Agreement of PROCAM and SCORE to assess cardiovascular risk in two different low risk European populations.

Abstract

The quality of cardiovascular risk assessment and subsequent medical treatment for those at elevated risk is dependent on the precision of resource allocation. Several confounders may lead to unnecessary treatments, while those at elevated risk may remain untreated. Confounders occur both in calibration and discrimination. While calibration defines the threshold for individuals, which labels them as low, intermediate, high or very high subjects, discrimination describes the diagnostic accuracy of a test to detect those with future disease (Romanens et al., 2010). Once a test for cardiovascular risk assessment, such as the Framingham algorithm, has been derived from a population over decades, accuracy in different populations and continents still has to be undertaken (external validation). External validation is however difficult to obtain because contemporary performance can be assessed only after decades of observation of the outcome variable. Substitutions are however possible, e.g. with atherosclerosis imaging of carotid or coronary arteries (Romanens et al., 2017) or by measuring cardiovascular risk at the time of a cardiovascular event (Mortensen and Falk, 2017). Other important issues are the demographic or ethnic background of individuals assessed with such risk tools or the type of occupation. The working group on lipids and atherosclerosis (AGLA) recommends the use of cardiac or cardiovascular risk derived from PROCAM and SCORE respectively in Switzerland (Eckardstein, 2014). Based on predefined risk thresholds, a primary care subject is categorized into low, intermediate or high risk. For PROCAM, the categories are 0–9%, 10–19% and 20% or more; for SCORE, the categories are 0.0–0.9%, 1.0–4.9% and 5% or more (Piepoli et al., 2016). A major limitation to calculate global cardiovascular risk is present when single cardiovascular risk factors are high. Therefore, subjects with LDL above 5.0 mmol/l or systolic blood pressure above 160 mmHg are by definition high risk in Switzerland (for SCORE, the high-risk cutoffs are 8.0 mmol/l for total cholesterol and systolic blood pressure of 180 mmHg or more). The SCORE model based the risk algorithm on observations of fatal cardiovascular events in 12 European cohorts undergoing baseline examination between 1967 and 1991 (Conroy et al., 2003). In contrast, PROCAM was derived from working men, later extended to women using observations for myocardial infarction only (Assmann et al., 2007). Therefore, the accuracy of these risk assessment tools may be different in various populations. For the purpose of this study, we assess the agreement for a statin indication in German and Swiss subjects using the PROCAM/AGLA and the SCORE algorithm at various calibration thresholds.