Combining Cross-sectional Data on Prevalence with Risk Estimates from a Prediction Model

Death from cardiovascular diseases (CVD) is still the leading cause of death in Germany. About 271 people per ten thousand died of stroke and 179 of heart attack in 2008, both contributing to a combined percentage of about 37% of all deaths [1]. Well established risk factors for CVD include life style factors such as smoking, obesity, lack of physical activity, and diet factors [2]. More recently, research has focused on genetic risk factors [3]. Several models have been established to predict an individual’s risk of dying within the next five or ten years from CVD depending on the individual’s risk factors. The most important models for Germany are PROCAM [4] and HeartScore Germany [5], with the latter considering the important risk factors sex, age, blood pressure, smoking status, and cholesterol level. As some of these risk factors depend on personal life style choices, changing one’s life style, e.g. by giving up smoking, may prevent death from CVD. The magnitude of the avoidable fraction depends on the strength of the association and the prevalence of the modifiable risk factors. While quantification of the fraction of avoidable death from CVD has been investigated by several authors in other countries, such as the USA [6, 7], the Czech Republic [8] or Japan [9], and in particular in Combining Cross-sectional Data on Prevalence with Risk Estimates from a Prediction Model