To the Editor: Dr Ridker and colleagues1 have proposed the Reynolds Risk Score as an improvement in cardiovascular risk estimation for women. Several concerns regarding the authors' analysis and interpretation should seriously limit enthusiasm for this tool.
First, the equations were derived from a cohort of women health professionals enrolled in a clinical trial, not a population-based sample. Second, the design of the study included self-report of critically important covariates including weight, diabetes, smoking, and blood pressure. In the case of blood pressure, participants did not report measured blood pressure, only categories of blood pressure (ie, within 10 mm Hg increments). No published validation study has been done for the self-reported covariates in this cohort. Given the imprecise ascertainment, the present study probably underestimates the contribution of conventional risk factors to risk models. Third, the authors use the ATP III risk estimator to predict a composite end point of cardiovascular events and revascularizations. This estimator was originally designed with covariate weights to predict coronary disease events, not all cardiovascular events and revascularizations. When the authors appropriately readjust the weights of the ATP III covariates to predict the different outcome (Table 3), the ATP III covariates perform just as well as the Reynolds score.
Wang TJ, Kathiresan S, Lloyd-Jones DM. Algorithms for Assessing Cardiovascular Risk in Women. JAMA. 2007;298(2):175-178. doi:10.1001/jama.298.2.176-a