Comparison of relationships between four common anthropometric measures and incident diabetes

Crystal Man Ying Lee, Mark Woodward, Nirmala Pandeya, Robert Adams, Elizabeth Barrett-Connor, Edward J. Boyko, Mats Eliasson, Laercio J. Franco, Wilfred Y. Fujimoto, Clicerio Gonzalez, Barbara V. Howard, David R. Jacobs, Sirkka Keinanen-Kiukaanniemi, Dianna Magliano, Pamela Schreiner, Jonathan E. Shaw, June Stevens, Anne Taylor, Jaakko Tuomilehto, Lynne WagenknechtRachel R. Huxley, The Obesity, Diabetes and Cardiovascular Disease Collaboration

Research output: Contribution to journalArticlepeer-review

20 Scopus citations


Aims First, to conduct a detailed exploration of the prospective relations between four commonly used anthropometric measures with incident diabetes and to examine their consistency across different population subgroups. Second, to compare the ability of each of the measures to predict five-year risk of diabetes. Methods We conducted a meta-analysis of individual participant data on body mass index (BMI), waist circumference (WC), waist-hip and waist-height ratio (WHtR) from the Obesity, Diabetes and Cardiovascular Disease Collaboration. Cox proportional hazard models were used to estimate the association between a one standard deviation increment in each anthropometric measure and incident diabetes. Harrell's concordance statistic was used to test the predictive accuracy of each measure for diabetes risk at five years. Results Twenty-one studies with 154,998 participants and 9342 cases of incident diabetes were available. Each of the measures had a positive association with incident diabetes. A one standard deviation increment in each of the measures was associated with 64–80% higher diabetes risk. WC and WHtR more strongly associated with risk than BMI (ratio of hazard ratios: 0.95 [0.92,0.99] – 0.97 [0.95,0.98]) but there was no appreciable difference between the four measures in the predictive accuracy for diabetes at five years. Conclusions Despite suggestions that abdominal measures of obesity have stronger associations with incident diabetes and better predictive accuracy than BMI, we found no overall advantage in any one measure at discriminating the risk of developing diabetes. Any of these measures would suffice to assist in primary diabetes prevention efforts.

Original languageEnglish (US)
Pages (from-to)36-44
Number of pages9
JournalDiabetes Research and Clinical Practice
StatePublished - Oct 2017

Bibliographical note

Funding Information:
This work was supported by the National Health and Medical Research Council of Australia [grant number 1103242].

Publisher Copyright:
© 2017 Elsevier B.V.


  • Body mass index
  • Diabetes
  • Waist circumference


Dive into the research topics of 'Comparison of relationships between four common anthropometric measures and incident diabetes'. Together they form a unique fingerprint.

Cite this