[This Transcript is Unedited]
DEPARTMENT OF HEALTH AND HUMAN SERVICES
NATIONAL COMMITTEE ON VITAL AND HEALTH STATISTICS
SUBCOMMITTEE ON QUALITY
February 26, 2009
Hubert Humphrey Building
200 Independence Avenue SW
CASET Associates, Ltd.
Fairfax, Virginia 22030
- Justine Carr, M.D., Co-Chair
- Paul Tang, M.D., Co-Chair
- Carol McCall, F.S.A.
- Larry Green, M.D.
- Garland Land, MPH
- Blackford Middleton, M.D.
- William Scanlon, Ph.D.
- Don Steinwachs, Ph.D.
- Mark Hornbrook, P.D.
TABLE OF CONTENTS
- Welcome and Introductions: Justine Carr
- Defining Meaning of “Quality” in Quality Subcommittee
- Innovative Data Sources and Data Streams – Carol McCall, Blackford Middleton
P R O C E E D I N G S (9:06 a.m.)
DR. CARR: I am Justine Carr, co-chair of the subcommittee.
MS. KANAAN: Susan Kanaan, writer
DR. STEINWACHS: Don Steinwachs, member of the subcommittee.
DR. MIDDLETON: Blackford Middleton, member of the committee.
DR. GREEN: Larry Green, member of the subcommittee.
DR. HORNBROOK: Mark Hornbrook, member of the full committee.
MS. MC CALL: I am Carol McCall, on the full committee and member of the
DR. CARR: Paul is en route. I think that we can start. As you can see, we
put together an agenda based on our last call. As you know, this afternoon we
are going to be hearing from a number of folks on novel sources of data.
What we had outlined today as talking about, when we say quality, are we
talking about quality of care, quality of data, quality of what. Then we are
going to talk about innovative data and sources. Carol is going to share some
of her experience, and Blackford as well, and then planning for our
freestanding quality hearings.
But I think I want to raise one issue. Yesterday we heard a lot about the
Recovery Act, the dollars available and then a number of new issues raised in
that, PHRs on privacy, on HIPAA, on stewardship, a number of themes that we
have seen before. So I think it is worth pausing for the moment to say, did we
hear anything yesterday that we feel should call for a mid-course correction of
something we need to do immediately, or are we good to go with the themes that
we have established already.
So your thoughts on that?
MS. MC CALL: I think it is a great question. The thing we have to ask
ourselves is, given where we are, is there anything we would change about what
There is nothing that I would change in the immediate term. I will reserve
the right to change my mind by the end of the meeting. The hearing that we are
getting set to have I think is still the right one. I was glad to hear themes
around data stewardship. I think it has to do with new sources of information
and broadening the lens with which we view data, its sources, its uses, what it
means to quality, or what quality means with respect to health care, health
data, all those different types of things. We will hone in on those.
I don’t think we should change our path. I still think the next step is the
right one. We may add some different textures to it. It may help inform exactly
what we ask, what we do, who we bring.
DR. MIDDLETON: I’m so sorry I had to miss yesterday, but I was gainfully
employed reviewing some stuff with AHRQ down South. Could someone characterize
the summary if you will of the presentations yesterday? I have heard a lot of
things from a lot of different sources about how the stimulus bill is going to
impact IT obviously, and potentially quality and patient safety and care, et
One of the things that I raised on the list via e-mail earlier was, how
does our role in NCVHS change or not? Given the call for new FACA committees
from the legislation, what do we do as NCVHS? Does that change our mission or
goals or activities?
DR. CARR: We talked some about that yesterday. I think that — others chime
in, but I think our sense is the same sort of presence we have had for 60 years
will continue. A lot of this is about health IT, and we are about some of that;
we are also about other, broader questions. So I don’t think there is any sense
I think there was a call yesterday to put together here are the things that
NCVHS is well equipped to do, and get that out there. I think there is some
skepticism about whether these two FACAs will be able to be pulled together in
90 days and implemented, especially because we are still waiting for a
I think also, well, I don’t know, we will see, I am just reflecting on
conversations, but it may be that they are going to be very implementation
focused, the standards in particular. I think the question that I heard more of
is what does it mean for the new AHIP 2 committee. Did others heard anything
different? Are we going to be impacted, is NCVHS impacted?
MS. MC CALL: I actually had to leave right when that question was being
asked, and I heard a hard stop and I didn’t get to hear the answer. But I’ve
got the same question, which is, are we elevated, demoted, or are we in a
different position, so that what we should be doing —
DR. MIDDLETON: Two out of three of those are bad, demoted or pushed to the
MS. MC CALL: Right.
DR. MIDDLETON: The specific questions I might ask the committee to
consider, either the subcommittee or the full committee, is just looking at the
legislation, now that it is law, things like the focus on HIE. Not only has it
standards impact, and will draw upon a bunch of the standards thinking, I
think, but there are clearly implications to data integrity and quality as we
think about their transit source, authentication, non-repudiation, et cetera,
and how they are used for measurement or care delivery.
Then secondarily, this whole issue of meaningful use, that is, the
Medicare/Medicaid funds will be allocated based upon demonstrations of
meaningful use of HIT. So far, all that means is, DRX, decision support and
quality reporting. So we own a third of that right there, quality reporting;
how can we help outline strategy in the standards or quality data issues for
quality reporting, but further, might we think also about measurement of these
MS. MC CALL: Thank you. The things that I saw woefully missing, and I
brought it out both in the full committee and then offline, and I think is the
task that the Populations Subcommittee just walked away with, which is, there
is no money there for the health statistics enterprise for the 21st century, I
mean none. That is one of the biggest misses that if we don’t make sure is
shored up, we are going to regret it, maybe not today, maybe not tomorrow, but
soon and for the rest of our lives.
DR. STEINWACHS: So Carol, you shouldn’t have left the meeting, because your
name came up several times.
MS. MC CALL: There you go. But I think our role with respect to that is to
say what else needs to be done. So it is not just quality of data. There is
more that needs to be done around metrics, what is a good metric. Nobody really
talks about that, but they matter and they matter a lot. What they are, how
they are couched, are they consistent.
Then, even a metric is not just reporting, it is analytics on same. There
are going to be more ways to analyze them than there are different elements and
bits. There is very little talk about dichotomies, there is very little talk
DR. CARR: Some of the work that Paul did with high tech was to say, here
are all the things that are out there, and here are the things that we can make
I think that taking that one step further is to think about these
electronic health records, what are the structured fields, looking toward what
will we want to be measuring, because right now, because you can get it on your
computer doesn’t mean that — it is a little bit electronic, but in terms of
queryable fields or creating registries for measurement, it is not quite there.
So I think that might be one thing that interfaces with that, what are those
Over my now five years of being on the committee, there is a balance that
we need to strike between granularity and high level. I think again, it is
teeing up the issues, asking the questions that are answered by others.
One of the things that in the NEHIC presentation that caught my attention
yesterday is the high level ideas articulated, followed by a — I forget what
they call it, but a specialist, kick the tires on the concept. We need to
remember that even though a number of us have a lot of expertise in this area,
we meet four times a year plus some hearings, so we can’t be doing the complete
drill down. So I think striking that balance of how you tee up the important
questions and hand it off.
We heard yesterday from the CDT that did a project on de-identified data.
We had raised it as part of our secondary uses hearing, but we realized we did
not have enough information. They brought all the thought leaders together.
They haven’t answered everything, but they brought it farther than we were. I
think that is the model that we want to think about.
So as we think about these things today, maybe what we do is hear a little
bit from you and Blackford about things that you are doing, and then come back
to this issue at the interface or the areas of focus in the upcoming Recovery
Act that relate to quality and how we would interface, where we think we should
be. This afternoon, we are going to hear about data from non-traditional
sources. That is interesting, but it may be not intersecting as much as we
So if you could talk a little bit about information that you were talking
about with the pharmacy and predicting adverse events.
MS. MC CALL: Actually there were a couple of things I wanted to do. One of
them was talking about data creation and data use, and then some
non-traditional players that I see in general, certainly not a complete
landscape, but just some examples.
On data creation, there is a lot that we do at Humana. I am part of our
Innovation Center there, which is something that we started back in 2000. It
was a time of big change at Humana, but also in the industry toward
The Innovation Center is this investment in being intentionally different.
It has been this Chinese calendar year of different themes and all of that.
Where we are today, we have done a variety of different things.
Around health behavior change, there is a strong belief that that is one of
the keys, and it is really all about how do you engage people, how do you
identify them, all of that. There is a lot that we do with personal
technologies, there is a lot that we do with incentives and rewards.
We have launched health games which are absolutely fascinating. There is
one in particular I will tell you about. We have a lot more emphasis on social
networks and social programs.
Some examples. This is a spinoff company, it called Sensei. It is a
personal coach on your cell phone around diet, exercise and fitness. It allows
you to customize, personalize not just diets, but prompts based on the
particular issues that you have, whether you tend to be the midnight raider or
whatever it is.
This is something that is not just for Humana members. You could go out and
buy it today. So the whole idea is to have something that is personal, intimate
and always on you, so it is really quite delightful.
Another is called health models. We created this with Virgin Life Care,
Richard Branson, Virgin and all of that. It is an incentive and reward program.
I can earn, and I am part of it, $450 a year by essentially keeping track of
not what I eat, but in this case, what I do with my feet. There are contests
and there are programs, and there are winners and losers and all this kind of
stuff. You can compare yourself against others, and it is really interesting.
There is data in there that is being generated and tracked, and you can compare
and contrast. We have tracked the impact, to your point, Blackford, and we have
seen changes in weight, we have seen changes in blood pressure. You step in a
health zone and it will keep track of some of this stuff. It is very easy.
Another one is called Horsepower Challenge. This is a health game
literally, for kids in schools. There is a pedometer. They wear it on their
shoe when they walk in the building. It uploads — those steps earn points. The
points allows you to pimp your horse that is now part of the game. The horses
are now racing each other. The best school wins, school on school, and the
winner gets to go to the Derby, in a whole big splash.
The feedback on that one is that it harnesses the kids’ natural desire to
be creative, what they do with their horses on this game; they compete. But
what the teachers say is, it not only increases awareness of activity, which it
does, it changes their relationship to it. It also changes it not just in
school and with each other, but in their families. The parents are saying they
are having different conversations. So it becomes harnessing what is naturally
fun to do, where the emergence is healthier behavior.
Last but not least in this data creation and programs is essentially a
social bike program called Freewheeling. It is spun out to a company called
B-Cycle. It is just like the bike sharing program in France, but unlike that
program, it also allows you to accumulate points, a la Health Miles. So you can
keep track of your own activity. There are kiosks that are around. You come in,
you sign up, you use it, you drop it off at the other kiosk. We put one here in
Washington, D.C. It was launched at the Democratic National Convention. We also
had it at the Republican National Convention. We have sold the city of Denver,
and we sold it in Miami Beach.
But the whole point is that these types of things from a groundswell up
become about health and health behavior. So if you think more broadly about the
data that is being generated and how it is going to be used, there are
implications everywhere. People are going to want to track and compare. So it
steps even outside the PHR, at least in the way most people think about it.
So those are some novel data creations. I can give you some more examples,
but those are delightful highlights.
In terms of data uses, what I do in my world is, I do scary stuff with
data. I have got a Health Services Research Center that does the things that
you might traditionally think it does. It takes advantage of the data that we
have, administrative though it is.
We do some interesting things in terms of fusing quantitative and
qualitative, but those are recognizable sets of activities. What we do that is
kind of unusual, we have a portfolio of predictive models. I was telling
Justine of one that we have created which is really quite delightful, folks.
What we did was we took all of our claims data. We then processed it to
create a lot of meta-data, a lot of tags about conditions and situations,
everything from, yes, you do have cardiovascular disease or not, things you
would recognize, but also indicators of depression, et cetera. We fused it with
a knowledge base of every known possible drug, serious adverse drug reaction,
drug or disease, et cetera.
When we fused those, we came to the point of view of every person who
believed had a serious adverse drug reaction over a two-year period, across our
entire Medicare population. It added up to be quite a sum. It added up to be
over $500 million of impact just for the inpatient admission or the ER visit
period, no follow-on costs, no other costs.
Then we build a predictive model off the whole mess. This is a tough, tough
predictive problem. These are rare event issues. They don’t happen often. They
are categorical data, longitudinal, all kinds of hard computational science,
and they did it.
So they have an area under the ROC curve for certain types that is over 80
percent. That is very good. That is with the time boxed, can you be better than
random at all. There are a lot of situations where they will come back around.
They will tune, they will refine, et cetera.
So that is the type of thing that could become — it is not prefect, it is
far from, but what do you do with something like that? You could make it the
beginning part of a clinical decision support that says, your doctor, if he
could put it into some of these information exchanges.
Humana has a company, they are called Availity. It is a nonprofit, all
payor, all provider. It uses claims data, snap-together payor based health
records, and you could have this sitting on top.
Another thing we do is, we use our data, but also now our members, to power
up clinical studies, observational life science/biomarker studies. So we are
powering up what we call the bioimage study, which is part of a huge set of
research around high risk plaque. They are closing in in about 18 months. They
will have a blood test that they are ready to launch that identifies people at
high risk near term of first heart attack or stroke.
What we have done is, we have said, we will have 7300 people into a
clinical trial by June of this year, that have been identified as qualifying,
outreached with a call center, enrolled, and they come to a mobile clinic.
So it is a really exquisite example of a novel and different use of, in my
world, administrative data to massively speed up, scale up and diversify the
moment from science to solution. So it is a very novel use. No other health
plan has done this in this way at this scale, and it makes possible things that
would traditionally be much slower or not at all.
Last but not least around all things geeky, we are pursuing a simulation.
Think about the Sim City of health policy. One of the predictive models that we
built is an obesity inference model. We don’t have height and weight and
therefore BMI on our membership, we just don’t. But for a small group of
people, we did get it. We got it because we built an HRA, and we got that we
built that off of some other predictive models that we have. But to the HRA we
then asked other questions.
So for about 100,000 people, we had a whole bunch of stuff about health
behavior and height and weight and all that. We said, okay, we can build an
inference model that says, for people that we don’t have height and weight, I
can use your claims data and I can figure it out. That has an area under the
ROC of 287, so it is pretty good.
What you can then do is take that, run it on your entire adult population,
and you can infer obesity in every single adult. You can now create, and we
will be creating, an obesity index, a health index that ties it to things I
know about metabolic syndrome. Do you have diabetes, what is my opinion on
that, do you have hyperlipidemia, do you have high blood pressure, are you in
fact obese. You can then combine these in different ways to understand, broad
side of a barn, who is likely pre-diabetic. You can go out and you can get
them. BPP says, one thing you want to do is have them reduce their body weight.
We have the interventions. So you can use data in novel ways.
I want to take it one step further. What I want to do is do some
simulation. I want to take a particular city and I want to simulate the health
of a city. The way you do that is, you grow a city. So we will go back, and I
will be doing some data fusion that is really cool. I want to take a city, it
will probably be Louisville, and I want to be able to take all this different
information about it over the last 20 or 30 years, everything from what is the
walkability index, how many sidewalks do they have, what have they done with
public transportation, what is happening in public schools. These are all of
the forces that come to play on the health of the city.
You use those as guideposts. You can use agent based modeling and other
simulation techniques to grow that city and its forces to where it is today.
Where it is today health wise is going to be how I represent it in this health
index. But that health index is an emergent property. It is not an input, it is
an output. So there are ways that you can use these other forces to get you to
Then what you have is, you have the ability for policy makers to understand
the major influencers on health as they go around making decisions about urban
planning and how they handle their businesses. So it really does become the Sim
City of health policy.
It is fusing data at levels that were never meant to go together, so you do
it very carefully. Some of it will be at a very high level, others will be at
this level, and you basically do good enough, so that you can have a model that
is good enough to make better decisions.
So those are some non-traditional uses of our data being fused with other
data. We expect the health index to be ready second quarter. We expect the Sim
City of health policy to be ready third quarter. So that is data uses,
There is some good stuff here in Tab 6, 6 or 7, that for the first time
talks about some of the health data players. So in the packet, social networks
come to health. There is discussions in here about companies like Patients Like
Me, James Heywood, Ben Heywood, they are in Boston; they are delightful. These
are places where people come and put in some of their own data, and can
volunteer for clinical trials, all kinds of stuff.
There are some other things out there. Adam Bosworth who used to head
Google Health, he now stepped out to start his own company, Keiss. Full
disclosure; I am on the advisory board of that. That one is going to do some
interesting things around what I call do it yourself labs. There will be a way
for people to order their own labs and step around the gatekeeper, of having to
go to your doc and pay that cover charge. It will all be within regulation, no
360 network behind it. So it is not an issue of regulation; it is the fact that
there is going to be a fundamental sea change in terms of people’s own ability
and accountability for their own health.
Another technology that I have seen is truly a do it yourself lab kit. This
is something that once shrinkwrapped, it becomes a do it yourself kit that you
can buy literally at a Walgreen’s. It takes a drop of blood. It is to perform a
lot of the know your numbers tests, whether it is cholesterols or a variety of
things that you might normally get in one of the comprehensive panels at a
doctor’s office, and literally bypass even Keiss entirely.
There is a lot out there on body sensors, everything from heart rate that
is done remotely, sitting on your bed, to how you are sleeping, to motion, to
gate. These are things by Intel and Natrills Research Center, Foster Miller,
Physic Ventures, Unilever.
DR. CARR: Carol, you are always a wealth of rich ideas. It is fantastic.
Having heard that, Blackford, I want to give you a chance to talk about other
perspectives. We are just talking about what we have said out to do and a
little bit about what was talked about yesterday, about the structure fields.
DR. TANG: Can I ask two questions of Carol?
DR. CARR: Yes.
DR. TANG: One, in discovering — I don’t know whether ADEs or all kinds of
MS. MC CALL: No, ADEs.
DR. TANG: You said you sort of heard things. Did you infer them, or did you
use claims data that also pointed out that this in fact was a complication due
to an ADE?
MS. MC CALL: Only six percent of our claims are coded that way. It is in
peer review with people at the University of Arizona. We have had I know are
ghosts in the data, so it all depends on how well those are accurately defined.
But the way it was done was, you made sure to strip out the underlying
incidence of admission. It was only something that the literature said was
coded up. It said, if you see these combinations, then you could see an
admission for X. So only if it was an admission for X, and then even then, some
of those got stripped out.
So there are still some ghosts in there. We believe that 80 percent of them
are more —
DR. TANG: Six percent is about the right number, actually. The other
question is, you described very innovative ways of using public data fusion. It
is the use of that information that would determine whether this is a positive
thing for the individual or not. How do you determine that?
MS. MC CALL: At this point, once you actually get the end of the result,
there are a number of different ways you could do it. Can you see an individual
in — this is single play, just like a video game; do you want to be able to
see individuals in it or not. If the answer is no, then all of that detail will
get stripped out.
DR. TANG: So there are a couple of ways to do it. I could go find the
individual to go help them, renting the bike or whatever, or I could say, you
ain’t getting insurance from us. How do you decide?
MS. MC CALL: There is the Sim City of health policy, which can strip out
all the detail. You can strip all of the individual detail out of that, or you
can present it in a way that can say, I can show you globs or clumps of people.
So that is one thing that you can do. But then you quickly transitioned
over to, what do I do with the information to decide if I am going to insure
DR. TANG: You have used these things in novel ways. You have gotten new
information. You have to separate what is good and what is not, but anyway, new
sources of data. Depending on whether you are a provider or an insurer, you
could do one or the other thing to with it.
MS. MC CALL: Having been cast in the role of evil empire, I understand, but
you are going to have to get more specific because of the implications. Are you
saying it is impossible?
DR. CARR: Is this how we want to spend our remaining 25 minutes? I think it
is very interesting, but we are not going to get through the whole
DR. TANG: I want to hear from Blackford in response to this offline.
DR. MIDDLETON: Well, a hard act to follow, so I am going to go from 100,000
feet and dive in.
It strikes me thinking about this is that we are fundamental in the midst
of a sea change. A provider oriented view of data, a provider oriented way in
which data is created, is going to be supplanted by a patient oriented view of
data and a patient oriented means by which many data are created or acquired
and used, et cetera.
So I go back, Paul and Justine, to the original NCVHS depiction of patient
and provider and public health. I think we need to redraw it as a patient
centered view of health care, self and proxy, workplace sources of data, home,
and then social populations, social networking 2.0, whatever you want to call
it. It is a completely different depiction of the way we think about these
data. I think that allows us to think about the principles by which these data
have to be considered in their multiple uses and infusions, et cetera.
I think the same principles apply about accuracy and precision and
reliability, validity issues of data, as well as when repudiation issues will
be perhaps more important. I will always trust your data. I will always assume
that it has not been changed in transit. However, if a patient is accessing
their data, or even creating some of their data, I may be very interested in
changing it. I wish I could change the number of zeroes on my bank account. But
it is that level of non-repudiation, if you will.
So I think that raises a set of issues about how we think about patient
centered data and how we might articulate the principles that we have to
consider around these issues of validity, reliability, accuracy, precision,
integrity, maintenance of validity across time and space.
Then finally, what are the value metrics that we will use to decide these
data of this type are worth gathering and sorting and using. There may be other
purposes and other businesses and all the rest of it and we might be interested
in those data, but we might provide some guidance on this.
This came up the last time we met. I am particularly interested in value
and value measurement and assessment.
DR. CARR: Yes, thank you.
DR. MIDDLETON: So I think we have to bring that on. So after that high
level guide, then let me go a little bit more down to the ground level.
What we have been doing, Partners has had a long interest in gathering data
from a wide variety of sources, not only across the enterprise and all the
typical health care side of the equation way.
We have done as much as we possibly can to create an integrated data model
and gather these data in two primary forms. One is a clinical data repository
to the transactional needs of care management, but also importantly a research
patient data registry, a separate denormalized, more population oriented view
of these very same data that have been cleaned up dramatically from their
clinical source and messiness to be available and used for research.
Now, the third primary focus after clinical and research focus on these
data is to think about how to organize the genetic genomic data, how to marry
the genotypic representations we all have with the phenotypic representations,
and do so in a meaningful way that has value from the get-go.
In addition to that, very tactically, we have had a bunch of experiments
ongoing with personal health records and patient source data capture,
specifically things like showing the patient an abstract of their record in the
form of a personal health record, tethered to the EMR, tightly coupled to our
EMR. We find interesting things. Just showing the patient the med list can
detect discrepancies in the doctor’s understanding of the med list from the
patient’s actual use of meds. In a significant number of times from around
randomized control trials, physicians are informed by just having the patients
review their own med lists, so extremely interesting stuff.
In addition, we find — and this replicates some other findings — that in
diabetes care, not only can you activate the patient to be more engaged in
their own diabetes care in all the usual and right ways, by engagement and
participation in data gathering and submitting a journal to the physician for
their review, perhaps most importantly, that patient involvement activates the
physician. The physician then is more likely to actually do things to manage
the diabetes beds.
MS. MC CALL: We have seen the same thing.
DR. MIDDLETON: So this is a very funny thing; how do we overcome inertia?
Maybe we have to stop throwing water at the provider, and use the patient to
activate the physician.
Other kinds of experiments. The Center for Health has done a bunch of home
monitoring for QBNI and CHF and all this different kind of stuff, and the same
kinds of findings are coming to be.
So I don’t have anything more detailed to offer. I would be happy to send
in more detail if people are interested. But I think we have to reorient — we
have talked about this before — 180 degrees from the provider view of data to
a patient centric view. Maybe it is a concentric set of circles, or maybe it is
five bubbles, but patient, community, social network, and maybe somewhere out
here is provider. I would love to do that.
DR. CARR: Or maybe from your observation it is providers, since we are
using the patient to activate the provider.
MS. MC CALL: Although I would still go back and say patient in the center,
and then maybe there are some different rings around it.
The other thing that strikes me is that given NCVHS and given the realities
of changes in the landscape, but also the realities of our own frequency, et
cetera, we cannot engage in trench warfare on what the right fields are. It is
not our job; we will lose the battle anyway. There are people that have a lot
more knowledge about what the right things are.
I think part of our role is to stay a couple of steps ahead and say, so
glad you guys have made the investment in HIT. I am going to assume you are
going to get that one figured out. Congratulations.
Now, where do we need to go next and start essentially creating a landing,
a beachhead, for what the next issues are going to be. I do believe it is a
telescoping out from a health care and a provider centric to a health and a
person centric, and start basically doing the work to create a point of view
around that. So that is one.
The other is your whole bit on measurement and hey, did it work. This is
CER for everybody. It is not just head to head competitions on giant class
effect in studies. This is any program.
DR. GREEN: I really like what I have been hearing. It can save us a lot of
time, if you want to see a third set of examples that are practical, from the
ground. The November issue of the American Journal of Preventive Medicine.
DR. CARR: Which journal?
DR. GREEN: The supplement to the November issue, 2008, to the American
Journal of Preventive Medicine reports out six years of RWJF work that sounds
like what these guys were just talking about. You can see untraditional sources
of data that, you will laugh when you start trying to do this. It comes from
the YMCA, it comes from unusual places.
One practice group replicated health behavior counseling with LISS. LISS of
course was a computer. One of the data points that came out of that one was
that there was a patient that used LISS every night at 1:30 in the morning.
Talk about a quality measure and access and availability and the patient can do
Where I really want to go is, I think I want to pretend I am on the NCVHS
for a minute as a member. You have got key questions on the agenda here. I did
not set out to organize the set of comments here, Justine.
I think quality is an attribute of health care, is the fundamental issue.
Quality of health is very, very important. I think the Committee for National
Bio Statistics, we have a monitoring function for health, but that is not what
I think the word quality actually means. I think the word quality is in play in
our country and our world in the stimulus bill, and where we are right now. We
have got to do something about an unsustainable, unaffordable, run amok health
care delivery system, and we have got to get some money out of it to do the
sorts of things that Carol is trying to get to.
So quality as far as I am concerned, we should be thinking about as an
attribute, not a real result outcome sort of thing. Quality is a means as far
as I am concerned to the end of improved population health. That is the way I
have started thinking about it.
I have been in a number of discussions at the IOM and elsewhere. We can
spend the rest of our life arguing about this, too, and I don’t want to argue
about it, but I just want to come clean. I am just confessing.
I think what I heard this morning is the visionary leadership that is
uniting bio medicine and the socioecological framework in our time, in our
place, in our country right now. We have really solid data that you don’t get
to health if you don’t address both.
You are the only person that said genetics so far this morning. We believe
that is 30 percent of the variance in whether you live your full life span or
whether you suffer when you didn’t have to. We think that about 40 percent of
the behavioral stuff that you highlighted in 14 different ways, that is 70
percent of the variance. We think that health care is about ten percent of the
But we also have pretty solid evidence that health care serves as a nexus.
It is a point of integration when a lot of things can happen where you can
connect it with the school, with the church, with the work, and personalize it
and prioritize it. That is the answer to the question about the interface with
person centered health care, not person centered health. We have learned about
that and we know about that.
Our structure in the NCVHS is displayed this morning, I think. There is no
really clean way to isolate for me the work of the Quality Subcommittee from
the work of the Standards Committee and the privacy group and the Populations
Committee meeting that we just had. The good news that I see in this meeting is
that they are coherent, and they are tending to be convergent right now.
So I am struggling to understand where the Quality Subcommittee —
MS. MC CALL: Yes, but I am less worried about organizational design right
now, committees and all that, as I am what I heard you start with, which is,
the focus should be on capturing, measuring and then trying to assure quality
of health care. That is what I heard you say, which is very different than
taking a view to say we are going to go beyond health care.
DR. GREEN: I don’t want you to hear that. That is not what I am trying to
say. But we do have new opportunities to assess health care that we need to
But the way this meeting was structured as I read our book, and what we are
going to be listening for this afternoon from my perspective is, we sense, we
smell, we have examples at the table that you can unite administrative data,
the survey data that is population based and that the National Center for
Health Statistics operates, and clinical data through the new world. It needs
to flow in the Nationwide Health Information Network, and out of that circle
that Bob Kolander showed us over and over again, deriving that is our
assignment, as far as I am concerned. Using data to develop the information is
what we should be talking about.
We are enriched with the tension and the conflict between all of these
measures being intensely personal and there being intense interest in keeping
it completely private.
DR. CARR: I think what might be helpful is to say what we are not about,
and then what we are about. I think what I have heard is, there is going to be
a huge influx of dollars into HIT, and there will be lots of granular HITSP
type looking at all these different things and merging information. It is going
to be fast paced and specific. We with our time frame, our expertise and so on,
don’t see ourselves as participating in that level of granularity, so that is
MS. GREENBERG: When you say we, you are talking about this subcommittee?
DR. CARR: Right. But what we do think is that our role is to go where the
puck will be, as Carol has said three years ago from that book. We are hearing
today about some of that remarkable leveraging of new technology, blending of
databases and imposing thoughtful integrative questions on top of data.
I like what Larry said. As we think about where we are, health care is ten
percent of the variance. Genetics is 30 percent and behavior is 40 percent. So
if we are looking to make a difference, what Kayla is talking about is
behavior, and what Blackford was talking about was genetics. That is not part
of the conversation today.
So health care, all the ways of delivering care is getting lots of
attention, but it is only going to be ten percent of the solution, and behavior
and genetics are 70 percent of the solution.
MS. GREENBERG: But I also really liked Larry’s point about — because that
is my song — all of our money, all of our focus, all of our research is on
health care, and it really isn’t the biggest thing that influences health care.
But on the other hand, we certainly know it has got to have an influence.
I really liked your concept of data access. I think it could be. If you had
a person centered health care system, it could be the nexus. I don’t think it
necessarily is for most people, but it certainly has that potential.
DR. CARR: And maybe the medical home and all of that. But I think what is
intriguing is, it is the nexus informed by genetics and the behaviors. I think
that is so key. If it is just about ACE inhibitors and ARBs, we are doing that
with the core measures and all that, but there is this whole other thing about
activity, and even designing cities. We have the note on the stairway telling
you it wouldn’t be a bad idea to walk. Just those simple things makes me walk.
MS. GREENBERG: I want to make a case that this Quality Committee’s work
really could be galvanized around examining the union that Carol illustrated
this morning. She talked about, she just had administrative data, she didn’t
know what something meant, or she didn’t have the rest of the clinical data.
What we believe is that we see a future where she could have that, too.
MS. MC CALL: Oh, yes, I am all about fusion.
MS. GREENBERG: When you do the fusion, the opportunity to assess the
quality of health care and its outcomes, and to get to the value equation
We have got people sitting around this committee that envision how that
could happen. It undergirds measuring the quality. It is a measurement issue.
It is about quality. It is about getting the right data in the right place,
getting these intersections. It was on our agenda.
Part of our chore to listen for today, as we look at these models, what can
come out of the administrative data, where are the holes and where are the
gaps. I think our committee should be making contributions to envisioning how
we get the data we need to assess the quality of the health care delivery
system and the health of the nation in the world that is coming.
MS. MC CALL: We have got it written down here, so let’s not rewrite our
MS. GREENBERG: Let’s stick with the charge.
MS. MC CALL: Right, so let’s stick with the charge. I think what I am going
to do is push us back toward health, monitoring the nation’s health data needs
to identify emerging health data issues, methods, technologies, put a star next
to technologies, to identify strategies for evolution from single purpose,
narrowly focused to more multi purpose integrated shared data.
DR. MIDDLETON: There is a tension though I just want to draw out, that I am
feeling, that I wonder if others feel, or if I am missing something.
On the one hand, I think there is a clear drive toward m fusion and
aggregation and pulling together data from multiple data sources, and then
using it, exploring those large data sets. At the same time though, there is a
pressure towards hyper segmentation or mass customization that has come up
clearly in our conversations with the ACME group in the recent meeting.
It may be in the end that there is a treatment for Blackford Middleton that
is different than yours, Larry, and different than yours. When we think about
this from a traditional epi small cell numbers problem, then there are
different issues about how we trust and use the data integrity and reliability.
So I don’t want to lose that micro data focus, while we think only about
the data fusion kinds of issues.
MS. MC CALL: I understand. I would encourage us to go back and look at this
page, guys. The question is, rather than reinvent this wheel.
DR. CARR: What tab is it?
MS. MC CALL: It is under Tab 4, and it is the last page in the tab. It is
the charge of the Subcommittee on Quality. It says we are focused on two areas,
and that we have near term priorities that are specifically two things, and ask
ourselves whether or not, given the current state, we believe this has changed
or needs to change.
DR. TANG: I would guess that what we heard both in the main committee and
this morning is a reaffirmation of where we are headed. One is person centered,
and two is health and health care.
I think that the novel uses just gives you more hope that you can do
something, one, by combining the data and two, by sharing it back with the
individual. We shouldn’t all of a sudden stop and focus back on the ten
`I think the ten percent, rather than being a nexus, right now it is end of
life tail of the dog, and why chase that? But it could be if we refocus on the
patient centered health, we could have an infrastructure that allows us to
permeate rather than be the end of life tail. That just changes the whole —
and I think it is still where we need to go.
MS. GREENBERG: It is so hard in this committee to talk about the redesign
of health care, which is the first transformation since 1830, without getting
that reaction. I wish we could put that to bed. There is not really a conflict
here. There is not really a conflict about wanting to know if the redesign of
the largest health care delivery in the country works or not. There is not
conflict in wanting to know the answers to that, having the data sets available
to do that. We are taking a person approach that acknowledges the
socioecological framework of what you do at work and what you do at school
What we are looking at here is to quit segregating these things as if they
are different, and to put them into the center of the table and say yes, all of
those matter. This genetics thing is so neglected, and it is going to be blow
things up really soon.
DR. CARR: I am going to be the Chair here. What is the output and what —
MS. MC CALL: What do we need to come away with today, guys?
DR. TANG: One is reaffirmation. I think that is pretty good.
MS. MC CALL: Yes, but we have a hearing to plan.
DR. CARR: And the audience is the Secretary, the customer.
MS. GREENBERG: Can I just make one very quick statement? That is, I
wouldn’t worry either about this organizational thing. You can talk better in
small groups, in subcommittees. But this group can be more the visionaries to
some degree, not that they are concrete. But I don’t think you have to worry as
much as some of the others do about hitting the stimulus package, so it is
DR. TANG: I think we can reproduce the substance of this discussion for the
DR. MIDDLETON: So Paul or Justine, just a question following up on
Marjorie’s comment. This was said earlier. I kind of don’t get it, that we are
removing ourselves from the stimulus part of the package, which is about
quality data reporting. I don’t get that.
DR. CARR: I think that maybe we are speaking too narrowly. There are things
we have done before, stewardship, privacy, HIPAA, those kinds of things, the
connectivity of HIEs, stuff like that. But this piece of it I think is very
noble to put forward as an area of focus and development, taking what we have
with the administrative data and blending it with other data sources.
DR. MIDDLETON: But I am just worried now at the very tactical level. NQF,
we are doing our thing with high tech two, with the quality data set and the
work flow. There are others thinking about the AMA collaborative measures and
what is going to come out of an EMR. I don’t know if NCVHS has ever really put
a stake in the ground about HIT based quality data reporting, and it couldn’t
be more relevant or germane than right now.
DR. CARR: What would that look like?
DR. MIDDLETON: I think a well thought considered piece, white paperish
thing on the principles surrounding HIT based quality data reporting to give
guidance to all the effort that is going to flow right now about how to make
that part of meaningful use really be meaningful.
MS. MC CALL: Can I ask a question, a very naive question? Is that job not
already taken by somebody? What I don’t want to do is wrestle with somebody
over whose job it is to do.
DR. MIDDLETON: It is a fair question. I thought you were going to ask, is
MS. MC CALL: No, I know it is not done.
DR. MIDDLETON: So is it being done by someone else.
MS. MC CALL: Or has somebody else been charged to do it. What I don’t want
to do is scale a hill, only to find out that somebody gets to the top and I go,
what the heck are you doing here. It gets to our role.
DR. MIDDLETON: I guess my observation — Paul, I would be interested in
your response — my observation is that there is a lot of related and
disjointed efforts that may or may not be pointing all to the same goal. If
NCVHS’ mission in life is to define the goal, in this case what would be
quality data reporting from HIT, that is our bread and butter, I think.
DR. CARR: Let me ask this. Susan has helped us put together a primer on
data stewardship. It is nothing earth shattering; it comes off of our earlier
hearings, but it is taking what this group said, here are all the observations,
and putting it together as a compendium, not making a recommendation, but
creating a compendium. Would that serve this purpose?
DR. MIDDLETON: That is one part of it, maybe even a small part frankly,
DR. CARR: I don’t mean data stewardship, but what represents quality. It is
DR. MIDDLETON: I might try elevating its stature from primer to white paper
DR. CARR: We have testimony today or presentations today, the beginning of
a 2.0 hearing about that, and we were talking about going in that direction.
The two things that we have heard about, what Carol is doing, what you have
talked about, to me are things that we don’t hear about elsewhere, and that
need truly to be elevated and that we need to hear more about. So I feel like
that would be a reason to focus right here.
DR. TANG: So Blackford, I think high tech two can cover what you just said.
That was our discussion a couple of days ago, for clinical data for quality
measures, and quality measures that have to be up to snuff to accept clinical
I think that plays a little bit to what Carol was just saying. We certainly
can accept that kind of work as part of our quality charge, but I think what
you were talking about this morning is new and not being done by anybody else.
MS. MC CALL: These are less about the specific tactics. The big theme in
here is whether or not we see an unmet need that we can do in a big way around
DR. CARR: So we will listen to the presentations this afternoon. Cynthia
will set up a call in ten days, and we will then come up with a plan.
(Whereupon, the meeting was adjourned at 10:03 a.m.)