Related Experiment Videos
Development and Evaluation of a Rules-based Algorithm for Primary Open-Angle Glaucoma in the VA Million Veteran
Cari L Nealon1, Christopher W Halladay2, Tyler G Kinzy3,4,5
1Eye Clinic, VA Northeast Ohio Healthcare System, Cleveland, OH, USA.
Ophthalmic Epidemiology
|November 25, 2021
Summary
Researchers developed a computable phenotype using electronic health records to accurately identify primary open-angle glaucoma (POAG) in veterans. This method, validated with genetic data, enables better study of complex eye diseases.
Area of Science:
- Ophthalmology
- Genetics
- Health Informatics
Background:
- Electronic health record (EHR)-linked biobank data offers research potential for complex ocular diseases.
- Accurate computable phenotypes for imaging-diagnosed ocular conditions are often unavailable in EHRs.
- Primary open-angle glaucoma (POAG) diagnosis typically requires imaging, posing a challenge for EHR-based research.
Purpose of the Study:
- To develop and validate a computable phenotype for identifying POAG using the Department of Veterans Affairs (VA) Computerized Patient Record System (CPRS) and Million Veteran Program (MVP) biobank.
- To create and refine algorithms for POAG case and control identification based on clinical, prescription, and diagnosis data.
- To assess the accuracy and reproducibility of the developed algorithms in classifying POAG.
Main Methods:
- Utilized CPRS clinical ophthalmology data from VA Medical Center Eye Clinic (VAMCEC) patients.
- Developed and iteratively refined POAG algorithms using clinical, prescription, and ICD-CM codes.
- Validated algorithms through chart review at multiple VAMCECs and assessed predictive values (PPV, NPV) with expert clinical diagnosis data.
Main Results:
- The final algorithms demonstrated high accuracy and reproducibility for POAG classification, with positive predictive values (PPV) greater than 83% and negative predictive values (NPV) greater than 97%.
- Algorithms performed consistently across different racial groups, including Black or African American and White Veterans.
- Application to the MVP biobank and subsequent genetic analysis of a known POAG locus further validated the algorithms' effectiveness.
Conclusions:
- The developed computable phenotype using combined EHR and genetic data is a viable approach for studying complex diseases like POAG.
- This method enhances the accuracy and reproducibility of disease classification in large biobank datasets.
- The approach facilitates research into complex ocular diseases by overcoming limitations in EHR data accessibility for imaging-dependent diagnoses.