Comparing the predictive ability of the Edmonton Obesity Staging System with the body mass index for use of health services and pharmacotherapies in Australian adults : a nationally representative cross-sectional study

We assessed the value of the Edmonton Obesity Staging System (EOSS) compared with the body mass index (BMI) for determining associations with use of health services and pharmacotherapies in a nationally representative sample of participants in the 2011–2013 Australian Health Survey. A subsample of participants aged 18 years or over, with at least overweight (BMI ≥ 25 kg/m2) or central obesity (waist measurement of ≥102 cm for men; ≥88 cm for women), and who had provided physical measurements (n = 9730) were selected for analysis. For statistical significance of each predictor, we used logistic regression for model comparisons with the BMI and EOSS separately, and adjusted for covariates. For relative explanatory ability, we used the Nagelkerke pseudo R 2, receiver operating characteristic curve, and area under curve statistic. The EOSS was significantly better than the BMI for predicting polypharmacy and most of the health service use variables. Conversely, the BMI was significantly better than the EOSS for predicting having discussed lifestyle changes relevant to weight loss with the primary care physician. Clinicians, health care professionals, consumers, and policy makers should consider the EOSS a more accurate predictor of polypharmacy and health service use than the BMI in adults with overweight or obesity.