[The indication to statin therapy in primary prevention patients with dyslipidemia: implications for using national risk functions in the Italian population].

Background: Risk functions for cardiovascular risk estimation, specific for the Italian population, have recently been developed. It is possible that using them, instead of the Framingham algorithm, to assess risk and define the indication to cholesterol-lowering therapy might substantially change the rate of statin prescription in primary prevention.

Methods: In this study, two different national risk functions, the CUORE Project algorithm and the risk function incorporated in the software Riscard 2002, have been compared to the Framingham algorithm in a cohort of 517 dyslipidemic asymptomatic patients consecutively addressed to a lipid clinic. Contingency tables and kappa value estimation have been used to assess the extent of concordance between them in classifying patients into risk categories, as well as in identifying among them those deserving statin therapy, according to two different sets of guidelines, such as the Adult Treatment Panel III and the reimbursement criteria for statins set by the Italian National Health System.

Results: Both national algorithms gave lower risk estimations, in comparison with the Framingham algorithm. A low concordance was found even between the two national algorithms, with lower risk estimates by Riscard 2002. As a consequence, less patients were selected for treatment according to national risk functions. However the prescription rate was more strongly affected by the set of guidelines used to assess the indication to treatment, independent of the method used to estimate risk.

Conclusions: Our study confirms that using different risk functions can substantially change risk estimation in dyslipidemic patients, with some implications for statin prescription rate. However, the latter is mainly influenced by the set of guidelines used to identify patients for treatment. Furthermore, the two national algorithms so far available for risk estimation in the Italian population strongly differ in risk estimates, suggesting the need for further testing their accuracy.