Keywords

1 Introduction

In the United States and across the world, the number of older persons compared to younger has reached a level never before seen in human history. The United Nations has described the current trend in global population aging as unprecedented, pervasive, and enduring, and has stated it will have profound implications for many facets of human life (UN, 2001). In the United States alone the population of adults aged 65 and older is expected to nearly double from 37 million in 2005 to 72 million in 2030 (He, Sengupta, Velkoff, & DeBarros, 2005). With the benefit of extended lifespan comes a longer amount of time when people may experience age-related health conditions. Ideally, the extended years the current and future generations are expected to experience will be vibrant and marked by extended independence. However, to achieve this goal society and technology will need to be adapted to the higher percentage of older adults in the population, particularly technology that focuses on the management and amelioration of age-related diseases and their effects on physical and psychological health.

Chief among age-related diseases is type 2 diabetes, estimated to affect ~ 27 % of older adults in the United States in diagnosed or undiagnosed form (Cowie et al. 2009). Furthermore, it is also estimated that an additional 40 % of older adults have prediabetes (Cowie et al. 2009), a state of elevated blood glucose that does not yet qualify as diabetes but indicates an increased risk of developing diabetes as well as an increased risk of heart disease and stroke. Due in part to diet and in part to the larger percentage of older adults, type 2 diabetes is increasing worldwide, noted by the World Health Organization as the “slow-motion catastrophe” of a rising trend in chronic, noncommunicable disease (Chan 2011). It is currently the seventh leading cause of death in the United States (CDC 2011). With two-thirds of the older adult population of the United States estimated to either be at risk of diabetes or to already have the disease, it is one of the greatest health concerns facing older adults today and in the foreseeable future.

Type 2 diabetes has serious health consequences if not managed and treated, as the pancreas fails to regulate insulin in response to dietary sugars. Over time, elevated blood glucose raises the risk of heart and kidney diseases, stroke, blindness, limb dysfunction, amputation, and chronic pain as the result of nerve damage. A four-community epidemiological study found that older adults with diabetes experienced substantial comorbidity with report visual problems, major physical disability, and hospitalizations during the past year (Moritz et al. 1994). Low blood sugar levels bring their own dangers as well, including high rates of Alzheimer’s disease and dementia (Yaffe et al. 2013). Despite these dangers and side effects from the disease, there is hope for those suffering from type 2 diabetes. Studies of self-management find diabetes symptoms can be controlled and the onset of the disease delayed in high-diabetes risk persons who keep tight control of their blood sugar levels through diet and exercise (DPPRG, 2002; Lindström et al. 2003).

1.1 Challenges of Self-Management

However, persons with type 2 diabetes do not tend to control their blood sugar levels well, including even the most health-literate older adults (Boren 2009; Klein & Meininger 2004; Shigaki et al. 2010). A meta-analysis of ~ 10,000 patients found that even after a self-management training program, improvements in blood sugar levels were small - indicating that type 2 patients need more than education (Klein et al. 2013).

Problems with effective self-management appear to be both cognitive and motivational in nature. Type 2 diabetes is notable because of the necessity for patients to manage their own condition by actively engaging in self-management behaviors (Skinner et al. 2006). These behaviors include self-monitoring of blood glucose via at-home glucometers, medication adherence (including for some, administering insulin shots as needed), adherence to a diet appropriate for the individual’s level of insulin resistance and medication use, physical activity, and visiting healthcare practitioners for diabetes-related health checks. Some of these self-management behaviors are cognitively complex and involve processes like problem detection and identification, sensemaking, decision making, and planning/replanning (Klein & Lippa 2008). Because of these cognitive requirements, successful self-management is not an easy rule-based procedure. It necessitates “the fitting of complex and sometimes contradictory information into a coherent picture that generates a reasonable action strategy” (Klein & Lippa 2012), a dynamic control task that patients are often unable to understand and perform (Klein & Lippa 2008).

Persons with diabetes who experience trouble with this dynamic control task will likely experience even greater trouble in the future due to the apparent causal link between poor diabetes control and cognitive decline. Poor diabetes control is associated with both diagnosed and undiagnosed cognitive dysfunction (Munshi et al. 2006), and current findings suggest that poor diabetes control precedes cognitive decline in older adults. A four-year prospective study of cognitive change in older adults found that women with impaired glucose tolerance at baseline had four times the risk of major cognitive decline on a verbal fluency test after the four year period when compared to women with normal glucose tolerance at baseline (Kanaya, Barrett-Connor, Gildengorin, & Yaffe, 2004). A meta-analysis of prospective studies of cognitive decline in persons with diabetes found a 1.6-fold increase in the odds of future dementia (Cukierman, Gerstein, & Williamson, 2005). These findings suggest that poorly managed diabetes can rob patients of the abilities most needed to manage their disease well.

Furthermore, because the patient must choose to engage in these behaviors on his or her own, the emotional and motivational factors affecting self-management behavior are especially important when considering how to improve diabetes management. Major depression in persons with diabetes is associated with worse self-management, including less physical activity, poorer diet, and lower rates of medication adherence (Lin et al. 2004). This is particularly concerning since diabetes is in itself associated with a higher risk of depression than that of the nondiabetic population (Anderson, Freedland, Clouse, & Lustman, 2001; Peyrot & Rubin 1997). The stress of managing diabetes may directly contribute to depression; more frequent blood glucose self-monitoring can be associated with negative psychological well-being outcomes including higher levels of distress, worries, and depressive symptoms (Franciosi 2001). There is also some evidence that improving individuals’ capacity to cope with stress can improve diabetes management. A randomized controlled trial found that teaching acceptance coping strategies to adults with diabetes improved both self-reported self-management behaviors and HbA1c values, an indicator of glycemic control over time (Gregg, Callaghan, Hayes, & Glenn-Lawson, 2007).

2 Supporting Self-Management via Technology

The development and use of smartphone applications (“apps”) to assist with diabetes self-management is a rapidly growing area, driven in part by increases in smartphone adoption. Smartphone ownership has consistently increased since their introduction and as of 2011, over 85 % of Americans owned one (Tran, Tran & White, 2012). As adoption increases, smartphones have become more affordable and utilized across the socio-economic spectrum (Liang, et al., 2010; Nundy, Dick, Solomon & Peeka, 2013). Related directly to diabetes self-management, smartphones allow users to track and manage their diabetes in a variety of environments. Indeed, users noted they would rather use a mobile phone than be “tied down” to a home computer (Harris, et al., 2010). Such immediacy of tracking encourages consistent and frequent measurements for self-management (Baron, McBain, & Newman, 2012; Lyles, et al., 2011; Nundy, et al., 2013). An additional benefit comes from the displays on these phones, as they allow graphical and tabular displays of information, which can aid understanding of the complex and dynamic system that is diabetes management (Årsand & Tatara, 2010; Harris, et al., 2010). When compared to paper logbooks, a traditional method of self-management, smartphone users made fewer errors, were more likely to log information, reported ease of noticing trends (Harris, et al., 2010), and found smartphones more motivational (Rao, Hou, Golnik, Flaherty & Vu, 2010). Smartphone benefits are not exclusive to diabetes management with the FDA predicting that by 2015, 500 million people will be using mobile health applications (El-Gayar, Timsina, Nawar & Eid, 2013).

Studies have shown smartphone apps improve diet and exercise (Tran, Tran & White, 2012), increase the frequency of blood glucose logging (Lyles, et al., 2011), and provide much needed feedback on a more frequent basis from healthcare providers (Harris, et al., 2010). When feedback was provided automatically via smartphones, users reported elevated motivation to self-manage (Nundy, Dick, Solomon & Peeka, 2013). The connection with health care providers works both ways, as there are fewer needed appointments when communication of levels and management comes through an app (Lyles, et al., 2011). However, despite the promise of smartphone apps to improve diabetes self-management, a number of challenges remain regarding their use. This is particularly true when considering older adult users, and the design of existing apps often fails to take older users’ capabilities and limitations into account (Whitlock & McLaughlin, 2012).

2.1 General Design Recommendations

It is possible to describe the desirable features of a technology to support the tracking and management of blood glucose levels from a review of the literature. First, the technology should be mobile and usable in a variety of environments, particularly environments linked to eating and exercise. Designs should consider age-related changes in perception, cognition, and movement control and be tested with a representative sample of the target population. Second, the technology should provide accurate and up-to-date information regarding available carbohydrates and their potential effects, medications, and blood sugar historical trends. Third, the technology should encourage good self-management behaviors beyond simply providing information to the user, reducing stress related to the disease when possible. Fourth, the technology should provide or be a conduit of emotional support during self-management decision-making. Last, the technology should scaffold the user during decision-making in a complex and dynamic system. These features can be broken down into specific recommendations for blood glucose tracking applications on smartphones (“apps”).

In terms of content, the app must track blood glucose levels, nutrition, medication use, and physical activity (Baron, McBain, & Newman, 2012; El-Gayar, Timsina, Nawar & Eid, 2013; Liang, et al., 2010). Optimally, blood glucose levels would be tracked in immediate form and historical data/trends would also be available. Nutritional information should include tracking of foods eaten with amounts and a count of the carbohydrates (Årsand & Tatara, 2010). Tracking of physical activity should be detailed enough to assist with weight management but also provides inputs useful to informing patterns of blood glucose levels. Tracked data must be displayed in uncluttered, high-contrast graphical visualizations that allow the user to discern relationships and patterns between tracked variables and resulting blood glucose levels.

In terms of education, adaptive training should be offered through the app. This could be as simple as automated messages from health care providers (Harris, et al., 2010) to connections with other users best able to answer specific questions in a timely manner. This mimics the benefits that have been found for the use of online forums by persons with chronic health conditions (Eysenbach et al. 2004).

It is also desirable that apps provide a support structure, both for emotional support and decision support. One study found that among older adults with diabetes, those who reported greater social support were likely to have fewer impairments on the ADLs and IADLs, to have better self-rated health, shorter duration of diabetes, were less likely to feel depressed and to have trouble with stress, and less likely to have had a heart attack (Zhang, Norris, Gregg, & Beckles, 2007). The support structure should include a reminder system that prompts the user for readings and tracking information (Baron, McBain, & Newman, 2012; El-Gayar, Timsina, Nawar & Eid, 2013) but with the understanding that reminders alone do not guarantee adherence (Brath, et al., 2012). Support should extend beyond the app and social networks of the user and also allow communication with health-care providers that are part of the team helping the user to manage type 2 diabetes (Baron, McBain, & Newman, 2012; El-Gayar, Timsina, Nawar & Eid, 2013; Liang, et al., 2010).

The app should motivate the user in the face of an ambiguous, complex, dynamic task that offers few if any instances of immediate feedback. Elevated blood sugar levels tend to be severe before symptoms are noticed by patients, meaning that an older adult can be off optimal levels for long periods of time before acute symptoms act as feedback. Time spent off optimal levels permanently damages the body. Thus, apps can step in to provide the feedback and rewards not offered by the condition itself.

The last recommendations for app design center on usability. The app should be adaptable to the changing health, knowledge, and performance of the user (Årsand et al. 2012). To improve the mobility of the app, it should sync across devices as to be readily available and accessible even when a smartphone is not. Usability should be tested with older adults with type 2 diabetes throughout app development. Adherence to older adult design guidelines assists in the initial designs and directions (Pak & McLaughlin, 2010; Rogers, Fisk, Charness, Czaja, & Sharit, 2009) but there is no substitution for testing representative and long-term tasks with the target population. Pak and McLaughlin offer guidelines for usability testing with older adults (Pak & McLaughlin 2010).

2.2 Analysis of Existing Apps

We examined three apps selected on the basis of published usability ratings identifying them as among the most usable of diabetes self-management apps available on the Android market (Demidowich et al. 2012). The three apps were Glucool Diabetes, OnTrack Diabetes, and Dbees.com. Glucool Diabetes is available as both a free and a premium ($4.99) version, with 126 reviews and an average user rating of 3.6 stars. OnTrack Diabetes is available for free, with 5,455 ratings and an average user rating of 4.4 stars. Dbees.com is available for free, with 415 ratings and an average user rating of 4.0 stars.

A commonality shared by all of these apps was that each tended to offer major features that users expect in current diabetes tracking apps, for example tracking multiple variables (blood glucose, food, physical activity, etc.) and exporting data to a spreadsheet. However, features were not always implemented in accordance with usability principles. Graphs were sometimes cluttered (Fig. 1, right) and often did not contain appropriate labels (Fig. 1, right; Fig. 3, right). For one app the website implementation did not load correctly when viewed on the mobile device used for testing (Fig. 2, right).

Fig. 1.
figure 1

Glucool Diabetes app, with tracking input screen (left) and historical graphical summary of blood glucose (right).

Full size image
Fig. 2.
figure 2

Dbees.com app, with daily summary screen (left) and failed attempt to view graphical data on the integrated website on a Samsung Galaxy S4 mobile device (right).

Full size image
Fig. 3.
figure 3

OnTrack Diabetes app, with daily summary screen (left) and historical graphical summary of blood glucose (right).

Full size image

No app was designed using guidelines for the needs of older users, and older users who experience physical or cognitive side-effects of diabetes are likely to find their use especially problematic, e.g. reading tiny labels on graphs. Some apps provided both tracking and up-to-date information on dietary choices, but none included information analysis support to help users interpret causes behind trends in the data. No apps provided the social support or communication with healthcare providers beyond data transfer that is recommended in the literature. No apps provided emotional or decision support to the user and showed no adaptation. No apps provided education during their use.

It is important to note that despite their occasional usability challenges, these apps are highly useful to many users, as evidenced by their ratings and number of downloads. Many of them are the efforts of programmers highly invested in the management of diabetes, who have dedicated a great deal of time to the features of these apps. We are grateful for these efforts, as they are a first step toward diabetes management applications created using a human factors and participatory design process.

2.3 Future Directions

To advance the current state of mobile apps for diabetes self-management we advise the development of features that lower the cognitive and time costs of making better-informed decisions about self-management. Consider the scenario of a user with diabetes going out to a restaurant for breakfast. Current apps require the user to spontaneously remember to use the app to look up nutrition information before ordering, placing an unnecessary burden on the user’s prospective memory. This may be particularly problematic for older users given the tendency for prospective memory to decline with age (Huppert, Johnson, & Nickson, 2001).

To address this problem, app developers could take advantage of smart phones’ GPS function to detect when users enter a restaurant and provide just-in-time intervention Once the app detects users have entered a restaurant it could prompt them with a push notification to view a restaurant-specific menu database (Fig. 4, left). Because GPS detection is not guaranteed to be accurate, users should be able to correct the app when its location information is wrong, and have it remember the correction for future visits (Fig. 4, right).

Fig. 4.
figure 4

Prototypes of push notification feature (left) and screen allowing user to choose to view the menu or correct the GPS location (right).

Full size image

App developers could also utilize crowd-sourced diabetes knowledge by allowing users to rate restaurant menu items in terms of appropriateness for a healthy diabetes diet (Fig. 5, left). Users can consult the ratings to quickly determine the best choices on the menu. After selecting a menu item, users are given the nutrition information with carbohydrate content highlighted, and can add it to their food log with a single touch (Fig. 5, right).

Fig. 5.
figure 5

Prototypes of crowd-sourced menu nutrition ratings (left) and menu item nutrition information with carbohydrates highlighted (right).

Full size image

Although accuracy of crowd-sourced information is a potential concern, a previous study of a diabetes internet forum found that recommendations given by users on the forum were in agreement with best practice clinical guidelines 91 % of the time (Hoffman-Goetz et al. 2009). This suggests that user-sourced information in the context of diabetes may reach high accuracy. Crowd-sourced menu ratings could also be independently assessed for accuracy by professionals, e.g. nutritionists, or by “super users” singled out for their high level of diabetes knowledge.

Features like these that reduce the cognitive burden of making informed self-management decisions are likely to be particularly helpful for users experiencing age- and diabetes-related cognitive decline. We furthermore recommend that developers utilize the lessons learned from building apps that track and display food, exercise, and blood glucose levels to make apps that also:

  • Are created using design guidelines for older adult usability

  • Are tested with older users

  • Allow two-way communication between users and healthcare providers

  • Contain motivational features such as gamified rewards and social support

  • Contain educational features, such as tutorials and online social networks of other patients and healthcare professionals

  • Provide just-in-time decision support when diet, exercise, or medication decisions must occur