Introduction: The rise in T1D, T2D, and obesity make it increasingly important to understand the role of BMI and lifestyle on glycemic control and IR. While it is known that weight and exercise affect these, the extent to which the effects of exercise are direct versus mediated by BMI has not been determined.
Methods: We analyzed data from the 2013-2016 NHANES dataset in participants >18 yrs of age (n=1168). Those on glucose-lowering medications were classified as T2D and those on insulin only as T1D. Participant reports on physical activity intensity and duration were converted to metabolic equivalent of task (MET)-hours/week. HOMA-IR, a measure of IR, was calculated in participants with insulin measurements not receiving insulin (n=335). Linear regression models were fitted for mediation analyses to evaluate whether BMI mediated associations between MET and HbA1c or HOMA-IR.
Results: In participants with T1D (n=173) and T2D (n=995), the mean ± SE age (56.8 ± 0.8 and 61.1 ± 0.3 yrs), BMI (32.8 ± 0.8 and 33.1 ± 0.3), and MET-hours/week (38.8 ± 4.5 and 38.3 ± 3.8) were reported, respectively. Further, 48.7% and 46.3% were females and 51.3% and 53.7% were males, respectively. The majority (62.9%) were Non-Hispanic White. HbA1c was significantly different between participants with T1D (8.1 ± 0.2) and T2D (7.4 ± 0.1), p=0.0008. After adjusting for age, race, and sex, the independent effect of MET-hours/week on HbA1c was nonsignificant in both cohorts (p=0.384 and p=0.167, respectively), suggesting BMI mediation analysis was not appropriate. The same was true with the effect of MET-hours/week on HOMA-IR (p=0.265).
Conclusions: Our results show that there were no BMI mediated associations between MET and HbA1c or HOMA-IR. However, participants with high HbA1c could be actively engaging in more physical activity to improve glycemic control and there is potential for recall bias. A future prospective study with larger sample sizes may be needed.
P. Y. Soni: None. Q. Tang: None. H. M. Ismail: None.
