Approaches to Estimating Childhood Obesity Prevalence from Electronic Health Records

• Activity Number: 445 - GOVT CSpeed 2
• Type: Contributed
• Date/Time: Thursday, August 12, 2021 : 4:00 PM to 5:50 PM
• Sponsor: Government Statistics Section
• Author(s): Erin Tanenbaum* and Raymond King and Dawn Heisey-Grove and Samantha Lange and Devi Chelluri and Susan Paddock and Scott Campbell and Jason Boim and Peter Mork and Andy Gregorowicz
• Companies: NORC at the University of Chicago and CDC and MITRE and CDC and NORC at the University of Chicago and NORC at the University of Chicago and NORC at the University of Chicago and NORC at the University of Chicago and MITRE and MITRE
• Keywords: Electronic Health Records; non-probability based sample; administrative records; Health informatics; social determinants of health; race imputation

Abstract

Public health surveillance of obesity relies on probability-based suveys, which may be expensive and struggle with response rates, geographic coverage, and timeliness. Some systems also depend on self-reported weight and height which can result in biased estimates. Here, we present methods and tools to for generating timely population-based estimates of obesity using non-probability-based electronic health record (EHR) data. EHR estimates are statistically weighted to the census population based on individual and population-level social determinants of health. Race is imputed using census information as well as patients’ known chronic conditions. These methods and tools were applied to IQVIA’s Ambulatory Electronic Medical Record (AEMR) data to generated population-based obesity estimates for children age 2-19. We will discuss the methodology and share comparisons of national AEMR weighted non-probability-sample estimates to population-based Nation Health and Nutrition Examination Survey (NHANES) estimates. The NHANES youth obesity estimate was 18.5% (2015-2016) compared to 17.2% in AEMR. Demographic comparisons will be presented and discussed.