The Right Way to Allocate Block Time Webinar transcript

SANJEEV AGRAWAL: Good morning, everyone. Good afternoon if you’re on the east coast. I just wanted to start by putting in context what this webinar is about. As you know, or at least some of you know, we’ve been having a full series of webinars around all our efficiency. And this is the last of that series. The other webinars are all available on our website as well. So a couple of months ago, we launched this series where we started with an overview on August of 23rd of how to get the most out of your OR capacity. And then September 13, we talked about how you right-sizes blocks using a methodology known as collectible time, instead of using block utilization as a way of measuring surgeon and block honor performance because that is a broken metric. So if you’re interested in that, that webinars available. 

And then a week after that, we talked about how do you create an. OpenTable-like capability to both create open time and release and request open time, which by far is one of the most practical ways of creating capacity in your ORs and sharing it with much more lead time than most hospitals we see do. We then talked last week about how do you get surgeons and administrators to look at reporting, to look at metrics, to understand them, to give them any credibility or importance through 2018 ways of reporting as opposed to generic reports that we’ve seen most health systems have as well as the EHR reporting that seems to be widespread, but has different degrees of efficiency and efficacy. 

And today is the last of the series where we’re talking about a fun mathematically complex, but really the right way to manage what is an expensive perishable commodity. And so this webinar is about how do you right– not just right-size blocks, which we talked about a month ago, but this is more by saying given the data we have, how can we come up with a better lock allocation to begin with. Now, our experiences based on working with many, many health systems that have deployed a number of tools under the iQueue umbrella. So 30 seconds on us as a company, LeanTaas as a company that uses predictive analytics mobile and web tools to digitally transform core processes around block allocation, around block right-sizing, around block release and request. 

And all of these tools are being used by about 500 ORs across this country. Across 14 institutions that you see on this page, a mix of academic, community, EHR-independent whether it’s Epic, Cerner,. Meditech, it doesn’t really matter. There’s a tremendous amount of great data in all of these EHRs and. Timestamps of how the law OR been used that can be effectively used to do whatever you’re trying to do– right-sized blocks, allocate capacity right, release, and request, and do 2018 reporting. So let’s just start with the problem statement, and for those who’ve been part of other webinars, this is going to be a little bit of a repeat. But the story I always love talking about is my visit to a. Children’s Hospital where I asked a senior surgeon whether how long it would take if I brought my child in to get them admitted and operated on for an elective case. 

And they told me, well, I have block time on Wednesdays, and I’m fully booked out. You know, we’re one of the top institutions in the country, and I probably couldn’t fit them in because there is no open time either at least that I can see for six to seven weeks. And so what I ask that gentleman was, if I walk through your today and tomorrow and the day after and next. Monday, every day of the week, are you telling me that the person that has reserved capacity, i.e. The block allocated to them is the person in the OR, doing cases. And are you telling me that there is never any capacity left on the table every day of the week? And he smiled wryly, and he said, of course, you know that’s not true. And so part of me wanted to say, well, this feels a little bit like I stood in line to buy a Super Bowl ticket. I stood for six hours, spent $10,000, and then I show up at the game, and a third of the stadium is empty. 

So this inherent capacity management paradox that, for our time, is so precious that it’s never available. And yes, there is reserved time allocated to individuals and groups and service lines that own it that it’s left on the table. How can both of them be simultaneously true? So at the beginning, and for the first 10 minutes, I’m actually going to talk about why that actually happens mathematically speaking, and what the underlying problem is when you are operating something with high or even small degrees of unpredictability. But the punchline here is that if we can improve the way in which we allocate time, we can improve the way in which we right-size blocks, release, and request time. 

The upside, not just the financial upside, but being able to see my kids sooner and get them into the OR, the quality of patient care, the access, really this is about access to the OR goes up a lot. So none of this is new news for operators and others that are on this webinar, but I just wanted to frame the problem as too from a patient perspective what this feels like. And that’s the problem we’re trying to solve. So as we did in every other webinar before this, we said there is a common theme to why the problem happens. So for example, when we talked about right-sizing block, we talked about the fact that block utilization is a meaningless metric to compare surgeon performance or block on a performance. 

The theme today is that the way blocks are allocated reserving expensive capacity is like creating permanent carpool lanes for those who live in California or anywhere in the country if you experience this notion of reserve lanes on the highway. There is an inherent inefficiency to it for the public good. And the same logic for why that happens applies to the OR. And in fact, there are 50 such examples of highly expensive asset-intensive industries where reserving capacity sounds like a good thing, but if you do it beyond a certain point, you essentially create a humongous amount of inefficiency. So let’s start there. Let’s actually start with a couple of practical real-world examples. 

So imagine that we took a highway, and we took all its lanes. And we said, only Mustangs could drive into left lane, and only Toyota is in the second lane, and only Fords in the lane after that and, only GM cars in the lane after that, et cetera. When you think about it when we allocate block that is essentially what we’re doing. We’re taking an interest group at a service line, a block on our surgeon, or a group and saying, you have Wednesdays, you have Thursdays, you have two blocks a month, you have five blocks a month. Now, what is the basic mathematical problem with that? The most basic mathematical problem of that is much like on the highway. If you’ve ever experienced a situation where you’re in a non carpool lane watching a carpool lane that’s half empty, wondering why the heck am. I stuck behind 2000 people when these folks who have access to a lane fully reserved for them aren’t using theirs. 

In the traffic world, the problem is, that it is impossible to predict exactly how many Fords or Mustangs or GMs or Toyotas are going to be traveling in a small interval. If we could actually predict that each lane would be used well by minute, by 10 minute increments, in fact, there’s nothing wrong with allocating capacity based on the volume of cars we expect in each time interval. In the OR world, really what we’re doing is, when we’re allocating capacity– let’s say we’re having an. OR committee meeting today, and you say, Sanjeev, we’ll give you Wednesdays because Dr. Agarawal, the. OR big and important doc, we want to keep you, we’re going to give you Wednesdays. Underlying that decision is an inherent assumption or a series of assumptions. 

First, I will see the right number of patients in clinic over the next 52 weeks. A certain percentage of them will translate into patients or that need surgery. And then I will magically be able to fit them only on Wednesdays, and the length of their cases will be such that I will neatly be able to place them in Tetris block increments, so all my Wednesdays are filled. That is the expectation and hope that you have of me when I have been granted the Wednesday blocks, right? Well, we’re not manufacturing cars. We’re not a manufacturing plant. We are an inherently unpredictable system called surgery, where the volumes of cases, the lengths of cases, the patient volume into my clinic is predictable to a certain extent but not precisely. 

Same thing in the highway situation, we are taking a perishable commodity, which is a long time if you don’t use it. It’s like an airline seeking moovit, you don’t make money, the patient doesn’t get access. And so if you are mathematically 100% sure of something, how do you deal with that situation? And in the OR– the airlines have done a great job of it, believe it or not. Many other industries have done a good job of it, as we’ll talk about. That’s the reason hotels don’t give you a specific room when you call them. They say, yes, by and large, I can give you, there is capacity in my hotel. So whether it’s traffic, whether it’s highways, or think about a busy parking area. How many times have you ever driven around a parking lot where there are reserved spaces, where there’s no one in? But the rest of the parking lot is completely full, so you can’t find a spot. And there are 19 parking spots that no one is using. 

Or think about what would happen if runways on an airport were allocated by airline, only Delta could use runway 1. Only Southwest could use runway 2. Well, how do you know how many flights are, can will Delta always be able to fill its runway? Would Southwest always be able to fill its runway? You’d have the situation where we’d all be sitting around waiting on maybe one packed runway while the other runway is empty. So whether it’s hotels, whether it’s– any of these systems share three common characteristics. One is it’s really expensive to build highways. It’s really expensive to build hotels. It’s really expensive to build OR. Second, this is not like a gold brick. This is perishable commodity. This is an expensive perishable asset. You use it or lose it. 

So if a Southwest plane takes off with a third of the plane empty, you’re never going to get that revenue back or be able to supply that many passengers. Same thing with ORs. And finally, probably the most defining commonality about these systems is in to throw out mathematical gobbledygook. It is a stochastic system and not a deterministic one. Meaning, there is volatile changing unpredictable demand for both the volume of cases and the length of cases. That’s what’s common about these systems, and that’s the reason why unlike ORs, hotels, traffic highways, et cetera, try hard not to reserve too much of the capacity. So reserving such assets will always lead to suboptimal system utilization. 

And just to give you some data behind this. You can do this in your own ORs. This is studies we’ve done across the customers we work with where I just happened to pick one service line that happens to be orthopedics, where each of these colored lines is a surgeon’s volume, the number of minutes they’ve used over time. If a decision was made to allocate capacity in say March of 2014, there was no way in March of 2014 that someone could precisely predict that Dr. Purple was going to these many minutes in the OR, or Dr. Blue is going to do that these many minutes in the OR. And so, and the reason is, of course, you know life happens. There are seasonal trends, orthopedics goes up, and Q4 is not that exciting in Q2. Oncology has less of a seasonal trend. People take vacations. There are clinic conflicts. We have teaching responsibilities, we’re speaking at hems. 

Or God forbid, some of our surgeons don’t have as many cases as we thought they were going to bring into the OR, but we’ve given that time away, right. So this is one of the reasons why this happens where volume is not precisely predictable. It’s predictable within a range. The second reason it happens is we have this magical notion, that somehow, we’ll be able to predict the length of a case. The number of folks that want better case length estimation is like 100%. And I’ve talked about 100 perioperative leaders at this point over the last couple of years. And yes, you can do a better job of case length estimation. But by definition, before you walk into an OR, and open someone’s brains up and put them back together if you’re a neurosurgeon. Or even if you’re doing more regular surgeries, there is always going to be variation in how long it’s going to take, and being able to precisely predict how long the case is going to take is almost impossible. 

So there are three basic reasons behind the volatility. The mathematical reasons are. I cannot predict how many patients are going to show up in my clinic precisely. I cannot predict how many cases that will translate into precisely. I cannot predict if I will actually be able to fit them all in on the days that. I want to fit them in. And then when I do, I don’t really know exactly how long it’s going to take. So that, now, resembles the highway problem and the hotel problem, et cetera, et cetera. So then you might ask the question, well, why do we have block at all? Shouldn’t this just be a free for all? Shouldn’t everybody just be able to get open time and everything should be like open table where I reserve capacity? That’s now swinging the pendulum way too far on the other side. There is inherent efficiency in giving block time as we all know. And here are the four simple reasons that might be others. 

But first of all, there are only certain days of the week that surgeons can do cases because they have clinic on other days. That’s the most obvious issue why certain days of the week are better for everybody. The second is, if I’m going to go to the OR, obviously, I want to be able to do a number of cases at once. Why would I want to go at 10:00 in the morning and then 4:00 in the afternoon today, and then 3:00, just surgeon efficiency in ease. The third is, if I do a certain type of case, then having the same equipment, the same team, the same room, if I do a robot case, and doing them back to back on the same day. And by the way, me and other folks in my service line share that commonality. And so giving a service line, or a surgeon that’s doing a similar types of cases a block of time makes a lot of sense. 

And then of course, from in. OR management perspective, your ability to say, I need anesthesia, I need staffing, I need to be able to provide the tools and resources, the supplies, becomes a lot more efficient that if you give people blocks. So what is the answer? We believe the answer is to give blocked time, but not go fully blocked out. Have a mix of open time and block time, and create that mix by looking at historical data to be able to identify what is the baseline volume of block you should guarantee people. What does that mean? Let’s go back to the highway analogy. The basic problem with giving reserving capacity, and saying the left lane was for Toyotas, why that I couldn’t predict how many Toyotas would show up in 10 minute increments. 

But if I looked at historical data, and I knew for a fact that the minimum number of Toyotas in the left lane between 8 and 9 is at least 50. Then providing enough reserve capacity for 50 toyotas makes a lot of sense. Now, someday it’s at 60, and someday, it’s at 90. But if there was open lanes for those Toyotas to go in, I would be able to fill them in. So very similar analogies applies here, that if I can look at historical case volumes and lengths of cases, I can predict what’s the minimum or baseline amount of time I should actually reserve for a block owner. And then to the extent that there is volatility and they need more, create enough open time for them to be able to manage. Hopefully, this will become clear in a second. 

Now, let’s take the first part of that problem. How do we see what’s the minimum amount of time that I’m going to need? And when you introduce a concept that companies like FedEx, companies like. Southwest, many others have mastered this concept in mathematical terms is known as Bin Packing. So if you think about what the problem we’re trying to solve, Bin Packing goes as follows. If I am FedEx, and it’s October, sometime over the summer, I have to decide how many trucks will I need in December? What’s the minimum number of trucks I need to deliver all my packages from point A to point B in December. Now, for them, it’s a very expensive proposition to have too many trucks. It’s an even more expensive proposition to have too few. 

So the way they think about it is to some extent applicable to the. OR, so what they do is that they go and look at historical volumes, weights, sizes, and routes of packages they have delivered. So they say, OK, let’s look at every December and the relationship between December and every other month of the year to predict how many packages of what size, how many one by one by one cubes with watches. And how many to five by one by three cubes with pies because December is all in a month. And so I can tell based on historical data, what the volume and mix is going to be. And then what they do is a mathematical optimization called constraint-based optimization where they say, OK, given the volume and mix of packages that I expect by a sophisticated algorithm that misforecasting how much I should expect this December? How is the economy doing? How was this August relative to last August? And that should give me a seasonal trend as to what December is going to do. 

That’s number one, what’s the volume and mix? Second, what are my constraints? I can’t have a very heavy truck on a highway or a very light truck. So I need to impose the fact that the minimum weight and the maximum weight that. I’ll be carrying in a truck is x and y. What does that mean for the OR? Let’s remember the objective that we’re trying to fulfill is to figure out, by block owner, what’s the baseline number of block I should give? And how much time should. I have open to absorb the inherent volatility in demand? All right. So let’s– let me show you one way of doing this. Imagine that I’m looking at the orthopedic service line, I have data, or this could be done for an individual surgeon, but certainly, it can be done for a service line. 

I have data in– through your EHR, you have data through your EHR that says if I look at the orthopedic service line, how many cases that were between 0 and an hour long did they do? How many cases that were between 2 and 3 hours long did they do? How many cases that are between 3 and 5 hours long that they do? So when I finally say, if you look at each of these colored boxes, it’s showing the raw number of cases of 0 to 60 minutes. This is showing there are a number of cases of saying hour of 3. And then if I predict how much I expect the next 90 days to produce, how much do I expect orthopedics to do over the next 90 days? That’s taking all these data points for the volume and mix historically, and projecting into the next 90 days, and coming up with an algorithm that’s basically saying there is inherent seasonality in this mix. 

There is inherent growth in this mix. There is an inherence, and I know, by the way, if there are external shocks to the system, if I’ve hired three more orthopedic surgeons or a couple of orthopedic surgeons have left the system. All of that can be baked in, so when you come up with a forecast of how much time I should give orthopedics, you’re starting with zero-based budgeting. You’re saying what is the total volume and mix of cases that I expect orthopedics to do. And by the way, it’s backed by gazillions of data points because I have all of these data points about every orthopedic case that was done, say it’s a 2, 3, 4 years in the past. So when you add up the case link category, you’ve come up with a total volume forecast. Then what do you do? 

And this, by the way, is real data, but it’s anonymous, so we’re protecting the innocent. And I’ll show you what this is saying. So now, if I look at every service line, and I do this kind of a forecast. So imagine doing this sort of a forecast of how much time is going to be required by every service line, which is based on historical evidence. Which you will generally find, is that the total number of allocated blocks that service line currently has, is higher than what the forecast would say. OK. And the reason is most health systems, when they allocate time, look at this kind of a trend, and try and give enough buffer capacity to everybody. This is going back to the carpool lane, and I know there’s going to be 50 cars that are at a minimum. But there are times that there are hundreds, so let me give them 80 cars worth of capacity. 

The problem is a lot of the times, they won’t use more than 50. And then, when you add up all this capacity that’s being wasted because you’ve over-allocated block, you have no time left for open time. And by the way, every day, we have this phenomenon where many carpool lanes are actually empty. So going back to how this works. Imagine doing this kind of a forecast. And by the way, for the data geeks amongst us, you can get very sophisticated with forecasts. You can await the near data points more than the further data points through instruments like AREMA. Just trust me for a minute, that it’s possible to do this kind of a forecast of volume and mix, and be fairly accurate about it. 

So this is actually a study that we did, where we looked at the current amount of allocated block across a 25 OR system. 1600 block, most of them were allocated very little with open time. And then in the next three columns, you are seeing what this forecast said. So what the forecast said was, by service line, you’ve allocated this amount of capacity, but if I actually did a sophisticated forecast, it would show how much they actually need for the next 90 days, OK. And I’ll come to these other two columns in a second. But right off the bat, notice that the amount of capacity that the forecast was saying should have been reserved. Which 220 blocks less than the total capacity, theoretical capacity of the. OR, which was 1600 blocks. That’s problem number one, OK. 

Problem number two was to say, now, if I am going to– so there’s to concepts. I’m going introduce. The first is what we call open-open time. When you go and talk to General Surgery, and say, you’ve got 345 blocks, we think you should have 287. The first thing you’re going to hear is we’ll never get our cases done. Of course, we know that. But then if you tell General. Surgery that, by the way, I’m going to give you 287 blocks, but I’m going to create 220 open blocks next quarter for you to be able to put your hands up. And if you need to fill them, give you a simple tool that we discussed three weeks ago. 

So access to the OR beyond the capacity that I’m giving you is going to be very easy. You might have a shot of winning that argument. Because squeezing capacity out and creating a baseline level of allocation for each service line is really, really helpful because it creates open time across the system. The second level of efficiency is to say, now, within general surgery, I’ve got 50 surgeons or 100 surgeons. Some of them have more predictable volumes than others. Some of them do high volumes more predictably than others. Others do high volumes, but they’re not very predictable. They’re very volatile. So if I go back here, back to this example, each surgeon had the characteristic of how predictably they have done cases. How volatile is their case mix over time? And are they guaranteeing me a certain amount of business every quarter? 

So if I took the purple line, for example, I could draw a line through it saying, I can guarantee that Dr.. Purple is at least going to use 8,000 minutes in the OR, OK. So then allocating him 8,000 minutes, I can be rest assured, it’s the minimum number of Toyota cars, 50, that will travel on the highway. But there are also open lanes for that doctor to use, or that car to use. So the same logic applies when you now say, let’s create a second level of efficiency, by saying, let’s take that 287 general surgery blocks. Look at the data of the performance of every surgeon within that, and come up with an individual allocation underneath that. And create what we call service line open time. So if you take the full pie of blocks, slice them into those that are given the full 14, 79 across service lines to create 220 open time blocks. Then go and create a base level of allocation that you are very, very confident that the underlying data supports that surgeons will actually use. 

You’re able to create service line open time, which allows you to now decentralize the management of block time. It basically says, if I’m the head of general surgery, I’m actually being given 287 blocks, which the system is recommending be fully reserved. 235 is reserved to individual surgeons, but 52 is open time within my service line to manage as I choose. Now, if I need more time than that for general surgery, remember there’s 220 truly open-open blocks that I can reserve. Obviously, this works really well when there is a simple way of accessing open time. We talked about that three weeks ago, that you can create an OpenTable metaphor. We call that exchange of open time. But the proof in the pudding, really the last two columns are as important as anything else because when we actually measured how many blocks were needed by each individual service line, we found that for almost everybody. 

Then the amount of blocks that’s needed that was needed was less than even what the forecast said. And the reason why that’s the case is the forecast very forgiving. Where the forecast basically says is if I find surgeons that use six out of eight hours most of the time, we’ll just round it up to a full day. So it’s rounded to the block, and not to a minute because the smallest unit of time in an OR is actually not a minute, it’s the smallest length of case you can pack into the OR. So to even within this allocation if you will, we find for the same reason seasonality happens that our clinic conflicts that are conference conflicts, people don’t have enough of a book of business. This is, in some sense, a pretty lenient forecast, but it still can squeeze a lot of open time out of the system. 

All right, so that’s really the last of the sessions that we were planning. And I just wanted to summarize the lessons learned from the four webinars we’ve had. The three of which, we did in the last six weeks. The first, if you remember was, it’s really hard to take time away from it, when once it’s allocated. Once you give me every. Wednesday of the week, I’d rather give it pass it on and legacy to my kids, than give it back to you in the OR. There’s no way you’re going to extract them from me. Certainly, it’s going to be very hard for you to extract that time from me if you use block utilization as a metric because that’s completely not actionable. Instead, if you identify large chunks of time being left on the table in which a case could have been put in, the chances you’ll be able to right-size blocks using collectible time is much higher, but that was a topic for another day. 

Similarly, everybody has a block-release mechanism. Everybody has an order-release mechanism. Again, fundamentally flawed because it’s based on averages. 70 order-release, you know my head is in the freezer, my feet are in the oven, on average, my body temperature is OK. It’s the way averages work, so averages are completely broken because many people look way more than seven days in advance, and getting them to give their time up earlier is what you really want. And creating this mechanism to exchange time what we called OpenTable for open time is really the best way to create an open marketplace, that was also a topic for another day. The third is we all want to live in a data-driven culture. It’s impossible to live in one if people don’t believe the data, if the data isn’t really readily accessible if it’s not really easy to use and navigate through the data. And so much more user-friendly, provider-centric tools. 

And then today, this was about saying, OK, in addition to that, if you have to round out your portfolio of tools, and you want it to look at the way to start afresh and say what should my block allocation be, there is a way of doing zero-based budgeting by looking at the volume and mix of cases that each service line or each block owner has done in the past to create a forecast of an allocation with a view of only giving up enough reservation of capacity that you need to that creating enough open time for other for people to be able to share. So that’s really the set of lessons we wanted to impart based on our experience. And the way I want to round this out is talk about if you wanted to move to this model of doing these things, what does it take? 

The first thing it takes is to move away from what we call admiring the problem. The biggest issue we find in working with health systems is the legacy of believing that descriptive analytics is enough. My EHR tells me minutes used, volume utilization delays. And we find that because we’re using the wrong metrics, like block utilization to compare performance, and we don’t make much many decisions based on that, it’s hard to make decisions on that. A lot of our focus when it comes to analytics in the perioperative world seems to be around describing, or what I call, admiring the problem and trying to diagnose why that happened after the horse has left the bar. It’s not very useful to tell me if my stockbroker keeps sending me reports saying. I did terribly in Q2. Well that’s not very helpful. 

If they tell me what will they do for me in Q4 and Q1, and then measure themselves against it, that might be useful. The beauty of where we stand is everybody now has invested tons of money in the EHRs. We have all these timestamps. These timestamps tell us things like the volume and mix as you saw today. They tell us how long before they have surgery to people typically tend to book their cases that should give you a trigger to release time. And so, if you focus on predictive and prescriptive analytics, not for its own sake, but for the sake of transforming broken processes. That’s the four biggest processes that we believe that are broken in the way we manage or our capacity is, one, how we allocate it. 

Two, how we rightsize it and take time back. Three, how we release and request information. And four, how we report against it and use it to make decisions. So all this mathematical gobbledygook of probability theory machine learning stochasticity is all interesting. It’s only interesting if it can be packaged in a way that helps you change fundamental processes. It’s the people, process, and technology, but without the technology, we’ll continue to do garage sales when there is eBay. Without technology, we’ll continue to allocate capacity in ways, and then be surprised that we didn’t use it well, right? So the technology has to be a big part of it understanding that the people in process and the governance around it needs to move to a different place as well, but until you can use this to prescribe future action. 

So a part of my life. I spent at Google. And one of the things we launched at Google while I was there was Waze. And if you think about what Waze is doing, Waze is looking at historical data, and it’s saying what is going to happen. It really doesn’t help me to know that it took my mom 50 minutes from San Francisco airport to get to Los. Gatos where I live. It really helps me to know when I land, how long it’s going to take me. And by large, should. I take the highway, should I take the side roads. So the idea of predicting what’s going to happen and then prescribing what I should do because I can also make a lot of useless predictions. Like the sun’s going to run in the morning, three people are going to fall in your ER tomorrow, well, what are you going to do about it? 

So when you take historical data use it to make sense out of it. And then use it to make decisions, magic happens. More blocks are released, we right-sized blocks appropriately. We forecast was when you use how much time and we have a fact based discussion with highly data driven fact based argumentative people like me. I’m saying surgeons, service lives, where you throw garbage at me and I will not believe it. But if you go deep into the logic for why we’re making these decisions, it’s a lot easier to move tradition and culture. The last slide, the systems that are adopting tools like these are doing a few things well. One is they’re looking beyond traditional admiration of the problem and the descriptive analytics and dashboards to actually using the data to highlight opportunities for block right-sizing, block release, and request block allocation, reporting, et cetera. 

We talked a little bit about predictive is what predictive does. On our side of the fence, I know vendor is the dirtiest six letter word in the English language. There are many people who will tell you about predictive analytics. Problem is, you don’t really need to know about predictive analytics. What you need to know is how do I change block allocation? Now, the fact that we happen to use predictive analytics is interesting, but that’s sort of like trying to explain the Google search algorithm, instead of saying you know what, I’ll give you a web page put some words on it, and you’ll get magic out of it. So the underlying tools have to be very powerful. But from a user perspective, they have to be really easy. And they have to make change happen. 

The fourth is a touchy subject and I hope you take this in the right way, but health systems have spent so much on the EHRs. And frankly, I believe minimizing their investment in IT, and so what ends up happening is if the role of the IT function is essentially to keep the lights on as opposed to innovate, and even if it is to innovate, we’ve spent $50 million for example, building a platform called iQueue. It’s hard to imagine that any individual health system is going to spend that kind of money just for itself. And the reason why we don’t all build our own running shoes is because Nike has spent billions of dollars building it, so scaling innovation like this internally is very, very hard. 

Most of the time when IT is considered a cost center, being giving it $50 million to go do these sorts of things is just not practical for every health system. And then we run into a lot of situations where there is this belief, oh, my EHR is going to do it. Well, I’d like to know when. When is my EHR going to move to from the point of you know diagnostic analytics and descriptive analytics, and admiring the problem to actually solving it. And then finally, this idea of, do you have surge in leadership? Do you have perioperatives leadership that is willing to say,. OR time is precious. What we’ve done in the past isn’t necessarily going to work in the future. 

If you think about the volumes of– just as a country, if you think about the fact that we have 5,000 hospitals that have invested $200 million each in infrastructure, we’ve got a trillion dollars of infrastructure. And the volume of patients you know an aging population and the growth of how many patients we’re going to see is growing up dramatically. We cannot keep spending our way out of trouble by building more OR, so we’ve got to squeeze the maximum amount of inefficiency that we can from the ORs we have. And then, finally, every health system we work with or even don’t work with, have some internal project that’s done an Excel. Spreadsheet, that’s done on Tableau, that’s done through paper based reporting. 

But with all due respect, does not scale. What does that mean? I could build a little search engine on my computer today that works for me and 10 other people. Will it really scale to everybody in the office? Similarly, when we build these little tools that we have proprietary, one person knows how they work, three people are engaging in building the analytics. Will they actually work for 500 surgeons will they work in scale and change over time as our situations change, as we change service lines, as we hire surgeons, as we as surgeons, leave the system, as clinic schedules change, the OR schedules change. So these are some of the lessons we’ve learned in terms of the type of mindset, governance, culture that it takes to move to a world where we are able to rely on good math to make good decisions and this is just backing up what we all know that if you look at just the population growth, how it’s aging, and how many of us are already suffering from chronic diseases. 

The fact that the number of surgeries, the volume of surgeries we will do in the country will keep growing quite fast. And we just can’t keep up with that by building more alarms. So throwing supply at the problem is really not the answer. It’s actually making much better use of our allies. Let me stop here. Marianne, I don’t know if there are any questions. So the first question was, is your tool EHR-agnostic? Yes, this tool is EHR-agnostic. In fact, all of these to the research are agnostic because everybody has timestamps whether it’s Meditech, whether it’s Paragon, whether it’s McKesson. Certainly Epic, certainly Cerner. So if I go back to the logos I showed you, it’s a mix of half and half academic and community. It’s also half Epic, and the rest Cerner and Paragon and McKesson. 

So certainly that’s– it is the EHR-agnostic. The second question was, resulting from more capacity improvements, have any of your customers chosen to shut down any of their lives. Absolutely, so it’s not so much a question of shutting down as how many OR should I be running each day, so depending on how– we often talk about most health systems have a staffing shortage, be it anesthesia, be it nursing, et cetera. If you’re able to pack your OR better during business hours, your ability to both run fewer hours and not have overtime is something these tools certainly solve, happy to dive into more detail with you about them. The third is what best practices do you recommend to make collectible time a more effective policy at my hospital? 

So we discussed this one at length a few weeks ago, certainly I would point you to that webinar. What we have found is when this sort of tool, collectible time-type tool is exposed to all our committee, the bare adoption rate is pretty high because everyone has been struggling with block utilization as a metric to be able to right-size block when they’re presented with an alternative that makes a heck of a lot of sense because it’s identifying truly usable chunks of time. Then the collectible time portion of the logic makes a lot of sense to them. Then if you’re interested, we’d be happy to go out go dive deeper into collectible time. Although, the webinar wasn’t necessarily focused on that. And finally, what do you– how do you recommend having conversations with surgeons about collectible time when they feel their time is unfairly being taken away. 

Again, this refers to the collectable time webinar. I think there– underlying all of these tools is the idea of governance, policy, and buy in. What we have found is, when you actually present the logic of whether it’s predictive analytics to do right-sized blocks for collectable time or you present the logic for how do you should you allocate time, surgeons, anesthesiologists, data folks, are really quite interested in it because this is the promised land that data points in the EHR I have always promised to lead us to. And now, if you can actually make sense out of that data in a way that you can defend without– you know, I’ll call it a religious belief, block utilization because you know it conflates a 10-minute first case delay with a four hour chunk of time left on the table. There’s not much you can do with 10 minutes, but you can fit a couple of cases in four hours. Those two are not the same capacity left on the table. But happy to have that discussion in more detail when necessary. 

MARIANNE BISKUP: Thank you, all, for joining us today.