Statistical analysis

For descriptive purposes, means and standard deviations of baseline characteristics are shown by sex and tertiles of average daily sedentary time. Body composition metrics at age 16 and 23, and 7-year changes are presented by sex, and paired t tests were used to test for significant changes over time. Multivariable adjusted linear regression models were used to examine the associations between daily sedentary time and mean sedentary bout duration with each body composition metric. All analyses were performed in the overall sample and by sex due to sex-based differences in body composition measures at both study timepoints. Both average daily sedentary time and mean sedentary bout duration were analyzed continuously per standard deviation increments. Cross-sectional associations between sedentary behavior and body composition at age 16 years were conducted, followed by the longitudinal associations between sedentary behavior at age 16 years and body composition at age 23 years. Lastly, longitudinal associations between sedentary behavior at age 16 and change in body composition from age 16 to 23 years (e.g., BMI at age 16 years subtracted from BMI at age 23 years) was examined. Models were adjusted for sex (overall models only), socioeconomic status, dietary nutritional quality, maternal education, past 30-day alcohol use, past 30-day cigarette use, parental history of cardiovascular disease, and average minutes of daily MVPA. Models examining longitudinal changes in body composition further adjusted for the same body composition measure at age 16 years (i.e., models for change in BMI adjusted for BMI at age 16 years). Results from unadjusted models are included in supplementary materials. Although the focus of this analysis was sedentary time, we also conducted a sensitivity analysis testing whether average minutes of daily MVPA was associated with body composition at 16 years. Multiple imputation using chained equations with predictive mean matching was used to account for missing covariate data. Variables with missing values included maternal education (10.8%), past 30-day cigarette use (6.6%), past 30-day alcohol use (5.2%), and parental history of cardiovascular disease (5.2%). All analyses were conducted in SAS ^® Studio (v3.8, SAS Institute Inc., Cary, NC).