Cardiovascular Disease Mortality Risk Prediction in Educational Classes across European Regions, in the MORGAM Project Populations

Aims: Although socioeconomic status is a recognized independent risk factor for CVD mortality, the recommended European risk prediction equation for primary prevention does not consider it; an approach criticized by previous results in the UK and US. We aim to assess whether the SCORE project equation adequately estimates the risk in different educational classes, across several European populations. Methods: We considered 47 prospective population-based surveys from Nordic Countries (Finland, Denmark, Sweden), UK (Belfast and Scotland), Central Europe (France, Germany and Italy) and East Europe (Lithuania, Poland) and Russia. Baseline data collection and mortality follow-up (median time 10 years) adhered to standardized MONICA-like procedures. Three educational classes were derived from population-, sex- and birth year-specific tertiles of years of schooling. The individual SCORE risk was computed from age, total cholesterol, systolic blood pressure and smoking; the risk was recalibrated to the average observed risk in each population. We estimated age- and traditional risk factors-adjusted hazard ratios (HR) for 10 year CVD mortality (highest education as the reference), from Cox models. Moreover, the observed number of fatal CVD events by educational class was compared to the expected number, as estimated by the recalibrated SCORE function. Results: The cohorts summed-up 39,215 men and 29,240 women 40 to 64 years old and free from CVD event at baseline. Education was associated with CVD mortality in men (pooled age-adjusted HR for low vs high education: 1.6, 95% CI 1.4–1.9); the hazard ratios ranged from 1.3 (95%CI: 0.9–1.8) in Central Europe to 2.1 (1.6–2.7) in East Europe and Russia. The association attenuated after adjustment for SCORE risk factors and HDL-cholesterol. Among women, the association was significant in Nordic Countries only (age-adjusted HR for low vas high education: 1.7, 95% CI 1.1–2.6), but it was no more significant after adjustment for multiple risk factors. The original SCORE equation overestimated the risk at a population level, both in men and in women, except in East Europe and Russia. After recalibration, the SCORE equation overestimated the risk among the more educated men by 20% to 50% (in Central Europe, East Europe and Russia, respectively), but underestimated it in the less educated men by 7% to 23% (in Central Europe, East Europe and Russia, respectively). Conclusions: Our results, based on a well-harmonized study comprising several European populations, suggest the need to include country-specific socioeconomic status in the risk estimation equations.