Preferences of mHealth app features for weight management among breast cancer survivors from underserved populations

1. Background

A majority of U.S. adults own smartphones (77%), with similar ownership rates between gender and racial/ethnic groups.¹ mHealth has been gaining in popularity as scalable, personalized opportunities to provide effective behavioral change strategies for health promotion among cancer survivors.² A systematic review found digital interventions for cancer survivors resulted in improved moderate-vigorous physical activity behaviors and body weight; data was mixed for diet behaviors.² However, key questions remain in terms of how mHealth interventions should be optimally designed for breast cancer survivors from underserved populations based on their cultural and literacy differences³ and the cancer-related disparities they face.⁴ Obesity disproportionally affects underserved populations and excess weight is associated with increased risk of recurrence and decreased survival among breast cancer survivors.⁴ Reviewing the Behavior Change Techniques (BCTs) present in previous efficacious trials for weight loss among cancer survivors may provide a useful starting point to design mHealth interventions. We examined perceptions of evidenced-based features breast cancer survivors from underserved health disparity-facing populations would find important to include in an mHealth weight management app and to examine the usability of app prototypes.

2. Methods

2.1. Participants

We recruited female breast cancer survivors from a large urban safety net hospital. Recruitment methods included a hospital-based registry, a hospital-based breast cancer support group, flyers, and direct letters from a clinician. Interested patients were screened for self-reported eligibility criteria: reading/speaking English; female; ≥18 years old; ≥2 years post-breast cancer diagnosis of stage 0-3; ≥6 months since receiving cancer treatment including surgery, radiation, or chemotherapy; overweight or obese; and ownership of Android- or iOS-platform smartphone. This study (H-36585) was approved by the Boston University Medical Campus/Boston Medical Center IRB; informed consent was obtained from all participants.

2.2. Design

We examined a systematic review⁵ and identified 19 techniques (BCT Taxonomy version 1)⁶ present ≥2 times in interventions shown to be effective in weight-related trials among cancer survivors. We then developed a PowerPoint presentation with pictures that displayed how each BCT could be incorporated into an mHealth app. In phase 1, we held both individual and small group discussions during which participants completed a socio-demographic survey, were shown the presentation, and were asked to rate the importance of each BCT from 1 (not at all important) to 9 (very important). During each session, we then collected the participants’ individual rating forms and added the mean ratings from previous participants. Now considering their first rating compared to the group rating, each participant was asked to re-rate the features. The goal was to come closer to achieving consensus on which features to include through this modified Delphi method. The moderator also asked participants open-ended questions to explore: topics that stood out as most useful/least useful, how to brand apps for breast cancer survivors, and if there are any other features we should include. Lastly, participants were asked if they would like to be re-contacted for phase 2 of the study.

Based on the findings from phase 1, we identified the techniques deemed highly important and developed app wireframes exemplars using Balsamiq software. In phase 2, we showed the wireframes to participants on an iPad; participants assessed their usability by completing the 10-item System Usability Scale.⁷

2.3. Analysis

Quantitative rating data was examined descriptively. The audio-recorded sessions from phase 1 were transcribed verbatim. Using a content analysis approach⁸ assisted by NVivo 12 qualitative data management software, we created inductive codes to capture ideas pertaining to perceptions of mHealth apps. Two investigators independently coded several transcripts and met on multiple occasions to discuss and refine codes. The final codes were applied to the entire set of transcripts. In phase 2, System Usability Scores >68 were deemed ‘acceptable’.⁷

3. Results

In total, 13 women participated in seven Phase 1 small group discussion or interview sessions and nine women (7 from Phase 1 plus 2 new women) participated in Phase 2 wireframe usability evaluations. Among these 15 participants, mean age was 60.7 (SD=5.38) years; the majority belonged to a racial/ethnic minority group (African American (67%), Hispanic (14%)); had some college or less (54%); had Medicaid or Medicare (60%); and a mean BMI of 34.0 (SD=6.62). Most used their smartphone for email or Internet at least once a day (87%) and had ever downloaded an app (87%); but fewer had downloaded a health-related app (33%) or received text updates or alerts about health (40%). Mean years since diagnosis was 11.4 (SD=6.83) and mean years since the end of treatment was 5.9 (SD=6.15). Ratings of the 19 BCT behavior change techniques are presented in Table 1.

Table 1.

Modified Delphi ratings of the importance of selected Behaviour Change Techniques

Behavior change technique†	Round 1‡ mean (SD) range	Round 2‡ mean (SD) range
Feedback on behavior	8.6(1.08) 5-9	8.6(1.08) 5-9
Self-monitoring of the outcome behavior	8.5(0.75) 7-9	8.6(1.08) 7-9
Self-monitoring of the behavior	8.2(1.41) 6-9	8.5(0.75) 7-9
Reducing negative emotions	8.2(1.41) 6-9	8.5(0.75) 6-9
Instruction on how to perform the behavior	8.5(0.75) 5-9	8.5(0.75) 5-9
Review of outcome goals	8.1(1.69) 5-9	8.4(0.84) 7-9
Demonstration of the behavior	8.3(1.14) 5-9	8.3(1.14) 5-9
Review of behavior goals	8.2(1.41) 5-9	8.3(1.14) 5-9
Discrepancy between current behavior and goals	8.2(1.41) 5-9	8.2(1.41) 5-9
Goal setting	8.2(1.41) 5-9	8.2(1.41) 5-9
Biofeedback	7.7(1.98) 3-9	8.1(1.69) 3-9
Graded tasks	7.7(2.13) 1-9	7.8(2.08) 1-9
Problem solving	7.6(1.50) 5-9	7.7(1.49) 5-9
Action planning	7.2(1.62) 5-9	7.5(1.45) 5-9
Adding objects to the environment	7.2(2.32) 2-9	7.4(2.24) 2-9
Behavior practice and rehearsal	7.2(1.62) 5-9	7.4(1.39) 5-9
Nonspecific reward	7.2(2.68) 1-9	7.2(2.68) 1-9
Social support	6.4(2.37) 2-9	6.8(2.15) 2-9
Credible source	5.8(2.20) 2-9	6.2(2.14) 2-9

^†The Behaviour Change Technique Taxonomy version 1 (BCTTv1)

^‡Ratings followed a scale of 1[not at all important] to 9[very important].

In the open-ended discussion, we inductively coded three main topics: technology design, wellness, and targeting for breast cancer survivors (see online supplemental materials). For technology design, the ideas of simplicity and burden came up most frequently, along with preference for new/engaging information. For the second topic, wellness, participants noted the app should be inclusive of multiple wellness topics, including stress reduction and social relationships. For third topic, targeting the app to breast cancer survivors, the majority of participants wanted the app to include features geared towards survivorship issues, for example, coordinating with and showcasing local breast cancer-related events.

In phase 2, the mean SUS percentile score of the wireframe exemplars (Figure 1) was acceptable at 79.

Figure 1.

Screen shot of goals wireframe pertaining to goal setting and action planning behavior change techniques

4. Discussion and Conclusions

Breast cancer survivors from underserved populations rated most BCTs highly, with techniques related to feedback and monitoring, goals and planning, reducing negative emotions, and instruction/demonstration of behavior being most highly rated. Some techniques, such as social support, commonly included in trials for weight management⁵ had wide ranges of perceived importance; bringing down the mean rating in our small sample. This could be because social support features can take many shapes in mHealth interventions (e.g., Facebook group, coaching calls, message boards), which have varying levels of appeal.⁹ Future research should further explore which types of social support are most valued by breast cancer survivors from underserved populations. Open-ended discussion revealed women desire an mHealth app that is technologically simple, has low burden, and is engaging, which have implications for design features. For example, self-monitoring could rely on automatic uploading of data (e.g., steps, weight) when possible, or use less burdensome methods of self-monitoring of diet behaviors, such as monitoring food groups (vs. entering specific foods). These results share similarities with cancer survivors’ preferences for physical activity apps, in which survivors also desired less burdensome tracking methods and advice tailored to cancer experience, but perhaps had less emphasis overall on the need for low burden/technologically simple apps.¹⁰ Based on the high acceptability rating of the wireframe exemplars, future development, usability testing, and field testing of a complete app package are next research steps. Future quantitative study could examine patterns between participant characteristics and preferences for BCTs.

4.1. Study limitations

Although the modified Delphi process aimed to achieve consensus among participants’ ratings, the small sample size suggests the absolute mean rating should be viewed with caution. Future Delphi studies via online discussion panels may decrease potential for social desirability bias.

4.2. Clinical implications

Clinicians should encourage breast cancer survivors from underserved groups to utilize mobile apps targeted to survivorship issues, that are simple-to-use, engaging, and that contain evidenced-based features of behavior change.

• mHealth intervention features for cancer survivors from underserved populations have not been well specified

• We examined perceptions of Behaviour Change Techniques in apps

• Most Behaviour Change Techniques were rated highly, including self-monitoring and feedback

• Participants valued technologically simple apps that incorporated wellness and focused on breast cancer survivors

• Given good/excellent perceived acceptability, future research will involve complete app development and testing