Adjustment for obesity in studies of cardiovascular disease

To evaluate the validity and implications of using various obesity indices in adjusting or controlling for obesity, correlations were analyzed between six cardiovascular disease risk factors, age, cholesterol, log triglyceride, systolic blood pressure, diastolic blood pressure, and fasting plasma glucose, and weight (W), height (H), and five commonly used obesity indices, W/H, W/H2, 3 square root W/H, -H/3 square root W, and relative weight in a defined population of 4956 men and women. Subjects were residents of Rancho Bernardo, California and were surveyed in 1972-1974. Correlations of weight, height, and the obesity indices were also evaluated with each other. W/H, 3 square root W/H, and -H/3 square root W were highly correlated with weight (correlations = 0.96-0.997), but also correlated with height (correlations = 0.10-0.48). These very high correlations with weight and moderate correlations with height resulted in risk factor correlations with these three indices reflecting weight-risk factor correlations rather than obesity-risk factor correlations. W/H2 and relative weight were not quite so highly correlated with weight (correlations = 0.83-0.89), very highly correlated with each other (correlations = 0.999), and relatively uncorrelated with height (correlations = 0.17-0.01). W/H2 and relative weight risk factor correlations reflected true obesity-risk factor correlations and were significantly greater than risk factor correlations with weight, W/H, 3 square root W/H, and -H/3 square root W. These data strongly support the use of either W/H2 or relative weight for obesity adjustment in cardiovascular disease studies. Use of W/H, 3 square root W/H, or -H/3 square root W may result in underadjustment for obesity.