The type and intensity of physical activities performed by the veteran population changes considerably from active duty to postdeployment. This is often because of the relatively unstructured nature of postdeployment life, as well as service-connected illnesses or injuries [1]. Younger veterans involved in the Afghanistan and Iraq conflicts—Operation Enduring Freedom (OEF), Operation Iraqi Freedom (OIF), and Operation New Dawn (OND; OEF/OIF/OND)—may also have work-life balance issues related to childcare and elder care issues and additional civilian reintegration issues related to high physical and mental health burdens, which may present unique challenges to being more physically active (eg, chronic pain, mental illness, and substance abuse) [2,3]. The transition from being physically fit and active to being more sedentary can lead to rapid weight gain and an increased risk of adverse health outcomes including diabetes, heart disease, joint disorders, and some cancers. This rapid weight gain was documented in a prospective study by Litman et al (2013), where it was found that there is an increased rate of weight gain in veterans around the time of military discharge [4]. In a large cohort of OEF/OIF/OND veterans seen in the Veteran Health Administration (VHA) postdeployment, 65.8% of men and 46.7% of women were overweight or obese at their first visit [5]. Early intervention with individually tailored lifestyle programs targeting physical activity and prevention of weight gain has the potential to sustain the high level of fitness required for military service well past active-duty status and improve mental and physical health.

A potential strategy for increasing access to lifestyle modification programs for OEF/OIF/OND veterans and more technology savvy patients is through the use of mobile health (mHealth), which refers to the use of mobile computing, wearable sensors, and communication devices for the provision of health services and information [6]. Although mHealth apps are still emerging and evolving, there exists a solid foundation demonstrating that internet-mediated interventions [7], particularly when combined with the use of wearable physical activity monitors, tailored motivational messaging, and online coaching, can increase physical activity and have the potential to improve health outcomes [8-10]. Through leveraging the ubiquity of mobile devices, we can easily increase access to these types of interventions [6,11,12]. mHealth interventions have been built on findings from internet-mediated intervention trials and have likewise been shown to be effective in chronic disease self-management and promoting behavior change [13-22]; however, the evidence base for mHealth interventions is largely made up of small trials with short-term follow-up [6,18], and long-term engagement with mHealth programs is not always sustained [23,24]. A potential strategy for improving upon mHealth interventions is through the addition of health coaching, including telephone-based lifestyle coaching delivered by human coaches. Health coaching is a patient-centered, collaborative model grounded in theories of health behavior change, in which coaches work in partnership with patients to identify goals and action plans that maximize personal well-being and overall health. Many coaching interventions use techniques like motivational interviewing, goal setting, and problem-solving as key strategies. Across a wide variety of populations, health coaching has produced positive impacts on lifestyle modifications [25-27]. Adding automated coaching features, such as personalized messaging addressing barriers and motivators, as well as goal setting based on past outcomes gleaned from wearable devices, may further enhance the impact of health coaching [28].

We aim to evaluate the effect of adding automated (ie, automated tailored messaging and automated goal setting) and human telephone health coaching to Stay Strong, an mHealth app that seeks to improve physical activity in OEF/OIF/OND US veterans over 1 year. Stay Strong is a mobile app that is tailored to these veterans and tracks physical activity monitored by Fitbit Charge 2 devices and weight measured on a Bluetooth-enabled scale. Captured data are presented visually to participants to facilitate behavior change. We also aim to further assess the impact on the secondary outcomes of weight loss, depression, patient activation, and pain, as well as to test for potential moderating effects of individual characteristics (eg, demographics and familiarity with technology). We hypothesize that our enhanced version of the intervention Stay Strong + Coaching will result in greater improvement in physical activity (primary outcome) at 12 months compared with the base Stay Strong program.

This study is a randomized trial to test the effect of adding coaching (ie, automated tailored messaging, automated goal setting, and telephone calls with a health coach) to an mHealth intervention to improve and sustain levels of physical activity among a national sample of US veterans over 12 months. In this study, participants are randomized to receive either the base intervention (Stay Strong) or an enhanced intervention that includes coaching features (Stay Strong + Coaching). The decision to incorporate an active control group was guided by several considerations. First, there is sufficient evidence in the literature to suggest that this type of technology-mediated intervention for promoting increased physical activity is effective [8-10], even in a veteran population [29,30]. As such, we felt justified in not including a true usual-care control. Furthermore, physical activity tracking devices are highly ubiquitous in the general US adult population. On the one hand, the use of a wearable device by participants assigned to the control group would confound results. On the other hand, we did not believe it would be ethical to ask control-group participants to refrain from using one over the course of a 12-month intervention period. This study was reviewed and approved by the VHA Central institutional review board (IRB)–VHA IRB for multisite studies.

Stay Strong is a multicomponent mHealth physical activity intervention for iOS and Android platforms that supports physical activity and weight self-monitoring. Vibrent Health, a digital health company in Fairfax, VA, was engaged to develop the app, online portal, and server platforms necessary to support Stay Strong [31] (see Multimedia Appendix 1 for screenshots of the Stay Strong interface). The study team defined the functional system requirements to Vibrent, and then the team and Vibrent collaborated on the user-interface design. Feedback on the design was gathered from a convenience group of testers employed at the VA, several of whom were OEF/OIF/OND veterans, and the approach of utilizing a wearable activity tracker was endorsed by the target population in some of our previous work [32].

Figure 1 shows the theoretical framework underlying the design of the intervention. The theoretical orientation of Stay Strong is informed by the information-motivation-behavioral skills (IMB) model and self-regulatory theory [33-36]. Both models have been successfully applied to physical activity and describe processes of behavior change mediated through goal attainment and skill mastery, and both models acknowledge the central role that self-efficacy plays in sustained behavior change [37-42]. The IMB hypothesizes that cognitive and behavioral skills are a prerequisite for any health behavior change like increasing physical activity. However, skills are the product of information that is relevant to health problems and personal and social support for activation to change behavior. Thus, information interacts with activation to enhance self-efficacy and build skill mastery to facilitate sustained behavior change. Informed by the pilot data and the IMB, Stay Strong + Coaching will boost the positive reinforcing relationship between self-monitoring and feedback with goal setting and activation through a coaching support component. Human coaches will enhance activation and facilitate goal setting by (1) providing social support, (2) helping to develop specific plans of action, (3) helping with problem solving to avoid or mitigate barriers to goal achievement, and (4) enhancing engagement with the Stay Strong app by helping to interpret the individuals’ physical activity data and using it to monitor progress toward their goals. Automated coaching will further strengthen these supports by (1) providing personalized, realistic physical activity goals, grounded in the objective physical activity monitoring provided through the Fitbit Charge 2 device, (2) sending personalized messages to help overcome barriers to physical activity, and (3) sending educational and motivational messages via the Stay Strong app.

The theory of self-regulation also informs the core components of Stay Strong + Coaching. In self-regulation, individuals participate in self-directed behaviors. These self-directed behaviors are hypothesized to be managed through a dynamic feedback loop in which individuals’ self-monitoring and feedback about their past behavior are integrated into their goals and activation to change future behaviors [43]. Thus, aligned with IMB, self-regulation involves both cognitive and behavioral processes to facilitate goal setting and attainment. A key strategy of Stay Strong + Coaching is to enhance activation and facilitate goal setting though enhanced self-monitoring. The Fitbit Charge 2 device will provide detailed minute-by-minute self-monitoring information through the objective measurement of physical activity. The self-monitoring and feedback loop will act on self-efficacy via 2 pathways: (1) influencing participants’ goal setting and activation to adhere to goals and (2) providing data for the human coaches and for automated coaching to customize motivational messages about goal attainment and to set appropriate, realistic future goals. Increased self-efficacy will mediate positive changes in physical activity, which will, in turn, lead to secondary outcomes of weight control and improved overall mental (eg, depression) and physical health (eg, pain) status. This theoretical grounding guided the choice of intervention components, which have each been classified in terms of an established taxonomy of behavior change techniques (BCTs) [44] (see Multimedia Appendix 2 for a complete list of BCTs by group assignment).

The Stay Strong + Coaching intervention comprises multiple components that support physical activity and weight self-monitoring. Figure 2 illustrates the components with data flows and Table 1 provides a complete list of intervention components by study arm.

To be eligible to participate in this study, participants had to be an OEF/OIF/OND veteran, identify a VHA medical center and VHA health care provider responsible for his or her care, be interested in starting a physical activity program in the next 30 days, have access to a computer with an internet connection and a working Universal Serial Bus (USB) port, have a smartphone running a compatible iOS or Android operating system, and be younger than the age of 65 years.

The decision to limit our recruitment to individuals under 65 years of age was because of higher risk of cardiac issues in individuals over the age of 65 years. The individuals were excluded if they reported that a health care provider had told them that it was currently unsafe to exercise in an unsupervised or unmonitored setting, they had a history of eating disorders or a body mass index less than 20, they were not competent to consent for themselves to a research study, or have worn a physical activity sensor within the last 30 days.

Recruitment packets containing an invitation letter and a brief study overview sheet were sent in batches of approximately 200. First, the candidate participants were provided an individually assigned program code in their invitation letter and directed to a website. Second, the individuals completed a set of screening questions to assess study eligibility. If they were eligible, they were then directed to the online consent and Health Insurance Portability and Accountability Act (HIPAA) authorization processes and the baseline survey. Next, participants were instructed to install and register the Stay Strong app through Google Play (Android smartphone users) or Apple (iPhone users) stores. When the participants successfully installed the Stay Strong app on their smartphone, the study team was automatically notified via an online portal. At this point, the participants were considered to be fully enrolled if they completed all of these steps. All the participants were reminded to follow-up with their health care provider as needed throughout the study.

The participants who were fully enrolled were then shipped a package that included a welcome letter, devices (Fitbit Charge 2 & Bluetooth weight scale), a Fitbit USB dongle for syncing their Fitbit device, instructions, a frequently asked questions document, a written copy of the Study Information Sheet, and HIPAA authorization. The participants were required to authorize the Stay Strong app to sync and access data from Fitbit. When they were fully configured, the participants were instructed to use their devices for up to 2 weeks and to sync their data via the Fitbit Connect software, at least weekly. When 7 consecutive days having at least 5 valid days of data were synced, the individual was randomized to Stay Strong or Stay Strong + Coaching. A day’s worth of data were deemed valid if at least 5 lightly active minutes had been recorded. The individuals were instructed to sync data via the Fitbit Connect software only via a Bluetooth-enabled dongle, which required access to a computer using a USB port. This was necessary to comply with VHA data security and confidentiality standards.

If the participant had not met the minimum criteria to be randomized after 2 weeks, the study staff would attempt to call the individual to solve any technical issues that had arisen. A total of 3 attempts were made to contact the individual and receive the physical activity data. Our intent was to randomize the participants who successfully completed all the steps and met all the criteria. Randomization was stratified by gender, physical activity level (high vs low), and smart phone operating system (android vs iOS) into the 2 groups in a 1:1 ratio, with a target of randomizing 175 participants per arm. All the study staff were blinded to the randomization list, which was created by the study statistician using Stata 14.1 (StataCorp LLC) with random block sizes and uploaded into the study tracking database by the programmer. After an individual met the requirements for randomization, a staff member requested the arm assignment using an automated system within the study tracking database. After the assignment was made, the staff updated the mobile app to comply with the assigned arm. The study staff did not have any in-person interactions with the participants during the course of the study and had specific telephone contact protocols to protect against potential bias.

All the study participants will be provided with a US $25 Amazon gift card at 6- and 12-months for completing the study follow-up surveys. The participants will be permitted to keep the Fitbit device and scale that were issued as a part of this study.

Online survey data for this study are collected at baseline, as a part of the enrollment process, and again at 6- and 12-months postrandomization. In the event participants fail to complete the online assessments and are unreachable via phone, we will attempt to contact the participants via US mail. After completing the 12-month data collection survey, as well as a final Fitbit device upload, all participants’ study-issued Fitbit accounts are deleted and the participants are able to setup their own personal account at their discretion. The Stay Strong app for both groups will remain active for an extra 30 days to ensure that sufficient endpoint physical activity and weight data are synced.

Within this study, we will collect a variety of outcome measures, which are summarized in Table 2.

The primary analytic cohort will be intent-to-treat, with the proposed primary and secondary analyses focusing on the effect of Stay Strong + Coaching compared with Stay Strong alone. The 1 exception is with women who self-report pregnancy at any of the 3 assessment times (baseline, 6-month, and 12-month); they will not be included in the analyses for the primary outcome of physical activity or for weight loss.

Descriptive statistics, including graphical displays, will be used to summarize all study variables overall and by arm. Evidence of between-arm imbalance in baseline characteristics that are potentially related to change in physical activity level (eg, baseline weight and full-time work status) will be noted and sensitivity analyses adjusting for these baseline characteristics will be considered to ensure that an observed intervention effect is not because of this baseline imbalance. Summary statistics will also be reported for primary and all secondary outcomes by arm, including unadjusted changes in daily active minutes from baseline to 12 months as means and 95% CIs.

Between-group comparison of active minutes will be examined using all longitudinally assessed weekly averages of active minutes, including the baseline active minutes, based on the longitudinal data mixed-effects model (75) with time (weeks since randomization), treatment group, and interaction of time by treatment group as primary predictors, participants and slopes as random effects, and autoregressive correlation within the person. The model will be adjusted for stratification factors of sex, operating system, and baseline activity level (high vs low). Time may be parameterized appropriately if a nonlinear slope over time is suggested in graphical examination of the outcome measures over time. On the basis of the model, predicted active minutes at 12 months will be compared with 2 treatment groups, and a test of significant slope of the interaction term will also be used to test if the change from baseline in active minutes differ between treatment groups. Weights will be analyzed similarly. Although assessments will be done only 3 times, the secondary outcomes of depression and pain will be analyzed similarly as for the analyses for active minutes. To test for moderation of the coaching effect by gender, the model specified above will be expanded to include the gender by Stay Strong + Coaching interaction. We also plan to conduct analyses to examine patterns of intervention utilization and dose-response relationships. All the statistical analysis will be completed by the study team, independent of the funder and vendors involved in the study.

The participants are instructed to report any changes in their medical condition to their primary health care provider first, followed by a report to the study hotline. Moreover, all participants receive automated in-app reminders every 90 days to notify the study staff of health changes. Any reported adverse event will be followed up by the study staff for additional information and will be classified as serious or not serious, related or unrelated, and anticipated or unanticipated, as well as severity, by the study staff and study physicians. Any adverse event that is categorized as serious is subject to IRB required reporting, and medical suspension and reclearance processes will be initiated as required.

This study marks a significant contribution to the mHealth literature. Despite the increasing ubiquity of wearable physical activity tracking devices among consumers, recent reviews identified the need for randomized trials with a sufficiently large sample size and longer duration (longer than 12 weeks) interventions [6,28]. Moreover, the lack of fully described theoretical frameworks that underlie mHealth-mediated behavior change interventions has been noted [79,80]. Stay Strong is a fully powered trial with an active comparator, which follows participants for 12 months [28]. It features a theoretically driven behavior change intervention. Furthermore, most trials using wearables include samples with a minority of men, whereas participants in this trial are predominately male (267/357, 74.8%), which is an especially important contribution to the literature. Also, we seek to assess if coaching provides additional benefits via deeper engagement with wearable activity monitors compared with activity monitoring alone, a unique contribution to the literature.

This study is also significant in its reliance on a national sample of OEF/OIF/OND veterans who were consented and enrolled online with no in-person assessments or interactions. On the whole, OEF/OIF/OND veterans are younger than other veteran cohorts. Publicly available data show that post-9/11 veterans have a median age of 35 years; thus, many are digital natives born after 1980, having grown up with computers and other forms of information technology [81]. Recent estimates indicate that 98% of American adults aged 30-49 years have a cell phone, and 89% have a smartphone, suggesting that our smartphone-based intervention is well suited to our targeted population, important for increased adoption and uptake [82].

The response to mailed letters was much higher than anticipated based on our team’s past experiences with recruiting for internet-mediated interventions. This has been especially striking, given that it can be challenging to engage veterans in lifestyle interventions as demonstrated by our work and the work of others. The average age of participation in VHA’s weight loss program (also known as MOVE!) is 57 years [83], compared with 37 years for Stay Strong participants. Thus, mHealth interventions have significant potential for engaging this cohort of veterans to improve lifestyle behaviors like physical activity. The Stay Strong trial is the first national trial of an mHealth intervention in the VHA aimed at increasing physical activity.

This study has limitations. Our primary outcome measure of physical activity is focused on active minutes, as measured by the Fitbit device. As previously mentioned, active minutes is a proprietary measure that is not precisely defined, and any potential changes to how active minutes are calculated by Fitbit are out of our control. Although this is a limitation to our approach, there is also strength in the fact that our use of this measure allows us to ensure that this nationally-conducted randomized trial can be conducted entirely without bringing participants into a clinic or research setting for data collection. This pragmatic nature opens the door to a more nationally representative sample and reduces the ultimate burden placed on research participants. In this vein, another limitation to our approach is our reliance on self-reported data for some of our collected measures. We have taken steps to gather objectively measured data whenever possible (via study issued devices). Another limitation of our approach centers on the fact that not all coaching features persist after the first 2-3 months of the intervention. Although our automated features continue through the duration of the program, contact with human coaches is focused on in the first several weeks. Future work should seek to determine whether prolonged human-coach contact has an effect on participant engagement and participant outcomes. We also have a methodological limitation because we seek to determine the effectiveness of adding coaching features to the Stay Strong intervention and we are predominately focused on quantitative measures. Although we understand the richness that a qualitative or mixed-methods approach would add to this study, budgetary constraints preclude additional measures or qualitative data collection. Additionally, our approach is limited by the fact that participants are instructed to sync their data using a dongle attached to a laptop or desktop computer rather than syncing directly using the Fitbit app on their smartphone. This was necessary in our setting to help ensure privacy of personal data and comply with VA regulations. This decision will likely affect our user engagement and compliance with syncing protocols because of the increased burden in using this approach. The reliance on laptop or desktop syncing reduces generalizability and potentially disproportionally excludes lower-income individuals, who are more likely to rely on cell phones for internet connectivity and are less likely to own desktops or laptops. Lastly, though we intended to block-randomize participants by gender, physical activity level, and smartphone operating system into the 2 groups in a 1:1 ratio, the first 283 participants were randomized without blocks because of a programming error. However, the last 74 participants were block randomized. This led to unequal distributions in a few of the cells. However, analyses showed there were no differences in baseline characteristics.

This randomized control trial will provide much-needed findings about whether the addition of telephone-based plus automated coaching and personalized goal setting will improve levels of physical activity compared with using a wearable device and smartphone app alone. This year-long trial will also provide insights on feasibility and acceptability of interventions like Stay Strong among OEF/OIF/OND veterans.