What is already known on this topic

Depression and anxiety in young people are major causes of disability. There is potential for prevention by using effective treatment strategies for the disorders, but previously effects for depression and anxiety have not been aggregated. In this paper, the combined effects of prevention programmes on depression and anxiety are examined.

Methods of the study

MEDLINE, PsycINFO and the Cochrane Library of Systematic Reviews were searched for systematic reviews from 1980 to 2014, and randomised controlled trials (RCTs) from August 2010 to August 2014. No search for unpublished studies was undertaken. For the systematic reviews, these included reviews of RCTs of programmes to prevent the onset of either depressive or anxiety disorders in children aged 5–18 years, published in English, with outcomes that included valid diagnoses or symptom measures of anxiety or depression and data that could be extracted. The later search for RCTs was carried out to ensure studies published after the systematic reviews were included and used the same criteria. Included studies were independently identified by two of the authors.

Analysis included multivariate meta-analysis and random effects meta-regression, using the I² index to test heterogeneity and the Q statistic to test statistical significance. There was no examination of potential publication bias. The primary outcomes were composite measures. An ‘internalising disorder diagnosis’ was calculated from the average of the natural logs of the depression and anxiety risk ratios, their SE and by calculating their covariance before generating the inverse log to produce final estimates. An ‘internalising disorder symptoms’ outcome was created by taking the mean of the depression and anxiety Cohen’s d values and calculating their covariance. This calculation yielded composite measures of diagnosis and symptoms that takes into account comorbidity of anxiety and depression.

What does this paper add

• The use of combined estimates of depression and anxiety prevention programmes to assess effectiveness is new. Previous reviews have reported effects on depression and anxiety separately.

• Meta-regression showed that psychological approaches are better than educational (t₇=−2.51, p=0.04) and interventions delivered by teachers result in larger effect sizes (t₇=−2.64, p=0.04)

• A similar study on depression prevention did not find that the intervention effect was moderated by who delivers the intervention.1 The inclusion of anxiety disorders in this review may have influenced this outcome as anxiety programmes are mostly delivered by teachers and these studies have bigger effect sizes.

• The effects for universal interventions in this study are larger than previously reported.2 ,3 Targeted interventions have been shown to have larger effects than universal, as would be expected, but in previous reviews2 ,3 the type of population (universal or targeted) did not modify the overall effect statistically.

• The finding that the impact of prevention programmes decreases over time (eg, the effect size for internalising symptoms was −0.26 postintervention, −0.23 at 6–9 months and has decayed by 12 months) is consistent with previous findings (eg, depression scores the effect size post is −0.21 and by 12 months is −0.12.)2 ,3

Limitations

• There is limited power in meta-regression to enter all of the potential confounders into the same analysis or test interactions; it is possible that the variation in intervention effects suffers from potential confounding from known or unknown covariates.

• When there are few trials in a subgroup (eg, type of intervention) there is an increased chance that spurious associations may be observed. Consistent findings across a number of trials make it more likely that there is a real effect and that the findings are not just due to chance.

• In this paper there is no consideration of the effect of the type of control group. The placebo response is high in depression and anxiety and in previous reviews placebo or attention controlled trials have shown no significant effect of intervention.2 ,3 Positive findings in studies where there is no attention control, for example, a wait-list control group, must be interpreted more cautiously than those with active control. There are very few studies in depression or anxiety prevention that use an active control condition, making it difficult to rule out a placebo effect.

• There is no differentiation between psychological programmes. Other reviews have found the type of intervention may affect outcome.1 ,3 It will be important to take this into account in future work.

What next in research

• Studies should investigate potential mechanisms of action so that transdiagnostic processes can be targeted to allow a reduction in depression and anxiety (and potentially other symptoms).

• More robust study designs are required that include longer term follow-up as well as using attention placebo control groups. It is important that control groups should be designed to ensure that the effect is not related simply to time and attention, as the placebo response rate is high in anxiety and depression.

Do these results change your practices and why?

No. The effect sizes of the prevention programmes on symptoms are small, the cut-offs for diagnosis may result in false positives, the possible placebo response remains problematic, effects rapidly reduce over a period of 6–9 months and there is a high level of heterogeneity in findings. This study, would not lead us to recommend a change in practice, nor the widespread introduction of anxiety and depression prevention programmes without robustly designed effectiveness trials with positive results.