To investigate relevant acceptance factors, we chose the qualitative method of guided interviews, which added a qualitative dimension to the two prior quantitative substudies. Our aim was to understand the subjective perspective of older diabetes patients toward apps. We consciously opted for a personal approach with the study participants using personal interviews instead of questionnaires. This approach was used since a certain degree of insecurity toward the research topic was to be expected among the participants, due to a possible lack of previous experience. This approach enabled us to adapt the interview guideline according to the individual survey participants and their previous experiences in handling smartphones, tablets, and apps. Furthermore, the assessment of factors promoting or inhibiting the acceptance of diabetes apps by the elderly is a complex issue, and a qualitative approach was better suited to examine this topic. This study was approved by the Ethics Committee of the Technische Universität (TU) Dresden (reference number: EK 241072013), prior to the initiation of interviews. The interviews were conducted between July and December of 2013.

We developed a theory-based and uniform interview guideline with open-ended questions as a basis for the study (see Multimedia Appendix 2). It provided the foundation for the comparability of the answers. At the end of each interview, we asked the patients to test and evaluate two existing diabetes apps to reveal difficulties in (first) use:

The tested apps needed to satisfy the following criteria: (1) have German content, (2) be among the top 10 most commonly installed diabetes apps in the respective app store at the time of their selection (July 2013), and (3) be multifunctional (ie, able to combine several functions within one app). The following functions were offered by both apps: documentation function, analysis function, reminder function, and data forwarding/communication function. These functions are described in more detail in Arnhold et al [1]. Figure 1 shows screenshots of the start screens of both apps in order to illustrate their range of functions. In the run-up to the test, the participants did consciously not receive any form of introduction to the apps or the devices on which they were presented. The order in which the apps were presented was randomized to prevent an impact by the presentation order. Both apps were tested on tablets to increase the user-friendliness for the target group [48]. A pretest of the interview guidelines was performed prior to the commencement of the field work. Based on the results of the pretest, the guideline was slightly revised.

We decided to include both type 1 and type 2 diabetics in the study, as currently available apps address both types equally and rather differ in terms of the range of functions. Additionally, we included both participants with and without prior experience with mobile devices and apps to include the perspectives of both groups. The test subjects were recruited from diabetics’ self-help groups, diabetics’ associations, specialty shops for diabetics, general medical practices, diabetologists’ practices, and pharmacies. The following inclusion criteria were defined: (1) aged 50 years or older, (2) diagnosed with type 1 or type 2 diabetes mellitus (via patients’ self-disclosure), and (3) sufficient cognitive abilities to participate in a 60-minute interview. Persons who did not meet the inclusion criteria were excluded from this study.

All interviews were transcribed verbatim. The transcripts were the foundation upon which consecutive data analysis was performed. To analyze the data, we used the structured content analysis by Mayring [49], which allows for an association between the deductive and inductive creation of categories [49,50]. The analytical focus was on designing a system of categories and subcategories, as well as their characteristics [49], which in turn served as a structural dimension. We started the analysis in accordance with the structure of the guideline [50]. The interview questions were partially transcribed directly into analysis categories. Based on the content, all relevant text passages were extracted and put into their specific categories [50]. Developing the categorical system and the deductive category application represented the initial framework with which we structured the content and the analysis. During the process of examining and analyzing the individual interviews, we tested, modified, and specified the categorical system. Missing, but relevant, categories were developed and added inductively based on the text material that we collected. This was done up to the point at which the system became saturated (ie, no new categories emerged) [50]. This method complies with the principle of openness within qualitative research.

The chosen data collection method enabled us to quantitatively analyze the data due to its systematic and rule-based approach. The individual steps were processed with the qualitative data analysis (QDA) software MAXQDA (2011 version).

The following steps briefly explain the specifics of the method used for analyzing the gathered data:

Overall, we interviewed 32 people with diabetes aged 50 or older. The recommended sample size for guided interviews is set at 30 people [51]. The interviews lasted between 15 and 90 minutes. The following represents a selection of the survey results. An overview of all results can be found in Multimedia Appendix 3.

There were an equal number of female (16/32, 50%) and male (16/32, 50%) participants in the interviews. The mean age of participants was 68.8 years (SD 8.2). Of all the participants, 44% (14/32) successfully completed vocational education, 13% (4/32) held a degree from a technical college, and 34% (11/32) had a university degree. Within their last or current job positions, 81% (26/32) had been working as employees, 13% (4/32) had been self-employed, and 6% (2/32) were employed as skilled workers. The majority of participants (17/32, 53%) were between 65 and 74 years of age. Of all the participants, 78% (25/32) have had diabetes mellitus for more than 10 years. Of the participants, 66% (21/32) had type 2 diabetes, 31% (10/32) had type 1 diabetes, and 1 patient out of 32 (3%) was afflicted with a hybrid form of both type 1 and type 2 diabetes.

Of all the participants, 34% (11/32) described themselves as highly interested in new technologies. Of the participants, 53% (17/32) were open-minded in terms of technology, provided that it entails an additional benefit to their treatment, such as having a positive effect on their therapy/blood glucose values or improved convenience of self-management. Only 13% (4/32) of the interviewed diabetics described themselves as having no interest in innovative technologies, whatsoever.

At the time of the survey, 25% (8/32) of the interviewed diabetics older than 50 years of age owned a smartphone or tablet and 47% (15/32) knew apps. However, only 2 diabetics out of 32 (6%) already used a diabetes app for the purpose of documentation of blood glucose values, namely the OnTrack Diabetes app for Android (Medivo) and the DiabetesPlus für Typ 2-Diabetiker app for Android and iOS (SquareMed Software GmbH).

Within this study, the essential aspects influencing the acceptance of diabetes apps were the self-reported reasons for or against the utilization of mobile devices and apps, as well as the obstacles that emerged during the actual app test. The following sections will describe the most influential factors in more detail.

We decided to consider the reasons against using a smartphone, tablet, and apps in one section, as the use of a portable device is the essential prerequisite to assessing apps. Additionally, the principles in handling these devices are similar. The main obstacles during the app test were a lack of additional benefits—for smartphones/tablets (10/18, 56%) and apps (4/8, 50%)—as well as finding the initial training and the handling phase for smartphones and tablets to be too complicated (6/18, 33%). One participant quoted the following:

Out of 18 participants, 4 (22%) of them stated that the financial cost-to-benefit ratio of smartphones/tablets was unacceptable, especially when the device would only be used for diabetes treatment. Of 18 participants, 5 (28%) had concerns regarding the protection of their private data, as illustrated in the following quote:

Additionally, the interviewed diabetics were concerned with a lack of interoperability between different devices and apps both for smartphones/tablets (2/18, 11%) and apps (1/8, 13%).

In addition to the self-reported obstacles, an app test was conducted to determine obstacles during the actual practical use of mobile devices and apps, which in some cases was the very first contact with such technology. On average, the participants tested each of the two apps for 11 minutes (22 minutes in total). Participants without previous experience spent more time on testing (26 minutes on average) compared to those who already used a smartphone or tablet (19 minutes on average). Figure 2 presents the most common obstacles. They were either observed by the interviewer or self-reported by the interviewees. Of the 32 participants, 3 (9%) were not able to participate in the app test section, as the interviews were conducted via telephone.

Out of the remaining 29 participants, 26 (90%) felt that neither functionality nor usability of the apps were intuitive or easy to grasp. This was especially true for those with no prior experience, who required additional aid (eg, how to enter data into the app). The following is an excerpt from transcript of the interview between interviewer and participant:

Help was requested as to how to make the keyboard appear only after touching sections that ask for data to be entered, which wasn´t the case with previous generations of mobile devices. This is illustrated in the following interview excerpt:

An additional difficulty was in understanding the symbols and functions of the keyboard, which were not intuitive and complicated the process of entering data even more. For the sake of completeness, it is important to mention that the difficulties encountered while using an onscreen keyboard cannot be counted as app-specific issues, as this may be the result of a lack of experience with smartphones and tablets, in general.

Of the 29 participants, 19 (66%) felt insecure and uncertain in terms of navigating through the menu within the apps, especially when switching between different layers of the menu. This is illustrated in the following interview excerpt:

The participants faced issues such as adapting to different devices, operating systems (Android vs iOS), or apps and their individual layouts. This was particularly the case when attempting to save the progress on different devices, and this led to insecurity for 34% (10/29) of participants. Occasionally, it resulted in data loss. This issue is illustrated in the following interview excerpt:

Of 29 participants, 14 (48%) had difficulty in identifying and pressing touch-sensitive areas on the touchscreen. Another obstacle presented itself when users had to switch keyboard layouts, between numbers and letters, to enter data into the app. This also led to insecurity for 59% (17/29) of the participants. Criticism was also voiced about the small size of font, the space between letters and representations (14/29, 48%), the insufficient color contrast (8/29, 28%), and the inability to flexibly adapt the size of the font and representations to individual needs (4/29, 14%). Moreover, some of the participants felt that the options and possibilities of the presented apps did not fit their personal needs in terms of their diabetes treatment. They expressed that important functions were missing (11/29, 38%), for example, the possibility to manage polypharmacy or comorbidity. Some also had the impression that existing features were irrelevant (10/29, 34%). A medication-related issue is illustrated in the following interview excerpt:

Nearly half of all participants (14/29, 48%) repeatedly stated that neither of the two tested apps offered an additional benefit to their regular diabetes treatment. Of the participants, 34% (10/29) stated that it would be too difficult and time-consuming to obtain the skills required to work with this technology.

During the course of testing, the participants also reported positive impressions. Of 10 participants, 5 (50%) were of the opinion that using diabetes apps would have a positive effect on their self-reflection and the monitoring of their therapy. Of 10 participants, 2 (20%) were impressed by the clear arrangement and presentation of the entered data, as well as the simplicity and speed of the documentation. Both patient groups, those with and those without prior experience in using smartphones and tablets, reported positive impressions.

In the course of this survey, we asked the participants what features they would expect to find in a helpful diabetes app (ie, which features would entail individual additional value). Of 28 participants, 9 (32%) expressed their need to be able to add personal remarks to the measured blood glucose values (see Figure 3). They said this would be useful should it be necessary to recreate the situation under which data was aggregated (ie, extremely low or high blood glucose values). A quarter of the participants (7/28, 25%) said they would appreciate a reminder feature for medication or blood glucose measurement. The same number of participants (7/28, 25%) felt that it would be useful to be able to define individual thresholds for blood glucose values and to be able to highlight deviating values. This is illustrated in the following quote:

Of 28 participants, 6 (21%) would like to find a kind of reference book in which to look up information about diabetes mellitus, its treatment, and medication. Furthermore, they would appreciate finding information regarding nutritional facts for meals, which one would consume in a restaurant, or meals that have not been prepared by oneself (4/28, 14%). In order to support the documentation of values, participants would prefer to find all information (ie, blood glucose values, medication, and annotations) on a single page or table (6/28, 21%). This reveals their perception and preference of already existing and utilized blood glucose diaries. This is illustrated in the following quote:

Altogether, the tested apps did not provide any of the features the patients would expect to find in a helpful diabetes app, except for the reminder feature.

In addition to distinctive features, characteristics of the design of an app are crucial for their acceptance and usage among diabetics 50 years and older. The special usability requirements provided by elderly users of mobile apps have been presented and evaluated by Arnhold et al [1].

For questions related to using modern technology—in general, or specifically for diabetes treatment—family members were the primary choice for 61% (19/31) of the participants, particularly their children, grandchildren, and partners. The second most frequently asked group of people were friends or people from their peer group (11/31, 35%). The least favorite source for information was the Internet or online forums used to provide solutions for technical difficulties (2/31, 6%).

When interviewing study participants, there is always the possibility that their answers are influenced by social desirability, which in turn could lead to biased results. To tackle this issue, we opted for an open interview setting, gave participants the chance to ask questions, and kept the number of people present during the interview to a minimum.

The choice of guided interviews as the method for our data collection was done with regard to an open interview setting and the chance to ask further targeted questions, in case particularly interesting or relevant topics arose. However, we restricted the evaluation of the gathered data to the unambiguous statements provided by participants during the interview. Naturally, we could not have drawn any conclusions on aspects and needs that might be relevant to the participants, but were not presented to us.

At the time of preparing this article, there were no other studies investigating the acceptance of diabetes apps by patients aged 50 or older. Hence, this study makes an essential contribution toward a better understanding of the promoting and inhibiting factors that influence the acceptance and usage of diabetes apps. We consciously decided to use a qualitative research method in order to have an open approach toward this field of research, and to put focus on the relevant subjective aspects of the participants. In combination with the work by Arnhold et al [1], it is now possible to conceptualize diabetes apps that are tailored to the needs, skills, and usability requirements of the target group, diabetics aged 50 or older. In addition, the results can be used as a starting point for quantitative studies in this field with a larger sample size.

This study was the first to examine the factors that have an impact on the acceptance of mobile diabetes apps by patients aged 50 or older. The key factors that emerged for acceptance were the perceived additional benefit and the perceived ease of use. Less influential factors were previous experiences/knowledge, current health status, available support, trust in own technical abilities, perceived data security, expected reliability/fault tolerance, and joy of use. Furthermore, we showed that the needs of the investigated target group are highly heterogeneous due to differences in previous knowledge, age, type of diabetes, and therapy. Therefore, the contents of a helpful diabetes app should be individually adaptable. Personal contact persons, especially during the initial phase of use, are of utmost importance to reduce the fear of data loss or erroneous data input, and to raise acceptance among this target group.