Panel discussion at the 8th “Monitor Health Care Research” specialist congress. Prof. Dr. Thomas Wilke, Dr. Michael Happich, Prof. Dr. Dieter Paar, Dr. Ilona Köster-Steinebach, Corinna Schaefer, Prof. Dr. Reinhold Roski, Prof. Dr. Wolf-Dieter Ludwig, Prof. Dr. Bertram Häussler, Prof. Dr. Matthias Schrappe and Prof. Dr. Gerd Antes.
Roski: I recently read that the country of Israel wants to release all health data stored there for international research: The government expects this to bring significant benefits, both for the care solution itself and economic benefits, combined with the hope that that companies that work with this data and want to pay for it settle in Israel.
Häussler: This is similar to what happened in Iceland around ten years ago, when the genetic database was made public. This gave a boost to Iceland and to international research as a whole. In our case, however, I see a great deal of fatigue, which is also expressed in the fact that the coalition agreement mentions digitalization 271 times. Mr. Schrappe, you just said that the patient file will not lead to an integration of the health system. Maybe you are right that in ten years we will say that nothing has happened again. But I don't actually believe that this time, because there are at least three huge potentials at work - Google, Amazon and Apple - that will make offers in this environment that, in the worst case, there is nothing to counter apart from rigid bans.
Antes: I - not surprisingly - have an extremely different opinion. Of course there was a kick in Iceland back then. But where is the evidence of benefit for this? Especially now that the world is being reorganized, so to speak - even if Facebook has just fallen on its face in the US Congress - we need proof of benefit and, above all, the weighing of benefit against risk, which we currently completely lack. In particular, we have a huge conflict of interest on the part of computer science and also on the part of my colleagues. By the way, science is currently completely failing to position itself on the topic of “big data”.
Schäfer: In connection with collecting a lot of data, it depends on the concepts on which this collection is based and the benefits I expect from it. The Quantified Self scene, in which many people store their data and hope to derive insights from it, is very indicative of this. These insights that one hopes to derive are based on assumptions that are not well supported by evidence. Especially not through causal evidence. Here we are faced with the question of which social system we operate in when collecting data, which in the worst case scenario can lead to the idea of solidarity being undermined. First you collect a lot of data, then we assign causality to it and at some point explain that the blame lies with you and that you can no longer take responsibility for it in solidarity. Quantified Self is a beautiful example of what can happen. This must be taken into account when demanding an assessment of benefit and harm.
Schrappe: From an economic point of view, it's primarily about making this data tradable. This is of course one of the last fields in which privacy exists. The question is whether we can even turn this back. I have the numbers pretty clear in my head: In the US, $26 billion of healthcare spending is spent on health research, $15 billion - well over 60 percent - of that on data in computer science, data analysis and genetics. The publications on this increased by almost 1,000 percent from 1994 to 2014, and in PubMed in general only by 175 percent. We are therefore seeing an orientation in this direction across the entire scientific scene. You can approve of that or not. However, I would like to point out that we actually need something like this if we want to set up electronic medical records. I'm not saying that no one would put anything in the file, but I'm saying: we're being too naïve if we expect so much from it. The whole thing is not as simple as it seems. The English started it in 2004 or 2005 and had to end their project after a year and a half because people simply didn't take part. That's why we need research in this area that observes and analyzes the environmental conditions, otherwise we will end up on our knees even with these hopeful approaches.
Ludwig: Mr. Schrappe mentioned in his presentation that digitalization was mentioned 271 times in the coalition agreement. I added that the term drug therapy safety only appeared once, and that was in view of demographic developments and multi-medication. For 15 years we have been talking about electronically assisted prescriptions, interaction detection and sensible pharmacotherapy. Today, in 2018, we are exactly where we were at the beginning of this discussion. There are many reasons for this. Partly, of course, it's the fault of the companies that were supposed to deliver the programs but haven't. Also at Gematik, which failed spectacularly with its electronic health card. If someone had told me 15 years ago that in 15 years you would be roughly where you are now, we would probably have been very desperate back then. The other point is – and I am extremely grateful to the speakers for this – that no one mentioned the term “personalized medicine”. This is the first event I've attended where this has happened. If we look at what personalized medicine, especially in areas such as oncology, where it plays a major role, has contributed in this decade from the amount of data that we now have about diseases, genetic changes and certain dispositions Patients can raise concrete improvements in medicine, which is very manageable. This is another example of how we may not benefit as much from big data alone. I would be happy if I was wrong.
Pair: There was a World Congress on Diabetes years ago where a poster was presented by patients who programmed their insulin pumps themselves using the sensor. It's somehow a bit embarrassing for us as an industry when young Type I diabetics start programming their devices themselves. Couldn't it just be that with all this big data, maybe we need to think a little more optimistically.
Stegmaier: Mr. Schrappe, you mentioned the term research agenda in your lecture. Who would you see as responsible for setting up a research agenda? Would that possibly be the G-BA?
Schrappe: I am not of the opinion that the G-BA, although it performs a central function in the Innovation Committee, would be the right address to set up a research agenda. I don't necessarily think that the government bureaucracy is the right one either. A more correct approach would be associations of scientific societies that could set up a committee of experts for a certain period of time to develop such an agenda. I don't think the Federal Joint Committee would be able to do this because it would filter everything - even what you see all the time - according to the respective interests. As a result, the innovative element that we urgently need would suffer.
Antes: If I may formulate a utopia? I would separate medicine from the BMBF, where everything from forest extinction to anywhere else is researched. Medical research belongs in the BMG, which would give us something similar to that in Great Britain. Anyone who interlinks research and care can also automatically prioritize, at least technically. But that's certainly not something I'll ever experience again.
Ludwig: As a clinician, I would like to say something about the role of patients. In all of our discussions and lectures, an important ethical principle is missing: that is patient autonomy. That's exactly what we want to achieve, especially in oncology. I have to say: Even after 35 years of clinical activity, we are still very early on and have achieved very little.
Köster-Steinebach: In principle, I share Mr. Schrappe's assessment of the G-BA's suitability for developing a research agenda. If you break down the complex intervention of creating a research agenda into steps, the first would be to ensure that we are given a platform to articulate, present and collect the important questions in order to eventually move forward. I don't see such a possibility at the moment.
Stegmaier: Do you have the impression that, as part of the early benefit assessment, the patient representatives in the G-BA can advance certain things a little?
Köster-Steinebach: I do think it makes a difference when patients sit at the table. Even if we are repeatedly outvoted because applications are rejected at a ratio of 13:1, 13:0 or 12:1. If you have to look the representation of patient interests in the eyes, some things may simply no longer work out the way they did without them. It's not quite the effect you'd want, but it's better than nothing. I was very grateful to Mr. Schrappe for calling for an objective need through research in our system.
Stegmaier: Mr. Happich and Mr. Paar: You have shown wonderfully that you can extrapolate if you want to. But is this even recognized if you want to move from a narrow study population to a broader population closer to care? Is this appreciated in Germany or have you not tried it yet?
Happich: We're not that far along yet. First of all, we are generally in the process of clarifying acceptance. That was the first question as part of the IMI project that I presented, whether regulatory or HTA authorities consider this approach relevant for the future and whether the instruments meet the relevant quality criteria. In my opinion, the method itself has great potential that needs to be viewed from a European perspective. Europe may simply have more budget to push forward with such a project. To do this, however, we have to manage to perhaps step away from the isolated perspective in Germany a little, then the things that we can implement in Europe might have a little more traction.
Wilke: I want to be a bit more direct. I would say: Dear G-BA, just do it. Ask a submitting company what your results would look like if you applied real-world patient characteristics. I really don't understand why you don't just do that.
Couple: And how do we then deal with the “blue field”* that you showed in your lecture?
Wilke: That's why I say patient characteristics instead of population characteristics, because the latter are certainly methodologically more demanding. But we are now very methodologically confident in the area of patient characteristics. That's why for me it would be a useful approach to show how an innovative preparation works when real-world patient characteristics apply. I see no reason why this shouldn't be considered
G-BA could require.
Stegmaier: Ms. Dr. Köster-Steinebach, you took part in many public negotiations as a patient representative in the G-BA for eight years and also took part in secret subcommittees. As a patient representative, if you had known that such a thing existed, would you have called for these types of studies?
Köster-Steinebach: First of all, I would like to answer Mr. Wilke because I know why the G-BA doesn't do something like that. The answer is very simple: because it is not in the law. If it were in the law, the G-BA would do it. But because it's not in there, he's afraid to do it. And when you asked whether I would have appreciated this approach: absolutely!
Stegmaier: But you could have financed that yourself with your own budget.
Köster-Steinebach: Absolutely. But that is another question. But that leads us to the question of why studies are not closer to patient characteristics. Anyone who makes this demand that such extrapolation projections be made may have to do so on the basis of very small numbers of cases, which must be given high weight. This again has to do with high levels of uncertainty, which will certainly also have an impact on the quality of the RCT. I have often experienced in negotiations that it was said afterwards that the available study data would not help answer our questions. That's why I advocate a stronger educational effect when designing such studies. If we are going to use money for this, these studies should also be such that relevant questions can be answered afterwards.
Ludwig: In the lectures we heard two examples from the range of oncological diseases. 20 or 25 years ago, when I first treated patients with this disease, it was one large population pool that included, so to speak, all patients with non-small cell lung cancer. Today, this pot changes from year to year as new genetic changes are discovered, all of which have prognostic significance. That's why we would have to simulate everything again every year, which I consider to be almost unfeasible - because it is very costly. I would like us to work on the quality of the studies themselves before we simulate too much.
Roski: There doesn't seem to be much controversy between RCT and other methods. Everything here is combined so elegantly and politely.
Antes: Everything in the area of health care is very intersected. I would even start with the Akademia, which I once referred to in an editorial as the “Axis of Evil”. Because everyone – from the ethics committees to the members of the Bundestag – are perpetrators and victims at the same time. The only people who got off well in my editorial were the industry. Because she is the only one who behaves rationally and does what she actually wants to do: make money. Everyone else, however, has a hidden agenda. But because the whole topic is so complex, we have such enormous problems with it.
Roski: Can you name a few points?
Antes: Take medical biometrics, for example, where a few chairs have just been folded. But now there are so-called data scientists who largely don't know how to evaluate complex studies. And this is exactly what promises the future through the evaluation of large amounts of data. When I wrote the first article on this topic in the “Laborjournal” in 2015, a colleague approached me and said: “In principle, you are not entirely wrong, but what should I do now? I am exactly in the time window where I can get a permanent position and a professorship and have built my career on this topic.” This is the result of research funding from the BMBF. We pushed the whole ship in that direction, and even if we wanted to get out now, it wouldn't be quick.
Schrappe: That's exactly how it works. It's also the same with health services research. Programs have been set up here that are also linked to the fate of life. That's why you don't have to blush when you talk about scientific elements being included in the law books. This increases the sovereignty or ability to interpret the boring reality by representatives of these scientific disciplines, be it EBM, health services research or whatever. That's why everyone sticks to it, of course. In my observation, however, scientists are very bad at dealing with political influence because they are simply unable to understand the difficulties of political decision-making. Instead, they think they can really influence things with their advice, but they underestimate the complexity of the processes. But because they want to hold on to their virtual influence, in some cases they are even prepared to make certain compromises in terms of - let's put it this way - scientific purity.
Häussler: The question was not whether, but what interests lay behind it. Let's just look at ourselves and start with Mr. Antes, who seems to think that data scientists are a snake that grabs at everything. I ask: “Can only biometricians handle clinical data?” Or could it not be that the academy just assumes that everything the others do is the devil because it tries to defend its topics against others? She does this by simply claiming: “The others are not experts.” Now I come into my own and explain: “My interest is to make much more use of all the data sets that statutory health insurance hoards, for example without constant restrictions and control.” My demand: “Just let it run its course and see what comes out.” And the questions that follow: “Why don’t you just let us off the chain?” “Why don’t you let us have correlations Calculate until it smokes.” Afterwards, we can still ask each other what connections exist and how they can be validated. We could also do time series analyzes to get even closer to causality. We should not leave this to countries like Israel or Iceland or, above all, the United States or even China.
Köster-Steinebach: We actually also have a crisis of independence from science. As an applying institute, you know that if you get results that the contractor doesn't want, you don't need to apply a second time. This ultimately means that all results are always produced based on interests. I therefore advocate a stronger separation between the stakeholder and the client of evaluations. If we don't manage to do that, then we will ultimately have an activity that misses the core of science.
Ludwig: What you mean has also become relatively clear in some projects within the framework of the Innovation Fund.
Schäfer: First of all, Mr. Häussler, why you shouldn't be let off the chain. I can tell you this: From my point of view, because when we update the next guideline, we will first of all find 135,000 systematic reviews on the correlation between asthma and psychological problems in mothers during pregnancy. We have to ask ourselves how action-guiding this is. We can certainly find a lot of correlations everywhere, but we always have to sort out everything that doesn't guide action. Of course, you will also find a lot of things that lead to wonderful research hypotheses and that you can then try to research further. But ultimately it's about treating patients well. To do this, we must mobilize all the resources we have, and not just in any way, but mobilize intelligently. That's a problem with big data: we simply don't have unlimited resources to somehow process all this data, which is why we leave this work to machines. But our clinical and scientific thinking may fall by the wayside. That's why the core question of everything we do is: What specifically helps patients? But not if I tell him you have asthma because your mother had mental problems during pregnancy.
Häussler: The questions that we are dealing with are not l'art pour l'art questions. An example: We now believe that we have a hepatitis epidemic due to the hepatitis virus. But what if diabetes played a very important role here? And what if insulinization of the population played a very important role? If we don't know this because we don't research it, then your reviews are going in the wrong direction.
Couple: We can only make money with approved drugs, but without an RCT there is no approval.
Ludwig: That's not entirely true, there are many approvals without an RCT.
Couple: When it comes to cardiovascular indications, it doesn't work without RCTs. Our last one cost 24 million euros. I do believe that real world evidence can help us do better studies. “Letting go” also means that under certain circumstances you realize that what you previously thought may suddenly be questioned because you say to yourself: “Somehow this can’t be true!” That's why I would advise trusting the experts on this question, who will ensure that no nonsense correlations are calculated.
Hildebrandt: I would like to suggest a change in perspective. What I find flawed in our German system is that there is a divergence between the interest in optimizing care and health benefits and the respective partners' own interests. The problem is that health insurance companies do not have a 100% interest in optimizing the health benefits of patients. What they do have, on the other hand, is an economic interest in getting appropriate resources from the health fund and spending less. The same applies to every player in every single sector. For the practicing doctor who behaves contextually differently than we would like precisely because he is financed differently. This also applies to the hospital. So why don't we consider how we can create new levels of organization that generate economic interest in maximizing health? If we could organize this politically, we would also be interested in better studies. <<
