>> One of these gaps is the definition of what “research data” is and, in particular, whether routine data from the statutory health insurance system is such. Clarifying this is now the task of other laws, such as Book 5 of the Social Security Code (SGB V), the Research Data Act, the Health Data Use Act and, at European level, the European Health Data Space.
For health care research and everyone who works in or for the health system, the question of whether access to routine data from statutory health insurance is simplified and clearly regulated is of paramount importance. To put it clearly with Helmut Kohl: “What matters is what comes out at the back.” Outlining the long road to unlocking the potential of this data, the current dead ends in politics, society's ingenuity in developing solutions, the unsolved problems and potential ways out - that is the subject of this article, which can only touch on it and is the starting point for a series of in-depth publications.
Providing medical care to a population is much more than the application of medicine. No matter how advanced medicine is, whether and how it is received by a population and what benefits it brings can vary greatly. “A lot helps a lot” does not automatically apply here. Lichtner and Plant showed this at the beginning of the 1970s with regard to appendectomy (Lichtner and Plant 1971) and also that the frequent performance of this operation may also be responsible for the increased mortality from appendicitis. This publication, which was groundbreaking at the time, was the impetus to search the reality of care for deviations that can no longer be explained epidemiologically (Neuhaus et al.
1985), and to raise the question of (social) medical rationality and economic efficiency: “Do the insured get the services they need? And do they need the services they get?” (Swart and Ihle 2005).
The first attempts to use routine data date back to the 1970s, when the local health insurance companies in Bavaria began to transfer their paper billing documents to digital data carriers (Debold 1976). In the 1980s, the Federal Ministry of Labor, which was responsible for health insurance at the time, supported model experiments in which several different health insurance companies at various locations in West Germany were involved (Meye and Schwartz 1984). In the context of these projects, work was created that dealt with the measurement and comparison of service quantities, but also with analyzes of care that dealt with questions of quality and evidence-based care (Federal Ministry of Labor and Social Affairs 1990). As a result of the projects and model experiments of the 1970s and 1980s, a lasting connection emerged with the new term “healthcare research” (Häussler 1990).
The fact that research activities with routine data died down after reunification until the mid-2000s is certainly due to the fact that suitable data sets from health insurance companies were practically not available until then and that only a few researchers asked for them. This may be due to the fact that reunification also shaped the discussion in the healthcare sector. In addition, topics such as “quality assurance”, “disease management” and “evidence-based medicine” filled the intellectual space of the healthcare reform debate between 1990 and 2005.
A new dynamic for the use of routine data for health services research only emerged when a smaller health insurance company entered into a cooperation with an academic institution. This was done against the background of the GKV Health Modernization Act, which came into force in 2004, which with the new paragraphs 303a to 303f SGB V on “data transparency” at least signaled the will of the legislature that transparency through GKV data was desired. A further driver was the development of “morbidity-oriented risk structure compensation” (Morbi-RSA) (Reschke et al. 2005). During this time, the first edition of the manual for secondary data analysis “Routine Data in Health Care” (Swart and Ihle 2005), which provided the first systematic methodological presentation, was published. The “German Network for Health Services Research” was founded in 2006 and in 2010 the German Medical Association also emphasized its interest in routine data (Swart et al. 2010).
It was not until the mid-2010s that the use of routine data (claims data), as measured by relevant publications, increased sharply (see Fig. 1), after smaller increases in relevant publication activity from 2005 and 2010 were to be observed. The frequency of publications on “healthcare research” was significantly ahead of the increase in publications with “claims data”, which indicates that the statutory health insurance system does not meet the need for access to routine data, which has been known since the 1980s covered.
How is it that it was practically only after 2015 that strong publication activity on research work with routine statutory health insurance data began? Since everything that concerns the statutory health insurance is known to be highly regulated by SGB V, and because the issue of “data protection” makes this even more serious, it is necessary to take a look at the development of the relevant regulation.
In the “Health Insurance Cost Containment Act” in 1977, Section 223 was included in the Reich Insurance Regulations (RVO), according to which a “health insurance company (…) in appropriate cases, in cooperation with the statutory health insurance associations, the hospital authorities for the respective area and the medical examiners, primarily covers cases of illness with regard to the services used” and “can inform the insured person and the treating doctor about the services used and their costs” (Federal Ministry of Labor and Social Affairs 1988, p. 56). On this basis, the model tests already mentioned were carried out, the results of which resulted in the Health Reform Act of 1988, in which the RVO of 1912 was transferred to SGB V. Essentially, it was about comprehensive regulations on billing and performance audits and the necessary data documentation, but not about topics that can be assigned to the context of “healthcare research” but which were discussed in the projects.
It took until 2004 when the idea of data transparency was given a legal basis with the “Health Modernization Act” in paragraphs 303a to 303f SGB V, which now includes, among other things, “institutions of health care research” among those who are allowed to use statutory health insurance routine data. “to the extent that they are necessary for the fulfillment of their tasks” (Federal Ministry of Health 2004). Publications based on this data remained rare in the following years. The central associations of health insurance companies that still existed at the time and the National Association of Statutory Health Insurance Physicians were responsible.
This did not go unnoticed by the Federal Ministry of Health (BMG), so Federal Minister of Health Daniel Bahr issued the first “Data Transparency Ordinance” (DaTraV) in 2012 (Federal Ministry of Health 2012).
The main idea behind it was to name the “German Institute for Medical Documentation and Information” (DIMDI), an institution that has a more neutral attitude towards the task of creating data transparency than the representation of practicing doctors, whose aversions to transparency The supply process did not predestine them for a positive attitude right from the start2. Ultimately, the health insurance side also had no intrinsic interest, as the health insurance companies still often take the view that GKV routine data belongs only to them and should only be used by them.
In practice, this idea has not borne fruit. The great inconvenience of the practical implementation has ensured that so far only two publications have been documented that state that the routine data according to DaTraV have been made available.
The BMG probably also saw it that way and dissolved the DIMDI (for whatever reasons) as an independent authority and incorporated it into the BfArM. In addition, Federal Health Minister Jens Spahn had a completely revised version of the DaTraV drawn up in 2020, which came into force on July 10, 2020. Significant change: “The Robert Koch Institute carries out the tasks of the trust in accordance with Section 303c of the Fifth Book of the Social Code. (…) The tasks of the research data center in accordance with Section 303d of Book Five of the Social Security Code are carried out by the Federal Institute for Drugs and Medical Devices.” (Federal Ministry of Health 2020, p. 1). The responsibilities have now migrated to the third pair of responsible authorities, this time to authorities that are subordinate to the BMG. The BfArM will receive a research data center (FDZ), which will probably be built in part from the remains of the DIMDI, which was awarded to the BfArM.
It has not yet been proven whether the new constellation regulates the availability of routine data better than the “solutions” of the last 20 years. In any case, it is not yet clear that the processes are running at the desired speed. The FDZ website at the beginning of November 2022, more than two years after its legal creation: “The new health research data center is under construction. (…) For the first time in calendar year 2024 (sic!), the entire data set specified in the law will be transmitted with the reporting year 2023 to be delivered. (…) Faster application processing: This is ensured, among other things, by an electronic application process. Applications can probably be submitted online in 2023.” (Federal Institute for Drugs and Medical Devices (BfArM) 2022). It will therefore take at least four years from the entry into force of the reformed Data Transparency Ordinance in 2020 to its first use.
This means we are facing the fact that no homogenized data set that represents the entire statutory health insurance fund will be available at least until 2024. If we go back to the beginning of these analyzes (see above), almost 50 years will have passed before this central research infrastructure will be available. Legislature has been working on this issue for twenty years without any visible success.
Against this background, one is astonished to observe the steep increase in relevant publications (see Fig. 1).
This development can be easily explained by the fact that the need for the use of GKV routine data was so high and that responsible decision-makers in the GKV use their resources and share them with other interested parties.
From the analysis of the publication data already mentioned, conclusions can be drawn about the data providers involved, the research companies involved in the analyzes and whether studies and publications are also carried out on behalf of or with the participation of industrial companies.
Overall, a differentiated picture emerges of the different constellations in which routine data projects take place, although projects that are only carried out with a subset of GKV data are not considered here (e.g. statutory medical billing data in accordance with Section 295 SGB V or service data from the Hospitals according to §21 KHEntgG). Qualitative aspects of the analysis are shown in Table 1, quantitative ones in Figure 2. The results in brief:
• In all constellations considered, data is used by health insurance companies. The spectrum ranges from large health insurance companies that carry out their own projects or provide them in cooperation, to providers who can directly access homogenized and anonymized data sets from various insurance companies.
• Individual large and medium-sized health insurance companies have dedicated departments or even institutes that specialize in such analyses.
• Research service providers are often involved, most of whom have extensive experience.
• The options for industrial companies are more or less limited depending on the data source. The need for data for studies with “real world evidence” is certainly greater than the supply.
• In some cases, analyzes can be paid for, which often means that they can be processed quickly and in a predictable manner.
A quantitative representation of the temporal development of the different constellations is only possible for some of the publications and is only intended to provide insights into the temporal dynamics. Project-by-project access to projects from the Innovation Fund is hidden here. It can be seen that the relative importance
• large and medium-sized health insurance companies are decreasing as a data source (1. and 2.) and
• the number of providers that have more direct access to a data pool is increasing (3 and 4).
It can be assumed that only the possibility of remuneration for analyzes creates the conditions for building up and maintaining appropriate capacities and thus being able to serve increasing demand.
The self-help obviously worked quite well, especially because many questions can be answered without a data set that includes all those insured by the GKV. Of course, this would be more meaningful because questions such as: B. result from cash changes.
Nevertheless, there are numerous shortcomings in the status quo for which a solution at the overall statutory health insurance level would be attractive:
• Insufficient capacity leads to waiting times.
• Too little price competition between data pools in constellations 3 and 4.
• Potential unequal treatment of requests depending on institutional background or research questions.
It is worth taking a look at two large neighboring European countries with different health systems but comparable data structures: France's national health insurance (Caisse Nationale de l'Assurance Maladie - CNAM) and the National Health Service (NHS) in England. Both the CNAM and the NHS provide very detailed data3 that depicts inpatient and parts of outpatient care on an individual, pseudonymized patient basis.4
In both countries, there is an opportunity for public and private organizations to purchase licenses that enable analysis using these datasets. In England, a total of 337 licenses limited to one year were awarded in 2021, including around a third to the administrations of administrative regional units and around ten percent to private individuals, including around fifteen consulting companies, which primarily provide analyzes to pharmaceutical and medical technology manufacturers .
The acquisition of the licenses is subject to high requirements, on-site checks are carried out and the purposes that may be pursued with the analyzes or are excluded are defined. Licensees must maintain external committees that review the purposes for which the data is used. The licensees are published, with their measures and instruments to secure data and prevent misuse being published (NHS Digital 2022).
Against this background, it is not surprising that a good 1,900 publications can be identified in England between 2005 and 2021 that were carried out with routine data (HES), more than three times more than the 596 studies with routine data that come from Germany. The licensing model, which is also used in France, should therefore be considered as a serious variant for further development in Germany. Ultimately, the constellations documented in Section 4 (particularly Groups 3 and 4) have functional equivalents already installed, so that we do not have to start from scratch within Germany.
Against this background, we look at the political projects announced by the current government. Its most important elements are the “Research Data Act” (FDG), the “European Health Data Space” (EHDS) and the “Health Data Usage Act” (GDNG).
What is the potential and status of these initiatives?
• The FDG is under the responsibility of the Federal Ministry of Education and Research (BMBF) and is particularly important for the university academic area. It is intended to make the vast number of data sets accessible for further uses, regardless of whether it is public or private data. To date, there are no publicly available documents from the BMBF that provide an idea of the status of the legislation. In principle, however, a high potential for conflict must be expected because the goal of creating economic value with data is not compatible with the fact that this data can be released at the same time.
• The GDNG is also mentioned in the coalition agreement and is being prepared under the responsibility of the BMG. It is intended to provide connectivity to the EHDS and, in particular, to make electronic patient records (ePA) usable as data sources across European borders. Furthermore, it should also enable routine data to be merged with register data using uniform identification numbers. There are no documents available on this yet either.
• The EHDS is a European framework initiative similar to the General Data Protection Regulation (European Commission 2022). This sets guidelines for the member states and is intended to structure national legislation, such as the GDNG or the FDG. The Commission has already presented a first draft law and wants to get it underway by 2025.
Realistically, one must currently assume that the use of routine data for health research depends primarily on the implementation of the second version of the Data Transparency Regulation of 2020. The reform initiatives mentioned can make a “top” contribution to this. Nevertheless, the main wish at the moment is that the elementary data transparency in SGB V becomes a reality after 50 years. The time perspective is therefore 2024. If it were possible to set up a European-style licensing model in parallel, a big step would be achieved in the long term.
Having taken part in the decades-long procession along this Via Dolorosa, one cannot help but feel a certain bitterness. The state and legislature were certainly not good leaders. So far you don't have much more to show than a ruined building in the SGB V. If it hadn't been made makeshiftly livable by many inventive people, we would be even more embarrassed in front of ourselves and our neighbors when it comes to the use of digital in public infrastructure. The state must now deliver and, as a result, bring itself to allow the creativity of institutions and companies. <<
by: Prof. Dr. Bertram Häussler, IGES Institute
Data is the raw material of science because researchers can use it to test and further develop their ideas about reality. In a present where the availability of information has increased unimaginably, the raw material is mutating into a treasure: “Data is the new gold” (Lenzen 2018). In the age of “evidence-based” economic success also depends on data because you can no longer sell anything that cannot be proven to be useful.
“We want to use the unused potential that lies in numerous research data more effectively for innovative ideas. We want to comprehensively improve and simplify access to research data for public and private research with a research data law and are introducing research clauses.”
(SPD, ALLIANCE 90/THE GREENS and FDP 2021)
The traffic light coalition has made data one of the objects of their policy for good reason. They should be used and access to them should be improved.
