The authors have declared that no competing interests exist.
Provided edits/additions to the manuscript: LAG SJW JLS MPB BB TNR JC. Data cleaning/verification: LAG SJW JLS JC. Conceived and designed the experiments: ACK EBH MPB BB TNR. Performed the experiments: EBH MPB LAG SJW JLS JC. Analyzed the data: EBH LAG. Wrote the paper: ACK EBH.
Mobile devices are a promising channel for delivering just-in-time guidance and support for improving key daily health behaviors. Despite an explosion of mobile phone applications aimed at physical activity and other health behaviors, few have been based on theoretically derived constructs and empirical evidence. Eighty adults ages 45 years and older who were insufficiently physically active, engaged in prolonged daily sitting, and were new to smartphone technology, participated in iterative design development and feasibility testing of three daily activity smartphone applications based on motivational frames drawn from behavioral science theory and evidence. An “analytically” framed custom application focused on personalized goal setting, self-monitoring, and active problem solving around barriers to behavior change. A “socially” framed custom application focused on social comparisons, norms, and support. An “affectively” framed custom application focused on operant conditioning principles of reinforcement scheduling and emotional transference to an avatar, whose movements and behaviors reflected the physical activity and sedentary levels of the user. To explore the applications' initial efficacy in changing regular physical activity and leisure-time sitting, behavioral changes were assessed across eight weeks in 68 participants using the CHAMPS physical activity questionnaire and the Australian sedentary behavior questionnaire. User acceptability of and satisfaction with the applications was explored via a post-intervention user survey. The results indicated that the three applications were sufficiently robust to significantly improve regular moderate-to-vigorous intensity physical activity and decrease leisure-time sitting during the 8-week behavioral adoption period. Acceptability of the applications was confirmed in the post-intervention surveys for this sample of midlife and older adults new to smartphone technology. Preliminary data exploring sustained use of the applications across a longer time period yielded promising results. The results support further systematic investigation of the efficacy of the applications for changing these key health-promoting behaviors.
The major killers of adults in the U.S. and many countries worldwide–non-communicable diseases such as cardiovascular disease, cancer, stroke, and type 2 diabetes–have been demonstrably linked to a small number of key health behaviors, including regular physical activity
Most evidence-based interventions to promote regular physical activity have used in-person instructional formats delivered in clinical or community settings. Among the constraints of such approaches are staff time, intervention fidelity challenges, transportation and venue costs, and reduced intervention personalization when group formats are applied. Such resource-intensive approaches can be especially problematic for sustaining health behavior change over time.
“Tele-health” and other mediated approaches to health behavior change provide an empirically supported, convenient, and potentially lower-cost alternative for reaching large proportions of the public over a longer period of time
Advances in built-in smartphone activity sensors (i.e., accelerometers) provide a timely and potentially cost-efficient means for enhancing ongoing physical activity participation. In addition to continuous activity monitoring, such devices can be programmed to provide automated, behaviorally and contextually tailored information to facilitate health behavior change throughout the day and across a variety of settings.
There has been an explosion of smartphone health promotion applications (apps), including physical activity apps, which have been projected to be in the thousands (e.g., a search in the Apple App Store for “Health and Fitness” resulted in more than 2000 apps available for the iPhone). Yet, relatively few apps have drawn explicitly from relevant behavior change theory or evidence or have undergone systematic evaluation
This first-generation feasibility study aimed to apply a behavioral science-informed user experience design (BSUED) process
The Stanford University School of Medicine Human Subjects Institutional Review Board approved this study. All participants provided written informed consent.
An interdisciplinary team of behavioral and exercise scientists, health experts, computer scientists, and engineers collaborated in constructing distinct physical activity smartphone applications using behavioral science-informed user experience design (BSUED)
Three behavior change apps to promote regular physical activity and reduce sedentary behavior, based on three distinct motivational frames drawn from behavioral science theory and evidence, were constructed and iteratively tested. One physical activity/sedentary behavior app applied an
A second physical activity app applied a
A third physical activity app applied an
After arriving at the three behavioral frames, a “design thinking” approach was used to rapidly iterate through concept exploration, prototypes, and ethnographic testing of the user-application interfaces to inform the interaction design architecture for the three apps.
In addition to the above three behavior change apps, an app to compile, analyze, and integrate the built-in accelerometer data being collected on a continuous basis from the project smartphones with the intervention apps was developed. This app, which was programmed to provide “just-in-time” feedback to users of all three behavior change apps using algorithms based on the national recommendations for physical activity (i.e., 150 minutes or more per week of moderate-intensity physical activities such as walking)
The target population consisted of community-dwelling adults ages 45 years and older who were insufficiently physically active (i.e., engaged in less than 60 minutes of moderate or more vigorous physical activity per week that increased heart rate, breathing, or perspiration), reported typically sitting for 10 or more hours per day, were able to participate safely in a physical activity program based on responses to the physical activity readiness questionnaire
During the period from January 2010–March 2011, the interdisciplinary project team undertook initial app design, programming, and iterative user testing. Twelve individuals meeting the project eligibility criteria described above participated in formative evaluation and user testing of the project apps. Among the activities initiated during this phase of the project was determining the most powerful motivational frames to test based on the behavioral science literature and user interviews, along with the most relevant evidence-based behavior change strategies and techniques accompanying each frame
The Android smartphone platform was utilized in light of its capabilities with respect to “live wall paper” displays, ease of programming, and ability to run the continuous built-in accelerometer in the background simultaneous with other apps. To prolong the smartphone battery life sufficiently to allow for continuous accelerometer data capture throughout the day, the phone's default battery was replaced with an extended life battery. For all three apps, the data being collected via the smartphone's built-in accelerometer were available within a smartphone-based database for use by the three apps to provide individualized feedback throughout the day. These data were transmitted, via an encrypted protocol, to the project's local servers each evening for data storage and to allow researchers to monitor the quality of data while the study progressed (see below).
The behavioral components for the three apps are summarized in
Analytic | Affect | Social | |
“Push” component (i.e., notifications) | X | X | X |
“Pull” component (i.e., information found via participant selecting an icon) | X | X | X |
“Glance-able” display | X | X | X |
Passive activity assessment | X | X | X |
Real-time feedback | X | X | X |
Self-monitoring | X | X | X |
“Help” tab | X | X | X |
Goal-setting | X | ||
Feedback about goals | X | ||
Problem-solving | X | X | |
Reinforcement | X | X | X |
Variable interval reinforcement schedule | X | X | |
Attachment | X | ||
“Play” | X | ||
“Jack pot” random reinforcement | X | ||
Social norm comparison | X | ||
Competition/collaboration | X |
In addition to the basic elements described above, each of the three behavior change apps had distinct elements based on the theoretical and empirical literature underlying the specific motivational frame being applied. For the
For the
Similar to the analytic app, a history tab was also available whereby participants using the social app could see a visual summary of the overall history of their physical activity and sitting time. In contrast to the analytic app, however, all personal data in the social app were displayed in reference to group averages (e.g., for daily physical activity level, a participant would see one line representing his/her physical activity for that day along with two other lines in the graph representing the average physical activity for his/her own group and the other group). This history configuration was used to further emphasize the social comparison aspects of this app.
A participant electronic “message board” was also available to users of the social app, where participants could post, in real-time, comments, suggestions, or other information they deemed appropriate to the other individuals assigned to the social app. The message board was not officially moderated (i.e., all posts were immediately shared with others participating in the social app arm of the study). However, to positively influence the injunctive norms and etiquette within the message board, study confederates posted, at least weekly, comments and information that included similar types of material that were incorporated into the informational “tips” and problem-solving strategies used in the analytic app, but written in a style that would conform with the conversational language typically used on electronic message boards. When participants were randomized to the social app, they were given instructions on how to post to the message board and were asked to post a brief introduction of themselves, without divulging any identifying information (e.g., name, exact age, address, phone number, etc.) to ensure confidentiality.
For the
Positive reinforcement occurred in two ways. First, whenever a person reached pre-specified thresholds related to physical activity or trajectories of sedentary behavior (e.g., through engaging in “breaks” from prolonged sedentary behavior throughout the day), the bird would appear on the live wallpaper and give the person a “thumbs up” while making a melodious sound. These “reinforcers” were delivered on an expanding variable interval reinforcement schedule, demonstrated in the literature to be useful for promoting ongoing maintenance of behavior
Similar to both the analytic and social app, there were options for the participant to view his/her progress and history. Specifically, with each progressive activity level increase, the bird flew higher up on the screen. The highest level attained on the screen was used as a visual “history” of activity across the day. In addition, as the participant accumulated more daily activity, following which the bird reached more distant destinations of interest, pictures of those destinations were added to a “travel” tab whereby all of the places the bird visited were displayed.
Following the app development activities, we conducted feasibility and fidelity testing of the three apps with respect to their capabilities for impacting initial physical activity and sedentary activity levels. We recruited participants in two waves (n for wave 1 = 27; n for wave 2 = 41). For each wave, individuals meeting the eligibility criteria were randomly assigned, using a computerized version of the Efron procedure
To assess the feasibility of increasing regular physical activity and decreasing sedentary activity throughout the day using the three apps, participants completed standard self-administered questionnaires at baseline and at the end of the 8-week intervention period. These questionnaires were used as the primary assessment instruments in this initial app feasibility and fidelity testing procedure because they are straightforward to collect and analyze, and are among the most commonly used outcome measures in the physical activity/sedentary behavior field for the target age group. Using such measures also allowed us to compare the behavior changes observed using our apps to those reported in other physical activity intervention studies, including those using other information technologies, in the same age group.
To assess physical activity levels, the CHAMPS Physical Activity Questionnaire was used. This instrument has been found to provide a valid and reliable estimate of usual physical activity behavior, including walking, in middle- and older-aged adults
To assess sedentary behavior levels, the Australian sedentary behavior questionnaire (referred to as the Measure of Older Adults' Sedentary Time [MOST]) was used
To evaluate user acceptability of the apps, participants completed a user satisfaction survey at the end of the 8-week intervention period. The survey, adapted from similar user satisfaction surveys in this age group
To evaluate the feasibility of each app for improving initial physical activity and sedentary behavior patterns, pre-post paired-comparison
The 68 adults participating in the intervention app feasibility testing protocol were an average of 59.1±9.2 years old (range = 45–81 years), with 73.5% women. Seventy-six percent had a college degree, 51.4% had an annual household income of $70,000 or greater, 48.5% were working full-time, and 39.7% reported being currently married. Sixty-nine percent were non-Hispanic White, 13% were Hispanic/Latino, and 12% were Asian. Mean body mass index (BMI) was 29.6±6.2. A third of the sample was randomized to each smartphone intervention app (Analytic n = 22; Social n = 23; Affect n = 23), with no significant between-group baseline differences found in the demographic variables or baseline physical activity or sedentary behavior variables (
While all but one participant was successful in using their assigned smartphone app through at least 5 weeks of the 8-week protocol, 7 participants were missing post-test physical activity or sedentary behavior questionnaire data (i.e., 10.3%). Missing questionnaire data were due to participant time constraints or not properly filling out the questionnaires. Within the constraints imposed by analysis of subgroups with small n's, independent
Participants across all three apps reported significant mean increases in weekly minutes of brisk walking across the 8-week intervention period (paired
Study participants also reported significant decreases in the daily amount of discretionary time they spent sitting in front of the television (paired
In general, participants reported positive experiences with the three apps. The majority of the sample (87%) reported that they found the apps easy to use; 77% reported that the length of time needed to use the apps “was about right”, and only 11% reported that the number of contacts with the apps “was too many”. Only 16% reported having a hard time remembering to use the apps. Over two-thirds of participants (69%) reported that the apps motivated them to be more physically active and to sit less (74%), and the majority of participants reported that the apps helped them remember to exercise regularly (71%) and made them aware of their sitting time (87%).
After using the smartphone apps for an 8-week period, this initially smartphone-naïve sample of midlife and older adults reported generally more positive than negative attitudes related to smartphones in general. For example, 91% agreed that smartphones are a fast and efficient means of gaining information, and 85% agreed that smartphone applications have unlimited possibilities that have yet to be thought of. Relatively few participants reported that smartphones made them uncomfortable because they did not understand them (9%), or were intimidating because of their complexity (18%). No significant between-app differences were discerned, but small sample sizes reduced power to detect such differences.
With respect to difficulties with the apps reported during the study, we found such user difficulties to be, for the most part, relatively minor and readily resolved. The most common difficulties experienced by users included questions concerning whether the app was registering physical activity consistently (44%; after ensuring that the app was working properly, staff checked the phone to ensure that it was being used and worn properly, and participants were provided with some additional instructions in phone use and the importance of attaining the moderate-intensity or more vigorous levels of activity upon which the feedback was based); reports that the phone with the extended battery was heavy to carry (29%; participants were given carrying pouches or belts on which to attach the phones if they did not have appropriate ones); and reported difficulties using some of the general smartphone features (e.g., making and receiving calls, retrieving voice mail, etc.) (23%; staff produced simplified instruction sheets addressing the most frequently asked questions in this area). Typical of mobile phone use more generally, 18% reported dropped calls or poor mobile phone coverage from time to time, and 9.5% reported some difficulty reading the mobile phone screen. None of the above difficulties led to participant loss to follow-up.
To begin to explore how long participants would be willing to continue using the custom apps following the 8-week study period, 12 participants enrolled in the latter stage of the study (4 from each app group) were allowed to continue to access their assigned app if they so chose until the investigators collected all smartphones on day 233 post-study. Participants were approached in consecutive order just prior to their 8-week study end date and were invited to continue using their assigned apps until the number that agreed reached 12 (4 from each app group). Of the 15 participants approached, 80% were willing to continue using their assigned app. The reason the three participants gave for declining further participation concerned their disinclination to continue being “connected” this intensively to their mobile phones.
These 12 participants continued to use their apps for a mean of 191±33 days post-study (range = 120–233) (Analytic: M = 211.0±19.0 days; Social: M = 199.3±27.8 days; Affect: M = 162.0±33.5 days). Over half (53.5%) of these participants, who completed an additional user satisfaction survey at the end of this maintenance period, indicated that they would be willing to use their assigned app for an additional 6 months or longer. In addition, 70% indicated that they would recommend the app to others.
While there has been a steady rise in mobile device applications aimed at promoting regular physical activity and related health behaviors, few have drawn systematically from behavioral science theory or evidence. This first-generation feasibility study applied a behavioral science-informed user experience design (BSUED) process
The apps also appeared to be useful for decreasing the average number of daily minutes participants spent sitting in front of the television—a highly prevalent discretionary sedentary behavior among individuals in this midlife and older age group
We found all three apps to be generally easy to use and acceptable by the current sample of participants, who had no prior experience with smartphones. Given participants' initial levels of inactivity, careful instruction on the overall physical activity goals targeted in the three apps, i.e., accumulating physical activity of at least moderate intensity (akin to brisk walking) for 10 minutes or more at a time, occurred at the beginning of the interventions and when participant questions arose concerning the feedback they were receiving from the apps.
While the sample was well educated, 25% were from racial/ethnic minority groups, a significant proportion were women, and all were from the aging population segment—groups that traditionally have been under-represented in information technology-based health behavior research
Among the methodological limitations of this study is the lack of an appropriate control group against which to directly compare the effects of the three smartphone apps. The next step in this line of research is to investigate systematically the efficacy of the smartphone apps relative to such a control
In conclusion, integrating behavioral science theory and evidence with an iterative user-oriented design process may enhance the potency of mobile device applications aimed at promoting behavior change in key health areas such as physical activity and sedentary behaviors. The current results set the stage for systematic investigations of such applications within the context of experimental studies as well as in comparison to commercially available programs.
We thank Martin Alonso for his design expertise, and Frank Chen, Elizabeth Mezias, and Anthony Nguyen for their programming activities for the project.