Authors' reply

We welcome the opportunity to reply to House & Owens' two observations. We will address their concerns in the order in which they present them: the title and the limitations associated with the methodology, respectively.

First, the title does not say that suicidal behaviour was changed. The title states that the aim of the intervention was ‘to reduce subsequent suicidal ideation and behaviour’. Moreover, we insisted on the term ‘exploratory’ remaining in the title against the suggestion of a reviewer and explicitly did not present the 6-month outcomes as main findings.

Second, House & Owens are correct that there are methodological limitations to the study, but these are recognised in the paper. Indeed, the final sentence of our conclusion chimes with the points they raise: ‘Further research is required to replicate the findings with a more complete data-set and objective outcome measures’. In total, we highlighted six points of caution with respect to the interpretation of the findings. Methodological limitations are inherent in exploratory studies of this kind, but we endeavoured to address them by making weaknesses explicit and adjusting analyses to mitigate the effects of limitations as far as possible.

We believe our approach of last observation carried forward is a good example of adjustment to the limitation of missing data here. As House & Owens note, the observation carried forward method is not perfect and may introduce bias, which we assumed most readers would be well aware of and therefore did not flag in our ‘limitations’ section. In terms of the present study, the observation carried forward was pre-intervention, rather than post-intervention. Given that our randomisation check demonstrated no differences between conditions pre-intervention, any score carried forward would, if anything, bias results against finding an effect of the intervention. Had we chosen instead to measure outcomes immediately post-intervention and carried these observations forward, then the findings would have been biased in favour of finding an effect of the intervention. We would therefore argue that, cognisant of the limitations of last observation carried forward per se, we adopted the most rigorous approach possible within the given context (a busy emergency department in a low- and middle-income country).

Another example of our attempt to make the limitations of our study explicit concerns the points we make about the measurement of the main outcome variable. It was not possible to record re-attendance following non-fatal self-harm, and so we had to rely on the use of the Suicidal Behaviors Questionnaire – a self-report measure of suicidal ideation and behaviour. In the discussion, we state that ‘it would be useful to have a more objective outcome measure, such as future hospital admissions, although this is not currently possible in the Malaysian context’. It behoves us to conduct research such as ours in low- and middle-income countries, even if it can only be exploratory at the present time.

In sum, we stand by our conclusion that this paper details a promising approach to intervention that is worthy of further research.