Elsevier

Preventive Medicine

Volume 49, Issues 2–3, August–September 2009, Pages 148-152
Preventive Medicine

Measurement error and ethnic comparisons of measures of abdominal obesity

https://doi.org/10.1016/j.ypmed.2009.06.023Get rights and content

Abstract

Background

Methods of estimating central obesity are important because of the increasing frequency of obesity related diseases worldwide. Here we evaluate the precision of measuring waist circumference and the waist to hip ratio with comparisons across ethnic groups.

Methods

The third Australian Risk Factor Prevalence Study (1989) of 9279 adults recorded height, weight, and Body Mass Index together with duplicate measurements of the waist and hip circumferences, the waist to hip ratio, and blood pressure levels using clearly defined survey techniques. Measurement error and precision for these variables were calculated, and comparisons were made across ethnic groups.

Results

Coefficients of variation for the waist circumference and the waist to hip ratio were less than 1% indicating good precision in comparison with quite large variability for systolic and diastolic pressure readings. Waist circumference showed increased variability in subjects with larger body build in comparison with waist to hip ratio. Equivalence tests across ethnic groups indicated that the waist to hip ratio was independent of ethnicity.

Conclusion

Waist to hip ratio provides a superior measure of central obesity with low measurement error, high precision, and no bias over a wide range of ethnic groups. We believe that it is essential to standardize methods in the assessment of central obesity. Assessment criteria should be based on waist to hip ratio rather than waist circumference.

Introduction

The rising prevalence of obesity in most nations is causing an increased burden of diabetes, cardiovascular diseases and certain cancers (World Health Organisation, 1998). The measurement of obesity is under review because of increasing evidence that the distribution of body fat and specifically central or abdominal obesity is more important than the degree of excessive weight for a given height, as assessed by the Body Mass Index (BMI). The waist circumference (WC) and its refinement, the waist to hip ratio (WHR), are preferred predictors of the metabolic syndrome (Snijder et al., 2004), type 2 diabetes (Wang et al., 2005) and cardiovascular risk (Megnien et al., 1999, Chen et al., 2007). A very large cross-sectional study of risk factors associated with acute myocardial infarction has confirmed that measures of central obesity and specifically the waist to hip ratio show a much stronger association with coronary heart disease than BMI, increasing the population attributable risk of obesity by more that 3 fold across all ethnic groups (Yusuf et al., 2005). Prospective studies (de Koning et al., 2007, See et al., 2007), including our own (Welborn et al., 2003, Welborn and Dhaliwal, 2007, Dhaliwal and Welborn, 2009), also demonstrate the superiority of WHR and to a lesser extent WC over BMI. The global burden of obesity is considerably underestimated by BMI. But there is a lack of substantive epidemiological data on WC or WHR in ethnic groups, and the technologies are not standardized. Very little has been reported on the precision and reproducibility of the methods used.

This study uses repeated measures of WC and WHR from the Australian Risk Factor Prevalence Study of 1989 to evaluate the precision error of these techniques. The relative utility of WC and WHR will also be assessed in relation with comparisons across ethnic groups and with implications for specific standards in assessing obesity.

Section snippets

Methods

The third Australian Risk Factor Prevalence Study conducted in 1989 by the National Heart Foundation of Australia (Bennett and Magnus, 1994) recruited residents aged 20–69 years who, were registered on the Commonwealth electoral rolls of December 1988 from nine metropolitan cities across Australia: North Sydney, South Sydney, Melbourne, Brisbane, Adelaide, Darwin, Hobart, Perth and Canberra. Subjects were chosen by systematic probability sampling of sex and 5 year age groups. Representative

Measurements

Physical measurements included a single record of height and weight in light summer clothing without shoes. Height was measured to the nearest centimetre and weight to the nearest tenth of a kilogram (1 kg was deducted from actual weight as an allowance for the clothing). The Body Mass Index (BMI) was calculated, being the weight (in kilograms) divided by the height (in metres) squared. Strictly standardized methodologies were followed while measuring waist and hip circumference (Alexander and

Statistical analysis

Measurement error and within subjects % coefficient of variation of the repeated measurements of obesity measures (waist circumference and waist to hip ratio) and blood pressures values were calculated (Bland and Altman, 1986). The measurement error is used to determine the changes that will be considered significant in clinical practice, and is defined by the British Standards Institution as the value below which the absolute difference between two measurements will lie with 95% probability (

Ethical approval

Australian Institute of Health Interim Ethics Committee, following a consultation with the Commonwealth Privacy Commissioner issued the ethics approval for the 1989 survey. The current Ethics Committee of the Australian Institute of Health and Welfare approved the analysis of the survey data.

Results

The baseline characteristics, measures of obesity and country of birth for the 9279 subjects who participated in the 1989 survey are summarized in Table 1. The survey consisted of 4552 males and 4727 females with age range from 20 to 69 years. About 95% of the subjects were born in Australia, United Kingdom and Ireland, Northern Europe, Southern Europe and Asia.

The precision error of waist circumference, hip circumference, waist to hip ratio, systolic and diastolic pressure in both male and

Discussion

This report uses the Australian population data to evaluate the precision error in waist circumference measurements and waist–hip ratio and to compare these measures across various ethnic groups. Our findings confirm waist to hip ratio as a superior measure of central obesity as its measurement error was not associated with the size of the individual as compared to waist circumference. WHR also showed a high degree of precision, similar to that of waist circumference in both males and females.

Conflict of interest statement

The authors have no conflict of interest to declare.

Acknowledgments

The authors are grateful to the National Heart Foundation of Australia for permission to use data from the 1989 Risk Factor Prevalence Survey. We would like to thank Thaila Bejoy for her research and administrative assistance, especially in relation to sourcing the papers and referencing.

References (32)

  • BlandM.

    An Introduction to Medical Statistics

    (2000)
  • BlandJ.M. et al.

    Statistical methods for assessing agreement between two methods of clinical measurement

    Lancet

    (1986)
  • BoyleC.A. et al.

    Waist-to-hip ratios in Australia: a different picture of obesity

    Aust. J. Nutr. Diet.

    (1993)
  • Precision of Test Methods. 1: Guide for the Determination and Reproducibility of a Standard Test Method (BS5497, pt 1)

    (1979)
  • ChenM.M. et al.

    Intraobserver and interobserver reliability of waist circumference and the waist-to-hip ratio

    Obes. Res.

    (2001)
  • ChenL. et al.

    Anthropometric measures and absolute cardiovascular risk estimates in the Australian Diabetes, Obesity and Lifestyle (AusDiab) Study

    Eur. J. Cardiovasc. Prev. Rehab.

    (2007)
  • Cited by (37)

    • Variability of biomarkers used for the classification of metabolic syndrome: A repeated measurements study

      2022, Nutrition, Metabolism and Cardiovascular Diseases
      Citation Excerpt :

      It is more difficult to put a tape properly than measuring weight on a digital scale. In agreement with another study including double measurements of more than 9000 individuals the intra-individual CV was slightly higher for waist than for hip circumference in our study [26]. In comparison to data from the control arm of a study focusing on anthropometric changes in response to exercise, the intra-individual standard deviations for body weight and waist circumference were about half as much in our study [11].

    • Association between body mass index and clinical outcomes of patients after cardiac arrest and resuscitation: A meta-analysis

      2018, American Journal of Emergency Medicine
      Citation Excerpt :

      Notably, BMI value as the only measurement for obesity is insufficient. Waist-to-hip ratio (WHR) and waist circumference (WC) as measures of central obese are better indicators for predicating mortality with low measurement error and high precision [31,32]. A cross section survey for US population revealed that people with central obese had higher mortality risk compared with a man with similar BMI but no central obese, indicating central obesity defined by WHR was more effective to predict clinical outcomes [33].

    • Body mass index and all-cause mortality among type 2 diabetes mellitus patients: Findings from the 5-year follow-up of the MADIABETES cohort

      2017, European Journal of Internal Medicine
      Citation Excerpt :

      Firstly, waist-to-hip ratio and waist circumference are superior to BMI in predicting all-cause mortality [41]. The waist-to-hip ratio, in particular, shows a low measurement error and high precision, and provides unbiased estimates over a wide range of ethnic groups [42]. Secondly, the BMI reference category that has been used in this study (the values of the percentile 5–25) is not the conventional range.

    • The effects of daily consumption of grapefruit on body weight, lipids, and blood pressure in healthy, overweight adults

      2012, Metabolism: Clinical and Experimental
      Citation Excerpt :

      Individuals performing measurements were not blinded to treatment condition, which may have introduced bias, although they were blinded to previous recorded measures. The Australian Risk Factor Prevalence Study, a study of 9279 adults, showed that waist circumference is subject to measurement error of approximately 1.84 cm and concluded that a change greater than 2 cm is statistically significant [39], findings that suggest that the change observed in our population is not due to measurement error alone. In addition, physical activity was not measured in the present study.

    View all citing articles on Scopus
    View full text