This analysis tested the hypothesis that physical activity (PA) improves insulin sensitivity beyond that related to weight loss in adolescents with severe obesity, post-metabolic bariatric surgery (MBS).


We used StepWatchTM activity monitors to objectively measure PA in 88 participants from the Teen Longitudinal Assessment of Bariatric Surgery (Teen-LABS) study. Primary outcomes included absolute change in fasting insulin, HOMA-IR and fasting glucose from baseline to 6, 12, 24, and 36 months. We used SAS PROC TRAJ to generate activity trajectory groups based on probability and individual participant step count trajectories. We used linear regression models with generalized estimating equations for the analysis. Models included adjustments for baseline measurement, visit, surgical procedure, sex, and % change in BMI.Separate models were created adjusting for % change in iliac waist circumference (IWC) or % body fat (BF), measured by bio-impedance, instead of BMI.


PROC TRAJ identified two activity trajectories: more active (MA, n=13) and less active (LA, n=75). The MA trajectory mean daily step count was >6,000, increasing to >9,000 at 2 years. The LA trajectory mean daily step count remained stable near 4,000. No participant had more than 4 minutes of moderate step activity (cadence of 80 steps/minute). Still, fasting insulin and insulin resistance, as HOMA-IR, were found to differ in association with activity trajectory. Fasting glucose did not. MA was associated with greater absolute decrease in fasting insulin (-7.8 uU/mL [95% CI: (-11.8, -3.7)], p =<0.001) and greater decrease in HOMA-IR (-1.9 [95% CI: (-3.0 -0.7)], p = 0.001), when adjusted for % change in BMI. The significant independent effect of activity remained when adjusted for % change in IWC or % BF.


More PA appears to independently enhance improvement in clinical measures of insulin sensitivity beyond that attributable to weight loss or change in % body fat post-MBS in adolescents with severe obesity.