Heterogeneous associations between domain-specific physical activity and sarcopenia risk in diabetic adults: A causal forest analysis.

BackgroundModerate-to-vigorous physical activity has been associated with lower sarcopenia prevalence, whereas sedentary behavior may be associated with elevated risk. However, evidence remains limited regarding whether these associations persist in diabetic populations, whether dose-response relationships are maintained, and which subgroups exhibit the most pronounced associations with sarcopenia risk.MethodsCross-sectional associations between domain-specific moderate-to-vigorous physical activity, sedentary behavior, and sarcopenia risk were examined using progressive multivariable linear and logistic regression, with continuous skeletal muscle mass index scores and binary sarcopenia status as outcomes. Restricted cubic spline analyses characterized dose-response relationships. Stratified subgroup analyses with interaction terms assessed consistency across age, sex, body mass index, and race/ethnicity. Causal forest algorithms estimated individual-level heterogeneity in association magnitude, reported as conditional average association effects rather than treatment effects.ResultsBased on 1230 participants, sarcopenia prevalence was 18.46%, with notably higher rates observed among individuals with obesity, those of Mexican American ethnicity, and those with lower educational attainment. Following full covariate adjustment, both occupational moderate-to-vigorous physical activity (β = -0.020, 95% confidence interval: -0.034 to -0.006, p = 0.007) and leisure-time moderate-to-vigorous physical activity (β = -0.018, 95% confidence interval: -0.032 to -0.003, p = 0.018) demonstrated independent inverse associations with sarcopenia risk, where β denotes multivariable linear regression coefficients for the continuous skeletal muscle mass index outcome (negative values indicating lower skeletal muscle mass index and higher sarcopenia risk). Dose-response analyses further identified activity thresholds beyond which progressively more favorable associations were apparent. Causal forest estimation indicated that total physical activity exhibited the most pronounced inverse association with sarcopenia risk (mean conditional average association effect: -0.0547). Exploratory subgroup analyses suggested potential heterogeneity in association magnitude, with point estimates directionally consistent with more pronounced inverse associations among adults aged ≥51 years, females, and individuals with obesity. These patterns are preliminary and would benefit from prospective verification before more definitive conclusions are drawn.ConclusionsIn this cross-sectional, observational analysis, physical activity was inversely associated with sarcopenia risk across diverse subpopulations of adults with diabetes. Exploratory causal forest analyses suggested potentially more pronounced inverse associations among older adults, females, and individuals with obesity, although these patterns should be regarded as hypothesis-generating because the cross-sectional design precludes any inference about temporality or causality. Future longitudinal studies and randomized controlled trials are needed to verify these findings.