Growth Curves, BMI, and Childhood Obesity: An Assessment of Measurement Issues from NYC Public Schools

• Activity Number: 251
• Type: Contributed
• Date/Time: Monday, August 5, 2013 : 2:00 PM to 3:50 PM
• Sponsor: Health Policy Statistics Section
• Author(s): Stuart Sweeney*+ and Kevin Konty and Sophia Day
• Companies: Univ of California and NYCDOHMH and NYCDOHMH
• Keywords: Childhood obesity ; BMI ; LMS ; Growth curves ; Quantile regression

Abstract

U.S. obesity prevalence continues to trend upwards and policies have thus far failed to achieve widespread change. Within this broader context, child and adolescent obesity has also trended upwards and public schools have begun to monitor trends and to serve as a test bed for nutritional outreach and fitness programs. New York City has been particularly aggressive in this regard with a large data collection effort since 2005 and extensive interventions targeting the schools. This paper reports on measurement issues that have emerged to date, particularly with the CDC implementation of the LMS method to define obesity prevalence based on BMI. The current LMS approach results in a non-trivial number of BMI scores being graded as "implausible" but those values persist longitudinally suggesting that it is not simply measurement error. We explore the utility of quantile regression growth curve modeling of longitudinally linked records as an alternative basis for targeting and evaluating health policy interventions.