Keywords

1 Introduction

The ongoing demographic change puts healthcare systems under pressure. The aging society causes increasing incident rates of chronic, degenerative, and age-related diseases. At the same time, a growing shortage of physicians leads to the insufficient availability of medical care possibly making it next to impossible to provide everyone with adequate medical care in the future. Especially those suffering from chronic diseases requiring special treatment will be affected [10, 12]. However, while the number of diseased people is increasing, a digital change is transforming the health care sector. One very promising improvement supporting the health care system by using technical information and communication technologies (ICT) are mobile health applications (short: mHealth apps) [1]. Because mHealth apps are easily accessible by Smartphone – of which almost everyone owns one these days – they can support and improve a patient’s care with features like medical reminders, the monitoring of vital parameters, or emergency call functions. Hence, the patients’ quality of life and care in the aftercare of surgery can be improved whilst relieving healthcare systems [20, 39].

However, the introduction of new digital technologies also introduces concerns, especially regarding privacy since personal and medical information is collected, analyzed, and stored [41]. Additionally, particularly older adults may seek assistance in the interaction with digital technologies. Therefore, it is of great importance to understand the individual motives and barriers associated with a new technical application in medical aftercare treatment. Factors that enhance the willingness to use and increase the acceptance of mHealth apps in aftercare need to be studied. This allows the healthcare system and all stakeholders who are involved in the successful roll-out of mHealth apps in aftercare to consider apps especially tailored for target.

With our study we contribute to the empirical state of research and offer practical recommendations for mHealth apps in aftercare by trying to understand privacy acceptance and the factors that influence the willingness to use digital assistance for aftercare.

1.1 mHealth for Aftercare

mHealth apps increasingly attract the interest from the healthcare system and potential users. The great potential that this technical achievement carries is multifaceted. Since the majority of people nowadays owns or has access to a Smartphone or other mobile device, access to apps – like mHealth apps – is easily obtained. The great advantages of such a technology are its constant availability from any place at any time, its easily transportable size, its equipment with sensors that allow features like, e.g., the measuring of vital parameters, and its possibility to address all kinds of user groups [11]. This enables the private use of mHealth apps for fitness or wellness purposes, but also its medical use. Through the latter, the state of health can be monitored remotely allowing early detection of diseases, supporting the health status, and prolonging life expectancy [12]. mHealth apps could thus contribute to relieving the overburdened healthcare systems [20]. mHealth apps are therefore a welcome technical achievement that can contribute to revolutionizing the healthcare system [21, 27] by enabling users to carry out part of their healthcare without the help of medical professionals. Enabling patients to take care of themselves more easily, reduces costs and allows the more targeted use of resources [11, 16]. Especially people living in rural regions with a shortage of doctors and poor mobility benefit from the resulting remote care.

However, the use of mHealth apps can only contribute to the health care system if they are of high quality and if both physicians and patients show a high level of willingness to use, and acceptance of, the technology. Since ill people and those in need of rehabilitation show a great desire to recover at home, it is of great importance to improve the scientific state of research on this topic and formulate practice-relevant recommendations for the possible roll-out hereof. All user groups should thereby be taken into account. Even though studies on the acceptance of mHealth apps, like remote monitoring in various fields of practice – e.g., breast and orthopedic surgery [6, 32] – already exist, a lot is still left uncovered. Especially the diseases that cost the most lives on a global scale – e.g., types of cardio-vascular disease – need to be studied to improve their treatment and decrease their death rate. It is therefore necessary to understand the motives and barriers which trigger people’s willingness to use and finally accept such a technical application.

1.2 Influences on the User Acceptance of mHealth for Aftercare

For the successful roll-out and sustainable use of a technology, its acceptance by potential users is required. From a temporal point of view, the adoption process of a technology starts with the knowledge and awareness by the potential users [28]. After a persuasion phase, a decision to adopt or reject is made. The here described Diffusion of Innovation Theory by [28] shows that different factors influence this process. Especially in the persuasion phase, the perceived characteristics of the innovation are important.

Several technology acceptance models summarize the important factors that influence the behavioral intention to use a technology from a user-centered perspective (e.g., the Technology Acceptance Model (TAM, [9]), the Unified Theory of Acceptance and Use of Technology (UTAUT [34]) and their extensions). With these models, the decision to adopt or reject a technology can hence be explained based on, e.g., the perceived characteristics of the technology (e.g., usefulness, ease of use), the situation (e.g., voluntariness of use, facilitating conditions), and user diversity (e.g., experience, age).

However, research has shown that these models do not properly fit in all contexts of technology use, especially regarding the medical context [15]. For one thing, factors that are important in this specific context are missing from the models. For example, privacy concerns are not considered despite the fact that they are an important barrier for the adoption of digital technologies, both in general [4] and in the medical and mHealth context [13, 19, 26]. For digital health technologies, it was shown that the most important privacy aspect for potential users was who has access to the data [24, 31]. As an important barrier, privacy requirements should thus be considered for mHealth apps for aftercare. Moreover, acceptance and privacy depend on the context and type of the technology [14, 23]. The features an mHealth app provides might therefore influence its acceptance and how this is affected by other factors.

Additionally, user diversity needs to be considered [36]. Factors like age, gender, and technical affinity can influence the perception and acceptance of technologies [35, 40]. Particularly, older adults – a main target group for mHealth apps for aftercare – represent a special user group regarding the acceptance and use of digital technologies. The perception of technologies is not only influenced by people’s upbringing and experience with technologies during their life [29], age also changes perceptual, cognitive and psychomotor skills. These – potentially in combination with illnesses and chronic conditions – can result in difficulties in interacting with technical devices [7]. Ease of use is a decisive barrier for the adoption of digital health technologies, not only for older adults [26, 30]. Therefore, technical support and assistance in getting to know and learning to interact with digital technologies are important requirements for mHealth apps.

At the same time, several user groups with different evaluations of acceptance and varying decision patterns may exist [5, 25]. For example, [25] found two distinct user groups that differed in their trade-off patterns between the benefits and barriers of Ambient Assisted Living technologies. Whereas privacy was the most important factor for both groups, their preferences regarding the exact wishes for privacy and other factors largely differed, showing very distinct decision patterns. [5] also found different user types in the acceptance of fitness trackers. For the largest user group, their decision for a fitness tracker was mostly based on the privacy design; for another group on the perceived utility; whilst for a third user group the motivational design was most important. These previous studies show that users are diverse regarding their preferences and decision patterns. Similar diversity may exist for the requirements for mHealth apps for aftercare.

1.3 Empirical Approach and Logic of Procedure

This paper addresses the question on which different aspects are relevant to understand the acceptance of mHealth apps in aftercare from the users’ perspective. To receive more robust results on this topic, a multi-method approach was chosen. In Fig. 1 an overview of our two-step study is depicted.

Fig. 1.
figure 1

Overview of the research process showing the qualitative and quantitative measures used to address our research questions.

Full size image

In the first step, qualitative data was collected through focus groups. Two focus groups of two different age groups were run. Very general questions were used to guide the group discussions to collect the intended diverse opinions and point of view of the participants:

  • Which features would you like to have when using an mHealth app in aftercare?

  • Which benefits come to your mind when thinking about using an mHealth app?

  • What kind of barriers can you think of?

With the identified requirements, perceived benefits, and barriers, a quantitative study was developed to evaluate and quantify these factors. The use of a Choice Based Conjoint (CBC) experiment allowed the evaluation of preferences for different concept alternatives. In that way, the preferences among different levels of three attributes – features of an mHealth app, types of data access, and assistance options for introducing the app – were determined. The choice of these three attributes was based on the results of the prestudy.

The research questions derived from our chosen Conjoint study were:

  • Which factor contributes the most to the decision to use an mHealth app in aftercare?

  • Are there differences in the response behavior between (groups of) participants?

The following section is structured according to our study process. First, the procedure and results of the prestudy are presented. Second, the development of the main study will be introduced by explaining the CBC approach in general and in particular related to our study.

2 The Prestudy

Two focus groups were conducted to identify the requirements, perceived benefits, and perceived barriers of potential users for mHealth for aftercare. The advantage of focus groups compared to interviews is the possibility to observe a discussion between participants regarding new topics for which they have been invited as potential users, not as experts. The focus groups were conducted as a part of a master’s thesis at the RWTH University in winter of 2018/2019. The participants attended voluntarily and were not compensated. In the following, sample, procedure, and results are shortly outlined.

2.1 The Sample

The participants were acquired from a local medical rehabilitation center and from the social circles of the authors with the aim to include participants of varying gender, with varying technical experience and skills, as well as varying experiences with aftercare and mHealth apps. In the first focus group, five younger adults between the age of 20 to 29 years participated. This age group is part of the generation of ‘Digital Natives’ [22] which has grown up with digital technologies. The second focus group was conducted with six adults older than 50 years, a generation that is – regarding their use of digital technologies – often called the ‘Silver Surfers’ or ‘Best Agers’ [3, 8]. Four of the six ‘Silver Surfers’, as well as one ‘Digital Native’ suffered from a chronic condition. Five participants (45%) had undergone aftercare of some sort (three as out-patient aftercare, two as in-patient aftercare). Regarding their experiences with mHealth, eight of the eleven (73%) participants had used mHealth apps before, mostly fitness apps. Non of the participants had used an mHealth app for aftercare before.

2.2 The Procedure

First, the participants were informed about data protection and the voluntary nature of the focus groups. After an introduction to mHealth apps in general and for aftercare, the participants discussed their own experiences with, as well as requirements for, such an mHealth app and the pro and contra arguments for the use of it. During the entire focus group, a scenario provided the setting for the discussion. The scenario asked the participants to put themselves in the situation of a patient coming home after a cardio-vascular operation with the option to use an mHealth app free of charge. Cardiovascular illnesses were chosen as they represent the number one cause for death globally [38].

In a practical part, all participants defined their own mHealth app with the features and options they desire. Additionally, each participant evaluated the importance of each characteristic using red (very important), yellow (rather important), and green (rather irrelevant) dots (see Fig. 2). After individually working on their personal preferences for mHealth apps, the participants shared their ideas and requirements with the group and discussed the relevance of the characteristics. The focus groups ended with a short wrap up.

After the focus groups, the participants filled out a short paper-and-pencil questionnaire with questions about demographics, their experiences with mHealth apps and aftercare, as well as their technical affinity. The focus groups were audio-taped and a verbatim transcript was written. A conventional content analysis was conducted to identify important requirements for mHealth apps for aftercare.

Fig. 2.
figure 2

Sketch of the task to define a personal mHealth app for aftercare (left) and the results of one of the participants (right).

Full size image

2.3 The Main Results

The analysis focused on the requirements for mHealth apps for aftercare. As important aspects, the desired features, privacy requirements, and (technical) assistance were identified besides other requirements. Perceived benefits and barriers were also analyzed. On overview of the main topics is presented in Fig. 3.

Fig. 3.
figure 3

The main requirements, benefits, and barriers identified in the prestudy.

Full size image

Perceived Benefits and Barriers: The motivation to use an mHealth app for aftercare stemmed from improved rehabilitation and fast help in emergency situations, a higher quality of life, time efficiency, cost savings in the healthcare system, feelings of medical security, the support of research, and monetary bonuses. The particular motivation to provide personal data to other stakeholders is based on the benefit for research, monetary profit, and improved help and recovery. Barriers against the use revolved around stigmatization, feelings of surveillance and privacy concerns, missing trust, and missing technical self-efficacy.

Desired Features: Users wish for the following features in an mHealth app: remote monitoring (e.g., of blood sugar, blood pressure, sleep, weight), medical reminders (e.g., for physical activity, drinking water, medication), information (e.g., about symptoms, allergies, personal medical data), recommendations and advice (e.g., about diet), as well as alarm services to call trusted persons, physicians, or the ambulance.

Privacy Aspects: Privacy requirements were very important to the participants. These included the control over data flow, particularly the consent to it (via legal contracts e.g., a declaration of consent, zipper clause, or a kind of a living will), as well as data security (e.g., encryption, verification of those who have data access) and the varying sensitivity of data types (participants distinguished sensitive from non-sensitive data in how much it needs protection). It was most important for the participants regarding their privacy to have control over who has access to the data (personal physician, care institutions, heath insurance, science, or trusted person).

Assistance: Additionally, the type of assistance for learning to interact with the mHealth apps was discussed as important requirement. The participants discussed who should be responsible (e.g., rehabilitation center/hospital, developer, physician, health insurance) and what type of assistance should be offered (e.g., personal consultation, online tutorial, courses, in-app support).

Additional requirements regarded the usability and user experience of the app, individualization options, and universal access to the app.

3 Method of the Conjoint Study

The results of the prestudy were used to develop a Conjoint study. The goal of the Conjoint study was to gain an understanding of which requirements are most important for potential users of an mHealth app and to analyze the user diversity of these requirements.

3.1 The Choice Based Conjoint Study

In contrast to traditional questionnaires, Choice Based Conjoint studies allow experimental variation of multi-factorial scenarios. The method was originally developed in market research, but has recently been successfully applied in social science for the modeling of technology acceptance [2]. Participants are asked to choose between concrete scenarios in which the characteristics of selected factors are varied. Each participant is presented with several decision scenarios that are randomly generated from the possible combinations. In this way, the influencing factors are not evaluated individually, as they are in a questionnaire, but in combination with, and in a trade-off between, each other. These decisions between the multi-factorial scenarios correspond to the actual decisions that users make about real technologies. Thus, the results are very meaningful and realistic.

Additional, the method provides the option to cluster participants based on their preferences. With this option, the ‘Latent Class Analysis’, users with similar preferences are identified and grouped in several user groups. Contrary to testing the influence of user diversity factors (e.g., age, gender) or differences between predefined user groups, the data-driven cluster analysis approach of the Latent Class Analysis allows the identification of user groups that differ based on their decision patterns in the Conjoint study. It thus reveals user diversity that may not be influenced by typical factors like age or gender. By combining a traditional questionnaire with a Conjoint study, additional information on, and attitudes of, the participants and the user groups are assessed and can be used to describe the identified user groups.

For CBC studies, a prestudy needs to identify the most important factors since only three to five factors can be included in a CBC. For the operationalization of each factor, three to five factor levels are set. As requirements for mHealth apps for aftercare, three important factor were identified on the basis of the related work and the prestudy: features of the mHealth app, privacy requirements, and technical assistance. In particular, it was important to the participants who has access to the data of the mHealth app (privacy) and what type of assistance for interacting with the technology is offered (assistance). Therefore, these three factors were included into the CBC study and for each factor three levels were determined. These are depicted in Table 1. The operationalization of the levels reflected realistic characteristics of an mHealth app for aftercare and were based on the answers of the participants in the prestudy. For each level, an icon was designed to help the participants to quickly recognize the levels within the Conjoint tasks.

Table 1. Factors and factor levels of the Conjoint experiment.
Full size table

In each choice task, the participants choose their favorite option out of three different scenarios. Each scenario consisted of a randomly chosen level of each factor. Figure 4 depicts an example of a choice task. Every participant completed ten choice tasks. The efficiency of the design of the CBC study was tested and confirmed as being adequate.

Fig. 4.
figure 4

Example of one Conjoint task consisting of three scenarios varying in the three factors: participants choose the preferred option.

Full size image

3.2 The Questionnaire

The Conjoint study was embedded into a questionnaire which assessed additional user data. The questionnaire started with an introduction to the topic, the voluntariness of answering the questions, and the consent to data provision. The first questions regarded demographics (age, gender, education), followed by health status and experience with aftercare. Participants were also asked about their use of popular digital technologies (e.g., smartphones) and of mHealth apps. Additionally, perceived privacy risks regarding the use of mHealth apps were assessed (using items by [18]) and the participants’ digital health literacy was evaluated (using the scale by [33]). The digital health literacy scale contains the subdimensions operational skills, navigation skills, information searching, evaluating reliability, determining relevance, adding content, and protecting privacy.

After the Conjoint study, the participants evaluated their intention to use an mHealth app for aftercare (using three items adapted from [34]) and their general attitude towards such an app (using a semantic differential with ten adjective pairs). All items were measured on six point Likert scales. The reliability of the scales was confirmed using Cronbach’s Alpha (\(\alpha > .7\)).

3.3 The Sample

N = 180 participants completed the questionnaire. The participants were recruited within the social network of the authors and through online discussion forums. The age of the participants varied between 15 and 75 (\(M = 33.96, SD = 13.4\)). 60.6% were women. The level of education of the participants was mixed. Table 2 in Sect. 4 provides an overview of the demographic characteristics of the sample.

Most participants perceive their health status to be rather good (32.8%) or good (41.7%). 13.9% evaluate it as rather bad to very bad. 24.4% have experienced professional aftercare. Regarding their experiences with technologies, 98.9% of participants use a smartphone. 70.6% of the sample use or have used digital health apps. On average, the participants do not agree, nor reject, to perceive privacy risks for mHealth for aftercare (\(M = 3.36, SD = 1.19\) on a scale from 1 (very low) to 6 (very high)). Their mean digital health literacy is high (\(M = 4.52, SD = .63\)).

4 Results

The results of the Conjoint analysis show, on the one hand, how important each factor is for the choice of the optimal mHealth app (relative importance scores), and one the other hand, the relative preference for the levels of each factor (part worth utilities). These values are relative and can only be evaluated in comparison to the other included factors, or factor levels of one factor, respectively. A negative part worth utility does thus not indicate rejection, but that this level is less preferred than the levels with higher part worth utilities.

The software Sawtooth Software was used for the development, acquisition, and analysis of the Conjoint study and the questionnaire. Additionally, analysis of variances and Chi Square Tests were used for the further analysis of the user groups identified in the Latent Class Analysis.

4.1 Average Decision Pattern

Who has access to the data is the most important factor for the decision to use an mHealth app for aftercare with an importance score of \(56.6\%\) \((SD = 16.9)\) (see Fig. 5). With a large difference, the type of assistance is the second most important factor (\(27.5\%, SD = 15.3\)). Which features the app provides is the least important of the three included factors (\(15.9\%, SD = 8.6\)).

Fig. 5.
figure 5

Relative importance scores of the three factors (n = 180).

Full size image
Fig. 6.
figure 6

Part worth utilities of the factor levels (n = 180).

Full size image

In the part worth utilities (cf. Fig. 6), the importance evaluation is reflected again: For the most important factor, the data access, the participants distinguish largely between the three factor levels. The health insurance is the least preferred party for data access (\(-82.5\%, SD = 42.2\)). Most preferred is medical personnel (\(66.2\%, SD = 39.7\)) and trusted persons score in between the other two (\(16.4\%, SD = 47.4\)).

As kind of assistance for getting to know and interacting with an mHealth app for aftercare, a personal consultation is most preferred (\(28.5\%, SD = 32.2\)). Least accepted is a preventative course (\(-17.8\%, SD = 30.6\)), closely followed by the online video option (\(-10.7\%, SD = 42\)).

Remote monitoring of vital parameters is the most preferred feature (\(6.5\%, SD = 21.9\)) and the emergency call function is almost as desired (\(3.1\%, SD = 21.1\)). The medical reminder function is less preferred (\(-9.6\%, SD = 23.5\)).

4.2 General Evaluation of mHealth Apps for Aftercare

In Fig. 7, the mean evaluation of ten adjective pairs is depicted, showing a generally positive attitude towards mHealth apps for aftercare. They are especially perceived as being useful and practical. Nevertheless, the participants still expect them to isolate them from people and to not be fun or trustworthy. The overall positive attitude corresponds to a rather high use intention with \(M = 4.4, SD = 1.04\).

Fig. 7.
figure 7

General evaluation of mHealth apps for aftercare on a semantic differential (n = 180).

Full size image

4.3 Privacy Supporters and Assistance Seekers

In the Conjoint results, large standard deviations indicate a high variation within the sample. To explain this variance by identifying user segments with differing decision patterns, a Latent Class Analysis was conducted. A two-group segmentation showed the best fit with the data (using the following indices: percent certainty, Akaike information criterion (CAIC), and chi square). The two user groups consisted of 36.4% (n = 61) and 63.6% (n = 119) of the sample showing adequate group sizes for further analysis. In the following, these two user groups are described regarding their decision patterns in the CBC experiments and then further analyzed regarding attitudes and user diversity factors.

Differing Decision Patterns. The juxtaposed relative importance scores for the three factors in Fig. 8 show the main distinction between the two user groups that lend them their names: The Privacy Supporters place even more importance on who has access to the data collected by an mHealth app for aftercare (73.3) than the mean sample and the second user group (47.0). In contrast, the type of assistance is less important to the Privacy Supporters (13.1). Only little difference can be identified regarding the importance of the provided feature (Privacy Supporters: \(13.6\%\), Assistance Seekers: \(9.0\%\)). In contrast, for the Assistance Seekers the type of assistance and who has access to the data are almost equally important (assistance: \(43.9\%\), data access : \(47\%\)).

Differences are also visible in the evaluation of the factor levels (see Fig. 9). Regarding the type of assistance, Assistance Seekers strongly prefer a personal consultation (78.1). In comparison, he online video is strongly rejected (\(-53.7\)), whereas this is the second best option for the Privacy Supporters (8.3).

Different decision patterns regarding who should have access to the data are also prevalent. Whereas the Privacy Supporters strongly prefer the medical personnel (105.7) and regard trusted persons as a relatively mediocre option (8.6), the Assistance Seekers prefer trusted persons (63.3) and the medical personnel not so much (14.5).

The distinction between the type of feature also varies between the two user groups. The Privacy Supporters prefer remote monitoring the most (18.3) and medical reminders the least (\(-22.4\)). In contrast, the Assistance Seekers prefer emergency call functions the most (12.4) and remote monitoring the least (\(-14.6\)).

Fig. 8.
figure 8

Relative importance scores of the three factors juxtaposed by the two user types (n = 180).

Full size image
Fig. 9.
figure 9

Part worth utilities (rescaled for comparability) of the factor levels juxtaposed by the two user types (n = 180).

Full size image

Characteristics of the Two User Groups. Table 2 shows the differences in age, gender, and education level between the two user groups. Analysis of variance show that the Assistance Seekers are significantly older (\(M = 37.2, SD = 14.6\)) than the Privacy Supporters (\(M = 32.3, SD = 12.5\)) with \(F(1,178) = 5.42, p < .05\). 67.2% of the Assistance Seekers are women as are 57.1% of the Privacy Supporters. There is no significant gender difference. The education level between the groups does also not differ significantly.

Table 2. Demographic characteristics of the sample and the two user types (N = 180).
Full size table

Furthermore, the groups can be described by their varying attitudes and skills. The Privacy Supporters show a significantly lower score of perceived privacy risk for mHealth apps (\(M = 3.23, SD = 1.2\)) than the Assistance Seekers (\(M = 3.61, SD = 1.13\)) with \(F(1,178) = 5.8, p < .05\). Also, the Privacy Supporters score significantly higher in operational skills, a subdimension of digital health literacy (\(M_{Privacy Supporters} = 5.6, SD = 0.66, M_{Assistance Seekers} = 5.32, SD = 0.85\)), (\(F(1,178) = 5.82, p < .05\)). Operational skills describe basic abilities to use a keyboard, mouse, and website features [33]. The two groups do not differ regarding their intention to use an mHealth app for aftercare, both in the given scenario and in the general evaluation through the semantic differential.

5 Discussion

Particularly in a future with shortages in healthcare resources, digital health technologies show a high potential to improve the quality of healthcare. Especially the use of mHealth apps on Smartphones and other mobile devices – which are nowadays available to almost everybody – is a very promising option to reach patients, e.g., to improve support in aftercare. For a successful roll-out of such technologies, the patients’ acceptance is a decisive prerequisite. Therefore, this study investigated the acceptance of mHealth apps for aftercare. The special focus lay on understanding which requirements are most important for mHealth acceptance and how these are influenced by user diversity.

In a qualitative prestudy, the following three factors were introduced as decisive requirements: a) the features and hence type of support offered by the mHealth app for aftercare, b) privacy requirements, for which who has access to the data was most important, and c) the kind of received assistance in getting to know, and learning to interact with, the mHealth app.

In a next step, these three factors were examined using a Choice Based Conjoint approach to not only study the individual influence of each factor on user acceptance, but also the trade-offs between the factors by using a multi-factorial design. The Conjoint study confirmed previous empirical work in that the privacy requirement regarding who has access to data is the most important factor for the decision to use digital health technologies [25, 31]. This result once again highlights the importance of privacy and the need to better understand and deal with privacy requirements in the context of mobile and digital health technologies.

The participants in this study did not differentiate much between the three offered features – medical reminders, remote monitoring, or emergency functions. The relatively low importance scores might thus indicate that the type of offered feature is not really important for the acceptance of mHealth apps for aftercare. Another explanation for this result could be that the type of feature is not irrelevant, but that the three used options are equally accepted, so that the choice between them does not influence the decision for an mHealth app. In order to understand which features are preferred and accepted, future work should examine the importance of features more closely by comparing other features that may show a larger distinction.

With a large gap between its importance and the importance of who has access to the data, the type of assistance in getting to know the mHealth app was the second most important factor. However, the high variance within the data indicated that the mean results hide a strong diversity of decisions patterns. For that reason, a Latent Class Analysis was conducted from which two user groups with differing decisions patterns could be identified.

For the first user group, the Privacy Supporters, who has access to the data was the only important factor for the choice of an mHealth app for aftercare. Surprisingly, at the same time the Privacy Supporters perceived less privacy risks for mHealth than the second user group – which should logically be associated with a lower importance of privacy within the Conjoint. This combination of results thus indicates that the other two factors are even less important to the Privacy Supporters. More research is needed to find which factors are important to this user group.

For the second user group, the Assistance Seekers, privacy is also very important, but the type of assistance is almost as important. This user group is significantly older and shows lower scores in a subdimension of digital health literacy. They least accept getting to know an mHealth app for aftercare through an online video only and instead prefer a personal consultation. This shows how important user diversity is: the Assistance Seekers need more assistance for the mHealth app and prefer personal consultations. Future research is needed to study the importance of assistance – a factor that has not yet drawn much research attention. For a successful roll-out of mHealth technologies, assistance and particularly the option for personal consultations should be provided especially for older target groups and patients who feel insecure in interaction with mobile technologies. Therefore, mHealth apps can support patients but should not substitute face-to-face care.

All in all, a generally positive attitude and high intention to use mHealth for aftercare was observed. This shows that mHealth apps for aftercare can be successful when the diverse patients’ requirements – e.g., for data access and assistance – are met. In light of the high user diversity, these may best be met by offering patients different options within a running system, e.g., a choice for who has access to data and how to get assistance for interacting with the technology.

That said, this study is not without shortcomings. The multi-method approach combined the advantage of qualitative research to identify important user requirements with the ability of Conjoint studies to quantify and experimentally modulate the importance and trade-offs between the important factors. But, in Conjoint studies only few factors can be included meaning that the results are only valid for this combination of factors and factor levels. Therefore, all results are restricted to the three included factors and cannot directly be compared to other studies.

Additional caution has to be taken considering the samples. The Conjoint sample included people of all age groups but was quite young on average. Many mHealth apps, especially for aftercare, are used by people of all age groups but show a particularly high potential for older adults. The results showed that older users have different requirements than younger adults. Future research should target older adults to analyze their requirements in more detail since this age group might still be quite diverse.

A last comment addresses the cultural setting of the study. Technology acceptance and privacy attitudes are known to differ between different cultures and in different countries [17, 37]. The present results are thus only valid in the German setting of the study. People of other cultures and countries may show different requirements for mHealth for aftercare.