Performance of PREVENT Cardiovascular Risk in Electronic Health Record–Based Clinical Practice
JAMA Network Open
Chuan Hong, PhD1; Mu Niu, BS2; Haoyuan Wang, MS; Daniel M. Wojdyla, MS; Nicoleta Economou-Zavlanos, PhD; Matthew M. Engelhard, MD, PhD; Michael Pignone, MD; Manesh R. Patel, MD; Michael J. Pencina, PhD
Summary
Question Do the Predicting Risk of Cardiovascular Disease Events (PREVENT) equations maintain 5-year cardiovascular disease (CVD) risk performance across subgroups under electronic health record (EHR) conditions with missing data?
Findings In this cohort study using data from the Duke University Health System EHR to create a cohort of 127 151 individuals with complete data and a cohort of 406 230 individuals with partially missing data, PREVENT showed strong discrimination with consistent subgroup performance. Original PREVENT equations modestly underestimated risk; local adaptation minimally improved calibration without affecting discrimination.
Meaning These findings suggest that the PREVENT equations can be applied to detect increased CVD risk in common clinical settings, including those with missing laboratory or vital sign data when relevant imputation is used.
Citation
Hong, Chuan, et al. “Performance of PREVENT Cardiovascular Risk in Electronic Health Record–Based Clinical Practice.” JAMA Network Open 9.4 (2026): e266838.