Errors in body mass index from self-reported data by sex and across waves of Add Health

Purpose: Body mass index (BMI) derived from self-reported height and weight is often used to study adiposity and its health implications. However, misestimates of BMI from self-reported data have been observed. This study adds to the literature by demonstrating how anthropometric misreporting patterns differed by sex and changed across time in a nationally representative cohort, as well as examining behavioral/psychological correlates of biases in BMI. Methods: Misreporting of height and weight (and thus BMI) from adolescence to adulthood in the United States was studied using the National Longitudinal Study of Adolescent to Adult Health (1996-2008). Behavioral/psychological characteristics possibly associated with errors in BMI were analyzed with fixed-effects models. Results: Different patterns of anthropometric misreporting resulted in larger underestimation of BMI among females than males at the beginning waves, but females saw a reduction by the last wave. Males did not see such a decrease, and their error, at 0.75 BMI units by 2008, was comparable to that of females. For both sexes, body image perception was a significant predictor of biases in BMI. Conclusions: From adolescence to adulthood, anthropometric reporting patterns changed, and its variation differed by sex. Nevertheless, errors in BMI were similarly associated with behavioral/psychological characteristics. (C) 2019 Elsevier Inc. All rights reserved.