Predictive performance of anthropometric indexes of central obesity for the risk of type 2 diabetes

Arch Med Res. 2005 Sep-Oct;36(5):581-9. doi: 10.1016/j.arcmed.2005.03.049.

Abstract

Background: In spite of several available anthropometric indexes, the relative merit of these indexes for the prediction of type 2 diabetes remains unknown. Considering that obesity and diabetes commonly coexist as co-morbidities, our objective was to directly compare the performance of measures of central and general obesity to predict the risk of type 2 diabetes.

Methods: We conducted a case-control study of type 2 diabetes on 150 cases and 150 age- and gender-matched controls. We directly compared the predictive performance of five anthropometric indexes: four related to central obesity--waist circumference (WC), waist/hip ratio (WHR), abdominal volume index (AVI) and conicity index (CI); and one related to general obesity--body mass index (BMI). We used various statistical approaches like area under (AUC) receiver-operating characteristic (ROC) curves, likelihood ratios, logistic regression and Shannon's entropy to compare the performance of the indexes in the study sample as well as bootstrapped samples.

Results: WC had the highest overall predictive accuracy that was gender insensitive (AUC=0.77 in males and 0.74 in females); a comparable information content as that of AVI (Shannon's entropy=1.81 for WC and 1.84 for AVI) and was a better predictor of the risk of type 2 diabetes than all the remaining indexes. WC also correlated strongly with the biochemical markers of diabetes like blood sugar and lipid profile.

Conclusions: WC is a simple, non-invasive and accurate predictor of the risk of type 2 diabetes that can potentially be used in screening programs in developing countries.

MeSH terms

  • Adult
  • Aged
  • Anthropometry*
  • Blood Glucose / metabolism
  • Case-Control Studies
  • Comorbidity
  • Diabetes Mellitus, Type 2 / blood
  • Diabetes Mellitus, Type 2 / etiology*
  • Diabetes Mellitus, Type 2 / physiopathology
  • Female
  • Humans
  • Male
  • Middle Aged
  • Obesity / complications*
  • Predictive Value of Tests
  • ROC Curve
  • Regression Analysis
  • Risk Factors
  • Sex Factors
  • Statistics as Topic

Substances

  • Blood Glucose