Journal of General Internal Medicine · Oct 26, 2026
Cardiovascular Risk Prediction in Postmenopausal Women With and Without Chronic Kidney Disease: An Observational Cohort Study
By Gallo RJ, Qin F, Hedlin H et al.
BACKGROUND: Chronic kidney disease (CKD) is an important risk factor for cardiovascular events, yet performance of cardiovascular prediction models are understudied in the CKD population, including women with CKD. OBJECTIVE: To investigate the performance of multiple cardiovascular prediction models, the Framingham, Adult Treatment Panel (ATP) III, ACC/AHA Pooled Cohort Equations (PCE), QRISK3, and PREVENT models, in women with and without CKD. DESIGN: Secondary analysis of a prospective cohort study (Women's Health Initiative Biomarker Studies). PARTICIPANTS: 21,548 postmenopausal women ages 50-79 without cardiovascular disease, including 1,095 with CKD defined by eGFR < 60 mL/min per 1.73 m². MAIN MEASURES: Model discrimination by C-statistic. Calibration by clinically relevant decision-making thresholds. Included events varied by model: Framingham model outcomes included myocardial infarction (MI), angina, or coronary heart disease (CHD) death; ATP III model outcomes included fatal and nonfatal CHD; PCE and PREVENT model outcomes included CHD or ischemic stroke; QRISK3 model outcomes included CHD, stroke, or transient ischemic attack. KEY RESULTS: Number of cardiovascular events varied based on outcome definition. For the non-CKD group, events ranged from 846 (4.1%) for ATP III to 1,435 (7.0%) for QRISK3; for the CKD group, events ranged from 93 (8.5%) for ATP III to 145 (13.2%) for QRISK3. Discrimination by C-statistic ranged from 0.57 for the Framingham model to 0.67 for PREVENT in the CKD group; in the non-CKD group, C-statistics ranged from 0.61 for the Framingham model to 0.71 for QRISK3 and PREVENT. Calibration was generally poor in both CKD and non-CKD groups, with underprediction for women with CKD across most models, as well as underprediction for women without CKD for the PREVENT model. CONCLUSION: Common cardiovascular risk prediction models have poor discrimination and calibration in women with CKD; the PREVENT model generally underpredicted risk for women with and without CKD. CLINICAL TRIAL NUMBER: Not applicable.