Background: To assess the reliability and accuracy of the Framingham coronary heart disease (CHD) risk function adapted by the Registre Gironí del Cor (REGICOR) investigators in Spain.
Methods: A 5-year follow-up study was completed in 5732 participants aged 35-74 years. The adaptation consisted of using in the function the average population risk factor prevalence and the cumulative incidence observed in Spain instead of those from Framingham in a Cox proportional hazards model. Reliability and accuracy in estimating the observed cumulative incidence were tested with the area under the curve comparison and goodness-of-fit test, respectively.
Results: The Kaplan-Meier CHD cumulative incidence during the follow-up was 4.0% in men and 1.7% in women. The original Framingham function and the REGICOR adapted estimates were 10.4% and 4.8%, and 3.6% and 2.0%, respectively. The REGICOR-adapted function's estimate did not differ from the observed cumulated incidence (goodness of fit in men, p = 0.078, in women, p = 0.256), whereas all the original Framingham function estimates differed significantly (p<0.001). Reliabilities of the original Framingham function and of the best Cox model fit with the study data were similar in men (area under the receiver operator characteristic curve 0.68 and 0.69, respectively, p = 0.273), whereas the best Cox model fitted better in women (0.73 and 0.81, respectively, p<0.001).
Conclusion: The Framingham function adapted to local population characteristics accurately and reliably predicted the 5-year CHD risk for patients aged 35-74 years, in contrast with the original function, which consistently overestimated the actual risk.