[This Transcript is Unedited]
National Committee on Vital and Health Statistics
Working Group on HHS Data Access and Use
June 20, 2013
National Center for Health Statistics
CASET Associates, Ltd.
Fairfax, Virginia 22030
TABLE OF CONTENTS
- Bryan Sivak, CTO, HHS
- Richard Dale, COO Optum Labs, Cambridge and Paul Wallace, MD, SVP Lewin Group
- Dwayne Spradlin, Health Data Consortium
- Review Needs of Communities from Joint NCVHS Roundtable, April 30-May 2
P R O C E E D I N G S (1:00 p.m.)
DR. CARR: Welcome everyone to the Working Group on HHS Data Access and Use. I am Justine Carr, Steward Healthcare, no conflicts. What we do for visitors is we go around the room and ask you to say who you are and where you are from. That’s for the transcript, and actually I would ask everyone to sit at the table.
MS. KLOSS: I am Linda Kloss a member of the Full Committee, member of the Privacy, Confidentiality and Security Subcommittee and the Standards Subcommittee.
MR. KELLY: Scott Kelly from the CTO’s Office.
MR. DAVIS: Damon Davis at the HHS CTO Office, no conflicts.
DR. CARR: Speakers, go ahead and introduce yourself.
MR. WALLACE: I am Paul Wallace from Optum Labs.
MR. DALE: Richard Dale, also from Optum Labs.
DR. KAUSHAL: Mohit Kaushal, part of the Working Group.
DR. ROSENTHAL: Josh Rosenthal, RowdMap, part of the Working Group.
MR. SPRADLIN: Dwayne Spradlin, Health Data Consortium.
MR. DOWNING: Greg Downing, I am the executive director for Innovation at the Department of Human Services. No conflicts.
DR. MAYS: Vicki Mays, University of California, Los Angles. I am a member of the Full Committee.
MS. BRADLEY: Lily Bradley, staff to the committee, ASPE.
MR. CROWLEY: Kenyon Crowley, University of Maryland, Center for Health Information and Decision Systems, member of the Working Group.
DR. FRANCIS: Leslie Francis, University of Utah, member of the Full Committee and member of the Working Group and no conflicts.
DR. CARR: Bryan is going to call in shortly with some overarching comments, but Greg I wanted to give you an opportunity if you wanted to make any leading comments.
MR. DOWNING: I just want to appreciate the work that’s been going on with the committee. I think we’ve been getting great feedback. Lily and other committee members, have engaged in social media, a breakthrough for NCVHS, as a means of sharing ideas. There have been certainly a lot of thoughts going on. Bryan will speak soon about the importance of where all of this data effort is going with the roll out of many aspects of the Affordable Care Act.
This year has been somewhat of a tipping point in terms of value creation elements that have come out of the data releases. We’ve asked Niall Brennan from CMS to join us again today. This group has played a major role in the establishment of new infrastructure within CMS and other parts of the Department to support new information products and analytic capacities within the Department to help support transformation healthcare throughout the country in a number of ways.
Many of you I’m sure have paid close attention to the various data sets that have been released by CMS in recent months, and the potential implications that brings to redesign of programs as well as consumer awareness of healthcare costs and qualities within their areas and what their providers rank relative to others. Many of you have commented about that. That is not the stopping point for where the Department is going with these efforts, but continued efforts to make new data products available from all corners of the Department.
We’re very appreciative of the thoughts that you all have brought to us. They present new challenges to us in terms of how we optimize our time and resources. Individually, we’ve had really great relationships working with the consortium and annual events that go on. Just during the break here, there were a number of members who have talked to us about upcoming code-a-thons, utilizing the data here. There’s a great deal of momentum in the
entrepreneurial community about that.
We’re very pleased by some of the representatives here who have come from that community and helping spread the word of healthcare is not a forbidden land to go to for innovation anymore. There’s a great many activities going on, and we’re very appreciative, Justine, of all your hard work as well.
DR. CARR: A couple of other members have joined us. Do you want to announce yourselves?
DR. STEAD: Bill Stead, Vanderbilt.
DR. TANG: Paul Tang, Palo Alto Medical Foundation.
DR. CARR: And Bill Davenhall.
MR. DAVENHALL: I am observing today.
MR. SIVAK: Hey, guys, it is Bryan Sivak here from HHS.
DR. CARR: Perfect timing, Bryan. Welcome, and we’re ready for your insights and comments. Thank you.
Agenda Item: Bryan Sivak, CTO, HHS
MR. SIVAK: I want to apologize first for not being able to be there in person and having to run relatively quickly. My schedule ever since the Datapalooza has been ridiculous, actually just before Datapalooza, so huge apologies. I have been getting updates on both the travails of the working group and on the meeting so far, and it sounds like everything’s going great. What I thought I would do to kick this off is to express a few thoughts of
encouragement, which will be me sort of translating from the Secretary and Deputy Secretary here as well as the White House. I think everybody sees the importance of the work. I wanted to give a couple of examples of that.
Last Friday I briefed the Deputy Secretary, Bill Corr, and Secretary Sibelius, on a lot of the activities that we’re engaged in, including some of the data work. Their reaction, frankly, was amazing to see. You guys all probably know the data set that Niall and company released recently, the two data sets actually, the DRG codes for hospital procedures and also the outpatient procedures for hospital pricing. I know Niall will tell you a lot more about that.
One data point that I think is quite interesting is that in the first 24 hours that the first data set was out there, it was downloaded over a 100,000 times, which obviously shows the level of interest in this information. When I told the Secretary and the Deputy Secretary that stat, Bill’s reaction actually was, that’s fantastic, what data set are we going to release next? Which I thought was the best thing he possibly could have said because that basically
gave me carte blanche to go to everybody else and say what data set are we going to release next?
I think it also shows the commitment that the leadership here at the Department has to this idea of open data and data dissemination to help transform the system that we’re all working so hard to change.
That dovetails nicely into the Presidential Executive Order on open data, which came out on May 10. I suspect that you guys are all well familiar with this one, but at a high level what it really does is set the default for the federal government to open data as opposed to closed data, but also getting data out there in machine-readable formats with standardized access, data catalogues that can be federated and can talk to each other, and much more. One of the more interesting components of it to me is the requirement that procurements have language in there that specifies the data generated by applications must be open. I think it will be interesting to see how that plays out over time.
At HHS we’re in great shape on this. We went through the Executive Order, and to be honest there’s not a lot we have to do because we’ve already done most of it. I think a lot of the EO was based on work we’ve done in the past. As I said, we’re in good shape there, but there are a few things we need to focus on. We do need to continue our pressure internally to keep this effort on track in general.
As you guys know, healthdata.gov is our main portal for getting information out there. I saw an email yesterday from Lily to the workgroup, and I know you guys have been having conversations about this, but we’ve been talking a lot internally about what we do with healthdata.gov.
Your voice is going to be critically important here to help us figure this out. We have upgrades planned, which I’m not sure if you’ve heard about yet, or you will, but these upgrades will help improve some of the functionality and set the table for some of the things we want to do going forward.
Some of those things are going to be, for example, the ability to sync with other external data catalogues, such as the one the New York State Department of Health has put out on their Socrata platform. In October, we want to bring in the CMS navigator data sets that Niall can tell you a lot more about. So a whole bunch of things like that are in the works. In order for us to really make healthdata.gov a one-stop shop for all kinds of health data, not just health data that HHS owns and operates.
One of the things with healthdata.gov, and actually with the community in general, that we’ve been spending a lot of time on, is this idea of building communities around data and data sets. To me, this is still a big open question. I think there’s massive potential to make healthdata.gov the center of this universe in a lot of ways, but I think there’s a good set of
philosophical questions to be asked about whether that’s the right direction to take.
There are some pros, definitely, to having this operating on a government property. There are also some cons. One of the big cons is that there are communities of people out there that already exist in other places. To build a community from scratch, or at least from a nascent community I should say, is a lot more difficult than going to where they already are. So, you lose some stuff if you don’t do it on a government site, so I think this is a discussion that has to happen, not just from the perspective of where to host it but also
what the thing is, what functionality do we surface and provide for people to actually have conversations about this data and this information.
That’s all the tactical stuff. The real question is: is this working? Is
this idea of putting data out there, making it available, teaching people how
to use it actually helping us drive a transformation of the public health in
this country and the healthcare system? I think in many ways, my view after
being here for nearly a year now and watching this firsthand is an unqualified
yes. That was really driven home to me a couple of weeks ago at the
Datapalooza. I participated last year– I attended.
It was before I started here, but standing up there on the stage on the
first day, kicking it off, looking out at a massive audience of 2,000 people
that were there for one reason and incredibly engaged, enthusiastic and excited
to be there was one of the more incredible experiences in my life. It really
hit home for me just how much this idea had grown from literally an idea and 40
people in a room and a pile of data to a movement, to a thing that people could
rally around that produces some tangible results. I think some tangible results
themselves came out of that conference.
You guys probably know we had a live code-a-thon at the Datapalooza. We had
ten teams that didn’t sleep for 44 hours and actually tried to solve a very
specific problem that we posed to them using real Medicare data. Those ten
teams actually created some really amazing stuff. The winners were a team of
four docs from MedStar that are actually now thinking about ways that this tool
that they built in 44 hours can be useful in the real world, to ACO groups, to
help those guys make decisions, which is exactly what the challenge was all
about. I think that’s super cool. If there’s any indication that this stuff is
useful, it would be the idea that after 44 hours of coding there was a product
that people will actually get real value out of with a little more development.
I don’t know what else you need to see.
Just to wrap this up, I know you guys have been working with Lilly and Susan
and I think that you’re really on the right track with the ideas that you have
around enhancing the value of the information and the data that we have here.
All of these ideas of new collaborations and crowd sourcing of ideas, these are
things we want to figure out how to integrate into the platform or into the
movement itself. I think the work you guys are going to do and are doing is a
really important art of getting those things discussed and everything collated
and put on the table so that we can move forward on these things.
This year is going to be an interesting one. As you guys know, as of October
1, we watch the health insurance marketplace, both here in the federal
government and across the country. The millions of Americans that are finally
going to have access to affordable care are going to change the landscape of
how care is provided in this country, in any number of ways. It’s our
fundamental belief that the data that we have here is critical to enabling that
transformation. My hope is that you guys will act as more than a group to help
us flesh out ideas, but also to ask some of the questions that need to be
asked, to push us to things we wouldn’t normally think up or do and keep
feeding off them. Having said that, thank you guys very much again for doing
this. Thanks for letting me say a few words in the beginning, and I have a few
minutes that I can answer any questions you guys have.
DR. CARR: Wonderful introduction, very helpful Bryan, and appreciate the
vote of confidence. I’ll open it up to any questions from the group.
MR. SIVAK: I am easy to find, email@example.com, and I know most of
you anyway, so feel free to email, call, whatever. I’m around.
DR. CARR: What we would like to do now, a couple other folks have joined and
then I’m going to turn it over to our speakers from Optum Labs.
MR. SCANLON: Jim Scanlon, from ASPE, HHS.
DR. COHEN: Bruce Cohen, a member of the Full Committee, member of the Data
Work Group, Massachusetts Department of Public Health, no conflicts.
DR. CARR: Thank you folks for coming today. Richard Dale and Paul Wallace
are going to speak to us about the work that is getting underway. As I
mentioned in the agenda, Optum has bought Humedica software analytics platform,
which facilitates the analysis of the huge amounts of data that Optum has, both
clinical EHR data as well as claims data. We look forward to hearing from you
now about what your thoughts are and what are you thinking of and provide an
opportunity for the workgroup to give you some feedback of what we see as
potential or potential obstacles.
Agenda Item: Richard Dale, COO Optum Labs, Cambridge and
Paul Wallace, MD, SVP Lewin Group
MR. DALE: Thank you very much. I am Richard Dale from Optum Labs. I’m the
chief operating officer, and along with my colleague, Dr. Paul Wallace, we’re
going to introduce what is going on at Optum Labs, which is a new group inside
Optum, which we founded initially with the Mayo Clinic. We’ll talk about our
models for collaboration and partnership across the healthcare industry and the
kind of research and innovation that we have to foster.
Again, I’m not sure of the protocol here. I’m certainly happy to take
questions as we go along. Just let me know if you have questions along the way.
You should have copies of the slides in front of you, both in black and white
and in color, so take your pick.
As I said, Optum, which I’m assuming most of you are familiar with Optum,
Optum working with the Mayo Clinic formed Optum Labs, which was announced at
the beginning of the year as an open collaborative research and development
center for the healthcare industry focused around improving patient care.
Although as you know Optum is a large commercial concern, Optum Labs itself,
our goal is not a revenue-seeking goal. It really is to provide a platform for
this research to occur with– I always hesitate to qualify the word
“unique”– somewhere between a reasonably unique and a fairly unique
and a very unique data asset. I certainly welcome your feedback on that.
As part of the founding of Optum Labs, the location of Cambridge,
Massachusetts was chosen symbolically for several reasons. One, to show that
not only was this not just a little tuck-in part of Optum, which is based in
Minneapolis, not only was this not just going to be about Mayo Clinic and Optum
in Minnesota, but this really was going to be a national resource. Optum Labs
was founded in Cambridge as one of the world’s great settings of research and
learning, both in the larger academic case, certainly in the healthcare world,
and also in the biopharma industry. The goal is for us to foster collaboration
and research innovation, and we’ll cover that during the presentation.
I’m going to briefly introduce a couple of the scientific key leaders who
are not here, and then I’ll introduce Paul Wallace. He and I are going to
switch off during the presentation so that you get the benefit of his expertise
about the key areas on the science and research side.
The CEO of Optum Labs is Paul Bleicher. Paul actually was most recently the
chief medical officer of Humedica, the founding chief medical officer there. As
was mentioned by Justine, Humedica was purchased by Optum at the beginning of
the year, and after that integration took place Paul moved over to take on
Optum Labs. You can see his bio here. Before he was at Humedica he was a
co-founding CEO of Face Forward. In fact, I was his co-founder, so there’s a
connection there. Paul started his career as an MD/PhD at Mass General. Justine
can tell you all about Paul in his young days.
Many of you may know Bill Crown who is the incoming president of ISPOR,
previously group president of health economics and outcomes research and late
phase research at Optum, and prior to that at Thomson Reuters MedStat. Bill for
many years was a faculty member of the Heller School at Brandeis.
It’s a pleasure for me to introduce Paul Wallace, who’s here today, who has
been senior vice president of comparative effectiveness research at the Lewin
Group, which is another Optum company and is moving over to the labs and is in
the middle of that transition now. Again, a little bit of his bio is here on
the board. I’m going to turn over to him now; the presentation will start to
cover some of the background of the labs.
DR. WALLACE: Thanks Richard. I want to add a little bit of additional
context. We really appreciate the chance to be here, and I think Bryan’s
comments really set the stage in a sense. The discussion now has moved to
thinking about communities of data. If we were to think back a couple of years
ago, the better metaphor would have been blind men and the elephant, trying to
figure out where the data was and what to do with it. I think things have moved
along really quickly to the point where now, Datapalooza being a great example,
where the discussion is now much more purposeful, and there’s much more
connectedness in the discussion. Where I hope we can contribute to this is
particularly using the metaphor now of communities.
A lot of the discussion from Bryan was talking about how the communities are
evolving using public data. I think there also are going to be opportunities to
think about what are the communities that are evolving in the private sector
and how do we think over time about the bigger community, or the
inter-community opportunities about thinking about what’s being learned in the
private sector, what’s being learned in the public sector, and how do we
actually translate that into good for patients that, frankly, could care less
whether it’s public sector or private sector. I think a lot of this is really
incumbent on us to think about what’s being learned within communities but also
where are the opportunities for inter-community collaboration.
What we really want to do is share with you a little bit of our experience
in thinking about what would a community look like in our sector, beginning
with some of the resources that have been accrued within the Optum businesses
and the Optum community, and share with you a little bit of our thinking, but
particularly to use that as a jumping off point for thinking about where there
may be opportunities to both leverage that learning, share that learning, but
also do things together. That’s really the context.
My background, the communities that I’ve come from, I hung out in Kaiser
Permanente for about 25 years. We were involved in implementing Epic in ’93
shortly after Paul, I guess, at Northwestern. It goes back a long ways, but
also have wrestled with these issues with a long time, and seeing community
there I also had been involved in Academy Health, and again Paul and I
intersect, where thinking about community health services researchers, who have
a particular perspective on the use of data and needs around data. I had worked
at the Lewin Group the last couple of years thinking about aspects of data and
the policy context. For instance, we’ve done a lot of the work using the United
date to support the IOM committee that’s looking at geographic variation to
really say what’s the commercial analogue of the Dartmouth Atlas(?) and to what
extent does variation in the community look different in the commercial world
as opposed to the publicly financed world, and the short answer is a lot, and
reconciling those issues I think is maybe one of those stepping off points to
think about how do we actually work through this.
We feel a responsibility since we have a lot of assets to think about how we
can apply those assets in a meaningful way, particularly apply them in a way
that will create new knowledge. I think as Richard mentioned, within the larger
United Health group, there are a couple of dominant enterprises. There’s United
Healthcare, which is the benefits and insurance business, which touches tens
and millions of people for their benefit coverage, but there also is Optum,
which is a services company, that deals in services and technology and
The Lewin Group, where I’ve been working the last few years, is actually
part of the Optum side. It originally was acquired by Ingenix, which
undoubtedly know, from the past. That’s the lineage for this, but it also puts
us in a position where there are a range of assets, where we also want to be
thinking about how do they best get presented and leveraged in the community.
As we look across that landscape, and we could all create our own list, but
there obviously are some huge opportunity areas or gaps, depending on whether
you think the glass is half full or half empty in healthcare.
I think where we want to go is to think about, okay, we have a lot of
services that we could do now to leverage our current understanding, but where
are the places we’re getting stuck, or where are the places that our clients
are getting stuck that they need new solutions? Are there aspects of, for
instance, having very large storage of data or new analytic tools or new types
of data, where we can get unstuck. What would it take to get unstuck if that
takes a little more flexibility and a little more space than you might have in
a traditional product development mode?
The vision for Optum Labs is to say, where can we deal with some of these
issues where people are stuck? The things that don’t work particularly well,
and this is no new list, but people in the healthcare system don’t play
particularly well together. Collaboration is a huge problem. I came out of an
organization that played pretty well together, but that’s because we owned
everything. That works pretty well for three percent of American healthcare,
and then what do we do with the other 97 percent when you can’t hire the people
that want to play that way? Thinking about collaboration across the whole
healthcare ecosystem is a huge problem.
Just thinking about how do we actually bring data forward in a way that is
an asset. Optum has invested half a billion dollars in creating a data asset,
but it’s not good enough to think about how would you use that today. We really
want to be thinking about where does that create opportunity down the road.
Particularly this whole idea of innovation, I think we’re all still trying to
figure out what innovation means, but I think part of what we know is that it
has a little different dynamic and production, or it has a little different
dynamic than operations. Other industries have thought through this, but I
don’t think healthcare has been particularly a leader in thinking about
innovation except perhaps in a few small centers.
The other part about it is this whole issue of spread and thinking about how
do we actually create spread, but another way to think about this is– I’ve
been in the quality improvement business for a long time, but quality
improvement has always had this metaphor if you just undertake this one change,
then life will settle back into being what you always hoped it would be. We’ve
always been in this metaphor of managing a change, but I think going forward,
is thinking about how do we create infrastructure and understanding and
competencies so that we can manage the rate of change but change is an
expectation. Thinking about that infrastructure has a lot of different aspects,
some of which are technology, some are what technology represents and conveys,
but a lot of it is also the social engineering.
It’s to Richard’s metaphor. You’ve got software, hardware, and web ware, and
managing the web ware and people and thinking about how everybody actually
collaborates is critical. I think what all of you know better than I is that
those things are interrelated. The great technology failures are when you
forget about the people.
Many of the gaps in people performance are when we haven’t leveraged
technologies. How do we work in between? I think we have what I suspect is a
quite similar belief to yours that one of the currencies that helps to connect
to those is actually how we think about data and how we think about
information. That’s the long-winded statement of why we feel motivated to do
As a doctor, I feel motivated to do this because I’m really tired of having
a patient in front of me who is an “n of 1” and being forced to fit
them into a box of “n of many” that they aren’t a good fit, either
because of their clinical characteristics, because of their socioeconomics,
because of their preferences and values. I want to get past this need to force
fit people and be able to match the “n of 1” in front of me to the
many “n of 1”s that are a good match to that person. I want us to
understand how we can create dynamically that reference population for any
patient. To me that’s patient-centered care, but thinking about how we get
there is again one of the major objectives that we have.
Our hypothesis about how we can create a community and move forward is on
this slide. Paul Bleicher is a– I’d say he’s an amateur physicist, but I think
he knows more about physics than most professional physicists. His metaphor for
the labs is to think of it being somewhat analogous to what the Hadron Collider
has done for physics. Basically creating something that’s big enough so that no
single entity could create it, but also catalyzing and attractive enough that
it brings in a diversity of talent and diversity of perspectives to generate
ideas that you wouldn’t necessarily get out of a single sector.
The goal is to create– and seeing also data as being the attractant for the
various participants. The metaphor again would be a table that would basically
be the opportunity to convene– the settings on the table would be probably
research topics or learning topics, but also thinking again about who would you
want at that table when you really wanted to ensure that you had a diversity of
ideas and diversity of solutions. It also builds on our past history of our
prior relationships within Optum. To think about can we actually construct an
environment where the different parties can actually begin to play well
together and is data actually a convening force for doing that? It’s a little
different calculus than perhaps exactly in the public sphere. For us, it’s
thinking about how do we actually create a place at the table for these
That’s what we’re in the process of doing, and it’s thinking about how do we
characterize the relationship that the different parties had both with the labs
as an entity but also with each other. There’s a lot of development work. We
also start with the data. Again, our belief is the data is going to be the
substrate for attracting folks. We have a rich data resource, and we’ll talk a
little bit more about that, but again, it’s the product of many years of effort
and many dollars of investment.
The other part about it is we see it as a national resource because it
reflects the national footprint. One of our obvious acts the reason we could
use them for thinking about national geographic variation is that the data
resource that we have is nationally representative.
The types of data that are in the data resource that we begin with include
the original Optum database, which started with claims, but it has much more
than claims. It now reflects the experience of about 149 million people,
obviously not a full lifetime of 149 million people. At any time within United
right now, United Healthcare, which is the dominant contributor to this
database, there are about 50 million full insured and about another 15-20
individuals who participate in self-funded plans.
There are some other data that also finds its way in there. There are also
some Medicare advantage data, but it doesn’t include, for instance, fee for
service. There’s a large amount of administrative data. It goes beyond claims.
It does include the usual suspect claims statement information from labs and
pharmacy and other ventures in the healthcare system, but there also are a
range of items that have been included in the database that get to income, get
to education level, a variety of SES stuff, and also a variety of geographic
That has recently been complimented by the Humedica data, which many of you
are probably familiar with, but Humedica takes a little different path.
Humedica– I think you’re familiar with the business model, but basically
Humedica works with the delivery systems that have EMRs to think about how data
being generated with EMRs can be transitioned into a common data model. It can
be relatively agnostic of EMR type. The data can be put into a common data
model and then fed back to the participants giving them both reporting but also
giving them benchmarking. Part of the negotiation for Humedica has also been to
also have a portion of that data made available, with the appropriate
permission of the provider groups, as de-identified data in an EMR based
We also know that the Humedica database overlaps pretty significantly with
the United database. There are about five million individuals who are
represented in both the Humedica data and in the United legacy data. We also
have– our design for the labs is to think about who are the other folks who
could contribute meaningful data, and more importantly insights into how to use
the data. That really is the root of our foundation partnership with the Mayo
Clinic and has an integrated delivery system as a research organization.
They bring both data, but they also particularly bring insight. They bring
insight into how to do quality research, but they also bring a lot of insight
into how to get research results into the hands of people who are in real time
practice. It helps us also in recognizing, as we talked about before, that
translation is not just the supply side of figuring out what works and doesn’t
work. It’s ensuring that the hypotheses that are tested are the relevant
hypotheses to the people that actually deliver care. Translation has to be a
two-way street focusing the hypothesis testing on things that are relevant. At
Mayo they have an evolving system around translation, but they also know that
if they hold a meeting to translate a research result and no practicing
clinicians show up, they drop it.
They recognize that people do vote with their feet, and they vote with their
time, and they vote with their clicks. If you can’t package research
information in a way that actually has uptake, it’s kind of like trees falling
in the forest. Part of the design here is to think about the whole continuum of
data use so that better science actually ought to have on one hand
publish-ability, but it ought to have action-ability. A lot of the design of
thinking about the community within Optum labs is to ensure that all of those
different perspectives are there.
It also means that a fair amount of the effort from a leadership mode is
facilitation. These are groups that will have different perspectives. They’re
there because they have different perspectives. The goal is to say, how can we
actually find where there are common elements and common agendas. One of the
other issues that all of us will face within our communities and ultimately in
an intercommunity mode is going to be governance and thinking about how do we
actually create governance that protects privacy on one level, but also
governance that gets to the point where the overall asset is leveraged in a way
that creates community good. That’s why it is not accidental that the mission
for the labs is about improving the health of patients. It’s also why the labs
are set up in a different structure than the rest of the Optum and United
enterprise, so that it creates an environment where a hypothesis can be tested
in a little different dynamic than might exist in the rest of the organization.
We made this available, and you guys see these lists all the time. The
message is there’s a lot of stuff in there. There is fairly robust
representation pretty much across the continuum. There are opportunities to
actually add quite a bit of additional detail that we have tactical plans for
how to actually proceed. This is actually a representation of what’s currently
in the Optum Lab’s data warehouse. The Humedica data is being actively
integrated in, and Richard will talk a little bit in a minute about our
approach to linking data across these different data sets. Hospital detailed
disability data is also being built in.
I should also mention that one of the other Optum companies that I’m sure
you’re familiar with is Quality Metric. Quality Metric is the legacy company
from John Ware’s work with the SF tools. Quality Metric is also a leader in
thinking about and pursuing patient-generated outcomes, so thinking about how
we create the standard 60 degree view of the patient and then make it available
in an accessible way is really at the root of this, but it’s really the
necessary but insufficient part. The exciting part is actually then having met
that substrate so we can actually have the follow up conversation on now what
do we really do with this stuff.
This is just a little bit more detail. What I can’t tell you is what are the
27 data elements that are in an advanced directive, but I have a high degree of
confidence that there are. I think all of you have had that experience of
trying to figure out how to store and move data around. There are a lot of
moving parts. We think we’re up to the task. We also have through the
acquisition and now alignment with Humedica a fairly robust capacity for doing
natural language processing.
Part of it is also to recognize this complimentarily between EMR data and
claims data. It’s a substantial portion of an individual’s care won’t show up
in the EMR in most settings, but there’s detail in the EMR that won’t exist
within claims data. Part of our early discovery is going to be understanding
what are the aspects of that actual complementarity and how do we leverage and
move that forward.
We are extraordinarily conscious, concerned, and focused on privacy and
security of the data. We’re very well aware that we won’t exist if we don’t
manage that. We also have to manage this tension about how do we achieve
protection of privacy and security while being able to actually address the
questions that are relevant on the front lines. The left-hand side of this is
really thinking about– obviously, the safest is to not do it, I guess, on some
level. That’s what my attorney would tell me, but the dilemma is that working
within the safe harbor actually has pretty limited potential for answering the
Then it’s really how can we move forward. We like everyone else have our
evolving approaches to think about how can we go down a couple of paths. One is
working directly with the IRB when that’s appropriate, but also working
indirectly with the IRB and other experts to think about how do we
appropriately de-identify these data sets. Again, people certainly know far
better than I, but the learning curve for thinking about how we statistically
de-identify is a steep one, and particularly when we start bringing in these
other sources of data.
It also may be one of the really important areas for us to share collective
learnings and thinking about what are the credible executable and durable ways
that we actually can achieve statistical de-identification as we attempt to
create these 360 degree views and thinking about how we do that in a way that’s
transparent and engaging of the patient community, not something that gets
sprung on them by some surprise.
We also then with that are thinking very much about how do we ensure that we
have appropriate engagement with all of the stakeholders, which run from
government as regulator on one hand to the patient who’s the actual beneficiary
of care on the other end, the practicing physician and other clinicians as very
important parts of that value change. That’s a bit about the assets, but I
think it’s also important for us to talk a little bit more about some of the
capabilities. Richard, why don’t I turn it back over to you?
DR. CARR: Rich, I want to pause a moment. Are there any questions, specific,
folks would like to bring forward now? Paul?
DR. TANG: You talked about the open collaborative. Does open mean people
have access to your findings and your data or conclusions? What’s open about
MR. DALE: I’ll answer briefly now. We are going to cover it later in the
deck as well. On the one hand we are actively interacting with people in
academia and industry, inviting partnerships for people to come, perhaps like
Mayo and contribute data as well as bring data projects and researchers to the
table. One of the guiding principles to the work we do is that any projects
that we undertake are, all other things being equal and in most cases, designed
to be the stuff, which if there’s interesting results at the end will be
These are not things which are designed for proprietary secret use by the
researcher by Optum. Indeed the opposite. There are many venues in industry,
and Optum is one of them, where proprietary research can be conducted. Optum
Labs is designed where the research will be disseminated and published. If it’s
an uninteresting result, we might just put it out on our own website as part of
an open disclosure of the research we’ve done.
DR. WALLACE: One of the advantages of doing this in the context of Optum is
if someone has a need to do proprietary stuff, there are a lot of other places
than Optum they can do it. We’re interested in finding the things that can be
generalized and are generalizable. We have to obviously think about a whole
range of issues that we could list out, but our goal is to have both the agenda
and the findings be available and to break down all of those barriers for doing
DR. CARR: So three other questions: Bruce, and then Leslie, and then Vickie.
DR. COHEN: I am not quite sure that I understood the answer to this last
question. If I were a researcher from Harvard and wanted access to your data,
how would I do it? That’s just a follow up for Paul. A couple of other
observations, these data sets seem to me to contain a huge amount of
information. It’s mainly based on an individual’s encounter with the medical
care delivery system. In our conversations when we talk about community health,
it’s not only those folks who are in the care system but understanding the
relationship between the clinical care system and public health more broadly
I’d like you to talk a little bit about how you see that intersection
occurring within the context of the lab. Finally, just a minor point, I see
that you’re getting your death information from the Social Security
Administration. That’s probably from the death master file, which is totally
outdated and incomplete right now. If your focus is on a research project, the
NDI is trying to expand its use. Actually, this is a plug for NAPHSIS; we’re
exploring the possibility of developing a more complete product that would
provide appropriate access for administrative uses for more complete and timely
MR. DALE: I will answer the first part of the question and then let Paul
answer the second part. To clarify my answer, if you’re a researcher at Harvard
or indeed anywhere we haven’t put up on the web to please call us if you have
research ideas, but that’s coming. There’s a certain amount of bandwidth that
we have, and we have governance processes about how we look at the research
ideas which were brought to us, but we will have an open door to encourage
researchers to come to us. We started to have conversations between the people
who are in Optum Labs with their colleagues, so Bill Crown has reached out to
the people he knows and Paul is reaching out and so on. Those conversations
with people we know will branch out to open houses and invitations to the
broader community to come to us.
DR. WALLACE: I wanted to second that it relates to the problem formulation.
The dilemma is that most of us researchers are restricted to formulating
problems around problems we can solve with the data at hand. A lot of it, where
this conversation is really valuable is a lot of the meta healthcare problems
cross the public and private sector. They also cross from the healthcare system
to the health system.
Part of the longer-term vision is to think about what are the existing
barriers to linking together or at least aligning those data resources and how
do we begin confronting those barriers. A lot of them are– one of the really
interesting questions to me, I think Bryan mentioned the exchanges, but what
proportion of people in exchanges are going to bounce into Medicaid in a given
year and bounce back, and how are we going to understand since many of those
patients will probably have chronic health conditions, how they’re doing with
their chronic health conditions if they’re existing in parallel universes even
within the same year. Thinking about what are the problem formulations that can
only be solved by moving this forward is probably a way to catalyze moving this
forward, if that’s helpful.
MR. DALE: One other comment, and I am not an expert on all of these things
but I have heard with regard to your comment about people who are not
encountering the healthcare system today but form part, obviously, of the
background of public health, the nice– a characteristic of the administrative
data is we do know who is enrolled within the United enrollment universe. We
can see at least a background of population whether or not they are submitting
claims. We do have the enrollment records as well, so that gives the beginning
of a view into that, but I’m sure not the end of it.
DR. FRANCIS: I very much appreciated your comment at the end about patient
engagement, because of that I want to make a mischievous observation and a
suggestion. The mischievous observation is the table. I tried to figure out
where patients are. Are they the food? Are they the kids hiding under the table
peeking out? Are they somehow represented by consumer organizations? I just
didn’t know. Are they going to get the seventh seat? Who even are they, which
is itself very difficult? The suggestion is this is a perfect place to try to
help think about those issues.
MR. DALE: I think part of the answer is that these slides, many of them
we’ve been using in talking to potential partners and not necessarily in this
setting, but when that slide went up I was thinking to myself, patients are not
DR. WALLACE: One of the great sources of creative tension that’s really
leading the charge of this is PCORI obviously. Being credible in that arena,
which is going to have a major influence on health services and other research
is going to mean actively confronting them. It would be presumptuous of us
right now to say that we have a solution to that, but we absolutely recognize
it has to be addressed. I appreciate your comment about the metaphor.
DR. MAYS: My question is going to piggyback a little bit on what Bruce
raised. I don’t know if it’s a question of how nimble or flexible you are or
just a philosophy about health. Looking at the data warehouse, it’s incredible
in terms of what’s in it. Part of the issue sometimes and what we’re interested
as well is the contextual variables of what got the person there. Knowing what
their primary language is, knowing– we’ve even begun to talk about how whether
or not in the health record we should be thinking about things like zip codes.
I’m sorry Bob Kaplan isn’t here, but we’re beginning to start talking about
what are the psychosocial variables that might be good in an electronic health
record. Some of that will also come from what’s collected. Are you flexible
enough that these kinds of things can go there or this is what makes sense for
you to collect? Where are some of these other things on the horizon?
MR. DALE: We’re actually very cognizant of those. The slides you have in
front of you describe what we have now and the immediate elements on the
horizon. We have talked indeed about exactly those things, and as somebody said
often the most important thing about an outcome is whether you have a ride to
the doctor. We are able to– one of the assets we have as Optum Labs working
inside the Optum universe is working with the Optum data management folks who
are used to working with lots of different kinds of data coming in at high
Insofar as we identify and have made available to us and find appropriate
other sources of data, which allow us to move in those directions, we are– I
don’t know whether it’s easy, but we are certainly able to go down those paths.
There’s nothing fixed about the boundaries of this universe of data. As Paul
said, it’s a big table. We’re hoping to invite lots of people to the table.
We’re talking about registries for example, which may be much smaller
populations but much more enriched for any number of reasons and combining a
small number of registry records in this large universe, those two sets of data
throw light on each other in interesting ways, so lots of thinking about that
in very nascent stages.
DR. WALLACE: One of the other framings of that which actually comes back to
the patient issue, too, is we think one of the advantages of expanding the
frame– it’s not just about how to turn data into knowledge. It’s really how to
turn knowledge into care. Even using the example of thinking about disparities,
in my past life if I wanted to get clinicians to the table I got more mileage
out of geo-mapping performance on a particular variable than I ever got out of
giving even physicians their own individual performance.
Then people get really creative because you bring them to the table in a
little different mode, and they begin to think about different solutions. You
end up with surgeons talking about food desserts. It’s amazing. That’s the test
of whether you’ve really engaged people in a meaningful way to me, but I think
it’s a great point.
MR. DALE: We have a few more slides to go through, and I don’t know what the
timing is here, Justine.
DR. CARR: I think we have some flexib– actually though, I think it is
helpful to make sure we get a couple of these questions in and we’ll get back
DR. TANG: Actually, piggybacking on Vickie in terms of the social
determinants to health, I noticed in your data model of the socio economic
status and you have income and net worth, life stage, life style indicators,
and health risk assessments. How do you get that data?
MR. DALE: So some of that is derived from geographic points that we have, so
we’re able, using standard enrichment– so it’s not based on this is Richard
Dale and this is what I know about Richard because I’m friends with the NSA
this week, but rather based on the enrollment address we’re able to enrich the
data with many of those things.
DR. TANG: So that would determine your– but that’s not direct from income
and net worth and lifestyle indicators or their HRA answers.
MR. DALE: The HRA stuff comes in from a different source. I don’t know
whether you know where that comes in from, but we can get that answer to you.
The socioeconomic stuff is available from third party vendors of information
based on census block and so forth.
DR. TANG: Lifestyle indicators, how do you get that from your address?
DR. WALLACE: That is from HRA stuff, health risk assessments.
DR. CARR: Josh?
MR. ROSENTHAL: This is really interesting. I have seen things like this from
WellPoint once upon a time, and Kaiser was even doing interesting stuff, and
more recently some of the big farmer folks are getting very interested in this
whether it’s Merck and others. The thing that immediately springs to mind is
what’s your end goal here? Is it to create environment for restricted access
much like the government does, so if you have approved qualification you get to
get into the data and play around with it, or is it actually creating public
use files that you want to put out?
When you say it’s for sharing, there are a couple of different lines that
folks in this room tend to think about. Restricted use you have an enclave. You
can prove you’re using for– you’re a reasonably intelligent person using it
for reasonable goals, you get access to it. Pure public data, you put it out.
Is that what you’re thinking or is it partnership and alignment with business
for the mother ship, or is there anyway you can disambiguate what you’re
envisioning versus what’s currently going on versus a couple of other visions
in the marketplace for folks like yourselves?
MR. DALE: So certainly the current vision is the first model we talked
about. It’s an enclave. It’s the opportunity for both academic and industry and
non-governmental and governmental organizations to comes forward and say we
would like to work on this research. We’re hoping to encourage collaboration
between those different people at the table doing that both in terms of our own
bandwidth and how we see the best way to steward the data and the concerns
about maintaining security and privacy to do that really in a controlled way at
I think we haven’t really done much concrete thinking beyond that other than
the answer we’re trying to give to those kinds of questions is that anything in
the future is possible. We haven’t yet really imagined those models. It’s
possible we could move toward some of the other models you discussed, but right
now we’re focusing on getting something stood up where we can show value in the
community, to the community around research where people come with qualified
research ideas. Part of that is because our first partner Mayo, and we expect
other partners who are contributing data, they also have concerns about
governance and how their data is used and they want to see that is properly
MR. ROSENTHAL: The thing behind my question I’m sort of driving at is if it
is an enclave model, and you look at government as an analogue, what’s the
purpose here? It’s a better mousetrap. You have better data, better partners,
better granularity, and you can do what you want to do. It serves a purpose
certainly. When I’ve seen this stood up by other private industry folks, the
criteria– there are only so many people that get restricted access. That’s
determined by bandwidth and stewardship and all these things, not usually by
I’ve seen it stood up by other folks like yourselves. That’s always been the
thing behind the scenes. The partnership criteria is commercial applicability.
It’s presented as if it’s open, which may or may not be the intent, but a few
years down the road it certainly doesn’t work out that way. One way of working
around that is actually creating a public data file, a de-identified file that
you can send out. Do you have any thoughts around that and disambiguating from
other folks that have tried things like this in the past? Does this make sense?
Do you know what I’m talking about?
MR. DALE: I know exactly what you are talking about. I think that part of
the answer will have to be how we walk the walk. I’m going to give you the talk
here, and I wouldn’t even do that, interpolate that, but I hope you’ll see some
of the partnerships we’re working on, which we hope to be able to announce
later this year will be people for whom there are clearly no commercial strings
attached, and that will exemplify the fact that’s not– we’re Optum, and who
knows? We’re going to find some of those partners are also Optum customers. I
think we’re also going to find that some of the partners are not in the
universe of customers, and we’ll be spending time and effort working with them.
MR. ROSENTHAL: You may not be opposed to creating a public file where you
just turn it loose to the wind, and anybody can access that–
MR. DALE: The answer to that is more about governance, security, and privacy
right now. We don’t see how to do that in a way– maybe you’ll come to us with
a great idea for how to do that.
DR. CARR: I have Walter, Mo, and then Jim.
DR. WALLACE: That will be also be driven by the requirements of the
partners. Our goal is to have a group– our hypothesis is that the more diverse
the group of partners, the more our governance will encourage us to make this
as broadly available as we can. That’s also why, Frank, to be blunt, why
creating some aspect of public-private partnership is probably important in
order to manage that. That’s part of how we think about it.
DR. SUAREZ: Sorry I missed the early slides. Of course I know Paul. We’ve
talked about this. The question I have is about the relationship of this data
warehouse with some of the other data warehouses that are built out there.
There’s the healthcare cost initiative or healthcare cost data– there’s all
the multi-claims database systems that are being collected. There are all-payer
claims databases that are being built in several states. How does that relate
to this initiative in terms of– is it the same data? Is it replicating some of
DR. WALLACE: I can answer in a couple of ways, some of which are tangential,
but we’ll back into it. One of which is we have been involved at Optum and at
Welland with the multi-payer claims database, which on one level was about a
technical initiative, but it’s actually been about a great deal is social
learning. What’s the both tacit and explicit knowledge that you bring forward
to these things when you can’t help go through that without having that also
influence how we would think about managing this kind of a data asset?
The second would be we want to put ourselves in positions where there are
opportunities to test the intercommunity relationships. For instance, one of
the opportunities that we would hope to be able to participate in would be
things as PCORI creates the network of delivery networks; one of the ways to
actually learn about the best way to interrelate networks is going to be on
those playing fields. We’re also interested in going after funding
opportunities, but really more of relationship opportunities where those things
are going to be actively in play. Some of that we can drive from the way that
we seek and attempt to create relationships.
There also are larger forces that are creating those relationships, and we
want to do both and figure out how we can at least compete. One of the tests of
our ability to credibilize our message is being able to compete in those
forums. It actually goes back to the earlier question to about how would we
make things available, how would we pull patients in? We want to put ourselves
in the setting where we have the internal creative tension to do that, but we
can also be at a table with other folks that are dealing with the same thing
who are in a position to share the information.
MR. KAUSHAL: In five years time, what does success look like, which you
might not know as yet? The second question: how do you envisage working with
us? How can we help?
MR. DALE: So I’ll answer the first question with some personal thoughts
based on some early discussions we’ve had around the table. Even we internally
have not yet stated those in an explicit way. I think that we’re excited
about– we would love to be able to say in a few years that this change in how
these particular patients are being treated is a result of research which was
done, projects which were undertaken by which that knowledge was diffused into
the clinical system.
I think that publication is the currency in science, but we would like to
move beyond that and really see some outcomes and therefore perhaps exemplify
some new methodologies, which allow other people to bring other knowledge,
which are created in other great institutions more quickly. The cartoons I’ve
seen about diffusing knowledge into the healthcare setting generally involves
skeletons falling to pieces with cobwebs on them in the corner while we wait
for knowledge to get out to the people.
DR. WALLACE: I think– this is an indirect analogue– but I think it’s like
the difference between how in quality we’ve measured hemoglobin A1C testing,
where now we’re thinking increasingly about medication possession ratios and
adherence. I think that honestly publication numbers are like measuring
hemoglobin A1Cs. It’s not the end game. It’s a means to the end, but the end
game to me is more around how robustly are we able to respond to front line,
real time questions that real clinicians and real patients have. I don’t know
what those metrics are. I think it’s going to be every bit as difficult as
figuring out how do we measure adherence.
I also think we’ve got to get in the hunt to figure out what it would look
like. On some level it has a structure, can we actually relate to the front
lines? It has processes about what’s the information flow, but it also has
metrics like if we actually change practice. I’ll cite another example, and I
think many of you are familiar with the work that Cincinnati Children’s has
lead around thinking about inflammatory bowel disease.
They have a really cool metric where they can show that kids that have
inflammatory bowel disease have improved remission rates because of the
influence of their network. That, to me, is the kind of outcome related science
we ought to be looking for. I would hope that our metrics have that kind of
outcome basis that’s complimented by publications and all the other things that
sort of move with the lever. I think we also have to figure out what those true
outcomes look like. That’s my thought about it.
MR. DALE: To the second part of your question, maybe we can reserve that.
Let’s look at the rest of the slides. To a certain extent, I think we’re here
to listen to what ringing bells here are most resonating.
MR. SCANLON, J: Back to the question of more or less comparable efforts
conceptually. As Paul knows, HHS has been working on a proposal for a
multi-payer claims database. Everyone thinks this is easy to do, and it’s just
immensely difficult. Do you envision in your model that you would be– and this
doesn’t have to be limited to claims, obviously. If you could do it with
claims, you could do it with electronic health record data when it becomes
available in the same partnership way. So, currently you have Mayo and you have
United claims, do you envision partnerships potentially with other payers who
have claims data and ultimately electronic health records, or is this something
that’s almost so difficult to do we have a lot of different–
MR. DALE: We do in fact envisage that, and there will be a slide in a
minute, which shows the Venn diagram with more bubbles.
DR. WALLACE: The other case, too, would be the Humedica data actually
reflects the EMR captured experience of 40 different delivery settings. We’re
thinking about how we can continue to grow that but also have specific focus
partners where Mayo clinic is–
MR. SCANLON, J: We thought having Medicare data in our database would help a
lot of others to come, but it’s still been very difficult because of commercial
arrangements and other things.
DR. WALLACE: I hope it would require a little along the way.
MR. BRENNAN: I don’t know if this is an appropriate time for me to comment
or not. This is Niall Brennan at CMS. We’ve always had a number of initial
conversations with the folks at Optum. We are heartened by their commitment to
Secondly, as a data hound, I find the merged electronic health record claims
data in particular very interesting. It would be great if we could figure out a
way for other payers to work with, awesome, but of course the devil is always
in the details.
DR. CARR: Thanks, Niall.
MR. DALE: The second part of slide 10, I’m going to beg your forbearance for
a moment and just touch on the technical aspects of how we both secure and link
the data, because that underpins some of what we’re bringing to the table here.
The slide itself, after you’ve seen it– it’s one of those things that after
you’ve studied it for several hours, it’s entirely obvious and clear. When
somebody makes a data contribution to Optum Labs, what we do is we provide them
with a standard cryptographic algorithm. It may even be an open source piece of
software, which uses the FIPS 180-4 SHA 512 algorithm, as it said here on the
slide. We give each contributor the same SALT code, which is the magic number
which starts the algorithm off.
They use the SALT code and the hashing algorithm to encrypt all of the
identifying fields, the key fields which we think of as identifying a patient,
so the name on the address, the social security number, and the data of birth.
What it means is if my record is in one database, and all of those fields which
refer to me are encrypted in a particular way, and if they’re in another
database, those same fields which identify me are encrypted in exactly the same
When they’re contributed to Optum, we do not know that it’s the Richard Dale
records, but we are able to see that there are two records from two different
sources, which match each other. In that way, we’re able to find those which do
match each other. Obviously there are some number of misses, because in one
record the address is done one way and in one the other way. We try and
standardize those and there are transposition errors as with everything else.
Because this is a research database, those false negatives matter less than
they would elsewhere.
We are able to find a good number of, an expected number of matches where we
can link those data together. So the 5 million lives that Paul mentioned before
which overlap between the clinical data we have and the claims data, are done
on that basis. Once we have made that link inside the Optum Labs database, we
then re-encrypt all those identifiers. If you’re a data contributor, you
obviously have seen both the unencrypted data and the encrypted data through
the first cycle. You could use that potentially to do what’s known as a
directory attack and work out what’s going on in the database.
By the time your researchers or you even as a contributor get to see your
data again, those encrypted fields have been re-encrypted a second time.
They’ve been linked and then re-encrypted. There’s no way for you to match what
you sent as an encrypted record with what you had as a source record. Finally,
and we’ll talk about this, we put the data into a virtual data environment
around which it’s very difficult for you to actually pull it out and rematch it
with your own data. We’ll talk about that in a moment. That’s all a bit
complicated, and I’m happy to stop again to answer any questions about how we
do that if it’s relevant.
DR. STEAD: What do you then do when additional data comes down out of
identified land for that person? Does it go back through the double stage
process and then get matched up, or is that not possible?
MR. DALE: We maintain an administrative view which no researchers have
access to and is maintained in this highly secure kind of de-militarized zone
so that we’re actually getting updates from our data contributors, both the
insurance data and the claims data on a regular basis earning new contributors.
It’s rematched each time. That data is secured to the same levels– Optum is
obviously used to maintain databases with fully identified data. That’s our
business for much of it, and so we’re using all of those standards and security
to manage that data.
MR. ROSENTHAL: Richard was actually one of judges at our boot camp for the
HDI things. I tend to give him the benefit of the doubt.
MR. DALE: Here is the answer to the question we just had before. We
absolutely are looking forward to inviting other data contributors to
participate in the partnership, as well as research partners and just
interested partners who want to have a seat at the table. We hope over time
that other kinds of data will be added and certainly other payers and other
providers will bring their data to the table in the way that United has and the
way that Mayo has. I hope certainly a year from now, if not in a much shorter
time frame, that there will be announcements that will indicate we’ve been
successful at bringing people to the table and making this happen.
DR. COHEN: Have you talked to the states that do have all payer claims
databases about access to those data, or would that be duplicative of much of
the information you have?
MR. DALE: It was not necessary be duplicative, and we would love to have
those discussions. I’ve been trained always to punt on specifics on who we are
or are not talking to, but it is absolutely in our agenda to have those
conversations and find willing partners. We’d be very glad to.
MR. DAVENHALL: My question relates to your Optum Labs data warehouse side.
I’m curious, since this is a lab, which is really great because you can
experiment in thinking about how you might incorporate crowd sourced data that
will be coming down the pike, other kinds of human services data that’s now
becoming available, will there be opportunities to build that into your model?
MR. DALE: As we were talking about before there are many different kinds of
data, and patient reported data is the first obvious next one, but those other
classes of data are relevant as well. We were talking before, Paul and I, about
natural language processing and how bringing data in from all sorts of places
could be exciting. We’ll talk in a minute or two about the governance of how we
bring data in and what’s important about how we do that. Again, at the moment
we’re limited only by our capabilities. Certainly our vision is to be open to
all of these possible conversations.
DR. WALLACE: I think it also gets at how all these things are dynamically
interrelated. For instance, I think we all think consumer data would be an
interesting thing, but how could you ever do that if you haven’t thought in
your governance about how you’ve ensured that you’re looking out for the
interest of the consumer. That was sort of the learning that we all had in
MPCD-2, in each step you go down the path, you have to go back and rework all
of your previous concepts about everything else you thought you knew before.
That’s the good news. I think if we want to go down this path we have to be
prepared to do that.
DR. CARR: Let me do a couple of administrative things. A couple of folks
have joined the table that we didn’t identify for the transcript.
DR. SUAREZ: Walter Suarez with Kaiser Permanente. I’m a member of the
DR. VAUGHN: Leah Vaughn, member of the Working Group.
MR. DAVENHALL: Bill Davenhall, ESRI, member of the Working Group.
DR. CARR: I know a number of folks are going to need to leave around 3:15.
We had just a very stimulating presentation, also by Dwayne, and I want to have
an opportunity to spend a little time thinking about how we work together.
We’ll have a break at 3:15, and I hope that a number of you can remain because
we have the work that Lily put together and some other overarching things to
discuss. What I’d like to say is as you look at your next set of slides if you
can pull out the ones that you really want us to respond to, and then we all
have hard copy and it’s on our website, so we know where you are and we can
give you additional feedback.
MR. DALE: Right, and we should be able to move through these fairly quickly.
So this slide is the obligatory slide with numbers, very big numbers, terabytes
and gigabytes and those sorts of things. I’ll be glad to talk about these to
anybody who wants to. This slide I will spend a minute on. We start with the
source databases at the bottom of the slide, and we go through the
de-identification and ingestion I described before to create what we call the
Optum Labs data warehouse, which is this linked logical database.
From that we go through a very specific process of defining research views.
It’s at that point that we get an outside statistical consultant, one of the
leading consultants, and we work with a few but one in particular, to certify
the views we are providing for researchers meet the standards of statistical
de-identification so that when you’re working with that research view and it
goes by the definition, which is written into the regulations that I won’t try
and remember, is statistically de-identified.
Every time we add new data, new columns of data, not more of the same data
but new types of data, and we create research views which provide access to the
linked database, we are focused on making sure that it provides the statistical
de-identification. I think this committee is more expert than I about this.
When I talk to my friends who are not so familiar with this, I say the way we
do this is we allow the researcher to chose the view which focuses in on those
attributes, which are important to their particular research so we can give you
more demographics and less geographics or vice versa so that there’s enough
ambiguity as it were about each individual record, which is shown in that view,
that it is statistically impossible or meets the statistical de-identification
threshold as certified by our outside consultant.
The examples are a state view, which might have– showing records to the
state level might have more richness of other attributes, whereas showing other
attributes to the zip-three level, which is more geographically specific will
have less identifiers than the other areas. If you’re focused on geography you
would head in that direction.
For each project we create what we call a sandbox, which has the research
view relevant for that project and the statistical tools the researchers are
looking for. Then we provide access through a virtual computing environment.
None of our researchers actually take the data away from the labs. They come to
the labs either physically or virtually and work with the tools and the data
inside a virtual computing environment which has the very high security
capabilities that we have, the auditability and the certification that they’re
not linking the data with other data, which might inadvertently allow for
re-identification. That’s one of the reasons that so far we haven’t gone down
the path of giving people cuts of the data.
DR. COHEN: So can they pull out individual level data or can they only
remove aggregate data from the lab?
MR. DALE: So you can only remove– at the end of the day you’ve done your
statistical analysis. The only thing we allow you to remove are summary tables,
and so on and so forth. In the sandbox, obviously, the individual rows are
there, but you can’t take them away and you can’t put other things in the
sandbox, which haven’t been certified to link under these research views.
DR. COHEN: That is very much akin to the research data center model here at
NCHS, which is quite successful.
DR. WALLACE: We array all the tools that you can imagine needing, and if it
isn’t there, we’ll get it for you, too.
DR. SUAREZ: Do you have an IRB?
MR. DALE: We don’t have an IRB because we don’t have PHI.
DR. SUAREZ: You don’t have PHI.
MR. DALE: The database is all de-identified.
DR. SUAREZ: You have PHI before it is de-identified. Once it’s de-identified
and has this unique identifier, some people still have PHI. I think it’s an
important consideration for purposes of determining research and usage of the
MR. DALE: We’re very cognizant of the regulatory environment in which we’re
working and the overall responsibility we have with all of this data. We’re
certainly intending to do all the right things in all the right ways around
that. We would probably need to get into the detail of what your concerns are,
and hopefully I believe we would be able to respond in a way that would make
you feel comfortable.
MR. CROWLEY: A quick question, within the environment how much flexibility
do you envision the researchers having to what they want to share, not share,
remain proprietary throughout the course of the collaboration?
MR. DALE: Just to clarify the question, what is the thinking behind a
researcher needing to keep something proprietary? Is this a matter of the race
to publish, or is–
MR. CROWLEY: That would be one example, or if there are things that are
forming or something that might be proprietary or being commercialized in some
instance, then maybe that part of the overall research agenda and research
program they may not want to openly share at some point.
MR. DALE: So we do have the flexibility to allow for projects to take place
under those kinds of frameworks. Each project when it gets considered and
approved, if there are considerations like that because somebody perhaps comes
to us with some half-baked intellectual property which use of this data can
really help with, we’re absolutely willing and able to consider allowing that
research to go ahead and adding that element to it. That’s all agreed up front
and clear up front rather than halfway through suddenly somebody changing the
The goal is that even though there may be– and we actually are actively
seeking both innovation, translation, and commercialization– that the core
research underlying those kinds of things will indeed be published over time
even if there’s also some commercialization which takes place. We hope to be
open about all of those activities. In the end it will be clear what we did and
why along the way. If there are good reasons for some of it to take place in
that kind of setting, that will be fine.
DR. WALLACE: I think a lot of these questions reflect that learning curve
that we’re all on where we have particular principles. We’re starting out with
the best policy we can have, but it actually goes to what are the metrics in
five years. I would argue that one of our metrics in five years is that all of
our policies have undergone significant evolution because we’ve had to deal
with a lot of these issues. I think it’d be presumptuous to assume that we have
the policy solution to every nuance here, but part of building the capacity and
governance is actually the ability to adapt and then be accountable back to
those core principles.
DR. MAYS: I just want to get a sense of the public-ness of the use of the
data. In a couple of instances where I’ve wanted as a researcher to use data
from a plan I have to sign a confidentiality agreement, and what it says is
that I can’t publish the data sometimes if it has negative findings, if the
data will in some way reflect negatively on the group. Do you have similar
MR. DALE: We don’t actually. One of the things we’re proud about is
certainly in the policy documents we’re still drafting we’re very clear that
the labs governance committee, which will have many of our partners involved in
it, will want to be able to review the publications for accuracy and so forth,
but we will not have a right to stand in the way of publication. We’ll ask for
a right to be able to say this is incorrect and you need to correct it.
If it comes down to a matter of he said, she said, the scientist has the
right to publish. We’re going to absolutely allow that. That works both ways.
First of all on us and on our partners that we’re not going to put those
conditions on, but also for people coming in that they understand that we want
them to publish, and we expect them to publish and we don’t want people doing
proprietary research which they’re going to hold to themselves, and so it’s a
two-way street there.
I’m not going to go through this slide in detail. There are great
opportunities, and we’d be glad to talk about this, these are what we call
building blocks to partnerships. In particular to the questions which were
asked before about flexibility, the next slide shows we’re very happy to build
flexible partnerships, and each partnership may look different, back to Paul’s
“n of 1”. It’s not just chose plan A, B, or C, but rather what data
do you have and want to bring to the table and contribute? What kind of
research do you want to do? Do you want to sponsor research for a particular
reason? Participate, collaborate in research? Do you want to be involved in
translation and commercialization? All of these different things can work
together, and we are looking forward to having many different kinds of
DR. WALLACE: We probably should mention one reason that we got into this
initially with Mayo is they’ve been very generous both with their time and
their thinking, but also with their opinions. Those of you– they’re not a
passive group. I think it goes to that model of creating collaborative
governance and being able to be flexible in how we partner.
MR. DALE: This is another slide, which I won’t go into detail on. It’s got
lots of great technical terms, all of which are fabulous. We’d be happy to talk
about these and certainly partners who are contributing data spend hours, days,
weeks, and months working with us ensuring their data will be properly and well
looked after. This is the point where we mention that Mayo has not trusted
outside partners with their data before and have become not only satisfied but
delighted with how we’re looking after their data and how we promised we would
and how it’s turned out we are.
We invite you to come and visit our wonderful space in Cambridge,
overlooking the Charles River. If you come in a couple of weeks on July the
4th, you’ll have a great view of the fireworks. More to the point
we’ve got great collaborative space for researchers to come and rub shoulders
and collaborate for symposia and events. We hope to have many public and
invitation-only events and would certainly welcome any and all of you to talk
to us about participating in those, sponsoring those, or leading events which
might be of interest, which fall into our mutual interests.
One of the things which we’re excited about obviously is the access to
resources we have in Optum. Certainly in the beginning as our community of
researchers starts small, but even as we go forward, Optum by virtue of the
business we’re in has many qualified scientists who work across the industry
who we can bring in when their expertise may be relevant to help us formulate
or conduct research and for people who are often in the day to day world of
business the opportunity for these people to come and participate in lab’s
research is very exciting for their own career and professional development. We
have a great opportunity to take advantage of that within the Optum community.
I’m going to turn it over for the last few slides here to Paul to talk
specifically about how the research process works. Hopefully we can do this in
a few minutes and then answer more questions.
DR. WALLACE: I think we can skip to the last slide. I think we’ve covered a
lot of the nuances. I just wanted to show one slide just because I think it
gets to the kind of stuff we hope to be able to do.
It’s really looking at the network analysis over there on the left hand
side. This is some work that Paul Bleicher did using Humedica data. It was
really looking at a community of primary care practices, mapping relationships
using referrals and other communications that were traceable within the EMR,
looking at how those practices aggregated and then asking, how do those
practices fall out in terms of readmission rates. With the greens having very
low readmission rates, the reds having very high readmission rates and then
everyone else in between, there are some really interesting patterns.
What we want to do is to figure out for things like that that we haven’t
been able to get at well with trials and other observational methodologies
typically say what is it about these guys that have a low readmission rate,
what are they doing in their practice, and can we use that for hypothesis
generation? We think this is a kind of technique we’re going to need to use for
things like multi-morbidity, where our sense is that multi-morbidity is
actually more of a tale of a curve problem than it is a convergence of common
What we really want to do is figure out who are the folks that have figured
out the calculus of resource use and conceptualization that allow them to have
whatever indicator we want to look at in a population of people that have
multi-morbidity. We think analyses like these that are going to get us there
much quicker than torturing current performance measures. That’s the kind of
stuff that we want to create an environment where tools compliment the data but
particularly are amplified by the people that we get to the table. That’s the
long and short of it.
DR. CARR: Dwayne, perfect segue, so whatever your question is but then I
would say let’s see if we can find convergence between what Health Data
Consortium is doing and what we heard here now.
MR. SPRADLIN: Probably more like a vote than a question, but I think it was
Albert Einstein that said if he had 60 minutes to save the world he’d spend 55
trying to ask the right questions, five minutes trying to answer it. You’ve
created what I think could be an enormously valuable national resource. It
feels like the infrastructure around how we identify the right questions, and
then you mentioned the large Hadron collider, and the image I have in my mind
is a radio telescope array or a super computer where we try to identify the
right sets of questions or make sure there’s a group in the research community
asking brand new questions.
Again, at least 20 percent of the hours per year on the super computer– how
do we make sure that we’re taking this and getting very good and in some cases,
novel kind of question asking into the system so that we actually see the value
make its way back into the health system?
MR. DALE: So as we said before, we are at the beginning stages of starting
to engage with researchers and academics and other members of the community,
encouraging them to ask questions. We have lists and lists of questions, and we
hope to throw most of them away and have them replaced by questions being
brought in from the outside. Initially, through working with Mayo as our
founding partner, but as other partners join in and partner with the labs, they
will be part of the team which are looking at this hopefully large volume of
questions. It won’t just be us. It won’t be Paul. It will be the group looking
at which of the exciting questions to ask.
We hope it will be 80 percent, not 20 percent, that will be the novel,
interesting questions. Again, it’s how we walk the walk. When you look at us
six months, a year, and two years from now you’ll tell us whether we go that
DR. WALLACE: To build on that, I think a lot of it is what we’re really
interested in, is where are people getting stuck. Particularly where have
people been stuck. A lot of what we’ve done in health services research is to
keep trying to run into the same wall and hope that we go through this time.
The challenge is here now that this might actually be a ladder where we can
climb over the wall. What are the things that we run into head on? You think
about where are the places where the healthcare delivery system is continuously
vexed? It’s multi-morbidity. It’s the combination of mental health and physical
illness. It’s how do we rationally understand and deliver end of life care.
I think it is how do we think about– there are a variety of other things
like that, but I think we want to at least devote a significant portion of the
portfolio to places where people are stuck. A lot of that means, too, that it’s
not so much we need to have the conversation with researchers, but we
particularly need to have the discussion with people that are closer to the
front line. I do a lot of guideline facilitation. It’s this continuing thing
for over 20 years now, where when you actually get the people delivering care
identifying their key questions that they would like to have answered in a
guideline, half of them the only way you can summarize the evidence is it
really sucks. At some point you kind of go that’s the demand side for
knowledge, and all of knowledge management is stacked up on the supply side.
A lot of what we’re interested in, and I think what you’re talking about, is
really how we rebalance a little bit of the demand side and the supply side. I
think the diffusion approach that we use in this country is trickled down
knowledge diffusion, and it works just about as well as trickle-down economics.
That’s a framework or a metaphor that I think we have to think about. Are we
really creating the stuff that people need on the front lines?
MR. SPRADLIN: Just a thought maybe to help set up the spirit of the next
section, but I didn’t use the metaphor of 20 percent of the time when the super
computer accidentally — I think it would be a wonderful thing to commit some
percentage of time on the collider and to make sure you’re getting the real
novel questions asked, maybe work with other organizations, some in this room,
that could help in a sense, run competitions or other things that could help
identify what some of those key questions are and essentially dedicate a
percentage of the time on the asset to ensuring that we’re getting those kinds
of research endeavors and important questions air time on the platform. I think
something like that could help also really present this as a public resource,
as opposed to one that’s essentially governed predominantly by the structures
you described earlier.
DR. WALLACE: It actually goes back to Bruce’s comment. I think a lot of
those are only going to be approachable if we think about a broader frame for
data being accessible, too. A lot of the community-based issues really require
us to take a broader view of the community. I think creating that agenda will
help us understand what really are the data needs to get at those things that
MS. KLOSS: I am looking at slide 20, the one we skipped over. I was
wondering if you’d thought about the unique opportunity you would have to begin
to develop some insight about the data themselves. We’ve always talked about
once we get enough data and electronic health records and begin to merge it
with administrative, we’d begin to understand what’s useful and not useful,
what’s vulnerable to quality, what some of those data issues are. As an
advisory committee on data, it seems to me there could be a whole
administrative simplification operations block here that would help us get
smarter about data and begin to look at ways to simplify.
DR. WALLACE: Do you have thoughts about how you’d begin to prioritize– what
are some of the–
MS. KLOSS: As a disclosure I do some work with Humedica. They’re clients on
the front end. When the data begins to be extracted out of the electronic
health records and gets cleansed, it shows the data quality problems in the
electronic health record. They’re just revealed to the world.
I think what health care organizations are doing is they’re going on to
analyzing data out of the cleaner data set but they’re not looping back to how
to improve the feeder system so that our data gets better over time. I think
it’s very difficult to do that in the day-to-day operational environment. There
may be some ways through the labs to kind of look at this aspect of the
DR. WALLACE: There’s a sort of built-in type 2 error in looking for certain
kinds of things because they don’t exist in the data. It’s like we’re never
going to find aspects of community services in healthcare claims data no matter
how hard we look. On one hand there’s how do you create the complimentarity,
but also the other part is going to be– looking at a problem through different
lenses is going to tell us– give us different results.
I think one of the other issues is going to be looking at– geographic
variability is probably going to look quite a bit different when you look at it
through the lens of EMR data than looking at it through the lens of claims
data. We’re going to have to make some decisions about where are those data
sets complimentary versus where are we actually burning effort and burning
cycles by collecting data we really don’t need.
MS. KLOSS: Even thinking about how we model the use of vocabularies,
classifications, other kinds of tools, that might be useful on the front end if
we can learn how to optimize how they get used. There might be another block on
DR. WALLACE: One of the other issues that was brought up before on data
quality is where are the better sources? We understand the issue with the death
files from Social Security. There are pragmatic decisions you make along the
way, but as there are better things that come along we absolutely want to trade
up to what are the better sources. That also means you have to trade out the
sources that aren’t contributing. Is that consistent with what you’re thinking?
DR. VAUGHAN: I wonder if you can answer a few things. One, can you clarify
does it say “for profit” or “non-profit enterprise? Who would
you regard as your principle competitors in this space and how are you
different are better? What is your ask?
MR. DALE: So Optum is a full private enterprise and Optum Labs is a
subsidiary of that, and we’ve talked about whether we could. It may be possible
or not to form a non-profit subsidiary. Those are difficult things to do, but
right now we’re for profit. The labs itself does not have a regular target or a
profit target within the unit, but legally that’s the structure. We haven’t
done a competitive analysis, who are the other people with interesting data who
we’re competing with.
We’d like to think that we’d have the opportunity, as Paul mentioned
earlier, to cooperate with other people who have or are developing pools of
interesting data. One of the informal mantras we have is if you’re talking
about competition you’re not thinking about the labs right. The rest of Optum
is thinking about its competition all the time. We’re looking for opportunities
to collaborate and we don’t think that– if somebody else has got a better data
set out there then we would love to make sure that ours is enriching that
and/or vice versa.
MR. DOWNING: I stand in collaboration for the people who can work at the
scale of this in population health, but one of the things that’s been striking
me over the last year or so is the science about the methodology that’s used in
these analytic frameworks. What struck me about this opportunity was the power
of numbers from different sets of variables. It seems to me that the science of
this to inform policy or in practice decision, what have you, in the future is
going to be based on some really well-calibrated principles about what models
and what data sets and what quality of data is needed to go into answering
I don’t know this body of research, but it strikes me– I haven’t seen it
yet, but what it reminds me of is in the early days of applying whole genome
analysis to biologic questions of relevance to human health that being able to
take different sequencing methodologies you would get different calls on things
based on what sequencing methodology or the sample prep that was used.
Then there are parameters of measure of accuracy of reads on each nucleotide
that came off the machines, and there were measures called, Fred and Frac that
would tell you about different variables of reliability of your data. To me it
seems like there is an analogy just from the sheer numbers aspect here that we
will ultimately need a science that gives us some higher level of certainty
about the things that we’re measuring in these ways, particularly since they’re
not longitudinal, hypothesis-based, and tested, that we’re doing lots of priors
on things that we don’t necessarily control.
To me, the thing that strikes about what you’re doing is there’s a great
capability to develop methodology and frameworks of measures that I’m just
salivating over, where I haven’t seen before, and a lot of it has to do with
the governance that enables you to do that sort of thing. You really want
highly talented math people– you’re in Cambridge, they’re probably walking
right by your door– that can answer some of these questions. It’s how we
calculate the stars and do the Magellan projects.
I think it’s in there, and I think we ought to be talking with NSF and
others about developing the high powered math capabilities that will generate
machine-learning futures to help decision making. I think it’s that, and I want
to try to find some ways to open doors for some of the best people to be able
to bring those questions in here. We’re very eager to talk with you about how
to bring that kind of science into the helm. Maybe I’m way off, but that’s what
MR. DALE: We would be very excited to have those conversations.
DR. WALLACE: If I had to think about, what would I want to be differentiated
on, it would be that we actually were able to create knowledge that was
actionable by real people facing real problems. I think it’s actually that
population health problem. The thing I learned about population health is that
the way that you create population health is one patient at a time. There are a
variety of opportunities perhaps to convey a common intervention to a lot of
people, but if you really want to manage the health of a population you have to
understand the nuance of every individual. I think that’s what clinicians do.
That’s also what we expect as patients. The knowledge base that we have now
doesn’t serve this purpose well.
I think that the opportunity is to think about how do we ensure that
population health is the sum of the health of every individual not that we
express in terms and means. It also means we have to develop new science. We’ve
already seen heterogeneity of treatment effect as kind of a bother. The
heterogeneity of treatment effect is actually the path forward. If I’m
heterogeneous, I want that respected. I don’t want to be bored in with
everybody else. I think that data like we’re having and the different types of
data allow us to better understand some of those nuances. I think we’re just at
the beginning of this. It’s going to be really complicated, and I think it’s
going to take a lot of people thinking hard about it.
DR. CARR: Larry, actually we need you to introduce yourself and your
question, and then I want to bring it back to Dwayne to have 20 minutes to talk
about the intersection. I welcome your continued participation. It’s been
tremendously rich and exciting.
DR. GREEN: Larry Green, I’m a member of the committee. This has really been
fun. I have a non-serious question. I want to know how Optum manages to live
with 377 medical health economists.
DR. WALLACE: One at a time.
DR. GREEN: I want to frame this. NCVHS is on a journey that over the last 18
months is actually convergent around one of the gaps. I really like, Paul, your
formulation about we’re trying to figure out how to close stubborn gaps. These
gaps– NCVHS is presently at a position where we suspect that activated
communities plus data equals the nexus in which a lot of gaps can start be
closed. Your very last comments just a minute ago about real people with real
problems in real places– how do they use this? If you build it, will they
come? Are you guys at a position to collaborate with us in trying to connect up
what you’re doing with the notion that communities have to become learning
DR. WALLACE: That is the only way to do it. We’ve done work together, and I
think what we’ve learned is that communities are beginning to move the needle
because that’s what they do. There’s a pretty compelling evidence base that how
people use data, how they collaborate, but also a lot of the places that are
beginning to peel back these really vexing problems are taking more
comprehensive approaches. I think that’s very much our interest. It partly goes
to our analytic interest in things like geographic variation. We don’t see
variation as something to stamp out. We see it as something to understand
because it actually gets at adaptability to get at needs of the community. It
also gets to innovation. Our drive is to understand what is it that
differentiates, but also what’s underneath that differentiation? Then how do
you actually clone it and take the next step?
DR. GREEN: So, that was a yes?
DR. WALLACE: That was a definite yes.
DR. CARR: We always have our minds blown at this meeting. I think just when
we think we understand things we hear yet another horizon that we hadn’t
thought of. Today is certainly no exception.
I think that we were very stimulated by what Dwayne presented this morning
in terms of how the Health Data Consortium seems to be a place where a lot of
these issues can come together, and communities can come together, to
understand what the issues are, understand what some of the solutions are, and
to learn together.
We spent yesterday and today talking about communities. We’ve actually had a
couple of sets of hearings and roundtables from communities, and we feel like
we have a lot of knowledge about that. The Datapalooza, seeing all kinds of
issues being brought forward, and you’re yet another dimension. It’s sort of
very large, very small, in between, all of this evolving in incredibly rapid
and dynamic fashion. I think we want to find that space of nimbleness that we
can keep up, contribute, and help coordinate, thinking really or constructs
about what are the needs and how we can help. Dwayne, back to you in terms of
what you’ve heard today and the intersection of NCVHS and the Consortium.
Agenda Item: Dwayne Spradlin, Health Data
MR. SPRADLIN: I guess from my point of view I’m still learning a little bit
about NCVHS and all of the great work that’s going on. I think that for me you
could see from the language that we used that we very much want to be a partner
and an actor in this with everyone else. We’ve been putting the strategy and
the early programs together, really this year, which I think will hopefully
have a tremendous impact down the road. We’re also pressure testing some of the
I think we’re operating on different timelines maybe than other groups.
We’re basically saying if we can have an impact in 24-36 months, those are the
things we’re focused on. We can’t see out much further than that. You get very
choicefull about picking the things you’re going to spend time on or not. I
guess I would reflect back a little bit and ask for a little bit of your
feedback, which I think will actually encourage some discussion here.
I’ve got three questions. For this group really representing a pretty vital
piece of this whole dialogue around health data and putting it to work, what
are the couple of things maybe that particularly resonated, good or bad, about
what I talked about from 11:30-12:00?
I’m interested in what are three to five things that we didn’t talk about
that you think represent unique opportunities for Health Data Consortium to
really move the needle, to have an impact in 24-36 months. Then I would say
maybe we can talk for a few minutes together around what are three to five
things that we potentially could explore doing together to deliver on the
goals. That would be my objective.
DR. CARR: Let me articulate the resonances that I saw. We have created a
framework for health data stewardship. You have a health data bill of rights,
having a discussion about both of those is an opportunity. In our work with
communities we have heard consistent themes. Our report actually from last
year, “What Do Communities Want”, they want to know about community
health indicators, expand their access to underlying data, help with
collecting, use of federal and state web-based data queries, expand technical
assistance, and then we just heard again resoundingly their interest is in
making better use of existing community-level data.
As we said before, the supply side is enormous, and the demand side–
they’re beginning to articulate. I think in our community hearings they were
very clear what’s out there, how can it help us, and how do we use it. I think
that Lilly has put together a wonderful summary of the thinking that has gone
on this year calling socialized HHS data. I think one of the things of interest
in this, and we could share this with you is who are the players and
stakeholders? What do they bring to the table? Bill has done some work on what
are the data sets while they’re out there, and we’ve made them available at
healthdata.gov, but where’s the ladder that gets you over– it’s there and how
do I use it?
We’ve talked about things that you’ve mentioned, your 50 top recognized
folks is exactly what we talked about yesterday. How do we get those best
practices? Where do we publish them so that people know to look at them and
learn from them? I think what you mentioned in terms of having a place where
communities can come and talk about what they’ve done, and I use the word
“community” not just as geographic communities but various
communities can talk about what they’ve done and our thoughts about getting
these best practices, and while we’re doing the work we can share our work with
you and maybe you take up the mantle of collecting this because of all the
various folks coming to that website. So let me stop there and ask others for
areas of intersection.
DR. COHEN: I’ve got a couple. Community groups by and large haven’t used
claims data, health record data at all. This is a new frontier for them. The
question is will they, could they, and how would they? Community groups– a lot
of the data the federal government promotes and uses around health surveys, the
feds think a small area is county, and that’s not actionable when we need a
The dual mandate of providing actionable information while making sure that
we protect the privacy and confidentiality rights of those for whom we collect
data– qualitative data, the feds have been very good at generating mortality
rates and paid claims data. In my community work and what we’ve heard from
others there’s a sense that we want to know more about our neighbors and what’s
going on in the community in a very different way. What kind of tools can we
provide to communities so that they can focus on issues? The nature of what
public health really means, today’s focus we’ve been talking about clinical
Public health is really improving the quality of our communities. It’s
housing, its public safety, its lighting, its civic engagement. In order for us
to be successful at improving communities somehow we have to link what we know
about using the medical care system with these bread and butter issues that
communities rely on. I think there’s lots of potential for us to think very
creatively as well as strategically about how we can work in this space
together. Those are the kinds of areas that I think would be helpful to us. Can
you continue to talk about it?
DR. FRANCIS: I wanted to underscore also the data stewardship and all the
questions that are raised in that and how to think through them as data get
released through data.gov. I also wanted to make an observation that really
goes to– I think it’s Lily’s second slide, it’s on page two– where if you
look at players, preliminary description of stakeholders, and their specific
data uses and data needs. One of the points that is made on that slide is that
depending on who the stakeholders are there may be different priorities in
which data you look at first in terms of release, how you set up the system to
support the needs, one or another of these.
My initial impression of the Datapalooza, and I went to the first ones, I
didn’t go to this last one, but I’ve also cruised around the website a lot is
there’s quite a major role played by venture capital, what will sell, what can
we make money out of. I’m not opposed to that, I just wouldn’t want to see it
eclipse in particular the public, public health, and possibly regulatory
interests in the data that show up lower down on the list, but if we use the
term “venture capital”, venture capital would have been at the
bottom. This is an alphabetical list. It’s not at all intentional that business
and developers and payers are at the top of the list, but I would really want
to underscore that point. I think this is a public good, and it should be about
how data can be used to ultimately benefit everyone’s health, not figure out
say how to design plans so that we get rid of the high cost users.
MR. ROSENTHAL: Ever so briefly, I have a very different opinion on that. One
of the things that I think Greg and Todd and company have done so fantastically
is, looked at what has worked in other areas where public good is at stake. You
take weather basically. You make weather data available. Rather than just
publishing it, you engage the market. You can’t innovate, you can’t come up
with as many creative things that will actually work as the market forces can.
Instead of just putting the data out there, you rely on the market.
I personally view it as a false dichotomy between public good and between
market ownership. I think if you’re very savvy you can actually use market to
drive public good. I think that’s what you guys have done ever so brilliantly
on the weather side, on the geo-location side, and are now doing with the
health data. I would encourage you to do that. I think it’s an old, false
dichotomy, in terms of MBA. I think you can absolutely achieve public good
while incentivizing market forces. Unless you want to go and try to do it on
your own, you’re going to be working with folks in the market place one way or
another. In that sense, what actually constitutes engagement?
I think the stuff Lilly did is absolutely fantastic. Version one is great.
You do weather, you do geo-location. It takes off. Now your phone tells you
should you bring an umbrella. You also have public good, and NOAA basically
tells you if something’s going to flood. That’s great. You do geo-location, the
same sort of stuff happens. With healthcare it’s tricky. There are perverse
incentives. It’s tricky every which way. What’s the first version is put the
data out there and see what happens. I think now with you guys getting involved
and Health Data Consortium, not only can you put the data out but can you
translate it into information.
I’d expand Lilly’s list just a little bit, and say the barrier to
interacting with the data right now is you have to basically be a developer, or
you have to be a technician, or you have to be pretty savvy at using some
stat-ware, being able to create a pivot table in Excel.
In order to engage the other folks out, market force or public good in the
communities, you want to lower that bar. You want kids coming in, kids
figuratively speaking, who might not be able to code or being able to pull
anything out of a LAMP stack or this, that, or the other thing. Again, you guys
have done so well looking to other models and other verticals. There are plenty
of other verticals where that happens all the time. If you’re familiar– one
I’d say, baking up the business and not just putting it in terms of a hacker or
developer or business person, but the need there is how can you connect the
data to a market need or a public good or a social need.
You have data. What are you trying to solve with it, regardless of what side
of the aisle you are? Any educational implementation really needs to focus
around that. That’s what we tried to do with some of this curriculum around the
little boot camp stuff. We’ve done it at Harvard, MIT, and Hopkins, and at HDI
as well. You guys participated, and thanks to the workgroup for doing that.
That was very nifty. Kind of taking that curriculum and expanding it– and the
other piece is how do you release this data and engage folks who don’t have the
technical capabilities of working it up to solve this business problem?
I’d encourage you to look to a couple of verticals where you see this all
the time. If you’re familiar with any of the public data browsers, and I know
we’ve been through a bit of this on the presentations, Tableau Public, or
Google Public, and Data Explorer. They did little silly things, kind of
public-private partnerships with ReadWrite Web. You want tech kids who aren’t
deep in healthcare coming in and playing with the data. You essentially put
your data up there, and it’s completely private, and people get to interact
with it in a visualization. They do a little contest, and they say look at this
and look at that.
The thing that– one, hundreds of thousands of users are interacting with
this, not just downloading, but kids here, there, and everywhere,
co-morbidities for diabetes. An 18 year old female who has no idea about it,
now she’s on her way to being an MPH person. Instead of just putting the data
out, potentially putting it into a browser and structuring the interaction of
users, not here’s how we can use the data, but here’s how you can actually
answer a question with the data. You can answer that question with no hacking,
no coding, no developing skills whatsoever.
That’s been a best practice in other verticals that might really be worth
looking at as you try to do it in the healthcare space. That’s basically it in
a nutshell. We’ve seen that, and by the way your data, as we looked at last
time, is already being used in those locations, so getting very proactive
around that. It’s in Google. It’s in some of these other areas, and it’s doing
some wonderful things, so getting out ahead of that and being able to interact
with that in a meaningful way is not such a bad idea, so moving up from data,
this is V1, Todd and company, to you guys getting involved, V2, using the
information basically to solve problems and looking at other verticals to look
at what you’ve done so well in the past around other cycles.
DR. CARR: I’m going to jump the line a little bit, because I want to point
out the adaptation of the sustainability model. One of the things that has
struck me with this workgroup and then also at Datapalooza is that there are
people like Josh who know unbelievable amounts about data and how to use it and
how to get it, and there’s– all of you– and there’s venture capital that
wants to fund it, whatever “it” is. There are communities that are
taking simple data, or even if you use the i-button, that’s my favorite, the
blue button iPad application where they take the complicated data, they do a
little translation, and then they add something really unsophisticated, the
name, address, and phone number of your doctor.
I think it’s a continuum. I think it’s intimidating to many people to think
about that, but then to other things, like you need the context experts, the
physicians, clinicians, community members, individuals. Anyway, I was trying to
do an adaptation of this sustainability model to say that you really need
everyone. You need the vision, first and foremost, because we’ve got lots of
stuff going around in service of what. You need the software specialist, the
clinical specialist, the informaticist capability.
You need the incentive, and I say, who’s paying for it and how does it get
sustainable? And then the resources, which I think is a lot about our HHS data
as well as this newly collected data, real time collected data. Then a key
question is who is the customer? You need to have a customer to keep it
sustainable. Then onto the methodologies, to data integrity, the privacy, the
security, all of those things. Then someone to give feedback, and maybe that’s
MR. KAUSHAL: I come to this as a reformed World Health Organization
individual, reformed physician, or reformed government employee now in this
world of venture capital. The areas where I see the highest piece of execution
is when you get teams of people with those combined skill sets, especially in
healthcare, which is so fragmented, so much asymmetry, so many malincentives. I
can name companies which are succeeding extremely well, so performing great
social value and also going in the right direction in terms of the market, but
they’re comprised of people with those backgrounds. I would argue just focusing
on one piece, whether it’s the business side or the non-profit side, it won’t
get us there. We have to get it together.
MR. ROSENTHAL: We got to market forces, and what will pop up next no doubt
is privacy, and I want to throw out something completely crazy there. I would
actually like to think about privacy a little bit different, potentially, as
long as all things are on the table, and perhaps even– I’m going to throw out
something crazy. How about a green button initiative, so green for
“go”? A family member of mine has cancer, and I’m able to donate bone
marrow. I can donate that, but I can’t donate my data in a research environment
to help him or her.
That strikes me– think of blue button, but green button. There are a lot of
people out there who would like to opt-in to their data, waive privacy rights,
de-identification, for research usage. You’re sitting around talking about this
massive blue button initiative. If you hear Todd tell the story compared to
initial healthy skepticism early on, no one would use it. Who would want it?
Massive demand. I’m putting another alternative out there and saying, of course
Just like you could get your medical record before blue button. You could do
it. It was possible. You’d write to your plan and see what happens. I’m not
talking about that. I’m talking about something on a wide scale initiative with
the same kind of narrative round blue button. Do you want to be able to
contribute your data– I understand we want rights to privacy. Can you waive
that essentially, do it on large scale, and then be able to use those as data
sets for some of these initiatives that we are talking about?
DR. CARR: I think these are the ladders that get us over the wall. I think
that one of the conversations we had with Greg early on was for this group to
think about here’s the problem to be solved and work backwards from there. I
think that’s a great use of the skills, so let me turn it to Vickie now.
DR. MAYS: Your green button is not that crazy. It’s not out there as much as
you think– no, this is what happened in the beginning of HIV research, where
people had very unique forms and they would go and find the researchers and
say, you can use whatever. We just need to scale it up.
I just want to respond to the various categories that you were talking
about. In terms of the one that resonates for me would be your capacity to be a
convener. I think who’s in your network maybe different than who’s in our
network. The ability to bring a diverse group to the table would be absolutely
wonderful. The other is, because you have that “connect and enable”,
but the other thing would be education. I’m not talking about traditionally the
way we do it, and that is thinking about a product-driven way in which to get
the information out, such as even to think about kids, thinking about
consumers, thinking about people in general.
We saw those great videos today. It’s like producing it in ways in which
people really use it would be that. In terms of the unique opportunities, I
think if you were able to develop something where community could use data
outside of secure data centers, so that means really thinking about
transmission privacy, et cetera, that would be absolutely wonderful. Even as
researchers going to secure data centers really slows us down, so if you’ve got
a way to think about that. Also the uniqueness would be what it is that you can
connect up around data.
For example I just saw a study about climate and poverty. It makes sense
after you think about where disasters are and who they keep hitting, et cetera.
If you think out of the box about data combinations, we’re being very health
focused, but I think there’s a lot of external stuff like mortgages,
foreclosures, or things like that that you could begin to explore a bit that
would help us get into the contextual issues that we often so want to explore
but have difficulty linking the data together.
DR. CARR: We have three more people whose name begins with “L”.
We’re going to get to them. I know you guys have planes and cabs. Do you have a
couple of more minutes to quickly go through?
MS. KLOSS: I am always brief. I have empathy for your earlier statement that
you’re starting up to get going. I think this is what can we do to collaborate
over the next 6-12 months. There I see providing input on the areas that you
identify as themes, sharing resources, having the members of the committee be a
resource as you ramp up on webinars and so forth. If there’s some expertise in
the group we can certainly help participate in presentations. I would think
looking forward to the next Datapalooza thinking about a way to expand what was
a very good session there on community engagement, but it was standing room
only, which says there was quite a lot of interest in that topic and it could
be broadened and perhaps made a little more point of focus next year. I’m
thinking of some things that are practical and short term.
MS. BRADLEY: I would like to point out there PowerPoint is really a summary
of everything that the working group members have been contributing over the
last few months. I’d really like to say to the Health Data Consortium how
excited I am about the opportunity there is to be the convener, the
disseminator, and educator, and to really untap the potential of the American
people. Clay Christensen talks about in the Innovator’s DNA about discovery
skills, and one of his associations I can randomly walk you through. On the
point of convening, I think that connecting us to other organizations that can
also help us disseminate would be fundamental.
So NACO, NACHO, the Governors Association, the Mayors Associations, these
are the kinds of groups we need to get to help with our rallying call to
deliver the message of the NCVHS full committee around communities as learning
systems. Another area within convening, there are two different activities
where I certainly heard a lot at the Datapalooza about building clinical trial
registries with control arms, so pharmaceutical companies. This is the idea
that Josh was briefly mentioning around I’d like to volunteer, consent now, for
my information to also be used later. I’m not sure who’s going to do that, but
the opportunity for you as a neutral party to be talking to folks and figuring
out who that player is who could get everyone on the same page.
Also on the all-payer claims data it seems like there’s a lot of interest
and yet not a lot of mutual agreement. As the information– so the NCVHS and
HHS produce so much great data, so many wonderful tools and recommendations and
best practices, and it’s just sitting there. I think that there’s from the top
and bottom, and then we need this activity going, very much as you described
the HDC, and it’s about maybe even just starting to teach within communities
about problem solving again, about how to ask questions, about how to find data
that relates to those questions, about learning how to use the data. These
things like looking at housing numbers, those are in the community health
rankings. They take a very holistic view that is fundamental to changing the
health of our communities.
The tremendous amount of potential that we have resides as Vickie was
describing literally within schools and churches, and there are so many
resources that remain untapped, and getting into the Code Academy and trying to
leverage these other platforms, the Khan Academy, and giving them this entry
way into seeing what’s out there, there’s really a lot of basic stuff they
could start doing before we even get into the really high powered analytics
that we know that NSF and folks like Optum will be able to help us do.
I think that as we look forward into the next decade and we understand how
we’re shifting into more of the sharing economy and Sidecar and Airbnb, and as
we move past luxury being conceived of as goods but rather as being conceived
of as these amazing experiences and experiential luxury and this is the new
thing that young people want and going and joining stuff like Tough Mudder,
it’s the socialization of the data that I think you’ll also see we can tap into
and help make communities stronger.
DR. FRANCIS: I wanted to underscore that this is a working group of a
statutory public advisory committee. To me the significance of that actually is
that we can be a wonderful sounding board for talking through the kinds of
questions that as we’ve been doing with the stewardship report– that report is
premised on the idea that you have to have good stewardship in order to enable
data use, not that the two of them are at logger heads or in controversy, but
that each helps the other and how to talk about how to do that.
It seems to me this is the perfect forum to be talking about how it is we
can combine the public interests and the market interests in a way that one
helps the other. I wasn’t meaning to say they’re at loggerheads, but I was
meaning to make the point that there’s sometimes market failure, and we need to
look at those kinds of questions. I would just encourage this as the public
forum for continuing those kinds of issues.
DR. CARR: What a rich afternoon, remarkable. I thank you, Paul, Richard, and
Dwayne for taking time out and spending time with us. I hope that we’ll
continue to stay in touch. I know that we will. With that, let’s take a 15
minute break, and we’ll reconvene and–
MR. SPRADLIN: I just wanted to say thank you for that. I can tell you I’m a
huge fan. That in the right settings– this is all about taking time off the
calendar. You want to compress how long it takes us to get to value, getting
this kind of feedback in hours as opposed to months of trying to ferret it out.
I’m sure you’ve got your own systems for doing that, but this is incredibly
valuable here. That was terrific.
There are a number of things that were mentioned here which I love because
it helps validate a few of the things we’ve been thinking about, in terms of
Khan Academy and using different kinds of tools the way people want to learn.
Could we create a repository of 60 minutes with you, 30 minutes with you, and
60 minutes with Deven McGraw and put that out there to help people learn 24
hours a day, because that’s the way people want to learn? Invaluable. I think
there’s a set of those things that we could look at, particularly if we can
count on all of you as content partners to some degree.
We could really scale that sort of thing and have an enormous impact. There
are a few things like that. I want to collect my thoughts. I did just want to
ask you two questions, and you don’t even necessarily have to respond. The
first one is it feels to me like we could play an important role. We said this
a couple of different ways, but G2 intelligence. So out there in the field when
we’re working with the affiliates, when we’re working with the different
groups, I think in a lot of ways we can aggregate market feedback, if you’ll
allow me to use the term that way, and in some ways some of the– some of that
should go to you, some of that should go to public comment, some of that should
be done in discussion boards, but I think we could be a powerful feedback
mechanism for you. If that’s helpful, I think that’s a role we’d like to play.
The second one is– we talked about green button and a couple of different
things. Something that’s really been percolating in my mind the last couple of
months is that as we think about PCORI and patient-centered and these consumer
circles and all the different groups I have yet to hear a compelling vision of
what it really means to put the patient at the center of their information
world. It feels like there might be a bit of a collaborative effort here that I
think could be very valuable, almost a futurist effort but not that far out, to
say here’s what patients are very soon going to expect and want. What does the
world look like around that that fulfills in those ways?
I do have to tell you along the lines of seeing change happen and happen
fast and aggressively in other industries, once you begin to power the
patient– there are two things here. One is they’re a lot smarter and a lot
more able to process some of this information than we give them credit for,
particularly if we can demystify some of the words. Secondly, I think there’s
an incredible amount of pressure that patients and the consumer side of this,
if you will, and I would also say physicians and primary care and the ones
right there, are just about to put back on the system just as soon as we begin
to unleash these tools and the knowledge.
The question is going to be why don’t I have that yet, or you can do that,
why can’t you do that? Some sort of envisioning around what that information
architecture needs to look like that’s patient centered I think could be a very
important piece of thought work that could guide a lot of this. For what it’s
worth, I’m thinking that might be the basis of some pretty interesting thought
DR. CARR: I don’t know if Paul Tang called in, but I’ll be eager to send him
the transcript. He’ll be ecstatic. I think you’re absolutely right, because as
we are now moving into Meaningful Use 2, a key component is bidirectional
dialogue for the patient getting their information on a portal and
communicating with the physician. It’s not enough to have the portal. You have
to have that dialogue. You’re exactly right. The expectations are going to be
all over the place. Many of the folks coming along are new at EHRs, just got to
meaningful use 1, and so providing some guidance on that front will be a
resource across the board. That’s just fantastic. We will take a break. We’ll
reconvene at 3:45. Those of you who are departing, safe travel.
Agenda Item: Review Needs of Communities from Joint
NCVHS Roundtable, April 30-May 2
DR. CARR: Okay, so we always just get smarter and smarter when we have these
groups. The richness of the conversation is just phenomenal. What I wanted to
do is throw a couple of themes out there that we can decide how we want to
proceed. I guess—we’re going to follow up with Dwayne and talk more about
how we can work together. He wants to identify one or two themes to start with.
We should think about that. Second, we have some wonderful work here both from
Lilly and from Bill Davenhall.
I want to look at that. We have a specific request from Bryan Sivak to look
at page 3 in Lilly’s work about where we go from here, how we might focus this.
I want Bill for you to give us a little update on the work that you had put
together, the drawings, if you could give me a framework on what it is and
where we want to go with it. We have about an hour. We’re going to hear from
Bill what this is and then we’re going to go back to Lilly’s framework and then
build on that, and maybe we can even talk about the sustainability model and
how we can incorporate that as well.
MR. DAVENHALL: These three graphics were inspired by yesterday’s discussion.
I put this together because I’m also looking for a place to land the airplane.
The first one I want to show, and I forget who on the full committee was
talking about birth and death. They brought this out yesterday in the meeting.
This is something I had seen from a senior engineering professor at Johns
Hopkins about a month ago who basically said we need to start off with a view
in our mind of what the heck we’re trying to do. He laid out this life cycle.
Essentially, this is what it looked like. There’s birth and there’s death,
and there’s a whole bunch of stuff in what he called the cloud. In that cloud
is all the stuff we’re talking about. Down at the bottom you have all the
factors that impact that cloud, and up at the top you have your life. You’re a
baby, a teenager, hopefully you make it that far, and then you’re a young
adult, middle adult, and a senior. He says almost everything you want to do is
constrained by this mental picture.
DR. SUAREZ: Bill, what’s the third factor?
MR. DAVENHALL: Environmental. Health, social, genetic, and environmental. I
would say I’ll probably have somebody spruce these up so they can be read by
mortals. All it was, it was really helpful when I first saw it. Sometimes it
takes somebody else out of another field to slap this in their face and say
this is really what you folks at health and human services keep talking about.
Pretend like you’re an engineer for a moment and figure out the goal here is
how do you industrialize what’s in this box? That’s graphic one.
Two, what we’re really talking about is health and social care, not health.
Health is probably—social care is probably a determinant of health. We
keep spending all of our time worrying about health and these minute health
factors, and glucose levels, and blood pressures, and so forth, when in fact
it’s the social context in which we live that’s having a major impact. We argue
as scientists about what’s the combination, what’s the formula for that.
Needless to say, when Larry has asked me to think about what the community
needs, every community that I’ve ever been in this is what they want to know.
Where are we going?
It’s up there, population dynamics, what is now and what will be. They
always want to know that. Where are we going? Are we growing, shrinking, more
babies, less babies, older people? They always want to know what resources do
we have in our community. This is the thing that’s so badly done across
America. There is no standardized way to collect resource information. What I
mean by that is there are a lot of standardized ways people have invented, but
there’s not a national approach to this. Anybody who wants to incorporate these
into some of the wonderful database labs that we’re talking about are going to
be really frustrated because there’s no way to do it. Where are the
church-based organizations, the volunteer groups?
They also want to know about need. What are the needs in our community? What
does our population need in the way of health and social care? They don’t
really want to spend all that time figuring out what it is. They want somebody
to tell them what are those problems. If you really think about it, that’s what
smart data does for you. Smart data will tell you what the age, sex, and race
is in a population. It moves dramatically and says, given this, this is what
this population is going to need. That’s the other thing, and they want to know
about their environmental environment, all those things that make up a
community from church, family, culture, factories, manufacturing,
industrialization. That’s what they want to know. Within this constellation is
all this stuff we’re talking about.
I just drew this picture to remind myself and you that that’s what we always
have to be thinking about as we start to get innovative about our solutions. I
say at the bottom we need to find a way to industrialize these processes. I
think that’s what the working group is about, to help industrialize SES. That’s
DR. CARR: Just work from more than a decade ago, with the committee
population health community attributes context, place and time, from shaping
health statistic vision for the 21st century.
MR. DAVENHALL: The third graphic is this ecosystem, which really gets at the
heart of the data. We talked a lot about vital registration. Now the problem
with all of this is every state has a different flavor of this. That’s one of
the problems we’re having. We learned yesterday from Bruce that not all states
do the same thing in vital registration. That means that these are still
one-offs. From a practical economic point of view, it’s hard to build
applications for this.
The only reason for this is, and I’ve probably forgotten some things here,
but if you look at it, there’s vital registration—I’ll go around the
clock—vital registration, and then all those things on the outside mean
that there are differing fingers in that pie. They’re all players. When we talk
about doing something with vital registration, you’ve got to talk to all of
these people because they all have a part of the process. Then you move over to
what I call the transactions, episodic transactions, and that’s basically where
this industry has been for about 40 years, wrapped around episodic
transactions. When you can stand up like Walter does and repeat x, x, xx, you
know “you be 16, you be 92”, you spend a world in episodic
Around the side are all these interests that are always around our table,
from hospitals, physicians, all the way down to companies and business. Then
you have the research community, and this is where probably most of the people
around the table have had their hearts for a very long time. We like research
or we wouldn’t be here. We have all this that’s circling around that. Then you
have what I call the community planners. This is like the ecosystem that we
have to play within when you say “help local communities” figure out
This is their ecosystem. On the edges that are half in the cloud, half out,
are people that are like patients and down here businesses in general. They’re
paying the tab. They’re paying an insurance premium, and they don’t really
understand what they’re getting or how to change it. They get up every day and
it’s still something they have to do and pay the piper for the insurance. Some
of them are more progressive about that. I just wanted to get that off my
heart, to say that it’s really important when we talk about these innovations
to keep reminding ourselves what in the heck are we doing all this for. That’s
my contribution on that.
DR. MAYS: This is absolutely wonderful in terms of the ways in which you
separated out the parts. It’s a good way to think about some of the data that
we need. Your comment about health and social care, some of the clinics that we
work with are very interested in not the health data we usually do, but they’re
very interested in how to get what they call enabling data. They’re very
interested in what’s the transportation data, how long does it take, what kind
of resources are there? I think we keep forgetting we need a lot of that stuff.
I like the way that you’ve done this, thank you.
DR. COHEN: Was this intentionally meant to look like the cloud?
MR. DAVENHALL: As a pilot, flying is 97 percent boredom and 3 percent stark
terror. I represent the 97 percent of boredom. Josh is the 3 percent of stark
terror. As a pilot you learn to do three things or you’ll die. You have to
manage your fuel in a crisis and you have to scan the instruments constantly.
You have to be able to stay ahead of your airplane. In other words, don’t let
the airplane fly you, you fly the airplane. That’s my job with Josh: manage the
fuel—honest, if you really think about it, the most exciting part for a
pilot is landing and taking off. It’s not in level flight. We’re actually
talking about landings and take-offs, which are the scariest parts. I think
every time Josh starts to talk I sit up a little straighter in my chair because
I know I have to do those three things.
DR. COHEN: So what can we as national committee do to move the conversations
along in this space? There are a couple of things that you highlighted. I
totally agree we’ve always operated in a classic deficit model, focusing on
disease and illness. We haven’t talked about health really, or community
wellness, an assessment of resources and assets and community strengths. There
are tons—as we’ve talked over the last week—there’s tons of data over
what’s wrong and very little data about what’s right. I’d love for us to figure
out how to transform the conversation into what makes communities healthy.
There are lots of data—there’s data on green space, successful
transportation systems, civic involvement. We’ve never thought of those as core
data for public health and for understanding what makes communities healthy.
This is the kind of stuff that needs to be integrated into our worldview. This
is where I think we can make a contribution. We are redefining what the public
health needs are, and in order to improve communities health the federal
government can provide assistance or connect you with these resources so we
have the picture of your health, not your lack of it.
MR. DAVENHALL: One of the things we need to do as a group is communicate to
the innovators in this space, which the working group members are part of, are
what are the things we do need? For example, I mentioned the community
resources. You desperately need these great minds to think about how we’re
going to reflect that kind of data and maintain it in a sustainable way easily
without asking foundations for billions of dollars and the government to spend
billions of dollars on it, or it won’t work.
Part of it is this technology obviously to me is capable of managing
crowd-sourced information. How can we use that technology to supplement all
that resource information to communities? These communities struggle with that
part of it. What happens is people can’t wait for their neighbors, so they go
ahead and do it anyway. They have ten groups in the community who just
collected information on the same ten hospitals, five clinics, and so forth.
It’s really a wasteful activity. It’s not that we’ve encouraged them to do
that, but the federal government hasn’t provided—I’m not saying the
federal government has to provide it—but there has been no leadership in
saying we have to look at a totally different way to do this because it doesn’t
exist. That’s one of the things.
DR. GREEN: My anxiety level is going up a bit when I look at the clock and
realize how fruitful the last four days have been and yet how we’re still
trying to figure out where to land. This is not even half-baked, but I want to
make a couple of observations, and then I want to throw out a suggestion for
consideration. The observations are that I think the most important thing on
Bill’s three diagrams is similar to what we just saw from Optum. It’s how many
fragments there are.
Don’t worry about what’s in the space, it’s just that it takes pages and
pages and slides and slides and segments of slides with words stacked up with
bullets inside, and just to try to get our arms around the pieces of the
puzzle. One way to frame the problems is the word “fragmentation”.
The healthcare delivery system is fragmented from the public health system; the
public health system is divided within itself. The mental health crowd can’t
get along with the substance abuse crowd. Poor people are different from rich
people, and it goes on and on. Social determinants—it just keeps going,
Another way to say that is our problem is we’re overly specialized. You said
you were thinking like an engineer. The big difference between engineers and
doctors is engineers are prepared to design within constraints. You’re thinking
about the constraints. This problem—in a way, NCVHS has an opportunity to
defragment. A word for that is “integrate”. We have agreed over the
years that one of our strong suits historically has been to envision and to
convene. That’s where I’m headed. I’d like to change the word
“envision” to “make visible”. I’m wondering if a space we
might go to next is to make visible what is possible in the new world, not the
old world, and not make visible all the barriers and the difficulties, the
administrative nightmares, the legal conundrums, all that sort of stuff, but to
try to make visible what is possible.
We’re having a very hard time talking about it because we can’t walk around
and touch it very well. Each of us is capable in our own imaginations to create
sort of a picture of it, but we still haven’t come to a shared understanding of
what we’re imagining. When I was listening to the stuff about the Datapalooza
and the excitement about it, I began to wonder about the following. How about
thinking about collaborating with people like Optum—I don’t want to
prejudice this to pick out any particular group—but what if we were to
identify with some of the contacts we already have a particular community that
has mud on its boots from trying to work a problem that matters to them and who
have tripped up and struggled? They’re not naïve. They’re an experienced
community that have tried to move the dial on individual and population health
some place particular and take their issue and present it very succinctly about
this is what they don’t like, this is what they’re concerned about, and this is
the desire to stay.
They’re trying to get from there to here. They’re wondering if they can use
data somehow or other to do that. What do they need to do to do that? And then
populate that additional space with hard-hitting, brief contributions to what
they could do with data from myriad perspectives, some traditional, some
non-traditional, to see if 45 minutes later you have whiplash, you suddenly
realize you never thought of it that way and see if that can trigger off a
recognition, a making visible that we’re at a point in history where these data
from these different sources that are pertinent and relative to community
health, that they actually can be tapped into.
They actually could be used. Out of that, I think that filters down and
starts identifying the remaining impediments, the things that need attention.
It might even locate where some of those are. You get my drift here now. I feel
as if we need to—I’ll stick with Bill’s metaphor. We either need to fly
this airplane or it’s going to fly us.
DR. CARR: I agree with you. We are learning more and more things. I think
that it would be terrific if we said here’s a problem that could be solved, and
here’s how to solve it. I look around the room, and we’re not configured for us
to solve it, but what Health Data Consortium does is they have 6-8 affiliate
partners, and he mentioned that in Nashville or somewhere they did the Cajun
Codefest. I wonder if we begin to think about it that way, rather than us doing
it, maybe we participate, maybe we don’t, but that we partner with them to
sponsor the kinds of things that Leah’s been talking about and Vickie is now a
DR. GREEN: I want to be quick to say I’m not proposing that we solve this.
I’m thinking about making it visible and helping convene the people that could.
Your comment right there about the code-a-thon—remember, I said this is
not even half-baked– but another possibility is to do the same problem three
different times with different people at the table to solve it.
DR. CARR: Work with me here. What I am struggling with is big ideas, and
then we leave here and we can’t quite do it. Lilly’s done an amazing job at
pulling our thoughts together, but now we’ve heard some partners, so if we came
out of the community thing—pick a community, pick a problem, pick
something, and then take it to the Health Data Consortium if they have some
upcoming code-a-thons or something like that, let them convene. Let us inform
and define, but I’m trying to land the plane. I like your idea, but I’m not
imagining all of us here for a long weekend trying to solve it, although we
probably could. We might be able to. I’m just trying to create sustainability.
This workgroup is a reactor group, an idea-generation group, and then we should
be able to hand off. It seems to me Health Data Consortium now has these
affiliates who could do it, am I right? We could give legs to what I’m trying
to solve and they could come back with three different ways to solve.
DR. GREEN: I think we should be cautious to recognize there are very few
problems that only have one solution.
DR. CARR: Right, so we do three. There are multiple groups trying to solve
DR. GREEN: How many ways could this community’s problem be solved? How many
ways can we expose how this problem can be solved?
DR. SUAREZ: I think the challenge that we still have is defining what is the
problem we’re trying to solve. I think these pictures, these are terrific
— to solve what? What is the problem? What is the issue that we’re trying
to address? Is it a lack of community resources? There are so many things that
can come out of this. Any of the circles have a problem, many problems. To me,
we just had this two-day event of the community health initiative that we had,
and I think we’re still struggling to define what it is that we’re trying to
do. I know what we’re trying to do in standards, and even then I still struggle
to keep things going. I’m trouble defining exactly what we’re trying to do with
respect to the community initiative.
DR. FRANCIS: So you and I are on the same page in the following way.
Sometimes what a working group does is it says to the group of which it is a
working group we need you to do this, this, and that. I’ve got some ideas about
what we need to do given what I’ve heard. A primary example of that would be
the question of how the privacy bill of rights that’s on the Data Consortium
either does or doesn’t fit with the things that this committee has said about
privacy and stewardship. That’s a task that I know how to do and would like to
have our subcommittee and then the full committee look at if that’s an
actionable kind of thing. Basically what I would try to want to understand in
terms of next steps are what do you want out of the rest of us? I’m on both.
DR. GREEN: To illustrate—man, I’m beginning to wish I hadn’t started
this, but maybe not– this could be a layered enterprise. There could be an
exercise around a real problem that a real committee had that answers Walter,
what’s the problem we’re trying to solve here? It’s childhood obesity in
That’s the problem we’re trying to solve, and then bring the different
fragments that are understood to be the underlying determinants and the data
sets that exist there show how they can be turned into knowledge that would
identify where the problem is located, the actual problem and its origin, that
sort of stuff, work that territory, maybe do it two or three times, have
someone like Leslie sitting there, have someone like Walter sitting there that
isn’t trying to solve that problem, they’re watching how the problem gets
To pull out the next step is now let’s talk about the things that you did
that would violate a code of civil rights for proper data use. Let’s show you
the things that broke existing law—
DR. CARR: Or let’s think about ways of doing it without—
DR. GREEN: Or things for which there are no standards, and this group solved
it with this way—if you harmonize those standards you would have gotten
DR. SUAREZ: The challenge is really identifying the problem. You just said
the problem we’re trying to solve is obesity. I don’t think we’re trying to
solve the problem of obesity.
DR. CARR: I want to read from our charter, because I think that will ground
us. Our charter says that based upon knowledge and familiarity with one, HHS
data resources portfolio; two, traditional and new information dissemination
strategies, development, and technologies, and social media and their
applications by the technology innovation community; and three, the needs for
data and information by major participants in the health system, the working
group will monitor and identify issues and opportunities and make
recommendations to HHS on improving data access and innovative use, including
content, technology, media, and audiences. Whether we solve obesity, rats,
whatever in the community, that is irrelevant. Our job is not to solve obesity.
Our job is to demonstrate innovation.
DR. SUAREZ: The key is to improve data access innovation. That’s the key in
the entire thing you read, the long thing that you just read. They keys are
data access and innovation. Those are the two key concepts. There’s nothing
DR. CARR: What I was responding to is maybe what Leslie was saying. We are
not driven to anchor on what was done in the full committee meeting. I think it
informs us, and this informs us, too, that as we’re looking at all that they’re
doing at Optum, they’re focusing on healthcare delivery basically, which is a
minor part of our lifecycle. That said, because we have an opportunity to start
somewhere, I think we heard a lot about communities. If we took a community
issue and the goal is to see what you can do to help address this issue.
MR. ROSENTHAL: I was going to say that’s what they’ve showed was very nice.
I have seen this multiple times. Two years ago WellPoint was in here, and it
was the same diagram except it had Watson with the jeopardy logo in the middle
of it. It was different version of it. Merck has theirs and it comes and goes.
In the diagram, it’s all the clinical side of it.
As I read the charter, innovation, data access, new data, I thought I heard
social media called out explicitly, and then communities not in the way we’re
talking about communities, but communities of technologists, innovators,
If it’s helpful at all, I’m meeting one. The way most entrepreneurs work is
with straw mans just throwing something out so people can react to it and they
get something done. I did that—it’s coming on a year now, and there are
seven things around these pieces if you want to revisit those or come up with
new ones on your own, I’d strongly suggest them. I’m not sure what the typical
protocol is, but per that charter to come up with a couple initial straw man
recommendations based on all the stuff we’ve heard over the past year.
DR. CARR: We just put one out. What do you think of it? The idea that we
have communities have identified issues, problems, and understanding their
thing. We want to set an expectation, deputize a group to demonstrate the power
of the combination of HHS, plus social, plus whatever else.
MR. ROSENTHAL: I think that’s worth doing for sure. I tend to kind of work
in blocks and say these are the seven things. Communities are doing it. We saw
it right? Louisville’s doing this on asthma. There are dozens of them doing it
in different ways.
DR. CARR: So we articulated the seven things. I don’t know if they’re
incorporated in this.
DR. COHEN: What are the seven things you’re talking about?
MR. ROSENTHAL: This was a document. I’m losing track of them now, but it was
information, education, curriculum around connecting the data to specific
public good, business use, access, actually some sort of other mechanism for
people interacting with data who don’t have data literacy, if it’s not a visual
data explorer whatever you want to come up with, but something that lowers the
bar so you don’t have to be a technologist.
Data access is an opt-in, the screen button thing. I threw these out there.
You can love them or hate them, but however you want to frame it up, the
ability for people to contribute the data, specifically the question
identifying the problem was privacy, privacy, privacy, and here’s a way to get
around it. Say, “contribute”, accept my data is the deal. That would
be another recommendation around the privacy piece. I think there are three or
four more. I could dig them up really quickly.
DR. CARR: So we could give that to Health Data Consortium and say here’s
some idea generation. What struck me today is now we have a place where lots of
people are going to go and look, and we can put those ideas out for discussion.
Again, as we think about NCVHS and we talked a bit yesterday about what else
do we need, what can help—communities are telling us every time we meet
with them, I don’t know what’s available, I don’t’ know how to use it, I can’t
do it. If we take a community, take their problem, deputize people in one of
these work-a-thons to solve it, we’ve then maybe helped solve it. We’ve
demonstrated the use of data. We’ve thought outside the box. We show
innovation, bring in social media, and now we have something for people to
MR. ROSENTHAL: That’s sort of what I’m getting at with the curriculum piece
instead of doing a hack-a-thon, do a solve-a-thon, or come up with a better
name for it.
MR. CROWLEY: Just to build on that, as we were trying to hone in on
something that’s packageable, doable, that we can attack, for the solve-a-thons
or code-a-thons or data jams, whatever you want to call them that are going on
all over the country, there’s not a well polished resource that can be reused
across these different instances, which make sure that when somebody comes in
at hour zero, day one that they can get up to speed immediately on solving
these particular community problems.
There are some disparate resources around data sets here, data sets there,
which ways they’ve been used, but there’s not a solid resource that you can
provide to these groups to immediately have them hit the ground running. Maybe
one of the things we should consider is what would that package of products, of
data sets, of advice for strategy to think about look like that could be
provided to these solve-a-thon participants either in advance of when it
starts. As Lilly’s hands are gesturing, there is the hackpad, which is a nice
start, which provides insights into how to run these. There are not data sets
associated with that–
MR. ROSENTHAL: I’m not dismissing that. What I don’t think you want to do is
just go in—I’ll be quiet after this—I don’t think you want to—it
can’t just be a hack-a-thon. That fails time after time after time. You have to
be engaging the community, for whatever it’s worth. What you can’t do is go in
and tell you all the places you’re wrong. Go to Louisville with Ted Smith, who
is former ONC and has reasonable dexterity around this sort of stuff, working
pretty cutting edge stuff around asthma and work with him and the university
and build your kit engaging all the participants.
DR. COHEN: I want to synthesize this whole conversation. We can—what’s
happened in the data workgroup and in the roundtable is everybody wants to work
with us. There’s enormous energy for partnering to help us address problems. We
have enormous capital to build on. We haven’t really figured out how to
leverage that capital. We need to do that. Essentially we create partnerships
of people who know the data, people who know the tools, people who are coming
from different perspectives around problems to figure out how people would
There’s been this disparate idea in my mind, some folks are focused on the
patient, others are focused on the community. I say let’s pick a problem and
empower these teams to address—let’s say diabetes is the issue. Somebody
starts with the patient, they have a high A1C, what does that conversation
build out to? They live in a community that has high obesity rates or elevated
diabetes; other teams start from that perspective, from that community
perspective. How do they reduce diabetes in their community?
The goal would be to understand the data flow that’s coming from both
directions that are needed to improve quality of individual health through the
provider system and quality of community health. I think it’s really important
to create these teams in a way they haven’t been created before. Folks, old
time data geeks like me together with folks who know community data with people
who have bright young ideas about doing things in these tool kits and where to
get different kinds of data I’ve never thought about.
DR. CARR: Just to follow onto that, I think there would be some
fundamentals. You have to use the HHS data. You have to use some social media.
They’re the “must-haves” to solve it.
MR. DAVENHALL: I wanted to react a little bit to this idea that we need
something really practical to keep hammering at here on data access and use. I
think that’s what we really ought to stick to. I’m a big nut about you can’t
get—you have to market this stuff. We sit here in a building that’s
probably got the largest collection, the most—in the world. What hasn’t
happened is these people have not exploited this data for the purposes of the
What I mean by that is they could do modeling with this data. They’ve got
enough data they can reconstruct samples. They can get sample sizes large
enough to do small area estimation. We hear a little about that mentioned but
not much. I don’t have much detail on it. The National Center for Health
Statistics is a suggestion in data access should be exploiting this massive
collection but change their focus from not just thinking about they’re only
going to supply data to states, that their job is to supply data to communities
without worrying about what the problems are at the moment.
It’s as Walter says, in my community it is child protective services, it’s
foster children. It’s not obesity, and it’s not asthma. It should be in some
cases, right? All I’m saying is in some ways what we ought to be doing is
making recommendations of where is this data and break it loose. The Agency of
Children and Families has massive amounts of data in human services. We’re in
the early days of getting that broken loose, but I’m saying we ought to
describe very specifically—go through all these HHS agencies and say what
is it we know you have and what is it you’re not pushing out—the health
people have now pushed more data out than we probably can use.
DR. CARR: I want to stay anchored, though. We have lots of ideas. I want to
build on one. I’m going to take yours as a friendly amendment to this idea to
somehow involve NCHS data if we can. We have to do one thing before we do
everything. By the same token, you update our data set, we need to get someone
to artistically portray this—we need to make sure Josh’s recommendations
from that early meeting are here, make sure Bill’s recommendation that NCHS
data needs to be explored in terms of how we can best use that, but I want to
stay on—we’ve got to build to one thing. We’re on a hack-a-thon to heaven.
MR. ROSENTHAL: If you do that, that will be cutting edge. Everyone does
these hack-a-thons. Just so you know, they don’t work. At the end of the day,
very little output, like .001 percent. Solve-a-thon, that’s kind of niche, very
DR. CARR: Solve-a-thon, but using the data, and if they can’t use the data
and there’s nothing they can do with it, that’s word back to HHS.
DR. VAUGHN: Picking up on Bruce’s point a little bit to give an example of
one of the most, really it ended up being a moving hack-a-thon, sponsored by
AT&T of all people. It was around autism. The challenge was to hack autism.
The way in which they came up with the questions was they actually asked
patients and their families what would be helpful, what do you need, how can we
help? Developing those questions with those communities of particular interests
came up with ten challenge statements which were very practical, very
thoughtful, a few of them kind of whimsical, but all had to be met in a way in
which the patients and their families felt was helpful.
It was who defined the problem and who defined it as being solved, and how
you got from point A to B involved several different steps. In the end, there
are still people who try to sell their stuff, but in the end it came up with
things that were not just practical and useful and solved real problems, but
that were really responsive to the communities they were representing.
DR. CARR: I’m cognizant of time, but Bruce we know we have all these
communities; they have volunteered to work with us. They have problems. Your
point is well taken. Make sure we have the voice of the community in what we’re
trying to solve.
DR. MAYS: Let’s also make sure that– there are several organizations of
clinic directors. I think it would be great to have them at the table,
particularly the group that is the federally qualified health systems. They
easily will come to the table.
DR. COHEN: So we’re building these teams, which is great.
MR. ROSENTHAL: I think we should have output. So what I hear is SWAT team,
case study, then output, so what did we learn, what did we get right, what
didn’t we get right, and the goal is to create a curriculum for the
communities, not just that we do it but that we actually create some collateral
that’s systemic and repeatable for other folks.
DR. CARR: So who are the customers to this?
MR. ROSENTHAL: The goal is to do the action, create a little paper around
what we learned, what we thought, what we got wrong, and then the goal is a
thing you can give to community number two.
DR. CARR: So there will be multiple customers. It’ll be the community, but
it will also be HHS because that’s our charge.
MR. KAUSHAL: It has to be scalable. I think helping two communities is
great, but the power of our impact is much bigger. So scalable, whether it’s
repeatable via those tactics, but whatever we decide to do, whether it’s the
data source we’re releasing or the way we’re working with communities, it has
to be able to have macro-impact.
MR. CROWLEY: For each iteration of the product, there should be a feedback
mechanisms with those participants. For each iteration, we know what was
helpful, what wasn’t helpful, what would have been helpful if they had that
data, and we iterate to have a continuously improving package for these folks.
DR. FRANCIS: So I want to look at something that Denise Chrysler said in the
workshop that we had, which was that sometimes things like the law are seen as
a barrier improperly because we haven’t understood the resources that we have.
I think it isn’t just—there are a lot of reasons why data haven’t been
released. Some of it has to do with fears, some misplaced, and if we say
improve data access and use, part of that is making sure that it’s accessed and
used responsibly. It seems to me that part of this picture ought to be
understanding the barriers and where they’re actually a mistake, as Denise
said, and where there’s something to them that we need to be understanding or
working with for responsible access and use.
DR. CARR: I think also we can go back to the framework. At the end of this,
we can say here is this, and the stewardship framework aligns or does not, and
then why not?
DR. SUAREZ: Two quick comments, one is there is already an initiative under
the Health Data Consortium to evaluate and grade for data openness agencies.
NACHS is already going to get a B+ or a C- on their access and availability to
health information, how open they are. That’s a major initiative they announced
at Datapalooza. There’s a lot of interest in trying to define the criteria by
which organizations—there’s already things underway around that.
I think what I see that is lacking, and it’s been building on the comments
from everybody, is this one place resource for communities. At some point I
mentioned in some other discussion the idea of establishing a portal, a
gateway, a national gateway one-stop to look at tools and resources. The
problem we have is there’s a lot of data, and it’s lot of places. A lot of
people are marketing that data, and if I’m in a community, I have so many
places to go to and so many things to look into that I get lost in the shuffle.
DR. CARR: I think that’s what we heard Health Data Consortium aspires to do.
I thought that was one of the things that they said, that they’d be a place
that can link you to what’s out there. If it isn’t, we’ll bring it back.
DR. SUAREZ: I am thinking about a community-driven, community-oriented,
community-focused portal. Health Data Consortium has so many things
that—if we really are focusing on the community, which I thought we were,
I think that should be a major separate unique point of access. There is
healthdata.gov, and a lot of people are going to be routed there. Communities
can go there–
DR. CARR: That’s going to bring us back to Lilly’s list, but what I would
say with that is first of all, Health Data Consortium in many ways is a bit of
a blank slate, and Dwayne was very interested in our ideas. I think this is
right; just to say here’s healthdata.gov we know is too overwhelming. Who are
the communities and then how are their resources found? I think we’re on the
same page on that. We need to look at socializing the data. We have a plan.
You’re going to help coordinate this. We have the expertise around the table to
tell us what we need for each element of this. You’ll be getting some calls
from Lilly, and we’ll follow up with Dwayne. I told him we’d talk to him in two
weeks on this. What I’d like to do is just have you say a few words about our
conversation with Ryan on page three, on that list, and what our homework is.
MS. BRADLEY: So this committee has talked a lot about socializing HHS data.
It’s looking at the examples of Wikipedia, Drupal. We can really crowd source
some of this stuff and make it open. The question that’s been going around
within the group and maybe HTC and within HHS is what does social community
around our health data look like? To help hone us in we did have a smaller
meeting. Bryan would like us to think about prioritizing these data audiences.
If we could go through slide three and pick one or two of those and then look
at the functionalities that we at the last workgroup meeting identified as most
needed by these audiences—actually they were functionalities–
DR. SUAREZ: Just a quick reaction on slide three, I don’t see
“community” as a staple.
MS. BRADLEY: So I’d love to proceed even if you just jot it down today, you
could name community, give examples, and then the specific uses or needs of
those communities. I might have thrown it under public health.
DR. CARR: Let’s flesh it out to reflect the work that we’ve been doing and
how we want to either add it, articulate it, make it separate, whatever it is,
but let’s hear it from Bruce, Walter, whoever, to say more about that.
DR. SUAREZ: I would not lump it under public health.
DR. CARR: I’m saying to create that section, break that out then as a group.
I think as we prioritize we already have a vested interest in that. They’re
looking at it as what else should they put on healthdata.gov that would make it
easier for people to use. The other question is do we really put it on
healthdata.gov, or do we really look at Health Data Consortium to do that part
of the work.
MS. BRADLEY: I think there’s an idea behind our website right now—it
does not have a clear audience. Let’s at least start working with one defined
audience. Let’s think about the persona of the person visiting our website and
help them navigate. Bryan actually did not ask us to tackle that question of
where should it be–
DR. CARR: No, he said that’s something that we’ll have to think about, but I
think as we think about what are the needs it will become clear with what’s the
capacity at HHS versus what’s the assignment for healthdata.gov.
MS. BRADLEY: Those functionalities, I think you can think about them
together. In slide three and six, if we have to start somewhere, where do we
go? Maybe someone more tech-y than I am would even propose that all of those
are really easy and we should do all of them all at once. They will be rolling
out—we will be rolling out related media in July 2013.
DR. CARR: I don’t want to give short shrift to this. I think maybe we do
this as takeaway and you follow up with people, to take a look at it, think
about what it might look like. We’ll probably need to have a bit more
discussion about it. I am energized about our plan to have a solve-a-thon for
the community and get one of the affiliates of Health Data Consortium to do
that. Are we all in agreement on that and everybody willing to contribute their
expertise in terms of framing this and some of us perhaps to be there?
MS. BRADLEY: Can I have another ask around? I have been trying to understand
why we say we don’t know what to do to help communities because there’s a lot
with CDC, AHRQ, and ASPE to conduct studies, CMS. We conduct all these studies
to look at which interventions work and which don’t. I don’t know if it’s a
matter of it not being tagged or just not marketed, but I don’t understand what
to do with that. Bill and I spoke about the foster kids, and I went through and
dug through the ACS data about foster children in Maricopa County and how it
compares to Arizona and what does the national ratings look like? Then I ended
up on the county health rankings website, and it’s great. It gives you your
instant report card.
I think it is pretty user friendly, and then you click on the button, and it
gives you recommendations on what you might want to do in your county based on
where you fall. It says that if you have a lot of foster kids, home visits are
kind of a good intervention. It gives you a ranking of the evidence. One of the
challenges we propose for the National Day of Civic Hacking was around
developing some tool that might facilitate home visits. This isn’t—I think
you could help HHS understand—this is going to be posted somewhere, where
do we put these learnings, how do we carry it from one place to the next?
DR. COHEN: I think you hit on a fundamental issue. We do a great job of
coming up with ideas. We have no idea how to disseminate this information and
communicate them to the people that we need to reach. It’s a mystery. It’s not
something that’s traditionally been in HHS’s bailiwick. OCR started to point us
in the right direction, but for all the great stuff that’s in this building,
the data are essentially cult data because they’re not getting out to the
people who need them. That’s an enormous challenge. The federal government is
not designed or hasn’t been aware that a lot of their problem is figuring out
how to disseminate the information the people need.
DR. GREEN: I hear your question, the analogue to the clinician’s question.
We’ve made the right diagnosis. We did the right test. We told them what to do,
and look, it’s not getting better. Scale that up to communities. It’s all
there. Why don’t they just do it? It’s because implementation is largely
ignored, unsupported, unstructured. There’s not an infrastructure. One of our
conclusion as a committee is we could use a little help. That’s another
derivative of what you’re talking about. If we can actually start getting
concrete, if we can land the plane somewhere, I believe we will be able to
derive from that exercise. Things such as nearly a curriculum for the community
assistance agency, this could have far reaching implications. That’s one of the
things that’s worth doing.
DR. FRANCIS: The point of community-based research is that you’ve got to
work with the community to understand what the community wants to know. One of
the issues that we’re grappling with here is we’re not entirely sure whether
the questions that we’re answering are the questions that actually are the ones
that if you asked the community what questions you would want to have answered
and even different communities you’d get the same thing.
DR. COHEN: My purple button is an HHS—and NCHS is funded from one
percent evaluation funds. Why don’t we establish one percent education and
dissemination funds off the top for all HHS agencies to promote—this is
our recommendation to the Secretary. Pull one percent or a half a percent of
the budget off for dissemination and education.
DR. CARR: So we are still defining the problem, and we’re defining the
problem through a use case, and we’re going to have– many of the things that
Josh has on his list of seven is going to be coming out of that as well. We’re
going to look at that, so we’re going to have convergence on what we heard a
year ago, and pulling it together, but we’re going to have the use case. We’ll
define what the challenges are and then we’ll come back with a report-out.
Everyone is going to have an assignment, because we need the expertise of
everyone in this room to create this solve-a-thon, to be meaningful, and to
help us address these issues.
DR. SUAREZ: What’s the time frame?
DR. CARR: I have to speak to Dwayne Spradlin and find out whether we can tap
into these affiliate groups, that Nashville group, that if he has another one
that’s coming up we’ll go with that.
DR. SUAREZ: The affiliates group is just the Health Data Consortium created
as a separate kind of activity that includes all the state or local initiatives
that are like the Health Data Consortium.
MR. DAVENHALL: Not exactly, these were self-identified people who said in
their particular area they would take responsibility for starting to develop,
essentially, an affiliation with the Health Data Consortium. Some of them are
one or two people.
DR. CARR: You are raising a good point. We have a concept, and we need to
get who will do this solve-a-thon. I threw out the idea because I had heard of
one group that was presented at Datapalooza. I’m sure Leah can help us with
DR. MAYS: You may want to think about the California Healthcare Foundation.
They may underwrite the whole thing and work with you. That’s one of his
partners, but you could probably go to them yourself directly. They’re the ones
that did the community health that we did. They’re really moving. They’re
really looking to solve some issues.
DR. VAUGHAN: Their code-a-thon was around vulnerable populations. All the
apps had to be developed for homeless people. They had to be inexpensive. They
had to be easy to use. They had to be for low literacy.
DR. CARR: I’m not sure we’re looking for an app, but we’ll
develop—we’re looking to solve a problem using HHS data and the other
DR. VAUGHAN: So use curated data sets from healthdata.gov and they develop
products for families who are homeless. They develop a secure way of uploading
a limited amount of medical information to electronic health record using a
text. They were pretty innovative.
DR. CARR: So this is where we need your expertise.
DR. MAYS: I would see them as someone who could be a funder, convener, a
MR. ROSENTHAL: The HSC stuff is great, and we should in terms of some of the
things we’re doing just pay attention to it. They had asked us, individual
folks, what they should be focusing on, so when they’re doing the list and
saying what they’re doing, that comes out of direct conversations around that
that we’ve contributed to. We should be cognizant of that. Number two, on where
to go or what to do, what to solve, Louisiana and Louisville—but just keep
that in mind.
MR. CROWLEY: So we had discussion a few minutes ago about the communication
and outreach and strategies for that. Is there a defined strategic marketing
plan for health data and healthdata.gov?
MR. KAUSHAL: It may be superfluous because Health Data Consortium has a huge
budget and capacity for just that piece. I don’t think we need to recreate the
DR. CARR: I think people have appreciated the ideas that have been generated
here. I think it will be very powerful if we can demonstrate this is what we
said, here’s the evidence.
MR. CROWLEY: For that plan if it’s something that somebody’s already done
and done well, if HDC is the right partner I don’t think we need to recreate
the wheel for a lot of this stuff. Use what’s there, leverage it.
DR. VAUGHAN: The foundation fund has funded the key data sets that are on
HDC’s site. There’s a separate piece from one of the presidential innovation
fellows that also started to retool it. That’s through the Office of Technology
and Science Policy.
DR. MAYS: Zero Divide is a tech foundation, and they really do get down to
the community level, so you might even ask Tessie.
DR. COHEN: All these organizations should be team members for this
DR. CARR: Okay, with that, we’re going to conclude. Thank you everyone,
brilliant as usual.
(Whereupon, the meeting adjourned at 5:00 p.m.)