Webinar Transcript: Inpatient bed capacity management – unlock bed capacity and reduce LOS (UCHealth)

At the LeanTaaS Transform Winter 2021 event, UCHealth Director of Operational Intelligence Isobel Handler, MBA, MHA, joined LeanTaaS Director of Client Services Pallabi Sanyal-Dey to discuss her flagship hospital’s journey in implementing iQueue for Inpatient Beds to precisely identify and address their roadblocks to effective inpatient bed capacity management. 

View the whole session, or read a blog summary.

Moderator: Welcome to Becker’s and LeanTaaS Transform: Better Healthcare Through Math Hospital Operations Virtual Summit. We hope you’ve enjoyed this event so far and we’re thrilled to have you join us for our next session, “Inpatient Bed Capacity Management: Unlock Bed Capacity and Reduce Length of Stay.” 

Before we get in, I’d like to cover a couple of housekeeping items. We’ll have time at the end of the presentation for questions and an answer session. You can submit any questions that you have in the q&a box that’s on your dashboard. Today’s session is being recorded and it will be available after the event to you on demand. Iif at any time you don’t see your slides moving or if you have any trouble with the audio our recommendation is just to try refreshing your browser. And so hopefully that’ll kind of help keep everything back in pace. And lastly, you’ll find a few other engagement tools on your dashboard. So be sure to check out the resources section and fill out the event survey. 

At this time it is now my pleasure to introduce our speaker for today. Isobel Handler is the director of operational intelligence at UCHealth, where she develops system level analytics to support the COVID 19 pandemic response, vaccine distribution, ambulatory patient access and inpatient capacity management, among other areas. Since joining UCHealth in 2017, Isobel has held various positions in clinical resource management, data analytics, and special projects. She has a Bachelor of Arts degree from Wellesley College and both an MBA and MHA from Cornell University. 

With that, I am pleased to turn the floor over to Isobel to begin our presentation. 

Handler: Thank you so much for that introduction. Today I’m going to be talking about how at UCHealth we addressed several challenges that we have with inpatient capacity management, using the LeanTaaS iQueue tool for beds, and I’m going to kind of organize those challenges in three groups. The first is how we were challenged with making consistent and data driven decisions when we didn’t have a lot of data, and how we ended up oftentimes making decisions more based on anecdote and feel, which led us to be kind of inconsistent with how we manage capacity depending on who’s working in a given day. 

The second challenge is around our habit of kind of being in crisis mode as it relates to capacity management, rather than being more proactive in managing capacity ahead of an emergency. And the third area is this optimization problem that a lot of hospitals struggle with where we’re trying to find the right bed for the right patient, among a very clinically diverse patient population, and then a hospital with many different types of beds and services that we can offer. 

Before I jump into those, I wanted to share a little bit about UCHealth as a system. So we’re based in Colorado, where we have 12 hospitals across the front range and almost 2000 beds among those hospitals. We have almost 140,000 inpatient visits, inpatient admissions, and observation visits per year, and you can see a few other operational metrics on this slide. We’re very proud of the recognition that we get both locally and nationally for our work in these hospitals. Our University Hospital is ranked number one in the state of Colorado and we have several others that are in the top 15. Three of our largest hospitals are magnet designated and we also have several quality-related accolades, for which we’re very proud. 

This is a map of how our hospitals are distributed throughout the state of Colorado, and we have divided ourselves into three regions, the northern Colorado region, Metro Denver and Colorado Springs. For most of today’s presentation we’ll be focused on the University of Colorado Hospital, which is over here just east of downtown Denver. 

The University Hospital has 703 beds today. We’re currently constructing a third inpatient tower so we’ll have upwards of 1000 beds in a year or two. And within this hospital, we see about 50,000 inpatients and observation patients per year. We have a very busy emergency department with almost 140,000 visits coming through there.

And at the University Hospital, we in the past really have really struggled with being consistent and strategic about the decisions we make about handling our high capacity. We’re building that inpatient tower because we’re often very full and we need to have more inpatient beds. And I wanted to kind of explain how difficult it is to handle capacity at a hospital of this size and of this complex of a patient population by showing some trends that we see just over a couple of over the course of a few weeks in our hospital. 

So there on the right side, I have a candlestick chart that’s showing our census over a given timeframe, where each of these bars represents one day and within any given bar there’s kind of the beat at the top and bottom which represents the maximum and minimum capacity or the maximum and minimum census on a given day. And then this thicker bar represents the open and close, where the bar will be red if we started the day with more patients than we ended with and so our census went down. With green the opposite is true. And there are a few things I thought were clear from this chart. One is that there are some days where we have really large fluctuations in our census, something like 10% of our total beds turned over in a given day. And other days, it’s more like 3% of our census changes. You’ll also notice that there’s a huge range in census even over the course of something as short as three days, so we had a range of 100 patients change in between these three days on this chart here. And one of the big challenges for some of our capacity management leaders who may be our house supervisors, patient placement nurses, or other people who make our big capacity decisions on a daily basis, is that it’s really hard to tell what type of day it is when you’re in it. 

And so without sophisticated data points that we had to tell our capacity management leaders what our strategy was going to be for the day, because before we knew what the day would turn out like, they really had to rely on feel and anecdote and some of their own experience to make really critical decisions. And there are a lot of decisions that we make about capacity in a given day. So for example, a patient placement nurse makes about 200 placements in a day. He or she will need to process a lot of information in order to make those decisions as optimized as possible. And without really good tools that’s impossible. So before we had the iQueue tool and some of the benefits that come along with it, these are some of the things that I would hear from our patient placement nurses or house supervisors. 

So in one example, I was talking to a patient placement nurse who had one open bed in our progressive care unit. And one patient from our cardiothoracic ICU who had a request for that bed. So the patient was requesting to be transferred from the CTICU to the PCU. And she decided to hold off on executing on that transfer because she had a feeling that she was going to need that PCU bed for another patient later. And the CTICU patient would just kind of hang out until she was sure whether that bed would be necessary for another patient. 

And another example, a house supervisor let me know that she was going to recommend turning on an RN a nursing incentive payments to get some nurses to pick up extra shifts because she predicted that we would be opening a surge area in the next day or two and she wanted to make sure that we had sufficient staff for those areas. And it wasn’t based on a number or an actual prediction. She just had a feeling that we would be surging in the next day or two. And so she turned on that incentive. And finally I had a conversation with someone about how she was making the decision to prioritize some placement of patients who were waiting in our PACU over patients who were waiting in our emergency department when usually I saw the opposite happening especially earlier in the morning. And her reasoning was that she had worked with Dr. Smith who was operating that day. And Dr. Smith usually finished her surgeries early. And so she was concerned that the PACU was going to begin boarding a lot of patients and to prevent that she was prioritizing getting the PACU decanted before the ED. In each of these scenarios, the decisions that they were making on anecdote and feel was actually right, it was definitely the right thing to do for the organization that day, but it’s not really a sustainable strategy to rely on a handful of people on the hospital being really good at this and really intimately knowing how hospital works to make these types of decisions because it’s simply not scalable, and it requires a lot of work from these teams, when it’s something that we could definitely automate. 

So when we adopted the LeanTaaS iQueue tool, we tried to move away from some of that anecdotal and feel-based decision making to making some more database decisions using iQueue. And here are some of our standard operating procedures summarized, I’m going to spare you some of the details that go behind these. But one of the things that is kind of a dogma for us is that we like to open surge areas in advance of needing them so that we’re able to fill them with patients who are fit for those surge areas. So for example, it wouldn’t make sense to put a patient with an infectious disease in a surge area, but it may make sense to put a patient who’s in observation there. And so we like to be proactive about this and our rule is that our open beds predicted in the next 12 to 24 hours, if it’s going to dip below 10, we’re going to open a surge area. Then on the other hand, if we have a surge area open and we expect to have more than 40 open beds in the next 12 to 24 hours, we’ll close it. It’s very clear. Anybody can action upon that and it’s standardized. We also have triggers for things like our transfer center. So in times of high capacity, sometimes we’ll go on delay protocols, we will ask patients to wait a couple of hours before we can find them a bed in our facility, or we’ll offer them a bed at another facility within our system that could accept them sooner. We don’t like to do that because we want to offer the patient the bed that they’re requesting, or outside the hospital the bed that they’re requesting, and so we have really clear and stringent rules for one we allow delay to happen in our transfer center. We also use the tool to closely manage our ICU beds to never let them dip below five, so that we could be sure that if we have very sick patients coming into the hospital, we would wait for them to go. We even use the tools for specific unit interventions. 

In the past, we actually on Tuesdays had two orthopedic surgeons who generated a lot of volume, and they just happen to have block time on the same day. Because we knew every Tuesday we have a surge of admissions that we’re looking for beds in the orthopedics unit, it was important to us to prioritize discharges from that unit as early as was appropriate. So we would apply some resources to that unit, and make sure that all the patients had rides, make sure they had the care management and nursing resources they had, so that when patients came out of the OR in the PACU, we had sufficient beds on the unit they would need. 

And finally, we even have used iQueue for some of our COVID management. So we’ll use the COVID census that we can see on the tool to let us know how many units we need for our COVID patients. When we dip below a certain level we’ll reduce our units from three COVID units to two and two to one. We have really clear rules about that just so we can make sure that we’re limiting the geographic scope of those patients as much as possible. 

Which brings me to challenge number two, which is getting us out of our crisis mode into more proactive capacity management. So here I have an image of the tool that we used before we had iQueue. It was an Excel based tool that we relied on, and asked our house supervisors to download spreadsheets from Epic, which is our medical record, on a daily basis to refresh this tool, which then we would update several times a day in order to check the status of the hospital and make some capacity management decisions.

Unfortunately, when we use this tool, there was a pretty high level of skepticism as it related to the predictions in this tool. So for one example, the source of discharge is listed here you can see we have the total discharges that were pended or confirmed for a given day. And you’ll notice this update was pulled at 8:30 in the morning. The pended discharges is based on the entry that charge nurses make at the start of their shift for patients that they expect to go home that day. And over the course of the day, as the charge nurses hear more updates on their patients and get to know the patients on the floor a little bit better, that pending number tends to increase pretty significantly. So in the morning, when our predicted discharges is at 91, I would have looked at this and thought, “I know that that’s going to be closer to 120 by the end of the day. So I’m not that worried that our bed balance is going to be negative 30 tomorrow morning, because I think we’re going to actually pick up 30 more discharges.” And I didn’t really trust the tool and therefore I wouldn’t really react to this number even though negative 30 beds in the morning means 30 patients boarding in the evening, which is definitely something that we don’t want to happen. 

Let’s say I was wrong, and we didn’t actually pick up 30 more discharges. We ended the day with around 90. We would kind of find out that we were in a capacity crisis when it had already happened. And so we know that we were short beds when the ED was already boarding and we didn’t have beds upstairs for those patients. And we’d start scrambling to open surge areas. But as I described before, not all patients are eligible for surge areas. And so I wouldn’t even be able to take advantage of those beds and fill them completely. 

So in my experience when we are already in a capacity crunch, and we’re kind of scrambling to activate some of the actions we can take to ease capacity, there’s really not a whole lot you can do once we’re already boarding. And we would kind of create all this chaos trying to escalate discharge delays, care management delays, staffed room cleans and that sort of thing. But our only real tool was escalation and at some point when everything’s priority, nothing’s a priority. And it was a pretty inefficient way to react to capacity. It definitely created a lot of chaos and it was something that I think was a pretty big staff dissatisfier because they were already tired and we were asking for them to provide us information about escalations. 

Now that we have iQueue, we can rely on more reliable predictions and we need to be much more proactive on how we manage capacity. The iQueue predictions are based on unit level detail which I’ll get into a little bit later. But they are much more reliable, much more trusted across the organization. So when we see bed balance is something that we believe and we use to make really important and proactive decisions about capacity. We’re also able to really effectively communicate what’s going on in the hospital and why we’re making decisions that we are using this tool. So having good predictions and having something that’s backed by data science, rather than kind of a boot strapped Excel spreadsheet has really transformed the way that we can make decisions about capacity and how we can manage it rather than react to it. 

So these are some of the proactive actions that we take. Over here I have a little snapshot of our ICU management tool. And I can see here in the snapshot that I have an ICU census of 145 patients out of my 152 beds. And within that census, I can look at the clinical level of care level for each of those patients. So I may be in an ICU bed and I may require ICU level of care or progressive level of care or floor level of care. So my clinical level of care is separate from where I am physically or geographically located. And sometimes it’s okay to have floor level of care patients and ICUs because many of our units are designed to be acuity adaptable, and so we should be taking care of a spectrum of patients but otherwise units are really only designed for ICU patients and they’re staffed accordingly. 

So we use this tool kind of in two ways. On one hand, as I mentioned before, it’s always a priority to make sure that we have a minimum threshold of ICU beds available, so that if we have a very sick patient coming in we have the bed that they so badly need and so we’ll use this tool, if we get dropped to a low number of open beds, we can take patients who are in a floor level of care in our ICUs and transfer them to floor level units.

On the flip side, often our main hospital has the problem where we actually don’t have enough MedSurg beds and we can have a small surplus of ICU beds. So when that happens, we’re strategic about how quickly we act upon downgrades for patients who were in the ICU. But have gotten a little bit better and are ready to move to that surge unit. And in that case, sometimes we’ll delay moving those patients so that we can preserve the med surge capacity that’s needed for patients who are waiting for a bed in the ED or so that we can strategically wait for the right bed to open up. 

So if I have a patient in an ICU who’s waiting for a bed on pulmonary, I may wait to downgrade them until I actually have a bed on pulmonary available, so I don’t end up placing that patient on an off-service unit. This allows me to track what tools and what levers I have at my disposal at any given time when I’m managing between the different levels of care. 

The next is the transfer summary. So at any given time our user placement nurses are looking at between 20 and 75 requests for transfer within the hospitals. These are patients who are in one inpatient bed or a unit and are requesting to move to another and we look at transfers as having different levels of urgency depending on their type. So the first thing that’ll stand up to here is the step up. So these are patients who are in a MedSurg unit but who require an ICU bed. And we really prioritize those because we need to make sure that those patients can receive the higher level of care that they need. You’ll rarely see anything more than one or two patients who are waiting to be transferred in the Step Up category, because we’d like to work those pretty quickly. Another area that we pay close attention to is the surge to floor group. So for patients who were placed in a surge unit when they were admitted, we don’t want them to stay in that surge unit for the duration of their stay, so we prioritize getting them to bed on a more standard inpatient unit after they spent one night in the surge area. And so those are definitely a priority for us to place. And then we have these middle three categories of patients who are stepping down, so like I described before patients who are in an ICU and ready for a floor level of care or those who are moving within a level of care within the unit. We don’t see these with quite so urgent and they tend to make up the bulk of the transfer request that we have. So we’ll come back later to how we prioritize among this group of transfers. 

The transfer summary helps us to make sure that we’re staying on top of those places and balancing those with new placements for patients who haven’t yet gotten an inpatient bed. The next area that we try to very proactively manage is our boarding population. So these are patients and our emergency department or our PACUs who have an admission order and are waiting in those areas to be placed in patients. And typically when we have very high capacity, we can see the number of patients waiting in these areas growl. And it’s important for us to keep an eye on those because as those censuses grow, it disrupts the intended operation of the ED/PACU. They’re not meant to be inpatient units, and we want to preserve their operation by not blocking them with patients who really need to be in another part of the hospital. So in addition to just tracking the total number of patients who are waiting for beds in each of these areas, we also really call out when we need an ICU or PCU level of care. Because that level of nursing and other services that are available in ICUs and PCUs are not always available, and EDs and PACUs so it’s more urgent to move those patients to an ICU then it may be for a floor level of care patient. 

And we also keep track of the maximum time that the patient has been waiting. So if we have a patient in the ED who has been waiting for an hour and a half for a bed, that’s more on the acceptable range of time but as we see that climbing to be higher and higher, we definitely want to prioritize getting patients so they aren’t spending a whole day or overnight in an area where it wasn’t intended for them to stay the night. 

So we’ll keep an eye on that even if our number of patients is low, we don’t want patients to be waiting for quite so long. 

Which brings me to challenge number three, which is when we have those patients and we need to determine where to place them in the hospital. How do we deal with that complex problem?

So in our University Hospital, we have on an average day, 21 units and 88 services. And the challenge that our patient placement nurses face is how you match up those 88 services to the units where they should go and how you can make sure that there’s a bed available each time they need one. And so here on the right I have mapping our critical care services. And you can see here in the case of acute care surgery trauma for their three services, their patients who require an ICU bed will primarily go to a surgical trauma ICU, and if there’s not a bed available there, our secondary choice will be the neuro ICU. And if there’s not a bed available in either of those units, then we’ll have to place the patient off service which is definitely not ideal. Because our hospital runs at such a high occupancy rate, usually in the high 90s, sometimes over 100%, we typically have several patients with active bed requests and just one bed at a time becoming available. So the chances that the bed that becomes available is the right bed for the patient who has been waiting the longest is pretty low, and our patient placement nurses have the challenge of determining how you kind of balance teaching placing patients in timely manner and also placing them on service, and by on service, meaning placing them on either their primary or secondary unit. 

It’s really important to us to please patients on service because it impacts the quality of care, how long they’re in the hospital, their patient experience, and it has a huge impact on staffing provider experience, because our providers don’t want to be traveling to several different units just to take care of one patient in each. And our nursing staff is really trained and specialized in a particular area of clinical care. And so those are the patients that should be going through their units. So the on service rate is definitely a key metric for us. In addressing this challenge, we have found that one of our capacity management dogmas is that capacity management gets very local very fast. And while many of the like metrics that we developed previously that were showing a mental dashboard I showed you they’re all at the hospital level, where most of the interventions and decisions need to be made using unit level of detail. And so this is a challenge that we communicated to LeanTaaS and they responded by creating us unit fingerprints. It’s basically an acknowledgement that units in and of themselves behave very differently. LeanTaaS took several different characteristics for these units and found that we can better predict their supply and demand for beds using a variety of different variables. 

So the size of the unit obviously matters, where their admission sources, whether it’s the ED or the OR matters and will have a big impact on the time of day when their demand will be coming in. We also know that the time at which services round and write discharge orders on a given unit will impact their demand and when they’ll have a supply of beds. And there are also less obvious variables that we incorporated into these finger prints such as are the nurses on that unit good at predicting where patients are going to go home or is it something that’s oftentimes not predictable Because it’s based on a last minute lab value or something like that. So, LeanTaaS took a long list of variables and created a fingerprint for each unit that would tell us by hour and by day of week, what the bed demand and supply would be so that we could tell whether we were going to have a surplus or a deficit. And this is something that they’re constantly updating using the historical data of the event. Data related to the specific patients that are there now, and other real time data such as patients who are not in the unit yet but who are likely to request a bed there after they’ve been admitted to the ED or the OR. So through this, we’ve been able to get to that granular local level of capacity management that’s really critical in order to make everything work. And here are some areas where the unit fingerprints are available in the tool.

The first is in the admissions queue, which is a relatively new enhancement that we put into place a few months ago. I’ll interpret this for you. The first, we have three units listed up here: our critical care annex, our burn ICU and our cardiac ICU. In each of these units, we have one patient with an admission order who’s requesting a bed on that unit. In the critical care annex, and I have my kind of purple patient with the active admission order here. Then I’m also anticipating that there’ll be three more patients over the course of the day that will be requesting beds on the CCA.

I also know that right now I have three clean and open beds available in the CCA and one dirty bed available that will soon be clean. And finally, I’m predicting that another fifth bed will become available from a patient who will discharge from the unit. So overall by the end of the day, what this is telling me is that I’ll have a one bed surplus after all of my admissions and bed cleans and discharges have taken place.

In contrast, in my burn ICU, I currently have no open beds, but I have one patient already requesting a bed and two more to come. 

So I need to make a plan for what to do with these two patients for sure. And potentially this third patient if they need to be placed more quickly before this bed becomes available to me.

And so using this detail, I can kind of troubleshoot the problem and say, I know that I’m going to need to place two to three patients from the burn service, not in the burn ICU. And so I’m going to consider actually cohorting them in the CICU because it’s more geographically adjacent to the burn ICU than the CCA is for example. And maybe for the burn population. It’s really important that they’re near to their provider team at any given time. I don’t want to have to ask that provider team to be traveling to the CCA.

So this allows us to kind of take a glance and quickly problem solve within a division when demand for one unit exceeds supply and for another we actually have excess supply. 

Another one of my new favorite enhancements to the tool is the transfer recommendations over here on the right.

And as I mentioned before, within our transfer request, the bulk of them fall into that category of steps down inter-lateral and intra-lateral transfers, which we don’t see as urgent as our step ups or our surge to floor transfers. And there tends to be a high volume of those so something like 20 to 70 and among those, it’s really helpful for us to help patient placement identify the ones that are those that we should act upon. 

So we do that by identifying win-wins, and this example, to patients in room 225 and 233 and my cardiothoracic ICU. That unit currently has one available bed, but I’m predicting by the end of the day, they’re going to have a negative one. And both of these patients are requesting a transfer to the pulmonary unit, and the pulmonary unit like CTICU currently has one open bed, at the end of the day, they’re expecting a couple of discharges leaving them with three ICU beds. 

And so I should prioritize these to step down placements from CTICU pulmonary, because it’ll free up the needed space on the CT ICU so they can go into the night with one open bed. And pulmonary will have two on-service placements as a result. So it’s a win-win. I’m just kind of highlighting here and letting our patient placement folks know that this is a transfer that we should act upon and prioritize even though it’s not in one of our other kind of highly urgent categories.

A third unit and intervention or tool is this unit detail page that we have an iQueue and it has a list of all of the units in the hospital with really important information like their bed count, their census, and how many patients are off service. Well, we expect their discharges and bed balance to be by the end of the day.

Previous to having this tool a lot of our capacity management huddles involved having our nurse managers in charge nurses come and report out on these metrics. And it really wasn’t an effective way of fostering collaboration. It was kind of more of a roll call and we never use the data to kind of problem solve. Whereas now I see our nurse managers and charge nurses collaborating on hey, I have a patient coming out of OR at noon and I need to make transfer three of the bed for them. It looks like you have a bed here, can we talk? Or hey, I noticed that you have a lot of floor patients on your unit today, do you think you can float the nurse to me? 

And so just increasing the transparency of these numbers, but anyone at the hospital who works on capacity or who would be curious to know can see them allows some really local decision making to occur via unit to unit which has a huge impact on the overall capacity of the hospital and has really increased engagement in the tool among a wide spectrum of people, whether it’s someone who’s working very locally on a unit or someone who is managing the overall ED census or overall capacity management.

So, in summary, we’ve covered our three big challenges and how iQueue has helped us to deal with them. And they were inconsistent decision making due to the lack of data, having a reactionary crisis-based culture and not having the intelligence we needed to make the best patient placement decisions for our patients. Using iQueue, we were really able to transform our capacity management strategy to be based on objective triggers rather than anecdote and feel. We planned more proactively for changes in consensus and really actually started managing it rather than reacting to it. We make much better decisions in patient placement and were able to prioritize the patients who need an intervention most and we’ve been really successful and being able to finally communicate our capacity status and the reasoning behind some of the decisions we make. 

Finally, one of our newer enhancements is that we’ve expanded the tool to all 12 hospitals in our system. And so we’re now able to leverage capacity in other nearby hospitals when one of our hospitals is full and maybe another has some bed availability. So we’re beginning to partner among our regions and even amongst the system more than we have in the past.

We chose LeanTaaS to work on this project with us because we have a really successful history  together. We’ve partnered with Lean past in our operating room and infusion space before we worked on beds. And we were really excited to work with them on beds because we have a lot of this operational expertise that I’ve described in some of our health supervisors and patient placement nurses with how they really intimately understand the complexity of the capacity ecosystem. And we were really excited to take that expertise and translate it into something that was meaningful and scalable outside of our University Hospital or even outside of UCHealth.

And we really appreciated that LeanTaaS recognized that expertise and spent a lot of time shadowing our house supervisors and our patient placement. They even did a lot of observation on our unit with our charge nurses so that they could begin to understand and appreciate the complexity of our hospital and the patient flow processes that we use every day.

And they were a fantastic partner for us during the pandemic. So we made several changes to the tool to help with our COVID specific operations. By the end of the pandemic, we were using those enhancements in iQueue for things like reporting federally what our census was and what our bed availability was. We used it for our hospital and system and clinic command meetings. And we even use the tool as part of the process that we stood up with Colorado State Transfer Center. As we were transferring patients across the state and managing the census in the state, even outside of our own system.

The results have been really positive so far. So we implemented iQueue in February 2020 Right before the pandemic and we went live with our other 11 hospitals in October just before a big surge. And I feel really lucky that we were able to accomplish this timing, I think this year may have been a lot harder for us if we didn’t have the tool.

Now we see it used across our hospitals in a variety of ways. Many of the hospitals use it in the morning for our bed meetings and some of our capacity management leadership kind of obsessively check the tool on their phones all day long. And as I mentioned before, we use it for several of our standard operating procedures and triggers when we’re making decisions about activating surge units or decisions about our transfer center protocols. So it’s definitely something that gets a lot of use in our organization. And I think the results of the tool really speak for themselves. Since implementing it, we’ve seen improvements in our ICU transfer time, our admit time, notably in our opportunity days associated with our inpatient length of stay and certainly in the confidence that our organization has and our capacity management, leadership, and decision making.

And all of these improvements happened during the pandemic. I think it’s really notable, that timing, because a lot of those who worked in hospitals know, we kind of stopped a lot of our process improvement and quality improvement projects and reallocated resources to the pandemic response. So the fact that we’ve been able to see this type of improvement, I think really speaks to just how valuable the tool is.

Finally, as we look forward, we’re now working on several enhancements to the tool that are very exciting. And aimed at making it even more prescriptive. Some of those that I’d like to highlight are enhancements to the discharge toolkit and engaging with our care management teams to see how we can leverage this tool to make their lives more data driven and to prioritize their work in a given day.

We also want to begin to provide more decision support to those patient placement folks who are making, you know, 200 decisions about where to place patients on a given day. So as much information and support as we can provide them would certainly be valuable.

We’re building out some daily retrospective analytics that will tell us based on what happened yesterday and the actions that we took, what was the effect on our hospital and our capacity management and how can we learn from that? What went well and what opportunities we have just so we can kind of be constantly learning and improving. And we also see some potential benefit in utilizing the tool for staffing. So I know a lot of hospitals have really burnt out staff, after what our nursing staff especially went through during the pandemic. And so to the extent that we can use the tool to make sure that we’re using our nursing resources as efficiently and reliably as possible, I think will make a really big difference.

With that, I would like to thank you for your time and open it up for questions.

Moderator: Excellent. Well, thanks so much for Isobel for that awesome visitation. Now we’ll go ahead and open the floor to audience Q&A. I’ll go ahead and get started here with some of the first ones. A quick reminder to our attendees. You can submit some questions, if you have them, using the Q&A box that you’ll see on your screen. Feel free to send us some in there.

The first question asks, can you please explain about the integration process that UCHealth has adopted to integrate data across different units in order to make conclusions about bed capacity?

Handler: I can take that one. I’ll give you a good example of some collaboration that I’ve seen across the unit and some of the high capacity hills that we’ve had recently. So as an example, about two weeks ago, we were particularly tight in ICU beds and our surgical trauma ICU. And so some of the services that typically go to that unit were having to overflow to their secondary unit which is our neurosurgery ICU. And as a result, we needed to make some downgrades for patients who were in the neurosurgery ICU to the neurosurgery floor. There are three units involved here: the surgical trauma ICU, the neuro ICU and the neurosurgery floor. And it’s not, I guess, as typical for the surgical trauma charge nurse to be collaborating with the neuro floor charge nurse, but they were able to kind of talk about what time of day they expected some of this volume to come in so that we could prioritize the timing of discharges that would need to happen from the neuro floor in order to make sure that there was a clean bed available for downgrades from the neuro ICU. And they were able to pretty quickly communicate about on a typical Wednesday, what their demand patterns are and using some numbers that were in the tool about what their ordinary demand would be and then how that was going to change based on the plan that we have to shift some patients from our surgical trauma floor to the neuro floors.

So I think the ability to quickly problem solve with very transparent and easily available data was key for them. And it was definitely something that was a conversation among charge nurses that was really effective and something that may not have happened as easily if it were occurring at a level more like our capacity management leadership. So I think the tool has been a really great way for us to help our local charge nurses and nurse managers to solve problems like that.

Moderator: Can you remind the audience which EMR you use at UCHealth? 

Handler: Happy to, we use Epic Systems. 

Moderator: Excellent. I’ll roll into one of our next questions here. Have you found the system has helped with better transparency and timelines, In room availability after discharge? So our attendee asked if you’re minimizing holding room after discharge to delay admission as a workload management effort?

Handler: Yeah, a big focus in the messaging that we’ve used in rolling out the tool and something that I think has resonated a lot with our nursing staff has been how we’re going to use this tool to make sure that we get patients to the best possible bed and get them to on service unit. And so I think, you know, we may have some experience in the past, where, if a unit is particularly busy, they may discharge the patient from a bed and a given unit and not document it for a little while just so they give themselves some time to kind of catch up. And that time when the bed is not being occupied and not being cleaned, is not productive for the organization. So it’s definitely something that we want to avoid.

But with the tool, and the importance of the importance that we’ve communicated and how the tool is helping us to level load demand and make sure that demand for certain types of specialty beds is met with supply for those beds, I think that our nurses are more eager to get the bed available for the patient who needs to come to their unit, rather than kind of preserving whatever time they can get from not having a new patient admitted to their unit. We’ve kind of shifted the conversation from you know, “I’m so busy and I need some time to get ready for my next patient,” to, “it’s really important for me to make sure that I’m getting the data available to the patient who needs to come to my unit next.” Because it matters to me that we have on service patients and it’s clear to me that the organization sees that as something really important as well.

Moderator: Absolutely, thanks so much for the insights there. Another question we’re seeing come in, just a little bit more interest in the predicted demand numbers. Could you talk to us a little bit about how you’re using that? I think specifically the attendees are interested in hearing a little bit about that kind of timing.

Handler: Yeah, so the predicted demand numbers are LeanTaaS has them based on historic data that goes back, seasonally, and they also use real time patient data. So it’s based on a very complex algorithm that I won’t describe as well as maybe someone from the LeanTaaS team, but it gives us a really good and reliable sense of what’s going to happen in the next 24 hours or so. And we’re able to kind of predict that on an hourly basis. So if something changes in the algorithm, it catches up really quickly to that change and we can respond more proactively.

But the predicted numbers, and then reliabilities definitely something that I think is one of the main wins of the tool.

Moderator: Absolutely, definitely a great feature. And looking to now, on the iQueue system, you talked about, are there any unexpected ways that you’re using that?

Handler: Yeah, it’s been kind of interesting to see different parts of the hospitals that are interested in using the tool. So I’ll say like food and nutrition for example, they have a large screen in their kitchen that displays the capacity level tools so that they can see different unit censuses and get a general idea of how busy the hospital is on a given day. And they’ve actually used that to drive some of how they prioritize their work and their delivery times by unit. So that’s something that I never expected, but I was really excited to hear about.

I think another way that it’s really helpful that I maybe also didn’t predict is that we use the iQueue tool as our source of truth for a lot of reporting that work we’ve had to do for Department of Health and Human Services as part of our COVID response. And the tool is able to provide us a real time report on what our bed availability is. We’ve configured it and some hospitals track very closely our COVID census and where those quotations are located.

So I think neither of those applications are something that we conceived of when we began building the tool, but it’s something that’s definitely been really valuable for those projects.

Moderator: For sure. And I think building off of that, can you explain to us how you’ve managed to admit patients more quickly using the iQueue tool, admit you know, the Coronavirus patients surge. 

Handler: When the surge was happening and in particular wave three that was occurring in late fall and early winter. We were much more strapped for inpatient beds than we had been in previous waves earlier in 2020, and the tool enabled us to reliably predict what our open COVID beds would be, and even to get a sense for what was going on across the system so that we could gauge our likelihood of needing to accept transfers locally or even from more rural parts of Colorado. So I think you know, the COVID era is very well characterized by how uncertain everything was and a real need to kind of prepare in advance for the worst case scenario. And I think that the tool allowed us to kind of balance being prepared for our worst case scenarios with also making sure that we were taking in as many patients as we could who so badly needed our care, based on what we knew was possible in the next 24 hours or so. And so I think UCHelath saw more COVID patients than any other system in the state of Colorado. And that’s something that we’re really proud of.

The tool kind of enabled us to do that with confidence, so that we were able to say yes to patients who needed to come to us and particularly from those rural hospitals that were really overwhelmed with their COVID census who needed an outlet like UCHealth to step in.

Sanyal-Dey: Isobel, I’m going to jump in. I wanted to address one question that came up – are there examples when iQueue systems didn’t perform as expected and actually, the way we’ve tackled this is more of the understanding of when do you use iQueue to look for certain information, when do you really rely more on the EMR or in your case, Epic, being that’s what you use in Colorado, what have you noticed from frontliners and from operational folks and how that distinction is helped to understand when to use the right tool?

Handler: Yeah, I think shifting our perspective from being an organization that relies on actual real time, data that’s reported in Epic such as this is how many discharges are, you know, confirmed for today. And shifting that to this is how many discharges we expect to occur between now and the end of the day, has taken some time for us to adapt to particularly in some of our smaller hospitals. So I think, you know, we’ve gotten feedback sometimes that the discharge numbers don’t look right. Sometimes, you know, there’s an opportunity for us to do a little bit of validation on the tool, but a lot of the time I think our patient placement nurses and house supervisors just aren’t as used to working with tools that are more prescriptive for them. And so I think balancing, this is kind of a snapshot of what our capacity is that we’re able to get in Epic and this is how many confirmed discharges I have in patients who are currently waiting for a bed in the ED. And kind of reconciling that information with what we can see in LeanTaaS, which is the total discharges you’re going to have by the end of the day. There’s a portion of those that aren’t in Epic yet because they haven’t happened yet. And while you have this many patients in the ED right now we actually expect that number to change over the next few hours is definitely been a change in our paradigm. And so something that we’ve needed to walk through and help partner with some of our smaller hospitals to think about how to react to those numbers differently from similar numbers that are available in epic today.

Moderator: Yeah, thanks so much, really hopeful to hear. To Jump into another attendee question that’s been said in how do you develop your placement portals? And how did you determine if they are optimal?

Handler: Great question. This was actually kind of my project that got me into capacity management a few years ago. We historically have just kind of assigned services in epic, like our hospital medicine, one service to a particular unit in the hospital based on what made sense, clinically, and before a few years ago, we never looked to see whether or not our bed complement matched our demand for beds.

And so our first try at this was basically to say, given our current matrix that says hospital medicine services should go here in general surgery should go here. We took that and asked the question, does the census for those patients, is it the right side for the units that they’re assigned to?

What we found is that in a hospital of our size with our high level of occupancy, the answer was no and as part of that, we made a couple of changes how services were assigned, because there were definitely some units that were kind of oversubscribed and we were set up for failure and that there was no way that you know, five hospital medicine services with 10 to 15 patients each would be able to fit on a single unit in a given day. And there are other services where they were kind of always taking off service patients because they tended to just have a little extra room.

We did some level loading to begin with and then we brought in this concept of having a secondary unit or a sister unit we sometimes call it, where if a bed isn’t available on the primary unit, then there’s kind of a backup plan. And that I think has helped us to think of patient placement in a way that’s very patient centered and very nurse centered, so that we’re organizing our care as best as we can given the constraints we have and given how full we are.

But it’s definitely something that we revisit quite often. And need to adapt as the censuses in our services change as we noticed that their censuses may vary over the course of the week. Like we talked about, what some of those unit fingerprint studies, and definitely something that we’re working to improve and kind of continuously monitor going forward.

Moderator: I think we’ve got time for just one more.

Sanyal-Dey: Yeah, totally. We can end with this last question, which is asking when does iQueue predict something that doesn’t feel like you should have expected? We’ve been pleasantly surprised with some of the comments coming from Colorado regarding the accuracy of some of the numbers. Can you comment a little bit about that?

Handler: Yeah, iQueue has changed the way that we behave in our approach capacity in a couple of different ways. So let me think of a good example of something that wasn’t effective. We have a new tool that I went over a little earlier that shows recommendations for unit transfers. And it will tell us, you know, based on these two units, it makes sense for you to transfer the patient because one unit is going to be short beds and the other one’s going to have a surplus and so making the transfer will kind of leave both units in a better place.

Previous to iQueue, we kind of had a rule that we never did lateral transfers, because it uses precious resources in our EDs, and it’s another patient handoff and it was just kind of easy to say that as a blanket rule, we’re never going to do that.

And something iQueue has provided is a way to surface some of those opportunities that make sense, and look at it in more of a nuanced patient-centered way. So if we have a patient who we expect to have a long length of stay, and they’re on the wrong unit today, it absolutely makes sense for us to do a lateral transfer, and we can decide to do that when it makes the most sense. So I think that is something that has definitely changed the way we think. As far as the tool is predicting a surge, that’s not something that was necessarily expecting, there are times we might pause and determine whether or not we think there’s something funny going on, and maybe hesitate to put some of our capacity actions into play, but to be honest I can’t think of a time recently where we’ve done that, I think the tool is pretty highly trusted, and when it’s pointing out something we weren’t expecting, usually that’s used to point out something we hadn’t noticed before. 

Moderator: It’s great to hear about that reliability. Unfortunately that’s all the time we have for today, but Isobel, thank you for joining us today. I always want to thank LeanTaaS for sponsoring the Transform event. We look forward to seeing you tomorrow – have a great rest of your day.

View the whole session here, or read a blog summary here.