Optimizing early prediction for antipsychotic response in schizophrenia

Area of Science:

• Psychiatry and Behavioral Sciences
• Clinical Psychology
• Biostatistics

Background:

• Previous prediction models for antipsychotic treatment response in schizophrenia had limitations, particularly low sensitivity.
• Early identification of treatment response is crucial for optimizing patient care and managing schizophrenia effectively.

Purpose of the Study:

• To develop and validate a novel statistical method for optimizing the early prediction of treatment response in schizophrenia.
• To create a prediction model applicable within standard personal computer systems, such as Microsoft Excel.

Main Methods:

• A cohort of 123 inpatients with acute schizophrenia exacerbation received risperidone treatment.
• The generalized estimating equation (GEE) logistic regression method was employed.
• Prediction models were established using treatment response data from the first two weeks of therapy.

Main Results:

• The model demonstrated high accuracy in predicting nonresponse at weeks 4 and 6 (80.8% and 81.8%).
• The model also accurately identified responders at weeks 4 and 6 (80.0% and 82.8%).
• Overall predictive power reached 80.3% at week 4 and 82.4% at week 6, with Positive and Negative Syndrome Scale scores yielding slightly better results than Brief Psychiatric Rating Scale scores.

Conclusions:

• Predicting treatment response at weeks 4 or 6 based on the first two weeks of risperidone therapy is feasible and accurate.
• The developed model offers acceptable specificity, sensitivity, and predictive power for clinical application.
• Further research is warranted to explore its applicability to other psychotropic medications, including antidepressants.