Implementing Electronic Health Records in Germany: Lessons (Yet to Be) Learned

Integrated care, integrated data

Health data integration is an essential part of almost any integrated care project today, as it has become the cornerstone of virtually every comprehensive attempt to establish an integrated health system [1]. Aiming at improving the quality and safety of health services, integrated care can be best described as an approach to overcome healthcare fragmentations [2]. In this line of thought, the systems that enable a continuum of care should include data structures that can link patients’ health information over time [3]. Interlinking the relevant parts of patient data and supporting the continuity of medical information through shared records enable delivering the right treatment at the right time [4]. The accumulation of information across time, location and sectors creates an uninterrupted patient journey, contributing to a more integrated health system and saving costs at a similar or better quality of care [3, 5].

Indeed, ensuring access to the right information at the right time improves the safety, effectiveness and efficiency of care [6]. In highly fragmented health systems, information transfer between in-patient and out-patient care sectors is vulnerable to leaks and failures at their interfaces; a vulnerability that can be overcome by integrating and securing the flow of health data [7]. By connecting relevant data across sectors, not only clinical services but also public health measures can be optimised. Using the available, yet administratively isolated or dispersed sets of data in a meaningful way can provide relevant insights for more effective treatment programmes and prevention measures at the population level, e.g. when monitoring disease outbreaks. For this reason, increasingly more health systems are taking advantage of the benefits of digitalisation and integrate data linkage into their public health surveillance activities [8].

A systematic and detailed collection of patient records, commonly known as electronic health records (EHRs), forms the core of the information system architecture in integrated health systems [9]. An EHR is a digital, longitudinal patient record that is available to healthcare professionals across a range of settings [10]. By enabling the flow of information between healthcare professionals, EHRs help improve the coordination of care, avoid complications and unnecessary hospitalisations, and increase cost-effectiveness [11], alongside contributing to guiding the clinical decisions of healthcare providers regarding diagnosis, management and treatment [12]. With the help of EHRs, patients’ up-to-date medication history and past interventions can be made available to healthcare professionals, enabling them to be informed about the prescriptions issued by their colleagues and recognise preventable adverse interactions and mortality risks [13, 14].

According to the World Health Organization (WHO), EHRs play a vital role in supporting the diagnosis and treatment of patients through the provision of rapid, comprehensive and timely clinical information at the point of care [15]. Indeed, using EHRs is deemed even a prerequisite for high-quality care, as it is impractical and unrealistic to expect that patients can be the primary source of their complete and accurate health history for each new provider at each new point of contact [3]. This is even more so for emergency patients, elderly with specific diseases such as dementia, and those with complex care needs. Multi-morbid patients are typically treated by different medical specialists in varying healthcare facilities, and it is particularly for this reason essential to apply an integrated approach to treat them successfully [16]. EHRs support healthcare professionals in providing the right treatment at the right time by capturing patients’ health history and information about past interventions, enabling a better coordination between specialised medical professions [3, 14]. With the availability of data on patient history across sectors, the quantity of medical interventions decreases whereas their quality should increase [17].

Apart from enabling better coordinated health service provision, EHRs are a valuable source of data that can improve the quality of health services and public health interventions [18]. Indeed, from a societal perspective a greater long-term benefit of EHRs is the information they can provide for population health research, and for planning, executing and evaluating policy interventions (secondary use or re-use of health data) [4]. Through increased data quality (higher accuracy, completeness, consistency, uniqueness, timeliness and reliability) and improved access to relevant information, EHR systems can contribute to better health services and policies. Combined with other datasets, aggregated clinical data from EHRs can build the basis for evidence-informed health policies that benefit the society as a whole [19]; understanding the determinants of ill-health, planning of health workforce and efficiently monitoring of disease outbreaks, to name but a few examples. In this way, robust Health Information Systems (HIS) can be established at the national level for the benefit of (healthy) citizens and patients [20]. With the help of HIS, decision-makers can monitor, evaluate and improve health system performance, and make timely, evidence-informed decisions [21].

Contextual factors affecting the use and re-use of health data in Germany

When it comes to the integration of health services, Germany ranks at the bottom in international comparisons [22]. This is mainly because the sectors concerned with providing health services in Germany are financed, organised and governed in isolated administrative silos, providing minimal incentives for vertical integration, i.e. better coordination between primary and secondary care levels [23, 24, 25]. This fragmented and uncoordinated health service provision leads to inefficiencies, diminished quality of care and high rates of patient admission [25]. In accordance with the disconnected health services working in silos, also health data are scattered across different administrative levels in Germany, exacerbating the pitfalls of its fragmented system [23]. Isolated datasets make a realistic assessment of the health system performance difficult and create a vicious cycle by hampering any improvement or optimisation of the health system and limiting the potential effectiveness of efforts to move towards integrated care [3, 26].

Against the background discussed above, Germany could have (and considering the drawbacks of its highly fragmented health system, should have) started working towards integrated, person-centred and preventive care with the help of EHRs. Yet for more than a decade, the implementation of a cross-sectoral, nationwide EHR system has been stagnating within this fragmented field. Offering a scheme for (health) policy analysis, Leichter [27] orders the factors influencing public policies into four different areas in his work that has been seminal in the health policy literature [28]: i) situational factors refer to impermanent events such as political reform or change of the government; ii) environmental factors are those that exist outside the boundaries of a political system such as international agreements, obligations and pressures; iii) structural factors, in contrast, are concerned with the unchanging elements of the society and polity such as political structure that covers e.g. form of government, and the number, strength and legitimacy of interest groups; and lastly iv) cultural factors encompass, among others, national heritage and traditional social values. As will be discussed below in detail, whereas the situational and environmental factors have pushed the agenda forward for integrated care and health data integration over the past few years (national health policies and international commitments such as European Health Data Space; EHDS), structural and cultural factors in Germany build obstacles in the primary and secondary use of ePA data.

To start with, one of the major structural issues impeding the successful implementation of ePA has been the role of the Federal Ministry of Health (MoH) during this process. Traditionally, the financing of Statutory Health Insurance (SHI) in Germany is decoupled from the federal budget, and government control measures in the health system are only indirectly possible [29]. Although recent developments have redrawn the political boundaries between the government and self-governance with increasing state intervention (especially since the early 2000s the regulatory interventions from MoH have considerably increased), the role of the federal government in health policies is mainly concerned with providing framework conditions and laws [30]. Most notably, while decisions on in-patient and out-patient care are taken at the federal level, the government itself is by law not involved in these negotiations [23]. The right to make decisions about medical care and the regulation of essential performance parameters on quality and efficiency lies with the actors of self-governing bodies of the SHI system; namely the sickness funds and representative organisations of healthcare providers at the Federal Joint Committee, in which power is concentrated. Similarly, the initial governance structure of gematik, the operating company for telematics infrastructure for the German health system, had been composed of the self-governing bodies of healthcare providers and sickness funds with equal ownership and voting rights at the time of its foundation in 2005 until its reorganisation in 2019. The principle of self-governance, combined with the federal government’s (limited) competency in the field of healthcare provision and the lack of an overarching framework to force the commercial healthcare industry to deliver the necessary telematics infrastructure resulted in a delayed ePA roll-out that could take place no earlier than 2021 [13].

Especially the self-governance of healthcare providers played a great role in this delay and represented their interests at the federal level by resisting progress in the EHRs project [31], blocking its implementation [32] and stating that they otherwise would refuse to provide health services [33]. The main reason for these actions can be sought in theories on coalitions and interest representation. Drawing on the findings of Alford from 1975 on ideological and interest group barriers to reform in healthcare politics, Döhler [29] argues that one of the interest coalition structures in healthcare is formed around market-shaped health policymaking; however, here ‘the market’ does not indicate the involvement of private industry but rather the lack of government involvement. This situation leaves the stage wide open for monopolists of the medical profession who aim to preserve their autonomy and fill the political vacuum, creating a challenging situation for other actors such as the public health administration, academia or health insurers. While these stakeholders may not necessarily oppose ‘the market’ dominated by the professional monopolists, they might endanger the autonomy of the medical profession that desires sovereignty and greater control over the healthcare sector. Hence, health policy decisions in Germany are mainly determined by the conflict between those two groups (monopolists of the medical profession vs. others), which results in blocking reform and maintaining the privileged status quo. Similarly, the stagnation of ePA roll-out was the reflection of a severe deadlock, as the main stakeholders of the self-governance model (medical associations and sickness funds) failed to reach an agreement at gematik for almost 15 years.

Implementation of EHRs

With the new coalition government in Germany since December 2021, the uptake of ePA may gather pace. As stated under the heading ‘Digitalisation in healthcare’ of the coalition agreement, the governing parties seek to accelerate implementation with a GDPR-compliant opt-out approach [56]. The experiences of neighbouring countries suggest that the opt-out model is a more straightforward way to tackle problems resulting from the systemic fragmentation of the German health system [57]. For instance, in France, only 20% of the population was linked to an EHR even 13 years after its introduction, which is the reason why France too decided to adopt an opt-out model from 2022 onwards [58]. A different example of an EHR introduction experience can be drawn from Austria; a country that adopted an opt-out policy from the outset. Today, in contrast to the French 97% of Austrians use EHRs, which include information about discharge letters, laboratory results, out-patient medications, e-prescriptions and medication allergies [18]. Indeed, given that a considerable number of countries in Europe have a higher level of digitalisation and data integration in their health systems than Germany [59], it might be helpful to focus on the ‘software’ (governance, stakeholder interests) instead of ‘hardware’ (financial affordability, technological advancement) components when discussing the adoption of ePA. In this context, the Austrian case is worth exploring, not only because of its preferred implementation approach (opt-out) but also because of its cultural and structural similarity to Germany.

In the beginning, similar to France and Germany, the idea of using EHRs led to some anxiety among Austrian physicians relating to the additional workload and cost burden [60], besides increased transparency, cost pressure and reduced confidentiality of patient data [31]. Criticism was raised especially by the Austrian Medical Chamber and data protection advocates [61]. In their comprehensive study comparing EHR deployment in Austria and Germany, Bogumil-Uçan and Klenk [61] offer two explanations for why those voices could not gain the upper hand in policy discussions. The first reason is the dominance of the Austrian hospital sector in a more integrated governance structure. Unlike Germany, healthcare provision in Austria is strongly focused on hospital care, making the conflict about fragmented vs. integrated care less pronounced. Through ambulatory pre- and post-hospital treatments in out-patient clinics, the in-patient and out-patient care sectors are better integrated in Austria than in Germany. Moreover, since many salaried physicians in Austrian hospitals also work as specialists in out-patient care, it is assumed that they were even in favour of the implementation of EHRs as a result of their experiences with digital health applications in hospitals. The authors’ second reason is that due to the pre-eminence of state actors in the hospital sector, the dominance of the in-patient sector in healthcare provision and the greater organisational integration of sickness funds, the main players in the Austrian health systems are the federal government, regional governments and the Social Insurance Association. They thus have a structural advantage over healthcare providers in influencing policy decisions, including the adoption of EHRs. Unlike in Germany, where the self-governing bodies of healthcare providers had veto power in gematik to stall the roll-out of ePA, healthcare providers in Austria were not even included in the decisions of ELGA Limited Liability Company, owned by the ‘main players’, all of them being in favour of a comprehensive, nationwide EHR system from the outset.

Although suggesting the same ‘main player’ structure for taking decisions on ePA would be overly ambitious, Germany could bring more stakeholders to gematik and tilt the balance of power in favour of the end users of ePA. For instance, the plans of the Coalition about accelerating the decisions of the self-governing bodies and strengthening patient representation with a reform of the Federal Joint Committee (stated under the heading ‘Patients’ rights’ in the coalition agreement) are praiseworthy and in line with international recommendations for designing policies to deliver people-centred health [62]. Yet, patient participation should be expanded to other relevant key organisations beyond the Federal Joint Committee, such as gematik, i.e. the new digital health agency. Patients, and in general insured persons, are ultimately the end users of crucial health innovations in the SHI system, including EHRs. It is important that patients, especially those that are multi-morbid and chronically ill, should be able to work with such technologies. Therefore, patient representation should be an inherent component in the overall strategy for digitalisation in healthcare. Indeed, giving decision-making power to patients and citizens can be an even more effective approach in digital health projects than a top-down governance intervention from MoH to self-governing bodies [63].

Secondary use of EHR data

Thus far in Germany, only billing data from sickness funds have been used for health services research, available only with a considerable time delay and financial resources [59, 64]. Starting from 2023, it will be possible for patients to make their ePA data available to an authorised data custodian, subject to their informed consent. Patients will be able to choose the scope of their ePA data to be shared with the authorised centre and have the option to limit the data to certain categories, datasets and specific areas of scientific research; these data are then to be pseudonymised and encrypted by the centre [65]. The research data centre of the Federal Institute for Drugs and Medical Devices (Forschungsdatenzentrum Gesundheit, Bundesinstitut für Arzneimittel und Medizinprodukte; BfArM) will serve as a research hub for ePA data for which patients in Germany grant access [47].

The national data-sharing policy to this end requires confirmation by insured persons to enable the use of their ePA data for research purposes (opt-in). Currently, claims data are being transferred automatically to the BfArM, and based on these data, the use of synthetic data (i.e. data artificially created through different algorithms that can mirror the statistical properties of the original data) is being tested to determine the optimal care through Artificial Intelligence (AI) solutions [66]. With an opt-in approach, the ePA data volume will be too low to create synthetic data and develop AI-supported solutions. Although good coverage across the entire population is essential in population health research to ensure that findings from the collected data are accurate, unbiased and representative, with an opt-in model, only the most engaged people who actively take steps to participate will be included in the data pool, restricting the chances of developing effective strategies for health at the population level [67].

Hence, considering the magnitude of data to be pooled and analysed, the opt-in requirement for the secondary use of health data will highly likely have a negative impact on the size and representativeness of the ePA data. A fictional example of this drawback is the following: in the case that 90% of the SHI-insured people in Germany choose to keep their ePA (opt-out) and 10% of those give consent for the re-use of their data for population health research (opt-in), the ePA data available for research will comprise only 9% of the SHI-insured population. Thus, the data collected may not represent the whole population accurately, making it challenging to develop evidence-based health policies at scale and draw lessons for improved clinical and prevention services. Similarly, although the right to delete ePA data may appear at a first glance to contribute to patient empowerment, it is not a good long-term strategy given the potential loss of valuable data that could be used in their clinical treatments or population health research. An integrated, person-centred and preventive health system should give equal importance to individual privacy and public health. Hence, a better approach to re-using ePA data could be to adopt an opt-out policy where patients can share their (pseudonymised) data for research unless they object, in line with the GDPR. Based on Art. 9(2) GDPR, EHR data in member states can even be collected for research purposes without any consent requirement and objection right as a standard policy when certain requirements are met [18].

However, an automated data transfer for the secondary use of ePA data might be politically difficult to introduce in Germany, a country where data security concerns are significant [68]. Indeed, the recommendations of the upper house (Bundesrat) to the Federal Government concerning Germany’s positioning on EHDS from September 2022, which included disapproval of some parts of the secondary use of health data in EHDS, indicate a rather cautious approach to data transfer [69]. In Germany, national data protection regulations are highly developed, hindering cross-sectoral data exchange and integrated care. For instance, there is no unique personal identification number that could allow cross-sectorial linkages in the German social security system, and a linkage between different datasets that contain personal information is only permitted with the prior consent of the person concerned [18]. Part of the reason lies in the culture and motives that are rooted in history. Data collection on and surveillance of citizens during the Nazi era and later the Soviet era left the country with deep scepticism of any action that seeks to gather personal data [70]. In the aftermath, the governmental system was rebuilt to ensure that numerous people and institutions have access to a wide variety of information, but no single actor to all information about one person [71]. Understandably, most Germans today show conservative and protectionist behaviour when it comes to automatic data collection or transfer.

On the other hand, developing health policies without taking advantage of the insights that ePA data might offer will highly likely exacerbate the fragmentation issue that Germany is facing. Working through the variety and complexity of social, environmental and individual factors that can shed light on the causes of diseases might require too much time on the part of patients, especially if they belong to older age groups, are multi-morbid and have lower levels of (digital) health literacy. In an opt-in policy for the re-use of ePA data, patients should actively seek information, first, to decide whether to give consent to share their data and, second, to choose which datasets could eventually serve the purpose. An opt-out approach with necessary information provided to patients would simplify this process and ensure a nearly population-level data collection, respecting at the same time informational self-determination of citizens, which is a constitutional fundamental right in Germany [12]. This step would be especially crucial if the interlinkages and causal connections for preventive measures could be identified not with human intellect but with AI through large amounts of information.

Still, in the case that the current opt-in policy for the secondary use of ePA data remains unchanged, it should be accompanied by several measures to increase the number of people sharing their data. Culturally embedded concerns regarding data privacy are understandable and might be difficult to overcome in Germany due to damaging experiences in its history. To convince the public of the benefits of sharing ePA data, massive educational and information campaigns will be necessary. In this context, the role of the mainstream media should not be underestimated. At best, journalists become partners in informing the public regarding the benefits (and risks) of the secondary use of health data. At worst, they take the advantage of the public’s underlying fears about data privacy in order to attract viewers or readers. Bold political moves that could challenge the entrenched national reflex to protect data might risk disapproval from the mass media that has the power to amplify anxiety, as was the case in France during the EHR deployment process [72]. Under these circumstances, the strategic organisation of federal-level information campaigns will be crucial, not despite but because of the risk of being perceived as negligent regarding citizens’ data privacy concerns.

For the secondary use of health data, if organising educational and information campaigns is one side of the coin, ensuring a high level of information literacy and trust of citizens on health matters is the other. Information literacy can be described as individuals’ ability to find, retrieve, analyse and use relevant information. Health literacy, as a specific field of information literacy, focuses on health sciences and health communication. It implies the achievement of a certain level of knowledge, personal skills and confidence to take action to improve personal and public health [73]. According to the statistics of a comprehensive country comparison report from Europe, only 28% of the population in Germany has sufficient or excellent health literacy; the lowest among the 17 European countries that participated in the comparative study and 27 percentage points below average [74]. Germany has been investing in improving the health literacy of the population through Nationaler Aktionsplan Gesundheitskompetenz, a project funded by the Robert Bosch Foundation.

At the intersection of information literacy, digital literacy and health literacy, a new term has been coined in the past years: digital health literacy. At a minimum, digital health literacy skills require individuals’ basic reading and writing proficiencies, working knowledge of using computers, and an understanding of how, why and when online health information is created, shared and received [75]. In line with the increasing trend toward the availability and use of digital health solutions across Europe, such as EHRs, there is a growing demand to improve the digital health literacy of populations [74]. The study mentioned above [74] also examined the digital health literacy levels of selected European countries. It conceptualises digital health literacy as the ability to search for, access, understand, appraise, validate and apply online health information and the ability to formulate and express questions, opinions, thoughts, or feelings when using digital devices. Similar to the health literacy outcomes, the digital health literacy level in Germany has been found to be the lowest among the 17 countries studied; 58% of the population in Germany report difficulties in digital health literacy, 20 percentage points below average [74].

Digital health literacy has a great impact on another essential element for the use of digital health services: trust. Research shows that digital health literacy is positively associated with greater perceived trust in health information from online health communication channels and information sources [76]. Indeed, (digital) health literacy is fundamental to developing societal trust in digital health services and data-driven healthcare [77]. In light of cultural differences between Germany and other European countries in terms of citizens’ trust in data sharing, scholars from Germany identify promoting digital health literacy as the key enabling policy action for integrating digital components into the German health system [78]. Within the concept of EHRs, a patient is an active partner who should access, add and manage health-related data, resulting in increased quality of care as well as the patient’s compliance [31]; hence, ensuring digital health literacy sets the foundation for patients’ participation, empowerment and self-determination in the German ePA [78].

Lastly, evidence also shows that in Germany a simple and safe interface design can be even more important for increasing usage of ePA than knowledge about their actual benefit or added value, which highlights the necessity to focus on the user-friendliness aspect of ePA data sharing options [78]. Due to a number of reasons such as i) the low level of information among the population regarding EHRs; ii) the culturally rooted concerns about health data security; and iii) a general disinterest in new and unfamiliar technologies, having a user-friendly ePA will be central for patients to manage their data [79]. For this reason, sharing and managing ePA data should be as simple as possible for the end users. Patients should be given the opportunity to simulate cases for managing their ePA data and become familiar with the available dataset options in online platforms through non-binding consultations. This aspect is especially crucial if they must opt in for the secondary use of their health data.