Analyzing cross-sectional data from 5,042 U.S. adults in the 2017–2020 NHANES, weighted logistic regression and restricted cubic spline models found that higher physical activity (PA) independently reduces Type 2 diabetes risk — but the magnitude varies sharply by body composition and age. A clear "BMI gradient" emerged: PA's protective effect was strongest in normal-weight individuals, attenuated in the overweight, and statistically absent in those with obesity. An "age stratification" mirrored this pattern, with middle-aged adults gaining the most benefit, young adults showing a borderline trend, and older adults showing no significant association.
This preprint, posted on medRxiv and not yet peer-reviewed, adds nuance to decades of broadly optimistic messaging around exercise and diabetes prevention. The findings suggest that recommending a fixed activity target uniformly across all body types and age groups may overstate real-world benefit for some populations — particularly people with obesity, who face compounding metabolic challenges that PA alone may not overcome. For middle-aged adults at normal weight, the data reinforce exercise as a high-leverage preventive tool. The cross-sectional design is a critical limitation: causality cannot be established, and reverse causation (sicker adults moving less) likely inflates the apparent age-related attenuation. The NHANES sample skews toward community-dwelling adults, reducing generalizability to clinical populations. If confirmed in longitudinal trials, these results would support precision-targeted physical activity prescriptions stratified by BMI and life stage — an incremental but practically meaningful refinement to public health guidance.
