The associations of physical activity patterns and the triglyceride-glucose index in US adults: a secondary data analysis of NHANES (2007-2018)

The triglyceride-glucose (TyG) index, a readily accessible surrogate biomarker for insulin resistance and metabolic health, has been extensively investigated across various ethnic populations. The critical role of physical activity (PA) in enhancing insulin sensitivity has been corroborated by numerous studies. Nonetheless, the relationship between different patterns of PA and the TyG index remains ambiguous. This study seeks to elucidate the association between different PA patterns and the TyG index. Participants with complete data on TyG index, PA patterns, and other covariates from the National Health and Nutrition Examination Survey database (2007-2018) were included in this study. Participant characteristics across PA groups were compared using Rao-Scott chi-squared test or Kruskal-Wallis test. Multivariate linear regression models were used to assess the relationship between PA patterns and the TyG index, with subgroup analyses, interaction tests and restricted cubic spline (RCS) regression analyses conducted to explore the stability and potential nonlinearity of PA-TyG index association, respectively. A total of 16,440 participants were included, with 12,909, 164, 118, and 3249 individuals being inactive, insufficiently active, weekend warriors (WWs), and regularly active (RA), respectively. Multivariate adjusted linear regression models showed that insufficiently active adults (β = - 0.05, P = 0.3524), WWs (β = -0.01, P = 0.9129), and RA adults (β = - 0.110, P < 0.0001) all had significantly lower TyG indices compared to inactive adults. Significant difference was observed between WWs and RA adults (P for trend < 0.0001). Stratified analyses and interact tests revealed that the PA-TyG index association in RA group was more pronounced in individuals being male, with higher education levels, and without diabetes. Additionally, a significant, non-linear, and negative relationship between weekly total PA and TyG index in non-inactive individuals was determined by RCS analysis (P for overall < 0.001, P for nonlinearity = 0.054). Being regularly active is linked to a substantial reduction in the TyG index, whereas weekend warriors and those with insufficient activity do not experience comparable benefits. Nevertheless, the weekend warrior approach can provide a practical alternative for individuals who are unable to engage in consistent exercise.