Hospital screening tool for suicide risk among self-harmers should be ditched

A screening tool used in general hospitals to detect suicide risk among patients who have self harmed should be ditched, concludes a study published online in Emergency Medicine Journal.

The technique (SADPERSONS Scale) fails to pick up most of those who require admission to a psychiatric unit, community psychiatric aftercare, or to determine those at risk of self harming again, say the researchers.

The SADPERSONS Scale was developed in the USA in 1983 as a means of assessing suicide risk among patients who had self harmed. It is based on 10 major risk factors, but has changed little since it was first devised, say the researchers.

National guidance in England stipulates that all patients who go to hospital after an episode of self harm should be given a full psychosocial assessment. But current pressures in hospitals to meet waiting time targets, combined with a reduction in the availability of mental health services, mean that emergency care staff are increasingly minded to use rating scales to find out which patients can be discharged without a full psychosocial assessment, say the researchers.

They therefore assessed SADPERSONS scores that had been recorded for 126 patients consecutively admitted to one emergency medicine department in a major general hospital in Oxford during the summer of 2011, to see how accurate it was at predicting how these patients were subsequently managed and treated.

This included admission to a psychiatric unit, a proxy for psychosocial assessment; the provision of community psychiatric aftercare; and bouts of repeated self harm in the following six months.

Self harm was defined as any act of poisoning or injury, irrespective of its purpose. Most of the patients (102; 81%) had taken an overdose; around one in 10 (11%) had cut themselves; and the remaining 10 patients (8%) had inflicted other forms of injury on themselves.

Admission to a psychiatric unit was required in five cases (4%) and community psychiatric aftercare in just over half (55%; 70). One in four patients (24.6%) self harmed again at least once.

The SADPERSONS Scale only picked up 2% of those requiring admission to a psychiatric unit, around 6% of those needing community psychiatric aftercare, and just over 6.5% of those likely to self harm again.

The authors point out that for the purposes of suicide prevention, any technique designed to spot potential suicide risk must have a low rate of false negative results—in other words, it must be accurate and not miss most of those at risk of killing themselves.

While the small numbers of patients in this study don't allow any conclusions to be drawn about the Scale's usefulness in predicting suicide risk, the scores did not pick up very accurately those most at risk of further self harm, which is particularly associated with suicide risk, say the study authors.

Twenty three out of 31 of the episodes of self harm occurred within the first three months of the first visit to emergency care. But only two of these patients had high scores on the SADPERSONS Scale; the rest had low to moderate scores, suggesting they were not at high risk.

"The results clearly show that the SADPERSONS Scale has a very limited role, if any, to play in the assessment of patients presenting to the emergency department following an episode of self harm," write the authors.

"Indeed, arguably, our results show that reliance on the scale for determining who should receive a psychosocial assessment or otherwise using it for prediction is not only misleading, it could be dangerous," they add.

The use of rating scales has become increasingly widespread in response to the need to standardise practice for ever increasing numbers of patients. But these tools often overlook individual dynamics, they say.

"A greater focus on clinical judgement is needed, accompanied by the necessary education, training and supervision, if we are to more accurately fully identify and intervene with those who are at greatest risk following self harm," they conclude.