Psychosocial factors associated with anxiety and depression in adolescents with epilepsy: A systematic review

Anxiety and depression are common in adolescents with epilepsy. Identifying psychosocial risk factors for anxiety and depression is essential for adolescents with epilepsy to receive appropriate support. This systematic review synthesised findings of studies examining the relationship between psychosocial factors and anxiety and/ or depression in adolescents with epilepsy. Outcomes were anxiety, depression, and mixed anxiety & depression. Six electronic databases were searched for studies which: used cross-sectional or prospective designs; quantitatively evaluated the relationship between psychosocial factors and anxiety and/or depression; presented results for adolescents with epilepsy aged 9 – 18 years; and used validated measures of anxiety and/or depression. Psychosocial factors were categorised as intrapersonal, interpersonal, or parent-specific factors. Sixteen studies (23 articles) were included. All but one were cross-sectional. Regarding intrapersonal factors, alternative mental health difficulties were consistently positively associated with all three outcomes. Negative attitude towards epilepsy, lower seizure self-efficacy, lower self-esteem and stigma were consistently positively associated with depression. Interpersonal factors (i.e., lower family functioning assessed from an adolescent ’ s perspective) and parent-specific factors (i.e., parental stigma, stress, anxiety and psychopathology) were positively associated with at least one outcome. Adolescent epilepsy management should exceed assessment of biological/biomedical factors and incorporate assessment of psychosocial risk factors. Prospective studies examining the interplay between biological/biomedical factors and the psychosocial factors underpinning anxiety and depression in adolescents with epilepsy are needed.


Introduction
Epilepsy is one of the most common neurological conditions in childhood [1], affecting around 22 million youth worldwide [2].Epilepsy accounts for approximately 13 million disability adjusted life years each year [3] and is responsible for approximately 0.5 % of the global burden of disease [4].Around 19 % and 14 % of youth with epilepsy 1 meet diagnostic criteria for anxiety and depressive disorders, respectively [5], 3 to 5 times higher than in the general youth population [6].In comparison to healthy control samples, youth with epilepsy also experience significantly higher anxiety (d = 0.57) and depressive (d = 0.42) symptoms [5].Anxiety and/or depression in youth with epilepsy is associated with poorer academic achievement, increased suicidal ideation, reduced quality of life (QoL), and higher health resource utilization [7][8][9][10].It is therefore imperative that youth with epilepsy have access to appropriate interventions to reduce anxiety and depression.To inform the development and implementation of appropriate interventions, identifying risk factors associated with anxiety and depression in youth with epilepsy is essential.
Research has primarily focused on identifying sociodemographic, ASM and epilepsy-specific risk factors [11,12] and has thus far produced mixed findings [13][14][15].While efforts to identify psychosocial risk factors has received less attention than the other three areas [12], there is growing evidence that psychosocial factors have a greater impact on anxiety and depression in people with epilepsy than risk factors across the other three areas [16][17][18][19][20].As all ASMs can trigger anxiety and depressive symptoms [21], switching to an alternative ASM following occurrence of anxiety and/or depressive symptoms is appropriate clinically.However, ASM side-effects alone do not explain the high rates of anxiety and depression experienced by youth with epilepsy [22,23].Moreover, as many sociodemographic and epilepsy-specific variables are not readily modifiable (e.g., age, seizure type) the clinical utility of identifying risk factors across these areas is questionable.Therefore, identifying psychosocial risk factors associated with anxiety and depression in youth with epilepsy appears a more clinically useful path [12,24].
Despite limited understanding of the psychosocial risk factors associated with anxiety and depression in youth with epilepsy, 10 trials have evaluated the efficacy of psychological interventions for anxiety and/or depression in youth with epilepsy [25][26][27][28][29][30][31][32][33][34].The psychological interventions evaluated in these trials aimed to modify a range of psychosocial factors including attitude toward having epilepsy, coping skills, illness appraisals, and family dynamics.Eight of the 10 trials were primarily designed to evaluate the feasibility and acceptability of interventions (i.e., phase I trials) [25][26][27][30][31][32][33][34]; while only two were fullscale trials primarily designed to test intervention efficacy (i.e., phase II trials) [28,29].Findings are mixed; seven trials reported a significant reduction in anxiety and/or depression from pre-to post-intervention [26,[28][29][30][31]33,35] and three reported a significant reduction from preintervention to 3-or 6-month follow-up [26,29,31].Alternatively, three reported no significant reduction in anxiety and/or depression from pre-to post-intervention [27,32,34] and one reported no significant reduction from pre-intervention to 3-month follow-up [27].These findings indicate that psychological interventions may reduce anxiety and/or depression in youth with epilepsy.However, as most of these trials were phase I intervention trials with underpowered samples, confidence in such findings is limited.Moreover, as none of the trials explored which psychosocial variables mediated treatment effects, it is unclear which psychosocial variables targeted for modification in these interventions were influential (or not) in the reduction of anxiety and/or depression.
Prior to conducting large-scale high-quality psychological intervention trials for youth with epilepsy, it is important to develop a better understanding of the psychosocial variables associated with anxiety and depression in youth with epilepsy, as this could help inform the development of theoretically-driven psychological interventions, considered best practice in intervention development [36,37].The psychosocial variables associated with anxiety and depression in youth with epilepsy can differ depending on the life stage of a young person (e.g., young childhood vs. adolescence) [38].As adolescence is a time of physical, social, and psychological change, developing an understanding of the psychosocial variables associated with anxiety and depression in youth with epilepsy during adolescence is a prerequisite to developing effective and age-appropriate psychological interventions for this group.While several studies have examined whether psychosocial variables are associated with anxiety and/or depression in adolescents with epilepsy [14,15,39], no systematic review has been conducted.The aim of the current review, therefore, is to systematically identify, appraise and synthesise the findings of studies examining the relationship between psychosocial variables (i.e., subjective psychological and/or social variables) and anxiety and/or depression in adolescents with epilepsy.

Method
This systematic review is reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement [40].The protocol is registered in the PROSPERO database (CRD42021293698).

Eligibility
Studies were included if they: 1) used a cross-sectional or prospective design; 2) conducted and reported findings of a quantitative analysis exploring the relationship between anxiety and/or depression and a psychosocial variable (multivariate analyses were included if anxiety and/or depression was the outcome variable); 3) reported findings specifically for adolescents with epilepsy aged 9-18; 4) assessed anxiety and/or depression using a validated self-report questionnaire (or subscale of a validated self-report questionnaire) or a validated structured diagnostic interview; and 5) were published in English in a peerreviewed journal.
Prospective studies were included if relevant analyses were conducted at baseline or if anxiety and/or depression was measured at follow-up.Intervention studies were included if relevant analyses were conducted pre-intervention (post-intervention data was excluded).Studies were excluded if all participants were specifically recruited based on a medical or neurological comorbidity (e.g., if having an intellectual disability or non-epileptic seizure disorder were part of the inclusion criteria for the whole sample).Commentaries, conference abstracts, case-studies, editorials, and review articles were excluded.
Outcome variables were anxiety, depression, and mixed anxiety & depression.Psychosocial variables were grouped into three categories based on the following definitions: 'intrapersonal factors' (subjective psychological and/or social characteristics located directly within the adolescent with epilepsy); 'interpersonal factors' (involving the relationship between the adolescent with epilepsy and another); and 'parent-specific factors' (subjective psychological and/or social characteristics located directly within the parent of the adolescent with epilepsy).

Search strategy
Medline, Web of Science, PsycINFO, CINAHL, psycARTICLE, and AMED were searched from their inception to July 2022 using a combination of terms related to epilepsy, emotional distress, and youth (see Appendix A for search terms).Search terms were limited to titles and abstracts and filtered by language (English) and document type (journal articles).Reference lists of included studies and relevant reviews [14,15,39,41] were hand-searched to ensure relevant articles were not missed.Searches were updated in March 2023 to identify additional relevant studies.

Screening and selection
Study screening was shared by three reviewers (JT, CD, & CM).One reviewer (JT) independently assessed all titles and abstracts; while two reviewers (CD & CM) each independently assessed approximately half of all titles and abstracts.At this stage, agreement between JT and the other reviewers (CD & CM) was high (91 %).Next, the full-text of all potentially relevant articles were retrieved and assessed for inclusion by one reviewer (JT).To check for consistency in selection, the other reviewers (CD & CM) each independently assessed a random 10 % of fulltext articles.At both stages, discrepancies were resolved through discussion between two reviewers (i.e., JT & CD; JT & CM).Any unsolved discrepancies were resolved through discussion with a fourth reviewer (PF or MGC).

Data extraction and synthesis
Using a specially-devised data extraction form (see Appendix B), data were extracted and tabulated from all eligible studies by one reviewer (JT).When studies recruited a broader sample, which included relevant analyses for a sub-group of participants meeting our eligibility criteria (i.e., adolescents with epilepsy aged 9-18), only data for the population meeting our eligibility criteria were extracted.
Extracted data included general study details, participant details, design and methodology details, and a summary of reported findings (including were possible relevant p, t, and F values, correlation coefficient values, standardized beta coefficients or odds ratios, and percentage of individual variance explained; R 2 values for overall models and unstandardized beta coefficients were not extracted).Articles that reported data from the same study were interpreted and referred to as a single study with all relevant articles listed.Each outcome variable (anxiety, depression, and mixed anxiety & depression) was examined separately.

Risk of bias
Risk of bias of included studies was assessed using a modified version of a quality assessment tool for observational studies developed by the Fig. 1.PRISMA diagram summarising the screening process for included studies.

J. Temple et al.
Agency for Healthcare Research and Quality [44] (see Appendix C).One reviewer (JT) independently assessed the quality of all included studies, while two reviewers (CD & CM) each independently assessed approximately half of included studies.Discrepancies were resolved through discussion between reviewers (i.e., JT & CD; JT & CM).Unresolved discrepancies were resolved through discussion with a fourth reviewer (PF or MGC).When assessing the risk of bias of analysis methods, we only assessed the risk of bias for the analyses included in this review (i.e., analyses evaluating the relationship between psychosocial variables and anxiety and/or depression).

Results
The electronic database search retrieved 8,716 articles.After removal of duplicates, 4,450 remained for screening based on title and abstract.Of these, 3,965 clearly did not meet inclusion criteria.The fulltext of the remaining 485 articles were assessed for eligibility.Overall, 23 articles corresponding to 16 studies were eligible and included (Fig. 1; see Appendix D for reference list of included articles).
Study characteristics are displayed in Table 1.Seven studies were conducted in North America and all but one study was cross-sectional.Nine studies (14 articles) measured one outcome (e.g., anxiety, depression, or mixed anxiety & depression), six studies (eight articles) measured two outcomes, and one study (one article) measured all three outcomes.Depression was the most frequently assessed outcome (13 studies, 20 articles); followed by anxiety (six studies, eight articles); and mixed anxiety & depression (five studies, five articles).Of the 13 studies measuring depression, 10 used self-report measures, two used self-report and parent-proxy measures, and one used a structured clinical interview.Of the six studies measuring anxiety, four used self-report measures; one used a parent-proxy measure, and one used a structured clinical interview.Of the five studies measuring mixed anxiety & depression, three used self-report measures and two used parent-proxy measures.The most used depression outcome measure was the Child Depression Index (CDI) [45], used in seven studies (10 articles); and the most used mixed anxiety & depression outcome measure was the anxiety/depression subscale of the Child Behavior Checklist (CBCL) [46,47], used in four studies (four articles).No anxiety outcome measure was used in more than one study.A glossary of the outcome measures used are shown in Table 2.Only six studies included multivariate analysis.
Participant characteristics are displayed in Table 3. Sample sizes ranged from 23 to 289.Mean sample age ranged from 11.8 to 15.6 years.Mean duration of epilepsy was reported in eight studies and ranged from 5 to 7.5 years.The proportion of participants taking ASMs was reported in 14 studies and ranged from 75 % to 100 %.
Only six studies (eight articles) included multivariate analysis investigating whether psychosocial variables are associated with anxiety and/or depression (in which anxiety and/or depression was the outcome variable).Of those six studies, there was considerable variation in entry method, and it was often unclear which variables were included in the final model.There was also considerable variability in the statistics reported for multivariate analysis.While only standardized beta coefficients are included in the narrative write-up, additional statistics (e. g., R 2 values) are included in Table 5, were possible.

Risk of bias
Risk of bias for the 16 included studies is presented in Table 4.The main limitations related to sample size calculation and control of potential confounders.Only one study [48] justified the sample size solely based on sample size recommendations and no study conducted a power analysis.Most studies did not control for confounders (as most conducted only univariate analyses).
Of the six studies (eight articles) conducting multivariate analyses, only three studies (across three articles) [19,38,49] controlled for all relevant confounders (i.e., variables significantly associated with anxiety and/or depression from univariate analyses and clinical/demographic variables associated with anxiety and/or depression in youth with epilepsy in prior reviews, i.e., age, gender, seizure frequency, number of ASMs, duration of epilepsy) [14,39].The other three studies (across five articles) [50][51][52][53][54] only partially controlled for relevant confounders.All studies recruited participants through neurology or paediatric clinics or epilepsy centres, increasing the likelihood participants had a confirmed epilepsy diagnosis.However, only three studies (four articles) [49,52,55,56] sampled patients consecutively, a method which reduces likelihood of selection bias.As no study conducted power analysis, general rules of thumb were used to decide if studies were adequately powered (i.e., n ≥ 50 for univariate analysis; n ≥ 104 + the number of IVs entered in the model for multivariate analysis) [57,58].Nine studies (across 12 articles) [19,48,56,[59][60][61][62][63][64][65][66][67] had an adequately powered sample to conduct their analyses; while seven studies (across 11 articles) [38,[49][50][51][52][53][54][55][68][69][70] conducted some or all of their analyses with an underpowered sample.Most studies used validated measures to assess psychosocial variables.[49,55] assessed attitude towards having epilepsy, perceived epilepsyrelated stigma, and the impact of epilepsy on adjustment and development.When entered in a multiple regression model with clinical, demographic and other psychosocial variables, none of these variables were significantly associated with anxiety.

General beliefs and attitudes.
One study [63] measured selfesteem and one [56] measured sense of self (a similar construct to self-esteem).Anxiety was significantly associated with both global and specific aspects of self-esteem (ρ = -0.22 to -0.48; OR = 1.13 to 1.29) but was not significantly associated with sense of self.

Other intrapersonal factors.
Two studies [56,64] assessed QoL.Eddy et al. (2010) found that QoL was significantly associated with anxiety (ρ = -0.40)[56].However, after correcting for multiple comparisons, this association was no longer significant.Lai et al. (2015) found that, when assessed cross-sectionally, QoL was significantly associated with anxiety (ρ = -0.29)[64].However, when assessed prospectively, Lai et al. (2015) found that mean change in QoL from baseline to 6-month follow-up was not significantly associated with mean change in anxiety [64].General life satisfaction was assessed in one study [56] and was not significantly associated with anxiety.[64].Anxiety was also significantly associated with quality of family and peer relationships (ρ = -0.29)but after correcting for multiple comparisons, this association was no longer significant [56].Anxiety was not significantly associated with satisfaction with one's broader social and cultural environment [56].

Parental epilepsy-specific beliefs and attitudes.
Parental perceived stigma towards epilepsy (i.e., parent's perception of their child being stigmatised) was assessed in one study [49].When entered in a multiple regression model with clinical, demographic, and other psychosocial variables, parental perceived stigma was not significantly associated with anxiety.

Parental mental health difficulties.
Two studies [38,49] assessed parental mental health difficulties and reported contradictory findings.When entered in a multiple regression model with clinical, demographic and/or other psychosocial variables, anxiety was significantly associated with parental anxiety (β = 0.35) [38] but was not significantly associated with parental psychopathology [49] or parental depression [38].from a parental perspective and one study from an adolescent and parental perspective).When assessed from a parental perspective, findings were mixed.One study [53] found that depression was significantly associated with 'developing competence and optimism' (ρ = -0.27);while the other study [50] found that after accounting for clinical, demographic, and other psychosocial variables, depression was not significantly associated with either 'positive coping' (i.e., developing competence & optimism, complying with treatment, seeking support) or 'negative coping' (i.e., being irritable and withdrawing).The same study (reported in a different article) [51] assessed the impact of the difference between mother's and father's perceptions of their child's 'negative coping'.When entered in a multiple regression model with clinical, demographic, and other psychosocial variables, the absolute difference between parent's perception of their child's 'negative coping' was significantly associated with depression (β not reported).
When assessed from an adolescent perspective, 'positive coping' (i.e., problem solving, cognitive restructuring, social support) was not significantly associated with depression but 'negative coping' (i.e., withdrawing, being self-critical, emotional dysregulation, blaming others, defeatist attitude) was, even after controlling for sex, number of ASMs, and seizure severity (β not reported) [53].

Epilepsy-specific beliefs and attitudes.
Three studies (four articles) [49][50][51]54] assessed attitude towards having epilepsy.Findings were mixed.Two studies (three articles) [50,51,54] found that attitude towards having epilepsy was significantly associated with depression after accounting for clinical, demographic, and/or other psychosocial variables (β not reported); while one study [49] found no significant association.All three studies assessing seizure self-efficacy found that it was significantly associated with depression (r = -0.32 to -0.58) [48,60], even after controlling for other psychosocial variables (β not reported) [54].Both studies assessing perceived epilepsy-related stigma found that it was significantly associated with depression (r = 0.48; OR = 4.35) [49,59], even after seizure frequency, number of ASMs, and other psychosocial variables were controlled (β not reported) [49].One study [55] assessed the impact of epilepsy on adjustment and development and found that it was significantly associated with depression (t-test only).
of the sample where White; h mean age at onset for those scoring above HADS-A cut-off; i Miniksar et al. (2023) [65] did not report the mean epilepsy duration of their sample but they did report the frequency of participants whose epilepsy duration was 1 year, < 2 years, < 5 years.Therefore, this information was included in the table; j only the 'adolescent' sample (aged [12][13][14][15][16][17][18] were included as the 'children' sample (aged 6-11) did not meet eligibility criteria; k 13 fewer participants were included in Wagner et al. (2012) [70], 2012 [53]), leading to slightly different participant characteristics.(continued on next page) 3.3.1.3.General beliefs and attitudes.Perceived locus of control was assessed in one study [50].After controlling for clinical, demographic, and other psychosocial variables, an external locus of control regarding social interactions and a general unknown locus of control were significantly associated with depression (β not reported).Self-esteem was assessed in two studies [62,63].Depression was significantly associated with both global (ρ = -0.51;β = 0.62; OR = 1.32 to 1.34) [62,63] and specific aspects of self-esteem (ρ = -0.28 to 49; OR = 1.22 to 1.37) [63].
Sense of self was assessed in one study [56] and was significantly associated with depression (ρ = -0.34).Negative self-evaluation was assessed in one study [65] and was significantly associated with depression (r = 0.62) but after correcting for multiple comparisons, this association was no longer significant.
Other mental health difficulties were assessed in single studies.

Other intrapersonal factors.
Two studies [56,64] assessed QoL cross-sectionally and one [61] assessed QoL prospectively.When assessed cross sectionally, Lai et al. (2015) found that QoL was significantly associated with depression (ρ = -0.43)[64]; whereas Eddy et al. (2010) found no significant association [56].When assessed prospectively, Lai et al. (2015) found no significant association between mean change in QoL from baseline to 6-month follow-up and mean change in depression [64].Two studies [54,65] assessed hopelessness and found that it was significantly associated with depression (r = 0.69) [65], even after controlling for other psychosocial variables (β not reported) [54].[53] are the same study; s it was unclear if analysis conducted was t-test or Kruskal-Wallis; t 3-item stigma scale adapted from a stigma scale developed by Jacoby et al. (1994) [71].

Interpersonal factors 3.3.2.1. Family factors.
Three studies (four articles) [38,50,51,55] assessed family functioning (two studies from a parental perspective and one study [two articles] from an adolescent and parental perspective).When assessed from a parental perspective, findings were mixed.Adewuya & Oseni, (2005) found that family functioning was significantly associated with depression (t-test only) [55]; while Puka et al. (2017) and Dunn et al. (1999) found no significant association [38,50].Haber et al. (2003) assessed the impact of the absolute difference between mother's and father's perception of family functioning on depression and found no significant association [51].When assessed from an adolescent perspective, family functioning was significantly associated with depression even after accounting for clinical, demographic, and/or other psychosocial variables (β not reported) [50,51].
Three studies (four articles) [38,[49][50][51] assessed family adaptive resources.When entered in a multiple regression model with clinical, demographic, and/or other psychosocial variables, none of the studies found a significant association with depression.

Other interpersonal factors.
Single studies assessed other interpersonal factors.Depression was significantly associated with social functioning (ρ = -0.49)[64] and interpersonal problems (β = 0.61; OR = 1.30) [62].Depression was also significantly associated with quality of family and peer relationships (ρ = -0.32)but after correcting for multiple comparisons, this association was no longer significant [56].Depression was not significantly associated with one's broader social and cultural environment [56].

Parental epilepsy specific beliefs and attitudes.
Two studies [49,50] measured parental perceived stigma towards epilepsy.After accounting for clinical, demographic, and/or other psychosocial variables, neither study found a significant association with depression.One of these studies (reported in a different article) [51] also assessed the impact of the difference between mother's and father's perceived stigma towards epilepsy and the impact of the difference between mother's and father's attitude towards epilepsy on depression.When entered in a multiple regression model with clinical, demographic, and other psychosocial variables, neither the absolute difference between parent's perceived stigma nor the absolute difference between parent's attitude towards epilepsy was significantly associated with depression.The one study [70] assessing parental seizure self-efficacy found it was significantly associated with depression (β not reported).

Parental mental health difficulties.
Three studies [38,49,50] assessed parental mental health difficulties.Findings were mixed.When entered in a multiple regression model with clinical, demographic, and/ or other psychosocial variables, depression was significantly associated with parental anxiety (β = 0.35) [38] but was not significantly associated with parental psychopathology [49] or parental depression [38,50].

Other parent-specific risk factors.
Single studies assessed other parent-specific factors.Depression was significantly associated with parental stress (strength of the association not reported) [69].Depression was not significantly associated with the general impact of epilepsy on parents [55].

Intrapersonal factors
Intrapersonal factors were only assessed in single studies.After controlling for clinical, demographic, and other psychosocial variables, mixed anxiety & depression was significantly associated with having a positive attitude towards epilepsy (β not reported) [50], an external locus of control regarding social interactions (β not reported) [50], and four illness perception domains: expecting epilepsy to last a long time (β = 0.38), perceiving oneself to have less personal control over epilepsy (β = 0.42), believing treatment can help (β = -0.36),and expecting epilepsy to have a high emotional impact (β = 0.33) [52].Mixed anxiety & depression was not significantly associated with the following illness perception domains: perceived consequences of having epilepsy, perceived understanding of epilepsy, and perception of identity due to having epilepsy (i.e., the name or label given to having epilepsy) [52].After controlling for clinical, demographic, and other psychosocial variables, mixed anxiety & depression was not significantly associated with a general external locus of control [50].

Interpersonal factors
Interpersonal factors were only assessed in single studies.After controlling for clinical, demographic, and other psychosocial variables, mixed anxiety & depression was significantly associated with family functioning when assessed from an adolescent perspective (β not reported) but not when assessed from a parental perspective [50].Mixed anxiety & depression was not significantly associated with family resources [50] or autonomous parental support and involvement [52].

Parent-specific factors
Parental perceived stigma towards epilepsy was assessed in two studies [50,67] and was significantly associated with mixed anxiety & depression (r = 0.26 [50], t-test only [67]).Mixed anxiety & depression was also significantly associated with parental psychopathology (r = 0.32) [68] but was not significantly associated with parental depression [50].

Discussion
This review critically appraised and synthesised the findings of studies examining the relationship between psychosocial variables and anxiety and/or depression in adolescents aged 9-18 years with epilepsy.Sixteen studies, reported across 23 articles, were included.A wide range of psychosocial variables were tested (37 for depression, 20 for anxiety, 14 for mixed anxiety & depression).At least one psychosocial variable was associated with anxiety and/or depression in each study, highlighting that psychosocial variables are consistently associated with anxiety and depression in adolescents with epilepsy.Intrapersonal factors were more consistently associated with anxiety and depression than interpersonal or parent-specific factors.Alternative mental health difficulties were the most frequently assessed variable and were consistently associated with anxiety and depression (e.g., anxiety was consistently positively associated with depression and mental wellbeing).This is in line with findings from a systematic review of adults with epilepsy [12] and with findings in other adolescent physical health populations [74][75][76][77].Attitude towards having epilepsy (significant in two of three studies [three of four articles]), seizure self-efficacy (significant in all three studies), and self-esteem (significant in two of two studies) were consistently associated with depression.Attitude towards having epilepsy and self-esteem were also respectively associated with anxiety and mixed anxiety & depression (both significant in one of one studies).This is in line with findings from a systematic review of adults with epilepsy, in which seizure self-efficacy was associated with depression; and self-esteem was associated with both anxiety and depression [12].Attitude towards illness, disease management selfefficacy (a similar construct to seizure self-efficacy), and self-esteem are also associated with anxiety and depression in adolescents with other physical health conditions [78][79][80].As negative attitude towards illness, low disease management self-efficacy, and low self-esteem negatively impact adherence to medical treatment in epilepsy and other physical health populations [81][82][83], the association between these variables and anxiety and/or depression may be underpinned by a shared pathway mediated by adherence to medical treatment.However, more robust research is needed to better understand the mechanisms underlying this relationship.Nevertheless, attitude towards having epilepsy and self-esteem may be important intervention targets for anxiety and depression in adolescents with epilepsy whereas seizure self-efficacy may be an important intervention target for depression.
Perceived stigma was associated with depression (significant in two of two studies) but not with anxiety (not significant in one of one studies).Several reviews suggest perceived stigma is likely an important risk factor for anxiety and/or depression in epilepsy [14,[84][85][86].However, in their systematic review of adults with epilepsy, Gandy et al. (2012) found that perceived stigma was only associated with depression in one of three studies [12]; and that perceived stigma only accounted for 0.26 % of the variance in anxiety.Thus, Gandy et al. (2012) concluded that the role of perceived stigma in the development of anxiety and depression in epilepsy may be overestimated [12].Our findings partly support this conclusion and suggest that the role of perceived stigma as a risk factor for anxiety in epilepsy may be overestimated.However, perceived stigma may still be an important risk factor for depression.
Findings from this review suggest certain interpersonal and parentspecific factors may also be important risk factors for anxiety and/or depression in adolescents with epilepsy, although confidence in such findings is limited.Regarding interpersonal factors, single studies found that when assessed from an adolescent perspective, perceived family functioning was associated with depression and mixed anxiety & depression.However, when assessed from a parental perspective, perceived family functioning was not associated with mixed anxiety & depression and was only associated with depression in one of three studies.The difference in perceived family functioning dependant on the informant (i.e., parent vs. adolescent) highlights the importance of assessing psychosocial variables and symptoms of anxiety and/or depression in adolescents with epilepsy from multiple perspectives [87,88].
Regarding parent-specific factors, parental perception of their child being stigmatised due to having epilepsy was associated with mixed anxiety & depression (significant in two of two studies); though this association was weak.Single studies also found that parental stress and parental anxiety were associated with anxiety and depression; and parental psychopathology was associated with mixed anxiety & depression.Support for the role of these variables as important risk factors for anxiety and depression are strengthened by similar findings in other youth physical health populations in which the interpersonal and parent-specific factors outlined above are associated with several mental health outcomes [89][90][91][92][93][94].Potential reasons for the associations outlined above are provided.
Parental anxiety about epilepsy is associated with 'overprotective' behaviours in parents of youth with epilepsy [95].Parental overprotective behaviour is a predictor for anxiety and depression in the general youth population [96][97][98].For those with epilepsy, adolescence usually involves the transition of responsibility of epilepsy management from parent to adolescent.This can lead to discrepancies between parent and adolescent about the adolescent's perceived level of autonomy [99].It may be that the associations between the parent-specific factors (i.e., parental mental health difficulties and parental perceived stigma) and anxiety and/or depression found in this study are mediated or moderated by a discrepancy in perceived level of autonomy between the parent and adolescent.Such a discrepancy may partly explain why the impact of perceived family functioning differed between parent and adolescent.
The association between parental mental health difficulties and anxiety and/or depression in adolescents with epilepsy may also be influenced by adolescents adopting beliefs similar to their parents.Adolescents with physical health conditions tend to seek information relating to their condition from those whom they have a close and longstanding relationship with [100].Thus, if parents are highly anxious and worrying about their child's epilepsy, then adolescents with epilepsy may adopt similar worrisome beliefs, potentially leading to increased levels of anxiety.
This synthesis provides a valuable insight into a broad range of psychosocial risk factors associated with anxiety and depression in adolescents with epilepsy and in turn suggests many potential interventions.Several limitations of the available studies preclude strong recommendations.First, as all but one study was cross-sectional, causation cannot be inferred.Identified risk factors may not lead to the development of anxiety and/or depression in adolescents with epilepsy but instead may be a consequence of anxiety and/or depression.Second, over half of the studies (n = 10) failed to control for clinical, demographic, or other psychosocial variables in their analyses.Without accounting for such variables, it is unclear whether identified risk factors are a consequence of other uncontrolled factors.It is also unclear how such variables may interact with each other.Third, only two studies measured adolescent anxiety and/or depression from multiple perspectives (i.e., parent and adolescent), whereas four relied solely on parent-proxy report.While assessing anxiety and/or depression from an adolescent perspective is priority [101,102], using a multi-informant perspective may provide more detailed insight into adolescent's experience of anxiety and/or depression [87,88,101].Fourth, although a range of psychosocial variables were tested, many were tested in few studies.This makes it difficult to conclude whether the lack of association between such variables and anxiety and/or depression is due to such variables not being important risk factors or due to there being insufficient research investigating the role of such variables.Finally, the included studies used heterogenous outcome measures and data analysis procedures, limiting confidence in conclusions drawn.Despite these limitations, the findings have important clinical implications.

Clinical implications
Anxiety and depression were consistently positively associated with each other as well as alternative mental health difficulties.This is unsurprising given that the co-occurrence of anxiety and depression in epilepsy is common [103,104].People presenting with anxiety and depression often experience more difficulties and respond less well to psychological and pharmacological intervention than those presenting with only anxiety or depression [105][106][107].This highlights the clinical importance of screening adolescents with epilepsy for multiple types of mental health difficulties and supports current clinical guidance which recommends screening for symptoms of both anxiety and depression in adolescents with epilepsy as part of regular review [108].
However, findings from this study indicate that clinicians need to go beyond screening for anxiety and depression and screen for a range of psychosocial factors also.This could help identify psychosocial risk factors which make adolescents with epilepsy susceptible to anxiety and depression and/or areas which could be a target of psychological intervention.Supporting this approach, Kazak et al. (2011) demonstrated that, compared to those receiving routine assessment only, screening for a range of psychosocial risk factors amongst newly diagnosed paediatric cancer patients led to patients and families receiving a wider range of psychosocial care corresponding to their identified needs [109].It has been recommended that at each epilepsy clinic visit, clinicians should, at minimum, enquire about changes to patient's mental health; while a more detailed assessment should be conducted for all new patients and at routine time intervals such as annually or following any recent changes to ASM protocols [21,110].During this more detailed assessment, it may be beneficial to screen for potential psychosocial risk factors identified in this review such as attitude towards having epilepsy, seizure self-efficacy, self-esteem, perceived stigma, family functioning and parental mental health difficulties.
A common approach in epilepsy clinics is to monitor those who do not meet clinical levels of anxiety and/or depression and refer them to specialist psychologist services for intervention once they do meet criteria [111].However, as most studies in this review assessed anxiety and/or depression on a continuum, the psychosocial variables associated with anxiety and/or depression in this review are not restricted to clinical levels of anxiety and/or depression (i.e., anxiety and/or depressive disorders).Thus, screening and identifying potential psychosocial risk factors associated with anxiety and/or depression amongst adolescents with epilepsy who do not meet diagnostic criteria for anxiety and/or depressive disorders could lead to such individuals receiving lower-intensity interventions targeting identified risk factors to reduce their likelihood of developing clinical levels of anxiety and/or depression.For instance, educating adolescents with epilepsy on seizure management, improving family dynamics, providing support to parents to reduce their mental health difficulties, and providing adolescents with epilepsy and their families with psycho-educational material related to epilepsy could increase seizure self-efficacy, reduce parental mental health difficulties, and improve family functioning [112][113][114][115].
When considering clinical implications for those who require specialist psychological intervention (i.e., for those who do not respond well to lower level interventions or those presenting with anxiety and/or depressive disorders), traditional cognitive behavioural therapies which target unrealistic appraisals of events, including how one appraises their illness may be beneficial given attitude towards having epilepsy (which included items about illness appraisals), perceived stigma, seizure selfefficacy, and four illness perception domains (expecting epilepsy to last a long time, perceiving oneself to have less personal control over epilepsy, believing treatment can help, and expecting epilepsy to have a high emotional impact) were all associated with anxiety and/or depression.However, as it is unclear from the studies included whether the appraisals of adolescents with epilepsy were unrealistic, and as studies only assessed associations, it is too early to conclude this.As appraisals made by adolescents with epilepsy may be realistic (e.g., "I am treated differently than my peers due to having epilepsy", "having epilepsy could prevent me from being able to drive"), challenging such appraisals might be of limited efficacy in people with physical health conditions such as epilepsy [116][117][118][119].

Future research implications
While the findings of this review highlight psychosocial factors consistently associated with anxiety and/or depression, the evidencebase is limited by the lack of prospective studies, which precludes identifying cause and effect relationships.To better understand which psychosocial factors lead to the development and maintenance of anxiety and depression, future studies need to look beyond associations and employ more sophisticated statistical modelling techniques such as path analysis and structural equation modelling.This would enable the causal role of psychosocial factors and the interplay between biological/ biomedical and psychosocial factors to be investigated within a welldefined theoretical framework.Given that adolescents with epilepsy often present with both anxiety and depressive symptoms, future research would also benefit from focusing on identifying psychological risk factors which cause and maintain both anxiety and depression.Exploring psychological factors which have been shown to predict both anxiety and depression in other physical health populations, such as worry and rumination, seems most appropriate [120][121][122].Moreover, most studies included adolescents across different phases of the adolescent trajectory.The adolescent trajectory involves several important developmental milestones such as identity development, transition from dependence on caregiver to becoming an independent adult, and seeking acceptance from peer groups [123][124][125].Adolescents with epilepsy contend with additional challenges such as transition of responsibility of epilepsy management from parent to child, increased fear of seizures in social situations, and increased recognition and realisation of restrictions accompanying an epilepsy diagnosis [14,126].Thus, the psychosocial risk factors associated with anxiety and/or depression in adolescents with epilepsy may differ depending on their developmental phase.Future research would benefit from focusing on adolescents at specific developmental phases across the adolescent trajectory.Finally, due to the unpredictability of many aspects of epilepsy such as seizures [127,128] and given that anxiety and depression can highly fluctuate over short intervals [129], future research should employ methodologies such as experience sampling methodology which accounts for this unpredictability and variability.

Limitations of the review
As this systematic review was restricted to published studies written in English, it is possible that relevant grey-literature studies and studies published in other languages may have been excluded introducing potential language and cultural bias.Moreover, all studies recruited participants through clinics or epilepsy centres.As youth with wellcontrolled epilepsy and those from minority groups are less likely to present in clinics and epilepsy centres [87,130,131], the generalizability of findings to the wider adolescent epilepsy population is unclear.Finally, included studies were restricted to those that reported findings specifically for adolescents with epilepsy aged 9-18.This decision was made pragmatically i.e., in the UK, adolescents generally transition to adult epilepsy services by 18 years of age [132].However, some definitions of adolescence extend to 24 years of age [133].Therefore, using a more liberal definition of adolescence may have resulted in the identification of additional relevant studies.

Conclusion
This review suggests that several psychosocial variables may be important risk factors for anxiety and/or depression in adolescents with epilepsy.This highlights that the management of epilepsy in adolescents needs to go beyond the assessment of biological and biomedical factors (e.g., age, comorbid somatic conditions, seizure frequency and severity) and incorporate assessment of psychosocial factors.To advance understanding of the psychological mechanisms underpinning and maintaining anxiety and/or depression in adolescents with epilepsy, more prospective research which explicitly tests the role of psychological mechanisms accounted for within theoretical models of anxiety and/or depression is needed.This would help guide the development of more efficacious psychological interventions for adolescents with epilepsy.

Table 1
Study characteristics.

Table 2
Glossary of anxiety and depression measures used in included studies.
[55]ve studies used the original 27-item version of the CDI & two used the 27-item Turkish version of the CDI; b two studies used the CBCL parent-proxy version & two used the youth self-report (YSR) version; c one study used the adapted Turkish version of the CBCL; d Adewuya & Ola (2005)[49]and Adewuya & Oseni, (2005)[55]administered the youth and parent-proxy version of the DISC-IV.The authors combined the information from the two versions.If either respondent (youth or parent) reported information that met criteria for the relevant psychiatric diagnoses within the last 12 months, the authors concluded that the relevant psychiatric diagnosis was currently present.

Table 3
Participant characteristics from included studies.
NR(continued on next page) J.Temple et al.

Table 4
Assessment of risk of bias of included studies.

Table 5
Summary of significant findings from included articles grouped by outcome.