Noncommunicable diseases (NCDs), also known as chronic diseases, are one of the major global public health challenges of the 21st century, being responsible for about 40 million deaths per year, 15 of which are premature (ie, between 30 and 69 years) [1]. The majority of these deaths occur in low- and middle-income countries (LMICs). According to the World Health Organization’s (WHO’s) global report, cardiovascular diseases, cancer, diabetes, and chronic respiratory diseases are responsible for 82% of all NCD deaths [2]. In Lebanon, the prevalence of NCDs accounts for 85% of all deaths. NCDs are the product of 4 main risk factors: tobacco use, physical inactivity, the harmful use of alcohol, and unhealthy diets, which, in turn, lead to 4 key metabolic changes (raised blood pressure, overweight and obesity, raised blood glucose, and raised cholesterol) [1]. With regard to overweight and obesity, globally, in 2014, there were more than 1.9 billion overweight adults, representing 39% of the world population. Of these, 600 million were obese [3]. In Lebanon, the prevalence of obesity and overweight is 65.4% (obesity accounting for 27.4% and overweight for 38%) [4]. A fundamental cause of obesity and overweight is an energy imbalance between caloric intake and caloric consumption. This imbalance is due to global trends of increased availability and intake of energy-dense foods that are high in sugar and saturated fats, and insufficient physical activity, due to the sedentary nature of many forms of work, modes of transportation, and increased urbanization [3]. All these behavioral risk factors are preventable, as they can be addressed by modifications in the environment and lifestyle of individuals (ie, increasing physical activity, reducing sedentary time, and following a healthy diet) [5].

The effectiveness of nonsurgical and behavioral weight management interventions has been demonstrated in several systematic reviews and meta-analyses [6,7]. For example, commercially available weight loss treatments, such as WeightWatchers, and pharmaceutical products, such as Qsymia, were found to be cost-effective strategies to achieve weight loss [8]. Effective behavioral interventions should include both physical activity and dietary components to reach larger and sustained effects [9]. Self-monitoring is one of the most effective change techniques [10] included in behavioral weight loss interventions, as evidence shows that people who report weight monitoring on a daily or weekly basis tend to be more successful in attaining weight loss goals [11,12]. Self-monitoring improves the awareness of caloric and food intake, increases self-efficacy, and permits the evaluation of any change or progress over time [13]. However, long-term weight loss and maintenance interventions that are delivered face-to-face usually require substantial work of a specialist workforce and sizeable resources, both from the participants and the service providers.

In the past decade, “health services and information delivered or enhanced through the internet and related technologies” (ie, “eHealth”) [14] have transformed the way patients interact with the health care system [15] and engage with their own health [16]. eHealth technologies allow users to monitor, track, and inform their health; communicate between health agencies and external stakeholders in terms of health; and collect, manage, and use health data [17]. In this context, new strategies for behavioral weight management interventions have been developed [18-22], with the aim to provide users with the support necessary to attain weight loss and, at the same time, contain costs [23]. The potential of eHealth and mobile health (mHealth) is especially relevant to LMICs, where phone ownership is rising rapidly, but access to health care services is often limited. Recent systematic reviews on eHealth behavioral interventions addressing NCDs and their risk factors in LMICs show promising results [24,25]. The attention toward mHealth apps is also justified by the high penetration rates of these technologies. According to a 2014 Pew Research Center survey, in Lebanon, the penetration rate of mobile phones was 86%, with smartphones reaching 45%, the highest penetration rate in the Middle East and North Africa (MENA) region [26]. Smartphone ownership is as high as 60% among adults aged 18 to 29 years and 55% among adults aged 30 to 49 years [26]. More recent data (October 2016 to November 2017) available from Net Marketshare show that mobile and tablet represent 61% of the market in Lebanon, with Android capturing 61% and iOS capturing 38% of the Lebanese market.

Mobile phones, and particularly apps, have been considered convenient intervention platforms as they are portable, appealing, and universal [27]. Mobile phones can be used for self-directed interventions for weight management. Self-directed interventions are those that “require minimal professional contact (eg, provision of initial instructions) or no professional contact and can be easily used with existing infrastructure and in the context of users’ everyday lives” [28]. Mobile apps started to show some suggestive evidence of effectiveness [18,29], with studies reporting positive effects in weight reduction when apps were employed as a supplement to telephone coaching [30]. However, small, nonsignificant effects were reported when apps were used as a standalone tool [31,32]. Smartphones are the ideal platform for delivering self-directed, Just-in-Time Adaptive Interventions (JITAIs), which are treatment programs that, as the name suggests, adapt to the patients’ progress, eg, when they attain goals or positively respond to treatment [33]. JITAIs can provide support through skill building (coping, making decisions, planning, etc), emotional support (encouragement, etc), and instrumental support (feedback, etc) when users need these features the most [33,34]. JITAIs are complex, algorithm-dependent interventions based on several design principles, which include decision points and outcomes that would inform the provision of tailored feedback (ie, tailoring variables), intervention options (based on if-then conditions grounded on behavioral theories, aimed at addressing short-term behavioral outcomes), and decision rules. For a more detailed overview of the JITAI framework, we refer the reader to the seminar report by Nahum-Shani et al [34]. Mobile devices are adequate platforms for delivering feasible and scalable JITAIs, given the smartphone technology advancements that are providing users with continuous monitoring and personalized coping strategies [34]. Nevertheless, little is known about the efficacy of mobile apps acting as main components of JITAIs for weight management.

According to the WHO’s global action plan for the prevention and control of NCDs [35], workplaces are one of the most important settings for health promotion as they are the gateway to a large number of people (about 65% of the world’s population is employed) [36]. In the last two decades, many public health efforts have been made, and numerous interventions have been conducted to tackle these problems. There is a wide evidence supporting the effectiveness of workplace interventions for the prevention of obesity [37-40], and many national governments adopted policy decisions for promoting health through workplaces. In response to the WHO call for action, the Lebanese Ministry of Public Health released a plan for the prevention and control of NCDs in 2016 [4], which provides a set of guidelines for action, despite excluding workplaces as a setting for priority interventions. This gap is being filled by the American University of Beirut (AUB) and its Medical Center (AUBMC) [41], which have expressed their commitment toward building a healthier campus and community through a long-term strategic plan endorsed in the Health 2025 vision.

The overarching goal of this study is to determine the feasibility and preliminary efficacy of a self-directed weight management intervention, targeting university employees and delivered entirely through mobile apps. Specifically, the study aims to (1) evaluate the acceptability and feasibility of using 2 commercially available mHealth apps for weight management; (2) evaluate the feasibility of implementation strategies, retention, and adherence to the study, fidelity to the protocol, assessments, and data collection procedures; and (3) estimate the effects of the interventions on weight-related outcomes (eg, weight, body mass index, and waist circumference), as well as on behavioral (physical activity and diet), and cognitive factors related to weight loss (motivation to lose weight, to engage in physical activity, and follow a healthy diet).

Participants are employees working at AUB and its Medical Center (AUBMC), located in Beirut, Lebanon. Employees include faculty (n=1182), of which 41.03% (485/1182) are female; nonacademic staff at AUB (n=1266) of which 46.99% (595/1266) are female; and AUBMC staff (n=3038) of which 52.00% (1580/3038) are female [41]. The employer did not influence the design of the study or its methodology. The recruitment strategy does not rely on the endorsement of any AUB or AUBMC authorities. Participants will be informed orally and through written consent forms that their participation is completely voluntary, free of charge, and no penalty will be pressed for those not willing to participate.

To be enrolled in the study, participants must (1) be employed full- or part-time; (2) be able to read, write, and understand English; (3) own a smartphone with either Android (v4.4 or above) or iOS (v8 or later); and (4) be interested in better controlling their weight (ie, losing weight, preventing weight gain, maintaining weight lost, or gaining weight in a healthy way). Exclusion criteria include the following: (1) being full-time students who cannot prove their status as full- or part-time employees at AUB or AUBMC; (2) not being able to read, write, and understand English; (3) not owning a smartphone with either Android (v4.4 or above) or iOS (v8 or later); (4) having physical disabilities preventing from exercising or walking; (5) being on a special diet for treatment of chronic conditions (eg, diabetes); (6) being diagnosed with anorexia or bulimia nervosa; (7) being on weight loss medications; and (8) having undergone bariatric surgery in the past 3 months. The trial is registered at ClinicalTrials.gov [NCT03321331]. The protocol of this study has been approved by the local Institutional Review Board (ref #FHS.MB.07/SBS-2017-0416).

Participants will be recruited through email invitations sent from the AUB Health and Wellness Center, and printed posters hanged on billboards on campus. A digital version of the poster with a shortened link (bit.ly/WaznApp) and QR-code linking to an eligibility screener survey will be displayed on monitors on campus TV screens. Invitations to enroll will also be circulated on social media using the research team’s personal and professional social networks (Facebook, Twitter, LinkedIn, and ResearchGate). Emails, posters, and social media postings will contain a link leading to an eligibility screener survey, which will first display an informed consent form. The survey, designed using LimeSurvey, is hosted on AUB servers. At the end of the form, employees will be instructed to complete their enrolment by visiting the University Health Services (UHS) clinics, where nurses will verify their eligibility and take basic anthropometric measurements (ie, height, weight, and waist circumference). Nurses will fill out paper-based forms, thereby cross-checking their eligibility to participate in the study. If participants are eligible, nurses will inform the participants that the research team will contact them via email with information about the next steps in the study. Nurses will provide a hard copy of the consent form to all participants.

Randomization procedures will take place after employees are confirmed to be enrolled in the study. A random sequence based on the minimization procedure will be generated using a computer program, following Altman and Bland’s approach [42]. The minimization procedure allows to balance the allocation of study participants to a prespecified number of treatment groups as soon as they enroll in the study, considering participant characteristics (ie, stratification by gender, age, and anthropometric features) collected during the eligibility phase. The program used to perform the minimization procedure is QMinim [43]. A statistician from the Faculty of Health Sciences will generate the sequence using the program. The random sequence will be uploaded to the REDCap (Research Electronic Data Capture) platform [44]. The research team will inform participants about their allocation after the visit, with a welcome email.

As we need to provide intervention participants with a link to download a full version of the app, allocation concealment and blinding cannot be applied. However, efforts will be implemented to reduce allocation bias. First, the clinical staff weighing participants at baseline and Week 12 will be blinded to treatment groups. Second, after the baseline clinic visit and the initial account set-up, minimization procedures will be undertaken by a statistician, not involving the principal investigator. Third, the intervention is completely delivered via mobile apps. The only contact is related to study procedures and data collection. Finally, data analysis will be conducted on masked data.

The WaznApp study is part of a larger project entitled “Can commercial mobile apps for weight management be used in interventions? Bridging the gap between usability, theoretical adherence, and user experience.” The project included a formative phase, which was based on a user-centered heuristic evaluation study, aimed at understanding how members of the employee community perceive 6 weight management apps (Lark, MyFitnessPal, SparkPeople, MyPlate, My Diet Diary, and My Diet Coach). The 6 apps were selected because they achieved the highest total scores on the Mobile App Rating Scale (MARS) [45], in a recent expert review [46]. In the heuristic evaluation study, 36 employees were randomly assigned to use one of these apps for 2 weeks. At the end of this period, they submitted their app quality evaluation using the user version of the MARS scale (uMARS) [47]. Three apps achieved the highest mean ratings for the total app quality score (Lark=4.0; MyPlate=3.8; MyFitnessPal=3.7). Lark clearly emerged as the app with the highest quality scores. In addition, this app has been recently employed in an observational study, involving 70 diabetic patients, who lost 2.4% of their weight at baseline after about 15 weeks [48]. Users of MyPlate reported several functionality issues, and its database is not as complete and responsive as its popular counterpart MyFitnessPal, which is one of the most downloaded apps for dietary tracking [49]. Compared with another popular app (Lose It!), MyFitnessPal allows users to automatically track activity through the phone or through integrations with many wearable devices [50]. For these reasons, we selected Lark and MyFitnessPal as trial apps.

The WaznApp study is a 12-week, prospective, single-center, parallel, randomized controlled trial with 2 study arms. The intervention arm will use Lark, a mobile coach, which provides interactive counseling through a chat-style interface by delivering motivational feedback, goal setting, and emotional social support. Lark acts as an adaptive intervention [33] and provides feedback on several outcomes (see, eg, the text and app-based Cell Phone Intervention for You trial [51,52]). The content of the messages depends on the way users interact with the app. The content is based on several behavior change techniques, such as goal setting, reviewing behavioral goals, feedback, and social support [46]. The users input their activities, foods or weight, but the app tracks automatically their active time using proprietary algorithms and phone motion sensors. The control group will use MyFitnessPal, a calorie-counting app that does not include JITAI components and relies on the user input for food tracking. MyFitnessPal acts as the control condition, as it provides limited social support, which is present in Lark, and it is an important feature that is generally lacking in calorie-counting apps [53].

In this study, primary outcomes are those related to the feasibility, whereas secondary outcomes are those related to the preliminary efficacy. The timeline for data collection is specified for each outcome below. We will collect data mostly through mobile-friendly, Web-based surveys. Participants will be invited to fill the questionnaires via email, which will contain personalized links to the surveys. Before the start of the study, all instruments will be pilot tested.

Except for the 24-hour recall/record done through the Web-based ASA24 platform, process and outcome measurements will be collected through REDCap [44], which is a secure, the Health Insurance Portability and Accountability Act compliant, Web-based survey application hosted on AUB servers. REDCap is an ideal program for managing longitudinal studies with multiple assessments.

Traditional sample size calculations are not typically undertaken in nonprobability sampling and in pilot interventional studies, as these are intended to test acceptability and evaluate the process as well as to inform power calculations for subsequent studies [78]. Nevertheless, researchers recommend that sample size justifications are provided also in pilot studies [78]. In the absence of similar studies in Lebanon and in the region, it is difficult to estimate precise effect sizes. However, estimates can be derived from the average effect sizes reported in similar mobile-based weight loss interventions that had minimal researcher interaction [79-81], included in Schippers et al’s meta-analysis [21]. In these studies, the effect sizes (SMD, standardized mean difference) based on weight change ranged from d=0.33 [79] to 0.37 [80]. According to Whitehead et al’s tables (for SMD=0.3 and 0.4, respectively) [78], a two-arm trial, designed with 90% power and two-sided 5% significance, will require between 305 and 181 participants per arm; a pilot study will require, respectively, 45 and 33 participants per arm [78].

Descriptive statistics will be conducted for the overall characteristics of the study population through presenting the numbers and percentages for nominal or categorical variables and means and standard deviations for continuous ones. For variables assessed using multiple items (eg, uMARS scales), Cronbach alpha and corrected item-total correlations coefficients will be used to assess the internal consistency of the measured constructs, before aggregating the information (ie, averaging). Bivariate correlations and chi-square tests will be used to explore associations among demographic and psychographic variables and main study outcomes.

Missing data are expected to be minimal for most variables. Depending on the number of participants, the number of missing data points, and the distribution of the outcome variables, we will decide which missing data strategy to use (eg, full-information maximum likelihood or multiple imputation) [82]. The missing data bias will be assessed by computing a binary variable reflecting the presence or absence of missing data for each variable in the model, and then, this binary variable will be correlated with all other variables in the model as well as an array of demographic variables.

Reliability of the uMARS scales will be evaluated using both indices of interrater agreement (IRA) and interrater reliability (IRR) [83-85]. Following the recommendations from the literature [86,87], IRA will also be measured according to Brown and Hauenstein’s aWG index [88], and the adjusted average deviation index ADM (adj) [89]. IRR indices will be based on intraclass correlation coefficients measuring test-retest reliability [90,91]. Similarly, test-retest reliability estimates will be used with the ASA24 Dietary Assessment Tool and the IPAQ-S [57].

Independent samples t tests, one-way analysis of variances (or nonparametric alternatives where appropriate) will be used to test differences in the relevant outcome variables (behavioral, cognitive, and weight-related data) between intervention groups and between pre- and posttest. Behavioral and weight-related data will be presented using appropriate confidence intervals as suggested by Lee and colleagues [92]. The estimation of preliminary efficacy will be based on the observed trends on data over time rather than traditional inferential statistics [93]. Subgroup analyses will also include intent-to-treat versus randomized analyses to detect whether differences in the outcomes are associated with adherence to the trial (per protocol).