The Pennsylvania "Big Data" Project
Understanding Obesity from Epigenetics to Communities
Project Lead: Brian Schwartz, MD, MS
This study extends existing data and ongoing research to understand the dynamics of childhood obesity in a large population of children living in 1,300 communities that vary in their land use, food, physical activity, and social environments.
Using a comprehensive electronic health record (EHR) from the Geisinger Center for Health Research, we have obtained up to 10 repeated body mass index (BMI) measures on over 163,000 children. This allows us to build models to understand how child characteristics, healthcare delivery and key community features affect trajectories of change in BMI. This is the largest study of longitudinal trajectories ever undertaken. We are using both traditional multilevel regression and newer system dynamics models to better understand the complex and dynamic interplay that may give clues about the origins of the obesity epidemic in children.
We are also conducting a nested sub-study of up to 600 children from a subset of 30 communities chosen from places that are both high (N=15) and low (N=15) in community features that could contribute to obesity. We are collecting new data on diet and physical activity from parents and children and mapping features of these communities using direct observation methods.
A key feature of this study is that we are obtaining saliva samples from children to measure DNA methylation, an important new area of study that reflects how obesity-related genes may be turned on and off by various community and behavioral triggers. This is among the first and largest studies of how community context impacts DNA methylation in loci that have been found to play a role in stress, appetite, and inflammation systems.
Click here to see the project's overview information sheet.
There are two phases to this study:
Phase One: Research using electronic health records (EHR) and secondary data sources for community assessment.
- Data obtained from Geisinger electronic health record on 205,000 children between 3 and 18 years of age.
- A geographical information system (GIS) was built by geocoding residential addresses of 163,820 children.
- Developed a massively scaled, longitudinal, multilevel model of BMI trajectories;
- Comprehensive data on community characteristics in 1288 communities on a 37-County area of PA.
- Identified important spatial variation in the proportion of over-weight children by community.
- BMI trajectories vary by many individual, healthcare delivery, and community features.
- There is spatial variation in the food environment.
- Used two different commercial data sources for food establishments (InfoUSA and Dun and Bradstreet).
Phase Two: Research with new, primary data collection.
- Up to 600 children separately recruited and enrolled in project from 30 communities with high (N = 15) and low (N = 15) rates of overweight children.
- Collecting new data from enrolled children and parents.
- Completing direct observation in communities of land use, food, physical activity, and social environments.
- Collecting saliva samples for epigenetic measurements.
- Antibiotic use and childhood body mass index trajectory.
- Community socioeconomic deprivation and obesity trajectories in children using electronic health records.
- Modeling US adult obesity trends: a system dynamics model for estimating energy imbalance gap.
- Attention deficit disorder, stimulant use, and childhood body mass index trajectory.
- Body mass index and the built and social environments in children and adolescents using electronic health records.
Please see here for all Faculty, Staff, Investigators, Students, and Post-doctoral Fellows involved in this project.
To learn more and find out how to get involved, please email the Global Obesity Prevention Center at Johns Hopkins directly here.