Risk assessment in the primary prevention of cardiovascular disease
Brindle and colleagues explored the external validity of the Framingham prediction rules in the primary prevention of cardiovascular disease, and its impact on clinical outcomes. They systematically reviewed studies that compared the predicted risk of coronary heart disease or cardiovascular disease, with observed 10-year risk. They found substantial under-prediction in high risk populations and over-prediction in lower risk populations.
In 2003 we systematically reviewed the internal and external validity of Framingham equation based prediction tools (Sheffield tables, New Zealand Charts, Framingham Categorical Risk Charts, Canadian Tables, Joint British and European Charts (1994)), the PROCAM risk calculator, the UKPDS risk engine and the SCORE risk charts in order to explore whether they can be used in informed or shared decision making in cardiovascular prevention (1).
Our focus was the middle European population. Although our electronic searches were not that sensitive, our results are consistent with the findings of Brindle et al. We found substantial overestimation of coronary risk with the Framingham-based charts. We did not find any external validation studies for the new European SCORE risk charts and the UKPDS risk engine. We identified one external validation study for the PROCAM risk calculator. We also observed that the application of these tools is widely recommended.
In our review we additionally scrutinised the discriminatory ability of the risk charts, which is to correctly classify subjects to their coronary risk. As the core of informed or shared decision making is the patient perspective, discriminatory ability is an important property of risk prediction, beyond calibration. The patient would like to know how his prognosis is, and whether it is accurate. We identified poor discriminatory ability of Framingham-based tools and the PROCAM-Risk calculator. We therefore decided to not apply these tools in informed or shared decision making for cardiovascular disease.
(1) Lenz M, Mühlhauser I. [Cardiovascular risk assessment for informed decision making. Validity of prediction tools]. Med Klin (Munich) 2004;99:651-61.