What 120,877 people’s diets looked like over four-year periods — food by food.
What the evidence says
- Across 120,877 adults in three cohorts, average weight gain was 3.35 lb per four-year period — with a 5th-to-95th-percentile spread of −4.1 to 12.4 lb, so the mean conceals a range in which many participants lost weight[1]
- Foods associated with GAIN, per daily serving increase: potato chips +1.69 lb · potatoes +1.28 lb · sugar-sweetened beverages +1.00 lb · unprocessed red meat +0.95 lb[1]
- Foods inversely associated with gain, per daily serving increase: yogurt −0.82 lb · nuts −0.57 lb · fruits −0.49 lb · whole grains −0.37 lb · vegetables −0.22 lb[1]
- Yogurt showed the strongest inverse association of any food tested — ahead of nuts, fruit, whole grains and vegetables[1]
- Lifestyle factors were independently associated: physical activity −1.76 lb across quintiles (the spread over the activity distribution, not a per-unit figure) · alcohol +0.41 lb per drink per day · television +0.31 lb per hour per day · sleep, with gain at both under 6 and over 8 hours a night[1]
- The authors’ framing: specific dietary and lifestyle factors are independently associated with long-term weight change — and they explain the inverse associations by displacement, not by diet quality beating calories: such foods ”cannot violate thermodynamic laws”, and increasing them ”reduced the intake of other foods to a greater (caloric) extent”[1]