Abstract
This study demonstrates the association between longitudinal BMI-percentile patterns and social determinants of health (SDOH) in early childhood. Five distinct clusters with different BMI-percentile trajectories are found and a strong association between these clusters and SDOH is observed. Our findings highlight the importance of targeted prevention and treatment interventions based on children's SDOH.
Introduction
Clinical research has traditionally treated childhood obesity as a static outcome, often relying on cross-sectional snapshots taken during individual pediatric visits. However, this approach overlooks the dynamic nature of growth. Two children may present with the same Body Mass Index (BMI) at age seven, yet their paths to that point—one maintaining a high BMI since infancy and another experiencing a rapid increase at age four—suggest different underlying etiologies.
Understanding these longitudinal "trajectories" is critical for identifying the influence of neighborhood-level Social Determinants of Health (SDOH). By mapping how factors like poverty, housing stability, and education levels associate with specific growth patterns, clinicians and policymakers can transition from generic interventions toward community-specific prevention strategies.
Methods and Data Sources
This study utilized a comprehensive dataset linking electronic health records (EHR) with census-level social data. The clinical cohort consisted of 36,910 children from the Nemours Children’s Health system, spanning Delaware, Florida, New Jersey, Maryland, and Pennsylvania between 2002 and 2019.
To ensure data quality and longitudinal relevance, the researchers applied a multi-stage filtering process. As shown in Figure 1, the initial dataset of over 68,000 children was refined to include only those with sufficient BMI observations over a five-year period, excluding underweight individuals to focus on overweight and obesity risks.
The analysis employed a two-step data-driven clustering approach:
- Heuristic Filtering: Identification of "homogeneous" groups who remained consistently within normal or obesity ranges (≥95th percentile).
- Latent Class Growth Mixture Modeling (LCGMM): Identification of hidden subpopulations among "heterogeneous" children whose weight status fluctuated over time.
Finally, multinomial logistic regression was used to calculate the odds ratios (OR) of belonging to specific trajectories based on 22 neighborhood-level SDOH variables derived from the U.S. Census Bureau American Community Survey (ACS) and standardized indices like the Child Opportunity Index (COI).
Results: Five Distinct BMI Trajectories
The modeling identified five distinct growth patterns, each showing unique associations with neighborhood stressors:
- Always Normal Weight (19.89%): The reference group for the study.
- Overweight Most of the Time (40.65%): The largest cohort, frequently linked to public insurance and minority backgrounds.
- Increasing BMI% (24.54%): Children who started at a normal weight but crossed into overweight or obesity categories. This group was strongly associated with crowded households (OR 2.5) and neighborhoods with high unemployment.
- Decreasing BMI% (13.70%): Children who began at a higher weight but moved into the normal range by age seven. This "positive" trajectory was strongly associated with higher neighborhood preschool enrollment (OR 1.26) and higher rates of bachelor's degrees.
- Always Having Obesity (1.16%): The highest-risk group, showing the strongest links to single-parent households (OR 7.1) and neighborhood unemployment (OR 5.7).
A key finding was the distinction between the "Increasing" and "Always" obesity groups. While the "Always" group was more common in neighborhoods with high housing density and immigrant populations, the "Increasing" group—which saw rapid gains during the transition to school age—had a disproportionately high prevalence of Black children (>40%). This suggests that systemic neighborhood stressors may trigger specific physiological responses during early childhood development.
Discussion and Limitations
The study underscores that neighborhood environments are powerful predictors of early childhood weight gain. Specifically, the protective power of education and the detrimental impact of housing instability highlight clear targets for policy intervention.
However, several limitations remain. The study relied on the last available address for each child, which may not account for the impact of relocation between different neighborhood types. Additionally, while the multi-state dataset is robust, it may not be fully representative of the entire U.S. population. Future research should incorporate multilevel analyses to explore the intersectionality between individual demographics and neighborhood stressors, as well as the role of biological factors and specific medical conditions.
By moving beyond static snapshots, this research provides a roadmap for more nuanced, trajectory-based pediatric care that accounts for the social context in which a child lives and grows.
Summary updated