Underestimating adolescent obesity

Previous studies have assessed the validity of adolescent self-reported height and weight for estimating BMI and obesity prevalence. However, these validity tests rarely take into account the missingness of self-reports -- that is, which adolescents fail to report height and weight in an interview context. In this paper, the authors analyze 663 adolescents ages 12-17 from the Los Angeles Family and Neighborhood Survey. The analysis examines the correlates of failing to report height and weight, of underestimation and overestimation of height and weight, and of the size of the error in reported height and weight. Obesity prevalence estimates based on self-report, measured height and weight for those who do report, and measured height and weight for those who do not report are then compared. Younger teens both overestimate and underestimate height and underestimate weight more often than older teens, and also have larger errors in self-reports compared to older teens. Consequently, obesity prevalence is substantially underestimated by self-reported height and weight (14.8% based on self-report vs. 22.1% based on measured height and weight for those who self-report, p<.001). In addition, younger teens are significantly more likely than older teens to fail to report height and weight, and those that do not report are significantly more likely to be obese. This leads to an incremental underestimate of 2.5 percentage points, for a total underestimate of 8.8 percentage points. The results have implications for data collection and analysis in obesity research and for obesity screening and prevention programs for adolescents.