Repeated self-harm is difficult to predict particularly in the long-term. Scales that help identify risk of repetition are widely used, and the most popular is the modified SAD PERSONS scale (mSPS). We aimed to clarify the accuracy of the mSPS in first time and non-first time patients presenting with self-harm with an extended follow-up period of observation.

We conducted the UK's first long-term prospective risk prediction study of patients to an Accident and Emergency (A&E) department. We followed up 774 patients for a mean period of 7.4 years. At the time of study 429 patients presented with their first episode of self-harm and 345 had a hospital record of past history of self-harm.

During 7.4 years of follow-up 54.7% of those who initially presented with self-harm represented with a repeat self-harm episode. Four hundred and twenty-one patients had a mSPS score less than 6 (low risk), 260 had a score between 6 and 8 (moderate risk) and 94 scored 9 or higher usually notated as high risk. The clinical utility of the mSPS was “poor” for predicting future self-harm but at cut-offs ≥7 it was “fair” for predicting non-repetition. Of the individual questions in the mSPS, a positive answer to “stated future wishes” was a true positive in 74.0%.

We found the mSPS was a relatively poor indicator of future self-harm risk in a long term follow-up of patients who self-harmed regardless of hospital record of previous self-harm history. However, mSPS did have modest value in predicting non-repetition.

The authors have not supplied their declaration of competing interest.