Model-Based Algorithms for Detecting Peripheral Artery Disease Using Administrative Data From an Electronic Health...

Area of Science:

• Cardiovascular Medicine
• Health Informatics
• Biostatistics

Background:

• Peripheral artery disease (PAD) affects millions, posing significant mortality and amputation risks.
• PAD is underdiagnosed and undertreated, hindering effective care and research.
• Identifying PAD patients in electronic health records (EHRs) is crucial but challenging.

Purpose of the Study:

• To develop and validate a model-based algorithm for detecting PAD patients using EHR data.
• To improve the identification of individuals with peripheral artery disease for clinical and research purposes.

Main Methods:

• An EHR query identified patients with PAD-related diagnosis codes.
• A logistic regression model with LASSO was built and validated using adjudicated patient data.
• The algorithm incorporated diagnosis codes, administrative, imaging, and procedure flags.

Main Results:

• The initial query identified 15,406 patients with PAD-related codes.
• A LASSO model using 108 code flags achieved an area under the curve of 0.862.
• The model effectively identified patients with peripheral artery disease.

Conclusions:

• The developed algorithm accurately identifies PAD patients from large EHR datasets.
• This model is advantageous as it uses broad patient data and does not rely on clinical notes.
• It enables PAD identification using only administrative billing data, facilitating large-scale cohort studies.