[This Transcript is Unedited]
DEPARTMENT OF HEALTH AND HUMAN SERVICES
NATIONAL COMMITTEE ON VITAL AND HEALTH STATISTICS
SUBCOMMITTEE ON QUALITY
THE MEANINGFUL MEASURE SUPPLY CHAIN –
BUILDING MEASURES THAT MATTER FOR OUR NATION’S HEALTH
October 14, 2009
National Center for Health Statistics
3311 Toledo Road, Auditorium A
Hyattsville, MD 20782
CASET Associates, Ltd.
Fairfax, Virginia 22030
Table of Contents
- Welcome – Justine Carr, Paul Tang, Carolyn Clancy
- Meaningful Measures of disparities
- Ernie Moy
- Kalahn Taylor-Clark
- Meaningful Measures of Value (including efficiency)
- Joachim Roski
- Michael Rapp
- Meaningful Measures of Integration, Population and Health and Health Issues
- Linda Harris
- Floyd Eisenberg
- Summary, Discussion and Next Steps
P R O C E E D I N G S
DR. CARR: Welcome to the second day of the NCVHS Quality Subcommittee
Hearing on Meaningful Measures. Yesterday we talked about the meaningful
measures supply chain and building measures that matter. Today, we are going to
continue that discussion but in light of national priorities we ended the day
yesterday with care coordination. Today, we are going to be talking about value
and efficiency, population, and disparities.
What I would like to do then is start by going around the room. I am Dr.
Justine Carr, Co-Chair of the Quality Subcommittee, Member of the Full
Committee and from Caritas Christi Health Care and I have no conflicts.
DR. TANG: Paul Tang, Palo Alto Medical Foundation, Member of the
Subcommittee and the Full Committee, no conflicts.
MR. QUINN: Matt Quinn, Agency for Health Care Research and Quality, Health
IT Group and I am Staff to the Quality Subcommittee.
MS. JACKSON: Debbie Jackson, NCHS, Committee Staff for NCVHS.
MR. MOY: Ernie Moy, AHRQ, Center for Quality Improvement and Patient
DR. CLANCY: Carolyn Clancy, AHRQ, I work with Ernie and Matt.
DR. FITZMAURICE: Michael Fitzmaurice, Agency for Health Research and
Quality, Liaison to the Full Committee and Staff to the Subcommittee on
DR. GREEN: Larry Green, University of Colorado, Member of the Committee,
Member of the Subcommittee, no conflicts.
DR. SCANLON: Bill Scanlon, Health Policy R&D, Member of the Committee
and the Subcommittee, no conflicts.
MR. REYNOLDS: Harry Reynolds, Blue Cross Blue Shield, North Carolina, Chair
of the Full Committee, Visitor to this Committee and I have no conflicts.
MS. KANAAN: Susan Kanaan, writer for the Committee.
MR. CARLTON: Tom Carlton, Blue Cross Blue Shield, North Carolina, Director
of Information Strategy.
MS. VIOLA: Allison Viola, American Health Information Management
MS. HOLMES: Julia Holmes, NCHS, I work in the Division of Biostatistics.
MR. ROSKI: Joachim Roski, Brookings Institution.
MS. CHRISTIANI: Jeannine Christiani, Contractor for the Committee.
DR. CARR: Paul and I, wanted to start off today with a little bit of a
recap of some of what we heard yesterday. We stated out the day hearing from
Helen Burstin and about work that is being done in the National Quality Forum,
with a focus on shifts to composite measures, measuring disparities in all that
we do, harmonizing measures across sites and providers, promoting shared
accountability and measurement across patient focused episodes of care, and
focusing on outcome measures, appropriate measures, and cost and resource use
measure, coupled with quality measures.
We also heard interesting work by the American Board of Internal Medicine
or American Boards. I think what was interesting about that was the work that
they are doing of having providers look at a population and look to see whether
they have achieved all of the appropriate care delivery elements. I think what
was so impressive was the engagement that came about because of that and
finding gaps and then addressing them. It weighs the issue about engaging
providers in these measures, having them be meaningful to the providers, and
also I think as Frank Opelka said, making them actionable so that when you look
at them you can deal with them.
Another point I think that came out yesterday also, was in terms of
actionable, to not just have the element, but to have the capacity to drill
down and stratify the groups and subpopulations. We also heard about in the
PCPI Model, integrating quality measures into EHR and focusing on the
timeliness of the data, having it available, also the incubator groups and
working through things.
I think that several questions emerged in our conversation at the end of
the day. I think one was is there a strategist for measurement? Who does that,
who holds that role? We heard actually many excellent presentations from
different groups but they were not the same, they were different approaches to
the same thing. The question is, how are we going to reconcile all good
approaches but not the same because without that standardization the vendors
cannot make EHRs that approach these. We know that NQF is updating the
endorsement criteria and measure developers are developing measures. I think
what we were wondering about is are these new measures, are they a retool of
old measures, therefore, the CHIM I think was very interesting. I think that
Matt brought up a great point at the end of the day. What about the world of
social networks because we are trying – we are gathering data and do we
get data from that.
I think a lot of the conversations came out at the end of the day when we
heard about work that is going on about coordination of care from NCQA. I think
it raised the question of what is a publically reported measure and what is
performance improvement. Performance improvement being done locally and public
measure being a simpler thing that is outcomes oriented, that engages and
raises awareness that is an important element. It is something that triggers a
response for improvement but is simple and compelling.
We had concerns about the prescriptiveness of what we heard yesterday
because it sounded more like QI and it sounded like it was going to take a long
time to get the multiplicity of measures that were being put forward.
The two other things, who drives and enforces standardization? That
includes not just the measures but for the risk adjustment algorithms. Then
also, as we heard more and more we really have been implicit in all of this
measured development – is the need for a data aggregator that even EHRs with
good registries may not have all of the data elements that are needed to create
the denominator or to risk adjust or to identify and exclude.
DR. TANG: So maybe one of the things that we tried to examine yesterday was
what are the attributes of a meaningful measure? Some of the characteristics
that I think that would drive our health system forward rather than backwards
is if the measure were inspiring, engaging, and actionable as Justine said.
Those are things that need to pull the health professional team forward as well
as the patient and be understandable. What we do not need, and I think that
some of this is tethered in the days when we only had administrative data are
measures that are prescriptive, that are burdensome, and that add cost to the
system in a sense. It has been said, no money, no mission. In an era of health
reform where we want to have an option other than paying for volume it may be
no measure, no mission, and perhaps no money.
I think that we really want – as we look forward to today
particularly, Carolyn’s remarks, in what is the vision for the future of
measurement and its role in driving and not just retrospectively reporting on
the health care that we provide. So it is with great pleasure that I introduce
Carolyn, I think to say that she is the head of the AHRQ would be an
understatement of her accomplishments. To say that she is a champion of
research and quality would equally be an understatement. I think she is a
champion and leader period of all things related to health and health care.
Probably the thing that most aptly characterizes her is when Mike Fitzmaurice
said yesterday that, when the Secretary announced that she was appointed to the
Head of the AHRQ during her administration and you had to tie people in their
seats to stop the applause. I think that is the indication of the respect and
adoration that folks not only in her agency but around the country have for
her. So it is with great pleasure that I introduce Carolyn Clancy.
DR. CLANCY: Good Morning everyone. First of all I have to say thank you for
all of the fan notes about Matt’s work with this particular hearing. We
know he is terrific but it is always lovely to hear and you really did put
together a fabulous agenda. Between Matt’s wrap up and your key points
just now I think that I have got a good flavor of what I feel a little bit
deprived about missing especially, since I spent a good chunk of yesterday
sitting like a sardine in an airplane.
However, if there was one take home message that I have is to be one of
excitement and humility; excitement because of the possibilities ahead,
humility because it is going to make us confront all of the weakness of what we
have been doing in assessing and improving quality. Peter Pronovost has
probably done more writing on this in a mainstream medical way than anyone that
I can think of. This issue of data aggregation and enforcing standards and all
of that, we do not do that at all. I mean we have got the threat or perceived
threat of the occasional audit. It is not even clear to me that the auditors
actually know what they are doing. I mean if we were to really get extremely
concrete about it a lot of what we are seeing and reporting is self reporting.
Now with that said it is a very powerful tool and it is probably one thing that
there is actually bipartisan agreement. I do not see this actually going away.
I am really thrilled to have an opportunity to share some thoughts.
I think the other message I would leave you with is it is really hard to
transition from a world where we have always asked the right questions. What is
important? What is the greatest burden of disease? Where can clinical care make
the most difference and so forth? Then it gets to that final funnel of what can
we measure. Feasibility drives all existing measurement activities. It is very
hard to transition from that to saying, wow, we can see right over the horizon
a health care world where data will be ubiquitous and the question we ought to
be focusing on right now, if not before this is what do we want to measure. My
own personal hope is the phase of retooling the existing measures or retooling
EHRs to capture some imperfect measures has assured us that we can make it
although, I think it is inevitable that it will be there.
I am going to hit on just a couple of key things here given the time and
really the incredible opportunity for conversation I am not going to spend a
lot of time on each slide in case any of you are nervous.
Obviously, a lot of very, very important activities, I am sure that you
have heard about the national priorities partnership from Helen Burstin
yesterday which although it is still 15,000 feet above the ground, we are down
from 50,000 feet above the ground. It has been an exhilarating experience to
participate in that, to really step back and say okay, what makes a lot of
sense here. The reason I think that is so important is what I just heard from
the two of you actually has an adherent conflict. Paul just described what I as
a clinician, believe in my heart of hearts, you give me information that makes
me say thank you and makes my job easier. Wow, that is fantastic. I have heard
clinician friends say that to me.
At the same time, many stakeholders think that there is actually quite a
bit of good to the just pure old accountability fashion and making sure that
people are actually checking the boxes. That is part of a democracy so most of
our efforts in getting to quality measures, approving them, and endorsing them,
and so forth have been a very large exercise in small democracy. In some cases,
we have really let go of what is very important. If you think about it what
makes clinical medicine intensely interesting is not saying, being a hamster on
a treadmill where you do exactly the same thing every single day. It actually
is taking scientific knowledge and tailoring and customizing that for the
unique needs of an individual patient. Most of our measurement enterprise does
not reflect that yet.
I was saying to Justine earlier, I think that some of the leading edge
institutions in terms of improving quality are starting to get that. I heard
someone tell me yesterday for discharge instructions, what many people actually
do is – or what a number of institutions have figured out to do is
everyone gets the basics and then they ask for a read back. Those people who
clearly get it and are clicking and do not need the extra supportive services
and so on and so forth may have less complex illness, they say, goodbye, good
luck and we hopefully see you elsewhere but not here. Other people, they use
that as almost a screening tool. I do not think we have been very smart about
that at all. Of course, we all want outcomes and so on and so forth. We have
given very little attention to either in the development of guidelines or on
measures thinking about workflow and I think that is very critical.
We do not know actually very much about the methods to do this and when we
get to the efficiency one of those e’s in the six dimensions of quality,
we know very little. By the way, for health reform, what everyone wants is high
quality, affordable care. They are not looking at us just continuing to spend a
whole lot of money and you know, measuring and making better report cards. The
question is when are these external, transparent comparisons helping our cause
and when are they not? I sometimes think that we have gotten so granular in
clinical quality measures NQA was endorsed over 500 I think. Some of that has
to do with the fact that there are so many physician specialties that it feels
to me again, like when the pilot gives you too much information. I do not
actually want to know what is the problem today. What I want to know is are we
taking off or not and do I need to make alternative plans. Clearly, I think
that there is a huge interest in making this information actionable. This is
from a slide about efficiency measures which I think is a huge, huge gap in the
landscape. I think that what we are very good at with the proprietary tools as
far as I can see, is actually pinpointing that sick patients cost more. Wow, a
pretty profound, I will leave it at that.
Frankly, I would say that we were much more excited about our efforts in
improvement. I pointed out the Keystone Project and NSQIP here because that
actually is closer to the vision that Paul just described. When people collect
enough data that is actionable and helps guide their efforts. But that is the
fundamental purpose for collecting the data. NSQIP started off in VA and we
actually gave the College of Surgeons a grant to expand it to a couple of
hundred of civilian hospitals. We are very proud of the CAHPS surveys but
sometimes it feels as if we are going to spin it off as an independent company.
Ultimately, we have got huge challenges vis-à-vis attribution, right? On
the one hand we are busily developing measures to assess the efforts of
individual members of the team and yet what gets people really excited is the
curriculum that promotes good teamwork, a skill that is terribly new to most
health care professions.
This little picture here is intended to be from the slide about the fact
that there are still institutions and organizations purchasing electronic
health records that think that somewhere in that box of instructions and all of
these computer guys who now live there full-time, is the instructions on when
you hit F7 to upload your quality measures. We have electronic health records
that support a transaction-based system that rewards volume rather than
quality. It is not remotely surprising that registries are an exception rather
than the rule and that the ability to actually give us any information about
like groups of patients is primitive as it is. It does not mean that we are not
there and we are very excited about some of the work that the Health IT expert
panel that Paul chaired that actually led us to begin to pave a roadmap from
our imperfect world today to a better place in the future.
Obviously, this is complicated stuff. I think the one bright promising
start here is the Recovery Act which gives us a huge amount of resources, both
in comparative effectiveness and in Health IT. I see, and we have gotten very
clear feedback from Congress, it is not intended as a job function but really
is intended as a down payment on the scientific infrastructure that we are
going to need to make health reform sustainable.
We know that Health IT could make a whole lot of this easier and sooner or
later, I must have heard this from Larry Green at some point, because it feels
like something that he would have said, sooner or later we are going to back
into the question of why do we collect data and what data do we collect as a
part of routine care makes sense anyway? It is an issue that in my view has
gotten very little study. Don Berwick once pointed out to our Advisory Council
that if you were to look at hospital charts today, the structure in terms of
how we record information has not changed in at least 50 years, the antibiotic
names are all different and so forth and more gizmos and technology that we are
dutifully recording numbers from, but the actual process of thinking about what
information do we need to provide care for a patient? What is the purpose
besides billing? It has not gotten any systematic examination at all.
We have been very pleased to be funding some of this work through our
Health IT portfolio and also through one of our search projects. Again, this is
largely retooling existing measures and trying to figure out how easy or
challenging it is to do that.
One very exciting area gets to the issue of data aggregation that one of
you raised. Two years ago, we actually funded some prototype distributive data
networks, at least one of which as I understand, has just gotten a boatload of
money from Recovery Act funding from NIH and I think that Larry Green has been
very connective to the DARTNet Project.
One of the strategic issues which we have not addressed in discussions
about Health IT, quality, or anything else, is do we imagine that data
aggregation in the future is going to be a centralized repository model or that
we will have regional aggregators with common standards that are audible and
all of that kind of stuff for those aspects of care that we do want to be
This is actually groundbreaking work in terms of figuring out how
challenging it is to develop common definitions and infrastructure as we are
focusing on trying to make electronic health records more useful for conducting
comparative effectiveness research. My understanding of why the clinicians in
DARTNet, this is a practice network of relatively small practices working with
some hospitals in Colorado, why they participate, what they get back. I think
that is going to be a very critical piece of this that we do not give enough
attention to at all. What do I as a clinician get out of this? What is the
incentive for participation? What they did was quality benchmarks which is very
helpful to them to know how am I doing and where can we actually improve our
efforts. Many people, including Farza(?), the Deputy Director at ONC, believe
that distributive data networks are definitely the future. I can tell you right
now, having done a request for information on a data stewardship strategy a
couple of years ago we have got almost no insightful comments; we got lots and
lots of comments about privacy, a ton of them. That was about 70 to 80 percent.
So we cannot let go of that in terms of making sure that that is factored into
the strategy as well.
One of the things that we are very attentive to at AHRQ these days is the
fact that it gets very easy to focus on data all of the time rather than
information that you want. In fact, I would have to say, I think this is an
area where the federal government excels. We are absolutely fantastic at
building databases that you can send us data to and if you are lucky we will
let you know that we received your submission but do not necessarily do
anything for you. I think right now we are hearing hospitals rebel a little bit
about this for health care associated infections, this is actually a fabulous
research database and for hospitals with a lot of infectious control folks it
is terrific. For small community hospitals it is kind of not working for them
and it is much more intense than they need to guide efforts to improve it. That
kind of interaction and figuring out what is the incentive for collecting and
reporting data I think may need a whole lot more attention.
One of the areas that we have been very excited and actually have some
clear resources is looking at CHIP and Medicaid. For a variety of reasons, all
of which have something to do with the fact that CMS has been able to use QIO
resources for much of what has been done in quality measure development and
they have done a fantastic job. But you cannot use that for people who are not
in the Medicare – you cannot use it for any work that is not potentially
relevant to Medicare.
Kids have kind of been left behind. So the CHIP Reauthorization Act lays
out a very thoughtful quality roadmap which effectively says, the Secretary
will release a list of measures to be voluntarily reported by states in January
2010. Not a lot of time to do this in 11 months. Not a lot of time for the
usual process. This is not going to NQF; this is leaning very heavily given the
economic climate that states find themselves in on measures that are already in
use. Because we have had an interesting transition here in terms of people
coming into leadership positions recently have stepped up to say, you know we
have got a terrific advisory council and we could create a separate advisory
group much like the AHIC used to do to focus on this. We have had lively,
intense multiple stakeholder discussions about, why don’t we blow this up
and start with much better measures on the one hand and states on the other
hand telling us you know, last year my quality department had 20 people, this
year I have 4 and I do not know if I get to keep them so I have got to lean on
what I am doing already.
The good news is after that effort there are actually clear resources
identified in the bill for moving forward, both for developing better measures
and for developing a pediatric electronic health record, whatever that is
exactly. I am here to tell you it is not a smaller one in difference to people
whose main passion in life is providing and improving child health care.
I think that there is a really important opportunity for synergy there so
your efforts could not be coming at a better time. A lot of money and
comparative effectiveness, the only details that I would call to your attention
is the Secretary’s money gives us a lot of opportunity to focus on
infrastructure that we will not ever again have an opportunity to make those
investments. So big investments coming and I cannot give you much detail now
because we are still in that intense multiple conversations with the OMB
period, but big investments coming in infrastructure, all of which I think or
much of which can actually be focused on assessing and improving quality of
care as well.
The Institute of Medicine gave good guidance here as well. It also focused
on some infrastructure needs as well as a list of the top 100 specific
questions that they felt were very helpful. Notice the word prospective
registry here? Clearly that is relevant both to conducting research and also to
assess quality of care. I think that what we have got to figure out is how can
electronic health records through the meaningful use incentives actually
pre-populate some of these registries? We have seen these be incredibly
successful are from those organizations that have the resources to pay a
dedicated data collector, you know it is a standalone effort but I think it has
got to be a much more organic connection with the delivery of health care.
Ultimately, what we are trying to get to I think is an information-rich
patient focus health care system. The glimpses of this that I have seen at some
organizations that I think have begun to take these challenges on very
seriously are that it literally transforms interactions from reactive, as in
what are you doing here today, in so many words; to an interaction that is
proactive, where the clinician and the patient start this conversation based on
how they have been doing, what preventive care they need, that kind of thing.
It is not that doctors do not aspire to do this. It is just finding that
information is painfully difficult because for those of you who are clinicians,
I am sure you have had many of the same conversations I have had with patients
over the years. I would just like to hear about this in your own words again,
to make sure I have a clear understanding of the problem, which really means we
cannot find the chart again. Clearly at AHRQ, we believe, and you will hear
much more about this from Ernie and Kalahn, that quality and disparity have to
be linked very tightly and that is also a strong focus for the investments we
are going to be making on behalf of the Secretary for comparative
We are talking local data here now, okay? We are not talking a huge
centralized repository because it is a given both from what we know about
disparities as well as what we see in the Dartmouth Atlas that the solutions
are likely to have to be fairly customized to particular communities. That does
not mean that there is not generalizable knowledge. I think it stands to reason
that, and the Mayo Clinic can tell us what works well in Rochester, Minnesota
does not necessarily export easily unmodified to Jacksonville or Scottsdale. I
think that we are going to see that from many aspects of quality improvement.
Clearly, with the reform momentum that we are seeing, with the opportunities in
CHIP, your timing could not be better. I really do want to say thank you.
The good news is we have got a lot of opportunities to improve which I
guess on some level means that it is hard to make a mistake. We have got a lot
of opportunities to connect the dots. If there is one area that makes me humble
it is that we rely a lot on Paul Tang and Janet Corrigan to be bilingual in the
language of Health IT and quality assessment and improvement. They need a lot
more company. Those two communities by and large do not overlap a lot.
I think that the meaningful use incentives will actually make some of that
happen kind of logically. A lot of organizations now are hip-deep in electronic
health record implementation but I will tell you what, one huge challenge with
electronic health records is still going to be as far as I can tell, there is
not a killer app that makes any clinician’s day easier because they are
using an electronic health record. You can say it is professional confidence
and I believe that with all of my heart but boy, it would be lovely to make it
fun as well.
I am thrilled that you are thinking about social networking. You know the
importance of a roadmap which is what I heard from both of your comments, I
think is self-evident. I will stop and I would like to just sit down so we can
have some back and forth if that is okay.
DR. CARR: Thank you so much Carolyn.
DR. TANG: One of the things that we forgot to mention about Carolyn is her
tremendous eloquence and intelligence, sheer intelligence. I always enjoy
listening to you. Some questions maybe about maybe the future and thanks so
much for your presentation and guidance.
We heard a lot yesterday about more measures about components and processes
and we sort of took a pause and said, look as medicine gets more complex, maybe
paradoxically, measures should get simpler and maybe even both the who is
queried and who benefits is different. So should we just outright ask the
outcome? Justine suggested, can’t we just ask the patients, were the
discharge instructions clear rather than, were they printed, were they handed,
were the et cetera. Then maybe we will get closer and closer especially in
today’s world – because I think in general the health literacy and
the appreciation of what it takes to get something out of your encounter with
your health team is getting more and more understood by consumers. Not that it
is just missing but that we do not still have complete health literacy across
the entire population. Just because of the internet people are more engaged.
Perhaps asking patients more about how we are doing is our version of social
networking because they are used to rating, they are used to reviewing, they
are used to providing each other information. Maybe we can just actually learn
from that. That is one piece.
Another piece is you have talked about what is the killer app? I certainly
have felt that in our practice, my killer app is actually getting that data
back. So you mentioned the bidirectional. There is this huge sucking sound.
Data goes in, 18 months later something comes out and has something to do in
theory with what you are paid, you know, your pay performance. Nothing is
relevant. You give data and we give data back but this is on a quarterly basis,
plotted and transparently so everybody can see everybody’s data. That
really riles them up. It almost does that magic killer app; you get that killer
app sense when it is probably going to change the next diabetic you see today
because of your score this quarter in a sense.
So I wonder if we do help some of those keys, and someone yesterday was
saying, you know what, well, we do not have a lot of power. I think that as you
said the whole Recovery Act and the stimulus money indirectly and directly
means that the people who are designing the measures upon which we are going to
be judged have a lot of power, a lot of transformative power as well. To draw a
couple of your messages; one, is looking for a killer app and maybe the
bidirectional, the feedback, the timing feedback may be one of those killer
apps because that is just how physicians are. But anyway, comments on getting
simpler measures, fewer of them, more outcomes oriented, and from the patients.
DR. CLANCY: So I think those comments are really terrific. You know one
killer app that I hear lots of doctors asking for is actually figuring out
could they get information about which of their patients is having challenges
adhering to medications. Most of the time we actually do not know that in a
clinical encounter so what do we do? We go into the same usual collusion
routine where both sides are pretending that everything is fine. Then we say,
we give patients the same kind of little sermon, you know, it is really
important to take those medications. Okay, see you in a few weeks. It would be
really terrific to know who is having challenges or not doing this. Now this
gets sensitive. Maybe some people do not want to but again, it would be
something that helps me customize what I need to do and also helps me know
where I need more effort.
I actually have moments where I think the key; the magic to fixing health
care is actually two things. One is some strategy in place to systematically
look across an organization or a practice or whatever the unit is, I do not
know weekly, pick your timeframe to say, where did we screw up this week. Where
are we dropping the ball? Who did we refer that got lost or whatever the issues
are. We do very little about it.
My favorite example is radiologists I gather can buy a CD with mammograms
because we know that volume has a strong relationship to quality and skill
there. I have not actually found anyone yet who systematically looks at their
own performance in practice which I think makes a lot of sense. To the extent
that board certification is promoting that. I think it is great but it is not
just about reading a mammogram. A lot of this is about people get lost. That is
actually what causes the lawsuits or at least from what I read. I do not know
if I have got a good database for that as much as whether the film was read
correctly. Yet, we have got a system that actually has people reading these
things at home without every looking at whether their system is working for
The second thing, for me would be something about being responsive to
patients. Now, with that said, I think we have got to be a little bit more
activist about it because I think, I cannot be the only person who has
periodically heard or even received things that basically say, if you rate me
well as a mechanic or whatever, I mean I am not even talking about medical
care, I will be really pleased and it is really important to me to get
recognized or rated well. I do not think that you want to push people into a
situation where there is a clear imbalance of power and say, how are we doing?
Read back has a lot of appeal to me. I did not actually learn that in
school or training at all. I learned it when I worked at the free clinic and I
heard a pharmacist do it one night on the other side of our dividing wall. I
did not even have a name for it but I knew it was powerful. He downloaded the
instructions and all of this and then said to the patient, tell me what you
heard. Wow, again, I did not know what it was but I knew it was really big.
When we began using that in our clinic it made a big difference.
DR. CARR: Just to add to that I think one of the models in safety is to say
the percent of patients who left their hospitalization with no errors, no near
misses, that everything went perfectly and it was a sobering number.
DR. SCANLON: Thanks very much. I think you touched on a number of important
things. I guess I also for one would like to highlight a sense of urgency,
particularly with health reform. In many ways, what we are doing now is an
evolutionary process which may end up in a very good place but it is taking a
long time. Justine pointed out the problem, well what are vendors supposed to
do in the interim as we keep changing what we sort of want?
I wanted to talk about something that actually is being discussed at MEDPAC
and it is likely to lead to a MEDPAC recommendation next month. As part of a
congressionally mandated study to look at the issue of comparing Medicare Fee
for Service with Medicare Advantage Plan, the idea is what we should be doing
is we should be asking for more information from providers, through using
electronic health records and meaningful use. It is asking not for measures but
in some sense the building blocks for measures because if the measures are
things that we are going to keep changing. What really need to think about what
is it that we can use as the component of measures?
One thing that comes up and this is an old MEDPAC recommendation, is we
should be getting lab values. HRQ is showing that we had a hearing a number of
years ago where this makes a difference in terms of risk adjustment and
obviously in payment policy that is a critical thing. I guess the question here
is kind of how much should we be moving down that path because it seems like it
is a faster path and one that the vendor can deal with more readily and there
would be fewer adjustments that would need to be made over the short term. Over
the longer term we may get a whole different set of building blocks that we
want get incorporated into the information that we have.
DR. CLANCY: I think you are definitely on to something there. I do not know
to what extent the Health IT expert panel got discussed here at all but the
general gist was to say, of the 500 odd measures that have been endorsed can we
find a subset that we think are probably more important? You know, care for
heart attacks, we want that to be good. You know some other very narrow focus
the process measures are not quite as important. If you create that subset can
you identify data types or components as you would say, that are important. I
sure would not want to steal Paul’s thunder here, he did a spectacular job
of leading this but I think it is exactly the right approach. There are some
big training implications here as well. This is not part of how you are trained
at all. Doctors are trained with two themes in mind. I think nurses and
everyone else says pretty much the same thing. You do the very best you can
with each patient and you move on to the next one. The notion of looking back
just is not there at all. The other major thing for doctors is they are not all
that great at peer review because I was not there. This doctor may have seen
something important that was not described and captured and all of that. I
think the components are just right. I certainly did not mean to diminish
urgency. I was more trying to say that –
DR. CARR: Yes, I mean just in terms of the components, just thinking about
if we collected times of everything. You know the measures we could construct
of timeliness would be huge. I mean we do not need to know which measures those
are but if we had the times we could do that.
MR. REYNOLDS: Thank you, always outstanding. So as we hold this hearing, as
you know, we usually move forward with some kind of comments, recommendations,
or something. So you get to write it this morning. What are our focus areas? We
might not get a chance to talk to you again before we do this. Give me a
framework where you think we could make the most difference. What we could do,
just those things.
DR. CLANCY: I think we need more work on the level of attribution. There
are two things inform my saying that.
Number one is to look at disparities you really cannot do that for very
many conditions at the units where people get care. It is too small. In
hospitals, only a fraction can even report on stratifying measures even for
heart disease because you just do not have a big enough sample size. So we say,
yes, we want to do that and then we go away from it, we have not said, then we
will report at the state level. I think Massachusetts is doing some of this
which is really terrific that they can say to you confidently a proportion of
Hispanic patients who report that they have got a usual source of care has gone
up dramatically. We are very excited about that. I think that has got to be
part of the roadmap moving forward.
I do not know what the rest of the recommendation is. I have to think about
it but I am in touch with Justine on a fairly regular basis and would certainly
be happy to share something. I also have not had a chance to do brief with Matt
because we have been in different places.
MR. REYNOLDS: So I am going to ask you just a little differently then so,
yesterday and I am sure later today, there is as Justine and Paul noted there
are a lot of people working on these measures. Where are the clear indicators
because if you have 10 working on it and we need to move it to speed that it
has been mentioned at, where should the focus be at least in the short-term to
kind of get a –
DR. CLANCY: I am very pragmatic. What I would do frankly is look at the
current drafts of health reform bills and figure out what are some common
elements. Secretary Sebelius, when she visited last week said she herself was
humbled by the number of sections of these bills that start off, the Secretary
failed… Now my interpretation of that having had lots of conversations
with folks on the Hill is the Secretary shall often precede an area we do not
know what to do. We do not know what the ideal data collection strategy is but
you know what, it would sure be helpful to do that.
I think if you were to take a look of some of those components of bills
which in the town hall meetings and that kind of intense period feeling have
gotten less attention; it is pretty impressive what is there. I would be using
that as a map for how this committee can help and know potentially specific
elements that are not there. That is okay for a bill but it will need attention
by HHS because someone is going to call this party. When you asked about who is
the strategist, what I have noticed in recent years with absolutely no
resources is everyone thinks that they are a strategist. I will include AHRQ in
that. Everyone is thinking about these grand designs and plans, now we actually
have an opportunity to make this real. That is incredibly exciting.
DR. GREEN: Carolyn, first of all, one comment and then Carolyn I would like
to ask you to just talk more about two or three other things that you
The comment is this conversation plus yesterday’s is merging for me
around an unstated assumption that I want to make explicit. It is the
assumption that we have to gather a bunch of data and put it somewhere in
proper fashion and way so that we then can use it to get the answers that we
seek. Of course, we do not agree with what the questions are. We actually do
agree that we do not know what the questions will be. Now we know the timeframe
is such that a development of a measure in this sort of assumption that
probably by the time that we are ready it does not matter particularly in the
face of health care reform.
Carolyn mentioned DARTNet briefly and I only know enough about this to
really be dangerous but there is a concept here that I want to get out on the
table. It is the concept that you go get the data you need at 2:00am in the
morning, when you know what the question is without ever acquiring the data.
That is the fundamental, transformative idea in DARTNet. What these guys are
doing is they can now do it for six or seven of the big EHR vendors.
While you sleep, the clinician or the patient at 2:00a.m., there is a data
query that goes to the EHR and gathers the seven measures from the clinical
database that enriches – we have only talked about how to enrich the
administrative database forever. It can unite those. It can query that
administrative database, the payment, the services, the CPT 4 numbers, the
codes, whatever they were. It can also go in and grab the lost charts and
non-lost chart, that record, and get the blood pressures for everybody that is
on this list. It leaves the data there. There is no consenting. There is no
IRB. There is no privacy and confidentiality. The d-identify comes back and
there are 29 of them and 28 out of 29 are in target range. One is not. From a
system level you now have a 28 out of 29 performance measures of some sort and
from the local accountable care organization or the accountable clinician you
can at 8:00 in the morning get the name of number 29 and it can have a little
flag on it.
That is a powerful idea for what we are talking about in my view. This
connects back to the exchange we had with Justine about where is the
aggregator? In my mind, that sort of really changes what we are talking about
around the aggregator. The aggregator is a DARTNet thing. There is no data
warehouse. Are you with me on this?
DR. CLANCY: There is no permanent data warehouse.
DR. GREEN: That is right. It is a virtual data warehouse that is constantly
– you mentioned the word prospective registry and that is one of the
things I wish you would say more about what you are thinking about prospective
registries because in a way this is like an ever evolving prospective registry
that can be customized.
So if you could just say more about prospective registry and could we go
back to this attribution thing again? I was not here for it and I hear you
saying that the discussion with the American Board of Medical Specialties, that
MOC stuff, what the boards were doing was useful. The boards, all of them are
struggling with the attribution issue. What do you mean by attribution?
DR. CLANCY: Okay, so first of all your comments on DARTNet I do not think
were dangerous but actually illuminating so thank you for that. I should point
out that Joachim Roski is here and he and Mark and others have been working
very hard on a set of distributive approaches to assessing performance and so
forth. They have actually come up with a strategy that this could be dangerous
and I will see if I can say this correctly, where you can get the information
that you need but the personally identifiable health information does not leave
the entity at any time. I think that is very valuable.
I do not know if aggregator is the right thing or what but even with
DARTNet you need an entity which is facilitating all of these queries and can
get you the information that you need and so forth. That is the genius of it. I
understand that everyone who prepares the presentations in the two groups gets
way more excited about DARTNet and so forth.
I do think that in the end there is probably something about being able to
merge the idea here of a distributive data network and prospective registries.
If you ever want to have fun you can Google the word registries. It is pretty
interesting. You get wedding gifts, sex offenders, a few other things but it
makes you realize that we kind of take this terminology for granted.
I would say in our current phase, in anticipating better electronic and
smarter, virtual strategy downstream, it is as much about getting a commitment
of clinicians and patients to say, we could do a better job here and we need to
be able to do so. I do not think that it has to forever be linked to a central
repository anywhere although, I think that is how SGS and others do their work.
The boards are struggling with the same issue. The attribution for me is which
aspect of performance are we trying to capture? Some of this is about
differences among stakeholders in terms of what they want.
When open season commences I get calls about what doctor should I have?
Now, the real answer to that question is you know, walking people through
contingencies and so on and so forth for everything from where you live to do
you have a family, et cetera. It is a fairly complex question but boy, they
want that individual.
I have never seen an issue of the Washingtonian or other magazine that
says, the area’s best doctors that do not actually do well on this stance
and so forth. Even though we know that measuring the contribution of an
individual clinician to the care of a patient, especially the patients where
you want some really good information about how we are doing is not so caught
up with and complex in terms of the interdependence among the members of a care
team that it gets very messy. So that means we have to get into very funny
rules about under the control of a clinician or health care system or practice.
Well that feels fair on one level. On the other hand, boy, does it limit our
imagination. What you end up seeing is if you were to peruse NCQA’s HEDIS
measures even organizations that are taking care of insured patients who have
electronic records for 30 years, decision support, the whole nine yards, 60
percent on intermediate outcomes of chronic illnesses, not rare stuff,
diabetes, cardiac risk factors, 60 percent is a high water mark. Seventy
percent would probably get you into Ripley’s. We just are not doing very
Now, we know why that is, right? We cannot make patients do things. This is
all about how they live with it. What we ought to be challenging ourselves is
this is a community challenge, right? Then how do we figure out how we can use
some of these tools to work more effectively with community organizations,
whether that is the health department, whether that is the church, whether that
is the Verizon store for that matter, I mean wherever it is that people go. I
have finally figured out that the one common denominator for the population is
probably motor vehicles and Verizon or other phone stores. You might want to
fill in the blanks for what is going to work in your particular community. Does
Well let me say it in a different way. You know what I would like not to
fund anymore? Studies trying to isolate how much is surgical skill and how much
is attributable to the team at a hospital. I want it all to be graded. I just
do not want people to be padding their CVs with more papers that address that
specific issue. It has all got to work.
Now for purposes of improvement you kind of do want to know. But for
reporting I do not think so.
DR. CARR: Michael –
DR. FITZMAURICE: Yesterday, we heard from a lot of different people with a
lot of different projects. It was very well put together by Matt, Paul,
Justine, and the rest of the staff. We heard a lot of good things. I mean it is
hard to say, well that is not so good, this is better but a lot of good things
but it was hard to piece them and fit them together. When we look for
leadership is it medical leadership? Is it someone representing the consumers
who say, this is what we want in terms of quality? Is it the states and the
federal government? We are looking for someone to put it together or to make
the recommendations to have all of this put together. So do we have the right
kinds of partnerships so far to move quality ahead? We have all different kinds
of partnerships or does it take time to play out as we get more findings and we
find out what works things will coalesce. What can NCVHS do to be the catalyst
to help this move faster?
DR. CLANCY: Again, I am going to come back to my pragmatic response a
couple of minutes ago and say, I would actually look at what is moving in terms
of shaping the landscape of opportunities and that is going to be health reform
legislation. I think that the multiple stakeholder input is incredibly
important. Part of the reason that I think that is in multiple meetings mostly
about safety, but about other aspects of quality as well, the game changer is
actually an articulate consumer sitting at the table who says, I know it is
hard but like, I am sorry why should I be permanently damaged from this
infection or whatever? Put to people well, it changes the conversation from
gosh, this is so hard and maybe we will go from 60 percent to 63 percent this
year that is our goal to yes, we can. That, I think you want to capitalize on
DR. MIDDLETON: Good morning, Carolyn. I apologize I came in late. Blackford
Middleton from Brigham and Woman’s Hospital, Partners Health Care, Member
of the Subcommittee, Member of the Full Committee and no conflicts.
Carolyn, first of all it is great to be with you again and thank you for
the comments, it was spot on. You will be happy to know that AHRQ funded
research at Partners Health Care found a tool that actually did seem to delight
doctors in use. This Smart Forms idea, several of us have been working on it
for a long time, tries to combine in the course of the clinical workflow all
that you need for kind of quality review, data review, documentation and
decision support. There are issues still with usability and some of the
technical underpinnings. I think there is a way to actually make the management
of health care information much more delightful and thereby to delight the
doctor and the patient.
That kind of leads into the second idea, which I think we need more work on
that, but the second idea, how do we conceptualize quality to ensure this new
world order so it is what I do for each and every patient that I see and for
each and every population that I am responsible for. I think that we have work
to do there to bring the quality management idea into the routine clinical
processes of care. Smart Forms may be one method of many.
The other ideas might include sort of something that came up yesterday, how
do we engage the patient further? What are the social networks that might apply
to the patient, to the provider? Certainly, if I am at peer group review and if
I have a quick question how do I manage that from kind of the social networking
point of view and from the participatory medicine point of view. How do we
engage the patient, not only to activate the patient but as we have seen to see
how the patient may further activate the provider in useful and interesting
I guess the question on the table I will ask you is you have already
pointed toward the interdigitation of this analysis with health care reform.
What are the couple of breakthrough ideas that you see that will take these
efforts on the way that you have been leading for years in HIT and quality
management and patient safety and hooked them to attributes in health care
reform that make them sustainable. What are the breakthroughs that we need?
DR. CLANCY: Well, that is big and profound. First of all let me say, I am
thrilled to hear about the Smart Form. I do not know that I would say the
legislation itself contains breakthrough ideas. In fact, I do not think
legislation ought to contain breakthrough ideas personally. I think it ought to
provide the opportunities and the space to create that. So between the
Resources and Recovery Act, which is very sketchy in terms of directives and
the Health Reform Legislation I think there is a use base in a very clear and
imperative – to move forward. I do not know what I think is the magic bullet. I
do think particularly for chronic illness that consumer engagement is pretty
I think that we have got to get smarter about how to do it. The Secretary
made a comment the other day that I totally loved. She said, you know,
isn’t this generational because after all, my father will never every
questions. Rumors are kind of in a different place and my kids think of them as
a legitimate source of a second opinion. It is not a one size fits all kind of
Injecting the notion and making it part of the fabric of care that you are
routinely checking to see how did I do, I think would be very powerful. How to
do that is less clear to me. I love the idea of social networking too so I was
thrilled to hear that Matt brought it up. I think I am really too old to know
what it means. I think my nieces and nephews do it. I am on Facebook but I am
about as boring as it gets. For them this is a different mode of interacting. I
have seen it at your place when you showed me how the providers working on
guidelines and some other activities interact asynchronously but again you are
solving a problem right? You know, to have a meeting and to bring in all of
these people and convene them together and all of that just is not going to
work. I am sure they have to meet sometimes but absent that it really fits into
their lives. Some part of this has got to be getting information and stuff that
consumers need and want to them when they want it. Not when they figure out how
to find us but when it is convenient for them.
DR. TANG: I want to pick on one final pearl that I heard in your
presentation and see how we can mine that. You talked about using the signs to
customize the treatment for an individual, how can we measure that activity or
DR. CLANCY: Oh boy that is a huge challenge for us going forward. The
freeway in my brain is very much about comparative effectiveness because both
the process is informing the Secretary’s allocation of resources for both
the Institute of Medicine and the Federal Council. Both groups independently
working expanded the definition to include something like care delivery
interventions. We do not even have good vocabulary much less definitions. Now
if you are prescribed a pill for your cholesterol we know what that is because
we happen to know if you are taking it but we actually know what the content
is. For disease management, care management, you name it we do not even
literally speak the same language across discipline and health care sectors
which does not mean that we cannot get there. I think it is going to be very
exciting. We will be making some investments over the next couple of years. I
am hoping that we can learn something about that.
DR. CARR: Carolyn, thank you so much for your time and also for your
dialogue over these questions. It is very helpful to us. Thank you.
DR. CARR: Okay, Ernie we are looking forward to your testimony today so
MR. MOY: Okay and I will try to be fast. So first to context, I work on the
National Health Care Quality Report and Disparities Report and so we have a
very specific kind of focus on disparities. We take regular old quality
measures of effectiveness, safety, et cetera and look at it by subgroup so that
is the perspective that I am going to take. There are other ways to look at it
but this is what I know so that is what I am going to show.
Also, when I first started to do this I asked meaningfulness, do you mean
meaningfulness like ONC? They said, yes so I took that literally and started
with ONCs perspective that meaningful use of data begins with capture but also
must ultimately improve clinical processes and improve outcomes. Applying the
disparities data, first of all realization, to me meaningful use is really
putting it in the context of quality improvement so not all of the other ways
we might want to look at data or disparities but for specifically honing in on
what is relative for quality improvement. Looking at disparities from that
perspective, the data sharing then obviously becomes data sharing now by
subgroups, different populations, and in particular populations that are
experiencing disparities in health care and they are amenable to quality
improvement, so specific kinds of disparities but perhaps not all disparities.
Ultimately, this must yield then improvements in clinical processes and
what does that mean? Disparities we know that there is a gap in the clinical
processes between different groups and if we are then going to reduce that gap
we must actually accelerate the pace for improvement among disadvantaged to a
rate that is greater than the rate of improvement among the advantaged group in
order to narrow that gap. That ultimately will be the acid test for improving
processes of care, to narrow the gap we must make the rate of change among the
disadvantaged faster than the advantaged group and ultimately to yield improved
outcomes similarly to reduce that gap you have to accelerate great improvements
among the disadvantaged group to a rate that is faster among the advanced group
in order to narrow that gap. Those are the ultimately then the ways of
assessing whether or not we have achieved meaningful use of disparities data
using an ONC kind of model.
Before I move further, I want to ask the question do you actually need
disparities and data to improve quality. Is it even meaningful or necessary to
look of the issue of disparities and my answer is in an ideal world I think you
do not because in an ideal world you know there are no disparities when 100
percent of people are getting 100 percent of the care that they need. In that
context you do not need to know really about disparities but we do not live in
the ideal world, we live in a non-ideal world. In our non-ideal world, I think
disparities in quality are related, but they are not the same things so
therefore, I believe it is meaningful to look at disparities and this is just
one illustration of it looking at geographic distribution in this particular
case on the left. Quality of care is one particular matrix that is very
commonly used, diabetics receiving hemoglobin A1c. States caught up into
quartiles from BRFSS, on the right you see a difference instead, a disparities
issue in this particular case, an income-led disparities issue. The main point
is it is not the same states that are doing really good or really bad in these
particular matrixes. Quality does not equal disparities when you are looking at
variation across states. They are distinct issues and in an imperfect world I
think it is meaningful to look at disparities of care to improve quality of
care. One way that we can use them is to use them together to target
I think that we are familiar with the quality chasm, the difference between
actual care and the high quality care we aspire towards. The problems that
intervene in effective care, unsafe care, et cetera. There is also acknowledge
of disparities gap, the difference between care received by the advantaged and
the disadvantaged, we see some of the barriers that exist there. Ultimately,
when you put them together maybe you get a really big gap which might be then a
good target for intervention. This is one of the ways that you can use
disparities data I think in a meaningful way to target interventions, to
improve quality of care efficiently.
One of the questions is how are we doing? First of all this is information
from one of our disparities reports a couple of years ago looking at data
capture related to different disparities populations. These are your OMB groups
race, ethnicity categories as well as looking at the poor. You see that these
are the areas where we do not have good data so you see from that very earliest
meaningfulness assessment of data capture we are not doing particularly good
from many of the OMB groups, so not so good.
How are we doing? If we are looking at the more stringent criteria of
actually reducing the gaps and processes and outcomes and again, a report a
couple of years ago, although not a whole lot has changed. You see that when
you are actually looking at the size of disparities and sorting them by those
that are getting better in the yellow and those that are getting better about
the same or getting worse – you see most stuff is not getting better. From
the more stringent criteria most of our processes and outcomes are not
narrowing based upon our observation of measurement activities related to
disparities. I think overall, how are we doing? Not so good.
How can we make disparities data more meaningful? That to me means how can
we make it so that they can actually be used for quality improvement? I think
there are a couple of different areas where they can be used to guide quality
improvement activities. First is targeting. I think I alluded to that already.
I think implicit in that is the use of disparities data is primarily making the
process of quality improvement more efficient. So if you can identify specific
process defects in specific geographic area, effecting specific subgroups you
can now target intervention for those folks, for everybody to try to improve
some – for me the benefit of disparities data for targeting is actually an
They can also be used to guide intervention once we have identified these
particular people that you want impact. I think disparities data can improve
effectiveness because I think most of what we do relates information and not
everybody gets information in the same way. If you know a specific group and
you know what kind of media they use, maybe knowing what kind of language is
also important, maybe knowing about some of the cultural time bombs that might
exist there might also be important. This might then prove the effectiveness of
specific kinds of interventions.
Lastly, again, tracking progress is important because if we say that
meaningful – using disparities data means making sure that it is really
having an effect on progress outcomes you need to track it. The disparities
data, because you only need to capture it for this particular subgroup, can
improve the efficiency of that data capture.
What geographic areas? I alluded to some of the variations across states.
This is now a network to kind of quantify the differences of disparities across
states. This is just one measure, colorectal cancer screening, the differences
between Hispanics, Non-Hispanic Whites. You see that there is a big variation
in that absolute gap across the states. You see the all state average in the
middle. Just by way of comparison, how big is this? This is the state gaps.
This is the difference between the best state and the worst state. You see a
lot of disparities in this case were actually bigger than the best state and
the worst state. They kind of put that into perspective, the magnitude of these
kinds of disparities.
How do you use this from the meaningfulness perspective? We you might want
to work on some of the states on that right-hand side that have really good
gaps. You might not have to then invest in working on it in those low-gap
When we start looking at state variations the next thing that people ask
you to do is look at variations across cities. You look at variation across
cities and there is variation across cities. Then you look at variations within
a city and there are variations within cities, overall, I think one of the
conclusions that we have come to in terms of trying to help people trying to
make disparities data meaningful is that the small unit that you can get to,
the more valuable they find it. I think that is one of the issues there.
Disparities data on one hand is more available at the higher levels but then it
is less actionable. Going down, it is more actionable but then it is really
harder to get to.
Very quickly another issue, what subgroups? I think that we stick to OMB
usually but when we start to parse out to different Hispanic subgroups in this
particular case by ethnicity, you see that there is variation. When you parse
it out by Asian groups you see that there is variation. The main point is the
way that we normally look at disparities from an OMB perspective is probably
suboptimal. There is a lot of variation across the granular ethnicities. In
fact, as a consequence, the IOM in their recent report recommended collecting
information about OMB granular ethnicity that goes beyond OMB but of course
using the CDC categorization there is a whole lot of different kinds of groups
which obviously creates a challenge. One of the issues is which ethnicity to
I very briefly alluded to language as also being important. You can see
that there is a lot of variation related to language proficiency, English
proficiency in this particular case, regardless of racial or ethnic group.
Again, OMB recommends collecting detailed language information but this is just
which I find very interesting. It is a number of languages in each of the
different states. You can see that all of these numbers are really, really
I think that it is appropriate and it is actually beneficial to define
meaningfulness of measures in terms of quality improvement. I think it is very
applicable to the issue of disparities.
How are we doing? So far the disparities measurement has been spotty. Some
optimal data collection has not really let to any improvement in processes and
outcomes to a significant degree. I think that you do see some examples where
there is some improvements but in general, it has not been very successful.
I think disparities data to be meaningful, should help to guide targeting,
development interventions, tracking interventions, and it should focus on
specific subgroups within specific geographic units. I think that gives a best
efficiency perspective. On the large number of granular ethnicities and
languages in the US is a challenge.
Carolyn did say that I could give you my personal recommendations that
reflects only my thought and has nothing to do whatsoever with AHRQ. She
actually has not seen these.
I think that it is important to support collection of disparities data
consistent with the OMB recommendations. We need to start collecting OMB race
and ethnicity and I think beginning to collect information on English
proficiency is kind of important to cover that language perspective.
I think the granular ethnicity detailed language is really hard and so the
first step, what I would recommend doing is identifying those that are
meaningful for national tracking. This is, I think we had a disclosure, this is
an area where obviously I do have a vested interest because it would help my
reports a whole lot if we could identify those ethnicities and languages that
are most meaningful for national tracking.
I think in that process we would probably develop methods that we could
then give to states and other private organizations to help them define what
ethnicities and what languages are most relevant to their local circumstances.
I think that as we think about quality improvement you need certain amount
of data. We need to collect enough data to support quality improvement for
every geographic unit that we are talking about.
I think that we ought to promote disparities measurement and the
disparities reduction approach. That is one way of improving quality. A lot of
times people say, well, you can just reduce disparities by improving quality. A
rising tide will sail boats. It typically does it just typically lifts them in
parallel and why not put the constant with disparity.
I think that disparities reduction by targeting specific subgroups in
particular geographic areas may be a more efficient way of actually improving
quality and at least folks should think about that.
Lastly, I think we need to assess our disparities data measurement
activities to see if they actually improve quality improvement, to see if they
are actually changing processes. If they are not, we consider whether we are
measuring the right thing and whether or not that activity is worthwhile.
DR. CARR: That was great. Can I ask you a quick question? The groups that
you identified as having a disparity are based on – my question really is
do you also look at income? In Massachusetts, we discovered that with their
health care insurance that we have a group of people who are poor but they have
insurance and they cannot afford the co-pay. It is sort of an allocation of
resources but it is sort of a new group that we have not looked at. They kind
of do not necessarily fall into the traditional groups here, any thoughts about
MR. MOY: Yes, so our mandate is to look at disparities related to race,
ethnicity, and social economic status so income is inherently one of the
disparities that we look at. It is very large. I mean there is a huge amount of
disparities related to income even when you stratify by insurance for the
reasons which you stated.
In terms of this meaningfulness discussion I think one of the things is it
does not, it is not really necessarily important how you define these
particular groups that are experiencing difficulties. They may very well be
defined by the intersection of race, ethnicity, income, and other kinds of
characteristics. The important thing is that they are identifiable and that
they can be targeted efficiently and that they are experiencing disparities.
MS. TAYLOR-CLARK: Just to add to that a little bit, one of the reasons
there is a focus on race, ethnicity is of course after you control for, and
often times after you control for socioeconomic position is what I would call
it, there are still disparities that exist among race and ethnicity. That is
part of why we actually have a focus also on race, ethnicity outside of
I want to thank you for inviting me here today. I also want to thank Matt
Quinn for inviting me and thinking of me. He saw at the ARHQ annual meeting
that we are very excited about the stuff that we are doing. I am going to be
speaking from my own perspective today, not from the perspective of my
institution, although I will give a shameless plug for the work that we are
doing at Brookings.
So today the roadmap is really to think, I have been asked and charged to
do three things. The first thing is to talk about current measures of equity.
The second is to provide some opportunities and challenges of current
measurement strategy. Finally, to have some insight I hope, on what makes
disparities measurement meaningful. In that inherently I am going to provide
some recommendations. I am not going to provide specific recommendations as
Just to bring us back to the beginning, the IOM has outlined six domains of
quality. One of which is equity. What we know is that there are measures for
safety. There are measures for effectiveness. There are emerging measures for
patient-centeredness, timeliness, and efficiency. What we also know is that
there are virtually no measures for equity in care. Currently what we do when
we are trying to measure equity in care is to stratify clinical effectiveness
measures. When I say that, what we are doing is we are collecting measures of
race, ethnicity, socioeconomic position, and others. As Ernie suggested, what
we then do is we take these measures of effectiveness, potentially even
patient-centeredness and timeliness and we then stratify them to develop some
understanding of what the disparity looks like. Then we may be able to develop
interventions to affect equity.
The question I then ask myself is what makes meaningful disparities
measurement measures. I am going to argue three things to you or submit three
things to you today that we could certainly talk more about.
The first is a problem that Ernie mentioned initially. That is we simply do
not have a discipline. I am talking about standard race, ethnicity, and
language but we can certainly talk about income or other socioeconomic
conditions and other demographic information. We have no standard race,
ethnicity, and language data across organizations. People, organizations, are
collecting this information in a number of different ways, using a number of
different categories. When I actually took this position I thought I would have
a really easy job. They said, well, you know, all you have to figure out is
first, how do we think about collecting the standard way race, ethnicity, and
language. Then how do we get that into the system and then spit some data out
so we can look at disparities. I thought this is going to be simple and as it
turns out that is not the case. We have a number of different ways that data is
collected and a number of different categories. I am going to talk a little bit
about how the Institute of Medicine has dealt with that.
The second submission that I will make to you is that we need to have, and
I have heard a lot about data aggregation this morning, but we have to have an
ability and a capability to integrate data systems. That is to say that where
we collect and require demographic data may be very different from where claims
sit, clinical data sits, lab data sits, so ultimately, our inability to
actually integrate those systems means that we have an inability to effectively
look at measure of effectiveness stratified by race, ethnicity, or other
The third point I am going to argue today, is that ultimately, one of the
reasons that we may not have measures of equity is because we simply do not
have incentives to collect or report or to utilize those data. I am going to
show you a couple of points that have been made by private organizations to
start thinking about a crediting organization in order to align incentives for
providers and health care organizations to actually acquire these data and
develop some measures of equity.
So the first point I made was that there are no standard race, ethnicity
data. The Institute of Medicine of August 29th of this year
developed a really interesting and really important recommendation that said we
absolutely need standard data. Now this is a very busy piece and let me see if
I can break it down.
What we are going to do now is to think about the local reality versus the
national reality. The local reality is exactly what Ernie points to and in fact
what Carolyn pointed to. It is to say that it is very nice that we have these
large OMB categories of race including Asian groups, and Black groups, and
White groups. Ultimately, at your local level we cannot act efficiently on that
information. I cannot say that if I am looking at someone who was
“Asian” whether their status as Cambodian or Vietnamese or Japanese
is actually more important to me at the local level. What I am trying to
develop is an intervention both in terms of language, in terms of culture, in
terms of other points that are really important to the historical data that you
live in now.
The issue here is that we have a local reality versus a national reality.
What IOM has suggested is look, it does not matter at the local level what data
you want to collect, it matters that we have some standard way to roll those
data up into these standard OMB categories. So that is in fact what they have
endorsed. They said, what we can do is collect this information but we need
some sort of map. We need some sort of map that will allow us to say you know,
in my local population we have a large number of Vietnamese folks, we have a
large number of Japanese folks and ultimately at the national level what we are
trying to do is track Asians. How do we actually roll those categories up in a
standard way among all health care organizations so that we can actually look
at these data appropriately?
In fact, that is the first argument that I make to you is that we
absolutely need to have that map in order for health care organizations to
follow it so that we are actually comparing apples to apples and oranges to
oranges when the data is done.
The second point that I am going to make is that we need an integrated data
system. Ultimately, you see here that there is a number of places where we get
data. For AHRQ’s purposes and the National Health Disparities Report we
are getting data from surveys of population, health care facilities, data
extracts from health care organizations, and ultimately surveillance and vital
At Brookings we are very specific in one area which is the health care
organization. I show you this very busy, but this is my favorite slide, a very
busy slide to say that we are getting this information mostly from the patient.
The patient gives a number of organizations, a number of entities their
information about a number of things. It is about their satisfaction of their
care, about their race and ethnicity, and other demographic data. Again, those
data systems and what you see and hear in terms of data exchange and transfer,
those data systems do not speak to each other well enough so that we can pull
the clinical or the claims or other administrative data and have them speak to
In terms of disparities, one of the things that we have actually noticed is
that there might be other data sources. One of the things that we are doing at
Brookings is actually working with employers as a potential data source. I am
not asked to speak about what are the current data sources but what might be
some new data sources. This is just one example that we are using. That is that
employers right now are acquiring these ethnicity data for the purposes of
equal opportunity. As we know they actually have to acquire those ethnicity
data. The question is how do we figure out, putting an identifiable flag on
those data so that we can then integrate a system in terms of health care that
would allow us to stratify some of the measures, some of the effectiveness,
patient safety, patient-centered measures by race, ethnicity so we could
actually look at disparity. The major difficulty that the health care system
and these other organizations have is actually getting the data. It is actually
acquiring the data from the patient and particularly race, ethnicity, and
language is a major challenge.
I wanted to talk a little bit about this aligning of incentives. I show
this again, very busy slide here to say that there are organizations, private
entities and I hope that HHS takes it up as well in terms of public entities.
In thinking about aligning incentives to not only collect, not only report but
also utilize and efficient and effective way as Ernie described, these data. So
what you will see is actually three organizations that I am going to describe
The Joint Commission, NQF, and NCQA are all – well the Joint Commission
and NCQA are trying to develop measures of equity so that their organizations
both hospitals and health plans are actually accredited based on some of these
standards. What we know about National Quality Forum is at this point they have
endorsed 45 practices to guide culturally appropriate and patient-centered
My argument to you today is that the best measure of equity that we can get
at this point, given that we actually have their measure of equity, are to
think about developing strong, structured, efficient patient-centered measures
that can be then stratified by race, ethnicity. That would be an argument that
I would love to hear arguments back to me on because I kind of toil with this
idea well, what would an equity measure really look like if we had our own
equity measure? At this point, my argument to you or my submission to you is
that we really need to think about patient-centered measures that would
actually lead to outcomes. That is one of the big pieces that we need to think
about, effective outcomes in terms of measurement.
NCQA, just to make this last point, is now developing new HEDIS measures
that will be based on the OMH CLAS standards which are culturally,
linguistically appropriate standards which includes data collection as well as
the use of these data.
I am going to actually end here by giving the shameless plug that I
promised to give which was that the racial and ethnic equity initiative, and
Dr. Roski is going to speak about, my colleague here is going to speak about
the high-value health care project at Brookings. This is something that we are
actually undertaking right now and we are really trying to look at
organizations and entities that are doing this work that will culminate in
March of next year in hopes that we can put forward for example, the
recommendations that your committee will make today.
I thank you for this opportunity and would welcome any questions.
DR. CARR: Excellent, thank you so much. I would like to open it up now for
DR. TANG: Thanks very much to both of you. I had the opportunity to hear
Kalahn at Brookings maybe twice because I am on a number of Brookings
One question I have is, as you will note we put it in the 2011 Meaningful
Use Measure, at least the capturing the data, and we are also tracking I think
the CDC version of that. I asked the IOM Committee to submit their comments
when it goes back to them for public comment because that came after the draft.
What ideas do you have about making these – so part of meaningful is
actionable. So what ideas do you have on how we can act on it one patient at a
time? I do not just offer up an example so it is not an example of what we are
trying to do. We have a wellness program and even wellness is not one size fits
all. One of the great examples, a physician in our group did, there is a
sizable population in the San Francisco Bay area of South Asians. They have a
higher risk of early cardiovascular disease. It would be inappropriate to
suggest that they change from their high carbohydrate South Asian diet to an
American diet of high fiber. What we have is instead customized both a test and
a video that talks about the conversion in South Asian terms. It seems like
that is the kind of “customization” that Carolyn talked about that we
have to do for each and every encounter. Different cultures have different
attitudes about medications or shots or diet or exercise, all of those things
and we almost have to do this – that would be the most effective effort at
least that is my simplistic way of looking at it.
Other ideas that you have in terms of how to make this data now that you
are helping us get the data, then we have to make it in front of ourselves so
that we can make it actionable. How do we study the effectiveness?
MS. TAYLOR-CLARK: Dr. Joe Reed at Harvard Medical School has actually
developed, and others, has developed programs for medical education. I think
that would really be the place to target some of the findings that we have to
say look, there are these differences in the cultures. There are these
differences in trust of the health care system that we know absolutely exist
among racial and ethnic subgroups. What we are really absolutely need to do in
the medical education point which I do not know is the charge of the committee
but it is to make those data available to students in medicine so that they can
actually develop culturally appropriate services, which is a term that I do not
absolutely love but I will use it for the sake of this point, so that they can
develop culturally appropriate services and interventions but really knowing
what their patients are, catering to their patient. That is why these
patient-centered measures might very well be so important because they are
MR. MOY: I think that you hit on just right on how disparities can be used
for that customization because it is inefficient to work with each individual
patient to refine their precise diet whereas if you can cluster the different
population subgroups then you can actually develop products that target that
particular subgroup. I think the important thing is maybe having some kind of
inventory systems that these things can be shared across institutional and
geographic settings where others in other parts of the country or in other
institutions in Boston might be able to take advantage of the same kinds of
DR. SCANLON: I have a question about how we get the disparity data that we
want. It seems partly about data flows. This committee actually has a
recommendation from about three years ago or so, we did a report on race and
ethnicity and recommended that there be an effort to collect the sort of
granular data that you talked about but not at the provider level. That is
where IOM seems to be saying that at the provider level we should be able to
have this kind of information. I think our logic was that collecting it once,
right, given the complexity of it is better than having everybody collect it,
submit it, and then have someone have to deal with the discrepancies that sort
of exist among them.
I guess my question is is IOM going that far in saying that once they
collect it is should also be transmitted which then I think we have this issue
of the discrepancies. Then there is also a question of is it really useful at
the provider level to have the granularity given that what you have talked
about in terms of the complexity of it and potentially that some higher level
aggregation is very important at the provider level but we do not need to
confuse things in the process.
MS. TAYLOR-CLARK: The IOM actually has made very clear that they think that
there is going to be better data submission in transfer protocols between
health care organizations so that is first. They also think that redundancy in
the system does not work. In Massachusetts right now we provide technical
assistance to the state to figure out how can we actually get from the
hospitals, and just to speak to the hospital point, acute care hospitals are
required at this point to acquire the information. They are doing a very good
job of it. Almost 98 percent in Massachusetts have been correct. The question
is how do we develop these data transfer protocols that would allow health
plans to acquire the data so that they do not have to reinvent the wheel. They
are also creating an all payer system which will be very helpful because that
demographic data will flow through it.
Your point to the provider level is a very good one and one that I think
even Carolyn made which is that the sample sizes would be too small for
providers to do anything with those data. Ultimately and unfortunately, from
the consumer perspective and we have done a lot of work with consumers, we know
that they are more comfortable with their providers and or hospitals collecting
and acquiring this kind of information. They can see clearly why you would
collect their race, ethnicity in terms of quality of care. What they do not
really get is why your health plan is acquiring this information. The entity
matters in terms of consumers.
MR. MOY: I also think that sharing is very important but I think that your
point as to collecting information that is actually meaningful to the person
who is going to be primarily using it is also very important. I think at this
point at least, I would not say we are at a point where they are collecting
tons of data and just not know how to collect it. I would encourage the
providers as a start to collect information on the groups that are particularly
meaningful to their practice. Then later on down the road once they have gotten
these back they may think about collecting more detailed data that might not be
meaningful to them as a practice but might be meaningful for a geographic unit
for aggregation. As a start, I would start with what is most important to them
that they can actually use in quality improvement.
DR. FITZMAURICE: I think it would be nice to have, more than nice to have a
study on population where you look at the disease incidents by a race,
ethnicity category. Then for the same population you look at treatment services
for that particular disease by race, ethnicity category. Then you look at the
outcomes information by race and ethnicity category so that you get an overview
of here is what we have. Here is what we are doing. Here is the result of what
we are doing. Then to follow up at a more individual level, maybe to form a
registry of certain disparate groups, with their permission of course, to
follow them over time so that we have the same kind of information, actionable
information to do something with as opposed to separate studies in different
part of the country for different areas where you do not make them together and
you are further away from cause and effect. Although, we think we have some
good evidence for cause and effect.
Are there areas where that is being done or is that a good model that is
also difficult and extensive to do we are not doing it?
MR. MOY: I think we try to do it in the reports, and I will say this is not
just specific to disparities, it is in measurement in general that typically
processes are collected in one place and outcomes are collected in a different
place, where health is collected in one place and health care is collected in a
different place. Then it is potentially somewhat less meaningful because you do
not know exactly where the lesion is. In the report we try to do that. We try
to link process measures with outcomes measures at least so that you can look
at the two in tandem, if they are going in the same direction you might have a
little better confidence in the processes are in fact effecting the outcomes.
I think it is a very valuable thing to do but whether or not that is
universally achievable is probably more difficult. They typically are collected
in different areas and different data collectors.
DR. GREEN: Ernie you had a slide where you color coded it on the right-hand
side. What the figure was showing was better, the same, or worse. What can you
teach us about the nature of the data that were necessary to get to that slide?
MR. MOY: You mean just taking the measure and cutting up the states into
better, same, and worse? It is fairly straightforward.
MR. GREEN: What are key attributes of the data and the data collection
process that allows you to summate that to the point that where you could say,
at this point in time things are getting better, they are not changing, they
are getting worse. Now that is the sort of matrix in my view that, to follow
Carolyn’s advice and we looked at the Health Care Reform Bill and I think
what we are going to want to know is are things getting better, are they
staying the same, or are they getting worse? What were the attributes of the
data that let you put a slide up there like that?
MR. MOY: You need to have the same measure that is being trapped over time
the same way using you know the same kind of sampling frame but besides that
PARTICIPANT: off mike.
MR. MOY: The same population, let’s put it that way that you are
sampling from are not the same people or the same units necessarily. In
general, I would say that is not that hard to achieve. We have many databases
that are continuous in nature and that we can track stuff over time. This
continuity has of course hurt us so when it does not get funded for a year that
hurts us. Most of the things are actually achievable. It is not impossible to
do, let’s put it that way.
DR. GREEN: I certainly take your point that you need to not allow the
variations and the answer there is is it getting better or worse be based on
the fact that you have totally change your data collection mechanism. I
understand that part but as we are trying to think about meaningful measures
that can be derived, I just wondered if you knew something else about what it
took to build up with what we have got to get to where we want to be quickly,
rapidly as opposed to deciding that it is going to take us eight years to
decide what the construct of race is and what we are going to represent it as
and how we are going to validate that. Is it going to be distinct from
geography or not? Is it economic or is that really – you know, you can get
the tone – I may be the Lone Ranger but I think that this Committee is
beginning to consolidate again around pragmatic ways for now. That is the
DR. CARR: So Larry was complementing you. Is that fair to say? Larry gave
you a compliment that this is very pragmatic information.
DR. GREEN: From Carolyn’s point of view, practicing medicine tomorrow,
you know you want to be really nice – I would like to know if my patients
are getting better, staying the same, or getting worse. I do not need 37 graphs
coming back for 16 of these and 9 of those and 2 of the others, that –
DR. CARR: Blackford will have the final question and we are very grateful
for this opportunity to talk with you.
DR. MIDDLETON: Thank you for presenting. It was really great to meet you
both and hear from you both. I guess the question crossing my mind is in a way,
culturally appropriate services as a reflection disparity is sort of a one
dimension of an analysis from many that might look at what is the right
combination of therapy or intervention for the patient at hand? Other
dimensions might include of course, what is the genetic makeup? Is the
hypertension therapy appropriate given a sequence analysis? What are the payer
appropriate services if you will which is the hardest one untenable of course
on face value but will this service be covered for this patient before me.
You have described a number of ways to assess culturally appropriate
services, I guess my question is in any of the work, it is not in my literature
at all, when you think about the disparities or the assessments of culturally
appropriate services, is there a way to take into account these other
dimensions or even one more patient’s preferences? Despite any cultural
orientation of course, patient preferences may dictate and sway decisions one
way or another. It seems like that is already being considered.
MS. TAYLOR-CLARK: That is in fact what – the submission that I gave to
you about equity and trying to develop measures of equity actually centered
around this idea of patient-centeredness which may in fact center around some
preferences. So that is, can we develop measures that are not necessarily being
stratified per se, but that get to equitable care? The only measure that I can
think of where we are not stratifying is to say that the patient came in and
they requested translation services and that becomes the denominator, all of
the patients that request translations services, why? That means that they are
limited English proficiency so they are now the denominator.
In the numerator, we have a combination. We have a combination of the
patients who received the services and potentially were satisfied with that
service. Some combination in that numerator and now we have a measure of equity
that is not being stratified but that actually allows us to look at those
things. I think that patient preference comes into this idea of
patient-centeredness if we can develop a measure that would look at preferences
that could be objective. That is the difficulty with these patient-centered
measures of course because what we know about different subgroups is that there
are some groups that are more trusting, some groups that have absolutely
different preferences based on their experience and their historical experience
in a health care system. It gets a little bit tricky to develop that measure.
That is why we are here. What we have got to figure out is how can we
actually develop a measure based on preferences that will allow providers and
others to make effective, actionable choices when they are developing
MR. MOY: This is one of the headaches we dealt with because the disparities
purists want to exclude patient preference from the equation and so when
iowa(?) did it on equal treatment disparities exclude you know, the people that
did not want the particular service. From a data perspective of course, it is
almost impossible to operationalize give what we currently have. We do not know
anything about patient preferences. We kind of ignore it in our reports because
it is the best that we can do.
However, I do note that people are increasingly are talking about maybe
tracking information about patient refusals so that when we are getting the
information about the different kind of screenings to report because the
patient refused it. That might actually be a kind of measurement that would be
used in tandem to incorporate the patient preference.
DR. MIDDLETON: Just to follow up briefly, I think that is a very
interesting way to look at it if you go from the compliment or the inverse side
being refused is an expression of preferences at some level. I guess where I
get concerned, and this is may be a pure Brookings sort of thing, from a value
perspective, values underline patient preferences at some level and values may
vary dramatically and be expressed through the preferences if you will. So even
within a single cohort culturally homogeneous – the values may be all over
MS. TAYLOR-CLARK: I will say that, and you may or may not back me on this,
and I do not dismiss this point of preferences but what we find in terms of
disparities and outcomes is that preferences are not driving many of these
disparities and outcomes and I will put that on the table. When they have done
smaller cohort studies they have found that preferences are not in fact driving
MR. MOY: I think one of the important concepts is making sure people are
expressing their preferences after they have received all of the information so
they are making an informed choice as opposed to you know an uninformed choice.
DR. CARR: Thank you very much for a wonderful presentation. We are going to
transition into our next group which is meaningful measures of value including
efficiency. Joachim Roski from Brookings, Managing Director of Health, Value
Health Care Project at the Engelberg Center for Health Reform will be our next
MR. ROSKI: Good morning, thank you very much for inviting both Kalahn and
myself to talk about some of the work that we have been doing.
What I am hoping to do this morning is you know trusting that Mike will
provide some more detail to comments, particularly on what CMS might be doing,
talk a little bit about just basic nomenclature so that we are all talking
about the same thing or at least use words in a consistent way. I found that to
be one of the key stumbling blocks in some of these discussions. Then I will
give a quick snapshot about how I see the environment in terms of measure
development. What is being deployed now? What are the advantages and
disadvantages? Then I want to talk a little bit about the infrastructure that
is required in order to put such measures forward in picking up on some points
that Larry has made earlier and that we heard Carolyn talk about, talk some
more about the distributive network opportunity and how that may be a pragmatic
and sophisticated way as a matter of fact to move forward in doing measures of
value and efficiency as well as measures of quality.
In terms of definitions, there are certainly lots of them out there. AHRQ
has commissioned a report that was published not long ago that goes into a lot
of detail about issues of efficiency. I am not claiming to be comprehensive in
my comments; all I am trying to do is sort of quickly set the stage for what
some terminology might be.
As you might imagine, the concepts of cost of care, efficiency of care, and
value of care are not synonymous although, in some discussions they are being
treated as if they were. I would like to call out how I see those differences.
These definitions are based on some work that the AQA, the Ambulatory
Quality Alliance has done a while ago but I think it is very useful to quickly
Cost of care really pertains to one part of a ratio if you will of where
the ratio is made up of some definition of quality and some definition of cost.
So cost is just one part of that and it measures total health care spending
including resources use and unit prices.
Efficiency of care would pertain to a particular cost of care for a fixed
and chosen degree of quality of care. You can take any of the IOM categories or
any other definition that you may have but the important part is some
stakeholder’s particular fixed view of what efficiency is. In today’s
discussion when we talk about efficiency it’s typically the efficiency as
the payer would see it or as the consumer might see it, actually it is more
likely the payer would see it efficiency.
Value, and that goes to Blackford’s point is really cost at a rated
preference of quality and that value of care may be different for different
people in the room. Just as some of us may value a Mercedes more than we might
value a Yugo even though both of these vehicles might get us from point A to
point B but we have different, other issues that we are concerned about when we
purchase cars. That is I would argue the same here.
If these are some of the definitions then the next question and I will
actually be focused largely, initially here on cost of care and how we measure
that. Then, when we think about cost of care they are actually different costs
to different people in the health care system. By now it has probably become
sort of commonplace to say somebody’s overuse is another person’s
It is important for us to be clear about what particular perspective we
adopt in talking about cost. For one there is the cost to the consumer.
Typically today, unless you do not have insurance and would aptly pay out of
pocket for everything, all you would see is the out of pocket cost above and
beyond what your insurance might cover. That may have little or nothing to do
with what the real thoughts are that go into all of the prices for receiving
For plans and employers, what they would see is discounted charges so
nobody pays as we all know the actual charges that are being put through, they
are being subject to negotiations between a plan or an employer and the
provider. That becomes then the cost, the unit cost multiplied by the
utilization if you will.
In addition, from the plan and employer perspective, we have the issues of
administrative costs on the plan administration side. I will not talk a whole
lot more about that other than to say it would be great to know a whole lot
more about the big problem from a pragmatic point of view as we do not have
very clear accounting standards for how to put that actually into buckets and
to know how for example, we should think about the cost that a health plan
might spend on a disease management program and how to cut a deal with that. Of
course, employers would be very concerned about issues of indirect cost and
productivity above and beyond what might be the outlay for the direct cost.
Now from the providers this turns out to be a very different part of the
equation because they have to be concerned about what costs am I going to lie
out in order to render a particular service? How many staff do I need to hire?
How many investments do I need to make in terms of sophisticated equipment and
so forth until you have fixed and variable costs that you have to be concerned
about? That is an important perspective.
From a societal point of view which is typically the perspective you will
see reflected in academic articles on how to think about cost, it is typically
not from a business perspective where some of these other perspectives –
here issues of disease burden, quality adjusted, life years and others might be
concurrencies in terms of how you might evaluate if something is or is not a
valuable service. For example, from a societal perspective one of the cheapest
interventions that we know of in health care is advising smokers to quit which
is a frustrating thing to be engaged in at a physician level typically because
you have very few people that actually take you up on it. It costs very little
to deliver and there is a certain percentage of patients that still respond to
it. On the other hand, mammograms for example turn out to be from the societal
perspective a relatively expensive intervention in terms of how many years of
life you might save across a population and so forth.
The next issue that I want to sort of drill in on is when we think about
cost what becomes a unit of cost that we want to think about? Here I will
particularly think about cost to the payers – in terms of the charges that
come through on the health care side. Just to put some nomenclature on that, I
can think of at least three ways you can think about a cost. One would be
unit-based, basically just counting up how many images are being done by a
particular physician per a population. That can be a general population or it
can be a very specific population with a particular condition.
You could think about episode-based approaches which are much more complex
but would be more in line with for example, a person-based approach to
measurement where you rather than counting up individual services, try to
string together a meaningful concept of what it means to have asthma or to have
a heart attack and to string together all of the costs that would be incurred
for treating a patient with that particular condition. That can be more or less
complex based on the type of conditions that you are dealing with, the type of
co morbidities that are associated with it and so forth.
If you engage in this sort of measurement unlike the former, you are much
more likely to be working in the relative cost area. What that means is you are
much more likely to take an approach where you measure this person is so much
more expensive or percentage wise more expensive in treating a particular
condition relative to some reference point which is typically the average that
is chosen for any particular collective of providers for example, as opposed to
dealing with absolute numbers. You would have to do some math to get back to
what the absolute numbers would mean.
Episode of care measures as some of you who are familiar with the work that
MedPAC for example has done, are not a panacea because they are also subject to
if you will, challenges that have to do with the data sources that we rely on
and what gets into the data system in the first place. Here essentially what
you are saying is the more effective that we need to treat a particular
condition, the relatively less costly you are to somebody else. We have all
heard about these examples where you know you can be incredibly efficient
treating patients who did not need the service in the first place. If you do
not separate that out you are going to have a one-sided point of view.
One way to balance an episode-based cost of use point might be a per capita
approach where not only would we measure how effective are you in treating a
particular episode, it is also keeping an eye on how many episodes are you
actually generating and how does that compare to somebody else that we are
In terms of measurement approaches, again on this side of the ledger in
terms of cost to the payers of the health care system I can think of for
example, three types of approaches. Some of them are more comprehensive and
enhanced than others. In my book what I think are probably are methodologically
the most advanced, I am not commenting on how useful they are, but
methodologically, the most advanced are in my mind proprietary episode-based
approaches. You may know them as episode treatment groups, ETGs, ERGs, as
promulgated by the firm enGenic, or you may have heard of MEGs, episode groups
promulgated by a firm formally known as MEDSTAT, now known as Thompson METSTAT.
It is very similar in approach trying to define essentially the universe of
diseases into particular types of episodes. That ranges from a universe of
about 500 some episodes to some much larger number and some smaller number that
is probably for this point of discussion not particularly important. They are
in wide use today by health plans around the country. CMS is experimenting with
some of these approaches and I am sure that Mike will talk about what they have
There are some other approaches – one concern with these approaches
outside of the utility of them in the first place is a concern about
transparency meaning that you cannot find out how actually enGenics is defining
particular episodes of care unless you buy the product and they will sort of
open the box. It is certainly not in the public domain. A head to head
comparison for example of these tools is not available although CMS has been
experimenting with that and maybe Mike can address that.
A second type of approach that is much more recent is basically trying to
reinvent the wheel trying to develop something in the public domain if you will
that exists in a quite sophisticated fashion and nontransparent way in the
public sector. For example, some of the work that we are affiliated with and
the American Board of Medical Specialties is trying to develop episodes for 12
conditions. Remember this is 12 out of a potential universe of 500. I am not
saying they are all equally as important but it is just a lot more limited. You
may have heard about the approaches by Promethius which is an effort that is
funded by the Robert Wood Johnson Foundation which has at its core an
episode-based measurement approach but they put some additional methods on top
of it basically trying to classify cost you would expect to occur in a health
care system or in a clinical encounter for an episode of care versus trying to
tease out potentially preventable costs of care.
These are typically based on costs that are associated with treating
particular conditions meaning that you are dealing with the costs that are
particularly incurred for treating a patient with diabetes. Different from that
for example would be costs that would be incurred for the same patient who
broke a leg where you would argue well, treating that really has nothing to do
with the diabetes for the most part. Those costs would not be included for
calculating diabetes related cost of care.
This third approach and I am not exactly sure if this the right
classification but I have called it a transparent condition specific per capita
approach that NCQA has adopted for six conditions. In full disclosure in my
previous life I was at NCQA and helped develop them but here the concept is
basically focus first on episodes that are relatively easy to define at least
timing wise. Basically, chronic conditions and it has this is indeed
cardiovascular disease, asthma, and so forth and try to count up the total cost
of care that it spends in treating these conditions. This would include not
only care that is specific to cardiovascular disease but include for example,
cost associated with a broken leg.
What are some of the key challenges that the measure developers have to
deal with? For one, it is notion of well are we dealing with costs? Are we
dealing with resource use? Are we dealing with paid amounts? The answer to that
can differ a little bit. If you deal with one payer only, what you would
probably deal with is the paid adjudicated amount for a particular cost of
care. If you are trying to do this across payers which for example, NCQA is
doing by comparing the two different plans, you have to find a different unit
because the fee schedule of health plans as you know is as securely stored as
gold at Ft. Knox. You will not get that. You will have to develop some
alternative way of putting a price on something that is not the actual price
paid by anybody in particular.
I can talk more about what some solutions are to that might be including
how you might extend based on some work or the fee schedule that Medicare has
and impute other stuff that they either do not have in their benefit package or
that is not defined there. Then you would talk more generically about resources
as opposed to any specific dollars.
In terms of the methods for episode-based approaches you will first have to
figure out well how do I define an episode? How long is it? How do I translate
a clinical concept that I have about what might constitute an episode and
operationalize that in my measurement environment in a way that I can come as
close to that or approximate that as best as possible. To date as you all know,
EHRs or clinical data systems are typically not used for measuring cost of care
mostly because they did not exist for the longest time. Second of all, the
accounting standards are not exactly clear in how you would put a dollar value
for example, on certain things.
Typically, the infrastructure that we are dealing with is administrative
data full cost. You still have the problem of how do I approximate out of
administrative data some clinical concepts of an episode of care that I want to
come to as close as possible? Just as with anything else we sometimes do pretty
well and we sometimes are way off and quite inaccurate. We would have to test
how close you are with your approach. You have a problem or you have an issue
relative to sensitivity and specificity. We all know that cost of care and
Carolyn with her eloquent brief about that this morning, sicker patients
generally cost more. When we define episode, we need to either on the
definition of the episode itself sort of rein in the availability in patients
to get to a homogeneous group of people that we want to compare in different
settings or you have to try to deal with it on the risk adjustment side and no
risk adjustment system that I know can really take care of a large amount of
unmeasured variation. You have to sort of deal with that.
You have to deal with issues of exclusivity and composites. We heard this
morning about how or to use a reference from yesterday how the future in
quality measurement and Paul had talked about that also is probably not the
789th process measure of was XYZ rendered within this timeframe as
opposed to did the patient walk X amount of time after they had hip replacement
surgery. You have the same issue here. If you want to construct episodes that
you can put together in terms of composites, meaning that we are not measuring
it only at the this micro level, but we want to be able to roll it up to have a
better point of view of what the cost of care might be, you then have to figure
out how they can stack.
Then you have to deal with the problem of exclusivity of assigning patients
two different episodes. That meaning that if you are, and this is obviously a
very common problem, if you have a patient who has diabetes and who has CHS and
who broke a leg last year, you at least have three episodes but you need to
figure out if you wanted to measure that doctors total cost of care you do not
count that person three times if you wanted to do that.
In the interest of time I will not go into a lot of detail on this risk
adjustment, I know you are all very familiar with the challenges with that.
In terms of the data sources measuring costs we already talked about that
administrative data is probably for the next few years be the only data source
that I know of that we have available to measure cost of care in this way.
Obviously, cost measurement systems for example on the hospital side who have
cost reports and so forth but in my estimation they have not proven to be all
of that fruitful and I do not hear a lot of people talking about building on
that in any particular sophisticated way. However, there are opportunities now
and I will talk about that in a little bit, how we might be able to get a
little bit more precise on our administrative data if we link in some clinical
data. For example, one of the problems with measuring through claims is that
often the risk adjustment or the precision on the denominator decision is not
there because clinical detail is missing. If we could connect some other
clinical data sources that might work provided that these data actually exist
in a somewhat ubiquitous way across the country.
We for example, currently have a project underway in California where we
are linking the registry data, the clinical data from the Society for Thoracic
Surgeons with WellPoint administrative claims in all California hospitals. Here
the opportunity that is aside from measuring quality is that we could use the
clinical data that exists on the registry side to better risk adjust our cost
measurements that come through on the administrative side. We can probably
think of many, many more examples of how that could work.
I talked a little bit about the problem with standardization already. One
of the issues for example that we have such a hard time getting a handle on how
high is a health plans administrative cost is this problem of standardization
and accounting standards for what exactly should go in what bucket. I think as
we all know there is a lot of wiggle room if you will to measure that at the
We talked about the issues of if you measure across payers how you have to
find multipliers I call them or a fee schedule that is not a proprietary fee
schedule but that could be used. Obviously, in terms of comprehensiveness and
this is generally true if you try to measure provider performance based on any
one health plan or payer’s performance or even CMS what you will encounter
is that you get a partial view obviously because that does not represent all of
the care that is rendered by that physician. Ideally, if you do use
administrative data try to shoot for all payer data that is informing that
In the interest of time I am going to skip over some of this. Key
challenges obviously are risk adjustment and that you do not want to over
adjust. You do not want to adjust away differences that are real that you would
like to get a handle on.
What we do not have to work on at all as far as I know is actually the
concept of efficiency in the sense of linking costs to a particular level and
fixed levels of quality. NQF as you may have heard has not endorsed any cost
measures as a matter of fact until not too few years ago this was sort of
considered anathema to their mission. With the arrival of Janet Corrigan that
changed luckily. There are now some frameworks that have been identified but I
think as Carolyn would say they are sort of at a 15,000 and 30,000 level foot
view. How to operationalize that is not exactly clear but we expect that an
endorsement process will go forward sometime early next year with measures
being submitted. A lot more work to be done on how exactly would we link
measures of cost with quality to get to an efficiency.
Then the issue of endorsement obviously has to do with trying to get a
consensus. That is what I want to get into in a little bit more is how in the
world are you even if we had this would we implement it in a consistent way?
What might be a practical and pragmatic way to think about the infrastructure?
Here I want to make a few points about why a distributive network or a
federated model might be viable here as well. We are actually trying to test
that in some projects both on the quality side and soon on the cost side.
I wanted to make a couple of overarching points in terms of how we got to
our conclusion. Basically what we are saying, and Paul said this before is with
the ARRA and with the arrival of more sophisticated IT technology and the funds
hopefully to pay for them, we will see data sharing between providers increase
hopefully significantly to improve care coordination around the country.
However, providers will come together with their business associates in a
variety of different arrangements and data sharing environments with probably
different capacities to do certain things. For example, you have providers
coming together in integrated delivery systems be they brick and mortar or
virtual in the future, that may be one arrangement. You may have another
arrangement where providers may work particularly close with health plans who
have figured out ways to support providers with information systems.
I think in Massachusetts there may be some early signs of how health plans
might be able to do that in a somewhat sophisticated way. You may have other
arrangements and this by the way might differ a little bit by condition. You
may have other arrangements where providers work very closely with registries
be they specialty based registries or be they regional registries that have
been set up. In very advanced environments which I cannot think of many other
places than Indiana and Massachusetts quite frankly where you would have
actually communitywide health information exchange where information is being
A lot of diversity and I think that from a national perspective what you
want to do is take advantage of the many issues that are out there and try not
to be too prescriptive about particular ways that you have to get to an end
point for which you have to be more prescriptive.
The basic notion here is that we are saying that you know, this is of
course preaching to the choir, providers who have come together in different
data exchange environment, that is the blue box and then out of that exchange
providers get value added information in terms of patients lists, decision
support, alerts that would allow them to act on particular issues.
Payers and we for example have seen this on the CMS side, on the disease
management side where for some time CMS tried to provide claims data into that
environment which for example informed them about re-hospitalizations and so
forth. What we are then saying is out of that environment, where PHIs being
exchanged, we should be able to extract in a consistent way, numerator and
denominator statement very much along the lines of what Larry had talked about
earlier. You can do this as frequently as you need it to be. That information
which does not contain any PHIs, it is just basically a rate of how many people
with diabetes got XYZ service can be used by public and private payers for
incentive payment. It can also be used to inform consumers about selection of
providers and by the way, consumers also can contribute information to this
This is what this whole picture looks like together. Basically what we are
saying here is this should be done in tandem with the several policies that are
being now advanced on the federal side both for meaningful use definitions of
IT, increases of value-based purchasing and so forth that the definitions of
measures and requirements should align and should ideally be the same. That
would give us the biggest bang for our buck. In terms of a practical path
forward we think that these distributive data models have a lot of promise. We
are now for example, experimenting with a way to consistently query health plan
data not just for sentinel events as for example that FDA is now doing but to
query them for some performance information which over time that data will be
increased with clinical information and so forth.
If we do that in a consistent way we can answer through these pilots a
number of very important questions that I think all you will have and again in
the interest of time I will not go into that but maybe you can read about them
afterwards on the slide, to answer critical questions about how to connect data
and how to get the right information out of the data and then act on the data.
We believe this requires leadership on the federal side or on the public sector
side in terms of coordination and planning. For example, your entity to figure
out the strategic vision methods and timeliness for setting the expectations
for what should happen. Then the implementation happens both on the public and
federal sector side in a coordinated fashion and the private sector side in
terms of agreeing on certain queries if you will, and then use those queries to
go after distributive data to bring back information that does not require the
sharing of personal health information.
With that I think I have gone over time and I apologize. Thank you very
DR. CARR: Thank you very much for a very rich presentation and we will
study in detail your slides. You have given us great food for thought. I want
to hold questions and ask Mike Rapp to pick up from there.
DR. RAPP: Good morning, thank you for the opportunity to present today on
the subject on value-based purchasing, combining costs and quality. I heard
reference to an interesting, pragmatic ways forward. I am appreciative of the
kind of conceptual framework that we had in terms of practical ways forward.
That is what of course we have to deal with at CMS because usually we do not
have too much time to implement many ideas that come forward, many of the ideas
from the private sector find their way into legislation so then it is our job
to implement them.
I am going to kind of try to go through a variety of things fairly quickly
just value-based purchasing conceptually. What our implementation is currently.
I want to focus on outcome measures so when we talk about combining costs and
outcomes so where are we on outcome measurements. I want to review
consideration and the use of these. I want to spend a bit of time on our thirty
day mortality measures and our thirty day readmission measures and reflect on
I will do a little disclaimer. When you heard that CMS might be
experimenting on measures, we do not like to think of ourselves as
experimenting but implementing or engaging in pilots and so forth. In any
event, what value-based purchasing means to CMS is transforming Medicare from a
passive payer to an active purchaser of higher quality, more efficient health
care. The various different concepts that Joachim talked about I think are
interesting and I do not disagree with him but we kind of use the term of
value-based purchasing and value a little bit more broadly. It refers to a
variety of tools that we use including measurement payment incentives, public
reporting, conditions of participation, coverage policy, QIO program. We have
pay for reporting, pay for performance, gain sharing, competitive bidding and
so forth. Currently our only program authority to pay differentially for better
quality is in ESRD. When MIPAA Legislation passed in July of 2008, Congress
also required by 2012 that we implement value-based purchasing for in-stage
They did it in connection with bundling payments, bundling payment
particularly with erythrocyte stimulating agents so those previously were paid
for separately and now they would be bundled so that the incentive would not be
any more to use it would be to not use it because you are getting paid for it
anyway. This idea then of value-based purchasing or measuring the quality and
having a differential payment of two percent based upon differential quality is
first authorized in that.
There is broad support for value-based purchasing. We implement it in many
ways in terms of demonstrations in particular. We have our pay for reporting
program; once again I will not go into detail on that.
In terms of measuring value, the basic concepts here are bringing together
cost and quality. In terms of the quality side of it we have different measures
available to us. Outcome, I think that most people would describe what we want
to really use if we can. Process measures, but then also other measures like
experience of care.
Potential costs to consider well, you can consider all the costs of care
for a particular beneficiary or you could consider the costs associated with a
particular professional. You get into all of these attribution issues, what is
fair to hold different parties responsible for?
Then what is your level of attribution? Are you going to focus on
individual doctors? That is always a challenge. I personally think we should
spend less time on that and more, at least in an initial start at groups of
professionals or even higher levels. Are we going to have accountability of
facilities, professionals, and how will we allocate among the facilities and
the professions. What kind of time period are we talking about? We have often
talked about episodes of care. Those are a little bit hard to implement I would
say in terms of identifying when it starts and when it ends. You might have
multiple episodes or you could relate it to a particular health care event. For
example, hospitalization which is a much brighter line and it deals with a
subject that is closely associated with the Medicare program and the costs
involved in Part A of course would be hospitalizations or post acute care, that
sort of thing.
Considerations in measuring value or efficiency we need to integrate the
quality and cost. It is not resource use alone as Joachim talked about. We do
have resource use reports that the Congress in the MIPAA Legislation required
us to share with doctors but those are not publicly reported and there is no
authority to do that, those are confidential reports that are shared but we do
not look at that as a value per se. It compares on a relative basis the use of
different things. Another way of integrating cost and quality are never events
and appropriateness criteria. You would say the cost is just not justified. In
that case, you can integrate them.
What are valid cost measurements? You have to have a valid cost measurement
and analysis. Think in terms of the same population. What are the scope of
costs that you are going to consider? What is a perspective, as Joachim talked
about the patient, the professional, the provider, the payer? Of course, we
always have to think about the adverse consequences in anything that we do. You
push on one part of the balloon and something else goes out. The health care
system, it definitely when you make payment adjustments or even publically
report information they adapt to it and try to deal with whatever is put into
place. What is the proper attribution? We will talk a little more about these
as we go further.
So as I mentioned, you could use various on the quality side of it, you
could use different measure types. There are advantages to each of them and
disadvantages. On the process measures in particular, what we have found is
they rapidly become topped out. At least if you have a measure which has to do
with do a certain thing, give a certain process of care. The ones that do not
tend to get topped out are the timing measures, that is hard to necessarily get
to optimum there.
You have to focus on the processes. You have to think about how they impact
the outcome too. With outcomes you of course, they are less available but the
advantages are significant in the breadth and scope. They are certainly less
subject to becoming topped out.
The experience of care measures are definitely good since it gives the
perspective of the patient. They also have a hard time being topped out. You
could also add structural measures to the mix if you wanted to.
I am going to go quickly through where we are with outcome measures since I
think that is where we would like to get if we can. I think most people would.
In the outcome measures used by CMS, I count them as 74 right now but do not
hold me to that number since I always have a hard time knowing for sure about
every last implementation. We have 28 in patient measures, 8 physicians, and so
forth. You see them and I will go through them quickly.
In our RHQDAPU and QIO Program we have paper reporting for hospitals, they
are given a two percent incentive or they avoid a two percent penalty I would
say if they report the measures to us. In the RHQDAPU Program currently as far
as outcome measures we have a 30 day mortality for AMI, heart failure, and
We have implemented selected AHRQ measures of selected medical conditions.
We use mortality and then AHRQ measures for surgical conditions and procedures
as you will see listed here, other complications of surgery.
We have our readmission measures which are outcome measures but we look at
them as a form of efficiency measure with the idea if you avoid readmission
then you are avoiding an unnecessary cost or you are avoiding a cost.
There is an all patient readmission rate that we use in our QIO Programs
and in our transitions of care theme which I will tell you a bit more about.
Then finally, we have intermediate outcome measures which is 6:00a.m.
A set of outcomes for the hospital, we include some outcome measures in the
Premier Hospital Quality Incentive Demonstration. Mainly they are inpatient
AMI, CABG, heart failure, have some outcome measures with regard to hip and
knee replacement. There is a plan as the Premier Demonstration continues on to
add some additional measures to test them further.
In the value-based purchasing plan that we were required to submit to
Congress again in 2007, we identified a number of tentative measures that we
include in that. That included a 30 day mortality measure as well.
Other ways beyond public reporting or showing these differences among
parties are our Hospital Required Conditions Policy. In the Deficit Reduction
Act, the Secretary was required to identify high-cost, high-volume for both
conditions to the extent that they were not identified through or present on
admission. Indicators having been present on arrival at the hospital then those
complications that occurred would not be paid. The basic payment would be made
but insofar as if there was a complication. Foreign objects retained after
surgery, air embolism, blood incompatibility, Stage III pressure ulcers, falls
and trauma. These things were felt not to – they were on the never event
list of NQF most likely – so the things that should not ordinarily occur. So
that was the test. The legislation did not require it to be something that
should never occur but ordinarily should not occur. They were put on the list.
What happens here is insofar as they do occur as complications or secondary
effects of the basic diagnosis that additional amount would not be paid.
Other items, manifestations of poor glycemic control, catheter associated
UTI, and vascular catheter associated infection, surgical site infections, deep
vein thrombosis, and PE with total knee replacement and hip replacement. So
this is an example that no one would ever suggest that a DVT/PE should
“never happen” and certainly that would be the goal but you could
hardly say that. Nevertheless, through rule making that was put on the list and
insofar as this is a complication of total knee replacement and hip replacement
there would be no additional payment given to the hospital. In a way this is
combining cost and the quality in one but in somewhat of a different way that
Joachim was discussing.
In terms of what this is going to do for the Medicare program and save it
from going bankrupt as you see here the estimated savings here are $21 a year.
That is not going to save the Medicare program or probably any other coverage
for health care. Nevertheless, that is where we are on that. Some of it has to
do with how good the present on admissions indicator works. How much time
people spend demonstrating that they do not have it.
This was extended for national coverage determinations for hospitals and
physicians so that in fact, not only is there no coverage for the complication,
there is no coverage at all if you operate on the wrong leg, wrong patient, or
wrong surgery on the patient. We do not have too many people arguing with that
but of course, if there is the wrong surgery on the wrong patient, how about
the right surgery and so forth.
In PQRI we have now some 170 measures or so in PQRI. We only have a very
few outcome measures for physicians. Diabetes, control HbA1C, LDL, and blood
pressure control that we use in NCQA measures. There is a CABG measure, deep
sterna wound infection, stroke/CVA, these come from the STS measures and they
would be collected through a registry.
Basically if you look at the hospital we have got an array of measures that
are outcome. Some reflect cost in a physician world where we do not have many.
In a physician group practice demonstration we are using those intermediate
outcome measures. We are required through MIPAA to provide Congress with a
physician value-based purchasing plan in May of 2010. So in working through
that plan certainly outcome measures would be under consideration.
Home health measures are mostly outcome measures. Things like acute care
hospitalization and following home care emergent care, whether they are
discharged to the community. Functional measures, pain medication management,
surgical wounds, complications, and incontinence, they are basically all
outcome measures. What we need there is a few process measures.
The nursing home for the long stay not a short stay necessarily we have
pressure sores, functional status, pain, incontinence, UTI, mental health, and
so forth. These are outcome measures.
In the short stay nursing home we have percentage with delirium, moderate
to severe pain, pressure sores and so forth.
With ESRD, we currently post patient survival, the amount of hemoglobin
control for ESA therapy and hematocrit below the minimum level.
In Medicare Advantage we have use outcome measures.
We have got an array of outcome measures. We have also a number of
considerations we have to think about in terms of getting data. A variety of
potential sources for that certainly electronic health record would be
something for the future. Now Joachim talked about distributive model of
getting this data because of the concern about privacy and so forth. Currently
that is not the way that we approach measurement. We collect the data. In the
hospital world, where the charts are extracted by the hospitals, all patient
data and that data is sent to CMS in the QIO program and the measures are
The advantage the distributive model has the advantages that he mentioned
getting the data together in one place, and we are talking about all of the
patient’s data on all aspects of the patient but just those data elements
for the measures themselves prevents one to first of all do risk adjustment. If
one has to deal with the issues across different settings you can bring that
data together. When you think about outcomes how are you going to get
information about outcomes without knowing who you are talking about? You at
least have some sort of pseudo identifier that does that for you. These are
kind of pretty significant issues. Certainly all payers get identifiable data.
Claims, that is the nature of claims when they talk about bringing claims
Those are very important issues. Certainly we would always want to maintain
a high-level of privacy on the part of the patients. In terms of the
practicalities, in terms of a distributive model, we do that in PQRI and our
registries in that we have, the model we have there is basically that each of
the registries calculates the clinical quality measures based upon the data
that they get from the physicians.
We have 74 registries but as you can imagine that means that all of these
registries have to work exactly the same. They have to use the same methods of
calculating otherwise you are going to end up things that you cannot bring
together. If your numerators and denominators, you have to be satisfied that
those are calculated exactly the same way and that your denominators are select
and so forth. You do not have that then you do not have any kind of data that
is comparable. Having that standardization in place I think is wonderful from
conceptual standpoint in terms of its practicality in the near term, I am
somewhat uncertain about that.
In our CMS 30 day mortality measures, these are risk standardized 30 day
all cause mortality and readmission measures for AMI, heart failure, and
pneumonia. They are NQF endorsed and implemented for the RHQDAPU program. This
is based upon Medicare claims alone. It is limited in that sense although large
portions of hospitals’ heart attack, heart failure, and pneumonia patients
are Medicare beneficiaries.
We are looking for ways to bring in other data. For example, the Veterans
Administration has contacted us and asked could we bring their data in and it
would be part of the overall thing. We have been contacted by some
representatives of health plans and asked could we bring in private sector data
for this. Again, to do that you basically have to bring the data together, you
have got to run your risk adjustment and your model, and come up with your
results. That is something that we are exploring.
You may be familiar with our 30 day mortality readmission measures but they
were developed by a Yale/Harvard team of “statistical experts.”
Harlan Krumholtz and Sharon-Lise Normand have been active on that. The measures
again are NQF endorsed and have gotten a lot of attention because they have
been published for a couple of years on the mortality side. This last summer we
posted the 30 day readmission measures. The USA Today picked up on this and
linked the data to their newspaper data server so people could search not only
on our site but through this USA Today site.
Why 30 days? Well it is a common time period that people use to look at
outcomes. It is not inpatient so it gets them out of the hospital for 30 days
from the start of the admission. We in fact, and with NQF endorsement, are
holding those hospitals responsible for what happens to them even when they get
out of the hospital and even the readmission issue. Although of course there
are a lot more parties involved in it than the hospital themselves.
Again risk adjustment, and I will show you how we display these but we have
now moved to calculating the performance rates based on three years’ worth
of data so that we are better able to show the differentiation among hospital
performance but based upon data rather than modifying such things as confidence
intervals and so forth. We use an interval estimate that is a 95 percent level
This shows you a distribution of AMI and heart failure rates. The way we
display them is this. We first of all show a diagram of something like this for
each of the hospitals. That center vertical line is the US national rate. So if
the US national rate for readmission for heart failure is 25 percent which I
think it is something like that, if at the 95 percent confidence interval
there, there interval estimate crosses the US national rate then we categorize
it as no different. If it is to one side or another it is even worse or better.
When we started displaying these we displayed them as buckets. When you
have only a years’ worth of data you do not have many in those buckets
where you can say with a high degree of confidence that are in fact different
than the US national rate. That was the way we did it the first year was
The second year we put the individual information more like this in terms of
where they would show what their point estimate is. So you will see for example
a heart failure ranges from about seven or eight percent up to – this is
the mortality, up to about seventeen percent. So now we show exactly what the
rate is but a point to recognize is whether you say it is different than the US
national average is based upon this 95 percent confidence interval.
There are a lot of things that you have seen. I know with the Dartmouth
Atlas and so forth about the geographic variations so we plotted out the
geographic variations on these different measures as well. Distribution of AMI
mortality you see here. We published something on this which basically pointed
out there was not specific correlation between the AM 30 day readmission versus
the 30 day mortality. In other words, it is not necessarily that one is better
than the other, they are better on both. There was no specific correlation.
Here you will see the mortality for heart failure. Then we have our
readmission. These are a little tighter in their distribution but still a
significant variation in – look at the heart failure and you see that the
average of the US national rate is 25 percent with a variation here of going a
little bit under 20 to over 32 percent.
I was at a party and I mentioned to somebody that they might be surprised
to know that heart failure patients, 25 percent of Medicare patients with heart
failure are readmitted within 30 days. He said, no, that does not surprise me
at all. They are sick. So you think that this grabs perhaps doctors in health
care professions like you know it does seem like we just had him in the
hospital and back in 30 days. You could have different points of view on this
and when you start getting into these kinds of issues about do you belong in
the hospital or don’t you belong in the hospital and you sort of point
this out I think people can legitimately have different reactions.
From a beneficiary standpoint, I guess we have to think about that too.
What do they think? They probably think if I want to go into the hospital, I
want to go into the hospital. I do not want somebody telling me that the
hospital does not want to do that because it has been less than 30 days and I
am going to be on a website or something. There are a lot of different
perspectives as Joachim was talking about.
Here is the distribution of readmission. I do not have the Dartmouth Atlas
overlaying this but I know that there is as it has been frequently pointed out
that Minnesota is a particularly advantageous place from the point of view of
efficiency. Well, if you look at AMI by readmission, it seems to be darker and
then if you go down to South Florida it does not seem so dark.
I have not correlated these with any other atlas. I think it would be
interesting to try to do that but the AMI readmission rates do not necessarily
correlate with the mortality and they do not necessarily correlate with other
things either. Here is the heart failure readmission and here is the pneumonia
So in any event with the 30 day readmission rates the AMI is at 19.9
percent, heart failure 24.5, and pneumonia 18.2. the goal of course is not zero
but we view it as overall the readmission rates are too high. Hopefully, this
would prompt lowering the curve or lowering the rate of readmission and all so
narrowing the distribution.
We have other hospital level measures like this that we have developed and
that have achieved NQF endorsement and would be available for implementation in
the future to us.
PCI 30 day all cause risk standardized mortality and also for lower
extremity bypass. There is interest – I think in MedPAC a few years ago
recommended building out measures in this arena.
I want to talk a little bit about moving to episodes and then I will wind
up here. Joachim was talking about the episode group or you start with some
kind of initiation and then you come to an end. The hospitalization itself is I
think – has a lot of advantages and does give you that bright line. It is
the thing that you have got to be at a certain level of illness to end up as an
inpatient in the hospital. If look at the hospitalization at the endpoint you
can then go out for some period of time and look at all of the care that takes
place and think of all of the service and all of the people involved with it.
It is an important starting point and possibly endpoint the AHRQ measures on
ambulatory sensitive conditions are designed to measure whether care –
which is likely to result in not having to go into the hospital is being
Of course, when you again as I mentioned for care not associated with
hospitalization is the challenge of defining when the episode of care started
or ended but of course we would be interested in that as well. Again, I think
it is just a bigger challenge unless one deals with the ambulatory sensitive
conditions. Beneficiaries not receiving care, I always like to remember those
because these are patients that presumably have health care coverage but they
do not see anybody so they do not have any claims and they do not have any
health record and so forth. They still may have high blood pressure or any
number of things. We probably need to spend a little bit of time on that
segment of the population.
If we use hospitalization as a starting point, I want to point out some of
the other things that we are working on that do not necessarily get a lot of
attention. That specifically is the post acute care demonstration that was
required in the Deficit Reduction Act. The idea here is in – when we talk
about data sources we go to claims or charts or so forth. The post acute care
time period is somewhat unique in that we routinely collect in the
SNFs(?)(B-1:43) and in the long-term care, every patient that is in a nursing
home has to have an assessment on a regular basis and that data is submitted to
CMS. It is that data source that drives the calculation or clinical quality
measures that we put up on Nursing Home Compare and so forth.
With that the problem with it is the instruments are different in the
different settings. We have the SNFs which use the MDS, the skilled nursing
facilities, the short-term benefit that Medicare pays for after which the
long-term Medicare does not pay for. In any event that MDS instrument does have
regular assessment in a variety of things.
We have in the home health a thing called the OASIS which does the same
thing for home health patients. Then we have inpatient rehab facilities as
well. The problem with that is you cannot compare across these settings, you
cannot compare the costs if you do not have the same kind of assessment. In
this demonstration they required us to develop a single instrument that could
be used in all of those settings that would start at hospital discharge. Then
it would follow that patient through that period of time. Then what the
demonstration is about is to look at the various assessments in terms of
functional status and other things that are included in that assessment and
then figure out the costs and compare them to that.
Interestingly, the way the payment system in the post acute world works is
it is based upon that assessment. That assessment drives the payment groupings
that are made. It is an interesting area that could possibly tell us something
about the future.
Now this instrument that is used in this demonstration is called the care
instrument. If you think about your data sources and you think about episodes
of care and you think about episodes of care that start at hospital discharge,
what would be better than an assessment that is individually made of the
patient starting at hospital discharge and carrying it through those post acute
settings over a standardized period of time. You will get not only things like
adverse outcomes but you get functional status which is a key ingredient of
what kind of services people need or do not need and if we carried that
We spent quite a bit of time thinking about this in some of the previous
HIT work that was done with the AHIC was the recommendations were made for a
standardized dataset which had in mind this care tool. As we go forward
possibly electronic health records could be – the inpatient record could
include these items in there. If that was the case then that would be the start
of the assessment which could be carried on further. This would also think
about coordination of care. It is a big issue that people have. How can we
coordinate care? You cannot coordinate care if you cannot share information. If
you have it in electronic form and you have a standardized set then they share
it in the post acute setting and so forth. It could deal with all kinds of
issues like medication reconciliation and so forth. You can do that and at the
same time you can couple the costs with it. You know what the costs are and you
can add the overall costs. Then you can go on to such issues as okay, who is
responsible? To what extent are we going to hold the hospitals or are we going
to hold the professionals?
One other thing that we have done, the care transition’s QIO theme in
the 9th scope of work deals specifically with the issue of
re-hospitalizations, both looking at conditions and geographic region. In kind
of an interesting way of approaching this in terms of attribution is that we
attribute the impact of an individual, professional, or provider based upon
proportion of transitions for the beneficiaries in that geographic region. So
just to go over that again, you have patients. They live in Washington, D.C.
They are Medicare beneficiaries but they do not necessarily get their care at
Providence Hospital or GW. They may go over to Johns Hopkins for example. The
doctors, where are the doctors located? What we have done in this is we have
linked up the beneficiaries with the providers of care. We weight basically
their impact on the care transitions based upon what extent that their care is
representative of transitions that the beneficiaries experienced.
In any event we get into these attribution issues but it is something that
we are working on through this and we will have the output of the scope of work
and measures and so forth. The intention here is to promote improved quality of
care within the setting, this for example heart failure, those things that
prepare the patient for going home knowing and understanding medicine and so
forth, coaching, following up with them afterwards. It promotes quality of care
within the setting, improves coordination processes, and community involvement
In conclusion, there is certainly active work being done I think to help
develop a value-based purchasing framework at CMS. I think a whole lot of
things potentially to draw on. I think in general, our preference would be to
the extent to which we could use them, outcome measures because of a variety of
benefits of that.
We have a lot of work to do in terms of getting measures in the positions
setting and we always are going to have the challenge and the physician setting
of the small numbers which leads I think us to want to at least tackle first
the group. This is my opinion. Everything that I say here is my opinion. I am
not speaking for the Agency or the US Government or anything else. It is my
Anyway, the idea of considering the group or other level of attribution and
I was kind of talking in that post acute setting you would see that when you
get in the hospital and things happen. You could try to weigh what degree of
involvement different parties have but you could perhaps have an aggregated
cost and an aggregated outcome at that point. Perhaps the attribution issue
would be significant but the measurement would be made less difficult I think.
That is basically a whirlwind through a lot of things that we are working
on and I would be happy to discuss it further and to get your ideas.
DR. CARR: Thank you very much. That was spectacular. Do you have some time?
Can you answer a few questions? I know we ran up against the time with Dr.
Roski also. Maybe if you could come up to the table. Harry, I think you had the
MR. REYNOLDS: Michael thanks again. The question I have – well first,
you have got a lot of good and focused work going on and we appreciate it. You
talked to us in a Meaningful Use Hearing. We heard yesterday from lots of
different people that everybody is looking at – you referenced some of
them today that you are picking up from other people. With Medicare being the
largest payer, Medicare moving forward at this speed, meaningful use being on
the table, why aren’t those measures just being the ones that are
selected? I know it does not cover the whole population. You do not deal with
every set of – and if something comes out as meaningful use is CMS going
to adjust or –
DR. RAPP: You are saying why don’t we use measures that we are using
in other settings for the meaningful use clinical quality measures?
MR. REYNOLDS: No, what I am saying is that you just listed how CMS sees
value and quality.
DR. RAPP: Well I went over a lot of things that we are doing that we might
MR. REYNOLDS: I guess all I am trying – as we recommend for the entire
ecosystem some suggestions and some considerations and so on, the fact that all
of these things are going on when we come up with one idea on how to do
meaningful use or there are sets of measures on how to do meaningful use, one
of the concerns that we heard yesterday is that payers be together at least in
some way. Just trying to understand as you see the rest of this going on how do
you see CMS playing in that positionally, functionally? If the measures that
come out are different do you pull back some of yours, kind of that statement
of positioning of that.
DR. RAPP: So first of all I cannot speak to the future. I can speak to the
past and the present. In terms of what is going to be supposed by the Secretary
in terms of the clinical quality measures that would be required for meaningful
use or for the other measures that is up to a lot of internal discussion and
clearance and sign off by the Secretary and so on and so forth. So you really
do not, even though I have some ideas of things that would be discussed one
does not really know until it is out. If I knew I could not tell you and I do
not know because nobody knows until the action of the agency takes place.
Going back to the subject of electronic health records which I think is
really the subject here and the reporting of clinical quality measures from
electronic health records and using electronic health records, we have been
interested in that for some time. We reflected most specifically in the
Physician Quality Reporting Initiative in that from the very beginning even
though the only way we could implement it in six months was to have
claims-based measures. We indicated that we were interested in the future of
moving to electronic health records and submission. We have followed up on that
by developing electronic health record specifications and seeking to work with
the standards organization and so forth so that we do not get out ahead of them
and that we are in sync and so forth. It provides somewhat of a foundation from
which wherever the Agency decides to go in the future with regard to electronic
health records at least that is somewhat informative I think at a minimum.
We have proposed under PQRI that in 2010, we quality those electronic
health record products that have sought to obtain such qualification and
testing it and if it works out then we would allow electronic health record
submission of clinical quality measures for PQRI in 2010. So again, that kind
of thing can inform a lot of work in the future.
In the hospital arena, we did the same thing in that we sought to have the
electronic specifications for a set of hospital measures developed and those
have just been posted on the HITSP website I think recently for public comment.
That provides again a foundation. There is a lot more work going on as well. I
think that is about as specific as I can be but it is not like high-tech
legislation is of course something that is brand new and providing a lot of
incentive to move in that direction. We have been interested in moving in that
direction all along.
What the final framework would be – there are issues that we have
talked about here in terms of distributive models and so forth. There is a lot
that has to take place. There will be regulations proposed that will be subject
to comment and no doubt the regulations will probably change somewhat based
upon that comment and also the input that has come from the advisory
DR. CARR: Dr. Roski I know you are time constrained. If there were a
question could you answer it or do you need to go? Was there a question for Dr.
Let me let you make your timetable then. Larry it is back to you.
DR. GREEN: More questions for Dr. Rapp, one about episodes. The question is
either your thinking or whomever you choose to represent or not represent, that
is all I can do. You used the word grouper and then you also discussed the
difficulties in deciding when they begin and end. There is an intersection this
committee, well not the whole committee has heard the last couple of years
relating to episodes about the challenges of defining but there is a core
intellectual distinction between an episode that becomes a reliable definition
when a computer reruns the same program again and drops the things in the same
bucket. Then there is the definition of episode that clinically meaningful. We
are interested in meaningful measures here and what is your thinking about
whether or not CMS gravitates toward clinically meaningful episodes versus a
statistical aberration that can be used?
DR. RAPP: Well I think that we want everything we do to be clinically
meaningful. We take advantage of the work that is done in the private sector
and brought to our attention. We rely on endorsement of the National Quality
Forum rather than going off on tangents of our own. But with regard to the
episode groupers, I am not the one in charge of trying to implement those in
the Agency but I do know of the challenges that they have presented in terms of
trying to define these episodes. I think that they have to do with you are
looking at when does an episode of a heart attack start? If you are not going
to start that at the hospital where would it start or other more less dramatic
acute care events. An episode of asthma, asthma is a chronic illness.
I think those present significant differences but in addition then what
costs are you going to throw in there? Are you going to throw in just the cost
that that doctor is specifically responsible for? Are you going to put in that
they came to me with a sore throat and pretty soon they ended up in the
hospital and I am the only doctor that they saw but they ended up having
pneumonia and all sorts of things and I am all of the sudden responsible for
I think the attribution is just as big of an issue as when this episodes
starts or does not start. That is why I think that those things will be
highly-challenging. Those are being used right now mainly and really only for
these confidential feedback reports. Congress has not authorized us to put up
things like that on a website for public reporting and so forth.
By contrast as I mentioned, things like readmission rates and those kinds
of measures that we have gotten through NQF were prepared and we can go forward
DR. GREEN: I do not recall us hearing any testimony or learning anything in
the last couple of years about the attribution issues in the episodes but in
terms of when they start and begin we have heard testimony and seen examples
where you ask the clinician, well this is a newer episode and when did it start
and when did it end and they will tell you. It is very reliable for asthma,
heart attacks, and all sorts of stuff. I just want to get that on the table.
The other question is –-
DR. RAPP: So you are advocating I guess in favor of those?
DR. GREEN: Well I think episodes seem to me to represent the insertion of
time into care.
DR. RAPP: Right, but you need to have that episode start – does it
need to start – the only issue I have which is I am not sure again, I am
not an expert on the episode groupers just from what I have heard there is a
lot of overlap on these things. It is not necessarily as uniform in terms of
that but be that as it may, we are interested in the time issue and in interest
in struggling with this issue it does just seem to me that it is easier to
start with something like hospitalization and go for a period of time
First of all you are dealing with the most expensive kind of care. If you
are in the hospital you have already spent quite a bit of money, post acute
care. There are all kinds of variations that take place here. People go into
the hospital, they go the nursing home, they go back in the hospital, they do
this, there are a variety of things that happen or the RTI wrote a report on
I do not really disagree with you. I just have not seen how we can
effectively implement those readily.
DR. GREEN: My other question was about home care in transition. This goes
back to yesterday afternoon again, about trying to understand our underlying
assumptions about the constructive care, the model in which we are operating.
So all of the home care slides that you had are really about home health care
providers as I read them, right? That is of course not the way most home care
is provided. Most home care is provided nonprofessionals and family members and
friends and neighbors and most of it is organized in other ways without
professional, certified home care providers.
Then it goes to the transition issue too. Both in your presentation and in
many others it is if we are assuming that transitions will incur in boxes.
There is Box A, there is Box B, there is Box C, and there is Box D and let me
tell you, the only person that is going to cross those boxes is that patient.
If we are lucky the patient may manage to cross them. There is no assumption
that there will be an integrating of some sort. This flies in the face of what
we are seeing from our work in the National Health Information Network and
stuff about the portability of digitized data and that sort of stuff.
We have heard stories of doctors rounding in rural Wisconsin while sitting
in Beijing with a $29 camera and Skype and a laptop. Are we – well I
started to say the word condemned that is a little pejorative isn’t it?
Are we stuck with the notion that we have to have measures of transitions that
presume that the care providers in the different settings are just going to do
their box and there is going to be a transition measure that has to be invented
for each one of them or is it possible to plan for a future in which – see
this relates back to the episodes, where the whole episode of care gets
integrated and connected?
DR. RAPP: That is really what I was talking about with the care and
assessment instrument. It is not limited when you talk about patients going
home we consider that a transition too, home without having involvement with
the home health agency. Those measures are for home health agencies admittedly.
DR. GREEN: Is the assumption that the doctor that took care of them in the
hospital will not see them at home or will not know about –
DR. RAPP: Well no, the doctor is part of the network as well.
DR. GREEN: Do you get my point? I mean we are going to start certifying
hospitalist doctors whose responsibility ends when they walk out the door. When
we moved towards patient-centered discussions we heard no one testify that
patients would not like to see their care coordinator to cross settings by
someone they know, who they are not a stranger. Are we headed towards these
measures? These measures can be so powerful. I mean Harry’s question was
really about when an 800 pound gorilla says I am going to measure this that is
gravity as far as I am concerned. We can discuss all sorts of things but that
is gravity. Will Medicare move towards transitions as if – well, they are
clean boxes, we have got them defined, this is where we are going to look at
them or how to cross the boxes?
DR. RAPP: Well I think that the point that you are making is to if I
understand what you are saying, is to focus on the patient and look at it as a
patient perspective and not look at it strictly speaking from the provider
setting. Just by way of explanation and not in defense is what happens is when
Congress passes laws they have paper reporting progress for doctors, for
hospitals, for home health agencies that is the way that they pass the
authority for us to do things. That does not mean that we should not look at it
from a patient perspective. That is the advantage of an assessment instrument
such as a care instrument.
Another thing I should mention in the Physician Quality Reporting
Initiative one of the things that we are interested in is that the doctors
participate in it since it is a voluntary thing so we can keep track of that. I
also ran some data recently to look at it from the patient perspective. Do not
look at the measures in terms of doctors and how many of those that reported
but look at it from a beneficiaries – how many beneficiaries did those
measures apply to? How many reports did we get with respect to the
beneficiaries? We have some information like that that we will probably be
I would certainly agree with you that the perspective is the patient. You
can decide that when a patient with pneumonia came in the hospital they got a
flu shot but we are interested more in what did the population of the
So I think you need to look from measurement, one needs to look at it from
a population level. I agree with you 100 percent. When you deal with
accountability though you necessarily have to deal with the different
participants in that system of care since that is what we are trying to
measure, the care. We want to know what the ultimate outcome for the
beneficiaries was. How they were impacted? Are the patients getting their flu
shots? Are they getting this kind of care or that? What is the mortality rate
for heart attack overall but then we want to dice it a little bit more and say
okay, what is the role of this particular provider in the health care system
and how well are they performing?
DR. CARR: Thank you, Bill.
DR. SCANLON: This is linked to Harry’s question about Medicare and
meaningful use and I guess kind of linked to what Larry was saying with how can
the gorilla be most useful? I guess I see two models in what CMS has done in
the past in terms of possibilities. One is MDS OASIS model. They actually
started not with the idea of what do we want – reporting done but they
actually started with the idea of what should a provider know about individuals
and start to provide good care. The MDS assessment goes back to 1987 long
before it became an issue of payment policy or nursing home compare or any
other kind of quality. What it has is that CMS or Medicare receives the
information and then it has got all kinds of uses that can be made versus I
guess the other model I would think of as the measurement approach. We decide
to hear some measures that we want and we ask for the relevant provider to give
it to us but then tomorrow we ask for another measure.
We keep doing this process and I guess I feel like over time we would be
better off with the first model in terms of getting basic information about an
individual that is relevant to the provider that they are seeing at that moment
and then having sort of a data capacity that we can ultimately build measures
on for the future. If Medicare were going in that direction say, this is where
we would want to go then it would be a powerful force. It would be very
meaningful use of EHRs because they I think are the tool that allows this to
happen with very minimal cost to the provider.
DR. RAPP: I think that if what you are kind of laying out and I mentioned
the care instrument, currently hospitals are required to do discharge plans but
they are not required to do an assessment discharge. The first assessment is
when they get to the nursing home if that is what it is. If you did the
assessment of hospital discharge and it was available to the physicians and to
the nursing homes or home health agencies, ESRD facilities, whoever it is then
you would track that same basic standardized information across time.
I agree with you. Then you would get to the doctor as well.
DR SCANLON: Well we actually need something at admission and we need
something from the encounter with physicians.
DR. RAPP: I think the assessment of discharge does not mean that you do not
get information about what happened in the hospital. It is merely – it is
the time point that the information is provided.
DR. TANG: Thank you for a comprehensive review of all of the stuff that is
You mentioned PQRI in 2010; CMS will accept directly if I heard you right,
the data that physician practices are reporting on?
DR. RAPP: I said that we had proposed to do that. The rule has not been
finalized. It is in the comment period right now. It will be finalized about
the first of November depending on whatever the Agency does. I cannot tell you
what the final result will be but what we proposed is for a set of 10 measures,
that we would accept those directly from EHRs for vendors that had demonstrated
that their product will do that.
There was a set of vendors that self-nominated in early 2009 so it is only
that limited set for 2010 assuming that the testing works. Just like the
registries we wanted to make sure that the registries could submit the data in
a proper format and so forth. Then we could take it and make sense of it.
DR. TANG: So you would get it directly from the end user or you would get
it from the company who had queried the end user?
DR. RAPP: One or the other or both.
DR. TANG: So you are getting the raw rates of numerator and denominator
rather than the G-codes.
DR. RAPP: The specifications are based on clinical parameters not –
like it would be the value of a hemoglobin A1c. It would not be CPT such and
DR. TANG: So does that demonstrate the capability to receive that kind of
information directly from EHRs for testing through their intermediaries for
quality measures in the future, well maybe concretely, practically for
meaningful use in 2011. Does that mean that you, CMS could receive the kinds of
measures that were in the grid?
DR. RAPP: Well, PQRI is a lot different than the high-tech and what has to;
it is a matter of scale for one thing. PQRI doctors primarily submit through
claims or they do registries. This is just a little, I look at this as a small
project that some EHR vendor products may be able to be accepted. It is not the
kind of comprehensive infrastructure that is necessary to deal with the whole
In the high-tech legislation there is only one way to you can send it in.
You cannot say well, I will skip the – you have to use, according to the
statute, use the electronic health record. In PQRI you can submit it by claims
or anything else.
DR. TANG: So let’s say for the vendors that you have tested or
certified or qualified –
DR. RAPP: We have not completed the testing yet – that may end up testing
that works –
DR. TANG: That could potentially scale – I mean what am I missing?
That could scale as long as you could test all of the certified vendors in
fact. You could call it the meaningful uses. Using a certified vendor we could
create criteria that they would fill the capability of transmitting to you. I
am just trying to see if that is –
DR. RAPP: I think I will leave it to you to draw your own conclusions. I
just want to describe what we have done and what we have proposed. Obviously,
that is a possible building block on other similar activities. I do not want to
mislead you or suggest to you that would be the way that the Agency would see
DR. TANG: So am I correct that that would be a pathway that if we could
scale and all of these other factors, if all of the vendors could be qualified
by the same way your PQRI is constructed that is conceivable?
DR. RAPP: Again, I would rather you draw your own conclusions from it. I do
not want to be interpreted as forecasting the future.
DR. TANG: I just want to know if I am on the right track.
DR. MIDDLETON: Thanks very much for really detailed and comprehensive and
very insightful overview. I only have one question I guess. I sort of struggle
a little bit with our current approach to snapshots of quality. Transitions of
care and episodes of care certainly reflect a different dimension if you will.
I am beginning to think of the temporal side of clinical decision making. The
clinician who is looking at the diabetic before him will say well geez, what is
the velocity of this disease? What is the vector of this disease if you will
because that actually is as important that diabetes log and all of the
synthesis that comes together?
I am wondering how can we introduce this idea of the velocity of disease
into measurement and can we supplement a snapshot or even bounded episodes of
care assessments where a broader notion of a real lifetime perspective on the
velocity of care? After all, all disease begins at birth. When does the episode
begin? Well, somehow, somewhere along the way. I wonder if there is a longer
view, if you have thoughts about measuring this notion of the velocity of care.
I think that the transitions could actually be expanded to cover a lot of it if
you were inclusive of your definition of transition or broaden the notion of an
DR. RAPP: I do not think there is any limit to the timeframe. So far with
our 30 day readmission and mortality we are at 30 days but you do not have to
limit it to 30. You could go to 60, 120 or beyond. We can measure it at the
population level. I think that is the first thing. What we can do at the
population level is a lot easier than what we can do when we try to hold
somebody accountable for that. The nature of value-based purchasing is you want
to pay for better quality. We want to have better quality and that we could
just measure whether we have got it. What are the outcomes that the patients
are experiencing? We could figure out ways to do that.
The next step though from a policy standpoint that seems to be of interest
is to somehow pay doctors, hospitals, and others differentially based on their
quality. That is the challenge and that is why you have, I brought up the
attribution issue that is pretty key. To what extent you have to weigh it, to
what extent are they responsible for it. I have heard talk in Pennsylvania I
believe implementing the AHRQ Ambulatory Sensitive Measures but toning it down
and saying that you cannot be completely responsible doing it at the plan
level. There are all kinds of issues but I think that attribution and the
accountability are major policy issues that go beyond just trying to get a
sense of what is going on at a particular geographic region.
DR. MIDDLETON: One quick follow up because you did not say yes or no. The
flip side of this is not only the doctor facing side or the organization or the
payer side but the flip side of this is the consumer facing part. What is a
roadmap if you will for quality care for my mom to understand? What are the
guardrails in which she can feel I am having safe care versus somehow I am out
of bounds. I need to see my doctor or see another doctor. What are your
thoughts on sort of the roadmap for quality from a very consumer point of view?
DR. RAPP: Personal views, I think that the patient definitely has to be
involved in this equation. Being a physician myself I know that most medical
care comes because the patient decides that they are going to go see somebody
and they present whatever problem they present which could be big or little.
Then the way that the health care system works is next thing the doctors will
go see this person and go see that person.
I read an article recently about somebody coming into the emergency
department room with herpes shingles of the eye and the doctor went in and I do
not know if you read this but it was fascinating. I know what is wrong with me.
I have got shingles. I need an antiviral and some steroids and boom I will get
that pain medicine and I will get better. He goes into the emergency department
and the emergency physician says, I agree with you 100 percent that is exactly
what is wrong with you but you know it could be something else. Maybe we should
have an ophthalmologist see you and the neurologist see you. So the doctor
patient said, well okay, why not? So pretty soon the neurologist comes in and
the ophthalmologist agrees with the diagnosis that is right. The neurologist
says, well why don’t you get an MRI? They get the MRI read and say maybe
it says something in the cavernous sinus. We should get a CAT scan. They get a
CAT scan. Well, the CAT scan is normal, maybe we should repeat the MRI. The
next day the final reading comes back well, that initial MRI was negative
For those of us that have been in health care I think that we can
understand how those things happen. We talked about preferences and values and
so forth but what is it in the system there that would counter that? If the
patient says well, I do not want something terrible to happen. I do not want to
have a stroke or whatever. The emergency physician says well, there is no
benefit to me missing a diagnosis and I can be sure if I have got my coverage
from these consultants that no matter what is wrong with him the neurologist
agrees with me and the ophthalmologist says I am solid. The patient said –
These are the kinds of conundrums that I think exist that if you do not
have the patient involved and the only levers that you try to work are the
doctors and the hospitals and so forth I think that from my own view you are
missing something and it will not ever get where we want to go.
DR. FITZMAURICE: You have a large portfolio of related quality measure
programs and demonstration. I really appreciate you coming here and telling us
about them. It is almost just overwhelming.
My question is what have you learned from your quality measure programs
that helped to inform the value-based purchasing and to guide the data
collection. What causes the hospital or physician group to improve its quality
of care? Is it the confidential reporting? Is it the public reporting, is it
not paying extra for never events? What causes change and what have you learned
that guides your data collection?
DR. RAPP: Well I think the public reporting has a very significant impact.
Hospitals will do whatever they can to try to improve that, both the process
measures, the mortality they are kind of probably – they complain we are
not telling them exactly what to do. Well, it is not our job probably to tell
them exactly what to do. We can just tell them that their rates are different
in terms of outcome.
Process measures, I think that we have learned that they get rapidly topped
out. People figure out how to do them but that does not necessarily have that
much to do with the overall outcome because there are too many other variable
For physicians we have not gotten into the public reporting. Everything is
confidential so far particularly at the individual level. I know that doctors
are very concerned about it and I do believe that is an issue. I think there is
a lot of reason to be but as a physician you have got your individual license
and reputation and once that is gone you do not have anything anymore. I think
on the physician side we would like to get the public report quality data but
we, it would seem to me that one would want to start at the group level so you
do not adversely impact the individual’s reputation and also the numbers.
I think public reporting is a very important driver. I do think that there
has been a tremendous change in the views of the physicians and others as to
whether that this is something that should take place and more accepting of it.
As far as how you can tie the money to it I think there is a lot more
willingness to publically report information than there is actually to tie the
money to it. Most of the proposals that you see in terms of saving money relate
to well people do not report data and they get a disincentive or there are a
bunch of neutrals. What we have not really figured out is that if you get these
quality measures to have improved performance and so forth will that actually
bring down the cost of health care. We have not figured out that one yet.
DR. CARR: Thank you very much. It was very much appreciated. We are going
to break now for lunch and we will resume at 1:30.
A F T E R N O O N S E S S I O N
DR. CARR: I would like to thank you for indulging us in our delay and
invite you to step forward to help us understand better meaningful measures of
integration, population health and health status.
DR. HARRIS: Thanks for having me today. I am not really sure who called me.
I kind of pondered what value I could bring to you all. I have thought that
perhaps there are a couple of things that those of us who are in the Healthy
People business might have to offer. I hope so. I think in the end what we have
to offer is a partnership over the next several years as we learn together
about how to measure the life and the health status of Americans so that we can
join in understanding how public health, population health, and clinical
outcomes might be viewed together.
With that thought in mind, I am today going to talk with you today about
the processes, some of the processes involved in Healthy People. A little bit
about the conceptual framework that has been used and is being used for 2020.
Then I am going to talk with you about the work of our group representing the
health communication and health information technology objectives development
part of Healthy People. Then a little bit about our conceptual framework
because we are I think of a number of the topic area in Healthy People some of
the most directly related to the kinds of things that you all are considering.
Although, there are other groups that I would encourage you to discuss, or have
these kinds of discussions with including the Access Group which is chaired by
Tricia Trinite at AHRQ.
I am going to start by telling you a little bit about Healthy People. I
assume you have heard about Healthy People. You may not know exactly what it
does but I am going to talk with you about it from the advantage point of
another effort to try to capture the essence of health at a national level and
a sort of set of lofty terms but then get down to the rigor of developing
measures, objectives that are measurable. It is that loftiness but grounded in
rigor that I believe can offer you all some lessons learned from what we have
been trying to do in Healthy People 2020.
First of all I want to say Healthy People is in its fourth decade so it has
legs. Each ten years is really an opportunity to rethink what the priorities
are, what the goals are, and obviously what the objectives will be for the next
decade. It is something for you all to consider when you think about how
overwhelming your own mission is to come up with meaningful measures. You do
not have to get it right the first time because it is an iterative process. We
are just doing the best that we can on each of the stages of our development.
Health People 2020, I am going to give you a little bit of insight about
how it is going to be different from the other decades because I think that may
be the most useful for you all. If you have questions about the past I will be
happy to answer them.
Like the rest of 2020, the goals or the mission is pretty much the same.
Instead of looking at health or health care at a micro system level it is
really trying to look at it as national, nationwide kinds of objectives. That
is meaningful in that it is not a federal set of goals. It is an effort to
bring from the grassroots, the best ideas in the country about what the
mission, what the goal, what the framework, and then ultimately what the effect
should be. So they are national objectives not federal objectives and that is
the engagement of the stakeholders. We have a forum just like you all do to
help bring in public comment. We also have a pretty aggressive effort to bring
public comment to the table.
Here is the timeline for 2020. As you see we are just about to launch the
public stakeholder period after all of the objectives that have been reviewed
by the federal interagency working group have been cleared. Next week these
almost 2000 objectives will be launched out for public review. After that
period we will have another opportunity to get public comment about those that
we have developed and do one last round before it is cleared through the HHS
process. It will be released next year, probably next December, maybe December
31st of 2010 to launch Healthy People 2020.
The 2020 Program is unique in all of the different versions of Healthy
People in that it is not limiting the sets of objectives to conditions. That
was not an easy thing to come about. For the first time, Healthy People is
taking what we lovingly call an ecological perspective. That perspective is
reflected in the context in which health occurs and where health takes place.
That is not just in the clinic but you will see that the health services are
one of a number of settings in which health happens or does not happen.
The determinants that have been recorded for Healthy People to date, and
this is still a little bit up for grabs, there are some opportunities to
redefine this a little bit. For the purposes of our discussion this is where we
are in our deliberations. These determinants are health services, biology and
genetics, individual behavior, social environment, and physical environment.
All of these have in some way some impact in the health outcomes. The goals
here as they are as of today, and I think they are pretty stable because they
have had a lot of input, is really primarily about obtaining high-quality,
longer lives in which people can get healthy and stay healthy.
I offer that to you as a goal that you all might want to consider too. I
heard this morning, one of the reasons I wanted to come is to kind of get what
is your language, what are your assumptions? Of course, as appropriate there is
a lot about illness, conditions, and how do you do the best job you can to
improve the quality of the way those conditions are managed?
Healthy People is really about preventing those conditions from happening
in the first place. So you will see some different assumptions and different
kinds of processes that are measured for achieving those sets of goals. In our
particular objectives in health communication and health IT, our lofty goal is
really to move us more toward a learning system. We view communication as a
dynamic in which there is a series of interactions in which feedback is an
important part of those interactions whether they are at the interpersonal
level or at the system level, organizational, or state or federal level. We
have we hope a patient-centered view but more specifically an interactional
view of health in the discussion that we are going to have today, and with
I am just going to tell you a little bit about the objectives that we in
our group have developed and how they fit into this ecological model. Before I
do that I will just give you a glimpse into the way we work in Healthy People.
Each one of the topic areas whether it is heart disease or access or health
communication and health information technology have working groups. Ours is
made up of three members of – three of us make up the leadership of this
working group, CDC and our office and Disease Prevention and Health Promotion
which I lead with our team and health communication and the Office of the
National Coordinator. We have tried to create an integrated view of health
communication and health IT. The idea is that the processes, the interactions
in a health care delivery system or any of these contexts are supported by the
tools that the health IT offers. That is kind of our orientation going into
I am going to go quickly through the different kinds of – this is the
list of individual behavior systems and the other kind of determinants in order
to get you a little insight into the objectives that we have chosen to reflect
important health communication and HIT objectives. I am going to tell you a
little bit about their measurement.
Measurement across the board for Healthy People is pretty much survey data.
I do not know if we have a lot to offer you on that. I am hoping that what we
have to learn with you together is more of the conceptual basis of what we are
Some of these have already been mentioned this morning. Health literacy is
really an important element for us to measure. We have measured it for Healthy
People 2010. That measure is no longer available so we are fishing for other
measures for ways that we can measure improving health literacy. We would love
to do that both from the population basis perspective but also from the systems
basis like what are the systems doing to try and make it easier for people to
understand the health and navigate the system?
A second objective that we associate with supporting individual behavior is
the increase, the use of electronic personal health management tools and Paul
you and your group helped us develop this perspective. For one thing, you all
helped us really see that it is not about the technology. In other words,
personal health records in some language but it really cannot be about the
technology. We were really pleased in our office to learn that the people at
ONC see it that way too even though they are technology focused. It is really
about the interactions and about how the tools can support those interactions.
As you will see we are very much interested in how this impacts the end use or
the patient or the consumer, the citizen.
The third one is about the quality of health websites. As you know lots of
people are going to the web for health information. We think that getting
accurate, transparent, easy to use, accessible health information from the web
is a meaningful way that people can get health information. Having access to
the internet is another important part of that access.
Social environment, that is just friends and family whether it is real
geographical areas or it is in virtual social networking, we know that if you
are isolated your health status suffers. The flip of that is also true that if
you have someone who you feel supports you in your health it makes a dramatic
difference. We want to measure the degree to which people have social support
and are being helped on that with HINTS which is the Health Interview National
Trends Survey from NCI.
On the physical environment there are lots of objectives that are I think
more relevant from a number of other topic areas. Ours has to do with
improving, increasing the quality of risk communication. That is a big deal at
the CDC. Right now the H1N1 flu, how do you get the word out that fits
everywhere you really need to get your shot? It is not a small, trivial kind of
effort. We are hoping that we can improve the best practices around our ability
to get those messages to the public.
In the area of biology and genetics, we are looking at an objective that
really has to do with personalizing health information. That has to do with
describing health information as you would prescribe medicine in a clinic. So
the information prescription idea is one where we are hoping people will –
that professionals will increase their willingness to prescribe reliable health
information and guidance for their patients. We will be tracking that though
the PEW Internet and Society Survey.
Biology and genetics, our contribution to that set of priorities is that we
will hope to increase the personalized health guidance and some of that has
been mentioned here. Personalized population health is a direction that we
would like to go. We had a hearing of our own on that and it was very helpful
for us to see the direction that we need to go and really want to track how
people want to feel, how personalized their information is including getting
the genetic profile and marking their progress toward health care and having
health care defined and informed by some ways by that information.
Health services, this may be the one that you all are most interested in
and obviously we have a number of objectives that relate to this. Improving
patient provider communication is one that is very important to us. It is a
placeholder for a lot of issues that we share with you all that is it is hidden
here a little bit but we want to look at that in terms of how patients can be a
part of the team. Not that any time there is separation between the patient and
the provider we want to try to eliminate those walls. We are hoping to create
some measures that get at that individuals perception of whether they feel like
they are a part of the decision making process of their team or is there even a
We also would like to know are they feeling like they are part of the
coordinated care around them. We are thinking it is not just enough to have
coordinated care but the patient should feel like they are a part of that
process as well. It is our aspiration to be able to measure that.
The second final objectives are objectives that are coming out of the
Office of the National Coordinator although we have worked on all of them
together specifically these two and the personal health record one.
The increase the use of health IT to improve individual population health
and the final one increase of advanced connectivity to improve individual and
population health are two objectives that will be operationalized according to
meaningful use that will be determined by the rulings that are being generated
by CMS and OMC today.
Those are kind of the top level view of the objectives. We think those are
many of the important objectives that help us think about what together could
create a learning system. I am going to tell you in a couple minutes and then I
am going to close and welcome your feedback.
One of the takeaways I would like for you to have is that when we think of
communication we do not think of it only as public affairs or social marketing
although that is often an important part of the communication process. We have
looked at communication more as I mentioned before as a series of interactions
between systems. It is the in between systems where decisions are made and
where change happens. To know, to really be able to use this as a unit of
analysis and to be able to capture the interactions at different levels helps
us to see what is the quality of the feedback that people are giving. Why are
they giving feedback whether it is provider to provider, on a team, or provider
to patient? Perhaps patient to patient when they are in a group visit and they
go out into the community and provide support for themselves in a social
network. We are really interested in this unit of analysis with feedback as
kind of the important mechanism where learning happens.
I want to offer you one particular model. Some of you may be familiar with
the Care model. We have seen this as a really promising direction to help
inform us in the way that we think about productive interactions. As you will
see at the bottom of the chart for those who are not familiar, productive
interactions from Ed Wagner’s perspective is really where we all want to
get. If we can according to the Care model continually have these productive
interactions with between an informed, activated patient and prepare a
proactive team then we really have something to work with. In our view, we
really have a learning system that if you can improve the quality of the
learning, then you really are we believe, at the heart of the kind of system
that can make the kind of adjustments and adaptations to the challenges on a
continual basis. That is why we think that the health communication and the
health IT perspective we bring to the party may offer you all something to work
with us on as you contemplate the same kinds of issues. Here I am just going to
lay out for you some of the assumptions we have about productive interactions
and how they offer we believe the promise for a meaningful set of measures.
Now we are right in the middle of our effort to finalize the measurement of
our objective. These are not cut in stone. We really do expect during the
public comment period that there will be some useful perhaps additions to these
objectives. We hope some efforts to help us measure them. We would really,
really like to have to compliment our survey-based data some clinical data,
that can together illustrate a bigger picture of how things are going at the
national level in terms of health status of Americans as well as how are the
processes that are either helping or making that creating some challenges for
that to happen. How are they working?
We would really like to join with you all and continue conversations. If
you think that is an appropriate way that we could learn together – as you
see this last sentence, it was so clumsy and I thought, I should really change
this sentence but it really reflects where we are. It is just sort of a jumble
of terms we want to make sure that these terms are in there. That it is really
about the interpersonal trust that happens in interpersonal relationships. It
is also the trust that happens at the larger system level. Trust is really
important even though it is not necessarily part of health information
We see the concepts from those personal interactions as really critical as
well as the more administrative care coordination teamwork and just data
sharing interactions. They need to have that human quality. Some of the
criteria that we all have come to understand as being vital to our own well
being to be incorporated into those.
As I said it is sort of lofty. A little bit extract but our bottom line is
we have to have the measures so we are looking right now to those of you who
are also looking at measures, that is why I brought three fellows and an intern
so that we can learn about the language and the perspectives that you are
taking. I welcome further discussions about this and I really appreciate you
inviting me here.
DR. CARR: Thanks, that was great, very clear and refreshing. I have a
question. In the measures that we have been talking about this morning or
yesterday and today, we have it linked to an accountable party, a physician, a
group, a hospital. Who do you see as the accountable party when you see
improvement or lack of improvement? Who do you turn to?
DR. HARRIS: We struggle with that on health literacy as a starting place
because the measure that we had for health literacy before was really from a
population perspective where nobody was accountable in our way of measuring it.
We would really like to have measures in which providers have some
accountability for changing the way they provide services such that their
protocols, their processes are designed to help people really navigate the
We are trying to have it both ways a little bit. If we could get the
measures to do that that is what we would do. Some of the responsibility is on
the citizen for becoming more literate and I am using this as an example, in my
opinion, a lot of this is just making the system more accountable so that the
design, whether it is interactions, websites, or the processes themselves.
DR. CARR: Just to follow up on that. So you assess this based on surveys.
Then if something got better, do you know why it got better?
DR. HARRIS: Well that is why we would like to program – exactly.
The social determinants approach is one big step towards getting there that
we can kind of identify what in general may be more or less affecting the
difference but you know with survey data you really cannot do that.
DR. CARR: Let me go around the room. Mike Fitzmaurice –
DR. FITZMAURICE: Linda that was a great summary of what you have been
working on for the past two or three years maybe a half of a lifetime. I am
wondering, there are measures being developed. We have had quality measures
from CMS today. We are experienced with meaningful use measures. I wonder if
some of those measures will fit into what you are trying to measure so that you
kind of do double duty or do piggybacking. I am thinking particularly of the
number of health services, health IT and communication objective and measures
They might be simple like do you have problem list? Do you have a
medication list? Do you have a personal health record? Things that are the
stepping stones to achieving what you want.
DR. HARRIS: Absolutely, when we look at the list of meaningful use and we
talked with the other Quality Subcommittee, as it looks like it may be coming
about, particularly the population health and interacting with families and
patients. These are the areas that are ripe for better understanding the common
ground, sweet spot between population health and health service delivery,
between personal health and population health. These are the kinds of things
that we have really though a lot about.
The personal health management I think is really going to be an important
one because it could be a lot of different things. We have not defined that
yet. It is an exciting time to help define it. One of the reasons that we have
not settled on anything yet is because it could be a social networking effort.
It does not have to be a record. If it is not a record it can be sort of
anything. We are hoping that – we are working with PEW on that. The PEW
Internet and Society folks are really close to the ground on that. They see the
social life of information as very real. We are going to be looking at that
DR CARR: Larry.
DR. GREEN: A brief editorial and then I have got two very specific
questions. I think Linda’s presentation really helped us a lot more than I
realized that it was going to around this measure independent of conditions. We
really largely track ourselves and our conversations around thinking about a
disease. I am so sick about talking about diabetes. We have this template model
that we keep using and it is disease-oriented. She helps us start thinking
about measures that are independent of conditions and I think this is really
important to us. It is fueled into practicality for me by work I know about
over the last or six or seven years where you try to get small practices to
address unhealthy behaviors and the pre-determinates of premature mortality,
the grid you show, where we know that it is doable because it has been done.
You can touch it, you can see it and you have these hookups between what goes
on in the community, what goes on in the practice with assessments linked
through website generated stuff that comes in for a visit that gets there ahead
of time that leads to responses and identification of risk leads to
That is a prescription for health as opposed to a prescription for a drug
or whatever. Then there can be feedback groups that occur asynchronously and
through community agencies. You got into that talking about your feedback. So
that sets up my first question is what is your thinking about measures of
feedback that a person gets about their health? How would you know that the
person was getting feedback? How could you detect it? How would you know that
it happened in some way or another?
DR. HARRIS: Well you know there is an interesting sort of theme about
episodes. I notice that episodes are very provider-focused in the discussion
that I heard this morning. I understand the attempt to be objective and that
that is important. There is actually quite a bit of research to suggest that
people think episodically in terms of patterns of their interactions. In other
words, they can recognize an argument. They can recognize joking. They can
recognize the beginning and the ending of their sense of whether they got
better or not. People can actually recognize reliably, predictably their own
episodes. I would suggest that that is another way to go, to think about not
just whether you got feedback but whether this was a quality interaction from a
patient’s perspective or episode or however you want to think about that
but the patterns of interactions that I think make for a useful unit of
analysis and what I am saying is that people can pretty reliably give us what
that episode means for them. We can ask them, did you get good feedback from
your provider overall rather than one – the way it is phrased now is
during your last visit. I think people are more capable than that. I think they
can give us more robust answers than that.
DR. GREEN: Thank you but let’s drill down into this just a bit.
Let’s for the sake of this question assume that health is won and lost in
the community not in the health care system, that unhealthy behaviors kill us
before our time. To change behavior requires constant feedback and
reinforcement. That is the type of feedback that I am wondering about you are
thinking about how you would measure whether or not individuals are getting
let’s call it reinforcement, for moving in the correct direction according
to the Care model for the chronic disease or for promoting their health. How
would you notice what they were getting?
DR. HARRIS: One way you would know is by measuring the kind of personal
health management tools people have. If a personal health record is designed
with feedback in it that tells you at least people who have access to the kinds
of personal health management tools that will provide the kind of feedback that
they need versus tools that can pacify patients. So access to those kinds of
tools is one way to measure it.
DR. GREEN: There is a list of those tools?
DR. HARRIS: There are personal health records that are being developed and
include those kinds of tools. They are not just a record, it is for example,
the use of cell phones to provide the kind of reminders that providers can
outsource to those who are developing or supporting those personal health
DR. GREEN: Actually this is analogous to what we heard from Medicare about
this. Where is the functional assessment and the status of the patient down at
the nursing home after discharge or before they leave the hospital? This is
analogous, that sort of proposition that could have items on it that should be
able to be detected, yet this is much easier. You said something, you made a
comment, you said but that measure is no longer available. What is that about?
DR. HARRIS: The Department of Education was measuring health literacy
because they were interested in literacy so because health literacy changes so
little they are only going to measure it every decade. That is not good enough
for us we have to have – which is kind of sad.
But Larry you know I think we have – about your question about the
feedback, this is the kind of thing I think we need to learn together on. I
think that Wagner is very interested in this. Others are interested in this. I
think the point of connection that we make between Healthy People and what you
all were doing, you all could maybe identify them as well as I could, where do
we learn together to really capture the essence of what we are trying to get
DR. GREEN: I will finish I promise Justine. I would like to create a memory
in our collective minds for further discussion about meaningful measures here
by creating a specific example. Long ago and far away when we wanted
nutritional measures that were meaningful and we went to the field, we had the
sort of discussion yesterday afternoon that dragged out over a year and a half,
we got all sorts of consultative help and we found more than 7000 measures,
nutritional measures. Skip a lot of intervening territory, cut to the chase, at
the end of the day this question – in a typical day how many servings of
fruits and vegetables do you eat? It started as part of Alice Ammerman’s
Start the Conversation Questionnaire down in North Carolina just to get into a
discussion of diet. It turns out you can use that sucker to screen entire
practice populations. It can be answered for every one over the age of two.
Once you screen them you can actually scale it, you can predict who has
unhealthy diets and who does not. You can use it as a follow up measure and you
can scale it up to your whole population and know whether the population is
generally moving toward a healthier diet or not, one question, a five point
I want to put that on the table as an illustration of what I view as a
really meaningful measure related to health. It flunks a lot of the criteria
that we were talking about yesterday afternoon in terms of its development and
how we got there. If you talk to people, just try to eat better, if you talk to
people in the health care delivery system who is trying to work with them they
love this sucker. It means something to them. I will be quite now I promise.
DR. CARR: We always welcome your input. It is very thought provoking.
Truly this was a great thing to end up on because I think it brings us back
to prevention and balances the equation.
DR. TANG: I just have one question sort of involved to what Larry said. As
you mentioned, you have 2000 measure out for public comment. Is there a thought
about condensing that some?
DR. HARRIS: I am really torn about to be the perfect fed or to tell you the
truth. I think the truth is everybody wants that. Nobody has stepped up to the
plate to do it. So I think it is really going to be an important thing for
people to do. I am hoping that will happen.
DR. TANG: To accept that these are survey-based which is I think what you
said and to take advantage of the tools and the kind of connectiveness with
everyone. It seems like if you did have a survey tool that had 20 or 50, one
more people would even take it once and you could convince people to take it
multiple times because then you could present things that could appeal to them
and that they could track. It might be susceptible to advice on changing
behavior and then watch the change. It almost seems self defeating to every
stakeholder’s purpose to have 2000 since no one, absolutely no one will
benefit in a sense. I am sure you have had this discussion but maybe what can
we do to help?
DR. HARRIS: Well if you want to take that on.
DR. TANG: It is not to reduce it but now that you have new tools where you
could consider even having self-administered it would only be contemplated if
it were really digested.
DR. HARRIS: But the challenge is to prioritize. It is to come up with the
10 most important measure. That is what we have not gotten anybody to step up
to the plate to do. I mean if you can do that then you are right, we have the
tools. It is the courage or the sense of authority to actually come up to
prioritize out of 2000 and thousands of stakeholders who are clinging to each
set of objectives.
DR. TANG: Well sometimes people change. In the meaningful use world we just
think back on the NPP process. I mean you can pick a number of processes but
there are still come contemporary health issues that – so maybe one
compromise is you have got a decade. Divide it up into four two and a half year
cycles and then at least you are starting with 500 and then keep whittling it
DR. HARRIS: Well you know I really think that – my bottom line is that
there is a lot of opportunity for cross fertilization among the advisory
committees to the Secretary. There is no reason why you all should not sit down
with Jonathan Fielding who is chairing the Healthy People and have this stock
kind of discussion. If you all do not do it I am really not sure, not you
personally, but the people who are advising the Secretary are in the position
to take on these kinds of issues and sometimes somebody from outside your own
tribe can make that happen by just suggesting it and taking a little bit of a
heat that you do not really experience so much because it is not your
stakeholder group. That is really what I am suggesting to you all as a result
of my coming here and spending a day with you. There is no reason why this
should be siloed kind of exercises.
DR. EISENBERG: Hi, this is Floyd Eisenberg and I have joined the call. Can
I make a comment? I think that one of the things that might help this effort is
at National Quality Forum we do have a set of priorities set by the National
Priorities Partnership and part of our next step is to have Tom Valic(?) and
Karen Adams working with the NCP goals plans to sit down with Healthy People
2020 to help coordinate how we can work across the six priorities and I guess
your 2000 measures, maybe there are more goals than measures so to figure out
how to work this through and to coordinate. I think that will help.
DR. HARRIS: That is right there has been some efforts and I did not mean to
be dismissive of the efforts to do this. It is who is going to adopt which set
of efforts to prioritize; IOM has offered some as well.
DR. CARR: Floyd thank you. I am glad you are online. Are you ready?
DR. EISENBERG: Yes, I do not have a whole lot to present and I apologize
that I do not have slides but I can give a bit of an overview of what we are
doing at the NQF in order to move towards the retooling and to move forward on
some of these areas that were discussed. Is it okay to begin?
DR. CARR: Please, yes.
DR. EISENBERG: The National Priorities Partnership, I think Helen Burstin
presented some of the goals yesterday. They are really patient and family
engagement, population health which includes preventive healthy living, and
community index measurement, safety care coordination, end of life care, and
overviews. There are groups being brought together over the next several months
to year to call for measures in these areas. Some of the Healthy People 2020
may fit right into this. In the process of our call for measures what we are
trying to do at this point is to encourage use of available electronic data.
Some of this I think may need some research. For instance, how to measure
social networking that shows engagement and education of families. How does the
PHR use it? Having it with certain functions is one thing but how do we know
the components are used for education or whatever we are looking for and to use
what. Some of that is what was discussed yesterday as the quality data set and
looking for the electronic information does address some of these issues. We
are trying how to best coordinate that into the measures from the electronic
data stream. At the moment what we are doing is we took the 17 measures that
are endorsed that map directly to the policy committee areas for measurement
and an additional somewhere around 53 measures that were requested through CMS
to have them retooled and we are in progress for the contracting of having that
done so that we can see all of them address the appropriate information in
electronic data. We expect that will be done, the retooling by March. So they
will be available for 2011.
We are also looking forward to adding a clinical decision support element
to the quality dataset and a coordination of care element around plans of care
so that we are identifying the appropriate information to measure for the
I do not have a lot more to discuss at this point. I was a late addition to
the panel. I am not as quick as Blackford to create a set of slides as
yesterday. I am free to discuss what you like to learn about.
DR. CARR: Can you say a bit more about the quality dataset? A little more
detail about what is in it or what is your vision to be in it?
DR. EISENBERG: Sure, the quality dataset framework starts with the data
element which is a concept that I am looking for something about a disease or a
condition. I am looking for communication with the patient or I am looking for
communication actually. We add to that communication, with whom as I use that
example, and that makes it a data type. So communication to a patient,
communication from a patient, each is a different data type. What communication
I am looking for is represented by – it might be I just need to know I
communicated. It might be I need to communicate about a specific subject and
that subject represented by a set of codes that says, this defines the subject
that I am looking for but communication with this patient on this subject.
What we have done is we have a set of terms around communication, so
communication to a patient to and from another clinical care provider. We have
almost any concept that you could think of for medication. Medications are
prescribed, they are administered, they are dispensed, they are declined for
laboratory studies and then they are ordered, they are performed, they are
resulted and if resulted a value. All of the information about a medication
such as the attributes such as dose and frequency come along with that
I apologize. I could have shown you a whole list of, the whole set of data
types. What we did try to do is address existing areas in the medical records
with all of those types and also new areas that we would look for such as
functional status, patient – I am sorry I am thinking satisfaction there
was another word we use for that, care experience.
What we have done is look for where in an electronic model of information
that information would be found. A lot of that work has occurred in HITSB
identifying for each of those elements where that kind of information would be
shared if it was sent from one EHR to another in a message or in a document. So
in a CCD or in an HR7 message what information represents that data type? That
then becomes the translation of what is in the EHR to the information needed
for quality or clinical care. Does that help?
DR. CARR: Yes, and actually this gets back to what Bill Scanlon was saying
about having a repository of data elements but then with those elements and you
know where they are then you can query how often something happened or that it
happened or what the result was.
DR. EISENBERG: Right and that is exactly what we are looking for is so now
that we have the framework now it is the issue of implementing it so we have
multiple data elements that can be reused either for clinical care or assessing
the care that was delivered through measurement or through data mining and also
for clinical decisions support to determine if a next step is necessary and if
This is in a sense a database of elements to be used for that.
Understanding this will in time grow and needs to be maintained and that is
part of the structure that we are trying to set up now.
DR. CARR: But does this also become the blueprint for the vendors in terms
DR. EISENBERG: It does. It becomes the blueprint – now we do hear from
vendors, and I would be interested in the Committee’s comment on that,
that they are concerned that if the data elements or this is prescribed exactly
how they represent everything in each of their products they all look the same
and they will all be commodities. They feel that they can provide better
innovation if they are given some leeway to improve the flow within their own
product as long as they know what part of that flow maps to and connects to the
information that is required for that data element.
We have created a concept more as – and I will use the Rosetta Stone
concept that through the using of data elements helps work with any EHR and
however they represented it, the same information so that it can be used.
So what I am interested in is does that make sense or does the Committee
feel there needs to be more prescription to what actually sits in the EHR?
DR. CARR: They do not know.
DR. EISENBERG: Actually we do not either in NQF. I do want to disclose that
I did work for a vendor it was actually a year ago, so I do understand that
perspective of one to innovate. I think there is some need there. I think there
is probably a balance and that will be sorted out in time but at least having
the standard types of elements and the standard list of them and the ability to
access them will help move all of this forward which ever direction that takes.
DR. CARR: Bill.
DR. SCANLON: Let me make a rash comment since I am not a clinician and
someone that is not directly involved with it. This notion that this
standardization is holding back sort of innovation is kind of like saying
innovation in word processing programs is held back by the fact that the spell
checker works against English. That is the standard. There are so many other
dimensions that you can think about for innovation and distinguish your
products that I just think that argument needs to be ignored very quickly.
DR. CARR: Mike Fitzmaurice.
DR. FITZMAURICE: Bill probably said it better than I could. I was going to
suggest that they compete on the ease and the flow of work of getting the
information into the recording mechanisms or into the electronic health record
but it all has to be recorded as apples not apples and oranges.
DR. EISENBERG: I actually appreciate both of you. I think when they –
often vendors or the ones that I have spoken with and my impression is they
listen to the discussion and they assume that the discussion is telling them
they have to have a standard interface and a standard flow and all have the
exact same questions in the same order as all of the others. I think the
efficiency of workflow is what vendors can compete on and can innovate on. The
data itself needs to be standardized.
DR. TANG: I guess I am still having trouble even understanding what they
are objecting to.
DR. EISENBERG: Sometimes I have similar concerns Paul. I guess we would
have to have the vendors in the room to –
DR. TANG: I guess I am very much with what Bill said. Maybe the world just
has to move forward without them.
DR. EISENBERG: I will not argue that. I did have one say to me if they
wanted to create their problem list dynamically from all observations, the
value of these observations, rather than have a separate list so every time a
physician or a nurse or whomever wanted to see a problem list that was created
and available for view. They want that ability. I am not sure if that works,
well I cannot kind of understand that concept too well.
DR. TANG: I think it is fine for a list of states that made up the scenario
that they can suggest to the physician in this case to put something on the
problem list. The problem does belong to the health care professional. There is
still going to be and is needed a problem list. It just really sounds like it
is taking a lot of effort to just not do it and I think we ought to just move
on and say it has got to be done.
DR EISENBERG: Actually that is the direction that we have taken is just
move ahead with the dataset and knowing that that is needed and not delaying
because of concerns about innovation.
DR. CARR: I think that concludes it. Floyd thank you again for your
flexibility and for you insights. We appreciate it very much.
DR. HARRIS: It ties together one of the things that we were talking about
and that is I would love to hear you all reflect on this for a second. One of
the standards of the objectives that we have is about quality health sites. It
also refers to a standardized interface; it does not have to be a website, to
improve health literacy. Should that be designed into, instead of that being a
requirement, should that be a part of the standardization of the tools that we
are making available for patients or maybe providers or maybe intermediaries?
That is just the kind of discussion I think that would be fruitful for us
to engage in together. It is certainly one of our important priorities and I do
not know that it has been a part of your discussion this morning.
DR. TANG: We had the discussion on disparities and so the disparities that
we can measure and that certainly can include language and health literacy.
Eventually we have to take a lot of time right now, and that is sort of what we
asked the earlier panel, with respect on trying to measure and measure
precisely with more granularity. That should prompt us to now figure out how to
act on those things and deliver. As an example, there is a meaningful use
requirement that says patient specific educational materials. That could be
pretty broad or it could become more stringent over time. I am not forecasting
the future I am just saying that you can imagine that to be patient specific it
would have to be in the language of the patient. Maybe it has to be adjusted to
the health literacy, it maybe culturally sensitive, a number of things.
Those are I think the important ways that vendors can help innovate what we
do more tailored to an individual.
DR. CARR: Okay, I think that we have about 20 minutes before we break to
organize or sort of categorize thoughts. Fortunately we have Susan here who is
working on this but why don’t we start with that? So Paul why don’t
you take the lead?
DR. TANG: I think the hearing has been productive in the sense that we have
learned a lot of information about the current activities. I think what we
expressed in the beginning of this morning was that it is not clear and it is
probably accurate to say that there is not any overarching strategy or
overarching framework to move these measures into what we think of as a new
era. The new era is all of the sudden there is a richer set of data that will
become available. There is a richer set of tools to make it displayed to people
who could be influenced by it. One of the points that I think Carolyn brought
up is it would be really nice if this was bidirectional, that we could use some
of the data that is analyzed in mind and put back in front of the people who
are effecting every day decisions.
In my mind, in her call for a killer app, I think that is almost a killer
app and I think Larry said something similar. That kind of information would
influence the very next person I saw. I think there is a real opportunity. I am
not sure it was necessarily expressed by the folks in the trenches right now
but maybe that is where we can contribute. That is what are the overarching
goals, vision, and whether there needs to be a framework and whether there
needs to be some kind of coordinating or hearing kind of group that helps guide
development of measures that are more meaningful to the providers, the
patients, and consumers not only for consumption for the way it is now. When it
is so late, which is usually one to two years late, it is almost not useful to
anybody but put to use by certain data consumers.
DR. SCANLON: I guess I will react to that killer app in a second but I also
think I have a different killer app that I worry about and I think Carolyn also
included it in her list too which is in some respects the health care system
and the fact that we have health reform on the horizon. Even without it we are
spending $2,500,000,000,000 and we are really worried about where that is going
and what we are getting for it. The question is what can we do about that? How
do we leverage information to a make a difference there? I mean after all the
IT that we are getting is being paid for to a great extent with public funds.
There is that aspect of it.
I think there is a lot that is feasible. We should not be deterred sort of
in any way. The killer app in terms of feedback – two weeks I made a
mistake on the internet trying to order something with my VISA card and my
phone rang within five minutes saying did you really try to do this? It was
actually the wrong expiration date and the card got rejected and they wanted to
know if somebody was using the card.
This is like within the capacity of current technology there is no issue
that things can be done. The information – my version of this is that we
extend the building blocks to the payers and they can give feedback so that if
the Medicare program gets information from their claims about the services
being provided there is no reason why within a day or within an hour
information in an aggregated for cannot go back to the provider. I mean that is
just a function of current IT capacity mechanics. It does not exist at CMS but
can exist at CMS. That is the kind of thing that we need to think about. I
really believe that the building of a database is key much more than the
current measure because I feel like the current measure we are still working on
incredibly hard to make them richer and that if we get hung up on what it takes
to build them we are going to end up having even more complaints from vendors
because they are going to be continuously saying, wait a minute, yesterday you
told me this is what you wanted and now you are telling me this. There is a
certain legitimacy to that but we have to overcome that by kind of being able
to build this kind of Swiss army knife of data so to speak. I think all of
these things are possible and we need to push forward on them.
This Committee has got a very valuable potential role in this because we
are the HIPAA Committee. There is a potential linkage between these information
flows that we may want and HIPAA. We are also sort of a very good
representative of the privacy community. That intersection with us not even
identifying the objective that intersection in terms of the meaning and what
the risks are et cetera are can provide very important advice for the
Secretary. I hope that we move forward in that and try to bring together not
just what we think about here in the Quality Committee or the Populations
Committee but also what our standards and our privacy sort of cuts across this
Committee as well.
DR. TANG: I can piggyback a little bit on one of the threats which are our
responsibility and our opportunity to give advice to the Secretary or at least
HHS. As Linda mentioned as far as the fact with the Healthy People and of
course we are a factor that is very related to the HIT Policy and Standards. I
think a lot of us have talked about how meaningful use is becoming a very
forceful tail that is wagging a much bigger dog that is part of the health
reform and that timing means everything. Just like we have pushed the measure
developers to a point, this is a real good opportunity to get some good
measures out there.
I think we have a moment in time also with respect to the Meaningful Use in
2013 and beyond there. Still the 2011 is open for public comment in December.
To the extent that we want some of this information on meaningful measures
which we picked up from the meaningful use phrase to have an effect, it seems
like we need to get our advice, our letter out – I mean it would be
idealistic to say November but when is our next meeting after that, February
that is actually too late because the public comment will be open.
DR. CARR: I think we should absolutely aim for November.
DR. TANG: That could be very potent. It will definitely go into the public
on 2011. It will also be forwarded to the coming 2013 deliberations but that is
how we would be coming in. I think that we have drawn certain conclusions. We
have talked about it offline that could be very potent and important to this
whole move in this direction. So that is the challenge to ourselves I think to
DR. CARR: So are you saying that there would be value in taking the
meaningful use proposal and intersect that with what we heard?
DR. TANG: That is one way HIT policy could take, maybe the other way is we
can digest what we think are measures that can be meaningful with respect to
measuring the health system and influencing it’s direction in health
reform. The Policy Committee can wrap that or CMS can wrap that into the whole
meaningful use construct.
DR. MIDDLETON: The idea of actually – the occasional interfacing of
the FACAs was extremely interesting for a full host of reasons. It seems
natural just from a management perspective to have those kinds of committees.
You know talk to each other periodically. I am sure it would lead to higher
quality and perhaps more interdigitating advice to the Secretary. There might
be a recommendation regardless of the current effort; there should be a
recommendation in general.
DR. TANG: I was going to pick up that and easily volunteer Justine to do
DR. CARR: Okay, next steps. I know I for one kind of would like to reflect
a bit more on this. I think that we have heard some important themes today. I
think perhaps if we could each list ten important themes and kind of send them
to all including Susan, I think that would be a helpful way of us saying what
were the things that we all heard.
DR. TANG: And then from that if we could lump them into five categories
then we could start saying something rather than –
DR. CARR: That was my – I tee up the lumping by splitting. I make the
elements like Bill and then I roll it up.
DR. GREEN: Justine, do you have in your mind a title of the letter? If you
throw a title forward our themes will follow.
DR. TANG: So that would be our recommendation for measures that would
meaningfully impact health reform.
DR. SCANLON: How about achieving the measures because I think it is
premature to say that we have the measures. I think what is critical is the
process and how we are going to get to the measures.
DR. CARR: Does it impact health reform or reform health?
DR. TANG: So the title at one point has to say, reaffirming the status quo
will actually get in the way of health reform. Actually I think I would delete
that one because it was my no measure, no mission. If the payment is now going
to be on measures then truly we have no measure, no mission, no measure, no
money. That is how crucial what we measure –
DR. CARR: No outcome, no income.
I guess we will get to the themes but I know I for one was struck by the
disparity in a sense of urgent action. I heard some great ideas today which
over the decade to come make come to fruition. I also heard some very concise,
incisive, ask one question and trigger a meaningful action. I think that in the
moment concept is true. I think timeliness is really an undercurrent in all of
this. Timeliness for the physician to see the results in the moment, timeliness
for us to find the one thing that will take and make us 25 percent better as
opposed to taking 25 years to make us 26 percent better.
When is our meeting? November 17th? I have it right here.
DR. TANG: So if we want something to submit into the public comment we have
to have it approved –
DR. SCANLON: When are the rights going to be released? It is in December
DR. CARR: We can approve it from the Subcommittee and then – well our
November meeting is the 19th and 20th of November.
DR. SCANLON: I guess the issue of approving it would have to be public
notice if this was going to happen with Paul it would have to be open in a
PARTICIPANTS: (remarks off microphone.)
DR. CARR: Well I think that, anyway we have had calls where we have had to
work on – I think the products will be transparent but I think the
alignment of our thinking, we will be talking about everything that was public
record and in the transcript today. It is a question of assembling it.
I actually think that we need to get our concepts that are the most urgent
concepts to be heard out by the 19th and that maybe we have a list
and a brief executive summary. Perhaps we then want to have a more in depth
letter or comment about taking it to another level. I think we definitely want
to have a distinct message ready for November. I think that we can do this on
conference call. I would say the top ten most important messages that you heard
coming out of this. Maybe what you thought after what you heard and then if we
can circulate them and then Cynthia if you could set up a call in the next say
10 days –
DR. SCANLON: One of the things that I think was apparent yesterday and
Harry’s communication as well in the topic today, is there exists a
leadership vacuum in getting this done. Maybe a way of putting that is
governance and accountability around this. I think that there are a lot of
people trying to do really good things. I know that is the case and doing
thoughtful things and many of them are actually collaborating and working
together. I mean yesterday we had two of the chairs of the NQF Initiative and
Helen Burstin talking about – they were there and working together. This
urgency and the timeline requires I think, a different sort of framework for
leadership and governance that does not exist today.
DR. CARR: So we have got the 10 things, 10 thoughts. Let’s say in the
next week and try to be specific with the measures. In seven days and I will
start off an email to all so we can just respond to all.
I just want to express my thanks to everyone for their wonderful
participation. Of course, thank you especially to Matt for working so hard on
this and also to our able staff for tolerating our last minute changes in so
many ways so thank you. I think we are done.
(Whereupon the meeting adjourned at 2:50 p.m.)