Measuring the bias, precision, accuracy, and validity of self-reported height and weight in assessing overweight and obesity status among adolescents using a surveillance system.

Adriana Pérez,Kelley Gabriel,Deanna M Hoelscher,Eileen K Nehme,Dorothy J Mandell

doi:10.1186/1479-5868-12-s1-s2

Abstract

BackgroundEvidence regarding bias, precision, and accuracy in adolescent self-reported height and weight across demographic subpopulations is lacking. The bias, precision, and accuracy of adolescent self-reported height and weight across subpopulations were examined using a large, diverse and representative sample of adolescents. A second objective was to develop correction equations for self-reported height and weight to provide more accurate estimates of body mass index (BMI) and weight status.MethodsA total of 24,221 students from 8th and 11th grade in Texas participated in the School Physical Activity and Nutrition (SPAN) surveillance system in years 2000–2002 and 2004–2005. To assess bias, the differences between the self-reported and objective measures, for height and weight were estimated. To assess precision and accuracy, the Lin’s concordance correlation coefficient was used. BMI was estimated for self-reported and objective measures. The prevalence of students’ weight status was estimated using self-reported and objective measures; absolute (bias) and relative error (relative bias) were assessed subsequently. Correction equations for sex and race/ethnicity subpopulations were developed to estimate objective measures of height, weight and BMI from self-reported measures using weighted linear regression. Sensitivity, specificity and positive predictive values of weight status classification using self-reported measures and correction equations are assessed by sex and grade.ResultsStudents in 8th- and 11th-grade overestimated their height from 0.68cm (White girls) to 2.02 cm (African-American boys), and underestimated their weight from 0.4 kg (Hispanic girls) to 0.98 kg (African-American girls). The differences in self-reported versus objectively-measured height and weight resulted in underestimation of BMI ranging from -0.23 kg/m2 (White boys) to -0.7 kg/m2 (African-American girls). The sensitivity of self-reported measures to classify weight status as obese was 70.8% and 81.9% for 8th- and 11th-graders, respectively. These estimates increased when using the correction equations to 77.4% and 84.4% for 8th- and 11th-graders, respectively.ConclusionsWhen direct measurement is not practical, self-reported measurements provide a reliable proxy measure across grade, sex and race/ethnicity subpopulations of adolescents. Correction equations increase the sensitivity of self-report measures to identify prevalence of overall overweight/obesity status.

Highlights

Evidence regarding bias, precision, and accuracy in adolescent self-reported height and weight across demographic subpopulations is lacking
body mass index (BMI) is often a critical variable included in worldwide surveillance systems and interventions to document outcomes of a program or policy, to describe epidemiology of childhood obesity, and/ or to quantify the magnitude of obesity status within and across populations
Some surveillance systems and other population-based studies of children and adolescents, including the National Longitudinal Study of Adolescent Health (U.S.), have incorporated ancillary studies where either all, or a subset of, participants’ heights and weights were directly measured and compared with self-reported estimates to examine validity. These comparison studies have been done in the U.S [5,6,7], Wales [8], Portugal [9], Germany [10,11], and Australia [12]. Results of these studies have shown that, while adolescent-reported estimates of height and weight are correlated with objective measurements, they typically generate a lower estimate of overweight and obesity prevalence [6,7,13,14,15,16]

Summary

Introduction

Precision, and accuracy in adolescent self-reported height and weight across demographic subpopulations is lacking. Some surveillance systems and other population-based studies of children and adolescents, including the National Longitudinal Study of Adolescent Health (U.S.), have incorporated ancillary studies where either all, or a subset of, participants’ heights and weights were directly measured and compared with self-reported estimates to examine validity. These comparison studies have been done in the U.S [5,6,7], Wales [8], Portugal [9], Germany [10,11], and Australia [12]. Results of these studies have shown that, while adolescent-reported estimates of height and weight are correlated with objective measurements, they typically generate a lower estimate of overweight and obesity prevalence [6,7,13,14,15,16]

Methods

Results

Discussion

Conclusion