From 3M Health Information Systems
Podcast Episode Transcript: Lowering readmission rates by breaking down silos
Dr. Gordon Moore: Hello, and welcome to 3M Inside Angle Podcast. This is your host, Dr. Gordon Moore, and today I’m talking with Teri Sholder. Hello, Teri.
Teri Sholder: Hello.
Dr. Moore: Teri, tell me a little bit about your role and the organization in which you work.
Teri: I am the Senior Vice President/Chief Quality Officer for BayCare Health System. I’m responsible for our clinical excellence strategy, deployment, insuring continuous improvement. Areas under the clinical excellence umbrella here at BayCare Health System in addition to the strategies include accreditation and regulatory compliance, performance improvement, our team of Lean and Six Signa engineers, as well as clinical documentation improvement.
Dr. Moore: You guys are located in Florida. Is that correct?
Teri: We are. We are one of the leading non-profit healthcare systems located in South Central Florida. We cover a service area of about four counties. We have 15 hospitals in multiple convenient locations, urgent cares. We have one of the largest home healthcare agencies in the nation. We’re one of the largest private employers in the area, estimated a little over six and a half billion in annual impact to not only our region, but also the state, and we’ve consistently over multiple years had a Aa2 credit rating, something we’re extremely proud of.
Dr. Moore: That’s pretty enviable. The reason I wanted to talk with you today is that I had the opportunity to hear you talk about readmission work that you guys have done and I was really impressed. I thought that it was interesting in terms of how data driven, how focused you were, how clear you were in measurements and understanding impact, and also thoughtful about the kind of innovations you brought to bear. I would love to ask you about how that work started and get into how you did it, lessons learned, and things like that.
Teri: All right. Thank you. Let me take you back to 2016. Around 2016 our Board of Directors set a goal for the health system to achieve top quintile in overall quality and safety compared to the nation’s large health systems. What we used was the annually published IBM Watson, formerly Truven, Fifteen Top Health Systems scorecard to kind of gauge our progress. They were the only scorecard that really would nationally rank you against similarly sized health systems. At that point in time we were at the 25th percentile, so not too good, sitting there in the bottom quartile.
Our journey has taken us quite a long way. We’ve seen drastic improvement in multiple measures and we’ve actually made it to the 86th percentile based on the most recent report that was published in April. Although that improvement has just been phenomenal and due to a lot of efforts from a very high performing team of employees and physicians, we were still having problems with Medicare spend per beneficiary and the 30-day readmissions. We know that these two are links, obviously, because in our current fee for service world we’re still getting paid for those readmissions, which is also included in the Medicare spend post discharge 30 days. That’s what we started looking into.
Of course, when the Hospital Readmission Reduction Program was introduced through CMS, this caused a lot of health organizations to say, “Well, we really need to get a grip on this,” not to mention the fact that what’s more important than that is that, you know, it’s a part of our mission, right? We need to serve the communities and having preventable readmissions is really just not acceptable and not aligned with our mission. So we really started putting a lot of focused efforts into how can we address these high risk populations and what we do know is when we look at the CMS Hospital Readmission Reduction Program the reports say, according to AHRQ, the Agency for Healthcare Research and Quality, readmissions have allegedly dropped eight percent since 2011.
One thing that’s really interesting, I say “allegedly” because earlier this year the AHIMA, American Health Information Management Association, and Modern Healthcare both published articles calling into question the real versus perceived reduction in readmissions since HRRP was introduced and apparently this was from a study out of JAMA previously, a year or so earlier than that, that showed that a 63 percent reduction in readmission rates based on the HRRP population was actually related to coding severity. What had happened was there was a change in how many codes, additional diagnoses, were being used in the HRRP program, which impacted the hospital’s Risk Adjustment Score, so it was allegedly resulting in the appearance of lower readmission rates. I just thought that was really interesting and it’s something that I think everyone should be aware of as we’re looking at readmission reduction rates: the impact that documentation and coding can have on this trajectory.
The study basically concludes that the nationwide estimates in readmission reduction since HRRP was introduced could possibly be about 50 percent less than what we thought, so I just thought that was kind of something interesting that people may want to know.
Dr. Moore: I was just thinking that it is interesting if you think about when we look at rates of things for populations how important it is to represent the true illness of the people because, of course, we would expect a healthy person to have a lower probability than somebody with multiple stable conditions, so if there’s an under-representation of that in the data set, then of course the rate may be high, but it may not represent the reality of what’s going on. That’s what I hope has happened since then, that maybe there’s been an improvement in the accurate representation of the total illness of people who are coming out of the hospital. On the other hand, I hope that it’s not gaming the system. Do you have any sense of if it’s one or the other?
Teri: I do have a sense just based on what we’ve done, but I’ll tell you what’s even more important than having that risk adjustment included in there is being able to identify those components that make that person high risk so that when we put programs in place to make sure they have a successful recovery that we’re meeting their needs post discharge, right? Understanding the severity of their condition and making sure that we have a comprehensive picture of that acuity really is going to impact how we set these folks up to be successful at home.
Dr. Moore: In the story that you’re telling about what sort of work you were doing, that sounds really interesting to me. Again, one part that fascinates me is the ability to use data to make intelligent decisions, because I think there’s often a mismatch between the application of resource and the need of the person and that’s waste, you know? We don’t need that. We need the right person to get the right care. How did you guys come to do that better over time?
Teri: Back in some of the early days basically we were kind of learning to crawl with readmission reduction. It was elusive. We had a hard time at that point in time even with the EMRs, we didn’t necessarily have a risk adjustment, so in the early days we were focusing on making sure that we understand who these folks were, who was at high risk, and what could we do for their discharge planning that would help them be successful. At that point in time, I think probably very common in the industry, we had good months sometimes followed by great months, but the bottom always inevitably fell out, so sustainability in those early years was really elusive. We were trying anything short of witchcraft to prevent readmissions at that point, but we just were not hitting the right things.
By 2017 we had put in place seven post acute programs. We had preferred provider skilled nursing facilities, automated follow-up phone calls to patients, telemonitoring. We had advanced practice nurse home visits among some other programs, and in 2017 we started to walk. We started to see that some of these programs were in fact being effective. What we didn’t know at this point in time was which ones. So we had these seven programs pretty much operating in silos, not sure which ones were providing the best results, so we did some in depth analysis to try to come up with the power combos. Which of all of these programs were having the greatest impact in preventing readmissions? With that we started to see a decrease in our readmission rate.
Going from 2017 we were at about an O:E ratio of 1.04 to 0.97 by about mid 2018, so we were starting to see an impact, but like I say, we weren’t sure which programs were working and which were not, so we did a deeper analysis on that to try and identify what effects they were having and actually ended up eliminating a couple of those that were proven to have minimal impact on preventing readmissions.
Dr. Moore: How did you know which of the programs were more or less impactful? What data did you use to come to that conclusion?
Teri: What we did was we looked at each individual program on its own merit. We would look at the readmission rate of patients that were impacted that just had that one program. Say they just had the ARNP visits at home, or say they just had home care with telemonitoring. We kind of segmented it that way. I can’t say that it was a perfect science, but it was very directional and helped us identify which combos of those programs were the most effective.
Dr. Moore: Were there certain places where the dominant intervention was in one particular program and therefore you could say, “Well, let’s look at this facility because they’re mostly doing this thing?” Was that one of the ways that you carved it out?
Teri: Yes. One of the things that we knew right off the bat was that our pharmacy transitions of care program was very effective in identifying patients that couldn’t afford their medications, so we would then help that population, maybe patients that were discharged on duplicate therapies, etcetera. So we did identify the pharmacy transitions of care was having the greatest impact on mitigating risk of readmissions and as we looked at these programs and we were saying, “All right, what is working the best and how can we figure this out?” a big driver was the fact that this was an approximately $13.5 million price tag on all of these post acute programs to try to mitigate risk of readmissions.
The needle still wasn’t moving like we anticipated or like we felt that it should, so our O:E ratio by mid 2018 was pretty close to what it was in 2017. However, what was really interesting was we did a control chart to see if all of our efforts had reduced the variation in any way. At this point in time what we had done is a control chart and what we discovered—the chart was our performance from 2016 through 2017, just to see if those post acute programs had had an impact in reducing variability in outcomes and it had. The upper control limits were decreased in 2018 to 1.12 compared to 1.22 in 2017, 1.44 in 2016, and then 1.48 in 2015, so basically we reduced that variability from 2015 going from 1.48 to 1.12 in about the middle of 2018. That tells us that our results are definitely consistent and stable. We saw this decrease in variation, but like I say, we still weren’t hitting that target.
According to AHRQ, 72 percent of readmissions are medication related and although we had the post acute pharmacy transitions of care where pharmacists were calling patients, following up, making sure they got their meds, we had not addressed the accuracy of the medication reconciliation process in the hospitals, so we were finding med errors post discharge with the pharmacy transitions of care, but the thought was, gosh, you know, let’s being that inside. Let’s make that proactive versus reactive and have the pharmacists involved in the medication reconciliation before the patient leaves, with a goal of just improving that discharge medication reconciliation by using guideline directed therapy, evidence based medicine, watching for duplicate medications, suboptimal dosing, patient comprehension and adherence, and ability to afford their medications.
By bringing that all in the hospital, our thoughts were that we could proactively approach it, giving the patient a better chance when they discharge home of a successful recovery. To do that, of course, is resource intensive, so what we decided to do was a pilot. We selected one of our hospitals and we randomly selected a little over 300 Medicare patient charts. The clinical pharmacist would review for the guideline directed therapies, looking for those things that I just mentioned, no duplicate therapy, optimal dosing, adjustments for, say, renal failure or any other special recommendations. The results were pretty compelling. Out of the 303 records we found 87 to have duplicate therapy, 32 had inappropriate directions for use, 62 had omissions from what the specialist had recommended. It was just so compelling, so we went right from that into establishing the process where the pharmacist would continue to do those medication reconciliations and then compare that to a control group with the outcomes we were trying to achieve. The objective was to reduce readmissions.
While the intervention group had the med rec done by pharmacy, the control group got the usual standard of care with nursing transcribing and the physician signing off. The difference in readmission rates was pretty compelling. The intervention group readmissions were 7.63 percent and the control group without any intervention was 15.5 percent, statistically significant P-value less than 0.05. We were thinking this is absolutely effective. It’s the best thing to do with our patients, so we took this to our executive leadership team. We took it to our board and were approved to add 13 pharmacists across the health system to do a remote version of medication reconciliation.
Currently we have a 97 percent physician acceptance rate of the pharmacists’ recommendations. The patients are going home on the right things. What happens in the process, the physician will review the recommendations by the pharmacist, may modify, and then approves or declines. In 2019 year to date our readmission O:E ratio is at 0.90, which is the lowest it’s been since prior to 2016.
Dr. Moore: Wow.
Teri: Yeah, that’s how we use that data to drive readmissions. Go ahead. I’ll stop and give you a chance to ask a question.
Dr. Moore: The results are huge. That’s absolutely terrific. I’m also thinking about the price tag and how you sell that to hospital leadership. How does that happen?
Teri: The price tag? Yes, it was pricey, but when we go back to our mission to serve the community, it’s the right thing to do for patients. It’s tough. We are in a fee for service world and we do get reimbursed for these readmissions, but the right thing to do is to make sure these folks have a successful discharge and our leadership team here at BayCare, the C suite, as well as the hospital leadership team and our board, are absolutely committed first and foremost to clinical excellence. We’re in a great position to be able to serve the community in that capacity and make sure that we’re doing the right thing by our patients.
Dr. Moore: And that implies that there’s not necessarily an ROI in doing this work; it’s really based on doing the greater good, or is there also an ROI?
Teri: It’s absolutely at this point in time based on doing the greater good, because, Gordon, we, like I say, it’s that fee for service. You reduce readmissions, you’re going to be reducing a hospital’s revenue, so it really is about doing the right thing.
Dr. Moore: Wow. If the policy was standing next to me, I’d want to kick it in the ankle for that. That’s unfortunate, but certainly I’m pleased to hear about the positive results. Has that—you’ve achieved that level in 2019. Is that pretty fresh or has that been around long enough that you think it’s going to stick?
Teri: I think it’s fresh. I think it’s going to be around, so our first four months were—January was the lowest January we’ve ever had. I think we came in at about 0.81 in January. February started to climb up just a little bit. Here in Florida it’s not uncommon to have a higher acuity, of course, in the winter months because we have people that are down here living that do not live here in the summer months. Yeah. I feel like it’s sustainable. Of course, we’re going to keep an eye on it, and there’s so much more to do. We still have not been able to effectively address the Social Determinants of Health. The question is, how do we identify these folks? We have our community needs assessment, so we know pockets, areas, zip codes where we have opportunities such as food insecurities, transportation, housing.
There’s so much about readmission reduction strategies that are pretty much archaic. I heard Dr. Eric Coleman speak recently and it was so compelling because there’s just so much more to do in discharge planning. He was mentioning that with discharge planning we tend to think about the historic information. We tend to think about that person’s goals being, all right, you reduce your sodium by this so that you don’t have heart failure, dah dah dah, and what was so interesting is that thinking not about our goals, but what are the patient’s goals? Sometimes the patient’s goals might not have anything to do with their heart. He was telling a story about a woman named Mabel whose goal was to be able to go to church. It wasn’t anything about her heart. It was about being able to fit into her shoes, because she refused to wear slippers to church. She was embarrassed.
So, you know, if we think about tying the goals to what’s important to our patients and important to our communities, then maybe we can start going down the path of supporting those goals versus thinking like clinicians all the time and believing that just because we know it’s about the patient’s heart the patient thinks it’s about their heart, right? Yeah. It was really good. Some other advice that he gave was thinking about the discharge plan and do that discharge plan with the end user in mind. You know, making it more about the future versus the historic. Does the patient know what to do if the symptoms occur off hours? Do they know who to call? Do they know what resources are available?
Despite the fact that all of us have probably been to tons of presentations on readmissions, I think there’s so much more to explore around social determinants and kind of getting to the bottom of how we can help these folks at high risk.
Dr. Moore: So is that where you guys are going next?
Teri: It is. We’re working on strategies through our ACO—we have a very successful ACO—looking at how can we leverage our physician offices to collect important information, maybe electronically be able to get advanced directives and have the conversations in the physician’s office, to activate the family, making sure that we’re meeting the family’s needs, the caregivers’ needs. Caregivers are often overlooked when we’re doing a lot of this discharge planning, so what can we do to help them, as well as the patients themselves? What resources are out there?
Dr. Moore: Are you guys doing any testing at this point or are you still in the planning phases?
Teri: We’re actually still in the planning phases with the social determinants. Like I say, part of the challenge is identifying who these folks are and then we can add another challenge, too. We can add assessment criteria to identify food insecurities or to identify other social determinants, but we can’t do that without having the resources in place to follow up on it, right? Do we have the food banks identified? Having to understand how we’re going to make sure that there’s a closed loop process takes a lot of planning. That’s exactly what we’re in the middle of.
Dr. Moore: I really appreciate your recognition of having the resources in place. I was part of a diabetes and depression collaborative in New York City years ago and one of the frustrations of the practices participating was the screening led to a lot of unmasking of depression, but they didn’t have the resources to address the unmet need and, therefore, felt like they would almost rather not know, which is really not where we need to be in healthcare, but it does point out that the more you’re able to front load the resources to meet the unmet need, the better off you’re going to be. I think about that in terms of how you assess for factors that get between people and outcomes that are important to them.
The reason I use that phrase is that it’s more than just social determinants because there are so many other things, like Mabel with fitting into her shoes. It’s really the value of what’s important to Mabel. How do you recognize those things? How do you systematically get them on the table and how do you make that part, then, of an information stream across your entire enterprise? That’s an impressive amount of work.
Teri: It is. We all talk about being patient centric, right, but being able to do that and identify those needs is truly a patient centric approach. We really need to be family centric, too, right? It’s like I mentioned. We cannot forget the caregivers and make sure that we’re meeting their needs, as well.
Dr. Moore: In terms of other lessons learned or recommendations, as other hospitals or health systems are hearing this, what would you recommend, think about this, don’t try that?
Teri: I think probably one of the things I would recommend is taking a look at the risk assessment tool. I know that a lot of the readmission risk tools are based on LACE criteria, but there’s a lot missing in there. There’s a lot of things in those risk readmission tools that stratify the patients into high, moderate, or low risk for readmissions, but making sure that the social determinants are in there, that you can capture them as much as possible. Like I say, you can’t really ask the patient if they have food insecurities and then not be able to follow up with helping them if they do. I think getting an accurate risk assessment for the patients really is what will help us design a discharge plan that will be effective.
Dr. Moore: That sounds like it’s not so much the typical medical dataset; it’s all the other things that you’re identifying as being important.
Teri: It is, but I also want to go back to the medication. If we at our health system, one of our hospitals, had that many errors in the medication reconciliation it would not surprise me if other hospitals do, too. I think it’s something to look at. The initial intent of the medication reconciliation was to insure that the patient was getting the right medications, the right dosages, etcetera. The whole intent was for the physicians to sit down and really do a thorough medication reconciliation and we know that’s not happening. You know, the nurses fill it out; the physician will come and check it.
Our pharmacists who are doing these medication reconciliations are spending about 30 minutes on average with each record. Of course, they’re seeing the high risk patients, but spending a significant amount of time. The thing is, the pharmacists have the expertise and they understand the medications. They understand the interactions, so that would probably be something I would encourage everyone to think about. Think about that med rec and how we’ve historically done it and really who needs to be doing that reconciliation for these high risk patients.
Dr. Moore: The other thing that I appreciate about the way you guys came to this was, I’ve been reading a lot of medical literature around how hospitals and health systems engage in and are successful at quality improvement and one thing that just rings out of the literature is that it’s very important to have leadership buy-in and to use a framework for quality improvement that’s evidence based. I heard you reference Lean and Six Sigma. I’m not saying one is better than the other, but to have an evidence based framework, to use data to understand the impact and the outcomes, and to use a Pilot and Spread approach are all highly successful pathways to achieving quality improvement. It sounds like you guys are really on that.
Teri: Yeah, we are. We use a quality model with a methodology built on first the customer needs followed by process focused improvement and then continuous improvement, so our performance improvement team has really been a catalyst in escalating our performance. The performance improvement engineers are involved in making sure that process change is not only sustained, but spread across the health system, so this most recent Fifteen Top Health Systems report that came out showed that our alignment across our organization is at the 86th percentile, as well as our performance, and that’s quite a feat.
I know there was an article in Modern Healthcare recently talking about the larger healthcare systems are the more difficult it is to be aligned, particularly from a quality and safety standpoint, so I feel that our performance improvement methodology, our quality model has really been central to that alignment.
Dr. Moore: Teri Sholder, given the outcomes you guys have achieved, that’s impressive given the size of your system. I want to thank you very much for your time today.
Teri: Thank you. I appreciate the opportunity to share our journey. We still have a lot of work to go and a long road to go, but I really believe that we are on the right track.