Authors' reply

Kripalani et al raise an important issue by suggesting that those who repeatedly self-harm may constitute a distinctive clinical subgroup. We initially avoided including this group in our study because there remains considerable uncertainty about an appropriate definition. However, following discussion with Dr Kripalani we agreed a three-way classification of repeated self-harm among young people: repeated self-harm (19 out of 89), with self-harm both in the past and currently or using several (three or more) methods (since it is unlikely that multiple methods of self-injury refer to a single incident); a single incident (17 out of 89), with an explicit statement of a transient incident; unsure (53 out of 89), which constituted the remainder. The crude repetition rate of 20% is typical for self-harm (Bennewith et al, 2002). We proceeded to re-analyse the data from our original paper for repeat self-harm (results available on request).

In summary, we can confirm that repeated self-harm was unrelated to gender, or social class of origin, but was related to current labour market position, with youth outside the labour market more likely to self-harm repeatedly. Young people who repeatedly self-harmed were more likely to use all methods except taking pills and more violent methods, which were common to all groups. Those who repeatedly self-harmed were far more likely to do so to relieve negative emotions (anger, anxiety or to punish themselves), but self-harm with intention of killing oneself was common to all groups. Taken together this confirms that those who repeatedly self-harm are more likely to use self-injury as a coping mechanism. With regard to service use, those young people were nearly twice as likely to have used emergency services and over three times as likely to have used psychological services from the age of 11.

This suggests that young people and adults who repeatedly self-harm are heavy users of both health services in general and psychiatric health services in particular, and this is compatible with the assertion of Kripalani et al that a small proportion of clients may account for a large proportion of resources. Distinguishing between repeated and other forms of self-harm could provide useful clinical information, provided that both researchers and clinicians can agree on a clear definition.