Food-based healthy diet metrics, type 2 diabetes and insulin resistance in Brazilian Longitudinal Study of Adults Health (ELSA-Brasil): a prospective study of over 12 000 Brazilians.

The Global Diet Quality Score (GDQS), the Global Dietary Recommendations score (GDR), Nova ultra-processed foods score and Minimum Diet Diversity for Women were listed by the UN as promising food-based metrics to track dietary quality. The aim was to evaluate those diet metrics performances for monitoring diet-related type 2 diabetes risk and insulin resistance in comparison with the Alternative Healthy Eating Index (AHEI). The study included 12 254 participants (56 % women; 35–74 years) of the Brazilian Longitudinal Study of Adults Health (ELSA-Brasil) with available dietary, biochemical, socio-demographic and lifestyle data. Diet quality scores were derived from a validated FFQ covering the previous 12 months. Incident diabetes and changes in insulin resistance (HOMA-IR) over time were evaluated, and mediation by BMI was assessed. After an average 8·2 years follow-up period (17 % attrition; n 10 191), with 1-standard deviation increase in GDQS, GDR and AHEI, a 5 %, 6 % and 7 % decrease in incident diabetes was observed with BMI mediation effect of 39·6 %, 74·8 % and 59·4 %, respectively (P < 0·001 for all the analysis). HOMA-IR increase rate over time was higher in the lower quintile of GDQS (P = 0·002) and GDR (P < 0·001), compared with the upper quintile. As AHEI, GDR and GDQS had similar performances in monitoring diet-related type 2 diabetes risk, food-based metrics, such as the GDR and the GDQS, can be advantageous in lower resources settings and in nations where there is no food composition data availability.