The study is a 3-arm RCT with adolescents with uncontrolled asthma (Figure 1). The study will be conducted at the Arkansas Children’s Hospital (ACH), with children receiving asthma care at ACH primary and specialty care clinics. A sample of 370 adolescents (age 12-20 years) and their caregivers will be enrolled. Participants will be randomized 1:1:1 to 1 of 3 groups: (1) a group using PEAKmAAP alone, (2) a group using PEAKmAAP with data sharing (PEAKmAAP-DS), and (3) an eUC group using an app with nonasthma daily reminders. Randomization will be stratified according to 2 strata based on asthma severity status (mild persistent and moderate-to-severe persistent) in order to minimize differences in baseline severity that may potentially bias treatment effects. A 1:1:1 permuted block randomization scheme with varying block sizes will be independently generated by our biostatisticians using nQuery Advisor 7.0 within each stratum to conceal allocation. The randomization list will be kept independent of the research staff to ensure allocation concealment.

The study is registered at ClinicalTrials.gov (NCT03842033). SPIRIT (Standard Protocol Items: Recommendations for Interventional Trials) reporting guidelines were used to create this manuscript (Multimedia Appendix 1) [25]. Adherence of this report to the SPIRIT guidelines was supported by a tool developed under R01LM014079.

The inclusion criteria are as follows: (1) age ≥12 and ≤20 years; (2) physician-diagnosed persistent asthma using national guidelines criteria [2]; (3) uncontrolled symptoms, as measured by an ACT score of ≤19 [26,27]; (4) prescription of a preventative (controller) asthma medication in the 6 months prior to enrollment; (5) access to a smartphone compatible with the PEAKmAAP app; and (6) access to the internet.

The exclusion criteria are as follows: (1) any significant underlying respiratory diseases other than asthma, (2) current smoker, (3) moderate-to-severe developmental delay that interferes with the participant’s ability to self-monitor their asthma, and (4) participation in any asthma clinical trials in the 6 months prior to enrollment.

The study protocol was reviewed and approved by the Institutional Review Board (IRB) of the University of Arkansas for Medical Sciences (UAMS; protocol #206110). All protocol modifications will be approved by the IRB.

Informed consent and assent will be obtained from caregivers and participants, respectively. All participants will be provided with an opportunity to opt out of any study procedures or withdraw from the study at any time.

The principal investigator (PI) will carefully monitor study procedures to protect the safety of research participants, the quality of the data, and the integrity of the study. The PI and unblinded study staff will have access to the data throughout the project. Each participant will be assigned a unique ID. Only study staff will have access to the IDs and information that identify the participants in this study. The electronic and hard-copy files with the participant IDs and study information will be kept separate and stored in a locked office at Arkansas Children’s Research Institute (ACRI) and will be accessible to research personnel only. Only deidentified data will be used for data analysis. At the end of the study, the records will be destroyed in compliance with ACRI and UAMS research policies and guidelines. Electronic records will be stored on the ACRI/UAMS servers, which are password-protected.

To ensure confidentiality, several features have been included in the PEAKmAAP and Nutrimap apps to protect participants’ health information in the event their peers look at their phones:

Each participant will receive a US $50 incentive for completing the baseline and 3-month visits and a US $100 incentive for completion of the 12-month visit. Participants will also receive US $15 for completion of each monthly survey and up to US $15 each month for daily logging on PEAKmAAP or Nutrimap (US $0.75 per day up to a maximum of US $15 per month). Participants completing all study visits and earning the maximum amount for daily logging will receive a total of US $560 for 12-month study participation. Participants will be offered transportation vouchers, if needed, to travel to the ACH for any in-person visits at baseline, 3 months, or 12 months.

Prior to the study’s start, the research team convened a Community Advisory Board (CAB) consisting of adolescents with asthma and caregivers of adolescents with asthma. The CAB partnered with the research team and met monthly over a 12-month period to review the findings of the pilot studies [21,22,24] and provide input in an iterative user-centered design process. The CAB conducted beta testing with the PEAKmAAP prototype and provided feedback to the research team regarding functionality and advised the team on the final app’s design prior to participant recruitment. The CAB also provided content ideas and feedback on educational videos and asthma tips sent via push notifications from the PEAKmAAP app.

The intervention is based on the Chronic Care Model (CCM) [28]. This model was chosen because it relates specifically to chronic illness management, and we developed our intervention to include components that foster improved self-management, integrate guideline-based principals into daily life, and support productive interactions between patients and health care teams [29]. CCM components have been used in many interventions, with improved outcomes among patients with asthma [30,31]. Including multiple components of the CCM is associated with stronger effects on medication adherence outcomes among patients with asthma [32]. Our intervention incorporates 3 components of the CCM (Table 1) based on input from the CAB in a method that is convenient, acceptable, and interactive with adolescents on a daily basis.

Medication nonadherence is the primary barrier to asthma control among adolescents [10,11]. Medication adherence is a complex behavior, and behavior change is most likely to occur when a patient is motivated and feels confident in their ability to change [33]. The latter notion refers to self-efficacy. The intervention incorporates the Health Belief Model (HBM) to affect behavior change, for example, increase self-efficacy and increase medication adherence (Figure 2). The HBM is a commonly used explanatory theory of health behavior that has been applied to adolescents [34,35] and specifically to beliefs about medication adherence and asthma management among those with asthma [11].

All enrolled participants will receive care from their usual primary care or specialty care provider at the ACH. At baseline, all participants will receive a printed paper copy of their personalized AAP from the electronic health record (EHR), as prescribed by their health care provider. Trained research personnel will provide each participant with training on how to use their AAP according to their health care provider’s instructions. All participants will use a smartphone app during the 12-month enrollment (details described later).

We will use the Research Electronic Data Capture (REDCap) program. Data from the study will be checked for errors, inconsistencies, and missing data. We will examine the data graphically using scatterplots for continuous outcomes and frequency tables for categorical data. Unresolved data anomalies will be sent to the primary investigator for verification. Obvious outliers due to data entry errors will be corrected before data analysis begins. Different data sources will be linked using a unique identifier assigned to each participant.

All subjects will be monitored for AEs related to study procedures. Study investigators acknowledge that participants may experience asthma symptoms or exacerbations throughout the course of the study; however, exacerbations are expected due to the nature of the disease and not due to study-related procedures. Due to an intrinsic exacerbation risk to all patients with asthma, we will advise participants to continue to use medications as prescribed by their PCP and to keep their regularly scheduled visits. The research team will also conduct research review meetings on a weekly basis to monitor the progress of the study, ensure participant safety, discuss any unanticipated problems, ensure treatment integrity, and ensure that recruitment and study goals are being met, as well as solve any issues pertaining to enrollment, recruitment, or completion of study procedures. A Safety Monitoring Committee (SMC) will independently monitor participant safety. The SMC, consisting of a biostatistician and an independent clinical research investigator, will meet no less than once a year and as needed to discuss any potential safety concerns related to the research. All participants will be screened for AEs and serious adverse events (SAEs) at each study visit. They will be asked about asthma- and non–asthma-related health issues. All clinically significant AEs will be brought to the attention of the PI, who will categorize the AEs as “expected” or “unexpected” and as “likely” or “unlikely” to be related to the study. The PI will monitor all related AEs and SAEs. All AEs and SAEs will be entered into the study database and, as required, reported to the University of Arkansas for Medical Sciences Institutional Review Board, the SMC, or the National Institutes of Health (NIH). Because the interventions are adjuncts to ongoing clinical care and all participants are encouraged to continue to engage with their PCP, no formal ancillary or posttrial care is planned. Similarly, given the low risk of the interventions and continued PCP care, there are no criteria for discontinuing or modifying allocated interventions beyond a participant’s personal decision to discontinue participation in the study.

Primary models using the intent-to-treat (ITT) principle will be performed for all primary and secondary outcomes. All subjects will be considered part of their assigned treatment groups irrespective of their adherence to treatment, and all randomized subjects will be included in the analysis. Secondary models using a per protocol (PP) population including only participants who adhered to the protocol will be conducted. As appropriate, models will be adjusted for the baseline value of the outcome, with other baseline confounding or covariate variables included. Secondary subgroup analyses will stratify by baseline asthma control classification (poorly controlled vs not well controlled) defined by ACT scores. Holm-Bonferroni correction will be used to control the overall familywise error rate under 5% for analyses with multiple comparisons. Missing data in primary and secondary models will be handled by multiple imputation if “missing at random” is a reasonable assumption. Results will also be reported discarding missing values as a sensitivity analysis.

Using these models, the primary outcome of the proportion of participants with ACT scores of >19 at 3 months will be analyzed using linear models with a logit link function and contrasts to compare the 2 PEAKmAAP groups combined with the eUC group. Secondary analyses of the proportion of ACT scores of >19 at additional timepoints (6, 9, 12 months) will be assessed with contrasts for the group effect (to measure the average effect across the different timepoints) and the group-by-time interaction effect (to measure the change in effect over time) between groups (PEAKmAAP-DS vs PEAKmAAP; PEAKmAAP groups combined vs eUC). Since the ACT score will be measured at 3, 6, 9, and 12 months (all participants will have baseline scores of ≤19), these correlated repeated measures data will be analyzed with a binomial distribution and a logit link function for modeling with an appropriate correlation structure (eg, compound symmetry or first-order autoregressive). Additionally, minimally important improvements in ACT scores (3-point increase) and CASI (1-point decrease) will be similarly modeled as binary outcomes across timepoints, contrasting eUC versus the 2 PEAKmAAP groups combined or PEAKmAAP versus PEAKmAAP-DS. Other secondary outcomes (health care use, rescue medication use, lung function) will be analyzed using similar models for nonnormal data using appropriate distributions and link functions.

Medication adherence rates at 12 months will be compared using a logistic regression model after adjusting for baseline values as a covariate, contrasting eUC against the 2 PEAKmAAP groups. Mean asthma self-efficacy scores at 3 and 12 months, including the subscales for prevention and management, will be compared among the 3 groups using the analysis of covariance model after adjusting for baseline values as a covariate in the model. The proportion of adolescents who receive step-up therapy will be compared between the 2 PEAKmAAP groups, and the 2 PEAKmAAP groups together will be compared with the eUC across different timepoints using a similar logistic model.

No interim analyses are planned, and thus, there is no planned stopping rule for the study.

The sample size was determined to detect a clinically meaningful difference between PEAKmAAP and PEAKmAAP-DS groups combined versus the eUC group for the primary outcome measure, which is the proportion of adolescents with ACT scores of >19 at 3 months. We estimate that the proportion of participants with ACT scores of >19 at 3 months will be approximately 15% higher in the PEAKmAAP and PEAKmAAP-DS groups compared to the eUC group based on pilot work [24,40]. Using a generalized estimating equations (GEE) model for binary longitudinal data, we estimated that a sample size of 107 participants in each group will yield at least 80% power to detect the difference mentioned before. After adjusting for a 15% attrition rate, the final sample size needed will be 124 in each group (372 total). The sample size will also have at least 80% power to detect a change from a baseline of 4 between the eUC group and PEAKmAAP and PEAKmAAP-DS groups combined for the asthma self-efficacy score. The SD of the asthma self-efficacy score was determined to be 7 [21,22]. The sample size determination was performed using the GEESIZE macro in SAS Software and PASS [46-48]. All sample size calculations assumed a 2-sided significance level of 5%.

Adolescents with asthma are at increased risk for adverse health outcomes related to asthma and are a group traditionally difficult to engage in self-management strategies for chronic illnesses. Smartphone use is prevalent and has been cited as a preferred and convenient method of communication among adolescents. mHealth delivered via smartphones can provide health education, medication tracking and logging, and chronic disease management integration into the daily lives of patients with asthma [49]. The current literature examining the role of mHealth apps for adolescent asthma is limited. Most studies are small pilot trials with a small number of participants or a short study period. We will test PEAKmAAP’s effectiveness among a large cohort of high-risk adolescents with uncontrolled asthma over a 12-month period. Based on the CCM and by virtue of its convenience, ability to provide immediate feedback, and ease of integration into a daily routine, we will examine the PEAKmAAP intervention as a tool to improve asthma self-management, improve medication adherence, and reduce asthma-related morbidity among adolescents at risk for future exacerbations. Our central hypothesis is that the PEAKmAAP and PEAKmAAP-DS groups will have reduced asthma morbidity compared to the eUC group. We further hypothesize that PCP data sharing will facilitate sustained asthma control and that significantly more adolescents in the PEAKmAAP-DS group will have continued asthma control at follow-up visits compared to PEAKmAAP alone and the eUC group. This novel approach has a high impact potential as it may be applicable to other age groups and patients with other chronic medical conditions.