[music playing]

Oby: From the campus of Harvard Medical School, this is ThinkResearch, a podcast devoted to the stories behind clinical research. I'm Oby.

Brendan: And I'm Brendan, and we are your hosts. ThinkResearch is brought to you by Harvard Catalyst, Harvard University's Clinical and Translational Science Center.

Oby: And by NCATS, the National Center for Advancing Translational Sciences.

Brendan: During medical training at Massachusetts General Hospital, Erica Shenoy saw how infection control protocols, necessary to reduce the risk of hospital transmission of infection, could also impact patients in hospital units in unexpected ways. Patients with a history of infection with MRSA, a bacteria that requires patient isolation, were often delayed in moving out of acute care and into the rehabilitation facilities they required. Leading up to the COVID-19 pandemic, much of Dr. Shenoy's work on deisolation of patients with MRSA primed her to address both the implications of the need to isolate large numbers of patients as well as the urgency to deisolate patients safely to ensure appropriate care and continued hospital capacity.

In the spring of 2020 during the peak of the COVID-19 pandemic, physicians and infection preventionists were working 18-hour days determining who should be isolated and who didn't need to be. To improve hospital operations and patient care, Dr. Shenoy and colleagues developed an algorithm based tool to facilitate appropriate isolation and deisolation for patients being worked up for possible COVID-19 infection. This tool, called CORAL, short for COvid Risk cALculator, resulted in a 95% reduction in physician workload and has since expanded to outpatient locations and is in use at 12 additional facilities across the Mass General Brigham health care system. Dr. Erica Shenoy is the associate chief of infection control at Massachusetts General Hospital. Well, Dr. Shenoy, thank you very much for joining us, and welcome to ThinkResearch.

Erica Shenoy: It's great to be here. Thanks for having me.

Brendan: So you are the associate chief of the infection control unit at Mass General Hospital, and your research focuses on hospital operations and infection control. Can you tell us what that means?

Erica Shenoy: Sure, so infection control, first of all, is very different than infectious diseases. I am an infectious diseases physician, but in my role in infection control, much of what we do intersects in many ways with hospital operations. So for example, when you have patients who have either suspected or confirmed infections that need specific isolation or use of personal protective equipment, that can mean that they are placed, for example, in particular parts of the hospital or certain sorts of rooms, or if the infection is confirmed, we can sometimes cohort those patients with another patient who has the same infection when we have double rooms.

And the reason I talk about that is because that has lots of implications for the capacity of the hospital. So do we use our beds in the most efficient way possible and also use of resources, training of frontline providers, clinicians, and support folks to implement the correct infection control? So if I give an example of how this might affect operations, one example is an infection called MRSA, which is methicillin-resistant Staph aureus. And patients can be infected with it, and they can also be what we call colonized. So they might have it for a while on their skin.

If a patient comes in and we need to match them and we need to put them into a room where there's someone else who also has MRSA in order to use our beds efficiently, we have to match them on the gender-- so we're going to put male-male, female-female-- and also some of their other needs. If they need orthopedic care, they might be on orthopedic floor versus another sort of floor. So you can end up-- because of infection control policies and procedures, you can end up with blocked beds in a hospital at a time where every bed is needed to meet our capacity demands, but we've got some blocked beds because we haven't been able to do that matching piece. And we have blocked beds in the hospital. So that's kind of where the intersection between my interests as infectious diseases, how long do people really remain colonized or infected, intersects with what does that mean in terms of the use of resources within the hospital as we're implementing policies and procedures to prevent possible transmission in the hospital.

Brendan: And this is an area that you became interested in as a resident. Can you tell us about how you became interested in this intersection?

Erica Shenoy: Yeah, so there were some formative experiences. And I do remember one very specific one. I was as a resident very early on interested in infectious diseases, so I knew I was going towards infectious diseases. But I didn't really have any, I don't think, inkling, and as I learn more about infection control, I realize how little I knew about infection control at that point. I basically thought it was follow the sign. Wear the gowns and gloves and whatever when they tell you, and that's what I thought of infection control.

But when I was a resident, one of the jobs on our particular team was to try to arrange for patients to be discharged from the hospital, and there was, just like there is now again, capacity constraints. We need to make sure we have beds available for patients coming into the hospital, and there was an urgency to make sure that we can get patients, for example, to a nursing facility or rehabilitation facility. And I remember in that role that I had, my job that day was to go through a patient list and figure out who was going home, who was going to facilities.

And there was a patient who had been on our service for days, and we were actually doing very little for that patient from a medical perspective. So I was talking with case management as to why that patient couldn't leave the hospital to go to the rehab, and the case manager said, well, look, they've got this label of an infection. And because of that, at the rehab, they need to figure out if they can cohort this patient, and when I have a who has one of those labels, she told me, I have to refer them to many more rehab facilities or nursing homes because the chance of them getting a bed is much less.

And so this patient had a label on them. It was a label of a particular infection. It got me thinking from the kind of clinical perspective how long do patients remain what we call colonized with specific organisms and also the operational impact that this was having because this patient needed to move on. They needed rehabilitation care. They didn't need acute medical care, but they were stuck with us until we could get them to the right place.

And that bed could be used for other patients, for example, waiting for admission to the hospital. So that really got me really interested in both the clinical piece and the impact on the hospital, and so I reached out to one of the ID physicians. And I said, who can I talk to about infection control? Because I really want to learn more about this and how we can learn about this and perhaps improve our current state.

Brendan: And so you started learning more, and what was your next step after that? Where did that take you?

Erica Shenoy: Well, I started-- I met with the person who ended up being my mentor here, Dr. David Hooper, and learning about one particular infection, MRSA, and how long patients remain colonized and over the next many years went through several studies looking at duration of colonization, rapid screening for patients. We did a randomized controlled trial. As a fellow, I led that, and then we did an implementation study.

And we got to the point where we were thinking about keeping people isolated for the time that was required, but as soon as possible, when they did no longer need to be isolated, the importance of deisolation and opening up beds and having people labeled correctly. And I think that really primed me in terms of the understanding, the practical implications of infection control, and specifically these labels that we put on patients related to infections that they have, and then eventually, they will be cured of the infection but may still have that label on them. And also, a lot of the working relationships with operations folks and clinical folks on the ground so that when COVID came around in the spring and we were really faced with a crunch in terms of capacity in the hospital and trying to understand how long patients, as you were working them up for COVID, needed to remain isolated until you took COVID off the differential-- how to think about that operationally but also to leverage some of the informatics approaches we had to work on that.

And the end result, which I think we're going to talk about in terms of CORAL, I think it was above and beyond what I thought was possible. But in the situation of a pandemic with a lot of really smart people working on a problem, I think we were able to create something pretty interesting. It's actually opened my eyes, and I'm coming back to the MRSA question and thinking about how I can do that better in the future.

Brendan: What did it look like for you at the beginning of the pandemic, and how did that lead to what you created, which is the software tool called CORAL?

Erica Shenoy: Yep, so we take us back to March 2020, and that's when things started just heating up quite a bit here in terms of the volume of patients who came in as suspected COVID. At the time, we had a lot of constraints on our personal protective equipment, so we needed to conserve that. We had constraints on our testing supply and the turnaround time related with tests for COVID.

We also know-- and this has not changed-- that our tests are imperfect. So a test for COVID or these PCR tests, for example, they're not perfect. They're quite good. In a hospital setting, there's actually quite high stakes if it's a false negative, which you know that there will be at some level.

At the same time, keeping patients in isolation because they can't be cohort for a long period of time is also high stakes not just for that patient, because being in isolation may delay the sorts of care that they might need. In some ways, that can happen, but also, because we need to make sure that we're using every bed wisely. The second piece that was going on during that time is we were just learning so much about the epidemiology of COVID, what the clinical presentation was like, how to work patients up, and it was just evolving very quickly. And so the way we were managing patients and deciding the workup was done-- a single negative test, and we're done with the workup or if we need to do additional testings.

We had actual infectious disease physicians reviewing every single chart every day. They were working from 6:00 AM to midnight. Basically, these are clinicians of ID-trained physicians, some of the most senior physicians in our division all working together, having three-times-daily kind of rounding where they presented the more challenging cases, and based on their review, the decision was you're done with the workup. You can deisolate the patient. Or you need to do some additional steps, and that might be testing the patient, getting some additional imaging or some additional diagnostic.

It was incredibly labor intensive. I mean, even with that kind of work effort, in the end, we calculated over 5,000 hours of physician time, and that didn't even incorporate the infection control practitioners who were joining the rounds and helping implement the recommendations. It still resulted in, at least our institution, about a hundred patients in the hospital each day who were in that suspect COVID category, and it was an unsustainable sort of activity.

And we were kind of getting through this first peak, and we were learning, too. And that was one of the really exciting things about this work, is that the teams and the clinicians got really, really good at sorting through the patients and figuring out in this patient with this epidemiology, this symptomatology, this negative test is good. We're done with the workup. We can move on. And then in these patients, we really need some additional diagnostics to sort out for sure that they don't have COVID.

And based on all of that work, we developed as a team a scoring system that we would score these various features-- the epidemiology, the symptoms, the combination of symptoms, the imaging findings. And in the background, that score would say, if the score is less than this amount, the workup is done. Deisolate. Take them off of precautions and resolve that label on their chart.

If they were over this amount, we need to do another step in the workup, and here's what the next step of the workup is. And it became protocolized. And then we built this in the electronic health record, so the clinician, by answering a few questions-- and it would pull in the results that we knew about the patient automatically-- it would tell them at the end of answering these questions right there at the bedside in the chart of the patient what the next step was.

And it could be done 24/7, didn't require infection control or infectious disease to review the chart with the exception of when CORAL said, you've done the entire workup. This is still a really perplexing case, a challenging case, a complicated case. We're going to take this case and have infectious diseases review it and do that focused chart review, so that instead of reviewing hundreds of the cases, we ended up reviewing 5% of the cases.

Brendan: Right, and that's, I mean, obvious. To take 95% of the human labor out of any process like that is a major accomplishment. So I mean, besides that aspect of just freeing up people to not have to work 18 hours a day anymore, what do you think is the biggest impact? What has been the biggest impact of the introduction of this tool?

Erica Shenoy: Well, certainly, I mean, people are attracted by the fact that we reduced that labor. I mean, that labor had to be used elsewhere, so we were not going to be able to support that sort of process. But there are many other advantages to CORAL. So we studied it in the pre-post first surge, and we looked at stewardship of the laboratory resources. So we found that post CORAL, we were doing second tests less often, and we were doing additional diagnostics less often.

We did not see any increase in patients testing positive after CORAL had resolved them, so we looked for some safety signals to make sure that CORAL was at least as good as the clinicians. But some of the major impact, I think, is really at the patient and the clinician level. So first of all, the patients-- we found a reduction in the time from when the test came back to them being isolated. So pre-CORAL, on average, they were in this kind of limbo land of being worked up of on average about 36 hours. After CORAL, it went down to about 19 hours.

So that's a substantial reduction. That opens up beds, and it increases capacity. And it streamlines the workflow.

The second piece is something that I didn't really anticipate but is pretty exciting, which is that it was actually a teaching tool. So the clinicians at the bedside in a way, because this was built from the ID workup perspective, how an infectious disease physician thinks about a workup, they were actually doing the workup by answering the questions. And because they got used to the sorts of questions we were going to ask, I think that in their evaluation of the patient, they got pretty routine to think about where they were coming from-- was it a nursing home? Was it a private home-- getting the symptoms down.

And I think that piece has been really important, but also, because the way we built it and the way we've maintained it over time is that we have modified the logic in the background. As the epidemiology has changed, as vaccines have come into the picture, we have started incorporating that into the background logic, and it doesn't require the frontline clinicians to learn everything over again or to learn things. When things are changing, we can change the scoring in the background and make these modifications that change kind of the pendulum of who's going to pass and who's going to need more additional testing because we're following the science. We're following the public health recommendations, and really, at the bedside, it is more seamless.

We've also gotten a lot of feedback from the frontline users who are reporting things. Well, when I try to do CORAL, it did this, or why is CORAL doing that? And that has been kind of a learning process and allowed us to improve the user interface when possible, reduce clicks.

And we expanded it over time to include many different hospitals in our systems, so community hospitals, small hospitals that maybe have 20 beds, to larger community hospitals. We brought this into the ambulatory environment, so primary care docs and others working up patients can use it. We built this to help people decide when to take the label off for COVID patients when they're deisolated.

Public health recommendations have been changing the entire time, and it's really hard to keep track of it. But we keep track of it for you in CORAL, so you just work through. And that you're applying the latest guidance at that time.

So I think part of it brought the information, and the piece of information that was coming through is like a fire hose. And with everything going on at the front line, at the delivery point of care, I feel like CORAL is able to shield the clinicians from it and kind of just ask them the most relevant information and help make that decision for them. And they can do it 24/7, and so that, I think, streamlines things and really from our perspective has improved the decision making at the bedside.

And at this point, we looked last week, as one of my colleagues was presenting some of this data, and it's been used over 100,000 times. And that was a little bit mind boggling to me, but it's in use every single day. I mean, people use the phrase, did you coral them? Yes, I coraled them. And so it's really gotten into both the workflow and, I guess, the vernacular.

Brendan: Yeah, I was going to ask how much it's being used, and I think once your noun becomes a verb, then you've really caught on. Who says, I'm going to search the internet for something? It's, I'm going to google it.

Erica Shenoy: Right.

Brendan: One of the other questions I wanted to ask was how what you're doing now compares to what you're doing before the pandemic. I wonder if you could talk about kind of that idea, the pandemic as a springboard for innovation that helps the emergency that we're in but also kind of the pre-existing and continuing issues that are outside of the scope of the pandemic.

Erica Shenoy: Sure. Well, I mean, we started talking a little bit about MRSA. That's how I got into a lot of this, and actually, coming full circle, I think CORAL-- I guess I could encapsulate it that when we were rolling out CORAL-- and one of the things we did with CORAL, it was like, CORAL starts today.

We are turning off the pager. You cannot page infectious disease. It's over. This is the tool.

And when we showed it, some of the clinicians were like, I can't believe we can do that in the electronic health records. There were bells and whistles. Now there are a lot of bells and whistles, but even then, it was seamless. And it really captured that there's a lot that we can do in the electronic health record to support clinicians and make things easy and do it in a very efficient way.

And so coming back to MRSA, we started developing ways to have the frontline clinicians do that evaluation for the label. We have had policies in place for a while. They say if they haven't had it for this long, you can get this screening test, and you can do that. And then you can ask infection control to remove the label.

It's very hard to get that done. Everyone is so busy, but if you can truly integrate it into the workflow-- and then you get the benefit, which is if you, as the frontline clinician-- you've done what you get prompted to do. And at that moment, you click the button, and it takes the MRSA label off. And you can put them in a regular patient room. I think there is kind of a feedback of this being really efficient and beneficial to the patient and beneficial to my workflows.

And so we are in the process right now of developing-- we have a name for it. It's another acronym, but we're still debating. And so I don't know if it's going to be the final acronym to make it catchy for the frontline clinicians-- that would essentially, for patients presenting with MRSA, it would say, the person is eligible.

This person is eligible. They haven't had MRSA in a while. Here's the test to order. When it comes back, type dot whatever in the system, and it's going to lead you through those simple questions. And at the bottom, you click the button. If it makes all the checks, we're going to take that MRSA label off your patient, take the isolation off. And so we're doing that with MRSA, and probably by the time this airs, we'll have that live in production.

But we're also thinking about it more in workups that are pretty standard to help the clinician to triage things and take you as far as you can go in the workup. So an example might be if you suspect a patient has tuberculosis, and you know the tests you generally need to order. But if we were able to integrate some of the epidemiological risk factors into an evaluation, we could then tell you at the edge, you know what? This person should see infectious diseases.

Or you know what? You've completed the workup. You're done with that workup.

And there are lots of opportunities where there are guidelines, but it's hard to build that into your normal workflow and to steward people through the right laboratory tests. And part of this work has really also had this added benefit of instructing people, explaining to people what the next step in the workup is, and it amplifies kind of the infectious disease and the infection control expertise by bringing that right into the frontline clinicians. And then it gets to us eventually, but we get to that 5% or whatever that number is where it makes sense at that point for infectious diseases or infection control to be involved. I think that has really good patient potential and provider-clinician satisfaction because it's integrated into the workflow.

Let's say you've worked up a patient for COVID. You've got your negative first test. They did have some symptoms. Maybe they had some epidemiology that might make you think that you might need a second test.

You're in the patient chart. There's a blank note section, and you literally type a period and CORAL. And up pops a form with little checkboxes or things to click off.

It asks you, where do they live? Click off these symptoms. What was the chest X-ray?

And in your screen, it pops up the recommendation. And if it says you're done, you sign it. It documents that in the chart, and the patient's deisolated.

If it says, you know what? You need to keep working this patient up. It actually presents you with the next order, and you can keep going in the workup. So that's the way, I think, the seamlessness of this could be really useful to clinicians. And then if the guidelines change, we can update the algorithm in the background, and everyone is implementing the most up-to-date workup for that patient.

Brendan: Yeah, it sounds like this is why the EHR exists is because it's a tool. Everybody's using the same tool. We're not relying on one person's knowledge and a policy that not everybody's read.

Everybody goes through the same workflow, and by integrating it all into this system that the entire hospital is using, everybody has the same information and is doing the exact same steps. It's not like, oh, I happen to have a fellowship in infectious disease, so I know how to do this. It's really amazing to see that this is possible.

Erica Shenoy: I mean there's no doubt that the EHR can be burdensome.

Brendan: I'm sure it is.

Erica Shenoy: I know that as a frontline clinician.

Brendan: Yeah, I'm sure it is. Yeah.

Erica Shenoy: I think one of the principles, too-- in the team that's been developing it, we've got programmers. We've got a lot of clinicians, and one thing we try to do is we were doing our bells and whistles was to reduce clicks. And so any time we're thinking, oh, we would really like to incorporate this question, we're thinking, OK, that's adding a click. How do we take away a click? How do we make this as user friendly as possible?

I think the second piece is I don't want people to think that it's so protocolized that we don't capture the individuality of any one scenario. In fact, our algorithm, at least for CORAL, it's using combinations of the features that are being documented by the clinician to make the ultimate decision, but also, there's always this pop-up valve.

And the pop-up valve is we can't in CORAL say for sure that you're done the workup. We're going to bring you to the next step, which is you raise it to the level of the ID physician. If you think about the EHR, you want to be augmenting the frontline provider to be working at their highest level and all the support functions to be working at the highest level so that people are using their brains the way they're meant to use their brains and not doing a lot of busy work or doing things that could be protocolized.

So I think a lot of the support for CORAL has come because frontline clinicians give us feedback. We think it through. We incorporate when it makes sense. We give that feedback back.

So it's a living system, and it's obviously had to grow over the course of the pandemic in ways and try to make-- when we see something that people are doing where they're kind of giving us feedback over and over, it's the same topic. We're like, OK, this is a pain point. This is something we can work on to try to fix for people when we can.

And I think if we could do it for COVID with the barrage of new information that comes all the time and make it work in that sort of system, then that opens a possibility for more routine things that actually still have a lot of impact on clinicians and patients but where we can have pretty simple approaches that leverage all the information that's in the EHR, bake it into the workflow, so it's really seamless and that you see the benefit up front as you're doing it.

Brendan: Yeah, so what do you see as future applications of a tool like this?

Erica Shenoy: So I mentioned we're doing MRSA. It's funny. At a staff meeting the other day in infection control, one of our infection preventionists said, I keep seeing people getting an extra test for this infection. Is there any way we can develop something like CORAL to guide them through when that next test is necessary? Because it has all these unintended consequences, and there are guidelines about when you should get the second test.

And I'm thinking thinking, ding ding, OK, that's our next disease on the list. But there's certainly a lot of opportunities here. We're studying it. So there's a couple of studies that have been published from our team looking at false negative tests and how we can modify our algorithm. We're doing some additional analyzes.

I am really interested in outside of the acute care setting in ambulatory or outpatient care how we can use systems like this where there's not a lot of infection prevention support in ambulatory. It's kind of one of those areas where there are resources within the main hospitals, but a lot of clinics out there don't really have that sort of support. And if we can both build infection control and infectious diseases support within the EHR, maybe we can really help those kinds of environments in a way that perhaps they haven't been able to have access to those resources before. So I'm interested in not just acute care but ambulatory.

And in fact, we've got CORAL operating and four rehab facilities as well right now, so we're learning about how it's used in different sorts of patient populations and then, I guess, as I mentioned, expanding to other sorts of infections and kind of workups in this area. I kind of mentioned the team throughout, and I have to say, it is a very special team. We have a combination of infectious disease physicians. We've got programmers, some of whom are also clinicians, and they build within the electronic health record, the infection preventionists who are kind of the front lines of hearing lots of questions that come from the frontline.

And it's exciting because it works, and it's useful. It's got patient improvement. It's got clinician and operations improvement. So I've really learned a lot from the team, and it's opened my eyes to some of the opportunities that we have to make the system work better for everyone.

Brendan: Great. Well, Dr. Erica Shenoy, the associate chief of infection control at Mass General Hospital, thank you very much. It was a pleasure to have this conversation with you.

Erica Shenoy: Thanks so much.

[music playing]

Brendan: Thank you for listening. If you've enjoyed this podcast, please rate us on iTunes and help us spread the word about the amazing research taking place across the Harvard community.

Oby: To learn more about the guests on this episode, visit our website, catalyst.harvard.edu/thinkresearch.