Search

Purpose: Unsupervised item-response theory (IRT) models such as polytomous IRT based on recursive partitioning (IRTrees) and mixture IRT (MixIRT) models can be used to assess differential item functioning (DIF) in patient-reported outcome measures (PROMs) when the covariates associated with DIF are unknown a priori. This study examines the consistency of results for IRTrees and MixIRT models. Methods: Data were from 4478 individuals in the Alberta Provincial Project on Outcome Assessment in Coronary Heart Disease registry who received cardiac angiography in Alberta, Canada, and completed the Hospital Anxiety and Depression Scale (HADS) depression subscale items. The partial credit model (PCM) based on recursive partitioning (PCTree) and mixture PCM (MixPCM) were used to identify covariates associated with differential response patterns to HADS depression subscale items. Model covariates included demographic and clinical characteristics. Results: The median (interquartile range) age was 64.5(15.7) years, and 3522(78.5%) patients were male. The PCTree identified 4 terminal nodes (subgroups) defined by smoking status, age, and body mass index. A 3-class PCM fits the data well. The MixPCM latent classes were defined by age, disease indication, smoking status, comorbid diabetes, congestive heart failure, and chronic obstructive pulmonary disease. Conclusion: PCTree and MixPCM were not consistent in detecting covariates associated with differential interpretations of PROM items. Future research will use computer simulations to assess these models' Type I error and statistical power for identifying covariates associated with DIF. [ABSTRACT FROM AUTHOR]

Purpose: Patients with coronary artery disease (CAD) experience significant angina symptoms and lifestyle changes. Revascularization procedures can result in better patient-reported outcomes (PROs) than optimal medical therapy (OMT) alone. This study evaluates the impact of response shift (RS) on changes in PROs of patients with CAD across treatment strategies. Methods: Data were from patients with CAD in the Alberta Provincial Project on Outcome Assessment in Coronary Heart Disease (APPROACH) registry who completed the 16-item Canadian version of the Seattle Angina Questionnaire at 2 weeks and 1 year following a coronary angiogram. Multi-group confirmatory factor analysis (MG-CFA) was used to assess measurement invariance across treatment groups at week 2. Longitudinal MG-CFA was used to test for RS according to receipt of coronary artery bypass grafting (CABG), percutaneous coronary intervention (PCI), or optimal medical therapy (OMT) alone. Results: Of the 3116 patients included in the analysis, 443 (14.2%) received CABG, 2049(65.8%) PCI, and the remainder OMT alone. The MG-CFA revealed a partial-strong invariance across the treatment groups at 2 weeks (CFI = 0.98, RMSEA [90% CI] = 0.05 [0.03, 0.06]). Recalibration RS was detected on the Angina Symptoms and Burden subscale and its magnitude in the OMT, PCI, and CABG groups were 0.32, 0.28, and 0.53, respectively. After adjusting for RS effects, the estimated target changes were largest in the CABG group and negligible in the OMT group. Conclusion: Adjusting for RS is recommended in studies that use SAQ-CAN to assess changes in patients with CAD who have received revascularization versus OMT alone. [ABSTRACT FROM AUTHOR]

Purpose: The objective of this systematic review was to describe the prevalence and magnitude of response shift effects, for different response shift methods, populations, study designs, and patient-reported outcome measures (PROM)s. Methods: A literature search was performed in MEDLINE, PSYCINFO, CINAHL, EMBASE, Social Science Citation Index, and Dissertations & Theses Global to identify longitudinal quantitative studies that examined response shift using PROMs, published before 2021. The magnitude of each response shift effect (effect sizes, R-squared or percentage of respondents with response shift) was ascertained based on reported statistical information or as stated in the manuscript. Prevalence and magnitudes of response shift effects were summarized at two levels of analysis (study and effect levels), for recalibration and reprioritization/reconceptualization separately, and for different response shift methods, and population, study design, and PROM characteristics. Analyses were conducted twice: (a) including all studies and samples, and (b) including only unrelated studies and independent samples. Results: Of the 150 included studies, 130 (86.7%) detected response shift effects. Of the 4868 effects investigated, 793 (16.3%) revealed response shift. Effect sizes could be determined for 105 (70.0%) of the studies for a total of 1130 effects, of which 537 (47.5%) resulted in detection of response shift. Whereas effect sizes varied widely, most median recalibration effect sizes (Cohen's d) were between 0.20 and 0.30 and median reprioritization/reconceptualization effect sizes rarely exceeded 0.15, across the characteristics. Similar results were obtained from unrelated studies. Conclusion: The results draw attention to the need to focus on understanding variability in response shift results: Who experience response shifts, to what extent, and under which circumstances? [ABSTRACT FROM AUTHOR]

Purpose: Because physical-mental comorbidity in children is relatively common, this study tested for response shift (RS) in children with chronic physical illness using a parent-reported measure of child psychopathology. Methods: Data come from Multimorbidity in Children and Youth across Life-course (MY LIFE), a prospective study of n = 263 children aged 2–16 years with physical illness in Canada. Parents provided information on child psychopathology using the Ontario Child Health Study Emotional Behavioral Scales (OCHS-EBS) at baseline and 24 months. Oort's structural equation modeling was used to test for different forms of RS in parent-reported assessments between baseline and 24 months. Model fit was evaluated using root mean square error of approximation (RMSEA), comparative fit index (CFI), and standardized root mean residual (SRMR). Results: There were n = 215 (81.7%) children with complete data and were included in this analysis. Of these, n = 105 (48.8%) were female and the mean (SD) age was 9.4 (4.2) years. A two-factor measurement model provided good fit to the data [RMSEA (90% CI) = 0.05 (0.01, 0.10); CFI = 0.99; SRMR = 0.03]. Non-uniform recalibration RS was detected on the conduct disorder subscale of the OCHS-EBS. This RS effect had negligible impact on the longitudinal change in externalizing and internalizing disorders construct over time. Conclusions: Response shift detected on the conduct disorder subscale of the OCHS-EBS, indicated that parents of children with physical illness may recalibrate their responses on child psychopathology over 24 months. Researchers and health professionals should be aware of RS when using the OCHS-EBS to assess child psychopathology over time. [ABSTRACT FROM AUTHOR]

Purpose: Mixture item response theory (MixIRT) models can be used to uncover heterogeneity in responses to items that comprise patient-reported outcome measures (PROMs). This is accomplished by identifying relatively homogenous latent subgroups in heterogeneous populations. Misspecification of the number of latent subgroups may affect model accuracy. This study evaluated the impact of specifying too many latent subgroups on the accuracy of MixIRT models. Methods: Monte Carlo methods were used to assess MixIRT accuracy. Simulation conditions included number of items and latent classes, class size ratio, sample size, number of non-invariant items, and magnitude of between-class difference in item parameters. Bias and mean square error in item parameters and accuracy of latent class recovery were assessed. Results: When the number of latent classes was correctly specified, the average bias and MSE in model parameters decreased as the number of items and latent classes increased, but specification of too many latent classes resulted in modest decrease (i.e., < 10%) in the accuracy of latent class recovery. Conclusion: The accuracy of MixIRT model is largely influenced by the overspecification of the number of latent classes. Appropriate choice of goodness-of-fit measures, study design considerations, and a priori contextual understanding of the degree of sample heterogeneity can guide model selection. [ABSTRACT FROM AUTHOR]

Purpose: Item non-response (i.e., missing data) may mask the detection of differential item functioning (DIF) in patient-reported outcome measures or result in biased DIF estimates. Non-response can be challenging to address in ordinal data. We investigated an unsupervised machine-learning method for ordinal item-level imputation and compared it with commonly-used item non-response methods when testing for DIF. Methods: Computer simulation and real-world data were used to assess several item non-response methods using the item response theory likelihood ratio test for DIF. The methods included: (a) list-wise deletion (LD), (b) half-mean imputation (HMI), (c) full information maximum likelihood (FIML), and (d) non-negative matrix factorization (NNMF), which adopts a machine-learning approach to impute missing values. Control of Type I error rates were evaluated using a liberal robustness criterion for α = 0.05 (i.e., 0.025–0.075). Statistical power was assessed with and without adoption of an item non-response method; differences > 10% were considered substantial. Results: Type I error rates for detecting DIF using LD, FIML and NNMF methods were controlled within the bounds of the robustness criterion for > 95% of simulation conditions, although the NNMF occasionally resulted in inflated rates. The HMI method always resulted in inflated error rates with 50% missing data. Differences in power to detect moderate DIF effects for LD, FIML and NNMF methods were substantial with 50% missing data and otherwise insubstantial. Conclusion: The NNMF method demonstrated comparable performance to commonly-used non-response methods. This computationally-efficient method represents a promising approach to address item-level non-response when testing for DIF. [ABSTRACT FROM AUTHOR]

Purpose: The Seattle Angina Questionnaire (SAQ) is a widely used patient-reported measure of health status in patients with coronary artery disease. Comparisons of SAQ scores amongst population groups and over time rely on the assumption that its factorial structure is invariant. This study evaluates the measurement invariance of the SAQ across different demographic and clinical groups and over time. Methods: Data were obtained from the Alberta Provincial Project on Outcome Assessment in Coronary Heart Disease registry, a registry of patients who received coronary angiogram in Alberta, Canada. The study cohort consists of adult patients who completed the paper-based version of the 16-item Canadian version of the SAQ (SAQ-CAN) 2 weeks and 1-year post-coronary angiogram between 2009 and 2016. Multi-group confirmatory factor analysis was used to assess configural, weak, strong, and strict measurement invariance across age groups, sex, angina type, treatment, and over time. Model fit was assessed using the comparative fit index and root mean square error of approximation. Results: Of the 8101 patients included in these analysis, 1300 (16.1%) were at least 75 years old, while 1755 (21.7%) were female, 5154 (63.6%) were diagnosed with acute coronary syndrome, 1177 (14.5%) received coronary artery bypass graft treatment, and 3279 had complete data on the SAQ-CAN at both occasions. There was evidence of strict invariance across age, sex, and angina type, and treatment groups, but partial strict invariance was established over time. Conclusion: SAQ-CAN can be used to compare the health status of coronary artery disease patients across population groups and over time. [ABSTRACT FROM AUTHOR]

Purpose: This work is part of an international, interdisciplinary initiative to synthesize research on response shift in results of patient-reported outcome measures. The objective is to critically examine current response shift methods. We additionally propose advancing new methods that address the limitations of extant methods. Methods: Based on literature reviews, this critical examination comprises design-based, qualitative, individualized, and preference-based methods, latent variable models, and other statistical methods. We critically appraised their definition, operationalization, the type of response shift they can detect, whether they can adjust for and explain response shift, their assumptions, and alternative explanations. Overall limitations requiring new methods were identified. Results: We examined 11 methods that aim to operationalize response shift, by assessing change in the meaning of one's self-evaluation. Six of these methods distinguish between change in observed measurements (observed change) and change in the construct that was intended to be measured (target change). The methods use either (sub)group-based or individual-level analysis, or a combination. All methods have underlying assumptions to be met and alternative explanations for the inferred response shift effects. We highlighted the need to address the interpretation of the results as response shift and proposed advancing new methods handling individual variation in change over time and multiple time points. Conclusion: No single response shift method is optimal; each method has strengths and limitations. Additionally, extra steps need to be taken to correctly interpret the results. Advancing new methods and conducting computer simulation studies that compare methods are recommended to move response shift research forward. [ABSTRACT FROM AUTHOR]

Purpose: Perceived social support is known to be an important predictor of health outcomes in patients with acute coronary syndrome (ACS). This study investigates patterns of longitudinal trajectories of patient-reported perceived social support in individuals with ACS.Methods: Data are from 3013 patients from the Alberta Provincial Project for Outcome Assessment in Coronary Heart Disease registry who had their first cardiac catheterization between 2004 and 2011. Perceived social support was assessed using the 19-item Medical Outcomes Study Social Support Survey (MOS) 2 weeks, 1 year, and 3 years post catheterization. Group-based trajectory analysis based on longitudinal multiple imputation model was used to identify distinct subgroups of trajectories of perceived social support over a 3-year follow-up period.Results: Three distinct social support trajectory subgroups were identified, namely: "High" social support group (60%), "Intermediate" social support group (30%), and "Low" social support subgroup (10%). Being female (OR = 1.67; 95% CI = [1.18-2.36]), depression (OR = 8.10; 95% CI = [4.27-15.36]) and smoking (OR = 1.70; 95% CI = [1.23-2.35]) were predictors of the differences among these trajectory subgroups.Conclusion: Although the majority of ACS patients showed increased or fairly stable trajectories of social support, about 10% of the cohort reported declining social support. These findings can inform targeted psycho-social interventions to improve their perceived social support and health outcomes. [ABSTRACT FROM AUTHOR]

Purpose: This study describes the characteristics and quality of reporting for published computer simulation studies about statistical methods to analyze complex longitudinal (i.e., repeated measures) patient-reported outcomes (PROs); we included methods for longitudinal latent variable measurement and growth models and response shift.Methods: Scopus, PsycINFO, PubMed, EMBASE, and Social Science Citation Index were searched for English-language studies published between 1999 and 2016 using selected keywords. Extracted information included characteristics of the study purpose/objectives, simulation design, software, execution, performance, and results. The quality of reporting was evaluated using published best-practice guidelines.Synthesis: A total of 1470 articles were reviewed and 42 articles met the inclusion criteria. The majority of the included studies (73.8%) investigated an existing statistical method, primarily a latent variable model (95.2%). Most studies specified the population model, including variable distributions, mean parameters, and correlation/covariances. The number of time points and sample size(s) were reported by all studies, but justification for the selected values was rarely provided. The majority of the studies (52.4%) did not report on model non-convergence. Bias, accuracy, and model fit were commonly reported performance metrics. All studies reported results descriptively, and 26.2% also used an inferential method.Conclusions: While methodological research on statistical analyses of complex longitudinal PRO data is informed by computer simulation studies, current reporting practices of these studies have not been consistent with best-practice guidelines. Comprehensive reporting of simulation methods and results ensures that the strengths and limitations of the investigated statistical methods are thoroughly explored. [ABSTRACT FROM AUTHOR]

Purpose: Patient-reported outcome measures (PROMs) are frequently used in heterogeneous patient populations. PROM scores may lead to biased inferences when sources of heterogeneity (e.g., gender, ethnicity, and social factors) are ignored. Latent variable mixture models (LVMMs) can be used to examine measurement invariance (MI) when sources of heterogeneity in the population are not known a priori. The goal of this article is to discuss the use of LVMMs to identify invariant items within the context of test construction.Methods: The Draper-Lindely-de Finetti (DLD) framework for the measurement of latent variables provides a theoretical context for the use of LVMMs to identify the most invariant items in test construction. In an expository analysis using 39 items measuring daily activities, LVMMs were conducted to compare 1- and 2-class item response theory models (IRT). If the 2-class model had better fit, item-level logistic regression differential item functioning (DIF) analyses were conducted to identify items that were not invariant. These items were removed and LVMMs and DIF testing repeated until all remaining items showed MI.Results: The 39 items had an essentially unidimensional measurement structure. However, a 1-class IRT model resulted in many statistically significant bivariate residuals, indicating suboptimal fit due to remaining local dependence. A 2-class LVMM had better fit. Through subsequent rounds of LVMMs and DIF testing, nine items were identified as being most invariant.Conclusions: The DLD framework and the use of LVMMs have significant potential for advancing theoretical developments and research on item selection and the development of PROMs for heterogeneous populations. [ABSTRACT FROM AUTHOR]

Background: Response shift (RS) has been defined as a change in the meaning of an individual's self-evaluation of his/her health status and quality of life. Several statistical model- and design-based methods have been developed to test for RS in longitudinal data. We reviewed the uptake of these methods in patient-reported outcomes (PRO) literature.Methods: CINHAHL, EMBASE, Medline, ProQuest, PsycINFO, and Web of Science were searched to identify English-language articles about RS published until 2016. Data on year and country of publication, PRO measure adopted, RS detection method, type of RS detected, and testing of underlying model assumptions were extracted from the included articles.Results: Of the 1032 articles identified, 101 (9.8%) articles were included in the study. While 54.5 of the articles reported on the Then-test, 30.7% of the articles reported on Oort's or Schmitt's structural equation modeling (SEM) procedure. Newer RS detection methods, such as relative importance analysis and random forest regression, have been used less frequently. Less than 25% reported on testing the assumptions underlying the adopted RS detection method(s).Conclusions: Despite rapid methodological advancements in RS research, this review highlights the need for further research about RS detection methods for complex longitudinal data and standardized reporting guidelines. [ABSTRACT FROM AUTHOR]

Purpose: Response shift (RS) may mask true change in health-related quality of life in longitudinal studies. People with chronic conditions may experience RS as they adapt to their disease, but it is unknown whether fluctuations in disease activity will influence the presence of RS. The study purpose was to test for RS in individuals with inflammatory bowel disease (IBD), a condition characterized by periods of symptom flares and remission.Methods: Data were from the Manitoba IBD Cohort Study (N = 388). Multi-group confirmatory factor analysis (MG-CFA) and a RS detection method based on structural equation modeling were used to test for reconceptualization, reprioritization, and recalibration RS in participants with consistent active, consistent inactive, and inconsistent disease activity over a 6-month period on the SF-36.Results: The MG-CFA revealed that a weak invariance model with equal factor loadings across groups was the best fit to the baseline SF-36 data. Reconceptualization, uniform recalibration, and non-uniform recalibration RS was detected in the consistent active group, but effect sizes were small. For the consistent inactive group, recalibration RS was observed and effect sizes were small to moderate. For the inconsistent disease activity group, small-to-moderate recalibration RS effects were observed. There was no evidence of reprioritization.Conclusions: Individuals with a chronic disease may exhibit RS even if they are not actively experiencing symptoms on a consistent basis. Heterogeneity in the type and magnitude of RS effects may be observed in chronic disease patients who experience changes in disease symptoms. [ABSTRACT FROM AUTHOR]

Background: When patients experience a substantial health state change, they may undergo changes in the underlying meaning of their self-report to standardized outcome measures. These response shifts can reflect changes in the patient’s internal standards, values or conceptualization of quality of life. We investigated the presence of changing values (reprioritization response shift) in a longitudinal cohort of spine surgery patients. Methods: Spinal decompression surgery patients (mean age 52 years; 39 % female, 36 % working) provided visual analogue scale (VAS) back and leg pain items, the Short-Form-36 (SF-36v1), and the Oswestry Disability Index (ODI) data pre- and post-surgery ( npre = 169; n6weeks = 102; n3months = 106; n6months = 68). Improved and No-Effect patient groups were compared using the VAS minimally important difference (±15 points) as a cutoff. Reprioritization response shift detection was based on change in the relative importance of the SF-36 domains for group discrimination pre- and post-surgery. Results: The Improved group evidenced significant post-surgery differences from the No-Effect group on bodily pain, general health, physical functioning, social functioning, vitality, and the ODI. The relative importance analysis showed a differential effect with bodily pain ( p < 0.01) and physical functioning ( p < 0.05) becoming more important, and role physical ( p < 0.01) becoming less important post-surgery in distinguishing the Improved group as compared to the No-Effect group. The Improved patients also evidenced stronger associations between bodily pain and physical functioning, vitality and general health ( p < 0.05). The No-Effect group evidenced increased inter-correlations of bodily pain with social functioning, mental health, and general health ( p < 0.05). Conclusions: Patients who report clinically significant change in leg and back pain post-surgery using VAS pain scores are also distinguished by increased importance of bodily pain and physical functioning, and decreased importance of role physical. Bodily pain is primarily reflective of physical item response post-surgery among Improved patients, but reflects physical, social, and emotional item response among No-Effect patients. These changes in values may reflect a “moving goal post” in outcome assessment that complicates the interpretation of mean differences over time on standard spine outcome measures. [ABSTRACT FROM AUTHOR]

Purpose: Patient-reported outcome measures (PROMs) such as the Neurological Disorders Depression Inventory in Epilepsy (NDDI-E), a 6-item epilepsy-specific PROM, is used to screen for major depressive disorder symptoms for patients with epilepsy (PWE). The validity and interpretation of PROMs can be affected by differential item functioning (DIF), which occurs when subgroups of patients with the same underlying health status respond to and interpret questions about their health status differently. This study aims to determine whether NDDI-E items exhibit DIF and to identify subgroups of PWE that exhibit DIF in NDDI-E items.Data were from the Calgary Comprehensive Epilepsy Program database, a clinical registry of adult PWE in Calgary, Canada. A tree-based partial credit model based on recursive partitioning (PCTree) was used to identify subgroups that exhibit DIF on NDDI-E items using patients’ characteristics as covariates. Differences in the identified subgroups were characterized using multinomial logistic regression.Of the 1,576 patients in this cohort, 806 (51.1%) were female, and the median age was 38.0 years. PCTree identified four patient subgroups defined by employment status, age, and sex. Subgroup 1 were unemployed patients ≤ 26 years old, subgroup 2 were unemployed patients > 26 years, subgroup 3 were employed females, while subgroup 4 were employed male patients. The subgroups exhibited significant differences on education level, comorbidity index scores, marital status, type of epilepsy, and driving status.PWE differed in their interpretation and responses to questions about their depression symptoms, and these differences were a function of sociodemographic and clinical characteristics.Methods: Patient-reported outcome measures (PROMs) such as the Neurological Disorders Depression Inventory in Epilepsy (NDDI-E), a 6-item epilepsy-specific PROM, is used to screen for major depressive disorder symptoms for patients with epilepsy (PWE). The validity and interpretation of PROMs can be affected by differential item functioning (DIF), which occurs when subgroups of patients with the same underlying health status respond to and interpret questions about their health status differently. This study aims to determine whether NDDI-E items exhibit DIF and to identify subgroups of PWE that exhibit DIF in NDDI-E items.Data were from the Calgary Comprehensive Epilepsy Program database, a clinical registry of adult PWE in Calgary, Canada. A tree-based partial credit model based on recursive partitioning (PCTree) was used to identify subgroups that exhibit DIF on NDDI-E items using patients’ characteristics as covariates. Differences in the identified subgroups were characterized using multinomial logistic regression.Of the 1,576 patients in this cohort, 806 (51.1%) were female, and the median age was 38.0 years. PCTree identified four patient subgroups defined by employment status, age, and sex. Subgroup 1 were unemployed patients ≤ 26 years old, subgroup 2 were unemployed patients > 26 years, subgroup 3 were employed females, while subgroup 4 were employed male patients. The subgroups exhibited significant differences on education level, comorbidity index scores, marital status, type of epilepsy, and driving status.PWE differed in their interpretation and responses to questions about their depression symptoms, and these differences were a function of sociodemographic and clinical characteristics.Results: Patient-reported outcome measures (PROMs) such as the Neurological Disorders Depression Inventory in Epilepsy (NDDI-E), a 6-item epilepsy-specific PROM, is used to screen for major depressive disorder symptoms for patients with epilepsy (PWE). The validity and interpretation of PROMs can be affected by differential item functioning (DIF), which occurs when subgroups of patients with the same underlying health status respond to and interpret questions about their health status differently. This study aims to determine whether NDDI-E items exhibit DIF and to identify subgroups of PWE that exhibit DIF in NDDI-E items.Data were from the Calgary Comprehensive Epilepsy Program database, a clinical registry of adult PWE in Calgary, Canada. A tree-based partial credit model based on recursive partitioning (PCTree) was used to identify subgroups that exhibit DIF on NDDI-E items using patients’ characteristics as covariates. Differences in the identified subgroups were characterized using multinomial logistic regression.Of the 1,576 patients in this cohort, 806 (51.1%) were female, and the median age was 38.0 years. PCTree identified four patient subgroups defined by employment status, age, and sex. Subgroup 1 were unemployed patients ≤ 26 years old, subgroup 2 were unemployed patients > 26 years, subgroup 3 were employed females, while subgroup 4 were employed male patients. The subgroups exhibited significant differences on education level, comorbidity index scores, marital status, type of epilepsy, and driving status.PWE differed in their interpretation and responses to questions about their depression symptoms, and these differences were a function of sociodemographic and clinical characteristics.Conclusion: Patient-reported outcome measures (PROMs) such as the Neurological Disorders Depression Inventory in Epilepsy (NDDI-E), a 6-item epilepsy-specific PROM, is used to screen for major depressive disorder symptoms for patients with epilepsy (PWE). The validity and interpretation of PROMs can be affected by differential item functioning (DIF), which occurs when subgroups of patients with the same underlying health status respond to and interpret questions about their health status differently. This study aims to determine whether NDDI-E items exhibit DIF and to identify subgroups of PWE that exhibit DIF in NDDI-E items.Data were from the Calgary Comprehensive Epilepsy Program database, a clinical registry of adult PWE in Calgary, Canada. A tree-based partial credit model based on recursive partitioning (PCTree) was used to identify subgroups that exhibit DIF on NDDI-E items using patients’ characteristics as covariates. Differences in the identified subgroups were characterized using multinomial logistic regression.Of the 1,576 patients in this cohort, 806 (51.1%) were female, and the median age was 38.0 years. PCTree identified four patient subgroups defined by employment status, age, and sex. Subgroup 1 were unemployed patients ≤ 26 years old, subgroup 2 were unemployed patients > 26 years, subgroup 3 were employed females, while subgroup 4 were employed male patients. The subgroups exhibited significant differences on education level, comorbidity index scores, marital status, type of epilepsy, and driving status.PWE differed in their interpretation and responses to questions about their depression symptoms, and these differences were a function of sociodemographic and clinical characteristics. [ABSTRACT FROM AUTHOR]