According to the World Health Organization, approximately 280 million people suffer from major depressive disorder (MDD) worldwide. 1 In the USA alone, 49 000 people died by suicide in 2022. 2 People living with epilepsy (PWE) have higher rates of psychiatric comorbidities, with depression being the most common. Reference Gilliam, Kuzniecky, Faught, Black, Carpenter and Schrodt3 Suicidality, including ideation and attempts, occurs more frequently in PWE compared with the general population. Factors that have been associated with more significant depression among PWE include greater seizure frequency and recurrence, Reference Liu, Chen and Zheng4 both of which portend a worse quality of life (QoL). Reference Ioannou, Foster, Sander, Dupont, Gil-Nagel and O’Flaherty5 Additionally, a bidirectional association exists, with people experiencing mood disorders and suicidality being at greater risk for developing epilepsy. Reference Hesdorffer, Ishihara, Mynepalli, Webb, Weil and Hauser6
Risk factors
People living with epilepsy may have additional factors beyond seizure occurrence and which may differ from sociodemographic risk factors for depression risk in the general population. While several studies have demonstrated increased prevalence of depression among women in the general population, studies have been inconsistent among PWE with comorbid MDD. Reference Parker and Brotchie7 Shorter epilepsy duration has been associated with less severe depression, while older age in PWE has been associated with more severe depression. Reference Yang, Yang, Shi, Wang and Jiang8 Studies have shown that MDD can worsen medical comorbidities and, similarly, that a greater medical burden would lead to more severe depression in PWE. Reference Yang, Yang, Shi, Wang and Jiang8
Access
One important area in which PWE may be vulnerable to mental health comorbidity relates to place of residence, possibly connected to social isolation and care access. Treatment disparities exist in rural communities because access is often challenging, particularly for mental healthcare. Reference Newman, Hawrilenko, Jakupcak, Chen and Fortney9 Self-perceived stigma around mental health among rural residents compounds this problem because rural residents are less likely to seek mental healthcare, resulting in higher rates of hospitalisation. Reference Stewart, Jameson and Curtin10 Given the already substantial stigma burden related to having epilepsy, there is a critical need to support PWE living in rural settings and who may be particularly vulnerable to depression.
Aim
This analysis is part of a larger randomised controlled trial (RCT) examining the outcomes of an epilepsy self-management programme, Self-Management for People with Epilepsy and Negative Health Events (SMART), which prioritised enrolment of adults with epilepsy residing in rural and semi-rural regions of the USA. We performed a cross-sectional analysis of baseline data to examine demographic and clinical correlates of depressive symptom severity in this sample of PWE. Based on the extant literature, we hypothesised that women, younger individuals and those with more frequent seizures and a greater medical burden would have greater depressive symptom severity.
Method
This study is a cross-sectional analysis of baseline RCT data in PWE who had experienced at least one negative health event (NHE) in the past 6 months, with NHEs defined as seizures, hospitalisations, emergency department visits and self-harm attempts. This trial utilised a community advisory board to promote recruitment from a rural region in the USA. Study inclusion criteria included English-speaking adults ≥18 years of age diagnosed with epilepsy, having had ≥1 NHE in the past 6 months and with the capacity to provide informed consent. Participants were excluded if they were actively suicidal or homicidal, had neurodegenerative disorder precluding the ability to understand written informed consent or they were pregnant. Informed consent was obtained from participants both in person and over the telephone; each participant was sent this document detailing the risks and benefits of participating in the study, under the Institutional Review Board of University of Iowa (no. IRB ID 201910824). A more detailed description of the RCT methods has been described elsewhere. Reference Ghearing, Briggs, Cassidy, Privitera, Blixen and Sajatovic11 A standardised depressive symptom severity scale was used to examine demographic and clinical correlates of depression. For this analysis we focused on frequency of seizures in the past 30 days and past 6 months, excluding participants who had >500 seizures per day. For this analysis we included only those RCT participants with complete depression severity scores at study baseline.
The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional committees on human experimentation, and with the Helsinki Declaration of 1975 as revised in 2013. All procedures involving human subjects/patients were approved by the University of Iowa Institutional Review Board. Participant data were de-identified and are stored in the Managing Epilepsy Well Network database.
Measures
Each participant completed the nine-item Patient Health Questionnaire (PHQ-9), a self-reported symptom survey based on the DSM-IV diagnosis of MDD. This instrument serves as both a screening instrument and a tool to grade depression severity. PHQ-9 has previously been validated in PWE. Reference Rathore, Jehi, Fan, Patel, Foldvary-Schaefer and Ramirez12 Depression severity based on total PHQ-9 score is rated as follows: 1–4 for minimal depression, 5–9 for mild depression, 10–14 for moderate depression, 15–19 for moderately severe depression and 20–27 for severe depression. Patients scoring 0 on PHQ-9 were rated as having no depression. A score of 10 or greater is typically used to identify clinically significant depression in practice settings. For this analysis we elected to collapse cut-off scores of 0–9 to denote no/mild depression (NMD), 10–19 for moderate depression (MOD) and (20–27) for severe depression. The analytical measures used in this study are now described.
Originally adapted by Dilorio and colleagues, the Epilepsy Stigma Scale (ESS) is a validated, ten-item scale designed to measure the extent to which individuals believe that epilepsy is viewed negatively and impacts their social interactions. Reference DiIorio, Osborne Shafer, Letz, Henry, Schomer and Yeager13 ESS is scored on a seven-point scale, where higher scores indicate a greater perceived stigma.
The Epilepsy-Related Self-Efficacy Scale (ESES) was used to measure self-efficacy related to epilepsy. This instrument consists of 33 items, with scores ranging from 0 to 330. Higher scores reflect greater self-efficacy. Reference DiIorio, Hennessy and Manteuffel14
The 38-item Epilepsy Self-Management Scale (ESMS) evaluates the competency of epilepsy self-management across multiple life domains. Scores range from 1 to 190, with higher scores reflecting more effective self-management of the condition. Reference DiIorio, Osborne Shafer, Letz, Henry, Schomer and Yeager13
The Multidimensional Scale of Perceived Social Support (MSPSS) is a 12-item scale used to measure perceived social support, with scores ranging from 12 to 84; higher scores indicate greater social support. Reference Zimet, Dahlem, Zimet and Farley15
The Charlson Comorbidity Index (CCI) uses the self-rated CCI, which weighs different comorbidities to predict the risk of death within 1 year of hospitalisation. Reference Chaudhry, Jin and Meltzer16 A higher score predicts a greater risk of mortality and, as such, we chose to quantify the number of medical conditions within this index.
The Rapid Estimate of Adult Literacy in Medicine, Revised (REALM-R) is a screening tool used to assess health literacy. It is a word recognition test in which participants are asked to pronounce various words by which they are graded. Reference Baker, Williams, Parker, Gazmararian and Nurss17 The test is scored from 0 to 8, with a score of <6 indicating low health literacy.
The Quality of Life in Epilepsy Inventory-31 (QOLIE-31) scale consists of 31 questions comprising social and cognitive functioning. Reference Cramer, Perrine, Devinsky, Bryant-Comstock, Meador and Hermann18 This survey is scored from 0 to 100, with higher scores indicating a higher QoL.
Functional health status was assessed using the 36-question Short Form Health Survey (SF-36v2), a generic measure of functional health status and relative burden of disease. Reference Ruta, Garratt, Abdalla, Buckingham and Russell19 SF-36v2 yields two psychometrically based components, a mental component score (MCS) and a physical component score (PCS). The former measures emotional well-being, with domains including vitality, social functioning and mental health. The latter measures physical functioning and covers health problems such as chronic pain that would limit one’s daily functioning. Each summary is scored from 0 to 100, with lower scores indicative of worse functioning.
Rurality of residence was determined using the Rural Urban Continuum Code (RUCC) created by the U.S. Department of Agriculture. Reference Davis, Cromartie, Farrigan, Genetin, Sanders and Winikoff20 This system classifies counties according to population size, level of urbanisation and proximity to metropolitan areas. Each US county is assigned one of nine codes, where one represents the most urban and nine the most rural. For simplification we elected to group codes by population: ≤20 000 encompassing RUCC 6–9 and ≥20 000 encompassing RUCC 1–5.
Data analysis
Descriptive statistics were used to describe demographic and clinical variables of interest. The distribution of continuous variables was compared across the three PHQ-9 categories using the non-parametric Kruskal–Wallis equality-of-populations rank test, while categorical variables were compared using Fisher’s exact test. Using this same statistical approach, moderate and severe depression categories were compared with the mild depression category. Variables demonstrating a suggestive association (P < 0.1) with depression status in the bivariate comparisons were considered in a multivariable ordered logistic regression that aimed to identify variables independently associated with depression status. A stepwise backwards approach was applied to the fully parameterised model, retaining variables with suggestive associations (P < 0.1). This multivariable model was repeated, excluding SF-36v2 MCS and PCS, because these were moderately correlated with more specific epilepsy measures that were also included in the initial model. A two-sided alpha of 0.05 was considered statistically significant.
Results
The demographic characteristics of study participants are listed in Table 1. The analysed sample consisted of 159 PWE with a mean age of 39.46 years (s.d., 12.15). The majority of participants were women 67.3% (n = 107) and White 82.4% (n = 131). African Americans comprised 8.81% (n = 14) of participants, with Hispanics comprising 6.92% (n = 11). Nearly half (47.8%) of the study population was unmarried or not cohabitating; 39.6% (n = 63) of participants had a bachelor’s degree or greater level of education, and nearly 40% were unemployed and earned less than US$25 000 per year. Twenty-nine participants (18.2%) were from rural populations of <20 000 people. The average number of seizures over the past 6 months across the entire study population was 20.64 (s.d., 41.16), with a past 30-day seizure frequency of 6.43 (s.d., 21.18).
Demographic characteristics and depression severity in people living with epilepsy (PWE) as measured by the nine-item Patient Health Questionnaire (PHQ-9)
NMD, no depression to mild symptoms; MOD, moderate depression; RUCC, Rural Urban Continuum Code; MCS, mental component score; PCS, physical component score.
Bold denotes statistical significance <0.05.
Distribution of 9-item Patient Health Questionnaire (PHQ-9) baseline data in people living with epilepsy. NMD, no depression to mild symptoms; MOD, moderate depression; SD, severe depression.
The distribution of PHQ-9 scores is shown in Fig. 1. Mean total PHQ-9 score was 10.46 (s.d., 7.003), consistent with moderate depression. A total of 48% (n = 76) of participants met the criteria for NMD, while 41% (n = 65) met those for MOD and 11% (n = 18) met those for severe depression. Health insurance coverage differed among groups, with 83.3% (n = 15) of those with severe depression, 52.3% (n = 34) of those with MOD and 50.0% (n = 38) of those with NMD having either public or no insurance. Participants in the severe depression group had had significantly more seizures in the past 30 days (mean, 19.77; s.d., 55.43) compared with the other groups (P = 0.026 and P = 0.017). Furthermore, the severe depression group scored higher on ESS (mean total, 51.3; s.d., 13.86), reported lower social support (mean total, 55.05; s.d., 20.85), worse QoL (mean total, 26.75; s.d., 11.51), poorer emotional well-being (mean total, 21.93; s.d., 12.38) and worse physical functioning (mean total, 38.42; s.d., 10.22). Participants in both the MOD and severe depression groups had lower self-efficacy compared with the NMD group, 229.64 (s.d., 51.45) and 234.27 (s.d., 41.47), respectively. While >90% of the NMD and MOD groups identified as White, only 61.1% (n = 11) of the severe depression group did so. Regarding seizure type, focal to bilateral tonic–clonic correlated with depression severity.
Table 2 shows the clinical and demographic variables associated with depressive symptom severity in the study sample. Participants with more seizures in the past 30 days had an increased likelihood of having depression (odds ratio 1.05). Higher mental functioning was associated with less depression (odds ratio 0.86).
Multivariable analysis of demographic and clinical variables associated with depression symptom severity in people living with epilepsy (PWE) participating in an epilepsy self-management clinical trial a
Bold denotes statistical significance <0.05.
a. Stepwise backward model selection (retaining variables with p < 0.1) was applied to a full parameterised model that included variables that had any evidence of a suggestive bivariate association (p < 0.1) in Table 1: race, employment, income, insurance type, birthplace, CCI count, seizure count in last 30 days, focal to bilateral tonic–clonic seizure, tonic–clonic seizures, ESS, ESES, ESMS, Multidimensional Scale of Perceived Social Support, QOLIE-31 T-score, mental component score (MCS) and physical component score (PCS).
b. Initial model excluded 36-question Short Form Health Survey PCS and MCS, as they were correlated with QOLIE-31 T-score (R 2> 0.4).
Discussion
Over 50% of PWE in this trial had a depression severity score in the moderate or severe range, with seizure frequency in the previous 30 days correlating with depression severity. Our initial hypotheses, that women, younger individuals and those with more frequent seizures and a greater medical burden would have greater depressive symptom severity, were only partially confirmed. We found worse depressive severity in those with more frequent seizures, but we did not find significant associations between depressive severity based on gender, age or medical burden.
The mean daily seizure frequency was close to 20 in the past 30 days among those with severe depression symptoms when compared with just over 4 in the past 30 days among the group with mild or no depression symptoms. Having more seizures in the past 30 days also correlated with increased risk for more severe depression. However, seizure frequency in the past 6 months did not correlate with depression severity. Aligned with study inclusion criteria, ≥95% of participants had a NHE that was a seizure in the 6 months prior to study enrolment. The total sample had a relatively high pre-study enrolment seizure burden, which may explain why the 6-month seizure frequency was not significantly different among the three groups. Of note, while we categorised epileptic seizures based on type, we did not record information on non-epileptic events including psychogenic non-epileptic seizures (PNES). Studies are now emerging, including that by Bornovski et al, which show that a diagnosis of PNES alone places individuals at a higher risk of suicide. Reference Bornovski, Jackson-Shaheed, Argraves, Hitchins, Tolchin and Galluzzo21 Furthermore, a comorbid diagnosis of PNES increases the risk of a suicide attempt in PWE. Reference Faiman, Hodsoll, Young and Shotbolt22 Thus it is possible that participants with comorbid PNES may have accounted for a percentage of the severe depression group.
Contrary to our initial hypothesis, greater medical burden, as measured by CCI, was not associated with greater depressive severity. Among the reasons for this may be that our sample had a low mean CCI score – 1.02 (s.d., 1.61), indicating minimal medical burden. This may be due to the low average age of the overall sample, 39.46 years (s.d., 12.25). Furthermore, while 67% (n = 107) of participants were women, gender was not associated with depressive severity in our baseline sample, as we initially hypothesised.
Depression in epilepsy carries a higher rate of suicide than that in the general population. Reference Harris and Barraclough23 Other authors have found an increased risk of suicide with a monthly seizure frequency of at least one, and in PWE who require a loading dose of an anti-seizure medication. Reference Kim, Kim, Jeon, Kim and Lee24 It is important that clinicians are aware of the importance of suicide screening and prevention alongside seizure control in PWE. The significant correlation between worse depression severity with greater 30-day seizure frequency in our analysis sample suggests that those presenting with recent seizures are a population who would benefit from not only depression screening, but also effective interventions to help treat depressive symptoms.
In addition to significant correlations between depression severity and seizure frequency, we also found significant associations between depression severity and QoL. Consistent with the literature, Reference Boylan, Flint, Labovitz, Jackson, Starner and Devinsky25 the severe depression group had a worse QoL score. However, cross-sectional studies such as that presented here do not provide data on how QoL changes over time. Notably, prospective studies, including that by Hamid et al, measured QoL over 5 years and found that it changed over time, even with partial seizure control. Reference Hamid, Blackmon, Cong, Dziura, Atlas and Vickrey26 Furthermore, the effect of depression on QoL may diminish over time. Functional mental status component was also found to be correlated with depression severity. Other variables found to be associated with worse QoL in PWE, such as older age, lower education level and longer duration of epilepsy, Reference Shakir and Al-Asadi27 did not correlate with depression severity in this sample. Our sample overall was highly educated, with almost 70% (n = 123) having at least some college education or a college degree. In addition, the relatively small sample size and higher seizure burden may explain the lack of correlation with factors noted in previous studies.
A recent study by Mayo Rodriguez et al showed that QoL was worse in patients with focal epilepsy. Reference Mayo Rodríguez, Parejo-Carbonell, Sanz Graciani, Romeral Jiménez, De la Cruz Moreno and Lastras28 Notably, participants in all groups in our study who suffered from focal seizures that transitioned to bilateral tonic–clonic activity had worse depression; however, QoL was not assessed by seizure type in our study. Suicide has been reported to be 25 times more common in those PWE who have the most common type of focal epilepsy, temporal lobe epilepsy. Reference Cai, He, Luo, Gui, Wei and Lu29 These subjects are at greater risk for postictal psychosis, which may explain the greater risk of suicidality. Reference Gattaz, Valente, Raposo, Vincentiis and Talib30 Lastly, suicide is five times more common in PWE undergoing epilepsy surgery as opposed to those undergoing treatment with medication. Reference Verrotti, Cicconetti, Scorrano, De Berardis, Cotellessa and Chiarelli31 A limitation of this study is that seizure type was based on patient self-reporting. The most common seizure type reported was generalised tonic–clonic, which may underrepresent the true number of people who had focal to bilateral tonic–clonic seizures. Furthermore, patients identifying as having focal to bilateral tonic–clonic seizures may be more likely to have undergone evaluation for surgery, including video electroencephalogram monitoring, which resulted in a better understanding of seizure localisation than for other participants in this study.
While the severe depression group reported the lowest perceived social support as measured by MSPSS, both the MOD and severe depression groups had low self-efficacy. Other authors have shown that the moderating effect of self-efficacy on epilepsy was stronger in PWE who had suffered a seizure in the past year as opposed to those who were seizure free. Reference Lee, Im, Jung and Seo32 Furthermore, a 2006 study showed that the ability to manage seizures, their intensity and the severity of depressive symptoms were the most significant factors impacting self-efficacy levels among people with epilepsy. Reference DiIorio, Shafer, Letz, Henry, Schomer and Yeager33 Therefore, psychosocial interventions are imperative to improving QoL and self-efficacy and and reducing symptom burden in PWE. Several self-management interventions exist, including MINDSET, and from our own group SMART, which have been shown to improve QoL in PWE by both reducing seizure frequency and improving behavioural outcomes. Reference Begley, Shegog, Harding, Goldsmith, Hope and Newmark34
Perceived stigma is highly prevalent among PWE and can result in increased feelings of isolation, embarrassment and stress, reduced opportunities for education and employment and adverse health outcomes. Reference Akbari, Mohammadi and Hosseini35 Our findings revealed that the severe depression group experienced the most significant perceived stigma. These are consistent with previous findings from our group, which has established a link between higher perceived stigma and increased depressive symptoms; Reference Blixen, Ogede, Briggs, Aebi, Burant and Wilson36 and with findings from other authors who have noted that stigma is associated with depression and suicide risk. Reference Zhao, Liu and Xiao37 It is worth noting that health literacy was not correlated with PHQ-9 scores. Plausible explanations for this may include the fact that REALM-R has not been validated in PWE, and that our sample was relatively well educated. Furthermore, contrary to the literature, our study did not find a relationship between PHQ-9 scores and ESMS; Reference Begley, Shegog, Liu, Tatsuoka, Spruill and Friedman38 because ESMS is a secondary outcome, our study was not powered to detect this finding.
Implications for future practice
Our study population carried a high burden of depressive symptoms, with 52% of PWE experiencing moderate to severe depression. This highlights the need for routine screening strategies to identify depression, which can be employed at a routine neurology appointment. Because nearly one in five PWE in our sample lives in a rural setting, telehealth or other remotely delivered care approaches that can address both general health and depression are greatly needed to support PWE who are vulnerable to mental health comorbidity. Further interventions aimed at improving self-efficacy and QoL in PWE should be investigated. Future studies might explore factors among PWE with focal seizures that could contribute to increased depression and suicidality.
Limitations
Our study has several limitations, including the relatively small sample size that limits both generalisability and the statistical power to detect modest effect sizes (f < 0.025). The cross-sectional design precludes causal inference. While PHQ-9 is validated in PWE, it is not specific to this population, in contrast with the Neurological Disorder Depression Inventory for Epilepsy, Reference Gilliam, Barry, Hermann, Meador, Vahle and Kanner39 that was designed to account for anti-seizure medication effects. Furthermore, we did not collect data on the numbers of psychotropic and/or anti-seizure medications taken by each participant, which may have informed our findings in regard to depressive severity groups; and we did not collect information on additional psychiatric comorbidities, which would have informed a broader clinical picture. Finally, seizure types and frequencies were self-reported, which may have introduced inaccuracies.
In conclusion, people living with epilepsy and suffering from more severe depression reported a higher frequency of seizures within the past 30 days, diminished QoL, reduced self-efficacy, greater perceived stigma and poor overall functioning. These findings underscore the critical need for routine depression screening in PWE, and the integration of mental health support into epilepsy care to address these interconnected challenges.
Data availability
The data that support the findings of this study are available from the corresponding author, M.S., upon reasonable request.
Author contributions
R. Branning: conceptualisation, writing and visualisation of the manuscript; G.R.G.: funding acquisition, supervision, conceptualisation, investigation and writing of the manuscript; F.B.B.: formal analysis, methodology, validation and writing of the project; M.L.: writing of the manuscript; D.S.: writing of the manuscript; M.T.: data curation and visualisation of the manuscript; J.B.: data curation and project administration; J.K.-M.: data curation; N.F.: data curation; C.A.: data curation and project administration; R. Barigye: data curation; M.S.: funding acquisition, investigation, supervision, conceptualisation and writing of the project.
Funding
This publication is a product of the Health Promotion and Disease Prevention Research Center, supported by Cooperative Agreement Number 6 U48DP006389 from the Centers for Disease Control and Prevention (CDC). The findings and conclusions in the report are those of the authors and do not necessarily represent the official position of CDC.
Declaration of interest
M.S. has received research grants from Neurelis, Intra-Cellular, Merck, Otsuka and Alkermes; is a consultant to Alkermes, Otsuka, Janssen, Lundbeck and Teva; and has received publication royalties from Springer Press, Johns Hopkins University Press, Oxford Press and UpToDate. We confirm that M.S., a member of the BJPsych Open editorial board, did not take part in the review or decision-making process of this paper. G.R.G. has received grants from CDC. F.B.B. has received research grants from NIH and is a consultant for The Michael J. Fox Foundation and iConquerMS. The other authors have no disclosures to report.
Open access
eLetters
No eLetters have been published for this article.