>> Professor Pfaff, in the cover interview of the last issue of
“Monitor Health Services Research”, their two colleagues Braithwaite and Mannion have found very clear words for what they expect from health services research as a science internationally and also in Germany. Mannion also accuses healthcare research – especially that from Germany – of being undertheorized. Aren't you a little embarrassed?
No, Mannion is right. For this reason, I have invited him as a keynote speaker at the upcoming 20th German Congress for Health Services Research. It is perhaps embarrassing that from an international perspective Germany has a strong theoretical tradition, but this is hardly visible in health services research. But we are not alone here, because evidence-based medicine also works quite well without theory.

Like that?
Every science has its methods. The methods of healthcare research are usually based on the methods of empirical social research and inferential statistics. With this toolbox of methods you can get pretty far. This is why health care researchers are often referred to as “methodologists” by medical professionals.

Where is the problem?
You can test theses using good methods, but the problem is: these theses may or may not be derived from theories. The latter is the case when practitioners – such as clinicians – have a good care idea and want to have this idea evaluated. At the beginning of health care research, using theories as a starting point for research was not absolutely necessary in order to achieve the necessary advances in knowledge.

The statement that evidence-based medicine is theoryless will certainly not be easily accepted by many people.
Evidence-based medicine is not theoryless per se; it can also be practiced on the basis of theories. The fact is that the methods of evidence-based medicine can also be used without having to have any theory. It is enough to have an assumption that is then tested in a well-conducted randomized controlled trial. Evidence-based medicine is a very good method, nothing more. Their high methodological quality almost tempts you to give little thought to the complex theoretical connections. You simply have to face this truth, as bitter as it may be for some.
Are you perhaps making it too easy?
The causal relationships that are usually tested in randomized, controlled studies are relatively simple in nature - such as: “A leads to B”. In health services research, the situation is often more difficult because we are almost always dealing with complex interventions in complex and less constant environments. In order to understand all this and then intervene well, we need more theoretical knowledge about these connections in health services research.

Because instead of exact knowledge, a mere guess is enough to – to put it somewhat disrespectfully – research “without it”?
Certainly some practitioners approach the matter in a more casual way, along the lines of: “That’s a good idea, let’s see what comes out of it.” This “trial and error” process can lead to very innovative results, but also lead to long and costly trial-error loops with end results that often come too late because the intervention has already been outdated by technical and medical progress.

The approach doesn't seem to have worked so badly so far if you look at the triumph of health services research over the last 10 or 20 years.
This is also the case. In the first two decades of healthcare research in Germany, the main aim was to clarify issues arising from politics and practice through good methodological research. But at some point you will reach the point of what is difficult to explain or, in the case of completely new events like Covid-19, the point of what is not yet explainable: then at the latest you will need theory. This point has now been reached.

Let's take an example: What is the theory when it comes to the question of what a pilot brings?
They'll have a hard time finding them. An international review was recently able to identify nine functions from the literature. In general, a separate theory is usually required for almost every function of a supply structure - such as the pilot here. For example, a Loste can primarily have a supporting function. The theory of social support can be used to explain this function. If, on the other hand, the controller is primarily a coordinator, one would have to use a coordination theory, such as Habermas's theory of communicative action. And so forth. However, a controller can and usually will perform different functions at the same time, which of course makes the whole thing more complex.

Accordingly, a theory of the “metasystem pilot” would also be needed, which would then have to be supported in a context- and/or indication-specific manner.
This is going in the right direction. What is particularly important is to formulate a theory based on observations or to use an existing theory to explain it. But that alone is not enough, because you have to continually check the validity of a theory in health services research projects.

But they set the bar very high.
It's not me who sets the bar high, it's science. Science is more than just good empirical research. Science is characterized by the interplay of theory and empiricism.

Your core requirement?
Health services research must become more scientific. Not only to help establish better care in a timely manner, but also to be able to catch up with the other sciences that have long since mastered this interplay. To take the example just mentioned: To test the thesis that the pilot performs a support function, one would have to - as mentioned - use the theory of social support and, on this basis, plan an intervention in which the pilot only performs this one function. The randomized controlled study that would then be carried out would enable this theoretically supported thesis to be tested. A confirmation of this thesis would then not only have practical significance, but would also prove the usefulness of the underlying theory. This results in scientific progress little by little.

If you followed that, you would have to do three years of research just to practically test the support function of a controller. If you were to look at the nine functions of a controller - there are certainly more - in isolation, you would be busy for almost 30 years.
You have to do it on a larger scale
See world frame. If nine research groups around the world were to focus on one of the nine pilot functions, after three years we would have a rough idea of which pilot functions are particularly important and which are not. Then we will basically know better in a relatively short period of time. On this basis, much more effective pilotage models could be developed in the future. With this approach we are slower at the beginning, but faster at the end.

To do this, you first have to know the possible theories.
This is made even more difficult because there are a relatively large number of theories and because not every theory solves every problem. If there is a specific supply problem, one must ask which theory is suitable and best suited to explain this problem. However, health services research - especially in Germany - has not developed in the lofty tower of science, but rather from very practical questions.

For example?
For example, clinicians approach medical sociologists and ask what reasons there are for hospital doctors not applying the existing medical guidelines and what can be done about it. Medical knowledge of the body is not sufficient to clarify this question. Investigating the behavior of doctors genuinely falls within the remit of sociology, psychology or behavioral economics. To a large extent, health services research is not health research, but rather behavioral research.

This could certainly explain a lot of things, such as the question of why there are so many hip operations or suddenly too few intensive care and ventilation beds or at least a considerable discrepancy in the reported numbers.
If you have the relevant theories, yes. Incentive-based theories of organizational behavior could be used to answer your first question. The phenomenon of reactive measurement can be used to clarify your second question regarding the correct reporting of the given supply structure. We know, for example, from empirical social research that when reports are financially or otherwise relevant, some actors are enticed to report in such a way that the desired effect, such as financial support for infrastructure, occurs.

What do we learn from this?
There must be no interest-sensitive measurement of structures, processes and results in care. If you want to make rational socio-political decisions, you need indicators that cannot be directly influenced by interest-driven actors. One such indicator would be, for example, the Covid-19 incidence determined on the basis of representative cohort studies.
denz rate. Such an indicator cannot be directly influenced by health actors, but the number of reported intensive care and ventilation beds can, in principle, be influenced. Incidentally, the Federal Audit Office only recently pointed out this last point. We learn from this that indicators that cannot be directly influenced by interest-driven actors should be used as a basis for political control. In addition, these indicators should best be collected by other actors.

For example from the health care researchers?
For example. Or even from the epidemiologists. If this ideal of interest-independent measurement is not practical, we must specifically address the actors in the health system and critically analyze their behavior. This could then be described as critical health services research. This often occurs when the health system and its actors are themselves the subject of a questioning investigation.

Will this not make health services research vulnerable?
The health and medical system must be able to withstand evidence-based criticism. As a scientist, I also know that this path is not easy, both for those who are being judged and for those who judge critically and are in turn criticized for it.

Knowing full well that most health services research institutes are located in or connected to university hospitals.
We have to talk about it openly. Critical health care research is certainly a double-edged sword. There may be harsh backlash. However, it is a path that health services research should also take in the future, among many other paths. Health services research must learn to become more self-confident and independent.

You have already seen what happened in the public health scene when it wanted to become an independent “School of Public Health” outside of the “School of Medicine”.
Yes that is correct. The idea of establishing “Schools of Public Health” outside of the “Schools of Medicine” in Germany and the USA has not really taken off. That is why I advocate taking a cautious approach with the declared goal of establishing itself as an independent healthcare science at universities and medical faculties.

Hand on heart: The majority of health services research to date isn't exactly a game changer or even a system buster, is it?
From my point of view, it was an important first step to optimize the existing system with the help of health services research. If we can do this on a broad basis, we will have achieved more than was previously conceivable. However, health services research should also ask system-critical questions when it comes to stubborn problems and dysfunctional structures. This is the only way real system innovations can emerge. But that's exactly what you need, not just methods, but also profound theories.

Which brings us back to the topic: which theories?
Just as an example: I recently took part in a conference of German health sociologists, where the figuration theory1 by Norbert Elias, the structuration theory2 by Anthony Giddens and the structural-functional theory3 by Talcott Parsons were used to explain health care situations. You can learn a lot from health sociologists here. The same applies to psychology. In health services research, for example, the behavioral theories4 of Fishbein and Ajzen and the health behavior model5 of Schwarzer can be profitably applied. That's exactly what Braithwaite and Mannion say in the MVF interview: making the existing theoretical tradition more usable for health care research and care practice. That is the goal.

How would that work?
The first step is to set up impact models for a planned intervention. To put it simply, impact models are used to formulate theses about the short, medium and long-term effects of the individual intervention components. It would be ideal if these theses were also derived from higher-level theories. Then the test of the intervention would also be a test of the thesis and partly also a test of the theory on which the thesis is based. The starting point of the approach should be the impact model of the intervention.
Which is nothing more than a theory.
True. In this case, I would still prefer the term impact model. In a publication6 by the Innovation Committee's expert advisory board, we recommended the use of such impact models.

Were you successful with your initiative?
My impression is that the awareness of making impact models the starting point for healthcare innovation is gradually gaining ground. However, in my opinion there is still a lot of room for improvement.

Where does it come from?
Let's take the example of the Innovation Fund. Many projects in the Innovation Fund were initiated by clinicians. They come from practice, recognize a gap in care and try to close it with an often very well thought-out care intervention. This approach in itself is welcome.

The but?
This care intervention takes place without recourse to a
Theory, we are in trial and error mode. It may be that you had the right idea and this is confirmed by the evaluation. However, it can also be the case that there is a negative result. In both cases, without theory, one cannot properly explain why the intervention worked or why not.

Isn't every person a theorist? Plus a national football coach and recently also a virologist or epidemiologist?
Secure. All people are lay theorists. We would not be able to cope without assumptions about how the individual areas of life function. These lay theories may be wrong from an objective scientific perspective, but as long as people believe in them, they guide their actions. This is what the Thomas theorem7 suggests. It says that what man believes to be true is true in terms of the consequences that man derives from it. Incorrect lay theories regarding health can lead to incorrect health actions.

They are then called Covidiots, among other things.
Corona deniers are in fact an example of people who have a particular form of lay theory regarding a specific disease. This lay theory then entails the well-known consequences for the actions of this group of people. It is the task of science to compare and convey “evidence-based” theories to these lay theories.

But scientific theories also have their limits.
Of course. A theory is only valid and useful for a certain area. One can put it this way: A theory is a tool for the scientist and the practitioner. If, in the course of a health services research project, it turns out that a theory is not really suitable for solving a specific health care problem, you have to – to stay with the picture – look in the “theory toolbox” to see which other theory would serve better as a tool.

If you were to set up every healthcare research study like this...
...that would be ideal from the perspective of theory-oriented health care research.

Does this mean that health services research should dare to do more sociology?
More sociology and more psychology. Clearly. In any case, health services research must become more theory-oriented, because nothing is more practical than a good theory, as Kurt Lewin8 said. This also means losing your fear of theory itself.

You probably hear that often.
Clear. When clinicians hear the word theory, they often fear that the practical relevance is lost. But that doesn't have to be the case, because as I said: Nothing is more practical than a good theory. However, this must be “translated” for the practitioner in such a way that he can understand it and implement it in practice.

Let's take the problem of the above-mentioned insufficient application of well-known guidelines in hospitals, which has actually changed little to date.
If you approach this problem from the perspective of a systems theorist, you will see these acceptance problems as evidence that we are dealing with an autopoietic system9 in the medical profession. Such systems create themselves, regulate themselves and only accept suggestions from outside, but cannot be controlled directly. Autopoietic systems can only be controlled indirectly, for example by setting a financial framework. Systems theorists call this context control10. We as healthcare researchers therefore need to better understand overall what makes the doctor as a person and the medical profession as a social system “tick”. For this we need theory-oriented health services research.

This would answer the question of why you are only making this maximum demand now, on the 20th anniversary of the German Congresses for Healthcare Research, of which you are the first and 20th president.
We first had to methodically secure and advance health services research. That was important in the first step, because good methods can get you relatively far. But at some point there inevitably comes a point where you realize that this methodological approach alone cannot explain all the important phenomena in care and that you often have to accept learning processes that are too lengthy and unsystematic. That is why it is now time in health services research to take the second step and place theory orientation alongside method orientation. I would call this health care research 2.0. We are now igniting the second stage, so to speak.

Maybe also because some health care researchers have had the feeling for a long time that no matter how good a method, you can only go so far.
That's exactly the point. We have come a long way, but we now need to continue to improve scientifically. This is one reason why we at the University of Cologne do not call the master's degree program in health services research, but in health care science. Health care science is health care research plus health care theory plus critical health care research.

Why do you call the 20th German Congress on Health Services Research rather banally “Understanding the healthcare context – promoting practice transfer”?
I don't find the title that banal. If you want to understand the context, you need a theory about the context. Anyone who wants to promote practice transfer also needs a theory about practice. In principle, theory is, so to speak, in the subheading of the conference title.

Why not in the title?
The title of a congress is the result of a participatory decision-making process. I noticed that there is something like a fear of the concept of theory. The clinicians among the healthcare researchers then immediately see the ivory tower of science in front of them and hesitate. But this is only the 20th Congress. We're now starting to ignite the second stage. As my former teacher Bernhard Badura once said, it will take at least two decades before a new orientation is finally established. We are working on it and want to venture into more theory. <<

Professor Pfaff, thank you for the interview.

The interview was conducted by MFV editor-in-chief Peter Stegmaier.