Equalization of four cardiovascular risk algorithms after systematic recalibration: individual-participant...

Area of Science:

• Cardiology
• Public Health
• Epidemiology

Background:

• Cardiovascular disease (CVD) risk estimation relies on algorithms with debated optimal performance.
• Primary prevention guidelines recommend several risk assessment tools, but their accuracy varies across populations.
• Recalibration is a method to adapt risk algorithms using population-specific data.

Purpose of the Study:

• To compare the performance of four major CVD risk estimation algorithms head-to-head.
• To evaluate the impact of recalibration on the accuracy and clinical utility of these algorithms.
• To assess the effect of recalibration on the identification of individuals eligible for preventive therapies, such as statins.

Main Methods:

• Individual-participant data from 360,737 participants without CVD across 86 prospective studies were analyzed.
• Four algorithms were compared: Framingham risk score (FRS), Systematic COronary Risk Evaluation (SCORE), pooled cohort equations (PCE), and Reynolds risk score (RRS).
• Risk discrimination and calibration were assessed before and after algorithm recalibration using target population data.

Main Results:

• All four algorithms demonstrated similar risk discrimination capabilities.
• Before recalibration, FRS, SCORE, and PCE over-predicted CVD risk, while RRS under-predicted it.
• Recalibration reduced the proportion of individuals aged ≥40 years classified as high risk from 29-39% to 22-24%, improving the targeting of preventive interventions.

Conclusions:

• The clinical performance of CVD risk algorithms varies significantly before recalibration.
• Simple recalibration substantially improves and nearly equalizes the performance of these algorithms.
• Recalibration enhances the precision of preventive actions, aligning treatment decisions with clinical need.