Real-World Effectiveness of the Bright Bodies Healthy Lifestyle Intervention for Childhood Obesity

Abstract

Objective(s):

To examine the extent to which Bright Bodies, a high-intensity, family-based pediatric weight management intervention, improved body mass index for participants since publication of the RCT establishing efficacy in 2007, and to describe adaptations to the program.

Methods:

For participants enrolled from 2008-2018, we used mixed linear mixed effects models to estimate monthly change in BMI expressed as percentage of the 95^th percentile (%BMIp95) during participants’ first beginner-level program.

Results:

The sample included 396 youth (mean ± SD age 11.7 ± 2.8 years, 61.6% female, 37.1% non-Hispanic Black, 26.3% Hispanic or Latino, 53.8% with public insurance, 80.1% with severe obesity). Across the 11 years, participants’ %BMIp95 reduced on average by 1.63% (95% CI, 1.44 to 1.82) per month during their first program (mean duration 10 weeks) after adjusting for age, sex, season and year, starting %BMIp95, race and ethnicity, and insurance category. Greater reduction in %BMIp95 was associated with male versus female sex, spring/fall versus winter seasons, enrollment in 2008-2018 versus 2015-2018, and higher starting %BMIp95 (p-value for all <0.001). Adaptations since 2007 included pragmatic changes to increase engagement and address funding shortages.

Conclusion:

Our results suggest sustained clinical effectiveness of Bright Bodies in the context of real-world adaptations.

INTRODUCTION

More than 1 in 3 U.S. children ages 2-19 years are affected by overweight or obesity, with an alarming increase in the rates of severe obesity over the last few decades.^1–3 Obesity in childhood is associated with substantial burden of chronic diseases in the short-term and long-term.^4–6 Behavioral interventions remain the cornerstone and first line of treatment for youth with obesity.⁷

Bright Bodies is a family-based, multidisciplinary, intensive lifestyle intervention developed at Yale to meet the needs of predominantly low-income, racial and ethnic minority children and adolescents with overweight and obesity in New Haven, Connecticut.⁸ A randomized controlled trial (RCT) published in 2007 showed reduced body mass index (BMI) among 8-16 year-old children with obesity randomized to Bright Bodies versus clinical control at 6 and 12 months, with sustained effects at 24 months.^9,10 The U.S. Preventive Services Task Force (USPSTF) 2017 Report cited Bright Bodies as one of only a few effective interventions for youth with obesity published with a clinically-meaningful improvement in BMI.¹¹ Since 2008, Bright Bodies has continued as part of routine clinical care. Adaptations to the Bright Bodies program—including efforts to increase participant engagement, pragmatic changes to address budgetary constraints, and updates to the curriculum to keep content current—have been implemented over time to sustain the program.

Among the few behavioral interventions for childhood obesity with demonstrated efficacy, little is known about the ongoing effectiveness of these interventions following completion of clinical trials. There have been increasing calls for evidence to inform implementation of efficacious obesity treatment programs in diverse, real-world settings.^12–15 Additionally, there is a growing body of literature that underscores the importance of assessing healthcare interventions as they are tailored to local contexts to improve fit and sustainability.¹⁶ Studies of adaptations after completion of randomized controlled trials are needed to explore how program modifications may both positively and negatively impact intervention fidelity, effectiveness, implementation, and sustainability over time.^17,18

The primary aim of this real-world effectiveness study was to assess BMI changes among first-time Bright Bodies participants between 2008 and 2018. We hypothesized that, despite adaptations to sustain the program, Bright Bodies would demonstrate continued effectiveness in improving BMI. We additionally examined the extent to which specific participant and program characteristics were associated with BMI changes. Finally, we described program adaptations since 2007 using a well-established framework for study of adaptations from the field of implementation science.¹⁹

METHODS

Participants and Setting:

We included all participants who (i) enrolled in Bright Bodies and attended ≥1 orientation, nutrition education, or physical activity class between 2008 and 2018; and (ii) had an initial BMI measurement ≥85^th percentile for age/sex. We excluded participants (i) for whom first program data was unavailable or (ii) who concurrently enrolled in an RCT assessing the efficacy of Bright Bodies for pediatric subjects with early signs of prediabetes/impaired glucose tolerance.²⁰

Bright Bodies Intervention:

As previously described,^9,10,20 the Bright Bodies intervention included nutrition education classes, behavior modification classes, and supervised physical activity. The nutrition education/behavioral modification component (40 minutes, once weekly) utilized a non-diet approach that emphasizes nutrient-dense foods of moderate portion sizes. The behavioral modification component included several lessons about techniques such as self-awareness, goal setting, coping skills training, and cognitive behavioral strategies. Participants and their parents/caregivers received a standardized workbook, The Smart Moves Workbook, for use throughout the program. Sample topics included: “Better Food Choices: a Non-Diet Approach,” “What’s in Your Drink?,” and “Risky Business: Identifying and Coping with a High-Risk Situation.” An exercise physiologist or physical therapist led the physical activity component (50 minutes, twice weekly). Participants met in the evenings at a local public school with programs scheduled 3 times per year in the winter, spring, and fall seasons. Participants were permitted to enroll after the start of the program. Bright Bodies was originally designed to be delivered in successive 12-week programs at beginner, intermediate and advanced levels.⁸ However, due to scheduling limitations (school holidays, weather-related closures,) program duration held from 2008 to 2018 ranged from 6 to 12 weeks.

Data Collection and Measures:

We obtained demographic, anthropometric, and participation data routinely collected as part of the program. Staff recorded all program data electronically using standardized templates containing demographics data, height, weight, body fat percentage, and class attendance. To identify unique participants across multiple programs completed over 11 years, we used name, date of birth, sex, and contact information. For discrepancies, we confirmed demographic data in the electronic health record (EHR), when available. We extracted parent-reported race and ethnicity (categorized as non-Hispanic White, non-Hispanic Black, Hispanic or Latino, or other/unknown) and current health insurance type (public, private, or other/unknown) from the EHR. We included these covariates based on prior literature suggesting variation in effectiveness of pediatric weight management interventions by age, sex, season of program enrollment, race, ethnicity, and insurance type.^21–31 The date of enrollment was considered the date of first available recorded weight measurement. To determine starting BMI category, we defined overweight as age- and sex-adjusted BMI percentile ≥85^th to <95^th percentile and obesity as BMI percentile ≥95^th percentile, with Class 1 obesity defined as BMI between 100% and 119% of the 95^th percentile and severe obesity as BMI between 120% and 139% of the 95^th percentile (class 2 obesity) or ≥140% of the 95^th percentile (class 3 obesity).⁵ We recorded enrollment season (winter, spring or fall) and year (early [2008-2010], middle [2011-2014] or later [2015-2018]).

At the time of enrollment and on the final day of the program, staff measured weight (participant in socks and light clothing) to the nearest 0.1 kg and body fat percentage using a medical weight scale/body fat analyzer (TBF 300, Tanita, Arlington Heights, IL, used in previous Bright Bodies trials due to its high correlation with dual-energy X-ray absorptiometry in children).⁹ A Harpenden stadiometer (Cambridge, MD), calibrated in 0.2-cm intervals, was used to determine height. These were the same standard protocols followed in the original RCT for measurements of weight and height.⁹ Weekly weights were additionally measured using the same approach and scale, zeroed and calibrated before each weight. Linear extrapolation was used to estimate the expected weekly height. If only starting height was available, that height was carried forward for all subsequent BMI calculations. The primary outcome was monthly change in BMI expressed as percentage of the 95^th percentile (%BMIp95) during participants’ first Bright Bodies beginner-level program. We chose this metric based on recent reports revealing limitations of using BMI z-score to assess change in adiposity over time, particularly for children and adolescents with severe obesity³² and recommendations to use %BMIp95 to assess change in weight status in youth in intervention studies ^32,33 Additional outcome measures included monthly changes in BMI and body fat percentage.

To describe adaptations made to the Bright Bodies program since 2007, P.H. and S.L.S. interviewed the program director (M.Sa.) and two prior program coordinators. Utilizing Stirman et al’s well-established framework for describing adaptations and modifications to interventions, we categorized adaptations according to relevant categories, including when modifications occurred; whether adaptations were planned/proactive or unplanned/reactive; what was modified; and goals and reasons for modification.¹⁹

Statistical Analyses:

Baseline individual and program characteristics were summarized using mean and standard deviation for continuous variables and frequency and proportion for categorical variables. To examine monthly change in the outcomes, we conducted growth curve analysis using linear mixed effect modeling. We included participants with at least 1 BMI measurement available to better pursue an intention-to-treat analysis. All available weekly BMI measurements for enrolled participants’ first program were included in the model. Time, expressed as days from enrollment, was the main fixed effect to capture the slope of outcomes. A random intercept was used to account for within-subject correlation among repeated weight assessments. Unadjusted univariable models were first used to explore the relation of each covariate with overall means of outcomes. Multivariable models included adjustment for age, sex, race and ethnicity, insurance type, season and year of first Bright Bodies program, and starting %BMIp95 or BMI as well as their interactions with days from enrollment. Linear contrasts were constructed to compare the slopes by subgroup defined by each covariate. Slopes of BMI categories were estimated using mean baseline %BMIp95 or BMI of participants in each category.

For the subset of participants with complete attendance data available, we evaluated the effect of contact hours received in a separate analysis, stratifying the fully-adjusted models of monthly change in %BMIp95 and BMI by tertiles of total contact hours with an interaction term. Baseline characteristics were compared between participants with and without available attendance data using t-test or Chi-square test, as appropriate.

At the time of publication of the initial Bright Bodies RCT in 2007, methodological concerns with using BMI z-score to assess change in adiposity (noted above) had not yet emerged, and BMI z-score was commonly used in studies of weight management interventions. To facilitate direct comparison of our primary analysis to outcomes reported in the initial Bright Bodies RCT, we performed a secondary data analysis of the RCT data to examine change in the newer, recommended %BMIp95 outcome metric.

All statistical analyses were performed using SAS Software, version 9.4 (SAS Institute, INC., Cary, NC). The significance level was set as p<0.05, two-sided. The study was approved by the Yale University School of Medicine Institutional Review Board (IRB).

RESULTS

Sample Characteristics:

Of 33 Bright Bodies programs held during the 11 years from 2008 to 2018 (3 programs per year), we located data from 27 programs (82%). While all program data from available spreadsheets included attendance records for nutrition/behavior modification classes, 16/27 (59.3%) contained attendance records for physical activity classes. We excluded participants for whom anthropometric and attendance data were missing from their first Bright Bodies program, presumably because they initially enrolled either prior to 2008 or during one of the 6 programs for which spreadsheets were missing (n = 60), and participants who concurrently enrolled in an RCT for patients with impaired glucose tolerance during the study period (n = 23).²⁰

In the final sample of 396 participants, mean (SD) age was 11.7 (2.8) years; 60.6% were female, 37.1% non-Hispanic Black, 26.3% Hispanic/Latino; and 53.8% had public health insurance (Table 1). The majority of participants (80.1%) had severe class 2 or class 3 obesity at enrollment. The 27 programs included a mean (SD) of 14.7 (5.5) new participants per program, with a mean (SD) duration of 69.2 (15.3) days per program.

Table 1.

Characteristics of individual participants (N = 396) and 27 Bright Bodies sessions from 2008-2018

A: Participant characteristics
Participant Characteristics (N = 396)	No. (%) or Mean (SD)
Age at enrollment, years	11.68 (2.78)
Sex, female	240 (60.61%)
Race and ethnicity
Non-Hispanic White	90 (22.73%)
Non-Hispanic Black	147 (37.12%)
Hispanic or Latino	104 (26.26%)
Other	16 (4.0%)
Unknown	39 (9.8%)
Health insurance type
Public	213 (53.79%)
Private	136 (34.34%)
Other	6 (1.5%)
Unknown	41 (1.0%)
Baseline anthropometric measures
BMI expressed as percentage of the 95th percentile (%BMIp95)	142.82 (27.33)
BMI z-score	2.45 (0.37)
Body fat, % (n = 352)	45.16 (8.25)
Baseline BMI category
Overweight	18 (4.6%)
Class 1 obesity	61 (15.40%)
Class 2 obesity	118 (29.8%)
Class 3 obesity	199 (50.25%)
Season of first session
Winter	158 (39.90%)
Spring	88 (22.22%)
Fall	150 (39.90%)
Year of first session
Early (2008-2010)	141 (35.61%)
Middle (2011-2014)	145 (36.62%)
Late (2015-2018)	110 (27.78%)
B: Program characteristics
Characteristics of Bright Bodies Programs (N = 27)	No. (%) or Mean (SD)
Year
Early (2008-2010)	8 (29.63%)
Middle (2011-2014)	9 (33.33%)
Late (2015-2018)	10 (37.04%)
Season
Winter	10 (37.04%)
Spring	8 (29.63%)
Fall	9 (33.33%)
For all 27 sessions with available data:
Number of new participants enrolled in session	14.67 (5.53)
Duration of session, days	69.15 (15.35)
Total number of nutrition/behavior modification classes held in session	9.52 (1.63)
For 16/27 sessions with available exercise class attendance data:
Total number of nutrition/behavior modification classes held in session	9.94 (1.29)
Total number of exercise classes held in session	18.25 (2.96)
Total number of classes (nutrition/behavior modification + exercise) held in session	28.19 (4.09)
Total number of contact hours offered in session	21.83 (3.21)

Effects of the Bright Bodies Intervention:

Across the 11 years, mean %BMIp95 decreased by −1.63% per month (95% Confidence Interval (CI) −1.44, −1.82) during participants’ first Bright Bodies beginner-level program, in mixed-effects linear regression models adjusted for age group, sex, year and season of enrollment, starting %BMIp95 category, race and ethnicity, and insurance type (Figure 1). The adjusted mean reduction in BMI was −0.29 kg/m² (95% CI −0.24, −0.33) per month. Table 2 shows unadjusted analyses for monthly change in %BMIp95 overall and stratified by the same subgroups. For the subgroup of participants with body fat percentage measurements available at both the start and end of the program (n = 200), body fat percentage improved by −0.8% per month (95% CI −2.38, −0.79) after adjusting for the same covariates.

Figure 1.

Monthly change in percentage of body mass index (BMI) at the 95^th percentile (%BMIp95) for participants during their first Bright Bodies program, adjusted for age group, sex, year/season of first session, starting %BMIp95, race/ethnicity, and insurance type.

†: p-value type-3 omnibus test

‡: Slopes of categories were estimated using average baseline %BMIp95 of participants in each category; Interpretation for the effect of baseline % BMI percent of 95%: For each additional unit increase in baseline % BMI percent of 95% we expect a decrease (improvement) of 0.012, (95% CI −0.018, −0.007) in % BMI percent of 95% per month

Table 2.

Monthly change in percentage of body mass index at the 95^th percentile (%BMIp95) during participants’ first session; bivariate (unadjusted) analyses overall and stratified by age, sex, year/season of enrollment, starting %BMIp95, race/ethnicity, insurance type

	Monthly Change (95% CI)	P-value†
Overall	−1.54 (−1.68, −1.40)	<0.001
Age group		0.062
Older; > 12 yr	−1.37 (−1.60, −1.15)
Younger; ≤ 12 yr	−1.64 (−1.82, −1.47)
Sex		<0.001
Male	−1.91 (−2.13, −1.68)
Female	−1.33 (−1.51, −1.16)
Season of first session		0.005
Spring	−1.81 (−2.24, −1.39)
Fall	−1.74 (−1.95, −1.53)
Winter	−1.30 (−1.50, −1.09)
Enrollment year		<0.001
Early (2008 - 2010)	−1.88 (−2.11, −1.66)
Middle (2011 - 2014)	−1.43 (−1.66, −1.21)
Late (2015 - 2018)	−1.19 (−1.47, −0.92)
Starting %BMIp95 ‡		<0.001
Overweight	−0.88 (−1.15, −0.61)
Class 1 obesity	−1.12 (−1.32, −0.92)
Class 2 obesity	−1.40 (−1.54, −1.25)
Class 3 obesity	−1.86 (−2.03, −1.68)
Race/ethnicity		0.535
Hispanic or Latino	−1.36 (−1.64, −1.09)
Non-Hispanic Black	−1.60 (−1.84, −1.35)
Non-Hispanic White	−1.62 (−1.89, −1.36)
Other/unknown	−1.57 (−1.91, −1.23)
Insurance type		0.84
Public	−1.57 (−1.77, −1.38)
Private	−1.48 (−1.72, −1.25)
Other/unknown	−1.57 (−1.92, −1.22)

^†:p-value type-3 omnibus test

^‡:Slopes of categories defined by average baseline %BMIp95 of participants in each category; Interpretation for baseline %BMIp95: For each additional unit increase in baseline %BMIp95 we expect a decrease (improvement) of 0.015 (−0.020, −0.010) in % BMI percent of 95% per month

We observed statistically significant decreases in %BMIp95 in all subgroups of covariates of interest. The magnitude of these differences was modified by sex, season, initial enrollment year, and starting %BMIp95 (p<0.001). Figure 1 displays the greater magnitude of %BMIp95 reduction we observed among males compared with females, spring and fall versus winter seasons, earlier (2008-2010) versus middle and later year of enrollment and more severe versus less severe obesity class. We did not observe significant differences in monthly change in %BMIp95 by age group, race/ethnicity, or insurance category.

Of the 396 participants in the total sample, 191 participants from 16/27 program cohorts had complete nutrition/behavior modification and physical activity class attendance records available. Those with versus without attendance data available did not significantly differ with regard to age at enrollment, sex, race or ethnicity, health insurance type, or starting anthropometrics. The subgroup with complete class attendance data available attended a mean (SD) of 5.7 (2.8) nutrition/behavior modification classes and 9.9 (5.2) physical activity classes -- totaling 12.0 (6.0) contact hours received by participants during the program (Table 3). After adjusting for covariates, the magnitude of BMI change differed by level of contact (p<0.001), with the greatest reduction in %BMIp95 (−2.27% [95% CI −1.96, −2.58] per month) among those in the highest attendance group who received at least 15 intervention contact hours (Table 4).

Table 3.

Attendance of Bright Bodies participants in their first session

Characteristic	No. (%) or Mean (SD)
For full sample (N = 396)
Present at orientation class	224 (56.57%)
Present at last nutrition/behavior modification class held in session (“completer”)	213 (53.79%)
Nutrition/behavior modification
Number of classes attended	5.18 (2.80)
Percent of classes attended	0.54 (0.28)
For 16/27 sessions with available physical activity class attendance data (N = 191)
Present at orientation class	121 (63.35%)
Present at last nutrition/behavior modification class held in session (“completer”)	115 (60.21%)
Nutrition/behavior modification
Number of classes attended	5.68 (2.82)
Percent of classes attended	0.56 (0.27)
Physical activity
Number of classes attended	9.88 (5.22)
Percent of classes attended	0.53 (0.26)
Total
Number of classes (nutrition/behavior modification + physical activity) attended	15.55 (7.76)
Percent of classes attended	54 (26)
Number of intervention contact hours completed (nutrition/behavior modification + physical activity classes)	12.02 (6.03)

Table 4.

Monthly changes in %BMIp95 and BMI, stratified by attendance tertile (N = 191)

	N	%BMIp95		BMI
		Adjusted Monthly Change* (95% CI)	P-value†	Adjusted Monthly Change* (95% CI)	P-value†
Contact Hours Received			<0.001		<0.001
Low (1.5 – 9.33 hr)	65	1.33 (0.62, 2.04)		0.45 (0.28, 0.62)
Medium (9.5 – 14.83 hr)	65	−1.42 (−1.81, −1.03)		−0.27 (−0.37, −0.18)
High (15 – 24.83 hr)	61	−2.27 (−2.58, −1.96)		−0.46 (−0.53, −0.38)

^†:p-value type-3 omnibus test

^*Models adjusted for age, sex, season and year of enrollment, starting %BMIp95 or BMI, race/ethnicity, and insurance type

Our re-analysis of data from the original Bright Bodies RCT data using the %BMIp95 metric found within-group change in %BMIp95 among Bright Bodies participants (n=105) of −10.7% (95% CI −8.8, −12.7) at 6 months, −11.9% (95% CI −9.2, −14.7) at 12 months, and −11.1% (95% CI −7.7, −14.5) at 24 months from baseline. Assuming a constant monthly rate of %BMIp95 change, this extrapolates to a monthly change of −1.8% per month over 6 months, −1.0% per month over 12 months, and −0.5% per month over 24 months for participants in the Bright Bodies group. For the clinic control group (n = 65), in which participants received visits with a multidisciplinary weight management team every 6 months, %BMIp95 changed by +1.3% (95% CI −1.3, 3.9) at 6 months, −0.1% (95% CI 3.5, −3.7) at 12 months, and −5.4% (95% CI −0.8, −9.9) at 24 months. Assuming a constant monthly rate of %BMIp95 change, this extrapolates to a %BMIp95 change of +0.2% per month over 6 months, −0.01% per month over 12 months, and −0.2% per month over 24 months. Overall, in the original RCT we observed statistically significant treatment effects for change in %BMIp05 (intervention group – control group) of −12.0% (95% CI −15.3, −8.8; p <0.001) at 6 months, −11.8% (95% CI −16.4, −7.3; p <0.001) at 12 months, and −5.7% (95% CI −11.4, −0.1; p =0.04) at 24 months.

Adaptations to the Bright Bodies Intervention Since 2007:

Adaptations to the program since the initial trial have included efforts to increase engagement and motivation, pragmatic changes to address funding shortages, and updates to the curriculum (Table 5). Some modifications, such as curriculum updates and introduction of “Bright Bodies Bucks” for raffle prizes/incentives, were proactive measures taken to keep the program content current and encourage participation. Others were unplanned responses to participant requests or funding shortages. In 2009, two pragmatic changes were made due to funding shortages: 1) more of the staff were trainees rather than fully certified professionals and 2) intermediate and advanced levels were combined into one group. In 2013, the physical activity curriculum became more game-focused in response to participants’ requests.

Table 5.

Adaptations to the Bright Bodies program since initial randomized controlled trial, 2007-2018

Implementation Year	Adaptation	Were Adaptations Planned?1	What Was Modified?1	Description and Reasons for Adaptation	What Was the Goal?1
Efforts to increase engagement and motivation
2007	Completion certificates	Planned/Proactive	Implementation and scale-up activities	Prizes for “biggest loser” (2 for younger group and 2 for older group) were discontinued. Instead, all participants received completion certificates at the end of each session.	Increase engagement; reduce cost; increase satisfaction
2008	Bright Bodies “Bucks”	Planned/Proactive	Implementation and scale-up activities	Introduced 3-4 raffle prizes throughout session. Sought to increase recognition of all participants who demonstrated progress and not just “best” outcomes. Bright Bodies “bucks” for raffle drawing were earned for a variety of activities demonstrating engagement with the program (e.g., attendance; submitting physical activity logs)	Increase engagement; reduce cost; increase satisfaction
2013	Brief exercise class added to orientation	Planned/Proactive	Content (lengthening/extending)	Brief exercise class added in response to perceived eagerness of participants to start exercise during orientation class.	Increase engagement; increase retention
2013	Inclusion of more games in physical activity curriculum	Unplanned/Reactive	Content (adding elements; removing elements)	In response to request by participants and goal of increasing participant “ownership” in the program, more interactive games were added to the physical activity classes.	Increase engagement and satisfaction; improve fit with participants
2015	Inclusion of siblings	Unplanned/Reactive	Contextual (population)	Siblings including those without overweight/obesity permitted to attend program in response to families’ requests. Sought to improve retention and family-based practice.	Increase retention; improve feasibility; improve fit with recipients; increase satisfaction
2016	More modifications included during exercise instruction	Planned/Proactive	Content (tailoring/tweaking/refining)	Modifications for varying fitness levels shown more frequently during class to accommodate all participants and prevent injuries.	Improve fit with recipients
Pragmatic changes
2009	Increased use of trainees as staff	Unplanned/Reactive	Contextual (personnel)	Utilized more nutrition and exercise physiology students in response to: (1) high volume of requests for precepting trainees from local universities; and (2) decreased funding for seasoned staff.	Reduce cost
2009	Restructured group levels from 4 levels to 2 (Beginner/Intermediate)	Unplanned/Reactive	Contextual (format)	In response to group size and experienced staff shortages, combined Intermediate I, Intermediate II, and Advanced groups to one group, Intermediate/Advanced.	Improve feasibility; reduce cost
Updates to curriculum
Ongoing	Curriculum updates	Planned/Proactive	Content (tailoring/tweaking/refining; adding elements; condensing/shortening; substituting; removing or skipping elements; reordering of intervention elements or modules; changes in packaging or materials)	Nutrition and behavior modification topics added annually to address new topics of interest and keep material current with pediatric weight management literature (e.g., “Break the Fast with Breakfast”; “Stress Management for the Stress Eater”)	Improve effectiveness or outcomes; increase engagement; address cultural factors

DISCUSSION

In this study of 396 participants in 27 sessions of the Bright Bodies program held between 2008 and 2018, we observed significant overall monthly reductions in %BMIp95 during participants’ first beginner-level program. In subgroup and stratified analyses, the magnitude of %BMIp95 improvement varied by year of enrollment, participant sex, season of enrollment, and attendance as measured by contact hours. We observed the greatest magnitude of %BMIp95 reduction in those with more severe (class 3) obesity. We did not observe differential effects in older versus younger participants, by race and ethnicity, or by insurance category in unadjusted (Table 3) or adjusted (Figure 1) models.

In the Bright Bodies RCT published in 2007, participants in Bright Bodies had a reduction in mean BMI of −2.1 kg/m² at 6 months compared with a mean BMI gain of +1.1 kg/m² among children randomized to usual care, yielding a treatment effect of −3.1 (95% CI: −4.2, −2.1).⁹ In our re-analysis of the original RCT data, participants in the Bright Bodies group had a reduction in mean %BMIp95 of −10.7% (95% CI −12.7, −8.8) at 6 months compared with a mean %BMIp95 gain of +1.3% (95% CI −1.3, 3.9), yielding a treatment effect of −12.0% (95% CI −15.3, −8.8) at 6 months. Body fat percentage at 6 months improved by −3.2% compared with a mean increase by +2.0% in the control group. If we assume a constant monthly rate of change, these results can be extrapolated to improvements in BMI of −0.35 kg/m² per month, %BMIp95 of −1.8% per month, and body fat percentage of −0.5% per month over 6 months in the original RCT. Thus, our real-world findings from 2008 to 2018 of overall mean monthly reductions in BMI, %BMIp95, and body fat percentage of −0.29 kg/m2, −1.6%, and −0.8%, respectively, during participants’ first program (with mean duration 69.2 days) suggest downward trajectories in BMI and adiposity that are comparable with the 6-month results from the original RCT. In addition, our finding of first-program overall mean monthly %BMIp95 reduction of −1.6% appears promising as likely to yield a clinically meaningful change over 3-6 months of participation based on recent literature. A study of youth with obesity enrolled in 31 sites participating in the Pediatric Obesity Weight Evaluation Registry (POWER) published in 2018 found that a 5-percentage point decrease in %BMIp95 at 6 months (approximately −0.83% per month over the initial 6 months of participation) is associated with improvement in cardiometabolic risk factors.³⁴

We observed greater magnitude of decrease in %BMIp95 and BMI in earlier years (2008-2010) of the Bright Bodies program compared with middle and later years (2011-2018), although notable reductions were significant in all year categories. This finding may be explained, in part, by the concept of ‘voltage drop’ whereby the observed benefit of the intervention diminishes as the intervention moves from the more optimal environments of efficacy trials to effectiveness studies.³⁵ Additionally, we hypothesize that some of the adaptations identified in our study may have adversely impacted program effectiveness in more recent years. In particular, increased use of trainees for teaching classes to address the lack of funding and reimbursement by insurance payers and/or the inclusion of more games in the physical activity curriculum at the request of families may have unintentionally decreased the program’s fidelity over time. Despite this observed decrease in effectiveness over time, it is notable that even in the later years (2015-2018) during which improvement in BMI was lower compared to earlier years, the monthly change in %BMIp95 was −1.38% per month (Figure 1), which remains likely to yield clinical benefit if maintained over 6 months of participation.

Our observation that males and those enrolling in spring or fall had greater %BMIp95 reduction compared with females and first-time participants in the winter sessions, respectively, may be explained, in part, by the program’s emphasis on physical activity delivered twice weekly as part of the program and encouraged outside of classes. A previous study demonstrated greater improvements in BMI in males with an intervention that was primarily exercise-based.³⁶ Another showed greater improvement in %BMIp95 at 1 year for participants initially enrolled in an obesity management clinic in the spring/summer, perhaps due to relative ease of increasing physical activity outside during these months.³⁷ Studies of seasonal variation in children’s physical activity have generally found decreased physical activity in the winter; Bright Bodies participants’ overall activity levels may have been lower during the winter months, mitigating program effectiveness.³⁸

In our study, those with severe obesity had the greatest magnitude of %BMIp95 reduction which is promising in light of several previous studies showing greater effectiveness of behavioral obesity treatment interventions among youth with less severe obesity/overweight; in some cases children with severe obesity continued to gain weight.^39–41 These findings are particularly timely as recent studies show that the greatest increase in childhood obesity prevalence has been observed in the severe obesity category.¹ Moreover, these findings are consistent with results from the POWER study in which %BMIp95 improved by 1.88% over 4-6 months, with greater improvement in participants with class 3 severe obesity in comparison to class 1 obesity. A review for the USPSTF of 42 trials of lifestyle-based interventions to reduce excess weight in children and adolescents found that key components of successful interventions, including Bright Bodies, were high intensity and family involvement.⁷ Bright Bodies has likely remained successful over time, as compared to other lifestyle-based interventions, because the program model has retained its core functions of high intensity—with meetings held twice per week—and strong encouragement of family participation. Further studies are needed to determine which unique aspects of the Bright Bodies intervention—such as its overall high intensity, the emphasis on physical activity, or incorporation of cognitive behavioral strategies—may account for the greater magnitude of effectiveness we observed among participants with severe obesity, and to determine whether these initial improvements can be sustained over longer-term follow-up.

Strengths of our study include the use of over a decade of real-world effectiveness data following an efficacious RCT to assess maintenance of effectiveness and to describe adaptations in a sample with a high proportion of youth from populations disproportionately impacted by obesity, including racial and ethnic minorities and recipients of public insurance.^42,43 It should be noted that the program’s real-world effectiveness was despite the inclusion of youth taking medications with potential to cause significant weight gain such as antipsychotic medications. By contrast, these youth were excluded from the initial RCT. An additional strength of this study is the identification of program adaptations and the underlying reasons for the changes. It is notable that most of the adaptions identified (Table 5) were pragmatic responses to funding shortages. This finding is timely given that limited resources and low insurance reimbursement for delivering such high-intensity lifestyle interventions remain barriers to broad-scale program dissemination and access.^44,45

Limitations of our observational study include the absence of a control group which impedes our ability to assess the incremental benefit of the Bright Bodies intervention above changes due to other factors such as regression to the mean, secular trends such as policy interventions, or other interventions such as medications that may positively or negatively impact weight trajectories. The findings of the original RCT demonstrating BMI gains in the clinical control group⁹ and the continued rise in obesity prevalence nationally^1,3 suggest that this potential source of bias is likely to underestimate the effectiveness of the intervention. As expected with the real-world nature of our data, there was variation in the number of BMI measurements available for each participant, resulting from (i) the varying duration of sessions, (ii) attrition; and (iii) rolling enrollment policy in real-world practice. In contrast to the initial year-long Bright Bodies RCT—in which participants in the intervention group attended the program twice weekly for 6 months and every other week for an additional 6 months, allowing for primary outcome measurements at 6 and 12 months—many new participants since 2008 did not return and/or were missing data beyond their initial program. Therefore, we focused our analyses on first-program changes, and we were unable to estimate longer-term changes in %BMIp95 due to small sample size. While results for the subgroup of participants with complete attendance data available suggested greater magnitude of %BMIp95 improvement in those who received the highest contact hours, consistent with previous studies showing a general dose-response pattern for pediatric lifestyle modification interventions,⁷ our analyses were limited in determining if this is a true dose effect or due to unmeasured participant characteristics.

In conclusion, our results suggest ongoing real-world effectiveness of the Bright Bodies intervention for youth with overweight and obesity, including those with severe obesity. These findings highlight the need for further study and efforts to support longer-term program maintenance, sustainability, and implementation of Bright Bodies in new settings to facilitate broader-scale dissemination and impact. While we observed clinically-meaningful BMI improvements throughout the 11 years studied, we also observed decreased program effectiveness over time which may be attributable to adaptations made in response to constrained funding after the initial RCT. These findings speak to the importance and urgency of allocating adequate resources and insurance reimbursement for comprehensive, evidence-based treatment of childhood obesity.

Study Importance Questions:

What is already known about this subject:

• Despite demonstrated efficacy in clinical trials, many interventions fail to demonstrate effectiveness when continued as part of usual care and adapted to real-world contexts.

• An RCT published in 2007 showed that Bright Bodies, a high-intensity, multicomponent, family-based pediatric weight management program, improved body mass index, body composition, and insulin sensitivity among children and adolescents with obesity, with sustained effects at 24 months. However, effectiveness of Bright Bodies has not been studied since the initial trial.

What are the new findings in your manuscript?

• In this real-world observational study of 396 children enrolled in Bright Bodies sessions between 2008 and 2018, participants’ body mass index improved during their first beginner-level program.

• Our results suggest sustained clinical effectiveness of Bright Bodies in the context of real-world adaptations.

How might your results change the direction of research or the focus of clinical practice?

• Our findings suggest ongoing real-world effectiveness of Bright Bodies and highlight the need for further studies to support long-term program sustainability and implementation of Bright Bodies in new, diverse settings to facilitate broader-scale dissemination and impact.