Oxford Predicting Outcomes in Psychosis

Oxford Predicting Outcomes in Psychosis


The development of a prediction model of relapse following discharge from Early Intervention in Psychosis Services 

Aim of the Study

To develop a model to predict acute relapse within a year of discharge from Early Intervention in Psychosis (EIP) services to assist in discharge decision making. 

The Challenge

EIP services are the gold standard of care for people with first episode psychosis (FEP). EIP services are time-limited to three years of treatment, after which patients are either discharged to their GP or transferred to adult community mental health services. Risk assessment of who needs further secondary mental health care is difficult, and relapse is common in this patient group. We aimed to create a risk prediction tool to aid decision making. 

The Method

We identified all patients who had been treated by Oxford Health NHS Foundation Trust EIP services using the UK CRIS platform. We measured the risk of 1-year relapse using a logistic regression model, adjusting for previously identified risk-factors. 


The Outcome

We identified 831 eligible participants. The majority had non-affective psychoses (58.6%), 83.6% were discharge to primary care, and 79 (9.5%) were readmitted to hospital within one year of discharge. Performance of the model at predicting relapse was good, with a c-statistic of 0.795 (95%CI 0.794; 0.796). 


The Conclusion

The OX-POP prediction model could be of clinical usefulness in helping decision making at the time of EIP discharge. Further external validation of the model is currently underway in other UK CRIS sites. 

Find out how our data and consultancy can help you

Contact Us