Beyond the question of accuracy in measurement, the use of the BMI as predictor of various health risks also depends on the health risks at issue. For instance, some researchers [25] have shown that the BMI is just as good a predictor of the incidence of diabetes as a waist-hip ratio or waist circumference. By contrast, others [26] reported that the BMI performed somewhat less well than waist circumference as a measure of risk for cardiovascular diseases. Still, given the costs of data collection and the ability to reduce, even though not eliminate, systematic biases through adjustments using a few demographic variables, BMI values based on self-reported height and weight remain an important tool for population-based estimates of the health risks involved in obesity.
