2017 Public Health Grand Rounds 02/03

PUBLIC HEALTH GRAND ROUNDS Linking Research to Community Health Improvement Jointly sponsored by the Department of Public Health Sciences & URMC Center for Community Health

- Pesis-Katz and Dr. David Krusch and their talk today
will focus on population health analytics
and value-based health care payment.
Dr. Irena Pesis-Katz is a senior director
for population health management, informatics,
and payment innovation at the URMC.
As part of her role, she provides oversight,
leadership, and support for all provider payment
innovation contracting, population health management,
reporting, and analytics,
and value-based contracting strategy.
Dr. Pesis-Katz and her team support the development
of payer insurance contracting strategies,
develop a population health analytics capability
across the health care system, and translate,
integrate, and implement payment innovation
and informatics strategies in collaboration
with cross-functional areas.
Dr. Pesis-Katz is a social professor of clinical nursing
at the School of Nursing and a social professor
of public health sciences at the School of Medicine,
both at the URMC.
In 2009 she was to the Upstate Health Research Network
which has worked to reform
the national consumer reimbursement system
for out-of-network health care services.
As a native of Israel, she earned her PhD
in health service research and policy
from the URMC Department of Public Health Sciences
and a Bachelor of Science degree in Health Systems
Administration from the University of Ben-Gurion in Israel.
Dr. David Krusch has over 22 years of experience
in all aspects of clinical and operational applications
in a large multi-facility academic medical center.
He's responsible for the strategic planning,
design selection, implementation, and ongoing
operational management and the support of the most major
transformational health IT projects ever taken
at the URMC here.
His experience includes major academic leadership roles
and executive-level people
and process management and responsibilities.
As a practicing physician surgeon, his senior leadership
trans-la-tion-ist between the entire clinical constituency
and the clinical and management applications
that advance the business and the practice of medicine.
His expertise and strength lies
in providing strategic planning, vision, operations,
cross work, assessment, design development implementation
and post-implementation operations support
and optimization services related to process re-engineering
and information technology projects.
In his roles as the URMC Chief Medical Information Officer,
he oversees the e-record enterprise ecosystem
of clinical and business applications
that support all aspects of the medical center's operations.
Together with senior leadership, he's partnering
with Dr. Irena Pesis-Katz to design
and implement the UR medicine population health
management IT foundation.
So, welcome both.
- Thank you.
(clapping)
Can you hear me okay?
- [Audience] Yes.
- Thank you.
Okay, let me make sure
I can actually upload our presentation.
Okay.
Thank you everyone for joining us this Friday noon.
We are going to talk a little bit today
about, I'll start with what we're, this one is loud.
What we're doing in terms of value-based care,
where we're going, where the organization,
the medical center and the system in general is doing.
And then, we'll talk about the importance of IT
and analytics foundation to actually be answering
some of the questions that we have
and to change provider workflow and to help, actually,
the workflow to impact patient care.
I'll show, after Dr. Krusch presents, DIT foundation
and where we're going with this, and the roadmap.
I'll also show an example to what we're doing right now,
even before we have the full foundation in place,
because you really can't manage patient population
without having data information analytics
and have conversations with providers about all of this.
We will try to leave room for questions at the end,
but if that's okay with you, if you have questions
or clarifications or anything you wanna share,
just interrupt us.
I go really fast, so please, just raise your hand.
I'll notice it.
Before I even start to talk about the
value-based care and different contracts that we have,
I think it's very important that I describe the system.
I know that we're all, or, I'm sorry, most of us,
are at home, and part of the system, but
to understand who we are,
it's important to understand the structure that we have
and why we're going the way we're going
and why we have some of the complexities.
So, let me skip the University of Rochester
and URMC that we're actually in, the School of Nursing,
and talk to you a little bit about UR medicine
and then the entire network.
So, within UR medicine we have the hospitals,
we have our affiliates, and I already apologized
because this slide is outdated.
Noy-son-jones is supposed to be here on the yellow side.
We have our medical faculty group
and then, as part of our greater network,
we have the Accountable Health Partners,
which is our contracting entity for value-based care.
We include through that relationship
the different providers in the community,
PCP's as well as specialists.
This is just showing basically,
and that's the recent slide, I stole it from Mark Dowd-man,
of the size of our network and our services
and our system and what we provide.
Just so you understand, when I was showing
the size of HP, we're talking about slightly less
than 2,000 different providers.
About 300 of these are PCP's, and the rest are specialists.
Most of the specialists are employed
by the University of Rochester Medical Faculty Group,
and the rest are HP community specialists.
Before we started discussing the foundation
and what we need and the infrastructure,
we actually sat together and thought,
"What does it mean, population health management,
"at the University of Rochester Medical Center?"
Because if you ask what is population health management,
at different places or different people,
you will get different answers.
It was very important to us to start with a definition.
When we say here, population health management,
what are we talking about?
It's really operating and developing a system
that is using all of these.
It's using data, analytics, evidence-based medicine,
clinical decision support, and care coordination
to deliver care.
We're not talking here about delivering care
as a fee for a service or on a fee-for-service basis,
because for that we don't need population health management.
It's really to deliver care in episodic
and covered or attributed life basis.
So, if we have a population we are responsible for,
how do we deliver care for this population?
Not necessarily that it's coming to us already, by the way.
This is a snapshot
of all the contracts that we currently have,
or working on having, within population health.
It's pretty comprehensive.
What you have on the left is the different payers
or different types of contracts that we have,
the government, Medicare, Medicaid.
I'm sorry, commercial insurance, and direct to employers.
Across all these payers or payment mechanisms,
we always have fee-for-service.
And in fact, everything, even within population
or value-based care, at this point, at least,
is built on a fee-for-service chassis.
So, we start with a fee-for-service, and then at some point,
depends what is the payment mechanism
and what is the contract, there is a reconciliation
of all that is going on.
So within those, and I think,
especially when we talk to clinicians,
I think what is very important to understand
is that it's not,
when we move towards more and more risk
and risk-based contracts, it's not that we stop
whatever we were doing before.
In fact, everything in red,
we have those current arrangements,
and everything in blue is things that we think
or we're working towards, we think we will have
in the future or we're working
towards getting those in place.
So, when we talk about pay-for-performance,
or we talk about shared savings,
or we talk about bundled payments, which I will explain,
I'll give a specific example later in the talk,
we're not stopping what we were doing before.
So we can have a shared savings program with one payer
and a full risk program with another payer,
and actually we can have a health plan,
which is the U of R,
totally in addition and have an arrangement with that.
So, this spectrum is not
on a time baseline.
It's just a spectrum of services.
The more you go to the right, the more risk the provider is,
and the more important managing patient population becomes
and the urgency to really have a good foundation
increases.
So, just to quantify a little bit what we're talking about.
If you look at our entire book of business,
or basically our revenue from everything coming to us,
we wanted to understand how much
and how much did it grow over time.
How much of our
revenue to the system comes from value-based care?
What we have in 2013 is only 8%.
But what we're looking at today is about 40%
of all revenue
is based on risk-based contracts or value-based care.
That's, first of all, a big number.
That's a really big chunk of our revenue.
Second, it's a huge increase and very fast
if you look at just the number of years.
Again, just increases the urgency
of having a good foundation.
Just to briefly explain the difference
or what we're talking about in terms of the shift
from volume-based to value-based care,
the objective is different when we're talking
about value-based care, fee-for-a service world.
You're getting paid for everything you provide.
So, the more services you provide,
the more you're getting paid for those.
In population health management or value-based care,
you actually need to manage a population within a certain
contract, budget,
and you're responsible for that.
You don't want to provide too much,
and you don't want to provide too little.
You basically want to keep the patient
at the right level of care
and definitely not needing urgent type of services
or higher level of care.
So, if you're thinking, just to give you an example
to translate what I'm saying, if you're thinking
about our affiliates, as an example, if we have
community hospital affiliates, we don't want all the cases
from the regions to necessarily come to Strong,
which is an academic medical center, and do it here.
We want to try and keep it in those communities
as much as possible
so we get only the really complex cases.
Yes?
Yep?
- [Audience Member] So value-based plans...
- Volume, this is volume.
(chuckles)
Could be my accent too.
(laughing)
In terms of the different things that we're doing for
unwarranted variation, or to manage patient population
in a systematic,
consistent way that delivers
systematic and consistent quality and outcomes,
we have different, and this is not an exhaustive list.
This is just an example of the different things
that we're doing.
We have bundled payment for care improvement.
That's the CMS program.
I will give you that as an example later.
We have care management protocols and tools
across the continuum of care.
Again, we're not talking about, now,
managing patients in the office or the hospital.
It's really across the entire continuum of care,
regardless to where the patients are.
Utilization monitoring, out-of-network referral monitoring.
This is something that you may have heard
referred to as leakage,
or how much is going out of our system compared
to how much are we keeping in our system.
The importance of that is not only
to really keep the patients in system.
The importance of that is to provide continuity of care,
have all the information
in the system so when the patient is coming to us,
we really have all the data and all the information
related to that patient,
as well as
consistent quality of care across.
Yeah?
- [Audience Member] Could you go back a slide?
- Yeah.
Mmhm.
- [Audience Member] The population count, 32%...
- Yeah, this is somewhat old data, but that's, yeah.
This is not ours, no.
This is not our data.
This is Blue Cross Blue Shield of Maryland.
- [Audience Member] Okay, 'cause that's a very,
very big ratio.
- Right.
- [Audience Member] I've not seen that big of a ratio.
- That's a 2009 data.
I can't speak too much to it besides the fact
that I stole that.
The percentages, I can't remember now.
She's asking, "What's the percent
"of cost that is being generated by the top 2% or 5%
"of the population?"
It's pretty high.
I don't think it's 30%, but it's pretty high, yeah.
Primary care physician incentives
to achieve different standards of care.
We have those in place, and again,
there are different examples for that.
And clinical quality dashboards,
not only for clinical departments,
but also for primary care providers.
And the primary care role becomes increasingly
important when we're talking about managing the population,
because this is
the patient's home.
This is the patient's focal point and hopefully
the care management across the system is
at the primary care offices and provided by them.
We have what we call a portfolio
of our population health programs.
The portfolio is live and changing
depends on
what we have already established and what we're working on
and what we want to achieve in terms of our goals.
These are, we divided those different projects,
and actually each one is a set of project,
to three main components.
I'll talk about the last one separately.
The foundation,
the IT foundation to allow
and to be the infrastructure for everything
is creating a data warehouse.
I know there are many researchers here,
so the clinical data warehouse, some component of it,
once we have it established and implemented,
so we are in stages of implementing that,
will be available then as a research data warehouse
once we have that in place.
We're working on making that available.
The second component is the clinical decision support,
and we have different projects that fall into it.
Clinical variation is one of them,
leakage management or out-of-network referrals,
risk optimization, and all of these are very important,
especially again, when we are talking about taking risk
with the payers.
So before,
if we only care about fee-for-service,
it doesn't matter if we do not optimize reporting
the risk of our patients.
But now that we share premium,
we care about the premium that the payers are getting,
it's very important for us to actually be accurate
with reporting patient risk and optimize this.
In terms of care management, again,
we have different projects going on
with respect to care management work flow
and bundles, which I'll talk about.
And embedded in all these projects are two main things
and central things.
One is the provider
experience, so we are working with the providers
to make sure that whatever is being done,
whatever projects we're working on,
are actually viewed as helpful
and can help facilitate the care
that the providers need to provide.
The other thing, the other very important component
is patient engagement.
Again, across everything that we're doing,
we're trying to look at the patient
and what are the implications?
We're now looking at actually something
that will allow us, a tool that will allow us
to pull the patients for everything
that we're trying to implement,
so we're really putting the patient
in the center of population health management.
Dave?
- Thank you.
Thank you, Irena.
Can you guys hear me?
Irena, if you don't mind,
this slide generated so much interest.
I want to go back for just a second, okay?
One of the opportunities here is,
what's the opportunity?
These are the well patients.
You wanna engage them and you want to put programs
in place to keep them well.
These are the rising risk patients.
But this population here,
this is where your opportunity exists,
the opportunity to intervene, to create programs,
to gather information.
You don't wanna let this population get
to the high class population, 'cause by the time
you're here then you're managing sick patients
in the ICU with multiple chronic conditions
that have gotten beyond your opportunity
to intervene and prevent that.
When we're talking about managing populations
and intervening and putting programs in place,
yes, keep these patients well,
manage these patients so that they don't get there,
and then put a lot of programs in place
for care coordination and care management
to address this population.
That's where the value proposition or the opportunity is.
That's a tremendous amount of what population
health management is gonna be.
Now, let's get back to the IT component.
You can't manage what you don't have data.
You can't manage what you can't measure.
We have to be able to measure information
about the population.
The challenge with populations is not all
of our patients receive all their care in the same places.
They're not all under.
We think e-record is the final common solution
to everything that ails us.
Well, if we only had care delivered at URMC,
that might be case, and I said might be the case.
But that's the way, our populations are very mobile.
You need to be able to combine disparate data sources
from multiple places, including clinical data
as well as financial data.
You need to understand the variations in care
and the gaps in care.
From that, you can then derive, from those gaps in care,
what the best practices are to manage that population
and provide decision support where it really matters,
and that's to the health care workers at the point of care.
That's kind of the global scope of what the IT's support.
When you think about it, it's a virtuous cycle
because none of this is an end game.
First, you need to, with any, this is any business
proposition, define what your goals are
before you wanna measure something.
Then you identify the measurements
that are gonna help move the needle or bend the curve.
Once you identify the measurements that can move the needle
you can identify the data components.
Then you have to ingest the data.
You have to get the data from whatever sources it exists in.
Once you do that, you have to normalize the data,
because if we take data on populations
from multiple sources, it's not gonna necessarily
all mean the same thing.
I take a problem list from e-record and a problem list
from a community physician's Med-End EMR,
we're gonna have to normalize those
'cause it won't necessarily mean the same thing.
Unless it does, we're not gonna be able to act on it.
Once you normalize the data you can then map
different components as well, 'cause the patient
might be identified, the provider might be identified
differently in different sources,
so that's part of what we call
the semantic interoperability of the data.
Co-mingle it in that data warehouse.
Once you have it in the data warehouse,
we've done all that work, now you can begin
to analyze and derive these measurements
and key performance indicators.
Once you have the key performance indicators,
you can present the information in a dashboard
to the people that it matters to, the care managers,
the case managers, the providers, the social workers,
and then they can begin intervention.
Not only to here can you begin intervention,
you create the care coordination task,
you action on those tasks.
Well, then you wanna refine your goals and measurements
based upon the outcomes, 'cause if we don't look
at the outcomes based upon our actions,
we're not gonna get any place.
Once you refine those measurements,
you're gonna go back and redefine your goals.
Like I said, hopefully it's a virtuous cycle
versus the opposite of that, which is a futile cycle.
You don't like futile cycles.
So, what are we doing today?
What do we tangibly have in place to do this?
Obviously we have Epic, but not everyone's in Epic.
Irena's team has an in-house claims analytics group,
and our Accountable Health Partners,
which is our contracting and our care management arm
for the institution, has a data warehouse
that was purpose-built for the intent
of their contracts that we have called
Arcadia Healthcare Solutions.
So, we'll start to understand
how these add up to one another.
So, this is Arcadia.
Where did Arcadia come from and what does it mean?
So, we have
quality contracts, value-based contracts
with Excellus and others.
But let's start about the fact that in order
to manage these value-based contracts with Excellus,
we had to put into place a system to aggregate data.
At the practice level, you have all the practices.
I think there's 50 practices right now
with 11 different EMR's.
One of them is URMC and affiliates, there's one practice.
But it reaches out into other practices.
There's community-based PCP's.
There's 2,000 providers, community-based PCP's,
community-based pediatricians, community-based specialists,
and so all of these practices may be
on different electronic health records.
Arcadia, on behalf of AHP, connected these practices
with an electronic health record connector,
and went through
this process.
They identified the measurements.
That was not a trivial process.
They then identify what data needed to be ingested
and they mapped and normalized the data
from these practices, and they stored it
in a data warehouse.
From this data warehouse, we can then present
the key performance indicators, the quality measurements.
From those quality measurements, we can look
at how we can manage our practices to deliver
to the measurements that we're sharing the risk for.
We started with shared savings,
and then we went to shared risk.
We're in shared risk right now,
so we can optimize the care of the population
so we can deliver the value.
What's value?
I'm not an economist, but one thing I learned
from Roy, where's Roy Mil-in-isk, someplace here,
from Roy, that value is quality over cost.
We wanna deliver the highest quality
at the most reasonable cost.
Not the lowest cost, but the most reasonable cost
for that quality, and to do that,
you have to have those measurements in place.
This is what we call our data warehouse.
It's affectionately called the blue data warehouse
because that's the way this slide was colored.
But we also have another representation
of this that we have.
This data warehouse is only that population
that Excellus has contracted with us for.
The rest of the population's anonymized in there
so you can have a baseline community.
We also have a greater depth and breadth of data
on behalf of us, where we have all of our data
hosted by Arcadia through a BAA.
Nora?
Thank you, Nora.
Okay, that we can see 100% of the identifiable information
and we can add to that as well.
We call that our orange data warehouse.
You can understand why I put a little bit of emphasis
on that, because where we're going next with that.
This is purpose built for the value-based contracts
we have with Excellus and other payers
to facilitate
the work of AHP on behalf of URMC.
You can see that this is one of the Arcadia dashboards.
It shows a particular practice,
how many points they've captured for this population
and the quality measurements:
diabetic blood pressure control,
adolescent immunizations, hemoglobin A1c,
so on and so forth, and it shows them
where the gaps in care are.
Based upon those gaps in care,
they can look at the populations and the gaps in care
and they can look at the patients,
and then they can create interventions
for those patients and manage their care
through care coordination and outreach
so that they can prevent those populations
from that rising risk and getting it to that red zone.
The four major conceptual components,
multi-source data ingestion.
We wanna be vendor agnostic.
We wanna make sure whatever system we put into place
isn't just knowledgeable about data,
and it isn't just knowledgeable about data in meta,
and it isn't just knowledgeable about the RMC layout
or the ACM layout or the UMI radiology
or the bor-gon-ai radiology,
but it should be able to ingest information
from any particular vendor.
When it aggregates the data, we wanna make sure
we can normalize it and match the patients.
Then we wanna have on top of that an analytics foundation
that can identify those gaps in care.
But the most important component was
we wanna make sure this information
is embedded where it's gonna make a difference.
If you have multiple separate systems in front of you,
and you have to try to cognitively assimilate
the information, it's not gonna serve you as well
as if the tasks necessary for you to intervene
on the populations to improve and or manage their care
are embedded in the most relevant interface.
This would be the electronic health record
in which those clinicians deliver care to the patients.
It will have the greatest impact on the population.
But when you look at what the next step is,
this is kind of our
ice cream cone of population health management IT.
That's what you called it Irena, wasn't it?
Ingest the data.
We talked about that, but we can't say this enough.
Aggregate it, match, link,
semantically normalize the information,
identify, so here's the data sources.
EMR claims, patients, Arcadia.
Arcadia's another data source
'cause we're gonna grow beyond it.
Aggregate it,
provider, condition,
analyze populations at risk, identify gaps in care.
Here's a diabetic population, a COPD population,
and from these gaps in care you can intervene.
You can create actions and intervene to close the gaps
in care through preventative care management tasks
and improve the outcomes.
This is the ultimate common pathway.
But you can't achieve this without the information up front.
Now, this is not an I chart.
It might look like one,
but it's where is the final common pathway?
Not the final common pathway,
but where is the next generation pathway for us?
So, we know we have Arcadia,
but Arcadia is purpose-built for our contracts,
our value-based contracts with Excellus.
We wanna preserve everything we've done with Arcadia,
but this is what we call our population health data pyramid.
I'm sorry it's a little small, but it's kind of
the ice cream cone upside down.
It's a pyramid.
If you flip that in your mind, be creative,
think outside the box, use the right side
of your brain rather than the left side.
I'm much better at that, okay?
Epic used to call this cogito,
which is a Latin word which means, I think, therefore I am.
Then Epic did not like that word.
Epic is very touchy-feely, so they called it Star,
because it's this star in the sky.
Then Epic said, "Well, it's more than just
"a single star in the sky.
"It's everything about everything."
I haven't updated my slide because I don't like
using this word, but it's a touchy-feely word,
but what's everything about everything?
In Epic terms, that's the whole kit and caboodle.
Epic now calls
their population health data warehouse Caboodle.
(laughing)
I really don't like using that in a professional context,
but that's what we call it, okay?
This is their Caboodle data warehouse.
You can see
that at the base of the warehouse is the ingestion.
You can't
act on what you can't measure.
You have to ingest the data, aggregate it, normalize it,
co-mingle it in a data warehouse,
and then you can identify the gaps in care.
From those gaps in care, you can action.
We have two sides of this.
This is the community and this is URMC and affiliates.
For those of us that are using Epic,
we can directly derive all of these measurements
into the e-record EMR.
We can create registries which are groups of patients
based upon condition or procedures with inclusion
and exclusion criteria from those registries,
like a diabetic registry, or an AFib registry,
or a stroke registry.
We can identify gaps in cares from those registries.
Then the gaps in care,
we present them in dashboards within Epic.
Those dashboards can be directly action.
It's really neat.
You can actually see a population of diabetics
with hemoglobin A1c is greater than nine,
who haven't seen their PCP in six months.
In a mass action, you can send them all a communication
through their preferred communication methodology,
mail, MyChart, or fax,
and then you can schedule
an appointment for them all to come back.
That's why it has the greatest value if that data
is actually embedded in the EMR
with which they care for patients.
Then the patients can connect back
with the providers from MyChart.
But not everyone is on Epic.
For those that aren't on Epic, the same dashboards
that show the information can be presented
to the community base practitioners.
Although it's not directly embedded
in their electronic health records, their data managers,
these little red people in the community,
they can put those actions in the EMR's of the providers
in the community, then they can action on them
in their EMR, and then we can extract the data
from those EMR's.
Remember, we have to be vendor gnostic
so we can attract the data from those EMR's
through the pipes that have already been built
through Arcadia.
Yes?
- [Audience Member] She's got a question.
- Oh, I'm sorry, yes?
- [Audience Member] I realize some of these steps
you're never gonna be able to get rid of,
but you're talking about the benefits of basically
simplifying the EMR platforms.
- Mmhm.
- [Audience Member] Are there specific efforts aimed
at that, trying to-
- Like what you do here?
- [Audience Member] Yes, to move them
towards a single platform.
- Well, no, if everyone was on the same platform
and Judy Faulkner,
who's the sole proprietor and CEO of Epic,
would like to believe that everyone will be that way,
but it won't be.
In the future as the interoperability increases
and becomes more usable, I would actually say 10 years
from now, these components might be modular.
They might all talk to one another,
but we're not there yet.
So, the answer's yes, interoperability is trying
to get more robust to try to exchange information.
We are ingesting data from these EMR's
into our data warehouse electronically.
The question is, can these actions, can these gaps in cares,
the analytics, can the gaps in care
in the populations at risk be pushed into those EMR's?
Today we push them in with data managers,
but tomorrow we might actually be able to push actionable
tasks based upon the gaps in care and the population's risk
into those EMR'S.
Efforts are being made in that direction.
We're probably just on the cusp.
'Cause remember, population health management,
value-based care, is brand new.
You saw from Irena in 2013 it went from 8%
to 40% in 2017.
The technology is evolving at the same time
that the need for using the technology's evolving.
It's moving in that direction.
- [Audience Member] Does...
Create incentives to use certain EMR's...
- Well, AHP right now says in order to be part of AHP,
and you're a primary care physician,
you have to have an EMR.
Now, it just so happens our community
is relatively homogenous.
We have Epic in the major health systems,
both in major health systems, RRH and URMC.
Most of the primary care community has Med-End,
but it's not exclusive.
In our primary care community, right now there's 52
connected practices and there's 11 different EMR's.
We wanna try to not force them onto one platform,
but for new practices that aren't on an EMR,
we try to incentivize them either
to come up on Epic or Med-End.
So we have a little bit more homogeneity,
and that does make a lot of sense.
Then we ingest all that information back and,
once again, you have your virtuous cycle.
That's a little bit about the IT foundation.
That was a great question, 'cause you're right,
if it's not there embedded in the tool you're using,
you're not going to achieve the greatest value proposition
in terms of intervening on the population
and bending that curve.
So, back to my partner in crime.
Irena?
- Thank you.
Just to give you an example of how we use the data
and what we do and how we change
the way we
think about
population health and episodic care,
I want to talk about bundled care,
bundled payments for care improvement.
I have to say that those slides have been vetted by CMS.
Everything I present is my own opinion,
has nothing to do with CMS.
But they looked at our slides and they approve them,
so I have to tell you that.
It's very similar, the concept at least,
when we talk about bundle or risk contract.
In bundle,
basically
the provider is getting one payment
for all the care that is provided
from time of, depends on what triggers the bundle,
but in our situation from time of admission
until 90 days post discharge.
We had the opportunity of considering different time periods
and we took 90 days.
Everything that happens to the patient
from the time of admission, including what happens
during the hospitalization, until 90 days post discharge,
is the responsibility of the provider.
In a sense, we get one
payment for this entire time period.
If there is a complication, if the patient is readmitted,
if we choose a
level of care, whatever is being done
with the patient, there are, by the way,
for almost every condition 6,000 exclusions,
but other than that,
whatever is being done to the patient
is the responsibility of the provider.
Now, the big issue, and you'll see the data,
the big issue is really what happens
if something goes wrong where who bears the risk?
The full risk here is on the provider,
which is very similar to risk-based contracts
with payers, many times, although it's shared here.
There is a budget, and if we, as the providers,
perform the low budget, we get to share the gains.
If we perform above the budget number,
then we share the risk.
With CMS,
it's very much the same.
In fact, we have a bank account
that we set up with money in there that we can pay CMS.
We need to show them that we have the money there
so we can pay CMS in case there is a loss.
If there is a gain, every quarter there is settlement
and we get money from CMS for performing well
and performing below the budget.
And everything is included, by the way,
across the continuum of care,
regardless whether the care is provided by us,
by our system, or anyone else in the country,
as long as it's within those 90 days, it's all covered.
You can think about it as, I just heard in Albany
they have a plan that is called the warranty program,
similar idea.
Yes?
- [Audience Member] You could share the...
- Payers, yeah.
This is a contract.
I'm sorry, let me clarify.
This is a contract with CMS.
It's voluntary.
We decided that we're going to choose a couple episodes
that we are going to contract for.
It's only for the CMS fee-for-service population.
If they get any of these admissions into our system,
and I'll show you that we chose major joins and CHF,
then they are eligible to be part of a bundle.
They will be bundled by CMS.
This contract for bundled payments is only with CMS.
Yeah?
Right, it's all the services that CMS is built for,
yeah.
Medicare, yes.
Right, it's all Medicare fee-for-service,
and all the services included
within Medicare fee-for-service, yes.
Yeah?
Car accident.
If you were admitted for CHF
and then you have a car accident,
but honestly,
within the 90 days, then it's not included.
But honestly, the list of exclusions is very far,
really, a lot is included in it.
Unfortunately, no, with CMS it's non-negotiable.
Yes?
- [Audience Member] What portion of your patients
have Medicare fee-for-service?
- Good question.
In our community,
we are very heavily Medicare-advantaged.
That's one of the issues we have.
We don't have a big population.
That's an issue with smaller numbers.
That's how, actually, we chose,
that's a good segue to the next slide,
because that's how we chose the episodes
we wanted to take risk for.
We looked at two things as a first step.
What's the volume?
Do we have enough volume?
And, is there variation?
I'll show you the numbers, or a chart,
so you can actually see that.
Then we kind of wanted to know, is it worth it for us?
Because we have such a minimal penetration
of Medicare fee-for-service, or really higher penetration
of Medicare Advantage programs,
it was very hard to choose that.
We were limited by our options
because of the smaller numbers.
Then we wanted to see, or we wanted to choose
by really clinical leadership and clinical vision criteria.
Can we do it?
Do we have the clinical support?
Do we have someone who would lead the vision,
change clinical pathways, and can lead the team
and be successful in that?
We had the little RFP process,
internal RFP process going on,
and we ended up choosing CHF and major joint replacement.
When you look at
the graph of, this is over two years' period.
This is major joint replacement.
Each line here,
the full line represents one episodes.
There are overall 257 lines because we had 257 episodes
during that time frame.
In the blue, you see there is no variation here.
The blue is the DRG.
It's the initial admission.
That's what triggers the episode of care.
The yellow is home health.
The real big variation
is nursing home rehab,
SNF rehab.
At the beginning, we had very high percent of patients
going to SNF rehab.
That's why we identified this as an opportunity.
Very...
That's the cost.
The y-axis is the cost with just no numbers here.
- [Audience Member] Episode is-
- Yes, each episode is a patient, you're right, yeah.
This is the cost, so it's increasing,
and obviously higher nursing home stay,
higher increase in cost.
- [Audience Member] The way I read this is
the blue on the bottom, every single person
has a major joint replacement and you're paying for DRG.
- Yes.
- [Audience Member] Then some of these people
have yellow bars.
- Home health.
- [Audience Member] Right, and then some of these people-
- Some of them get readmitted, you see the red, yeah?
And some of them use SNF rehab.
It's one on top.
These are all cumulative cost.
Okay?
Yes, Steve?
- [Steve] The DRG is fixed,
but isn't there variation in cost within there?
- Yep.
- [Steve] Some DRG cost a different price.
Has that been already adjusted in this?
- No.
This is how CMS is seeing it.
This is claims data, everything that is billed to CMS.
CMS is the insurer's cost.
There is an opportunity within the DRG
that we also wanted to minimize,
to reduce the variation there,
which we were successful in doing,
but it's not reflected here.
Yes?
Yes.
What do you mean by processes?
- [Audience Member] I think that should be in there.
- That's within the DRG, yes.
Yep.
Other questions?
So, yes, Beth?
- [Beth] Of the multiple co-morbidities
for each of the cases...
That the variation in cost?
- We do, this is not risk adjusted.
This is just cost per episode.
Yep.
That's another, you just identified both issues
that we have with CHF, but yeah.
(chuckles)
Yep.
Here is the CHF picture, very different, right?
We have the initial admission,
and then the biggest variation and the biggest issue
that we identified is the readmissions here, the red lines.
Again, big variation in cost, big driver is readmissions.
That's what we were targeting within each
of those two episodes, very different things
based on what we saw in the data.
Within major joints, we really focused
on enhanced home care services
as opposed to SNF rehab.
That took a lot of education on both the provider's side
as well as the patient's side.
A lot of education, a lot of communication as well.
For the CHF, what we wanted to do is reduce readmission,
again, through advanced home care and rapid responses.
What we were able to do
is reduce the number of readmissions.
For both programs, we needed to establish
very clear clinical pathways,
make sure that we follow them,
and what I didn't tell you is that CHF is done
in both hospitals, Strong and Highland,
and for major joints, we only have Highland at risk.
Then we created a new
position in that is called Care Navigator.
The purpose or the
objective of this person is to follow the patient
throughout the system until the end of the episode.
She basically would introduce herself to the patient
and say, "Hi, my name is so-and-so
"and I'm your care navigator for the next three months."
They keep in touch with the patient.
They follow the patient.
They make sure that they coordinate their services
and they know where they are at any given moment,
even if they're in a nursing home.
Let me leave a couple minutes for questions.
I think we talked about most of these.
Dashboards to show providers and really help providers
track the patients as well, understand what is going on.
Here are some findings,
just so you see how we are doing on these.
For major joint replacement, we started with 74%
of the patients at baseline going to SNF for rehab.
We ended up with 15% during the risk period
and then home care increased from 22% to 74%.
At the same time, we were able to decrease ED visits
from 33% to 12%.
Pretty amazing results.
Yes?
- [Audience Member] The readmission...
- Yes.
This is within 90 days,
and it counts multiple readmissions if they are.
It includes everything within 90 days.
You're right, though, most of the readmissions
do happen within the first 30 days.
There's much less after that, but they still happen.
We look at every...
Just so you understand, it takes a village, okay?
We have a team of providers and nurses
and in-home care
people that really, also spectrum of providers,
and finance and analysts, and we all sit together
and trying to understand what happened this quarter.
We look at the data and we try to understand
what happened with the cost, what happened to the quality,
why are we seeing what we're seeing?
All of that is happening on a regular basis
to manage beyond
the day-to-day operations.
Yes?
- [Audience Member] Do you see variation...
I have really two questions.
Within those...
End up at a certain SNF,
are they more likely to have certain costs?
- Yes.
- [Audience Member] Then also, when they're going to SNF's,
are you writing them with certain choices because of that?
Then also, converting them to home care...
A lot more control...
- Yes, and I'll show you.
Okay, so there were lots of questions there.
We have...
SNF partners and we chose those partners,
it's a handful of partners, and we chose them based
on those SNF's that have higher quality,
can control variation, and really provide care
at somewhat more reasonable cost.
We had conversations with them.
We're working with our nursing home partners.
It's a small number, it's not everyone.
We do suggest
patients a list of,
we do provide them a list with our nursing home partners.
Then, with respect to home care,
I have a slide.
We reduced our costs by 33% for doing just that.
This doesn't include, Steve, I think you asked,
this doesn't include the internal cost savings.
For home care, this is within CHF.
Look at this difference.
If we send patient home without any home care,
this is after a CHF, the readmission rate was 47%.
If they had home care, but not by us, not U of R home care,
the readmission rate was 43%.
If they went to UR home care, or some of you know this
as BNS, the readmission rate was 18%.
Clearly, the collaboration, working together,
and making sure the communication is there was key
to being successful in that.
Just a couple lessons learned.
Obviously, the importance of data,
the importance of leadership in moving this along
and standing behind those programs.
Time requirements, obviously, as we think
about new programs and new things that we want to roll out.
And I think the last, most important,
we call it externality, it's a positive externality here
because what happened was those conversations
generated really a lot of interesting information and
conversations about other populations
that we may apply these clinical pathways to.
So with that, any questions or comments for both of us?
Yes?
- [Audience Member] This sounds like it went...
Nursing homes...
(chuckling)
- Great question.
I happen to sit on the board of the ju-es-en-ior-life.
It was a big concern, what will happen to their revenues
if they don't get basically those patients
after major joint?
The patient population of nursing homes is changing because
providers now want to take, as we take more risk
for populations, we really want to send only
the most complex patients that need the nursing home
to the nursing home, and otherwise try to find other rehab
or ways to take care of those patients outside the hospital.
We see this, at least locally, in the nursing home
also changing what they focus on,
and they now also try to provide different specialized
program to address that specific change
in what is going on in the rest of the market.
Yeah?
- [Audience Member] This is around these episodes of care.
If you're saying the Medicaid redefined model,
they have episodes of care,
they have complex, multiple specialty,
and then they have advanced primary care.
Have you started any work around that,
the advanced primary care,
and letting behavioral health into primary care?
Are we moving in that direction?
Data...
- Yes, and yes.
We have our district program.
We are part of the fli-ps regional program,
and we do all of that there.
Actually, when we talk about population,
and you saw Medicaid there, so it's part
of the state program as well
as we have a contract with Excellus,
pointing out,
to provide Medicare managed care
or care for that population.
We have all of these definitely embedded
as part of what we're doing.
Yeah?
- [Audience Member] We have time for one last question.
- Go ahead.
- [Audience Member] So expensive...
- Yeah, we're working with our clinicians
to identify that.
Someone already asked this question about CHF
and the complexity of the patients.
We are not, just full disclosure, for the CMS
fee-for-service population,
the Medicare fee-for-service population,
we were not as successful as we hope to be
because our population now,
the at-risk period, is much more complex
than our baseline population and we have smaller numbers.
That really is not in our favor.
However, we are working with our insurers
to include Medicare Advantage population in the mix
and provide bundled services for them
because it's really successful in terms of quality,
and we hope that cost would follow as well.
Thank you.
(clapping)
Hey, good to see you.
How are you doing?
Hi.
- [Audience Member] Nice to meet you.
- Nice to meet you, great questions.
- [Audience Member] Thank you.
Very interesting pres-
(audience chatting)