Marshall J D, Hazlett C B, Spady D W, Conger P R, Quinney H A
Department of Health Services Administration and Community Medicine, University of Alberta, Edmonton, Canada.
Hum Biol. 1991 Apr;63(2):137-53.
A validation study of convenient indicators of obesity was undertaken in 540 male and female subjects, aged 7-14 yr. Four adiposity measures that have commonly accepted obesity classification points [relative weight, relative body mass index (BMI), sum of five skinfolds, and triceps skinfold] were derived from measures of height, weight, and five skinfold thickness measurements. Body density measures were converted to percentage of body fat using Lohman's (1986) age- and gender-specific regression equations. Using greater than or equal to 20% body fat for males and greater than or equal to 25% for females as the standard for obesity, the diagnostic utilities (sensitivity, specificity, overall accuracy, and positive and negative predictive values) of the four obesity indicators at their commonly used obesity cutoff points were determined. Preliminary analyses demonstrate that use of these indicators should not be considered independent of the gender of the subject or without reference to the purpose for classifying subjects as obese. Secondary analyses, in which the obesity cutoff point in each indicator was altered to obtain a minimum specificity level of 95%, demonstrated that a sum of skinfolds was better at identifying true obesity than all other indicators in both males and females. There is potential for inappropriate labeling with all convenient indicators of obesity, and thus they should be used with caution.
