We conducted an extensive search across Google Play and the App Store to identify all available apps for HF self-care. The search was facilitated with the use of following key terms: HF management, HF manager, HF self-care, HF, and HF tracker. Apps were included in the review if they (1) were HF specific and (2) contained a self-care component (ie, medication, symptom management, reminder system, and behavior tracking). Apps were excluded if they were intended for use in a conference, for education, or for reference purposes.

In accordance with Chindalo et al’s reference architecture, we developed a list of 25 major functions that would promote HF self-care for older adults [19,22-24] (Table 1). These features were identified according to HF self-care and patient-engagement guidelines [19] as well as the expertise from our clinician authors (CD and KK). Before the start of the study, a design session was conducted where CD (cardiologist/HF specialist) and KK (family physician/clinical information technology architect) created individual lists for potential app features. A second design session was conducted with CD, KK, and SW to finalize the list of app functions as well as the specific functions required for app adequacy. All 25 major functions were not deemed required to adequately promote HF self-care but were more beneficial if included. App adequacy was determined if the following standard disease management features were included: (1) diagnosis, (2) weight, (3) behavior tracking, (4) self-care, and (5) notifications. These 5 features were chosen based on their ability to capture factors related to HF management protocol, personalized care for older adults, and health promotion [25,26].

Within each of the 25 functions, a list of descriptors was developed to help specify the components within the listed function. If the app included 1 of the descriptors, the feature was listed as present (Table 1). For example, the self-care feature consisted of 3 components including self-maintenance, self-management, and self-confidence [6,7]. Self-maintenance includes actions associated with treatment adherence, such as taking medication or following treatment regimens. Self-management includes the recognition of and response to changes in symptoms. Finally, self-confidence refers to the individual’s assurance in implementing necessary decisions during the management process. Self-confidence is not an explicit self-care behavior but has been recognized as an important moderator of self-care effectiveness [6,7,27]. Thus, self-confidence would not be captured in app functionality as clearly as self-maintenance and self-management, but instead could be expressed as a series of patient experience–related questions (Figure 1). Overall, if an app included any 1 of the 3 self-care components, the feature would be counted as being present.

A total of 2 reviewers (SW and KK) completed a preliminary screening of apps available in both Google Play and the App Store using the key terms mentioned previously. Following the search, apps were reviewed according to their title and summary description (Figure 1). This screening served as a method to separate the bulk of non-HF self-care apps from further evaluation.

Once screening was completed, a calibration session was held among the 8 reviewers (SW, KK, AG, AD, HW, HS, NN, and DL). During the calibration session, each reviewer was asked to evaluate and score the same 5 apps before the start of the session (Figure 1). Reviewers had a Google Sheet created for them with the list of functions for evaluation. During the session, the 5 apps were then reviewed on a summary sheet, and any discrepancies regarding scoring were discussed and resolved. This allowed us to standardize the training among all reviewers and ensure that we had at least an 85% agreement rate when evaluating the remaining apps. Additional information and comments from the calibration session were recorded and added to the final protocol. The evaluation sheet on Google Sheets was also revised for the final app evaluations.

After the calibration session, the remaining apps were assigned for evaluation, where 2 different reviewers evaluated each app. Each evaluation was completed on the revised Google Sheet with the respective app assignments. Once all the app evaluations were completed, the data were combined into a summary sheet for review. Each app score was then reviewed by the 2 assigned reviewers to ensure that they were within the 85% agreement rate. To determine if the scores met the 85% agreement rate, we conducted interrater agreement statistic and reviewed the kappa value. If the apps did not meet the 85% agreement threshold, the 2 reviewers completed an in-person or virtual evaluation session to review the app discrepancies. Both reviewers were required to provide evidence (ie, screenshot or quote) to support the presence of the feature, and a discussion was held until consensus was achieved. All supporting evidence was sent to SW for a final review. Following consensus, a postreview screening was then completed to filter out any apps that were not HF or self-care specific. The remaining apps within the inclusion criteria were analyzed through a descriptive analysis to assess the app search’s findings.

Reviewers evaluated each app based on the description, screenshots, videos, and reviews available on each app store website. Owing to the limited resources and to ensure a consistent method of app evaluation, we did not download the apps. Our rationale for not downloading apps was also based on the premise that users decide to download an app after reviewing it externally [28-31]. Many of the guidelines assisting patients with choosing a health app have urged users to become more meticulous with the apps they install and, in turn, have provided them with a series of questions to consider before downloading [29-31]. For example, in 2 articles, they suggested users consider the following questions before downloading an app: (1) does the app consider your needs (ie, symptoms and disease management), (2) is the app made by a health care system/physician or by a controversial company (ie, pharmaceutical company), (3) does the app have positive reviews (ie, on the Web or by physicians), (4) is the app regularly updated, and (5) are the app features relevant for you (ie, review screenshots and description of features) [29,30]. Currently, about 90% of most information for decision making about app adoption is available in its documentation [29,32,33]. Given the minimal incremental information available from the app itself, we felt that downloading the apps would not significantly change our evaluation.

In accordance with the guidelines for app review before download, reviewers also extracted the following data from each app: number of downloads, date of last update, cost, and developer [29,30]. However, considering that apps from the App Store do not publicly list their number of downloads or date of last update, reviewers omitted the number of downloads criteria for these apps and used the latest version date for the last update.

Each reviewer selected was equipped with postsecondary experience in the electronic health or health technology field to allow them to effectively evaluate the respective apps. Reviewers were required to follow the training protocol in accordance with the mHealth design architecture as well as attend the calibration session previously described [19].

Self-care is pivotal for HF patients to prevent worsening of HF, yet the majority of current HF apps available are neither HF or self-care specific. To our knowledge, this is the first study to identify and evaluate apps exclusively for HF self-care. We found 21 apps that were both HF and self-care specific. From the 21 apps, few contained key features such as behavior tracking. Apps that included the self-care feature were also listed as only being capable of self-maintenance. Thus, patients would be able to, at most, follow their treatment regimen but would not be able to respond to any changes. Potential features to expand on self-care could include medication titration algorithms to adjust medication doses according to weight fluctuations or the use of telehealth services to connect with a physician to modify their treatment regime [34,35]. Loop diuretics are currently used as the agent of choice for reducing symptoms of HF and controlling weight. Diuretics are traditionally adjusted by physicians. However, with self-directed medication titration becoming more commonly used for chronic disease management, this feature could be an opportunity to improve patient HF self-care in the home setting [34,35].

Our findings suggest that the current available apps are not able to support patients adequately with HF self-care; instead, are in need of further redesign or development. Many developers may have limited resources to accommodate all 25 features in a single app. Therefore, to appropriately engage patients in self-care, apps should at minimum have the following functions: (1) diagnosis, (2) weight, (3) behavior tracking, (4) self-care, and (5) notifications. However, from our systematic search, none of the apps even had these 5 functional features. Not only are these app features key for HF self-care but they can also be easily transferable to other health conditions, such as diabetes or asthma, as it captures the essential components for treatment management. The specifics detailing each feature will differ depending on the condition, but it provides a sufficient baseline category to allow the consumer, researcher, or clinician to incorporate key components for the app evaluation. The consumer, researcher, or clinician may list similar or different components within the 5 features, but with this, they are able to incorporate their perspective within 5 wider categories, while maintaining its relevance for multiple audiences. A prime example of an effective mHealth app is BlueStar from WellDoc Diabetes Management. The BlueStar app is a digital therapeutic for diabetes mellitus type 2 that serves as a virtual coach for patients, providing tailored guidance and facilitating the coordination of diabetes care with their existing care team [36]. BlueStar is a clinically validated tool developed by endocrinologists and clinical diabetes educators and has been evaluated in a clinical trial and reviewed in over 40 publications [37,38]. Patients who used BlueStar showed significant improvements in their diabetes management and reported a high satisfaction when using the app. These improvements are strongly linked to the diabetes expertise leading the development and evaluation of this tool, as their guidance helps ensure that the intervention is aligned with self-management principles integral to patient care [36,37]. Older adults with chronic disease already face many challenges with managing their condition, and the use of technology may further contribute to their difficulties if poorly designed. To ensure apps assist with patient self-care regimens, they should be developed in a manner similar to BlueStar, where specific disease management and patient usability criteria are used to both design and evaluate app effectiveness [37].

One surprising finding from this app search was that the lowest-scoring apps had a relatively higher number of downloads compared with the highest-scoring app (Table 4). These findings ultimately question whether the inclusion of more app features or just the standard features for disease management are more appealing for the end user. Features that patients and consumers view as valuable can vary depending on their self-care abilities. However, many studies have indicated that the primary reason apps fail to maintain user activity is because of the complexity of the app as a whole or the lack of growth in app functionality in accordance with user needs [39,40]. Thus, as effective self-care promotion is the primary goal of app usage, apps would require the inclusion of the standard disease management features, but their presentation should also be modified to accommodate for the challenges older adults face with app use. In previous literature, older adults have indicated that in-app customizable considerations improved the likelihood of their continued usage as they were able to modify their preferences according to their changing needs [20,21]. This could include characteristics such as customizing screen or font sizes and incorporating text to speech, audio cues, or in-app automation features as needed [21,22].

Nevertheless, it is also important to note that the fact that lower-scoring apps had higher downloads can also be attributed to several external factors promoting public app awareness. This includes factors such as effective marketing, links to a national health body, or the use of Web-based search engine optimization [29,32]. From our findings, we found that the app with the highest downloads (My HF) had a moderate score of 12, but it was developed by a privately owned pharmaceutical company that specializes in medication for cardiological conditions (Les Laboratories Servier). Although apps built by health care systems scored the highest, they had much lower download rates (Table 4). With these discrepancies between app scores and the number of downloads, our findings display how higher downloads may not be an appropriate representation for app effectiveness because of its potential ties to a developer with more marketing power.

The limitations of our study were as follows: (1) the apps were not downloaded but were reviewed based on app description, screenshots, videos, and reviews from current/past users, and there are good reasons to believe that the quality of the assessment is not severely compromised, as mentioned above; (2) descriptions for review varied in detail and quality (eg, download data not available for Apple iOS apps); (3) we were only able to review the number of downloads but could not quantify active app use; and (4) actual HF patients were not consulted to define the criteria for adequate HF self-care app.

Future studies should involve end users to better understand their needs with the design of an app to ensure the uptake and usability of an intervention. Specifically, the use of engagement strategies with HF patients and health care providers would be strongly desirable to ensure the findings of this study are congruent with what is experienced in reality. In our app evaluation, we incorporated app user reviews to assess user perspectives, but this method is limited to the feedback available and the quality of the responses on the Web. This study also did not evaluate the value of the 5 functional criteria for HF self-care compared with the remaining categories. Future studies should aim to understand the relative importance of each criteria in relation to patient outcomes, potentially with the use of focus groups or user testing to develop priority weightings for each function. In addition to this, as the potential rise in scale modification for disease specificity could lead to inappropriate features being selected for app shortlisting, there is a need to also evaluate the priority features in relation to other common chronic conditions (ie, diabetes and asthma). We believe there is value in the inclusion of the 5 minimum features for app evaluation, as it allows for specific components to be embedded within wider priority categories and provides a baseline mode to manage the use of multiple modified scales. However, before any scales can be managed or reviewed, future studies need to confirm the reliability of the 5 features for the management of priority app categories.

In summary, our study was the first to specifically evaluate HF self-care apps according to the criteria essential to promote HF self-care for older adults. We found that there was a lack of usable apps to promote HF self-care for older adults, and this is mainly because of the lack of a patient-centered design. With a rise in the aging population, identifying features pivotal for patient self-care will be crucial to increase their user experience and ensure the longevity of the app’s use.