Breakthrough Analysis: New Insights into Reducing Healthcare Waste (Webinar)

Advanced Efforts to Identify and Reduce Waste in Healthcare Delivery

January 16, 2014 – David A. Burton, MD

[Tyler Morgan]

…you who have joined us today to the 2014 Health Catalyst Webinar Series, and we’d like to thank you for joining this particular webinar, Advanced Efforts to Identify and Reduce Waste in Healthcare Delivery.  My name is Tyler Morgan and I will be your moderator today.  Throughout the presentation, we encourage you to interact with our presenter by typing in questions and comments using the questions pane in your control panel.  We will be answering questions throughout the presentation with multiple questions and answers time.  If we don’t have time to address your question during the webinar, we will follow up with you afterward.  We are recording today’s session, and within 24 hours after the event, you will receive an email with a link to the recorded on-demand webinar.  We will also send an email to you with the link to download the updated presentation slides in approximately two weeks.

I am happy to introduce our presenter today, Dr. David A. Burton.  Dr. Burton is a former senior vice president of Intermountain Healthcare where he served in a variety of executive positions for 23 years.  He spent the last 13 years of his 23-year career as co-architect with Dr. Brent James of Intermountain’s Clinical Integration Strategy, which was the primary sponsor of Intermountain Data Warehouse.  Dr. Burton was the founding executive vice president for Intermountain managed care plans, now known as SelectHealth, which currently provide insurance coverage to over 500,000 members.  He holds an MD from Colombia University College of Physicians and Surgeons and did his residency training at Massachusetts General Hospital in Internal Medicine.  He was a charter member of the American College of Emergency Physicians and was Board Certified in Emergency Medicine.  He practiced emergency medicine and was president of a single-specialty group of 20 emergency care physicians for nine years before joining the executive team at Intermountain.  Dr. Burton currently serves as an executive chairman of the board for Health Catalyst.

Dr. Burton…

[Dr. David A. Burton]

Thanks very much, Tyler.  It’s a pleasure to be with you today and we hope that the information that we will share will be helpful in thinking about how to reduce waste and how to get waste reduction to an actionable level.  Let me just apologize in advance.  Some of the analysis that we hope to share with you are not quite completed and that was why Tyler indicated that the final slides will be ready for download in about two weeks.  But you will see placeholders.  We’ll go through all of the methodology and then the frosting on the cake, which is the percents that reside in each of the forms of waste will be available as soon as we have a chance to make sure that everything ties out.

You can see now the composition of the group today, and so we’ve got a predominance of IT and administrators, a few clinicians and then some finance folks.

1.  Constructs for Understanding Healthcare Waste

Clinical Integration Construct

Clinical Programs – Ordering of Care

I’d like to begin by sharing a construct for considering waste and waste reduction.  This is the clinical integration construct that we began to develop at Intermountain and has since been significantly refined.  The vertical bars that you see here are what we call clinical programs, some of you will know them as clinical service lines.  The important concept is that it is these physician groups that are responsible for the ordering of care.

Clinical Integration Construct

Clinical Support Services – Workflow and Defects

The next dimension is the implementation of those orders or the delivery of care and they represent the clinical support services.  And you can see there are five groupings of those – diagnostic, therapeutic, the clinic care space, the acute medical area, and the invasive, which encompasses both the traditional surgical services, as well as the interventional medical services that you can see there on the screen.

Organization of Teams

Clinical and Technical

One of the importance, if not, the most important aspect of implementing improvement initiatives is the organization of clinical and technical teams and integrating within those teams the clinical leadership as well as the technical support infrastructure.  You can see here the structure that we developed over time with the senior executive leadership team that provides overall governance and prioritization initiatives.  And then under that multiple guidance teams for each of those vertical and horizontal bars what we just looked at, and then within each of those, multiple clinical implementation teams.

So let me give you an example.  A guidance team would be something like Cardiovascular or Women & Newborn’s or Acute Medical in the support services.  A clinical implementation team within the Cardiovascular would be, for example, a heart failure team or an ischemic heart disease, or rhythm disorders or vascular would be the four major ones within cardiovascular.

And then a work group is designed to integrate the technical and the clinical and to save the time and do the lab work for the clinical implementation team.  In parallel with that structure is a contents and analytics team, which consists of those technical infrastructure personnel, data architects, knowledge managers responsible for data stewardship, data quality assurance, and an oversight person could be called many different things – a chief knowledge officer, a chief analytics officer, sometimes it’s a chief medical information officer.  The important thing is that person has some ability to train and help as a mentor those who provide infrastructure support in the workgroup and across all of the clinical implementation teams or with a given domain like Cardiovascular.  That content and analytics team is responsible for data governance.  A chief knowledge officer is the overseer of all of those activities.

Technical Support Personnel

In the next slide here we have tried to portray the key roles in that technical infrastructure, those support personnel that provide the technical side of the workgroup.  So we have data capture.  And there we have in the green, the application administrators – that could be an expert in an EMR source system, it could be an expert in the financial source system.  These are the folks that get upstream and correct the data quality assurance problems.  You don’t want to correct those in the data warehouse.  You just keep correcting them forever.  Then the data provisioning is the fairview of the data architects and the data analytics, the data architects and the knowledge managers who are the data quality, data stewardship individuals.

Organization of Clinical Teams

With being all that together in this metaphor of a flatbed truck and you can see there down by the wheels, those three roles that we just described.  Driving the truck is a position lead and the glue that holds all this together, which is a clinical operations director, and in one of those vertical bars, that would be typically a nurse, in the horizontal bars could be a nurse but more often is a technical person, that is the co-pilot, if you will.

And then you see there as pockets on the truck those are clinical implementation teams, pregnancy, abnormal newborn, normal newborn, gynecologic processes.  So that’s our schema.  These are permanent teams, they’re not project oriented.  They have integrated clinical and technical members and the technical support people down there as the wheels support all of those pockets on the truck.

Repeatable System for Deployment

The other important thing in deployment, which is the most difficult of the three dimensions of improvement projects, is that you have a repeatable, scalable process, and this is the process that we use, the 7-step process.  You can see across the top here the various meetings from kick-off to development of the goal or an aim statement, the design of how we will implement the improvement initiative, the approval that we’re really ready to launch and then again measuring and reporting of results.  And you can see the steps that lead us along that pathway on the left-hand side there.

Population Health Management Construct

Population Health Management

Medicare fee-for-service payments by venue – 2011

I want to shift and talk about some other constructs; first of all, a population health management construct.  This is the view of the continuum of care beginning in the clinic, the outpatient, inpatient, and then post acute care, including skilled nursing facilities, long-term care, inpatient rehab facilities, home health, and hospice.  This data that you see at the bottom in the bar charts are actually from the nationwide 2011 Medicare fee-for-service payments that were put out by the innovation section of CMS and we took them and grouped them and analyzed them just to show the relative size of the dollars in most various venues.  That doesn’t necessarily mean the importance of the ones that are smaller.  They play an important role.  But you can see where you may want to focus some of your improvement efforts as far as waste reduction or this concern.

Population Health Management

Anatomy of Healthcare Delivery

This next slide is something I’ve worked on for about 20 years now and it is my attempt to put on one slide how the care delivery system works.  And so, a patient enters the system one of two ways – either they are a member of a health plan and they are encouraged to participate in screening and preventive activities, and out of that, if we find something that is positive or undesirable, that may lead us into the system; or more commonly patients enter the system because they have symptoms of disease.

The diagnostic workup can occur in an acute setting, like an emergency care unit, but it can also occur in a doctor’s office.  That diagnostic workup then leads to a provisional diagnosis and a triage to one of three major treatment venues – either the clinical care environment but this can be safely managed, as for example in the case of acute self-limiting conditions or chronic diseases, in the clinic or ambulatory world, or it’s acute enough that it needs to be admitted either to an acute medical general med-surg bed or to an acute medical ICU bed, or there is an urgent need to intervene either in an interventional medical process, like a percutaneous intervention to put in a stent by a cardiologist or a bypass surgery in the surgical side of that.

Those lead on down to the procedure on the invasive side, post procedure care.  The admission to the acute medical service leads to bedside care.  Both of those typically involve substance preparation, and in the clinic side, there are algorithms of care that may lead to referral to invasive or chronic disease subspecialist.

The next element of this has to do with the algorithms for prevention and treatment.  Those are knowledge assets, and for example, there would be prevention guidelines for different ages and gender, for screening for condition such as cancer.  There are admission order sets that pertain to either the acute medical or the invasive.  There are treatment and monitoring algorithms, what will I do, where do I start with a diabetic in terms of medications and monitoring, and then there are similar more sophisticated treatment and monitoring algorithms in the endocrinology clinic if the patient needs that level of service.  There are supplementary order sets in the acute medical here as we make rounds and update things and so on down.

In the bedside care, there are clinical operations protocols that have to do with assessments and preventing patient injury, bedside care, assistance with activities of daily living, those kinds of things.  In the invasive stream, there are likewise clinical operations procedures.  A timeout in an OR would be an example of that to prevent patient injury.  And then there are post procedure order sets, or if it’s the same day or outpatient surgery, a discharge process, likewise a discharge process from an inpatient.  And then into off in the post acute care, order set setting, standardized follow-up and back to the home.

Now, there’s another dimension to this and that is there are utilization management decisions to be made, and that diagnostic workshop should be done in algorithmic format that gets as quickly as possible to the appropriate diagnosis.  So there are things tested as specific to many of the conditions for which we provide care.  For example, in a heart failure, as we’ll see in a moment, we should be looking at an echocardiogram and a BNP in order to make that diagnosis.  And so, we’d like to get as quickly and efficiently to correct diagnosis and base our triage of treatment venue on that provisional diagnosis that’s as correct as possible.

At the triage level, the question there is “where can I care for the patient safely?”  And there are big differences.  If I decide that I can care for a pneumonia, community-acquired pneumonia patient, in an ambulatory setting and it turns out to be wrong, I may end up with an emergency admission to an ICU.  Where if I had triaged correctly, that patient could have been cared for in a general med-surg bed and the difference between those costs would be the waste, if you will.

There are also indications for referral.  If I proceed down that treatment and monitoring algorithm with a diabetic and I get to a point where after, let’s say, nine months my hemoglobin A1c, my primary indicator is still riding above 8%, then that may be time for me to get some help from an endocrinologist and see whether a more aggressive insulin therapy may be required.

There are also some hybrid boxes in here.  Oh one more, the yellow or orange box, and that is indications for intervention.  Because we’ve referred someone, either acutely or over from the ambulatory stream, to an invasive specialist doesn’t mean that they should necessarily get a procedure.  It may be for a more sophisticated diagnostic testing.  And so, there should be indications for intervention that we fulfill before we go ahead and do a procedure that (16:45) both the potential of morbidity, mortality, and increased cost.

Now, if you look at those combined combo boxes that have both orange and blue in them, those are within a case, not deciding whether or not to do a case, but there are options there, choices among antibiotics – deciding whether or not to use blood products.  So those are substance selection kinds of indications.

And then there is an important box there called clinical supply chain management.  This is where the decisions with regard to the goods sold to patients – things like stents, synthetic grafts, ICDs, rhythm disorder devices, prosthetics for hip and knee replacement, where we’re deciding both what to use and trying to standardize and use our volume to get the best possible price for the patient.  So that’s the anatomy of healthcare delivery.

Waste Construct

Utilization Management of Waste and Prevention and Treatment Waste

Now, if we go to the next area, I want to share with you a construct of how we think about waste, and that starts with a population view of waste and that’s the per capita management or population focus.  And there are metrics that tend to be something for a thousand members.  And so, how many admissions do I have, how many inpatient days, how many outpatient visits, procedures, emergency care visits, or readmissions.  There, I’m looking at a population usually by condition and the utilization management waste is run out by doing the appropriate things for those members by subpopulation if you go by a condition cohort, like diabetics for example.  Once I have decided that there is an indication to do something to have the patient back for the next visit, to admit them to the hospital, to do a procedure either on an outpatient or an inpatient basis, now, I’m not talking about a population, I’m talking about an individual and a per encounter or per case management focus.  There, my metrics are different, as you can see there in the list, some sample metrics.  And now, I’m trying to prevent prevention and treatment waste by making sure that I’m doing the right things for that patient within that case.

Three Forms of Waste

There are really three forms of waste and it’s necessary to dissect the overall waste into these three forms because the way we act in improvement initiatives to reduce the waste is different as you’ll see in each of these three types.

So on the left we have ordering waste, and there, we’re talking within a case about ordering tests, ordering care or substances, such as pharmaceuticals or nutritional supplements, blood products, and ordering supplies and the waste, as well as that don’t add value.  The workflow waste takes the form of variation and the efficiency of delivering test, the care and the procedures that are ordered.  And then the patient injury wastes are defects in the course of delivering that care, whether it be test of the care or the procedures.

Ordering of Waste

So if we focus first on the ordering waste, again just to keep our contract in mind, the ordering waste is the fair view of the clinical programs, the vertical bars.  It is the process by which we get to an analysis and quantification of that ordering of care waste.  First, we have to link up the administrative codes, the ICD9, the CPT codes, it could be at the APR-DRG level, already some of those (21:02) group, and using those administrative code linkages to the dollars, we’re able to define basic cohorts.  Those are based on administrative rules.

Clinical Integration Construct

Clinical Programs – Ordering of Care

Ordering of Care

Improvement Initiative Process

We then use those basic cohorts to get a directionally correct ranking of the cases by care process families.  So as we were talking earlier, that would be at the level of heart failure or ischemic heart disease or pregnancy, lower GI disorders, that level of granularity.

From that care process family, we lay out a care process model for those that make the 80/20 cut.  Once we have that care process model laid out, we can make more robust and clinically meaningful to our physicians and nurses the cohorts.  And so, those are advanced cohorts which add to the administrative rules and some clinical rules.  Those may have to do with medications that are specific to the treatment of asthmatics, for example.  It may have to do with lab tests or imaging studies, an echocardiogram, in the case of heart failure.  And so, we make that more robust, and the benefit of that is we are now applying our improvement strategies to a larger cohort of patients to which they should be applied.

Then we look at that care process model and begin to develop Aim statements and we provide starter sets for what we believe are the most important main statements that explain or help us reduce the variability.  That leads to prevention and treatment and utilization management starter sets and those are knowledge assets or protocols, if you will, order sets, those kinds of things, that then allow us to begin to improve the process and the final element is visualizations that track the improvements and tell us whether our good idea really did make any difference.

Ordering Waste Example

If we look more specifically at the ordering waste, here is our heart failure example.  And if I don’t have an echocardiogram to measure the ejection fraction, the percent of blood that is pumped out of the heart at each stroke, then I don’t really know whether they’re a heart failure or not, and a lot of patients are treated based on symptoms of heart failure without confirming the diagnosis.  The other thing that is released into the bloodstream is a peptide called BNP when muscle is stretched in heart failure and as that value rises, that’s a pretty diagnostic of heart failure.

There are other tests like a two-view chest x-ray that shows the size of the heart, arterial blood clot gases that show desaturation that contribute but are not unique or specific to heart failure.  And then if I’m jumping immediately to a CAT of the heart in order to measure the ejection fraction, that’s substantially more expensive, and I really shouldn’t be doing that unless I can get the information I need from a less invasive, safer, less expensive just like an ultrasound.

Workflow Waste

Clinical Integration Construct

Clinical Support Services – Delivery of Care

Workflow (delivery of care)

Improvement Initiative Process

If I look at the next form of waste, the variation and the efficiency of delivery, the workflow waste, now we are in the realm of the horizontal clinical support services.  And similarly, there is a scrubbing of departmental data now that we’re going to use, we need to map those departmental codes into cohorts, we do a similar Pareto analysis, 80/20 analysis of those clinical departments with the dollars attached.  And then we select a particular department, in this case, surgical services, and we lay out a value stream map and then we get into the tools that apply more appropriately to workflow such as an A3 and Aim statements and then similarly visualizations that measure process out and outcome metrics and balance metrics so that we could tell whether we’re doing any good or not.

Workflow waste – surgical services

Workflow Waste – Surgical Services

So here’s an example of a surgical services workflow, and you see we have drilled down inside.  If I go back, you can see the red rectangle here.  Now, what I’ve done is drill down inside and said, what can we learn about the turn time of cases, the cycle time, by drilling down and creating a more detailed value stream map with that particularly part of the process.  All of these flow charts are hierarchical.  So in each box I can drive more detail in and there’s another flow chart inside of that.

Defect Waste (Patient Injury)

And then finally if we go to defect waste, these are patient injuries where our delivery of the care goes wrong in some way.

Clinical Integration Construct

Clinical Support Services – Delivery of Care

And again, these occur in the delivery of the care, the horizontals, and they have a similar process.

Patient Injury

Improvement Initiative Process

In this case, most of the patient injury processes are defined the cohorts, if you will, by CMS.  And you can see right after the website there, the definition.  We then do an analysis of the frequency, the cost, the potential savings.  We have AIM statement starter sets.  We lay out kind of a hybrid of the care process model and the value stream map, so called the (26:49) Patient Injury Prevention Process.  And then we look at where that particular defect is likely to occur as far as the units within the hospital.  We don’t need to do pressure or alter surveys in the OR because the patients are not there long enough, but we do in the med-surg units and in the ICUs.  And then we focus in on one particular aspect of our flowchart, if you will, and we’re going to define criteria.  If I’m going to put a central venous catheter in, I need to be sure that there is a reason why I’m putting it in.  And then once I put it in, there is an intervention protocol that I need to follow in order to minimize the risk of harm.  And as with all of the others, we have visualizations that help us with the surveillance process, the tracking of those injuries and near misses.

Defect Waste – CLABSI Prevention

So here’s an example for the largest, as you’ll see in a minute, which is Central Line-Associated Bloodstream Infections.  I put a central line in, I get an infection and that becomes very expensive.

[Tyler Morgan]

Alright.  Dr. Burton, we are at a great place to stop.  We have had some questions come in.  First, we have had two questions regarding the presentations and people who joined late asking if they will get the presentation slides, and we would like to reiterate that we will make the slides available for download.  We’ll send an email out when those have been updated and are available in approximately two weeks.

[Dr. David A. Burton]

There’s a great question here, which I suspect comes from one of our, in one of the finance people.  “How do you allocate overhead to determine the true cost of any particular service?”  This, you’re highlighting here a problem in our industry.  Standard costing has been used in most verticals outside of healthcare for years.  Our problem in healthcare, and there are a number of us that are working on this, is that our largest input into the cost of the case in terms of the variable cost particularly is the nursing labor, and we yet we have no granularity in there.  And so, not only we do need to figure out how to allocate the overhead, but we need to figure out how to make more granular the actual cost.  And so, the idea there would be instead of billing as we do now for the nursing time in the room in care and averaging averages, we need to get down to describing in a true and event-driven or activity-based costing center.

Now, specifically to your question, the allocation of overhead, we tend to use direct variable cost, and the reason is that the clinicians have the greatest opportunity for influencing those by changing the clinical process.  They also know the most about the relationship with fixed and variable if they get a little bit of education.  And one of the things that we found in our clinical programs at Intermountain is that as they matured, they were more and more willing to tackle the fixed and variable split within the cost center, the chart of accounts.  And that was very helpful because, as you know, if you have variability and the allocation of that overhead, it’s going to create noise in your analysis.  And so, it takes you a larger number of cases and a longer period of time before you know whether the clinical changes you made actually made a difference.

So with that, let’s move ahead.

[Tyler Morgan]

Before you continue, I would just like to remind everyone that you can ask questions by typing in those questions into that section of your control panel.

2. Prioritization – Sample Healthcare Industry Analyses

Utilization versus prevention and treatment waste

Clinical Programs

Key Process Analysis (KPA)

[Dr. David A. Burton]

So let’s move now with that construct in mind to a question of prioritization.  Healthcare consists of hundreds, even thousands, of work processes.  We can’t work on all of them.  And so, we need to figure out which ones matter the most.  And Mr. Vilfredo Pareto comes to our rescue.

We looked at this earlier and what we’re looking at now in this particular analysis is the per case waste.  I’m not going to address today but hopefully we’ll in the future webinar the per capita for utilization management waste.

So if we look at what we call the key process analysis, and this goes back to Deming and Juran, Dr. Deming thought that if we want to improve a system, we have to figure out what the key processes within that system are.

Inpatient per case KPA

Inpatient per case KPA

And so, we look at those vertical dimensions again at the per case level and this particular analysis looks at the largest and which is this inpatient.  And there, we have used the variable cost and have looked at each of those care process families, so pregnancy, arthritis, and so on, as a percent of the total variable cost.  And we have run the analysis down, as you will see at the bottom there, to where we’re covering 80% of the processes that are assignable or groupable.  You can’t work on something that’s unassignable in whichever system you’re talking about.  And so, you can see the big ones there, but we can also see it graphically on the next slide.

Inpatient per case KPA

And here, each of those blue dots along the X axis there represents one of these processes that are on the spreadsheet.  And so, that first one is pregnancy in this particular analysis.  And then the red dot above the second dot is the first dot pregnancy plus the second dot, and so on for the accumulation.  The important take-away from the slide is, as you see, the top 10 care process families account for over 40%, 40.84%, of the opportunity based on variation analysis.  And the top 32 processes account for 80%.  Now, you can see there are a lot more dots in that and what we invariably find as we work with finance at first is that because usually of an enthusiastic passionate physician that’s working out a particular process, that we may be outworking on #75 and nobody is working on #2.  The challenge is that #75 requires about the same amount of technical support resources that #2 does but the opportunity for improvement is much much smaller.

Inpatient per case KPA

The next thing we do is create this bubble chart, and this is at a level of granularity lower than what we were just looking at.  So if I was looking at ischemic heart disease, inside that is CABG, PCI, acute MI, and so no, and I’m looking now at one bubble, the PCI process.  And fortunately, because of the APR-DRG grouper, I have the possibility of adjusting those cases, the PCI, for severity.

Inpatient per case KPA

So you see in the different colors in the (34:54) the severity adjustment, and each of those bubbles is one provider, the size of the bubble is the number of cases.  If I stay away from the most severe one where you expect a little more variation, I still see there an average or a difference of about $20,000 in average direct cost between the “highest cost” provider and the “lowest cost” provider and I have adjusted for severity.  So this means my patients are sicker.

Inpatient per case opportunity analysis

If I then take that analysis and I look at one of those physicians and I say, on average, Dr. J who had 15 cases, the average cost is $60,000 per case.  But the mean cost is $20,000.  If I were to bring him down to the mean, or her, that’s $40,000 times 15 cases.  And so, that represents an opportunity of $600,000.

Inpatient per case opportunity

Severity by Mean Variable Direct Cost by Provider (Percutaneous Cardiovascular Procedures)

If I do that same thing with all of those who are above the mean and begin to bring each one of those down, I begin to add up to real dollars.  Now, this will overstate a bit, the opportunity, and it will understate the opportunity.  To say that I have no opportunity for those cases at or below the mean to improve is not correct, but to say that I can bring everybody above the mean, all of those cases, is also overstating the opportunity.  But if you just count that, you can see there still is a very substantial opportunity in this particular case.

Inpatient per case opportunity*

Physicians variation perspective

And so that’s the methodology that we use.  So now if I take those (36:47), severity adjusted, and I look for the mean and bring those cases above the mean down to the mean, that’s the methodology that results in this opportunity analysis.  We’re taking the same 32 care process families in the same order that they were but you’ll see in the far right column that the percent opportunity using that adjustment methodology I just described is not necessarily upholstered in the top case.  So the total opportunity, you can see the numbers at the bottom, is about a 14% opportunity.

IP per case ordering waste

Opportunity Analysis

The important thing to note over at the right there and on this next slide is that that opportunity of per case waste includes all three types of wastes.  It includes ordering waste, workflow waste, and the defect waste of the patient injuries.

IP per case ordering waste

IP per case ordering waste

Sources of ordering variation within a case

If we look first step how would we think about and get at the inpatient per case ordering waste and do an opportunity analysis and just a reminder that these are the ordering physicians with vertical clinical programs, and what we’re talking about here as sources of ordering variation within a case are diagnostics.  Examples, large examples would be laboratory test variation and ordering things that aren’t contributing to the case, diagnostic imaging studies, very expensive tests.  Therapeutics could be therapies like respiratory, physical, occupational speech.  It could be our expensive substances – blood products, antibiotics, other classes of pharmaceuticals.  And it could also be in some cases, largest part of the cost of the case, is in fact the prosthetic, for in joint replacement cases.  Stents are also expensive, synthetic, bypass grafts, heart rhythm devices, all of those things, neurostimulator.  You can think of all of the things that fall in the cost of goods sold to patients category and that gives you some thoughts.

And so if we come back, what we’re really working on are admission and other order sets to reduce the variation in the diagnostics that are ordered.  We’re working on more intelligent substance selection, as well as clinical supply chain management.  That’s how we get at that ordering waste.

Per case workflow KPA

Sources of per case ordering waste

IP per case ordering waste opportunity

Per Case Workflow Waste Opportunity

Now, this is where we have a work in progress still.  The analyses are nearly done but they need to be checked.  We need to be sure they tie out before we send them out to you and say this is about proportion.  That’s very exciting to me and that may say something about my entertainment threshold, but to be able to tell you what percent of per case waste is in ordering versus workflow versus defect is pretty exciting.  I’ve worked on that a lot of years.

Reducing per case ordering waste

Order Sets and Indications

And so, how do we get at this?  Well, first of all, we start with evidence-based order sets for care processes that are in that Pareto list.  So try to reduce the variation and the ordering of simple diagnostic test.  Then as we get to focus on the more expensive things, we need evidence-based indications and cost information to standardize utilization of the imaging tests, the substances, and the major clinical supplies that come out of the supply chain management.

Reducing per case ordering waste

Health Catalyst Advanced Applications

We have and are developing additional ones, the High-Level Care Process map, where we talk about an Aim Statement Packet that typically has a major goal in 2 to 5 Aim statements starter sets, not meant to be prescriptive but meant to help people with the thought process.  A cohort definition that supports the Aim Packet, sometimes that simple administrative rules from most ICD codes often get supplemented for that Aim packet with some clinical observations like lab tests and imaging studies.  We have common metrics that apply across all of the particular conditions and additional outcome process and balance metrics that support the Aim Packet and starter visualizations, likewise that reflect those.

Workflow Waste

Opportunity Analysis

Per Case Workflow Waste Opportunity

Clinical Support Services

If we look at workflow waste and how we look at that now, reminder again, we’re talking about the horizontal clinical support services getting after this.

Per Case Workflow KPA – OSHPD

And similarly, this is a very large data set.  It’s the Office of State Health Planning and Development in California.  And so we took that, we scrapped the data and looked at what department really gives us the greatest thing for the buck.

Per Case Workflow KPA- OSHPD

And you can see this is more concentrated than the vertical ones with med-surg, acute care, accounting for those 20% of what we’re talking about.  They are color-coded for those (42:20) bars that you saw in the horizontal, clinical support services.  If I look at the (42:27), the biggest one is the acute medical.  So med-surg and ICU, emergency care.  Therapeutic is the next one.  Therapeutic is the next one, that gets into pharmacy particularly.  Invasive is smaller than I thought it would be, diagnostic, and finally the clinic care space.

Per Case Workflow KPA

Workflow Waste

We don’t have this analysis yet for the data set that we were looking at, the smaller data set that has the same denominator.  This will be in your updated slides that we will send you a link to download when they are available.  And similarly the same thing here in terms of the (43:07).  I suspect they will not be different.  And finally, the opportunity analysis that will be able to tell what percent of this.  This is a dry run that we did on the California data.

Reducing Per Case Workflow Waste

Here, the tools are different.  It’s a Value Stream Map, the clinical departments that are in that Pareto list.  One of the challenges is that almost everything that we do in facilities today, with Lean or GPS improvement systems, they are highly manual.  They (43:34) marks their A3s that are associate-run.  Automation is really essential to the ability to scale.  You can do manual processes for three or four departments.  When you get to the fourth or fifth, you begin to lose the games in the first.

The key to automating this is timely and complete capture of time stamps.  In some cases, we found pretty good accuracy, pretty good data quality in the time sense.  In other cases, we find a lot of missing time stamps that we need for that value stream map, start and stop times, for the value-added steps, and we also find that they are not hindered in a timely way because we still have the yellow sticky note paper towel syndrome going on.  And success in wringing out workflow waste will depend on figuring out how to improve the completeness and timeliness of the time stamps.

Reducing Per Case Workflow Waste

Health Catalyst Advanced Applications

Again, we have starter set packets.  I will not go through all.  They are similar to what I described for the ordering aspect.

Defect (Patient Injury) Waste

Opportunity Analysis

If we finally look at the defect or patient injury waste opportunity analysis, here we have defined for us, as you saw in that little thumbnail of this, by CMS, the inclusion and exclusion criteria with the exception of the ventilator-associated pneumonia which is a very large patient injury process and ADEs.  So we’re working on those to bring similar cohort definitions.

HAC Cohorts/Registries

California Data versus Medicare Data

Here is the similar analysis of those California data and also of the Medicare data that I showed to you in that nationwide innovation sample.  And the interesting thing here is, yes, we can tell you on these large data sets what the rank order is of the various types of patient injury.  You can see there that central line-associated bloodstream infections, vascular cath–associated infections, is by far the largest in terms of frequency.

Estimated Cost of Defects

2011 OSHPD Data

But more importantly, look at the cost here in terms of the step function.  50% plus of the cost of all those patient injuries is associated with those vascular cath-associated infections.  Again, we are in the process of doing and using the same methodology of applying it to our smaller data set, where we have more homogenous information.

Estimated Potential Savings

Patient Injury (Defect) Waste

And then you can see this when they come out but trust us that we did do a robust analysis to figure out the difference between the cost of cases that had a defect and those that did not within, in this case, the MS-DRG because that’s all we had in the California data set.

Patient Injury Waste Opportunity

And the interesting thing is on the left we have the total count of patients is 6,006, same on the right, and the total cost for each of those in the California data set.  And you can see the difference there is the improvement opportunity.

It is going to be smaller in your hospital and the defect or the ordering waste but the aura of safety is very important.  And so, this is what’s working on not because it’s going to be where most of your waste is but because you need to be able to demonstrate with all of the people that are profiling you that you are in fact a safe institution.

Focus on Workflow/Defect Waste

And in addition, you don’t want to be deemed, as I’ll show you in a minute, and this is new information to be implemented in 2015 and they at least got this partly right in the measurement domain no. 2, which is going to be weighted 65% in terms of the penalties.  You have central line-associated bloodstream infections and catheter-associated urinary tract infections.  And then you have a long list over on the left that is weighted only 35%.  So, this would suggest in terms of your financials and also the importance of the injury to the patients that we want to be sure that we are focusing on the things that matter the most.

Reducing Per Case Defect Waste

So, we have a process here again that we go through.  We define for each type of defect a Patient Injury Prevention Process.  So I don’t use the same flowchart or process to prevent central line-associated bloodstream infections that I do for pressure ulcers.  But I do follow the same schema.  I want to just decide what cohort of patients may be at risk that I should be screening.  Screening isn’t free.  It takes time and it costs money.  And so as I said before, we are going to screen all of the patients that are on in the ICUs and on the general med-surg units, probably excluding the post-partum unit.  Hopefully those patients are not there very long.  We are going to screen them using something like the Braden Scale tool to define which patients really are at risk.  And the reason we need to do that is find out a subset that, let’s say, have a score of 14 or whatever you want to set, 16 – is that the clinical operations protocol, the third bullet, that’s going to be implemented with those at-risk to try to prevent the injury is not cheap.  There is supplies, there is nursing time in terms of turning the patient frequently.  So we want to be sure that we are applying it where it’s going to make a difference.  And then we need a tracking system to detect injuries and near misses.

The thing that this approach allows you to do is to look at patient injury as a process failure to be subjected to root-cause analysis rather than as an “incident” to be reported.

In the one case, we’re looking for something to blame; in the other case, we’re blaming, which is most often the true place put to blame, the system or the process and we’re looking for process failures.  That is much more engaging than having forethought instant reports.

Reducing Per Case Defect Waste

What do we do in this, we have a Patient Injury Prevention Process map like we showed, the Aim statements, the cohort definition, the common metrics, and specific ones in the visualizations.

Why Focus on Workflow and Defect Waste?

IP Per Case Waste Reduction Opportunity

Facility Perspective

Focus on Workflow and Defect Waste

Now, why focus on workflow and defect waste?  As most of these are focused on ordering physician waste, we think that the size of the opportunity is going to be fairly large when we mention the aura of safety.  The other thing is, this is especially true with the workflow, is it helps with the payment structure schizophrenia that I will illustrate on the next slide.

Payment Structure Considerations

If we reduce workflow waste, it doesn’t matter how we’re paid, under all forms of payments, were going to benefit.  If we reduce defect waste, any place we’re paid on a per case basis, a perspective payment basis, we’re going to be benefited, and so all of the CMS patients who’s going to help us.

In addition, not only does it help us but it helps us avoid the CMS penalties.  And the other consideration is that it is easier to organize clinical operations teams than it is to get all of the physicians organized.  And so, this is a little less steep step.

CMS Penalty Considerations

Now, this slide shows you that across the top we could be paid a discounted fee-for-service through full capitation. And in workflow waste, when we eliminate that, it helps us in all of those areas that makes us more profitable on a per item or a per capita.

If we go down to the patient safety role, we can hurt ourselves as we eliminate defects, if we’ll pay discounted fee-for-service or per diem.  But under any kind of a per case for Medicare and if we negotiate with the commercial affairs, which they are usually happy to talk in per case or bundle per case, we are benefited.  And clearly when we get to capitated payment, we’re benefited.

It is the kind of the picture as we move out this year and the next three years in terms of the penalties.  And if you look at the hospital-acquired conditions, the defect waste, and the readmissions, which are a form of workflow waste, you can see that there’s a substantial penalty to be avoided if we are able to reduce that waste and get ourselves out of the penalty category.


Questions and Answers

[Tyler Morgan]

Alright.  We’ve now come to questions and answer portion of the presentation.  I would like to remind everyone that if you do have a question, please type that question into the questions panel on your control panel.

I would also like to take this opportunity while you’re typing those questions out to let you know that we will be at HIMSS 2014.  You can meet us at booth #6076.  We are actually sending a third of our company to address the many questions that you can ask and to show up some of our exciting new applications.  We are setting up 45-minute long meetings to go deeper with any of you that are interested.  We are very excited that two of our current clients, Stanford Hospital & Clinics and Texas Children’s Hospital, will be presenting.  Last year, we were overbooked.  We love to spend time with as many of you as possible.  So if you would like to meet with us at HIMSS 2014, you can call us at the number you see on the screen or you can go to

Alright.  Let’s get right to our questions.

Dr. Burton…

[Dr. David Burton]

So there’s a great question here, “How do you get buy-in from the medical staff that you approach?”  And that’s really the critical question.  We can have great analytics, we can have even good content, but if in fact we don’t have buy-in for the people who are the only people that can safely change the care, it doesn’t work.  And so, if we think back to those teams that we talked about, the purpose of that clinical implementation team, which was in the center just above the workgroup, it is to get buy-in.  And so, you’ve seen several places in the presentation today that we talked about starter sets.  We don’t ever take into a client a fully bait knowledge asset of any kind, an order set or a clinical protocol or whatever.  We take in a starter set.  And our expectation is that they are going to bid up on it and you want people to bid up on it because in the process of modifying it, if we think about a definition of the rules for a cohort or a registry, when the clinicians get in and modify that, then they begin to own it and it is their cohort, not a Health Catalyst cohort.  When they do that, that becomes material because those physicians and nurses that are sitting around the table in that clinical implementation team need to own that version of best practice because they are going to lead implementation among their colleagues back at their institution, whether that be a clinic, whether it be a hospital, or whatever.

So, of all of the things that you have to do to implement improvement initiatives, really the hardest one is the involvement of the clinicians.  I would say that if you have a good infrastructure and you’re able to produce those reports and visualizations like we looked at a few samples, it really engages the clinicians.  And if in fact you have a robust infrastructure, so that when the clinicians are generating improvement hypothesis or want to modify something, you can either do it on the fly in the meeting or you are back very quickly in the next meeting, and every meeting they see progress, then you start to really build some trust in relationships.

Now, there’s a question here, “Is it possible to copy the slides?”  The answer is absolutely yes.  We will be sending you a link to a download site as soon as we have those placeholder slides replaced with the updated source of truth.  And so, that would probably be within the next couple of weeks.  We’ve got a little more analysis and checking and making sure that it ties out before we put something out there that we claim is accurate.

Next question is, “How will the process of care family mapping be impacted with the implementation of ICD10?”  It’s a great question.  So everything that I’ve seen in ICD10 enhances what we’ve done in our mapping.  There is a translation table from ICD9 to 10, but as you know, there’s a large superset that is not in ICD9.  I have to say, of all of the different code sets that I met, the ICD code set is the most logically arranged.  So we are going to have to go and map all of those new codes into the clinical integration hierarchy, but I think that will only enhance the granularity and the accuracy as we use those code sets to try to look at the variation and the sources of variations.  So it’s a great question.  And by October we’re going to have that done.

“Could you suggest an open source of data for identifying benchmark cost to supplies by US region?”  I really don’t know of one.  I am sorry.  I can’t suggest that.  If your system is large enough – well let me take that back – one source, if you belong to a group purchasing organization, a DHA or a Premier type of organization, they should be able to provide you with some information.  And in fact, one of the values that they add is to be able to show you that and how much they are saving as a result of you throwing in with them to participate.  So they should be sources that would help you with that comparative.

Thank You

So that pretty much lines this up.  We’re at the top of the hour.  Thank you very much for your participation and we look forward to visiting with you again.

[Tyler Morgan]

Okay.  Thank you Dr. Burton.  Thanks to all who have joined us.  After this meeting closes, you will have the opportunity to take a short 6-question survey.  Please take a few minutes and fill out the survey so we can continue to bring you relevant content.  And within 24 hours, you will be receiving an email with a link to the recording of this webinar and then we will also send out an email with a link to download the updated presentation slides in approximately two weeks.

On behalf of Dr. Burton, as well as the folks here at Health Catalyst, thank you for joining us today.  Have a great day.  This webinar has now concluded.