Simple Demographic Model to Predict Multiple Cardiometabolic Risk Factors in Two Well-Known Observational Cohorts

Abstract

Listen

Translate article

Objective Determining the prevalence of multiple cardiometabolic risk (CMR) factors requires clinical and laboratory data not readily available to most health-care plans. We evaluated the ability of a simple model derived from the National Health and Nutrition Examination Survey (NHANES) using commonly available demographic information to predict prevalence of multiple CMR factors. Methods We defined CMR factors according to the National Cholesterol Education Program Adult Treatment Panel III (NCEP ATP III) and International Diabetes Federation (IDF) guidelines for metabolic syndrome classification in both the Atherosclerosis Risk in Communities (ARIC) and the Framingham Offspring Study (FOS) cohorts. Using NHANES data, a simple demographic model consisting of sex, race and ethnicity, smoking status, and the natural logarithm of age, generated the coefficients used to predict the prevalence of multiple CMR factors in ARIC and FOS. Predicted prevalences were compared with the observed prevalences in both cohorts. Results The ARIC and FOS cohorts consisted of 11,596 and 3532 subjects with a mean age of 62.6 and 58.8 years, respectively. The observed proportion of participants with metabolic syndrome in ARIC was 52.1% and 58.8% according to the NCEP and IDF definitions, respectively. In FOS the observed prevalence was 41.4% and 45.8% using the NCEP and IDF definitions. Predicted prevalences of metabolic syndrome for the NCEP and IDF definitions, respectively, were 51.3% and 53.5% in ARIC, and 48.2% and 51.4% in FOS. Differences between the observed and predicted prevalences for three of four additional risk factor sets, including abdominal obesity (AO) alone, AO plus diabetes, and AO plus diabetes plus dyslipidemia, were small (between 1 to 7 percentage points) in both cohorts. The model poorly predicted the prevalence of AO plus dyslipidemia. Conclusion A simple demographic model adequately predicted the prevalence of CMR factors. The model should help health-care plans lacking clinical and laboratory data to estimate prevalence of CMR factors in their populations. Future studies need to evaluate the influence of race, sex, and ethnicity on prevalence of these risk factors in various settings.