Screening for the risk of self-harm in an adult offender population

The reported incidence of self-harm is much higher among prisoners than in the general population. Along with the general increased risk among prisoners, there are also specific times of heightened vulnerability, such as on first reception into the prison environment. Currently in the UK, if an act of self-harm is carried out by a prisoner, if intent to do so is expressed, or if a prisoner is deemed vulnerable by a member of staff or a fellow prisoner, then this prisoner will be subject to a self-harm and suicide monitoring process known as an ACCT (Assessment, Care in Custody and Teamwork). As part of the ACCT process, a member of prison staff will carry out a prisoner assessment to ascertain the risk level of the prisoner, and whether the ACCT needs to remain open. However, there is currently no standardised risk assessment or clinical decision aid used as part of the ACCT process, which is perhaps largely owing to the paucity of validated risk assessment tools or clinical decision aids that are available to identify risk of self-harm in prisoner populations. The primary aim of the study was therefore to determine whether any pre-existing, standardised instrument would be suitable to use for the purpose of assessing the risk of self-harm among the specific ACCT population. This thesis describes various different elements of the study. Firstly, a scoping exercise was carried out in order to select the most appropriate pre-existing instruments that had the potential to predict self-harm events. A pilot study was then carried out to test the study process and to refine the instrument selection. Following this, a prospective cohort study was undertaken to assess the predictive validity of the selected instruments, where a six month follow-up determined self-harm occurrence since baseline, and area-under-the-curve (AUC) analysis examined the ability of the instruments to predict future self-harm. Alongside this, a thorough psychometric analysis of each of the instruments was carried out, in order to validate them among the specified ACCT population. Utilising the study dataset, an exploratory logistic regression analysis was carried out in order to identify individual background and instrument items that may prove effective in predicting future self-harm. Finally, structural equation modelling (SEM) was used to explore the relationships between some of the factors that influence self-harm, in order to contribute towards an understanding of the complexity of the issue of self-harm in prisons. The key findings from this programme of work can be summarised as follows: i. Self-harm is a common occurrence among the prison ACCT population, with 29.1% of the study participants deliberately self-harming during the follow up period, although this varied considerably across gender and participating prisons. ii. Four of the five selected instruments did display a certain level of psychometric validity among the study population; therefore validating the cut-points for the predictive analysis. However, all instruments required some refinement to meet the strict measurement criteria of the Rasch model. iii. Of the five pre-existing instruments that were selected for the study, none of these displayed a meaningful predictive validity. iv. Logistic regression analysis did reveal gender-specific item sets, producing predictive algorithms which were statistically significant in predicting future self-harm; however, the operational functionality of these item sets may be limited. v. Structural equation modelling revealed an insightful explanatory model of the process that may be involved in the culmination of self-harm in prison. Path analysis models supported the view that self-harm capacity and self-harm propensity are integral elements to the self-harm pathway, although the complete explanatory model is likely to be more complex.