Robert A. Harrington, MD: Hi. I'm Bob Harrington from Stanford University, and I'm here with my friend and colleague, Adrian Hernandez, who is a professor of medicine at Duke University and the [Outcomes &] Health Services Research Director for the Duke Clinical Research Institute. Adrian, thanks for joining us.
Adrian F. Hernandez, MD, MHS: Thanks for having me.
A Broken System
Dr Harrington: Adrian and I are going talk about some of the challenges associated with doing randomized clinical trials these days. Then, we'll talk about an exciting project that the two of us are involved with and that is sponsored by the Patient-Centered Outcomes Research Institute (PCORI): ADAPTABLE, a uniquely structured trial studying an interesting question in cardiology that we'll come back to.
But, Adrian, let's start with the background. Is the clinical trial system broken in this country?
Dr Hernandez: Yes. If you just think about patients facing daily questions and clinical decisions that they need to make, or that their clinicians need to help them make, we don't really have answers. We don't necessarily have [real-world] answers for what patients face every day.
Dr Harrington: These are common questions.
Dr Hernandez: That's right. Common questions. To get to those answers, our traditional trials network is very cumbersome, takes a long time. It's very expensive, and often it represents a very narrow slice of the population that we care for.
Dr Harrington: This is really the issue, isn't it? You and I have both been doing trials for years. We sometimes focus on a very narrow question. We build up the trial network. We get the trial answered over a very long period of time. And then that trial network—in many ways—goes away until the next thing comes along.
Dr Hernandez: We keep breaking it down, rebuilding it, breaking it down. We also have a redundant system. We have healthcare delivery where we're operating in [one] space while a parallel universe enrolls patients and collects the same data.
Dr Harrington: It's really astonishing, isn't it? If you think about it, we spend all of this money. Roughly one half—or maybe one third—of a coordinating center's budget for a clinical trial goes to monitoring, right? Of that budget, they're concerned about [whether] you put what was in the electronic health record or in the paper health record over here [in the research record].
That's how the errors occur. The errors occur going from here [points to head] to here [EHR], and yet we're monitoring, and people have made whole industries out of catching these relatively useless mistakes.
Dr Hernandez: Right. In simple terms, say approximately one half of the cost of a trial goes into things like monitoring that are not directly getting to the answer. By that equation, we could actually double the number of questions we could answer and/or broaden the types of drugs or devices to get to patients.
Dr Harrington: Michael Lauer [director of the Division of Cardiovascular Sciences], from the National Heart, Lung, and Blood Institute (NHLBI), likes to say that rather than spending $50 million conducting a clinical trial, he'd rather spend $5 million on 10 different questions. If we could actually get a level of efficiency that would allow that, wow! What a different world it would be.
Learning from the Swedes
Dr Harrington: One of the things that you and I talk about a lot is our friend [Drs] Stefan James and Lars [Wallentin, from Uppsala Universitet].
Dr Hernandez: The TASTE trial.[1]
Dr Harrington: My only regret about the TASTE trial is that our groups were not involved with it. But it was brilliant. Why don't you remind the audience what that was?
Dr Hernandez: The TASTE trial was really impressive. It was asking a question that's embedded within the registry. In this case, it was aspiration of thrombus in a case of myocardial infarction (MI), and what that has to do with clinical outcomes. Everything's already being collected as part of the registry, so it's end-serving randomization with two different treatments and then seeing what happens next.
Dr Harrington: In a registry, that's real-time, being collected at the point of care, and the registry is forming the patient record and forming the research record at the same time. It's truly single data, entered once, being used for multiple purposes.
Instead, what we're doing now is duplicating.PCORI's gotten interested in this, and you've been working with PCORI to understand how to tap into the infrastructure that they've created called PCORnet [the National Patient-Centered Clinical Research Network]. Talk to us about what PCORnet is, and why PCORI got involved in this.
Dr Hernandez: PCORI saw this as an opportunity. There's a need for national research infrastructure that unites clinicians, patients, health systems, and all the data that are being generated from patients every day as they have health encounters. [The infrastructure takes] the electronic records and billing claims and marries those two together, and uses that as an infrastructure for observational studies and pragmatic trials. [Importantly, this is] research embedded into healthcare. As patients are experiencing their care, we're constantly answering questions.
Dr Harrington: The learning healthcare system.
Dr Hernandez: Exactly. Right now, it has 13 Clinical Data Research Networks. That encompasses approximately 70 healthcare systems. That covers around 75 million lives. Then, we have 20 Patient-Powered Research Networks that bring patients together to enable investigators to answer their questions.
Dr Harrington: It's a unique way to marshal our country's size and diversity. Being able to bring in 75 million individuals almost assures that you're going to have a representative slice of America. Is that fair?
Dr Hernandez: Right. If you have specific questions that have never been answered for a certain population, then you can go target them. You can also have so-called "curated cohorts"—cohorts of well-characterized patients that have been asked, "Do you want to be a part of research?" So, when you can have a research question, you can go to them immediately and say, "Here's a study. Would you like to participate?"
Dr Harrington: PCORI spent the early years of its funding creating what I'll call the infrastructure. They've done that quite well, and they've got these data networks ready to be used. One of the things that I think we'd all agree on is that in order to study treatments or diagnostics, if you're going to compare A with B, you have to randomize people.
Dr Hernandez: That's right.
Dr Harrington: PCORI realizes this, and they've put out a request for proposal (RFP) for, what do they call it, pragmatic clinical trials? What were they trying to get at?
Observation Only Goes So Far
Dr Hernandez: Yeah. There's been a lot of concern about whether you can get to the answers that matter most through observational methods. The answer is, not quite yet.
Dr Harrington: You'll get part of the way, won't you? But you won't get all the way.
Dr Hernandez: Right. Invariably, randomization is a way to balance all of the confounding factors that you know about and don't know about. Here is a way to answer questions, especially centered on two or more different strategies of care—or situations in which patients have choices available to them—and randomly allocate those choices. Then we can find the answer that will matter for the whole population.
Dr Harrington: As I understand it, PCORI was very interested in either big public health questions or, conversely, questions involving niche populations that only they might have enough patients in their networks to answer.So, "pragmatic" doesn't mean big.
Dr Hernandez: Right. It comes down to impact. For a rare disease—say, a familial cardiomyopathy, or something else—that may have a very focused question. Hypertrophic cardiomyopathy, which has been poorly studied, is considered a rare disease; there is little evidence of what to do in randomized trials.
Dr Harrington: But [a trial on a niche population] could leave a big impact because it could affect, in a major way, a small group of individuals.
Dr Hernandez: That's right [and] then larger populations—patients with heart disease or coronary artery disease. Stroke is still the number one cause of mortality and disability, and there are [important] questions that can be answered in a simple fashion that have a large impact in the total population.
Dr Harrington: Population health. In the category of rare diseases, you might have a big effect in a small group. For common diseases, such as coronary heart disease, you might have relatively small effects—but given the size of the population involved, from a public health perspective, a big [impact].
Dr Hernandez: That's right.
ADAPTABLE
Dr Harrington: [PCORI]put out the RFP, and one of them that bubbled to the top is one that you and I have been involved with, and that concerns dose of aspirin. I've talked on Medscape Cardiology a number of times about a variety of antiplatelet agents, antithrombotic agents. I suspect you have done the same. We often say, "Well, we don't really know the dose of aspirin; we really don't know the proper dose of aspirin." Which is almost astonishing.
Dr Hernandez: It's amazing. Aspirin's been around for over 100 years, and when you look at the studies, we still don't know whether a higher dose of aspirin lowers the risk for mortality and heart attacks? Or does a lower dose improve the risk for bleeding and still have the same benefit? What's the risk/benefit ratio?
Dr Harrington: And is the risk/benefit ratio the same in men and women? Is it the same in diabetics and nondiabetics? Older [or] younger [patients]? African American? Caucasian American?We don't know.
Dr Hernandez: That's the concept of moving toward more personalized medicine. What's the right treatment for the right patient? You have to have large studies [conducted] in a simple fashion to really understand whether there are differences among these different groups.
Dr Harrington: Yeah. It's a great point, Adrian, because often when we talk about precision medicine or precision health, people immediately go to, "Well, if I understand their genome, and I understand their proteomics, then we can tailor that." We're going in that direction, but we also need to study large populations [stratified] into the various phenotypes.
Dr Hernandez: Just getting to the basics, if we can do that, would be great.
Dr Harrington: Aspirin is one of these questions that many of us have said, "We need to study this." When you look through all the literature, as you've rightly pointed out, there's never been—amazingly, given all the things that have been studied in cardiology—a sufficiently-sized direct comparison of low-dose (so-called "baby aspirin") vs high-dose aspirin, 325 mg.
Dr Hernandez: That actually turns out to translate (depending on which is the right answer) to up to about 20,000 deaths or MIs per year, or thousands of bleeds, depending on the answer. We really don't know.
Dr Harrington: These bleeds that we're talking about are not trivial: They result in hospitalizations, death, [and] blood transfusion. We're talking serious bleeding events.
Dr Hernandez: That's right. Actually, those bleeding events are much more common than [health concerns] we often see in the national news—concerns about vaccines, for example. [Bleeding events] are happening every day to patients, and we don't know how to prevent them, necessarily.
Dr Harrington: Let's jump then to the [ADAPTABLE Aspirin Study] that was proposed to PCORI, went through a series of reviews, and is now going to be PCORI's first pragmatic trial. What's the construct of the trial?
Dr Hernandez: One piece of background: This question was one of a series of questions, so about 28 of them were narrowed down to this. The reason it was chosen was because of its [potential] public health impact and how it could test the network in a way that hadn't been done before.
The way this will work is to, using these networks and their electronic health data, identify patients electronically, approach them electronically, consent them electronically, and then randomize them electronically. The follow-up will happen through PCORnet as patients have their healthcare encounters.
Dr Harrington: So, through their electronic health records.
Dr Hernandez: That's right. So, when study participants enter the hospital for an MI or bleeding, that will be captured as they enter through hospital doors, literally because of what is generated as part of their healthcare encounter.
Dr Harrington: The important thing for the audience to realize is that this is utilizing an infrastructure of data networks that has been created, as we talked about earlier. This is the first test [of the data network] beyond it being an interesting technology.
Dr Hernandez: It's still called a "demonstration" or "pilot study," but it's 20,000 patients.
Dr Harrington: What I've told people is that the 20,000 was picked in part because of the math. In 10,000 patients in each group, we should be able to discern relatively modest differences between the two groups with regard to the hard clinical outcomes of death, MI, [and] stroke. But [the 20,000] is also picked to give us power to start to look at subgroups.The 20,000 is also a demonstration of the power of these networks.
Patient-Centered Means Patient Involvement
Dr Hernandez: Actually, interesting [feedback] from patients: Some of them feel like 20,000 is not large enough. Because they see [that there are millions of] patients with heart disease, they ask, "Why are we studying only 20,000 patients?"
Dr Harrington: One of the things that I've found most interesting about working on this project is the patient involvement. Why don't you talk a little bit about that, because patients have been involved with us every step of the way. They're members of the protocol-writing committee. They're now members of the steering committee. I think, most interestingly, they're going to be members of the independent Data and Safety Monitoring Board (DSMB).
Dr Hernandez: Along with the clinicians, the patients actually decided that this was an important question. Further along is the patients actually deciding which of the outcomes matter most to them.
There are a couple other outcomes that were picked by the patients because it's what they care about in terms of their health status. From the very beginning, patients have been with us side by side [in] planning facets ranging from the engagement of patients to designing the consent form to make it patient-friendly, and even the logo.
The most interesting thing will definitely be the DSMB. We'll be training two patient participants of the study. They won't be in the study, but they will be similar to enrolled patients. [The two patients] will be on the DSMB. So we'll go through training, and then they will get locked in with the other DSMB members to look out for us on the study.
Dr Harrington: And go through those very, very tough exercises of looking at interim data when you know that the trial is still going on, and when you know that the data are not complete, and dealing with all that angst that comes with overseeing both the safety of the individual patients, but also the integrity of the research question. And how do you balance all of that?
Dr Hernandez: Right.
Dr Harrington: I think this is going to be a tremendous exercise for what we can all learn together, and I think—and I know you feel the same—that this is really going to elevate the quality of the research we do.
Dr Hernandez: Right. Also, we'll be able to translate results directly into practice. Because, from the hundreds of thousands of patients who are like [study participants] in these health systems, we'll be able to reach out to them directly and say, "Here's the answer."
We'll also have patient channels to enable participants to tell their story about why they were part of the study, and then also why [others] should care about their heart disease in this way.
Dr Harrington: [It's] a completely new way of doing research, certainly in this country—even different from what our friends in Sweden are doing in terms of how data are collected and patients are engaged. I think [that] is absolutely critical if we're going to get more public buy-in to the notion of creating a learning health system where we're constantly asking, answering, and refining questions and improving the system of care that we're involved with.
Dr Hernandez: The other thing is that as we set up this model for ADAPTABLE, you can imagine that in cardiology, just changing the questions in terms of aspirin, the outcomes will largely be similar, but for different conditions the outcomes will be different. The infrastructure will be able to answer—
Dr Harrington: One beta-blocker vs another.Do you need both beta-blockers and an angiotensin-converting enzyme (ACE) inhibitor, or might you get away with just one in certain subgroups?
Dr Hernandez: Or even [conditions] that have not been studied yet, but for which there are available therapies, and we just don't know.
Dr Harrington: And some of the common [practices] within the healthcare system. I love one that Rob Califf [deputy commissioner of the US Food and Drug Administration's Office of Medical Products and Tobacco] was telling me [about]: they're conducting a randomized trial of how frequently you bathe a patient in the hospital.
Dr Hernandez: Or dialysis: Longer or shorter dialysis? That can have an impact on mortality.
Dr Harrington: Take your blood pressure medicine in the morning or at night? There are many questions that should be answered in real time, but we need the system to do it.
Bargain-Priced
Dr Harrington: One of the things we haven't talked about is cost. You and I have been involved with publicly funded and privately funded trials, and the cost—tens to hundreds of millions—is not out of the question. We're at a fraction of that with this trial [of] 20,000 patients. What's the initial budget?
Dr Hernandez: $10 million.
Dr Harrington: Which is extraordinary.
Dr Hernandez: Right. If you compare it with other studies that have been conducted in a pragmatic fashion, even with the National Institutes of Health (NIH), that's usually about a $40 or $50 million study. At least. If you take other studies and you just add anything else that's required, then we're talking about hundreds of millions, 35 countries, and hundreds of sites.
Dr Harrington: So with 330-350 million people in this country, we ought to be able to answer a whole series of questions in a much more efficient way. I congratulate you and other leaders of PCORI and PCORnet for embarking on this. It's going to be an exciting journey, not just for the question—which is of major public health importance—but also for how we do the research.
Dr Hernandez: We're hoping that for each of these different concepts of building a trial, we actually have a learning research network. From how you develop [informed] consent that can be widely used and understandable to the electronic components of that—contracting all these different steps that cause everything to slow down.
Dr Harrington: You reminded me: We're actually going to do embedded randomization around adjudication within the trial.
Dr Hernandez: Right, and also follow-up.What's the right follow-up? No one knows.
Dr Harrington: We all think of those things in clinic, right? You see a patient, you say, "I'll see you in 3 months." Why aren't we studying these things?
Dr Hernandez: Three or 6 months. What matters?
Dr Harrington: Right. Why don't we just randomize and find out what the right answer is?
Dr Hernandez: That's right. That's why we want to be able to answer multiple questions: knowing what's the best method? What's the important health issue? Also, how do you measure different outcomes?
Dr Harrington: What are the systems that should support all of that?
Dr Hernandez: Exactly.
Dr Harrington: Adrian, this has been a terrific conversation. I want to thank you for joining me here on Medscape Cardiology, and I want to thank you, the listener, for tuning in.
© 2016 WebMD, LLC
Any views expressed above are the author's own and do not necessarily reflect the views of WebMD or Medscape.
Cite this: Robert A. Harrington, Adrian F. Hernandez. Will PCORnet Save US Clinical Research? - Medscape - Jan 25, 2016.
Comments