Do family history of coronary heart disease and social status improve 20-year risk prediction of first major cardiovascular event?

Introduction. Family history of CHD and low socio-economic status are well-established independent risk factors with the same level of evidence as biomarkers like high-sensitivity CRP or fibrinogen [1]. In addition, they are relatively easy to assess in clinical practice at a lower cost than biomarkers. However, their contribution to risk prediction beyond traditional risk factors has been examined to a lesser degree and with controversial findings; they are included in only a few risk equations, mainly from the UK. Aims. To assess whether family history of CHD and social status might improve long-term risk prediction in a Northern Italy population. Methods. N=3956 35-69 years old men and women free of cardiovascular disease were enrolled in three independent population-based cohorts conducted between 1986 and 1990 in Brianza (Northern Italy). Self-reported positive family history of CHD (prevalence: 27% in men, 34% in women) was ascertained at baseline. Three educational classes (high, intermediate and low education) were defined from age- and sex-specific tertiles of years of schooling. Absolute 20-year risk of first fatal or non-fatal coronary or ischemic stroke event during follow-up (MONICA validated) was estimated from gender-specific Cox models including age, total- and HDL-cholesterol, systolic blood pressure, anti-hypertensive treatment, diabetes and smoking (reference model). Model calibration (Grønnesby-Bogan goodness-of-fit test) and discrimination (Area Under the ROC-Curve, AUC) were estimated taking censoring into account. Changes in discrimination (Δ-AUC) and reclassification (Net Reclassification Improvement, NRI) defined the improvement from the reference model due to the addition of education and family history of CHD. Bootstrapped confidence intervals (CI) for Δ-AUC and NRI are also provided. Results. The estimated Kaplan-Meier 20-year risk was 16.8% in men (254 events) and 6.4% in women (102 events). All the models were well calibrated (goodness-of-fit p-value >0.20 in both genders). Education (2 df test p-value 0.03) and family history of CHD (hazard ratio: 1.55; 95%CI 1.19-2.03) were associated with the endpoint in men, but not in women. In men, the addition of both education and family history improved discrimination (Δ-AUC=0.01; 95% CI 0.002-0.02) and risk stratification (NRI=6%; 95%CI: 0.2%-15.2%). Considering men at intermediate risk (10%-20%) according to the reference model, NRI among cases was 12%, and the overall NRI was 20.1% (95%CI: 0.5%-44%). Conclusions. In this northern Italian population, two indicators of genetic risk and social status improve long-term risk prediction in men beyond traditional risk factors. Due to the low cost of assessment, they should be implemented in standard algorithms for risk prediction.