Background: Evidence on effects of carbohydrate intake with cardiovascular disease risk factors is inconsistent and has limited generalizability to diverse populations such as South Asians, whose carbohydrate intakes are typically high. Aims: We examined the association of the quantity and types of carbohydrate intake with two risk factors for major cardiovascular events– hypertension (HT) and hypercholesterolemia (HC). Methods: We analyzed cross-sectional data from participants aged ≥18 years in the South Asia Biobank in Bangladesh, India, Pakistan, and Sri Lanka. Using dietary data assessed with interviewer-led 24h diet recalls, we calculated intakes of total and subtypes of carbohydrates and several indices of carbohydrate quality. Primary outcomes were HT and HC, defined by medication use, doctor diagnosis, blood pressure and blood lipid levels. With mixed effects Poisson regression, HT and HC was regressed on quantiles of carbohydrate intakes by country to estimate prevalence ratios (PR) and 95% confidence intervals (CI), followed by meta-analysis pooling country-specific estimates. We examined interactions by diabetes status and adiposity. Results: Of 56,024 adults, 36.6% had HT, 16.1% had HC. In pooled multivariable analyses comparing extreme quantiles, total carbohydrate intake was not associated with HT or HC prevalence. Fructose, free sugar and free sugar-to-fiber ratio were associated with higher prevalence of HT (PRs Q5vsQ1 : 1.03-1.15). Higher intakes of cereal fiber (PR Q4vsQ1 : 0.96, 95% CI: 0.95-0.98) and legume fiber (PR Q4vsQ1 : 0.98, 95% CI: 0.97-0.98) were associated with lower prevalence of HT and higher cereal fiber intake with lower HC prevalence (PR Q5vsQ1 : 0.80, 95% CI: 0.75-0.85). Diabetes and adiposity were observed as effect-modifiers (p<0.001). There were null associations in adults with diabetes but in non-diabetic adults, free sugar intake was related positively with HT (PR Q5vsQ1 : 1.07, 95% CI: 1.05-1.09) while fiber intake was inversely related (PR Q5vsQ1 : 0.98, 0.95-0.97). Starch, sugar and free sugar intake were associated positively with HT prevalence in adults with high body fat (≥20% for men; ≥30% for women) and visceral fat (≥10%) levels (PRs Q5vsQ1 : 1.03-1.20); associations were null for low body and visceral fat levels. Associations varied by country (between-country heterogeneity I 2 >60%). Conclusion: Associations of dietary carbohydrates with HT and HC depended on the quality and type of intake in South Asians.
Read full abstract