Response to Andrew Carroll

Liezl Koen, Consultant Psychiatrist, University of Stellenbosch and Stikland Hospital, Cape Town, South Africa:

We are in total agreement with Andrew Carroll that the broadest possible definition for violence should be used when investigating violence with a view to studying risk markers. In the methodology section where we refer to violence ‘preceding’ admission we are in fact referring to any violent behaviour (meeting the stated criteria) that occurred during the lifetime history of the patient (as related by collateral, notes or the structured interview) and not just that in the immediate period prior to admission.