights (RI). The items are drafted as third-person statements with four frequency response options (never = 0, sometimes = 1, often = 2, always = 3). For its correction, the scale provides specific ABI yardsticks, allowing interpretation of the obtained results. Thus, direct scores obtained in each of the eight domains are converted into standard scores (M = 10; SD = 3) and percentiles. Moreover, the scale provides an overall raw QoL score (i.e., the sum of the scores obtained in each of the domains) that may vary from 0 to 192 (96 is the theoretical midpoint of the scale), where higher scores indicate higher QoL. This overall score may be converted into an easily interpretable Quality of Life Index (M = 100; SD = 15) by providing a QoL profile that allows professionals to elaborate person-centered support plans and rehabilitation programs.

This is a quantitative cross-sectional study. The research team contacted several organizations and health professionals that provide support and rehabilitation programs to the ABI population to obtain a broad and heterogenous sample. Interested participants were contacted by telephone and email to be informed about the study. The respondents had to score the 64 QOL-related items of the scale, estimating the frequency of the observable behaviors and circumstances that were described in them. Moreover, they completed a sociodemographic and injury-related questionnaire, based on the objective information available to each patient. Respondents had the possibility to complete the scales in a hard copy version or an online version. Then, the scales were collected, and the data were analyzed.

The Bioethics Committee of the University of Salamanca approved the research, which has complied with the ethical standards of the 1964 Declaration of Helsinki and its later amendments. Written informed consent forms were obtained from all the centers involved. These centers, in turn, have the consent forms of their patients and/or family members, which allow the use of their data for research purposes. Personal and clinical data were collected, stored, and protected, guaranteeing the confidentiality and anonymity of the participants.

Data were analyzed using SPSS.25 for Windows. Univariate and multivariate analyses were used to assess determinants of QoL scores. Due to the normal distribution of the dependent variable (QoL score), comparisons between groups (i.e., univariate analysis) were done using parametric tests: independent-sample t-tests were conducted to compare the total QoL score in independent variables with two levels (e.g., gender); one-way between-group analyses of variance (ANOVA) were conducted to compare the effect of the independent variables with more than two levels (e.g., civil status) in the total QoL score. In case of significant result, a Tukey post-hoc test was used to explore which means differed. A significance level of 95% was set (p ≤ .05), and effect sizes were analyzed using eta-squared (η2). The Cohen (1988) guidelines were followed to interpret the values (i.e., .01=small; .06=moderate; .14=large).

Multiple linear regression analysis was used to identify which independent variables significantly predicted QoL outcomes, and their contribution to the model. First, the sociodemographic and clinical variables that were significant in the univariate analysis were included simultaneously as an initial model (the enter method) to identify which variables were significant in the set. Factors with p ≤ .10 were retained. Once potential predictors were identified, the model was built by the backward stepwise method and then was checked by the forward method. Results are presented as R2, R2 change, F change, and standardized beta values. R2 was interpreted according to the Cohen (1988) guidelines (i.e., .02=small; .13=medium; .26=large).

Statistically significant differences were found in total QoL score by age, civil status, education level, prior employment status, type of home, level of supports, loss of legal capacity (i.e., the loss in the ability to make legally valid decisions and establish binding contractual relationships), recognized dependence and degree of dependence (i.e., people who, due to their age or because an illness or disability situation, cannot carry out their daily basic activities without the help of another person, can apply for long-term care), time since the injury, location of the injury, and presence of post-traumatic amnesia (PTA).

A moderate effect size was found in the loss of legal capacity, degree of dependence, type of home, level of supports, time since the injury, and dependence recognized variables, in decreasing order. The remaining significant variables obtained small effect sizes, indicating a small influence on the QoL outcomes. No significant differences were detected in QoL scores (p > .05) according to gender, age (grouped variable), etiology, severity of the injury (measured with the Glasgow Coma Scale, GCS), presence of coma and its length, and length of PTA. The results of the univariate analyses are presented in Table 2.

A multiple linear regression was carried out to examine the potential QoL predictors of ABI participants through the analysis of the variance in QoL scores, using first the enter method (introducing 12 factors into the model) and then the stepwise method (introducing five factors). Stepwise analysis showed five models, of which the last one was the one with the greatest predictive capacity (Table 3).

A significant model (F(5, 328) = 11.77, p<.001) predicted 15.2% of the sample outcome variance with a coefficient of determination (Adj. R2 = .13) considered medium according to the Cohen guidelines. The model fulfills homoscedasticity, and the residues are normally distributed. The resulting model contains five significant QoL predictors: loss of legal capacity (β=−11.91, t=−3.36, p=.001), time since the injury (β=−3.97, t=−2.65, p=.008), prior unemployment status (β=−10.56, t=−3.17, p=.002), location of the injury (β=−7.64, t=−2.73, p=.007), and degree of dependence (β=−4.33, t=−2.44, p=.015). The model predicted lower QoL for cases with loss of legal capacity, higher degree of dependence, prior unemployment status, longer time injured, and injured bilaterally (Table 4).