Impact of Post-Traumatic Epilepsy on Mental Health and Multidimensional Outcome and Quality of Life: An NIDILRR TBIMS Study

Abstract

Traumatic brain injury (TBI) and subsequent post-traumatic epilepsy (PTE) often impair daily activities and mental health (MH), which contribute to long-term TBI-related disability. PTE also affects driving capacity, which impacts functional independence, community participation, and satisfaction with life (SWL). However, studies evaluating the collective impact of PTE on multidimensional outcomes are lacking. Thus, we generated a model to investigate how PTE after moderate-to-severe (ms)TBI affects TBI-associated impairments, limits activities and participation, and influences SWL. Of 5108 participants with msTBI enrolled into the National Institute for Disability, Independent Living, and Rehabilitation Research TBI Model Systems between 2010 and 2018 and with seizure-event data available at year-1 post-TBI, 1214 had complete outcome data and 1003 had complete covariate data used for analysis. We constructed a conceptual framework illustrating hypothesized interrelationships between year-1 PTE, driving status, functional independence measure (FIM), depression and anxiety, as well as year-2 participation, and SWL. We performed univariate and multivariable linear and logistic regressions. A covariate-adjusted structural equation model (SEM), using the lavaan package (R), assessed the conceptual framework’s suitability in establishing PTE links with outcomes 1–2 years post-injury. Multiple parameters were evaluated to assess SEM fit. Year-1 PTE was correlated with year-1 FIM motor (standardized coefficient, β_std = −0.112, p = 0.007) and showed a trend level association with year-1 FIM cognition (β_std = −0.070, p = 0.079). Individuals with year-1 PTE were less likely to drive independently at year 1 (β_std = −0.148, p < 0.001). In addition, FIM motor (β_std = 0.323, p < 0.001), FIM cognition (β_std = 0.181, p = 0.012), and anxiety (β_std = −0.135, p = 0.024) influenced driving status. FIM cognition was associated with year-1 depression (β_std = 0.386, p < 0.001) and year-1 anxiety (β_std = 0.396, p < 0.001), whereas year-1 FIM motor (β_std = 0.186, p = 0.003), depression (β_std = −0.322, p = 0.011), and driving status (β_std = 0.233, p < 0.001) directly affected year-2 objective life participation metrics. Moreover, year-1 depression (β_std = −0.382, p = 0.001) and year-2 participation (β_std = 0.160, p < 0.001) had direct effects on year-2 SWL. SWL was influenced indirectly by year-1 variables, including functional impairment, anxiety, and driving status—factors that impacted year-2 participation directly or indirectly, and consequently year-2 SWL, forming a complex relationship with year-1 PTE. A sensitivity analysis SEM showed that the number of MH disorders was associated with participation and SWL (p < 0.001), and this combined MH variable was directly related to driving status (p < 0.02). Developing PTE during year-1 after msTBI affects multiple aspects of life. PTE effects extend to motor and cognitive abilities, driving capabilities, and indirectly, to life participation and overall SWL. The implications underscore the crucial need for effective PTE management strategies during the first year post-TBI to minimize the adverse impact on factors influencing multidimensional year-2 participation and SWL outcomes. Addressing transportation barriers is warranted to enhance the well-being of those with PTE and msTBI, emphasizing a holistic approach. Further research is recommended for SEM validation studies, including testing causal inference pathways that might inform future prevention and treatment trials.

Introduction

Traumatic brain injuries (TBIs) are a prevalent condition that significantly impacts individuals.^1,2 In the United States, TBI results in approximately 2.8 million visits to emergency departments annually.³ Of these cases, around 282,000 necessitate hospitalization, and about 56,000 individuals lose their lives due to TBI.³ Consequently, the societal expense is estimated at $76.5 billion per year.⁴ Post-traumatic epilepsy (PTE)—defined as one or more unprovoked seizures occurring more than 1 week after TBI⁵—occurs commonly after moderate-to-severe (ms)TBI. PTE constitutes ∼4% of focal epilepsy cases in the overall population and is the primary cause of epilepsy in young adults aged 15–24 years.⁶

The occurrence of PTE following TBI varies based on injury severity, ranging from 4% for moderate injuries to over 16% for severe injuries.^7,8 Severe head trauma with cortical damage and neurological complications, especially if the dura mater is breached, significantly increases the likelihood of PTE.⁹ Late seizures can be delayed but typically manifest within the first few years after the initial injury.^10,11 Approximately 40% of late seizures occur within the first 6 months, 50–60% occur within the first year, and 80% appear within 2 years. The risk of delayed seizures decreases over time, returning to baseline risk after 10–15 years.^7,12 About 86% of individuals having a late post-traumatic seizure will experience a second seizure within 2 years, emphasizing the importance of PTE prevention as well as diagnosing PTE after a single late seizure.¹³

A seminal PTE study in the prospective, nationwide observational TBI Model Systems (TBIMS) cohort reported a 16.8% cumulative PTE incidence in the first 2 years post-injury,⁸ and additional studies demonstrate PTE contributions to long-term mental health (MH) problems and disability.^5,14
–16 Similar to findings in the general population with epilepsy that psychiatric disorders are the most common comorbid conditions,^16
–18 those with PTE can have more frequent and severe displays of emotional dysregulation (e.g., irritability, agitation, aggression), more severe anxiety, and higher rates of comorbid depression and anxiety than TBI survivors without PTE.^5,14,19 PTE may also exacerbate the deleterious effects of TBI on cognition (particularly memory).²⁰ The impact of PTE extends beyond TBI survivors to their caregivers, who in one study reported higher caregiver burden associated with cognitive decline, apathy, and disinhibition, versus caregivers of veterans without PTE.²¹ The role of PTE in affecting cognitive skills suggests the potential utility of the functional independence measure (FIM), although its specific relevance to PTE remains underresearched.

The FIM was designed to evaluate functional independence across 2 primary dimensions as follows: FIM-motor and FIM cognitive (FIM-Cog). The FIM-motor scale assesses physical functions such as self-care, sphincter control, mobility, and locomotion, whereas the FIM-Cog scale assesses cognitive functions, including communication, social interaction, problem-solving, and memory. FIM-motor and FIM-Cog scores are linked to other functional domains such as participation, social integration, and quality of life.^22
–24 Higher FIM-motor scores are associated with greater independence in physical activities, and higher FIM-Cog scores correlate with better social and cognitive engagement, which in turn positively impact quality of life outcomes.^22,25

–28 Despite the well-documented utility of FIM, there is a relative paucity of data specifically examining associations within the context of PTE. Existing literature primarily focuses on general TBI populations,^29

–32 leaving a gap in understanding the unique challenges and recovery trajectories faced by those with PTE.

PTE both directly (as with driving ability) and indirectly (through MH and cognitive changes) may affect functional, participation, and quality of life (satisfaction with life [SWL]) outcomes post-TBI. Those with seizures are less likely to return to driving post-TBI,^33,34 a highly patient-valued functional outcome. A TBIMS study previously evaluated the independent contribution of driving ability after injury to participation in community-based activities (measured using the Participation Assessment with Recombined Tools, or PART).³⁵ Even after controlling for age, sex, race, education, FIM, depression, and living alone versus living with others, ability to drive accounted for meaningful variance in participation at 5 years post-TBI.³⁶ PTE may also contribute to MH conditions and participation restrictions that negatively affect SWL following msTBI.³⁷

Individuals with msTBI often face a significant decline in life roles making them particularly vulnerable to low or decreasing life satisfaction. However, limited information is available evaluating the potential detrimental effects of PTE following msTBI. Studies suggest that PTE is linked to diminished mental well-being following TBI.^{14,38

–43} The onset of PTE is linked to worse functional and psychosocial outcomes within the first 5 years post-TBI.³⁸ In addition, those with epilepsy report lower life satisfaction and self-rated health than expected, indicating the importance of these factors for their well-being.⁴⁴ Adults with epilepsy also have greater frustration versus those without epilepsy, particularly with meaningful accomplishments, social relationships, health, and energy levels.^45,46

Thus, quantifying the impact of PTE on driving ability, and its cascading effects on life participation, is essential. Limited research has examined the broader impact of PTE on outcomes after msTBI, and assessing long-term outcomes at 1 and 2 years following msTBI while accounting for confounding factors impacting outcome is crucial for a comprehensive understanding of PTE. Furthermore, elucidating the interplay between PTE and MH conditions may provide insights into their combined influence on SWL post-TBI. Filling these gaps may facilitate the development of targeted interventions and improve patient care for those with PTE.

Thus, the objective of this study was to generate and fit a conceptual model using structural equation model (SEM) that evaluates the multidimensional and temporal impacts that PTE has on 1- and 2-year outcomes after msTBI. We hypothesized that multiple direct and indirect negative associations between PTE and MH, physical and cognitive impairment, and driving status would influence later participation and SWL outcomes.

Methods

Participants

Our study included 5108 individuals enrolled in the National Institute for Disability, Independent Living, and Rehabilitation Research (NIDILRR)-funded TBIMS National Database injured between 2010 and 2018 with seizure-event data available at year-1 post-TBI. Participants, or legal proxies, provided written informed consent to participate. Each center's institutional review board approved data collection and analyses protocols, with data sharing overseen by the TBIMS National Data and Statistical Center (TBIMS NDSC). Data were gathered using standard TBIMS NDSC procedures.⁹⁹ Among this cohort, there were N = 1003 subjects with complete data on outcomes and covariates (Fig. 1). Variation in missingness was multifactorial, but largely driven by limitations in when specific tests of depression and anxiety were implemented. Covariate distribution is presented in Supplementary Table S1 as a sensitivity analysis of covariate missingness in the sample. There is a significant difference between the included and not included samples in year-2 participation and year-2 SWL. When evaluating the mean and median values, they are similar between the samples; their standard deviations are also similar. However, due to large class imbalance (N = 1003 included and N = 211 not included), the standard error for the mean differs considerably, indicating that the differences observed are likely due to class imbalance. Supplementary Table S2 presents the comparison of the analytic SEM cohort with N = 1003 subjects with the N = 5108 − 1003 = 4105 cohort that was not included in the analytic SEM due to outcome and covariate missingness. FIM scores, driving status, participation, employability (using the Disability Rating Scale (DRS) subscale), substance misuse, age, and residence had somewhat different distributions between these cohorts.

FIG. 1.

Structural equation model (SEM) cohort derivation diagram.

Each TBIMS center is affiliated with one or more Level 1 trauma centers and gathers data spanning acute hospitalization through inpatient rehabilitation. The TBIMS centers enroll individuals (≥16 years) with msTBI—defined as post-traumatic amnesia (PTA) >24 h, positive computed tomography scan findings, or a Glasgow Coma Scale (GCS) score <13—who present to a TBIMS acute care hospital within 72 h of injury and receive inpatient rehabilitation from a designated TBIMS rehabilitation hospital. Participants are interviewed at years 1, 2, and 5 and every subsequent 5 years post-TBI until death. Data for our main measures were collected during the 1- and 2-year follow-up periods.

Data collection

Data were gathered using standard operating procedures outlined by the TBIMS National Data and Statistical Center.^47,48 Self-reported measures were collected directly from participants or from a designated proxy if participants were unable to participate in an interview. Baseline data (i.e., pre-injury history, injury characteristics, acute care, and rehabilitation hospitalization) were collected through medical record review and participant/proxy interview. Follow-up data were collected via telephone, in-person interview, or mailed questionnaires.

Baseline variables

The following demographic and clinical variables were used as covariates: age, sex (self-reported male or female), years of education, residence (private home vs. health care setting or other), TBI severity (severe for GCS scores <9, PTA >1 week, or inability to follow verbal commands >1 day post-TBI vs. moderate), employability (not restricted/competitive vs. noncompetitive/not employable) as defined by the DRS,⁴⁹ and substance misuse (yes vs. no). When adjusting for follow-up variables such as residence, employability, and substance misuse in the SEM, we used the year-specific data as confounders (i.e., year-1 variables while modeling year-1 outcomes and year-2 variable while modeling year-2 outcomes).

Year-1 variables

Data collectors asked participants to self-report, or caregivers to proxy report (as yes/no), whether they had experienced a seizure over the year since rehabilitation discharge. Driving status was categorized as driving independently versus not (if participant had specialized or personal transportation, took public transit, or had no motorized transportation).

The FIM is a tool used in rehabilitation that assesses a patient’s functional independence through the following two components: FIM-motor and FIM-Cog. The FIM-motor component evaluates physical and self-care activities, such as eating, dressing, and mobility, whereas the FIM-Cog component assesses cognitive functions, including communication, social interaction, and memory. The FIM score is derived by summing the scores from the motor and cognitive subscales, yielding a total score that ranges from 18 to 126, with higher scores indicating greater functional independence.^50,51 Together, these scores offer a comprehensive view of a patient’s rehabilitation needs and progress. Ratings were based on the poorest performance over a 72-h assessment period. The distribution of both FIM motor and cognition subscales was left skewed, and Rasch adjustment was not sufficient to meet statistical assumptions for normally distributed data for use with linear regression.^52
–54 Hence, both subscales were categorized as independent versus dependent. Participants were classified as being independent if their FIM scores showed them, on average, to be either independent or modified independent across all items for each subscale.²²

Depression status was calculated using the Patient Health Questionnaire-9 (PHQ-9)—derived from 9 Diagnostic and Statistical Manual-IV (DSM-IV) criteria for major depressive disorder (MDD).⁵⁵ The PHQ-9 was not introduced to the TBIMS data collection protocol until October 2017, and the instrument documents multiple self-reported symptoms in the prior 2 weeks as follows: anhedonia, depressed mood, sleep disturbance, fatigue/low energy, appetite changes, worthlessness or guilt, concentration problems, psychomotor changes, and suicide ideation. Each of these are scored as 0 (not at all), 1 (several days), 2 (more than half the days), or 3 (nearly every day). Consistent with DSM-IV criteria for MDD and adapted for TBI participants were considered to have post-traumatic depression if they endorsed at least five symptoms “more than half the days” (suicidal ideation could be “several days”), with at least one endorsed symptom as either anhedonia or depressed mood.⁵⁶

Anxiety status was derived from the Generalized Anxiety Disorder-7 (GAD-7) scale and was not introduced to the TBIMS data collection protocol until October 2017. The distribution of the scale was right skewed and hence grouped into four categories in a manner consistent with the literature: 0–4: minimal anxiety, 5–9: mild anxiety, 10–14: moderate anxiety, and 15–21: severe anxiety.⁵⁷

The cumulative impact of multiple MH conditions (i.e., depression and anxiety) was evaluated in a sensitivity analysis by creating an ordinal variable differentiating between those with neither condition, those with one of the conditions, or those with both conditions as outlined previously by Juengst et al.¹⁴

Year-2 variables

Participation was measured using the summary statistic of PART-Objective (PART-O), an averaged total score, equating to the sum of the 3 PART subscale scores divided by 3.⁵⁸ The PART-O is a measure of societal-level functioning. It is used to evaluate long-term outcomes and assess intervention effectiveness in enhancing social and societal functioning. SWL was quantified using the satisfaction with life scale (SWLS)⁵⁹—a 5-item scale designed to measure the cognitive judgments of the participants’ satisfaction about life. Participants indicated agreement or disagreement with each of the 5 items using a 7-point scale ranging from 7 (strongly agree) to 1 (strongly disagree).

Analysis methods

A temporal SEM was developed to evaluate PTE relationships with long-term (year-2) participation and SWL using year-1 FIM (motor, cognition), MH variables (depression, anxiety), and driving status as possible mediators, while adjusting for age, sex, years of education, employability, residence, substance misuse, and TBI severity using the following conceptual diagram generated by the author team prior to analyses (Fig. 2).

FIG. 2.

Initial assessment plan conceptualization (before analysis): Temporal relationships with post-traumatic epilepsy (PTE) relevant outcomes. Covariate adjusted: education, age, residential status, injury severity, sex, substance abuse year 1/year 2, and employability. Dotted lines represent indirect effects.

Univariate regressions were initially performed to observe the cross-sectional relationships between each of the variables, and pre-SEM multivariable regressions (adjusting only for the main outcome variables of interest) were completed to determine possible mediation effects and the paths to be tested in the SEM. All pre-SEM regressions utilized the full sample size, including all complete observations for the variables involved, as it is standard practice to test initial relationships using as much data as possible to ensure that those tested in the full-scale SEM are the most promising. Linear regressions were conducted for continuous outcomes (e.g., SWL and PART-O) and binary logistic regression for dichotomous outcomes (e.g., driving status, FIM-motor, FIM-Cog, depression, and anxiety). The regressions presented tested associations for both the concurrent (year-1 outcomes vs. year-1 predictors) and temporal (year-2 SWL and PART-O vs. year-1 predictors) relationships hypothesized.

SEM can simultaneously analyze multiple, temporally dependent relationships while accounting for the residual covariances among the variables of interest, which overcomes the shortcoming of regular regression in the presence of complex correlational structure in the data.⁶⁰ In this study, we modeled multiple covariance structures to obtain a more refined relationship of PTE with other outcome variables. As the goal of this study was to understand the effect of PTE on these primary outcome variables, we tested the PTE relationship to each of the outcome variables in the SEM, irrespective of its significance in the multivariable regressions. However, other nonsignificant relationships in multivariate regression were dropped to simplify the model.

Since we had a mix of continuous and categorical variables included in the SEM, we used a diagonally weighted least squares (DWLS) method of estimation with a NLMINB optimizer (NLMINB: nonlinear minimization subject to box constraints).⁶¹ The root mean square error of approximation (RMSEA), standardized root mean square residual (SRMR), comparative fit index (CFI), and Tucker–Lewis index (TLI) were observed to test the SEM fit. The recommended CFI and TLI values are usually ≥0.95, and the suggested values for RMSEA and SRMR are typically ≤0.05.⁶² Values of standardized path coefficients (β_std) >0.50 indicate a large effect, 0.30 a medium effect, and 0.10 a small effect.⁶³ We presented standardized β’s (β_std) in the Results section to compare effect sizes across the relationships tested. We conducted the SEM analyses in R (version 3.2.5), using the lavaan package (version 0.5-23.1097).⁶⁴

Results

Univariate regression

The univariate regression results for the main outcome variables of interest (SWLS, PART-O, driving status, FIM-motor, FIM-Cog, depression, anxiety), stratified by PTE, are presented in Table 1. Building a full SEM with all paths can result in unstable coefficients due to limitations stemming from the available degrees of freedom per coefficient and the complex error covariance structure. Univariate regressions demonstrated that almost every variable of interest is related to every other variable, either directly or indirectly, indicating that a complex correlation structure (covariance) is present in the data.

Table 1.

Univariate Regressions of the Main Variables of Interest

Predictor	OR	95% CI	p value
Outcome: Year-1 FIM (motor) (independent vs. dependent)
Year-1 seizure (yes vs. no)	0.501	[0.420, 0.597]	<0.001
Year-1 FIM (cognition) (independent vs. dependent)	10.173	[8.907, 11.638]	<0.001
Year-1 Driving Status (yes vs. no)	13.529	[11.147, 16.572]	<0.001
Year-1 Depression Status (yes vs. no)	0.557	[0.414, 0.753]	<0.001
Year-1 Anxiety Status (yes vs. no)	0.526	[0.399, 0.698]	<0.001
Outcome: Year-1 FIM (cognition) (independent vs. dependent)
Year-1 Seizure (yes vs. no)	0.459	[0.386, 0.545]	<0.001
Year-1 FIM (motor) (independent vs. dependent)	10.173	[8.907, 11.638]	<0.001
Year-1 Driving status (yes vs. no)	7.771	[6.730, 9.003]	<0.001
Year-1 Depression status (yes vs. no)	0.344	[0.261, 0.452]	<0.001
Year-1 Anxiety status (yes vs. no)	0.319	[0.247, 0.414]	<0.001
Outcome: Year-1 driving status (yes vs. no)
Year-1 Seizure (yes vs. no)	0.379	[0.308, 0.463]	<0.001
Year-1 FIM (motor) (independent vs. dependent)	13.529	[11.147, 16.572]	<0.001
Year-1 FIM (cognition) (independent vs. dependent)	7.771	[6.730, 9.003]	<0.001
Year-1 Depression Status (yes vs. no)	0.464	[0.347, 0.614]	<0.001
Year-1 Anxiety Status (yes vs. no)	0.393	[0.297, 0.515]	<0.001
Outcome: Year-1 depression status (yes vs. no)
Year-1 Seizure (yes vs. no)	1.384	[0.880, 2.110]	0.143
Year-1 FIM (motor) (independent vs. dependent)	0.557	[0.414, 0.753]	<0.001
Year-1 FIM (cognition) (independent vs. dependent)	0.344	[0.261, 0.452]	<0.001
Year-1 Driving status (yes vs. no)	0.464	[0.347, 0.614]	<0.001
Year-1 Anxiety status (yes vs. no)	32.101	[23.234, 44.792]	<0.001
Outcome: Year-1 anxiety status (yes vs. no)
Year-1 Seizure (yes vs. no)	1.730	[1.152, 2.545]	0.007
Year-1 FIM (motor) (independent vs. dependent)	0.526	[0.399, 0.698]	<0.001
Year-1 FIM (cognition) (independent vs. dependent)	0.319	[0.247, 0.414]	<0.001
Year-1 Driving Status (yes vs. no)	0.393	[0.297, 0.515]	<0.001
Year-1 Depression Status (yes vs. no)	32.101	[23.234, 44.792]	<0.001

Predictor	β	SE	p value
Outcome: Year-2 participation (PART-O, Rasch 0–100)
Year-1 Seizure (yes vs. no)	−0.276	0.035	<0.001
Year-1 FIM (motor) (independent vs. dependent)	0.696	0.022	<0.001
Year-1 FIM (cognition) (independent vs. dependent)	0.636	0.021	<0.001
Year-1 Driving status (yes vs. no)	0.677	0.019	<0.001
Year-1 Depression status (yes vs. no)	−0.371	0.048	<0.001
Year-1 Anxiety status (yes vs. no)	−0.303	0.045	<0.001
Outcome: Year-2 quality of life (SWLS)
Year-1 Seizure (yes vs. no)	−2.079	0.451	<0.001
Year-1 FIM (motor) (independent vs. dependent)	3.978	0.326	<0.001
Year-1 FIM (cognition) (independent vs. dependent)	4.042	0.292	<0.001
Year-1 Driving status (yes vs. no)	3.996	0.261	<0.001
Year-1 Depression status (yes vs. no)	−7.661	0.590	<0.001
Year-1 Anxiety status (yes vs. no)	−7.104	0.554	<0.001
Year-2 Participation (PART-O)	5.236	0.170	<0.001

FIM, functional independence measure; SWLS, satisfaction with life scale; PART-O, participation assessment with recombined tools-objective; OR, odds ratio; CI, confidence interval; SE, standard error.

To address this correlational structure, and simplify our SEM, we next used multivariate regression to account for possible mediation effects to understand which paths should be tested in the SEM and whether to model the associations as covariances or regressions to appropriately represent the relationships.

Multivariate regression

The multivariable regression results for both concurrent (year-1 outcomes regressed on year-1 predictors) and temporal (year-2 SWLS and PART-O regressed on year-1 predictors) associations are presented in Table 2. This table contains a regression coefficient for each endogenous variable while adjusting for all other main variables of interest (not adjusted for covariates such as age, sex, race, education, and so on). These models are more refined than the univariate models presented in Table 1. The purpose of these models was to understand the paths (indicating both regression and covariance paths) that should be tested in the SEM, where each path of the SEM would be adjusted for all covariates.

Table 2.

Multivariable Regressions of the Main Variables of Interest (pre-SEM Regressions)

Predictor	OR	95% CI	p value
Outcome: Year-1 FIM (motor) (independent vs. dependent)
Year-1 seizure (yes vs. no)	0.792	[0.541, 1.171]	0.236
Year-1 FIM (cognition) (independent vs. dependent)	3.251	[2.456, 4.311]	<0.001
Year-1 Driving status (yes vs. no)	5.501	[3.893, 7.932]	<0.001
Year-1 Depression status (yes vs. no)	0.913	[0.578, 1.452]	0.697
Year-1 Anxiety status (yes vs. no)	1.037	[0.674, 1.608]	0.870
Outcome: Year-1 FIM (cognition) (independent vs. dependent)
Year-1 Seizure (yes vs. no)	0.916	[0.633, 1.335]	0.645
Year-1 FIM (motor) (independent vs. dependent)	3.254	[2.46, 4.312]	<0.001
Year-1 Driving status (yes vs. no)	3.620	[2.736, 4.823]	<0.001
Year-1 Depression status (yes vs. no)	0.589	[0.387, 0.900]	0.014
Year-1 Anxiety status (yes vs. no)	0.589	[0.396, 0.876]	0.009
Outcome: Year-1 driving status (yes vs. no)
Year-1 Seizure (yes vs. no)	0.357	[0.236, 0.531]	<0.001
Year-1 FIM (motor) (independent vs. dependent)	5.487	[3.879, 7.917]	<0.001
Year-1 FIM (cognition) (independent vs. dependent)	3.589	[2.711, 4.785]	<0.001
Year-1 Depression status (yes vs. no)	0.938	[0.596, 1.477]	0.783
Year-1 Anxiety status (yes vs. no)	0.574	[0.372, 0.876]	0.011
Outcome: Year-1 depression status (yes vs. no)
Year-1 Seizure (yes vs. no)	0.923	[0.516, 1.615]	0.782
Year-1 FIM (motor) (independent vs. dependent)	0.965	[0.608, 1.542]	0.880
Year-1 FIM (cognition) (independent vs. dependent)	0.594	[0.390, 0.908]	0.016
Year-1 Driving status (yes vs. no)	0.916	[0.595, 1.412]	0.689
Year-1 Anxiety status (yes vs. no)	31.864	[21.79, 47.256]	<0.001
Outcome: Year-1 anxiety status (yes vs. no)
Year-1 Seizure (yes vs. no)	1.438	[0.843, 2.392]	0.782
Year-1 FIM (motor) (independent vs. dependent)	0.987	[0.641, 1.532]	0.880
Year-1 FIM (cognition) (independent vs. dependent)	0.580	[0.392, 0.858]	0.016
Year-1 Driving status (yes vs. no)	0.557	[0.370, 0.833]	0.689
Year-1 Depression status (yes vs. no)	31.788	[21.747, 47.122]	<0.001

Predictor	β	SE	p value
Outcome: Year-2 participation (PART-O, Rasch 0–100)
Year-1 Seizure (yes vs. no)	−0.017	0.053	0.745
Year-1 FIM (motor) (independent vs. dependent)	0.268	0.045	<0.001
Year-1 FIM (cognition) (independent vs. dependent)	0.202	0.041	<0.001
Year-1 Driving status (yes vs. no)	0.392	0.037	<0.001
Year-1 Depression status (yes vs. no)	−0.264	0.062	<0.001
Year-1 Anxiety status (yes vs. no)	0.048	0.059	0.41
Outcome: Year-2 quality of life (SWLS)
Year-1 Seizure (yes vs. no)	−0.271	0.640	0.672
Year-1 FIM (motor) (independent vs. dependent)	1.797	0.560	0.001
Year-1 FIM (cognition) (independent vs. dependent)	0.308	0.505	0.543
Year-1 Driving status (yes vs. no)	1.498	0.466	0.001
Year-1 Depression status (yes vs. no)	−4.179	0.767	<0.001
Year-1 Anxiety status (yes vs. no)	−2.976	0.713	<0.001
Year-2 Participation (PART-O)	3.843	0.346	<0.001

SEM, structural equation model; FIM, functional independence measure; SWLS, satisfaction with life scale; PART-O, participation assessment with recombined tools-objective; OR, odds ratio; CI, confidence interval; SE, standard error.

In Table 2, we show that unlike the unadjusted models in Table 1, year-1 PTE is only a significant predictor of year-1 driving status when adjusting for other variables. From Table 2, we also observed that anxiety and depression have no significant relationship to FIM-motor scores; the FIM-motor subscale, in contrast, was not a significant predictor of depression or anxiety when modeled with other variables of interest.

However, FIM scores, depression, and anxiety showed strong relationships with each other, indicating that a covariance should be modeled for each of these relationships in the SEM. Depression and anxiety were significant predictors of FIM-Cog, and FIM-Cog was also a significant predictor of depression and anxiety, suggesting that pairwise covariances exist among these three variables and should be included in the SEM. Anxiety was also associated with driving status independent of other variables, but driving status was not associated with concurrent anxiety status, indicating a unidirectional regression path in the SEM. Conversely, depression was not associated with driving status, and driving also did not explain concurrent depression status, suggesting that these regression paths could be dropped from the SEM.

Relationships to year-2 PART-O and SWLS are also presented in Table 2. Anxiety at year 1 was not a significant predictor of year-2 PART-O, and year-1 FIM-Cog was not a significant predictor of year-2 SWLS. Thus, we also dropped these paths from the planned SEM. Considering the logical flow of these multivariable regressions, we constructed and tested all the paths presented in Figure 3, which we refer to as the SEM path diagram.

FIG. 3.

Conceptual path diagram for the proposed structural equation model (SEM) is based on multivariable regressions (solid lines) and hypothesized covariance (dashed lines) relationships. The inclusion of paths and their directional associations were informed by a series of multivariable regressions (Table 2).

SEM development

The SEM formulated from the relationships observed in Table 2 represents a highly complex system represented by the conceptual path diagram (Fig. 3), and the SEM generated with n = 1003 complete observations had overall satisfactory model fits (Fig 4). The CFI and TLI were 0.949 and 0.955, respectively. The RMSEA and SRMR were 0.064 and 0.097, respectively.

The SEM (Fig. 3) was adjusted for age, sex, years of education, residence, TBI severity, employability, and substance misuse. The paths significant at 5% and 10% levels are presented by solid and dotted lines, respectively. Standardized β (β_std) coefficients are presented to compare effect sizes across the various relationships tested with the SEM. Year-1 PTE was associated with year-1 FIM-motor (β_std = −0.112, p = 0.007) and marginally associated with year-1 FIM-Cog (β_std = −0.070, p = 0.079). PTE associations with FIM subscales were more prominent when tested in SEM compared with the multivariable regressions (Table 2), which may be due to covariance adjustment when generating the SEM. Year-1 FIM-motor and -Cog had a highly significant covariance (β_std = 0.406, p < 0.001). The covariance between year-1 depression and anxiety was even stronger (β_std = 0.844, p < 0.001). Year-1 FIM-Cog also had strong covariances with both year-1 depression (β_std = −0.386, p < 0.001) and anxiety (β_std = −0.396, p < 0.001). Those with PTE at year 1 were less likely to drive independently at 1 year (β_std = −0.148, p < 0.001), as were those with higher anxiety levels at 1 year (β_std = −0.135, p = 0.024). Participants with independent or modified-independent motor (β_std = 0.323, p < 0.001) and cognitive (β_std = 0.181, p = 0.012) FIM scores at 1 year were more likely to drive at year 1. PTE was not independently related to year-1 depression, anxiety, PART-O, or SWLS; however, PTE had an indirect link with these variables via its associations with FIM subscales.

The temporal relationships of year-2 PART-O and SWLS with the year-1 variables, when adjusting for covariates, showed that participants with depression at year 1 were also less likely to have higher scores in the year-2 PART-O scale (β_std = −0.322, p = 0.011) and SWLS (β_std = −0.382, p = 0.001). However, participants with independent or modified-independent motor function in year 1 were more likely to have higher scores on the year-2 PART-O scale (β_std = 0.186, p = 0.003) and SWLS (β_std = 0.189, p = 0.003). Participants who drove independently at year 1 had higher PART-O scores in year 2 (β_std = 0.233, p < 0.001), and higher year-2 PART-O scores were associated with higher year-2 SWLS (β_std = 0.160, p < 0.001). Although difficult to quantify given the complexity of SEM, these results indicate that FIM and driving status at year 1 may mediate PTE associations with SWLS. All SEM path coefficients (covariances and regression paths) are presented in Table 3.

Table 3.

Coefficients of the SEM

Variables	Estimate, β	SE	z-value	p value	Standardized estimate, β _std
Regression paths
FIM motor at year 1
Year-1 Seizure (yes vs. no)	−0.376	0.140	−2.676	0.007	−0.112
Employability (not restricted/competitive vs. others)	0.844	0.128	6.596	<0.001	0.258
Substance misuse (yes vs. no)	0.093	0.146	0.639	0.523	0.036
Age	−0.010	0.003	−3.208	0.001	−0.170
Gender (male vs. female)	0.275	0.111	2.476	0.013	0.109
TBI severity (severe vs. moderate)	−0.059	0.127	−0.461	0.644	−0.023
Years of education	0.017	0.019	0.854	0.393	0.044
Residence (private home vs. others)	0.264	0.271	0.976	0.329	0.035
FIM cognition at year 1
Year-1 Seizure (yes vs. no)	−0.235	0.134	−1.754	0.079	−0.070
Employability (not restricted/competitive vs. others)	1.065	0.137	7.798	<0.001	0.322
Substance misuse (yes vs. no)	0.183	0.135	1.354	0.176	0.071
Age	≅0 (negative)	0.003	−0.149	0.881	−0.008
Gender (male vs. female)	0.102	0.106	0.962	0.336	0.040
TBI severity (severe vs. moderate)	−0.414	0.127	−3.250	0.001	−0.161
Years of education	0.027	0.017	1.651	0.099	0.073
Residence (private home vs. others)	0.146	0.324	0.450	0.653	0.019
Depression status at year 1
Year-1 Seizure (yes vs. no)	0.124	0.151	0.818	0.413	0.039
Employability (not restricted/competitive vs. others)	−0.347	0.140	−2.470	0.014	−0.11
Substance misuse (yes vs. no)	0.269	0.125	2.147	0.032	0.110
Age	−0.003	0.004	−0.859	0.390	−0.059
Gender (male vs. female)	−0.129	0.127	−1.021	0.307	−0.054
TBI severity (severe vs. moderate)	−0.227	0.134	−1.693	0.090	−0.093
Years of education	−0.052	0.019	−2.681	0.007	−0.145
Residence (private home vs. others)	−0.416	0.345	−1.207	0.227	−0.057
Anxiety status at year 1
Year-1 Seizure (yes vs. no)	0.134	0.148	0.906	0.365	0.042
Employability (not restricted/competitive vs. others)	−0.306	0.138	−2.224	0.026	−0.097
Substance misuse (yes vs. no)	0.263	0.134	1.962	0.050	0.106
Age	−0.008	0.004	−2.307	0.021	−0.148
Gender (male vs. female)	−0.107	0.119	−0.899	0.369	−0.044
TBI severity (severe vs. moderate)	−0.138	0.132	−1.047	0.295	−0.056
Years of education	−0.053	0.019	−2.861	0.004	−0.147
Residence (private home vs. others)	−0.148	0.370	−0.400	0.689	−0.020
Driving status at year 1
Year-1 Seizure (yes vs. no)	−0.574	0.149	−3.858	<0.001	−0.148
Year-1 FIM (motor) (independent vs. dependent)	0.380	0.073	5.214	<0.001	0.327
Year-1 FIM (cognition) (independent vs. dependent)	0.205	0.082	2.499	0.012	0.178
Year-1 Anxiety status (yes vs. no)	−0.169	0.071	−2.381	0.017	−0.140
Employability (not restricted/competitive vs. others)	0.354	0.166	2.130	0.033	0.093
Substance misuse (yes vs. no)	0.200	0.136	1.476	0.140	0.067
Age	0.006	0.003	2.090	0.037	0.084
Gender (male vs. female)	0.077	0.103	0.748	0.454	0.026
TBI severity (severe vs. moderate)	−0.294	0.115	−2.556	0.011	−0.099
Years of education	0.045	0.016	2.75	0.006	0.104
Residence (private home vs. others)	1.375	0.515	2.667	0.008	0.157
Participation at year 2
Year-1 Seizure (yes vs. no)	0.064	0.062	1.022	0.307	0.031
Year-1 FIM (motor) (independent vs. dependent)	0.123	0.037	3.337	0.001	0.198
Year-1 FIM (cognition) (independent vs. dependent)	0.020	0.035	0.582	0.561	0.033
Year-1 Driving status (yes vs. no)	0.121	0.027	4.496	<0.001	0.227
Year-1 Depression status (yes vs. no)	−0.123	0.033	−3.782	<0.001	−0.189
Employability (not restricted/competitive vs. others)	0.487	0.061	7.976	<0.001	0.241
Substance misuse (yes vs. no)	0.050	0.050	0.991	0.322	0.034
Age	−0.007	0.001	−5.595	<0.001	−0.195
Gender (male vs. female)	−0.048	0.044	−1.098	0.272	−0.031
TBI severity (severe vs. moderate)	−0.007	0.052	−0.142	0.887	−0.005
Years of education	0.042	0.007	5.607	<0.001	0.179
Residence (private home vs. others)	0.631	0.172	3.659	<0.001	0.114
SWLS at year 2
Year-1 Seizure (yes vs. no)	−0.038	0.110	−0.344	0.731	−0.010
Year-2 Participation (PART-O)	0.309	0.071	4.364	<0.001	0.172
Year-1 FIM (motor) (independent vs. dependent)	0.153	0.057	2.659	0.008	0.136
Year-1 Driving status (yes vs. no)	0.052	0.043	1.227	0.220	0.054
Year-1 Depression status (yes vs. no)	−0.489	0.139	−3.515	<0.001	−0.419
Year-1 Anxiety status (yes vs. no)	0.072	0.138	0.518	0.604	0.062
Employability (not restricted/competitive vs. others)	0.466	0.100	4.660	<0.001	0.128
Substance misuse (yes vs. no)	−0.214	0.084	−2.547	0.011	−0.081
Age	0.002	0.003	0.644	0.520	0.025
Gender (male vs. female)	−0.099	0.090	−1.098	0.272	−0.035
TBI severity (severe vs. moderate)	−0.093	0.095	−0.979	0.328	−0.033
Years of education	−0.002	0.013	−0.140	0.889	−0.004
Residence (private home vs. others)	−0.214	0.234	−0.912	0.362	−0.021

Pairs	Estimate	SE	z-value	p value	Standardized Estimate
Covariances
FIM motor and cognition	0.406	0.054	7.510	<0.001	0.406
Depression and anxiety	0.844	0.029	29.087	<0.001	0.844
FIM cognition and depression	−0.386	0.058	−6.630	<0.001	−0.386
FIM cognition and anxiety	−0.396	0.056	−7.084	<0.001	−0.396

SEM, structural equation model; TBI, traumatic brain injury; FIM, functional independence measure; SWLS, satisfaction with life scale; PART-O, participation assessment with recombined tools-objective; SE, standard error.

As a sensitivity analysis, we performed an additional SEM analysis combining depression and anxiety into a single variable to capture their shared symptomatology and frequent comorbidity (Fig. 5). Among the SEM analytic cohort of N = 1003, 817 (81.5%) participants had no depression or anxiety, whereas 36 (3.6%) had depression only, 61 (6.1%) had anxiety only, and 89 (8.9%) had both depression and anxiety. The results showed that the associations between this combined MH variable and FIM-Cog had standardized β weights very similar to those observed when depression and anxiety were modeled separately. However, when modeling the associations of the combined MH variable with PART-O and SWLS, the β weights of the combined variable (Fig. 5) were modestly diminished for PART-O and SWLS compared with those of depression alone (Fig. 4). In addition, the distinct effect of anxiety on driving status observed in Figure 4, was not distinguishable in the combined model (Fig. 5), limiting the granularity of interpretation. Notably, with this ordinal MH variable structure, there is a direct relationship between increasing number of MH conditions and reduced likelihood to drive, suggesting a cumulative impact of MH conditions on driving status, which then indirectly impacted participation and life satisfaction. Direct associations between the number of MH conditions and participation, as well as life satisfaction, were also noted.

FIG. 4.

Summary results of the fitted structural equation model (SEM) with depression and anxiety modeled separately. Bidirectional arrows indicate covariance paths (where both variables covary), and unidirectional arrows indicate regression paths (where the outcome changes due to the change in the predictor). Dashed lines indicate at least trend level associations (p < 0.1).

FIG. 5.

Summary results of the fitted structural equation model (SEM) combining depression and anxiety. Bidirectional arrows indicate covariance paths (where both variables covary), and unidirectional arrows indicate regression paths (where the outcome changes due to the change in the predictor). Dashed lines indicate at least trend level associations (p < 0.1).

Discussion

This study explored the complex pathways by which PTE development within the first year after TBI might impact SWL and social participation while considering motor and cognitive dysfunctions, driving status, and MH disorders such as depression and anxiety. SEM allowed us to analyze multiple relationships simultaneously, addressing the complex interdependencies and temporal aspects in the data. Its suitability for handling diverse variable types was validated using DWLS with a NLMINB optimizer. Together, these findings underscore an intricacy of relationships among injury-related factors, functional abilities, MH, driving status, and subsequent outcomes in participation and SWL over time following msTBI. The CFI and TLI aligned with the typical guidelines regarding goodness of fit, whereas the RMSEA and SRMR slightly exceeded the suggested thresholds. However, when taken together, the fit indices suggest that the SEM constructed was satisfactory in modeling these complex relationships, especially given the simultaneous estimation of numerous paths/coefficients.

Year-1 PTE demonstrated a strong link with year-1 FIM-motor scores, while only demonstrating a marginal association with FIM-Cog. This SEM demonstrated more pronounced relationships between PTE and FIM subscales than what were reported using preliminary multivariable regressions, emphasizing the importance of considering these complex interrelationships together using SEM. Participants with PTE were less likely to drive independently at year 1, whereas those with better FIM scores were more likely to do so. Strong (bidirectional) covariance relationships were evident between FIM-motor and FIM-Cog, as well as between depression and anxiety at year 1. Despite lacking direct links, PTE indirectly influenced anxiety, participation, and SWL through its associations with FIM subscales. Year-1 depression significantly contributed to lower participation and SWL scores in year 2, whereas better motor function at year 1 directly contributed to higher year-2 participation and SWL scores. Furthermore, independent driving at year 1 directly impacted year-2 SWL, as well as indirectly contributed to higher year-2 SWL scores, via associations with year-2 participation.

Findings were generally consistent with the literature to date and suggest that PTE significantly impacts SWL for persons with TBI. Functional deficits, psychological health challenges, and overall diminished well-being is highest among those with treatment-resistant PTE compared with those with idiopathic epilepsy and/or is controlled.^{5,14
–16,65} Comorbid PTE following msTBI can be broadly understood by examining its intersections within the International Classification of Functioning, Disability, and Health (ICF) framework.^66,67 Mapping the primary relationships to the ICF framework highlights the multidimensional impact that PTE exerts on an individual’s overall functioning and well-being. The study findings illustrate how PTE affects various components of the ICF as follows: body functions and structures (motor and cognitive abilities), activities (driving independence), participation (life engagement), environmental factors (regulations and restrictions around driving), and personal factors (MH status). Importantly, the correlations between year-1 PTE and diminished functional independence, alongside FIM associations with anxiety, and restricted driving capabilities, demonstrate the interdependence of these ICF domains. These interactions illustrate how PTE permeates multiple layers of functioning and disability to impact an individual’s engagement in daily life and societal roles. After TBI, individuals returning to their communities need to adapt to substantial life alterations and reconstructing their sense of self. PTE exacerbates the already challenging impacts of msTBI and can further impede the process of societal reintegration.^68

–71 PTE onset, often months or years after the initial injury, frequently disrupts earlier progress toward recovery, and PTE presents as a significant obstacle that impacts autonomy, employment prospects, leisure activities, mental well-being, and safety. Thus, PTE management requires a multidimensional approach that considers both medical interventions and environmental modifications, MH support, and strategies to facilitate societal participation, aligning with the holistic tenets of the ICF framework.

In the complex landscape of maximizing clinical outcome post-TBI, identifying effective PTE treatments plays a key role in facilitating seizure management and preventing/mitigating other PTE sequelae. This critical goal affects the broader management goals of preventing cognitive changes, enhancing life participation, and improving overall SWL. PTE associations with patient outcomes may be influenced by the PTE itself, the medications required for its management, or a combination of both.^72,73 Antiseizure medication (ASM) administration for PTE may give rise to adverse effects on cognitive functions, mood, and behavior that influence these broader outcome constructs.^74
–76 Research by Raymont et al. showed this correlation by emphasizing how PTE serves as a predictor for overall cognitive capacity with a decrease in intelligence quotient scores among individuals affected by PTE.⁷⁷ In addition, the duration, frequency, and severity of PTE, including the shift from focal unaware seizures to focal to bilateral tonic–clonic seizures, also contributed to substantial cognitive deterioration.⁷⁷ In a prospective cohort study of 3044 participants with TBI, a 2.8% incidence of self-reported PTE was independently associated with significantly lower Glasgow Outcome Scale Extended scores, higher Brief Symptom Inventory-18 scores, and elevated Rivermead Cognitive Metric scores at 12 months post-injury.⁷⁸ Yet, Haltiner et al. demonstrated no significant differences in 12-month outcomes based on seizure presence when controlling for injury severity. Taken together, these findings may suggest that observed worse outcomes may be attributed to the brain injury effects causing seizures rather than the seizures themselves.⁷⁹

PTE was correlated with lower FIM scores. The primary cognitive consequences of ASM use include diminished attention, vigilance, and psychomotor speed.^74,76,80,81 Further, changes in psychomotor speed and motor function⁸² can affect FIM-motor scores, potentially impacting the ability to drive even when PTE is otherwise well-controlled. Despite the challenges involved with ASM treatment, effective seizure control can enhance cognitive function and patients’ well-being.⁶⁵ Assessing a patient’s cognitive state and psychomotor status before and during treatment initiation may aid in monitoring disease progression, as well as treatment effectiveness, medication side effects,^80,83,84 and effects on participation and SWL.^85
–87

In addition, higher FIM-motor and -Cog scores observed at year 1 were linked to independent driving, fostering positive effects on participation and SWL at year 2. Kolakowsky‐Hayner et al. examined the long-term effects of PTE in individuals 5–13 years post-TBI.⁸⁸ About half of 25 participants had suspended licenses, or abstained from driving, whereas some with licenses experienced seizures. Many faced limitations in social activities, particularly with respect to occupation and social integration.⁸⁸ Driving restrictions and poor seizure control are likely primary drivers for these findings.⁸⁸ Nonetheless, these challenges underscore the importance of comprehensive care and support systems for individuals navigating epilepsy after TBI and barriers to their transportation/driving needs,⁸⁹ including potential anxiety that may be keeping people from returning to driving. Anxiety and depression were significantly related to year-1 FIM-Cog, a factor impacting driving in our main model. Our sensitivity analysis suggests a cumulative direct impact of the number of MH conditions on driving. However, the unique contribution of anxiety to driving status was indistinguishable using this combined MH construct.

Anxiety is a common comorbidity affecting individuals with epilepsy, often exacerbating the challenges associated with driving.⁹⁰ Anxiety impairs cognitive functions and reaction times, which are critical for safe driving.⁹¹ In addition, the fear of having seizures can further heighten anxiety levels, leading to avoidance behaviors and reduced driving participation.⁹² The interplay between anxiety, FIM-Cog, and PTE significantly affects driving ability, thereby impacting overall functional independence, which has downstream impacts on life role participation and satisfaction.

Considering the distinct pathological dynamics between PTE and msTBI and their potential impact on SWL, there is a pressing need for future research priorities to delve into the functional and SWL outcomes observed after PTE. The literature highlights a noticeable correlation at 1-year post-injury,^78,79 indicating a complex interplay between seizure-induced neuronal activity, ongoing damage, ASM side effects, and initial injury burden, which likely exacerbates cognitive and neuropsychological impairments. Thus, future research priorities should investigate causal links between PTE and cognitive decline using blood biomarkers and neuroimaging techniques. The impact of ASMs on cognitive and neuropsychological function may complicate the identification of causal relationships between PTE and other outcomes.^93,94

However, this report increases our understanding of the multidimensional pathways between msTBI and their impact on both secondary conditions like PTE and functional outcomes, which may facilitate the development of effective treatment and prevention strategies. Biomarker-based investigations may inform the interplay between msTBI, PTE, and functional outcomes and may help uncover the biological basis for specific mechanisms and risk factors to support precision medicine treatment and strategies that reduce comorbidity burden and improve function, participation, and life quality. Finally, SEM validation against independent datasets will be important to mitigate model misspecification and quantify causal effects that may inform treatment trial design. Evaluating SEM performance with new independent cohorts that include depression/anxiety data will help refine and validate findings.

Limitations

While this study informs the complex connections between msTBI, PTE, functional outcomes, and MH, we also acknowledge several limitations. Our study focused solely on individuals with msTBI admitted to acute rehabilitation. Although the TBIMS national cohort is reportedly generalizable to the larger population admitted to acute rehabilitation,⁹⁵ the findings observed with the cohort may still have restricted applicability to those with PTE who do not meet the admission requirements for inpatient rehabilitation. The temporal nature of the study might not fully capture the dynamic changes in these variables within our 2-year monitoring period and over more extended periods. In addition, the limited sample with complete MH data limited power to detect smaller effect sizes between MH variables and PTE.

Our study is significantly limited by the absence of medication data, which hinders our understanding of how ASMs impact MH, as well as cognitive and motor function, which are key effectors of driving, participation, and SWL. While the SEM demonstrated favorable fit statistics with CFI and TLI, generally meeting the suggested criterion, the RMSEA and SRMR slightly exceeded the recommended 0.05 threshold. These fit indices indicate a generally acceptable model fit, but the slight deviations from the ideal thresholds may highlight potential limitations in the model’s precise representation of the tested relationships. Despite these encouraging fit indices, it is crucial to acknowledge limitations inherent to using an SEM approach. SEM’s sensitivity to sample size, particularly in complex models with mixed continuous-categorical variables, warrants caution in generalizing our findings. This is particularly pertinent given the imbalanced nature of our dataset, with only 12% of the sample having PTE, and the use of binary variables such as depression (yes/no). Binary variables typically require larger sample sizes to achieve stable estimates, adding another layer of complexity. However, this concern is mitigated by the robustness of the model, demonstrated through the analysis of fit indices derived from 1003 complete observations. While our model effectively captures multiple temporally dependent relationships, model complexity, coupled with the sample imbalance of those with versus without PTE, raises concerns about potential model misspecification and emphasizes the need for model validation in future work.

Conclusions

PTE in the first year after msTBI is directly and/or indirectly associated with many aspects of life, including driving ability, MH, and cognitive and motor skills, all influencing overall well-being. Managing PTE effectively during the first year after TBI, but perhaps also if seizures occur later, is crucial to prevent long-term negative effects. SEM revealed the temporal dynamics between PTE, functional outcomes, and SWL, emphasizing the need for integrated treatment approaches targeting epilepsy management alongside rehabilitation efforts. By integrating SEM results with biological data (e.g., neuroimaging or genetic markers) associated with epilepsy and TBI, personalized interventions may optimize long-term outcomes for individuals with PTE. Further research and validation studies are needed to explore these relationships and develop evidence-based interventions. A holistic approach to future work, one that considers medical, societal, and environmental factors that affect function in the setting of PTE, is necessary to improve the lives of individuals with PTE and msTBI.

Transparency, Rigor, and Reproducibility Statement

Study registration

This study is a part of the NIDILRR funded TBI Model Systems (TBIMS) Project and was not formally registered before implementation.

Analytic plan specification

While the analysis plan was not formally preregistered, the team members (A.K.W. and N.A.) with primary responsibility for the analysis certify that the analysis plan was prespecified.

Statistical power and sample size

A sample size of 1003 subjects with complete data was utilized for this study from the NIDILRR TBIMS database. Although determining the required sample size for a complex SEM is difficult, because of its relatedness to mediation models, we based our power analysis on a single mediator model for simplicity. The power of a single mediator model depends on the effect of the independent variable on the mediator (commonly denoted as α) and the effect of the mediator on the dependent variable (β), controlling for the independent variable. Based on preliminary data, we expected α to be medium-to-large and β to be small-to-medium, using Cohen’s criteria. Based on the simulation study by Fritz et al., the minimum sample size needed to detect the mediation effect in this case was 391.⁹⁶ Hair et al. provide practical guidelines for SEM, noting that as model complexity increases, the sample size should also increase. They recommend that for complex models a sample size of at least 300 is advisable, with larger samples often needed for more intricate models.⁹⁷ Westland provides formulas for determining minimum sample sizes based on model complexity. His work supports the need for larger samples (300–500 or more) for complex SEMs to achieve adequate power and model stability.⁹⁸ Based on these criteria for sample size estimation, we had adequate power to conduct these analyses.

Cohort derivation

Participants whose data were collected by the TBIMS network were used for analysis if they: (1) had msTBI, (2) had outcomes data collected between 2010 and 2018, (3) had a seizure event recorded during that time frame, and (4) had complete outcome and covariate data available. Using these inclusion criteria, a total of 5108 patients were screened. A total of 1215 subjects had complete observations for all outcome variables. Subjects totaling to 211 were missing at least one covariate, leaving data for 1003 participants to be used in the cohort.

Investigator blinding

Data collection was performed by research staff who were blinded to participant characteristics relevant to this study. Data analyses were performed by investigators who were aware of relevant participant characteristics; however, data analyses were performed on data without the use of patient identifiers.

Data acquisition, analysis, and assessment tools

The key inclusion criteria (e.g., primary diagnosis or prognostic factor) are established standards in the field. Data were collected using validated clinical tools as follows: FIM, PHQ-9, GAD-7, PART, and SWLS. Data were analyzed using a covariate-adjusted SEM, using the lavaan package (R). Four standard fit indices were used to assess goodness of fit for the SEM generated. All models are reported in the publication text. All equipment and software used to perform acquisition and analysis are widely available from The R Project for Statistical Computing.

Statistical assumptions, analysis, and multiple comparisons

Statistical analyses and review were performed by authors N.A., MS in Biostatistics, and R.G.K., PhD in Neuroepidemiology and MPH in Chronic Disease Epidemiology (respectively), with qualifications including advanced studies in biostatistics, statistical computing, and disease epidemiology. Correction for multiple comparisons was not relevant since SEM accounts for all relationships simultaneously when generating the parameter estimates and statistical significance of each individual relationship within the model.

Study replication/validation, data availability, and analytic code

Study validation by the study group is planned as a future analysis. Data from this study are available in protected repositories as follows: the TBIMS National Data and Statistical Center and Federal Interagency Traumatic Brain Injury Research Informatics System. Analytic code used to conduct the analyses presented in this study is not available in a public repository. However, code may be available by emailing the first primary author (N.A.) as of January 5, 2024.

Open access

The authors agree to provide the full content of the article on request by contacting N.A. or A.K.W.

Footnotes

Acknowledgments

The authors would like to thank Jessa Darwin for article preparation. The authors would like to thank the TBIMS National Data Center for facilitating data access and the TBIMS National Database Research Participants for their participatory contribution to this work.

Authors’ Contributions

Data integrity and access: N.A. and A.K.W. had full access to the study data, and they take responsibility for the integrity of the data and the accuracy of the data analysis. Concept and design: A.K.W. and N.A. Acquisition, analysis, or interpretation of data: A.K.W., N.A., R.G.K., and S.B.J. Drafting of the article: A.K.W., N.A., J.W., and S.B.J. Critical revision of the article for important intellectual content: N.A., J.W., R.G.K., S.B.J., K.D.O.C., M.S., R.D.Z., W.C.W., J.P.S., and A.K.W. Statistical analysis: N.A.

Author Disclosure Statement

No competing financial interests exist.

Funding Information

This article was developed with support from the Department of Defense W81XWH1810736 (A.K.W. and N.A.) and the National Institute on Disability, Independent Living, and Rehabilitation Research [NIDILRR grant numbers 90DP0041 (A.K.W.); 90DPTB0009 (K.D.O.C., R.G.K.); 90DP0033 (W.C.W.); 90DPTB0011 (R.D.Z.); 90DTB0013, 90DTB0025 (S.B.J.)]. NIDILRR is a Center within the Administration for Community Living (ACL), Department of Health and Human Services (HHS). The article contents do not necessarily represent the policy of NIDILRR, ACL, and HHS should not be taken as an endorsement by the Federal Government.

Supplementary Material

Supplementary Table S1

Supplementary Table S2

References

Maas

AIR

, Menon

, Adelson

, et al. InTBIR Participants and Investigators. Traumatic brain injury: Integrated approaches to improve prevention, clinical care, and research. Lancet Neurol, 2017; 16(12):987–1048; doi: 10.1016/S1474-4422(17)30371-X

Maas

AIR

, Menon

, Manley

, et al. InTBIR Participants and Investigators. Traumatic brain injury: Progress and challenges in prevention, clinical care, and research. Lancet Neurol, 2022; 21(11):1004–1060; doi: 10.1016/S1474-4422(22)00309-X

Taylor

, Bell

, Breiding

, et al. Traumatic brain injury-related emergency department visits, hospitalizations, and deaths - United States, 2007 and 2013. MMWR Surveill Summ, 2017; 66(9):1–16; doi: 10.1558/5/mmwr.ss6609a1

Centers for Disease Control and Prevention. CDC grand rounds: Reducing severe traumatic brain injury in the United States. MMWR Morb Mortal Wkly Rep, 2013; 62(27):549–552.

Semple

, Zamani

, Rayner

, et al. Affective, neurocognitive and psychosocial disorders associated with traumatic brain injury and post-traumatic epilepsy. Neurobiol Dis, 2019; 123:27–41; doi: 10.1016/j.nbd.2018.07.018

Annegers

. The epidemiology of epilepsy. In: The Treatment of Epilepsy: Principles and Practice. ( Wyllie

. ed.) Wolters Kluwer; 1996; pp. 165–172.

Annegers

, Hauser

, Coan

, et al. A population-based study of seizures after traumatic brain injuries. N Engl J Med, 1998; 338(1):20–24; doi: 10.1056/NEJM199801013380104

Ritter

, Wagner

, Fabio

, et al. Incidence and risk factors of posttraumatic seizures following traumatic brain injury: A Traumatic Brain Injury Model Systems study. Epilepsia, 2016; 57(12):1968–1977; doi: 10.1111/epi.13582

Caveness

, Meirowsky

, Rish

, et al. The nature of posttraumatic epilepsy. J Neurosurg, 1979; 50(5):545–553; doi: 10.3171/jns.1979.50.5.0545

10.

De Santis

, Sganzerla

, Spagnoli

, et al. Risk factors for late posttraumatic epilepsy. Acta Neurochir Suppl (Wien), 1992; 55:64–67; doi: 10.1007/978-3-7091-9233-7_17

11.

Chang

, Lowenstein

, Quality Standards Subcommittee of the American Academy of Neurology . Practice parameter: Antiepileptic drug prophylaxis in severe traumatic brain injury: Report of the Quality Standards Subcommittee of the American Academy of Neurology. Neurology, 2003; 60(1):10–16; doi: 10.1212/01.wnl.0000031432.05543.14

12.

Weiss

, Salazar

, Vance

, et al. Predicting posttraumatic epilepsy in penetrating head injury. Arch Neurol, 1986; 43(8):771–773; doi: 10.1001/archneur.1986.00520080019013

13.

Frey

. Epidemiology of posttraumatic epilepsy: A critical review. Epilepsia, 2003; 44(s10):11–17; doi: 10.1046/j.1528-1157.44.s10.4.x

14.

Juengst

, Wagner

, Ritter

, et al. Post-traumatic epilepsy associations with mental health outcomes in the first two years after moderate to severe TBI: A TBI Model Systems analysis. Epilepsy Behav, 2017; 73:240–246; doi: 10.1016/j.yebeh.2017.06.001

15.

Walker

, Ketchum

, Marwitz

, et al. Global outcome and late seizures after penetrating versus closed traumatic brain injury: A NIDRR TBI Model Systems study. J Head Trauma Rehabil, 2015; 30(4):231–240; doi: 10.1097/HTR.0000000000000127

16.

García-Morales

, de la Peña Mayor

, Kanner

. Psychiatric comorbidities in epilepsy: Identification and treatment. Neurologist, 2008; 14(6 Suppl 1):S15–S25; doi: 10.1097/01.nrl.0000340788.07672.51

17.

Selassie

, Wilson

, Martz

, et al. Epilepsy beyond seizure: A population-based study of comorbidities. Epilepsy Res, 2014; 108(2):305–315; doi: 10.1016/j.eplepsyres.2013.12.002

18.

Zeber

, Copeland

, Amuan

, et al. The role of comorbid psychiatric conditions in health status in epilepsy. Epilepsy Behav, 2007; 10(4):539–546; doi: 10.1016/j.yebeh.2007.02.008

19.

Kumlien

, Lundberg

. Seizure risk associated with neuroactive drugs: Data from the WHO adverse drug reactions database. Seizure, 2010; 19(2):69–73; doi: 10.1016/j.seizure.2009.11.005

20.

Muccigrosso

, Ford

, Benner

, et al. Cognitive deficits develop 1month after diffuse brain injury and are exaggerated by microglia-associated reactivity to peripheral immune challenge. Brain Behav Immun, 2016; 54:95–109; doi: 10.1016/j.bbi.2016.01.009

21.

Jabbarinejad

, Cohen-Zimerman

, Wagner

, et al. Determinants of caregiver burden in male patients with epilepsy following penetrating traumatic brain injury. Epilepsy Behav, 2021; 116:107768; doi: 10.1016/j.yebeh.2021.107768

22.

Milleville

, Awan

, Disanto

, et al. Early chronic systemic inflammation and associations with cognitive performance after moderate to severe TBI. Brain Behav Immun Health, 2021; 11:100185; doi: 10.1016/j.bbih.2020.100185

23.

Riggio

, Wong

. Neurobehavioral sequelae of traumatic brain injury. Mt Sinai J Med, 2009; 76(2):163–172; doi: 10.1002/msj.20097

24.

Walker

, Pickett

. Motor impairment after severe traumatic brain injury: A longitudinal multicenter study. J Rehabil Res Dev, 2007; 44(7):975–982; doi: 10.1682/jrrd.2006.12.0158

25.

Stineman

, Shea

, Jette

, et al. The functional independence measure: Tests of scaling assumptions, structure, and reliability across 20 diverse impairment categories. Arch Phys Med Rehabil, 1996; 77(11):1101–1108; doi: 10.1016/s0003-9993(96)90130-6

26.

Dodds

, Martin

, Stolov

, et al. A validation of the functional independence measurement and its performance among rehabilitation inpatients. Arch Phys Med Rehabil, 1993; 74(5):531–536; doi: 10.1016/0003-9993(93)90119-u

27.

Spitz

, Ponsford

, Rudzki

, et al. Association between cognitive performance and functional outcome following traumatic brain injury: A longitudinal multilevel examination. Neuropsychology, 2012; 26(5):604–612; doi: 10.1037/a0029239

28.

Sashika

, Takada

, Kikuchi

. Rehabilitation needs and participation restriction in patients with cognitive disorder in the chronic phase of traumatic brain injury. Medicine (Baltimore), 2017; 96(4):e5968; doi: 10.1097/MD.0000000000005968

29.

Rabinowitz

, Levin

. Cognitive sequelae of traumatic brain injury. Psychiatr Clin North Am, 2014; 37(1):1–11; doi: 10.1016/j.psc.2013.11.004

30.

Mani

, Cater

, Hudlikar

. Cognition and return to work after mild/moderate traumatic brain injury: A systematic review. Work, 2017; 58(1):51–62; doi: 10.3233/WOR-172597

31.

McGarity

, Barnett

, Lamberty

, et al. Community reintegration problems among veterans and active duty service members with traumatic brain injury. J Head Trauma Rehabil, 2017; 32(1):34–45; doi: 10.1097/HTR.0000000000000242

32.

Failla

, Juengst

, Graham

, et al. Effects of depression and antidepressant use on cognitive deficits and functional cognition following severe traumatic brain injury. J Head Trauma Rehabil, 2016; 31(6):E62–E73; doi: 10.1097/HTR.0000000000000214

33.

Novack

, Zhang

, Kennedy

, et al. Driving patterns, confidence, and perception of abilities following moderate to severe traumatic brain injury: A TBI Model System study. Brain Inj, 2021; 35(8):863–870; doi: 10.1080/02699052.2021.1934730

34.

Novack

, Zhang

, Kennedy

, et al. Return to driving after moderate-to-severe traumatic brain injury: A Traumatic Brain Injury Model System study. Arch Phys Med Rehabil, 2021; 102(8):1568–1575; doi: 10.1016/j.apmr.2021.02.006

35.

Bogner

, Bellon

, Kolakowsky-Hayner

, et al. Participation assessment with recombined tools-objective (PART-O). J Head Trauma Rehabil, 2013; 28(4):337–339; doi: 10.1097/HTR.0b013e31829af969

36.

Erler

, Juengst

, Smith

, et al. Examining driving and participation 5 years after traumatic brain injury. OTJR (Thorofare N J), 2018; 38(3):143–150; doi: 10.1177/1539449218757739

37.

Juengst

, Adams

, Bogner

, et al. Trajectories of life satisfaction after traumatic brain injury: Influence of life roles, age, cognitive disability, and depressive symptoms. Rehabil Psychol, 2015; 60(4):353–364; doi: 10.1037/rep0000056

38.

Bushnik

, Englander

, Wright

, et al. Traumatic brain injury with and without late posttraumatic seizures: What are the impacts in the post-acute phase: A NIDRR Traumatic Brain Injury Model Systems study. J Head Trauma Rehabil, 2012; 27(6):E36–E44; doi: 10.1097/HTR.0b013e318273375c

39.

Juengst

, Osborne

, Erler

, et al. Effects of fatigue, driving status, cognition, and depression on participation in a chronic sample of adults with traumatic brain injury. Physical Medicine and Rehabilitation, 2017; 1(1):1–13; doi: 10.3153/2/PhysMedRehabil.1.1.004

40.

Rabinowitz

, Chervoneva

, Hart

, et al. Heterogeneity in temporal ordering of depression and participation after traumatic brain injury. Arch Phys Med Rehabil, 2020; 101(11):1973–1979; doi: 10.1016/j.apmr.2020.05.026

41.

Boosman

, Winkens

, van Heugten

, et al. Predictors of health-related quality of life and participation after brain injury rehabilitation: The role of neuropsychological factors. Neuropsychol Rehabil, 2017; 27(4):581–598; doi: 10.1080/09602011.2015.1113996

42.

Erler

, Whiteneck

, Juengst

, et al. Predicting the trajectory of participation after traumatic brain injury: A longitudinal analysis. J Head Trauma Rehabil, 2018; 33(4):257–265; doi: 10.1097/HTR.0000000000000383

43.

Perrin

, Stevens

, Sutter

, et al. Reciprocal causation between functional independence and mental health 1 and 2 years after traumatic brain injury: A cross-lagged panel structural equation model. Am J Phys Med Rehabil, 2017; 96(6):374–380; doi: 10.1097/PHM.0000000000000644

44.

Kang

. People with epilepsy have poor life satisfaction and self-rated health: Findings from the United Kingdom. Front Psychol, 2022; 13:986520; doi: 10.3389/fpsyg.2022.986520

45.

Kobau

, Luncheon

, Zack

, et al. Satisfaction with life domains in people with epilepsy. Epilepsy Behav, 2012; 25(4):546–551; doi: 10.1016/j.yebeh.2012.09.013

46.

Elliott

, Richardson

. The biopsychosocial model and quality of life in persons with active epilepsy. Epilepsy Behav, 2014; 41:55–65; doi: 10.1016/j.yebeh.2014.09.035

47.

Mellick

. Traumatic Brain Injury Model Systems National Data and Statistical Center. 2018. Available from: https://osf.io/a4xzb/ [Last accessed: January 1 ].

48.

Dijkers

, Harrison-Felix

, Marwitz

. The traumatic brain injury model systems: History and contributions to clinical service and research. J Head Trauma Rehabil, 2010; 25(2):81–91; doi: 10.1097/HTR.0b013e3181cd3528

49.

Rappaport

, Hall

, Hopkins

, et al. Disability rating scale for severe head trauma: Coma to community. Arch Phys Med Rehabil, 1982; 63(3):118–123.

50.

Ottenbacher

, Hsu

, Granger

, et al. The reliability of the functional independence measure: A quantitative review. Arch Phys Med Rehabil, 1996; 77(12):1226–1232; doi: 10.1016/s0003-9993(96)90184-7

51.

Keith

, Granger

, Hamilton

, et al. The functional independence measure: A new tool for rehabilitation. Adv Clin Rehabil, 1987; 1:6–18.

52.

Heinemann

, Michael Linacre

, Wright

, et al. Measurement characteristics of the functional independence measure. Top Stroke Rehabil, 1994; 1(3):1–15; doi: 10.1080/10749357.1994.11754030

53.

Linacre

, Heinemann

, Wright

, et al. The structure and stability of the Functional Independence Measure. Arch Phys Med Rehabil, 1994; 75(2):127–132; doi: 10.1016/0003-9993(94)90384-0

54.

Wright

, Linacre

. Observations are always ordinal; measurements, however, must be interval. Arch Phys Med Rehabil, 1989; 70(12):857–860.

55.

Fann

, Berry

, Wolpin

, et al. Depression screening using the Patient Health Questionnaire-9 administered on a touch screen computer. Psychooncology, 2009; 18(1):14–22; doi: 10.1002/pon.1368

56.

Fann

, Bombardier

, Dikmen

, et al. Validity of the Patient Health Questionnaire-9 in assessing depression following traumatic brain injury. J Head Trauma Rehabil, 2005; 20(6):501–511; doi: 10.1097/00001199-200511000-00003

57.

Spitzer

, Kroenke

, Williams

, et al. A brief measure for assessing generalized anxiety disorder: The GAD-7. Arch Intern Med, 2006; 166(10):1092–1097; doi: 10.1001/archinte.166.10.1092

58.

Bogner

, Whiteneck

, Corrigan

, et al. Comparison of scoring methods for the participation assessment with recombined tools-objective. Arch Phys Med Rehabil, 2011; 92(4):552–563; doi: 10.1016/j.apmr.2010.11.014

59.

Diener

, Emmons

, Larsen

, et al. The satisfaction with life scale. J Pers Assess, 1985; 49(1):71–75; doi: 10.1207/s15327752jpa4901_13

60.

Duncan

. Introduction to Structural Equation Models. Academic Press: New York; 1975.

61.

Nash

. On best practice optimization methods in R. J Stat Soft, 2014; 60(2):1–14; doi: 10.1863/7/jss.v060.i02

62.

Cohen

. A power primer. Psychol Bull, 1992; 112(1):155–159; doi: 10.1037/0033-2909.112.1.155

63.

, Bentler

. Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Struct Equ Modeling, 1999; 6(1):1–55; doi: 10.1080/10705519909540118

64.

Rosseel

. lavaan: An R package for structural equation modeling. J Stat Soft, 2012; 48(2):1–36; doi: 10.1863/7/jss.v048.i02

65.

Gugger

, Kennedy

, Panahi

, et al. Multimodal quality of life assessment in post-9/11 veterans with epilepsy: Impact of drug resistance, traumatic brain injury, and comorbidity. Neurology, 2022; 98(17):e1761–e1770; doi: 10.1212/wnl.0000000000200146

66.

Leonardi

, Lee

, Kostanjsek

, et al. 20 Years of ICF-International Classification of Functioning, Disability and Health: Uses and applications around the world. Int J Environ Res Public Health, 2022; 19(18); doi: 10.3390/ijerph191811321

67.

Rauch

, Cieza

, Stucki

. How to apply the International Classification of Functioning, Disability and Health (ICF) for rehabilitation management in clinical practice. Eur J Phys Rehabil Med, 2008; 44(3):329–342.

68.

Bramlett

, Dietrich

. Long-Term consequences of traumatic brain injury: Current status of potential mechanisms of injury and neurological outcomes. J Neurotrauma, 2015; 32(23):1834–1848; doi: 10.1089/neu.2014.3352

69.

Hammond

, Perkins

, Corrigan

, et al. Functional change from five to fifteen years after traumatic brain injury. J Neurotrauma, 2021; 38(7):858–869; doi: 10.1089/neu.2020.7287

70.

Braaf

, Ameratunga

, Christie

, et al. Care coordination experiences of people with traumatic brain injury and their family members in the 4-years after injury: A qualitative analysis. Brain Inj, 2019; 33(5):574–583; doi: 10.1080/02699052.2019.1566835

71.

Ponsford

, Harrison-Felix

, Ketchum

, et al. Outcomes 1 and 2 years after moderate to severe traumatic brain injury: An international comparative study. Arch Phys Med Rehabil, 2021; 102(3):371–377; doi: 10.1016/j.apmr.2020.09.387

72.

Cho

, Park

, Baker

, et al. Post-traumatic epilepsy in zebrafish is drug-resistant and impairs cognitive function. J Neurotrauma, 2021; 38(22):3174–3183; doi: 10.1089/neu.2021.0156

73.

Lowenstein

. Epilepsy after head injury: An overview. Epilepsia, 2009; 50 (Suppl 2):4–9; doi: 10.1111/j.1528-1167.2008.02004.x

74.

Ortinski

, Meador

. Cognitive side effects of antiepileptic drugs. Epilepsy Behav, 2004; 5 (Suppl 1):S60–S5; doi: 10.1016/j.yebeh.2003.11.008

75.

Witt

, Helmstaedter

. How can we overcome neuropsychological adverse effects of antiepileptic drugs? Expert Opin Pharmacother, 2017; 18(6):551–554; doi: 10.1080/14656566.2017.1309025

76.

Loring

, Marino

, Meador

. Neuropsychological and behavioral effects of antiepilepsy drugs. Neuropsychol Rev, 2007; 17(4):413–425; doi: 10.1007/s11065-007-9043-9

77.

Raymont

, Salazar

, Lipsky

, et al. Correlates of posttraumatic epilepsy 35 years following combat brain injury. Neurology, 2010; 75(3):224–229; doi: 10.1212/WNL.0b013e3181e8e6d0

78.

Burke

, Gugger

, Ding

, et al. TRACK-TBI Investigators. Association of posttraumatic epilepsy with 1-year outcomes after traumatic brain injury. JAMA Netw Open, 2021; 4(12):e2140191; doi: 10.1001/jamanetworkopen.2021.40191

79.

Haltiner

, Temkin

, Winn

, et al. The impact of posttraumatic seizures on 1-year neuropsychological and psychosocial outcome of head injury. J Int Neuropsychol Soc, 1996; 2(6):494–504; doi: 10.1017/s1355617700001661

80.

Park

, Kwon

. Cognitive effects of antiepileptic drugs. J Clin Neurol, 2008; 4(3):99–106; doi: 10.3988/jcn.2008.4.3.99

81.

Dikmen

, Temkin

, Miller

, et al. Neurobehavioral effects of phenytoin prophylaxis of posttraumatic seizures. JAMA, 1991; 265(10):1271–1277; doi: 10.1001/jama.1991.03460100073027

82.

Eddy

, Rickards

, Cavanna

. The cognitive impact of antiepileptic drugs. Ther Adv Neurol Disord, 2011; 4(6):385–407; doi: 10.1177/1756285611417920

83.

Witt

, Helmstaedter

. Cognition in the early stages of adult epilepsy. Seizure, 2015; 26:65–68; doi: 10.1016/j.seizure.2015.01.018

84.

Lin

, Han

, Xue

, et al. Predicting cognitive impairment in outpatients with epilepsy using machine learning techniques. Sci Rep, 2021; 11(1):20002; doi: 10.1038/s41598-021-99506-3

85.

Karakis

, Cole

, Montouris

, et al. Caregiver burden in epilepsy: Determinants and impact. Epilepsy Res Treat, 2014; 2014:808421; doi: 10.1155/2014/808421

86.

Hoxhaj

, Habiya

, Sayabugari

, et al. Investigating the impact of epilepsy on cognitive function: A narrative review. Cureus, 2023; 15(6):e41223; doi: 10.7759/cureus.41223

87.

Khalife

, Scott

, Hernan

. Mechanisms for cognitive impairment in epilepsy: Moving beyond seizures. Front Neurol, 2022; 13:878991; doi: 10.3389/fneur.2022.878991

88.

Kolakowsky-Hayner

, Wright

, Englander

, et al. Impact of late post-traumatic seizures on physical health and functioning for individuals with brain injury within the community. Brain Inj, 2013; 27(5):578–586; doi: 10.3109/02699052.2013.765595

89.

Pingue

, Mele

, Biscuola

, et al. Impact of seizures and their prophylaxis with antiepileptic drugs on rehabilitation course of patients with traumatic or hemorrhagic brain injury. Front Neurol, 2022; 13:1060008; doi: 10.3389/fneur.2022.1060008

90.

Kanner

. Management of psychiatric and neurological comorbidities in epilepsy. Nat Rev Neurol, 2016; 12(2):106–116; doi: 10.1038/nrneurol.2015.243

91.

Classen

, Crizzle

, Winter

, et al. Evidence-based review on epilepsy and driving. Epilepsy Behav, 2012; 23(2):103–112; doi: 10.1016/j.yebeh.2011.11.015

92.

Gilliam

, Hecimovic

, Sheline

. Psychiatric comorbidity, health, and function in epilepsy. Epilepsy Behav, 2003; 4(Suppl 4):S26–S30; doi: 10.1016/j.yebeh.2003.10.003

93.

Pingue

, Mele

, Nardone

. Post-traumatic seizures and antiepileptic therapy as predictors of the functional outcome in patients with traumatic brain injury. Sci Rep, 2021; 11(1):4708; doi: 10.1038/s41598-021-84203-y

94.

Fordington

, Manford

. A review of seizures and epilepsy following traumatic brain injury. J Neurol, 2020; 267(10):3105–3111; doi: 10.1007/s00415-020-09926-w

95.

Corrigan

, Cuthbert

, Whiteneck

, et al. Representativeness of the traumatic brain injury model systems national database. J Head Trauma Rehabil, 2012; 27(6):391–403; doi: 10.1097/HTR.0b013e3182238cdd

96.

Fritz

, Mackinnon

. Required sample size to detect the mediated effect. Psychol Sci, 2007; 18(3):233–239; doi: 10.1111/j.1467-9280.2007.01882.x

97.

Hair

, Babin

, Black

, et al. Multivariate Data Analysis. Cengage; 2019.

98.

Westland

. Lower bounds on sample size in structural equation modeling. Electronic Commerce Research and Applications, 2010; 9(6):476–487; doi: 10.1016/j.elerap.2010.07.003

99.

TBIMS. Traumatic Brain Injury Model Systems National Data and Statistical Center. 2020. Available from: https://doi.org/10.17605/OSF.IO/A4XZB [Last Accessed; 17 December 2024].

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.02 MB