Randomization
Two YMCA sites, [A] and [B], were randomly assigned to face-to-face versus hybrid intervention formats using a random number table. Based on alphabetical order and an even number selected, [A] was assigned to receive the face-to-face format and [B] was assigned the hybrid format.
Data collection
Anthropometric, behavioral, psychosocial and physiological outcomes will be evaluated at baseline, immediately following the 12-week intervention, and at follow-up (24 weeks). At each of these time points, weight will be measured using an electronic calibrated scale (Seca 876, Chino, CA) and rounded to the nearest 0.1 kg with participants wearing light clothing and no shoes. Height will be measured by a portable stadiometer (ShorrBoard, Olney, MD) and rounded19 to the nearest 0.1 cm. Waist circumference, linked to metabolic syndrome in children [
34], will be measured at the umbilicus using standard protocols [
35]. All anthropometric measures will be taken in duplicate and averaged. Lacking a gold standard for measuring change in weight status in children, we will use the recommended change in percentage overweight [
36] calculated as percentage over the median BMI for age and gender. Child dietary intake will be assessed using two non-consecutive, 24-hour dietary recalls collected by trained nutritionists and entered into Nutrient Data System for Research (Minneapolis, MN, v. 2012) [
37] Changes in physical activity will be measured over a 7-day period using Actigraph GT3X accelerometers (Actigraph, Pensacola, FL); raw accelerometer counts will be processed and analyzed using youth-specific cut-points [
38].
Psychosocial measures will be evaluated to assess mediators of behavior change using validated questionnaires for youth including self-efficacy related to nutrition [
39,
40] and physical activity [
41], perceived competence in maintaining a healthy diet and exercising regularly [
42], and perceived parental support [
43]. Parents/caregivers will complete the Family Nutrition and Physical Activity Screening Tool [
44] designed to assess environmental support for healthy eating and activity (e.g. provision of nutritious food and opportunities to be physically active), and will also self-report food parenting practices using the Comprehensive Feeding Practices Questionnaire [
45]).
Physiological outcomes will be assessed through 12-hour fasting insulin, glucose, and lipids drawn at J2 Laboratories, Tucson, Arizona, by a trained phlebotomist and placed into serum separator vacuum tubes. Insulin resistance (IR) will be assessed using HOMA [
46], shown to more sensitive for identifying youth with metabolic dysregulation than an impaired fasting glucose threshold [
47]. Calculation of IR is based on modeling of fasting insulin and glucose concentrations using the formula: fasting insulin (μU/ml) × fasting glucose (mmol/L)/22.5 = HOMA. HOMA correlates closely to IR as measured by euglycemic clamp [
48], and a HOMA insulin resistance value of 2.6 will be used as the upper limit of normal [
49]. Blood pressure will be measured on the non-dominant arm using an automatic monitor (Omron HBP-1300, Hoofddorp, The Netherlands) and cuff-sizes appropriate for the mid-upper arm circumference. Systolic and diastolic blood pressure will be measured two times with 5 min intervals while the participant is seated and resting. The definition of hypertension will be adjusted for age per national guidelines [
50]. Individuals with fasting glucose values >126 mg/dL and no previous history of T2D will be referred to their primary care provider. The study medical director will be immediately informed and arrange appropriate follow-up depending on the level of blood glucose elevation and symptoms. Maturity will be assessed using Tanner’s validated self-report questionnaire which presents illustrations of developmental stages shown to agree with pubertal staging by a physician [
51,
52].
Feasibility data will include recruitment, enrollment, and retention rates, program attendance and engagement, and delivery costs. Program satisfaction will be self-reported by participants (parents and children) using brief surveys that inquire about relevance of content, promoters and barriers to attendance and engagement, and degree to which the family applied the intervention to lifestyle behavior changes. Adherence data will include attendance logs and time spent with mobile content (hybrid condition), and completion of self-monitoring and goal-setting activities (both conditions). Importantly, our mobile infrastructure [
28] permits capture of engagement with mobile lessons including frequency and duration of participant log-ins, communication with lifestyle coaches, and content downloads/views, which will allow us to characterize engagement and associated it with adherence metrics.
Data collection
Data collection and entry, and statistical analyses will be conducted by research staff who are blinded to treatment allocation. Participants will be assigned a study identification number upon enrollment. Study personnel involved in data collection will follow a strict written protocol that describes study measures for protecting confidentiality and privacy. All aspects of data collection and storage will be carefully monitored to ensure rapid detection of errors, inconsistencies or other problems. Data will be double entered with third party verification to insure data integrity, and kept in locked files and on a dedicated study computer and secure network. Access will be restricted to the principal investigator and designated staff. Files with information linking names and other personal data to participant identification numbers will be deleted at the end of the study, and written copies will be shredded.
Statistical analysis
Outcome measures will be stratified by delivery format to assess whether the mobile technology-supported delivery of the intervention is as efficacious as the face-to-face format. Descriptive statistics for change in endpoints (mean, median, and standard deviation) between baseline and 12-week measurements will be computed separately for each program format (face-to-face or hybrid mobile technology-supported program). The statistical significance of the change at 12 weeks will be assessed using a paired t-test combining both formats. One-sided statistical tests will be used, as the interventions approaches are of interest only if they decrease the percentage overweight and lead to improvements in diabetes risk factors. Statistical significance is set at alpha = 0.05. Sample size constraints do not allow adequate statistical power to formally test for differences between the two formats. However, exploratory models using linear regression analysis will estimate the magnitude of the difference between the face-to-face and combined formats, adjusted for baseline levels and the potential correlation between the participants within a given YMCA (expected to be small). These parameter estimates will provide effect sizes to appropriately power a larger study comparing the two formats. Since this is a randomized comparison, confounding is not an issue as long as there is non-differential dropout between the two formats. Differential dropout between the two formats will be assessed using a Fisher’s Exact Test. Similar analyses will assess whether improvements are maintained after the intervention ends, by comparing the 24-week versus 12-week values. We acknowledge the potential for differential loss to follow-up and missing data. Baseline values of those who complete the study versus those who do not will be compared using two sample independent t tests and Fisher’s Exact Tests, as appropriate. Additional sensitivity analyses will use multiple imputation for missing values and an intent-to-treat analysis with baseline values carried forward.
Anticipated results
We expect four to seven percent reduction in the percent of overweight youth after 12 weeks, similar to behavioral interventions in clinics [
5,
53,
54]; we also anticipate significant reduction in waist circumference over 12 weeks. Expected behavioral outcomes in youth include reduced energy-dense food intake including less sugary-sweetened beverages, increased intake of vegetable and whole grain servings, increased daily moderate-to-vigorous physical activity, reduced sedentary media screen time, and improved sleep time (i.e. closer to suggested amounts of 9 h per night). Child self-efficacy, perceived competence, perceived parental support, and intrinsic motivation is also expected to improve, as is parental support of environmental changes in the home related to nutrition and physical activity and use of effective food parenting practices. Reductions in insulin resistance as measured by homeostasis model assessment (HOMA) and maintenance of weight loss/healthy weight trajectories after 24 weeks are anticipated [
55]. Furthermore, evidence suggests mobile devices can support health behavior change [
56,
57], and frequent interaction (made more feasible with mobile tools) between program and participant is a positive predictor of adherence [
58]. Thus, we contend mobile devices provide similar opportunities for engagement and contact as face-to-face programs while reducing the burden associated with weekly attendance. We, therefore, expect similar results using a hybrid delivery (60 % face-to-face, 40 % mobile delivery) to be comparable to the traditional face-to-face instructor-led program.
Process evaluation and monitoring
The intervention has an a priori focus on scalability, replication, dissemination, and sustainability. Process evaluation and monitoring plans were created using the RE-AIM framework, a tool to assess the public health impact of health promotion interventions as described by Glasgow et al. [
59]. RE-AIM consists of five dimensions: reach, efficacy, adoption, implementation, and maintenance. These data will be collected using a combination of qualitative and quantitative measures and will involve our community partners (YMCA administrators and staff), the YMCA Lifestyle Coaches (delivering the intervention), and study participants. (Table
3) We will assess reach, impact, and engagement of participants and their families, and the feasibility of delivering the program using the YMCA infrastructure to inform future adoption, implementation and maintenance of the program. Efficacy will be measured both as the main study outcomes (anthropometric, behavioral and physiological, feasibility and acceptability, and attrition) and as the fidelity of the delivery of the intervention (e.g. training of and observations of curriculum delivery from Lifestyle Coaches). Quality assurance will occur with structured training and certification of YMCA lifestyle coaches, who will have continued access to the training team to discuss issues related to content or group moderation. The research team will review weekly session logs from YMCA instructors for potential departure from the EPIC Kids model, and conduct repeated observations of intervention sessions. Study staff trained in qualitative procedures will conduct interviews with YMCA staff and administrators to assess needs and resources for the program and for the YMCA at the program mid-point (November 2015). Engagement with administrators through interviews and surveys will inform activities related to the local adoption, implementation, and maintenance of EPIC Kids such as participant recruitment, branding and licensing, anticipated costs and resources associated with coach training, and program implementation and evaluation beyond the life of the grant.
Table 3
Process evaluation and monitoring plan
Reach | Screening form | • Inquiries • Non-eligible • Recruitment/referral source | Screening |
Baseline questionnaire | • Participant demographics • New or past YMCA members • Motivation to join • Willingness to participate (distance to travel) • Competing programs | Enrollment |
Efficacy | Lifestyle coach training survey & focus group | • Prior experience as a Lifestyle/DPPa Coach • Comprehension of material • Knowledge and skills acquired • Self-efficacy/preparedness • Perceived barriers/ potential solutions • Motivation to lead • Acceptability (likes/ dislikes) | Pre-study training (1 time) |
Telephone debriefing with lifestyle coaches | • Ease of implementation • Barriers/potential solutions • Perceived participation • Time management | After first lesson (1 time) |
Lifestyle coach weekly survey | • Curriculum comprehension • Ease of implementation • Fidelity of lesson (delivered as written & on time) • Perceived self-efficacy • Motivation to lead again • Perceived participant participation • Perceived barriers/solutions • Acceptability (likes/dislikes) | Weekly after each lesson (11 times) |
Attendance record | • Adherence (in-person) • Mobile data analytics (adherence for mobile-device delivered group) | Weekly (12 times) |
Participant weekly questionnaire (Child, parent) | • Comprehension • New knowledge • Usefulness • Motivation • Likelihood to adopt (short & long-term) • Enjoyment • Acceptability (likes/dislikes) • Suggestions for improvement | Weekly after each lesson (11 times) |
UA staff observation rubric | • Study fidelity (delivered as written and on time) • Participant engagement (quantity & quality) • Instructor communication/organization • Instructor classroom management • Instructor support of student needs | 4x each study site |
Lifestyle coaches final survey | • Perceived efficacy of intervention • Desire to continue program • Average preparation time • Time substitution (substituted activity if not involved in the intervention delivery) • Enjoyment & acceptability (likes/dislikes) | End of program (1 time) |
Participant final survey (Child and parent) | • Perceived effectiveness of instructor • Adoption of lifestyle changes due to program • Perceived long-term maintenance of changes • Likelihood to recommend program & perceived effectiveness of intervention for others • Use of YMCA for other purposes • Likelihood of future YMCA use • Cost of physical activity or food items due to interventionb
• Time substitution (activity substituted if they were not in the intervention)b
| End of program (1 time) |
Study outcomes | • Anthropometric (BMI-percentile, waist circumference) • Behavioral (24-hr dietary recalls, accelerometry, psychosocial (e.g., self-efficacy) • Physiological (fasting insulin & glucose, blood lipids & pressure) | Pre-post study and follow-up 12 weeks later (3 times) |
N/A – program data | • Cost • Lifestyle Coach and Research Staff Training requirements (time, effort, effectiveness) • Recruitment source success rate • Quality assurance/control | Throughout intervention |
Potential adoption, maintenance and implementation by YMCAc
| YMCA administrator interview | • Likelihood of adoption • Perceived needs to implement and maintain • Suggestions for improvement | Perceived efficacy of intervention |