Science Progress · Oct 28, 2026
Multisystem assessment of the association between physical activity and biological aging: Evidence from a national cross-sectional survey
By Du M, Zhang W, Xia L et al.
ObjectivePhysical inactivity is a key modifiable risk factor; however, evidence linking it to multidimensional biological aging (BA) remains limited, as most studies have focused on DNA methylation clocks. We therefore integrated four complementary clinical indicators (allostatic load [AL], homeostatic dysregulation [HD], phenotypic age acceleration [PhenoAgeAccel], and Klemera-Doubal method biological age acceleration [KDM Accel]) to assess the association between physical activity (PA) and BA.MethodsThis cross-sectional study was performed based on data from the National Health and Nutrition Examination Survey (NHANES) for the periods 2003-2010 and 2015-2018. Participants were categorized according to self-reported weekly PA volume (MET-min/week) into low (<600), moderate (600-3000), and high (>3000) levels. Associations were assessed using weighted multivariable linear regression, and restricted cubic splines were applied for dose-response analysis.ResultsA total of 13,379 middle-aged and older adults were included. After adjustment for covariates, moderate-level PA was independently associated with lower AL (β=-0.219, 95%CI: -0.308 to -0.130), PhenoAgeAccel (β=-0.524, -0.780 to -0.268), and KDM Accel (β=-0.638, -1.073 to -0.202). In primary fully adjusted models, high-level PA showed no significant associations with any indicator. Dose-response curves revealed nonlinear relations with AL, PhenoAgeAccel, and KDM Accel (all P for nonlinearity <0.001), with the most favorable estimates observed at moderate levels. Subgroup analyses confirmed these patterns and further demonstrated a consistent positive association between high-level PA and KDM Accel across age groups, as well as a reversed association for HD among cancer survivors. For AL, estimates may be conservative due to the inclusion of BMI as a covariate.ConclusionModerate-level PA was associated with more favorable BA indicators relative to low-level PA; however, associations varied across dimensions and subgroups. These findings underscore the multidimensional nature of the relationship between PA and BA and support the need for longitudinal and interventional research in the future.