Introduction: Shifting the population distribution of common risk factors has considerable potential to reduce disease burden. Peripheral artery disease (PAD) is a disabling, often life-threatening condition affecting 8.5 million U.S. adults, yet the potential impact of feasible shifts in the population distribution of adiposity on the associated burden of PAD has not been reported.
Methods: From the population-based, biracial ARIC cohort, 13,604 individuals (54% female; 27% African American; and mean age: 54 years) were examined after excluding 969 participants who at baseline had chronic conditions associated with weight change and 632 with prevalent PAD. Exposure [body mass index (BMI)] and covariates (smoking, hypertension, and diabetes) were ascertained at each of 4 triennial study visits. Incident PAD events were identified from active surveillance of hospitalizations and ICD-9-CM discharge codes. Diabetes and hypertension are time-varying covariates that may lie on the causal pathway between BMI and PAD, and are in turn affected by previous BMI. Using the parametric g-formula, we assessed the hypothesis that a reduction in the cumulative incidence of PAD would be observed following a hypothetical 5% yearly BMI reduction down to 24 kg/m2 for all individuals under 65 years of age with a BMI> 24 kg/m2. This approach allows for control of time-varying confounding by covariates that may act as both mediators and confounders, provided that we have longitudinal measures of those covariates. Participants were followed until the first incident PAD event (2%), study dropout (27%), death (7%), or end of follow-up (64%).
Results: During a median 12 years of follow-up, we identified 231 participants with incident PAD (31% female; 17% African American; 51% smokers; 45% with diabetes; and 77% with hypertension). The 12-year cumulative incidence of PAD was 1.89% (95%CI: 1.62, 2.17%) under the natural course, and 1.72% (95%CI: 1.46, 2.08%) following a 5% yearly reduction in BMI. Thus, we predicted a -0.17% (95%CI: -0.38, 0.13) change in the risk of PAD. The cumulative incidence of PAD following hypothetical shifts of various magnitude down to a BMI of 24 kg/m2 followed an expected dose response curve, although all estimates were within the 95% confidence limits of our study’s estimated cumulative incidence.
Conclusion: We observed an estimated reduction of small magnitude in the risk of PAD attributed to a feasible shift in the population distribution of BMI, consistent with the larger impact of PAD risk factors that are not influenced by BMI, such as cigarette smoking and age. There is need for characterization of the effect of reducing mid-life BMI on PAD with extended follow-up time, and to older ages when PAD risk is highest. Our results suggest that 9% of PAD cases occurring in this study population over the 12 years of follow-up could have been prevented by a 5% shift in the population distribution of BMI.