Abstract
Introduction: Cardiovascular disease (CVD) may prevent the adoption of healthy lifestyle habits such as physical activity, and thus may lead to weight gain and increased risk for lifestyle-related conditions such as diabetes mellitus (DM). Evidence suggests that medications commonly used in the management of CVD such as diuretics, β blockers, and statins are also associated with increased risk of diabetes. We examined the association of CVD with incident DM and investigated whether this association is mediated by change in BMI or medication use. Methods: Data from the HCHS/SOL Visit 1 (2008-2011) and Visit 2 (2014-2017) examinations were used to compare incidence of DM among persons with and without CVD at Visit 1. DM was defined using the American Diabetes Association criteria, and prevalent CVD using the Framingham Heart Study definition (i.e., self-reported history of coronary heart disease, cerebrovascular events, peripheral artery disease, or heart failure). Of the 8672 participants free of DM at Visit 1 and with complete data on DM at Visit 2 and on key covariates of interest, 2191 (25.3%) had prevalent CVD at baseline. A total of 2050 participants with CVD were matched to controls free of CVD at Visit 1 using 1:1 propensity matching. Covariates included in the propensity model were age, gender, educational attainment, marital status, Hispanic background, history of hypertension, family history of diabetes, health insurance coverage, chronic kidney disease at baseline, gestational diabetes, high cholesterol, metabolic syndrome, alcohol use, cigarette pack years, physical activity level, diet quality based on alternative healthy eating index, baseline BMI, baseline prediabetes. Matched pairs were analyzed. McNemar’s test was used to compare incidence of DM among cases and controls. The mediating effects of change in BMI and medication use were examined. Results: Covariate distributions were similar among participants with and without CVD. The incidence of DM among persons with CVD was 15.7% vs. 13.6% among those without CVD (p=0.06). No natural direct effect of CVD on DM was found. The association between CVD and incidence of DM was not mediated by change in BMI. On average, BMI increased by 0.38 (SE=2.56) and 0.33 (SE=2.50) among persons with and without CVD, respectively (p=0.55). The association between CVD and DM was indirectly mediated by the use of beta-blockers (percent mediated=24.8%), statins (percent mediated=9.8%), and diuretics (percent mediated=8.9%). Conclusion: There is slight evidence that medications used to treat CVD increase risk for new onset diabetes.
