The reviewers found eHealth interventions that incorporated behavioural interventions produced larger beneficial effects than solely educational eHealth interventions. Methodological flaws and a lack of information on study quality and the results of individual trials mean that the authors conclusions should be interpreted with caution and the results of the review may not be reliable.

To evaluate the impact of eHealth interventions on paediatric health-promoting and maintaining behaviours that are believed to impact the development of a physical disease and associated outcomes.

PsycINFO, PubMed and ERIC databases were searched up to and including 2009 for relevant studies in English; search terms were reported. Reference sections of review articles and book chapters were checked to identify additional references.

Published studies that included quantitative methods to evaluate the impact of an eHealth intervention on health behaviours known to moderate disease outcome in children or adolescents under 18 years of age were eligible for inclusion. eHealth interventions were defined in the review as applications of technology that sought to improve a patient's understanding of health information or use technology as a surrogate for a clinical in the delivery of treatment by delivering the active ingredient for treatment or as a substitute for face-to-face meetings with a clinician. The outcomes of interest were behavioural process variables that may lead to health improvements and health outcomes that were the target of behavioural process interventions. Case reports and studies of mixed adult and paediatric patients were excluded from the review. Telehealth interventions that used technology as a method of communicating between patients and clinicians were excluded.

Intervention types included in the review were internet, CD-ROM, computer games, virtual reality, cell phone and combined phone and internet. The number and duration of sessions varied between studies. Interventions were used for weight control and nutritional health, symptom control for asthma, smoking cessation, healthy sun behaviour and diabetes care. Where stated, the comparators were face-to-face sessions and telephone conversations with health-care professionals, group therapy and day camps.

The authors did not state how many reviewers performed the study selection.

The authors did not state that they assessed methodological quality.

Two reviewers extracted data to calculate Cohen's d-statistic to represent effect sizes from pre-to post intervention and 95% confidence intervals (CI). Interventions were classified as either educational (primarily focused on communicating some form of knowledge about health behaviour to children or adolescents, such as knowledge about disease control or replacement behaviours) or behavioural (where technology was used to provide a behavioural intervention, such as self-monitoring, or goal setting). A coding system was used to demonstrate agreement. Any disagreements were resolved by discussion until 100% consensus was reached for all studies.

The results of the studies were weighted by sample size and pooled effect sizes and 95% CIs were calculated using the Hedges and Olkin weighted least squares approach in a random-effects model. Effect sizes of 0.20 to 0.49 were rated as small, effects of 0.50 to 0.79 were regarded as medium effect sizes and those above 0.80 were large effect sizes. The Q-statistic was used the evaluate statistical heterogeneity. The authors calculated fail safe n sizes to investigate the impact of unpublished studies and studies that had not passed the peer review process on the results.

Thirty-three studies (7,690 participants) were included in the review: 29 randomised controlled trials (RCTs), two non-randomised between-group studies and two non-randomised single group pre- and post-test studies. Sample sizes across the studies ranged from 10 to 1,578 patients. Post-treatment assessments were conducted from two weeks to 52 weeks post intervention.

Small but statistically significant benefits of the eHealth interventions were observed across all trials (d=0.118, 95% CI 0.066 to 0.171; 33 studies) with significant heterogeneity across the results (Q=62.851, p<0.001).

Effect sizes for the interventions that incorporated behavioural components were also small and statistically significant (d=0.354, 95% CI 0.232 to 0.475; 20 studies). There were no significant differences between educational eHealth interventions and control treatments (d=0.033, 95% CI 0.037 to 0.103; 13 studies). The reviewers stated that the Q statistics within each of the behavioural and educational interventions did not suggest statistically significant heterogeneity.

The fail safe n size calculation was that 230 studies with null findings were required to negate the observed findings.

Larger effect sizes were produced for eHealth interventions that incorporated behavioural methods such as self-monitoring, goal-setting, immediate feedback and contingency management than interventions based solely on education.

The review addressed a question that was broad in scope. Some criteria for study inclusion were outlined. Appropriate electronic databases were searched for relevant studies. The restriction to published studies in English risked publication and language biases. Steps to avoid errors and bias in the review process were reported for data extraction, but not for study selection. There was no assessment of methodological quality.

Statistical pooling of the results of many different intervention types, patients and study designs may not have been appropriate. In particular, the results of non-randomised studies were associated with potential biases and may result in an overestimation of effect.

Methodological flaws and a lack of information on study quality and the results of individual trials mean that the authors conclusions should be interpreted with caution and the results of the review may not be reliable.

Practice: The authors stated that interventionists should carefully consider how eHealth interventions should incorporate behavioural strategies such as goal setting, self-monitoring and provision of immediate feedback. Interventions in eHealth that used only education should not be used as the sole mechanism of action in behaviour change programmes.

Research: The authors stated that component analyses of future eHealth behavioural interventions may help to identify causal mechanisms within different modalities and potentially streamline interventions. Mobile devices had untapped potential and their ubiquity among adolescents meant that there may be value in further investigations of similar interventions. The aggregate impact of electronic interventions on the cognitive processes in pain ratings or coping should also be considered in future work.

None 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.