Estimating Childhood Obesity Prevalence in Communities Through Multi-institutional Data Sharing

Area of Science:

• Public Health
• Epidemiology
• Health Informatics

Background:

• Local childhood obesity prevalence data are crucial for targeted prevention efforts but often unavailable.
• Electronic Health Records (EHR) offer a potential data source for estimating local obesity rates.

Purpose of the Study:

• To establish a multi-institutional system for sharing EHR data to generate childhood obesity prevalence estimates at the census tract level.
• To adjust these prevalence estimates based on population demographic characteristics.

Main Methods:

• Utilized EHR data from six healthcare organizations in Denver County (2014-2016) for children aged 2-17.
• Calculated Body Mass Index (BMI) from height and weight, defining obesity as BMI ≥ 95th percentile.
• Determined census tract location and imputed missing race/ethnicity data for adjusted prevalence calculations.

Main Results:

• Obesity prevalence estimates were generated for 89,264 children, representing 73.9% of the estimated population.
• The county-level adjusted childhood obesity prevalence was 13.9%, with higher rates in males, older children (12-17), and Hispanic individuals.
• Significant variation in adjusted obesity prevalence was observed across census tracts, ranging from 0.4% to 24.7%.

Conclusions:

• A multi-institutional EHR data-sharing system is feasible for producing local childhood obesity prevalence estimates.
• This system can provide valuable, localized data to inform and support community-based obesity prevention initiatives.