This review concluded that there was no evidence that additional psychosocial interventions following self-harm have a marked effect on the likelihood of subsequent suicide. The conclusions appear to reflect the review findings but limitations, such as the lack of a quality assessment of the included studies, make it difficult to confirm the reliability of the results.

To determine whether additional psychosocial interventions following an episode of self-harm reduce the likelihood of subsequent suicide.

MEDLINE (from 1966), EMBASE (from 1969) and PsycINFO (from 1967) were searched up to February 2005. In addition, meta-analyses and systematic reviews and the reference lists of relevant papers were screened. Authors were contacted to obtain data on suicide deaths where this was not reported in the original papers.

Study designs of evaluations included in the review

Randomised controlled trials (RCTs) were eligible for inclusion. The duration of follow-up in the included studies ranged from 3 to 36 months.

Specific interventions included in the review

Studies comparing additional or enhanced psychosocial interventions with a control intervention or standard care were eligible for inclusion. The included studies investigated individual psychotherapy such as cognitive-behavioural therapy, interpersonal psychotherapy and dialectical behaviour therapy of varying intensity (between 2 and 50 sessions) and at various settings, or the change in organisation of services to enhance the uptake of psychosocial treatments or to facilitate contact with services at times of crisis.

Participants included in the review

Studies with patients who had harmed themselves prior to participating in the study were eligible. Most, but not all, of the included studies were in adults. Where reported, the suicide attempt took the form of self-poisoning or self-harm and patients mainly presented to the accident and emergency department. Several studies had further specified inclusion criteria, such as more than two suicide attempts or suicide attempters who had missed the follow-up appointment.

Outcomes assessed in the review

The studies had to report death from suicide to be eligible. The review reported the impact of the intervention on suicide and on all-cause mortality.

How were decisions on the relevance of primary studies made?

Two reviewers independently assessed the papers for inclusion. Any disagreements were resolved by discussion with a third reviewer.

The authors did not state that they assessed validity.

The data were extracted in duplicate. The reviewers calculated the rate of suicide in each arm of the study to model the numbers of deaths in the treatment and control groups and determined the proportion of participants in both arms who died by suicide. The difference in death rates within each study was compared with critical values of a standard normal distribution and effect sizes were calculated for each study.

How were the studies combined?

The effect sizes (difference in square roots of the rates) were combined in a fixed-effect meta-analysis and presented together with the 95% confidence interval (CI). Funnel plot asymmetry was examined to determine the potential for publication bias.

How were differences between studies investigated?

Statistical heterogeneity was assessed using a chi-squared test.

Eighteen RCTs (n=3,918) were included in the review.

The funnel plot did not indicate publication bias.

There was no statistically significant difference in the rates of suicide in the intervention and control groups when pooling all 18 RCTs (difference in root rates 0.00, 95% CI: -0.03, 0.03, p=0.86). There were 37 suicides in total: 18 per 2,165 person-years amongst those who received enhanced treatment and 19 per 2,237.5 person-years amongst those who received control treatment. None of the included studies showed a statistically significant difference between treatment arms in root rates. No statistically significant heterogeneity was found (p=0.99).

There was no statistically significant difference in mortality in the intervention and control groups when pooling all 12 RCTs that provided complete mortality data (difference in root rates -0.02, 95% CI: -0.06, 0.02, p=0.26). There were 29 deaths in total: 11 per 1,527.5 person-years amongst those receiving additional psychosocial interventions and 18 per 1,422 person-years amongst those receiving control treatments. No statistically significant heterogeneity was found (p=1.00).

The meta-analysis did not provide evidence that additional psychosocial interventions following self-harm have a marked effect on the likelihood of subsequent suicide.

The review stated a clear question and clear inclusion criteria. The searches encompassed more than one electronic database and screening of references, but other than that no explicit attempt to locate unpublished data was reported; the risk of publication bias was, however, assessed. Measures were taken throughout the review process to reduce reviewer errors and bias. The quality of the included studies was not assessed, which made it more difficult to comment on the validity of the pooled results.

The interventions varied greatly in their approach, setting and intensity and the pooling of these diverse studies is controversial. The choice of effect measure was justified but may make the effects difficult to compare with other pertinent publications. In their discussion, the authors correctly pointed out that no evidence of a treatment difference did not equate to evidence of no difference. Overall, the conclusions appeared to reflect the evidence presented but limitations, such as the lack of a quality assessment of the included studies, make it difficult to confirm the reliability of the results.

Practice: The authors stated that the view that a range of public health measures for preventing suicide should be pursued, including population-based strategies such as restricting access to means of suicide and encouraging responsible discussion of suicide in the media, was supported.

Research: The authors did not state any implications for further research.

Imperial College London.

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.