Fig. 3: Pooled odds ratios (95% CI) for the association of total carbohydrates (%E) with likelihood of newly diagnosed T2D and newly diagnosed prediabetes in the ICMR–INDIAB study (n = 18,090).

a, Newly diagnosed T2D. b, Newly diagnosed prediabetes. A fixed-effects model was used to synthesize data from the regions. Two-sided P values were used in calculating ORs and 95% CIs through survey logistic regression. All dietary exposure variables (refined cereals (g d⁻¹), milled whole grains (g d⁻¹), added sugar (g d⁻¹)) were adjusted for total energy using the residual model. Models were adjusted for PSU, state code, population weight, age (in years), sex (male/female), family history of T2D (yes/no), tobacco use (never/current/past), alcohol use (never/current/past), PAL (sedentary/moderate/vigorous), time period the study was conducted, education qualification (primary school/high school/college), BMI (kg m⁻²), place of residence: urban versus rural, intakes of total energy (kcal d⁻¹), fat (%E), fruits (g d⁻¹), vegetables (g d⁻¹), tubers (g d⁻¹), milk and milk products (g d⁻¹), alcohol (g d⁻¹), pulses and legumes (g d⁻¹), added salt (g d⁻¹) and edible oils and fats (g d⁻¹), hypertension, dyslipidaemia and abdominal obesity. Models with food sources as the main exposure were adjusted for the remaining food sources while models with total carbohydrates (%E) were adjusted for total fat (%E). Newly diagnosed T2D was defined as HbA1c ≥6.5% and/or if fasting CBG ≥126 mg dl⁻¹ (≥7 mmol l⁻¹) or 2-h postglucose load CBG ≥220 mg dl⁻¹ (≥12.2 mmol l⁻¹). Newly diagnosed prediabetes was defined as HbA1c 5.7–6.4% or if fasting CBG was ≥100–125 mg dl⁻¹ (≥5.6–6.9 mmol l⁻¹) and/or if 2-h postload CBG value was ≥160 and <220 mg dl⁻¹ (≥8.9 to <12.2 mmol l⁻¹). CBG, capillary blood glucose.