Development and validation of nomograms including individual- and area-level variables to predict risk of fatal and non-fatal cardiovascular diseases among Russian population

Abstract

Introduction Cardiovascular diseases (CVD) are the greatest threat to health worldwide and in Russia. Our study aimed to use Cox proportional hazards models to develop cardio-vascular risk scores and nomograms based on prospective data from studies conducted in Russia. Methods All materials used in this study were obtained from the epidemiological study “Epidemiology of Cardiovascular Diseases in the Regions of the Russian Federation” (ESSE-RF): ESSE-RF (2012-2014) and ESSE-RF2 (2017). A total of 18,454 individuals without CVD aged 25–64 years were included in our study. The participants were randomly divided into a training and testing set at a ratio of 7:3. The Russian deprivation index and its components (social, economic and environmental) were used as area-level predictors. To select the best potential predictive variables for our models, the random forests variable selection algorithm based on minimal depth was used. To predict three- and five-year CVD-free survival, four prognostic nomograms were developed from the results of multivariate analysis. Results The nomograms had considerable discriminative power, calibrating abilities and clinical effectiveness. The time dependent AUC was > 0.7 for the prediction of CVD-free survival in both the training and testing sets. Conclusion For the first time, the nomograms have been created that include area-level predictors (socio-economic and environmental) and lipid spectrum indicators (triglycerides, high-density lipoprotein cholesterol and low-density lipoprotein cholesterol) and assess the probability of fatal and non-fatal cardiovascular events among the Russian population.