Date of Award

January 2025

Document Type

Open Access Thesis

Degree Name

Medical Doctor (MD)

Department

Medicine

First Advisor

Mona Sharifi

Second Advisor

Stephanie Samuels

Abstract

Bright Bodies (BB) is a family-based intensive health behavior and lifestyle treatment (IHBLT) for pediatric obesity, with demonstrated efficacy and real-world effectiveness. Our study involved two aims: 1) to assess the extent to which real-world effectiveness varied by participants’ neighborhood characteristics, and 2) to evaluate the impact of the COVID-19-associated transition of BB to virtual delivery on participant engagement, particularly for those disadvantaged by inequities in social determinants of health. To achieve Aim 1, we included first-time BB participants 2008-2018 who had an initial BMI ≥85th percentile and available addresses. Addresses were geocoded using ArcGIS Pro and assigned a Child Opportunity Index (COI) score (Low, Moderate, and High), a robust census-level multidimensional index of neighborhood characteristics. Participants were also assigned a COI Healthy Environment (COI HE) score, an index subcategory which accounts for factors highly relevant to BB and obesity outcomes (e.g. healthy food retailer density, neighborhood walkability). We used two mixed effect linear models to examine differences in monthly change in BMI, measured as a percent of the 95th percentile (%BMIp95), by COI or COI HE, adjusting for age, sex, season/year of first session, and baseline %BMIp95. In the COI HE model, we additionally adjusted for ethnicity and insurance type. The study sample included 391 youth (mean ± SD age 11.7 ± 2.8 y, 60.1% female). Participants’ neighborhoods were categorized as 80.8/66.5% Low COI/COI HE, 10.5/22.5% Moderate, and 8.7/11.0% High. Participants’ %BMIp95 reduced on average by 1.65% (95% Confidence Interval [CI] 1.41, 1.89) per month overall in the COI model; results from the COI HE model were comparable (1.84 [95% CI 1.61, 2.07]). COI was not significantly associated with change in %BMIp95 (p = 0.050). Participants from both High and Low COI neighborhoods had greater change in %BMIp95 compared with those from Moderate neighborhoods (-0.68 [-1.28, -0.07], p = 0.03; -0.44 [-0.85, -0.04], p = 0.03 respectively). In contrast, COI HE was significantly associated with change in %BMIp95 (p = 0.004). Those from High COI HE neighborhoods had greater change in %BMIp95 compared with those from Moderate (-0.65 [-1.18, -0.13], p = 0.014) and Low neighborhoods (-0.79 [-1.25, -0.32], p-value = 0.001). These results suggest that BB is effective across all levels of COI/COI HE, yet may be more so for those from High COI HE neighborhoods. Future modifications to BB and similar pediatric IHBLTs should seek to directly address inequities in access to built environments that facilitate healthy, active living. To achieve Aim 2, we included new participants in the traditional in-person BB (1/2018-12/2019, pre-pandemic) and virtual BB (4/2020-3/2022, during-pandemic) who enrolled within a sufficient timeframe to attend ≥10 classes. Addresses were geocoded and assigned both COI and COI Housing Resources (HQ) scores; the latter accounts for factors uniquely relevant to virtual programming (e.g. limited broadband access, crowded housing). Using Chi-square and independent t-tests, we compared baseline participant demographics, as well as enrollment timing, COI/COI HQ distribution, and geographic spread for in-person vs. virtual participants. Using Kruskal-Wallis, Wilcoxon Rank-Sum, and Pearson correlation tests, we investigated variation in attendance for in-person vs. virtual participants, assessing for the influence of enrollment timing, COI, COI HQ, and geographic spread. New in-person (N = 67) and virtual participants (N = 73) had similar baseline characteristics (mean age 11 ± SD 2.6y, 55% female), with no meaningful differences in overall attendance, late enrollment rate, or COI/COI HQ distribution. Those in virtual programs had a longer median travel time to the in-person Bright Bodies location by 2.7 min (p = 0.002). We did not observe a meaningful difference in attendance between in-person vs. virtual Bright Bodies among those who enrolled on time (75.0% vs. 74.3%, p = 0.98), but among those who enrolled late, attendance was substantially lower in the virtual group (56.9% vs. 31.3%, p = 0.02). We also did not observe meaningful differences in attendance in in-person vs. virtual BB among those from Low COI (60.0% vs. 50.0%, p = 0.18), Moderate COI (90.5% vs. 84.2%, p = 0.42), or High COI neighborhoods (42.7% vs. 50.0%, p = 0.96). Nor did we observe meaningful differences in attendance among those from Low COI HQ (59.2% vs. 46.7%, p = 0.07), Moderate COI HQ (51.0% vs. 71.4%, p = 0.35), or High COI HQ (80.0% vs. 60.0%, p = 0.32) neighborhoods. Travel time to the in-person BB location was inversely associated with attendance among in-person (r = -0.29, p = 0.02), but not virtual participants (r = 0.06, p = 0.63). These findings suggest that virtual BB appeals to youth with similar demographics compared with in-person BB and has the potential to engage a more geographically diverse population. A tailored orientation for late enrollees may improve engagement specifically in virtual BB. Barriers rooted in housing resource inequities do not appear to hinder participation more in virtual BB compared with in-person BB.

Comments

This is an Open Access Thesis.

Download
Open Access

This Article is Open Access

Share

COinS