294 - Understanding heterogeneity in pediatric obesity treatment response using the electronic health record

Publication Number: 294.6930

Cody D. Neshteruk, Duke University School of Medicine, Durham, NC, United States; Sarah Armstrong, Professor of Pediatrics, Chapel Hill, NC, United States; Emily M. D'Agostino, Duke University School of Medicine, Durham, NC, United States; Sophie Ravanbakht, Duke University School of Medicine, Durham, NC, United States; Asheley C. Skinner, Duke University, Durham, NC, United States

Background: Youth obesity treatment is characterized by a high degree of heterogeneity in individual response to treatment.

Objective: We aimed to identify predictors of weight change and changes in cardiovascular health among youth receiving obesity treatment.

Design/Methods: Data were drawn from the electronic health record of patients attending an obesity treatment clinic from 2013-2023. Patients aged 2-18 years at the enrollment visit (baseline) with ≥ 1 follow-up visit after baseline were included. Child predictors included baseline age, sex, race/ethnicity, insurance status, and number of comorbidities. Neighborhood environment was included as a predictor by linking geocoded patient addresses at the census track level to the Child Opportunity Index (COI), a measure of neighborhood quality. Height, weight, sex, and age were used to calculate BMI relative to the 95th percentile (BMIp95). Cardiovascular outcomes included systolic and diastolic blood pressure percentiles, glycated hemoglobin (HbA1c), cholesterol, and triglycerides. Repeated-measures linear models were used to examine the relationship between predictors and change in outcomes from baseline, controlling for obesity medication usage, baseline BMIp95, and time (months).

Results: The sample included 3930 youth (mean age: 10.9y; 54% female; 36% Hispanic, 34% Black). Older age at baseline (-0.24; 95% CI: -0.35, -0.13) and higher number of comorbidities (-0.39; 95% CI: -0.75, -0.03) were negatively associated with change in BMIp95 over time. Black youth demonstrated an increase in BMIp95 (0.99; 95% CI: 0.20; 1.78) compared to White youth. Hispanic youth demonstrated greater improvements in systolic and diastolic blood pressure, (-1.47; 95% CI: -2.03; -0.91; -2.45; 95% CI: -3.17, -1.72, respectively) compared to White youth. Patients with public vs. private insurance experienced greater improvements in systolic and diastolic blood pressure (-0.50; 95% CI: -0.93, -0.08; -1.29; 95% CI: -1.86, -0.73, respectively). COI z-scores were negatively associated with change in BMIp95 (-10.03; 95% CI: -19.7, -0.30). No significant predictors were observed for HbA1c, cholesterol, or triglycerides.

Conclusion(s): Findings can inform tailoring of child obesity treatment to the unique needs of individual patients, which may promote greater retention and adherence as well as improve treatment outcomes. Additional research is needed to understand environmental factors (e.g., family, built environment) that may influence individual response to obesity treatment.