Assessing the risk of self-harm in an adult prison population

Paper 57: Introduction and Aims: In the UK, over two-thirds of the prison population have a history of drug or alcohol dependence and many have co-morbid mental health problems. In response to a perceived need for screening instruments to identify the risk of self-harm, we undertook a prospective cohort study in three UK prisons to identify potential screening instruments, or their items, and determine their predictive validity. Design and Methods: Four hundred and fifty prisoners under "suicide watch'' were administered a set of standardised questionnaires, including the PriSnQuest, Borderline Symptom List-23, SelfHarm Inventory, PHQ-9, and the Clinical Outcomes in Routine Evaluation - Outcome Measure, and followed up for a maximum of six months. Area under the curve analysis examined the ability of the instruments, or their items, to predict future self-harm, and Cox Proportional Hazards models examined the multivariate predictive ability of the scales and various socio-demographic and sentencing factors with regard to time-to-self-harm. Results: During the follow up period 28% of participants selfharmed. The screening instruments failed to show predictive value when applied within area under the curve or Cox models. However, good predictive values were shown for gender-specific sets of questions derived from the screening tools. Such questions related to history of alcohol dependence, previous history of self-harming behaviours and recent threats of violence towards others. Discussion and Conclusions: Aggregating such questions into easily applied screening questionnaires could provide guidance for future care pathways. Implications for Practice, Policy and Translational Research: Existing screening tools are poor at predicting future self-harm but individual questions within the tools have good predictive validity. Different screening questions for future risk of self-harm should be used for male and female prisoners.