Reliability of self-reported height and weight measurement in the evaluation of adult overweight and obesity

Introduction: The validity of Body Mass Index (BMI) based on self-reported height andweight is of critical importance in the proper assessment of studies that rely onquestionnaire-derived data. In this study, we evaluate the accuracy of adolescentoverweight and obesity classification bpased on self-reported height and weight factorsand propose a regression model for predicting the true BMI.Methods: This cross-sectional study has been conducted on 1546 (+18 years og age)living in the south of Iran. The participants were chosen using a multistage samplingscheme. We compare the prevalence of overweight and obesity as well as the mean andstandard deviation (SD) of BMI based on self reported and true height and weight. Amultiple regression analysis was carried out to build a regression model to predict the trueBMI based on sex, age group, and self-reported BMI.Results: The overall mean±SD of BMI for self reported and real data were 25.2±1.9and 26.3±2.1, respectively. Estimated prevalence of overweight and obesity were 39.3%and 8.7% based on self reported data, and 60.8% and 15.7% based on the true BMI. Onaverage the true BMI was 1.1 kg/m2 and 1.3 kg/m2 higher for men and women,respectively. Consistently over all age groups and weight classes, BMI values computedusing exact information were larger than BMI values computed using self-reported data.This finding was more evident for female and obese participants. Regression modelingrevealed that sex, age group, and self reported BMI are the most reliable factors forpredicting the true BMI.Conclusion: In etiological studies and studies that analyze the relation between variousdiseases and obesity and overweight, it is preferable to use true values of height andweight. However, in the absence of true data, the suggested regression model can be usedto predict the true BMI with a narrow 95% confidence interval and a desirable 95%prediction interval.

کلیدواژه ها:

Body Mass Index ، Prevalence ، Obesity ، Overweight ، Regression Model