Cardiovascular risk scores underestimate risk in women with CKD

Design

Secondary analysis of a prospective cohort

Population

21,548 postmenopausal women without cardiovascular disease, 1,095 with CKD

Primary outcome

Discrimination and calibration of five risk models

Effect

C-statistic 0.57–0.67 in CKD; most models underpredicted

Using the Women's Health Initiative biomarker cohorts, investigators tested five risk models (Framingham, ATP III, Pooled Cohort Equations, QRISK3 and PREVENT) in 21,548 postmenopausal women aged 50 to 79 without cardiovascular disease, of whom 1,095 had an eGFR below 60.

In women with CKD, discrimination was poor: C-statistics ran from 0.57 for Framingham to 0.67 for PREVENT. Calibration was poor in both groups, with most models underpredicting risk in women with CKD, and PREVENT underpredicting in women without CKD too.

A risk score that reads low in a woman with CKD may be falsely reassuring. CKD itself is a strong cardiovascular risk factor, and a low calculated score should not on its own be the reason to withhold a statin or blood pressure treatment that would otherwise be considered.

• Treat a low calculated cardiovascular risk score with caution in women with an eGFR below 60.
• Weigh CKD, albuminuria and other kidney markers directly when deciding on preventive treatment.
• Record which risk model you used, since estimates differ widely between them.
• Revisit risk as eGFR or albuminuria changes, not only at fixed intervals.

Why it matters

Risk calculators are making treatment decisions for a group they were not built for.

Don't overread it

This cohort was postmenopausal women in the United States; performance in men or other populations may differ.

The statistics, in plain English

A C-statistic of 0.5 is no better than chance and 1.0 is perfect; values around 0.6 mean the model ranks women only slightly better than chance. Underprediction means more events occurred than the model forecast.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.