Eating healthier was linked to lower teen BMI. Barely.

Cohort study. 22,131 kids and teens. diet and BMI surveys 1996-2001 (GUTS I) and 2004-2011 (GUTS II); up to seven years of data per cohort, periods at least two years apart.

People were watched for years. Nobody was told what to do. A clue, not proof.

Who: US children of Nurses' Health Study II participants from all 50 states, aged 9-14 (GUTS I, 1996) or 10-17 (GUTS II, 2004) at enrolment; analysis limited to ages up to 20; analytic baseline 22,131 (GUTS I 6,621 boys + 7,612 girls; GUTS II 3,743 boys + 4,155 girls); >90% non-Hispanic White; mostly healthy weight (70.6-78.1%)

• Boys: +10 diet points-0.09BMI, tiny change
• Girls: +10 diet points-0.04BMI, tiny change
• Dairy: +1 point+0.02BMI, both sexes

More to know, from our read of the paper:

• Observational (fixed effects remove stable differences between children but not time-varying confounding): 'linked to', never 'causes'
• HEI-2020 is scored per calorie (density): it rates the MIX of foods, not how much is eaten. Authors suggest energy intake may be an important pathway; the study did not test calorie intake
• Not representative: children of nurses, >90% non-Hispanic White, higher diet scores than average US youth (baseline about 57-60 vs 49.6 nationally in 2016)
• BMI from self-reported height and weight; diet from a self-administered food frequency questionnaire
• HEI-2020 (2020-2025 guidelines) applied retrospectively to 1996-2011 data

Researchers followed 22,131 US kids and teens, aged 9 to 17 at the start and mostly white children of nurses, for up to seven years using food questionnaires and self-reported height and weight. When a child's Healthy Eating Index score rose 10 points out of 100, BMI was 0.09 lower in boys and 0.04 lower in girls, which the authors call not clinically significant. The link was stronger in kids with overweight or obesity, but still small. A better dairy score went with slightly higher BMI, and pizza and burritos were among the top dairy sources. Linked, not proven.

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.