The Letter by Large & Ryan (Reference Large and Ryan2012) correctly points out that the positive predictive value of 30% is for both non-fatal and fatal repeat self-harm acts, which is the outcome we based our performance measures on. We hope that this will not be misinterpreted as the measure for suicide alone. We acknowledge that suicide is a rare outcome and that aiming to predict only those repeat acts that end fatally would not be feasible. Throughout the development of ReACT we were mindful that there needed to be a balance between correctly identifying the relatively small number of suicides and the large proportion that re-attended with non-fatal self-harm. We acknowledged that ‘no risk assessment measure can be accurate enough to assume a patient assessed as low risk will not repeat self-harm or complete suicide’ (Steeg et al. Reference Steeg, Kapur, Webb, Applegate, Stewart, Hawton, Bergen, Waters and Cooper2012) and the risk of suicide is markedly elevated for anyone within the self-harming population relative to the general population (Cooper et al. Reference Cooper, Kapur, Webb, Lawlor, Guthrie, Mackway-Jones and Appleby2005).
However, we proposed that a tool drawing together risk factors from a large prospective cohort, based on population-level data and real outcomes provides some evidence to inform risk categorization according to a patient's likelihood of further self-harm. We support the NICE guidance (NICE, 2004) that this is only a part of a wider assessment of a patient's psychological and social needs. With any clinical decision tool it is important to be clear about the proposed utility as well as its statistical validity and performance. With tools designed for use in mental health settings, such as the ReACT tool, discussion around the proposed clinical use becomes more important than with those designed for physical conditions. As Large & Ryan highlight, specific diagnostic tests are often carried out following the result of screening tools. While the course of action is not so clear-cut when treating self-harm patients presenting in an emergency situation, the use of screening tools as part of mental health risk assessment has been recommended (DoH, 2007). We therefore welcome the opportunity to expand on this further.
Screening is the beginning of a process and we were not suggesting a four-question tool can be used alone to determine a patient's outcome in terms of non-fatal repetition or suicide. The four factors identified may act as a ‘red flag’ to emergency-department (ED) staff treating the patient in the early stages of the presentation. We agree that specific interventions would not be determined solely on the basis of the ReACT tool. Any intervention would be based on a comprehensive assessment of psychological and social needs. Risk is an important area to consider as part of the wider assessment, particularly when considering immediate management, but would not inform decisions on interventions alone. However, mental health clinicians may use this awareness in consideration of the potential benefits of certain treatments and as part of a dialogue with the patient about treatment options.
The provision of mental health care for patients presenting to EDs with self-harm is not consistent. For example, the proportion of ED presentations assessed by mental health specialists varied considerably by hospital in a national study in England (Bennewith et al. Reference Bennewith, Gunnell, Peters, Hawton and House2004). NICE (2004) recommend that assessment in the ED should be performed by ‘appropriately trained staff’. We agree a full psychosocial assessment by a mental health specialist before discharge would be the ideal.
Ultimately, the tool may be useful in forming the risk profile of a patient, but is just one component of overall management.
