This study was an experimental, pre-post single-arm trial. The evaluation lasted for 12 weeks.

The study took place in the eastern part of Norway during spring and early summer 2017. Patients were recruited from two rehabilitation centers. One rehabilitation center offered a 12-week CR program and the other offered 1-week and 4-week programs. Approximately one-third of the participants were recruited from each of the three CR programs for this feasibility trial, the same proportions planned for the upcoming RCT.

Eligible patients were women and men over the age of 40 years who completed CR in one of the three programs during a period of two weeks. They had to own and use an Android or iOS mobile phone and be able to read and understand Norwegian or English. Exclusion criteria were restrictions regarding exercise intensity for any reason due to the primary end point in the planned RCT, which is intended to be maximal oxygen consumption (VO_2peak). Descriptive data collected at baseline included sex, age, diagnosis, treatment, history of smoking, educational level, exercise habits last year, and VO_2peak.

Patients received the app after attendance in CR. The app was developed to guide and help individuals change behavior and maintain habits. The follow-up was based on the transtheoretical model of behavior change [19]. According to this model, health behavior change involves progress through six stages of change: precontemplation, contemplation, preparation, action, maintenance, and termination [19]. In this connection, motivational interviews are used to help people access motivation to change a particular behavior through collaboration, evocation, autonomy, and exploration [20]. The patients are supposed to set goals that are small, important to them, specific, and realistic to achieve [21]. The app used in this study permits the user to create and set such goals (Figure 1) with tasks and accompanying reminders. A supervisor has access to an administrator interface (Figure 2) and can monitor the goals and tasks of each patient. In addition, the patient can write reflections in the app that the supervisor can read in the administrator interface. The app itself provides reminders and evaluations of tasks and weekly goal achievement that automatically pop-up. In these evaluations, the patients must reply with a red or green face depending on whether they have completed the planned tasks or not and rate the weekly goal achievement on a scale from 0 to 100.

At baseline, a supervisor guided the patients in setting individual goals by using elements from motivational interviewing. The supervisor in the study was a physiotherapist specializing in cardiovascular and pulmonary physiotherapy with five years of experience in CR. Each patient was encouraged to set a minimum of two goals with related tasks to be able to reach each goal. The patient decided when and how often reminders of the tasks should appear on their mobile phone. During the follow-up period, the patients received short, tailored, individualized motivational feedback directly through the app 1 to 3 times a week and comprehensive individual feedback through email once a week. Patients could submit questions to the supervisor at any time, receiving an answer within 2 working days. If the question was medically related (eg, changing medication or chest pain), patients were advised to contact their general practitioner. Patients were followed for 12 weeks by the same supervisor who included the patients at baseline.

In order to determine whether follow-up of patients after CR through an app was feasible in an RCT, we chose the following criteria for success:

Based on Treweeks’ [27] recommendations for pilot and feasibility trials, we needed 10-15 patients to be able to have confidence in the conclusions drawn from the data. Data were analyzed using SPSS Statistics for Windows version 24.0 (IBM Corp). Descriptive statistics are reported for each case and in means and standard deviations for the whole group. Differences in outcome variables (baseline to 12-week) were analyzed using nonparametric tests due to the small number of patients. Significance was set to P<.05. In case of missing data, we used the last observation carried forward method.

The Regional Committee for Medical Research Ethics, Region Eastern Norway, reviewed the study and found that approval was not required. All participants provided written informed consent.

In total, 24 patients were available for inclusion in the study (Figure 3), and 17 (71%) were eligible for the potential RCT. Ultimately, 14 patients were enrolled in this study.

Half (50%) of the patients were iOS users, and half were Android users. Approximately one-third had attended each of the three CR programs—12 weeks, 4 weeks, and 1 week. Table 1 provides the baseline characteristics: 71% (10/14) were men, mean age for all participants was 60.1 (SD 8.5) years, and mean VO_2peak was 27.6 (SD 6.2) mL/kg/min.

All patients used the app regularly throughout the 12 weeks. Additionally, all patients answered all tasks within a week. Table 2 provides the app use for all patients. All patients had goals or tasks related to exercise or fitness training, and 9 of 14 had goals or tasks related to maintaining or improving their dietary habits. The mean numbers of individual goals, tasks, and weekly reminders were 1.9 (SD 0.5), 3.5 (SD 1.1), and 10.3 (SD 4.5), respectively. The mean SUS score for all patients was 84.8 (SD 12.8). iOS users scored higher in SUS than Android users (90.4 [SD 7.7] vs 79.3 [SD 15.0]) (Mann-Whitney test, P=.12), but the difference was not significant. Patients scored 96.8 (SD 7.2) on the questions on usefulness of the app and 91.3 (SD 14.2) on questions regarding their own motivation. Most of the patients (9/14, 64%) reported it to be useful to use the app for 6 to 12 months after attendance in CR. All patients felt it was very important that they were closely monitored by the supervisor during the first months. After that, monitoring could be less frequent.

Only minor problems with the technology appeared during the study. Patients could not report that they had completed tasks for a 9-hour period, and for 4 weeks, patients could not save the score on the weekly perceived goal achievement question that appeared in the app.

The supervisor spent approximately one hour to include each patient to the study. During this hour, the supervisor obtained written consent; collected sociodemographic data; created a user for the app in the administrator interface; helped the patient download the app and set realistic, specific, important, and individual goals and tasks; and trained the patient to use the app. Thereafter, time spent monitoring patients was, on average, 6 minutes per patient per week for the 12 weeks. In addition, on average 7 minutes each week was spent answering patient emails and 9 minutes was spent talking to the service provider about bug fixes and update needs of the app.

The domain physical fitness in COOP/WONCA improved from 2.2 (SD 1.0) to 1.9 (SD 0.9), P=.046. There were no statistically significant changes in any of the other domains. The domain pain and discomfort in EQ-5D improved significantly, from 2.1 (SD 1.1) to 1.8 (SD 1.1) (Wilcoxon signed-rank test, P=.046). There were no statistically significant changes in any of the other domains. On the SF-36, no statistically significant changes were found in any of the domains.

Mean scores with standard deviations for both baseline and 12 weeks with P values of the changes are presented in Table 3 in addition to minimum and maximum observed scores and percent of ceiling and floor effects for each questionnaire at baseline, with associated domains.

Mean scores of perceived goal achievement, week by week, are presented in Figure 4. There was a statistically significant improvement in goal achievement from baseline to week 12 with a mean change of 41.2 (SD 39.0) (P=.002). None reached ceiling or floor effect. Distribution of scores in COOP/WONCA and EQ-5D at baseline and after 12 weeks are presented in Multimedia Appendix 2 and 3, respectively.

To our knowledge, this is the first study to evaluate the feasibility of using an app as a tool to promote and monitor adherence to a heart-healthy lifestyle after attendance in CR with predefined criteria for success. Results demonstrated a high recruitment rate and high adherence to use of the app. In total, 71% of available patients were eligible and wanted to participate in the study, and all patients used the app during the entire intervention period and answered all tasks. The supervisor spent in average of 6 minutes each week to give individualized feedback to each patient. Quality of life, health state, and self-perceived goal achievement improved; however, we observed a ceiling effect in questionnaires measuring quality of life and health status. The strengths of feasibility studies are to report the possible pitfalls of a large RCT, weigh strengths against weaknesses of the intervention, investigate the feasibility of patient recruitment and outcome measures, and come up with solutions on how to conduct the RCT [28]. A strength of our study was its clear eligibility criteria and rigorous protocol, which ensured that the sample included the targeted patient population and accurate data collection at predefined study time points.

Our findings are in line with results from studies evaluating mHealth interventions for chronic disease management. In a systematic review [10], 62 of 107 included studies evaluated usability, feasibility, and acceptability of mHealth interventions. The most used mHealth intervention was text messaging. The number of studies including mobile phone apps is not specified, but it is stated that 25 used specialized software or a mobile phone app. Generally, the review concluded that the usability, feasibility, and acceptability of mHealth tools were high in connection with chronic disease management. Both patients and providers appreciated the mHealth tools [10]. Specific results from studies with mobile phone apps are not presented. In addition, none of the 25 studies that used specialized software or a mobile phone app were conducted in heart patients [10]. A review by our group determined that mobile phone apps seem to be most common in studies conducted in patients with diabetes mellitus [17].

According to the technology acceptance model, usefulness and ease of use are key factors that influence whether people accept or reject technology and thereby influence adherence to a technology-based intervention [29]. In this study, the overall satisfaction with technology measured with the SUS was 84.8. This score is considered as very high with a high degree of usability [30]. There was a difference in SUS score between Android and iOS users in favor of the iOS users. This is not surprising as the iOS operating system has a unified user interface for all mobile phones, whereas the Android operating system comprises several user interfaces due to the wide range of mobile phone producers using the platform. Because of the difference in SUS score between Android and iOS users, it is necessary to make the Android version of the app more stable for the upcoming RCT.

Patients reported clear and realistic goals for a heart-healthy lifestyle that they were able to evaluate weekly in the app. It was surprisingly easy to guide the patients in setting goals. This can be explained because goal-setting is often used as an approach in CR programs [31]. Both CR centers from where the patients were recruited emphasize goal-setting in their CR program. In addition, the supervisor’s background and experience in CR may have contributed to the effective goal-setting. Continuation of focus on their own goals for a heart-healthy lifestyle in the follow-up after attendance in CR may have contributed to patients’ perceived usefulness of the intervention and perceiving it would be useful to be followed up for a longer period of time. Most patients reported that it would be beneficial to use the app for a year. Whether the patients actually preserved or improved their exercise capacity or nutritional-related goals is still uncertain, but the results from this feasibility study support moving on to the RCT.

Several points of guidance from the patients in the study have been adopted and incorporated into the final design. First, although satisfaction with the technology was high, some potential improvements were discovered. One to three times a week, individualized motivational messages were sent to each patient. These messages appeared on the patients’ mobile phones as a push notification. It turned out that several of the patients were not familiar with push notifications, and therefore these messages were lost without some of the patients having read the content. Based on the feedback, the app has been adjusted, and individualized motivational messages are saved in the app. Each patient can then decide when they want to read and delete them. A technical problem with weekly goal achievement was fixed during the fifth week of the study, and it is now fully functional. In addition, there were some options in the app with regard to when a task should start. In example, patients could create a task with any start time they wanted. This function did not work properly, and patients reported that they didn’t need it. Therefore, the functionality has been deleted in order to keep the app as simple and easy to use as possible.

Quality of life was evaluated with the questionnaires SF-36 and COOP/WONCA. It turned out that more than 20% of the patients achieved ceiling effect on 50% or more of the domains in these questionnaires, which makes it difficult to detect any improvement in these domains. Floor effect was achieved only in the domain physical role functioning on the SF-36. The high number of patients reaching the upper limits may have been a result of the non–disease-specific questionnaires that were used. Additionally, the included patients were relatively young, not in any acute phase of disease or illness, and had just completed an extensive rehabilitation program [32]. To be able to evaluate possible changes in quality of life in the upcoming RCT, we have decided to use the HeartQoL health-related quality of life questionnaire. HeartQoL has been found to be both valid and reliable in patients with the primary diagnoses the CR patients normally have (eg, angina, myocardial infarction and heart failure [33], stable coronary artery disease [34], atrial fibrillation [35] as well as in patients with implantable cardioverter defibrillators [36] and patients following heart valve surgery [37]). On the EQ-5D, ceiling effect was reached for more than 30% of the patients in all domains. Again, this can be explained by the patients’ relatively young age and the inclusion of nonhospitalized patients [32] and is in line with other research on this population [38,39]. Despite this, we have chosen to keep the EQ-5D in the planned RCT due to its ability to conduct health economic statistics and because HeartQoL doesn’t include an overall health status like EQ-VAS.

In line with the general Norwegian population according to Statistics Norway [40], 92% of the patients at CR were owners and users of mobile phones. Although 71% of the patients were eligible for inclusion in the study and therefore for the upcoming RCT, it is not likely that these patients would sustain participation if they perceived that randomization to the control group resulted in an inferior intervention.

Although the results from this feasibility study are promising for the upcoming RCT, we have to be aware that the patients in the study were only followed for three months, and it is reasonable to believe that there will be dropouts when the study runs over a year. This must be taken into account in the calculation of how many participants will be needed to detect an effect in the RCT, and we have added 20% for possible dropouts.

Based on preset criteria for success, our study shows that an intervention with an app that allows individualized monitoring after attendance in CR is feasible. All patients used the app to get help for preventive activities such as exercise and dietary change. Implementation of mobile phone apps as a tool to promote adherence to preventive activities after CR is a novel approach. Since research in this area is warranted, this paper may serve as a foundation for other upcoming RCTs as well and inform the development of RCT management.