A Computable Phenotype for Autosomal Dominant Polycystic Kidney Disease

Area of Science:

• Nephrology
• Medical Informatics
• Genetics

Background:

• Computable phenotypes are algorithms for identifying patient groups in electronic health records.
• Accurate identification of autosomal dominant polycystic kidney disease (ADPKD) patients is crucial for research.
• International Classification of Diseases (ICD-9/10) codes are commonly used in EHRs.

Purpose of the Study:

• To assess the accuracy of a computable phenotype for identifying ADPKD patients using ICD-9/10 codes.
• To evaluate the phenotype's utility in clinical trial recruitment.

Main Methods:

• Reviewed four patient samples from the EHR database based on ICD-9/10 codes.
• Included patients with and without ADPKD codes, and with/without nephrology clinic visits.
• Determined ADPKD status using internationally accepted diagnostic criteria.

Main Results:

• The computable phenotype demonstrated high sensitivity (99% for nephrology patients, 97% for non-nephrology patients).
• Specificity was also strong (84% for nephrology patients, 82% for non-nephrology patients).
• The algorithm accurately identified ADPKD patients across different clinical settings.

Conclusions:

• A computable phenotype utilizing ICD-9/10 codes can effectively identify ADPKD patients.
• This tool offers acceptable accuracy for screening healthcare records for ADPKD cohorts.
• Facilitates efficient patient identification for research and clinical studies.