Validation of an ICD-Code-Based Case Definition for Psychotic Illness Across Three Health Systems

Area of Science:

• Psychiatry
• Medical Informatics
• Health Services Research

Background:

• Electronic health records (EHRs) increasingly use diagnostic codes for psychosis detection.
• Limited data exists on the accuracy of psychosis diagnostic codes.
• This study assesses the validity of International Classification of Diseases (ICD) codes for psychosis.

Purpose of the Study:

• To evaluate the positive predictive value (PPV) of ICD codes for psychosis.
• To assess the accuracy of psychosis diagnostic codes in EHRs across different health systems.
• To determine if the first documented code reliably identifies first-episode psychosis (FEP).

Main Methods:

• Utilized EHR data from three health systems.
• Grouped ICD codes for psychotic and mood disorders with psychosis into five categories.
• Sampled 1133 records for chart review to calculate PPVs.

Main Results:

• Overall PPVs exceeded 70% across health systems (0.72-0.83).
• Schizoaffective disorder codes showed the highest PPVs (0.80-0.92); major depressive disorder with psychosis varied (0.57-0.79).
• PPVs were substantially lower (0.08-0.62) when identifying first-episode psychosis.

Conclusions:

• The first psychosis diagnostic code accurately captures psychosis episodes.
• This code is not a reliable indicator of first-episode psychosis (FEP).
• Findings impact the development of risk prediction models for undiagnosed psychosis.