A plethora of prediction models have been developed across medicine. Some have had significant clinical impact, particularly in cancer and cardiovascular medicine - translating evidence from large datasets into stratified healthcare. As clinical adjuncts, simple validated tools can improve communication, consistency and transparency of decisions around care. However, models developed vastly outnumber those adopted. This can be due to methodological limitations in development, lack of validation in relevant patient groups, or overlooking clinical feasibility and acceptability. Based on a literature review and new empirical research in general psychiatry, we will outline key principles in applying prediction model approaches to three important adverse outcomes in psychiatric settings – suicide, violence, and cardiometabolic risk. This will include two recent studies validating prediction models in psychosis services; “OxMIV”, for predicting violence, developed using large register data and validated in 1,145 individuals presenting to Early Intervention in Psychosis Services (EIS) alongside clinical feasibility work; and “PsyMetRiC”, for predicting the risk of developing metabolic syndrome in young adults with psychosis, developed and externally validated using routine electronic health record data from three UK EIS. Prediction of suicide in high risk groups will be examined, and new evidence on the challenges of accuracy and links to interventions.
Chair: Professor Belinda Lennox, University of Oxford
Predicting violence risk in first episode psychosis: validity, feasibility and acceptability of the "OxMIV" tool - Dr Daniel Whiting, Department of Psychiatry, University of Oxford, and Oxford Health NHS Foundation Trust
"PsyMetRiC”: a cardiometabolic risk prediction calculator for young adults with psychosis - Dr Benjamin Perry, Department of Psychiatry, University of Cambridge, and Cambridgeshire and Peterborough Foundation NHS Trust
Evidence for risk assessment for suicide in high-risk groups - Professor Seena Fazel, Department of Psychiatry, University of Oxford, and Oxford Health NHS Foundation Trust