Using Parks and Recreation Providers to Enhance Obesity Treatment: A Randomized Controlled Trial

Pediatric obesity is a complex, multifactorial disease with genetic, socioeconomic, and environmental influences.¹ As of 2020, 14.7 million United States youths aged 2–19 years were affected by obesity,^2,3 and the current obesity prevalence is almost certainly higher as there was an observed doubling of the rate of youth body mass index (BMI) increase during the COVID-19 pandemic.⁴ Obesity is unlikely to resolve without treatment and leads to significant health and social consequences.^5–7 The 2023 American Academy of Pediatrics (AAP) Clinical Practices Guideline for the Evaluation and Treatment of Children and Adolescents with Obesity outlines safe and effective treatment options, including intensive health behavior and lifestyle treatment (IHBLT), pharmacotherapy, and metabolic and bariatric surgery.⁸ IHBLT is the cornerstone of treatment, providing intensive nutrition and physical activity opportunities for children and their families (ie, ≥26 hours of in-person contact over 3–12 months).⁹ However, IHBLT is limited by poor access, particularly in underserved communities, and the limited capacity of clinics alone to deliver the recommended treatment intensity.¹⁰ Additionally, little is known about the effectiveness of IHBLT among low-income and racial and ethnic minority groups.

One proposed solution has been for health care systems to partner with community organizations to increase capacity and improve access to IHBLT.¹¹ Public parks and recreation (P&R) centers provide low-cost or free access to physical activity and are accessible to diverse populations. Access to P&R facilities is associated with higher levels of physical activity among children and adults.^12,13 We have previously developed a model of pediatric obesity treatment called Fit Together that pairs health care organizations with local P&R providers to deliver IHBLT.^14,15 In the Fit Together model, health care partners screen for obesity, provide counseling, and treat comorbidities while P&R partners provide IHBLT in the community setting. In pilot testing, Fit Together has demonstrated feasibility in engaging a diverse population of youths, the ability to meet the 26-hour threshold, and effectiveness in improving self-reported physical activity and quality of life.^15,16 Based on these preliminary data, we sought to test Fit Together in a randomized controlled trial (Hearts & Parks). Therefore, the aim of the current study was to compare the Fit Together model with clinical care alone to determine the effectiveness of Fit Together on youth BMI and cardiorespiratory fitness among a diverse sample of youths with obesity.

Hearts & Parks was a 2-arm, randomized controlled crossover trial that aimed to evaluate the effectiveness of the Fit Together model of youth obesity treatment in comparison with routine primary care obesity management (ClinicalTrials.gov ID: NCT03339440). A detailed description of the design and methods is available elsewhere.¹⁷ This study and all procedures were approved by the Duke University Health System Institutional Review Board (Pro00086684). The Consolidated Standards of Reporting Trials reporting guidelines were used in this study (Supplementary File 1).¹⁸ 

Participants were recruited from four primary care clinics in Durham, North Carolina, from 2018 to 2020. The eligibility criteria included (1) age 5–17 years and BMI at least at the 95th percentile for age and sex at the time of referral, (2) youth and caregiver ability to speak/read English or Spanish, and (3) caregiver ownership of a device capable of sending and receiving text messages. The exclusion criteria included (1) families living more than 20 miles from the community program, (2) endogenous or genetic causes of obesity, (3) taking medication that causes weight gain or loss, (4) participation in an obesity treatment program within the previous 12 months, or (5) either the caregiver or child having a significant health problem that would limit participation. During well-child visits, primary care physicians (PCPs) were alerted to potentially eligible participants by a best-practice alert (BPA) in the electronic health record. The BPA was programmed specifically for the study and would pop up if a patient met the inclusion criteria. PCPs then had the option to refer the patient to the study (on family consent), opt the patient out (eg, if they had severe obesity that would prevent them from participating or urgent comorbidities), or decline on behalf of the family. Trained study members confirmed eligibility, contacted families regarding participation, and proceeded with consent/assent before the baseline study visit.

Fit Together is a 6-month intervention that includes clinical visits with a pediatric clinic and lifestyle sessions at a local P&R center. In Hearts & Parks, the Fit Together intervention included clinical visits at the Duke Healthy Lifestyles pediatric obesity treatment clinic and access to Bull City Fit, a community program delivered at a Durham P&R center.¹⁴ At Healthy Lifestyles, participants received care from a team of medical providers, registered dietitians, physical therapists and a licensed counselor. Providers worked with patients and families to set and monitor health-related goals and monitor patient safety. Patients were encouraged to visit Healthy Lifestyles once a month. Visits with each provider lasted 30 minutes, and patients often saw multiple providers on the same day. Participants were also encouraged to attend Bull City Fit at least twice a week. Bull City Fit was offered 6 days a week and, at each session, there were opportunities for unstructured and structured physical activity. A weekly cooking class and nutrition education session was also offered. To promote family engagement, at least 1 adult household member was required to attend and participate with the child, and siblings were encouraged to join.¹⁵ 

The waitlist control group received nutrition and activity counseling from their PCP during standard well-child visits and at additional follow-up visits at the discretion of the PCP. All PCPs at the 4 practices received training on motivational interviewing and written materials on nutrition and activity counseling. Families also received a non–obesity-related literacy intervention consisting of a monthly $15 gift card to a locally owned bookstore. At the end of the 6-month waiting period, they were invited to participate in the Fit Together program.

The target sample size was 270 participants based on previous research showing that a meaningful difference in mean 6-month BMI z score is −0.10.⁷ This sample size calculation accounted for a 30% loss to follow-up rate at 6 months and was computed using a 2-sided t test at an alpha significance level of 0.05.

Participants were randomized to the intervention or comparison group in a 1∶1 allocation ratio. Randomization was conducted using the Research Electronic Data Capture system by the study biostatistician. Research coordinators notified patients of their assigned group following the baseline visit. Given the nature of the intervention, neither patients nor providers could be blinded to the treatment assignment.

Data were collected at baseline and at 6 months at the clinic or community site. The study was powered to detect differences in 2 primary outcomes: youth BMI and cardiorespiratory fitness.¹⁷ 

Height was assessed using a Seca stadiometer (Seca, Hamburg, Germany), and weight was assessed using a Tanita body composition scale (Tanita, Tokyo, Japan) using standard protocols. Measures were taken in triplicate and averaged. BMI was calculated based on age- and sex-specific reference values from the Centers for Disease Control and Prevention growth charts.^19,20 Because of the known challenges in assessing changes in BMI among youths with obesity, height and weight measures were converted to BMI relative to the 95th percentile (BMIp95).²¹ Additionally, the following classes of obesity were calculated: class I obesity (BMI ≥95th percentile or BMI ≥30), class II obesity (BMI ≥120% of the 95th percentile or BMI ≥35), and class III obesity (BMI ≥140% of the 95th percentile or BMI ≥40).

Cardiorespiratory fitness was assessed using a submaximal 3-minute step test, a valid and reliable test of cardiorespiratory fitness in youths.²² A baseline heart rate (HR) measure was recorded before the start of the test using a pulse oximeter. Using a metronome, participants were then asked to step up and down on a platform for 3 minutes at a pace of 96 beats per minute. Another HR measure was taken immediately following the completion of the stepping protocol as a measure of peak HR (ie, 3-minute HR). Then the participant was directed to sit in a chair for 1 minute, where their HR was taken again to assess submaximal HR (ie, 4-minute HR). The 4-minute HR values were included in the analysis.

For intervention participants, attendance at clinic visits and Bull City Fit sessions were recorded to assess if participants were able to achieve the recommended 26 hours of treatment. Each visit to Healthy Lifestyles was counted as 1 engagement hour, and each Bull City Fit visit was counted as 2 engagement hours. Total engagement was calculated as the sum of all hours during the first 6 months of the study. Engagement hours were classified into the following categories: 0, less than 5, 5–25, or more than 26 hours.

At baseline, caregivers completed a demographic questionnaire that included child sex, race, ethnicity, annual family income, caregiver employment, food insecurity, and language spoken at home.

On March 15, 2020, all in-person study visits were halted because of COVID-19 mitigation measures. At this point, enrollment was concluded; however, to complete follow-up study visits, a remote data-collection protocol was established. Digital scales and stadiometers were mailed to participants’ homes. Study visits were conducted through direct video observation of child height and weight measures. The 3-minute step test was discontinued because of the inability to complete it remotely. The intervention was adapted to an online format, in which Healthy Lifestyles visits were conducted by telehealth, and Bull City Fit sessions were moved to an online platform, which included synchronous Zoom sessions or asynchronous recordings.

Descriptive statistics were computed to summarize sample characteristics at baseline. Means and SDs were calculated for all continuous variables and frequencies (%) for categorical variables. Differences among the groups at baseline were not statistically tested.²³ To account for the COVID-19 disruption, the cohort was divided into 2 groups: (1) “not affected” and (2) “affected” for the primary analysis. The not affected group included participants with a 6-month visit where at least 1 of the primary outcomes was measured before March 15, 2020 (ie, not affected by protocol disruptions). Participants who did not have a 6-month visit were also considered not affected if their baseline visit was before September 15, 2019 (ie, 6 months before the COVID-19 shutdown). The remaining participants were considered affected owing to some exposure to COVID-19 adaptations of the study protocol during the active study period.

For each group (ie, affected and not affected), generalized estimating equation models were used to assess the effectiveness of the intervention on the primary outcomes (ie, BMIp95 and submaximal HR). An unstructured covariance structure was used in all models to account for correlation between repeated measures. Within-family correlations were weak and thus were not accounted for in the models (sensitivity analyses not shown). The models for the primary analyses included a “time*intervention” term to test for an intervention effect. Additionally, the model for the BMIp95 outcome was adjusted for race and ethnicity, and the model for the 4-minute HR outcome was adjusted for race and ethnicity, age, and sex. Sensitivity analyses were conducted with all participants regardless of COVID-19 disruptions, and additional models were also tested for subgroup differences by race, age (<12 years vs ≥12 years), and sex. The results of subgroup analyses should be interpreted with caution because the study was not sufficiently powered for these analyses. All analyses were conducted using an intent-to-treat approach. Descriptive statistics were used to describe participant engagement with the intervention. All analyses were performed using SAS version 9.4.

The flow of participant recruitment is shown in Figure 1. The final sample included 255 youths with a mean (SD) age of 10 (3.0) years. Nearly half of the participants were male (47.8%), and the majority were either Hispanic (38.8%) or non-Hispanic Black (38.4%). More than half of the participants had class I obesity (54.5%). Full baseline youth and family-level characteristics are shown in Table 1. Slightly fewer than half (112/255, 44%) were affected by COVID-19 disruptions to study protocols. Youth and family-level characteristics of participants not affected by and affected by COVID-19 disruptions are shown in Supplementary File 2.

The effect of Fit Together on BMIp95 and fitness by COVID-19 affected status is shown in Table 2. For youths not affected by COVID-19 disruptions, intervention youths experienced a significant decrease in BMIp95 (β = −3.05; 95% confidence interval [CI], −5.08 to −1.01) as compared with control. However, we did not observe a significant BMIp95 difference between intervention and control youths affected by COVID-19 disruptions (β = −3.25; 95% CI, −7.98 to 1.48). In a sensitivity analysis comparing the entire cohort, intervention youths had a significant decrease in BMIp95 compared with control youths (β = −3.32; 95% CI, −5.69 to −0.96). Because of the discontinuation of the fitness assessment, comparisons are only available for the not affected group. There was no significant difference between intervention and control youths in 4-minute HR (β = −7.18; 95% CI, −16.12 to 1.76).

In exploratory subgroup analyses, there were no significant between-group differences by age category (ie, <12 years vs ≥12 years) and sex (ie, female vs male) for either BMIp95 or 4-minute HR (Table 3). For youths affected by COVID-19 disruptions, there was a significant difference such that non-Hispanic Black youths had decreased BMIp95 (β = −8.58; 95% CI, −15.42 to −1.72).

There were also several significant within-group differences for youths not affected by COVID-19 disruptions. Younger (β = −2.83; 95% CI, −5.25 to −0.42), male (β = −4.83; 95% CI, −7.91 to −1.74), and non-Hispanic white (β = −6.19; 95% CI, −10.33 to −2.05) intervention youths had significant decreases in BMIp95 compared with control youths. Female intervention youths had a significant improvement in 4-minute HR compared with control female youths (β = −12.51; 95% CI, −23.75 to −1.28).

Mean (SD) engagement hours for intervention participants was 8.9 (15.5) hours (range, 0–80 hours). Participants not affected by COVID-19 disruptions had a higher mean number of engagement hours compared with those affected by COVID-19 disruptions (11.5 ± 17.1 vs 5.4 ± 12.5 hours, respectively). In total, 10% (13/131) of participants received a high-intensity dose (ie, ≥26 hours) and nearly one-third of participants (30%, 39/131) received a moderate-intensity dose (ie, 5–25 hours). The hours of engagement for intervention participants are shown in Figure 2. No adverse events related to the study protocol were reported.

This study found that Fit Together, a pediatric obesity treatment intervention delivering IHBLT through a clinic-community partnership, led to improved weight outcomes in youths with obesity; however, no effect on cardiorespiratory fitness was detected. In addition, Fit Together was able to reach a diverse group of youths (>80% from marginalized racial and ethnic groups). The findings highlight the effectiveness of this clinic-community model and will inform future pediatric obesity medicine practice and research.

Recommendations from the AAP and US Preventive Services Task Force support the use of IHBLT to treat pediatric obesity.⁸ In meta-analyses examining the effectiveness of behavioral treatment programs, trials with at least 26 contact hours demonstrated an increased intervention effect on BMI (mean difference, − 1.4) compared with trials with less than 26 hours (mean difference, − 0.3).²⁴ However, few interventions have been able to deliver the recommended intensity, even in clinical programs designed to treat pediatric obesity.²⁵ Findings from this study show that the Fit Together model provides an avenue to implement IHBLT and achieve BMI reductions. Of note, few youths in this study met the at least 26-hour IHBLT recommendation, suggesting a need to further investigate the dose-response relationship and heterogeneity in treatment response within IHBLT interventions.

Findings in this study are similar to those of previous work demonstrating the effectiveness of engaging community organizations in the treatment of youth obesity. For instance, the Mind, Exercise, Nutrition, Do it (MEND) program engaged youths with obesity and their families in community-based physical activity and education, which yielded significant decreases in youth BMI in several trials.^26,27 MEND was adapted to the Healthy Weight and Your Child program that has been widely disseminated through YMCA organizations.^28,29 This suggests that the Fit Together model could potentially be scaled and sustained within community settings.

Although findings were in the expected direction, there was no significant effect of Fit Together on youth cardiorespiratory fitness. The most likely reason is insufficient power because of discontinuation of the fitness assessment during the COVID-19 pandemic. Additionally, it is likely that youths who experienced reductions in BMIp95 also made dietary changes that contributed to reductions in weight, which may explain the reduction in BMI without changes in fitness. In subgroup analyses, there was a significant difference in fitness among girls not affected by COVID-19 disruptions, which may be attributed to improved physical activity in a population that traditionally has lower physical activity; however, this should be interpreted with caution.^30,31 Given the cardiometabolic benefits of improved fitness, even in the absence of weight loss, it is necessary for future obesity treatment interventions to continue to explore the effect on youth fitness.³² 

Prior pediatric obesity treatment studies have often failed to reach priority populations, including marginalized racial and ethnic groups and low-wealth populations that experience a greater obesity burden.^33,34 In the current study, we were able to engage youths and families from predominantly priority populations, which may be attributed to the access provided by partnering with Durham P&R. However, subgroup analyses only revealed a significant effect among non-Hispanic white youths, suggesting a need to further understand the effect of this intervention among marginalized racial and ethnic groups.

This study had several limitations. First, COVID-19 safety measures affected both data collection and intervention delivery. Assessment of youth fitness was discontinued because of difficulties administering the protocol virtually, leading to missing data for 1 of the co-primary outcomes. The intervention was shifted to an online format, and although both synchronous and asynchronous options were available, engagement with the online program was low and likely did not match the intensity of in-person session. Additionally, although the Healthy Lifestyles–P&R partnership was feasible in this study, it may not be generalizable to primary care. Healthy Lifestyles is a specialized tertiary care obesity treatment clinic affiliated with an academic medical institution. However, other studies suggest that this model can be implemented in primary care settings.¹⁴ 

A partnership among health care and P&R providers demonstrated effectiveness in improving BMI within a program designed to deliver IHBLT for youths with obesity. Future research should seek to further understand the heterogeneity of treatment response with respect to BMI reduction and other key health outcomes as well as avenues for dissemination and the sustainability of this approach.

We would like to thank all the youths and families who participated in this project. Additionally, we would like to acknowledge Emily D’Agostino, Jessilyn Dunn, Janna Howard, Neha Pagidipati, Eric Peterson, and Charlene Wong for their contributions and insights.

AAP

American Academy of Pediatrics

BMI

body mass index

BMIp95

body mass index relative to the 95th percentile for age and sex

BPA

best-practice alert

CI

confidence interval

HR

heart rate

IHBLT

intensive health behavior and lifestyle treatment

MEND

Mind, Exercise, Nutrition, Do it

P&R

parks and recreation

PCP

primary care physician