Self-harm before and during imprisonment: cohort study of males in prison linking population-based routinely collected data in Wales.

BACKGROUND: Self-harm among UK prisoners has risen over the past decade. AIMS: To explore self-harm risk factors and mental health conditions in prisoners, pre- and during imprisonment, compared with the general population. METHOD: This retrospective cohort study linked electronic health records and Ministry of Justice data for Welsh male prisoners (2019), and a comparison general population cohort. We examined imprisonment likelihood based on prior self-harm and mental health conditions using logistic regression. We also studied self-harm risk up to three years during imprisonment through Generalised Estimating Equations and time-stratified Cox regression, using a pre-imprisonment comparator (3 years before). RESULTS: Prisoners (N = 6095) had higher rates of self-harm and mental health conditions pre-imprisonment compared with non-prisoners (e.g. self-harm odds ratio: 2.1 (1.9, 2.2)). Self-harm risk was 5.25-6.47 times higher in prisoners than non-prisoners, both pre- and during imprisonment. Risk was highest shortly after incarceration, then declined, becoming lower than pre-imprisonment after 7 months. While most conditions correlated with higher self-harm risk during imprisonment (e.g. drug use, hazard ratios: 1.5-3.0), some (e.g. depression and alcohol use) showed weaker links in prisoners than non-prisoners, particularly from 7 months after imprisonment. Self-harm risk was seemingly higher in prisoners on remand compared with those sentenced. CONCLUSIONS: Pre-imprisonment, self-harm in male prisoners is already high compared with the general population, potentially driving a saturation effect, where known general population risk factors have a weaker effect in prisoners. Self-harm prevention should target people in contact with criminal justice, irrespective of imprisonment. In prisons, prevention efforts deployed at inception should target those with prior self-harm, drug use, learning difficulties, bipolar disorder and those on remand.