Evaluating Consumer m-Health Services for User Engagement and Health Promotion: An Organizational Field Experiment

Introduction Chronic disease is prevalent and costly in the U.S. (Tu \& Cohen, 2009). Poor eating habit is one factor that account for risk of chronic disease (Arsand, Varmedal, \& Hartvigsen, 2007). Smartphone technology has been promising to improve preventive health outcomes. However, its great potential has not been widely applied to people's eating behaviors and its impact is unknown.

Professional and peer supports can improve health status (Elkjaer et al., 2010; Lorig et al., 1999; Perri, Sears, \& Clark, 1993). However, the former is usually delivered didactically or passively with limited use of smartphones. There is also little evidence of the effect of peer support delivered by smartphones in the domain of healthy eating. This research aims to study what smartphone technology can do to upgrade professional and peer supports, and to evaluate the impact of these mobile-app enabled supports on people's behavior of healthy eating and user engagement.

Hypotheses

According to Social Cognitive Theory, we hypothesize the following:

1. Mobile-enabled self-monitoring approach improves users' healthy eating behaviors through improved self-efficacy
2. Professional support improves users' healthy eating behaviors through improved self-efficacy
3. Peer support improves users' healthy eating behaviors through improved self-efficacy
4. The amount of support is positively correlated with the change in behaviors and their determinants

Experiment Design The hypotheses will be examined bya 4-month randomized field experiment. 375 subjects will be recruited and assigned to one of the five arms to receive the corresponding tool for diet management at no cost.

Arm 1: a mobile App with both professional and peer support Arm 2: a mobile App with peer support only Arm 3: a mobile App with professional support only Arm 4: a mobile App without any support Arm 5: a non-mobile web App In addition to the App usage data, five surveys are conducted at baseline and the end of each month. Respondents will be compensated by $8 and a chance to win $200 for each completed survey.

Interventions

1. All subjects receive the following interventions: an education package includes the importance of healthy eating, concept of MyPlate, personalized daily food plans; reminders throughout the study; goal setting capabilities;
2. Self-monitoring provided by an Android App: a heuristic approach inspired by MyPlate to record their food consumption which allows users to record their meals by images and doesn't require estimations in cups and ounces; daily reports and trend reports
3. Self-monitoring provided by the web App: a traditional approach to record their food consumption which requires estimations in cups and ounces, and no images are allowed; no daily reports and trend reports are provided
4. Professional support provided by a registered dietitian via the Android App: the supports include the following:
5. Reply to users' messages regarding healthy eating
6. Provide meal-specific comments on subjects' meal consumptions: one meal per week
7. Provide feedback on the subjects' consumption patterns: once per week
8. Peer support provided by other subjects via the Android App: the App provides platforms for subjects who have the same interest to communicate to each other. The actions the subjects can do in the platforms include:
9. Post images or texts related to healthy eating
10. Share meals with their ratings
11. Like/dislike others' posts
12. Comment on others' posts e Create groups which allow subjects who have the same interest to join

Measurements

1. Dependent Variables
2. Eating behavior: a score of healthiness of a meal will be assigned by a dietitian. The scores obtained by the same subject along the experiment compose the subject's eating behavior
3. Engaging behavior: this is measured by the number of meals recorded by the subject in a week. The numbers for the same subject along the experiment compose the subject's engaging behavior
4. Independent Variables
5. Level of peer support: this will be measured by a score representing the number of posts, likes, and comments received and given by a subject in a week
6. Level of professional support: this will be measured by a score representing the number of messages or comments exchanged with the dietitian in a week
7. Mediator Variables: self-efficacy, outcome expectation, and impediments will be measured by survey instruments

Data

1. User Survey: characteristics and perceptions
2. Mobile App: time, location, and contents of goal setting, meal entries, communications with professionals and peers
3. Web App: contents of goal setting and meal entries

Analysis Confirmatory factor analysis and statistical modeling such as structural equation modeling and mixed models will be conducted to test our hypotheses.