A Disruptive Approach To Optimize Workloads in a Crisis Environment
@ ESICM LIVES 2020
Matthias Johannes Merkel M.D., Ph.D, Sr. Associate Chief Medical Officer Capacity Management & Patient Flow, Oregon Health & Science University (OHSU) and Gerald Dunstan, MBA, PhD., Principal, GE Healthcare, UK
Hear from Matthias Johannes Merkel M.D., Ph.D, Sr. Associate Chief Medical Officer Capacity Management & Patient Flow, Oregon Health & Science University (OHSU) about the clinical and operational benefits that his organization experienced by moving capacity management to digital and its use during COVID-19.
Also, hear from Gerald Dunstan, MBA, PhD., Principal, GE Healthcare, UK, about how Real Time Data is Supporting Caregivers and Driving Outcomes.
So, thank you very much, Dr. Dunstan for this presentation. I believe we will have time at the end for a couple of questions, so I am now giving the mic to Professor Merkel. Again, that's a great privilege having you with us, Professor Merkel. Thank you very much for sharing your experience. Thank you.
Matthias Johannes Merkel M.D., Ph.D:
Thanks Marc. Thanks for having me. I will show you our journey here, or just you in the Pacific Northwest and the United States, and how we went from paper to digital and use it during this COVID-19 pandemic. I have nothing to disclose. I want to point out that our mission control center runs on a GE technology platform, as you just have seen. We using some of these tiles, and the reason why I got involved as an anesthesiologist and intensivist, is largely because in 2016, when we started our journey, the ICUs and admission to the ICUs became a bottleneck, and I happened to be there at that point, the ICU medical director.
As you go into this, it's really critically important to step back and think about what is patient flow and capacity management founded on. One thing is, what you see here is sort of a simplified version how patients come into the system as scheduled and unscheduled admissions. At large, a hospital exists of an ICU and acute care. Yes, we have to acknowledge there's a lot of variability around this and then patient discharges. The fitness of the errors here shows where our patients here at OHSU is the most part of that. In the hospital, we have essentially three elements we have to think about.
One is the staff, one is the space where we can place patients, and one is the time how patients spend in the hospital like the length of stay. The day to day and week to week variability in our sense is it is really contributed by the scheduled admission. So this is a really key, important component that we causing this to our system ourself. And then we have the unscheduled one, which kind of is much more predictable volume. So for example, at OHSU, we know that our emergency department on average produces one inpatient admissions per hours. So that is an important variable you have to build in as you manage your capacity and all around that.
So why did we create mission control? And one of this is really since 2012, we had engaged in several initiatives and you can see some of them resulted in publication. We worked on synchronizing how our transfers out of the ICUs in our acute care setting work across all units, surgical specialty units and medical ICU.
We worked with the surgical areas, the operating rooms, how the patient arrives into the ICU using our electronic healthcare record with automated pages so that we really used smart functions so that the right team was there, think of a formula one pit stop kind of approach. And then as an institution, we also created what we call a value stream as we transformed into reducing waste as our system and using lean methodology, like the Toyota principle in improving access to our system and our length of stay.
It's fair to say that we had successes in these pockets, but it was really focused on these areas. And we were lacking the large-scale progress. And then, to make matters worse, our problem persisted. Here, I'm going to let run a video which shows our acute care units. And it mimics our 2016 years. If we would have brought in every patient immediately after we accepted and placed it in the primary unit where we want this patient to be.
So the yellow is the patients who were accepted into our system or should have been accepted in our system. And here on the side, you can see how often we exceed the physical footprint in this unit.
It's fair to say that it's pretty clear this is not working, and we knew this. But when you look at some of the data in 2016, our average midnight census was 88% and we thought we were on fire and bursting. We knew that our problem is largely an acute care problem. We had patients boarding in the ICU resulting in an ICU decline, but if we would have had an acute care bed, we could have freed up an ICU bed.
We declined 568 transfers out of the state of Oregon and the Pacific Northwest in 2016. And we are a health system, which I explain you a little bit here, where we had two community hospitals, all of them in a radius of about 25 miles here in around the Portland city area who had an average midnight census of 40 to 60%. So it was clear we need to do better to fix this problem.
And how did we try to fix this problem in 2016? So this is a snapshot from October 12th, one of the many, many days that year, where we had a crisis. Our demand of inpatient beds was way higher than our actual available beds. So our bed flow team and our house supervisor, which is a nurse, went through the hospital and really counted how many beds do we have. They had it on a piece of paper and we transcribed it on a white board and we met, which commonly happens somewhere between 9 and 10 in the morning.
By the time this was written on the wall, it was already outdated because patients left the ICU, new patients were accepted. It was a very dynamic process. To highlight what our actions commonly were on this, our emergency room went on ambulance divert, we messaged in our transfer center that we are really extremely full and made a case by case decision.
We kept it open as a trauma. We are level one, one of two level one trauma centers, the highest trauma level of care you can get in our state. We try to never cancel but had to delay procedures because we couldn't really manage that. The most common intervention was meeting again at 2:45 in the afternoon, seeing how bad it really became. Everybody went off their duties and we hope for the best. And we did this several weeks in a row and realized that doing the same thing and expecting a different outcome is just not going to happen. So that's where we created our OHSU Mission Control.
It was very clear to us we need to really re-engineer our capacity management. We need to build what we call Mission Control. That's our command center. And after exploration, we decided to go before GE Technology. Our metrics what we wanted to achieve is really like we wanted to be able to accept all Oregonians who needed our level of care out of the state, either to our academic center or to our partner hospitals.
So instead of a white board, now we have this. It's a snapshot from early in November. It shows our current centers, available beds by acute care, critical care, and other specialty areas, adults here, women and children hospital down there. It also shows what we have in our two community partners here, one is the HMC and the other one is Adventist. And this gives us real time data and it's updated automatically out of our EMR documentation. And it's really helpful for the team to manage around that.
The other component is because we are so heavily relying on transfers and our state relies on us bringing in patients, we really created this visual map, which is a map of the state of Oregon. It shows you where a patient in our state distributed the referral location, and if you drill down, it shows you when was the patient accepted? How long are they waiting? Where is the referring hospital in relation to any of our hospitals here in our health system? And how long a transfer will help.
This allows the team, which is the most critical part, to really make decisions in real time, supporting the patient's need. And the team is a complex operation. So this our Mission Control team. Before we started, we had only the transfer center coordinator, the patient placement coordinator, a bed flow manager, and administrator on duty, but they were in different departments and divisions.
What we did is we put this all together, we connected it with our health system. We created these huddles several times a day, how these teams interact and use their knowledge and expertise to really run the health system as a four hospital system most effectively.
In order to be successful around this, it was clear for us we need some guiding principles. So this is a snapshot of pre-COVID of our transfer center that shows our big walls where we have these screens you have seen earlier. And the three guiding principles we use is utility, fairness, and fidelity. And this is really used every single day. And the more we moved into the pandemic, the more it became critically to really have guiding principles so that a diverse team of nurses here and a physician, so we have a physician in our command center 24/7. It's a group of 16 physicians from emergency physician to family medicine, to vascular surgeon, urology, intensivist. It's really a handpicked group who help us to manage that.
And you can anticipate the power and the signal what this team can give is if the vascular surgeon negotiates the next internal medicine patient placement because that patient tops out on the utility and needs to come in right now. We also went into the capacity management at the unit level. This is a snapshot where in each of our units, we have a monitor where the charge nurses can look at a drill down what's going on in the unit.
This is a makeup, but like it shows a new patient assigned to this patient who is supposed to discharge, and the order is already written. It shows that how long the wait time is, there's a color coding around that. It also shows some of the risk elements of patients who are existing in continuing their care. And some of the alerts you're showing like patients who will need some diagnostic images or IV therapy placement and things like that.
We also enhanced this with our business analytics team and we developed a forecasting tool, and this is a snapshot from just earlier this morning. It gives you the occupancy in our adult acute setting last week. This is today. And what do we expect it to be before that. This is a very good example, because as some of you might know, Thanksgiving is coming up, so it's essentially a four-day holiday. Thursday, Friday, Saturday, Sunday. So there's a dip on our elective scheduled admissions. This is based on machine learning going back to 2012.
Where this sometimes fails us, is that it isn't built into the pandemic. So, we need to fine tune that. What you see here in pink is the 80% upper confidence interval and you can see how often we actually would hit, or are at risk of hitting our physical capacity limit.
So this is really important for us to make decisions around the day-to-day operation and opposite to how capacity management was done in our institution from changing in the moment and accepting that we can't predict the future, we introduced tools to predict the future better, and we are able to change it. And what you can see here that last week by actively managing our elective scheduled admissions, we changed the trajectory and bent the curve here, which looks the same like this if you would go back.
So here's our success story. So each bar graph here on the left shows you the number of declines in our transfer center based on capacity. I mentioned in 2016, almost 600 patients. Each year, we've got better. We opened our Mission Control Center in December, 2017. So you can see gradual improvement, and then where it's fully operational, it has been substantially better.
These are fiscal years, which for us are from July until June the following year. At the same time, we increased our total patients transferred into our health system substantially up there. And when you calculate that, that equates to a 20 bed unit with 90% midnight occupancy. So a huge success for us.
And then this happened. The all-familiar coronavirus hit the world. In our state here in Oregon, we had the first hospitalized COVID-19 patients on February 28th. And we really went into a high gear mode with that, realizing that the Mission Control is becoming a central hub to manage the crisis for us as a health system.
We use discrete event simulation to better understand the potential impact early on, especially on the mandate by the governor office to stop all elective surgeries to make sure that we preserved PPE and have space for the initial perceived threat of COVID coming down into our hospital system. And then we switched gear very quickly to learn how to manage our capacity constraint in this new world with COVID-19, and I will show you about this a little bit more.
So, predicting the impact. So, the first two weeks we were working hard, we didn't have a single COVID-19 patient in our health system, but what we learned is that if you look at our elective, non-emergent cases who require ICU admission over the course of the week and acute care, we estimated that we can free about 5 to 13 patients per day in the ICU setting and could be around 50 patients a day on the acute care setting. You can also see that how our day to day variability was very present in this at the beginning of COVID-19.
So how did this play out for us? So, on this day here, we stopped our elective surgical procedure in March. And then we went into a modified operation with our emergency center on, and then here in beginning of May, we started to slowly rebuild it. So, we had the prediction and it happened, and the census actually fell because we did not need to replace our patients with COVID-19. And I will show you a graph in a little bit showing the COVID-19 hospitalization on the state of Oregon.
So we have done really well. And then we slowly built it up and you can see here, this goes until mid-November, our midnight census, both in acute care and critical care. So there's some variability, but overall, we rebuilt it and we remain extremely full as a hospital.
So as we learned the impact of COVID-19 and understood that our patients spent more time in the hospital, we have less staff available, and our acute care areas are reduced available. We use this discrete event simulation to look at what's going to happen.
So this is using our data to simulate how often we would get bottlenecks in acute care or in the ED and in the transfer center highlighted here for patients waiting and it shows down there. It shows the occupancy in acute care and critical care here. And it just shows how significant the pressure point is. And it became very clear that we need to change our approach to that and mitigate how the future will look like. So for us, this was the start to really more actively manage our scheduled admission so that we have a better handle on that.
We also brought in some new technology components to support how we manage this crisis. And for OHSU Mission Control, one is this COVID tile which shows you the COVID positive cases in our health system separated by the academic center here and our two community partners. It also shows how many ventilators at this given moment are used, it's updated in real time, and how the testing looks like from tests sent and positive tests over the course of more than a week. And then, we really asked the most critical questions, like how can we make all hospitals in the state of Oregon act as one hospital system during this pandemic?
This is a map of the state of Oregon, and it shows you the dimension of the state. It's often referred to as a urban to rural state because most of the population in our state is concentrated here about the Portland metro area. But it's a large area, actually very beautiful.
On the right side, you can see my daily commute here in the background, Mount Hood. In the state, we have 62 hospitals, 25 of them are critical access hospital, which means they have less or equal to 25 beds and a 24/7 staffed emergency room. We have eight large multi-hospital system, including OHSU. We have about 4.2 million people, a little bit above that. This is a 250,000 square kilometer area we serve. And we serve the entire state as the only academic center around that. And the PQC here in the background is our highest volcano 3,429 meters.
So the question was really early in this pandemic, how can we make all 62 hospital work as one health system for the benefit of every patient who needs us for COVID or for non-COVID?
We sketched out this in collaboration with GE Healthcare. And it's just like, okay, if we get the data feeds from all of these hospitals, electronic healthcare records, we can house them in our data center here in the academic center, we can use the tile technology to display it on desktops and any smart devices or in our Mission Control wall.
And we can display it as an entire state, in different region, which makes sense to health systems, to hospital down to the individual unit. And initially this was like, "Oh, it would be cool if we can do this." We were able to pull it off relatively quickly. Within four weeks, we had the initial data feed established and we have now this live data tracking available.
Here, you can see it sorted by the various region in our state, following the state regional trauma system. It shows you how many adult ICU beds, down to how many psychiatric beds we have, how many neonatal ICU beds, negative pressure rooms, and how many ventilators for the systems who were able to share that kind of data detail for us. As of November 19, we had 60 of 62 hospitals agree to participate.
The two hospitals who are not participating are technology in a situation where they can't currently share the data. All major health systems are actively participating, which was a critical piece to really make it functional for this pandemic. And we are now tracking almost 8,000 beds in real time which accounts for 92% of the bed in our state.
And then, we had to test it for a non-COVID 19 use. This is a five day series in late September. It's the same picture taken over five days, the air quality in the Portland area became the worst in the world. I'm sure a lot of you have read about it. It was unbelievable, and I've never seen something like that.
There were also health systems, hospitals affected and had to be evacuated. And relatively quickly, Mission Control at OHSU was put on the spot in guiding the entire state how we can do this. We are the regional resource hospital, which means that we manage the emergency room crowding with what we call soul management, where we distribute the ambulances across all the emergency rooms in our Portland metro area. We manage mass casualty incidents out of our Mission Control transfer center, but we also were immediately available to manage capacity across a larger region.
And we combined this with a critical resource team where we identified that if one region is under distress because they need to evacuate their resource hospital, we really wanted to be able to partner across this and coordinate this around that. And the tile allowed us to have a platform where we can look about how many patients do you have available in that.
So how did Oregon do during this COVID? Again, for the first two weeks after the first patient was there, we didn't have a single patient in our health system. And then slowly the admission increased here, it ebbed, and then there was a second wave, and now we are in the midst of a third wave here. And it's steadily climbing and we can see this, we track this on our COVID tile, and it's real, it's there, and we had to curtail elective admissions to remain open for emergency interventions, like stemmies and strokes and others like that. And the trauma care we provide for the entire state.
The Mission Control has been a central anchor point to manage this 24/7. So it has been really helpful for us.
Future opportunities for us in Mission Control is we are working with state regulators to transition the statewide tile to a permanent platform solution so that it's available for the long-term, everybody likes it, and we will advance along this. And within our health system, we're working on removing existing blind spots, such as nursing, staffing, and other critical parts or workload assessments in real time so that we have a holistic picture how we are doing right now, how we anticipated to do at the end of the shift, and how are we doing into next week, combining our data, our team, and the interaction around this.
With this, thank you for letting me share what we do here in Oregon. And I'm happy to answer any questions. And if anyone wants to reach out to me, this is my email I can be reached. Thank you very much.