Adverse Short-Term Cardiometabolic Outcomes in Psychosis Early Intervention Services: Which Risk Prediction Algorithm?
Child And Adolescent Mental Health
Adverse Short-Term Cardiometabolic Outcomes in Psychosis Early Intervention Services: Which Risk Prediction Algorithm?
Child And Adolescent Mental Health

Adverse Short-Term Cardiometabolic Outcomes in Psychosis Early Intervention Services: Which Risk Prediction Algorithm?

Early Interv Psychiatry. 2025 Dec;19(12):e70116. doi: 10.1111/eip.70116.

ABSTRACT

INTRODUCTION: Weight gain and other adverse cardiometabolic changes occur early in severe mental illness (SMI) but may be preventable. It is unknown whether existing cardiometabolic risk prediction algorithms can be repurposed to predict early cardiometabolic changes after first-episode psychosis, especially in younger populations.

METHODS: Data from newly enrolled patients (16-65 years) to the Cambridgeshire and Peterborough CAMEO Early Intervention Service for Psychosis (EIS) (March 2022-Feb 2024) were used to explore the accuracy of population-based (QRISK3, QDIABETES) and SMI-specific (PsyMetRiC-full, PsyMetRiC-partial, PRIMROSE) cardiometabolic risk prediction algorithms for early cardiometabolic changes (weight, total cholesterol, triglycerides). Risk scores were calculated using baseline clinic data, and outcomes assessed at three-monthly follow-up. Multiple imputation was used where appropriate. Predictive accuracy was assessed using the coefficient of determination (R2) and root mean squared error (RMSE). Sensitivity analysis was conducted in younger (16-35 years) patients.

RESULTS: We included n = 74 participants (male = 50%, white European = 75%, mean age = 32.0, median follow-up = 112 days). Mean body weight increased by 3.58 kg, total cholesterol by 0.64 mmol/L, and triglycerides by 0.84 mmol/L. PsyMetRiC-full was the best predictor of weight (R2 = 0.54, 95% C.I., 0.36-0.69, p < 0.001; RMSE = 1.91; 95% C.I., 1.58-2.37); and triglyceride changes (R2 = 0.36; 95% C.I. 0.12-0.56, p = 0.004; RMSE = 1.98; 95% C.I., 1.43-2.63). No algorithm accurately predicted total cholesterol changes. General population-based algorithms performed poorly. Sensitivity analysis results were more extreme in favour of the PsyMetRiC algorithms.

CONCLUSION: PsyMetRiC is likely to be better than alternatives at predicting short-term cardiometabolic changes in early psychosis, particularly in patients aged 16-35. Stakeholder engagement is now required to co-decide ‘high-risk’ thresholds and intervention strategies to reduce the cardiometabolic impacts of psychosis and its treatment.

PMID:41305929 | DOI:10.1111/eip.70116