The EPICARDIAN-score is based on a large cohort of 3,474 old Spanish citizens. Previous scores as Framingham and European SCORE are based on population recruited long time ago, before the 1980s, when the percent of risk factors was lower at population level. In contrast the EPICARDIAN cohort was assembled more recently and provides risk estimation for people treated with and without antihypertensive and other cardiovascular drugs. The EPICARDIAN-score overcomes also the problem of applying ß-coefficients derived from studies in middle-aged populations to older persons, which may be inadequate for risk estimation in the elderly.
Age is the strongest predictor of cardiovascular risk in older men and women. However, when using age as a time-scale variable, rather than as a standard variable, the performance of Cox model improved, particularly for women. For example, diabetes emerged as an independent predictor in women (in the original model diabetes was borderline).
A different CVRF pattern was observed in males and females. In men, diabetes was the strongest predictive factor, and high blood pressure treatment had a greater impact than in women. Smoking shown also a significant effect on CVD, but it was higher in women than in men. Similar to previous studies [21, 22], the contribution of TC to CVD act in the reverse direction after the age of 70. The absence of a positive predictive value for TC raises the concept that risk estimation in the elderly may warrant a different approach than in middle-aged individuals, and confirms the predictive power of certain classic risk factors as hypercholesterolemia diminish with age.
Current charts used to predict CVD in old people mainly rely on calibrated risk function from the original Framingham  and the European scores . Important limitations of these scores are the small number of people older than 70 years, the occupational origin of some cohorts, and the consideration of CVD mortality as the only main outcome in the predictive model. In contrast, the EPICARDIAN score includes an important proportion of individuals older than 75 years, and provides total CVD risk, a more comprehensive estimate of the individual total risk in a Mediterranean country as Spain, with low rates of CHD. Moreover, previous European charts for the elderly do not include diabetes or, as the Framingham algorithm, prediction is based on a small number of diabetic people (4%), not based on the current International definition of diabetes. Finally, the EPICARDIAN-charts uniquely accounted for the use of antihypertensive drugs, the most common treatment in older people, which allows solving a major gap in previous risk assessment tools .
Recently, a new European SCORE system for old persons, SCORE-OP, has been published . We applied the SCORE-OP to our population, and found the 10-year average CVD risk was significantly lower than the obtained with the EPICARDIAN charts, both in men and women . This is most probably due to the fact that the mean age of the EPICARDIAN population is older, and the prevalence of diabetes and smoking higher than in the cohorts included in the SCORE-OP, mainly coming from Northern and Central Europe. Also, the two equations do not include exactly the same parameters. For instance the EPICARDIAN equation includes hypertension treatment as a variable in the model, while the SCORE-OP includes c-HDL. Finally we also have shown that the SCORE-OP identifies fewer patients with 65 year and older, at high or very high risk, than the original derived SCORE function for low-risk countries, as Spain .
The EPICARDIAN charts consider only the major CVRF. Other factors as family history of CVD, fibrinogen level, physical activity, waist circumference or C-reactive protein, can theoretically modulate the CVD risk in the elderly. In contrast, as mentioned above the predictive value of classic risk factors, as hypercholesterolemia or systolic hypertension, weakens with age [8, 21], and the absolute risk estimations in some cells, particularly in those groups of older age with few individuals with very high levels of TC and SBP could be less accurate because of this phenomenon.
Our study has also strengths, as a sample recruited from urban and rural settings, different geographical locations, wide range of old ages, and large proportion of diabetic individuals. The EPICARDIAN cohort provides the first prospective association between total CVD and several CVRFs for the elderly in Spain, a country with low incidence of CVD. The EPICARDIAN system allows the direct estimation of total CVD risk based on the background contemporaneous risk of the elderly population, rather than based in foreign scores calibrated for low risk European population. Predictions based on these calibrations may not work if a long time and a marked change in CVD mortality occurred between the risk score was derived and applied to different populations.