Health Catalyst Overview: A Platform Approach For Transforming Healthcare
Health Catalyst Overview: A Platform Approach For Transforming Healthcare
May 25, 2016
[Jared Crapo, Vice President at Health Catalyst]
Thanks Tyler. So today, we are going to talk a little bit about Health Catalyst. I will give you some quick background about our company and what we do, and then we will show you how you can use Health Catalyst Solutions to deliver better outcomes for your patients.
We believe that with the right evidence, analytics, and methods, providers can transform healthcare [00:27]
So, at Health Catalyst, we have a core belief that with the right evidence and the right data analytics and the right transformation methods, healthcare providers can transform healthcare. And when we say transform healthcare, we really mean improve the quality of care that they deliver for the patients that they serve. And as a company, we have been fortunate enough to work with many of our clients to help them accomplish this mission.
Health Catalyst Clients [01:08]
Just a brief overview of some of our clients and where they are located. You will notice we have about 40 clients spread throughout the United States of various categories. We have a number of accountable care customers. We have lots of customers that are larger integrated delivery systems, we have a handful of clients who are Children’s Hospitals and we have quite a few academic medical centers as well. These clients range from very large organizations, like Kaiser Permanente, who has 40 hospitals and a large insurance plan, to small community hospitals, like Gulfport Memorial Hospital in Gulfport, Mississippi.
Health Catalyst Leads the Enterprise Healthcare BI Market in Improving Outcomes In Latest KLAS Report [01:58]
Recently, KLAS, which is the healthcare market research firm. I suspect many of you have heard of KLAS. They released a report on enterprise healthcare business analytics and one of the questions that they asked health systems who had business analytics tools deployed is whether or not their business analytics vendor products and services contributed to better outcomes. And as you can see in this little chart, Health Catalyst customers said that the tools and services that they were using from Health Catalyst had a compelling contribution to better outcomes in their institution and we are pretty pleased that our clients feel that way about our products and services.
Questions the 3 Systems answer [02:58]
So Health Catalyst has a systematic approach to this healthcare transformation and we believe that outcomes improvement really consists of three core components, and you can think about these components by asking one question for each one. The first question is what should we be doing? What are the best practices? What does the evidence in the literature say is the best way to deliver care for a particular condition. The second component is how are we doing, how do we measure the current care that we are providing? And then the third component can be described by asking yourself the question, how do we transform? How do we implement change? How do we become better at delivering the care that we are providing for our patients? And when you combine those three components together, Health Catalyst believes that that yields better outcomes for patients.
Often, we are not equally strong in all three of those areas, and when that occurs, that often limits the improvements that we are able to provide for our patients. For example, if we are very strong at keeping current in what the literature is, so we are current on our best practices and we have a really strong measurement system or analytic system but we are not very good at deploying transformation or deploying change broadly throughout our organization, then we might have pockets of excellence but we do not have a systematic broad scale improvement across our entire organization because we have a more difficult time getting everybody to do the right thing.
So, today in our presentation, we are going to talk a little bit about how these various components work together and we will show you some of our products, so that you can see how to apply those products to deliver better outcomes.
What does Health Catalyst offer? [05:29]
So, Health Catalyst has a suite of products and services that we provide for our customers to help them with all three of those areas. First, we have our Late-Binding ™ Data Warehouse which brings together data from your acute EHR, from your ambulatory EHR. You might include claim data. Probably if you have a cost accounting system, we usually bring that in. We often also bring in other sources of financial information, like your general ledger system. We often bring in supply chain data. Many of our clients bring in timecard data, so they know how their staffing is related to clinical outcomes. And by bringing all that data together in a data warehouse platform, that allows us to comprehensively and systematically measure many different disease conditions, many different care delivery processes. Once we have that data brought together in a data warehouse, then Health Catalyst provides a wide range of analytic applications. We have almost a hundred of these applications today and many of these applications are focused on very specific disease conditions or specific care delivery processes. For example, we have an application focused on heart failure readmissions. One of our applications in our accountable care suite focuses on per-member-per-month spending. We have an application focused on CMS bundled payments. We have an application focused on sepsis. So a broad range of these analytic applications that you can bring to bear to help solve specific problems. And then finally we have improvement services where we teach our clients a systematic approach to deploying change throughout their organization, to how do we work with physicians to actually affect the best practices that we know we should be doing, how do we make data available to clinical teams so they can use it to improve the care that they are delivering. And so, Health Catalyst helps our clients with all three of these areas.
Late-Binding ™ Data Warehouse Platform
Ignaz Semmelweis [08:15]
We are going to take a pause from talking about Health Catalyst for a minute and I am going to tell you about this man right here. His name is Ignaz Semmelweis and he was a physician in the 1840, a young doctor in his early 30’s, and he is Hungarian and he went to Vienna to work at the Vienna Hospital, which was a fairly prestigious institution. And in the Vienna Hospital, they had two clinics that were labor and delivery clinics, where in the first clinic, they trained physicians how to deliver babies and care for mothers who deliver babies, and in the second clinic, they trained midwives. And so, Ignaz Semmelweis was appointed the chief resident in that first clinic.
Puerperal fever, Yearly mortality rates [09:26]
Now, as he started looking at the women who came to these clinics to deliver their babies, he noticed that the mortality rate for the mothers was much higher in the first clinic than it was in the second clinic. And most of these mothers were dying from a condition that they called puerperal fever. And shortly after birth, these women would get this purplish rash and then they would die a couple of days later. And they did not really know why these women were dying, and Ignaz Semmelweis wanted to see if he could figure out what the difference was between the first clinic and the second clinic, why were so many more women dying in the first clinic. And, you know, the prevailing scientific theories of the day attributed disease to ill winds to climate. Some people thought that old buildings could make people sick. But most of these prevailing theories did not really apply here. These clinics were in the same building and the weather was the same.
Monthly mortality rates 1841-1849 [10:48]
And so, Ignaz Semmelweis started tracking meticulous data to try and figure out what was causing death for these mothers. And his data is so good that we can go back today and apply modern statistical methods to his data. And so, here is a chart of the maternal mortality rate in these clinics. And you can see some months the maternal mortality rate was as high as 30 percent, one in three women who delivered babies were dying. And in sum of his writings, Dr. Semmelweis describes the women who would come, they had a policy at the hospital that they would admit women to either clinic 1 or clinic 2 on alternative days. So if you came in on an odd day, you went to the first clinic; on an even day, you went to the second clinic. And he describes women who would beg to be admitted to the second clinic because they knew that their chances of dying were lower. And in fact, there were some women who would rather give birth in the street than be admitted to the first clinic. And Dr. Semmelweis actually tracked the mortality rate for women who gave birth in the street and it was lower than women who gave birth in the first clinic. So he was determined to try and figure out what was the cause.
After a couple of years of working on this problem, Dr. Semmelweis’ friend, Jacob, who was also a physician there, he was elsewhere in the hospital, he cut his hand with a scalpel while he was working on a cadaver. And a couple of days later, his friend Jacob died with symptoms that looked very similar to the puerperal fever that these women who gave birth in the clinic were dying from. And Dr. Semmelweis wondered if there was some connection between what he called this cadaver as particles and this puerperal fever. And so, as he was thinking about it, he noticed that in the first clinic, where the mortality rate was high, that was where the physicians worked, and the physicians were the ones who were dissecting cadavers.
And so, Dr. Semmelweis decided that he would try and institute a policy to get all the physicians to wash their hands with chlorinated lime before they came to deliver babies, and the reason he chose chlorinated lime is because it was the thing that he knew, the best thing he knew that got rid of the smell on his hands after working on a cadaver. And so, he thought maybe that would be a good thing to try and use.
Monthly mortality rates 1841-1849
Chlorine Handwash [14:00]
So we instituted a handwashing policy in the first clinic and began to track, and he continued to track the mortality rate. And as you can see, the mortality rate dropped dramatically after he instituted this chlorinated handwashing policy. Well, Dr. Semmelweis had made an important discovery. But the other doctors in the clinic did not believe that this chlorinated handwash had anything to do with it. In fact, most of his colleagues ridiculed him and told him that he was crazy. And in fact, he was forced to leave Vienna and go back to Budapest.
Ignaz Semmelweis [14:46]
And 20 years later, Dr. Semmelweis died, about the same time that a guy, named Louis Pasteur, proposed his germ theory, and it was yet another 20 years later after Louis Pasteur proposed his germ theory that a guy, named Joseph Lister, really instituted widespread antiseptic practices in clinical practice. So 40 years after Semmelweis’ discovery before we had widespread antiseptic practices. And it turns out that Joseph Lister’s name lives on today with this antiseptic that we call Listerine.
So what can we learn from the story of Dr. Semmelweis? Well, he had great data. He had one of the three things that you need. He had fantastic data to show how his chlorinated lime handwashing policy matched up with a reduction in mortality rate. He had great data, he had discovered a new best practice and proved that it worked, but he was not able to get widespread adoption of his idea. And so, after he left the clinic, women continued to die from puerperal fever.
Sepsis Facts [16:11]
Well, today, we would call puerperal fever condition, now known as sepsis, and it is still an enormous problem in healthcare today. In the United States, we spend more than $20 billion each year taking care of patients who have sepsis. About 17 percent of deaths that occur in hospitals have sepsis as a contributing condition. It is not always the only thing but have sepsis as a contributing condition. What that means is that during this one-hour webinar that we have together today, 22 people in the United States will die from sepsis. And it is a fast-moving condition. Just as in Dr. Semmelweis’ time, where women would die within a couple of days after giving birth, it still is a rapidly progressing disease. And we have learned that the earlier that we can recognize and properly diagnose sepsis and then give antibiotics, that has an enormous reduction in mortality rates. In fact, for every hour that antimicrobial treatment is delayed, there is a 7.6 percent decrease in survival rate. So it is hugely important to have early identification and early intervention to have successful outcomes in sepsis.
So, why do I tell you this big long story about sepsis? What does that have to do with anything anyway? Well today we would like to use sepsis, that specific condition, as a way to demonstrate for you how Health Catalyst products and services can help improve healthcare.
Product Demos [18:13]
And so, now I am going to turn it over to my friend, Sam Turman, who is going to describe how some of Health Catalyst products can be used and have been used by our clients to help them improve the care that they provide for their sepsis patients.
KPA (Key Process Analysis) [18:30]
Now, sepsis is just one of many different clinical conditions, but in this one-hour webinar, we have chosen to focus on sepsis as a way to describe how Health Catalyst helps our clients deliver better care.
So, Sam, why don’t you show us how Health Catalyst products can help contribute to better sepsis care.
Perfect. Thanks so much, Jared. My name is Sam Turman. I am a Solutions Consultant here with Health Catalyst and I am based out of Atlanta. I tell you, (19:02) there are a little bit of significance as we go through kind of our first case study. So we have on our website over 70 success stories and I am going to pull from two of my favorite ones. I knew today we are focusing on sepsis. I am going to show you a path that a couple of our clients have found a lot of really good success with improving outcomes. There were a number of applications, three of them particularly. One of them, you are looking at now. It is a Key Process Analysis. What this tool does, it allows us to apply a Pareto Principle, meaning that 20 percent of our processes are consuming 80 percent of our resources. And find out where is there some room for improvement for our length of stay, financially where is their variation in care, and where can we really make a difference for our patients, as well as some cost savings.
Once we decide that, in this case we will have a point of sepsis, how do we define that population? Our next application will be a Population Explorer. What can we explore about the population, what do they have in common, which facility should we be looking at, and how can we validate that data to make sure that it is accurate within the data warehouse.
And then finally, we have an advanced sepsis applications. What specific best practice interventions can we build in in use of a point of care to really improve outcome and improve, in this case, adherence to a 3-hour bundle or 6-hour bundle for our different patients.
One of the stories that I like to share is from Piedmont Healthcare. They are located in Atlanta in Northern Georgia. They are significant to me because they are about – one of their campuses is about six and a half miles from my home. And if I was to be admitted to the hospital, it would be one of their six hospitals and one of their 125 physicians who would be caring for me. Before they engaged with Health Catalyst, they had a sepsis program in place and I am going to tell you a little bit about it. They had three employees who spent every morning looking through three different spreadsheets and trying to figure out which patients should be included in their sepsis program. Once they found those patients, they then went and manually extracted up to 480 charts from each one of these six hospitals. And Piedmont’s estimate, this was about one full time employee’s equivalent job is just gathering this data.
Alright. So they get the data, it’s gathered. Their next issue is they then bring that data to the providers. Well, at this point, it is 30 to 60 days old, and the providers say, “You know, I really can’t trust this data. It’s old and I don’t even really remember exactly what I did 30 months ago and I am not going to take the time to look back to the chart to see what I could have done better to comply to that 3 or 6-hour bundle.” So there had been an attempt, they had tried. Eventually, it just kind of phased out and was deemed the sepsis project.
When they engaged with Catalyst, we were able to look through this KPA tool and see that as they suspected there is a lot of variation around sepsis and it is a high spend area for them, it would be great for their patients to improve that source of care, as well as be able to improve some cost savings opportunities there as well. They started to go through and validate the data through the population explorer and take that data to every four of their hospital that it pertains – so every acute floor, the ICU, the emergency departments, and show it to the providers and say, “Does this data look accurate?” And then gave the providers the opportunity to say, “Yes, that looks good. No, that doesn’t. I prefer this filter. Or this is more helpful for me to see. This is insignificant. This is significant.” So now, all of a sudden, the providers have seen where the data comes from, they buy into the data being accurate and they have had some input as to what information goes into eventually a sepsis application. So no longer are we dropping an IT project off at the Piedmont physician’s desk and say, “Hey, use this.” But in their mind, they are dropping my application off that I had some input in at my desk.
They adopted a team with a strong and active leadership support, a really good program manager, expert clinical leadership, and an engaged physician champion. And as they developed this, they introduced the sepsis application. It looks over four different key elements of improving sepsis and we will go over those here in a bit. And in time, and actually only a 3-month period, they are able to reduce their sepsis mortality rate, they are able to gain confidence in the trusted good data from the doctors, and be able to have almost real-time experience and showing what has affected the patients today, what affected the patients yesterday, not 30 or 60 days ago. So, how are we doing in our 3 and 6-hour bundle, how are we doing on our length of stay, how are we doing on our mortality, how are we doing on our cost. And ultimately, they saw 97 percent improvement in their time to reporting, what used to take days and one FTE a full month now took five minutes. We have automated that 480 chart review by each location, each hospital. There is no wait time for reports and all manual sepsis reports were completely eliminated, and they continue to refine this process. Through this team is how we engage with our clients to that adoption model. And what I am about to show you in our application is how we leverage our analytics from our data warehouse with built-in best practices, the whole three systems coming into play.
So jumping into the actual applications now.
First, I am going to show you our Key Process Analysis and kind of give you a (25:14). All of our applications have some things that are similar, some things obviously that are different, but I want to show you what you will be looking at so you know as we review these in three to four minutes. It will be pretty quick demonstrations.
KPA (Key Process Analysis)
We always have a welcome screen. Of course, the title, what version, who owns it, how recent is the data.
And really important is our information tab. Not just an overview of what the tabs are in the application and what the application does, and this point showing that we leveraged that 80/20 rule to identify areas of clinical variation and clinical potential focus, but also definitions.
Definition – Data sources and Definitions [25:51]
How are we defining in this one application a process, a program, a family, how do those work together. In this case, a program is a clinical program and this will be looking into general medicine. There is a care process family within that program of infectious disease and one of those individual processes in there is sepsis. How are we defining and calculating total cost, variable cost, length of stay. When we look at some of our bubble charts, what are some of the metrics that are built in there and how are they defined. That is in every one of our applications.
How to Navigate [26:28]
A little details with how you navigate.
Related Content [26:30]
And some other relevant content. All this can be made unique to your specific organizations.
Pareto Analysis [26:36]
And then we jump into the actual application. There is a number of tabs up from left to right. They go from more general to more specific topics all the way down to that individual patient line item detail. They are drillable dashboards, if you will. And as we drill in, it highlights filters here on the left-hand side. So that is the way (27:00) the Key Process Analysis. What we are looking for first in the scene is our Pareto Analysis here by care process families. What we are seeing down towards the bottom is our individual clinical programs – Hematology-Oncology, General Medicine – and they are listed by rank order of what is our highest, in this case total cost. We could list them as total variable cost, length of stay, charges. Any metric we can pull out of the data warehouse, we can slice and dice by here. Some key process metrics across the top are discharges, mortality rates, average variable cost, length of stays. And then the red line is our cumulative total. So if Hematology-Oncology is taking up 5.89 percent of our total variable cost, and General Medicine 5.83, that is a total of 11.72. so we have this exponential arc all the way to 100 percent.
Pareto Analysis [28:00]
Now, maybe we are interested in this case just in our top 10 process families that have the highest variable cost. I can see them below here. And these Pareto metrics that we can filter by are also included in these columns. If they are red, they are within the top 5 of those columns. So we are looking for areas that potential bleed across the page. In addition to Hematology-Oncology, we have General Medicine and Cardiovascular that are very ride. In this case, we want to look into General Medicine or Infectious Disease but we can now prioritize and get the data placed at the table of discussion and decision making as in which direction can we get the most bang for our effort, if you will, with these improvement teams and our efforts.
Focus Area [28:43]
And we can start to focus now on our specific focus areas within those top 10. So instead of looking at care process families, in the focus area, we are looking at the individual care process, the very lowest, most detailed layer. And what we see on our legend is all these different colors for the different clinical programs, like we saw on the other side. So Cardiovascular are all of our green dots and the orange dots are Surgery, the blue dots are General Medicine. We can filter again by a number of metrics or risk models. But in this case, as we are looking, we see on the X axis our total variable cost. The further a bubble is to the right, the more of our variable cost that is gobbling up. The higher it is on the Y axis, according to the Elixhauser Risk Model, the higher risk in variation of the care of the patient. So what we are looking for is really this upper right quadrant here for expensive areas that are also very variable in their care. And as we hover over top of these dots, each dot represents a care process. The size of the dot represents how many of those cases we have, we see a care process for coronary artery disease, and then we see a care process here that has an area of variation of septicemia.
Focus Area – Septicemia [30:02]
As we click that dot, we can drill into it specifically. And look, for septicemia patients, how are we doing here on that risk model of variation in care. The higher the risk, the more the variation you expect to see. So if I took this to a doctor, the first thing they are going to tell me is, “Of course, I have the most variation. My patients are the sickest.” And indeed we see that here. We have some high-risk people that might have high comorbidities.
Focus Area – Septicemia – Level 1 through 3 Risk Level Patients [30:32]
But let us look specifically at our level 1 through 3 risk level patients. What we are seeing in each of these dots is an individual provider. The larger the dot, the more cases that provider has. Where it falls on the axis is their total variable cost. And we can see in an ideal world, if we remove all variation, we should be stacking at the mean and get rid of all of these outliers over here to the right. That would be ideal. But there are some variations and some room for process improvement here, so much that when we calculate our average total variable cost minus the median cost gets a sum opportunity and we are looking at a little over half a million dollars of opportunity of variable cost savings, as well as bring our quality of care to our patients.
As promised, what we have looked at so far, just to recap, we have looked at what areas are consuming the most resources. We have then turned and now we have looked where areas have for septicemia has the most variation, and that there is variation in that care and confirms that.
Patient Detail [31:32]
And now we can go and look at the Patient Detail for those 772 patients and export them to a work list of some sort or click into their exact individual information based upon their demographics or what information it would surface from the data warehouse.
So again, the KPA is one of our exploratory applications. It is meant to be a directional to let us know what direction we could potentially run in and improve an outcome. So what we have determined here is that septicemia is an area of high cost variation and of clinical variation that we would like to then drill into and see some more data. So just like in our Piedmont example, they realized that this was a high area that could affect their bottom line and their patient outcomes and they may want to start to verify the data and explore that population of patients.
Population Explorer [32:26]
And we do that with one of our tools, called the Population Explorer.
Again, as we drill in here, we have that same welcome tab, we have that same informational tab.
Outcome Dashboard [32:42]
And then we start from that high level executive style dashboard. So we can look and see our KPIs across the top, our discharges, mortality rates, average variable cost, length of stays, readmission rates, how we are performing. And in this case, we have no date filters how we are doing over time.
Outcomes Dashboard [33:00]
Maybe we are most interested in 2015 in the third quarter and our discharges are now down to just under 20,000. How we are performing in those different months, as well as which departments these patients are in.
But we are most interested in our septicemia patients, which through our clinical program, we know are under General Medicine.
Under that medicine, we know that they are under Infectious Disease Care Process Family.
And take into that individual septicemia set of patients. So we have gone from 569,000 discharge records and gone down to 101 discharges. So we are filtering through all of our data from our disparate sources in this visualization. And maybe we are just interested in those who are in inpatient care, not outpatient or emergency room.
We can then go in and do a quick comparison of these patients – where are they potentially located, what is the difference between this cohort if we change just a few different things about them.
What you are seeing here in the comparison tab is a blue and a grey cohort. The blue cohort has all the filters we previously applied playing forward and the grey have none. So we see down here there is absolutely no filters.
If we want to compare apples to apples, we can click to copy the blue to the grey cohort and then change just some information. For example, where should we be focusing the septicemia effort where our most of the patients at.
If we go to discharge facility, we can look at our Millrock Health Eagle location and we can compare that to our other discharge facility on record, which is Granite Hospital. We are going to see that most of our septicemia patients are seen at the Millrock Health Eagle location, and see there are 69 discharges as compared to the 6. Length of stays, variable cost, etc. So we want to focus potentially on validating the information within the Millrock Health Eagle location, not so much the Granite Hospital.
Patient Detail [35:11]
And again, now that we are down to these 69 patients of our most recent patient data, we can go to our Patient Detail. See, all these patients here have the opportunity to export this list easily to Excel. And we can take it back and take it to our providers and actually compare it with the medical record and make sure that we are matching up on all of our data points, so we can get buy-in on the accuracy of the data, so we can move forward with the sepsis project itself.
We can also build dynamic reports here, where specifically if I am interested in my visit detail, I would like to know the facility count ID, so my providers know how to look up this information, Admit Source, Discharge Source, Risk Model, potentially inpatient length of stay in hours, not days or midnights, and then export that to Excel.
So recapping all the way from the beginning with our KPA, we have identified areas of potential improvement. We have now identified some of that population and working on validating that information. Once it is validating, we get buy-in from our providers, we can start to work on the specific septicemia advanced application.
In addition to getting buy-in from our providers for our septicemia application, our septicemia patients, we want to take them into consideration. We are considering the interventions to build into this application. Do you agree with the three-hour bundle and what is our goal time we want to get the first (36:41), where should we specifically make this and how would you like to engage.
Sepsis Summary [36:48]
Starting again from a summary, it is a high-level overview. So what we are seeing here specifically is our discharges, mortality rates, and our average variable cost by monthly, quarterly, or an annual basis. This is a high-level executive dashboard to get a quick glimpse into how we are doing for even our different discharge groups. Maybe we are just interested in our sepsis group, our septic shock group, or severe sepsis, those who have expired and those who have not.
Sepsis Summary [37:22]
And we would like to filter on just sepsis. So highlight that, you can see it highlights the filter here on the left as well. And now, we are down to 435 records.
Sepsis Summary [37:29]
Maybe we are most interested in just what has happened in the fourth quarter of 2014. And now we are looking at 65 discharges.
Sepsis Metrics [37:37]
In addition to our executive dashboard, we can go over to the metrics. And again, see how we are performing over time on process metrics, how are we doing on our process overall for our 3-hour bundles, our lactate results, our antibiotic administered and our fluid, and how is that affecting our outcomes. So we are able to pull that data together and quickly visualize it.
[Sepsis 3 Hour Bundle: Overall [38:00]
Where this application gets really exciting is in our 3-Hour Bundle tab. So how are we performing overall for these 66 patients in Q4 of 2014 with the sepsis severity, not septic shock or severe sepsis, and how are we doing overall.
Sepsis 3 Hour Bundle: Lactate [38:20]
And individually click through and see how we are doing on our lactate intervention. We have a goal of 60 minutes of getting this done. Our first lactate order is happening in the first 30. We are having that collected within the first 43 and the results are coming back above our 60-minute goal line of 79. What can we do to potentially improve that.
Let us look and see which are the ordering providers. What are those ordering providers’ average minutes, what is their median minutes, what is their, what is the total number of orders. If they only had one, where they potentially having an off day. What can we do to help that going forward.
Sepsis 3 Hour Bundle: Blood Culture [38:57]
Same for our blood culture, how are we performing over time, how quick is that order coming in, how quick is it being collected, and how quick is that second blood culture being collected and resulted, who is ordering it.
Sepsis 3 Hour Bundle: Antibiotics [39:12]
Antibiotics, who is ordering it, who is administrating it, how are we doing bumping up against that zero to 180-minute with that 3-hour bundle goal line there.
Sepsis 3 Hour Bundle: Fluid [39:25]
And of course how are we doing on our fluid resuscitation here. We can view all these numbers either on average or median.
Sepsis 3 Hour Bundle: Overall [39:38]
And then our overall compliance be able to drill down to that line item level. Now that we are looking at these few patients here, maybe I am interested in specifically why we are not completing a blood culture for this cohort of patients.
Sepsis 3 Hour Bundle: Overall [39:56]
If I highlight these patients here, I am now looking at just the eight of them.
Sepsis Summary [40:02]
I can go back and see again their executive summary, see how those seven patients are performing.
Sepsis Metrics [40:06]
How their process metrics may be affecting their outcomes metrics.
Sepsis 3 Hour Bundle: Overall [40:12]
What is the timing, is there any commonality amongst their ordering antibiotic providers and administrating providers.
Patient Detail [40:22]
And then again who are those individual patients. We can surface them here directly and be able to click in to them and see some of their demographic information, as well as their provider information, hospital information. Any information that is available in the data warehouse, we are able to surface here very quickly.
So through the whole process, we have been able to identify areas of potential outcome, verify the populations, and then actually use specific interventions to drive down the mortality rates and provide better care for these patients. Not only have they done this at Piedmont but also at a number of our other clients. One that comes to mind specifically is Thibodaux Regional Medical Center in Thibodaux, Louisiana. One thing I love about the story is Thibodaux was already doing a fantastic job with sepsis. They are below the national average. But when they went through and did their KPA findings that are Key Process Analysis, they realized there is still quite a bit of variation. And as they drill through the variation, there are still areas of improvement. And was it enough just to be below the national average? They decided it was not. They started a care team where they are able to drive sepsis again to get even better and saw 37 percent increase in timely antibiotic delivery, 96 percent improvement in fluid administration, and 19 percent improvement in lactate compliance, and they improved their blood culture compliance by 55 percent, and they followed the same exact practice and it is something that the providers are able to get behind through validated data and really push the outcomes improvement for their patients. And really probably the best statistic there is the first week that they introduced this, within the first three days, they attribute two lives being saved to the sepsis advanced application, which is certainly an outcome improvement for the facility but also for those patients and their families.
I will turn it back over to you, Jared.
Awesome. Thanks, Sam.
Questions the 3 Systems answer [42:39]
Sam just showed us how Health Catalyst applications can combine data analytics with best practices and how those tools together can help our clients to widely adapt and widely have access to that data and those best practices, so that they can improve the care that they deliver. And today, we sort of focused on an example of sepsis but Health Catalyst has many other clinical conditions that we have helped our clients worked on. They follow these same principles and this same process. We have clients who have worked on labor and delivery. Many of our children’s hospitals have done work related to appendectomies. We have clients who have done work in non-clinical areas. For example, we have clients who have worked on improving their revenue cycle by focusing on discharges that are final but have not yet been billed and why have they not been billed, and usually it is because the discharge note has not yet been signed by the physician, but not always. We have clients who have done work on labor management, how did they better align their staffing with demand for services.
So, we have many different areas where we have been fortunate enough to help our clients apply these principles to yield outcomes improvement.
So we have a few questions that I would like to address, a couple that sort of came in during Sam’s demo. One of them was in the key process analysis where we were showing cost. The question was, “Was that actual cost? Is that charged to mail from a claim? Is that paid amount?” As a best practice, we encourage our clients to actually use their actual cost, which is what usually would come from a cost accounting system, something like EPSI or Strategas, because looking at your actual cost as opposed to what your paid amounts were gives you more accurate insight to how you are consuming resources. Now, some of our clients do not have a cost accounting system. And so, we have to use a different metric. And so, often we will use the paid amount from a claim instead. We hardly ever use charges because there is hugely a pretty big gap between the charged amount and the paid amount. But if you remember, back in the demo, there are also other proxies that we can use. For example, we can use length of stay as a proxy for cost because if the length of stay is longer, it is going to take us more actual cost. So we might use length of stay as a proxy. So that was one of the questions that we had come in.
Are there any other questions? If you have got other questions, you can put those in the comment box and we will try and address those.
Alright. Well, as we give everyone some time to type in some questions, we have got a couple questions for you. Particularly, we have our registrations to give away to the Healthcare Analytics Summit that will be held on September 6th through the 8th in Salt Lake City.
Are you interested in attending the Healthcare Analytics Summit in Salt Lake City?
(Single registration) [46:31]
So, we are going to launch this first poll, and this is for the single registration. Are you interested in attending the Healthcare Analytics Summit in Salt Lake City? And with the limited space that we have left, we would like to encourage you, if you are pretty sure that you can make it, please select ‘Yes’ on here. And there will be also a limited period of time in which we will be able to redeem the winner to redeem the giveaway.
So we are going to go ahead and close this and now give you the opportunity to enter for our team of three registration.
Are you interested in attending the Healthcare Analytics Summit in Salt Lake City as a team? (team of three registration) [47:05]
The reason why we give a team of three rather than just a single, but we are going to both, is because we believe in talking about analytics having a team there, being able to be with other members of your organization is really the most effective ways to really talk about healthcare transformation from an analytic perspective and really to be able to drive outcomes improvement.
Okay. We are going to go ahead and close that. Thanks everybody for responding.
How interested are you in someone from Health Catalyst contacting you about a demonstration of our solutions? [47:35]
And we just have one last poll question, one that we ask almost every time. If you are interested in someone from Catalyst contacting you about a demonstration of our solutions, please respond to this.
Alright. We will leave this up. We have got some great questions coming in now. This is wonderful.
|To what degree do you find that accuracy of available data is a potential issue?
|Nobody’s data is perfect. The good news is that you do not need perfect data to do meaningful improvement. And I will give you specific example. Many of our teams use their first improvement effort and improvement in data capture. For example, we have a lot of clients who have done work on labor and delivery. And for nearly any project in labor and delivery, you need to know the gestational age of the unborn baby. And if you think about all the different ways and methods and places that that is often captured, the ultrasound technicians have a template, the maternal fetal medicine team has a template, the obstetricians have a template, use a different assessment when a patient presents in the hospital for delivery, and we do not often have consistent data capture. And so, because most labor and delivery things require a uniform way of assessing that, we often have to just simplify or standardize the way we are capturing that data.
So, accurate data is important but you can often make lots of improvements with the data that you already have.
|How long does it take to wire all these sources of data to the warehouse?
|As we have worked with our clients, we find that it typically takes about six months to bring in a core set of data sources into the data warehouse. And when we say a core set of data sources, we mean your acute EHR, your ambulatory EHR, your cost accounting system, if you have one, your patient satisfaction data, if you have that, and then we also usually bring in some financial data so we know things about all of your discharges. That process of bringing the data into the warehouse and mapping it so that we understand what we have usually takes about six months.
|Can you speak more to data mapping process and more broadly to implementation? So you said six months for that process. Are there things involved with implementation?
|Yes. So the first phase of an implementation is usually bringing some data into the warehouse and mapping it into our metadata repository, which is part of the Health Catalyst Data Warehousing Platform. And that process of data mapping and data ingest for a core set of sources usually takes about six months. Once you have those core data available in the warehouse, then you can start to utilize those applications that Sam demonstrated, so you can decide what you are going to work on first. We generally find that our customers are able to map in a core set of data, choose their first improvement, and then realize their first improvement from that area. So if they are going to work on sepsis, they can choose sepsis and get their first batch of improvements usually within 12 to 14 months after beginning the implementation. And then those improvements tend to balloon out from there because we already have a core set of data sources. And so in year two we will find that they have multiple improvement teams and each team has been able to realize several different improvements as they mature their implementation.
|I understand that some healthcare provider organizations are moving right past the data warehouse. Some are storing all data at its most granular level and aggregating on the fly, so to speak, thereby eliminating much of aggregations and latency involved in the building of a data warehouse. What is Health Catalyst’ view on this topic?
|There are lots of entities that are sort of skipping past the data warehouse. There are lots of organizations that use the reporting that comes as part of their existing transactional systems. For example, your core EHR might provide lots of good analysis of the data that is contained there. At Health Catalyst, we believe there is a lot of value in collocating that data in a data warehouse. There are some kinds of analyses that are very difficult to perform on the fly. For example, Health Catalyst has some pretty sophisticated predictive models that are fairly computationally expensive and we cannot perform those on the fly. So if you want to apply those predictive models to a data set, that is pretty difficult to do on the fly.
There are also some data sets that require that cannot be ingested on the fly. For example, many claim data sources are only available after a 30-day time period. So, while there are some benefits of skipping over a data warehouse as it relates to how fast can you process the data, we believe that you get a much richer analysis when you collocate the data in a data warehouse.
|Given the complex data display, how do you extract the meaningful message?
|As you noted in the applications that Sam showed, yes, we have lots of complex measures, but up at the top, our applications tend to surface the three or four, sometimes five, key metrics that really matter. And so, that is one thing that can help with extracting the core message. But a more important topic is how do we teach our organization to raise their level of data literacy? And that is an education and an experience issue, not a technology problem. And the best way we know how to do that is to start to surface data that providers can trust so that they then get broader exposure to data so they become better at assimilating the key messages from that data.
|If you are capturing claims data, what is the most typical source? The CMS, commercial payers?
|All of the above. We have many clients who love their CMS claim feed, for example, from the Medicare Shared Savings Program, who load those claims from CMS into our data warehouse. Many clients have their own small health plans and we often load that data, but we also ingest data from commercial payers. So it is all of those sources.
|How does master data management fit within Health Catalyst Solution? Do you require data to be mastered prior to acquisition? Do you master the data or create a golden record within the ingestion process? Your thoughts on this topic.
|We could do a whole webinar just on this topic. The short answer is that we do not impose master data management at any particular point. We try and utilize data that has already been mastered. For example, many of our customers already have a master patient index that they have made substantial investments to reconcile patient identity. And if they can send us a golden master patient identity, which is often the patient’s EMR number, or the medical record number, then we can utilize that within the data warehouse. So for data that is already mastered, the data warehouse can just trust that data.
Now, in some cases, that data has not yet been reconciled. And so, we have the ability to reconcile data within the data warehouse. Generally as a principle, Health Catalyst believes that you should master that data when you need it, not when you ingest it. And so, the gestational age example that I gave earlier is a great example of let us do that when we have a team of clinicians who are ready to do work in this area, then let us worry about what is the quality of that data and what additional analysis or processing do we need to apply to it to make it more useful. We generally are not big proponents of doing all that work before you load the data into the warehouse.
Thank You [57:38]
Alright. Well, we are at the top of the hour. Thank you so much, Jared. Thank you, Sam. I would just like to let everybody know, shortly after this webinar, you will receive an email with links to the recording of this webinar, the presentation slides, as well as the winners of the summit giveaway. Also, please look forward to the transcript notification we will send you once it is ready. On behalf of Jared Crapo and Sam Turman, as well as the rest of us here at Health Catalyst, thank you for joining us today. This webinar is now concluded.
[END OF TRANSCRIPT]