RISA -IPD ReducIng Self-harm in Adolescents: Individual Patient Data meta-analysis


Self-harm is common in teenagers. In surveys, 10% report self-harm in the past year. Teenagers who self-harm are at risk of repeating self-harm.

Suicide is the second commonest cause of death in 10 to 24-year olds, after road traffic accidents. Teenagers who self-harm do so for a variety of personal, family and social reasons. Those who have self-harmed are likely to experience more emotional difficulties and difficulties relating to people later in life. Their families and those close to them report experiencing considerable distress.

If we had effective treatments to reduce the likelihood of repeat self-harm we could save lives, reduce distress and improve life chances. Unfortunately, despite a number of different research projects, we still have no clear evidence of an effective treatment intervention that will reduce the likelihood of further self-harm if someone has already self-harmed. Much of this existing research has involved relatively small samples.

In addition, it has included groups of young people who have self-harmed in different ways and for different reasons - but despite these differences everyone in a specific trial will have been offered the same treatment. It is possible that the treatments tried so far have not been shown to be effective because those treatments are helpful for some young people who have self-harmed and not others and these benefits have been hidden in the overall result.

To address this possibility, we will gather information from research that has already taken place, combine this information and conduct further analysis. We will focus on a type of research known as a Randomised Controlled Trial. This is a research study that randomly puts participants into two or more different groups.

The participants in each group are given a different treatment and the results of the treatments are analysed to see which is more effective. By combining information from these types of study we will have data on a large group of young people who have received a range of different interventions.

Dividing this large group into smaller groups (for example, boys vs girls, those with depression vs those without, those using different methods of self-harm, those receiving individual treatments vs other types of treatment etc.) may help identify 'sub-groups' of young people who have self-harmed that do better (or worse) than others on particular treatments.

This might enable us to make recommendations, 1) to clinicians to guide the sort of treatment to offer to particular groups, and 2) to researchers about targeted treatments that could be further evaluated for specific sub-groups. Our research team is well placed to conduct this research.

We have conducted many of the randomised, controlled trials of interventions in self-harm in the UK and therefore already have access to and an understanding of the data. We have also published research on the treatment of self-harm in adolescents that involves finding all the relevant studies by searching databases of all published studies and then combining their results.

Between us we have clinical child mental health expertise and statistical expertise in analysing large and complex data sets. We will also be working with an expert Patient and Public Involvement group of young people, set up specifically to support this study, to ensure that we take into account the views of service users in designing, conducting and sharing the results of the research.


If the meta-analysis indicates clearly that certain sub-groups of young people do better (or worse) with certain types of intervention, we would expect significant changes in the way that services are delivered to those groups of young people.

If the evidence is less clear-cut it is possible that avenues of future research are suggested using more tailored interventions for sub-groups of young people, leading to new and better targeted randomised controlled trials to confirm (or refute) the hypotheses raised by our results.