|Efficacy of exposure versus cognitive therapy in anxiety disorders: systematic review and meta-analysis
This review found no evidence that cognitive and exposure therapy differed in their efficacy for treating panic disorder, post-traumatic stress disorder and obsessive compulsive disorder, but there was strong evidence that cognitive therapy was more effective for social phobia. There was potential for bias in the methods, and the author's findings on social phobia require cautious interpretation.
To compare the efficacy of cognitive versus exposure therapy for anxiety disorders.
PsycINFO, MEDLINE and EMBASE were searched for articles from inception to May 2010, without language restriction. Search terms were reported. The reference lists of articles retrieved and relevant clinical guidelines were checked, and experts were approached to enquire about unpublished studies.
Eligible studies were randomised controlled trials (RCTs) of exposure (with or without relaxation training) versus cognitive therapy (with or without behavioural interventions) for treating anxiety disorders. Trials with pharmacological interventions were eligible provided the co-intervention was the same for both groups. The outcome of interest was efficacy, preferably reported in each trial as a pre-specified primary outcome. Continuous outcomes had to be reported using a standardised (preferably published) rating scale. The reviewers excluded trials in which one arm received both interventions, unless they reported a comparison between cognitive and exposure therapies alone. Trials with more than 30% attrition, with clearly non-blinded outcome assessment, that did not assess "pure" therapies, or with therapist rating were excluded. Cognitive therapy and exposure therapy were defined in detail in the review.
The included trials were set in Europe or North America. Most participants were Caucasian adults, with a mean or median per trial age of 28 to 38 years, where stated. The percentage of male participants ranged from zero to 65. Participants had obsessive compulsive disorder, post-traumatic stress disorder, panic disorder with or without agoraphobia, social phobia, hypochondriasis, or generalised anxiety disorder. A few patients were allocated to a comparison treatment or control condition, such as waiting list, rather than one of the two interventions. In most trials, the intervention consisted of eight to 16 sessions, with equal intervention time for both groups, delivered in either an individual or a group setting. Trial outcome measures differed according to the condition studied, and for some conditions two or more measures were used. Outcomes were reported at short-term and long-term follow-up.
Full-text articles were selected by two reviewers, who settled disagreements by consensus.
Assessment of study quality
Allocation concealment, attrition, a priori outcome specification, adequate reporting of relevant efficacy outcomes, adequate blinding of outcome assessment, and self-reporting of outcomes were assessed. The author did not state how many reviewers conducted the assessment.
Both raw and transformed data were extracted. Odds ratios were calculated for dichotomous data and standardised mean differences were calculated, using Hedges' g, for continuous data, both with 95% confidence intervals. If no primary outcome measure was specified, the reported outcome that most closely assessed clinical efficacy was selected by the reviewers. Trial authors were contacted for additional data, if necessary.
The author did not state how many reviewers extracted the data.
Methods of synthesis
The data were synthesised using a fixed-effect model to calculate pooled odds ratios or standardised mean differences, along with 95% confidence intervals. Ι² was used to assess heterogeneity. If it was over 50% and heterogeneity could not be explained, a random-effects model was used. Funnel plots and the trim and fill method were used to investigate publication bias in comparisons for each condition.
Results of the review
Twenty-two RCTs were included in the review, of which 20 with over 1,300 participants (range 16 to 162) were pooled. Allocation was reported as blinded in two trials, and assessor blinding was adequate (or a self-report tool was used) in 11 trials. Most trials did not pre-specify the primary outcome, eight only included completers in analyses, and attrition ranged from 3% to 29%. The long-term follow-up was at 26 weeks or more in most trials (range none to 104).
There was no statistically significant difference in efficacy between cognitive and exposure therapy for: obsessive compulsive disorder in the short term (five RCTs; 290 participants; Ι²=49%) or in the long term (four RCTs; 181 participants; Ι²=51%); post-traumatic stress disorder in the short term (five RCTs; 287 participants; Ι²=31%) or in the long term (five RCTs; 226 participants; Ι²=23%); and panic disorder with or without agoraphobia in the short term (seven RCTs; 274 participants; Ι²=62%) or in the long term (six RCTs; 247 participants; Ι²=69%; random-effects model).
There was a statistically significant difference in efficacy favouring cognitive over exposure therapy for social phobia, both in the short term (SMD 0.52, 95% CI 0.37 to 0.74; three RCTs; 128 participants; Ι²=45%) and in the long term (SMD 0.46, 95% CI 0.29 to 0.73; two RCTs; 75 participants; Ι²=0).
There was no indication of publication bias, but the number of trials in each comparison was small. The findings of one trial of hypochondriasis and one trial of generalised anxiety disorder were not reported.
There was no evidence that cognitive and exposure therapy differed in their efficacy for treating panic disorder, post-traumatic stress disorder, and obsessive compulsive disorder; but there was strong evidence that cognitive therapy was more effective than exposure therapy for treating social phobia.
The objectives and inclusion criteria were clear, except for participants, which diminishes the reproducibility of the review. There was potential for subjectivity in determining which trials had "pure" interventions. Relevant sources were searched, without restriction by language. No publication bias was identified, but (as the author noted) there were too few trials to assess this risk reliably. Inclusion was restricted to higher quality trials. It was unclear whether sufficient efforts were made to minimise the risk of reviewer bias and error in the processes of data extraction and quality assessment.
Appropriate methods were used to combine the trial data and assess statistical heterogeneity, but there were inconsistencies between the figures and the text for heterogeneity. Where high heterogeneity was found it was not investigated further. As the author noted, the findings were open to bias due to the low quality of most trials, the small number of trials with small samples, the potential overlap between the interventions, and potential researcher allegiance. The author commented that the generalisability of the findings might be limited by the setting and delivery of the interventions.
The author's findings on social phobia require cautious interpretation, as they were based on only two or three small trials.
Implications of the review for practice and research
Practice: The author suggested that clinicians should note the better efficacy of cognitive therapy over exposure for social phobia and that it should be incorporated in the relevant clinical guidelines. Policy makers could use this evidence to consider the relative need for training in cognitive therapy or exposure therapy.
Research: The author stated that it would be useful to explore the additional value of combining exposure and cognitive therapy for treating different anxiety conditions and to determine whether exposure therapy had any useful role for obsessive compulsive disorder.
Supported by the Psychiatry Research Trust.
Ougrin D. Efficacy of exposure versus cognitive therapy in anxiety disorders: systematic review and meta-analysis. BMC Psychiatry 2011; 11:200
Subject indexing assigned by NLM
Anxiety Disorders /therapy; Behavior Therapy /methods; Comparative Effectiveness Research; Humans; Implosive Therapy; Odds Ratio; Randomized Controlled Trials as Topic
Date bibliographic record published
Date abstract record published
This is a critical abstract of a systematic review that meets the criteria for inclusion on DARE. Each critical abstract contains a brief summary of the review methods, results and conclusions followed by a detailed critical assessment on the reliability of the review and the conclusions drawn.