Health Catalyst Overview: How Improving Outcomes Can be Like an Awesome Ski Trip (Webinar)

Health Catalyst Overview
How Improving Outcomes Can be Like an Awesome Ski Trip
March 31, 2015

[Tom Burton]

Thanks Tyler. I’m glad to be with you all today. This webinar is a little bit different. Normally, we’re focused more on talking about an educational topic. But today we get a lot of questions about, well, what exactly does Health Catalyst do. And so, about once a year we will do kind of an overview of our product roadmap, and this time we thought we’d use a ski analogy. Being from Utah, I love skiing and I’m a big avid skier. And so, we’re using a ski analogy today to kind of talk about what we do as an organization and really how improving outcomes and the work around outcomes improvement can be like a really awesome ski vacation or ski trip.

Health Catalyst at a Glance [00:55]

So just to give some of you a little bit of background, we serve a lot of different types of organizations, from integrated delivery systems to accountable care organizations, to community hospitals, to academic medical centers and children’s hospitals, and what we found is that these principles that I’ll be going over today really apply across the board. Whether you’re a large integrated delivery system or a small community hospital, there’s some core principles and some core tools that we think are needed to really help you improve outcomes.

And so, as we talk today, I hope that this message will resonate of things that we see as really critical on your outcomes improvement journey.

Health Catalyst

The Greatest Outcomes on Earth [01:41]

So if we are to take this ski map or trail map, and those of you who are skiers will recognize, you’ve seen lots of trail maps like this. We were able to get one of our local ski resorts here to allow us to use this with their permission. And if you think about a ski trip, there’s some things that you’ll need before even you get to the slopes. So you’ll need to have the right equipment. And we kind of make the analogy that having the right equipment, the right gear, is like having a good analytics platform and having the right clothing, helmet, goggles, gloves, etc. So there’s a huge aspect of an analytics platform that’s key to skiing here.

The second is you’ll want some lessons. If you’ve never skied before, you’ll want some lessons, and there could be private lessons or beginner lessons, or even a workshop. And so, we’ll talk a little bit about some of the services that Health Catalyst provides and how that might relate to those types of ski school opportunities.

But the real heart of what we do at Health Catalyst is improving outcomes. And we’ve represented all of the different types of outcomes that we improve and applications that we have to improve outcomes represented by different parts of the mountain. And to get to those different parts of the mountain, you can take different ski lifts. And so you can see on our map here we have population health which takes you to a lot of different areas of the mountain where you can ski the heart failure run or the back and spine run or the community care run or accountable care where we have things like managing your referrals to a financial area where you manage your rough cycle to the operational efficiency area, where you might be thinking about supply chain or practice management.

So, the analogy here is that if you have great skis and great boots, great equipment, if you’ve learned how to ski well, there’s really any part of the mountain that’s open to you. And the same skis can really help you traverse lots of different types of terrain.

So what we thought we would do here is now go into a little bit more in detail kind of on each of those areas that I just mentioned – your equipment, your training, and ski school and then actually skiing the mountain.

Equipment Rental or Purchase (Hosted or on-Premise Data Warehouse Platform [04:17]

So if we start, as Tyler mentioned, Health Catalyst is an enterprise data warehouse and analytics company. And we kind of categorize that, it’s two major areas, the Health Catalyst Analytics Platform and our Shared Framework Tools. And in our analogy, we have certain tools that could be acquainted to skis, boots, bindings, and poles. So if you think about the problem that a lot of us face with data in healthcare, today, that data is scattered across a lot of different EMRs, financial systems, HR systems, administrative systems, patient satisfaction systems, and even multiple EMRs, multiple financial systems, especially as we see accountable care groups start to form, where a lot of primary care physicians may have a different EMR.

So, some of the tools that Health Catalyst have developed is, first, what we call our Source Mart Designer tool. That allows you to connect to all of these different transactional systems and be able to bring that data into the data warehouse, collocate, and integrate that data. And so, the Source Mart Designer tool helps to automate and accelerate that process. Once the data is collocated, there are a lot of common tasks that need to be performed on that data, things like attribution, how do I connect which patients belong with which providers, or modelling risks, or sharing data between organizations. So we call that our shared frameworks and tools that are commonly used. Regardless of what part of the mountain you’re skiing, you’re going to need to do some common tasks when it relates to integrating and manipulating data.

And then finally, we have a tool called Subject Area Mart Designer, which allows us to get into a specific clinical area and start to interpret and analyse the data. And so, our Subject Area Mart Designer tool allows for a lot of different clinical and operational areas to be analysed, to build predictive models, and then to take prescriptive actions based on models that can be built. And so, that tool is really a flexible tool that can help you define cohorts that can help you design calculations for metrics and then integrate all that data together, and so show correlation and causality.

Late-Binding ™ Data Warehouse [07:02]

Just a brief reminder, many of you may have seen this like before in some of our previous presentations, but Health Catalyst uses what is called the Late-Binding ™ Data Warehouse modelling technique. And yet here is there are a lot of different transactional systems that will need to be integrated. And rather than trying to integrate those right upfront when the data is initially brought in, which is within an early binding model technique, we prefer a Late-Binding ™ technique where we bring the data in in a fairly raw format. It almost mirrors the transactional system it comes from. Then a small number of those elements are linked together – so patient identifiers, provider identifiers, location identifiers. But not everything is normalized early on.

And only when a use case presents itself for a particular clinical or operational area do we do what’s called the binding of a specific data element to a concrete definition. And because we hold and wait to do that binding late in the process, multiple bindings exist for similar concepts. This allows for a lot more flexibility and decisions can be made later in the process when you’re actually working with physicians and nurses rather than when you’re initially bringing that data together. We found this to be way more effective than some of the other techniques, like dimensional modelling and an enterprise data model approach, and found that this works very well with complex data that is frequently changing and where definitions of what is in control versus out of control diabetes, for example. As national societies change, their definition, as new medical knowledge is discovered, we found that a flexible architecture is one of the most important attributes of your data warehouse.

Platform Update [09:20]

Just to give you a little bit of an update of some of the enhancements that we’ve made to our Health Catalyst Analytics Platform over the last year, we have four major releases of platform in 2014. We added a bunch of extensibility points, which allows data architects and product engineers to extend our platform at predictable points. So, different processing routines, different algorithms can be added into the Health Catalyst platform without having to change or go outside of the platform. So that was a major enhancement. We also added four new different source formats that we now support, and we enhanced our auditing and diagnostic tools.

In the shared frameworks area, we added Eventalytics, which is an app that we use for any kind of presentation where you want to get real-time feedback. For those of you who attended our Health Catalyst Analytics Summit, we actually used that to get real-time feedback during our keynote presentations and throughout the conference, so that we could get feedback from organizations. In the future webinars, we will start to be using that mobile app as a way to get more detailed and granular feedback.

We have also added Geospatial Support, Risk Profiling, and Predictive Modelling Support in our shared frameworks.

We have a lot of work going on in our source marts. We now have 11 EMRs that we support and then we have several more that are underway. We did about 84 different source marts in 2014. We have a total of 55 source marts, some of those we did multiple times obviously. We’ve seen that the tools, like Source Mart Designer, are helping us to enhance how quickly we can implement these source marts. So for example, one of our clients, Crystal Run, we completed 9 source marts where we integrated data from EMRs, financial systems, etc. in a matter of about 9 weeks. So, the tools that really help us accelerated how fast we can bring that data together.

In addition, we have launched our hosting environment, and that allows some of our clients who choose to do a hosted environment rather than an on-premise environment and we put very significant security in place in order to do that. So those are some other enhancements that have happened to the platform.

Share Frameworks & Mobile [12:06]

If we think about some of the things that are on our roadmap that we’re planning on building, shared frameworks and mobile are two areas we’re focusing a lot on. Those shown here in green are things that we are currently doing, where it’s deployed at at least one customer. Those in blue are those where we’re under development, where we may be beta-ing with our first customer on that particular area. And those in grey are things that are in our roadmap, we’re planning on doing but we don’t have it deployed at our individual customer site yet, it’s still in our product development area.

So, I won’t go through reading all of these but just to highlight a couple of areas that we’re focusing on, one is our collaborative analytics framework and exchange and we are pleased that children’s hospital association has joined the Health Catalyst customer family. And this is an area where we are going to be spending a lot of time and we’ve started developing here, where rather than just high-level comparative analytics, we think collaborative analytics will allow our customers to share data at a much more granular level, where I can see not just who is doing better than me but I can actually drill down to the individual interventions, the individual order sets, and understand what is this other organization doing differently than I am to achieve better or worst results than I’m achieving. And so, we are putting some emphasis into our collaborative analytics framework and exchange and want that to be an environment that not only data outcome results can be shared but process results, clinical content, operational content can be shared so that we can accelerate improvement across the country.

The other one that I want to highlight is an area that we have been trying out on ourselves for some time and we call it our Get Fit Stay Fit application and we’ve been using it with our own internal team members now for several quarters. And basically it’s a way for employers to engage their members in healthy behaviors, such as tracking both what you’re eating, how you’re exercising, and we’ve added some gamification features that have made a huge difference where we’ve had a lot of our team members lose significant weight, get their diabetes under control, and we’ve seen it go well beyond most wellness programs. So we see it being a great long-term tool to help engage patients and their families in their own health. So that’s a new area that we have gotten into and that’s again built on a mobile framework that we’ve designed to be built right into the platform.

Ski School = Health Catalyst Services to accelerate improvement [15:11]

So, we’ve talked a little bit about the equipment that you’ll need for skiing, the platform and the shared frameworks that are necessary regardless of what particular part of the mountain you’re going to ski on. The next area I’d like to cover is the ski school, which are our Health Catalyst services that we provide to accelerate improvement. So we’ve got and broken this down into beginner lessons, private lessons, and ski workshops.

So beginner lessons are something we just include whenever we sell software, which is installing the software, installing the platform, installing various applications, and getting them up and running and wired up.

But that’s not where we stop. Probably the most important work comes in the private lessons, where we have three major areas. One is actually assessing an organization’s readiness for significant improvement work. Think of this as kind of a physical that you would go to to say am I ready for some hard-core skiing. And a lot of times organizations don’t realize that they may have an injury that’s preventing them from skiing some of the harder runs. And so, we provide assessment services to help understand just how ready is the organization to really engage in significant improvement efforts.

The second area is governance and understanding that this is not a technical problem only that we’re trying to solve. It’s really a change management problem. Getting people to adopt different and better ways of delivering care has a lot more to do with change management than it does to do with technology.

And finally, we’ve now entered into many contracts where we are going at risk with our customers, where we have embedded improvement team members working alongside our customer on clinicians, where if we don’t help actually achieve outcomes improvement, we don’t get paid. And we think that’s a great model to align incentives where we both go at risk for the improvement of the care that we’re delivering to patients. And so, we have improvement services that follow both integrating data and helping to redesign care delivery.

Finally, we have what we call our ski workshops and that’s really quality improvement training, getting everybody a common vocabulary where we can talk about things like variation and care delivery. We can talk about process improvement and understand different ways to improve and learn different skill sets. And so we have what we call Health Catalyst University.

Private Lessons = Assessment Services

Evaluation of organizational capacity for change, current capabilities, and gaps [18:05]

To just go into a little detail, maybe on private lessons and the ski workshops, this is an example of an assessment that we give, looking at the three major areas that I’ve described today – the analytics, the content, and the deployment. Again, the analytics would be like the ski equipment, content would be the different runs, different parts of the mountain you could ski, and the deployment is all about how well are you trained to do improvement work.

And so, we can do an assessment. This is a self-assessment that one of our clients did where we had 12 major categories that they looked at and then they self-assessed, are they just starting, are they mid-journey, or are they mature, and the center of the target there is the mature. And you can see that there is a difference of opinion in a lot of different areas, which was a great opportunity to start a discussion around well what do we really need to do to get mature and what does it mean to be mature in these key competency areas.

Private Lessons = Governance Improvement Services

Data Governance/Data Stewardship and Advanced Organizational Governance & Prioritization


Another area where we provide significant services is in helping to set up data governance, data stewardship and advanced organizational governance and prioritization. This includes setting up guidance teams which meet on a regular basis to prioritize which aims or which improvement opportunities to focus on first, midsized teams and small teams that design the actual innovation, and then broad teams that implement the innovation. And we implement concepts that Everett Rogers originally wrote about in his book ‘Diffusion of Innovation’. Part of this is identifying the right people to put into leadership roles, they should be innovators and early adopters, that will help the early and late majority get broad adoption of new ideas and new ways of delivering care and measuring how that care is delivered. This is probably the hardest thing to do and where we spend a lot of our time working with our clients, helping them to design the teams.

Organize AGILE Teams [20:15]

One of the things that we think about in designing a team is getting the right people on the truck. We use this analogy of a flatbed truck driven by physician and nurses but having the wheels be certain key support roles, such as application administrators, data architects, and knowledge managers. Application administrators can make changes to the underlying data capture transactional system. So if a change is needed for the EMR or a change is needed in one of the other departmental systems, the application administrator can actually make the change to that transactional system. The data architect’s role focuses on moving that data from the transactional system into an area where it can be interpreted, writing those extraction routines, implementing, using the tools that I mentioned earlier as we were talking about the ski equipment. And a knowledge manager is typically someone who has both clinical and technical capabilities. They understand clinically what’s going on. Typically, this person is a nurse.

Now, we stress the importance of organizing permanent teams and implementing an AGILE methodology or an iterative methodology where on a weekly basis the small team meets together and works on defining cohorts, designing metrics and understanding how care delivery could change. They then take those drafts of changes to metrics and changes to clinical operational workflow to a broader team that has representation across the clinical organization, and get buy-in and get modifications to that draft in order to get broad acceptance of the principles that they are trying to implement.

Private Lessons = Embedded Improvement Service (At Risk) [22:10]

This is an example of the improvement methodology that we leverage, which includes concepts from LEAN and PDSA, from AGILE software development and this is really meant to get systematic outcomes improvement. There are activities such as identifying well what are the best practices and having a starting point based on evidence if it’s available or expert consensus. There’s clinically defining cohorts of patients, there’s understanding how to design AIM statements, and they’re selecting measurement and metrics that will help understand how that’s going across the system. Then there’s designing the intervention and rolling it out and then measuring the progress. We provide services around all of these activities and have mentors for our client’s counterparts who will be long and permanently playing these roles. And it’s only in this adoption process do we see broad outcomes improvement, not just at one hospital or one clinic.

Ski Workshops= Accelerated Practices Program [23:15]

Finally, we provide accelerated practices program, which is like a ski workshop. We have several different flavors of that quality improvement training. First is an intense training program that we are opening up. Our first cohort is going through this training right now. Our next cohort will be in October. And this is an 8-session course taught over 4 to 6 months where they spend about 2 days per month in intense training, understanding quality improvement theory and using that theory on an actual project. We also provide an executive training where we do a short 2-day executive course at different locations across the country. And then also we have just-in-time training, which is a library of 10 to 15-minute modules that can be used and leveraged just in time as a particular permanent team needs to refresh its memory on a particular quality improvement principle or topic.

Alright. Now that we have our ski equipment and we’ve gone through some of the information about the ski school, let’s move on to actually getting to the skiing, actually doing improvement work.

Types of Ski Runs = Improvement Types [24:27]

There are a lot of different types of improvement. So, we could start with just data provisioning improvement. This could be like kind of taking the tow rope. Reducing the amount of time that analysts and data architects search and hunt and gather data and getting that into a centralized data warehouse is a real key and it’s really the first step.

Next is identifying opportunities for improvement. There’s a lot of different things you could work on but understanding where the bigger opportunities and where the smaller opportunities is really key.

Next is process improvement, actually making improvement to steps of the process. Ultimately then leads to our black diamonds or outcome improvement. And they are sorted here at the top or the most difficult runs and then they are sorted as towards the bottom or are the easier things to achieve. The reason outcomes improvement is hard to achieve is it takes a lot of skill to not only get the technical solution in place but change clinical and operational behaviors. Examples of outcome improvement would be things like actual reduction in the mortality rate, hard cost dollar savings or improved health function of the patients we’re trying to serve.

The Bunny Hill = Key Process Analysis (KPA)

Data Driven Opportunity Analysis [25:59]

So, one of the very first lifts that we recommend everybody go on is what we call the key process analysis or the Bunny Hill. Again, this is a green circle. This is an area where we’re not actually improving any care but we’re identifying where are our biggest opportunities to improve. What’s interesting is with all the organizations we have worked with, we’ve seen this 80/20 rule play out. So, a small number of the processes, we think it’s about 13 in most cases, make up a third of the opportunity. When you get to the top 40 care processes, it’s 62%. And when you get to the top 45 or top 85 processes, which is about 20% of the processes in healthcare, it represents about 80% of the opportunity. And so, the Pareto principle or the 80/20 rule definitely applies in healthcare. This helps to prioritize what to work on first.

Less Effective Approach

“Punish the Outliers” [27:05]

Now, once we start to work on an area, there’s a couple of approaches. A common approach, which is less effective, is to set a minimum standard and then try to shame or guilt those clinicians who are below the minimum standard into performing above the minimum standard. Well this is a “punish the outliers” or “rank and spank” or “cut off a tail” kind of approach. And while you will get some movement, it doesn’t really shift the mean that much.

Effective Approach to Improvement:

Focus on “Better Care” [27:40]

What we prefer and what we recommend and our services help to achieve is tightening the curve, reducing the variation in care deliveries by adopting standards of best practice or shared baseline and then the entire curve shifts towards excellent outcomes and becomes much tighter.

Cost Per Case, Vascular Procedures [28:01]

Part of what we can do is think about this in a little analogy. So if we have a single physician and they’re performing 15 invasive cases of a particular vascular procedure and their average cost is around $60,000, yet the average for the entire physician group is around $20,000, there’s an opportunity if that physician can operate closer to the average. And so, if you multiply the difference by the number of cases, that could turn into a $600,000 opportunity. As you look at then all of the physicians or all of the hospitals or clinics, and we can look at this multiple different ways, you can see that as we add these up at the shift, each clinician or each physician or each hospital or clinic down to just the average of what the system as a whole is doing, this can turn into very large opportunities.

And so, as we look at prioritizing which areas to work on first, we want to look at the areas that have a widespread variation, in other words, from hospital to hospital or clinic to clinic or physician to physician. There’s a lot of variation in the cost of the care and the clinical outcomes of the care.

Improvement Approach – Prioritization [29:20]

We can categorize this into four quadrants. So if we think at the bottom of the graph of resource consumption and the Y axis as variability, we can create a four-box matrix. Large processes with significant variation are a top priority. And then large processes which are consistent, but maybe were consistently good, not consistently excellent, are a second priority.

Internal Variation versus Resource Consumption

(Excel Example shown) [29:51]

And so, we can graph those out and we can see which are our largest processes and which are our biggest opportunities for reducing variation.

Internal Variation versus Resource Consumption

(Key Process Analysis App Example shown) [29:59]

This is done in a tool which we typically implement in the first couple of weeks working with our clients, called the Key Process Analysis tool and this allows us to get information, again, not to improve but just to identify the opportunity and that’s that first type of improvement, the green circle run.

Health Catalyst

The Greatest Outcomes on Earth [30:20]

So, if we now, we skid this one lift, the Key Process Analysis, this helps us to kind of decide which lift we’re going to get on to go after real quality improvement or real process improvement.

So we’re going to now dive into each of these different lifts and talk a little bit more about what’s included in the applications and the services related to the different lifts and the different parts of the mountain that they serve. So we’ll start by going up the population health lift.

Population Health Lift

What’s included in skiing a great Pop Health Run? [30:52]

So what’s included in skiing a great Population Health Run? Well first is an understanding of the map of the process, then understanding the potential improvements, clinically defining a precise patient registry, adopting standardization aids and then producing actionable visualizations.

Care Process Improvement Map

Sepsis (Severe Sepsis and Septic Shock) [31:15]

This is an example of a care process improvement map that identifies common problems for sepsis and common metrics to measure those problems or measure compliance to best practice.

Outcome & Process Improvement [31:30]

Once we’ve picked a particular problem area to work on, we might write a process AIM statement. For heart failure, for example, it might be we’re going to improve medication reconciliation or we’re going to try to improve our follow-up visit scheduling. Most of those process improvements will lead to an outcome improvement. So if I improve my follow-up visit scheduling and my medication reconciliation, that can have the outcome improvement of reducing my readmission rate.

Standard Patient Registry

Start with administrative codes [32:01]

We start with standard registries that help us define maybe an administratively-based cohort, but quickly we’ll need to move to clinically-based definitions of cohorts.

Precise Patient Registry

Move to clinically defined cohorts [32:17]

One example of that occurred when we were working with the Children’s Hospital, looking for asthma patients. And we stared with just the billing code for asthma but quickly we realized we needed to add wheezing and problem list codes, as well as look at specific medications that only asthma patients were on. When we combined all of that, we realized that instead of just 29,000 patients that we need to be thinking about, it was over 100,000 patients. So clinically defining a cohort is very important.

Adopt Standardization Aids or Knowledge Assets [32:52]

Then we look at implementing standardization aids or knowledge assets. These come in three categories – who should get the care, what care should be included, and how can the care be delivered efficiently. And so, if we’re looking at who should get the care, this might be an indication for referral. Should this patient be seen by their primary care physician or by a specialist. For order sets, we may be standardizing the pre-procedure or post-procedure order set. And for efficiency or workflow, we may be looking at a transfer checklist or bedside practice guideline, how do we make sure that care gets delivered in the most efficient manner.

MD Population Knowledge Assets [33:35]

These knowledge assets lay out across the continuum of care delivery, including the ambulatory setting, the intensive setting or the invasive setting. And so, many times when we start working on one of these care areas for population health, there’s a lot of variation in these different colored boxes. Sometimes it’s unknown what different positions are doing. And so, the first step is really taking an inventory of what’s going on in each of those categories.

Payment Structure Considerations [34:06]

What’s interesting is depending on your payment structure, improving one of these things, which will be the right thing to do for the patient, could have a positive or negative impact on your bottom line. So if you work on an intervention indication, perhaps getting more precise about who should get a spine fusion versus go to physical therapy, if you’re on a fee-for-service payment model for the majority of your payers, that could actually hurt your bottom line as your care delivery organization, and you may need to re-negotiate your payer contracts for that condition of lower back pain and move towards a full capitation or condition capitation payment model. So that when you make an improvement, it actually has a positive impact on your bottom line for your organization.

Actionable Visualizations [34:60]

Finally, included in the population health category are actionable visualizations. Now, I had a question submitted of what type of visualizations does Health Catalyst support. Today, our platform is agnostic. We provide some starter sets in either Tableau or QlikView but really we’ve had clients use all sorts of different visualization tools. The visualization pieces are merely the last piece of showing the data and there are different places where you might want to use different methodologies of visualization. In some cases, a static report is enough. In other cases, you want a more interactive dashboard that’s actionable.

This is an example of a heart failure dashboard used by care managers, showing which patients show the highest at risk so that they can prioritize which patients to follow up with first. And so, it’s using a predictive model for readmission combined with demographic data and discharge data so that they can actively take action on those patients that need contact the most.

Population Health Lift

What’s runs are currently open? [36:16]

So, we’re now going to just take a look at all of the different runs. I kind of talked about what’s included in each run. But these are the different runs that we’ve been working on at Health Catalyst. So those again with the green next to them are available today. They have been implemented at one or multiple clients. Those under development are being groomed right now just like grooming a run where we’re working on it with our first client but it hasn’t been deployed broadly. And then there are a lot that are on the roadmap where we plan on working with them and we’re actually actively looking for a client to engage with to work on that together.

And so, you can see they are both population health and patient injury prevention applications that include not just the visualization but also all of those other areas that I just covered – the care process map, the intervention knowledge assets, the definitions of cohorts, inclusion and exclusion criteria. They really help to jumpstart these teams that will be working on the improvement.

Alright. Let’s move to another part of the mountain.

Accountable Care Lift

Key Concepts in Shared Risk Management [37:32]

If we go up the accountable care area, which is closely related but slightly different from our population health area, there are two major areas for improvement. The first, we’ve already talked about, which is long-term care delivery improvement, which is really accomplished by skiing other parts of the mountain. But if you’re in at-risk contracts, there are some specifics that you need that are shown here in orange. This includes at-risk contract monitoring and management, network management, care management, performance monitoring, and improvement prioritization.

Analytics for Accountable Care [38:14]

So in this area, we have a tool called ACO Explorer that helps to monitor at a high level all of the contracts that have per member per month risk, looking at leakage across the network and so forth. And then we have specific tools that dive into each of these major areas. So for example, with contract risk management, we can look at our per member per month analyser to drill into the claims data to identify opportunities for improving high cost per member per month situations. And then we have another tool for evaluating the 48 episodes of care with bundled payments. For network management, we look at both leakage and referrals to know when people are going out of network. We also have a tool which understands which areas your network may be complete or you may need to contract or acquire a specialty group where you don’t have geographic coverage to handle the population that you’re at risk with. And then Attribution Modeller is a tool that we have had for some time that allows you to create different models to link patients and providers together.

In care management we have patient risk stratification, being able to look at high-risk high-cost patients and identify and predict future risk. And then we have approximately 400 out-of-the-box patient registries in our cohort builder that allow for designing specific registries of patients that you can go at risk with. Finally, performance monitoring, the ACO 33 metrics is an area here where you’re able to monitor and improve your performance on the CMS measures and look at areas for improvement in those areas. Finally, improvement prioritization, having a claims-based population explorer and key process analysis allows you to just look at claims data to see what are the biggest opportunity areas which I showed a little bit earlier.

Accountable Care Lift

What’s runs are currently open? [40:24]

So these is a status of which of those runs are currently open, which are being groomed and which are on the roadmap and you could see that a number of those are currently deployed at clients being used actively. Several more are in development right now and then patient engagement is on the roadmap and I talked a little bit about that before at Get Fit Stay Fit that we have been testing on ourselves first internally.

Incidentally, we use the Health Catalyst Analytics Platform to run our entire business ourselves and that’s something we really believe in. If the tools that we’re selling aren’t good enough to run our own business, then we’ve got a problem. And so, Health Catalyst runs its entire business on a Health Catalyst Analytics Platform. We have an internal team of data architects of corporate analytics that allows to really understand our business. We looked for variation in our outcomes improvement and analytics delivery and look at ways of improving that.

Translational Research Lift

What’s runs are currently open? [41:28]

So because of time’s sake, I am going to kind of go quickly through the other lifts. Translational research is an area we’re just getting into. This one includes tools like research assessment and de-identified cohort builder, allowing researchers to see do I have a big enough sample size using de-identified data from the data warehouse for my study. Also, being able to look at a study recruitment where I can see actually in the hospital right now that I could recruit based on the criteria I set for my study. So a lot coming out here and we’re very excited about this new area that we’re getting in to have research and using analytics to really help accelerate the research agenda.

Financial Analytics Lift

What’s runs are currently open? [42:15]

In financial analytics we have several products around revenue cycle, financial management, understanding in a much more detailed way some of the analytics around financial performance. We’re also actively working on an activity-based costing system which we’re calling CORUS®, which stands for Clinical Operations Resource Utilization System, and this will eventually integrate real-time location services and be able to passively capture data about when and how long things are taking on an activity level, not just on a billing level, which has been needed for a long time in care delivery.

Operational Efficiency Lift

What’s runs are currently open? [42:55]

If we go over to our operational efficiency lift, we have many departmental systems like lab, radiology, pharmacy and different areas where we want to improve the operational efficiency or through a particular department or typical care area. So we have a patient flow explorer that helps to understand exactly how patients are flowing through the system. And we have a labor management explorer that shows how resources are being utilized at the departments of a hospital. In addition, we put a lot of effort into our practice management area, being able to understand patient access, professional billing, and really the work around efficiency within a clinic. Those are all deployed at different client site. There are others that we’re working on in this area to expand operational efficiency.

Health Catalyst

The Greatest Outcomes on Earth [43:57]

So I hope this has been a helpful analogy as we think about the three major areas. First, making sure you’ve got great equipment and gear, the enterprise data warehouse platform, understanding the right helmet, goggles, gloves that you’ll need, regardless of what runs you’re going to ski. Second, we talked about the ski school, getting trained on quality improvement and having embedded resources that actually help you achieve the improvements, and are actually willing to go at risk to make sure that you get down the mountain safely and have a great time. And finally we talked about the different areas of the mountains all the different things that an analytics platform can help you to improve. And we talked about population health accountable care, translational research, financial analytics, and operational efficiency. I know we covered a lot in a very short amount of time.

Health Catalyst Catalog App [44:55]

And so Health Catalyst has created a mobile app and I’d like to invite each of you to download this free app. It’s at and Tyler will actually put the link up right now. We’d like to invite each of you to download that app which has much more detail on each run on our ski map, if you will, and it will include a description of the applications in each category, what improvements we anticipate that you will get by implementing the analytics, as well as the services associated with that particular area, as well as how will you know you’re successful in that area. And we talked about the different types of improvement process and outcome improvement. On any of these, you can click the checkbox next to the app, that information will be sent back to Health Catalyst, and we will start to gather information about which areas people are most interested in. And if you’d like, there’s a place in the app where you can request more information and we would be happy to schedule a demo on any of those particular apps you’re most interested in.

So I hope each of you will take the opportunity to download that free catalog which gives a much more detailed description than what I’ve been able to cover in the last 45 minutes or so, and hopefully that can answer questions you might have about the Health Catalyst platform, the applications and improvement areas we cover and our services that help you achieve those improvements.

Skiing Great Runs = Great Outcomes Improvement [46:38]

In closing I just want to say I love to ski but even more I love to see our client organizations get great outcomes improvement. That’s really what we’re all about – helping our customers achieve the greatest outcomes on earth. I hope this has been a helpful session and we’ll now open it up for some questions.


[Tyler Morgan]

Thank you so much, Tom. I would like to remind everyone that you can enter your questions into the questions panel of your control panel. I would like to note that I did put that link that you gave us, Tom, into the chat window of your control panel. You can click from there. But also that link will be included in the follow-up email we provide to our registrants and attendees. So, we also would like to thank those who responded to Tom’s question over the weekend, asking for questions and such. We hope that those of you saw in Tom’s presentation, him responding to that. So let’s get started with the first question here.

Can you speak to any real-time analysis with real-time feedback, with real-time interventions during hospitalization, like sepsis alerts or other critical clinical alerts real-time? Sure. So sometimes we’ll refer to this as closed-loop analytics where the analytics actually drive alerts in the system. This can be accomplished in two ways. One, you can actually write an interface back into the EMR system and we’ve done that at some organizations. Allina up in Minneapolis has done that in several situations. In fact, they have a predictive model on which patients are most likely to be readmitted and that predictive model then organizes a work cue used on the units to go and do interventions before the patient is discharged, which may be patient education, it may be getting a social worker involved, etc. etc.   And so, that’s an example of closed-loop analytics, where they’ve used the Health Catalyst platform to interpret and predict data but then they closed the loop and presented information and the EMR itself through interfaces to show that right in the workflow of the clinician.   That’s one way of doing it.   That’s a little bit more difficult because it typically requires you to modify the EMR.Another option is to use the analytics to simply design an alert in the EMR system, and we’ve seen that done successfully as well, where I’m maybe just trying to figure out what are the right thresholds, so I avoid alert fatigue. And so, we’ve seen people trying to look at how lab people get orders, for example, what’s the right point of setting the threshold for alerting versus not alerting. And you use the analytics kind of in the backroom, interpret that data and then use the analytics coming out of that to set alert thresholds.


So both ways we’ve seen done with our clients successfully and the Health Catalyst platform does support both of those methods.


When are you expecting the predictive modeling functionality to be an active offering to Health Catalyst? Let’s take into account providers’ past performance and probability for behavior change. Are most hospital systems that are adopting Health Catalyst committed to an ACO effort? That’s a great question. So right now we have several clients using some of our predictive modeling functionality. I mentioned Allina. Crystal Run has done some really interesting things with predictive modeling and had implemented or in the process of implementing a tool that we call “Patients Like Me” and that’s where a physician can sit down with the patient, look at their current behaviors, and then from the analytics, from the data warehouse, find similar patients with similar conditions, similar demographics and show them if there was a change in their behavior, whether it be stop smoking or improve their BMI or start taking their medications more consistently. This is the predictive model showing how your health improved and potentially your long-term costs of your healthcare will improve. Now, they have been using that for several months now and have really liked that functionality. It’s not as (inaudible) as I want it to be but we’re going there and we’re putting a lot of effort into expanding those frameworks into a lot more of our applications.The second part of your questions were how many of our customers are in accountable care situations. Well I think all of our customers are seeing pressure from their payers to move to more value-based ones. I think we have three or four of the original 16 pioneer ACOs who were Health Catalyst customers. And so, a good chunk of our customers are actually in ACOs. But not the majority. I think the majority are just starting to get into that, they are kind of dipping their toe on the waters, saying is this for real, is it really going to stick. But as we talk with payers, regardless of what happens with the actual legislation, I think value-based payments, paying for the value of the care not just for the service, you know, I see a huge coming away from fee-for-service towards value-based care payment strategies. I hope that answered the question.


If you could talk about the difficulty and complexity a team of healthcare leaders work through to become aligned on what is truly meaningful and impactful information. And how do we fairly measure it and report it out? That’s a great question and perhaps the hardest part of care transformation and improving outcomes is not the technology. It’s the actual adoption of change. So part of what we feel helps in that area is to develop a language where people can talk about variation and talk about process failure rather than people failure.   I love the quote that says, “the system is currently perfectly designed to get the result it’s currently getting.” And so, we’re not trying to change people, we’re trying to change the process. And if we talk about it in those terms, if we identify process failure, not people failures, we seem to be much more successful at seeing behavioral change in clinicians. A lot of times everyone is trying to do their very best but they don’t have the information at the right time or it’s not easy to do the best practice. And so, a lot of process redesign is all about, first, having a language that you can talk about care improvement together, having forums or those clinical teams, that flatbed truck I talked about, organizing a permanent team, not just the SWAT team that goes in and fixes a problem because you’ve got written up on your last joint commission visit. But a permanent team is going to own the quality and own the processes of care delivery long-term. That’s where we really see organizations making a difference.We actually measured the variation with our own clients, who’s able to achieve the most outcomes improvement.   Those who have established permanent team that own a clinical care domain and are focused on the quality, cost, and the patient experience, permanently have some separate quality departments. It’s actually the clinical department that does the care delivery. They have the most success and outcomes improvement.   Organizations like Texas Children’s, Stanford, Allina, they made investment in the training, doing things like Health Catalyst University, where you have an on-site improvement training course and then organizing permanent teams. Those are really really key components and we see variation from that principle and those that don’t follow that principle don’t have nearly that good results but not extraordinary results. Great question.


If you could talk about achieving data normalization when pulling from multiple vendors and systems, not to mention the many different semantics in use across the industry, how to best manage this evolving issue? Yeah, this is a really tough question and it’s one that’s coming up more and more. So it’s fairly easy to accomplish data integration if everybody is on the same EMR, the same patient satisfaction system, the same, you know, it’s one system across the board. Well that’s becoming more and more rare as we see these integrated delivery networks forming and these accountable care organizations forming that really don’t have the time more than money to switch everybody in the same EMR. So, we found the Health Catalyst Analytics platform to be very helpful in that regard, in that we can bring in very quickly 5 to 10 different EMRs in their raw format. We don’t have to integrate all of the data at once but we can get it collocated very quickly.Then as we’re working on a specific improvement, let’s say we’re at risk for diabetes patients or heart failure patients, now we can integrate just the data that we’re really trying to integrate around that care improvement initiative. And that’s the difference with a Late-Binding ™ versus an early binding approach. An early binding approach would say we have to normalize all the data in every field across all of these different EMRs right at the beginning. A Late-Binding ™ approach does know just collocate that data, get it in the central data warehouse repository, link the key identifiers that we know we’re going to link, like patient, provider, medications, things that are very common that every EMR has and then wait for the specific use cases to arise when we’re actually trying to improve something to get that Late-Binding ™ or tighter integration around clinical or operational definitions.


So again, the Late-Binding ™ approach really helps solve that problem in a meaningful way, where you can focus on the bindings of a normalization on the things you’re actually trying to improve, not just the whole universe of data elements that have been captured.

[Tyler Morgan]

Alright. Well, thank you so much, Tom. We’re so glad that all of you could join us for our webinar today. But before we close the webinar, we do have one last poll question. While we rarely focus on Health Catalyst, in this webinar, we have spent more time on what Health Catalyst does and what our products are. So, if you are interested in someone from Health Catalyst contacting you to schedule a deeper dive demonstration, please take the time to respond to this last poll question. Shortly after this webinar, you will receive an email with links to the recording of this webinar, the presentation slides, and also the link to the product catalog. Also, please look forward to the transcript notification we will send you once it is ready.

On behalf of Tom Burton, as well as the rest of us here at Health Catalyst, thank you for joining us today. This webinar is now concluded.