skip to Main Content

How Emory Healthcare Optimized Infusion Scheduling to Significantly Decrease Wait Time Webinar transcript


JEN WOOD: Welcome to the Association of Cancer Executives. Hot Topic webinar. My name is Jen Wood,. ACE administrative and meeting manager. And I have a couple of ACE updates before we begin today’s webinar. ACE’s first international meeting, the interventional oncology leadership conference, will be held in London, England at the Conrad St. James Hotel from November 12th to 14th. Information and registration details are available at We have a discount code available for anyone interested in attending. Please contact ACE headquarters for more information. 


The 2018 Annual Meeting is being held in Portland, Oregon at The Nines Hotel from. January 28th the 30th. And registration will be opening shortly. If you are dialing in to today’s webinar from a phone, please be sure to put your phone on mute. And do not put the phone on hold if you need to step away from this webinar. We will have a question and answer period at the end of this webinar. Or you can answer enter your questions in the comment box on the. Go to Meeting control panel. 


I would like to thank ACE platinum sponsor, LeanTaas, for assisting in bringing this webinar to ACE today. Now, I’d like to welcome our speakers. Sheryl Bluestein is Vice President of Operations for Emory. University Hospital Midtown. And Mohan Giridharadas is. Founder and CEO of LeanTaas. They will be discussing how. Emory Healthcare optimize infusion scheduling to significantly decreased wait times. Now I’m turning it over to Sheryl and Mohan. 


SHERYL BLUESTEIN: Right. Thank you so much, Jen. Just to let you all know how the next hour, this is Sheryl Bluestein. We’ll go through an overview of our Winship Infusion Center at Clifton. I will also discuss why the current approach to infusion scheduling results in long wait time. We’ll then go into our actions taken for Emory Winship. And then we’ll open it up to the Q&A. With that, we’ll go ahead and jump in. So just to tell you little bit about Winship and our infusion centers. So Winship is an NCI comprehensive cancer center that operates five oncology and fusion centers in Metro Atlanta and the surrounding area. Our total chair count for the five sites is 150 chairs. 


Emory Healthcare also operates three non-oncology infusion sites that have 12 chairs. Today, we’re just going to focus on our infusion center that’s located at our main Emory campus. And that one has 69 infusion chairs. And it’s broken up into eight days. It also has a separate phase 1 unit as well as an injection clinic. We typically– at that site, for the 69 chairs, we see about 45,000 patients per year. We’ve actually grown about 10% year over year, so a very high volume center. We typically see about 15 to 20 add-on patients per day, a lot of those being blood or hydration. We are open seven days a week. We are open from 7:30 AM 7:00. PM, Monday through Friday. And then we’re also open a more limited day on Saturday and Sunday. 


We are on a Cerner medical record. And we are still on paper chemotherapy orders, but working to get to CPOE soon. We do have an on-site pharmacy at this infusion center and upgrading it to a USP-800 pharmacy. We have about 50 nursing FTEs. And that includes our nurse clinicians, our MAs, Our LPNs. We also have two APPs as well as our shift nurse managers and a unit director. With that, we will move into some of our challenges. 


So as a site with that many chairs and that much– with the staff, we often have long patient wait time. We’ve averaged about 30 minutes from our appointment time to chair prior to some changes that we’ve made. We also have– we end up with some slow mornings for our 7:30 AM chairs. And then we get the big wave of patients between about 10:00 AM and 2:00 PM. Many patients are coming to us getting their labs drawn, seeing the physician, and also having their infusion all done on the same day. So trying to have all of these appointments running on time is particularly challenging for us, which results in some patient dissatisfaction. 


Our Press Ganey scores for the wait time and registration area is about an 83.4 where our goal has been a 90.3, at least, in the 50th percentile. We often do service recovery with patients who have waited at times over an hour, still having to do a good amount of service recovery. Staff dissatisfaction we’ve had, it’s challenging for our nursing staff to get to lunch many times as they’re available in the mornings. But then the patients come right right around the same time and right around their lunch time. And then resulting in lower than desired employing engagement for that center. 


We’ve done a good amount of prior process improvement in lean work. We’ve worked with a team from Georgia Tech Institute to come in into a full process flow map. We’ve implemented a visibility system. So given a 69-chair infusion center, rather than having our shift nurse managers consistently walking around to see which chairs are available, we created a system that has– by color coding to see which chairs are available or which chairs are dirty, and we’ll soon be able to seat. We added pagers for nurses to be alerted when their chemotherapy is ready. So rather than our nurses walking to our pharmacy, getting in the window, and staring at the pharmacist, we’ve implemented pagers to let them know when their drugs are ready. 


We also, about two years ago, opened up an injection clinic, which has been a huge patient satisfier, as well as staff satisfier. This is a three-chair quick turnaround treatment site. And it’s staffed by a LPN. And we do a lot of. Prolias, Xgevas, and some of those quick treatments there without those patients having to go back into the larger infusion center. But with all these improvements, we still we’re not getting to the wait times that we really needed eliminating, really, the wait time. So we’re really trying the goal of smoothing our schedule, so not having those peaks and valleys like we’ve had. So with that, we had been in conversations with Mohan and team in understanding the IQ product. And then we had some goals going into that. 


When we started talking with Mohan about a year ago, our goal is being really to unlock capacity without having to add chairs. So one of our challenges is that there’s only so– we really didn’t have the faith to continue to add infusion chairs that we really needed to become more efficient with the chairs that we did have. We had another goal, obviously, of improving our patient satisfaction. We were– our goal being at 95.6 to recommend. And we were not there. We were around the 92, 93. And then also, another goal of certainly improving our employee satisfaction. Somebody is not on mute. We’ll give time for you to mute your phone. 


We also have a low– had a low employee engagement before that we were really looking to improve, that we needed to improve. Our goal being to get to 75% engagement index through Press Ganey. So a lot of goals that we set forth as we began working with. LeanTaas team. And with that, I will pass it to. Mohan to talk a little bit more about it. 


MOHAN GIRIDHARADAS: Great. So thank you, Sheryl. Let me just start by describing why chair utilization and wait times are so tightly linked and why the current approaches to reduce in scheduling leave you with no choice but having long wait times. So let me just start by defining chair utilization. So at any point in time, if you’ve counted how many patients are in chairs versus how many chairs are available and empty, you get a sense for utilization. And if you just click the animation, what you see is the typical pattern of chair utilization in any infusion centers. It starts out low in the morning and then it peaks in the middle of the day. 


So Sheryl, just click one of these, yeah. So it starts out low in the morning, rises to a peak. That peak usually occurs sometime between 11:00 and 2:00 PM. And then it drops off slightly more simply in the afternoon. We’ve now go to over 150 infusion centers. And this pattern is true almost 100% of the time. So why does this matter? If you go to the next page, there’s a key concept on the next page with a little border gap. So when you think about chair utilization and patient wait times, it’s not a linear curve. Meaning, as you start approaching full utilization, the wait time grows exponentially. Meaning, a 10% increase in volume could result in a 200% increase in wait times. So it starts to go up really, really steep. 


Now, the key thing to understand is the linkage between variability and utilization. Pretend, for a moment, this was a very predetermined exact process. Meaning, every patient took exactly one hour, no more no less. If it was that predictable, you could push your chair utilization all the way to 80% or 90%. And there wouldn’t be a problem. But the reality of life in infusion is variability is very, very high. Some people have a one-hour treatment, some have a two, some have a six, some have an eight. And even all two-hour treatments are not exactly two hours. 


Some could be as little as 30 minutes if they don’t get their blood counts right. And two hours can go as long as four or five hours if they suddenly need something in between. As this variability kicks in, the point in time at which you hit the brick wall becomes earlier and earlier. In fact, in the extreme, it could be at 60%, you start to experience this exponential lift in wait patterns. So mathematically understanding utilization, variability, wait time, and the nonlinear equations becomes key. 


So if you go to the next page, you start to see how this plays out in more [INAUDIBLE] So as I said, the treatment length varies by regimen. It varies by patient. You’ve got lots of layers that are not predictable or controllable. And patients who often arrive late need extra time. And nurses sometimes call in sick. As a result, if you laid out the patients arriving on time, the length of the. Tetris block, being the duration of the treatment. Most infusion centers play the Tetris game like this, where you can see lots of holes and the fact that it climbs all the way to the peak. Now why is this a problem? It’s a problem for two reasons. 


One is, the peak only lasts two or three hours. But your nursing shift is eight or 10 hours long, which means, by definition, there’s a mismatch. So if you start for the peak, you will have too many nurses in the off-peak, which is an FTE cost that’s unnecessary. If you start for the off-peak, you’ll have insufficient nurses during the peak and the weekends would become even worse. And so, you’ve got a catch-22 situation where neither alternative is very good. 


But the bigger problem, the second problem is whenever a peak approaches capacity, a system becomes mathematically unstable. So what do I mean by mathematically unstable system? It means that every metric will go into the tank. The easiest way to think about this is the freeways during rush hour. During rush hour, there’s not enough capacity in the freeway to put even one more car on the go. Therefore, all metrics are in the tank. A 20 minute drive now takes 100 minutes to beat. A fender bender that should take 10 minutes to clear takes two hours to clear. And a fender bender that should delay 10 people delays 10,000 people. So every metric begins 5x worse, 10x worse, 100x worse, 1000x worse. 


The reality of life is infusion is a series of fender benders waiting to happen. The pharmacy will run late, the mix will run late, the labs will back up, a patient will call– a patient will react badly, a patient will show up late, a nurse will call in sick. If any one or more of those fender benders happened early morning or late evening, it’s OK. You’ll recover in 10 minutes, and it’s fine. If any of them happened in this middle period of the peak, it’s like the big-rig crash on the freeway during rush hour traffic. The system will be a mess for the next three to four hours. By the time you dig out of it, it will happen again. You know, then it’s the end of the day. And this will happen again tomorrow like “Groundhog Day.” This is what inclusion centers have to deal with. 


So how do you solve this problem? To go to the next page, the way you solve it is by doing the following. First, you’ve got to predict in a mathematically robust way what the volumes are on. Monday, on a Tuesday, or a Wednesday are going to be. Second, you have to predict the mix. Of the 100 patients that come on a Monday, how many are going to need a one-hour treatment? How many will need a two-hour? How many will need a three-hour? So get the durations right. 


Third, you have to factor in how accurately that particular center can forecast durations in the first place. So if Winship said it is a three-hour treatment, we could look at the data and say, how accurate is that? It is really somewhere between two hours and 45 minutes, and 3 hours and 15 minutes? In which case, we’d say four is very, very accurate. Or is it between any numbers in two hours and eight hours? In which case, you’d say, their ability to predict the three-hour is not that accurate, which means you could make some adjustments. So once you make the mathematical adjustment, the trick is to lay out the Tetris blocks in a way that gives you the flattest profile possible and builds all of the uncertainty into the lake of the Tetris blocks. Why is this magical? 


For four reasons. One, it unlocks capacity. So when you unlock capacity, what that does is it walks you away from the edge of the tip. So it’s still rush hour traffic. But it’s like rush hour traffic on a bank holiday or a school holiday, where, yes, there are lots of cars on the road, but it feels much, much better. This is magic for flexibility. Meaning, if you now suddenly get patients showing up late or add-on patients at the last minute, you’ve got chair capacity to put them in. The second thing that really helps out is it flattens the workload. Meaning, nurses no longer have the period of too much time early morning and late afternoon, and not enough time in the middle of the day, so consistent workload throughout the day. And it fits the nursing schedule. Nurses who starts early leave early. Nurses that start late leave late. And so this lets you run the infusion center in a much more strategic way, which then reduces the wait time and improves the flow. 


So this is kind of how you think about the mathematics in infusion schedule. So in order to do this, what you have to do is you have to pre-engineer the Tetris block so you get a flat profile. So as you animate each step, the first thing we do is look at the data and figure out what are the right logical duration buckets across the top? One-hour, two-hour, three to five, six to eight, nine plus. Having done that, we then set up appointments at 10-minute intervals. Most infusion centers will often set up starts of infusions every 30 minutes. If you start infusions every 30 minutes, it’s a bit like JFK airport in the ’60s, would have flights taking off every 30 minutes. They could only do 400 flights a day. Now they have flights taking off every 90 seconds. And they do 4,000 flights a day. 


The airspace doesn’t change the number of [INAUDIBLE] but it unlocks a lot of capacity when you have [INAUDIBLE] started with this. The third thing we do, which it turns out is actually quite important, is sort out how many simultaneous starts you can do. So at 8 o’clock, you can start two three-hour to five-hour appointments and two [INAUDIBLE] and so on. Just because we’ve got start times every 10 minutes, it doesn’t mean you have to start appointments every 10 minutes. That just gives the patients more choice. Now, what this does is it changes the dialogue between the scheduler and the patient just ever so slightly. 


Rather than telling the patient, when would you like to come in in the morning, and just finding when a chair is open, the scheduler now talks to the patient and says, oh, Mrs. Jones,. I see you need a six-hour to eight-hour treatment. I can offer you 8:00 AM, 8:10 AM, 9:10 AM, 10:00 AM, or 10:50. Can we make one of those work? You don’t have to get all of them, right, but the more you steer the right patient population bucket to the right start time, the better engineered your Tetris blocks become, and the more likely you are to get a flatter profile. So this turns out to be one of the key secrets to getting infusion schedules optimized. Sheryl, you’re going to have to click through several animations and then get to the next page. You just [INAUDIBLE]. Yeah. Back to you. 


SHERYL BLUESTEIN: All right. Thank you, Mohan, for that overview. So now to get into our– so Emory Healthcare, we went forward with the IQ system. We signed our contract over back in December, and we had our kickoff meeting in January. So within our kickoff team, we made sure to include all of our partners from the clinic, the lab, pharmacy, all those folks who either impact the infusion center, or are impacted by the infusion center. We also worked on the set-up parameters. So one of the things that the iQueue team needed was basically, how is our infusion center currently set up? What are our hours of operation? How is our staffing by day of week? Now, that includes both nursing and pharmacy. Also, about nurse touch times. 


So how much time does each nurse want to spend with the patient one-on-one one before they begin treatment? Either before and after their appointment or their chemotherapy. We also talked– as. Mohan was mentioning, about appointment start times. We were scheduling, previously, every 30 minutes. So we changed that, and we’ve actually moved down to the 15 minutes, so increasing the number of patients I can get in. So that was one of the big changes there for us. We also did a good amount of training and scripting for our clinics and infusion centers, working on– because previously, patients who had been coming at a certain time, that appointment may no longer be available, when we went live with our new schedules. So we had to do a good amount of scripting around it. So basically from kickoff to go live was about five months. So we went live in early June of 2017. 


And actually, our first day I had to take a picture of our waiting room and send it over to Mohan because I don’t– we were a little shocked. Our waiting room at 2:00 PM on a– I believe it was a Monday– was empty. And I don’t think we have previously had a empty waiting room at 2:00 PM, so it was quite exciting for us. Now, we did– you know, there were days since that we have had patients back in that waiting room, and it hasn’t been all smooth, but it was really exciting that first day, and it really– our staff and team were very excited to see that. Initial results. So I think our story is just now being told. 


So we are about four months out from our go live, so there is certainly a lot more that we expect to see. We’ve had some challenges around the data. So each day there’s a daily huddle and a history that comes out at 6:00 AM, my favorite email of the day. So one of them talks about the history of how yesterday– how you did yesterday and your compliance. And then the other one is the huddle preparing for today. We did have some issues with the data feed coming out from Emory, so we have since gone in and worked to correct that, but it did take us a second to get that data corrected. So anyway, on initial results, looking at three months prior to our go live, and then three months post. 


We did see a 21% decrease in our patient wait time while volumes were relatively flat. So really being at about a 9-minute decrease for our average wait time, which to us and to our patients is huge. And then during that peak time, that 10:00 AM to 2:00 PM, we actually saw a 10-minute decrease in our wait time. We saw improvements in our. Press Ganey patient satisfaction scores, and that question that talks about wait and registration area, we’ve seen an increase in that score as well of at least a point. And we’re also seeing a decrease in our overtime hours from prior year, so on the financial component while [INAUDIBLE] we’ve seen that decrease in overtime. 


We also, before we went live, sent out a survey that was really targeted at staff satisfaction, mostly focused around our nurses. And talked about getting lunches and satisfaction with the job in general. We have not done our post. I have not got the results back from our post go live survey from them, but expect to see that our employee engagement and their response to those questions will be very favorable. And our next slide here, I know this is a little bit of a busy one, but it really talks through our wait times. And you can see our biggest improvements are earlier in the week, that. Monday, Tuesday, Wednesday where we’ve had some really significant drops in wait times. Even where our volumes has decreased, you can see our Monday average is dropping about 28% average wait times, 25% on. Tuesday, and Wednesday, 25% as well, even where volumes actually were up on Monday and Wednesday. So really good results there. Fridays continue to be our challenge. 


Our volumes have gone up on Fridays, and so one of the things that we can see is where we need to better shift some volume. But we’ve also seen some decrease over on Fridays as well, in terms of wait time. This one I think is the slide that I’m most excited about. And you can see our biggest bucket for wait times, which is– I hate to say this– was our 60-plus waiting time, 60-plus minutes. After our implementation, that biggest bucket shifted to our earlier– or getting them– our patients back before– even before their appointment time. So really some huge improvements. You can see on the screen the change in the– from the before with the yellow and the orange to the after and just a really nice increase in the– 57% increase in our patients waiting between 0 to 15 minutes. So really good improvement here. 


So some of our next steps, then, moving forward is better incorporating the daily huddle. I think we got a little shy when some of our numbers were a little off of how much we were going to push this out to our whole infusion center. We wanted to make sure that the data was 100% accurate that we were sharing, and so we really– we believe we’ve corrected a lot of those issues. So one of those things we’d really like to do is better utilize that data that’s being sent out on a daily basis at 6:00 AM. It does a look-ahead, I think it’s about five days so you can also see, I’ve already looked and it looks like Friday,. October 20 has already been flagged and looking like it’s going to be a challenging day. 


So some of the things our schedulers can go ahead and do is start looking at that Friday, not adding on to that. Friday’s schedule, seeing where we maybe even could shift some patients who don’t have a doctor’s appointment that day and maybe could switch to another day. So really using that data that’s coming out and better planning for our day. It also gives us a kind of a score of how we did on overbookings and underbookings for that day, and we’ve– I believe we’ve been around that 80% range, so really trying to make sure that we are utilizing the template most appropriately. 


Strengthen the discipline of capturing the timestamp data. So one of our challenges is that we found out with this that we weren’t doing as good a job as we could have been when the patient actually arrives to the chair, as well as when the patient leaves. So making sure that we are getting those timestamps so that our data is accurate. And then also reviewing the templates for continuous improvement as the volumes and mixes of patients evolve. So we had our– we’ve had our template, then, for, I guess, four months now. And we begin to see where sometimes on Fridays, as I was mentioning, it’s becoming a little bit of a challenge. 


We’ve also– are adding more physicians, so knowing that we will likely work back with the LeanTaaS and iQueue team to adjust our template so that they continue to fit our centers’ needs. And another thing that we’re doing, so I mentioned in the beginning that we have five oncology infusion centers. We only went live with the one at our Clifton Center, our largest infusion center. We are working with Mohan and team to put together a package for it going forward with our other infusion centers. 


Because there’s– the staff at those sites have heard about this and have heard from another nursing team that they’re actually getting to go to lunch and things like that, so really looking to spread these improvement opportunities to all of our other sites throughout the Emory system [INAUDIBLE]. So with that, I’d love to open it up to questions, and answer any questions that you may have. 


ANGELA: Hi, this is Angela. Hello? 


SHERYL BLUESTEIN: Hello. ANGELA: Hi. I was curious about nurse-to-patient ratio. 


SHERYL BLUESTEIN: Yeah, so typically our nurse-to-patient ratio, we have eight chairs per bay, and we– our goal is to have three nurses within that– with those eight chairs. So it’s about a three– ends up being usually about three to one [INAUDIBLE]. 


ANGELA: OK. Great, thank you. 


SHERYL BLUESTEIN:. But then we get [INAUDIBLE] to share– we have two nurses in a bay, but that’s our goal. 


ANGELA: Got it. Thank you. 


SPEAKER 1: When you say two– three to one, can you elaborate a little bit on that? You’re saying three nurses to one patient in an hour? Three nurses to one patient– 


SHERYL BLUESTEIN: Yeah. One nurse to three patients. 


SPEAKER 1: OK, so one nurse to three patients for eight-hour shifts, or one nurse to three patients in four hours? 


SHERYL BLUESTEIN:. Throughout that we just try to ensure that the nurses who are covering each other that– typically a lot of our nurses do work 12-hour shifts. So throughout the day, goal being one nurse to three patients, but sometimes that fluctuates when they’re at lunch and it can go up to four. Even it’s gone up to five. Are there are other questions? 


SPEAKER 2: Do you use– do you have nurses who are doing your infusion scheduling, or do you have an assigned infusion scheduling staff that takes care of this, or who is doing the scheduling? 


SHERYL BLUESTEIN:. Sure, good question. So we have a set infusion schedulers. They’re patient account reps, they’re not clinical, and they are– they work together to field our calls coming from the clinics and scheduling our infusions. We also have implemented chair-side checkout, so we have somebody from the clinic who works in the infusion center and schedules their infusion appointments as well. But it isn’t– our nurses are not doing the scheduling. 


SPEAKER 3: I know you mentioned that you have labels indicating dirty chairs and available chairs. How easy was that to build in your system? 


SHERYL BLUESTEIN: Yeah. So our original goal with that was to use our Cerner and– everybody has FirstNet– like a FirstNet-type system like is used in the. ED so that we could see more of a tracking board. We had challenges due to our high volumes of infusion patients. We actually have a reoccurring encounter too. We have a lot of open encounters that we were told that if we went up on that system, that we could possibly cause a lot of slowness for areas outside of infusion. So rather than– we ended up having a team of Georgia Tech students come in and they worked with us to really build a web-based system, home-grown type of system that basically has the dots for each one of our infusion chairs. 


They’re all white, and then when a patient is assigned to it, it turns it red. We just click on the dot through the system and turns red, and then when the patient leaves, the nurse can mark it– click on it, it turns brown, and then once it’s ready, it’s been cleaned, just click on it again and it turns green so that our triage team– so we actually have a team of usually about two nurses sitting up at the front of our infusion center– are monitoring that. So that they see when it’s turned green and that that chair is back available. So a little more manual than we would have liked. 


But rather than waiting on the IT component, we went ahead and did a homegrown system. And actually so far, it’s working really well. And just given the– before, we had our charge nurses just walking, and it’s a pretty big space with these chairs, so they were continuously walking around with pen and paper noting which chairs were open and which chairs were dirty. So it’s worked well for us. 


SPEAKER 3: I have one more question. In regards to your– I saw one slide that had a color-coded time slot with blue, maybe indicating the certain time. That– is that showing the availability of chairs? That slide was very fast so. I couldn’t capture it all. You had a slide that had the whole day from 8:00 AM to 5:00 PM, and within that slide it showed– 


MOHAN GIRIDHARADAS: It’s the template grid shell. It’s the grid page. 




MOHAN GIRIDHARADAS:. Yeah, this one. This page, right? 


SPEAKER 3: That’s correct. Yeah. 


MOHAN GIRIDHARADAS: Right. So let me explain what you do with this page. So with the mathematical optimization that’s done is figured out the right way to sequence appointments of varying durations. So have you noticed every hour of every day is different. So 8:00 AM to 9:00. AM on a Monday does not look like 8:00 AM to 9:00 AM on a Tuesday. The different mix of [INAUDIBLE].. 


So what you’re seeing here is by figuring out the durations and figuring out when we should start appointments of those durations, we’re creating choices for patients all through the day, regardless of the duration they’ve got. We’ve mathematically modeled to make sure we get the right number. This gets translated into the Cerner templates. So when the schedulers are talking to patients, when they know this patient needs a two hour treatment, they can automatically realize what the potential best options to put that two hour treatment would be. And if they comply with this guideline 80% or 90% of the time, they will get the results that Sheryl and the others are getting. Does that answer the question? 


SPEAKER 3: Yes, it does. Thank you. Was it difficult to predict that or were you able to do it with your system? Did it come easy? I guess that’s my question. 


MOHAN GIRIDHARADAS:. Now it comes easy, but it comes easy after four years and possibly $20 million in mathematical investments. But now it comes easy. So that’s the product. 




SHERYL BLUESTEIN:. So this system– 


SPEAKER 4: I have a quick question regarding– how does the system account for no-shows, people that just don’t show up for appointments. Because we have a fairly significant right here, probably about 15% no-show rate. Ho would it account for that? 


MOHAN GIRIDHARADAS: Yeah, that’s a great question. So what we do is we– because we’ve done pattern analysis from the last six months of data and we get data every single day automatically, we’re constantly monitoring the no-show rate. And so usually no-show rates have a consistent [INAUDIBLE] why that would be challenging. So what we do is we assume that the no-shows, we gross up the number of appointment slots [INAUDIBLE] to include the no-shows because the no-show appointment gets made and then broken. So you have to have enough slots to accommodate all the no-shows. We then look to see whether the add-ons counterbalance the no-shows. Meaning if you take hour by hour, what’s the net impact of no-shows versus add-ons? 


Because add-ons add to the number of people turning up and no-shows subtract the number. If hour for hour it sort of balances out, you can be OK. If there’s big swings, like there’s a two hour window where it’s mostly no-shows and not any add-ons and then another two hours where there’s no no-shows and lots of add-ons. This could work for the balance of it, that’s going to [INAUDIBLE] 


SHERYL BLUESTEIN:. And one of the things we’ve seen at Emory with– before, we weren’t able to schedule our add-ons. So we would find out that they needed blood and we would say, OK, send the patient down, or for fluid, send them down and we’ll get them in as soon as we can. But we didn’t really have a slot for them. So our add-on patients would wait two hours, three hours, and just sitting in our waiting room. 


Now with this tool, we actually have slots where they are available and we actually schedule our add-ons into an actual 3:00 slot. And that we’re able to give them a time just like they had gotten an appointment the other way. And it’s worked much better for our add-on patients and their patient satisfaction. 


SPEAKER 4: So that sounds really good. My other question is how do you– and this is for, I’m sorry. I can’t remember your name– from Ann Marie– how do– 




SPEAKER 4: –you account for like that arrive late? See, we’re very clinic– our infusion center is right now separate. We’re going to be joining, but we’re separate from the physicians’ offices, so we’re very dependent on them. And our patients can run sometimes, I mean, between 30 minutes and two hours behind their actual appointment time, which really, as you can imagine, causes a problem. I don’t know if you have any issues like that and I don’t know if this could help that at all. 


SHERYL BLUESTEIN: Yeah, we certainly have– 


SPEAKER 4: Just work with the [INAUDIBLE] center? 


SHERYL BLUESTEIN: Sorry, say that one more time? 


SPEAKER 4: I guess that’s my question. And I don’t know if this– if you’re using this in both the clinic and the infusion center or just the infusion center. Because I don’t know how you could do it with one without the other. 


SHERYL BLUESTEIN: Yeah, no that’s exactly right. And we actually have done something similar. We did work in our lab too knowing that lab, physician’s office, infusion, they’re all dependent on each other. But for the accounting for those patients that arrive late in our historical data, as Mohan was saying, it took into account when timing wise for arriving. But we actually have seen improvements around our patients arriving closer to their appointment time based on how we are doing the scheduling. So we have seen some good benefits of patients getting closer to their appointment times. I don’t know, Mohan, if there’s other sites who are seeing anything– 




SHERYL BLUESTEIN: –different. 


MOHAN GIRIDHARADAS: Yeah. Let me address that. The whole issue of linked appointments, where someone sees an oncologist first and then an infusion, is a particularly complicated problem that current methods are not solving correctly. Here’s what I mean by that. Think of it like a connecting flight, where the first leg of the flight is the oncologist and the second leg of the flight is infusion. 


The reason connecting flights work in real life is because the on-time performance of the first leg is very, very good. Much of the old complaint about airlines, we’d miss probably less than five connecting flights in our life because the first flight is usually on time. When the first flight is the oncologist, it becomes an unreliable system because the oncologist is [INAUDIBLE] for a variety of reasons [INAUDIBLE].. 


Therefore, thinking of it like a connecting flight is mathematically an imperfect model. So what have we done with this? We think of it as a nonstop flight followed by a taxi ride. So now you’re no longer are dependent on the first flight being on time, because you arrive when you arrive and you take a back seat and you show up. Guess what? If you think of it that way, then the nonstop flight is the oncologist, the taxi ride is the infusion. If you are the owner of the taxi fleet, what should you do? You should look at the schedule and decide whether you need to have 20 cabs waiting outside or 200 cabs waiting outside. And so once you understand the schedules and then you start looking at the patterns for what the arrivals are, 7:00. AM, 7:10, 7:20 on Monday, then on Tuesday and so on, you get the flow of taxis and the mix of [INAUDIBLE] that you need just right. At this point, guess what? 


You don’t care if the passenger stepping onto the curb for a taxi came in on an on-time United Flight or on a delayed American flight. So what our mathematical modeling now does is by monitoring the arrival of patients into the infusion center and [INAUDIBLE] wait time profiles with the arrival times, we no longer care whether the patient is arriving from the breast clinic on time or from the cervical clinic on a delayed basis. It doesn’t matter. And because of the optimization, you unlock capacity. So the chances of having a chair available, regardless of whether the patient is delayed or not, have now gone up by an order of magnitude. 


So that’s the way you model it. Because linked appointments, the moment you print out the patient’s itinerary of 8:10 you’ll be at labs and 8:20 you’ll see the physician and 9:40 you’ll have the infusion, the moment that itinerary comes off the printer, it’s already wrong. And so that’s the reason why it needs to be modeled in a different way. AMY: So in this template model, are you kind of saving spots then for treatments of different duration? And then is there an automatic release on that template? So that if you get to be two days out and you still haven’t scheduled that six to eight hour infusion, you have an open slot there, you can use that for three of the two hour infusions? 


SHERYL BLUESTEIN:. You can do that. We’re not really saving– we don’t really do freeze and unfreeze or anything like that, or saving spots. But there is some– if you get to the point where you have a shorter duration and you don’t have a slot for it, you can go ahead– as it gets closer, we do go ahead and put that in a longer slot. So that’s some of the work we have done. But we don’t do the releasing and freezing because typically if we have open spots going in, our add-ons will fill those spots. 


MOHAN GIRIDHARADAS: Right, that’s exactly right. The other thing is if you’ve got a four hour slot, you can’t substitute that as four one hour slots. It looks like the same thing, but it’s not. Because it has different loadings on the nursing staff. The nurses, if they manage one four hour treatment has a lot more workload than a nurse who has to manage four back to back one hour infusions. Also, the probabilities of running long or running short are wildly different for four hour appointments than one hour appointments. And so the way we deal with this [INAUDIBLE] is by having rules that allow you to put smaller appointments in longer slots. Or even sometimes a longer appointment within a smaller slot, but finding a neighboring slot that’s unused. So there are ways to make the mathematics work out without going through the release and reassign process. 


SHERYL BLUESTEIN: Are there other questions? 


SPEAKER 5: Hi. Can you tell us how your scheduler knows how and where to put a patient if they’re not a chemo nurse? 


SHERYL BLUESTEIN: Yeah, so the schedulers– so I’m not sure how many have Cerner Scheduling. We use Millennium. Scheduling, which has a view in how it’s been broken out, and it has the duration. So basically, it’s a visual. They’ve got a seven hour appointment because we have times based on the regimen that we’ve create– or we’ve done based on history of how long that duration is going to take. And then they take that duration and they put it into– if it’s a seven hour duration, we’ve got a six to eight hour slot open, they’ll just plug it into there. So they just basically use our duration document and then plug into the opening slot. So I think our schedulers have really appreciated this because before, it was just a wide open chair. And they would use some logic of the short appointments and try to get some earlier in the day. 


But really, they were working with a wide open chair and putting them in wherever they thought it would work better. But now, it’s more clear cut of where they should plug that patient appointment into. So that’s been– I think they’ve really appreciated this tool. 


SPEAKER 6: Hi. I have two questions. One is, is there a notice of the difference in pharmacy times as far as pharmacy prep or bringing out products sooner? Is there any data on that? 


SHERYL BLUESTEIN: Yeah, so we need to dig more into our data. One of the things we did notice in terms of the pharmacy prep was that we were getting our patients back into the infusion chairs quicker, but that we were then having some issues around pharmacy and getting those drugs out just in the same frame. So that was one of our challenges. So we are looking at where we had improved patient satisfaction time and wait in registration area, we had decreased once the wait in the chemo area. So that pharmacy is now working to tweak their workflow as well. But that’s one where I don’t have good time stamps right now, we are digging into. 


MOHAN GIRIDHARADAS: One thing we’ve found from other infusion centers is the following. Part of the reason that pharmacy turnarounds become long is because they get a [INAUDIBLE] bunch of orders at one time. So sometime around 10:30, 11:00, they’ll suddenly get fix, six, eight orders. And each order is complicated and the mixing process is difficult, and so they start to get backed up. And once you get backed up, it’s difficult to get back on track. So by level loading the schedules, what you are doing is pushing a new order to the pharmacy every 15 minutes or every 30 minutes and eliminating this big rush hour kind of 10 orders coming in in a short window of time. And therefore without needing to change the pharmacy [INAUDIBLE],, you usually find a fair amount of improvement in the turnaround time of drug delivery. 


SHERYL BLUESTEIN: Yeah,. I think one of our– we changed our double check process. So we put that in with going there, so their pharmacy check process actually started to take a little bit longer too. So that was one of our challenges as well. But we also– yeah. Pharmacy is one where we’re working. We added a fast track pharmacist as well, so we have a good number of those injections are under one hour, just quick treatments. And so we pulled one of our regular pharmacists out and we had them working on just those quick treatments to turn those around quicker. And we’ve seen some good benefits with that as well. 


SPEAKER 6: And you said your quick– your fast track is just injections. Is that correct? 




SPEAKER 6: You don’t do hydrations or anything like that? 


SHERYL BLUESTEIN: Not in our– not to our fast track pharmacist typically, but it goes into our others. 


SPEAKER 7: Now, you had mentioned patterns. Is that the data that comes out from these huddle reports? Is this something that you guys started off with this template and then in a week or two time, you’ve realized it didn’t work and your tweaking? Does the company help you guys with that part? Or how does that work exactly, like when you first started and went live? 


SHERYL BLUESTEIN: Yeah so we– before we went live, we worked with Mohan’s team and we got– we probably went through the templates and had maybe up to version 10 or 12 of really just looking at them, having our nurses weight in, different things we may have missed, wanting to have a few earlier morning slots, and various tweaks back and forth. And that was before we actually implemented them into our Cerner systems. So back and forth a good amount until everybody kind of signed up and said, yes, we’re comfortable with these for now. So we implemented those templates. And then we’ve now, about four months in and seeing as the daily huddles come out, some opportunities on Fridays. And then we’re also adding about three new physicians. And so I have a feeling we’ll be in touch with their group very soon to tweak them again. So I think it’s usually, I would say Mohan, I don’t know, maybe a quarterly tweak that occurs. 


MOHAN GIRIDHARADAS: Yeah. Most of our customers tweak their templates once a quarter. And sometimes on the quarterly tweak, they might say let’s tweak. Mondays and Wednesdays but leave Tuesday,. Thursday, and Friday alone. And then in a subsequent tweak, do that as well. So the tweaking is meant to be kind of continuously monitored, but typically every quarter. So the analysis on how well the templates are working is happening automatically every time. And so I don’t know, Sheryl, if you’ve got your email handy, if you brought up what a huddle report email looks like or something like that just so people get a sense for what the email traffic looks like. 


SHERYL BLUESTEIN:. Yeah, I probably just don’t have one handy, but I’m glad to certainly share one with the group at another time. But basically, we can see it’s got– it’s color coded so you can see when you’re under your capacity by two is yellow. And then you could see it gets over capacity, it turns red. And you can really see where your pain points are going to be in that day. And then as you look at the outlook over the next five days or so, you can see where you’re projecting this schedule above. And they flag those days for you so you can go ahead and start looking at those days and where you could make some changes. Are there other questions? Amy? AMY: So with this system then, is this a stand-alone scheduling system, or is this something that you built these templates within your EMR scheduler so it would work with any kind of software system? Or how did you do that exactly? 


SHERYL BLUESTEIN:. Yeah, so we built them within our regular scheduling template. So we use Cerner, it’s a Millennium scheduling, that we just laid them over into what we were currently already using. So not a not a lot– no real big training changes to our schedulers or anything like that. There was a little IS component on tweaking– or not tweaking, but having to update making those more set time frames within our Millennium schedule. So that’s how it works with Cerner. I don’t know of other Epic or Mohan, other clients that you’ve worked with. 


MOHAN GIRIDHARADAS: Well that’s always the case. So it’s independent of your scheduling system. You’re intended to use your current scheduling system. All we do is help you get the templates loaded into your scheduling system exactly as you scheduled. So it’s not meant to disrupt any [INAUDIBLE] systems. 


AMY: Can I ask what hours your infusion center is open? And are you a seven day a week kind of facility or– 


SHERYL BLUESTEIN: Sure. So we are a seven day a week facility. That’s again, this main location here we’re talking about. We are open at 7:30 AM to 7:00. PM, Monday through Friday, and then Saturday 7:30 to I think it’s 4:30, and Sunday 7:30 to 3:30. And then major– we’re closed on the major holidays, so just Christmas, Thanksgiving, but open on day after Thanksgiving, Christmas Eve, those days. 


SPEAKER 8: How long have you had operations running on the weekend? 


SHERYL BLUESTEIN: So I’ve worked with infusion for four years and we’ve had it from there. So I think, gosh, probably since– a long time. [LAUGHS] Now our other site, so we have some smaller sites, four other smaller sites. Those do not currently operate on the weekend, but we may have to look to change that, at least in some of our more northern sites. [INTERPOSING VOICES] 


AMY: I’m sorry. And how do you meet with the physician offices to assure that if they’re having an appointment with the oncologist’s office first and then coming to the infusion center, how are you able to coordinate those times? We do same-day scheduling as well here, and we have difficulty sometimes finding an opening on the oncologist’s schedule that matches with the length of time we need in our infusion center. 


SHERYL BLUESTEIN: Yeah, that’s a great one and that has been a tricky one for us. One of the things we really try to encourage is scheduling that infusion appointment first because it’s a little bit more challenging, the constraints there. So we work to schedule the infusion appointment first. If we do schedule the physician appointment, but typically we’ve gone about an hour after that physician appointment is when we will look to schedule. 


But we have had to make some tweaks to that because with the tool, when we implemented IQ, when we were scheduling the physician’s appointment first and then having to do infusions, we were having some pretty big gaps between maybe two hours, three hours, which is really frustrating for our patients. So we have balanced those. Anytime we get a situation where we couldn’t schedule the infusion appointment first and we’re looking at a possible gap of longer than two hours, we escalate that to our system director who oversees infusion scheduling and just see where we can make some changes to decrease that gap between the two. Knowing that sometimes it does work out because the physician is running so late that there isn’t really a gap, but there are times where there’s been some significant gaps that we’ve had to address. 


MOHAN GIRIDHARADAS:. That’s a great way you address it Sheryl, is going back and forth both ways. Some other ways we’ve seen, for instance, are putting a tentative gap which is longer than you’d like, but then using the no-shows and cancellations to get people in earlier by directly texting them when they can show up earlier. Huntsman keeps a running a waiting list, for example, when people are there for more than two hours, and does everything in its power to get them in sooner. And nine times out of 10, they do. 


SHERYL BLUESTEIN: Are there other questions? 


MOHAN GIRIDHARADAS: Sheryl, with five minutes left, maybe we should just flip to the last page so people, if they think of questions later on, they will have the email address to drop it on the deck. 




MOHAN GIRIDHARADAS:. The last page. 


SPEAKER 9: You said the system works with any EMR? Any EM– [LAUGHS] Any EMH– It works with any of them? 


MOHAN GIRIDHARADAS: Any of them. So far we’ve got deployments with Epic, Sono, Mosaic, [INAUDIBLE],,. Flatiron, Aria, I’m sure I’m missing a few. 


SPEAKER 9: How about Meditech? 


MOHAN GIRIDHARADAS:. Meditech as well. 


AMY: So just to clarify, so if I [INAUDIBLE] would analyze the data over a set period of time, someone from that team would come up with the mathematical calculations of where [INAUDIBLE] 




AMY: –durations and then that [INAUDIBLE] 


MOHAN GIRIDHARADAS:. That’s exactly right. And then the product will continue to monitor [INAUDIBLE]. So it’s continuous cross product, right, in that sense. It’s not a one time thing to get it done. But it’s more a one-time calculation to get to the optimum [INAUDIBLE] and then daily support to keep you running optimally. 


SHERYL BLUESTEIN:. Well I’ve put up the screen that has our contact information. You’re welcome to reach out to me. My email’s at the bottom. I’m glad to answer any questions either IQ related or collaborate on some of our infusion challenges that I think most of us share as we work together to improve. I want to thank everybody for joining the call Yes, and I would like to thank. Sheryl and Mohan for taking time to join us today. And this webinar recording will be posted to the members-only section of the ACE website. Thank you again and. I hope everyone has a fantastic rest of their day. 


SPEAKER 2: Thanks. 


SPEAKER 5: Thank you. 


MOHAN GIRIDHARADAS:. Thank you, everybody. Thanks, Sheryl.

Back To Top