Design and Validation of a Questionnaire to Measure the Acceptance of Telemedicine by Healthcare Professionals in Germany

Background: Telemedicine allows healthcare professionals to manage patient treatments remotely and to benet from the interaction of telemedicine centers. Even though telemedicine responds to the current burdens in healthcare systems, these complex infrastructures depend on technology, nancing, organization, policy, legislation, and, not least, on their corresponding acceptance by the user. Little is known in this context about the drivers for this acceptance by healthcare professionals giving treatment. Germany, with the highest healthcare spending in the world, is particularly lagging here. As there is less focus on guidance for building acceptance constructs, and, in particular, little research on the acceptance of interactive telemedicine by healthcare professionals in Germany, the objective was the collection of the relevant factors inuencing German healthcare professionals in the implementation of telemedicine service programs (TSSs). Acceptance constructs were built and operationalized in a questionnaire, taking into consideration the gold standard technology acceptance model of Davis. Methods: A literature review was conducted to identify the acceptance factors that had already been detected and frequently conrmed. These factors were evaluated according to the coverage of the interactive infrastructures with telemedicine centers and how accurately they tted Germany. The identied factors were adapted, and missing factors were designed, via an idea collection for items. The reliability of the developed constructs was tested using a eld test to check the constructs using item revision analysis. Results: Frequently supported external variables, taken from 17 ltered studies covering the technical, organizational, social, legal, and individual context, were identied. However, the review showed that there was less focus on the German setting and the interactive aspect, so the market-related variables were adapted and new constructs of “trust in the telemedicine center” and “relationship to patient” were added. Ten variables were proven to have a reliability, measured with Cronbach’s Alpha, of more than 0.7. Conclusion: This paper enhances the already existing technology acceptance studies in healthcare by covering German specications and the interactive character of TSSs. A corresponding questionnaire has been developed ready for future research.

. Technology acceptance research is a broad eld and looks at the area of information systems. The gold standard for evaluating technology acceptance is the technology acceptance model (TAM) of Davis [23]. Davis claimed that the factors perceived usefulness (PU) and perceived ease of use (PEOU) determine attitude, and that this again in uences behavioral intention (BI). Further, PEOU affects PU, and PU independently affects BI [24]. The term BI describes what leads to actual use and can be understood to account for acceptance [24]. Later, during the evolution of the TAM, the model was extended in TAM 2, which removed attitude but added a variable covering social in uence, with the factor known as subjective norms (SN) that directly in uences PU and BI [25,26,27,28] Venkatesh et al. [28] later investigated the uni cation of acceptance studies and developed the Uni ed Theory of Acceptance and Use of Technology (UTAUT). The most critical step was adding the facilitating conditions coming from the general research into IT acceptance as one determinant of BI [26,29]. The TAM has reached the gold standard, as reviews of this widely-used model have shown that the TAM is a reliable model for analyzing acceptance by individuals [25,26,30], not least because of its simplicity and clarity [31,32,33]. However, several revisions to the TAM have been proposed, and the model has been questioned; in addition, many other determinants in uencing the core constructs have been tested in the past [26]. It is certainly the case that the results for the factors, relationships and effects differ by system types and users [31,34]. Looking at healthcare technology, the TAM has been the most commonly used model to study acceptance [18,35] and validated through questionnaires that are able to explain the reasons why a particular technology is used [36].

Methods
For the questionnaire construction, a literature review was conducted to collect the factors that build and in uence acceptance. A frequency analysis identi ed the most common con rmed factors, and these were then clustered. After reviewing the content of the factor clusters, an item collection was performed to adapt and build the constructs for the German interactive healthcare context. The constructs were operationalized in statements, and the questionnaire was formed. A test of the questionnaire was conducted, and an item revision analysis checked the reliability of the constructs.

Literature search strategy
A manual literature review was performed to capture the constructs to explain the factors in uencing acceptance of an interactive telemedicine structure and to check the existing studies on technology acceptance of the German healthcare system. International studies published between January 2000 and January 2020 were selected from PubMed by keywords. To ensure that the papers were of high quality, only studies that had been peer-reviewed were included. The search keywords were telemedicine OR telehealth AND technology acceptance model. Studies in languages other than German and English, as well as studies that did not have open access or full text availability, were not included. The review approach and ndings are illustrated in Fig. 1, which follows the PRISMA guidelines for reviews [37].

Study exclusion criteria for the literature review
The following exclusions were made to produce a clear ltered pool of factors. After manually reviewing the clinical trials, qualitative research (e.g. interviews, pure literature reviews, and research on technology assessment (e.g. effectiveness and feasibility)) was excluded. Analyses of patient acceptance or acceptance by other non-healthcare target groups were not considered. Most importantly, research on the assessment of healthcare IT and on programs without interactive telemedicine aspects (by which we mean communication with a telemedicine center and/or the patient) such as documentation software, single devices, electronical patient records or general technology or information technology, was taken out. Only studies that used the TAM, UTAUT or the extension variables were considered, to remain with the core constructs and make a similar basis for the item creation, to ensure compatibility. Moderating factors, especially in terms of socio-demographics, were excluded. Out of 253 records, 17 studies were ltered out and passed the eligibility test. Frequency analysis and clustering of acceptance factors for studies in the review Fifty-eight different construct relationships could be identi ed in these 17 studies [34,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54] Only those constructs that were supported more than twice in a construct relationship were picked out. Clearly, the review shows that the core constructs from the original models, like behavioral intention (BI), impacted by attitude (ATT), and in uenced by perceived ease of use (PEOU) and perceived usefulness (PU), were also identi ed in these studies as essential variables. Further analysis of the external variables showed that they can be clustered into ve contexts [55], as set out below.
The technological context explains the handling of the system, including PEOU, and the performance impact, including PU. Next, the organizational context is covered by the facilitator variable, which describes whether there are suitable resources and a suitable environment for the clinical task and the system implementation, as well as the compatibility variable, which describes the t to the current tasks. The facilitator factor is in uenced on the behavioral level by the variable of perceived behavioral control, and these can be merged. The individual context describes the con dence and competence in handling a system, using the variable habit (HAB). The social context is covered by the subjective norms (SN) variable, which describes the in uence on the user of other people. Another nding is the in uence of market-related factors. The review showed that there were economic variables that were focused on nancing, such as rewards in terms of nancial support or compensation for medical fees [38,39,42]. Another market factor is the regulatory or legal context. In this area the perceived importance of data security (IOD) stands out as an in uence on the behavioral intention [40,45,48]. However, for a comprehensive review there were several gaps. First, a view on the interaction framework and the service context variables was missing. Secondly, investigations focusing on the German market were absent. Therefore, the variables needed to be analyzed and equivalent indicator dimensions built. It was necessary to build the missing dimensions by an idea collection of items. A su cient item pool for each construct was created. Table 1 shows the origin of the items and the de nition of each construct, as derived from the analyzed studies. The degree to which the HCP believes that using TSS is free of efforts / TSS is di cult to use [34,75].

PEOU1
Learning to operate TSS is easy for me. [76] PEOU2 I nd TSS easy to use. [27,74,76] PEOU3 I think using TSS is easy for my patients. [76] PEOU4 I nd TSS to be exible when I interact with it. [76]

PEOU5
My interaction with TSS is clear and understandable. [27,74,76] Facilitators The degree to which the HCP believes that organizational and technical resources and environment exist to support the implementation and use of the system [28,74].

FAC1
I have the necessary resources to use the system. [74] FAC2 I will use TSS if I receive appropriate training. [45] FAC3 I will use TSS if I receive the necessary technical assistance. [45]

FAC4
Given the resources, opportunities and knowledge it takes to use the system, it is easy for me to use the system. The degree to which TSS is perceived as being consistent with the existing needs, values and past experiences of the HCP, who is a potential adopter of the system [76].

COMP1
Using the TSS is compatible with all aspects of my work. [28] COMP2 I think that using the system ts well with the way I like to work.

COMP4
The use of TSS implies major changes in my clinical practice. [41,45,50] COMP5 The use of TSS promotes my clinical practice. [50] Subjective Norms The HCP's perception that most people who are important to him / her think he / she should or should not perform the behavior in question [73].

SN1
People who in uence my behavior think that I should use the system. [27,74] SN2 People who are important to me think that I should use the system. [27,74] SN3 Patients welcome me using TSS. [50]

SN4
Other colleagues welcome me using TSS. [50] Habit Degree of IT a nity and perceived competence according to the routine handling of IT [45,47].

HAB1
I have used telemedicine in the past. [45] HAB2 I already use telemedicine technologies. [45] HAB3 I feel comfortable with information and communication technology. [45,47] HAB4 I feel comfortable with using and consulting technology for my patients in my daily business. [45,47] Importance of Data Security Degree of perceived importance of security of patient data without loss of therapy quality, of ethical considerations and of transparency of information [45].

IOD1
Technical standards for the handling of patients' medical data are necessary. [40]

IOD2
Committing to standards for the handling of patients' medical data is necessary for my practice. [40]

IOD3
It is important to me to be able to inform my patients in detail about the use of their medical data. Idea collection

Importance of Standardization
Degree of perceived importance of conducting and recording the treatment in a standardized way without loss of quality [40].

STAN1
I am in favor of standardized documentation for medical practices. [40]

STAN2
In the case of standardized treatment processes, the administrative effort exceeds the medical bene ts. [40]

STAN3
Daily work in medical practice is too heterogeneous for important processes to be standardized. [40]

STAN4
Evidence basis The TSS needs to analyze data based on the evidence and to show that transparently.

Relationship to Patient
Degree of fear of losing contact and relationship with the patient because of changes in the means of communication and involvement of telemedicine experts.

REL1
Loss of patient I fear losing patients as a result of TSS, because they will not need me anymore.  Adapting variables in the context of the German healthcare environment Diseases like cardiovascular disorders (which take the biggest share of medical expenses and have the highest mortality) and pulmonary disease, as well as diabetes (which has the highest increase in prevalence) impact heavily on the German healthcare system [56,57]. Thus, a crucial in uence when choosing the relevant item dimensions to t the German context were the statements on telemedicine and digitalization made by the German diabetes association (DDG) [58], the German association of cardiology (DGK) [14], and the German association of pneumology (DGP) [59]. These position papers are addressed to the political situation and are considered in the legislative procedure. All the parties involved are working intensively on telemedicine topics.
Taking these statements into account, a call for clarity on the nancing of treatments was noted throughout [14,58,59,60]. For this reason, the ideas of Hwang et al. (2014) were extended by further items. It was assumed that the nancial evaluation of a digital service compared to an analog service (see item FIN3) is an in uencing factor (Ickarth 2018). Speci c to German healthcare is the question of including the service in the standard nancing of care (see item FIN6).
The payment scheme (see item FIN5) for patients is also unclear: the scheme needs either to be paid for privately (see item FIN7) or to be covered by reimbursement from insurance (see item FIN8) [61]. Lastly, the funding (see item FIN4) for the speci c cost of implementation in a medical facility (see item FIN9) could possibly be an in uencing factor [58]. One other eld for action mentioned in those position papers is the importance of developing digital standards to ensure high-quality patient care [14,58,59,60]. The items concerning the importance of standardization were oriented using the ideas of Dünnebeil et al. [40]. One item regarding the development of evidence-based standards (see item STAN4) was added. Along with the requirements for technical standards and for transparency towards patients in relation to data handling, the mandatory adherence to the principles of data management was consistently mentioned. In this context two other items were added to the section on the importance of data security, to cover assurance of the treatment quality (see item IOD3) in the digital environment and the responsibility of associations to discuss the legal, ethical and scienti cal framework for the data (see item IOD4) in such service systems [58].
Adding new variables in the context of the interactive telemedicine structure When it comes to an interactive system, with the patient and the healthcare professionals in the telemedicine center as additional players, fewer studies have concentrated on this new mode of communication, which might be a determinant of the behavioral intention of a physician. On the one side, the change in communication might mean that the relationship to the patient changes and the responsibilities for the treatment decisions and pathway need to be adapted.
This becomes important because patients become better informed and the physician-patient relationship changes accordingly [58]. On the other side, new methods of communication with medical colleagues in the telemedicine center can be seen. Looking into the literature, aspects of this digital collaboration emerge that are similar to those for eConsult platforms (exchange of knowledge about patient cases between HCPs via a digital platform). These aspects were considered for the development of this dimension [62,63]. When dealing with interactions between people, trust is an important form of social capital [64,65,66]. Consequently, two new constructs -"relationship to patient" and "trust in the telemedicine center" -were built. It is assumed that the perceived change in the patient relationship is determined either by the fear of losing the patient and losing control of the clinical treatment pathway (see items REL1 and REL3) because of the involvement of another healthcare professional at the center (see item REL2), or by the hope of strengthening the patient relationship (see item REL4) because of increasing contact points in the mixed analog and digital communication (see item REL5). The quick monitoring of inaccessible (see item REL7) and immobile patients (see item REL8) can be seen as an advantage for the patient-physician relationship. The goal would be to increase the treatment adherence (compliance) of a patient (see item REL6) [58,62].
Trust in the telemedicine center is assumed to be regulated by general trust (see item TRU1), following the model originally proposed by Yamagashi and Yamagashi [67]. Trust can also be created by certi cation (see item TRU2) and standard processes (see item TRU8), as well as by xed and stable personal contacts (see item TRU4) with the HCPs involved (see item TRU5), which can be described as institutional trust (see item TRU3) [65]. Also, distrust can appear when there is the fear of losing control of the treatment (see item TRU6) or of not getting su cient information when passing on responsibility to colleagues (see item TRU7) [62,63].
Thus, for the completely new variables TRU and REL, more than eight items were developed in order to produce a set that could undergo item revision after the eld testing. The operationalization was framed by the wordings from the original TAMs that had already been established and validated, and from the most commonly used scales in the reviewed studies, to pick out formulations for the medical context [68]. A minimum of four items were de ned as the indicators for each construct. The construct for attitude was therefore extended to cover the aspects of the generic attitudes to general telemedicine (see item ATT3) and digitalization (see item ATT4).
The corresponding statements were then translated into German (see Additional File 1). The eld test assessed possible distortions or di culties in comprehension, and the German wording for BI, PEOU, PU and SN was aligned with the corresponding proposals [69]. The participants were asked to rate the statements on a 5-point Likert scale from strongly agree (5) to strongly disagree (1), as in the original TAM approach [24]. Before the respondents were asked to undertake the rating and were given instructions for the questionnaire, a TSS was described in general terms to give them an idea of the functionality and method of interaction with such a TSS system and service offering (see Additional File 2). At the end of the survey the respondents were asked for personal data.  Cronbach's Alpha was measured to test the reliability of each construct and its internal consistency. The goal was to end up with between four and six items for each construct with an acceptable Cronbach's Alpha above 0.7. Items were eliminated to achieve a reliability improvement higher than 0.5. In total, 40 datasets were included. The age distribution of the participants was between 21 and 63, with an average age of 40 years. 35% of the participants were male, 63% were women and one participant answered other. Most of the respondents (75%) worked in an employment relationship, with 22 working in a doctor's o ce, 17 in a clinic and one in another institution. The healthcare professionals worked in different medical disciplines: most in physiotherapy or psychology, and some in general medicine with a focus on nutrition/diabetes. Table 2 shows the reliability analysis of the constructs. All the constructs except STAN showed a high reliability after necessary amendments for COMP, FIN, REL, and TRU. After removing particular items, the total reliability shows a Cronbach's Alpha of 0.914. As a consequence of the item revision analysis, the construct of the importance of standardization was removed, because the internal consistency was lower than 0.7. A separate analysis for the more experienced participants (those aged above 30 years) also showed only a slightly higher reliability of 0.386.

Discussion
The di culty in the implementation of telemedicine is the consideration of the complexity of all mandatory settings. All the named burdens and drivers may in uence the acceptance of this possible user group. Many studies have investigated patient acceptance [70], which seems insu cient on its own for this system interaction, as only a little is known about HCPs' acceptance. In this context, the treating healthcare professional is the initiator of the treatment and has the main responsibility for the medical care. Thus, it is critical to analyze the attitude of the treating HCP, who is the gatekeeper to the TSS [71]. Therefore, a glance at the technology acceptance research was needed to identify the relevant determinants for acceptance. The literature review shows the constructs of the external variables, covering the technical, organizational, social, legal, and individual context, that have been frequently supported, but only one study focused on the German healthcare market [40]. This made it necessary to adapt the constructs to the regional conditions in terms of the nancial and legal aspects. Additionally, the literature review identi ed a gap in the consideration of the aspect of the interaction and cooperation with telemedicine centers and the active involvement of the patient, so that two new constructs, "trust in the telemedicine center" and "relationship to patient", were necessary to map out these structures. The test analysis con rmed the robust quality of ten acceptance constructs and a high reliability in total. Nevertheless, the proof of reliability needs to be replicated in further investigations and the variable STAN needs to be explored again. Several test participants mentioned in the free text option that the acceptance also depends on the opinion and ability of the patients. There needs to be an investigation into whether the attitude of the patients is also an important factor affecting the behavioral intention to use a TSS.
Since the focus on the healthcare professional and the required interactivity was missing from the literature, as was a German setting, this should be a special point in future research.

Limitations
There are several restrictions to the generalization of the ndings. The literature review shows a high heterogeneity in studies of healthcare settings and disciplines regarding data analysis as well as the medical and regional focus. Nationwide studies focusing on single settings and areas need to be conducted to check on possible commonalities and differences. The review was limited to PubMed, so this research could only validate the reliability of the constructs for this selected variable set, but it drew a rst picture of the relevant factors. Different data collection techniques and exploratory approaches to sharpen the factors are recommended. The new constructs have not yet been tested regarding their possible relationships with other variables and their signi cance for a Further, the test only included a limited set of treating HCPs, who might also have been in uenced in their health area and profession by the Covid-19 situation, which made changes to communication with patients necessary.
The test described an example TSS which leads to a prognosis issue; subsequent research on the dynamic time measurement of technology acceptance is needed [19,54].

Conclusion
Understanding the complexity of the factors in uencing the acceptance by HCPs of the German telemonitoring market is crucial. This study developed a conceptual framework by identifying and designing multifaceted variables and operationalizing them in the validated questionnaire. This delivers a robust foundation that can be transferred into speci c medical disciplines and professions in which telemonitoring systems with an involvement of telemedicine centers are considered. Furthermore, the factors of "relationship to patient" and "trust in telemedicine center" are a new basis for capturing the interaction of such TSS infrastructures. The adapted and new constructs were optimized through the testing and reached an acceptable reliability for further research.
However, the results need to be combined with the attitudinal perspective of patients to pick up all relevant players in the system.
There is a great need for further research on German HCP acceptance in telemedicine, to provide evidence to identify the barriers and the potential, which will lead to a statement of the requirements of technology, legislation and nancing of telemedicine structures in Germany for providers and policy, especially from the professional society view. This understanding is essential to develop the appropriate conditions for such complex programs and to close the existing gaps.
Last but not least, the Covid-19 crisis and the rst reaction to ease telehealth nancing in Germany [72] shows the urgency and relevance of this topic.