Longitudinal Trends of Depression in Traumatic Brain Injury: The Role of Individual Heterogeneity in Clinical Prediction

Longitudinal Trends of Depression in Traumatic Brain Injury: The Role of Individual Heterogeneity in Clinical Prediction | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Longitudinal Trends of Depression in Traumatic Brain Injury: The Role of Individual Heterogeneity in Clinical Prediction Nelofar Kureshi, David B. Clarke, Abraham Nunes, Cindy Feng, Syed Sibte Raza Abidi This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6463585/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Background Depression affects approximately 30% of individuals after traumatic brain injury (TBI), yet long-term depression trends and their determinants are poorly understood. This study aimed to model depression trajectories over ten years post-TBI, compare the predictive performance of population-level only versus both population and subject-level effects, and assess the model's clinical utility for predicting depression in unseen and existing patients. Methods Data were obtained from the Traumatic Brain Injury Model System (TBIMS) National Data Bank. Depression was measured using the Patient Health Questionnaire-9 (PHQ-9) and collected at 1, 2, 5 and 10 years after injury. Covariates included age, sex, race, employment, education, functional measures, injury severity, pre-injury mental health, and substance use. Linear mixed modelling was used to identify depression trends and factors associated with depression. Predictive performance was evaluated using mean squared error, coverage, and precision. Results The sample comprised 19,397 individuals (mean age 43). Depression scores showed a small decrease over time among those with pre-injury mental health treatment history, but this change was not clinical meaningful. Significant predictors of Year 1 depression included pre-injury mental health treatment (β=1.6), female sex (β=0.86), and prior head injuries (β=0.4). When predicting depression for existing patients using early depression scores, the model achieved precision of 3.7 points, whereas for new patients, the model's precision was 6 points. Conditional predictions outperformed marginal predictions. Conclusion Depression trajectories following TBI exhibit substantial individual heterogeneity. Population-level models alone inadequately capture this complexity, while models incorporating both population and subject-level variations significantly improve predictive performance. This modeling approach demonstrates the potential for predicting depression trajectories in clinical settings, thereby facilitating individualized assessment and intervention. traumatic brain injury depression longitudinal mixed model prediction Figures Figure 1 Figure 2 Introduction Depression is a common consequence of traumatic brain injury (TBI), affecting approximately 30% of individuals with TBI, particularly during the first year post-injury (Fakhoury et al., 2021). The etiology of post-TBI depression is complex, involving pre-injury factors, injury characteristics, and post-injury experiences. Despite its prevalence, the long-term trends and determinants of depression trends following TBI remain insufficiently understood. The global burden of depression has increased from 1990 to 2019, with higher incidence and disability in females and older age groups, and variation across socioeconomic development levels (Liu et al., 2024). This report indicates that the rates of the two subtypes of depression, dysthymia and major depressive disorders, have remained largely stable globally and regionally over the study period, with the majority of patients suffering from major depressive disorder (Liu et al., 2024). Several studies indicate an overall increasing trend in depression prevalence over time within the general population. A systematic review and meta-analysis found a predominant increase in the likelihood of experiencing depression, with a pooled odds ratio of 1.35, suggesting a significant rise in depression prevalence over time (Moreno-Agostino et al., 2021). In the United States, data from 2005 to 2016 revealed an increase in severe depression, especially among adults aged 65 and older, and moderate depression among those aged 20-39 (Yu et al., 2020). The majority of research on the impact of TBI on depression relies on cross-sectional data collected during the early post-injury period. Across various studies, the prevalence of depressive disorders post-TBI ranges widely. In a systematic review, the prevalence of depression was found to be 17% in the first year after TBI, increasing to 43% over the long term (Scholten et al., 2016). Longitudinal studies provide additional insights, showing variability in depression rates over time. For instance, Bombardier et al. reported that 29% of TBI patients experienced a major depressive episode in the first year post-TBI, with relatively stable rates during follow-up, while minor depressive episodes decreased over the same period (Bombardier et al., 2016). Other studies indicate that depression post-TBI is dynamic, with some individuals developing new depressive symptoms while others experience recovery. Approximately 26% of individuals not depressed at one year developed depression by the second year, and nearly one-third of those with minor depression progressed to major depressive disorder (Hart et al., 2012). The elevated risk of depression persists beyond the first year, with studies reporting prevalence rates of 42% at 2.5 years post-injury (Jorge et al., 2004). The three month prevalence of depression is reported to be 56% (R. Singh et al., 2018), 42% at one year and 38 % at 10 years (R. K. Singh et al., n.d.). These findings are largely based on single-institution studies, which limit generalizability to diverse TBI populations and geographic regions. Additional limitations include small sample sizes and relatively short follow-up periods. While prior research, using various methods from finite mixture modeling, have provided insights into the diverse patterns of depression following TBI, these approaches may not accurately reflect the individual journeys of recovery (McInnes et al., 2024). Individuals with TBI are a remarkably heterogeneous population. Some individuals demonstrate resilience and experience significant improvements in their depressive symptoms over time, while others struggle with persistent or worsening symptoms, despite similar demographic and injury characteristics. However, population-level estimates of depression prevalence and overall trends are still useful as a starting point for understanding the broader impact of TBI. The challenge lies in capturing both the general patterns and the individual variations around those patterns. Linear mixed-effects models offer a powerful approach to address this challenge by simultaneously estimating fixed effects, representing population-level trends, and random effects, which capture individual deviations from those trends. By modeling both the average trajectory and the individual departures from it, we can gain an understanding of the complex and heterogeneous nature of depression following TBI. Importantly, while linear mixed models can incorporate random effects for personalized predictions, health research frequently isolates fixed effects when deriving predictions applicable to broader populations (Brown, 2021; Welham et al., 2004). These population-level prediction may be appropriate where generalizable conclusions outweigh individual heterogeneity, however this is not the case in TBI population where the concept of an 'average' TBI patient is inherently problematic (Anderson, 2025; Covington & Duff, 2021; Rabinowitz et al., 2020). For a more accurate understanding of long-term depression following TBI, predictions derived from linear mixed models should incorporate both population-level fixed effects and individual-level random effects. This nuanced understanding is crucial for identifying individuals at high risk for depression, and for tailoring interventions to meet their specific needs. The current investigation addresses critical knowledge gaps in understanding long-term depression trajectories in TBI patients. The study objectives were to: (1) characterize the longitudinal course of depression over 10 years post-TBI; (2) identify sociodemographic and clinical characteristics associated with depression trajectories; (3) compare the model's predictive performance using population-level information alone (representing average trends) against predictions that incorporate both population-level and subject-specific information; and (4) evaluate the potential utility of the model for predicting depression trajectories in a clinical setting. We hypothesized that population-level predictions alone would demonstrate poorer performance compared to models incorporating both population and subject-level effects, and that the magnitude of this difference would be substantial, highlighting the critical importance of accounting for individual heterogeneity in predicting depression trajectories following TBI. Methods The study methodology is illustrated in Figure 1. Data source: This retrospective analysis utilized data from the Traumatic Brain Injury Model Systems (TBIMS) National Database, a prospective, multicenter longitudinal database tracking outcomes from injury through 30 years post-injury(Tso et al., 2021). The TBIMS systematically collects data at standardized intervals: initial injury, rehabilitation discharge, and follow-up assessments at 1, 2, 5, and every 5 years thereafter. This study analyzed the publicly available dataset. Sample size: Sample size calculations were performed to ensure sufficient power for a linear mixed-effects model. We assumed a small effect size (f² = 0.05), power of 0.8, and a significance level of 0.05. Given the multilevel design, we accounted for an intraclass correlation coefficient (ICC) of 0.4 and an average of 4 measurements per subject. Based on these parameters, the required sample size to detect statistically significant effects was 12,560 subjects. Study population : Participants were individuals with traumatic brain injury enrolled in the TBIMS database. Eligible participants met at least one of the following clinical criteria: - Glasgow Coma Scale (GCS) score below 13 in the emergency department - Post-traumatic amnesia (PTA) exceeding 24 hours - Presence of intracranial neuroimaging abnormalities - Loss of consciousness (LOC) greater than 30 minutes Additional inclusion criteria specified age ≥16 years and receipt of inpatient rehabilitation at one of 16 participating TBIMS centers. Variables and measurement: The analysis included predictor variables selected based on established associations with post-TBI depression outcomes. Demographic characteristics considered were age (continuous), sex (male/female), marital status (single, married, or divorced/separated/widowed), education level (less than high school, high school graduate, or college degree), and employment status at injury (employed, retired, student, or unemployed). Injury-related characteristics encompassed injury mechanism (fall, motor vehicle collision, violence, or other), injury severity indicators using the Glasgow Coma Scale (GCS), Functional Independence Measure (FIM) score (continuous), and length of stay (continuous). Pre-injury clinical factors included documented history of mental health treatment, substance use disorders, psychiatric hospitalizations, and suicide attempts. The outcome of interest was depression as measured through the Patient Health Questionnaire-9 (PHQ-9). The PHQ-9 is a 9-item measure of depression based on diagnostic criteria (Kroenke et al., 2001). Respondents rate the frequency and severity of depressive symptoms over the last two weeks. Each item is scored using a Likert-type scale from 0–4 where 0 indicates “not at all”, 1 is “several days”, 2 is “more than half the days”, and 3 is “nearly every day”. Total scores range from 0 to 27, with higher scores indicating higher levels of depression symptomatology. The PHQ-9 has a sensitivity of 88% and a specificity of 88% in the diagnosis of major depression when the cut-off score of ≥10 is used and has been validated in the TBI population (Fann et al., 2005). The PHQ-9 was not administered at discharge from acute or rehabilitation care; it was available for follow-up time points. Inverse probability weighting: Estimates of baseline covariates on PHQ-9 scores may be underestimated due to differential attrition in the TBIMS over the study period. The analyses included weights for the inverse probability of attrition to account for this attrition (Weuve et al., 2012). Specifically, we estimated follow-up time-specific weights based on the inverse probability of being observed at each time point. These weights were then applied to each observation to minimize potential biases in estimates of baseline covariates on PHQ-9 scores. Data Analysis: Baseline characteristics were compared using t-tests for continuous variables and chi-square tests for categorical variables. Discrete variables are reported as counts (percentages), and continuous variables as means (standard deviations). Longitudinal Modeling: We employed linear mixed-effects models to analyze longitudinal PHQ-9 trajectories. The model specification was: PHQ9_ij = β₀ + β₁(Time_ij) + β₂X_i + β₃(Time_ij × X_i) + b₀i + b₁i(Time_ij) + ε_ij Where: i denotes individual j denotes measurement occasion β represents fixed effects b represents random effects X represents covariates ε represents residual error The longitudinal trend in depression over time (measured at 1, 2, 5, and 10 years) was assessed using linear mixed models to account for the correlation between repeated measurements within the same individual and to include all available data. A random intercept was included for each subject to account for individual differences in depression levels at baseline. A random slope for the effect of time was also included to capture individual variations in the rate of change in depression over time. Only significant interactions are reported. As PHQ-9 scores were first measured at the Year 1 visit, the time variable was centered at Year 1. This allowed the main effects in the model to reflect covariate effects on PHQ scores from baseline to Year 1, and the covariate-by-time interactions to represent the effects of covariates on the rate of depression in subsequent years. Model diagnostics included visual inspection of standardized residual and quantile-quantile (Q-Q) plots, which indicated approximately normally distributed residuals with some deviations at the extremes. Missing data for the outcome variable were not imputed, as a likelihood-based approach was used to handle missing data under the assumption that it was missing at random. Only covariate data were imputed, and the results presented are based on a pooled analysis of 10 imputed datasets. Minimal Clinically Important Difference : We evaluated the effect of time on PHQ scores using two established methods to classify participants' changes based on the proportion of individuals achieving a minimal clinically important difference (MCID). The MCID represents the smallest change in PHQ scores perceived as clinically meaningful. The widely accepted MCID for the PHQ-9 is a 20% change in score or 5 points and is commonly used to assess intervention effects in depression trials (Carlo et al., 2021; Löwe et al., 2004). It is important to note that the dataset does not contain information on whether participants were receiving active treatment or interventions for depression during the follow-up period. Therefore, given the observational nature of the current study, we also adopted a more conservative estimate of MCID based on the standard error of measurement (SEM) as a criterion for clinically relevant change (Łakuta et al., 2022; Turkoz et al., 2021). To account for intra-individual variability and establish a 95% confidence interval, the SEM for the PHQ-9 was multiplied by 1.96. Using this approach, the conservative MCID for the PHQ-9 was determined to be 4 points. Model assessment: We distinguish between fitted values (predictions on observed data used to build the model) and predictions (on unseen data). We examined model performance for fitted values at two levels: Level 0 (Population-level): Predictions based solely on fixed effects, representing population or average trajectories. Level 1 (Subject-specific): Predictions conditional on the Best Linear Unbiased Predictions (BLUPs) of the random effects, capturing individual deviations from the population average. For each level, we assessed the residual distributions and prediction metrics (mean squared error [MSE], coverage, precision). MSE was calculated as the average of the squared differences between the observed PHQ-9 scores and the predicted PHQ-9 scores. A lower MSE indicates better overall prediction accuracy, with smaller average deviations between the predictions and the actual observed values. Coverage was defined as the proportion of realized PHQ-9 observations within the 50% prediction interval. Precision was the average width of the 50% prediction interval for predicted PHQ-9 values. Predictive performance assessment: To provide a proof of concept demonstrating the potential clinical utility of the developed linear mixed-effects model for predicting individual depression trajectories, we designed two scenarios designed to simulate real-world clinical applications: (1) prediction of depression trajectory for new patients and (2) forecasting the depression trajectory for existing patients using initial PHQ-9 scores. It is important to note that these scenarios are intended to illustrate the feasibility of using the model for individualized prediction and to generate hypotheses for future research, rather than to provide definitive evidence of its clinical effectiveness. Scenario 1: Prediction of future depression scores using early data This scenario aimed to evaluate the model's ability to predict future depression severity based on early longitudinal data. The model was trained on data from Years 1 and 2 to capture initial depression trajectories. This trained model was then used to predict individual depression scores at Year 5, using observed PHQ-9 scores from Years 1 and 2, along with baseline demographic and injury-related characteristics. Scenario 2: Prediction of depression trajectory for new patients This scenario aimed to evaluate the model's ability to generalize to individuals not included in the model fitting process. We created a test set of participants by excluding a random subset of individuals from the original dataset. The model was then trained on the remaining participants and used to predict individual depression trajectories (PHQ-9 scores at Year 2) for the individuals in the test set, using only their baseline demographic and injury-related characteristics. For both scenarios, we compared two types of predictions: marginal predictions based on fixed effects only (population-level estimates) and conditional predictions that incorporated both fixed and random effects (subject-specific estimates). This comparison allowed us to assess the added value of including subject-level variation in a prediction setting. The predictive performance of models from both scenarios was quantified using metrics of MSE, coverage, and precision. Sensitivity analyses : To ensure the robustness of our findings, we conducted a sensitivity analyses. we excluded participants whose injury was caused by gunshot wounds to evaluate whether these cases influenced the overall results. The following R packages were used for data analysis and visualization: nlme for fitting linear mixed-effects models, Jmbayes for model prediction in longitudinal data emmeans for post-hoc pairwise comparisons of estimated marginal means, and mice for multiple imputation of missing data. Results Sample characteristics Table 1 presents the demographic, injury-related, and pre-injury clinical characteristics of the overall cohort (n = 19,397). The mean age was 43 years. The majority of participants were male (74%) and identified as White (66%), followed by Black (18%) and Hispanic (11%). Regarding marital status, 46% were single, 33% were married, and 21% were separated, widowed, or divorced. In terms of educational attainment, 40% had completed college, 36% had completed high school, and 24% had less than a high school education. At the time of injury, 61% were competitively employed, 18% were retired, and 12% were unemployed. Table 1 Baseline characteristics of the study population Characteristic N = 19,397 1 Age 43 (20) Sex Female 5,107 (26%) Male 14,278 (74%) Race Asian/Pacific Islander 539 (2.8%) Black 3,548 (18%) Hispanic Origin 2,181 (11%) Native American 104 (0.5%) Other 237 (1.2%) White 12,745 (66%) Marital status Married 6,440 (33%) Sep_Wid_Div 4,081 (21%) Single 8,828 (46%) Income =>100,000 687 (5.6%) 20–49,999 3,671 (30%) 50–99,999 1,872 (15%) Not Competitively Employed 5,973 (49%) Education college 7,654 (40%) HS 6,789 (36%) less than HS 4,637 (24%) Employment Competitively employed 11,694 (61%) Other 648 (3.4%) Retired 3,420 (18%) Student 1,172 (6.1%) Unemployed 2,327 (12%) Cause Fall 5,588 (29%) MVC 8,905 (46%) Other 2,758 (14%) Violence 2,146 (11%) GCS Chemically Paralyzed or Sedated 4,469 (23%) Mild 6,342 (33%) Moderate 2,271 (12%) Severe 6,259 (32%) Acute Payor Medicare 3,012 (16%) Other 255 (1.3%) Private 10,874 (57%) Public 5,001 (26%) Rehab Payor Medicare 3,061 (16%) Other 228 (1.2%) Private 10,637 (55%) Public 5,387 (28%) Drugs 3,891 (21%) Drink Category Abstaining 6,813 (41%) Heavy 2,538 (15%) Light 3,276 (20%) Moderate 4,075 (24%) Unknown 2,695 Mental health treatment 2,575 (22%) Psychiatric hospitalization 834 (7.1%) Suicide 610 (5.2%) FIM Motor 66 (19) FIM Cognitive 24 (7) LOS Acute care 21 (18) LOS Rehab care 27 (26) Pre-injury TBI 0.39 (0.89) 1 Mean (SD); n (%) Motor vehicle collisions (MVC) were the most common cause of injury (46%), followed by falls (29%), and violence (11%). Based on the Glasgow Coma Scale (GCS) distribution, 32% experienced a severe TBI, 12% had a moderate TBI, and 33% had a mild TBI. For the remaining 23% of participants, GCS scores were unavailable due to chemical paralysis or sedation at the time of assessment. Regarding pre-injury clinical factors, 21% reported drug use, and alcohol consumption was categorized as light (20%), moderate (24%), and heavy (15%). Additionally, 22% reported pre-injury mental health treatment, 7% had been previously hospitalized for psychiatric causes, and 5% had a history of suicide attempts. Factors associated with depression : Table 2 presents the results of the linear mixed-effects model assessing various factors related to changes in the PHQ-9 total score over time. Several baseline covariates were significantly associated with PHQ-9 scores at Year 1. Table 2 shows the changes in PHQ-9 scores and effect sizes for the model covariates. The largest main effect and effect size was for pre-injury mental health treatment (β = 1.6, 95% CI: 1.21–1.97). A higher number of head injuries prior to the index TBI (β = 0.4, 95% CI: 0.31–0.52) and being female (β = 0.86, 95% CI: 0.63–1.09) were also associated with higher PHQ-9 scores. Table 2 Linear mixed effect model of PHQ-9 scores Characteristic Beta 95% CI 1 p-value Fixed effects Intercept 5.511 4.06–6.96 0.000 Time -0.006 -0.03-0.02 0.662 Mental health treatment No — — Yes 1.592 1.21–1.97 0.000 Age -0.024 -0.03-0.01 0.000 FIM Motor 0.006 -0.00-0.01 0.121 FIM Cognitive 0.006 -0.01-0.02 0.549 LOS Acute care 0.004 -0.00-0.01 0.283 LOS Rehab care -0.004 -0.01-0.00 0.132 Pre-injury TBI 0.418 0.31–0.52 0.000 Sex Male — — Female 0.863 0.63–1.09 0.000 Race White — — Asian/Pacific Islander -0.092 -0.71-0.53 0.771 Black 0.702 0.42–0.99 0.000 Hispanic Origin 0.564 0.21–0.92 0.002 Native American 1.449 0.19–2.71 0.024 Other 0.648 -0.29-1.59 0.177 Marital status Married — — Sep_Wid_Div 0.558 0.27–0.85 0.000 Single -0.952 -1.24-0.67 0.000 Employment Other — — Competitively employed -0.502 -1.10-0.09 0.099 Retired -0.237 -0.89-0.42 0.479 Student -0.975 -1.68-0.27 0.007 Unemployed 0.064 -0.59-0.72 0.847 Cause Other — — Fall -0.058 -0.39-0.28 0.733 MVC 0.155 -0.14-0.45 0.300 Violence 0.618 0.20–1.04 0.004 Acute Payor Other — — Medicare -0.507 -2.04-1.03 0.518 Private -0.485 -1.71-0.74 0.437 Public 0.191 -1.05-1.44 0.763 Rehab Payor Other — — Medicare -0.364 -1.94-1.21 0.650 Private -0.172 -1.43-1.09 0.789 Public 0.140 -1.13-1.41 0.830 Drugs No — — Yes 0.837 0.57–1.10 0.000 Psychiatric hospitalization No — — Yes 0.188 -0.40-0.78 0.526 Suicide No — — Yes 0.852 0.10–1.60 0.027 Drinking Category Abstaining — — Heavy 0.194 -0.13-0.52 0.242 Light 0.164 -0.11-0.44 0.246 Moderate -0.332 -0.60-0.07 0.015 Education college — HS 0.536 0.30–0.76 0.000 less than HS 1.066 0.78–1.35 0.0000.000 GCS Mild Chemically Paralyzed or Sedated -0.054 -0.34-0.24 0.715 Moderate 0.094 -0.26-0.45 0.602 Severe -0.185 -0.47-0.10 0.204 Time * Mental health treatment -0.143 -0.21- -0.07 0.000 Random effects Random intercept 18.50 Random slope 0.13 Covariance between intercept and slope -0.49 Residual variance 13.83 ICC 0.57 1 CI = Confidence Interval Racial differences were significant, with Black (β = 0.70, 95% CI: 0.42–0.99) and Hispanic (β = 0.56, 95% CI: 0.21–0.92) patients having higher scores compared to White individuals. Patients who were single (β = -0.95, 95% CI: -1.24- -0.67) had lower depression scores compared to those who were married, whereas patients who were separated/widowed/divorced had higher scores (β = 0.56, 95% CI: 0.27–0.85). Those with less than a high school education (β = 1.1, 95% CI: 0.78–1.35) and high school education (β = 0.54, 95% CI: 0.30–0.76) had higher depression scores compared to college graduates. Patients who were injured by violent causes had higher depression scores (β = 0.62, 95% CI: 0.20–1.04) compared to those injured by other causes. Other significant factors included substance use, with individuals who reported pre-injury drug use having higher PHQ-9 scores (β = 0.84, 95% CI: 0.57–1.10). Those with a history of suicide attempts had significantly higher scores (β = 0.85, 95% CI: 0.10–1.60). Longitudinal trend of depression : There was a significant interaction between follow-up time and pre-injury treatment of mental health conditions indicating that the effect of time on depression scores differed by the presence of psychiatric history. Specifically, for individuals with a history of pre-injury mental health treatment, depression scores demonstrated a small decrease over time (β =-0.14, 95% CI: -0.21- -0.07). The interaction was analyzed by decomposing the effects to explore differences in PHQ scores based on pre-injury mental health treatment status, while accounting for the influence of other covariates by averaging over their levels in the model. Figure 2 illustrates a decline in PHQ scores over ten years among patients who received mental health treatment prior to injury, whereas those without prior treatment exhibited stable scores over the same period. Table 3 presents the marginal means for PHQ scores stratified by pre-injury mental health treatment status. Although the decline in PHQ scores among the treated group was statistically significant across the study period, this decrease amounted to 1.3 points from Year 1 to Year 10. Table 3 Marginal means, standard errors, and 95% confidence intervals for PHQ-9 scores by pre-injury mental health treatment status Follow up year Mental health treatment = No Mental health treatment = Yes Means SE 95%CI Means SE 95%CI 1 7.13 0.27 6.59–7.67 8.72 0.27 8.19–9.26 2 7.13 0.27 6.59–7.66 8.57 0.26 8.06–9.09 5 7.11 0.27 6.57–7.64 8.13 0.26 7.62–8.63 10 7.08 0.29 6.51–7.64 7.38 0.32 6.76–8.01 Heterogeneity in baseline depression and trends In addition to the fixed effects, the model included random effects, indicating heterogeneity in Year 1 depression scores and their trends across ten years. The variance of the random intercept (subject-level variation in PHQ-9 scores) was estimated at 18.50 (SD = 4.30), reflecting substantial individual variability in baseline PHQ-9 scores. This suggests that some individuals had much higher or lower initial depression scores than the average participant. The variance of the random slope for time (variation in the rate of change in PHQ-9 scores across individuals) was 0.13 (SD = 0.36), indicating variability in the rate of change of PHQ-9 scores over time across individuals. In other words, some participants' depression scores improved more rapidly than others, while some showed little to no change, or even worsened over time. The intraclass correlation coefficient (ICC) was 0.57, suggesting that 57% of the total variance in PHQ-9 scores was attributable to between-subject differences. Model performance : Model performance was assessed at two levels: Level 0 (population-level), representing predictions based solely on fixed effects and reflecting population-average trajectories, and Level 1 (subject-specific), incorporating both fixed and random effects to capture individual deviations from the population average. Model performance was evaluated using MSE, coverage, and precision. For Level 0, the MSE was 31.17 and the coverage was 23%; i.e., a small proportion of observed PHQ-9 scores fell within the 50% prediction intervals. The model precision at Level 0 was 3.76 points. Visual inspection of the residual plot for population-level predictions revealed a tendency for the model to under-predict higher PHQ-9 scores and suggested potential heteroscedasticity. Level 1 fitted values demonstrated a substantially lower MSE of 7.75, indicating a considerably smaller average squared difference between the observed and predicted individual depression scores compared to the Level 0. For Level 1, the coverage improved to 58% within the prediction intervals and model precision was 3.75 points. Visual inspection of the residual plot for Level 1 (BLUP) predictions suggested some remaining heteroscedasticity, though less pronounced than at Level 0. These similar precision values suggest that while both approaches had comparable interval widths, Level 1 fitted values achieved this precision with a greater accuracy in predicting individual outcomes, as evidenced by the lower MSE and improved coverage within the 50% prediction intervals. Predictive performance assessment To evaluate the potential for clinical application, we assessed the model's ability to predict future depression scores (Year 5) using early longitudinal data (Years 1 and 2). This within-sample prediction scenario reflects a setting where repeated measurements are available for the same individuals (Table 4 , Scenario 1). The conditional prediction using random effects yielded an MSE of 0.014, and the coverage within the prediction intervals was 100%. The precision, as measured by the average width of the 50% prediction intervals, was 3.7 points. This indicated excellent subject-specific predictive accuracy when sufficient longitudinal information was available for the individual. Table 4 Prediction metrics for clinical scenarios Prediction level Scenario 1 Scenario 2 MSE Coverage Precision MSE Coverage Precision Marginal prediction 30.48 18 2.5 29.07 33 5.1 Conditional prediction 0.014 100 3.7 19.57 64 6.0 In the second scenario, we assessed the model’s generalizability to new patients by evaluating its ability to predict depression scores in a hold-out test set (Table 4 , Scenario 2). The model was trained on a subset of participants and applied to new individuals, simulating prediction in an unobserved population using only their Year 1 scores to forecast Year 2 outcomes. In this out-of-sample setting, the conditional prediction achieved an MSE of 19.57. The coverage within the 50% prediction intervals was 64%, and the average interval width (precision) was 6 points. Conditional predictions outperformed marginal predictions across both scenarios, with lower MSE and higher interval coverage. In the sensitivity analyses, excluding participants with gunshot wounds, yielded results consistent with the main model; the beta coefficients for key predictors differed by less than 5%. Discussion In the current study, we employed linear mixed modeling to account for both fixed effects (population-level trends) and random effects (subject-level deviations) to capture the heterogeneity in depression trajectories following TBI. This approach treats depression as a continuous phenomenon, providing a more nuanced representation of depressive symptom evolution over time. When examining depression trends, we distinguished between statistical and clinical significance, as small numerical differences in depression scores may yield statistically significant results in large samples without translating to meaningful clinical change [12]. To address this limitation, we utilized the minimally clinically important difference (MCID) to identify meaningful changes in PHQ-9 scores. This study is unique in its demonstration of the limitations of relying solely on population-level information when predicting depression trajectories in patients with TBI, highlighting the critical need to account for individual heterogeneity. As hypothesized, models incorporating random effects demonstrated substantially improved predictive performance compared to those based solely on fixed effects. Specifically, while population-level estimates captured only a small proportion of the variance in PHQ-9 scores (marginal R² = 0.07), incorporating random effects led to a marked increase in explained variance (conditional R² = 0.58) and a substantial improvement in coverage within the 50% prediction intervals (23% vs. 58%). These findings underscore the inherent variability in depression symptoms following TBI; ignoring this individual-level variability may lead to inaccurate predictions and potentially ineffective clinical decision-making. The primary goal of this study was to gain a better understanding of the dynamics of depression over time and to highlight factors associated with depression scores among patients with TBI. Our findings revealed that there is considerable heterogeneity between individuals for initial depression scores. Depression trends varied by pre-injury mental health treatment. Patients with a history of pre-injury mental health treatment exhibited a small but statistically significant decrease in depression scores over the follow-up period. However, this reduction did not reach the threshold for minimal clinically important differences, suggesting that while trends may differ statistically across clinical factors, the observed changes are unlikely to be clinically meaningful. The overall stability of depression trends observed in our study is consistent with findings from other studies. In the general population, the prevalence of dysthymia and major depressive disorders has remained relatively stable between 1990–2019, despite an overall increase in the global burden of depression (Liu et al., 2024 ). While longitudinal studies examining depression trends specifically in the traumatic brain injury (TBI) population are limited, trajectory modeling studies have provided valuable insights; a seminal study employing group-based trajectory modeling identified three distinct mental health trajectories among adults with TBI over 2.5 years post-injury: low-stable, medium-stable, and high-stable (Feldman et al., 2020 ). Shen et al identified two main trajectories in adolescents with TBI: a low-stable group (85%) and a high-increasing group (15%) over a 10-year period (Shen & Wang, 2023 ). Other studies have identified stable, improved, and delayed onset trajectories (Bombardier et al., 2016 ; Gomez et al., 2017 ; Heath et al., 2023 ). Collectively, these findings suggest that the course of depression after head injury is heterogeneous, with the majority (70–80%) of participants exhibiting stable trajectories. In addition to identifying longitudinal trends, this study investigated the demographic, clinical, and injury related factors associated with Year 1 depression scores. Consistent with broader epidemiological trends, sex and age appear to shape the vulnerability to depression. Among demographic factors, females in the current study were more likely to experience higher depression scores than males. In accordance with previous studies which have found that older adults have lower depression scores than their younger counterparts (Bombardier et al., 2010 , 2016 ; Passler et al., 2022 ), we also found that there was a small deceasing depression trend with increasing age. Results from the current study demonstrate that Black and Hispanic races reported higher scores than White patients, a finding supported by several others authors (Arango-Lasprilla et al., 2012 ). Several factors contribute to the increased risk of depression in racial minorities. Individuals from marginalized racial backgrounds may face economic instability which can exacerbate mental health issues after an injury (Stein et al., 2019 ). Additionally, racial minorities are reported to have less access to quality mental health care, which can hinder their ability to receive timely and effective treatment for post-injury psychiatric conditions (Brenner et al., 2020 ). Pre-injury psychiatric history and substance use are well established risk factors for increased depressive symptoms after head injury (Bockhop et al., 2023 ; Delmonico et al., 2022 ). In line with this finding, we observed that pre-injury mental health treatment, which may serve as a proxy for mental health conditions, had the strongest association with initial depression scores. In addition, suicide attempts, and a history of drug use also increased depression. Given this elevated risk, it is important for healthcare providers to implement targeted screening for affective disorders in patients with a history of psychiatric issues after TBI. Although several studies have highlighted the bivariate association of injury severity with depression, the majority of studies have not found an association between injury severity and symptoms of depression in multivariable analysis. Similarly, we also did not see an effect of GCS on depression scores, suggesting that clinicians should anticipate and address depression during recovery from physical trauma, regardless of injury severity (Versluijs et al., 2022 ). Although other studies have found that higher levels of cognitive impairment at discharge predict elevated depression trajectories (Cariello et al., 2020 ) and lower functional independence is linked to poorer mental health outcomes (Carmichael et al., 2023 ), we did not find any association with the motor and cognition subscales of FIM. The analysis of predictive performance revealed that the model's ability varied depending on the specific clinical context – specifically, whether the task involved forecasting future scores for existing patients or predicting scores for unseen individuals. When predicting PHQ-9 scores for new patients based on baseline PHQ-9, demographic and clinical data, the model's 50% prediction interval had an average width of 6 points (± 3 points around the predicted score), indicating that half of the observed outcomes are expected to fall within this range. This prediction interval is wider than the commonly accepted MCID of 5 points, indicating uncertainty in identifying new patients who are likely to experience meaningful changes in their depression symptoms. The relatively infrequent data collection of PHQ-9 (Years 1, 2, 5, and 10) in the study cohort may partially explain the model's limited precision for new patients, as the long intervals between assessments could miss important temporal dynamics in depression symptoms. Regular data collection may yield more precise predictions, particularly for new patients without established trajectories. In contrast, when predicting Year 5 PHQ-9 scores for existing patients using their prior data (Years 1 and 2), the model demonstrated substantially better predictive performance. The conditional prediction in this scenario achieved an average prediction interval width of 4 points (± 2 points around the predicted score). This improved precision likely reflects the benefit of incorporating individual longitudinal data, which enabled the model to better capture subject-specific patterns in depression trajectory. Importantly, this prediction interval is narrower than the commonly accepted MCID of 5 points, suggesting the model may have sufficient precision to detect clinically meaningful changes in this longitudinal prediction scenario. Across both scenarios, conditional predictions outperformed marginal predictions reinforcing the importance of accounting for individual-level variability when predicting depression scores. Overall, the predictive performance results suggest that with further refinement, the model has the potential to aid clinicians in making informed decisions regarding depression treatment planning and resource allocation. The current study has important public health and clinical implications. First, the overall depression trend indicates the need for sustained, targeted interventions shortly after injury to achieve clinically meaningful improvements in mental health outcomes. In the study sample 44% of patients experienced symptoms consistent with mild, moderate, or severe depression, emphasizing the necessity for systematic screening and management protocols for this substantial proportion of TBI patients. Secondly, from a clinical standpoint, we identified and confirmed several demographic, injury, and clinical factors associated with higher initial depression score, including sex, education, employment, marital status, and TBIs prior to the index injury, which can help clinicians identify high-risk patients requiring enhanced monitoring and intervention. Finally, our predictive modeling approach demonstrates potential clinical utility, particularly for existing patients with longitudinal data where the model achieved sufficient precision to detect clinically meaningful changes. This could support more personalized depression management strategies and resource allocation in longitudinal TBI care. Strengths and limitations This study has several strengths. Our modelling approach enables us to estimate both population-level effects, quantifying how fixed factors influence depression scores across the population, as well as individual-level variability in baseline depression scores, capturing how individuals deviate from the population's average score at baseline. We used data from the TBIMS which represents the largest prospective TBI outcome study in North America (Dams-O’Connor et al., 2018 ). Therefore, our results may generalize to other TBI populations, particularly within other regions of the United States and Canada. The identified model was developed using data collected immediately upon arrival in the ED and upon hospital discharge. All information included in the model is easily collectable and could be routinely included from electronic health records. In contrast to many applications of mixed-effects models for prediction, which often utilize only the fixed effects components, this study leveraged Best Linear Unbiased Predictors (BLUPs) to incorporate individual random effects, an approach we demonstrated is essential for achieving accurate and clinically meaningful predictions of depression trajectories following TBI. Though this study yields important findings regarding the evolution of depression during the first 10 years after TBI, it is acknowledged that there are limitations that warrant cautious interpretation. As the TBIMS databases collect follow-up information at 1, 2, and 5 years, and every 5 years thereafter, it is common to have incomplete data on patients due to participant withdrawal or loss of follow-up. In addition, missing information may also occur for specific time periods that a given variable had been collected/included by the data set. To account for this limitation, we used a modelling approach which accommodates missing data that is common in longitudinal study designs. To account for potentially informative attrition in our analyses, we estimated weights to apply to each observation in modelling process. By using follow-up time-specific weights, we ensured that the analyses accounted for varying probabilities of observation across different time points, providing a more accurate representation of the relationship between baseline covariates and PHQ-9 scores. We did not include time-varying covariates in our analysis, as many of these factors, such as changes in marital status, employment, or social support, have a reciprocal relationship with depression. For instance, worsening depression can lead to reduced employment opportunities or social isolation, while changes in these factors may also contribute to shifts in depression severity. Including such covariates would complicate the interpretation of their influence, making it unclear whether they are predictors or consequences of depression. True baseline depression is not captured in TBIMIS at the time of injury or discharge from hospital, therefore we could not account for baseline PHQ-9 scores in the model. Finally, we did not have data on whether participants were receiving treatment for depression during the follow-up period. This limitation restricts our ability to directly evaluate the impact of clinical care or therapeutic interventions on changes in depression scores over time. Despite these limitations, the study uniquely provided ten years of prospective data on depression recovery for a cohort of individuals with TBI. If replicated, our results have important public health implications and can inform strategies to optimize depression screening and intervention in the early post-injury period. Conclusions To our knowledge this is the largest study to model the impact of multiple factors on depression trends for a large population of individuals who sustain a TBI over a ten-year period after injury. Although the population-wide depression scores exhibited a small decrease over the study period, there was substantial variability between individuals in terms of their baseline depression as well as the rate of change over time. We identified several demographic and clinical covariates associated with initial depression scores. Importantly, we demonstrated that population-level models alone inadequately capture the complexity of depression experiences following TBI, whereas accounting for subject-level variations significantly improves prediction. Our modeling approach has the potential to predict depression in patients being followed longitudinally as well as identify depression trajectories in new patients. Declarations Acknowledgements: The Traumatic Brain Injury (TBI) Model Systems National Database is a multicenter study of the TBI Model Systems Centers Program and is supported by the National Institute on Disability, Independent Living, and Rehabilitation Research (NIDILRR), a center within the Administration for Community Living (ACL) and Department of Health and Human Services (HHS). However, these contents do not necessarily reflect the opinions or views of the TBI Model Systems Centers, NIDILRR, ACL or HHS. Funding: This project was supported by the Canadian Institutes of Health Research (CIHR) through the Canada Graduate Scholarships-Doctoral (CGS-D) Award. The funder had no role in the preparation of this manuscript. Conflict of interest: The authors have no conflict of interest to declare. Consent for publication: Not applicable. Availability of data and material: Traumatic Brain Injury Model Systems Database (TBIMS) Ethics statement: This study is secondary data analysis of a publicly available dataset. Informed consent was signed by the patient, family or guardian. Author contributions: NK contributed to the conception and design of the study, acquisition and analysis of the data, and drafting of the manuscript. DBC, AN, SSRA and CF contributed to the design of the study, data analysis, and review of the manuscript. This work was part of the PhD thesis of NK in the Interdisciplinary PhD program at Dalhousie University under the supervision of SSRA and CF. All authors interpreted the data, revised the manuscript critically for important intellectual content, and provided final approval of the version to be published. References Anderson, J. F. I. (2025). Heterogeneity of health-related quality of life after mild traumatic brain injury with systemic injury: A cluster analytic approach. Disability and Rehabilitation , 47 (2), 347–356. https://doi.org/10.1080/09638288.2024.2345278 Arango-Lasprilla, J. C., Ketchum, J. M., Drew, A., Hammond, F., Powell, J. M., Kreutzer, J., Lim, P. C., & Carr, C. (2012). Neurobehavioural symptoms 1 year after traumatic brain injury: A preliminary study of the relationship between race/ethnicity and symptoms. Brain Injury , 26 (6), 814–824. https://doi.org/10.3109/02699052.2012.655360 Bockhop, F., Cunitz, K., Zeldovich, M., Buchheim, A., Beissbarth, T., Hagmayer, Y., von Steinbuechel, N., & CENTER-TBI Participants and Investigators. (2023). Influence of Sociodemographic, Premorbid, and Injury-Related Factors on Post-Traumatic Stress, Anxiety, and Depression after Traumatic Brain Injury. Journal of Clinical Medicine , 12 (12), 3873. https://doi.org/10.3390/jcm12123873 Bombardier, C. H., Fann, J. R., Temkin, N. R., Esselman, P. C., Barber, J., & Dikmen, S. S. (2010). Rates of major depressive disorder and clinical outcomes following traumatic brain injury. JAMA : The Journal of the American Medical Association , 303 (19), 1938–1945. https://doi.org/10.1001/jama.2010.599 Bombardier, C. H., Hoekstra, T., Dikmen, S., & Fann, J. R. (2016). Depression Trajectories during the First Year after Traumatic Brain Injury. Journal of Neurotrauma , 33 (23), 2115–2124. https://doi.org/10.1089/neu.2015.4349 Brenner, E. K., Grossner, E. C., Johnson, B. N., Bernier, R. A., Soto, J., & Hillary, F. G. (2020). Race and ethnicity considerations in traumatic brain injury research: Incidence, reporting, and outcome. Brain Injury , 34 (6), 801–810. https://doi.org/10.1080/02699052.2020.1741033 Brown, V. A. (2021). An Introduction to Linear Mixed-Effects Modeling in R. Advances in Methods and Practices in Psychological Science , 4 (1), 2515245920960351. https://doi.org/10.1177/2515245920960351 Cariello, A. N., Perrin, P. B., Agudelo, Y. R., Olivera Plaza, S. L., Quijano-Martínez, M. C., Trujillo, M. A., Arango-Lasprilla, J. C., & Begali, V. (2020). Predictors of longitudinal depression trajectories after traumatic brain injury in Latin America: A multi-site study. NeuroRehabilitation , 46 (2), 205–212. https://doi.org/10.3233/NRE-192972 Carlo, A. D., Basu, A., & Unützer, J. (2021). Associations of Common Depression Treatment Metrics with Patient-Centered Outcomes. Medical Care , 59 (7), 579–587. https://doi.org/10.1097/MLR.0000000000001540 Carmichael, J., Hicks, A. J., Gould, K. R., Ponsford, J., & Spitz, G. (2023). Ten-Year Cohort Study of Emotional Distress Trajectories After Moderate-Severe Traumatic Brain Injury. Archives of Physical Medicine and Rehabilitation , 104 (7), 1081–1090. https://doi.org/10.1016/j.apmr.2023.02.008 Covington, N. V., & Duff, M. C. (2021). Heterogeneity Is a Hallmark of Traumatic Brain Injury, Not a Limitation: A New Perspective on Study Design in Rehabilitation Research. American Journal of Speech-Language Pathology , 30 (2S), 974–985. https://doi.org/10.1044/2020_AJSLP-20-00081 Dams-O’Connor, K., Sy, K. T. L., Landau, A., Bodien, Y., Dikmen, S., Felix, E. R., Giacino, J. T., Gibbons, L., Hammond, F. M., Hart, T., Johnson-Greene, D., Lengenfelder, J., Lequerica, A., Newman, J., Novack, T., O’Neil-Pirozzi, T. M., & Whiteneck, G. (2018). The Feasibility of Telephone-Administered Cognitive Testing in Individuals 1 and 2 Years after Inpatient Rehabilitation for Traumatic Brain Injury. Journal of Neurotrauma , 35 (10), 1138–1145. https://doi.org/10.1089/neu.2017.5347 Delmonico, R. L., Theodore, B. R., Sandel, M. E., Armstrong, M. A., & Camicia, M. (2022). Prevalence of depression and anxiety disorders following mild traumatic brain injury. PM&R , 14 (7), 753–763. https://doi.org/10.1002/pmrj.12657 Fakhoury, M., Shakkour, Z., Kobeissy, F., & Lawand, N. (2021). Depression following traumatic brain injury: A comprehensive overview. Reviews in the Neurosciences , 32 (3), 289–303. https://doi.org/10.1515/revneuro-2020-0037 Fann, J. R., Bombardier, C. H., Dikmen, S., Esselman, P., Warms, C. A., Pelzer, E., Rau, H., & Temkin, N. (2005). Validity of the Patient Health Questionnaire-9 in assessing depression following traumatic brain injury. The Journal of Head Trauma Rehabilitation , 20 (6), 501–511. https://doi.org/10.1097/00001199-200511000-00003 Feldman, B., Shen, J., Chen, C., Shi, J., & Xiang, H. (2020). Perceived health after adult traumatic brain injury: A Group-Based Trajectory Modeling (GBTM) analysis. Brain Injury , 34 (6), 741–750. https://doi.org/10.1080/02699052.2020.1753111 Gomez, R., Skilbeck, C., Thomas, M., & Slatyer, M. (2017). Growth Mixture Modeling of Depression Symptoms Following Traumatic Brain Injury. Frontiers in Psychology , 8 . https://doi.org/10.3389/fpsyg.2017.01320 Hart, T., Hoffman, J. M., Pretz, C., Kennedy, R., Clark, A. N., & Brenner, L. A. (2012). A Longitudinal Study of Major and Minor Depression Following Traumatic Brain Injury. Archives of Physical Medicine and Rehabilitation , 93 (8), 1343–1349. https://doi.org/10.1016/j.apmr.2012.03.036 Heath, L. M., Kidwai, M. R., Colella, B., Monette, G., Tselichtchev, P., Tomaszczyk, J. C., & Green, R. E. (2023). Predictors and Functional Outcomes Associated With Longitudinal Trajectories of Anxiety and Depression from 2 to ≥36 Months After Moderate to Severe Traumatic Brain Injury. Journal of Neurotrauma , 40 (21–22), 2311–2320. https://doi.org/10.1089/neu.2023.0003 Jorge, R. E., Robinson, R. G., Moser, D., Tateno, A., Crespo-Facorro, B., & Arndt, S. (2004). Major Depression Following Traumatic Brain Injury. Archives of General Psychiatry , 61 (1), 42–50. https://doi.org/10.1001/archpsyc.61.1.42 Kroenke, K., Spitzer, R. L., & Williams, J. B. (2001). The PHQ-9: Validity of a brief depression severity measure. Journal of General Internal Medicine , 16 (9), 606–613. https://doi.org/10.1046/j.1525-1497.2001.016009606.x Łakuta, P., Krankowska, D., Marcinkiewicz, P., Bociąga‐Jasik, M., & Komorska‐Błażewicz, A. (2022). Enhancing well‐being and alleviating depressive symptoms in people with HIV/AIDS: An intervention based on if–then plans with self‐affirming cognitions. Applied Psychology. Health and Well-Being , 14 (3), 899–919. https://doi.org/10.1111/aphw.12357 Liu, J., Liu, Y., Ma, W., Tong, Y., & Zheng, J. (2024). Temporal and spatial trend analysis of all-cause depression burden based on Global Burden of Disease (GBD) 2019 study. Scientific Reports , 14 (1), 12346. https://doi.org/10.1038/s41598-024-62381-9 Löwe, B., Unützer, J., Callahan, C. M., Perkins, A. J., & Kroenke, K. (2004). Monitoring depression treatment outcomes with the patient health questionnaire-9. Medical Care , 42 (12), 1194–1201. https://doi.org/10.1097/00005650-200412000-00006 McInnes, A. N., Olsen, S. T., Sullivan, C. R. P., Cooper, D. C., Wilson, S., Sonmez, A. I., Albott, S. C., Olson, S. C., Peterson, C. B., Rittberg, B. R., Herman, A., Bajzer, M., Nahas, Z., & Widge, A. S. (2024). Trajectory Modeling and Response Prediction in Transcranial Magnetic Stimulation for Depression. medRxiv , 2024.05.30.24308258. https://doi.org/10.1101/2024.05.30.24308258 Moreno-Agostino, D., Wu, Y.-T., Daskalopoulou, C., Hasan, M. T., Huisman, M., & Prina, M. (2021). Global trends in the prevalence and incidence of depression:a systematic review and meta-analysis. Journal of Affective Disorders , 281 , 235–243. https://doi.org/10.1016/j.jad.2020.12.035 Passler, J. S., Sander, A. M., Temkin, N. R., Barber, J., Gardner, R. C., & Manley, G. T. (2022). Depression in Older Adults 12 Months After Traumatic Brain Injury: A TRACK-TBI Study. Archives of Physical Medicine and Rehabilitation , 103 (1), 83–89. https://doi.org/10.1016/j.apmr.2021.08.021 Rabinowitz, A. R., Chervoneva, I., Hart, T., O’Neil-Pirozzi, T. M., Juengst, S. B., & Hoffman, J. M. (2020). Heterogeneity in Temporal Ordering of Depression and Participation After Traumatic Brain Injury. Archives of Physical Medicine and Rehabilitation , 101 (11), 1973–1979. https://doi.org/10.1016/j.apmr.2020.05.026 Scholten, A. C., Haagsma, J. A., Cnossen, M. C., Olff, M., van Beeck, E. F., & Polinder, S. (2016). Prevalence of and Risk Factors for Anxiety and Depressive Disorders after Traumatic Brain Injury: A Systematic Review. Journal of Neurotrauma , 33 (22), 1969–1994. https://doi.org/10.1089/neu.2015.4252 Shen, J., & Wang, Y. (2023). Correlates of Longitudinal Trajectories of Depressive Symptoms in Adolescents With Traumatic Brain Injuries. Journal of Pediatric Psychology , 48 (12), 1021–1029. https://doi.org/10.1093/jpepsy/jsad073 Singh, R. K., Humphries, T. J., Dawson, J. F., Tiupin-Szulc, J., Mason, S., & Lecky, F. E. (n.d.). Changes in depression symptoms over 10 years after TBI; a long-term prospective study. Brain Injury , 0 (0), 1–8. https://doi.org/10.1080/02699052.2024.2413639 Singh, R., Mason, S., Lecky, F., & Dawson, J. (2018). Prevalence of depression after TBI in a prospective cohort: The SHEFBIT study. Brain Injury , 32 (1), 84–90. https://doi.org/10.1080/02699052.2017.1376756 Stein, M. B., Jain, S., Giacino, J. T., Levin, H., Dikmen, S., Nelson, L. D., Vassar, M. J., Okonkwo, D. O., Diaz-Arrastia, R., Robertson, C. S., Mukherjee, P., McCrea, M., Mac Donald, C. L., Yue, J. K., Yuh, E., Sun, X., Campbell-Sills, L., Temkin, N., Manley, G. T., … Zafonte, R. (2019). Risk of Posttraumatic Stress Disorder and Major Depression in Civilian Patients After Mild Traumatic Brain Injury: A TRACK-TBI Study. JAMA Psychiatry , 76 (3), 249–258. https://doi.org/10.1001/jamapsychiatry.2018.4288 Tso, S., Saha, A., & Cusimano, M. D. (2021). The Traumatic Brain Injury Model Systems National Database: A Review of Published Research. Neurotrauma Reports , 2 (1), 149–164. https://doi.org/10.1089/neur.2020.0047 Turkoz, I., Alphs, L., Singh, J., Jamieson, C., Daly, E., Shawi, M., Sheehan, J. J., Trivedi, M. H., & Rush, A. J. (2021). Clinically meaningful changes on depressive symptom measures and patient-reported outcomes in patients with treatment-resistant depression. Acta Psychiatrica Scandinavica , 143 (3), 253–263. https://doi.org/10.1111/acps.13260 Versluijs, Y., van Ravens, T. W., Krijnen, P., Ring, D., & Schipper, I. B. (2022). Systematic Review of the Association Between Trauma Severity and Postinjury Symptoms of Depression. World Journal of Surgery , 46 (12), 2900–2909. https://doi.org/10.1007/s00268-022-06750-3 Welham, S., Cullis, B., Gogel, B., Gilmour, A., & Thompson, R. (2004). Prediction in linear mixed models. Australian & New Zealand Journal of Statistics , 46 (3), 325–347. https://doi.org/10.1111/j.1467-842X.2004.00334.x Weuve, J., Tchetgen Tchetgen, E. J., Glymour, M. M., Beck, T. L., Aggarwal, N. T., Wilson, R. S., Evans, D. A., & Mendes de Leon, C. F. (2012). Accounting for bias due to selective attrition: The example of smoking and cognitive decline. Epidemiology (Cambridge, Mass.) , 23 (1), 119–128. https://doi.org/10.1097/EDE.0b013e318230e861 Yu, B., Zhang, X., Wang, C., Sun, M., Jin, L., & Liu, X. (2020). Trends in depression among Adults in the United States, NHANES 2005–2016. Journal of Affective Disorders , 263 , 609–620. https://doi.org/10.1016/j.jad.2019.11.036 Additional Declarations No competing interests reported. Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-6463585","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":456109436,"identity":"ecdbbefa-1abf-456f-83c4-86da18e8c5e3","order_by":0,"name":"Nelofar Kureshi","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA+UlEQVRIie3OvWrDMBDA8RMGZ0kyXzC4r+DiIc3bSBTsyVOXDqFcCGhyPCdv0TGjRCFeDFkztdbSuV1CvISqhUI7WKFbB/0XfcBPJwCf718WENzbBT/3jDAGZW/cMYLmB0lBsb8REHSJTAeLRavmzzBZF8Z025u8qjXBad5PZqVeJmp3BxHmaTpqsNg0gli56yfJQcjoLeQQYxZGTGLxqAQFEDrIi5Gozl9k0HUS82RvLDm7ptiXteT2Y1kII4nczqWASQdpxDLRFR9OytcgsuR6czCkV5WD1E+6VUceY52x904+XI33t7o9HfvJd8NfJ3UZ+Hw+n8/VB2rFVzWoGxoFAAAAAElFTkSuQmCC","orcid":"","institution":"Dalhousie University","correspondingAuthor":true,"prefix":"","firstName":"Nelofar","middleName":"","lastName":"Kureshi","suffix":""},{"id":456109437,"identity":"8dee97fa-5f84-4660-a63f-9fbe079737f2","order_by":1,"name":"David B. Clarke","email":"","orcid":"","institution":"Dalhousie University","correspondingAuthor":false,"prefix":"","firstName":"David","middleName":"B.","lastName":"Clarke","suffix":""},{"id":456109438,"identity":"53f2afda-6674-4063-9e73-c1e30f2b074e","order_by":2,"name":"Abraham Nunes","email":"","orcid":"","institution":"Dalhousie University","correspondingAuthor":false,"prefix":"","firstName":"Abraham","middleName":"","lastName":"Nunes","suffix":""},{"id":456109439,"identity":"17d04387-21e0-4f41-92cc-6c99e9f68967","order_by":3,"name":"Cindy Feng","email":"","orcid":"","institution":"Dalhousie University","correspondingAuthor":false,"prefix":"","firstName":"Cindy","middleName":"","lastName":"Feng","suffix":""},{"id":456109440,"identity":"aa5d57fd-b3da-4f84-8bb9-3c7623ebcd6e","order_by":4,"name":"Syed Sibte Raza Abidi","email":"","orcid":"","institution":"Dalhousie University","correspondingAuthor":false,"prefix":"","firstName":"Syed","middleName":"Sibte Raza","lastName":"Abidi","suffix":""}],"badges":[],"createdAt":"2025-04-16 12:38:19","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6463585/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6463585/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":82887039,"identity":"20c87170-055f-40e7-b1e9-ce9299016889","added_by":"auto","created_at":"2025-05-16 11:55:16","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":77154,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eStudy methodology\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"Figure1studyflow.png","url":"https://assets-eu.researchsquare.com/files/rs-6463585/v1/929e2b10b1abd811d8ce2b17.png"},{"id":82885549,"identity":"013b9497-1832-4f8b-9d53-f1a6e35f1c04","added_by":"auto","created_at":"2025-05-16 11:47:16","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":14366,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eInteraction probe for pre-injury mental health treatment and time\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"Figure2MHtimeinteraction.png","url":"https://assets-eu.researchsquare.com/files/rs-6463585/v1/fd7167d90a1aef9695167ff0.png"},{"id":87932359,"identity":"b752123a-3c67-4e4b-b105-fa16b760d17b","added_by":"auto","created_at":"2025-07-30 13:53:19","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1393865,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6463585/v1/f9434954-1538-4469-9afe-6861d00031c7.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Longitudinal Trends of Depression in Traumatic Brain Injury: The Role of Individual Heterogeneity in Clinical Prediction","fulltext":[{"header":"Introduction","content":"\u003cp\u003eDepression is a common consequence of traumatic brain injury (TBI), affecting approximately 30% of individuals with TBI, particularly during the first year post-injury (Fakhoury et al., 2021). The etiology of post-TBI depression is complex, involving pre-injury factors, injury characteristics, and post-injury experiences. Despite its prevalence, the long-term trends and determinants of depression trends following TBI remain insufficiently understood.\u003c/p\u003e\n\u003cp\u003eThe global burden of depression has increased from 1990 to 2019, with higher incidence and disability in females and older age groups, and variation across socioeconomic development levels (Liu et al., 2024). This report indicates that the rates of the two subtypes of depression, dysthymia and major depressive disorders, have remained largely stable globally and regionally over the study period, with the majority of patients suffering from major depressive disorder (Liu et al., 2024). Several studies indicate an overall increasing trend in depression prevalence over time within the general population. A systematic review and meta-analysis found a predominant increase in the likelihood of experiencing depression, with a pooled odds ratio of 1.35, suggesting a significant rise in depression prevalence over\u0026nbsp;time (Moreno-Agostino et al., 2021). In the United States, data from 2005 to 2016 revealed an increase in severe depression, especially among adults aged 65 and older, and moderate depression among those aged\u0026nbsp;20-39 (Yu et al., 2020).\u003c/p\u003e\n\u003cp\u003eThe majority of research on the impact of TBI on depression relies on cross-sectional data collected during the early post-injury period. Across various studies, the prevalence of depressive disorders post-TBI ranges widely. In a systematic review, the prevalence of depression was found to be 17% in the first year after TBI, increasing to 43% over the long\u0026nbsp;term (Scholten et al., 2016). Longitudinal studies provide additional insights, showing variability in depression rates over time. For instance, Bombardier et al. reported that 29% of TBI patients experienced a major depressive episode in the first year post-TBI, with relatively stable rates during follow-up, while minor depressive episodes decreased over the same period (Bombardier et al., 2016). Other studies indicate that depression post-TBI is dynamic, with some individuals developing new depressive symptoms while others experience recovery. Approximately 26% of individuals not depressed at one year developed depression by the second year, and nearly one-third of those with minor depression progressed to major depressive disorder (Hart et al., 2012).\u0026nbsp;The elevated risk of depression persists beyond the first year, with studies reporting prevalence rates of 42% at 2.5 years post-injury\u0026nbsp;(Jorge et al., 2004). The three month prevalence of depression is reported to be \u0026nbsp;56%\u0026nbsp;(R. Singh et al., 2018), 42% at one year and 38 % at 10 years\u0026nbsp;(R. K. Singh et al., n.d.).\u003c/p\u003e\n\u003cp\u003eThese findings are largely based on single-institution studies, which limit generalizability to diverse TBI populations and geographic regions. Additional limitations include small sample sizes and relatively short follow-up periods. While prior research, using various methods from finite mixture modeling, have provided insights into the diverse patterns of depression following TBI, these approaches may not accurately reflect the individual journeys of recovery (McInnes et al., 2024). Individuals with TBI are a remarkably heterogeneous population. Some individuals demonstrate resilience and experience significant improvements in their depressive symptoms over time, while others struggle with persistent or worsening symptoms, despite similar demographic and injury characteristics. However, population-level estimates of depression prevalence and overall trends are still useful as a starting point for understanding the broader impact of TBI.\u0026nbsp;The challenge lies in capturing both the general patterns and the individual variations around those patterns. Linear mixed-effects models offer a powerful approach to address this challenge by simultaneously estimating fixed effects, representing population-level trends, and random effects, which capture individual deviations from those trends. By modeling both the average trajectory and the individual departures from it, we can gain an understanding of the complex and heterogeneous nature of depression following TBI. Importantly, while linear mixed models can incorporate random effects for personalized predictions, health research frequently isolates fixed effects when deriving predictions applicable to broader populations (Brown, 2021; Welham et al., 2004). These population-level prediction may be appropriate where generalizable conclusions outweigh individual heterogeneity, however this is not the case in TBI population where the concept of an 'average' TBI patient is inherently problematic (Anderson, 2025; Covington \u0026amp; Duff, 2021; Rabinowitz et al., 2020). For a more accurate understanding of long-term depression following TBI, predictions derived from linear mixed models should incorporate both population-level fixed effects and individual-level random effects. This nuanced understanding is crucial for identifying individuals at high risk for depression, and for tailoring interventions to meet their specific needs.\u003c/p\u003e\n\u003cp\u003eThe current investigation addresses critical knowledge gaps in understanding long-term depression trajectories in TBI patients. The study objectives were to: (1) characterize the longitudinal course of depression over 10 years post-TBI; (2) identify sociodemographic and clinical characteristics associated with depression trajectories; (3) compare the model's predictive performance using population-level information alone (representing average trends) against predictions that incorporate both population-level and subject-specific information; and \u0026nbsp;(4) evaluate the potential utility of the model for predicting depression trajectories in a clinical setting. We hypothesized that population-level predictions alone would demonstrate poorer performance compared to models incorporating both population and subject-level effects, and that the magnitude of this difference would be substantial, highlighting the critical importance of accounting for individual heterogeneity in predicting depression trajectories following TBI.\u0026nbsp;\u003c/p\u003e"},{"header":"Methods","content":"\u003cp\u003eThe study methodology is illustrated in Figure 1.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eData source:\u003c/em\u003e\u003c/strong\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003eThis retrospective analysis utilized data from the Traumatic Brain Injury Model Systems (TBIMS) National Database, a prospective, multicenter longitudinal database tracking outcomes from injury through 30 years post-injury(Tso et al., 2021). The TBIMS systematically collects data at standardized intervals: initial injury, rehabilitation discharge, and follow-up assessments at 1, 2, 5, and every 5 years thereafter. This study analyzed the publicly available dataset.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eSample size:\u003c/em\u003e\u003c/strong\u003e Sample size calculations were performed to ensure sufficient power for a linear mixed-effects model. We assumed a small effect size (f\u0026sup2; = 0.05), power of 0.8, and a significance level of 0.05. Given the multilevel design, we accounted for an intraclass correlation coefficient (ICC) of 0.4 and an average of 4 measurements per subject. Based on these parameters, the required sample size to detect statistically significant effects was 12,560 subjects.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eStudy population\u003c/em\u003e\u003c/strong\u003e\u003cem\u003e:\u003c/em\u003e Participants were individuals with traumatic brain injury enrolled in the TBIMS database. Eligible participants met at least one of the following clinical criteria:\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e- Glasgow Coma Scale (GCS) score below 13 in the emergency department\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e- Post-traumatic amnesia (PTA) exceeding 24 hours\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e- Presence of intracranial neuroimaging abnormalities\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e- Loss of consciousness (LOC) greater than 30 minutes\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eAdditional inclusion criteria specified age \u0026ge;16 years and receipt of inpatient rehabilitation at one of 16 participating TBIMS centers.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eVariables and measurement:\u0026nbsp;\u003c/em\u003e\u003c/strong\u003eThe analysis included predictor variables selected based on established associations with post-TBI depression outcomes. Demographic characteristics considered were age (continuous), sex (male/female), marital status (single, married, or divorced/separated/widowed), education level (less than high school, high school graduate, or college degree), and employment status at injury (employed, retired, student, or unemployed). Injury-related characteristics encompassed injury mechanism (fall, motor vehicle collision, violence, or other), injury severity indicators using the Glasgow Coma Scale (GCS), Functional Independence Measure (FIM) score (continuous), and length of stay (continuous). Pre-injury clinical factors included documented history of mental health treatment, substance use disorders, psychiatric hospitalizations, and suicide attempts.\u003c/p\u003e\n\u003cp\u003eThe outcome of interest was depression as measured through the\u0026nbsp;Patient Health Questionnaire-9 (PHQ-9). The PHQ-9 is a 9-item measure of depression based on diagnostic criteria (Kroenke et al., 2001). Respondents rate the frequency and severity of depressive symptoms over the last two weeks. Each item is scored using a Likert-type scale from 0\u0026ndash;4 where 0 indicates \u0026ldquo;not at all\u0026rdquo;, 1 is \u0026ldquo;several days\u0026rdquo;, 2 is \u0026ldquo;more than half the days\u0026rdquo;, and 3 is \u0026ldquo;nearly every day\u0026rdquo;. Total scores range from 0 to 27, with higher scores indicating higher levels of depression symptomatology. The PHQ-9 has a sensitivity of 88% and a specificity of 88% in the diagnosis of major depression when the cut-off score of \u0026ge;10 is used and has been validated in the TBI population (Fann et al., 2005). \u0026nbsp;The PHQ-9 was not administered at discharge from acute or rehabilitation care; it was available for follow-up time points.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eInverse probability weighting:\u003c/em\u003e\u003c/strong\u003e Estimates of baseline covariates on PHQ-9 scores may be underestimated due to differential attrition in the TBIMS over the study period. The analyses included weights for the inverse probability of attrition to account for this attrition (Weuve et al., 2012). Specifically, we estimated follow-up time-specific weights based on the inverse probability of being observed at each time point. These weights were then applied to each observation to minimize potential biases in estimates of baseline covariates on PHQ-9 scores.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eData Analysis:\u003c/em\u003e\u003c/strong\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003eBaseline characteristics were compared using t-tests for continuous variables and chi-square tests for categorical variables. Discrete variables are reported as counts (percentages), and continuous variables as means (standard deviations).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eLongitudinal Modeling:\u003c/em\u003e\u003c/strong\u003e We employed linear mixed-effects models to analyze longitudinal PHQ-9 trajectories. The model specification was:\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cem\u003ePHQ9_ij = \u0026beta;₀ + \u0026beta;₁(Time_ij) + \u0026beta;₂X_i + \u0026beta;₃(Time_ij \u0026times; X_i) + b₀i + b₁i(Time_ij) + \u0026epsilon;_ij\u0026nbsp;\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eWhere:\u0026nbsp;\u003c/p\u003e\n\u003cp\u003ei denotes individual\u0026nbsp;\u003c/p\u003e\n\u003cp\u003ej denotes measurement occasion\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u0026beta; represents fixed effects\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eb represents random effects \u003c/p\u003e\n\u003cp\u003eX represents covariates\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u0026epsilon; represents residual error\u003c/p\u003e\n\u003cp\u003eThe longitudinal trend in depression over time (measured at 1, 2, 5, and 10 years) was assessed using linear mixed models to account for the correlation between repeated measurements within the same individual and to include all available data. A random intercept was included for each subject to account for individual differences in depression levels at baseline. A random slope for the effect of time was also included to capture individual variations in the rate of change in depression over time. Only significant interactions are reported.\u003c/p\u003e\n\u003cp\u003eAs PHQ-9 scores were first measured at the Year 1 visit, the time variable was centered at Year 1. This allowed the main effects in the model to reflect covariate effects on PHQ scores from baseline to Year 1, and the covariate-by-time interactions to represent the effects of covariates on the rate of depression in subsequent years.\u003c/p\u003e\n\u003cp\u003eModel diagnostics included visual inspection of standardized residual and quantile-quantile (Q-Q) plots, which indicated approximately normally distributed residuals with some deviations at the extremes. Missing data for the outcome variable were not imputed, as a likelihood-based approach was used to handle missing data under the assumption that it was missing at random. Only covariate data were imputed, and the results presented are based on a pooled analysis of 10 imputed datasets.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eMinimal Clinically Important Difference\u003c/em\u003e\u003c/strong\u003e\u003cem\u003e:\u003c/em\u003e We evaluated the effect of time on PHQ scores using two established methods to classify participants\u0026apos; changes based on the proportion of individuals achieving a minimal clinically important difference (MCID). The MCID represents the smallest change in PHQ scores perceived as clinically meaningful. The widely accepted MCID for the PHQ-9 is a 20% change in score or 5 points and is commonly used to assess intervention effects in depression trials (Carlo et al., 2021; L\u0026ouml;we et al., 2004). It is important to note that the dataset does not contain information on whether participants were receiving active treatment or interventions for depression during the follow-up period. Therefore, given the observational nature of the current study, we also adopted a more conservative estimate of MCID based on the standard error of measurement (SEM) as a criterion for clinically relevant change (Łakuta et al., 2022; Turkoz et al., 2021). To account for intra-individual variability and establish a 95% confidence interval, the SEM for the PHQ-9 was multiplied by 1.96. Using this approach, the conservative MCID for the PHQ-9 was determined to be 4 points.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eModel assessment:\u003c/em\u003e\u003c/strong\u003e We distinguish between fitted values (predictions on observed data used to build the model) and predictions (on unseen data). We examined model performance for fitted values at two levels:\u003c/p\u003e\n\u003cp\u003eLevel 0 (Population-level): Predictions based solely on fixed effects, representing population or average trajectories.\u003c/p\u003e\n\u003cp\u003eLevel 1 (Subject-specific): Predictions conditional on the Best Linear Unbiased Predictions (BLUPs) of the random effects, capturing individual deviations from the population average.\u003c/p\u003e\n\u003cp\u003eFor each level, we assessed the residual distributions and prediction metrics (mean squared error [MSE], coverage, precision). MSE was calculated as the average of the squared differences between the observed PHQ-9 scores and the predicted PHQ-9 scores. A lower MSE indicates better overall prediction accuracy, with smaller average deviations between the predictions and the actual observed values.\u0026nbsp;Coverage was defined as the proportion of realized PHQ-9 observations within the 50% prediction interval. Precision was the average width of the 50% prediction interval for predicted PHQ-9 values.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003ePredictive performance assessment:\u003c/em\u003e\u003c/strong\u003e To provide a proof of concept demonstrating the potential clinical utility of the developed linear mixed-effects model for predicting individual depression trajectories, we designed two scenarios designed to simulate real-world clinical applications: (1) prediction of depression trajectory for new patients and (2) forecasting the depression trajectory for existing patients using initial PHQ-9 scores. It is important to note that these scenarios are intended to illustrate the feasibility of using the model for individualized prediction and to generate hypotheses for future research, rather than to provide definitive evidence of its clinical effectiveness.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eScenario 1: Prediction of future depression scores using early data\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis scenario aimed to evaluate the model\u0026apos;s ability to predict future depression severity based on early longitudinal data. The model was trained on data from Years 1 and 2 to capture initial depression trajectories. This trained model was then used to predict individual depression scores at Year 5, using observed PHQ-9 scores from Years 1 and 2, along with baseline demographic and injury-related characteristics.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eScenario 2: Prediction of depression trajectory for new patients\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis scenario aimed to evaluate the model\u0026apos;s ability to generalize to individuals not included in the model fitting process. We created a test set of participants by excluding a random subset of individuals from the original dataset. The model was then trained on the remaining participants and used to predict individual depression trajectories (PHQ-9 scores at Year 2) for the individuals in the test set, using only their baseline demographic and injury-related characteristics.\u003c/p\u003e\n\u003cp\u003eFor both scenarios, we compared two types of predictions: marginal predictions based on fixed effects only (population-level estimates) and conditional predictions that incorporated both fixed and random effects (subject-specific estimates). This comparison allowed us to assess the added value of including subject-level variation in a prediction setting. The predictive performance of models from both scenarios was quantified using metrics of MSE, coverage, and precision.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eSensitivity analyses\u003c/em\u003e\u003c/strong\u003e: To ensure the robustness of our findings, we conducted a sensitivity analyses. we excluded participants whose injury was caused by gunshot wounds to evaluate whether these cases influenced the overall results.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe following R packages were used for data analysis and visualization: \u003cem\u003enlme\u003c/em\u003e for fitting linear mixed-effects models, \u003cem\u003eJmbayes\u003c/em\u003e for model prediction in longitudinal data \u003cem\u003eemmeans\u003c/em\u003e for post-hoc pairwise comparisons of estimated marginal means, and \u003cem\u003emice\u003c/em\u003e for multiple imputation of missing data.\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003e \u003cstrong\u003eSample characteristics\u003c/strong\u003e \u003cp\u003eTable\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e presents the demographic, injury-related, and pre-injury clinical characteristics of the overall cohort (n\u0026thinsp;=\u0026thinsp;19,397). The mean age was 43 years. The majority of participants were male (74%) and identified as White (66%), followed by Black (18%) and Hispanic (11%). Regarding marital status, 46% were single, 33% were married, and 21% were separated, widowed, or divorced. In terms of educational attainment, 40% had completed college, 36% had completed high school, and 24% had less than a high school education. At the time of injury, 61% were competitively employed, 18% were retired, and 12% were unemployed.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eBaseline characteristics of the study population\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"2\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCharacteristic\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eN\u0026thinsp;=\u0026thinsp;19,397\u003csup\u003e\u003cem\u003e1\u003c/em\u003e\u003c/sup\u003e\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAge\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e43 (20)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSex\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFemale\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e5,107 (26%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMale\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e14,278 (74%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRace\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAsian/Pacific Islander\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e539 (2.8%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBlack\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3,548 (18%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHispanic Origin\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2,181 (11%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNative American\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e104 (0.5%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eOther\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e237 (1.2%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eWhite\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e12,745 (66%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMarital status\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMarried\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e6,440 (33%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSep_Wid_Div\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e4,081 (21%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSingle\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e8,828 (46%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eIncome\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e=\u0026gt;100,000\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e687 (5.6%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e20\u0026ndash;49,999\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3,671 (30%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e50\u0026ndash;99,999\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1,872 (15%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNot Competitively Employed\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e5,973 (49%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eEducation\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ecollege\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e7,654 (40%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHS\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e6,789 (36%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eless than HS\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e4,637 (24%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eEmployment\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCompetitively employed\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e11,694 (61%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eOther\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e648 (3.4%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRetired\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3,420 (18%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eStudent\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1,172 (6.1%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eUnemployed\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2,327 (12%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCause\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFall\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e5,588 (29%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMVC\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e8,905 (46%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eOther\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2,758 (14%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eViolence\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2,146 (11%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGCS\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eChemically Paralyzed or Sedated\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e4,469 (23%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMild\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e6,342 (33%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eModerate\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2,271 (12%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSevere\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e6,259 (32%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAcute Payor\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMedicare\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3,012 (16%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eOther\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e255 (1.3%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePrivate\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e10,874 (57%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePublic\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e5,001 (26%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRehab Payor\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMedicare\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3,061 (16%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eOther\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e228 (1.2%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePrivate\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e10,637 (55%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePublic\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e5,387 (28%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDrugs\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3,891 (21%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDrink Category\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAbstaining\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e6,813 (41%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHeavy\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2,538 (15%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLight\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3,276 (20%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eModerate\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e4,075 (24%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eUnknown\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2,695\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMental health treatment\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2,575 (22%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePsychiatric hospitalization\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e834 (7.1%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSuicide\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e610 (5.2%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFIM Motor\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e66 (19)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFIM Cognitive\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e24 (7)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLOS Acute care\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e21 (18)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLOS Rehab care\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e27 (26)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePre-injury TBI\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.39 (0.89)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e \u003cp\u003e\u003csup\u003e\u003cem\u003e1\u003c/em\u003e\u003c/sup\u003e\u0026nbsp;Mean (SD); n (%)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/p\u003e \u003cp\u003eMotor vehicle collisions (MVC) were the most common cause of injury (46%), followed by falls (29%), and violence (11%). Based on the Glasgow Coma Scale (GCS) distribution, 32% experienced a severe TBI, 12% had a moderate TBI, and 33% had a mild TBI. For the remaining 23% of participants, GCS scores were unavailable due to chemical paralysis or sedation at the time of assessment.\u003c/p\u003e \u003cp\u003eRegarding pre-injury clinical factors, 21% reported drug use, and alcohol consumption was categorized as light (20%), moderate (24%), and heavy (15%). Additionally, 22% reported pre-injury mental health treatment, 7% had been previously hospitalized for psychiatric causes, and 5% had a history of suicide attempts.\u003c/p\u003e \u003cp\u003e \u003cb\u003eFactors associated with depression\u003c/b\u003e: Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e presents the results of the linear mixed-effects model assessing various factors related to changes in the PHQ-9 total score over time. Several baseline covariates were significantly associated with PHQ-9 scores at Year 1. Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e shows the changes in PHQ-9 scores and effect sizes for the model covariates. The largest main effect and effect size was for pre-injury mental health treatment (β\u0026thinsp;=\u0026thinsp;1.6, 95% CI: 1.21\u0026ndash;1.97). A higher number of head injuries prior to the index TBI (β\u0026thinsp;=\u0026thinsp;0.4, 95% CI: 0.31\u0026ndash;0.52) and being female (β\u0026thinsp;=\u0026thinsp;0.86, 95% CI: 0.63\u0026ndash;1.09) were also associated with higher PHQ-9 scores.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eLinear mixed effect model of PHQ-9 scores\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"4\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCharacteristic\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eBeta\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003e95% CI\u003csup\u003e\u003cem\u003e1\u003c/em\u003e\u003c/sup\u003e\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003ep-value\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFixed effects\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eIntercept\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e5.511\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e4.06\u0026ndash;6.96\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.000\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTime\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e-0.006\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e-0.03-0.02\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.662\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMental health treatment\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1.592\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1.21\u0026ndash;1.97\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.000\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAge\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e-0.024\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e-0.03-0.01\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.000\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFIM Motor\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.006\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e-0.00-0.01\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.121\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFIM Cognitive\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.006\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e-0.01-0.02\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.549\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLOS Acute care\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.004\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e-0.00-0.01\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.283\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLOS Rehab care\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e-0.004\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e-0.01-0.00\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.132\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePre-injury TBI\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.418\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.31\u0026ndash;0.52\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.000\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSex\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMale\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFemale\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.863\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.63\u0026ndash;1.09\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.000\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRace\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eWhite\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAsian/Pacific Islander\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e-0.092\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e-0.71-0.53\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.771\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBlack\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.702\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.42\u0026ndash;0.99\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.000\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHispanic Origin\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.564\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.21\u0026ndash;0.92\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.002\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNative American\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1.449\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.19\u0026ndash;2.71\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.024\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eOther\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.648\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e-0.29-1.59\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.177\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMarital status\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMarried\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSep_Wid_Div\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.558\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.27\u0026ndash;0.85\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.000\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSingle\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e-0.952\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e-1.24-0.67\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.000\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eEmployment\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eOther\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCompetitively employed\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e-0.502\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e-1.10-0.09\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.099\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRetired\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e-0.237\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e-0.89-0.42\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.479\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eStudent\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e-0.975\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e-1.68-0.27\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.007\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eUnemployed\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.064\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e-0.59-0.72\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.847\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCause\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eOther\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFall\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e-0.058\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e-0.39-0.28\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.733\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMVC\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.155\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e-0.14-0.45\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.300\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eViolence\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.618\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.20\u0026ndash;1.04\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.004\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAcute Payor\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eOther\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMedicare\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e-0.507\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e-2.04-1.03\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.518\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePrivate\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e-0.485\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e-1.71-0.74\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.437\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePublic\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.191\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e-1.05-1.44\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.763\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRehab Payor\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eOther\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMedicare\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e-0.364\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e-1.94-1.21\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.650\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePrivate\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e-0.172\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e-1.43-1.09\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.789\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePublic\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.140\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e-1.13-1.41\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.830\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDrugs\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.837\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.57\u0026ndash;1.10\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.000\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePsychiatric hospitalization\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.188\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e-0.40-0.78\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.526\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSuicide\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eYes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.852\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.10\u0026ndash;1.60\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.027\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDrinking Category\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAbstaining\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHeavy\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.194\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e-0.13-0.52\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.242\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLight\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.164\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e-0.11-0.44\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.246\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eModerate\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e-0.332\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e-0.60-0.07\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.015\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eEducation\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ecollege\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHS\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.536\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.30\u0026ndash;0.76\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.000\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eless than HS\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1.066\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.78\u0026ndash;1.35\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.0000.000\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGCS\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMild\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eChemically Paralyzed or Sedated\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e-0.054\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e-0.34-0.24\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.715\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eModerate\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.094\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e-0.26-0.45\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.602\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSevere\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e-0.185\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e-0.47-0.10\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.204\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTime * Mental health treatment\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e-0.143\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e-0.21- -0.07\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.000\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eRandom effects\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRandom intercept\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e18.50\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRandom slope\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.13\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCovariance between intercept and slope\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e-0.49\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eResidual variance\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e13.83\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eICC\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.57\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"4\" nameend=\"c4\" namest=\"c1\"\u003e \u003cp\u003e\u003csup\u003e\u003cem\u003e1\u003c/em\u003e\u003c/sup\u003e\u0026nbsp;CI\u0026thinsp;=\u0026thinsp;Confidence Interval\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eRacial differences were significant, with Black (β\u0026thinsp;=\u0026thinsp;0.70, 95% CI: 0.42\u0026ndash;0.99) and Hispanic (β\u0026thinsp;=\u0026thinsp;0.56, 95% CI: 0.21\u0026ndash;0.92) patients having higher scores compared to White individuals. Patients who were single (β = -0.95, 95% CI: -1.24- -0.67) had lower depression scores compared to those who were married, whereas patients who were separated/widowed/divorced had higher scores (β\u0026thinsp;=\u0026thinsp;0.56, 95% CI: 0.27\u0026ndash;0.85). Those with less than a high school education (β\u0026thinsp;=\u0026thinsp;1.1, 95% CI: 0.78\u0026ndash;1.35) and high school education (β\u0026thinsp;=\u0026thinsp;0.54, 95% CI: 0.30\u0026ndash;0.76) had higher depression scores compared to college graduates.\u003c/p\u003e \u003cp\u003ePatients who were injured by violent causes had higher depression scores (β\u0026thinsp;=\u0026thinsp;0.62, 95% CI: 0.20\u0026ndash;1.04) compared to those injured by other causes. Other significant factors included substance use, with individuals who reported pre-injury drug use having higher PHQ-9 scores (β\u0026thinsp;=\u0026thinsp;0.84, 95% CI: 0.57\u0026ndash;1.10). Those with a history of suicide attempts had significantly higher scores (β\u0026thinsp;=\u0026thinsp;0.85, 95% CI: 0.10\u0026ndash;1.60).\u003c/p\u003e \u003cp\u003e \u003cb\u003eLongitudinal trend of depression\u003c/b\u003e: There was a significant interaction between follow-up time and pre-injury treatment of mental health conditions indicating that the effect of time on depression scores differed by the presence of psychiatric history. Specifically, for individuals with a history of pre-injury mental health treatment, depression scores demonstrated a small decrease over time (β =-0.14, 95% CI: -0.21- -0.07).\u003c/p\u003e \u003cp\u003eThe interaction was analyzed by decomposing the effects to explore differences in PHQ scores based on pre-injury mental health treatment status, while accounting for the influence of other covariates by averaging over their levels in the model. Figure\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e illustrates a decline in PHQ scores over ten years among patients who received mental health treatment prior to injury, whereas those without prior treatment exhibited stable scores over the same period. Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e presents the marginal means for PHQ scores stratified by pre-injury mental health treatment status. Although the decline in PHQ scores among the treated group was statistically significant across the study period, this decrease amounted to 1.3 points from Year 1 to Year 10.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eMarginal means, standard errors, and 95% confidence intervals for PHQ-9 scores by pre-injury mental health treatment status\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"7\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFollow up year\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"3\" nameend=\"c4\" namest=\"c2\"\u003e \u003cp\u003eMental health treatment\u0026thinsp;=\u0026thinsp;No\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"3\" nameend=\"c7\" namest=\"c5\"\u003e \u003cp\u003eMental health treatment\u0026thinsp;=\u0026thinsp;Yes\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMeans\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eSE\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003e95%CI\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eMeans\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eSE\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003e95%CI\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e7.13\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.27\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e6.59\u0026ndash;7.67\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e8.72\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.27\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e8.19\u0026ndash;9.26\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e7.13\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.27\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e6.59\u0026ndash;7.66\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e8.57\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.26\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e8.06\u0026ndash;9.09\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e7.11\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.27\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e6.57\u0026ndash;7.64\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e8.13\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.26\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e7.62\u0026ndash;8.63\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e10\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e7.08\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.29\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e6.51\u0026ndash;7.64\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e7.38\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.32\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e6.76\u0026ndash;8.01\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003cstrong\u003eHeterogeneity in baseline depression and trends\u003c/strong\u003e \u003cp\u003eIn addition to the fixed effects, the model included random effects, indicating heterogeneity in Year 1 depression scores and their trends across ten years. The variance of the random intercept (subject-level variation in PHQ-9 scores) was estimated at 18.50 (SD\u0026thinsp;=\u0026thinsp;4.30), reflecting substantial individual variability in baseline PHQ-9 scores. This suggests that some individuals had much higher or lower initial depression scores than the average participant. The variance of the random slope for time (variation in the rate of change in PHQ-9 scores across individuals) was 0.13 (SD\u0026thinsp;=\u0026thinsp;0.36), indicating variability in the rate of change of PHQ-9 scores over time across individuals. In other words, some participants' depression scores improved more rapidly than others, while some showed little to no change, or even worsened over time. The intraclass correlation coefficient (ICC) was 0.57, suggesting that 57% of the total variance in PHQ-9 scores was attributable to between-subject differences.\u003c/p\u003e \u003c/p\u003e \u003cp\u003e \u003cb\u003eModel performance\u003c/b\u003e: Model performance was assessed at two levels: Level 0 (population-level), representing predictions based solely on fixed effects and reflecting population-average trajectories, and Level 1 (subject-specific), incorporating both fixed and random effects to capture individual deviations from the population average. Model performance was evaluated using MSE, coverage, and precision.\u003c/p\u003e \u003cp\u003eFor Level 0, the MSE was 31.17 and the coverage was 23%; i.e., a small proportion of observed PHQ-9 scores fell within the 50% prediction intervals. The model precision at Level 0 was 3.76 points. Visual inspection of the residual plot for population-level predictions revealed a tendency for the model to under-predict higher PHQ-9 scores and suggested potential heteroscedasticity.\u003c/p\u003e \u003cp\u003eLevel 1 fitted values demonstrated a substantially lower MSE of 7.75, indicating a considerably smaller average squared difference between the observed and predicted individual depression scores compared to the Level 0. For Level 1, the coverage improved to 58% within the prediction intervals and model precision was 3.75 points. Visual inspection of the residual plot for Level 1 (BLUP) predictions suggested some remaining heteroscedasticity, though less pronounced than at Level 0. These similar precision values suggest that while both approaches had comparable interval widths, Level 1 fitted values achieved this precision with a greater accuracy in predicting individual outcomes, as evidenced by the lower MSE and improved coverage within the 50% prediction intervals.\u003c/p\u003e \u003cp\u003e \u003cstrong\u003ePredictive performance assessment\u003c/strong\u003e \u003cp\u003eTo evaluate the potential for clinical application, we assessed the model's ability to predict future depression scores (Year 5) using early longitudinal data (Years 1 and 2). This within-sample prediction scenario reflects a setting where repeated measurements are available for the same individuals (Table\u0026nbsp;\u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e4\u003c/span\u003e, Scenario 1). The conditional prediction using random effects yielded an MSE of 0.014, and the coverage within the prediction intervals was 100%. The precision, as measured by the average width of the 50% prediction intervals, was 3.7 points. This indicated excellent subject-specific predictive accuracy when sufficient longitudinal information was available for the individual.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab4\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 4\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003ePrediction metrics for clinical scenarios\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"7\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003ePrediction level\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"3\" nameend=\"c4\" namest=\"c2\"\u003e \u003cp\u003eScenario 1\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"3\" nameend=\"c7\" namest=\"c5\"\u003e \u003cp\u003eScenario 2\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMSE\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eCoverage\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003ePrecision\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eMSE\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eCoverage\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003ePrecision\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMarginal prediction\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e30.48\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e18\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e2.5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e29.07\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e33\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e5.1\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eConditional prediction\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.014\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e100\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3.7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e19.57\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e64\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e6.0\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/p\u003e \u003cp\u003eIn the second scenario, we assessed the model\u0026rsquo;s generalizability to new patients by evaluating its ability to predict depression scores in a hold-out test set (Table\u0026nbsp;\u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e4\u003c/span\u003e, Scenario 2). The model was trained on a subset of participants and applied to new individuals, simulating prediction in an unobserved population using only their Year 1 scores to forecast Year 2 outcomes. In this out-of-sample setting, the conditional prediction achieved an MSE of 19.57. The coverage within the 50% prediction intervals was 64%, and the average interval width (precision) was 6 points. Conditional predictions outperformed marginal predictions across both scenarios, with lower MSE and higher interval coverage.\u003c/p\u003e \u003cp\u003eIn the sensitivity analyses, excluding participants with gunshot wounds, yielded results consistent with the main model; the beta coefficients for key predictors differed by less than 5%.\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eIn the current study, we employed linear mixed modeling to account for both fixed effects (population-level trends) and random effects (subject-level deviations) to capture the heterogeneity in depression trajectories following TBI. This approach treats depression as a continuous phenomenon, providing a more nuanced representation of depressive symptom evolution over time. When examining depression trends, we distinguished between statistical and clinical significance, as small numerical differences in depression scores may yield statistically significant results in large samples without translating to meaningful clinical change [12]. To address this limitation, we utilized the minimally clinically important difference (MCID) to identify meaningful changes in PHQ-9 scores. This study is unique in its demonstration of the limitations of relying solely on population-level information when predicting depression trajectories in patients with TBI, highlighting the critical need to account for individual heterogeneity. As hypothesized, models incorporating random effects demonstrated substantially improved predictive performance compared to those based solely on fixed effects. Specifically, while population-level estimates captured only a small proportion of the variance in PHQ-9 scores (marginal R\u0026sup2; = 0.07), incorporating random effects led to a marked increase in explained variance (conditional R\u0026sup2; = 0.58) and a substantial improvement in coverage within the 50% prediction intervals (23% vs. 58%). These findings underscore the inherent variability in depression symptoms following TBI; ignoring this individual-level variability may lead to inaccurate predictions and potentially ineffective clinical decision-making.\u003c/p\u003e \u003cp\u003eThe primary goal of this study was to gain a better understanding of the dynamics of depression over time and to highlight factors associated with depression scores among patients with TBI. Our findings revealed that there is considerable heterogeneity between individuals for initial depression scores. Depression trends varied by pre-injury mental health treatment. Patients with a history of pre-injury mental health treatment exhibited a small but statistically significant decrease in depression scores over the follow-up period. However, this reduction did not reach the threshold for minimal clinically important differences, suggesting that while trends may differ statistically across clinical factors, the observed changes are unlikely to be clinically meaningful.\u003c/p\u003e \u003cp\u003eThe overall stability of depression trends observed in our study is consistent with findings from other studies. In the general population, the prevalence of dysthymia and major depressive disorders has remained relatively stable between 1990\u0026ndash;2019, despite an overall increase in the global burden of depression (Liu et al., \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). While longitudinal studies examining depression trends specifically in the traumatic brain injury (TBI) population are limited, trajectory modeling studies have provided valuable insights; a seminal study employing group-based trajectory modeling identified three distinct mental health trajectories among adults with TBI over 2.5 years post-injury: low-stable, medium-stable, and high-stable (Feldman et al., \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). Shen et al identified two main trajectories in adolescents with TBI: a low-stable group (85%) and a high-increasing group (15%) over a 10-year period (Shen \u0026amp; Wang, \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). Other studies have identified stable, improved, and delayed onset trajectories (Bombardier et al., \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e2016\u003c/span\u003e; Gomez et al., \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2017\u003c/span\u003e; Heath et al., \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). Collectively, these findings suggest that the course of depression after head injury is heterogeneous, with the majority (70\u0026ndash;80%) of participants exhibiting stable trajectories.\u003c/p\u003e \u003cp\u003eIn addition to identifying longitudinal trends, this study investigated the demographic, clinical, and injury related factors associated with Year 1 depression scores. Consistent with broader epidemiological trends, sex and age appear to shape the vulnerability to depression. Among demographic factors, females in the current study were more likely to experience higher depression scores than males. In accordance with previous studies which have found that older adults have lower depression scores than their younger counterparts (Bombardier et al., \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2010\u003c/span\u003e, \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e2016\u003c/span\u003e; Passler et al., \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e2022\u003c/span\u003e), we also found that there was a small deceasing depression trend with increasing age. Results from the current study demonstrate that Black and Hispanic races reported higher scores than White patients, a finding supported by several others authors (Arango-Lasprilla et al., \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2012\u003c/span\u003e). Several factors contribute to the increased risk of depression in racial minorities. Individuals from marginalized racial backgrounds may face economic instability which can exacerbate mental health issues after an injury (Stein et al., \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). Additionally, racial minorities are reported to have less access to quality mental health care, which can hinder their ability to receive timely and effective treatment for post-injury psychiatric conditions (Brenner et al., \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2020\u003c/span\u003e).\u003c/p\u003e \u003cp\u003ePre-injury psychiatric history and substance use are well established risk factors for increased depressive symptoms after head injury (Bockhop et al., \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2023\u003c/span\u003e; Delmonico et al., \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). In line with this finding, we observed that pre-injury mental health treatment, which may serve as a proxy for mental health conditions, had the strongest association with initial depression scores. In addition, suicide attempts, and a history of drug use also increased depression. Given this elevated risk, it is important for healthcare providers to implement targeted screening for affective disorders in patients with a history of psychiatric issues after TBI.\u003c/p\u003e \u003cp\u003eAlthough several studies have highlighted the bivariate association of injury severity with depression, the majority of studies have not found an association between injury severity and symptoms of depression in multivariable analysis. Similarly, we also did not see an effect of GCS on depression scores, suggesting that clinicians should anticipate and address depression during recovery from physical trauma, regardless of injury severity (Versluijs et al., \u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). Although other studies have found that higher levels of cognitive impairment at discharge predict elevated depression trajectories (Cariello et al., \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2020\u003c/span\u003e) and lower functional independence is linked to poorer mental health outcomes (Carmichael et al., \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2023\u003c/span\u003e), we did not find any association with the motor and cognition subscales of FIM.\u003c/p\u003e \u003cp\u003eThe analysis of predictive performance revealed that the model's ability varied depending on the specific clinical context \u0026ndash; specifically, whether the task involved forecasting future scores for existing patients or predicting scores for unseen individuals. When predicting PHQ-9 scores for new patients based on baseline PHQ-9, demographic and clinical data, the model's 50% prediction interval had an average width of 6 points (\u0026plusmn;\u0026thinsp;3 points around the predicted score), indicating that half of the observed outcomes are expected to fall within this range. This prediction interval is wider than the commonly accepted MCID of 5 points, indicating uncertainty in identifying new patients who are likely to experience meaningful changes in their depression symptoms. The relatively infrequent data collection of PHQ-9 (Years 1, 2, 5, and 10) in the study cohort may partially explain the model's limited precision for new patients, as the long intervals between assessments could miss important temporal dynamics in depression symptoms. Regular data collection may yield more precise predictions, particularly for new patients without established trajectories.\u003c/p\u003e \u003cp\u003eIn contrast, when predicting Year 5 PHQ-9 scores for existing patients using their prior data (Years 1 and 2), the model demonstrated substantially better predictive performance. The conditional prediction in this scenario achieved an average prediction interval width of 4 points (\u0026plusmn;\u0026thinsp;2 points around the predicted score). This improved precision likely reflects the benefit of incorporating individual longitudinal data, which enabled the model to better capture subject-specific patterns in depression trajectory. Importantly, this prediction interval is narrower than the commonly accepted MCID of 5 points, suggesting the model may have sufficient precision to detect clinically meaningful changes in this longitudinal prediction scenario. Across both scenarios, conditional predictions outperformed marginal predictions reinforcing the importance of accounting for individual-level variability when predicting depression scores. Overall, the predictive performance results suggest that with further refinement, the model has the potential to aid clinicians in making informed decisions regarding depression treatment planning and resource allocation.\u003c/p\u003e \u003cp\u003eThe current study has important public health and clinical implications. First, the overall depression trend indicates the need for sustained, targeted interventions shortly after injury to achieve clinically meaningful improvements in mental health outcomes. In the study sample 44% of patients experienced symptoms consistent with mild, moderate, or severe depression, emphasizing the necessity for systematic screening and management protocols for this substantial proportion of TBI patients. Secondly, from a clinical standpoint, we identified and confirmed several demographic, injury, and clinical factors associated with higher initial depression score, including sex, education, employment, marital status, and TBIs prior to the index injury, which can help clinicians identify high-risk patients requiring enhanced monitoring and intervention. Finally, our predictive modeling approach demonstrates potential clinical utility, particularly for existing patients with longitudinal data where the model achieved sufficient precision to detect clinically meaningful changes. This could support more personalized depression management strategies and resource allocation in longitudinal TBI care.\u003c/p\u003e"},{"header":"Strengths and limitations","content":"\u003cp\u003eThis study has several strengths. Our modelling approach enables us to estimate both population-level effects, quantifying how fixed factors influence depression scores across the population, as well as individual-level variability in baseline depression scores, capturing how individuals deviate from the population's average score at baseline. We used data from the TBIMS which represents the largest prospective TBI outcome study in North America (Dams-O\u0026rsquo;Connor et al., \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). Therefore, our results may generalize to other TBI populations, particularly within other regions of the United States and Canada. The identified model was developed using data collected immediately upon arrival in the ED and upon hospital discharge. All information included in the model is easily collectable and could be routinely included from electronic health records. In contrast to many applications of mixed-effects models for prediction, which often utilize only the fixed effects components, this study leveraged Best Linear Unbiased Predictors (BLUPs) to incorporate individual random effects, an approach we demonstrated is essential for achieving accurate and clinically meaningful predictions of depression trajectories following TBI.\u003c/p\u003e \u003cp\u003eThough this study yields important findings regarding the evolution of depression during the first 10 years after TBI, it is acknowledged that there are limitations that warrant cautious interpretation. As the TBIMS databases collect follow-up information at 1, 2, and 5 years, and every 5 years thereafter, it is common to have incomplete data on patients due to participant withdrawal or loss of follow-up. In addition, missing information may also occur for specific time periods that a given variable had been collected/included by the data set. To account for this limitation, we used a modelling approach which accommodates missing data that is common in longitudinal study designs. To account for potentially informative attrition in our analyses, we estimated weights to apply to each observation in modelling process. By using follow-up time-specific weights, we ensured that the analyses accounted for varying probabilities of observation across different time points, providing a more accurate representation of the relationship between baseline covariates and PHQ-9 scores. We did not include time-varying covariates in our analysis, as many of these factors, such as changes in marital status, employment, or social support, have a reciprocal relationship with depression. For instance, worsening depression can lead to reduced employment opportunities or social isolation, while changes in these factors may also contribute to shifts in depression severity. Including such covariates would complicate the interpretation of their influence, making it unclear whether they are predictors or consequences of depression. True baseline depression is not captured in TBIMIS at the time of injury or discharge from hospital, therefore we could not account for baseline PHQ-9 scores in the model. Finally, we did not have data on whether participants were receiving treatment for depression during the follow-up period. This limitation restricts our ability to directly evaluate the impact of clinical care or therapeutic interventions on changes in depression scores over time. Despite these limitations, the study uniquely provided ten years of prospective data on depression recovery for a cohort of individuals with TBI. If replicated, our results have important public health implications and can inform strategies to optimize depression screening and intervention in the early post-injury period.\u003c/p\u003e"},{"header":"Conclusions","content":"\u003cp\u003eTo our knowledge this is the largest study to model the impact of multiple factors on depression trends for a large population of individuals who sustain a TBI over a ten-year period after injury. Although the population-wide depression scores exhibited a small decrease over the study period, there was substantial variability between individuals in terms of their baseline depression as well as the rate of change over time. We identified several demographic and clinical covariates associated with initial depression scores. Importantly, we demonstrated that population-level models alone inadequately capture the complexity of depression experiences following TBI, whereas accounting for subject-level variations significantly improves prediction. Our modeling approach has the potential to predict depression in patients being followed longitudinally as well as identify depression trajectories in new patients.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003e\u003cu\u003eAcknowledgements:\u003c/u\u003e\u003c/strong\u003e The Traumatic Brain Injury (TBI) Model Systems National Database is a multicenter study of the TBI Model Systems Centers Program and is supported by the National Institute on Disability, Independent Living, and Rehabilitation Research (NIDILRR), a center within the Administration for Community Living (ACL) and Department of Health and Human Services (HHS). However, these contents do not necessarily reflect the opinions or views of the TBI Model Systems Centers, NIDILRR, ACL or HHS.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding:\u003c/strong\u003e This project was supported by the Canadian Institutes of Health Research (CIHR) through the Canada Graduate Scholarships-Doctoral (CGS-D) Award. The funder had no role in the preparation of this manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConflict of interest:\u003c/strong\u003e The authors have no conflict of interest to declare.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication:\u003c/strong\u003e Not applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAvailability of data and material:\u003c/strong\u003e Traumatic Brain Injury Model Systems Database (TBIMS)\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cu\u003eEthics statement:\u0026nbsp;\u003c/u\u003e\u003c/strong\u003eThis study is secondary data analysis of a publicly available dataset. Informed\u0026nbsp;consent\u0026nbsp;was signed by the patient, family or guardian.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor contributions:\u003c/strong\u003e NK contributed to the conception and design of the study, acquisition and analysis of the data, and drafting of the manuscript. DBC, AN, SSRA and CF contributed to the design of the study, data analysis, and review of the manuscript. This work was part of the PhD thesis of NK in the Interdisciplinary PhD program at Dalhousie University under the supervision of SSRA and CF. All authors interpreted the data, revised the manuscript critically for important intellectual content, and provided final approval of the version to be published.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n \u003cli\u003eAnderson, J. F. I. (2025). Heterogeneity of health-related quality of life after mild traumatic brain injury with systemic injury: A cluster analytic approach. \u003cem\u003eDisability and Rehabilitation\u003c/em\u003e, \u003cem\u003e47\u003c/em\u003e(2), 347\u0026ndash;356. https://doi.org/10.1080/09638288.2024.2345278\u003c/li\u003e\n \u003cli\u003eArango-Lasprilla, J. C., Ketchum, J. M., Drew, A., Hammond, F., Powell, J. M., Kreutzer, J., Lim, P. C., \u0026amp; Carr, C. (2012). Neurobehavioural symptoms 1 year after traumatic brain injury: A preliminary study of the relationship between race/ethnicity and symptoms. \u003cem\u003eBrain Injury\u003c/em\u003e, \u003cem\u003e26\u003c/em\u003e(6), 814\u0026ndash;824. https://doi.org/10.3109/02699052.2012.655360\u003c/li\u003e\n \u003cli\u003eBockhop, F., Cunitz, K., Zeldovich, M., Buchheim, A., Beissbarth, T., Hagmayer, Y., von Steinbuechel, N., \u0026amp; CENTER-TBI Participants and Investigators. (2023). Influence of Sociodemographic, Premorbid, and Injury-Related Factors on Post-Traumatic Stress, Anxiety, and Depression after Traumatic Brain Injury. \u003cem\u003eJournal of Clinical Medicine\u003c/em\u003e, \u003cem\u003e12\u003c/em\u003e(12), 3873. https://doi.org/10.3390/jcm12123873\u003c/li\u003e\n \u003cli\u003eBombardier, C. H., Fann, J. R., Temkin, N. R., Esselman, P. C., Barber, J., \u0026amp; Dikmen, S. S. (2010). Rates of major depressive disorder and clinical outcomes following traumatic brain injury. \u003cem\u003eJAMA : The Journal of the American Medical Association\u003c/em\u003e, \u003cem\u003e303\u003c/em\u003e(19), 1938\u0026ndash;1945. https://doi.org/10.1001/jama.2010.599\u003c/li\u003e\n \u003cli\u003eBombardier, C. H., Hoekstra, T., Dikmen, S., \u0026amp; Fann, J. R. (2016). Depression Trajectories during the First Year after Traumatic Brain Injury. \u003cem\u003eJournal of Neurotrauma\u003c/em\u003e, \u003cem\u003e33\u003c/em\u003e(23), 2115\u0026ndash;2124. https://doi.org/10.1089/neu.2015.4349\u003c/li\u003e\n \u003cli\u003eBrenner, E. K., Grossner, E. C., Johnson, B. N., Bernier, R. A., Soto, J., \u0026amp; Hillary, F. G. (2020). Race and ethnicity considerations in traumatic brain injury research: Incidence, reporting, and outcome. \u003cem\u003eBrain Injury\u003c/em\u003e, \u003cem\u003e34\u003c/em\u003e(6), 801\u0026ndash;810. https://doi.org/10.1080/02699052.2020.1741033\u003c/li\u003e\n \u003cli\u003eBrown, V. A. (2021). An Introduction to Linear Mixed-Effects Modeling in R. \u003cem\u003eAdvances in Methods and Practices in Psychological Science\u003c/em\u003e, \u003cem\u003e4\u003c/em\u003e(1), 2515245920960351. https://doi.org/10.1177/2515245920960351\u003c/li\u003e\n \u003cli\u003eCariello, A. N., Perrin, P. B., Agudelo, Y. R., Olivera Plaza, S. L., Quijano-Mart\u0026iacute;nez, M. C., Trujillo, M. A., Arango-Lasprilla, J. C., \u0026amp; Begali, V. (2020). Predictors of longitudinal depression trajectories after traumatic brain injury in Latin America: A multi-site study.\u0026nbsp;\u003cem\u003eNeuroRehabilitation\u003c/em\u003e, \u003cem\u003e46\u003c/em\u003e(2), 205\u0026ndash;212. https://doi.org/10.3233/NRE-192972\u003c/li\u003e\n \u003cli\u003eCarlo, A. D., Basu, A., \u0026amp; Un\u0026uuml;tzer, J. (2021). Associations of Common Depression Treatment Metrics with Patient-Centered Outcomes. \u003cem\u003eMedical Care\u003c/em\u003e, \u003cem\u003e59\u003c/em\u003e(7), 579\u0026ndash;587. https://doi.org/10.1097/MLR.0000000000001540\u003c/li\u003e\n \u003cli\u003eCarmichael, J., Hicks, A. J., Gould, K. R., Ponsford, J., \u0026amp; Spitz, G. (2023). Ten-Year Cohort Study of Emotional Distress Trajectories After Moderate-Severe Traumatic Brain Injury. \u003cem\u003eArchives of Physical Medicine and Rehabilitation\u003c/em\u003e, \u003cem\u003e104\u003c/em\u003e(7), 1081\u0026ndash;1090. https://doi.org/10.1016/j.apmr.2023.02.008\u003c/li\u003e\n \u003cli\u003eCovington, N. V., \u0026amp; Duff, M. C. (2021). Heterogeneity Is a Hallmark of Traumatic Brain Injury, Not a Limitation: A New Perspective on Study Design in Rehabilitation Research. \u003cem\u003eAmerican Journal of Speech-Language Pathology\u003c/em\u003e, \u003cem\u003e30\u003c/em\u003e(2S), 974\u0026ndash;985. https://doi.org/10.1044/2020_AJSLP-20-00081\u003c/li\u003e\n \u003cli\u003eDams-O\u0026rsquo;Connor, K., Sy, K. T. L., Landau, A., Bodien, Y., Dikmen, S., Felix, E. R., Giacino, J. T., Gibbons, L., Hammond, F. M., Hart, T., Johnson-Greene, D., Lengenfelder, J., Lequerica, A., Newman, J., Novack, T., O\u0026rsquo;Neil-Pirozzi, T. M., \u0026amp; Whiteneck, G. (2018). The Feasibility of Telephone-Administered Cognitive Testing in Individuals 1 and 2 Years after Inpatient Rehabilitation for Traumatic Brain Injury. \u003cem\u003eJournal of Neurotrauma\u003c/em\u003e, \u003cem\u003e35\u003c/em\u003e(10), 1138\u0026ndash;1145. https://doi.org/10.1089/neu.2017.5347\u003c/li\u003e\n \u003cli\u003eDelmonico, R. L., Theodore, B. R., Sandel, M. E., Armstrong, M. A., \u0026amp; Camicia, M. (2022). Prevalence of depression and anxiety disorders following mild traumatic brain injury. \u003cem\u003ePM\u0026amp;R\u003c/em\u003e, \u003cem\u003e14\u003c/em\u003e(7), 753\u0026ndash;763. https://doi.org/10.1002/pmrj.12657\u003c/li\u003e\n \u003cli\u003eFakhoury, M., Shakkour, Z., Kobeissy, F., \u0026amp; Lawand, N. (2021). Depression following traumatic brain injury: A comprehensive overview. \u003cem\u003eReviews in the Neurosciences\u003c/em\u003e, \u003cem\u003e32\u003c/em\u003e(3), 289\u0026ndash;303. https://doi.org/10.1515/revneuro-2020-0037\u003c/li\u003e\n \u003cli\u003eFann, J. R., Bombardier, C. H., Dikmen, S., Esselman, P., Warms, C. A., Pelzer, E., Rau, H., \u0026amp; Temkin, N. (2005). Validity of the Patient Health Questionnaire-9 in assessing depression following traumatic brain injury. \u003cem\u003eThe Journal of Head Trauma Rehabilitation\u003c/em\u003e, \u003cem\u003e20\u003c/em\u003e(6), 501\u0026ndash;511. https://doi.org/10.1097/00001199-200511000-00003\u003c/li\u003e\n \u003cli\u003eFeldman, B., Shen, J., Chen, C., Shi, J., \u0026amp; Xiang, H. (2020). Perceived health after adult traumatic brain injury: A Group-Based Trajectory Modeling (GBTM) analysis. \u003cem\u003eBrain Injury\u003c/em\u003e, \u003cem\u003e34\u003c/em\u003e(6), 741\u0026ndash;750. https://doi.org/10.1080/02699052.2020.1753111\u003c/li\u003e\n \u003cli\u003eGomez, R., Skilbeck, C., Thomas, M., \u0026amp; Slatyer, M. (2017). Growth Mixture Modeling of Depression Symptoms Following Traumatic Brain Injury. \u003cem\u003eFrontiers in Psychology\u003c/em\u003e, \u003cem\u003e8\u003c/em\u003e. https://doi.org/10.3389/fpsyg.2017.01320\u003c/li\u003e\n \u003cli\u003eHart, T., Hoffman, J. M., Pretz, C., Kennedy, R., Clark, A. N., \u0026amp; Brenner, L. A. (2012). A Longitudinal Study of Major and Minor Depression Following Traumatic Brain Injury. \u003cem\u003eArchives of Physical Medicine and Rehabilitation\u003c/em\u003e, \u003cem\u003e93\u003c/em\u003e(8), 1343\u0026ndash;1349. https://doi.org/10.1016/j.apmr.2012.03.036\u003c/li\u003e\n \u003cli\u003eHeath, L. M., Kidwai, M. R., Colella, B., Monette, G., Tselichtchev, P., Tomaszczyk, J. C., \u0026amp; Green, R. E. (2023). Predictors and Functional Outcomes Associated With Longitudinal Trajectories of Anxiety and Depression from 2 to \u0026ge;36 Months After Moderate to Severe Traumatic Brain Injury. \u003cem\u003eJournal of Neurotrauma\u003c/em\u003e, \u003cem\u003e40\u003c/em\u003e(21\u0026ndash;22), 2311\u0026ndash;2320. https://doi.org/10.1089/neu.2023.0003\u003c/li\u003e\n \u003cli\u003eJorge, R. E., Robinson, R. G., Moser, D., Tateno, A., Crespo-Facorro, B., \u0026amp; Arndt, S. (2004). Major Depression Following Traumatic Brain Injury. \u003cem\u003eArchives of General Psychiatry\u003c/em\u003e, \u003cem\u003e61\u003c/em\u003e(1), 42\u0026ndash;50. https://doi.org/10.1001/archpsyc.61.1.42\u003c/li\u003e\n \u003cli\u003eKroenke, K., Spitzer, R. L., \u0026amp; Williams, J. B. (2001). The PHQ-9: Validity of a brief depression severity measure. \u003cem\u003eJournal of General Internal Medicine\u003c/em\u003e, \u003cem\u003e16\u003c/em\u003e(9), 606\u0026ndash;613. https://doi.org/10.1046/j.1525-1497.2001.016009606.x\u003c/li\u003e\n \u003cli\u003eŁakuta, P., Krankowska, D., Marcinkiewicz, P., Bociąga‐Jasik, M., \u0026amp; Komorska‐Błażewicz, A. (2022). Enhancing well‐being and alleviating depressive symptoms in people with HIV/AIDS: An intervention based on if\u0026ndash;then plans with self‐affirming cognitions. \u003cem\u003eApplied Psychology. Health and Well-Being\u003c/em\u003e, \u003cem\u003e14\u003c/em\u003e(3), 899\u0026ndash;919. https://doi.org/10.1111/aphw.12357\u003c/li\u003e\n \u003cli\u003eLiu, J., Liu, Y., Ma, W., Tong, Y., \u0026amp; Zheng, J. (2024). Temporal and spatial trend analysis of all-cause depression burden based on Global Burden of Disease (GBD) 2019 study. \u003cem\u003eScientific Reports\u003c/em\u003e, \u003cem\u003e14\u003c/em\u003e(1), 12346. https://doi.org/10.1038/s41598-024-62381-9\u003c/li\u003e\n \u003cli\u003eL\u0026ouml;we, B., Un\u0026uuml;tzer, J., Callahan, C. M., Perkins, A. J., \u0026amp; Kroenke, K. (2004). Monitoring depression treatment outcomes with the patient health questionnaire-9. \u003cem\u003eMedical Care\u003c/em\u003e, \u003cem\u003e42\u003c/em\u003e(12), 1194\u0026ndash;1201. https://doi.org/10.1097/00005650-200412000-00006\u003c/li\u003e\n \u003cli\u003eMcInnes, A. N., Olsen, S. T., Sullivan, C. R. P., Cooper, D. C., Wilson, S., Sonmez, A. I., Albott, S. C., Olson, S. C., Peterson, C. B., Rittberg, B. R., Herman, A., Bajzer, M., Nahas, Z., \u0026amp; Widge, A. S. (2024). Trajectory Modeling and Response Prediction in Transcranial Magnetic Stimulation for Depression. \u003cem\u003emedRxiv\u003c/em\u003e, 2024.05.30.24308258. https://doi.org/10.1101/2024.05.30.24308258\u003c/li\u003e\n \u003cli\u003eMoreno-Agostino, D., Wu, Y.-T., Daskalopoulou, C., Hasan, M. T., Huisman, M., \u0026amp; Prina, M. (2021). Global trends in the prevalence and incidence of depression:a systematic review and meta-analysis. \u003cem\u003eJournal of Affective Disorders\u003c/em\u003e, \u003cem\u003e281\u003c/em\u003e, 235\u0026ndash;243. https://doi.org/10.1016/j.jad.2020.12.035\u003c/li\u003e\n \u003cli\u003ePassler, J. S., Sander, A. M., Temkin, N. R., Barber, J., Gardner, R. C., \u0026amp; Manley, G. T. (2022). Depression in Older Adults 12 Months After Traumatic Brain Injury: A TRACK-TBI Study. \u003cem\u003eArchives of Physical Medicine and Rehabilitation\u003c/em\u003e, \u003cem\u003e103\u003c/em\u003e(1), 83\u0026ndash;89. https://doi.org/10.1016/j.apmr.2021.08.021\u003c/li\u003e\n \u003cli\u003eRabinowitz, A. R., Chervoneva, I., Hart, T., O\u0026rsquo;Neil-Pirozzi, T. M., Juengst, S. B., \u0026amp; Hoffman, J. M. (2020). Heterogeneity in Temporal Ordering of Depression and Participation After Traumatic Brain Injury. \u003cem\u003eArchives of Physical Medicine and Rehabilitation\u003c/em\u003e, \u003cem\u003e101\u003c/em\u003e(11), 1973\u0026ndash;1979. https://doi.org/10.1016/j.apmr.2020.05.026\u003c/li\u003e\n \u003cli\u003eScholten, A. C., Haagsma, J. A., Cnossen, M. C., Olff, M., van Beeck, E. F., \u0026amp; Polinder, S. (2016). Prevalence of and Risk Factors for Anxiety and Depressive Disorders after Traumatic Brain Injury: A Systematic Review. \u003cem\u003eJournal of Neurotrauma\u003c/em\u003e, \u003cem\u003e33\u003c/em\u003e(22), 1969\u0026ndash;1994. https://doi.org/10.1089/neu.2015.4252\u003c/li\u003e\n \u003cli\u003eShen, J., \u0026amp; Wang, Y. (2023). Correlates of Longitudinal Trajectories of Depressive Symptoms in Adolescents With Traumatic Brain Injuries. \u003cem\u003eJournal of Pediatric Psychology\u003c/em\u003e, \u003cem\u003e48\u003c/em\u003e(12), 1021\u0026ndash;1029. https://doi.org/10.1093/jpepsy/jsad073\u003c/li\u003e\n \u003cli\u003eSingh, R. K., Humphries, T. J., Dawson, J. F., Tiupin-Szulc, J., Mason, S., \u0026amp; Lecky, F. E. (n.d.). Changes in depression symptoms over 10 years after TBI; a long-term prospective study. \u003cem\u003eBrain Injury\u003c/em\u003e, \u003cem\u003e0\u003c/em\u003e(0), 1\u0026ndash;8. https://doi.org/10.1080/02699052.2024.2413639\u003c/li\u003e\n \u003cli\u003eSingh, R., Mason, S., Lecky, F., \u0026amp; Dawson, J. (2018). Prevalence of depression after TBI in a prospective cohort: The SHEFBIT study. \u003cem\u003eBrain Injury\u003c/em\u003e, \u003cem\u003e32\u003c/em\u003e(1), 84\u0026ndash;90. https://doi.org/10.1080/02699052.2017.1376756\u003c/li\u003e\n \u003cli\u003eStein, M. B., Jain, S., Giacino, J. T., Levin, H., Dikmen, S., Nelson, L. D., Vassar, M. J., Okonkwo, D. O., Diaz-Arrastia, R., Robertson, C. S., Mukherjee, P., McCrea, M., Mac Donald, C. L., Yue, J. K., Yuh, E., Sun, X., Campbell-Sills, L., Temkin, N., Manley, G. T., \u0026hellip; Zafonte, R. (2019). Risk of Posttraumatic Stress Disorder and Major Depression in Civilian Patients After Mild Traumatic Brain Injury: A TRACK-TBI Study.\u0026nbsp;\u003cem\u003eJAMA Psychiatry\u003c/em\u003e, \u003cem\u003e76\u003c/em\u003e(3), 249\u0026ndash;258. https://doi.org/10.1001/jamapsychiatry.2018.4288\u003c/li\u003e\n \u003cli\u003eTso, S., Saha, A., \u0026amp; Cusimano, M. D. (2021). The Traumatic Brain Injury Model Systems National Database: A Review of Published Research. \u003cem\u003eNeurotrauma Reports\u003c/em\u003e, \u003cem\u003e2\u003c/em\u003e(1), 149\u0026ndash;164. https://doi.org/10.1089/neur.2020.0047\u003c/li\u003e\n \u003cli\u003eTurkoz, I., Alphs, L., Singh, J., Jamieson, C., Daly, E., Shawi, M., Sheehan, J. J., Trivedi, M. H., \u0026amp; Rush, A. J. (2021). Clinically meaningful changes on depressive symptom measures and patient-reported outcomes in patients with treatment-resistant depression. \u003cem\u003eActa Psychiatrica Scandinavica\u003c/em\u003e, \u003cem\u003e143\u003c/em\u003e(3), 253\u0026ndash;263. https://doi.org/10.1111/acps.13260\u003c/li\u003e\n \u003cli\u003eVersluijs, Y., van Ravens, T. W., Krijnen, P., Ring, D., \u0026amp; Schipper, I. B. (2022). Systematic Review of the Association Between Trauma Severity and Postinjury Symptoms of Depression. \u003cem\u003eWorld Journal of Surgery\u003c/em\u003e, \u003cem\u003e46\u003c/em\u003e(12), 2900\u0026ndash;2909. https://doi.org/10.1007/s00268-022-06750-3\u003c/li\u003e\n \u003cli\u003eWelham, S., Cullis, B., Gogel, B., Gilmour, A., \u0026amp; Thompson, R. (2004). Prediction in linear mixed models. \u003cem\u003eAustralian \u0026amp; New Zealand Journal of Statistics\u003c/em\u003e, \u003cem\u003e46\u003c/em\u003e(3), 325\u0026ndash;347. https://doi.org/10.1111/j.1467-842X.2004.00334.x\u003c/li\u003e\n \u003cli\u003eWeuve, J., Tchetgen Tchetgen, E. J., Glymour, M. M., Beck, T. L., Aggarwal, N. T., Wilson, R. S., Evans, D. A., \u0026amp; Mendes de Leon, C. F. (2012). Accounting for bias due to selective attrition: The example of smoking and cognitive decline. \u003cem\u003eEpidemiology (Cambridge, Mass.)\u003c/em\u003e, \u003cem\u003e23\u003c/em\u003e(1), 119\u0026ndash;128. https://doi.org/10.1097/EDE.0b013e318230e861\u003c/li\u003e\n \u003cli\u003eYu, B., Zhang, X., Wang, C., Sun, M., Jin, L., \u0026amp; Liu, X. (2020). Trends in depression among Adults in the United States, NHANES 2005\u0026ndash;2016. \u003cem\u003eJournal of Affective Disorders\u003c/em\u003e, \u003cem\u003e263\u003c/em\u003e, 609\u0026ndash;620. https://doi.org/10.1016/j.jad.2019.11.036\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"traumatic brain injury, depression, longitudinal, mixed model, prediction","lastPublishedDoi":"10.21203/rs.3.rs-6463585/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6463585/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eBackground\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eDepression affects approximately 30% of individuals after traumatic brain injury (TBI), yet long-term depression trends and their determinants are poorly understood. This study aimed to model depression trajectories over ten years post-TBI, compare the predictive performance of population-level only versus both population and subject-level effects, and assess the model's clinical utility for predicting depression in unseen and existing patients.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethods\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eData were obtained from the Traumatic Brain Injury Model System (TBIMS) National Data Bank. Depression was measured using the Patient Health Questionnaire-9 (PHQ-9) and collected at 1, 2, 5 and 10 years after injury. Covariates included age, sex, race, employment, education, functional measures, injury severity, pre-injury mental health, and substance use. Linear mixed modelling was used to identify depression trends and factors associated with depression. Predictive performance was evaluated using mean squared error, coverage, and precision.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe sample comprised 19,397 individuals (mean age 43). Depression scores showed a small decrease over time among those with pre-injury mental health treatment history, but this change was not clinical meaningful. Significant predictors of Year 1 depression included pre-injury mental health treatment (β=1.6), female sex (β=0.86), and prior head injuries (β=0.4). When predicting depression for existing patients using early depression scores, the model achieved precision of 3.7 points, whereas for new patients, the model's precision was 6 points. Conditional predictions outperformed marginal predictions.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusion\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eDepression trajectories following TBI exhibit substantial individual heterogeneity. Population-level models alone inadequately capture this complexity, while models incorporating both population and subject-level variations significantly improve predictive performance. This modeling approach demonstrates the potential for predicting depression trajectories in clinical settings, thereby facilitating individualized assessment and intervention.\u003c/p\u003e","manuscriptTitle":"Longitudinal Trends of Depression in Traumatic Brain Injury: The Role of Individual Heterogeneity in Clinical Prediction","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-05-16 11:47:12","doi":"10.21203/rs.3.rs-6463585/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"5df40855-b0b4-4f19-a8b5-42b80b1eed3b","owner":[],"postedDate":"May 16th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2025-07-30T13:53:10+00:00","versionOfRecord":[],"versionCreatedAt":"2025-05-16 11:47:12","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-6463585","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6463585","identity":"rs-6463585","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}