Obesity is an established risk factor for hypertension in high income countries, but its importance is less established in low- to middle-income countries, particularly rurally. Many established methods for measuring adiposity are not feasible in rural settings, so it is important to establish the reliability of feasible methods in these settings. We aimed to test the validity of various methods for assessing body fat in a cohort of rural Indians living in poverty.Design and Method:
In 1026 adults in rural Andhra Pradesh, we measured blood pressure according to a strict protocol. Body fat was measured using skinfold thickness, transformed using age-specific and general equations, and electrical bio-impedance analysis (BIA). Agreement between the techniques of BIA and skinfolds was assessed using the Bland-Altman analysis, while the presence and nature of bias was assessed using ordinary least products (OLP) regression.Results:
There was good agreement between methods, with BIA estimating 5.6% greater body fat than the age-specific equation. More people were categorized as obese using BIA (39% men, 9% women) than via skinfold method either using the general equation (3.9% men, 0.7% women) or the age-specific equation (5.2% men, 1.2% women). The slope of the OLP regression between percent body fat generated by skinfold (age-specific equation) and BIA was 1.00 (y-intercept = 5.65, r2 = 0.47). Similar results were obtained using the general equation (y-intercept = 7.45, r2 = 0.32). Importantly, no proportional bias was detected between the methods. BIA displayed a similar association with hypertension (Odds Ratio (OR) 1.09, 95% Confidence Interval (95%CI) 1.05–1.13) to the skinfold technique (OR 1.04, 95%CI 1.02–1.07).Conclusions:
The ability to assess body fat accurately is important for determining the risk of hypertension. Given the relative ease by which BIA measurements are made we recommend BIA as a valid method for determining percentage body fat in population-based studies of hypertension in rural settings.