Improving Clinical Research and Patient Care with Health Informatics: The Critical Feedback Loop
Ann Nguyen:
Hi! This is Ann Nguyen, Senior Associate Conference Producer with Cambridge Healthtech Institute. We have a podcast interview today for the Clinical Genomics and Cancer Informatics conferences, which are part of the 2016 Bio-IT World Conference & Expo, returning to Boston, Massachusetts this April 5-7. We're pleased to be talking with Dr. Louis Fiore, Executive Director of MAVERIC, the Massachusetts Veterans Epidemiology Research and Information Center with the Veterans Affairs Boston Healthcare System.
Hello Lou, thank you for your time today.
Louis Fiore:
Hello Ann. Pleased to be here.
Ann Nguyen:
You have board certifications in Internal Medicine, Oncology and Hematology. Why Veterans Affairs and programs as your focus? How did you come to work in that area professionally and what's its value for clinical genomics research scientifically?
Louis Fiore:
Well Ann, I've been working for the Department of Veterans Affairs for 35 years, my entire career. Started as an intern and resident and what drew me initially and then has kept my interest in working for the VA, is that it's an accountable care organization that has embedded within the clinical care ecosystem research opportunities. So there's lots that one can do professionally in a research mode while caring for patients and there's a lot that one can do to improve the care of the patients, because by virtue of the fact that research and clinical care are embedded. Additionally, there's no money that changes hands in the VA. As an accountable care organization everything is controlled internally so there's no profit motive which has always been very important to me. But I think the most critical aspect of working in the VA across medicine and hematology and oncology is the enabling nature of the electronic medical record. VA has been digitized now since the '80s and we have a wealth of electronic health record data at our fingertips.
Additionally, it's a national system so that there's 150-plus hospitals that are linked together. You can imagine the power of aggregating those data and really being able to detect signals that you couldn't detect unless you had such a large database. And then finally each hospital is staffed with folks who are engaged in both patient care and research. So it's easy to design programs and to have them taken up by individual sites. And since the patients are veterans who are used to volunteering from their service days, you have the ideal environment to conduct research. So in brief, you have an infrastructure with electronic medical record that's able to do really cool stuff and you have willing participants.
You mentioned genomics research. If you add to the mix genomic findings, then that's the special sauce the VA has, makes that really fly because most genomic research requires access to EHR data and access to EHR data of large numbers of patients. And so what better environment to do that, and for the past 5 years that's been my primary focus.
Ann Nguyen:
You have particular expertise in the reuse of Electronic Health Record data for quality improvement, discovery and validation of genomic knowledge. How do you see this area evolving in the next decade?
Louis Fiore:
That's a great question and I think that it's a very exciting area and almost certain, and I'm not almost certain of many things, that the area is going to become increasingly important. There's a catch here. There's data that's contained in EHR provided your healthcare organization can aggregate it as I just discussed that VA can, you have the potential to really cool stuff. But there's also an application layer of the electronic health record and in order to be really effective, to effect change at the point of care, you need to be able to introduce work screens into your electronic health records that facilitate discovery and validation. Simply having access to the data is wonderful and necessary but it's not sufficient. So that when patients are being cared for in a facility, there has to be some way to generate knowledge from the database certainly, but then to return back into the healthcare system information that helps take care of patients and improves care. That's the critical feedback loop. I think that when that happens, you'll enable learning within a healthcare system, so-called learning healthcare system activities.
Current electronic medical records do not take this into consideration. They facilitate longitudinal recordkeeping of patients, typically using text, not structured data, and very few of them have abilities to generate analytics to support decision support and most importantly to apply decision support. The backend of EHRs is what has to change and I think that in the industry that will change and the incentives to do that are going to be from all different directions, both improving quality of care as well as generating new research findings. For genomic research, again, it's most critical. When someone has a genomic profile done, be it germline or tumor, that information is extensive and will reside outside of the EHR. So EHR flexibility has to not only be able to serve to feed back to patients and clinicians what to do and how to do it more effectively, but have to be able to access disparate data sources that contain, for example, genomic data. And further, the analytics have to able to operate on both genotype and phenotype data. Those are all big changes. None of that is happening in any one place today. Lots of it's happening in different places and bringing it all together. That is what will be really exciting.
Ann Nguyen:
What's so transformative about MAVERIC's Point of Care Research Program, which uses health informatics to enable clinical research and enhance clinical care?
Louis Fiore:
Let me pick up where I left off at question 2. That's a great segue question here. The VA Point of Care Program is attempting to do what I just described as sort of the end game. The VA Point of Care Program touches patients when they're admitted to a hospital or to a clinic area, consumes their electronic health record data, operates on it and returns to the point of care a recommendation on how to better take care of the patient but it doesn't -- not necessarily as decision support, but it reintroduces the question as a can we flip a coin between these two apparently similar outcome treatments and randomly assign patients, thereby making the value of their follow-up data far higher because it's not observational data but rather it's randomized data to make that real, because that’s a lot of talk. Patients being admitted to the hospital and they require intervention with drug A or B, drugs A and B are both used equally. No one knows which is better. That means there's clinical equipoise about which of the two to use. Too expensive to do a clinical trial on every decision A or B because there's thousands of those.
Imagine then, the patient admitted to the hospital needs either A or B. When the clinician goes to pick A or B, the electronic medical record says, "Let us pick one of these two for you if you have no problem and the patient agrees." If that proceeds in that fashion, then the patient is randomized between A and B, and clinical care continues as it would have had that interaction not have happened. And then we simply cull data from the electronic medical record database and see which patients do better: those who randomly got A or those who randomly got B. And if we see a trend over time that A is outperforming B, then we simply remove B as an option in the medical record and thereby improving the quality of care, pays on the dollar what it would have cost to do a clinical trial. That is the essence of the Point of Care Program.
Why is it transformative? It's transformative because we're learning from every patient encounter what works and what doesn't and how to take better care of the next patient. That's what transformative. Instead of doctors having to learn in their brains what seems to be working and what seems not to be working, we augment their brain with a computer system and we introduce randomization which allows for scientific reasoning and eliminates or reduces bias to the observational data alone.
Finally, let me say that once we know what's working better, A or B, we can simply turn off the inferior option and therefore decreasing the time it takes for a finding of importance like A is better than B, to be translated into clinical care. This time that it takes for an important finding can actually be translated into clinical care. It's called the T2 translation gap and it's due because doctors have to read this stuff, they have to digest, they have to believe it. It takes years and years for a finding to be finally absorbed and embraced by a community. You can reduce that time here to a simple writing of a line of code which no longer offers the inferior option. Will healthcare providers rebel against this? Well, not if they're part of the decision-making and have been informed about the A or B comparison in real time, which they would be because they're asked to consent to the process. In a minute or two, I can't possibly summarize what's transformative about it in sufficient detail. That's just an example of how we're thinking.
Ann Nguyen:
Finally, when you share your “Update of the Department of Veterans Affairs Precision Oncology (POP) Program” at the conference on April 6, what's the takeaway message for your audience?
Louis Fiore:
Well Ann, at that conference we're not going to have 2 minutes to summarize how transformative what we're doing is. We're going to have 25 minutes and then 5 minutes for questions and I embrace that opportunity. What I'll do is talk about the Precision Oncology Program and how it gets at the aspects that we've been discussing so far and how at the point of care you can empower clinicians to make informed decisions using data that's been aggregated over the past patients that that clinician has seen as well as any other clinician in the VA system and how we can have new offerings or treatments for these patients based on understanding from a database perspective what's worked and what hasn't and what should we try next. I'm going to wrap all that up. I'm going to talk about the social aspects of that. I'm going to talk about the scientific aspects about it and the informatics requirements to make that happen. And I think that the takeaway message for the audience -- I hope it's "Holy mackerel, there's a brand new world out there on how we can actually take care of patients learning as we go, and boy, the standard of care should be to offer patients novel treatments because we already know what works and what doesn't and now we're looking to determine what works even better."
I think that if those points are made, I think that most of the folks in the audience will understand how their particular niche applies to the big puzzle because this is teamwork. This is team science. I don't even want to use the word science. It's quality improvement all the way to science and the stakeholders that are required to make this work range from Chief Financial Officers who make decisions about supporting these kinds of activities in medical centers on one extreme to hardcore technicians on the other extreme who understand what the best way to move data around and keep it secure is. I’m just going to paint a landscape I think of where this is going to be in 5 or 10 years and it's not going to be theoretical. It's going to be grounded in a program that we're actually implementing. I firmly believe we're the only healthcare system in the country that could do it because of previous discussions, and I'm very proud to present where we're at. That's going to be my take-home messages. I guess they're going to be unique for each person there who understands how they can contribute to such an effort.
Ann Nguyen:
Thank you Lou. It sounds like there are going to be a lot of exciting developments in your space and we'll look forward to learning more from you when you're at the conference in the spring.
Louis Fiore:
Ann, thank you for giving me a few moments to talk about what's going to be presented and I look forward to this meeting as I have to past meetings. I find them to be enjoyable and informative, and networking and meeting new folks is always a highlight for me.
Ann Nguyen:
That was Dr. Louis Fiore of the Veterans Affairs Boston Healthcare System. He'll be speaking during a shared session for the Clinical Genomics and Cancer Informatics conferences (Tracks 10 and 12) at Bio-IT World Conference & Expo running April 5-7 in Boston. To learn more from him, visit www.bio-itworldexpo.com for registration info and enter the keycode “Podcast”.
This is Ann Nguyen. Bye and thank you for listening!