C-Reactive Protein and Parental History: Reynolds Risk Score

C-Reactive Protein and Parental History Improve Global Cardiovascular Risk Prediction. The Reynolds Risk Score for Men

Paul M Ridker MD^*, Nina P. Paynter PhD, Nader Rifai PhD, J. Michael Gaziano MD, and Nancy R. Cook ScD

From the Donald W. Reynolds Center for Cardiovascular Research and the Center for Cardiovascular Disease Prevention, Divisions of Preventive Medicine and Cardiovascular Medicine, Brigham and Women's Hospital, Harvard Medical School, and MAVERIC, VA Boston Health Care System, all in Boston, Mass.

^* To whom correspondence should be addressed. E-mail: pridker@partners.org.

Background—High-sensitivity C-reactive protein and family history are independently associated with future cardiovascular events and have been incorporated into risk prediction models for women (the Reynolds Risk Score for women); however, no cardiovascular risk prediction algorithm incorporating these variables currently exists for men.

Methods and Results—Among 10 724 initially healthy American nondiabetic men who were followed up prospectivelyover a median period of 10.8 years, we compared the test characteristics of global model fit, discrimination, calibration, and reclassification in 2 prediction models for incident cardiovascular events, one based on age, blood pressure, smoking status, total cholesterol, and high-density lipoprotein cholesterol (traditional model) and the other based on these risk factors plus high-sensitivity C-reactive protein and parental history of myocardial infarction before age 60 years (Reynolds Risk Score for men). A total of 1294 cardiovascular events accrued during study follow-up. Compared with the traditional model, the Reynolds Risk Score had better global fit (likelihood ratio test P<0.001),> Bayes information criterion, and a larger C-index (P<0.001). For the end point of all cardiovascular events, the Reynolds Risk Score for men reclassified 17.8% (1904/10 724) of the study population (and 20.2% [1392/6884] of those at 5% to 20% 10-year risk) into higher- or lower-risk categories, with markedly improved accuracy among those reclassified. For this model comparison, the net reclassification index was 5.3%, and the clinical net reclassification index was 14.2% (both P<0.001).> based on the Adult Treatment Panel III preferred end point ofcoronary heart disease and limited to men not taking lipid-lowering therapy, 16.7% of the study population (and 20.1% of those at 5% to 20% 10-year risk) were reclassified to higher- or lower-risk groups, again with significantly improved global fit, larger C-index (P<0.001),>reclassified. For this model, the net reclassification index was 8.4% and the clinical net reclassification index was 15.8% (both P<0.001).

Conclusions—As previously shown in women, a prediction model in men that incorporates high-sensitivity C-reactive protein and parental history significantly improves global cardiovascular risk prediction.

Key words: epidemiology • prevention • inflammation • genetics • risk factors