Skip to main content
Videos

EHR Integration: Strategies to Enhance Precision Medicine

Featuring Nate Wade, PharmD, MBA, BCOP

In a session at the Association of Cancer Care Centers 50th Annual Meeting & Cancer Center Business Summit, Nate Wade, PharmD, MBA, BCOP, associate director for clinical oncology at Flatiron Health, spoke about the role of electronic health records in precision medicine, as part of a session titled “Deep Dive: EHR Integration: A Key Component of Precision Medicine.”


Transcript: 

Nate Wade, PharmD, MBA, BCOP: Hey everyone, my name is Nate Wade. I'm a pharmacist. I'm associate director for clinical oncology at Flatiron Health.

Please give us a brief summary of the ideas you will be discussing during the “Deep Dive: EHR Integration: A Key Component of Precision Medicine” session at AMCCBS.

Dr Wade: Precision medicine is a pretty broad topic. And even when you narrow it down to just technology that can enable precision medicine. There's still a lot of ground to cover when I think about precision medicine, the way I think about tools for precision medicine is really it's not a single tool, but it what it is is multiple tools that kind of build on and then complement each other. When you think about precision medicine at the point of care, it's really a spectrum or a timeline. It starts with the clinician's decision to order a test, and then actually ordering that test, receiving the results, interpreting those results, and then finally taking some sort of action, whether it's initiating therapy or enrolling a patient in a clinical trial. And so in in a perfect world, we would have technology for each step in that process. In that process, full of precision medicine. In in a perfect world, we're going to have a tool or a feature within the EHR that will help the clinician either become more efficient or deliver better care at each time point within precision medicine.

So what I'm planning on talking about is 3 different things, really. The first bucket that I'll talk about is laying the groundwork to enable that suite of tools that we can provide along the timeline of precision medicine. So what that means is we're talking about integrations directly with NGS testing labs so that a clinician can order a test and then receive the results directly back into the EHR. And then also our process for normalizing and harmonizing structured data biomarker results so that they can be used and consumed by downstream products. In those 2 things we chose to start with those 2 things, and they because they really enable all of the downstream products that we want to do.

Then the the next topic that I plan on covering is the the actual products and features within the EHR that will help clinicians make decisions. And there's 2 things that in particular, that we'll talk about. Number one is Flatiron’s clinical trials matching tool that takes those structured biomarker results directly from the lab and then integrates them into the clinical trials matching tool, so that a clinician or trials coordinator can screen patients more effectively and look for open clinical trials that the patient may be eligible for. And then the second thing is helping the clinician select appropriate therapy based on the patient's disease, their stage, and any biomarkers that they have.

And then, finally, the last topic that we'll cover, hopefully it's the most exciting topic, is what can we do next? What is Flatiron working on. That's also going to help enable precision medicine, and that could cover a myriad of topics. We're looking at products that will help a clinician make sure that each patient is tested appropriately and at the right time. So they receive the right test and then interpret those results correctly. And then we're also looking at other ways that we can use those structured biomarker results to make the clinician more efficient, whether that's flowing into the clinical decision support tool or just building other types of features that will help alert a physician, that of a patient that is newly eligible for a new drug based on a certain biomarker that might have been tested 6 months or a year ago. So there's a lot of ground to cover and things that we can talk about. It's an exciting time to work in precision medicine and technology, because, like I said, we're getting close to that, to that vision of having a tool at each point in the timeline to help enable clinicians to provide better care. We're getting there. We've kind of accomplished it in some ways, but there's still a lot of work that remains.

Can you share specific examples of successful EHR integration initiatives that have improved the identification of suitable patients for precision medicine testing in the context of cancer care?

Dr Wade: Yeah, absolutely. It's a great question. I really like the way that you framed it, actually, by saying identifying suitable patients. And the word “suitable” is really key to me. You know. In my opinion, a very broad definition of a suitable patient could be just, you know, they have a diagnosis, and they have a stage, and here is the appropriate testing recommendation. And that's kind of the most simple way that you can identify patients for testing. And that's relatively easy to do. That technology already exists in our EHR, Onco EMR, and of course, EPIC and others kinda have similar types of alerts. If you ask me the word suitable goes a little bit deeper than that.

If you show an alert or a recommendation for every single patient that is diagnosed in a certain cohort of patients, you run the risk of over alerting and driving alert fatigue and for the clinician, and then, of course, if you have alert fatigue, then your alert is eventually gonna become less effective. So one thing that we've had successive one thing that we've had success in lately is building a machine learning model that understands whether a patient has had a full NGS panel run on them, run on their tumor, we've integrated that in our clinical trials matching tool, and that alerts the trials coordinator that says, Hey, this patient has an NGS testing result. You wanna make sure you double check that to make sure you're screening for all possible clinical trials. And the machine learning algorithm, it sends a result and it gives a confidence level, for whether or not it thinks that the patient has NGS testing results.

So you can see where I'm going with this. You know, I think that one thing that we're exploring is whether we can use that same machine learning algorithm or something very similar to suppress an alert for a patient that is being diagnosed with a standard of care, so that we can only show patients that truly need to see this alert. We can only show the alert for physicians that truly need to see it, not ones that have already been tested before. So you know, again using these types of alerts can be tricky, and it's our goal to make the alert as smart and as accurate as possible, and only show an alert or a nudge, or whatever it looks like when it's truly needed, and not for every single patient that is diagnosed with a certain disease.

Given the common challenge of molecular testing results being received as PDFs, what innovative approaches or solutions have you observed or implemented to ensure seamless integration of these results into EHRs?

Dr Wade: We've heard a lot about this lately, honestly over the past several years, and that is receiving these results as a PDF outside of the EHR is really problematic, it makes it difficult for the user to find it when they need it, and especially call up results that, from the past, that they might not be right at the top of mind for the for the clinician. So the first step to addressing that problem is what we've already done, is it by integrating with other labs to place the order in the EHR and then receive the PDF results directly back into the EHR. That solves one aspect of the problem. Just getting it right into the EHR, right? But we can't stop there. We're also receiving, we're going to begin receiving structured biomarker results, which, as I mentioned before, kind of unlocks a lot of other potential tools that could be used to enhance precision medicine. 

We hear a lot about the PDFs and how they're not standardized across vendors, and how information may not be displayed in in the best way, or it's it can be difficult to navigate some of those PDFs. I do wanna make sure that we give the labs credit, because I think we've heard that feedback, I'm sure they've heard that feedback as well, and I think they responded to some of that really valid feedback. As I look at some of the reports that that our NGS testing vendors are sending nowadays, they are much easier to read, and they do convey really important information. And so I'm not necessarily convinced that it's the EHR's job, or it's the EHR's role to make sure that everything is displayed in a standardized way. I think the best people to display the information are the labs that generated that information themselves.

So as we think about ways that we can enhance the clinician experience, reading those PDF results. Number one is getting it into the Hr. And we're mostly there. I think number 2 is just making it easier to find within the EHR and find the information that you're looking for. So we've come up with a couple of creative ways. They're relatively simple, but we found some ways to make sure that the clinician is able to find those results where and they're aware of those results when they do exist. So first is putting it right on the page. We call it the Treatment Plan page. But it's where clinicians spend the vast majority of their time. It's where they approve drug orders. They see treatment history over time. They see labs. And so we put it where the clinician is spending most of their time.

In addition to that, we also decided to make it easier to search for those types of documents. So obviously, it's pretty simple to be able to search for the name of a document, you just search for the name of the lab. We took it a step further by making sure that you could search the entire contents of the PDF document. So if you think maybe this patient, they might have had an EGFR mutation. They might be wild type. Let me double check that. You can search in our EHR. Search for EGFR, and it will automatically pull up all of the documents that contain that gene or any other gene that you might search for. So our goal was not necessarily to replace the PDF. But just make it more convenient to use. I think there's always gonna be a role for the for the PDF in addition to the structured biomarker results that we're gonna begin to use shortly.

Last year, identifying the right patient for the right testing was highlighted as a major barrier for physicians. Have there been advancements or strategies that effectively address this concern, and how have they impacted the overall process of precision medicine integration?

Dr Wade: We've heard this a ton, obviously, in the past few years. There have been advancements. I think there's still a lot of work to do, to be perfectly honest. I did mention previously about the machine learning algorithm that will build a really smart tool. We want to make sure that whatever we build and whatever the clinician sees is very effective, and they only see it at the right time for the appropriate patient. 

So when we think about the way to best implement that, the easiest way is to start with non–small cell lung cancer. And you can start with a relatively simple alert that just displays the the biomarker testing recommendations for metastatic or early stage non–small cell lung cancer immediately after the after the patient is diagnosed and staged. So so that's one way to do it. Again, I think I'd like to take it a step further, and as we expand that to different diseases, becomes much more granular. The testing recommendations come down to specific biomarkers, not just recommending a broad NGS panel for every patient. So we're making progress. We're exploring ways to make sure that we're building a really smart, alert you know, nothing is set in stone. We're still in the very early scoping stages of this, but we are making progress on how to improve that, because we heard that feedback loud and clear, from not just folks at ACCC but many other stakeholders as well.

Collaboration between EHR developers, testing labs, and providers is essential for overcoming challenges in precision medicine integration. Can you provide examples of successful collaborative efforts that have significantly improved the interoperability between these stakeholders?

Dr Wade: I think the easiest example of collaboration among stakeholders is the HL7 working group to develop the FHIR genomics standards. It's a really fantastic group of people that both organize in it or organize this group, and many stakeholders that participate in it, Flatiron has been a participant in the past to help develop these standards. And it does provide a lot of opportunities for collaboration. So that that's a really easy example to provide.

When we first started building our precision medicine program at Flatiron, and in building these integrations, the HL7 FHIR standards for genomics were really still in their infancy, and anybody will tell you, including them themselves, that they really weren't ready for prime time yet they needed quite a bit of work in collaboration to make sure that they met all the necessary use cases. We're getting much closer now. It's really exciting now, and I'll give credit both to HL7 working group and several of the NGS testing companies have approached us, and we've had preliminary conversations about changing over from the existing data schema for biomarker results over to a more FHIR-like standard. And so we're not quite there yet, we haven't implemented anything just yet, but those standards are there. I'd say they're ready for prime time, and by the end of the year we'll probably be live with several different labs using FHIR standards. So once the vast majority of stakeholders adopt those standards, it's going be really, really easy to make enhancements and build tools that will be able to consume those biomarker results.

© 2024 HMP Global. All Rights Reserved.
Any views and opinions expressed are those of the author(s) and/or participants and do not necessarily reflect the views, policy, or position of the Journal of Clinical Pathways or HMP Global, their employees, and affiliates.