Improving Women and Newborn Care With a Data Warehouse at North Memorial (Webinar)


Improving Women and Newborn Care With a Data Warehouse at North Memorial

[Dr. John Nielson]

Thanks for coming this afternoon. We have a little presentation in the next few minutes about quality and about Health Catalyst. My name is John Nielson, I’m a gynecologist from Minneapolis, and the Health Catalyst Group has come up to our neighborhood in the last couple of years and helped us develop a quality system that has revolutionized what we’re doing. I’m a gynecologist at heart and not a real techie guy.  So for me to embrace this stuff and advance it is saying something. So, the dumb guy I know is now here talking to you about health information system stuff.

North Memorial Health Care

So I work at a hospital called North Memorial Health Care. It’s in Minneapolis suburbs, the western suburbs of Minneapolis. It’s going through some changes.  One of them is it turned into an enterprise.  Another hospital got added to the system a new place three years ago that’s created some issues like any new system would have. So the main hospital, the North Memorial, is shrinking a little bit as the new growth model hospital takes off.

Maple Grove Hospital

The other place is called the North Memorial Health Care Maple Grove Hospital and it’s a big OB volume hospital.  We’ve increased our overall volume 20% in OB in the last two years, but the main shop has decreased about 50%. So we’ve made some transition.  And that’s created some issues.

I would represent our place in 2010 when this project began as facing increasing competition. Even in our primary service area our influence was shrinking. We had financial struggles and still do but are improving. We had a really poorly defined quality initiative and disjointed data sets. So we’re dealing with clinical, financial, managerial issues that didn’t speak to each other and we were kind of in a hurt locker.

We had a traditional quality process based on medical staff committees and traditional peer review, a lot of manual review, with a combination of constructive approach and punitive actions, and let’s just say we sent a lot of letters that didn’t help very much.

Scattered Analytics

So we realized we needed some change there. We had scattered analytics.  Down in here, you’re seeing a little bit of process improvement analysis, Excel spreadsheets. I got most of my data as medical director from ad hoc who did things manually and then send it to me every month and then say, “Well, that’s great but how does that fit with the rest of the system?” So we really had scattered analytics that weren’t getting us what we needed.

Historical Reporting

We had historical reporting, historical reporting that was all over the place too.

So it was very disjointed and when you’d ask a question, you spend a lot of wasted time trying to find, hunting for the data, putting it together, compiling the data. And there was a lot of what we call ‘waste in the system.’

Consultants

So, at that point, we needed help. And what do you do when you need help?  You’re calling the consultants.  And so we did that. We called in some consultants and a bunch of the senior managers weighed in.  I wasn’t quite involved in the early decision. I got involved just after the decision had been made.

So of the groups that came in, Health Catalyst at the time came in. Their mission was clearly to accelerate adoption and maturity of the analytics in healthcare and that’s kind of what we’re looking for.  Tom Burton and Dr. Burton were the people who energized us. Steve was around some but he didn’t have as much to do initially. And so, whenever we’d fade a little bit and Dr. Burton had come back in and say, “No, no, no. This is the stuff you guys have got to get working on.”

So that’s what we did.  So they got our interest about rewarding in a processed knowledge rather than punishing outlier behaviors. They talked about merging technical and clinical people and work together and create a system.

And so, we bought into that and looked at some more of the analytics, you know, build the baseline and then turn it into analytics, add clinical content, deployment and outcomes improvement.  I mean that was a pyramid that made a lot of sense to us.

And clearly around best science workflow and experience, realizing the metrics that are in the middle of the metrics are what make it happen. So we’ve got to tie it all together.

And so, as we embraced that philosophy and moved along, it got interesting. They actually came in through our finance department. I mean not many places are going to have a finance department bring in people to help you with quality. But that’s what happened.  So we identified we needed a data warehouse, a flexible adoptive data warehouse, started working on that.  And we also needed leadership, clinical integration with the corner office leadership – so enable physicians to be more involved in actual discussion and decision making.

COLT Concept

So the COLT concept, the Clinical Operations Leadership Team, taught us. That model taught us some process, created accountability physicians, integrated leader physicians and got us to the next step of decision-­‐making.

Organizational Structure Goals

So I’m just going to kind of go through the story how we started, how we moved along and how we got the progress.  We created an overall senior executive team, then the guidance team and clinical implementation team and work groups. These groups had implanted analytics people and doctors and nurses and made it multi-­‐disciplinary, not as much towers as it used to be. So it integrated the whole process together and it energized people a bit.

North Memorial Resources Consumed

We’d looked at all the different potential groups there could be and care processes and then we decided to start with Women’s and Children’s, which was my deal, and looked at metrics and analytics.  I’ll show you why.

We’d looked at resources consumed and did some numbers and this was early data that Health Catalyst provided for us.  So we realized that most of what we were doing, half of what we were doing was just in 7 care process families or service lines, and we looked further and saw 80% during the first 17.

Care Process Family Ranking Options

So then we put more data together and said, alright, let’s rank them on cases and length of stay, charge ranks, direct cost ranks.  And you see, the pregnancy model, the pregnancy service line or profit family ranking was #1. So that got us in the door we’re going to start with that.

Focus on Labor and Delivery First

And why we went further with that was it looked like it was a large process, we knew that there was a lot of variation within the service line, leadership readiness, which was me, I guess, and less complex process.  So it looked like something we could get our hands around and the COLT Group said, “Alright.  Here we go. Here’s your team.  Let’s see what you can do.”

So along the time that that was being brought along, we had created a data warehouse. So rather than have all these disjointed data sets coming in, they all came in to different source marts and we could pull up with the tools we had that Health Catalyst brought us and created, helped us create, now we can get information a lot easier and make clinical decisions without taking days and weeks and months.

So we kind of went around the concept: we can set up different pieces on what we call a “truck bed.” Here was a pregnancy care process, here’s a GYN care process, here’s a neonatal newborn process, and we can have all three of those different workgroups creating value and doing different projects under the guise of the Women’s and Children’s guidance team. So that is the baseline.   We got a director, we got data architects, we got business developers, actually is the shortest lady in the room hiding back there, but she’s our brain. She’s our data architect/data analyst.  When I started this deal, I said, what the heck is a data architect? I had no idea what that meant. Now I know all the stuff she does.  So that’s what a data architect does but it’s all complicated. So, we put all the pieces in place then to support the initiatives we were using.

We used a KPA tool that you can look at around here. A lot of you are familiar with that.  You can look at examples that are better than what I have here, but we, for example, looked at the APR-­‐DRGs and costs and looking at different groups and tried to look for the biggest bubbles and then looked for the variation from one end to the other on cost and say, alright, it’s a big deal and there’s a lot of variations, so what should we focus on and what brings quality to the process?

So we looked at C-­‐sections and Cesareans and vaginal deliveries, realized that there’s a lot of variability because it’s above this line and there’s a lot of different in costing. So how can we take that apart?

Care Process Model (CPM) Core Work Group

We’d looked further at it and we kind of put together then, here’s our pregnancy group, the core process model.  We didn’t call it our guidance team right away because this was the only team that was working when we started. So we just called it a Care Process Model Core Group. We put the people in place, we added a lot of nursing expert, subject matter experts, and started working.

Well, the first initiative we chose was to decrease the number of elective inductions and Cesarean sections which don’t have a medical indication. You know, that’s kind of one of the sexy things around the country now. Everybody is working on that because you’re going reduce the baby’s need to go to the NICU, it’s going to cut cost, it’s going to add quality. So it’s pretty clear that that was a reasonable thing to do. Other people in our neighborhood were kind of working on it.  And so, we took it at a challenge to see if we could jumpstart that and get ahead of them.

Quality,  Culture,  Science -­‐  First

In my mind, we were quality, culture, and this group, this work group, this model, we were thinking we had to go with the science first because there was a lot of science. And so, if we could show our docs the science, we’d be better off. We don’t have to change the culture because we got a lot of science behind it. The C-­‐section reduction that we talked about next was a little different.

Clinical Review Team

So we put together a clinical review team with all our main player groups. The review team in the middle would do the research, analyze it, and make suggestions. And then we bring them to the review team that met every 6 to 8 weeks to go over and analyze it further, with the support team in the background helping us put this together.

So we had periodic meetings, we did brainstorming, it encouraged engagement and implementation and accountability to bring those pieces of information back to your own group. So each representative was responsible for disseminating the information back to their group and it was constant learning. So that was a good process.

Data Capture, Data Provisioning, and Data Analysis

So what we did then, this slide talks about data capture, data provisioning and data analysis. Sounds complicated but when we started looking for the data to find out how many people or how many patients had inappropriate deliveries early, we realized that our Epic system had a lot of holes in the data set. The nurses weren’t adding this.  So we did a lot of Epic optimization issues. So we’d find things, we’d look at our data, we’d find it, we’d move it into the data warehouse and realize it was empty in some categories. So we’d fixed that.  We talked to the coders, we talked to the Epic folks and we cleaned it up, data mined it, cleaned it up, put it back up, and went around until we were pretty much getting all the information we needed and very little manual review was necessary. So we use that model and that works really well.

Epic Optimization

I could tell you a few examples of Epic optimization. That was a little bit of a hang-­‐up at first to the IT Department who said, “We got 400 things we’re working on, pal. We can’t do this.”  Well yeah, you can, because the boss says we can. So now we’re going to do that.  And so you get a system going and you get some impetus behind that. And we really did that.  It worked up pretty nicely.

Elective  Inductions  Pre–­39  Weeks  Workflow

Then we worked on the workflow. Starting with the physician clinic, they schedule a C-­‐section and the repeat of the C-­‐section or an induction. You’re starting the clinic, you’ve got to call the nurses to schedule things, there’s some yeses and no’s, labor and delivery, screen it, goes…And if it doesn’t agree with the list of medical indications, which we went 10 iterations on to get it cleaned up, so this is our final data set. This is the rules.  If people didn’t have an indication, then they’ve got to talk to me or one of the other medical directors or the medical staff officers and make their plea, and if it didn’t happen, it didn’t happen. So we had a hard stop and that’s pretty necessary.

Labor Scheduling Form

So this was a form that took so much work in here. So all the indications for medical indications, we prioritized them and we quantified these, so the nurse can, everybody had to send over this form, fill it out, the nurse looks at it, verifies it, schedules it if appropriate, if not, then there’s a waiting period and end of discussion.

So you can imagine, as this was implemented, people didn’t like to call me and try to make up a story about why this patient’s uncle was in the hospital and she had to be delivered because, you know, that kind of story. So we got rid of that kind of deal.  And people started understanding based on data and implementation.

We educated the patients about it. We had induction scheduling forms.  We had education pieces put out about why inductions are appropriate in certain settings to engage the patient also.

Wasted Time REDUCED

So over that process, we turned our wasted time into a reduced wasted time, where the way I describe it, I had time to be actually a clinical analyst rather than a data analyst. I mean I can actually, here’s the data, I can be clinical about it rather than have to worry about 3 weeks later getting the information to analyze. So it helped all of us and it reduced a lot of wasted time.

Workgroup Pregnancy – Reduce Indicated C-­‐Section & Inductions <39 Weeks January 2013 AIM Statement

So here’s the AIM statement that we initiated. Green is good. Recruit and Train kickoff, the AIM statement, implementation, design, launch approval, results review. We went through all those processes. AIM statement up there. What we’re trying to do was reduce non-­‐indicated C-­‐sections and inductions. Here’s our clinical process, here’s our accomplishments, next steps. And this dashboard really worked well for us – measures, timeframe. So I was a little resistant at first but it’s too small down here, so you can’t see those things, that’s the only problem, but the answers are there and that’s a really good process to use.

Goal: 99.2% Appropriate < 39 Week Elective Deliveries Outcome: 99.8%!!!

So what we did was we set a really aggressive goal for last year of 99.2% appropriate inductions and C-­‐sections – you know, the standard is 95%, 96%, 97%. People are getting there. We set a big one and we still did it.  So we created a lot of value there and it was a fun thing to do.

Physicians Culture Change

So I’d say we changed most of these based on data and changed the physician culture based on data, evidence-­‐based literature, opinions sought in appropriate forum, specific individual performance metrics.  So people got to see what they were doing. They got to see what everybody else was doing. We sent out a report every month with all these data points that was easy to read and disseminate. And I can’t stress enough how important the nursing engagement was.  I mean I tried to get the nurses to say, “Really? You’re going to do that?” You know, just ask a simple question. “Really?”  And if the doc comes back at you, we’ll deal with that. And then that went away too.  So it was kind of a team concept.  The nurses got so educated in the process that it was very helpful.

New Model OB Story

So for example, the new model story I would present, and this is a little bit hypothetical but we’ve all seen that patient, the “do what we want” patient. We’ve all seen the patient that comes in at 36 weeks, she’s tired of being pregnant, she’s not feeling well, and she went to the office, she asked for induction, the doc said, “No, you’re too early.” Comes back at 37 weeks, asks for delivery again.  “Alright, you can now – you can deliver. We’ll induce you at 38 weeks.”

So the provider calls to schedule, the nurse answers and says, What’s the indication? No indication.  Hard stuff.  Need to call the medical director. Well, alright.  She calls the medical director, have a discussion and there’s no medical indication. Some education is done but it’s no deal.  It’s over.  You’re not going to be able to do that.

So then the next day, we’d send out, the patient calls labor and delivery, we send educational materials, we explain it to her, and she realizes that it’s not a wise thing to do, and she goes back to be in normal and delivers at 40 weeks. You know, that’s a common story that before the initiative she would have been scheduled and delivered at 38 weeks, 37 weeks, and you know, 9 times out of 10 you’re fine but that 10th time the baby is in the NICU, it costs money, it’s angst, it’s a big problem every which way. So we’re proud at how we did this and we did it without a lot of real negativity from our physicians.

Reduce Prematurity Press Conference

We actually had a press conference. March of Dimes did a big deal on our neighborhood while in the state actually, and it came to our hospital and did a press conference about the success. A lot of what we did was on the basis of their research. And, you know, the March of Dimes is really involved in some good things.

North Memorial – AIM Project Status Summary January 2014

So then we said, okay, we’ve done that piece. What are we going to do next?  Now we’re going to look at C-­‐section. So we reduced the number of inappropriate deliveries, some of them C-­‐ sections.  Now we’re going to go look at that set.

NTSV Focus

And we did a lot of that and we said, alright, we’ll go with the national move again. We’ll go with the NTSV program.  We’ll look at the Nulliparous, Term, Singleton, and Vertex group. Now that’s the group that you think about it. If you reduce the C-­‐section rate in that group, it’s downstream, bingo.  Win. Win.  Win.  So focus on that.

Labor Management Teams

Then we looked at labor management. We set up three teams, a triage group, a first stage labor, a second stage labor. A lot of details and I’m not going to share it today.  But we had 50 nurses in our hospital system meeting in three different groups and they did most of the work with that, defining, laboring down, changing, looking, at the new changes in the labor curve and realizing if we do things a little differently, it’s going to add value. So that was really more nursing-­‐driven than physician-­‐driven early on and that was the culture part we’re talking about.

Workgroup: Pregnancy – Reduce NTSV C-­‐Section

So we did the same kind of thing. We used the AIM statement and we got all green and we ran for it.  You know, it’s always fun to tell about the mistakes you make along the way. Well, we made a mistake by – we were using BlueCross as metrics. They were promising us a bunch of dough, “if you can kind of do what we want you to do, here’s the number you’ve got to add, here’s what we do.”  We didn’t realize until halfway through the year that they didn’t have a pure NTSV group.  They already had breaches in it and it was kind of, it was a big surprise when we realized that.

Goal

So we had set up a goal that was kind of in the unachievable range. If we look at national average, local average, we had set our goal at 20.7%, thinking we were setting at 24%, and when we re-­‐did the numbers because BlueCross kind of – it was our mistake but they tricked us.  So we got this number to hit and I’ll be darned if we didn’t do it anyway. So I went in and told the CEO of the hospital, “Sorry pal, we’re not going to be able to make our goal here.” And he goes, “Well, so what, go away,” and we did it anyway. So it was kind of fun to do that but be sure you got your metrics right.

Labor Management Teams

Prodromal Labor Admission Data North Memorial Enterprise

Now, the other thing I had pulled out of these groups here, I was showing the triage group. This sounds like a funny thing, outpatient prodromal labor management. But what we did, one day, we were brainstorming and we said, well what if we look at the patients who come in in labor that are kind of not really in labor and prodromal labor? What if we look at the number of patients we sent home and then they came back within 24 hours and had their baby?

What’s the C-­‐section rate in that group? Because you know the patient comes in and you think there, well alright, we’ll break your water, we’ll do this and we’ll give you Pitocin, we’ll do all the active management things. Well in that group that we sent home and came back, we found out that their overall vaginal delivery rate was about 85% to 86% overall, and it was like, wow. I mean that’s pretty cool.  So that was something that we just came out of nowhere but it did have an effect on our overall numbers. So the brainstorming, the commonsense approach to that made a difference. So we’re going to keep looking at that.  That might be something that we can put in the literature and help people with because I didn’t see it anywhere when we looked.

So then we went – now the third part of the initiative is to reduce the overall C-­‐section rate and that again is to bring in VBAC, external cephalic versions, and keep moving with the information that we’ve already had and getting more benefit. So we kind of, things tied together and they moved from one workgroup to the next.

So if we go back to the truck again and say, alright, now what should we do? We got, we had maintenance is in here, what else should we do? We looked at the newborn group and we decided, okay, let’s go back to our metrics, we put a team together, what are we going to look at?

And we got out the KPI initiative again, we tried to study newborns and tried to see where the variation was, where the cost was, where the quality can be added. So we did a few passes at this and got pretty good information.

We decided to focus on premature and respiratory distress issues and how much are we using surfactant and how are we managing the first 15 minutes in the delivery room and put some metrics to that.

So we’re just in the middle of that process right now and are going to be adding value there.

Process Outcomes

So then you step back and you say, how do we, what do we do with these? Some of these initiatives we’ve created actually are going to decrease the financial performance for the hospital in the short term. You’re adding quality but you get paid more for a C-­‐section. So how do we manage that?  Well we go to the payers, we work on gain sharing programs and we say, this is what we’re doing, the payers want process. They want to see process.  They realized a lot of the outcomes aren’t quite refined enough. So the dollars can happen, the hospital can recoup on the quality side dealing with good payers, explaining process outcomes, and putting money on the line, and we are starting to make some progress there.

“Lessons Learned” – Pregnancy (Elective deliveries pre-­‐39 weeks)

So what lessons did we learn in this process, we learned how important the team is, including the nursing piece.  We realized that analysts “ROCK”, and I know our data architect is now and it’s kind of fun.  Best practices baseline and target. See, we had the wrong baseline and wrong target and it almost well, you know, it almost messed up our deal. The EMR is crucial for excellent workflow, so how do you optimize that? And nurses need to own the workflow, but can’t be “enforcers” of the clinical process. So they’re in the middle but you can’t make them the “enforcers”.

“Lessons Learned” – Pregnancy (Elective deliveries pre-­‐39 weeks)  DATA and ANALYTICS most important!!

The biggest thing we learned I think was how important data and analytics are. If you don’t have those, it’s pretty hard to convince a herd of cats without data. It’s kind of hard.

So my perspective, the two-­‐year journey for me where we are now is a bit interesting. I’ve been on the board of trustees forever, I’ve been the medical director and a regular OB-­‐GYN and I’ve seen it from all angles. And I just really think without the leadership that we got from a group like Health Catalyst, we would have not been in as good a position as we are right now.

So it’s been fun to see.  We still have a lot of things to do. My role then being the first person guidance team leader is to help the other guidance teams. The cardiologists are coming up, some of the acute medicine teams are being formed. We did, we had another, the other team last year on diabetes, and they’re going to focus on diabetes care in the hospital and it got too broad and the mistake was they tried to take on too much and it really didn’t, it didn’t work out as well. The outcomes were not so good, and so we’ve kind of put that on the sidebar for a little while.  So you have to do that. So part of my role is going to be to help the other groups not make mistakes from the lessons we’ve learned.

So we created the leadership focused based on quality. We integrated physician leadership with administrative leadership. We created a flexible data warehouse and we set the scene for further success.

So that’s kind of where we are. We have a dashboard.  We have all the tools.  We have dashboards now that can let us on a real-­‐time basis know where we are and help further progress.

So we’re going to increase the rapid cycle. These teams are going to be 8 weeks and then on hold for a while, and then 80 weeks more, we’re going to rapid cycle, we’re going to begin to add more operations to the guidance team activity, increase the numbers of clinicians involved and spread the lessons to new guidance teams.

Questions and Answers

Dr. Nielson:   Does anybody have any questions for my answers? Yes, sir?

Audience:      With that slide or a couple back, you were talking about some…essentially there you have showed how you were holding the gains.

Dr. Nielson:   The dashboard?

Audience:      Can you talk about that a little bit? Is that like real time or is that…

Audience 2:         It is real time if that was within the system.

Dr. Nielson:   So the question was what does this slide mean? What is this dashboard?  And yeah, if you look closer at it, here’s our pre-­‐39-­‐week results or outcomes. Here’s the external cephalic versions, attempted rate. This will be upgraded every month. NTSV C-­‐sections, overall C-­‐sections, VBAC, VBAC success rate. So we can look at, we can decide how many of those things we want to look at.

Audience:      If you want to go back?

Dr. Nielson:      Yup. Yup.

Audience:      And look at reducing the wasted time..

Audience 2:   So you can look at it overall and you could look at it from the last month to see if if changed, so that you’re not slowing down…

Dr. Nielson:      Yup.   So we would get out to all the providers a monthly report on what the year is, what last year was, what this month is.   And so we’re running 12-­‐week, 12-­‐ month  and  a  monthly…

Audience:      So that’s automated, so you don’t have to do all of this manually…

Dr. Nielson:   Well, you know, yes. That’s within the warehouse, all the data is there and Ashley is a specialist on that. She might have had to write all kinds of data programs or write or do computer programs in the past to do this. Now, it’s all tied together, so you know, even I can understand.

Audience:      That’s an important thing to do.

Dr. Nielson:   Well I know but it’s there. It’s good.  Any other questions? Alright.  Thanks. Have a good evening.

[End of Transcript]