Rationale and protocol for a cluster randomized, cross-over trial of recruitment methods of rural children in primary care clinics: A feasibility study of a pediatric weight control trial in the IDeA States Pediatric Clinical Trials Network

Abstract

A significant percentage of clinical trials fail due to poor recruitment. Despite this, few studies exist to evaluate clinical trial recruitment strategies using a randomized approach in any population, and none exist to test recruitment strategies for trials involving children or rural populations. For clinical trials focused on weight control, evaluating retention and dose are extremely important, as poor retention can lead to biased samples and existing research shows that dose (i.e. contact hours) is directly related to patient outcome. Finally, adequacy of blinding of assessment teams is rarely reported in pediatric trials, and unblinded staff may be more likely to inadvertently bias findings. Therefore, in this feasibility trial we aim to use rigorous clinical trial methodology to assess the effectiveness of two different recruitment strategies, as well as test retention, dose, and blinding. Specifically, we describe the rationale, design, and planned implementation of a feasibility study of a rural pediatric obesity treatment trial that will be implemented in four medical clinics in four states affiliated with the Environmental influences on Child Health Outcomes IDeA States Pediatric Clinical Trials Network (ECHO ISPCTN). The primary objective is to assess recruitment rate for consecutive recruitment (approaching recently seen eligible patients in consecutive order by date seen) compared to traditional recruitment (such as posters, flyers, tear-offs), as well as to assess retention, dose, and blinding. If successful, this trial will support the implementation of a large multi-state trial directed at addressing obesity in rural children and their families recruited from their primary care clinics.

1. Introduction:

The most common reason that clinical trials end without result is a failure of recruitment.[1, 2] Specifically, 50–63% of all trials fail to achieve recruitment targets or require extended recruitment periods, which can increase costs and/or leave studies underpowered to achieve their study goals.[3, 4] As such, there is increasing attention to methodological research on recruitment.[5–9] Despite this expanded interest, there are only rare examples in the literature of randomized trials of various recruitment strategies (randomization is a marker of scientific rigor), and none of these studies focus specifically on the best methods for recruiting children or the best methods for recruiting rural individuals. Further, no pilot studies in advance of larger trials have been published with the explicit aim of increasing the likelihood that a larger trial will recruit successfully.[3]

Children are a protected population[10] requiring parental consent, and in certain situations child assent, and undergo a higher level of regulatory scrutiny. There is research to indicate that schools may be good partners in recruiting children to some types of research,[11] but if parents need to participate as well, this can be a challenge to school based research.[12] Recruiting for clinical trials through primary care clinics as opposed to non-medical venues has the advantage of routine medical screenings and the expectation of receiving medical care.[13] When recruiting children through medical clinics, a frequently cited barrier is lack of clinician knowledge and experience in undertaking pediatric research.[14, 15]

Rural families also face many barriers and have lower participation in clinical trials.[16, 17] Challenges to recruiting rural populations are multifactorial, and difficulty in recruiting the clinics that care for rural patients may be at least part of the explanation for why rural participants are under-represented in clinical trials.[18] Some research indicates that engagement of local resources and providers as research sites can increase participation,[19] and that rural clinics may be more likely to engage in research in affiliation with a network. Specific to treatment outcome research, rural families may be more likely to see researchers as outsiders, and to believe researchers lack important information about rural culture.[20] In addition, community-level barriers in rural areas may also play a role such as population size and density, poverty, and lower educational attainment.[20]

The importance of being able to successfully conduct clinical trials in the intersection of these two difficult to recruit populations, rural populations and children, is highlighted by the health disparities they experience. Children living in rural areas have poorer health, difficulty with health care access, and increased mortality when compared to urban children.[21, 22] According to the 2015 National Survey of Children’s Health, children living in rural areas are more likely to have overweight and obesity when compared to urban children (38% vs. 30%),[23] although research indicates their caregivers may be less likely than their urban counterparts to perceive their children as having overweight/obesity.[24] Although there have been some attempts to address these health disparities through intervention studies, specifically in the area of pediatric obesity,[25, 26] recruitment into these trials has proven challenging and research has not explored how to improve the recruitment of these rural children and their families.

Other markers of methodological rigor are also of heightened importance in pediatric obesity trials. Regarding retention, research indicates that trials including components recommended by the Expert Committee Guidelines (such as including parents and children together and focusing on nutrition and physical activity) are more likely to face retention challenges.[27] Regarding dose, a recent review by the United States Preventive Services Task Force indicates that planned dose is directly related to treatment outcome, with trials including 26 planned contact hours or more consistently demonstrating mean reductions in weight as compared to usual care.[28] Other research indicates that adherence to these planned contact hours ranges from 60–80% in most pediatric trials.[29] Regarding blinding, only 26–40% of all clinical trials include blinding of assessors,[3, 30] and some data indicate there have been no trials conducted to date that actually test the blind.[31]

A goal of the Environmental Influences on Child Health Outcomes IDeA States Pediatric Clinical Trials Network (ECHO ISPCTN) is to involve rural children and their families in state-of-the-art clinical trials.[32, 33] The network was founded in 2016 in 17 states by the National Institutes of Health to provide medically underserved and rural populations access to state-of-the-art clinical trials.[34] Before a larger effectiveness or efficacy trial of a pediatric obesity intervention could be conducted in the ECHO ISPCTN, it was important to determine the ability of the network to recruit and retain children as participants through physician’s offices in rural areas.

The purpose of the current manuscript is to describe the design of a cluster randomized, cross-over trial of recruitment methods in the context of the feasibility of a rurally tailored[35] obesity treatment intervention (called iAmHealthy)[26, 36] in the ECHO ISPCTN. Specifically, our primary objectives focus on an assessment and comparison of two recruitment methods, as well as an assessment of retention, dose and staff blinding; secondary objectives focus on obesity related changes in child and parent health behavior. The goal of the feasibility trial in four clinics is to assess the feasibility of conducting a large, fully powered treatment-outcome trial of iAmHealthy. To our knowledge, this is the first randomized trial of recruitment options, as well as the first pilot study of recruitment and retention conducted in advance of a planned larger trial.[3]

2. Study Design

2.1. Primary Aims (Table 1)

Table 1.

Study Primary Aims.

1.	Participant Recruitment
	A. Evaluate the two proposed recruitment options to determine which are feasible for recruitment of rural participants
	B. Compare the two proposed recruitment options
2.	Participant Retention
	Evaluate the percentage of randomized participants who remain in the study through the final measurement time point (sixth month) in both arms (i.e., iAmHealthy behavioral intervention and newsletter only)
3.	Dose
	Evaluate the percentage of participants in the iAmHealthy behavioral intervention arm and retained through the final measurement who receive a sufficient dose (i.e., 80% [20.8 hours] of the 26 planned contact hours) of the intervention
4.	Staff Blinding
	Evaluate the agreement between the blinded assessor’s estimation of participant arm assignment and actual participant arm assignment

Objective 1. Participant Recruitment -

A) Evaluate the two proposed recruitment options to determine which are feasible for recruitment of rural participants. Hypothesis: The percent of randomized participants, among those contacted with each recruitment option, will be greater than 20%.

B) Compare the two proposed recruitment options. Hypothesis 1: The Consecutive recruitment option will yield a higher percent of participants randomized than those contacted through the Traditional recruitment option. Hypothesis 2: The Consecutive option will require less “time to full recruitment” than the Traditional option.

Objective 2. Participant Retention - Evaluate the percentage of randomized participants who remain in the study through the final measurement time point (sixth month) in both arms (i.e., iAmHealthy behavioral intervention and newsletter only). Hypothesis: The study will retain greater than 75% of the randomized participants through the final measurement (sixth month) in each arm.

Objective 3. Dose - Evaluate the percentage of participants in the iAmHealthy behavioral intervention arm and retained through the final measurement who receive a sufficient dose (i.e., 80% [20.8 hours] of the 26 planned contact hours) of the intervention. Hypothesis: 80% of the participants randomized to the iAmHealthy behavioral intervention arm who are retained through the final measurement will receive a sufficient dose (i.e., 80% [20.8 hours] of the 26 planned contact hours) of the intervention.

Objective 4. Staff Blinding - Evaluate the agreement between the blinded assessor’s estimation of participant arm assignment and actual participant arm assignment. Hypothesis: The study will achieve blinding in both arms, as determined by calculating the New Blinding Index score with a 95% confidence interval that includes 0.[37, 38]

Secondary Objectives

The secondary objectives are to obtain preliminary estimates on the effectiveness of the iAmHealthy behavioral intervention compared to the newsletter-only intervention with respect to changes from baseline to 6 months in 1) child BMIaz; 2) child BMI; 3) caregiver BMI; 4) child nutrition; and 5) moderate to vigorous physical activity in the child.

2.2. Participant recruitment

All methods are approved by the Institutional Review Board at the University of Arkansas for Medical Sciences through the SMART IRB system, with all other sites relying on UAMS as the central IRB. Prior to participation, parents provide full written consent, and children age 7 years or older provide full written assent.

2.2.1. Clinics

Of the 17 sites within the network, 9 sites responded affirmatively and submitted the necessary information to be considered for participation in the feasibility study. Information included the name and contact information for their participating rural clinic (one per site), as well as pediatric clinic volume and rurality of patients seen at the rural clinic (at least 500 patients who meet inclusion/exclusion criteria seen over the past 1 year). For the purposes of this study, rural is defined by the United States Department of Agriculture (USDA) Rural Urban Commuting Area (RUCA) codes (greater than or equal to 4). We will use the RUCA ZIP Code crosswalk version 3.1 available from the University of North Dakota to determine rurality.[39] Clinics were also to have space for recruitment/measures, as well as a locked cabinet to store research equipment. Based upon the submitted information, four clinic sites will be selected, as well as two “back-up” sites. The purpose of back-up sites is to be able to step in and participate should any of the initially selected sites not proceed through IRB and/or contracting on the timeline required. Set a priori, each clinic is required to recruit at least 16 participants for those participants to move forward to randomization.

2.2.2. Participants

Inclusion criteria are that participants must be 6–11-year-old children, have a BMI%ile≥85^th, live in a rural area (RUCA ≥ 4), and receive primary care at one of the four participating clinics. Also, both the child and parent(s) must speak English, and the family needs to be available at times the intervention is offered. Exclusion criteria include: children with a known physical limitation or who receive an injury which significantly limits physical mobility, children with a significant known medical issue (i.e. cancer), children and parents with significant developmental delay/cognitive impairment known to the clinic that could affect compliance, children already enrolled in a weight-loss trial, and children who have a sibling already enrolled in the program.

2.3. Recruitment Option/Clinic Randomization

Clinics will implement both recruitment options consecutively (4 weeks per option), with the order of recruitment options randomized (see Figure 1). Consecutive recruitment includes clinic personnel approaching potential participants based on past (retrospective) or current (prospective) appointment lists provided to the research team in consecutive order by date seen, as defined by prior research.[17] The consecutive retrospective appointment list includes patients seen over the past year that is then narrowed based upon age, child BMI percentile, and rurality (based on home zip code). The consecutive prospective appointment list includes patients coming in over the subsequent 1–2 weeks who met the same criteria. The Traditional recruitment strategy includes hanging flyers in clinic, distributing flyers at check-in, newsletters, social media and other methods focused on participant self-identification. As indicated above (section 2.2.2.) children are required to be existing patients at the clinic to participate. Clinic staff will ensure that components of the previous recruitment option are fully dismantled prior to implementing the successive recruitment option. This two-month randomized recruitment period will be followed by a planned two week “catch up” recruitment period, during which clinics will use their most successful recruitment options to attempt to reach their maximum allowable recruitment (n=32), resulting in a total of 10 weeks of planned recruitment.

Figure 1.

Study Design.

2.4. Consent/Assent and Baseline Period

Initial procedures include obtaining consent in person at the participating rural clinic by trained ISPCTN research staff at each participating ISPCTN site. Following proper consent and assent (if appropriate), a baseline period of one-month is planned. Baseline measures will include a demographic form, child height/weight, primary caregiver height/weight, a 3-day 24-hour dietary recall (child), and moderate to vigorous physical activity (child, via actigraph). These activities will be conducted at the rural clinic.

2.5. iAmHealthy Behavioral Intervention and Newsletter Intervention

Within each clinic site, families will be randomly assigned by the statistical team to either the iAmHealthy Behavioral Intervention arm or the Newsletter only arm.

2.5.1. The iAmHealthy Behavioral Intervention

is a family based behavioral group intervention designed to be delivered via interactive televideo to caregivers and children from rural areas.[26] The intervention has been tailored to rural families based upon previous qualitative work[40] to include topics such as child exercise without peers, eating at social/group gatherings, increased attention to self-esteem and decreased focus on eating fast-food. The intervention is composed of 26 contact hours; 15 hours of group sessions and 11 hours of individual sessions. Each group session contains up to 15 identified families from a participating clinic, who are randomly assigned to this arm. Groups meet weekly for 12 weeks, followed by monthly for 3 months, for a total intervention period of 6 months. iAmHealthy group sessions are led directly from a treatment manual and focus on nutrition, activity and behavioral parent training that has been described elsewhere.[26] The individual health coaching sessions focus on the same topics but allow the professionally trained team member to assist the family in implementing the curriculum as planned and assist in tailoring the content to their unique familial, cultural, financial, and regional needs. All intervention sessions are delivered remotely via a tablet provided to participants by the study team.

2.5.2. Interventionists and Fidelity

Interventionists for the group sessions are Ph.D. psychologists with experience in health behavior change intervention, and they are assisted by dietitians who also serve as interventionists for the individual health coaching sessions. Interventionists follow the iAmHealthy treatment manual for the group sessions and receive training with the PI prior to the start of their intervention group sessions. Treatment fidelity will be measured for 20% of sessions by comparing a digital video recording of the sessions to a checklist of 8–10 topics from the treatment manual. Both the group session leader and an independent coder will demonstrate inter-rater reliability during a training period in order to code for fidelity. Weekly supervision will also be provided to the entire intervention team by a senior psychologist with expertise in weight management and pediatric psychology.

2.5.3. Technology

Families receive tablets (iPad 7^th Gen 32GB WiFi+Cell) with data regardless of owning computers or iPads/tablets. This ensures uniformity of experience across users. These tablets are chosen to allow multipoint connections in order to display multiple sites simultaneously. The tablets utilize an institutionally vetted videoconferencing solution (Zoom Video Communications Inc, Overland Park, Kansas) that has been successful with previous home-based videoconferencing interventions with families.[26] The approach is also compliant with the Health Insurance Portability and Accountability Act (HIPAA) and allows the families to securely connect to the intervention team and other families in the convenience of their homes. Uniform data plans will be purchased for participants with tablets to ensure consistent connectivity; this choice was made due to variable connectivity across rural sites. Lost or broken tablets will be replaced if this occurs during the study period. At the end of the study, subjects will be required to return the tablets to researchers for use by future network studies.

2.5.4. Newsletters

Families assigned to either intervention group will receive a monthly newsletter focused on general child health topics, as developed by the American Academy of Pediatrics (healthychild.org). These newsletters will be mailed using USPS monthly for 6 months.

2.6. Primary Outcome Measures

2.6.1. Objective 1- Participant Recruitment Rate.

For each recruitment option, recruitment occurs over four weeks resulting in two values for each recruitment option: potential participants approached, and participants randomized. The numerator will be the number of participants randomized, and the denominator will be the number of participants approached. For Consecutive recruitment, the number for potential participants approached will be the number of potential participants approached by ECHO ISPCTN site coordinator. For Traditional recruitment, the potential participants will be the number of self-referred potential participants who contact a study or clinic representative.

2.6.2. Objective 2 – Participant Retention.

We will calculate participant retention as the percent of randomized participants who remain in the study through the final measurement time point. The numerator will be the number of randomized participants who complete the final measurement time point and the denominator will be the total number of randomized participants. We will then calculate the retention rate for each clinic site and by treatment arm across all clinic sites.

2.6.3. Objective 3 – Dose.

Among participants randomized to the iAmHealthy behavioral intervention arm who are retained through the final measurement, we will calculate the proportion of the participants who receive at least 80% of the planned intervention (dose). The planned dose is 26 contact hours, and 80% of the planned dose is 20.8 planned contact hours.

2.6.4. Objective 4 – Staff Blinding.

We will assess blinding with the New Blinding Index.^33,34 This index estimates the percentage of correct guessing beyond the level expected by chance in each treatment arm. Specifically, the ECHO ISPCTN site coordinator and a designated backup will remain blinded and serve as the blinded assessor. We will ask the blinded assessor at the time of the final measurements to enter into an EDC system her or his estimation of the treatment assignment for each participant. The options for assessors to select from will include treatment, control, and do not know (options required by the New Blinding Index).

2.7. Participant Measures

Unless noted otherwise, all participant-based measures are collected at baseline and post-treatment (6 months; Table 2).

Table 2.

Measurement Schedule.

	Baseline	6 months
Demographic information	X
Child height/weight	X	X
Primary caregiver height/weight	X	X
3-day 24-hour dietary recall (child)	X	X
Moderate to vigorous physical activity (child)	X	X

2.7.1. Demographic Information

Basic demographic information is captured via electronic data capture system and includes participant name, primary caregiver name, household income and other basic information.

2.7.2. Child height/weight

Child height and weight are obtained for screening (to ensure inclusion criteria are met) and again for baseline and post. Measures are taken in clinic on a SECA High Capacity Digital Flat Model 813 scale and with a Detecto Free-Standing Portable Height Rod stadiometer by trained research staff, in triplicate. These measures are used to calculate child BMIaz (body mass index adjusted z score[41]), and child BMI (weight in kilograms divided by the square of height in meters). While it is common to use BMI for age and sex z score (BMIz) in obesity projects, BMIz is not does not perform well for severe obesity. For that reason we opted to use BMIaz and BMI for this project.[41]

2.7.3. Primary caregiver height/weight

Parent height and weight are obtained in the same manner as child height/weight. These measures are used to calculate primary caregiver BMI (weight in kilograms divided by the square of height in meters).

2.7.4. 3-day 24-hour dietary recall

The 24hr food recall will be conducted via telephone one day at a time using a standardized five-pass method, developed by the US Department of Agriculture for use in national dietary surveillance. Although there are weaknesses to every method of dietary assessment, this approach was selected as it is widely used in large trials, which serve as models for the current project,[42–45] and data suggest it is the most valid and reliable method of dietary assessment for children.[46] We collect two weekdays and one weekend day at each assessment point for the participating child, as is recommended. The data are self-report and collected using highly standardized probes by fully trained research staff. Regarding school lunches eaten away from home, per typical procedures, parents review the school lunch menu with their child at the end of each day, having the child report to the best of their ability what they consumed. Recalls are analyzed with the Nutritional Data System for Research (version 2019; University of Minnesota, Minneapolis, MN). Although a plethora of information is available from this analysis, the current study specifically focuses on servings of sugar-sweetened beverages per day, number of “red” food items per day (foods with ≥7 grams of fat and/or ≥12 grams of sugar), and number of servings of fruits and vegetables per day, as calculated by the computer software.

2.7.5. Moderate to vigorous physical activity

For the current trial, child physical activity is assessed using the triaxial wGT3X-BTLE ActiGraph accelerometer (ActiGraph LLC, Pensacola, FL). The ActiGraph is a small, lightweight activity monitor that is worn on an adjustable belt over the non-dominant hip for seven consecutive days. The ActiGraph has been shown to provide valid assessments of physical activity, across a range of intensity levels, for adults and children during both laboratory (treadmill walking/running)[47, 48] and daily living activities.[49] Prior to data collection, participants and parents receive detailed instructions on wearing and caring for the monitor. Accelerometers are initialized to begin data collection at 12:00AM and raw data are sampled at 40 hertz. After the data collection period, participants will return the accelerometers via a prepaid envelope and data are downloaded and reintegrated to a 15 second epoch via ActiLife software prior to data processing. Data will be screened for non-wear using the Choi algorithm,[47] and participants with ≥8 hours per day on ≥4 days are retained for analysis. The primary outcome is average time (across valid wear days) spent per day in moderate to vigorous physical activity (MVPA), as well as % time spent in MVPA using Evenson cutpoints. Re-wears during the screening/baseline period will be allowed as long as they occurred during the data collection window. Planned procedures include distributing and collecting activity monitors at clinic visits, with supplemental exchange by United States Postal Service (USPS).

2.8. Analysis

2.8.1. Power analysis

We based the sample size on the primary efficacy endpoint, percentage of participants randomized among those contacted by recruitment option. ECHO ISPCTN site coordinators will approach and contact up to 560 potential participants (280 per recruitment option) to consent up to 224 participant and randomize up to 112 total participants. For each recruitment option, the target proportion of participants randomized among those contacted as 20%. To test the null hypothesis that the proportion randomized is 13% without implementing the recruitment strategy against the alternative hypothesis that the proportion will improve to 20% using the recruitment strategy required 280 participants at the two-sided 0.05 significance level with power of 0.90.

2.8.2. Quantitative Analysis Plan

We will first use the intent to treat population (all child/caregiver pairs who are randomized) to conduct all analyses, and we will consider this the primary set of analyses on which to base inferences. We will also conduct analyses with the per protocol population (child/caregiver pairs who completed at least 50% of the assigned intervention and had no major protocol deviations, defined as a protocol deviation that has a major impact on the participant’s rights, safety, or well-being; or the completeness, accuracy, and reliability of the study data). We will separately analyze each treatment arm, and we will summarize descriptive statistics for continuous data by mean and standard deviation or median and interquartile range, as appropriate. We will summarize categorical data by frequency and percent. We will investigate any outliers detected during data review, and we will define methods for handling outliers or data transformation in the statistical analysis plan.

Objective 1A. Participant Recruitment: Individual Recruitment Options.

We will separately calculate the percentage of participants randomized among those contacted with each recruitment option across sites. For participants contacted through more than one recruitment option, these participants will be analyzed based on the option in which consent occurred. For each recruitment option, we will use a one-sample test of proportions to test that the participant recruitment rate (i.e., percentage of participants randomized among those contacted) will be greater than 20% (alternative hypothesis) versus 13% (null hypothesis). We will also calculate corresponding 95% confidence intervals for each method.

Objective 1B. Participant Recruitment: Comparison of Recruitment Options.

We will compare the percent of participants randomized among those contacted between each recruitment option using a generalized linear mixed model (GLIMMIX) with a binomial distribution and logit link function. We will include a month and recruitment option as fixed effects and clinic as a random effect in the model. Using Traditional recruitment as the reference, we will also calculate odds ratios comparing recruitment options and their corresponding 95% confidence intervals. Comparison of time to full recruitment across the two recruitment options: Four clinics will implement the two recruitments options (in random order). Each clinic will recruit 14 participants per recruitment option. The time to full recruitment per recruitment option for each clinic is defined as the number of days from start of recruitment for first participant to screening and then consent, at which point diet and activity monitoring begin for last participant. Thus, for each clinic, we will have a paired set of time to full recruitment corresponding to the two recruitment options. The DCOC statistical team will use the Wilcoxon-signed rank test to compare the difference in time to full recruitment across the two recruitment options.

Objective 2. Participant Retention.

We will separately calculate the percentage of randomized participants who remain through the final measurement for each intervention arm (iAmHealthy behavioral intervention, newsletter only). For each intervention arm, we will use a one-sample test of proportions to test if the participant retention is greater than 75%. We will also calculate the corresponding 95% confidence intervals for each condition.

Objective 3. Dose.

We will calculate the percentage of families in the iAmHealthy behavioral intervention arm who receive a sufficient dose: at least 80% (20.8 planned contact hours) of the 26 planned contact hours. We will use a one-sample test of proportions to test if the percent of the families who receive this dose is greater than 80%. We will also calculate the corresponding 95% confidence intervals.

Objective 4. Staff Blinding.

We will measure blinding by evaluating the agreement between the blinded assessor’s estimation of participant assignment and actual participant assignment (e.g., iAmHealthy, newsletter only). To assess the agreement, a weighted kappa, we will conduct corresponding 95% confidence intervals and hypothesis test by using the Blinding Index, proposed by Bang et al (New Blinding Index).[50, 51] Hypothesis testing will be whether the 95% confidence interval excludes zero. We can estimate the treatment-specific New Blinding Index and corresponding variance for treatment (i) by using the following formulas:

$$\widehat{\mathit{n}\mathit{e}\mathit{w}}\mathit{B}{\mathit{I}}_{\mathit{i}}\mathrm{=}\mathrm{(}\mathrm{2}{\hat{\mathit{r}}}_{\mathit{i}|\mathit{i}}-\mathrm{1}\mathrm{)}\ast \mathrm{(}{\mathit{n}}_{\mathit{i}\mathrm{1}}\mathrm{+}{\mathit{n}}_{\mathit{i}\mathrm{2}}\mathrm{)}/\mathrm{(}{\mathit{n}}_{\mathit{i}\mathrm{1}}\mathrm{+}{\mathit{n}}_{\mathit{I}\mathrm{2}}\mathrm{+}{\mathit{n}}_{\mathit{I}\mathrm{3}}\mathrm{)},\text{for}\mathit{i}\mathrm{=}1,2$$

$$\mathit{V}\mathit{a}\mathit{r}\mathrm{(}\widehat{\mathit{n}\mathit{e}\mathit{w}}\mathit{B}{\mathit{I}}_{\mathit{i}}\mathrm{)}\mathrm{=}\{{\mathit{P}}_{\mathrm{1}|\mathit{i}}\mathrm{(}\mathrm{1}-{\mathit{P}}_{\mathrm{1}|\mathit{i}}\mathrm{)}\mathrm{+}{\mathit{P}}_{\mathrm{2}|\mathit{i}}\mathrm{(}\mathrm{1}-{\mathit{P}}_{\mathrm{2}|\mathit{i}}\mathrm{)}\mathrm{+}\mathrm{2}{\mathit{P}}_{\mathrm{1}|}{}_{\mathit{i}}\mathit{P}_{\mathrm{2}|\mathit{i}}\}/{\mathit{n}}_{\mathit{i}},\text{for}\mathit{i}\mathrm{=}1,\mathrm{2}.$$

The numerator, $\mathrm{(}\mathrm{2}{\hat{\mathit{r}}}_{\mathit{i}|\mathit{i}}-\mathrm{1}\mathrm{)}\ast \mathrm{(}{\mathit{n}}_{\mathit{i}\mathrm{1}}\mathrm{+}{n}_{i2})$, of the New Blinding Index estimates the number of people who guess the treatment correctly beyond chance level. The denominator considers the number of assessors who thought the participants were assigned to the iAmHealthy behavioral intervention, newsletter-only, or “I don’t know.” We will implement the calculations separately to answer the questions: 1) Was the treatment arm blinded? and 2) Was the placebo arm blinded? The following hypotheses will be tested: H₀: P_1|1 = P_2|1 versus H_A: P_1|1 > P_2|1 for the treatment arm and H₀: P_2|2 = P_1|2 versus H₀: P_2|2 > P_1|2 for the placebo arm, where P_J|i(i,j=1,2) is the “conditional” probability, and we can estimate it by ${\mathit{P}}_{\mathit{j}|\mathit{i}}=\raisebox{1ex}{${\mathit{n}}_{\mathit{i}\mathit{j}}$}\!\left/ \!\raisebox{-1ex}{${\mathit{n}}_i$}\right.$. P_j|i=P(guess j|assigned treatment i) for i=1 (iAmHealthy, newsletter-only) And j=1 (iAmHealthy), 2(newsletter), 3 (I don’t know). We will conduct this test by using the normal approximation to the binomial distribution. The New Blinding Index takes on a value between −1 and 1, where a null value of 0 indicates the most desirable situation under successful blinding. A positive value implies failure in masking above random accounting (i.e., majority of participants correctly guess child/caregiver pairs treatment allocation), and a negative value suggests success of masking or failure of masking in the other direction (i.e., more individuals mistakenly name the alternative treatment).

3. Conclusions

This paper describes a cluster randomized, cross-over trial of recruitment methods of rural children in pediatric primary care clinics, along with methods for studying retention, dose, and blinding. The current feasibility trial is designed to study these key factors prior to a larger, fully powered effectiveness or efficacy trial focused on treating pediatric obesity among rural children and their families. The work represents the first randomized trial of recruitment methods in pediatrics. The methodological focus on comparing recruitment approaches specifically in the rural pediatric population will support the best recruitment options for the future effectiveness study (and other pediatric rural studies like it) moving forward. The current trial also extends this feasibility work with a focus on participant retention, participant dose, and staff blinding in these rural children and their families. The results of this iAmHealthy ECHO ISPCTN feasibility trial may support the conduct of a future network wide study which could significantly impact the way we care for rural children and families with obesity.