This review examined the effectiveness of internet-based cognitive behavioural therapy (CBT) for health problems and concluded that the Internet will increasingly be studied and used to deliver CBT to patients with health problems. Despite data limitations including clinical and methodological heterogeneity, small samples sizes and small number of studies for each health problem, the authors’ cautious conclusions appear reasonable.

To examine the effectiveness of internet-based cognitive behavioural therapy (CBT) for health problems.

PubMed, PsycINFO, EMBASE and Cochrane Central Register of Controlled Trials were searched up to February 2007 for published articles; Digital Dissertations was searched for unpublished data. Search terms were reported. In addition, reference lists of retrieved articles and relevant reviews were also searched.

Randomised controlled or comparative trials of CBT techniques that delivered at least 50 per cent of the intervention via the internet were eligible for inclusion if they were aimed at changing behaviour in patients with existing disorders or health problems. Studies of patients with mental disorders and studies that focused on preventative interventions for lifestyle factors such as smoking and obesity were not eligible.

Included studies were conducted in Sweden or the USA. Most patients were women. Children were also recruited from the community. Mean age ranged between 12 and 57 years. Patients presented with pain, headache or other health problem (tinnitus, physical disability, chronic disease, breast cancer, insomnia and paediatric brain injury). Interventions were either self help (with or without support) or individual/group online therapy, and compared CBT (including psychoeducation, exercise/relaxation or problem solving) with a control (mostly waiting-list). Interventions were delivered over four to 20 weeks.

The authors did not state how many reviewers selected studies for relevance.

Validity was assessed using previously published criteria, including items on: allocation concealment, blinding of assessors, completeness of follow-up and intention-to-treat analysis. The authors did not state how many reviewers performed the validity assessment.

Average pre- and post-test scores were extracted for intervention and control groups to calculate effect sizes (d). An effect size of 0.8 was considered large, 0.5 was considered moderate and 0.2 was small. The authors did not state how many reviewers performed the data extraction.

A random-effects model was used to pool mean effect sizes for pain outcomes, with 95% confidence intervals (CIs). Where pooling was not possible, effect sizes were reported as a narrative synthesis and in tables. The Q-statistic and I² tests were used to assess for heterogeneity.

Twelve studies (13 papers, n=1,704 patients; 841 receiving intervention and 863 controls) were included in the review. All studies appeared to be randomised controlled trials. Sample sizes ranged from 22 to 958 patients. One study accounted for 56 per cent of the participants. Six studies reported a mean follow-up of 7.67 months. Allocation was conducted by an independent party in three studies. Concealment of random allocation to patients and assessor blinding was not possible or not reported in any studies. Drop-out rates ranged from two per cent to 61 per cent. Only one study reported using intention-to-treat analyses.

There was a moderate to large mean effect for internet-based CBT compared to control for pain (three studies), 0.58 (95% CI: 0.25, 0.92, p<0.001). There was no evidence of statistical heterogeneity. Effect sizes were small to moderate in patients with headache, ranging between 0.19 and 0.56 (three studies) and varied in patients with other health problems, ranging between 0.10 and 0.70.

The Internet will increasingly be studied and used to deliver CBT to patients with health problems.

The review question was clear and was supported by adequate criteria for participants and interventions. Criteria for study design were broad and no inclusion criteria were defined for outcomes. A comprehensive literature search was undertaken to identify published and unpublished articles, thus reducing the risk of publication bias. However, it was unclear whether attempts were made to minimised language bias. Validity was assessed, but quality was not optimal and some studies reported high drop-out rates. The review process for validity assessment, data extraction and study selection was not reported, thus reviewer error and bias cannot be ruled out. Appropriate methods were used to investigate statistical heterogeneity. Appropriate methods for synthesis were used (when pooling of effect sizes was not appropriate). There was clinical and methodological heterogeneity. Several limitations with the included studies were acknowledged by the authors, including small samples sizes for most studies and small number of studies for each health problem. Given the limitations, the authors’ cautious conclusions appear reasonable.

Practice: The authors stated that the results may not be generalisable to the general population in the clinical setting.

Research: The authors stated that future research should focus on care as usual or other control groups, face to face treatment comparisons, clinical recruitment strategies and target populations including children and older adults. Further research was warranted to assess how CBT interventions via the internet should be presented (in particular to reduce drop-out rates) and assess how internet-administered CBT could be incorporated into stepped-care models of care.

Not stated.

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.