Total carbs weren't linked to diabetes. The kind of carb was.
People say: “How much carb you eat is what counts” We read the study.
Cross-sectional study. 213,704 people. up to 36 years of follow-up (baselines 1984/1986/1991 to 2019/2020); 5,628,955 person-years.
One snapshot in time. Nobody was told what to do. A clue, not proof.
Who: US female nurses (NHS n=74,325; NHSII n=90,403) and male health professionals (HPFS n=44,232), free of diabetes, cardiovascular disease and cancer at baseline; 93-99% white; biomarker subset n=41,687
- Cereal fiber-23%most vs least
- Whole fruit-20%most vs least
- Sugary-drink sugar+22%linked, not proven
More to know, from our read of the paper:
- Observational (linked, not proven): residual confounding possible; authors list observational design as a limitation
- Comparisons are highest vs lowest fifth (quintile) of energy-adjusted intake, not per serving. Cereal fiber Q5 median 7.6-10.3 g/d vs Q1 2.5-3.4 g/d
- Corresponding author: Frank B. Hu, Harvard T.H. Chan School of Public Health, fhu@hsph.harvard.edu (printed correspondence line)
The kind of carb mattered more than amount.
- Whole fruit, not juiceWhole fruit -20%, juice sugar +14%.
- Pick grains with fiberCereal fiber linked to 23% lower risk.
- Skip sugary drinksTheir sugar: 22% higher risk.
Researchers followed 213,704 US nurses and health professionals for up to 36 years and tracked their diets with food questionnaires. Total carbohydrate intake showed no clear link to type 2 diabetes once body weight was accounted for. Eating the most cereal fiber was linked to 23% lower risk and the most whole fruit to 20% lower, while sugar from sugary drinks was linked to 22% higher risk and fruit juice sugar to 14% higher. This is an observational study in mostly white US adults: linked, not proven.
AlEssa HB, Xia T, Mousavi SM, Sawicki C, Willett WC, Bhupathiraju SN, Hu FB (2026). Optimal Measures of Carbohydrate Quality to Lower the Risk of Type 2 Diabetes: Findings from Three Prospective Cohort Studies. The American Journal of Clinical Nutrition.
Design: Cross-sectional study · pooled analysis of three prospective cohorts (Nurses' Health Study I and II, Health Professionals Follow-up Study); diet by repeated food frequency questionnaires, Cox models comparing highest vs lowest fifth of intake (linked, not proven); plus a cross-sectional biomarker sub-analysis
DOI: 10.1016/j.ajcnut.2026.101337
Goals:
We read the abstract and the full text of this paper and checked every number on this page against it (18 facts, checked on 2026-10-29). The study design is named above. Cohort and other observational studies are described as linked, not proven. Found an error? Email theis@onestudyaday.com.
Join the waitlist: one checked study a day for your goal, by email
Pick up to three goals. We email you once when the daily study starts. The goals for this study are ticked already.