Pharmacist University of British Columbia UBC Vancouver, British Columbia, Canada
Background: Recent evidence indicates that the widely used QRISK3 score underestimates the risk of cardiovascular disease in patients with chronic obstructive pulmonary disease (COPD), raising concerns about the accuracy of similar risk assessment tools employed in North America. We examined the performance of three risk equations– simplified Predicting Risk of Cardiovascular Disease EVENTs (PREVENT), Pooled Cohort Equations (PCEs), and the 2008 global Framingham Risk Score (FRS; widely used based on the Canadian Cardiovascular Society dyslipidemia guidelines) to estimate 10-year total cardiovascular disease (CVD) risk in individuals with COPD.
METHODS AND RESULTS: Individuals ≥40 years of age with COPD were identified from five longitudinal, community-based epidemiologic North American cohort studies (ARIC, MESA, Jackson Heart Study, Cardiovascular Health Study, and Framingham cohorts). The 10-year risk estimates from all three models were derived based on definitions as used in the original model development studies. COPD was defined according to the Global Initiative for Chronic Obstructive Lung Disease guidelines as persistent airflow obstruction based on a post-bronchodilator FEV₁/FVC ratio of < 0.70 at baseline or study follow-up. Discrimination (using time-dependent area under the receiver operating characteristic curve) and calibration (using observed to the average estimated risk ratio (O/E)) were assessed.
We included 5,770 COPD patients in the analysis of PREVENT and the PCEs, and 5,522 in the analysis of the FRS. PREVENT demonstrated the highest discrimination (c-statistic 0.76) followed by PCEs (0.66), and FRS (0.57). The PREVENT underestimated risk (O:E ratio 1.25), whereas the PCEs (0.73) and FRS (0.67) significantly overestimated risk, by approximately 30%. These discrepancies varied by age and sex, with a more pronounced underestimation in younger adults with PREVENT (O/E ratio: 1.4, 95% CI: 1.3, 1.5) and overestimation in older adults with PCEs (O/E ratio: 0.54, 95% CI: 0.44, 0.64; Table 1).
Conclusion: Primary-prevent CVD risk models exhibited variable and modest discrimination, and significant miscalibration in patients with COPD. Recalibration of existing models or development of COPD-specific CVD prediction models may improve risk prediction and CVD primary-prevention decisions.
