Cognitive behavior therapy for depression in people with epilepsy: A systematic review


  • Milena Gandy,

    Corresponding author
    1. The School of Psychology, University of Sydney, Sydney, New South Wales, Australia
    2. The Centre for Emotional Health (CEH), Department of Psychology, Macquarie University, Sydney, New South Wales, Australia
    • Address correspondence to Milena Gandy, The School of Psychology, University of Sydney, Brennan MacCallum Bld (A18), Sydney, NSW 2006, Australia. E-mail:

    Search for more papers by this author
  • Louise Sharpe,

    1. The School of Psychology, University of Sydney, Sydney, New South Wales, Australia
    Search for more papers by this author
  • Kathryn Nicholson Perry

    1. Centre for Health Research, School of Social Sciences and Psychology, University of Western Sydney, Bankstown, New South Wales, Australia
    Search for more papers by this author


Cognitive behavioral therapy (CBT) is a recommended treatment for depression in people with epilepsy (PWE); however, a recent Cochrane review found that there was insufficient evidence that any psychological therapy is effective. This conclusion provides little help to clinicians who provide interventions for depressed PWE. The aim of this review was to systematically and qualitatively review the literature on the efficacy of CBT for depression in PWE based on randomized controlled trials (RCTs) and case series. We aim to determine patterns in the literature to inform the type of CBT, if any, that should be offered to PWE who are depressed. Databases MEDLINE, PsycINFO, and the Cochrane EBM Reviews were searched via OVID. Selection criteria included the following: (1) participants with epilepsy; (2) use of CBT; (3) valid depression outcome measure; and (4) published in peer-reviewed journal in English. Inclusions of studies were assessed by two independent researchers. We identified 14 outcome papers for 13 CBT trials including 6 randomized controlled trials (RCTs) and 7 case series. Positive effects of CBT on depression were reported in three of six RCTs. A review of their content revealed that all effective RCTs specifically tailored CBT to improve depression. Conversely, two of three RCTs that failed to find depression-related effects focused on improving seizure-control. This pattern was also observed in the case series. Although limited in number and having methodologic limitations, the treatment studies included in our review suggest that interventions tailored toward improving depression are possibly efficacious, whereas those that focus on improving seizure control do not appear to be. However, this review highlights that there is need for further RCTs in this area in order to confirm the possible efficacy of CBT for depression in PWE.

Despite the high prevalence and significant consequences of comorbid depression in people with epilepsy (PWE), it often remains untreated (Hermann et al., 2000; Jones et al., 2003; Barry et al., 2008; Kanner, 2011). Physician reluctance to place PWE on antidepressant medication, due to fears of lowering seizure threshold and adverse drug interactions between antidepressants and antiepileptic drugs, has been identified to be a major barrier for the treatment of depression in PWE (Barry et al., 2008; Cotterman-Hart, 2010; Noe et al., 2011). As a result, effective psychological therapies aimed at improving depression in PWE would be of particular value (Barry et al., 2008; Ramaratnam et al., 2008; Kerr et al., 2011).

A recent Cochrane review assessed the findings of 16 randomized controlled and quasi-randomized trials of psychological treatment in PWE (Ramarantnam et al., 2008). The review found no reliable evidence that any psychological treatment—including relaxation therapy, electroencephalography (EEG) biofeedback, and cognitive behavioral therapy (CBT)—led to significant improvement in seizure control or improvement in quality of life (QoL). The treatments included in the review were varied in therapeutic content, rated as methodologically weak, and yielded mixed results (Ramarantnam et al., 2008). Despite the lack of conclusive evidence for effective treatments for depression in PWE, the recent international consensus clinical practice statement for the treatment of neuropsychiatric conditions associated with epilepsy recommended CBT as the psychological treatment of choice for depression in PWE (Kerr et al., 2011).

CBT has been found to be an effective treatment for depression in the general population (Tolin, 2010; Hofmann et al., 2012) as well as in a number of health populations, including multiple sclerosis (Mohr et al., 2001), chronic pain (Eccleston et al., 2003), rheumatoid arthritis (Sharpe et al., 2001), and chronic obstructive pulmonary disease (COPD; Livermore et al., 2010). One of the significant benefits of CBT as a form of therapy is its ability to be adapted to address specific issues for different health populations (Krishnamoorthy, 2003). However, despite these numerous recommendations, the effectiveness of CBT to improve depression in PWE remains unclear.

The recent Cochrane review was limited to randomized controlled trials (RCTs) and included only three RCTs of CBT for depression. An additional three trials have since been conducted in this area. Therefore, the first aim of this systematic review was to provide the first review of all the published evidence for CBT interventions for PWE to improve depression outcomes. The second aim was to assess whether the literature allows any recommendations as to which CBT strategies are most strongly empirically supported for the treatment of depression in PWE, which could help clinicians provide treatments for PWE who have mood disturbances that are more strongly evidence based.


Using the Meta-analysis of Observational Studies in Epidemiology (MOOSE) criteria as a guide (Stroup et al., 2000), we conducted a literature search of published studies in April 2012. The electronic databases MEDLINE, PsycINFO, and the Cochrane EBM Reviews—Cochrane Central Register of Controlled Trial were searched via OVID using different combinations of the following words: epileps*, seizure, depression, depressive symptoms, mood, cognitive therapy, behaviour therapy, problem solving therapy, mindfulness therapy, and psychotherapy. The search strategy modeled but expanded upon the Cochrane review of psychological treatment in epilepsy search strategies (Ramarantnam et al., 2008). There were no limits placed on the years for searched articles. All medical Subject Heading (MeSH) terms were exploded to broaden the search for relevant studies. In addition, the ancestry method, which involves hand searching the reference list of empirical studies and reviews, was used to identify further studies. To meet the inclusion criteria, the study needed to: (1) involve participants with a diagnosis of epilepsy, (2) use a form of CBT, (3) use established measures of depression as outcome measures, and (4) be published in a peer-reviewed journal in English.

Figure 1 displays the process of study selection for the review. We identified 133 articles. All titles and abstracts were reviewed to determine their relevance, by one reviewer (MG), who is a registered psychologist and PhD candidate. Reasons for exclusion are outlined in Box A. A random selection of 25% (33) of abstracts was assessed by an independent assessor (LS), who is a professor of clinical psychology, to ensure interrater reliability, where there was 100% agreement. Following this, 32 abstracts were retained, and full copies of the articles were reviewed. Two raters (MG and LS) independently assessed each article for eligibility, as outlined in Box B. Three additional articles were retrieved via the ancestry method. A list of excluded RCTs is included in Table S1. Disagreement between reviewers was resolved by consensus; interrater reliability had a kappa coefficient of 0.86.

Figure 1.

Search process utilized and the outcomes obtained in identifying articles for the review. CBT, cognitive, behavioral therapy, PNES, psychogenic nonepileptic seizures, PWE, people with epilepsy.

All RCTs were rated using the PEDro-P scale. The PEDro-P is a modified version of the PEDro scale, which was initially developed to rate the quality of RCTs for the purpose of systematic reviews, on the Physiotherapy Evidence Database (Maher et al., 2003). The scale has been modified by researchers who manage the Psychological Database for Brain Impairment Treatment Efficacy (Psychbite), to tailor it to the analysis of treatment trials in the area of brain impairment, including epilepsy. The PEDro-P was selected because it provides a quality score based on strict internal validity criterion for RCTs, which were established by strict Delphi technique procedures (Verhagen et al., 1998). The PEDro-P quality score allows for comparison between RCTs and has been found to have good reliability (Maher et al., 2003). The scale contains 11 items, one question assesses external validity and the remaining 10 assess internal validity. The PEDro-P items are scored as yes (1) or no (0). A copy of the items and the PEDro-P scores for each RCT have been reported in Table S2.

Given that case series provide only Level IV Evidence for the efficacy of an intervention, according to the National Health and Medical Research Council (NHMRC) of Australia definitions (2009), we include them in this review in order to determine whether the patterns that are identified from RCTs are also supported in the case series that appear in the literature. However, we have not assessed case series for quality, since by definition internal and external validity are likely to be relatively low.


Trial design

Fourteen articles reporting 13 CBT studies were identified. One paper was a long-term follow-up analysis of one of the RCTs (Chaytor et al., 2011), and was therefore not considered as a separate study. A detailed summary of the studies is provided in Table 1. Six of the studies were RCTs (Davis et al., 1984; Tan & Bruni, 1986; Martinović et al., 2006; Ciechanowski et al., 2010; Thompson et al., 2010; McLaughlin & McFarland, 2011), thus providing NHMRC Level II Evidence (NHMRC, 2009). Four of the six RCT designs employed simple comparisons of CBT with a waitlist control (WLC) or treatment as usual (TAU). One trial compared CBT with both WLC and TAU and another compared CBT with relaxation. The remaining seven articles were treatment trial case series designed studies (n = 7).

Table 1. Summary of RCTs reviewed
Author (year, country)ParticipantsCondition and interventionAssessedSignificant group × time interactionap-Valueb
SampleCBT male (%)CBT age (years) Mean (SD)Study designTherapy deliveryTotal period and amountDepression outcome measureDropout rate (%)
  1. BDI, Beck Depression Inventory; CBT, cognitive behavioral therapy; CESD, Centre for Epidemiological Study of Depression Scale; CIDI-Auto, Composite International Diagnostic Interview-Computerized; FU, follow-up; GDS, Geriatric Depression Scale; HAMD, Hamilton Depression Scale; HSCL, Hopkins Symptom Checklist; h, hours; mo, Month; MDD, major depressive disorder; N, number; NR, not reported; RA, research assistant, SC, standard care; TAU, treatment as usual; UC, usual care; vs., versus; wk, week; WLC, waitlist control.

  2. a

    There was some discrepancy between studies in the reporting of statistics. Where reported, we include the actual p-values, but where not available we have simply included reference to whether or not they are significant.

  3. b

    Pedro-P ratings scored out of 10.

  4. c

    Mean visits for PEARLS intervention was 6.2 visits = 6.2 h (SD = 3; median = 8, range = 0–8; four had no visits, 72.5% had ≥6 visits).

  5. d

    Thirty-two participants took part in at least half of the sessions = 4 h.

  6. e

    This study also used a modified version of the measure of the BDI, which included a positive response category for each item. As this measure contained the same information as the BDI, and produced the same outcomes, it was not considered as a separate measure.

  7. f

    Participants attended an average of 7.4 sessions = 14.8 h.

  8. g

    Although this study also used the Depression Adjective Checklist and Generalised Content Scale, these are not validated measures of depression and thus excluded for the review.

McLaughlin & McFarland (2011, Australia)N = 37; CBT n = 18, ≥60 years old4468 (7)CBT vs. RelaxationGroups run by psychologist6 weekly × 2 h sessions = 12 hCIDI-Auto (depression and dysthymia) GDS0Pre, post, 3-mo FUNo 3/3 measures at post and 3-mo FU6

Ciechanowski et al. (2010, U.S.A.)


FU Chaytor et al. (2011)

N = 80; CBT n = 4052.543 (11)CBT vs. UCIndividual home based by masters level social workers

8 × 1 h visits over 19 weeks = 8 h+c

5–10 min phone calls

Monthly over 12 months


HSCL Suicide items (measured at 12 months)

12.5 at 6 and 12 mo FUPre, 6 and 12-mo FU

No 1/1 (p = 0.096) at 6-mo

Yes 2/2 HSCL (p = 0.003)

Suicide items (p = 0.025) at 12 mo:

No 2/2 measures at 18-mo

Thompson et al. (2010, U.S.A.)N = 32, CBT n = NR2336 (NR)CBT vs. WLCHome-based groups via net or phone run by layperson with epilepsy and RA8 weekly × 1 h sessions = 8 hdBDIe36Pre, post (8 wk)Yes 1/1 measure (p = 0.001) at post4
Martinović et al. (2006, Serbia)N = 30; CBT n = 154017 (3)CBT vs. TAUIndividual by NR

8 × 1 h sessions over 2 months = 8 h+

4 × 1 h boosters sessions monthly = 4 h




6Pre, 6 and 9-mo FU

Yes for 3/3 measures (p ≤ 0.050) at 6 mo:

Yes 3/3 measures (p ≤ 0.050) at 9-mo

Tan & Bruni (1986, Canada)N = 27; CBT n = 837.533 (11)CBT vs. SC vs. WLCGroups run by NR8 weekly × 2 h = 16 hfBDI10Pre, post, 4-mo FUNo 1/1 measure at post and 4-mo5
Davis et al. (1984, U.S.A.)N = 13; CBT n = 812.533 (11)CBT vs. WLCGroup run by social workers.6 weekly × 2 h = 12BDIg13Pre, post, 6-wk FUNo 1/1 (p = 0.100)2

Trial quality

All trials had a total score of 1 for external validity on the PEDro-P, as they each specified their trial eligibility criteria. For internal validity, the PEDro-P scores ranged from 2 to 7 (M = 5, standard deviation [SD] = 1.8). All trials used random allocation, with only one failing to conceal allocation (Tan & Bruni, 1986). Two studies failed to report baseline comparability between their treatment groups (Davis et al., 1984; Tan & Bruni, 1986). Only two studies used blind assessors to assess at least one key outcome measure (Tan & Bruni, 1986; Ciechanowski et al., 2010). Three had dropout rates in excess of 12% (Davis et al., 1984; Ciechanowski et al., 2010; Thompson et al., 2010). Only two trials reported an intention-to-treat analysis (Tan & Bruni, 1986; Ciechanowski et al., 2010). Between-group statistical comparisons were reported in all but one trial (Davis et al., 1984). Point estimates and variability for treatment effects were reported in only half of the trials (Martinović, 2001; Ciechanowski et al., 2010; McLaughlin & McFarland, 2011).

Randomized controlled trial details


When the six RCTs are considered, a total of 247 participants entered these trials (mean 41 (SD 22); range 15–80); of these 219 (89%) completed the majority of treatment (mean 37 (SD 23); range 13–80). The mean percentage of male participants in CBT groups was only 25% (SD 15; range 12.5–52.5). The sample had a mean age of 38 years (SD 17; range 17–68).

Only three trials stipulated specific epilepsy criteria. One included participants with newly diagnosed epilepsy (Martinović et al., 2006), another excluded newly diagnosed patients (Thompson et al., 2010), and one included only patients with inadequate seizure control (Tan & Bruni, 1986). One study specifically treated adolescent patients (13–19 years old; Martinović et al., 2006) and another older adults (≥60 years old) only (McLaughlin & McFarland, 2011).

Treatment delivery

Of the six RCTs, four examined group therapy (Davis et al., 1984; Tan & Bruni, 1986; Thompson et al., 2010; McLaughlin & McFarland, 2011), one using “virtual groups” via the internet or telephone (Thompson et al., 2010). Of the four programs that described the qualifications of facilitators, one used lay therapists (Thompson et al., 2010), one used psychologists (McLaughlin & McFarland, 2011), and the remaining two were facilitated by social workers (Davis et al., 1984; Ciechanowski et al., 2010).

The mean total treatment duration for the RCTs was 9 weeks (SD =5, range 6–19 weeks) and the mean number of hours in treatment was 11 h (SD 3, range 8–16 h). All trials offered weekly or every two week therapy. Two trials provided booster sessions and support for rehearsal of skills learned in the primary intervention phase (Martinović et al., 2006; Ciechanowski et al., 2010).

Depression measures

Three (of six) RCTs specified depression status criteria. One only included participants with a formal diagnosis of depression (Ciechanowski et al., 2010). Another included participants defined as having “subthreshold depression,” whereby patients reported elevated symptoms of depression on measures (Beck Depression Inventory [BDI], or the Center for Epidemiological Study of Depression measure [CES-D], and Hamilton Depression Scale [HAMD]), but did not necessarily meet diagnostic criteria for depression (Martinović et al., 2006). The third trial selected participants based on their scores on the CES-D (Ciechanowski et al., 2010; Thompson et al., 2010), such that only those with scores in the range of >13 to <38 were included (Thompson et al., 2010). Only one of the six studies used a diagnostic interview (McLaughlin & McFarland, 2011), although another used the HAMD to assess clinician-rated symptom severity (Martinović et al., 2006). The remainder relied on patient self-report scales, including the BDI (4/6; Davis et al., 1984; Tan & Bruni, 1986; Martinović et al., 2006; Thompson et al., 2010), CES-D (1/6; Martinović et al., 2006), Hopkins Symptom Checklist-20 (HSCL-20 [1/6]; Ciechanowski et al., 2010), and the Geriatric Depression Scale (GDS; [1/6]; McLaughlin et al., 2008). Only one of the six studies assessed suicidal ideation (Ciechanowski et al., 2010).

RCT CBT components focus

Table 2 provides a detailed description of the CBT components in each of the treatment trials. Five (of the six) trials explicitly noted that the aim of the study was to evaluate the benefits of CBT for improving depression outcomes (Davis et al., 1984; Martinović et al., 2006; Ciechanowski et al., 2010; Thompson et al., 2010; McLaughlin & McFarland, 2011); one trial instead referred to improvements in psychopathology (Tan & Bruni, 1986). Three (of six) trials also aimed to evaluate the benefits of CBT to improve seizure control (Tan & Bruni, 1986; Martinović et al., 2006; McLaughlin & McFarland, 2011) and four (of six) to improve quality of life or psychosocial wellbeing (Tan & Bruni, 1986; Martinović et al., 2006; Thompson et al., 2010; McLaughlin & McFarland, 2011). Thompson et al. (2010) also aimed to assess the effects of CBT for increasing knowledge/skills and self-efficacy. An important distinction can be made in terms of the primary focus of these CBT interventions. Four (of six) RCTs primarily tailored CBT to improve depression and coping in PWE (Davis et al., 1984; Martinović et al., 2006; Ciechanowski et al., 2010; Thompson et al., 2010), whereas the remaining two focused on seizure control.

Table 2. CBT components of RCT
RCTSignificant effects on depressionCBT seizure-focused contentCBT depression-focused content
  1. CBT, cognitive behavioral therapy; MBCT, mindfulness based cognitive therapy; N, no; Psychoed, psychoeducation; RCT, randomized controlled trial; UPLIFT, using practice, learning to increase favorable thoughts; Y, yes.

  2. 1 = McLaughlin & McFarland (2011); 2 = Ciechanowski et al. (2010); 3 = Thompson et al. (2010); 4 = Martinović et al. (2006); 5 = Tan & Bruni (1986); 6 = Davis et al. (1984).

1NReiter behavioral intervention: Self-monitoring; psychoed of physical and emotional healthy lifestyle and medication issues; understanding epilepsy and identifying auras; identifying, avoiding and controlling seizure triggers; CBT and seizures 
2YPhysical activity schedulingPEARLS program: problem-solving therapy; Social and Physical activity scheduling
3Y Project UPLIFT (MBCT): psychoed for depression and epilepsy, thought monitoring; identifying cognitive distortions; self-esteem; problem identification; goal setting; identifying supports; relaxation exercises; mindfulness activities, e.g., attention to breath
4Y Activity plans; relaxation; identifying and challenging thought distortions; social skills and problem-solving training
5NSelf-monitoring of stressful events and seizures; relaxation and coping skills training for stress and seizure control (stress inoculation approaches; cognitive restructuring; assertiveness and social skills rehearsal; problem-solving; discussion of future plans and vocational issuesReading about coping with depression and stress
6N Increasing pleasant events and physical exercise, increasing assertiveness; identifying self-talk (fault thinking); thought-stopping; increasing positive cognitions and coping statements

Effects of the intervention on depression

Because of the poor reporting of some trials we were unable to perform a meta-analysis. Instead the results are summarized in relation to those that reported significant and nonsignificant effects on depression outcome measures. Across the six RCTs, there were 11 outcomes of depression reported. Positive results were described in 6 of 11 of these measures at one or more assessment point. The six observations came from three (of six) trials. The first RCT found significant positive long-term effects of CBT compared to TAU (Martinović et al., 2006). Large treatment effects were found for the BDI (d = 0.85), the CES-D (d = 0.86), and the HAMD (d = 1.5) at 6 and 9 months (BDI [d = 0.85], HAMD [d = 1], CES-D [d = 0.65]). The second RCT also found a large treatment effect (d = 0.75) in HSCL-20 scores at posttreatment, and in suicidal ideation at 12-month follow-up (Ciechanowski et al., 2010), but not at 6- or 18-month follow-up (Chaytor et al., 2011). The final RCT to report positive findings reported significant improvements on the BDI following a group based CBT program compared to WLC (Thompson et al., 2010).

Of the three RCTs with nonsignificant effects of treatment on their depression measures, the most recent failed to find significant changes on three measures of depression in older adults who had received a CBT program compared to those who received a relaxation program (McLaughlin & McFarland, 2011). This included the GDS immediately and 3 months posttreatment, and the Composite International Diagnostic Interview (CIDI-Auto) for a diagnosis of depression and dysthymia at 3 months. The trial did, however, report main effects for time, whereby both the CBT and relaxation program experienced significant improvements on the GDS and CIDI dysthymia diagnosis. The other two RCTs failed to find significant differences in BDI scores in participants who received CBT compared with control groups (Davis et al., 1984; Tan & Bruni, 1986).

Comparison of effective and noneffective trials

As displayed in Table 2, a significant pattern emerges when the CBT intervention focus of effective and ineffective RCTs is compared. It is important to note that all three effective RCTs were focused primarily on improving symptoms of depression as opposed to seizure control (Martinović et al., 2006; Ciechanowski et al., 2010; Thompson et al., 2010). On the other hand, two (of the three) ineffective RCTs were focused primarily on seizure control (Tan & Bruni, 1986; McLaughlin & McFarland, 2011). Also displayed in Table 1, all three RCTs that failed to find significant effects were group-based interventions (Davis et al., 1984; Tan & Bruni, 1986; McLaughlin & McFarland, 2011), whereas two (of three) effective RCTs used an individual therapy approach (Martinović et al., 2006; Ciechanowski et al., 2010). There were no systematic differences in length of therapy, and it is important to highlight that two (of three) effective RCTs utilized booster therapy sessions. Although overall there were no clear differences in PEDro-P scores between effective and ineffective trials, it is notable that the only trial to tailor CBT to improve depression that subsequently failed to find any therapeutic benefit was the lowest quality of all the trials (Davis et al., 1984).

Case series details

A summary of the case series studies is included in Table 3.

Table 3. Summary of case reviews
Author (year, country)SampleMale (%)Age (years) Mean (SD)Therapy deliveryPeriod and amountDepression outcome measureSignificant improvements in depression
  1. AED, antiepileptic drug; BDI, Beck Depression Inventory; CBT, cognitive behavioral therapy; CDI, Children's Depression Inventory; FU, follow-up; GDS, Geriatric Depression Scale; HADS-D, Hospital Anxiety and Depression Scale-Depression; h, hours; JME, juvenile myoclonic epilepsy; MDD, major depressive disorder; N, number; NR, not reported; POMS, profile of mood states; TLE, temporal lobe epilepsy.

  2. a

    Participated attended an average of 15 session = 22.5 h.

  3. b

    Participants required to attend 6 of 10 sessions = 12 h.

Crail-Melendez et al. (2012, Mexico)N = 16 TLE outpatients with MDD634 (9)Group run by CBT therapist and psychiatrist16 weekly × 1.5 h sessions = 24 h aBDIYes 1/1 measure from pre- to mid-treatment (session 8; p = 0.0001) and posttreatment (session 16; p = 0.001)
Elsas et al. (2011, U.S.A.)N = 82541 (8)Individual by epilepsy therapist1 h weekly meetings over 6 months = 24 hPOMSYes for 1/1 measure pre to post (p ≤ 0.001), including the depression-dejection scale (p ≤ 0.05) but not pre to FU (6 months)
Macrodimitris et al. (2011, Canada)N = 183941 (11)Groups run by clinical psychologist and social worker10 weekly × 2 h = 20 hbBDI-IIYes in 1/1 measure between pre and post (p ≤ 0.001)
Wagner et al. (2010, U.S.A.)N = 9 children and parents55.513 (2)Child group run by pediatric psychologist and parent groups by social worker8 weekly × 1 h + combined session = 8.5 hCDINil 1/1 measure
Snead et al. (2004, U.S.A.)N = 7 adolescents4315 (18)Adolescent group and Parent group run by a psychologist6 weeks × 1 h 6 hCDINil 1/1 measure
Goldstein et al. (2003, United Kingdom)N = 6 poorly controlled seizures039 (16)Individual run by clinical nurse specialist12 weekly × 1 h = 12 hHADS-DNil 1/1 measure
Spector et al. (1999, United Kingdom)N = 714.340 (NR)Groups run by psychologist and occupational therapist8 weekly × 2 h = 16 hHADS-DNil 1/1 measure


The characteristics of participants in the case series were similar to those within the RCTs. There were 89 participants recruited into seven studies (mean 12 (SD 6); range 6–23) with a 19% dropout rate. A similar proportion of males (26%; SD 21; range 0–55.5) of a similar age (32 years; range 13–41) was evident. Most studies (n = 4) included participants with specific epilepsy criteria, temporal lobe epilepsy (TLE; Crail-Melendez et al., 2012), and poorly controlled epilepsy (Spector et al., 1999; Goldstein et al., 2003; Elsas et al., 2011). Recruitment was usually through tertiary referral services (6/7). Two studies excluded participants with depressive disorders (Wagner et al., 2010; Elsas et al., 2011), whereas one only included participants with major depressive disorder (MDD) not currently taking antidepressants (Crail-Melendez et al., 2012).


In terms of the treatments administered, the majority were group interventions (5/7) of slightly longer duration than in the RCTs (12 sessions over 16 h). Two (of seven) specifically targeted children with epilepsy and their parents (Snead et al., 2004; Wagner et al., 2010), and the remainder (5/7) targeted adults.

All studies relied on symptom measures of depression, including the BDI (n = 2; Macrodimitris et al., 2011; Crail-Melendez et al., 2012), the Hospital Anxiety Depression Scale (HADS [n = 2]; Spector et al., 1999; Goldstein et al., 2003), the Children's Depression Inventory (CDI; [n = 2]; Snead et al., 2004; Wagner et al., 2010), and the Profile of Mood States (POMS; [n = 1]; Elsas et al., 2011).

Case series

Each of the seven case series had a single depression outcome measure point. Positive effects of CBT were reported in three of seven studies (Elsas et al., 2011; Macrodimitris et al., 2011; Crail-Melendez et al., 2012), a proportion similar to that of the RCTs. There were no clear patterns in terms of participant characteristics or length of intervention. However, case series studies were explored to determine whether the same pattern of results was evidenced as discussed above. Two (of three) effective case series were depression-focused CBT interventions (Macrodimitris et al., 2011; Crail-Melendez et al., 2012), whereas, only one (of five; Elsas et al., 2011) that focused on seizure frequency was effective (Spector et al., 1999; Goldstein et al., 2003; Snead et al., 2004; Wagner et al., 2010). Furthermore, this study involved a high dose of CBT (6 months) and effects were lost at 6 months posttreatment (Elsas et al., 2011).


Although previous systematic reviews have concluded that there is insufficient evidence to support psychosocial treatments for depression in PWE (Ramarantnam et al., 2008), this is the first review to focus exclusively on CBT recently recommended for use in PWE (Kerr et al., 2011). This systematic review draws slightly more optimistic conclusions. We found a pattern of results that suggested that there was limited support for any beneficial effects on depression from CBT that focused on seizure control. That is, of the two RCTs and five case series that tested this variant of CBT, only one case series demonstrated positive effects on depressive symptoms. Furthermore, this effect was found following a significantly higher dose of CBT (6 months) compared to other trials where effects failed to remain 6 months posttreatment. On the other hand, CBT focusing specifically on depression did appear, based on the majority of RCT studies (three of four), to result in improvements in depressive symptoms for patients with epilepsy. Furthermore, this pattern of results was supported from the available case series. It should be noted that these observations are tentative and require further investigation. Before considering these findings in further detail, a comment on limitations of the literature need to be acknowledged, since they render any conclusions preliminary.

The RCTs included in this systematic review had a range of limitations. On average, trials were rated as being of moderate quality on the PEDro-P. Sample sizes were small, with some studies including fewer than 10 participants in the CBT arm (Davis et al., 1984; Tan & Bruni, 1986), with an underrepresentation of male participants. Dropout rates were variable, and in some studies were quite large (e.g., 36%). Within a few studies, participants often failed to attend all treatment sessions, resulting in questions about whether all completers were exposed to a sufficient dosage of the therapy. The trials also failed to comment on adherence protocols relating to their CBT interventions. Furthermore, only two (of six) trials reported intention-to-treat analyses, and no studies systematically investigated the characteristics of those who dropped out during treatment. In addition, it is possible that the findings of this review are influenced by a publication bias toward studies reporting positive outcomes, meaning that the results inflate the likelihood of positive results.

In addition, there was a range of differences in the therapy that was provided, including mode of administration, qualifications of the facilitators, provision of booster sessions, and different session content. In fact, in one study, CBT was augmented with antidepressant medication review where participants failed to respond to the initial course of therapy (Ciechanowski et al., 2010). It is therefore difficult to confidently ascribe differences in outcome to a single factor. It is important to note that the majority of trials utilized symptom measures of depression as opposed to a clinical diagnosis. The effects of CBT on well-defined depression criteria therefore are lacking.

Notwithstanding these issues, the fact that there were some consistent differences between those RCTs that found positive benefits for participants and those that did not is important in guiding clinicians and researchers. Our review indicated a relatively consistent finding, that the majority of CBT trials (n = 2 of 2) and case series (n = 4 of 5) that were focused on improving seizure control were ineffective in improving depression. This finding is not surprising when considered among research reviews that suggest the relationship between depression and epilepsy variables, including seizure frequency and severity, is weak (Lambert & Robertson, 1999; Hermann et al., 2000; Marsh & Rao, 2002; Seethalakshmi & Krishnamoorthy, 2007). Given the current poor evidence of a strong relationship between seizure frequency and depression in PWE, it is not surprising that treatment aimed at reducing seizures would fail to improve depression-related outcomes.

In contrast, the relatively consistent pattern that effective RCTs (n = 3 of 3) and case series (n = 2 of 2) were those that tailored CBT to improving depression in PWE suggests that these forms of CBT intervention are likely efficacious. This is particularly supported by the finding of medium to large effects for CBT interventions on various depression outcomes over long-term follow-up periods in two of the three effective RCTs (Martinović et al., 2006; Ciechanowski et al., 2010). However, before firm conclusions about the effectiveness of depression-related CBT programs in PWE can be made, further RCTs in this area are necessary. Of importance, CBT was found to be effective in adults and adolescents with and without depressive disorders across the three effective trials (Martinović et al., 2006; Ciechanowski et al., 2010; Thompson et al., 2010). This is important, because many PWE experience postictal, atypical, or subclinical symptoms of depression that fail to meet diagnostic criteria for a formal depressive disorder or are short-lived, but may still contribute to poorer QoL or place them at risk for developing clinically identifiable depression if left unmanaged (Kanner, 2009). Unfortunately, because the trials reviewed here did not provide diagnostic information, it is unknown whether these varying presentations respond equally to intervention and future research is needed to address this issue.

The findings of Martinović et al. (2006) suggest that CBT can prevent the occurrence of future depressive disorders in adolescents with new-onset epilepsy. Reserving the use of treatment protocols for depressive symptoms only among those meeting diagnostic criteria is not indicated according to the available evidence. Furthermore, the potential preventative role of CBT in adults with epilepsy needs to be assessed further.

Our comparison of effective and ineffective trials also suggests that CBT may be more effective if delivered individually, and that long-term benefits are most likely with the inclusion of booster sessions. Booster sessions provide opportunities to revise and practice CBT skills, which may be particularly important for adults with epilepsy who are known to be at a high risk of memory impairments that may affect the retention of CBT skills delivered in therapy (Elger et al., 2004; Vingerhoets, 2006).


Depression in PWE is a highly prevalent and disruptive condition that often goes undetected and untreated. Our review suggests that CBT interventions aimed at seizure control are unlikely to confer depression-related benefits. Alternatively, our review suggests that interventions tailored toward improving depression are likely efficacious, thus providing some support for the recent international clinical practice recommendations that CBT be used to treat depression in PWE (Kerr et al., 2011). However, it should be noted that these suggestions are preliminary and require additional research support before firm conclusions regarding the potential superiority of depression-focused CBT over seizure-control focused can be established. The pattern of results also suggests that individual therapy with booster sessions is most likely to be effective for PWE. Although a more optimistic conclusion than the recent Cochrane review (Ramarantnam et al., 2008), we also found that the treatment trials reviewed had numerous methodologic limitations that preclude definitive conclusions as to the current level of effectiveness of CBT for improving depression. These limitations must be addressed in future trials before firm conclusions about the efficacy of CBT for depression in CBT can be reached. In particular, trials are needed that have larger sample sizes (with equal numbers of men and women), well-defined depressed data via the use of diagnostic interviews, and that provide intention- to-treat and effect size data.


Ms Milena Gandy is supported by the generous scholarships of the National Health Research Council of Australia and the Molly McDonnell Foundation of the Epilepsy Society of Australia for this research.


None of the authors has any of interest to disclose. 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.