Models to predict relapse in psychosis: A systematic review

Area of Science:

• Psychiatry
• Clinical Epidemiology
• Data Science

Background:

• Limited evidence exists on the accuracy of psychosis relapse prediction models.
• Accurate prediction of relapse is crucial for timely intervention in psychosis.

Purpose of the Study:

• To systematically review existing literature on the accuracy of psychosis relapse prediction models.
• To identify reliable predictors for psychosis relapse.

Main Methods:

• A comprehensive literature search was conducted across multiple databases (Embase, Medline, PsychINFO, etc.) up to September 2016.
• Prediction modeling studies were assessed for risk of bias and applicability using the PROBAST tool.
• Included studies focused on models developed and/or validated for psychosis relapse prediction.

Main Results:

• Two studies involving 33,088 participants were eligible.
• One model using prodromal symptoms had high risk of bias, explaining 14% of relapse variance.
• A second model using administrative data demonstrated moderate discriminative capacity (C = 0.631) with low risk of bias.

Conclusions:

• High-quality evidence is lacking, precluding specific recommendations for psychosis relapse prediction models.
• The utility of prodromal symptoms for predicting relapse remains unclear.
• Developing prediction models using routine data appears promising but requires further investigation.