The Smart AF app was developed by Health Tech Innovation Center, in association with Kyoto Prefectural University of Medicine, funded by the BMS/Pfizer Japan Thrombosis Investigator Initiated Research Program (JRISTA). There are 3 features of the Smart AF app (Figure 1).

There are 7 components to the patient educational program. With videos that are approximately 1-2 minutes in length, patients can learn about AF and learn self-management methods, including how to detect and treat AF, the importance of anticoagulation, and the treatment of comorbidities.

After inputting information about the characteristics of their AF history, other medical history, medication information (eg, type of antithrombotic medication or other concomitant medications), and lifestyle behaviors, each patient’s CHADS2 score is automatically calculated, enabling physicians to easily understand their risk for stroke. Furthermore, the app can take inputs on step counts, sleep time, and presence or absence of symptoms and, by sharing this information with health care providers, the proactive involvement of the patient in their own care can be promoted and supported.

A reminder is automatically transmitted through the app in the morning and evening daily to prevent forgetting to take medication.

Follow-up was planned at 1, 3, and 6 months after enrollment, and a reminder email for the survey was automatically sent to the patient's app at the time of follow-up. The self-reported 8-item Morisky Medication Adherence Scale (MMAS-8) was used as the survey instrument [18-20]. The MMAS-8 score assesses patients' self-reported adherence to their anticoagulant medication. According to the MMAS-8 score (range 0-8), adherence was defined as high (score 8), medium (score 6 to <8) or low (score <6).

Patient enrollment was carried out by obtaining informed consent from patients with AF attending Kyoto Prefectural University of Medicine hospitals between May 2019 and September 2020. All participants provided electronic informed consent and were assigned a password. Inclusion criteria were as follows: documented diagnosis of AF; current prescription for OACs (ie, dabigatran, rivaroxaban, apixaban, edoxaban, and warfarin) for at least 3 weeks; and ownership of a mobile phone. Individuals less than 20 years of age, those with valvular AF, and those who had been taking OACs for less than 3 weeks were excluded. This single-center prospective observational study was approved by the Institutional Review Board of the Kyoto Prefectural University of Medicine (ERB-C-1429).

Statistical analyses were performed using R version 3.6.1 (R Foundation for Statistical Computing). Continuous variables are expressed as mean and SD, and categorical variables as number and percentage. MMAS-8 scores at 4 time points (baseline, 1, 3, and 6 months later) were plotted to illustrate changes in these variables over time. The Wilcoxon signed-rank test was used to evaluate differences in MMAS-8 scores between baseline and 1, 3, and 6 months. For the 96 patients who were able to complete the follow-up up to 6 months, the change in adherence from the time of enrollment to 6 months is reflected graphically by a low, middle, and high MMAS score. The percentage of patients who activated the app at least once per day is expressed as the mobile app retention rate, and the mean retention rate with SD per week at 1, 3, and 6 months was calculated and graphed. Associated factors were also identified using simple regression analysis. “Low retention,” defined as the percentage of days in which the app was activated at least once, was ≤10% during the observation period. Differences with P<.05 were considered to be statistically significant.

This study was funded by JRISTA. Permission for use of the MMAS-8 scale and its coding has been acquired, and a license agreement is available from MMAR, LLC, Donald E Morisky, ScD, ScM, MSPH, 294 Lindura Ct., United States.

The Smart AF app has been developed, which integrates patient education, reminder alarm, and patient engagement strategies without increasing the burden on health care providers. Elderly patients with AF who used the Smart AF app demonstrated significantly improved MMAS-8 at 6 months compared to baseline. The Smart AF app demonstrated a significant decrease in the low adherence group and an increase in the high adherence group at 6 months compared with baseline.

Current clinical guidelines for AF advocate incorporating patient preferences regarding treatment and support and involving patients in management decisions [21,22]. The participation of patients with AF in the process of developing and refining the app through patient involvement resulted in the Smart AF app. For example, to facilitate the use of the app, age-related aspects considered to be useful in the design of mHealth (mobile health) tools, including large screens, large fonts, and ease of navigation, were incorporated into it. As a result, we adopted a simple layout and large navigation buttons that are easy to use, even by elderly individuals. Among the few studies that evaluated the effectiveness of smartphone apps in improving medication adherence in those with AF, research investigating the mAF app revealed that it had a complex design with a great deal of content, such as education programs, clinical decision support material, patient involvement in self-care, structured follow-up, and many areas that users needed to manipulate manually [17]. Regarding the Health Buddies App, there are some problems, such as a lack of interest in the characteristic games of the app, to the point that grandchildren who become “Buddies” must also participate; moreover, although achieving tasks and goals inside the app was fun, long-term behavioral change was difficult because these achievements do not correlate directly with patient health conditions [16].

The Smart AF app was developed with a focus on patient education, reminder alarms, and patient engagement in self-care, and we tried to keep the app simple without any other features. In fact, the FACILITA study also revealed that a mixed intervention, consisting of patient education and a simple calendar reminder of drug intake, was an effective strategy to improve adherence to dabigatran (to 91% and 89% at 6 and 12 months, respectively, compared with 65% and 63% for the control group) [10]. Therefore, despite the simplicity of the app’s contents, we believe that not only did the reminder alarms encourage patients to take their daily medication, but by having them record their medication in the app's calendar, we could encourage their participation and make them more aware of their engagement in self-care by managing their daily health.

Previous studies have also reported that tailor-made educational interventions can significantly improve anticoagulation management of warfarin [23]. In addition, the IMPACT-AF (Integrated Management Program Advancing Community Treatment of Atrial Fibrillation) study, which investigated the use of oral anticoagulation in patients with AF, found that it was improved by a multifaceted, multilevel (including at medication initiation as well as at follow-up) educational program implemented by physicians [24]. The Smart AF app has educational contents to help patients quickly solve their questions about AF. Thus, the smart AF app can enhance patient education and medication reminders at any time and place. In addition, we believe that we have succeeded in encouraging long-term patient engagement in self-care by avoiding complex operations and time-consuming input as much as possible.

The decrease in the retention rate of the app over time is an important issue to be considered. The mean retention rates at 1, 3, and 6 months were 0.48 (SD 3), 0.36 (SD 3), and 0.27 (SD 1), respectively. A decline in app use over time is also a concern that has been highlighted in previous reports [25].

Although the app offers the advantage of completely remote recruitment and enrollment, lack of human communication may mean less motivation for the participants to continue compared with studies conducted face-to-face.

The Smart AF app was designed to pop up notifications (reminder) every morning and evening, even when the app had not been activated or opened. Such pop-up feature may have succeeded in improving medication adherence regardless of whether the app was activated or not. However, long-term use is essential for the app to affect users. Clearly, additional efforts to improve retention rates are necessary. The characteristics associated with the retention time of apps have not been well studied. In a study of patients with asthma, being female and older was related to longer retention [26]. In our study, 95 patients enrolled in the study had an HADS-T score of 20 or higher (major anxiety and depression), and most of them had anxiety and depression. As shown in Table 2, older patients, patients with a regular diet, and patients with anxiety had higher retention rates. Conversely, younger patients, those with irregular diets, and those without anxiety had lower retention rates. To deliver mHealth effectively, it is important to identify patient domain factors, such as psychological factors, dietary regularity, and age, that mark suitable candidates for the app.

In the future, to further increase continued app usage rates, we believe that we must work to incorporate an interactive design; more specifically, design elements that respond immediately to patient operations and behaviors.

There were limitations to our study and findings. First, this was a single-center study, and the patients were not prospectively randomized into intervention (the app group) and usual care groups. Second, while important in this type of study, the reliance on measurement of self-reported results is a challenge often encountered. The data obtained from these indices may be supported by more objective observations. For example, we used a self-reported medication adherence tool to examine medication adherence. Older adults taking multiple medications may not be able to accurately report their medication use status due to poor memory or confusion. Therefore, it may be useful to complement self-reported measures using more objective measures of medication adherence (eg, medical record review and pharmacy documentation). Third, other new interventions, strategies, and technologies designed to enhance long-term adherence to DOACs need to be developed and investigated because patients with AF are a large and diverse patient population, and not all will have access to newer mHealth tools. Nonadherence is often caused by a multitude of factors, indicating the necessity of providing patients with tailored and more personalized tools. Lastly, this app only showed an improvement in adherence for 6 months; therefore, future studies are needed for long-term improvement.

The Smart AF app improved medication adherence among elderly patients with AF. In the realm of medication management, an approach using an mHealth technology that emphasizes education, automatic reminder, and patient engagement may be helpful. The challenge that emerged, however, was the decline in the rate of persistent use of the app over time; therefore, continuous doctor-patient interaction via the app will be necessary in the future.