Objective: To examine the accuracy of self-reported height and weight data to classify adolescent overweight status. Self-reported height and weight are commonly used with minimal consideration of accuracy.
Data sources: Eleven studies (4 nationally representative, 7 convenience sample or locally based).
Study selection: Peer-reviewed articles of studies conducted in the United States that compared self-reported and directly measured height, weight, and/or body mass index data to classify overweight among adolescents.
Main exposures: Self-reported and directly measured height and weight.
Main outcome measures: Overweight prevalence; missing data, bias, and accuracy.
Results: Studies varied in examination of bias. Sensitivity of self-reported data for classification of overweight ranged from 55% to 76% (4 of 4 studies). Overweight prevalence was -0.4% to -17.7% lower when body mass index was based on self-reported data vs directly measured data (5 of 5 studies). Females underestimated weight more than males (ranges, -4.0 to -1.0 kg vs -2.6 to 1.5 kg, respectively) (9 of 9 studies); overweight individuals underestimated weight more than nonoverweight individuals (6 of 6 studies). Missing self-reported data ranged from 0% to 23% (9 of 9 studies). There was inadequate information on bias by age and race/ethnicity.
Conclusions: Self-reported data are valuable if the only source of data. However, self-reported data underestimate overweight prevalence and there is bias by sex and weight status. Lower sensitivities of self-reported data indicate that one-fourth to one-half of those overweight would be missed. Other potential biases in self-reported data, such as across subgroups, need further clarification. The feasibility of collecting directly measured height and weight data on a state/community level should be explored because directly measured data are more accurate.