This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in JMIR mHealth and uHealth, is properly cited. The complete bibliographic information, a link to the original publication on https://mhealth.jmir.org/, as well as this copyright and license information must be included.
During the COVID-19 pandemic, but also in the context of previous epidemic diseases, mobile apps for smartphones were developed with different goals and functions, such as digital contact tracing, test management, symptom monitoring, quarantine compliance, and epidemiological and public health research.
The aim of this study was to explore the potential for the acceptance of research-orientated apps (ROAs) in the German population. To this end, we identified distinctive attitudes toward pandemic apps and data sharing for research purposes among smartphone users in general and with a focus on differences in attitudes between app users and nonusers in particular.
We conducted a cross-sectional, national, telephone-based survey of 1003 adults in Germany, of which 924 were useable for statistical analysis. The 17-item survey assessed current usage of pandemic apps, motivations for using or not using pandemic apps, trust in app distributors and attitudes toward data handling (data storage and transmission), willingness to share coded data with researchers using a pandemic app, social attitudes toward app use, and demographic and personal characteristics.
A vast majority stated that they used a smartphone (778/924, 84.2%), but less than half of the smartphone users stated that they used a pandemic app (326/778, 41.9%). The study focused on the subsample of smartphone users. Interestingly, when asked about preferred organizations for data storage and app distribution, trust in governmental (federal or state government, regional health office), public-appointed (statutory health insurance), or government-funded organizations (research institutes) was much higher than in private organizations (private research institutions, clinics, health insurances, information technology [IT] companies). Having a university degree significantly (
These findings indicate a high potential for the adoption of ROAs among smartphone users in Germany as long as organizational providers engaged in development, operation, and distribution are state-funded or governmental institutions and transparency about data-using research institutions is provided.
After the outbreak of the COVID-19 pandemic, various governments and the European Union decided that digital solutions, most notably smartphone apps, should contribute to pandemic management and research [
In recent years, there has been an increasing amount of qualitative, quantitative, and mixed methods studies in an international context that have explored behaviors, motivations, and perceptions about mobile phone–based apps for health. They focus on health apps in general [
In the context of health apps in general and pandemic apps in particular, current debate is mainly focusing on privacy concerns and perceptions toward sharing health data [
This study explored the potential for the adoption of ROA among the German population. We focused on smartphone users and aimed to identify specific challenges for app usage. To our knowledge, this is the first nationwide, telephone-based survey study in Germany since the first pandemic apps (“Corona-Warn-App” and “Data-Donation-Apps”) were released nationally. It is also, to our knowledge, the first study focusing primarily on individual data sharing via smartphone apps for pandemic research. Leading research questions were (1) “Which sociodemographic and personal factors influence the use of a pandemic app among smartphone users in the German population?” and (2) “How do users and nonusers of pandemic apps differ in their motives, attitudes toward pandemic apps, and willingness to share data with researchers?” The objectives of this paper were therefore to identify distinctive attitudes toward pandemic apps and data sharing for research among smartphone users in general and with a focus on differences in attitudes between app users and nonusers in particular. The results can inform empirically based ethical recommendations for the future development, design, and implementation of ROA.
We designed a survey comprised of 17 question units (see
A representative phone-based survey seemed more appropriate than online panels to reach people who are not internet-savvy or avoid online surveys. Especially when it comes to questions of public acceptance of modern technologies, such as smartphone apps, a broader sampling strategy seemed more appropriate to make statements about the whole population.
Inclusion criteria were people (1) aged ≥18 years, (2) with a registered address in Germany, and (3) who were literate in German. The sample population was comprised of private households in Germany with at least one landline connection and people with at least one mobile phone connection. The population survey was conducted by the company Kantar GmbH. It took approximately 15 minutes to 20 minutes to complete the questionnaire. A dual-frame sampling approach (ie, taking into account both landline and mobile phone numbers) was used. The landline (n=703) and cell phone (n=300) samples were then combined by statistical weighting according to the demographic statistics of the German population (see
A representative telephone-based population survey with 1003 people in Germany aged 18 years or older was conducted between December 10, 2020 and January 18, 2021. A sample size of 1000 interviews was originally planned; 3 further interviews were conducted due to already arranged appointments with target persons. The German population aged ≥18 years currently is around 69.4 million. In current survey research, 1000 respondents have proven to be a practicable and statistically acceptable sample size for representative population surveys in Germany. We can refer to the seminal national survey study by Richter et al [
The survey instrument for the phone questionnaire contained closed-ended question types. The questionnaire encompassed 17 question units in the German language and entailed the following domains: (1) current usage of smartphone and pandemic apps, (2) motivations for using or not using pandemic apps, (3) trust in app distributors and data storage, (4) willingness to share coded data with research institutions using a pandemic app and attitudes toward data handling, (5) social attitude toward app use, and (6) demographic and personal characteristics (
Descriptive statistics were calculated for all items. Since we were interested in pandemic app usage and attitudes toward data sharing with research institutions, we focused on the subsample of smartphone users in our statistical analysis (n=778). Chi-squared tests were used to identify differences between users and nonusers of the app regarding the willingness for data sharing, social attitudes toward app usage, perceptions of the trustworthiness of the app provider, and the preferred location for data storage. To quantify the factors influencing app usage, logistic regression analysis was used. For this purpose, sociodemographic variables such as age, gender, education, place of residence, and migration background were included in the model. In addition, personal experience of being directly infected or knowing someone who has been infected by the COVID-19 virus was elicited (personal affection). Statistical significance was determined by
Of the 924 participants that were included for statistical analysis, 84.2% (778/924) stated that they used a smartphone. The following analysis refers to this subset of smartphone users, as we deemed smartphone usage a condition for technology-specific considerations on pandemic app usage. We assumed that inclusion of non-smartphone users in the statistical analysis would have engendered a mixed sample of distinctive versus hypothetical usage attitudes.
Sociodemographic profile of survey participants (n=924) and the subsample of smartphone users (n=778).
Characteristic | Survey participants (n=924), n (%) | Smartphone users (n=778), n (%) | |
|
|||
|
Female | 474 (51.3) | 404 (51.9) |
|
Male | 450 (48.7) | 374 (48.1) |
|
Nonbinary | 0 (0) | 0 (0) |
|
|||
|
18-30 | 153 (16.6) | 153 (19.7) |
|
30-39 | 120 (13.0) | 119 (15.3) |
|
40-49 | 175 (18.9) | 168 (21.5a) |
|
50-59 | 183 (19.9a) | 153 (19.6a) |
|
60-69 | 128 (13.8a) | 94 (12.1) |
|
≥70 | 165 (17.9) | 91 (11.7) |
|
|||
|
None/still in school | 51 (5.5) | 44 (5.7) |
|
Without A-level | 570 (61.7) | 444 (57.1) |
|
A-level | 138 (14.9) | 133 (17.1) |
|
Academic degree | 166 (17.9a) | 157 (20.1a) |
|
|||
|
Yes | 170 (18.4) | 156 (20.1) |
|
No | 754 (81.6) | 622 (79.9a) |
|
|||
|
West German states | 765 (82.8) | 645 (83.0a) |
|
East German states (including Berlin) | 159 (17.2) | 133 (17.0a) |
|
|||
|
Yes | 315 (34.1) | 294 (37.8) |
|
No | 609 (65.9) | 484 (62.2) |
aDue to the fact that we used weighted data, the sample size may differ by ±1 in some analyses due to rounding effects. Calculating with weighted data also has the effect that percentages can deviate minimally in the decimal place compared with the quotient n/N in natural numbers. For a detailed description of the weighting, see
Our analysis of attitudes focused on 2 major topics. First, we report attitudes toward app providers, by which we understand organizations and institutions involved in the development, provision, and operation of pandemic apps (see
Attitude responses toward (A) preferred location for data storage and (B) trustworthy app provider among smartphone users (n=778).
Attitudes among people willing to share data (“data sharers,” n=653) for research via an app among smartphone users (n=778).
Attitude responses among data sharers (n=653) | Results, n (%) | |
|
||
|
Don’t know/none of these | 10 (1.5) |
|
Data collected by a fitness watch | 215 (33.0b) |
|
Continuous data (ambient temperature) | 298 (45.6) |
|
Health-related data | 330 (50.5) |
|
Location and movement data | 366 (56.1b) |
|
Contacts with other people | 436 (66.8) |
|
Data manually entered in the app | 447 (68.5) |
|
Test results | 553 (84.8b) |
|
||
|
Don’t know/none of these | 8 (1.2) |
|
Calling a video hotline of the research institute | 58 (8.9) |
|
Calling a telephone hotline of the research institute | 101 (15.5) |
|
Sending the data via SMS | 123 (18.9b) |
|
Sending the data via email | 169 (25.9) |
|
By entering the data on the website of the research institute | 207 (31.7) |
|
Sending the data automatically to the research institute | 379 (58.1b) |
|
By enabling data sharing in the app each time | 437 (66.9) |
|
||
|
Not so important/not at all important | 158 (24.2) |
|
Very/rather important | 495 (75.8) |
aMultiple answers were possible.
bDue to the fact that we used weighted data, the sample size may differ by ±1 in some analyses due to rounding effects. Calculating with weighted data also has the effect that percentages can deviate minimally in the decimal place compared with the quotient n/N in natural numbers. For a detailed description of the weighting, see
Attitudes among people not willing to share data (“non-data sharers,” n=125) for research via an app among smartphone users (n=778).
Attitude responses among non-data sharers (n=125) | Results, n (%) | ||
|
|||
|
Don’t know | 3 (2.2b) | |
|
Other reasons | 9 (7.2) | |
|
I am worried that the data will be leaked. | 78 (62.2b) | |
|
I doubt that this data will help research. | 78 (62.4) | |
|
I am concerned about unknown third parties using my data. | 85 (68.2b) |
aMultiple answers were possible.
bDue to the fact that we used weighted data, the sample size may differ by ±1 in some analyses due to rounding effects. Calculating with weighted data also has the effect that percentages can deviate minimally in the decimal place compared with the quotient n/N in natural numbers. For a detailed description of the weighting, see
Multivariable correlates of pandemic app usage.
Variable | Coefficient | |
Constant | –0.564 | .02 |
Male (vs female) | 0.069 | .65 |
Age groups | 0.055 | .27 |
University degree (vs no university degree) | 1.081 | <.001 |
Eastern Germany (vs western Germany) | –0.494 | .02 |
Immigration background (vs no immigration background) | –1.242 | <.001 |
Being affected (vs not personally affected) | 0.279 | .09 |
Different attitude responses toward (A) preferred location for data storage and (B) trustworthy app provider among app users (n=326) and nonusers of pandemic apps (n=452).
Different attitude responses toward (A) data sharing for research and (B) the statement, "the use of pandemic apps is a social duty" among app users (n=327 and n=326, respectively) and nonusers of pandemic apps (n=452 and n=451, respectively). Due to the fact that we used weighted data, the sample size may differ by ±1 in some analyses due to rounding effects. For a detailed description of the weighting, see
Our survey revealed that German smartphone users demonstrated a strong preference for state-funded or governmental organizations with regard to storage of app data as well as app providers when asked which providers the participants considered most trustworthy. We considered the statement of preference for a certain app actor for data storage to be an indication of trust in that app actor (or one's smartphone) for data storage and management. As
Interestingly, German software companies were classified twice as trustworthy (344/778, 44.2%) as international companies (161/778, 20.8%*).
Less than half of participants stated that they used a pandemic app (326/778, 41.9%).
We used logistic regression analysis to determine which sociodemographic characteristics influenced the probability of using or not using a pandemic app. Having a university degree significantly (
The high level of trust in state-funded app providers is matched by the fact that 83.9% (653/778) of smartphone users were willing to provide their app data for state-funded research. We called that subsample of people willing to share data with research institutions “data sharers.” When asked about what kind of data people were willing to share for research, the vast majority indicated test results (553/653, 84.8%*), followed by contact tracing data (436/653, 66.8%). In contrast, only one-third (215/653, 33.0%*) were willing to share data via a pandemic app that were originally collected by a digital mobile device, such as a fitness watch (
The 3 most frequent reasons why people were not willing to share their data for research (125/778, 16.1%) were concerns about lack of control of app data (85/125, 68.2%*; ie, the concern that third parties were using the data without consent), that data would be leaked (78/125, 62.2%*), and doubts on whether app data really would bring research forward (78/125, 62.4%;
We found substantial differences in attitudes toward pandemic app providers between users and nonusers of pandemic apps (
As indicated in
This study examined pandemic app usage and attitudes toward data sharing with research institutes among a sample of smartphone users, which represented a subsample of a representative population survey in Germany. Our study provides several important findings. In the following sections, we focus on 4 of them. First, the results showed a high willingness to share data with state-funded research institutes for pandemic research, but the willingness for data sharing went along with a strong need for self-controlled data handling and transparency about the involved data-analyzing research institutions. Second, there was a remarkable decline in trust toward private providers and organizations involved in data storage and distributing pandemic apps when compared with state-funded organizations. Third, regression analysis showed app usage is positively correlated with a higher level of education. Fourth, our study revealed significant differences in trust attitudes between app users and nonusers.
One of the aims of this study was to elicit peoples’ attitudes and concerns toward sharing data collected by pandemic apps. The overwhelming support for data sharing via pandemic apps for research purposes among smartphones users in Germany is consistent with previous surveys that examined the willingness for “data sharing” or “data donation” [
In the international context, the picture on the willingness to share data in general is quite heterogeneous. For example, Abeler et al [
When informing policy makers about affirmative attitudes among smartphone users toward willingness to share data with research institutes, those findings should be contextualized with stated preferences for criteria of data sharing and processing. In our study, we explored different criteria that allow for a more nuanced picture: (1) the kind of data to be shared (eg, health data, location data), (2) transparency about data-receiving research institutions, (3) the mode of data collection, and (4) data transmission (
In summary, we conclude that those willing to share data for research purposes express a strong interest in a self-determined way of data sharing. This means that mechanisms of manual handling such as activation of data transfer, a set of selected kinds of data, and comprehensible and detailed information about data processing would likely increase willingness for data sharing with research institutions. However, since automatic data transmission is also endorsed by a large portion of participants, the picture is more complex. An ambiguous tendency in attitudes toward data transmission was also reported by Becker et al [
One important finding in our regard is the extent to which attitudes toward state-funded and private app providers vary among smartphone users: Private providers were considerably less trusted with data storage and providing an app. Here, we interpreted the preference of a storage location as an expression of trust in this specific organization. We found that trust in state-funded research institutes and governments for the storage of app data is very high (almost two-thirds of smartphone users). This is an encouraging message for state-funded research intuitions even if ongoing public discussions about privacy, data security, governmental surveillance practices, and centralized versus decentralized storage solutions for pandemic apps might give the opposite impression [
The large gap between state-funded and private providers poses a challenge for the reality of pandemic app development, which is mainly achieved via public-private partnerships. Considering that pandemic research is of extremely high public health relevance and therefore differs from many (not all) other areas of mobile health (mHealth) where health behavior or health research addresses a smaller population of patients, research on pandemic apps can clearly benefit from a strong emphasis on the public partner. However, the high level of trust in government and state-funded research institutes as app providers can be gambled away if there is an increasing reliance on private-public partnerships in which tech companies co-determine the technical and design solutions, as was the case when Apple and Google offered governments their common exposure notification application programming interface (API) [
In line with the large longitudinal survey from Munzert et al [
Regarding the factors of age and gender, other studies found that both were no or only weak predictors for pandemic app usage [
Our study indicates a basic typology differing between users and nonusers of pandemic apps, which relies mainly on attitudinal features and less on sociodemographic factors. Type one—the data sharer—is characterized by high trust in state-funded research institutions and app providers, high willingness to share data, and seeing pandemic apps as useful for pandemic research as well as agreeing that there is a societal duty to share data to help with pandemic containment. The other type—the data-sharing skeptic—can be characterized by lower trust in state-funded app providers, decreased willingness for data sharing with research organizations, and considerably lower agreement with the view that using pandemics apps is a societal duty. These empirical findings can help to improve our understanding of who future app researchers would want to address. As problematized in other fields such as organ donation [
The findings in this report are subject to at least 3 limitations. First, with regard to inclusion criteria, the participants of the survey were all residents of Germany aged 18 years or older (which is also a common ethical-legal requirement for this type of survey) and accessible via a landline or mobile phone number, so no statements can be made for people younger than 18 years or people without any telephone connection or using call blockers. Since, for example, the national pandemic app (Corona Warn App) is available for teenagers aged 16 years and older, our sample of smartphone and app users is not exactly representative of all potential app users. However, since age was not a statistically significant factor for app uptake in our survey, the question arises whether including younger populations would have had statistically significant effects on public attitudes toward pandemic app usage. Regarding people using call blockers or people without landline or cell phone numbers, we could only speculate that these populations may have a rather skeptical attitude toward sharing app data.
Second, our decision to focus our statistical analysis on smartphone users (n=778) was based on considerations that eventually non-smartphone users may not have accurate conceptions and no concrete opinions about specific applications and app data, so answers by non-smartphone users about app details could have had a rather speculative character. The characteristics of the smartphone user sample slightly differ from the German population in 3 aspects: Smartphone users are somewhat younger (–3.35 years), slightly higher educated (4.4% more with A-Level or university degree), and slightly more often personally affected by current COVID-19 disease (3.7% more are affected). Therefore, the generalizability of the present results to older people and the overall German population is limited. Nonetheless, we consider our sample more informative for app developers and governance policies than surveys based on online panels involving a convenience sampling.
Finally, due to the time limitation for telephone surveys, we opted to not provide a definition for the “use” of smartphones and pandemic apps in our questionnaire (see
The rapidly expanding field of apps in mHealth is very diverse with respect to architecture, features, and purposes. Smartphones users might be confused about different types of pandemic apps [
Our study indicated that trust in and trustworthiness of different app providers for data storage and app distribution, self-determination of data storage and transmission, and the social attitudes toward pandemic management are also crucial for such governance. Furthermore, lay-accessible information—also considering various sociocultural groups and different levels of eHealth literacy—should be part of future frameworks. Future research, (eg, on the incentivization of app adoption and data sharing or “data donation” [
In order not to gamble away the high willingness to share data via an app with state-funded research institutes, the life cycle of pandemic apps and all organizational providers involved in it should be made transparent.
From an ethical point of view, public-private partnerships for app development and app operation might be reconsidered because public and private app providers are perceived very differently among smartphone users. This applies all the more when it comes to public health emergencies such as pandemics when digital solutions are rapidly recommended to fix challenges in management and containment. At least, we assume, transparency of the engaged sectors and parties can help to engage as many citizens as necessary for valid ROA deployment.
Questionnaire translated into English.
Weighting model.
Flow chart: recruitment and sample.
Attitudes among people willing to share data (“data sharers,” n=653) for research via an app and people not willing to share data for research via an app (“non-data sharers,” n=125) among smartphone users (n=778).
Pandemic app use in West and East German Federal States.
application programming interface
Bundesministerium für Bildung und Forschung
Coordination on mobile pandemic apps best practice and solution sharing
eHealth Literacy Scale
Ethical, Legal, Social Issues
information technology
mobile health
Forschungsnetzwerk der Universitätsmedizin zu COVID-19
research-orientated app
The interview study was conducted as part of the Coordination on mobile pandemic apps best practice and solution sharing (COMPASS) project. We particularly thank Johannes Huxoll from Kantar GmbH for his valuable support in pilot-testing the questionnaire and providing supplementary methodological material on the survey data. We especially thank Rüdiger Pryss and Dagmar Krefting for valuable input in planning the study and discussing preliminary results. We acknowledge Theresa Petzold for her support with the first descriptive statistical analysis and with data maps as illustrations. Ruben Sakowsky and Julia Perry deserve thanks for their thorough language editing. The COMPASS project is part of the Nationales Forschungsnetzwerk der Universitätsmedizin zu COVID-19 (NaFoUniMedCovid19) and is funded by the German Federal Ministry for Education and Research (Bundesministerium für Bildung und Forschung [BMBF]; Grant number 01KX2021). The BMBF had no role in the study design, conduct, nor writing of the article. The work is licensed under the Creative Commons Attribution License CC-BY 4.0.
None declared.