Association between carbohydrate quality index and cardiovascular disease: Tehran Lipid and Glucose Study

While emerging evidence from Western populations suggests that higher carbohydrate quality (assessed by indices incorporating fiber, glycemic index, whole grains, and solid carbohydrate sources) is inversely associated with cardiovascular risk factors, prospective evidence from Middle Eastern populations regarding the association between carbohydrate quality and cardiovascular disease (CVD) remains limited and inconsistent. We included 2,648 participants free of CVD, from the participants of the 3th phase of Tehran Lipid and Glucose Study (TLGS). Dietary intake was assessed at baseline (2006-2008) using a validated 168-item food frequency questionnaire. Carbohydrate quality index (CQI) was calculated based on four components: dietary fiber intake, dietary glycemic index, whole grain to total grain ratio, and solid carbohydrate to total carbohydrate ratio. The CQI and its components were calculated and categorized into tertiles. Incident CVD was defined as coronary heart disease, stroke, or cardiovascular death occurring during follow-up. Cox proportional hazards models were used to estimate hazard ratios (HRs) and 95% confidence intervals (CIs), with sequential adjustment for demographics, lifestyle factors, dietary factors, and potential mediators (BMI, diabetes, hypertension). Sensitivity analyses were conducted by (1) omitting the whole grain ratio from the CQI to yield a 3-component Modified CQI, and (2) constructing a Population-Adapted Index (PAI) replacing whole grain with legume intake. During a median follow-up of 10.6 years, 171 CVD cases (6.5%) were identified. After multivariable adjustment, no significant association was observed between CQI and incident CVD (T3 vs. T1: HR = 0.83, 95% CI: 0.57-1.21). Individual CQI components also showed null associations. Sensitivity analyses using a Modified CQI (without whole grain) and a Population-Adapted Index (incorporating legume intake) yielded consistent null associations (Modified CQI T3 vs. T1: HR = 0.66, 95% CI: 0.39-1.13; PAI T3 vs. T1: HR = 0.93, 95% CI: 0.57-1.50, fully adjusted). In this prospective study of Iranian adults, we observed no statistically significant association between CQI and incident CVD. However, statistical power was limited for detecting subtle associations (MDHR = 0.801 per SD at 80% power), and larger prospective studies are needed.