Apps Seeking Theories: Results of a Study on the Use of Health Behavior Change Theories in Cancer Survivorship Mobile Apps

Background Thousands of mobile health apps are now available for use on mobile phones for a variety of uses and conditions, including cancer survivorship. Many of these apps appear to deliver health behavior interventions but may fail to consider design considerations based in human computer interface and health behavior change theories. Objective This study is designed to assess the presence of and manner in which health behavior change and health communication theories are applied in mobile phone cancer survivorship apps. Methods The research team selected a set of criteria-based health apps for mobile phones and assessed each app using qualitative coding methods to assess the application of health behavior change and communication theories. Each app was assessed using a coding derived from the taxonomy of 26 health behavior change techniques by Abraham and Michie with a few important changes based on the characteristics of mHealth apps that are specific to information processing and human computer interaction such as control theory and feedback systems. Results A total of 68 mobile phone apps and games built on the iOS and Android platforms were coded, with 65 being unique. Using a Cohen’s kappa analysis statistic, the inter-rater reliability for the iOS apps was 86.1 (P<.001) and for the Android apps, 77.4 (P<.001). For the most part, the scores for inclusion of theory-based health behavior change characteristics in the iOS platform cancer survivorship apps were consistently higher than those of the Android platform apps. For personalization and tailoring, 67% of the iOS apps (24/36) had these elements as compared to 38% of the Android apps (12/32). In the area of prompting for intention formation, 67% of the iOS apps (34/36) indicated these elements as compared to 16% (5/32) of the Android apps. Conclusions Mobile apps are rapidly emerging as a way to deliver health behavior change interventions that can be tailored or personalized for individuals. As these apps and games continue to evolve and include interactive and adaptive sensors and other forms of dynamic feedback, their content and interventional elements need to be grounded in human computer interface design and health behavior and communication theory and practice.


General I nst ructions for Coding Mobile Health Applications
Please review the mHealth theory and behavior change taxonomy before coding the mobile health applications (mHealth apps). Discuss the techniques with co-coders to ensure that all coders interpret these materials, defini tions, and techniques similarly. Conduct at least one practice coding session jointly with practice materials that are comparable (but different from) the final study materials and discuss these results with the team before beginning coding the study applications. The following suggestions assume that the selection of the mHealth apps has already been made and that there is agreement among the team that all of the selected apps meet the stated criteria for inclusion in the study.
Suggestions for optimal coding of health behavior mhealth apps: • Read all selection and coding material before coding.
• Scan the different health behavior techniques presented in the coding table or Coding Scoring Sheet as these may differ by different types of apps.
• Print out the Behavior Change Theory Definition Mat rix and one Coding Scoring Sheet for each mHealth application. Put the name and source of the mHealth Application on the Coding Scoring Sheet • Start coding using the Coding Scoring Sheet. In case of any doubts between the techniques, please re-read the descriptions in the Theory Analysis and Definition Mat rix After you have finished coding a mHealth application, please review the completed coding scoring sheet to make sure that you have scored the correct techniques and that you have crossed out any techniques that are not use in the mHealth app under consideration.
Request: If you have suggestions for improvements or extensions of this coding manual please make them prior to beginning to code. The coding manual is a work in progress, but for consistency in the study all coders need to begin and end with the same set of instructions.

General Techniques in m Health Applications
Each scoring sheet has specific areas to score or annotate the presence of personalization, tailoring and participation.

1.Personalization:
This is the provision of opportunities in the mHealth application to make elements of the application personal by the selection of tools or elements that are specific to the individual using the application. An example would be the ability to select a disease type from among several available in the application and then to follow a specific path or set of tools or systems. For example, being able to select "breast cancer" and then being provided sets of information specific to that type of cancer. Another example would be the abili ty to select to receive emails or texts of a specific nature. The choice of "yes" or "no" to a specific capability of the application would be considered personalization.

Tailoring:
Coders are asked to annotate the score sheet for each mHealth application to indicate the app's capability to include an intervention element or component that is specific to the characteristics of the person using the app. Coders will be asked to score tailoring at three different levels in the ini tial assessment of the mHealth application (see Coding Scoring Sheet): 1) Macro-tailoring at the group level. In this instance the mHealth application can be adapted to adjust the intervention materials (including information) that the participant receives based on pre-tested characteristics. For example an app may ask the user if he/she wishes to receive texts and/or assessments on diet, on exercise or smoking cessation. 2) Meso-tailoring at the individual level. The amount or type of in tervention depends on the individual needs of the participant. For example, the participant could select between texts delivered once a day versus once a week.
3) M icro-tailoring at the individual level. Specific techniques in the mHealth application are tailored to the unique individual. For example personalized goal setting and reporting tailored to the individual's own needs and desires for physical activity. Or GPS t racking and reporting of an individual's walking or running activities.
Note that all of these general techniques may be used in one mHealth application. I t is possible to have personalization, macro-, meso-and microtailoring techniques. To score these general techniques the user or participant must be prompted to select an answer or provide input and make decisions in relation to the techniques. Scoring is accomplished by marking the technique with a 1 or 0 in each element or section of the sheet. A "1" indicates that the technique is present in the app, a "0" indicates that i t is not present. Personalization and tailoring scoring are provided as additional elements for each major determinant.

P rovide information about health behavior linkages.
Basic information about cancer and cancer survivorship, diagnosis, t reatment, and/or availability of resources for clinical or non-clinical purposes. If Personalized, the user is prompted to select or provide personal answers about type of cancer or stage of survivorship for example. If Ta ilored, the user is required to select actions or elements specific to the intervention and the way information or activities are delivered to them as a result of these choices.

P rovide information on action/behavior and consequences.
Information is provided about the cost/risks/benefits of action or inaction with respect to certain cancer survivorship behaviors. This scoring would also consider r isk-communication strategies such as persuasive communications for example post t reatment health screening, smoking cessation or adherence to f lu-shot recommendations.

I n tention
3. P rompt I n tention Formation. The mHealth application includes suggestions for general behavior setting or formulating desired outcomes of a behavior for healthy survivorship, e.g., maintain a healthy weight, exercise regularly, eat 5 fruits and or vegetables daily. I t may be sometimes difficult to distinguish this from knowledge or awareness, but coders should look for language that indicates a specific action or activity. Also, note that this technique is different from the actual setting of a goal or behavioral objective to facilitate change or adherence.

4.Provide I nstruction. This technique involves telling or showing the user
or participant ways to facilitate behavior change. For example explaining "SMART" goal setting, or how to use an app's function to record questions on a mobile phone to ask a provider during an appointment. The function of the instruction must be directly related to the improvement or behavior change, not for general use of the phone or the app. The facilitation may be in the form of wri t ten instructions, videos on YouTube that link from the app or images or cartoons that show a step-by-step instruction.

P rovide mate rials for education/information.
The app provides the cancer survivor with specific materials and information that are suggestions for behavior change. These differ from 1. Knowledge/Awareness in that the education is specific to a behavior change or an action. For example, information on late effects of cancer t reatment with prompts of when to contact a healthcare professional. Another example would be educational information about health screenings specific to cancer survivors, for example breast mammograms for female survivors t reated with whole body or mantle radiation starting at an early age. I n tention 6. P rompts for Specific Goal Setting. This involves planning and setting a specific goal for what a person would do within a specific time and includes the specific context within which a behavior will be performed. This would include selecting or writ ing down (micro-tailoring) of a specific goal for example setting a personal goal to "engage in physical activity for 150 minutes each week." Goal setting would include information on when, where, how to act in a specific behavior. 7. Review of general or specific goals. This would involve using the mHealth app in reconsideration of previously set goals or intentions and would require an indication of behavioral performance resulting from selfmonitoring or t racking. An example would be review of t racking a goal setting for intake of a specific amount of calories per day or number of minutes of physical activity for a week. Another example might be noting a set of questions to be asked at a healthcare provider appointment regarding levels of pain or emotional distress during a past week or month.
Self-Efficacy 8. P rompt self-monitoring of behavioral goals. The mHealth app suggests that the person record brief notes or keep a journal or diary to record behaviors and actions related to health behaviors. Examples might be a journal of physical activity or pain or distress monitoring and actions taken to alleviate such as meditation or self-talk.

Persuasion (verbal or w r itten)
The mHealth app delivers messages (may be personalized or tailored) designed to strengthen efficacy/control beliefs related to the execution of target or suggested behaviors. Examples might be often-used successful strategies (e.g., "park at the far end of the parking lot", or "use the stairs instead of the elevator to increase physical activity") or general tips. New beliefs may be induced and/or new information provided to the participant or user to create new control or behavior beliefs.

P rovide information about peer behavior (Peer passive).
The mHealth app provides information about what other cancer survivors do and think in relation to targeted behavior change. This can be provided in the form of wri t ten anecdotes YouTube videos and may be presented as interviews or case studies.

P rovide opportunities for social comparison (Peer active)
The mHealth app offers participation in Facebook, Twit ter or other social media and networking in which discussion and social comparison may occur. The focus is on providing social reference for the behavior change or activity. Only score this technique when examples of group or peer discussion including personal stories of behavior are shared. For example, a participant cancer survivor shares that "setting my own goals for physical activity and sharing those with my Facebook friends really helped me make my goals." Or "wr it ing down my concerns about pain helped me communicate more effectively with my doctor."

Mobilize Social Norms ( Important Others)
The mHealth app provides exposure to the social norms of important others in relation to a healthy survivorship activity or health behavior change. Important others may be valued and t rusted experts such as a recognized healthcare professional, a celebri ty cancer survivor or a recognized cancer survivorship researcher or advocate (e.g. Nancy Brinker of Susan G. Komen for the Cure or Cathy Giusti, founder of the Multiple Myeloma Research Foundation.