Discussion
In this study, a sample of diabetes-related apps was coded for public health characteristics and health literate design strategies for mHealth apps. These apps were rated highly by users, and most were classified as appropriate for children and adults. Consistent with other studies of health apps and diabetes apps, most of the diabetes apps in this sample addressed diabetes management and therapeutics. Paid apps were more likely than free apps to use health literate design strategies such as using plain language, labeling links clearly, and having a "back" button to help with organization.
One explanation for these differences is that with paid apps, perhaps more effort was undertaken to conduct formative research and usability testing before product launch. Those activities may have identified functions in the app for which the user experience could be improved to increase understanding and ease of use.
Because low health literacy is more likely among people of low socioeconomic status, the cost of apps may be prohibitive for people with low health literacy. If these people are more likely to use free diabetes apps, then they are more likely to have apps that lack features that enhance usability and understanding. Further research can identify diabetes app characteristics, including functionality, cost, and ratings, that may influence potential users to pay for an app instead of downloading one for free. Because user ratings of free apps and paid apps did not differ significantly, it would also be helpful to conduct usability tests to directly compare levels of satisfaction for free and paid apps. Also, by understanding which types of diabetes apps people with low literacy would choose to use regardless of cost, we could identify where and how resources for improving the health literacy of mobile diabetes apps would be best used.
Our study has several limitations. Although we used Health Literacy Online as a tool to rate existing apps, its original purpose was to help guide the design of health websites, including strategies for testing usability. Not all Health Literacy Online strategies and actions, such as those requiring knowledge of the app developers' target users and usability processes, were included in the codebook for our study because of a lack of information. Because the information coded in our study could be ascertained only by viewing the app description and ratings information and using the downloaded app, we did not have enough background information to know the history of each app's development or what, if any, usability testing was done before its launch. Health Literacy Online is one of several tools that can help guide the creation of health literate mHealth applications or to assess the health literacy of health information materials, including those that are digitally based (eg, mobile apps, websites, computer applications). Using this tool for existing apps may not be as appropriate or useful for usability outcomes, but it is a starting point to help evaluate the health literacy of diabetes apps.
Another limitation of this study is the generalizability of the sample of diabetes apps selected. The use of a simple random sampling strategy yielded 4 of the top 10 most popular diabetes apps in the App Store (as of February 2015). In addition, the search terms included only diabetes-related terms, because we were interested only in apps that self-identified as diabetes-related through the app name or description. Other search terms such as "glucose" or "blood sugar" were not used. However, an additional search using the terms "glucose" or "blood sugar" yielded 294 apps, 176 of which also appeared on the list of diabetes-related apps.
Finally, the sample of apps examined included only iOS apps and did not include any Android-compatible apps. Because African American cellular telephone owners are more likely than whites or Latinos to own an Android telephone instead of an iPhone (42%, African Americans; 26%, whites; 27%, Latinos), this may limit the study's applicability to more diverse audiences.
In general, the findings of this study indicate that additional work should be done to improve mHealth apps. In particular, encouraging a development process for free diabetes-related apps to make them more user-friendly and accessible to diverse audiences could potentially increase their use and understandability among audiences, especially people with low health literacy.