Depressive and anxiety disorders in epilepsy: Do they differ in their potential to worsen common antiepileptic drug–related adverse events?


Address correspondence to Andres M. Kanner, Department of Neurological Sciences, Rush University Medical Center, 1653 West Congress Parkway, Chicago, IL 60612, U.S.A. E-mail:


Purpose:  To compare the effect of anxiety disorders, major depressive episodes (MDEs), and subsyndromic depressive episodes (SSDEs) on antiepileptic drug (AED)–related adverse events (AEs) in persons with epilepsy (PWE).

Methods:  The study included 188 consecutive PWE from five U.S. outpatient epilepsy clinics, all of whom underwent structured interviews (SCID) to identify current and past mood disorders and other current Axis I psychiatric diagnoses according to Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, Text Revision (DSM-IV-TR) criteria. A diagnosis of SSDE was made in patients with total Beck Depression Inventory-II (BDI-II) scores >12 or the Centers of Epidemiologic Studies-Depression (CES-D) > 16 (in the absence of any DSM diagnosis of mood disorder. The presence and severity of AEs was measured with the Adverse Event Profile (AEP).

Key Findings:  Compared to asymptomatic patients (n = 103), the AEP scores of patients with SSDE (n = 26), MDE only (n = 10), anxiety disorders only (n = 21), or mixed MDE/anxiety disorders (n = 28) were significantly higher, suggesting more severe AED-related AEs. Univariate analyses revealed that having persistent seizures in the last 6 months and taking antidepressants was associated with more severe AEs. Post hoc analyses, however, showed that these differences were accounted for by the presence of a depressive and/or anxiety disorders.

Significance:  Depressive and anxiety disorders worsen AED-related AEs even when presenting as a subsyndromic type. These data suggest that the presence of psychiatric comorbidities must be considered in their interpretation, both in clinical practice and AED drug trials.

The presence of depressive symptoms has been found to have a negative effect on the severity of adverse events (AEs) related to antiepileptic drugs (AEDs) in persons with epilepsy (PWE) (Cramer et al., 2003; Ettinger et al., 2004; Perucca et al., 2011). For example, in a population-based study, PWE and depressive symptoms compared with asymptomatic PWE were found to endorse AEs related to AEDs of greater severity(Ettinger et al., 2004). Likewise, a recent study found a negative effect of major depressive episodes (according to Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, Text Revision [DSM-IV-TR] criteria) on AEs (Perucca et al., 2011).

Comorbid occurrence of depressive and anxiety disorders are relatively frequent in PWE (Tellez-Zenteno et al., 2007). To date there are no data on whether depressive and anxiety disorders have a different effect on AEs, whether a worse effect may result from the joint occurrence of depressive and anxiety disorders, or whether subsyndromic forms of depressive episodes have a lesser negative effect than clear-cut depressive or anxiety disorders. The purpose of this study is to compare the effect of depressive and anxiety disorders when occurring alone or together on AED-related AEs in a large cohort of outpatients with epilepsy. In this study, the diagnosis of depression and anxiety disorders had to meet the diagnostic criteria of the DSM-IV-TR (American Psychiatric Association, 2000). The second aim of this study was to determine if the effect of subsyndromic forms of depression on AEs differed from that of major depressive episodes (MDEs), anxiety disorders, or a combination of both disorders.


Patient selection

Two hundred five consecutive patients from five outpatient epilepsy clinics in the United States were initially enrolled in the study. Twelve patients had to be excluded because of incomplete datasets. Therefore, the study included 193 patients: 131 women and 62 men with a mean (± standard deviation, SD) age of 39 ± 11.7 years. The mean duration of their seizure disorder was 18.5 ± 12.8 years. The inclusion and exclusion criteria have been published elsewhere (Gilliam et al., 2006).


To identify current and past mood disorders, every patient underwent a structured interview with the mood disorders module of the Structured Clinical Interview for a DSM-IV Axis I diagnosis (SCID) (First et al., 1995). To identify other current DSM-IV Axis I diagnoses, patients were given a second structured psychiatric interview with the Mini International Neuropsychiatric Interview (MINI) International Neuropsychiatric Interview (Sheehan et al., 1998). Patients also completed two self-rating screening instruments to identify symptoms of depression: the Beck Depression Inventory-II (BDI-II) (Beck, 1987) and the Centers of Epidemiologic Studies-Depression (CES-D) (Weissman et al., 1977).

The presence and severity of AEs in each patient were recorded with the Adverse Event Profile (AEP) (Baker et al., 1994; Gilliam et al., 2004). The AEP is a self-rating, 19-item instrument that assesses the presence and severity of the more frequent AED-related AEs during the previous 4 weeks. The severity of each symptom is rated on a 4-point Likert scale (1–4), where 4 is indicative of worst severity. It should be noted, however, that of the 19 items of the AEP four can also be symptoms of depressive or anxiety disorders. This raises the possibility that high AEP scores may be due to actual symptoms of depression and anxiety and not reflect worse AEs. To avoid the possibility of this circular reasoning, total scores of the AEP were calculated separately with all the 19 items and with 15 items, which excluded items that could potentially be rated positively as an expression of the mood or anxiety disorder. The two scores were first compared between asymptomatic patients (n = 103) and patients meeting criteria of any of mood and anxiety disorder considered below (n = 85). Next, comparisons of the two AEP scores were conducted among the following diagnostic groups: (1) No psychopathology (N = 103), defined as the absence of any DSM-IV-TR Axis I diagnosis and total scores of the BDI-II < 12 and of the CES-D < 16; these are the cutoff scores suggested by the developers of these instruments as being indicative of a symptomatic state (Beck, 1987; Weissman et al., 1977); (2) current subsyndromic depressive episode (SSDE), defined as total scores of the BDI-II ≥ 12 and/or of the CES-D ≥ 16 in the absence of any DSM-IV-TR diagnosis of mood disorder identified with the MINI or SCID (n = 26); (3) current MDE (n = 10); (4) current anxiety disorders (this category included any patient with one of the following conditions: generalized anxiety disorder [GAD], social phobia, simple phobia, agoraphobia, panic disorder, obsessive compulsive disorder [OCD], or posttraumatic stress disorder) (n = 28); and (5) current mixed MDE/and anxiety disorder (n = 21). Five patients with a diagnosis of current dysthymia were eliminated because of the small number, leaving a total of 188 patients for all analyses.

Patients in group 1 were considered to be asymptomatic and those in groups 2–5, symptomatic. Given that patients could have more than one type of anxiety disorder, group 4 was subdivided into two subgroups: 4-A with a single anxiety disorder (n = 16) and 4-B with more than one type of anxiety disorder (n = 12). Likewise group 5 was subdivided into two subgroups: 5-A with current mixed MDE/and 1 type of anxiety disorder (n = 12) and 5-B with current mixed MDE/>1 type of anxiety disorder (n = 9). This allowed us to establish if having more than one type of anxiety disorder had an effect on AEs. Therefore, two sets of comparisons of total AEP scores were carried out. The first included the five initial diagnostic groups, whereas the second included groups 1, 2, 3, 4-A, 4-B, 5-A, and 5-B.

Data analyses

Comparisons of the three sets of AEP scores among the five and seven diagnostic groups were carried out with ANOVAs followed by post hoc analyses with Bonferroni. Chi-square statistics were used for comparisons of dichotomous variables.

To identify the variables predictive of high AEP scores, we conducted a step-wise linear logistic regression analysis with the total AEP score being the dependent variable, while the independent variables included: (1) psychiatric status (symptomatic [groups 2–5, n = 85] vs. asymptomatic [group 1, n = 103]); (2) type of pharmacologic regimen (monotherapy [n = 102] vs. polytherapy [n = 86]); (3) concomitant use of antidepressant drugs for a minimum of 3 months (n = 43); and (4) seizure control (seizure free for the last 6 months vs. persistent seizures). For this analysis, data were available for 173 patients (n = 83 were seizure-free for at least 6 months and n = 90 had persistent seizures).

Finally, we assessed which items of the AEP were most affected by being symptomatic with a depressive and/or anxiety disorder. To that end, we conducted a correlation of the AEP score of each item between symptomatic and asymptomatic patients, using the 15-item AEP scale. Given the number of correlations, a p-value of 0.01 was accepted as significant.


Table 1 summarizes the type and prevalence of current mood and anxiety disorders identified in the 188 patients. Of note, among the 26 patients who met criteria for SSDE, 10 had total scores in both the BDI-II and CES-D at or above the cutoff score; 10 met criteria only on the BDI-II score and 6 only on the score of the CES-D. The two AEP total scores and the total scores of the BDI-II and CES-D of each diagnostic group appear on Table 2.

Table 1.   Type and frequency of mood and anxiety disorders (according to DSM-IV-TR criteria)
Psychiatric disorderN
  1. Note that patients may be in more that one of the above categories.

Current anxiety disorder28
 Generalized anxiety disorder20
 Social phobia17
 Panic disorder4
 Obsessive compulsive disorder5
Current mood disorder31
 Current dysthymia5
 Current major depressive episode31
  With melancholic features15
  Due to general medical conditions2
Past mood disorder54
 Past major depressive episode54
 Past manic episodes5
 Past hypomanic episodes6
 Subsyndromic depressive episode26
 Patients without history or presence of above psychiatric disorders103
Table 2.   Comparison of BDI-II, CES-D, and QOLIE-89 scores among the various diagnostic groups
Psychiatric disorderNAEP score
All items (n = 19)
AEP score
No psychiatric items (n = 15)
Total BDI score (cutoff score: 12)Total CES-D score (cutoff score: 16)
(A) Groups for primary analyses
  1. AEP, adverse event profile; BDI-II, Beck Depression Inventory-II; CES-D, Center for Epidemiologic Studies-Depression.

  2. aANOVA.

 None10332.2 ± 7.626.3 ± 6.53.3 ± 3.33.4 ± 3.6
 Subsyndromic depressive episode2645.1 ± 9.636.4 ± 7.615.6 ± 6.519.4 ± 11.8
 Major depressive episode only1049.6 ± 9.339.0 ± 7.422.8 ± 9.621.7 ± 13.1
 Anxiety disorder only (1 or >1 type)2845.5 ± 10.835.1 ± 8.314.8 ± 12.516.6 ± 14.0
 Major depressive episode/anxiety disorder2152.8 ± 9.240.2 ± 6.628.4 ± 10.731.1 ± 13.3
 F, p-valuesa 38.8, <0.000128.7, <0.000168.1, <0.000153.8, <0.0001
(B) Additional groups for secondary analyses
 One type of anxiety disorder1644.9 ± 9.834.3 ± 7.612.8 ± 12.215.6 ± 14.0
 >1 type of anxiety disorder1246.3 ± 12.336 ± 9.517.5 ± 13.018.0 ± 14.6
 Major depressive episode/one type of anxiety disorder1248.0 ± 7.637.2 ± 5.524.8 ± 8.427.3 ± 11.0
 Major depressive episode/more than one type of anxiety disorder959.4 ± 7.244.2 ± 5.934.7 ± 11.735.9 ± 14.9
 F, p-valuesa 28.3, <0.000120.5, <0.000150.2, <0.000137.3, <0.0001

Effect of psychiatric diagnosis on AEs

Comparison among the five diagnostic groups

Analysis of variance yielded significant differences in the two sets of AEP scores (p ≤ 0.0001 in both; see Table 2A). Yet, Bonferroni post hoc analyses revealed that these differences were accounted for only by higher scores among the four symptomatic groups (2–5) relative to the asymptomatic group (p < 0.0001, in each comparison), whereas no significant differences were found among the four symptomatic groups. We cannot exclude the possibility that the lack of difference between patients with more severe forms of depression (e.g., MDE) and SSDE may be a function of the small number of patients in the MDE group. Likewise, although an ANOVA of BDI-II and CES-D scores yielded significant differences (p < 0.0001), post hoc analyses with Bonferroni corrections also failed to show differences of their total scores between patients with MDE and SSDE.

Does the presence of more than one type of anxiety disorder have an effect on AEP scores?

ANOVA of the two sets of AEP total scores among the seven diagnostic groups (1, 2, 3, 4-A, 4-B, 5-A, and 5-B) yielded significant differences (see Table 2B). Post hoc Bonferroni analyses revealed that these differences were accounted for by higher scores of each symptomatic groups relative to the asymptomatic group (p < 0.0001 in each comparison). Additional differences were based on higher scores in patients with mixed MDE/>1 anxiety disorder relative to those with SSDE (p = 0.001), a single anxiety disorder (p = 0.001), and >1 anxiety disorder (p = 0.026). Post hoc Bonferroni analyses of AEP scores without the psychiatric symptoms yielded the same results, with the exception that the scores of patients with mixed MDE/>1 anxiety disorder did not differ from those of patients with >1 anxiety disorder (data not shown).

As shown in Table 2, the total score of the AEP with all 19 items were higher than those of the 15-item AEP. Nonetheless, exclusion of the four psychiatric symptoms that are common to AED-related AEs, depression and anxiety disorders, did not change the results of any analysis. Given these findings, we only used the scores of the 19 item-AEP for the logistic regression analysis below.

Predictors of high AEP scores

Among the 188 patients, 85 were in the symptomatic groups, 86 were on a polytherapy regimen, and 43 were taking an antidepressant medication. Among the 173 patients for whom seizure frequency was available, 90 had persistent seizures in the last 6 months. Linear logistic regression identified the following variables as being associated with a higher AEP score (worse perception of AEs): (1) being symptomatic (AEP score: 46.6 ± 10.6, vs. 32.1 ± 7.8, F = 74.1, p < 0.0001); (2) having persistent seizures (41.6 ± 11.4 vs. 35.8 ± 11.3, F = 8.36, p = 0.004, t = 3.46, p = 0.01); and (3) being on antidepressant medication (46.5 ± 10.1 vs. 36.7 ± 11.2, F = 11.5, p = 0.009). On the other hand, polytherapy AED regimen was not associated with a worse perception of AEs (40.9 ± 11.5 vs. 37.0 ± 11.6, F = 0.2, p = 0.5).

We examined the possibility that patients taking antidepressant medications and patients with persistent seizures had higher total AEP scores because of being symptomatic with a mood and/or anxiety disorder. To that end we compared the total AEP scores between symptomatic and asymptomatic patients in those on and off antidepressant drugs and in patients who were seizure-free and patients with persistent seizures. Our results revealed that the significantly higher AEP scores among patients taking antidepressant medication or with persistent seizures were a function of having a psychiatric disorder. For example, the AEP score among asymptomatic patients did not differ whether they were seizure-free (n = 51, AEP: 30.3 ± 7.7) or not (n = 42, AEP: 34.4 ± 7.6). Likewise, the AEP score among symptomatic patients did not differ whether they were seizure-free (n = 32 AEP: 44.7 ± 10.7) or not (n = 48, AEP: 47.9 ± 10.5).

Among the 43 patients taking antidepressant medication, 31 (72%) continued to be symptomatic (AEP 49.3 ± 9.1). This score did not differ with that of the 55 symptomatic patients who were not taking antidepressant medication (AEP 45.6 ± 11.1). Likewise, there was no difference in the AEP score among the asymptomatic patients on (n = 12, AEP 37.7 ± 8.2) and off antidepressants (n = 90, AEP 31.3 ± 7.3). Furthermore, although the choice of antidepressant drug was correct in all 31 symptomatic patients treated pharmacologically, the dose was suboptimal in 26 (84%). Yet, these findings do not totally exclude the possibility that in some selected cases selective serotonin reuptake inhibitors (SSRIs) may exacerbate adverse events of some AEDs when used in combination, including tremor, sedation, weight gain, as well as adverse events the patient may be unaware of such as hyponatremia.

Effect of symptomatic state on individual items of the 15-item AEP

Significant correlations were found in all 15 items, with the higher correlations being identified in cognitive (difficulty with concentration [r = 0.45, p < 0.0001] and memory problems [r = 0.46, p < 0.0001]) and neurovegetative symptoms (disturbed sleep [0.45, p < 0.0001] and sleepiness [r = 0.46, p < 0.0001]). Correlations of the remaining symptoms were as follows: tiredness (r = 0.43, p < 0.0001), dizziness (r = 0.42, p < 0.0001), unsteadiness (r = 0.42, p < 0.0001), double or blurred vision (r = 0.34, p < 0.0001), upset stomach (r = 0.32, p < 0.0001) and to a lesser degree, shaky hands (r = 0.26, p < 0.0001) hair loss, (r = 0.27, p < 0.0001), headache (r = 0.25, p = 0.001), trouble with mouth and gums (r = 0.24, p = 0.001), problems with skin (r = 0.22, p = 0.002), and weight gain (r = 0.2, p = 0.006).


The findings of this study demonstrate that patients with SSDE, MDE, anxiety disorders, or mixed MDE/anxiety disorders endorse worse AED-related AEs than asymptomatic patients. Exclusion from the AEP of the four psychiatric items that can be as well the expression of a mood and anxiety disorder did not change the findings obtained with the original instrument, which demonstrates that the higher AEP scores in symptomatic patients were not a function of circular reasoning.

Our results suggest that the severity of AEs did not differ between patients with anxiety disorders and MDE. However, the AEs worsened when the two disorders occur together in patients with more than one type of anxiety disorder. On the other hand, the significant effect of SSDE on AEs highlights the fact that this phenomenon is not restricted to the more severe forms of mood disorder. The similar effect of SSDE and MDE on AEs was surprising and counterintuitive and may be related to the small number of patients with MDE. Therefore, a repeat study with a larger number of patients is needed to answer this question.

In univariate analyses, having persistent seizures and being on an antidepressant drug were associated with a higher AEP score. However, these differences were found to be related to experiencing a psychiatric disorder. Our data, however, cannot exclude the possibility that SSRIs can worsen adverse events caused by AEDs, as indicated in our results.

The data from this study suggest that the presence of comorbid mood and anxiety disorders should be investigated in patients reporting frequent AED-related AEs. Likewise, when using the AEP to track severity of AED-related AEs, the presence of high scores should prompt clinicians to rule out comorbid mood and anxiety disorders.

Furthermore, these data can have significant implications in regulatory AED trials, as comorbid mood and anxiety disorders are relatively frequent in patients with treatment-resistant epilepsy who are enrolled in most of these studies. Indeed, these patients appear to be more likely to report AEs, which consequently may lead to false-positive results and an inaccurate toxicity profile of a drug. Furthermore, placebo-controlled drug trials do not mitigate this risk because while most trials exclude patients with severe forms of psychiatric illness, they do not exclude those with SSDE or anxiety disorders, and randomization to active drug or placebo does not take into account the presence of these psychiatric comorbid conditions. Clearly, it is imperative that the effect of mild psychiatric comorbidities on AED-related AEs be factored-in in the interpretation of AED-related AEs in regulatory double-blind placebo controlled trials.

Screening for psychiatric comorbidities should not be limited to depressive episodes (as most studies do at the present time), but should include as well screening instruments for anxiety disorders. As shown by our data, comorbid anxiety disorders are as frequent as mood disorders, as 49 (57.6%) of the 85 symptomatic patients presented with one or more than one anxiety disorder with and without concurrent MDE. The prevalence of anxiety disorders in this cohort was 26%, which is comparable to data reported in population-based studies (Tellez-Zenteno et al., 2007).

Clearly, these data suggest that PWE should be screened for symptoms of depression and anxiety disorders. Different self-rating instruments to screen for depression can be used, including the BDI-II and CES-D, as used in this study. If the intent is to screen for MDE, specifically, the Neurologic Disorders Depression Inventory-Epilepsy (NDDI-E) may be appropriate (Gilliam et al., 2006), whereas the Patient Health Questionnaire Generalized Anxiety Disorder is an acceptable option to screen for symptoms of generalized anxiety (Kroenke et al., 2007), and the Hospital Anxiety and Depression Scale (HADS) is a self-rating instrument that can be used to screen for both depression and anxiety (Zigmond & Snaith, 1983).

The small number of symptomatic patients is one of the limiting factors of our study, particularly the small number of patients with MDE. Clearly, these data must be replicated in larger samples of patients. In conclusion, the presence of comorbid depressive and anxiety disorders has a negative impact on AED-related AEs and must be investigated in the clinic as well as in regulatory trials.


This study was funded by an unrestricted grant by GlaxoSmithKline. Dr. Kanner has received research funding from GlaxoSmithKline, Novartis, and Pfizer. Dr. Meador has received research funding from GlaxoSmithKline, Eisai, Marius, Myriad, NeuroPace, and SAM Technology. Dr. Meador is a consultant for the Epilepsy Study Consortium for which funds are received by Emory University toward his salary; he has consulted in this regard for NeuroPace and Upsher Smith Laboratories. Dr. Barry has received research funding from GlaxoSmithKline and Forest Pharmaceuticals and grants from the American Epilepsy Society and the Epilepsy Foundation. Dr. Hermann has nothing to disclose. Dr. Gilliam has served on a scientific advisory board for Cyberonics, and has received speaking honoraria from Ortho-McNeil and Eisai. We confirm that we have read the Journal’s position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.