Sensitivity and Specificity of Body Mass Index as a Marker of Obesity in Elderly Patients on Hemodialysis

J Ren Nutr. 2016 Mar;26(2):65-71. doi: 10.1053/j.jrn.2015.09.001.

Abstract

Objectives: Diagnosing obesity by body mass index (BMI) may not be reliable in elderly individuals due to the changes in body composition. We aimed to analyze the accuracy of BMI thresholds by World Health Organization (WHO) and Nutrition Screening Initiative (NSI) to diagnose obesity in elderly patients on hemodialysis (HD).

Design: Multicenter cross-sectional study.

Setting: Six dialysis facilities.

Subjects: 169 elderly on chronic HD (70.4 ± 7.1 years; 63.9% men).

Main outcome variable: Total body fat percentage (BF%) was assessed by the sum of skinfold thicknesses and abdominal fat by waist circumference (WC). Both were used as reference to test the specificity and sensitivity of BMI thresholds (WHO: ≥30 kg/m(2); NSI: >27 kg/m(2)).

Results: The prevalence of obesity according to NSI-BMI, WHO-BMI, BF%, and WC thresholds were 31%, 13%, 27%, and 29.6% in men, respectively, and 36%, 15%, 13%, and 75% in women. Compared to BF%, the sensitivity of NSI-BMI was moderate (65.5%) for men and high (100%) for women, whereas that of WHO-BMI was low (31%) for men and high (87.5%) for women. Compared with WC, NSI-BMI had good (75%) sensitivity for men and moderate (47.8%) for women, whereas WHO-BMI had moderate (43.8%) sensitivity for men and low (19.6%) for women. The best agreement with BF% was observed for NSI-BMI in men (kappa = 0.46) and for WHO-BMI in women (kappa = 0.80). For WC, the best agreement was for WHO-BMI for men (kappa = 0.63) and NSI-BMI for women (kappa = 0.31).

Conclusions: BMI thresholds do not accurately diagnose adiposity in elderly on HD. Therefore, using BMI may lead to misclassifications in this segment population.

Publication types

  • Observational Study
  • Research Support, Non-U.S. Gov't

MeSH terms

  • Absorptiometry, Photon
  • Adipose Tissue
  • Adiposity
  • Aged
  • Body Composition
  • Body Mass Index*
  • Brazil / epidemiology
  • Cross-Sectional Studies
  • Female
  • Humans
  • Male
  • Middle Aged
  • Obesity / diagnosis*
  • Obesity / epidemiology*
  • Prevalence
  • Renal Dialysis*
  • Sensitivity and Specificity
  • Skinfold Thickness
  • Waist Circumference