Importance: Assessment of the burden of mortality due to excess body weight in a population and its subgroups is important for designing health policies for interventions. Mendelian randomization (MR) studies can provide an opportunity to correct for unmeasured confounding bias present in observational studies, but such evidence has not been used to assess population burden of mortality due to excess BMI.
Objective: Combine results from a recent Mendelian randomization (MR) study and data from the National Health Surveys to estimate preventable fraction (PF) of 10-year all-cause and cause-specific mortality by different degrees of BMI reduction in the US adult population and underlying risk strata.
Designs: We use cross-sectional data on the distribution of BMI and other risk factors of mortality from the National Health and Nutritional Examination Surveys (NHANES) across two-time spans (1999-2006 and 2017-2018). We use linked data from National Death Index to characterize the observed risk of 10-year mortality associated with BMI and other risk factors based on the NHANES 1999-2006 cohort. We further import results from an external MR study on linear and non-linear effects of BMI and use novel methods to estimate preventable fraction (PF) for deaths under different counterfactual scenarios of BMI reduction in the NHANES population.
Settings: Primary analysis is restricted to the NHANES non-Hispanic white population (age range 40-69 years) due to the unavailability of MR studies in other groups, but projections are provided for the African American population under the assumption of homogeneity of causal effects.
Outcome: Preventable fraction for 10-year all-cause mortality and cause-specific mortality due to 50% and 100% reduction of excess BMI (BMI>25.6 kg/m2) for the US adult population in the age range of 40-69 years.
Results: Nearly 33% and 43% of the NHANES 2017-2018 target population are overweight (25.6 kg/m2≤BMI<30.7 kg/m2) and obese (BMI>30.7 kg/m2), respectively, according to WHO definitions. Estimates of relative risks for different BMI categories (relative to normal BMI) from the external MR study range from 1.05 (25.6 kg/m2 ≤ BMI < 27.8 kg/m2) to 5.95 (BMI> 42.4 kg/m2). We estimate PF for 10-year all-cause mortality due to 50% and 100% reduction of excess BMI for the population to be 24% (95% CI: 14 - 34) and 35% (95% CI: 22-48), respectively. The estimate of PF of death due to heart disease and cancer for this population reaches up to 48% (95% CI: 25-71) and 18% (95% CI: -2-38), respectively. Partitioning of PF shows that 60% of all BMI-attributable deaths arise from only 12% of the population who are at the highest risk due to obesity and a combination of other risk factors.
Conclusions: Nearly one in three deaths in a contemporary US adult population can be attributed to overweight and obesity. A substantial fraction of these deaths are likely to be preventable through pragmatic and targeted BMI interventions.