Introduction:
Estimating effects of population-level increases in obesity on diabetes has important implications for public health policy. Therefore, we used the population attributable fraction (PAF) to quantify trends in the burden of obesity on incident diabetes in key sex and race/ethnicity subgroups over time.
Hypothesis:
PAFs for obesity attributable incident diabetes will be greatest for women, differ by race/ethnicity and increase over time.
Methods:
Using data from the Multi-Ethnic Study of Atherosclerosis (MESA), we estimated unadjusted and adjusted hazard ratios (HRs) of obesity on incident diabetes mellitus (DM). We included non-Hispanic White (NHW), non-Hispanic Black (NHB) and Hispanic-Mexican (MA) MESA participants with available data on body mass index and key covariates and without DM at baseline. Next, we estimated the prevalence of obesity in four pooled groups of NHANES survey cycles from 2001-2016 with similar characteristics to MESA participants (age 45-80 years, NHW, NHB, and MA without cardiovascular disease). Lastly, we combined estimates from MESA and NHANES to quantify unadjusted and adjusted PAFs (adjusted for age, income, education, physical activity and diet). Leveraging the objective longitudinal assessment of incident DM in MESA and a nationally representative sample of NHANES, we estimated broadly generalizable PAF estimates over time.
Results:
Of 3869 MESA participants, mean age was 61 (± 10) years with 47% men, 56% NHW, 31% NHB, and 13% MA. Incidence of DM was 11% with overall adjusted HR of 2.75 (95% CI 2.26, 3.34) for obesity. Prevalence of obesity increased from 34% (32, 37) in 2001-2004 to 41% (39, 44) in 2013-2016. Overall adjusted PAFs ranged from 0.23 to 0.52 over the study period (TABLE), with greatest estimates in NHW women.
Conclusions:
In conclusion, over time, the proportion of diabetes due to obesity has increased across multiple subgroups of the population, further highlighting the health burden of obesity in the population.