[This Transcript is Unedited]
DEPARTMENT OF HEALTH AND HUMAN SERVICES
NATIONAL COMMITTEE ON VITAL AND HEALTH STATISTICS
February 26, 2009
Hubert Humphrey Building
200 Independence Avenue SW
CASET Associates, Ltd.
Fairfax, Virginia 22030
Table of Contents
- Call to Order, Welcome
- Implementing Standards: A View from the UK and IHTSDO
- Health Statistics for the 21st Century – Update
- NCHS/BSC Update – Vital Stats
- Emerging and Innovative Sources of Health Data
- Committee Discussion for Follow-Up
P R O C E E D I N G S (10:00 a.m.)
MR. REYNOLDS: If everybody could please take their seats, we need to get
started. Are we on the internet and everything? Okay, good.
Good morning. I’d like to call this meeting to order. This is the second day
of meetings of the National Committee on Vital and Health Statistics. The
National Committee is the main public advisory committee to HHS on national
health information policy.
I am Harry Reynolds, work at Blue Cross Blue Shield North Carolina, and
Chair of the Committee. I want to welcome Committee members, HHS staff, and
others here in person, also welcome those listening in on the internet. I would
like to remind everyone to speak clearly and into the microphone.
Now let’s have introductions around the table, and then around the room. For
those on the National Committee I would ask if you have any conflicts of
interest related to any issues coming before us today, would you please so
publicly indicate during the introduction. And I have no conflicts.
MR. SCANLON: Good morning, this is Jim Scanlon, I am Deputy Assistant
Secretary for Planning and Evaluation at HHS, and I’m the Executive Staff
Director for the Full Committee.
MR. BLAIR: Jeff Blair, Director of Health Informatics, Lovelace Clinic
Foundation, member of the Full Committee. And LCF is one of the nine original
states and regions that have received an NHIN trial limitations contract.
DR. SCANLON: Bill Scanlon, Health Policy Research and Development, member of
the Committee, no conflicts.
DR. STEINWACHS: Don Steinwachs, Johns Hopkins University, member of the
Committee, no conflicts.
DR. FITZMAURICE: Mike Fitzmaurice, Agency for Healthcare Research and
Quality, I’m not a member of the Committee.
DR. SUAREZ: Walter Suarez with the Institute for HIPAA/HIT Education and
Research. I am also a member of several of the national projects, including a
Board member of the Health Information Technology Standards Panel, a member of
the Technical Advisory Panel for the Health Information Security and Privacy
Collaborative, and a member of the CCHIT Policy Privacy and Compliance Advisory
Committee, and insofar as that might create a conflict, I so state.
MS. MILAM: Sallie Milam, West Virginia Healthcare Authority and West
Virginia Health Information Network, member of the Committee, and an NHIN
DR. MIDDLETON: Blackford Middleton from Partners Healthcare Clinical
Informatics Research and Development and the Center for IT Leadership, member
of the Committee, no conflicts.
MS. MCCALL: Carol McCall, Human, member of Committee, terribly conflicted,
but no known conflicts.
MR. LAND: Garland Land, National Association for Public Health Statistics
and Information Systems, member of the Committee.
DR. FERRER: Jorge Ferrer, Veterans Health Administration.
DR. OVERHAGE: Marc Overhage, Regenstrief Institute, Indiana Health
Information Exchange, a member of the Committee, and I see no conflicts with
DR. HORNBROOK: Mark Hornbrook, Kaiser Permanente, member of the Committee,
and no conflicts for today’s discussions.
DR. GREEN: Larry Green, University of Colorado, member of the Committee, no
DR. CARR: Justine Carr, Caritas Christi Healthcare, member of the Committee,
DR. TANG: Paul Tang, Palo Alto Medical Foundation, member of the Committee,
DR. WARREN: Judy Warren, University of Kansas School of Nursing, member of
the Committee, and no conflicts.
DR. FRANCIS: Leslie Francis, University of Utah, member of the Committee,
and no conflicts.
MS. GREENBERG: Marjorie Greenberg, National Center for Health Statistics,
CVC, and Executive Secretary to the Committee.
DR. SONDIK: Ed Sondik, Director of the National Center for Health
MR. REYNOLDS: Welcome everyone. You thought yesterday was exciting, hold on,
we’ll start with our first speaker, and I think seatbelts might be required. We
look forward to that. What is always exciting, as we try to do, is it get
people that have done in other places, we did that with a lot of our ehealth
records and so.
Today, although Professor Severs introduced himself, I’ll give him a formal
introduction as the speaker. And he is currently the Chairman of the NHS
Information Standards Board, which is the single assurance and approval body
for the use of information standards within the NHS, and with its communicating
partners in England. As well as being a practicing consultant, Martin is also
Associate Dean of Clinical Practice within the science facility at the
University of Portsmouth, and the Higher Education Institutions Representative
Non-Executive Director on the Hampshire and Isle of Wight Strategic Health
Authority. He has over 15 years experience in health informatics and was the
original chair of the medical group initiator of the professional information
advisory groups in their first iteration. He is also the UK representative of
the Management Board of the International Health Terminology Standards
Development Organization, and is its Chairman, IHTSDO, owns and manages SNOMED
We have asked Professor Severs to come in today and give us a sense of how
they have made this work over in the UK. So with no further ado, thank you. He
is in the states for some other reason, and we were able to borrow his time, so
we consider ourselves extremely fortunate, so thank you very much.
MR. SEVERS: Mr. Chairman, and members of the Committee, thank you for
inviting me. I do have a conflict. I’m quite happy in the discussion period
afterwards to tell you how I manage it. On the one hand I’m the gamekeeper, and
on the other hand I’m the poacher, in terms of being Chairman of a standards
development organization internationally, and in the UK I’m the Chairman of the
body that formally approves standards. I can tell you how we deal with that,
I’m kept out of the room and I take no part in it, and the appraisal, I take no
part in any of the processes that go on.
Without further ado, I’m going to make the presentation over four parts in
twenty minutes, and give you twenty minutes in which to cross-examine me,
question me, or engage in discussion, whichever takes your fancy. I’m going to
spend a little bit on context, because I believe the context that I’m going to
set is very different to the context that you’re used to.
I’m going to give you a perspective, a present perspective on it from the
Information Standards Board for Health and Social Care for England. Then I’m
going to give you a perspective from an International Health Terminology
Standards Development Organization, an international developer. Then I’m going
to give you some personal conclusions, I’m going to stress these are personal
because they’re not exactly straightforward and may be slightly controversial.
Then we’ve got twenty minutes for discussion.
The first thing I want to say is that within the UK we have three levels of
standards which we think about: framework, or of the high level overarching
structure within which standards at levels can be derived or developed; the
fundamental standards, which are the standards which populate the operational
standards, which are the detailed and precisely defined standards for
operational use within specific areas of the NHS.
The thing that will strike you as different during the presentation is that
we belief the operational standards should have a clearly defined requirement
that should be approved, it should have the specification of what the standard
is, and every standard should be associated with organizational, technical, and
clinical and user implementation guidance. If you want standards to be
effective you have to implement them. You cannot just vote on them by
consensus, you have to implement them.
In order to show that you’ve implemented them, there’s got to be some
derived conformance criteria to establish that they are being used properly,
both organizationally, clinically, and technically.
Having set that context, the Information Standards Board for Health Social
Care was setup originally in the year 2000. It is quasi-independent, in that my
accountability and that of the domains leads in the various specialist areas
goes to the Director of Information for the NHS, although my offices and
support organization are hosted by one of the agencies NHS connecting for
health – independence.
It has structurally evolved during that period, but I’m not going to go into
the details, but one of the things I would urge to think about in any approval
system is that it must evolve with time and with learning. We’re the only
approval body to the NHS and Social Care in England. We approve standards based
on appraisal, that is explicit assessment criteria, which we publish plus value
judgment, standards are often at the forefront, and you can’t just raise an
explicit assessment process, in the end you do need experts involved as well,
so we call that appraisal.
And the evidence that we demand is from implementation evaluation. Does it
work? We have appraisals from six different domains, and we don’t necessarily
use all domains on all standards, but we try to. These different perspectives
are important because there’s a need to manage tension. Political perspective
might be at odds with the managerial perspective, which might be at odds with
the public health perspective. The only way standards can be implemented out
and effective is if you manage that tension of these different perspectives.
We focus on three key points: implementation, can it be implemented,
implementability would be a better description of where we’re at,
interoperability, and safety. We internally manage the tension between the
major stakeholders at the Board level where we have professional
representation. The main organizations for auditing and scrutinizing the NHS
organizations are also present, as well as the development side of the
technical and statistics organizations within the NHS. They are all managed
within the same Board.
One of the key messages, and I’ll drill this one and I make no apologies for
drilling on this one, the value of a standard only lies in effective
implementation. The evidence is quite overwhelming that that’s the case.
Therefore, there are two different perspectives of what a standard is. The
classic standards developer view of a standard dominantly it’s a document
specifying the nationally or internationally agreed properties of a good
consensus process where everybody agrees.
We take a different view as a standards approver. A standard is a quality or
measure serving as a basis or principle to which others conform or should
conform or by which the quality of others is judged, i.e. you can only approve
a standard if you’ve got some evidence that it can be implemented.
The NHS does take the second view, so the ISB is not acting in a way that’s
outside where the NHS view is of standards, but it is a different view to what
you may have in your heads about what a standard is.
In terms of the work done by the Information Standards Board, you can see
from this diagram that our workload is increasing year on year out. That’s the
volume of standards that’s come through ISP since 2002. The majority of those
standards are what I would describe as record classification standards, that is
utilizing the content of a record for national purposes. If that is not done by
automatic computer processing, that means that you have to have human
transcription. Human transcription is a significant cost. And I’ll come back to
this theme as we go further down the talk. The bottom line is the demand is
The second issue is the nature of the standards that are going through. At
the bottom of your screen you’ll see big blue blocks, those are requirements.
And you can see when we started off people would come to us with large numbers
of requirements. As you can see over time when we said that’s fine, but when
you present the draft standard this is the evidence that we would require based
on this. The number of people coming with requirements has dropped
You will also note that the top gray boxes are significant in that they are
getting larger. That is as we evolve and our understanding gets better, change
to existing standards is a major piece of our work as we go forward.
These are just two examples to give you a flavor of the type of operational
standards. The first you would probably be very comfortable with, which is
utilizing data for payment for hospitals, ‘payment by result’ is the word
we use in England, and ‘SUS’ is the secondary users service, which is a
large database we have for all the data required that goes forward to the
Center. Operational standards work well in the area of classification, and
we’re able to pay hospitals.
The other example I’d like to show is the example of the payment of GPs,
family physicians would I think be the equivalent in the United States. You
will notice that the language is different, we are now talking about referral
patterns, we are now talking about outcomes to individual patients, we are now
talking about prescribing habits to individuals, we are now talking about
quality of care. We’re at a level of granularity that is below where we were
before. All this is based on a terminology called Read Codes Version 2, and as
you’ll know, Read Codes Version 2 are an integral part of SNOMED CT, SNOMED CT
being the next generation on from the Read Codes.
But what you get for increased granularity of data, is you start to get at
really key issues like quality of care, what prescribing habits are for
individuals, outcomes for individuals, with a more sophisticated tool.
If I may move on to the International Health Terminology Standards
Development Organization, this is just some basic information. And I would like
to just thank this Committee for its report in December of 2002 where it did an
analysis of various terminologies, and I’m delighted to say SNOMED CT came out
very favorably in your report. But it did have a big question mark of the
governance arrangements. I believe the governance arrangements now should meet
your criteria more than adequately.
I’m not going to go through this slide, but to say we have an organization
that gives you a flavor of what the nature of that organization is.
What I would like you to note is a number of key things about this. The
association of SNOMED CT, not the individual members, but the individual
members which are jurisdictions, and I am delighted to say the United States of
America is one of those jurisdictions, govern the association through the
Articles of Association. All authority is derived from that General Assembly:
one member, one vote. They have a Management Board which I have the pleasure to
Chair, and I have to appear before my General Assembly twice a year and account
for the Management Board to actions over the last six month period. I have to
present annual accounts, I have to face them and they scrutinize me, and that
is something I find personally extremely uncomfortable, but I think it’s
absolutely essential in that standards development organization that is going
to produce product that is going to be used for the care of our citizens, our
family. I think the governance arrangements are absolutely essential.
Most of the decision making is devolved to the Management Board, and the
reason for that is you need slick and quick efficient management, you cannot
wait six months for decision making. So there’s a separation between strategic
management and good governance.
The Management Board is accountable, the General Assembly can if it so
wishes at anytime say we will sack the Management Board forthwith, and can do
so without notice. The intent here is that members do not have a property right
in SNOMED CT that they could exert or exploit to disadvantage others. In
international standards this is important.
Secondly, the IHTSDO decision making arrangements can be effective,
responsive, agile and accountable both upwards and downwards to the community
of practice. I think at least one of your members on this Committee has got
experience of making sure that happens.
The vision and values are part of the explicit governance of the
association, you cannot just say the nice words of openness and transparency
and fairness, you have to live them. Members are countries through their
designated agents. Members pay fees that are calculated on a trusted third
party World Bank metric based on wealth and based on the ability to repay
loans. We don’t negotiate fees, it’s by the trusted third party World Bank
metric, there is no discussion about doing deals, I don’t have any bureaucracy
or any staff negotiating prices for membership fees. It’s there, it’s on the
web, everybody can see it. There is one license, a single end user license, no
marketing, no strategy, one single license, i.e. there is a commitment to the
health of the citizen, not paying people to do administrative tasks if at all
Affiliates within a country pay no fees to the IHTSDO whatsoever as part of
the membership fee. There are no exclusions to that, anybody in the United
States can use SNOMED CT if they have an affiliate license.
There are changes set by the IHTSDO for non-member nations. The reason for
that is that we wish to not have loads of affiliate licenses and lots of
payment schemes to individual vendors, we prefer a very simple one nation, one
membership fee, and get rid of as much administrative overhead as possible.
So what have we achieved in less than two years? One, is that we’ve
maintained the timely release of SNOMED CT twice a year, and a smooth
transition from old to new owners. We have had to set up an international
organization in the Danish legal system, which we’ve recruited and
independently funded, which I think is also important in all of our offices.
We have designed and acquired and are about to implement a replacement
tooling infrastructure for designing and building a SNOMED CT. And we focused
all of the support service effort of improving the quality of SNOMED CT, which
is a never ending task, but clearly one which people have said oh that little
bit’s wrong, that little bit’s wrong, you have to listen to customer feedback,
you have to improve the quality.
I just want to note that translation and RefSet modules, not so much about
the RefSet, but translation and cultural differences are really important to
address in an international standard. If countries don’t feel that they have a
voice, that there is cultural sensitivity to their needs, they will not
We have to demonstrate and I believe do demonstrate a primary citizen focus.
We have already set up a broad based license fee exemption for 49 low income
countries. We intend to continue to have innovation around the primary citizen
focus. We have attracted two new members, Singapore and Cyprus, and one other
potential new member has made it public that they wish to join and that is
Spain, and they’re in the process at the moment.
The other point I would like to stress on this slide is that we have
demonstrated a commitment to harmonization. Working to build a standards
organization is not easy, both for them and for us, this is not a one-sided
issue. There has to be give in order to rationalize the standards market, and
with active liaison in eight organizations and practical work ongoing, which is
why I’m in America with six others. Maintain financial balance and demonstrate
openness, transparency, and fairness.
In the UK we’re looking to have SNOMED CT in primary care systems, they’re
already starting to be rolled out at this point in time, which will continue.
Within acute care the system actually applies, the big system is going at the
moment – SNOMED CT inside.
Within health systems themselves, I believe that there are major risk cost
benefits from international standard. For example, at the moment we are trying
to support major policy initiatives around maternity and child health. Those
sets of data need to be encoded in SNOMED CT such that the data can be
collected from real patients in real clinical settings and used to help
understand it at top level government, whether or not they’re meeting the needs
of the population with maternity and child health.
Now, that clearly is not an unusual agenda across the world, for example, in
the United States. If the UK has already spent many hundreds of thousands, say
millions, on that process, within the IHTSDO we have to make available
resources of that nature to other countries. Once that is made available to
another country, they can then utilize that data for start – and it cuts down
their costs, and actually cuts down quite a lot of risk in terms of
establishing whether that dataset can be collected, can be implemented in
systems, and can deliver the information at the national level you need. That
sharing is an important aspect of an international standards development
So some personal conclusions, Mr. Chair. The first is: What processes are in
place to ensure that a country has confidence in its own data? This in the UK
was brought out by the Kennedy Report, which investigated major issues around
data in the UK.
Does this variation in data reflect real differences in practice, or is it
related to the information standards, or both, and how would you tell? I’m
going to give you a little bit more detail on this in a while, and I’m happy to
go into even finer detail in answer to your questions.
First, on fundamental standards. I believe the fundamental standards should
ideally be international and internationally controlled. There should be as few
of them as possible, and the owners, the organizations that own these, in my
view should have certain attributes, which I’ll go into. Why? I believe risk
sharing for countries is important. Risks should decrease over time, not
Secondly, cost sharing, cost reduction per country is really important. If
one country has done an investment in a guideline for myocardial infarction,
using for example SNOMED CT, it is ludicrous not to share that with other
countries and give them a start.
The third thing is benefit sharing. We can utilize the operational work of
other countries to help other countries, in fact give them away, dare I say it
to other countries, to use as a start.
I think there has to be, and I’m not going to go through this slide so don’t
worry about that, but there has to be some duties of owners of international
fundamental standards. There must be a strong governance process that ensures
that it is safely maintained and remains fit for purposes. There must be a
business model that ensures a funding stream for continuity. There must have a
management process and resources that deliver timely, safe, and culturally
sensitive, quality assured products that meet business needs.
I believe the requirements process should be open, transparent, and fair, as
should the prioritization process. One or more countries, and indeed the
managers, cannot dominate, there has to be good governance.
And there should be an electronic distribution service that allows for
controlled access to all parts of the standard. There is no point putting a
technical point out and pretending you’ve given it away free. Then actually the
documentation enables you to implement it effectively costs an arm and a leg,
and you don’t get access to it. You have to put all parts of the standard out,
such that everybody can access all parts needed for implementation.
Equally, I think there is a responsibility for any country that involves
themselves in an international standard to engage with it. And I think overlaps
in fundamental standards should be actively discouraged, and contradictions
proactively removed. Therefore every international standards development
organization should have somewhere in its articles of association, or whatever
terminology that fits within a given jurisdiction, it has to stretch out to
other standards organizations and work with them.
In terms of the big picture, one of the big learning points that has come
out from the UK is this one. That we’re all very comfortable in the human world
with fashion or practice, in my case medical practice, but the whole issue of
professional practice information management is a bit of a dark art. Is there
anywhere in the world a national curriculum for recordkeeping? Everybody has a
very eloquent policy statement for team-based care. Show me any country that
has the professional practice recordkeeping standards for team-based care.
Organizational management people everybody’s very aware of, but in essence
that’s about classification of data that already exists within records. Policy
information strategies from policy needs to be though through.
My challenge at the fundamental technical level is I believe recordkeeping
standards are as much an important part of fundamental information standards as
anything else. I think a nation needs to be explicit about its fundamental
record content standards, and I put SNOMED CT as one of those, I’ll put HL7
Version 3 as another one of those. I think also there needs to be fundamental
record classification standards, ICD-10, in England OPCS 4.3, are good
examples. But there has to be a relationship that’s machine processible,
between the record content standards and the record classification standards,
unless explicitly you want human transcription. And that is the decision that
has to be taken on a national level.
Equally, the operational standards, datasets, electronic guidelines, care
pathways, clinical communications, registers, all need to utilize the agreed
national, and ideally international, fundamental standards.
More controversially I believe there should be a single point of approval
within any nation both of the fundamental and the operational standards.
I believe, Mr. Chairman and members of the Committee, that benefits from
information standards only accrue from appropriate and effective
implementation. I believe implementation has technical, organizational, and
user dimensions. Attention to acknowledging all these dimensions, in my view,
personal view, in current information standards activities could be improved,
in particular the absence of clinical governance is a significant concern.
MR. REYNOLDS: Professor, thank you very much, we appreciate it. And now I’m
going open it up for questions from the group. Paul, Judy. Go ahead, Paul.
DR. TANG: Outstanding presentation. I think a key message on your summary
slide is the notion that it is not a useful standard unless it’s effectively
implemented, and all the structures and policies and practices that make it so
and how to evaluate it, the standard. That is really key and something we need
The other piece is the clinical governance, and presumably the clinical
participation. And it goes back to your context, which is the clinical user.
That is one of the problems we have in our SCOs is how do we get real clinical
users to have the time and wherewithal and the patience to put up with this
process so that they can be dutifully involved, but also so that you can have
effective implementation and be driven that way.
DR. SEVERS: The challenge you set out is a big one. My issue here is that
the science of clinical practice at the profession level has outstripped the
science of recordkeeping at the profession level, such that the informaticians,
the politicians, the managers, have no where to go for an authoritative
decision. Let me just push this with some stories, because I think the stories
are quite helpful.
What is the professional definition of a diagnosis, for example? Simple
question. And where is the authority for that at the record content level?
There is a definition of what a diagnosis is, for statistical purposes, for
record classification level in some instances. So where do you go to get help?
What is an established professional practice?
Another good example of the clinical definitions, one of my early
experiences is from the very senior person in the UK civil service, taunts some
big manuals and says look at the variation in death rate from myocardial
infarction, how would you explain that with your extremely good data knowledge?
And I said, What is the definition of myocardial infarction? Where is it
written down? Where is it taught to any clinician? Where would I go to find
out? There is good evidence from the research literature in England, there are
at least three definitions in current use in the heads of clinicians, which
could account for a 30 percent variation in the incidence of myocardial
infarction, for example.
If we are going to have rigor, money, and effort into the computational
element, we’ve actually got to have some profession-based. So it’s not just
individual, Dr. Tang, I think there’s an issue here about the professions. If
you’re going to be registered as a nurse, as a doctor, as a physiotherapist,
pharmacist, what are the standards of recordkeeping? What are the clinical
definitions that are taught to you? Where do you go to find out where they have
My challenge is not to the individuals. I think we have many, many, many
able individuals. My challenge is where nationally do you get the formal
professional authority that this is the way we go. And where do you go
internationally for the same information.
Sorry about the long answer, but I think it was a difficult question.
MR. REYNOLDS: Judy.
DR. WARREN: Mine was just an acknowledgement. For those of you who may or
may not know, I sit on the Quality Assurance Committee of IHTSDO, and had the
pleasure of working with Martin for the last two years, and was the former
chair of the nursing working group of the organization. And some of the things
that Martin showed you on how organizations should behave I think are critical,
because in the evolution of a U.S. organization owned terminology, to one that
was owned internationally, there was a lot of conflict, and certainly huge
amounts of fear. And I think some of the lessons that this organization has put
together and has migrated from has alleviated most of the fear. And what you
will begin to see coming out of SNOMED is a far better quality product, and
there really has been a huge emphasis on creating the quality for that.
So as you take a look at these slides, and for our own use, I’d really focus
in on that. And then just a personal thank you to Martin, because he has paid
attention to me when I was very upset, calmed me down, assured, listened, and I
think we’re in a much better place both as an international organization and
just as a standard as well.
MR. REYNOLDS: One last question, and then Blackford.
We feel your pain with Judy, Professor, we do, we truly can understand that.
MR. REYNOLDS: So you looked at what you do, you’ve probably looked at what
we do from afar. Can you make any comments on – so congratulations we just gave
the job of billing us here – what would be the things that you would say are
the things that we would have – would currently or would struggle the most
with to get to the kind of things that you’re talking about?
DR. SEVERS: I think a listing of the fundamental standards, both at
recordkeeping, record content, record classification. And then managed
interrelationships would be the key thing that I would insist upon at national
I would also recommend, because normal people don’t understand any of the
language of this, it’s just way beyond them, in my view there needs to be a
service so that if a department, a government department with a legitimate
right, or in our case, I don’t know what the equivalent is, the National
Institute of Clinical Excellence, the organization that produces guidelines,
needs to say we need this to be implemented in systems. There is a service that
knows what the recordkeeping standards are, what the clinical definitions are,
what the fundamental content standards are, such that this guideline, dataset,
register, can be built and be used by humans and processed by machines, is a
key service line structure, that I believe can transform the way ordinary
people think about health informatics and about information standards. Because
at the moment is still a little bit of a dark art. We need to get down to the
implementation front and help people develop things that can be used in real
Those would be the two top recommendations I would make.
MR. REYNOLDS: Blackford. Thank you very much.
DR. MIDDLETON: I would like to echo Paul’s comments. Thank you for an
I guess at Partners Healthcare you may know Howard Goldberg, and of course
we’re actually working on this problem of implementing SNOMED as the
terminology service for the clinical information system infrastructure.
My question pertains to two dimensions of real live implementation of this
now in practice. On the one hand the terminology, of course, still may not meet
the needs of the average end user clinician. That is there are significant
human interface issues, a long list look-up, and finally appropriate
terminology across multiple axes, as well as compositional issues of putting
things together in a way that makes sense clinically.
I agree with you wholeheartedly, and to hear you say the fundamental
requirements start at the top with a business requirement and a clinical issue,
integration issue, and a technical integration issue, you know, I commend you
for that. I think however we have a long way to go, and I see you agree.
The second though is that there’s a fundamental relationship between the
standards and knowledge representation. As soon as someone begins to use the
standard for defining knowledge, if you will, for example, we use the SNOMED
infrastructure to define a subset of people we call diabetics, now here’s the
trick. It’s not so hard to find the algorithm or to find a reasonable way to
define the subset called diabetics, what gets tricky is when the reference
architecture, if you will, from SNOMED changes under our feet. There are block
shifts of terms from tree to tree, or axis to axis, and there is then this
relationship between terminology management and knowledge management, which I
think is not yet adequately discovered, explored. I would look to you for help
in ways to make that easier as we are called upon to adopt these standards and
use them increasingly in an in-depth way in our systems, how do we manage
terminology and al the related parts like knowledge, first and foremost.
DR. SEVERS: First of all, thank you very much. I entirely agree with all
your comments. That’s not good enough. You needed an answer.
Can I just say just by way of a preamble, my top priority for the first two
years at the Management Board was to build it and get where we are, and get a
tooling infrastructure in place that enables to design and build and release
mechanism, because the existing structure wasn’t there, and I’ve got to get
that implemented in next year. You will note that every document that will now
be emerging from the IHTSDO as implementation, there is set of papers going
– example, about refocusing around implementation.
There are two ways we’re tackling this. One, we have the affiliate forum,
which dominantly goes with an affiliate license. We’re also setting up at this
point, and we’re not sure how to balance this, what we call an implementation
special interest group, and that is all the – what we see going forward is
that the majority of our effort is going to be in discussion with people like
your good self. And the reason is, is implementation, the best source of real
life issues that are going to improve implementation are from the people that
are doing it, okay for real, and therefore those are the people we need to
speak to, listen to, in order that we can then get implementation tools, we can
get proper strategies of when we change the basic structure so there’s a long
warning time so that people can then work through it.
What I’m saying is I’m not laying out a set plan, what I am laying out is a
set of processes of an organization that knows it has a challenge, is willing
to shift its ground, is focusing its strategic intent on the implementation
issue, and the dialogue with people who have real life business needs and
priorities around implementation, in all of that we can focus our not
insubstantial voluntary intellectual resource and our smaller financial
resource into that very issue.
And I’m sorry I can’t give you a definitive answer.
DR. MIDDLETON: One quick follow-up. I think this sounds great, and I’m happy
to engage, and of course everyone around the table is equally happy to engage.
I think just to put a fine point on the related question. It may be that
actually terminology to management actually has to have a broader view of its
role with respect to knowledge management. What I mean simply is suppose
actually it was the terminologist’s responsibility to define subsets of
diabetics, then there would be a central authority, a central standard, and all
the rest of it, as opposed to each implementation worrying about this issue. In
that way we might propagate terminology and knowledge in a much more efficient
fashion than terminology and then knowledge.
DR. SEVERS: Can I come back, Mr. Chairman, on that?
Your suggestion in essence is around an international health content
standards development organization. I couldn’t possibly comment in an
organizational sense, but on a personal sense I receive it positively.
The second issue is around knowledge management. One of the issues that you
will see with SNOMED is that I’m trying with my Management Board colleagues to
encourage established international organizations that are viewed by the
clinical communities as the authoritative international organizations to become
involved in SNOMED as a special interest group. Because I believe that we
cannot do the type of work that you are suggesting without good clinical
governance of this is the right way to represent this, this does make sense
clinically and this one doesn’t.
So in that respect I’ve already got the World Association of Societies of
Pathology and Laboratory Medicine special interest group. We’re having ongoing
discussions with the World Organization of Networks of Clinical Academes of
family medicine in general. And we’re also in the early stages of discussions
with the International Council of Nurses.
MR. REYNOLDS: I have Mike, Paul, and Jorge, and then we need to move onto
the next speaker.
DR. FITZMAURICE: Two questions. One you may have already answered, in which
case the answer would be quick. How should the standards be structured? Is
there a design? Should you be able to have a computer discover what you tell
the computer you want? Like a clinical record Google, but in a standards
development document, not in medical record.
Secondly, better development and use of medical records you’re promoting, I
like that. How would this use of coding affect the health of the patient, and
patients like that one, I think would be paramount in most researchers’ minds
and maybe even in clinical minds. But you also ask how well does this describe
the patient systems, diagnosis of symptoms, diagnosis and treatment, and my
income. All these questions asked in the United States, is the same true in the
UK, and/or in SNOMED that financial payment considerations, which is in need of
some users, or maybe resource use, are considered in dividing categories and
setting categories and coding.
DR. SEVERS: That is quite a portfolio of questions. It all depends on how
you wish to pay your clinicians. If you wish to pay your clinicians on detailed
clinical outcomes, then it does need to be recorded in the record. If you want
to pay your clinicians on certain statistical labels, then you use the record
classification system. I try to demonstrate that we are paying the GPs in
England based on what’s in their records, based on good quality of care and
using a terminology. We didn’t need that for the hospitals where you only need
a classification of diagnoses in terms of primary diagnoses to match to an HRG
system that we can use RCD type map to a grouping system, align it with a
costing and pay that way.
I’m trying to be fair to your question, but the reality is if it’s important
that SNOMED is used for that purpose, then we need to demonstrate it is fit for
that purpose given the specific use case that that individual member country
wants to pursue. That is the first thing. So yes it is within consideration.
But if you look at our articles, it’s about the health of humankind. It is
what’s in the record. SNOMED is about populating the record for the health of
the individual. It’s about that trusted relationship between the nurse and the
patient, or the doctor and the patient, or the healthcare team. It is about
recording accurately my health. Now, if you said to me should that be available
to the patient, I would say yes. And I think one of the biggest stimulants that
any system could have is that post-consultation the records should be available
to the patient.
And one of the good things about having a structured record is that’s
enabled, and third-party data can be hidden because it’s structured. So data
that might cause harm and third-party data, which shouldn’t be disclosed, can
be hidden because it’s structured and you can do that. So that is one of the
And about the first bit of your question, because I missed that –
DR. FITZMAURICE: When we put out standards, whether it’s SNOMED or whether
it’s HL7, is there a structure that you would recommend for the standards so
that you know where to go look for things, that there’s a heading like this
pertains to the diagnosis, this pertains to –
DR. SEVERS: Yes, okay. Because this is a spectrum between the clinical
profession recordkeeping standard, say in England all the Academies of Royal
Colleges have just set a standard saying that if anybody is admitted to
hospital these sections need to be in the record. If the new dataset is going
to be produced by NIST on a guideline, I would expect it to be structured in
the same way such that it can be implemented in the record, such that you could
already pre-populate a record electronically, with the required data items. Now
we then change clinical practice from recording everything, to recording some
things and validating other things, because we’re already pre-populated it.
So, for example, the spironolactone debacle, where we actually had the
fourth line agent name spironolactone for treatment of heart failure, all the
research which was beautifully well done had a large number of exceptions. If
they had been corded up when the guidelines, the use spironolactone came in,
you could have had the decision support system, as my colleague Dr. Middleton
just pointed out, built into the system such that all the exceptions could have
been flagged, and we wouldn’t have got the situation whereby through no fault
of the clinician or no fault of the policymakers the 199 exceptions which no
human being could ever remember, would have got inside the system. So I believe
the use of terminology, especially around decisions that can be of enormous
help. Equally the chief medical officer may make a statement, another benefit,
that everybody with chronic respiratory disease, for example, should have a flu
vaccination. Great, what does that mean?
Unless you have a terminology which every instance in the record that could
be described in a respiratory disorder, or chronic respiratory disorder, so you
may then have your chief medical officer instead of writing out just a paper
document may send out a piece of software that enables every user, every
general practitioner in the country to load it up and identify all those
patients with chronic asthma in the record, chronic bronchitis, emphysema, all
of the record content items which are necessary to have which would map to the
global chronic respiratory. So there are massive benefits in terms of using
detailed highly granular information.
MR. REYNOLDS: We have Paul then Jorge and then we’re cutting it off.
DR. TANG: So Blackford talked about going a level down from the SNOMED
terminology to feed into the clinical report. And you talked about getting the
information in the system. Now I want to talk about getting the information
into the system. And I am also going to pick on the risk carrying, risk down
the time, costs down the time, and sharing the work of others.
So to do this I’m going to take an example from an organization I don’t
belong to, Kaiser, for two reasons. One, I can then say something good about
it, and two, since it’s not my asset I can give it away.
They have something called CMT as an interface terminology to SNOMED, as an
example, and that is to facilitate getting information into the system, and
then by which we can derive value. Do you have a way of having Kaiser share its
what’s called utilize the operational standards of others, using Kaiser’s
operational standard to help enhance the overall value to all countries?
Clearly, Kaiser would have to be rewarded for that.
MS. MCCALL: It’s a reward unto itself.
DR. TANG: That’s right. Is it possible to encourage that? Is there a way to
DR. SEVERS: Our Articles of Association are 75 pages long. But what you are
describing is a culture, and within the Articles of Association if that
interface was owned by the United States then there is a way of bringing it up
to the international level. So extensions of SNOMED can be brought from a given
member to the international arena. But with that right becomes responsibility.
This is a difference. So anything that we take internationally we have to pay
for and maintain, but we can do it.
But the question you wanted was a level below that, which was Kaiser. And
Kaiser has an interface technology, and the issue is that if – first of
all it would need to be deemed to be a very useful thing for the United States,
and if Kaiser then took it to the national level and everybody was happy then
there is a mechanism by which we can get that shared internationally within our
Articles of Association.
I say that with a caveat because I don’t know exactly what we’re talking
about, but the mechanisms exist within the Articles of Association for taking
things up. But we can’t just go to a colleague who has developed something a
derivative of SNOMED, we can ask for it, they can say no, we can ask for it and
they can say yes that’s fine, but we cannot go in there and take somebody’s
life’s work who have done a derivative of SNOMED and spent millions and just
take it off them, no we can’t. We can ask them, but we can to a member, we can
to the United States, we can to the United Kingdom. Actually we’ve never had
to. As soon as somebody else says oh that would be really useful to me, people
want to share and want to make it available because actually it helps everybody
else because it’s a quid pro quo.
DR. FERRER: – hair you have with your responsibility, but Harry kind of
asked you a global question, asked a strategic way to follow in the U.S. with
regards to standards. My question to you is based on your experiences,
oftentimes we speak about core standards, building blocks that must be there
for you to build upon sort of long term, but it seems that your view of your
experience has been even though you recommended the interdependencies and
relationships that you are more focused in making the core framework, house, if
you will, interests built, as opposed to having to do all the standard work
harmonization. Is that a correct assumption? Or based on what you’ve seen in
the UK, how would you explain that?
DR. SEVERS: With my UK hat on, unique identify the NHS number, we have
worked very hard through the information standards to make that happen, because
it’s a fundamental standard, it was a requirement, and we pushed very hard for
give us the implementation evidence that you can do it and then we’ll approve
it, and the NHS has responded through NHS connecting for health. And there are
a set of project milestones in which the NHS number unique identifiers is
With my ISB hat on there are a number of other standards that I’m interested
in making sure come to fruition.
With my international hat on, I spend about a third of my time on
harmonization. The harmonization is tough. I can write the minutes, I can
produce documents, I can setup teleconferences, I can throw the weight of
– I can have enthusiastic colleagues, but if the other organization
doesn’t want to play I can’t make headway. This is a relationship between two
parties. And even if my members say, look, it’s absolutely fundamental, we have
a relationship with this standard, get it sorted, and I work all hours God
sends to try and do that, if the other organization doesn’t want to play
there’s nothing you can do. And that is where national pressure can start to
work in the international – if you’ve got an organization that’s made up
of national members, and every national member is saying you need to work with
this other organization, this other organization won’t play ball, they can
start using their national might together to influence that organization. But I
spend I’d say a third of my time is on trying to work with other organizations
to make it easier for the implementers and the users to get benefit from
standards, such that they’re not overpowered with the complexity of making two
standards work together.
MR. REYNOLDS: Judy, I understand you have a clarification, not a question.
DR. WARREN: I think this speaks to some of Blackford’s concerns. There are
strategies for getting changes into SNOMED, and one of those strategies is
belong to a working group, which is open to anybody who shows up, we’re always
willing. The other strategy is to go through the U.S. terminology entity that
is part of the country’s working group, and that is being housed in Betsy
Humphrey’s office at the National Library of Medicine. So you can contact her
about funneling in concerns, requests, etcetera, that needs to go forward up to
the international level.
DR. MIDDLETON: Even better, we have Howard Goldberg, one of my guys on the
MR. REYNOLDS: Professor Severs, outstanding. Thank you. We felt we were
fortunate before you spoke, that we had you here, and I think you lived up to
your billing, and I’ve heard a lot of good compliments.
MR. REYNOLDS: Next we’re going to talk about the health statistics for the
21st Century and get an update from Dr. Daniel Friedman, who
introduced himself shortly ago and was a previous member of NCVHS, so he knows
the drill here.
Then I understand we have Dr. Parrish on the phone; is that correct, Dr.
MS. JACKSON: He hasn’t plugged in yet.
MR. REYNOLDS: Dr. Friedman, please feel free to begin when you’re ready.
DR. FRIEDMAN: Thank you. I would like to be respectful of the Committee’s
time, so it would be helpful if you could give me an approximate timeframe for
how long you want me to chat.
MR. REYNOLDS: How long is Dr. Parrish going to chat?
DR. FRIEDMAN: He’s not going to be chatting at all.
MR. REYNOLDS: Okay. If you can talk between now, you get the next 25
minutes. And that will leave us some time to have some questions. I want to
make sure what you have done in the past, are doing now, and are going to tell
us, and we have a chance to ask some questions.
DR. FRIEDMAN: Very good, thank you. That will be more than sufficient.
I’m going to be speaking about three things. First of all, quickly reviewing
the 2002 report on Shaping a Health Statistics Vision for the
21st Century. Second, even more quickly, describing the project
that Gib Parrish and I have been asked to undertake. Then third, asking
Committee members to engage in a very brief exercise to review the
recommendations that the Committee itself put forth in 2002.
The 2002 report, Shaping a Health Statistics Vision for the
21st Century, was a product of three partners: the Committee
itself, NCHS, and the DHS Data Council. And Ed Sondik approached the Committee
in 1998 or 1999 and asked us to undertake this process. The report was written
by Gib Parrish, who at the time was a senior epidemiologist at CDC. Gib has
since left CDC, and he and I do most of our independent consulting together. It
was written by Ed Hunter, who at the time was Assistant Director of NCHS and is
now the Acting Director of the CDC Washington office, and by me.
The process that we undertook to develop that 2002 report spanned around
three years. It started in 1999, and on your left you will see a list of
roughly ten professional meetings, at which we presented the various stages of
the report and the report processes. At all these meetings we sought feedback
on what we were doing.
In 1999 the Committee asked the National Academy of Sciences to do a
workshop on the topic. The Committee commissioned a series of papers from
authors in the U.S. and Canada. In 1999 and 2000, the Committee undertook a
series of local discussion groups around the topics of what ended up in the
report, and also a series of expert discussion groups here in Washington.
In 2000, an interim report was developed. Based on that interim report there
were four regional public hearings held throughout the U.S. During those public
hearings and afterwards we formally solicited recommendations from the
participants in the vision process. Looking back on it, it did not seem like
such an exhausting process, but at the time it must have been pretty
Finally in 2002, the joint report of the Committee NCHS and the Data Council
was produced, and then included in that report but separately were the NCVHS
recommendations. In that report, and I just want to spend a couple of minutes
on definitions, because one of the things we found throughout this process is
that there was a dearth of the most basic definitions. So one of the things
that the Committee did, the partners did, was start by defining health
statistics. Now, this seems very simplistic, but again, it was something that
really had not occurred in the literature previously. And we defined health
statistics as numerical data that characterized the health of a population and
the influence and actions that affect the health of a population. So this is
population based data, not individual based data.
Second, we developed a concept and defined the HHHhealth Statistics
Enterprises, the infrastructure and activities necessary to produce health
statistics, with its mission being to efficiently provide timely, accurate, and
relevant information that can be used to improve the nation’s health. The
Health Statistics Enterprise collects, analyzes and disseminates data on the
health of populations and the factors that influence health, again, a community
health focus, a population health focus, not an individual health focus. The
Health Statistics Enterprise includes public and private organizations and
individuals that perform the processes of health statistics at all geopolitical
levels. And the Health Statistics Enterprise, and it is currently constituted
in this country, is highly decentralized.
There were two conceptual frameworks that were developed through the
process, and that really formed the backbone of the report and formed the
backbone of the recommendations. The first conceptual framework was a
conceptual framework of influences on the population’s health. We developed
this framework in order to help maintain a focus on needed data, and data that
is needed for the Health Statistics Enterprise. As we will see in a moment,
that conceptual framework, again, it focuses on communities, it focuses on the
context of communities, it delineates major influences on community and
population health, and it emphasizes the level and distribution of health.
This is a high level view of that framework. In the center you will see Gib
and I and Ed Hunt call this the Rings of Saturn diagram. In the center is the
populations health, surrounding it is community attributes, then context of
community attributes. Then finally those are all put in place in time.
Now, purposely this conceptual framework does not include any causal arrows.
This was not an oversight on the Committee’s part, it was not an oversight in
NCHS’s part, it was not an oversight on the Data Council’s part. We purposely
left out causal arrows because we felt that for the most part causal arrows are
really highly speculative. And there are certainly other schematics of
populations health where you can find many, many causal arrows. This is just a
drill down view of that conceptual framework, and you will see in the center
the populations health defined in terms of level of health, distribution of
health, disease, functional status, and well-being.
We recorded a great deal of information through the consultative vision
process, and one of the things we did was using that conceptual framework come
to the conclusion that the health statistics in this country has a great deal
of unrealized potential. The causes of that included our using often a narrow
definition of population health, focusing on individual health rather than
community health, focusing on disease rather than functional status and
well-being in addition to disease. Limited vertical and horizontal integration
of data, insufficient longitudinal and lifecycle data, and then finally a real
disparity between data availability on the one hand and data use on the other.
The results of this were essentially abundant data, but much too little
information, and particularly too little information on cultural and political
context, community attributes, appropriate measures of health, and geographic
and sub-population data.
In addition to that conceptual framework of the populations health, another
conceptual framework that was developed by the Committee at the time was of
something we called the health statistics cycle. This was developed because we
felt that it was important to have a conceptual framework to help maintain a
focus on needed components of a rational health statistics cycle. And the
health statistics cycle as we defined it is really very simplistic and is quite
obvious. The health statistics cycle focuses on defining user data needs,
identifying data sources, translating data into useful statistics for users,
and evaluating the extent to which user needs are met. I just want to repeat
that last, which is evaluating the extent to which user needs are met.
In the center of that cycle, as you’ll see, is something we called an
integrating hub. And integrating hub would perform, it does not now perform,
but would perform functions of coordination, stimulating collaborative efforts,
setting standards, resetting visions for health statistics, generating new
methods and approaches, and establishing frameworks and standards for privacy,
confidentiality and security for the health statistics enterprise.
And again, the Committee found based on the consultative process, that the
health statistics enterprise in the U.S. has a substantial unrealized
potential. And in particular, that existing enterprise-wide coordinating and
integrating efforts, this is back in 1999 to 2002, were largely ineffective
with insufficient connections between data sources, data collectors, and data
users, insufficient attention to developing consensus approaches in standards
to simplify collection protection and use, insufficient resources, lacking in
overall confidentiality and data sharing framework flexibility and
adaptability, and enterprise-wide form. And then finally, this all resulting in
an undue burden, a substantial burden, and possibly an necessary burden on the
data sources themselves.
After the consultative process after reviewing the information, the
Committee and its partners developed a series of ten guiding principles for
recommendations. The first of those principles is focused on the need for
health statistics enterprise-wide planning and coordination. We spent a
substantial amount of time on that in the report, and I would say of all the
principles and the recommendations, those relating to enterprise-wide planning
and coordination were probably I don’t know if I would say the least popular,
but probably the most controversial.
The second guiding principle focused on the need for broad collaboration
among data users, producers and suppliers at all governmental levels. Third,
need for rigorous policies and procedures for protecting privacy,
confidentiality, and security. Fourth, flexibility to identify and address
emergent health issues and needs. Fifth, use of data standards. Sixth,
sufficient detail at different levels of aggregation, different governmental
and geographical levels. Seven, the need for integrated streamlined data
collection for multiple purposes. Eight, timely production of valid and
reliable health statistics. Nine, appropriate access to and ease of use of
health statistics. And ten, continuous evaluation.
NCVHS developed the recommendations based on those ten principles, and based
on the consultative process. And just to quickly repeat, the recommendations
were developed through an iterative process with participants in that three
year long process. We developed the recommendations through formal
consultations and formal solicitation of recommendations, and then finally
having the recommendations and the draft recommendations reviewed by the
participants. So it was I wouldn’t say it was a comprehensive process, but it
was a somewhat exhaustive process.
There were 30 recommendations that were produced for achieving the 10
guiding principles. Then there were 6 recommendations produced for moving
toward adoption of overarching conceptual frameworks. And the recommendations
related to the Health Statistics Enterprise structure itself, dataset
evaluation, and finally the research agenda. That is a very high level of the
2002 joint report, and the Committee recommendations flowing from that report.
Gib and I were asked to undertake a modest project to reconsider that
report, with the project focusing on assessing progress since 2002 on the
vision recommendations, and focusing on updating, reconsidering, and revising
the vision in light of national and international developments since 2002. The
first phase of this modest project has been funded, and we’re very pleased by
that, and it includes first of all a literature review, second, a series of
expert interviews, and third, iterative review of findings with the Committee
and with Committee staff.
The timeframe for this first phase is first of all in late May coming up
with a series of suggestions for updates to the report. In June, providing to
the Committee a preliminary assessment of progress since 2002 on those
recommendations, and discussing with the Committee possible changes to the
report. Then in September, a draft working paper on expected change to the
vision and recommendations for any future workshops or public hearings.
Actually producing any kind of an updated or changed report itself is not part
of this phase of the project.
One of the things I discussed with Don Steinwach and Marjorie and Debbie,
was trying to obtain some guidance from the Committee, hopefully today, on
which recommendations Gib and I should focus on. This would be very helpful to
us, because there are 36 recommendations there, as somebody said or implied in
the Populations Subcommittee this morning some of them are a bit dusty by now.
And we would like essentially off the cuff simplistic reaction from the
Committee members, and there is a little rating sheet, high, medium and low,
for each of the 36 recommendations that either has been passed out or will be
passed out, and we would like you for each of those recommendations rate each
of them in terms of your perception of continued importance. And the kinds of
criteria, and we’re not dictating criteria, but the kinds of criteria that we
would think of would include current relevance, intersection with e-health and
Initially we had this, or at least I suggested this very complicated
framework, three or four different dimensions for rating each recommendation.
My friend and colleague Gib, who is very good at reducing things to their
essence, basically said let’s keep it simple, let them use whatever criteria
they want, let’s just have them rate each as to high, medium, and low. Based on
those ratings and based on continued discussion with the Populations
Subcommittee and Marjorie and Debbie, we can then proceed to focus.
So thank you very much.
MR. REYNOLDS: So you want this filled out right now, is that what you’re
DR. FRIEDMAN: That would be terrific.
MR. REYNOLDS: What I would like to do is see if we can be letting people ask
questions as we’re doing that, because we’ll double-duty here.
MS. GREENBERG: Let me just say at the bottom of the third page it says your
name. If you don’t want to put your name, I don’t know why but you might not
want to, at least put whether you are a member or staff. The staff or liaisons
or whatever are more than welcome to fill this out also, but I think if you’re
not putting your name we want to know whether you are a member or not so that
we can at least separate that information out.
MR. REYNOLDS: What if we don’t know enough about some of the constructs to
actually vote on one item?
MS. GREENBERG: Leave it empty, I guess.
MR. REYNOLDS: That’s fine. But you’re asking us to do something, I want to
make sure I understand.
DR. FRIEDMAN: That’s fine.
MR. REYNOLDS: Okay. Jim, and then I got Carol, Leslie.
DR. FRANCIS: When I fill this out one thing I can be thinking about is how
important the criterion is. The other thing I could be thinking about is how
important it is to pay attention to it right now because I think it’s
inadequate. And I want to know which you want me to be thinking about. Because
something could be really high priority, but I could think that what got said
in 2002 was fine.
DR. FRIEDMAN: That’s an excellent question. What we’re looking for from you
is guidance about what we should spend our time on over the next several
MS. GREENBERG: You might put that slide back. But I think current relevance,
intersection with e-health, and long term importance are I think guides.
DR. FRIEDMAN: Any other notes you want to put down, I feel like this is
getting grades by my students, any other notes you want to put down feel free.
MR. REYNOLDS: Jim.
MR. SCANLON: I guess the only thing is I would, you know, recommendations
flow from a process of consultation, assessment, clear goals, assumptions, and
participation by all stakeholders. And then you have it, so you diagnose, you
assess, you make recommendations. And I would hope that these summary ratings
here are not going to be used in determinative, I think it is a good idea to
revisit in the light of current circumstances. But I wouldn’t want to see, I
mean we’re basically starting with the previous recommendations, and I hope
that’s not determinative in terms of they would be – recommendations as
well, and I don’t think most of the Committee members are familiar with that
either. I wonder if there’s not a more – in other words, this is fine for
sort of qualitative impression for you and Gib, but if you really want to know
what folks would like to see prioritized or maybe new things in the future, I
think you just need more consultation than the rating. I guess that’s my only
point. And you need to talk to stakeholders.
So I think this is meant to be a qualitative kind of input into the process,
but it shouldn’t be formal views of the NCVHS in terms of a vote. This is
purely qualitative –
MS. GREENBERG: No, I mean, these recommendations came out of a deliberative
process that you mentioned, Jim, but it was what, eight years ago, seven years
MR. SCANLON: That’s my point.
MS. GREENBERG: Because we don’t either have the resources or probably the
time or even the stomach to completely go back to every single one of these,
what Gib and Dan want to find out is – because we’ve been talking about
this vision and different people are more familiar with it than others, but for
years now is which of these you really think, you know, I was going to say we
have limited time to do that kind of deliberative process, and really the real
consultation, though there’s some consultation in this first phase, the real
consultation and maybe the workshops or hearings or whatever would be in the
second phase that we haven’t yet funded. But now where do you think they should
sort of focus their energies, and this is I think going to be impressionistic,
MR. SCANLON: And not the specific recommendation, I hope. It’s thematic,
again, you want to consult with the policy community and others as well, it’s
eight years later. So I guess with the understanding that it’s meant to be
thematic and qualitative, I mean you’re not literally asking people to rank the
old recommendations, why would we do that?
MR. REYNOLDS: I guess from a process standpoint, I would consider this
exercise that we have been asked to do as nothing further than input to the
Populations Committee to do their normal deliberation as they would do with
anything, and make a recommendation and bring it back.
So, Don, if you are requesting.
DR. STEINWACHS: I think what we thought was this was a way to get the
Committee quickly sort of thinking about what are the recommendations, what is
the content, and just to get some initial feedback. So I hope it is stimulative
to the members as well as it is input to the Populations Subcommittee.
MR. REYNOLDS: Right, but I want again back to process. And that is what I
think – this is going through the end to your guys, and then you’ll bring
it back up through the normal process. Okay. So with that, I have Carol,
Leslie, Mike, or Leslie’s done, Carol, Mike and then Justine. Justine.
DR. CARR: I just have a suggestion, that if we each put our top five
thematic important areas as high and then go pick the ones that are – and
then also maybe the five lowest and everything else, I think maybe that would
MR. REYNOLDS: With that please continue filling out your process here. While
we’re waiting for everybody to finish, I think with all the activities that
have occurred over the last month as far as how thing are going to be changing
in health IT and others, this is a really good time to take a look at how this
all comes together, so glad to have you working on it for us.
DR. FRIEDMAN: Thank you.
MR. REYNOLDS: Okay. When you are finished, are there any other questions for
Dr. Friedman before we move on? Okay. Then Don and Bill, we’ll assume you have
the ball working with Marjorie and the two doctors and Debbie, and we’ll be
anxious to continue to hear from you.
With that, let’s move onto the NCHS and BSC Updates. Jim, are you on the
phone? Jim Lepkowski.
DR. SCANLON: We were depending a bit on Jim.
MR. REYNOLDS: I can make a few comments, if you want to wait a minute until
he might get on. Bill, I have just a few regular comments, if you are counting
on him we’ll give him a minute more.
DR. SCANLON: Fine.
DR. STEINWACHS: Virginia said it’s a technical problem, he can’t un-mute his
MR. REYNOLDS: Okay, Bill, so you have the ball, and you get to run with it.
DR. SCANLON: What we were going to do, and Garland can help me on this,
which is one issue that we haven’t talked about before in any detail that BSC
was interested in bringing before this Committee was the use of electronic
health records in the vital statistics system. It is a topic that they
discussed at their last meeting, and the reason why I’m dependent upon Jim and
Garland to talk about this is that I wasn’t there for that discussion, so I
can’t report on it.
So, Garland, you may want to talk about this issue and then we can return to
MR. LAND: I wasn’t at the meeting either, but I can talk a little bit about
what I know about it.
I know the National Center and some of my staff have been involved in and
looking at this whole area, and developing a main analysis model, a functional
analysis of how to import medical record data into electronic birth and death
I think there’s a lot of potential, but I think there are a lot of issues
involved in that that really – I’m cautious, let me just say that. I can
give you several different reasons. Presently what happens is take the birth
certificate, the information comes from the medical chart, which can be both
that was from the prenatal record or from the record in the hospital. And also
it is from information that comes from talking to the other, so all the
information isn’t going to be obtained otherwise you would have to interview
the mother to get some of the information. So it really comes from three
different sources, so you have to take that into account.
I think what they’re coming up with and realizing is it’s highly unlikely
that all information can be transported from the electronic medical record into
birth certificate. There are legal reasons, there might be what’s in the
medical record may not be what you want the name of the individual to be on the
birth certificate. So you have to accept there are some legal ramifications of
that. There are issues of – that isn’t in the medical record, so as I
understand there is especially the fact that this thing might not be a 100
I had a lot of concerns about how do you edit the data. Right now since the
data is being entered manually into an electronic system, you can create online
edits as you are going along. But if you are pulling data in from someplace
else, once those edits are ticked off then it means going back and trying to
find where that information came from, and if somebody isn’t very careful they
will just go ahead and accept the data rather than now there is some manual
review going on. Since the birth certificate is not only a legal document it’s
also a statistical document, the editing function is very important, and it
could result in having data this is different in the electronic medical record
than what is in the birth certificate. And those issues have to be resolved. Is
that okay, is it not okay, and what are the legal ramifications of doing that.
On the death certificate side you don’t have causes of death typically in
the medical record, in the standard way that causes of death are entered onto a
death certificate. So that would mean changing the whole format of the medical
record to align it the way causes of death are entered. That is a whole other
process of thinking about what the medical record is suitable for, and is that
the best way to get it into the death certificate.
There have been several state experiments along these lines and they have
not been successful. And I can give you a lot of different examples of this. I
think that’s another cautionary note because this has been tried. Maybe it
hasn’t been tried to the degree of setting the standards and everything like
that. What’s going on now is trying to set up some HL7 standards for the
transmission of what’s in the medical record to the birth and death
For that reason I think it’s a process that needs to go forward and evaluate
as we’re going along, and certainly it’s going to take a couple of years down
the road just because it’s a HL7 standard setting process itself it takes time.
We just have to be cautious, and the substantive evaluation measures has to be
moved down this path, in my personal opinion.
MR. REYNOLDS: Jim, then Marjorie.
MR. SCANLON: Garland, how does this – whether or not the medical record is
the source of generating the vital statistics certificate, it doesn’t have to
be, by the way, as you know, you are still supporting the concept of electronic
MR. LAND: Oh, absolutely.
MR. SCANLON: So whether the medical record is the source or not is not so
much, you know, you can’t forward it particularly, it may not just be a fit.
MR. LAND: Yes.
MR. SCANLON: But secondly, how does this approach fit into the whole Patriot
Act requirements for security in national systems for electronic vital
statistics and so on? It envisions some sort of a system where every state
could, a standardized system where every state could check with every other
state, and that sort of thing, it doesn’t fit at all maybe.
MR. LAND: I’m not sure that the concept of electronic medical records going
to the birth certificates if it fits into the intelligence format, or Real ID
Act. Certainly there is a national need to secure birth certificates better
than what they are now, because the birth certificate serves as the breeder
document to get a driver’s license, to get a passport, and there is a lot of
fraud going on now in terms of birth certificates, unfortunately.
So we do recognize that there has to be more attention paid to that. Some of
that is through electronic means, and we are implementing an electronic system
called “EVE” through the Department of Homeland Security funding that
we just got actually this last week, that will validate a birth certificate. So
if somebody wants to know if this is really a legitimate birth certificate, we
have an electronic way of doing that now, and we’ll be implementing this system
over the next couple of years.
There is also a lot of standards that need to be developed in states around
the security of the paper when you’re issuing a birth certificate, in terms of
how internal standards within health departments and how they operate. And
we’re in the process of doing that, but we don’t have any teeth, that’s where
the regulations if they were ever promulgated would give some teeth to that
process to ensure that states are conforming to certain standards.
MR. REYNOLDS: Marjorie.
MS. GREENBERG: I just wanted to add that I think Garland has raised a number
of very legitimate issues which I don’t disagree with about what some of the
challenges will be in trying to use electronic health records for either birth
or death. I think to me it’s quite clear that they have a use, but that doesn’t
mean that it’s just whole scale going in. I mean if there is an electronic
prenatal record it seems that could very well inform the birth record. That
doesn’t mean that it would replace it by any means, or that it would meet all
the needs, unless it was engineered to do so. And that is the point that I’m
But the same thing, if there were a longitudinal health record for a patient
you think about if that were used by the person certifying the cause of death
that would be about by many orders of magnitude more information than most
people have when they certify causes of death. So that doesn’t mean the record
itself for that last hospitalization or whatever would tell you what the
underlying cause of death was, because in most cases it wouldn’t. I think the
use of them is there.
But the point I really want to make is that if we don’t design our
electronic health record systems to meet this critical need, which is really
what I call the bookends of your electronic medical records, your birth and
your death, with everything else being sort of in between, and don’t design our
electronic birth and death registration to benefit from those electronic health
records, then it will be very hard to retrofit.
So I think this has to be something that we’re looking at upfront rather
than later. Ed may want to add to that.
DR. SONDIK: The way I see it, we’re talking about something that is a system
that is fundamental to what we know about health, key measures. A recent IOM
report that I mentioned in the Population Health Group this morning came up
with twenty measures that are going to go into a national system of reporting
on a variety of aspects of life in America. Health is a very important part of
that. You look at the twenty, about half of those, probably about 40 percent of
those relate to the vital statistics.
We currently are not – ‘we’ being NCHS – we are only
minimally investing in that system to build a system, talking about the
21st Century, that really is 21st Century. There are a
number of issues that need to be dealt with to how to make the system more
efficient, how to move the data more rapidly, how to build timeliness, how to
increase the facility to do quality control, how to maintain security, how to
notify Social Security more rapidly, how to get information to a form where it
can be used by other agencies within the states and other agencies in the
federal government that matter, how to make it more accessible for research.
And we need to invest in it in order to do that, and we at NCHS have just
not had the resources to do that. At our last BSC, as we have on several other
meetings before that, we talked about where we were in the state of this, and
that we were talking about budget, and that for fiscal ’09 we had to make some
cutbacks in order to be able to – because under fiscal ’09 as of today
we’re on a continuing resolution, and in order to be able to meet that budget
we had to cut back on programs, and we did, we have done that in a number of
areas including the health interview survey.
One area that I felt strongly that I didn’t want to cut back was in vital
statistics, and at the same time I felt very strongly that we need to invest in
it for the future. We had proposed that we would in order to do that cut back
on the amount of data on the natality side that we were collecting, thereby
paying the states less, and use the difference between what we would have
normally paid and what we were paying to invest in the system for an electronic
vital statistics system, and also related to quality control. We had plans to
do that as of January. We in meetings with Garland, with NASAS and with others
and other stakeholders in this, they said they felt it was important that we
hold off on that, that the budget situation may change. I felt that was a
prudent thing to do. So we haven’t done that, we’re still in the process of
about I guess beginning negotiations related to our contracts with the states
this year, so we will see where we go.
I guess my very strong feeling is that we need to invest in this system.
There is potential for the Stimulus funds to be used for this. The constraints
on those funds, that they’re time limited, they’re not continuing funds, they
are not going to add to the base appropriation of any agency, that they relate
to jobs and will really be an economic stimulus, and of course that they have a
health impact, I think really all applies to this particularly given the fact
that one of the areas in the Stimulus funds is electronic health records. There
is no doubt, I completely understand what Garland is saying, I think as
Marjorie said we need to find a way for these to actually work together. It may
be that in some areas it will be more flexible than it is in others, depending
on what’s in the record.
I think Garland’s point is very well taken, and I honestly don’t see that
somehow the information will come out of the medical record and into the vital
statistics system. There are things that are just going to have to be added or
perhaps edited at that time, because of the special nature of the vital
But the timeliness and the quality of the data is of national importance. I
don’t have to get on a soapbox to say probably the two most fundamental
measurements about how healthy we are in the U.S. come from the vital
statistics system, and it’s important that that be of the highest quality that
it can be, and that we are able to link that information as much as we can to
other sources of information [microphone interference], on prevention and
wellness, actions, and status of the population, linking all of that to the
vital statistics. That I would see would require a considerable investment, but
a very feasible investment. And that’s what we talked about at the BSC.
MR. REYNOLDS: Larry then Paul, then we’ll break for lunch.
DR. GREEN: I want to allow Bill to fix what I’m about to say and sort of
make it contributory.
My take home from the Populations Subcommittee meeting this morning, was
really a spring off of Garland flagging yesterday at the end of our meeting,
the issues afflicting our vital statistics system. And my take home from our
discussion this morning was that we recognize the following: number one, the
state of vital statistics health system in the country is in jeopardy, it is
not healthy, it is not a healthy system, and it’s usually important.
A second thing I thought was a take home lesson was the Stimulus Bill
specifically makes resurrecting this issue timely, perhaps for several reasons.
One of which was there is going to be a great need for accountability and it is
stunningly difficult to think about fundamental accountability with no vital
statistics. So it connects to an important political issue at the moment. Also,
it’s been well-studied, it’s well-developed, there is a great deal of knowledge
about what needs to happen with the vital statistics system yesterday and as
soon as possible, and it could repair work and further development could be
done very quickly if there were action on this.
Then the third issue that I feel like is sort of just confusing this right
now is Populations Committee recognizes that Dr. Kolodner’s challenges and
issues, and the distribution of the EHR, and the development of the national
health information system is presenting us with uncommon, unusual, and maybe
unprecedented opportunities to enhance and improve the vital statistics
We were thinking we ought to do something like now, like this week. Now,
Bill, would you please fix that so I don’t do any damage.
DR. SCANLON: I don’t think you did any damage at all. I was going to end by
saying that we planned this afternoon if the opportunity is available to try to
bring forth a letter that would talk about which addresses these very issues.
And I think of it as going beyond the vital statistics system because I think
while that is essential it’s not enough. I mean there is this issue that we
could not have the adequate information to be able to assess either the health
or the healthcare provided in this country. And that was a problem that existed
five years ago, and it’s gotten worse instead of better over that time.
At the same time I don’t want to say that we need to hold sort of our
investments in HIT hostage until we figure out how to make the data flow for
assessing things at a populations level or a national level, because that is
going to be something that’s going to be a real detriment to progress. But we
have to think about as they’re moving forward how we make incremental changes
that are going to be helpful, how do we imaginatively find ways to take
advantage of what they’re doing regardless. But it’s a very broad issue.
What we’re focusing on today in potentially a letter this afternoon is the
issue of resources, because you can’t think about trying to resolve these
problems unless you’ve got the resources to invest in contemplating the issue.
MR. REYNOLDS: Do you have a draft letter now?
DR. SCANLON: Hopefully by 1:00 o’clock we will have a draft letter. We have
started on draft letters.
MR. REYNOLDS: What I would like you to do is as soon as you have it we would
pass it out while people are – while we’re going through the first session
people could at least get a sense of what it is so that we could be very
precise and expeditious at the 3:30 timeframe to try to at least have everybody
take a good look at it.
We have Paul, a question, and Carol, you still have a question?
DR. TANG: I just had another refrigerator magnet. Yesterday was, ‘Do
unto the data of others as you would have them do to your data.’ And I think
Garland said this, ‘Thou shalt go to the source that has the most vested
interest in its accuracy.’ I think you said to look at SSA to pick up some
birth information, perhaps, I don’t know.
MR. LAND: We provide information to SSA.
DR. TANG: Okay. I mean the other way to look at it perhaps is to pick up
some birth and death information, because they have a vested interest in
getting that information and getting it accurately, unlike unfortunately the
MR. LAND: And we do that, too, we provide death information to SSA.
DR. TANG: Okay. So you know that there was some Stimulus money to use that
for the VA disability as part of NHIN, so in other words there was just
something there, and it seems like this kind of imaginative use of other
MR. REYNOLDS: Carol, did you have a question? Okay, with that we’ll break
for lunch. See everybody back here at 1:00 o’clock.
A F T E R N O O N S E S S I O N (1:00 P.M.)
MR. REYNOLDS: If everybody could sit down please and we can get started. As
we have said before, and I think Larry touched on it again yesterday, this idea
of continually trying to educate the entire group on the types of things that
are out there. I appreciate that Justine and Paul and others, along with Matt,
that would be what would follow the word “along.”
And, Matt, I was trying to thank you and Justine interrupted me again, so
thank you for what you did, putting this together for us.
We have another Matt, who is called Matthew Holt, who is going to be the
person who is going to kick it off for us. Matthew, if you could tell us a
little bit about yourself.
Then Dr. Spranca, Dr. Halamka, and Dr. Mandl are going to follow that
MR. QUINN: Dr. Mandl, because of a family emergency will not be here today.
MR. REYNOLDS: So, Matthew, please.
MR. HOLT: Thank you very much, I’m on the mike, I hope.
I was down for a half hour, but I might have to go a little bit longer
because of the other cancellation. But those of you who have heard me before
know that’s not usually a problem. I know there’s also more time for
discussion, I would encourage you to stop me and yell, complain, interrupt,
raise points at any time during this. A couple of you in the room have known me
for quite a while, and I won’t belabor my background for this. But I’m a native
Californian, as you can tell by the accent. I’ve been living in California
since 1989, and signed up to be an American, therefore I’m a real one for seven
or eight years back now. Been living in the U.S. for about twenty years. I did
some graduate work at Stanford in health services research. Went to work for
the Institute of the Future in Menlo Park doing healthcare and healthcare IT.
I’ve worked with various people in the room, Blackford and Matt, and he and I
go back to the days at Medical Logic, maybe even before that, and Stanford as
well, didn’t even think about that. Since then I’ve wandered around and spent a
bit of time in the startup doing personal health records where I encountered
the other Matthew, Matt Quinn. As no good deed goes unpunished, because I gave
him a job in 2000 he made me come here today.
Probably the thing I’m best known for now best, is when I returned from the
post-startup lying on the beach in Asia phase that was very popular in 2002, I
had been running a travel blog and I thought somebody should be writing one
about healthcare, and I looked around and nobody was, so I started one. So I
own and operate the Healthcare blog, which now seems to have become a venue
mostly for people to fight about healthcare IT standards. It also covers
various other issues around healthcare, the spectrum healthcare business
market, policy, and what have you. And luckily because it’s well-read now I
don’t have to write on it everyday. There was one period of time when I did,
and that was hard work, I tell you.
As part of that I have always been following healthcare IT, much of my work
at Institute for the Future and elsewhere was around the adoption of EMRs and
the impact of the internet on healthcare. Around three or four years ago there
was a substantive argument which was going to change in the way that
information technology was being used in Silicon Valley, the emergence of what
has now been termed “Web 2.0,” there’s an argument about whether
that’s a real change or not. But there were enough people doing similar types
of things in healthcare that they started coming to me and saying hey you
should do something about this. But by accident I’m in the conference business,
and I hold a conference twice a year for Health 2.0, which I didn’t start the
movement from the conference, so full disclosure, some of the people I’m going
to talk to you about have sponsored the conference, or appeared there or what
have you, I don’t know if my conflict of interest statement will cover that,
but I’ll be happy to explain any of that later.
I’m going to spend about half the time, I’ll spend about half an hour
talking about Web 2.0 and Health 2.0, the general categories, and then I’m
going to narrow it down and sort of talk about the very, very early stage
issues of using what’s going on online, primarily in as a venue for data
collection and research, something that is your main mission, and I’ll be happy
to talk about that. I would warn you that is in incredibly early stages, there
is really conjecture and a bit of activity you’ve talked about, there’s no real
data about that. It’s something that we’ll see how it develops.
With that let’s kickoff to the PowerPoint. I will start with this one
because I’m best known as a blogger. This was done, this chart is from an
august organization, the BBC, that’s just a poll of people’s trust, and if you
look down at the bottom the internet blog is kind of lost. I, for one, am going
to be in Washington, D.C., right, and you people care about politics, and no
one cares about politics as you care about negatives here, so negative ratings.
So I thought let’s look at negative ratings, both actually have lower negative
ratings than the opinion of friends, family, and colleagues.
MR. HOLT: But in case you ever have any troubles at work on this Committee
and you perhaps value the opinions of colleagues and friends, perhaps you
should point them towards the blog.
Let’s dive into this Web 2.0 thing. It’s probably old hat for many of you,
but I do go around the country talking to people in healthcare IT a lot, and it
is quite surprising how little the sort of general movement in the Silicon
Valley IT world has moved, has been noticing the world of healthcare IT, which
is always a sort of separate island in IT. It also plays the other way, because
people from other IT world come into healthcare and assume they can change it
easily and find it not quite as easy as all that.
Let’s just very briefly run through this. You have probably seen this chart
with some of these terms like folksonomy, social software, Web 2.0, the long
tail, Open APIs and all that kind of stuff, AJAX. This is an ecosystem shot
which looks at, this is a couple of years old, estimated market value, these
numbers probably were a lot higher six months ago and now in terms of the
market value of some of these, but things like Wikipedia, Facebook, YouTube,
all these have had a significant impact in the way that people behave online.
And I also want to point out the speed at which this change has happened. I
happen to know this because the guy who is four doors down from me in my
apartment block is, I think he’s the guy who was, because there was a huge
renovation project and later I found out when I met him I found out he was one
of the cofounders of YouTube, and I figured that’s where the money came for the
renovation project. It is such an interesting thing, do you know when YouTube
was founded, the date, 2006. How quickly it’s been that YouTube has become this
phenomenon just out of nowhere.
So I think you will appreciate we’re in an area where the speed has really
happened very quickly. Blackford in his mind founded it in 2000, he wishes he
MR. HOLT: So very briefly, the overview of this Web 2.0 change. If you go
back to the mid-90s when we were first looking at the World Wide Web, and the
internet went from being this sort of weird science project in the general
collectiveness, it was really about publishing online and the ease of
publishing. And the big key is around content management, web masters,
regulating what was going on, and the dom letters were “e” and then
“I” and you put a dash in between the names so it became e-Health, or
e dash health or whatever, now eHealth 2.0 in some places in Europe.
Somewhere around 2004/2005, it really came into public consciousness since
the last couple of years, this concept that is now called “Web 2.0.”
And I don’t want to go overboard whether it’s the right nomenclature, and
whether or not it’s really a different thing, because a lot of the practices
that were going on in Web 2.0 were there in the early days of the internet and
even before in some of these online communities. But what you certainly have
seen is an enormous uptake in the level of activity of all these things, like
blogs, even amongst micro-bloggers. Is anybody twitting in the back there? The
twitter community is – Blackford’s doing it right now.
Obviously Wikis, I mean the power of Wikipedia, again, only created in 2002
I believe, if not yet it will be the dominant web address on the planet in
terms of just utilization of Wikipedia, an entirely user-generated content
system. The growth of these things has been really quite phenomenal.
What happens online is somewhat different. The other key difference is this
is not a phenomenon which has been created by individuals publishing the same
stuff they were doing offline before on paper. This is everybody getting to put
their own stuff up, and it brings on whole new rules of guidance, and then a
whole bunch of new ways that people put terminologies to these things. If you
look most web companies they don’t have a vowel in the name usually.
The main guy behind the concept of Web 2.0, putting names to this, is Tim
O’Reilly. I’m not going to read through all of these, but he has sort of a
pretty well known seven core competencies. I do want to pick up a couple of
these, which I think are very important. The second one here is control of a
unique hard to recreate data sources that get richer as more people use them.
Certainly a lot of what’s going on here is people creating more and more data
sources, and as you add more and more users to the network and to those data
sources the data in them becomes more complex, and in some ways it’s harder to
compete with in some businesses. In some ways that’s also done before in
The other one I think is particularly important to healthcare is the fifth
one here, which is leveraging the long tail through customer’s health service.
I’m not going to define the long tail. I’m looking around, I’m hoping this
august body has some concept of what we’re talking about here. But that may not
be the case, all right. [Laughing] The concept of the long tail here is that a
physical organization like a bookstore has to actually decide what it’s going
to put on the shelves, and they have a number of ways they can, and they’re
limited by shelf space and by having to create enough copies of Harry Potter to
pay the clerk and to pay the heating and the lighting bills and what have you,
so there is some constraints on them. There is no constraint on Amazon creating
a catalogue and then figuring out how to get you the rare book that you might
want to buy, which only happens every – which only gets sold a couple of
times a month.
The concept of the long tail is you get down into that long part of the tail
for those books that only sell rarely in a virtual world, in fact the total
number of those books sold is a fairly significant chunk of overall books sold,
like 20 to 40 percent depending on which market you are in. In other words,
there is a lot going on that doesn’t make the sort of first big chunk of
activity in any given market that is now enabled by essentially virtual online
activity. That is the concept behind the long tail. I hope that makes some
sense, because you can move that to healthcare very easily, particularly
through diagnoses of symptoms, you know, there are only a certain number of
symptoms, but we all know there’s an incredibly number of relatively rare
diseases which most doctors don’t see on a regular basis. But if you add up the
number of people with those disease conditions, eventually there are actually
quite a lot of them and enough to sustain certainly online communities.
I think a lot of what is going on in terms of research around especially the
rarer diseases, but also more common well-known disease states become revealed
to actually a collection of much rarer diseases, I think there’s a lot to play
here. I would bear that concept in mind as to where we head.
I’m going to power very quickly through what the hell is Health 2.0, and you
may still be wondering. I changed this chart because I started about two years
ago with this chart, which is my sort of best guess at the constituent parts.
Personalized search, makes the search much more personalized to the user
content search, is one part. Presenting data is integrated in a useful way to
consumers, and to all users including clinicians, because clinicians need data
and information presented to them in a better way as well. The third one is
particularly important for what we’re going to talk about later, which is
communities capturing the accumulated knowledge of patients and caregivers and
also clinicians. And then finally, creating more intelligent tools to deliver
I don’t think we need to talk about it too much today, but there’s a subtext
here about the locus of control of healthcare, changing from the system the
establishment providers to patients, and obviously that’s a long topic which is
not in the purview for today, but it is certainly a big part behind this
I’m a simple consultant, so here’s some simple charts, give you the rough
searches versus social networks versus tools are, searches carry the dominant
player. A small company in Mountain View, California, which figured out how to
make a few bucks out a of search, and the estimates of health search are
roughly they’re making a billion dollars in advertising associated with health
terms and such. Social networks and tools as well are growing. And what you’re
seeing in that little intersection between these three of them is where the
really interesting stuff is starting to happen.
There are two other points worth making, are in healthcare we have a lot of
transactions, an awful lot of transactions, and there’s a lot of content, and
we have done a really lousy job of putting the two together. One of the things
going on here is that using these types of tools we’re hoping to bridge that
gap. From a policy level foundation we’ve made a start putting these two things
together, but I think you’ll admit that’s a long way to go for everybody.
Then just a very quick run-through of some examples. This is a Google search
for diabetes. This is a search from a specialized company called
“Healia” which is based in Seattle now owned by the publishing
company called “Meredith.” You can see on the left-hand side of the
screen they have figured out ways to start filtering the search terms so the
results come back to you in a way that may be applicable for who you are as a
person. Very early stages of personalization search is going on here. A number
of other companies are taking a similar approach, Healthline, Praxion, a number
of other companies are doing this kind of thing.
I think you all see Google, and Microsoft setting up a pretty decent job as
well here, they purchased a company called MedHealth, but they purchased the
first company which does this kind of personalization getting search terms to
be more relevant. So some personalization searches going on.
Communities online are becoming more intelligent. There have been email list
serves for many years. This is a chart of an email list serve from the ovarian
cancer list serve. ACOR, which is the online cancer resource, I think it has
about 160 email list serves with people just communicating using email for
cancer information, very, very powerful patient community. But you are seeing
significant changes in how those tools for communities are changing.
Here’s one, I’ll go back to the this site again, it’s a screen shot from
patients like me, which in some ways is kind of the shining star from Health
2.0 sites. They collect very detailed information on their users. It’s a
community for people with very severe chronic diseases, ALS, Parkinson’s
disease, and some others like that. Just to give you an indication, it’s a site
in which you are exchanging all kinds of information, including recording
essentially your personal health record, although it’s not called that, a lot
of their day to day activities which would not be captured in other typical
Thirdly, in terms of content, this is a sort of in the name of replacing
something old with something new, this is some information sheet from the FDA
that is on the FDA website about a particular drug. But what would be really
helpful is if you could actually create a drug information system that really
reflected what you the patient were taking, and what symptoms you have, and see
if there might be some drug interactions that really matter to you, maybe in
for the long tail, maybe buried deep down in the text of a modem or a First
Databank monograph. In fact, it’s a company that I’m an advisor, who I begged
them to be an advisor, called Enhanced Medical Decisions, has some early
prototype of what called DoubleCheckMD. They have both a drug-based and a
treatment-based program using natural language recognition to come back and
bring simpler terms information that was buried in e-clinical monographs, so
presenting content in a much more interesting way.
I think this type of activity, these three things I’ve just shown you,
you’re going to see a lot more of that in the coming years as more and more
people throw computing power at content and information.
Very briefly, this is just some very brief outline data from the study that
Edelman, the PR agency did. What it is telling you is about 22 percent of the
population, according to their metrics, is becoming what they call “health
influentials,” which is sort of marketing speak for a group of consumers
who are not only looking at healthcare information online, but are also
starting to disseminate that kind of information to others. And I think you are
seeing the sort of evolution of a very smart group of consumers who in some
ways are running ahead of their providers.
By the way, they are actually like health expert blogs, I’m not sure the
health expert blog actually does get fairly good credibility if it’s coupled
with sort of getting good expertise behind it.
Just to try and give you a sense of where I think we’re heading here. I
think the issue for these online communities is to try to use these two
technologies. One is to comb, I don’t use that thing on the left much, but use
the one on the right, is to comb through all that content online and trying to
essentially get it to rise to the top, get the really important stuff to bobble
to the top, like on the top of a pint of Guinness, where the bubbles come up.
And I think that’s through the simpler way that we’ve seen these changes.
I think in terms of tools, when something becomes like not only a tool
rather than just a content source or a search database result, it’s when it
comes back to the consumer or the user and it’s much more personalized, it
gives them some analysis, it actually supports the decision they make and
actually helps them enable a transaction. I think several of the sort of
implementations that we’re seeing connecting the EMRs and PHRs and some of
these other tools with content around healthcare, Palo Alto Medical Foundation
is one of them, of course, are starting to resemble real tools for these
consumers. I think we’re going to see a lot more of that coming up.
I have changed just in the last few months my slides slightly. This is still
the same search, social networks and tools bubble chart. You saw the same
transaction data and content data out there, but there is clearly on behalf of
some provider groups in some health plans, I say the government’s a little bit
behind this, but is moving in this direction, an attempt to try and sort of put
these things together and to use transaction data in particular to support
search, search networks, and user tools by consumers. One example of this would
be Aetna is using its data from its claims database using the active health
management system to actually push the – consumers data about their
interactions day to day, and actually changing the way their webpage appears to
consumers based on the type of information about their conditions and the drugs
they’re taking or other things like that.
I think we’re starting to see the early stages of this whole thing sort of
mushing together, and the data that is out there is starting slowly to come
into support the information content seeking, which is really then a very
different place to now. But again, this is very early days.
I would say we’re seeing some slight changes to Health 2.0, even though it’s
early days. We are still seeing personalized search, but I think that search is
carrying a lot more about the user experience than it was. There is a lot of
experimentation going on how you present information content back to people in
a very useful way for them. This is sort of the UI, the user experience part of
that is user interface and user experience, I think it’s becoming increasingly
important, and I think that’s something that’s going on by the way on the EMR
side as well. There are some interesting new EMR companies who are looking at
things like Facebook and figuring out how do they do it right and how can we
use that kind of tools and technology in our development. I just saw one of
those this morning which hasn’t been publicly released yet, but I think it will
be very interesting.
We are now really wrestling with the issue of okay we have all this data
going on in these patient communities, a lot of activity. Somewhere around
about 30 percent of the people who are going online for healthcare, which is
pretty much everybody who goes online at some point or other, about 30 percent
of those people have been looking into user generated content, and about 10
percent have actually been creating user generated content in healthcare
contributed communities. So you got a lot of information, and we’ll talk about
that shortly. But the question is how do you parse through that and find the
relevant interesting stuff and present it back, and that is a really serious
I think increasingly some of these intelligent tools I talked about which
were basically getting people better more personalized content are starting to
be used for transactions, and used for simple transactions like setting
appointments with clinicians and what have you. At the moment I think that’s
increasingly going to be a part of what they’re doing. You are seeing this a
bit more on the financial side than on the clinical side. There are now some
pretty interesting new tools comparing health plans and trying to figure out
which health plan you should choose depending on which diseases you have. This
is something you might know, my old boss used to have a joke about how do you
figure out which plan to choose, it all started with you have to choose which
diseases you were going to have, it’s actually kind of coming true, you know.
Then, as I said, we’re also now starting to see these filers of data start
the walls coming down, start seeing them being attached to content, and there
is some really interesting stuff going on in some of the early work. Some of
the providers who are now working with Google and Microsoft in their health
force and Google health initiatives, are really starting to integrate some
really interesting data content, which is really much more than just
information about a disease, but actually what should you do given the
condition you currently have.
I think increasingly you’re finding much more activist patients, and that’s
going to be an ongoing theme. Actually I was at the meeting of the IOM
yesterday, which was about integrative care, which is interesting because
everyone on the panel was thinking about coordinating chronic care integration,
then one of them was thinking about acupuncture and natural healing being
integrated with care. But we’re talking about taking data from multiple silos,
integrating it and putting it into a more useful format, and patients
increasingly taking the lead in that.
Just to finish up the Health 2.0 section. Here is kind of a continuum of
care, continuum of Health 2.0. We are really on the left-hand side of this
continuum, which is using these Web 2.0 technologies for patients and
physicians to communicate with each other, but not really to connect to the
wide healthcare system. We’re just getting into that second box here, which is
tools communities created as part of this sort of Health 2.0 movement, starting
to connect to the healthcare system, which is simple things like appointments
and managing simple clinical events, getting lab tests, that’s just starting to
happen. But with things like Kaiser’s Health Connect and some of the other
tools around that’s now starting to happen.
Where we haven’t really seen much impact yet is the last two boxes, and
these are far more theoretical at this stage. But a couple things, if we’re
going to see with clarity into more data and understand more information, you
can expect at some point one would both hope and postulate to see changes in
the way that we pay for things, changes in the way we actually change our care
processes, brought about by more information. And I think if you talked to the
folks, I’m sure you’ve had them up before you in this group, from Kaiser
Permanente you have Health Connect where they now have a very large database
and they’re starting to run all kinds of charts, and I’m sure you’ve done this
at Partners and elsewhere.
We know that how we’ve been doing things isn’t the right way, and part of
the reason is we haven’t had really great data on how things get done, and we
haven’t been able to throw our computing prowess at this database. Once we do
all that, it fully stands to reason we can start things in a different way,
because every other business in America does it for good or for bad, as we’ve
been finding out.
Then the last part of this, and there are already some optimists running
with this, is the concept of using all this data to change discovery and
therapy. I think a lot of what we are talking about here is connected to that.
I don’t have the answer, but there are already people talking about using these
tools to do open source research, both clinical research on the bench in the
lab, but also amongst patients, and putting a focus a bit more than that later.
To me these are the continuum. Clearly if you look it on a sort of likelihood
versus ultimate impact, if you could change how this information changes
discovery, that would have the big ultimate impact. It’s less likely at the
moment, we already have these, that’s already happening, right now it’s growing
very fast, that’s very likely. But that is going to have less of an ultimate
impact than actually going all the way to having data change based on or
discovery based on changing the data.
So with that I’m going to start talking about sort of what you brought me
here to do with the research side. Are there any questions, comments, appalled
opinions about that? I’ll take a breath for a second.
MS. MCCALL: Are you halfway through?
MR. HOLT: I’m probably halfway on slides as far as this.
MS. MCCALL: I’ll wait until you just wrap it up.
MR. HOLT: Okay, good. I want to talk a bit about data sources, and as I said
at the start, this is very early days, so I don’t have a huge amount to say,
and I certainly don’t have the answer. But what is going on is that online
these patient people are talking about what’s happening and they are now being
given tools to rate what’s happening. So this is a patient daily string, and if
you look on the right-hand side here this is a page in which patients are
rating treatments for particular conditions. If you are halfway down the page
you might find it interesting that crying works 82 percent of the time, whereas
Lexapro only works 55 percent of the time. You might have heard on the news
this morning about Lexapro and crying.
So the quest is I have met with clinicians who had strong objections to
allowing patients to taking this information as useful from patients. We are
certainly seeing that in some ways the crowd is voting with what they think
works for it.
This is also happening among physicians. Sermo has now got over 100,000
physician members. Before we knew each other online, here they have a case I
just pulled out from last month which is a clinician conference case online
where clinicians get to vote about what they thought the result was, and you
can see the results are the votes on the right. There are not a huge number of
votes in this particular case, only 88 votes, but you get some sense of the
ability of power to get the clinicians through a clinical conference around a
particular case. And some of this has been used already by the FDA and some
other agencies, sort of used as an early warning signal for drug interactions
and ADEs and that kind of stuff.
I wanted to say a little about people who deliberately try to use these
communities for research. The key one is Patients Like me, which has actually
an entire segment that is devoted to research, it was started by two brothers
trying to save their third brother who had ALS, and it was started as a
research company that was then changed into an online community. They actually
had an entire research team, and they have various public presentations which
you could easily get hold of.
In particular, bubbling up through their communities is this sense that the
use of Lithium might have a positive impact on ALS. So they actually ended up
on this trial, and they had several hundred patients in a self-generated
clinical trial taking Lithium to see if it worked or not, and you had a whole
bunch of information available. This is available on the public website, over
122 of these patients have their information on the public website, freely
available, you don’t have to log in to go see it. And they were looking at to
see what was the impact of Lithium with these patients in everyday use on their
condition. 358 patients were taking Lithium here, and this is sort of further
down, here is one of them and you can actually see who this individual was. A
bunch of information here about the impact of going on this essentially
self-generated clinical trial.
You can even see the score, results of individual patients, and if you can
figure that out, this is functional, and I’m not saying you should be able to
understand this just by looking at the chart. But you can click on every one of
these lines, it’s a bit tricky, you get the functional status of those
individuals from the time they went on Lithium. As it happened interestingly
enough, the conclusion of this trial is that it didn’t have a positive impact.
But nonetheless, you were able to generate this trial out of nowhere very
quickly with this number of ALS patients, that probably would be almost
unthinkable to do this in any other community or scenario.
There has been one interesting new company that has come together, this is
really early days for this company, only have a couple of thousand people using
this site called Cure Together. It’s a geneticist who started this absolutely
on a shoestring. Here is a page where you can start and go to the site, and if
you have a condition, in this case it is depression, you can start recording
your symptoms on a very detailed level give information and put your
information on this site. The exact goal of Cure Together is to do develop
better information about symptoms and treatments and cures by basically
collecting information. It is essentially a volunteer database.
They have also got some way they can show some results. These are some
results about impact of treatments on migraine, whether you tried it, whether
it was effective. They are collecting huge amounts of data, very specific data
about individuals in a way that encompasses their whole life rather than just
the bit that the healthcare system sees. And Don Burbick(?) yesterday at IOM
was talking about, he was revealing, he said too much about himself, he was
talking about his surgery, a surgery which is quite well known probably in this
room, but actually lack of a knee surgery in that he had a steroid injection
instead. But that he was talking about his ability to go cross-country skiing
and to go to this place in New Hampshire where he can see – and he
eventually claimed no clinician has ever heard this story and knows about this
issue, but we’re talking here about neck massages as a treatment for migraine.
I suspect that there are very few medical records which record neck massage as
one of the treatments for migraine.
So the question is what works and what doesn’t. And this group is even
already, I won’t even try to pronounce this condition, but has already put
together over 100, maybe 200 women sharing their experiences and treatments,
and created this self-published book about all the, I won’t try to pronounce it
because I’ll get it wrong, but about a serious rare medical condition.
Then the last thing I’ll say about this is my dog is the one underneath on
the bottom right being attacked by the Boston Terrier, this issue around cats
and dogs. When I mentioned that the Healthcare blog is now being used between
people to argue about health standards, there is kind of an argument, and I
think we’ll come to this a bit later, about should we be putting electronic
medical records in the doctor’s office or should we be looking at trying to
capture patient outcomes, and sort of what should be rewarding. And even if
that argument is kind of finished with the Stimulus package rewards, at some
point we have to bring these two communities together. We have to figure out
how we make, as Bill Murray said in Ghostbusters, cats and dogs living together
in perfect harmony.
That may not happen, but if a cat is somebody who likes a place, and if dogs
like people, which is a common truism, if we’re going to follow people around
and try and record their information, we’ve got to try to somehow integrate
that back with what’s happening in the healthcare facilities where the care is
taking place. I don’t have an answer for that, other than we need to be
thinking about both ends of the spectrum as to how we generate data.
Then I just have some recommendations how to spend the $1.1 billion you guys
will control now, right? The numbers may not be quite right, you may not
actually have that. But I think with this money it’s cheap to go and figure out
some partnerships. Literally, Cure Together is a woman, a man, and a dog in a
garage, with not much, it’s unfunded, really with very little help. You don’t
need much to produce, it’s very cheap to put these sites together, you don’t
need much to help them keep going. And a little bit of research funding I think
would be very, very valuable. And I don’t think it should be in the business of
making more dot com millionaires, but I do think there are some little sites
here which could use a bit of help. And if they are having to rely on
advertising for the drug companies or what have you, most of these communities
don’t want that. They are interested in doing research and treatment and
self-help. And I think that’s an area where a few research grants funded
judiciously could go a long way.
I’m now stuck on this whole concept of integrated, integration, whatever it
means as of yesterday. But we do have to connect the big databases which are
based on EMRs, like what’s coming out of Kaiser and the VA and elsewhere, with
these other sources, we have to figure out a way to do integration between
them. I’m nowhere near clear enough to understand how that gets done, but I do
know that many other businesses have put together disparate datasets from small
algorithms, and a lot of computing power, and have got really interesting
answers out of this. It seems like very valuable industries like direct mail,
really as a society has managed to figure this out, so I think we could do some
of that in healthcare.
And the last piece, which is back to this cats and dogs issue, is if we are
going to spend all this money on especially CMS who’s going to direct this
money and reward providers who are doing the right thing, let’s think about the
whole panoply of patients and consumers out there and outcomes as a whole, and
everywhere they are and try to get information and reward reaching out to them
everywhere they are, and try and get reporting back from everywhere. But that
would be really useful information, because as has been stressed many times and
related this to you, healthcare is not what happens in one hospital admission
or in a seven minute office visit, healthcare or health is what happens to
people in their everyday life. We know that what impacts, you know, if we could
figure out what impacts their everyday life and the health status, that would
actually significantly change the numbers of those office visits, and also the
admissions we have to pay for. I think that’s the greater goal we’re all going
With that I’ve overstayed my welcome, and I’ll now shut-up and take any
MR. REYNOLDS: What is the desire of the group to – how soon is John
going to be joining us, how long?
MR. QUINN: He is scheduled for 2:30, but he’ll call here.
MR. REYNOLDS: For 2:30? No, that’s no problem. So here is what we have, Dr.
Mandl is not available, Dr. Spranca is not available, so Matthew is available.
MR. REYNOLDS: Let’s just go around the corner here.
DR. HORNBROOK: Matthew, have you seen any examples of what we can see in the
future? Let’s say some clinical scenarios of a parent with a kid with apparent
ear infection, and what is your vision of how they would treat their child and
stay out of the healthcare system? Then a vision of let’s say somebody with
something much more acute, like severe dehydration from vomiting and diarrhea,
and they’re starting to hallucinate and not sure what’s happening to them. Do
you view the web as being roles in these two kinds of cases?
MR. HOLT: In both of them. I think they are two different – so I think
in general what we need to be doing is replacing people with computing. Not on
a whole replace human beings with androids.
MR. REYNOLDS: You are being taped, don’t let it bother you.
MR. HOLT: That’s all right. I don’t work for the federal government, and I
have no federal grant, I don’t think.
MR. REYNOLDS: But the androids are listening.
MR. HOLT: If it’s being taped somebody somewhere is listening. What I mean
by that is there is surely no disagreement in this room that we don’t have
enough of the right people in the healthcare workforce, and we can’t easily
afford more people in the healthcare workforce. And we need to figure out a way
to drive decision making down to the cheapest and also most effective level at
the same time. This is Don Kemper from Healthwise I’m going to meet with
tonight has been banging on about this for many, many years. And I think my
solution to the sort of simple case would be a combination of smarter systems,
so something a bit like getting better personalized search to sort of
essentially deliver to the patient the same kind of guidelines that if they
called now and got a hold of the person like in Health Solutions, as Blue Cross
or wherever, and actually delivered and was able to get that information,
figure out what should I be doing, should I take my kid to the emergency room,
is it serious or is this something that can wait until the morning, get that
information and even share that with the group, or being able to chat with a
These sorts of things are starting to appear already. There is this company,
America Well, you’ve probably run into already, it’s been making a big splash,
we’ve got Blue Cross in Hawaii, we can go online and there are other things you
can do, which put that level of interaction into a – which isn’t urgent,
but is near term, but into a safe format that is useful for a – but I
think deliberate content first and then professional clinical care second, and
start it low end in terms of cost clinical care before you get to the emergency
For the second group, I’m not a clinician, my dad is a – the stuff I do
if my dad ever found out what I was doing, he’s an old world British
obstetrician gynecologist, and if he found out I was advocating patients having
rights I think I would be cut out of the will. I hope he’s not listening. But
he used to come home and discuss his clinical work all the time, and I never
heard him talk about the patient’s choice the whole time. But that might be a
reflection on 1960s Britain.
I think that what we need for those people, especially people who are
elderly, frail, sick, they’re not easy access – now, by the way, before I
get into this, a lot of the people or patients are incredibly frail, incredibly
ill, and have managed to figure out – and someone obviously tapping away
with neck movements. If any of you get a chance the movie about Stephen
Heywood, he’s the one of the Heywood brothers that unfortunately died with ALS,
really shows you incredible, you know, actually moving his neck and able to
take part in his community and use the computer, it’s phenomenal. That’s a
movie called So Much So Fast. I blogged it? Twitter user over here.
But to go back to the issue, for those people especially I think you’re
going to see, whether it comes from health plans or it comes from the federal
government, whether it comes from private sector consumer group, and whether
it’s the role of a medical home, I don’t know where it comes from, but someone
to be the navigator of the system using these kinds of tools I think is
And there was a case presented at another conference by the CEO of
DoubleCheck – where literally, now, it’s very early days again for this,
and she’s a clinician, she’s a physician, and she used this tool, but she’d
been called by her friend in Los Angeles who said my mom is going to the
emergency room, she’s 85 years old, she has multiple co-morbidities, the
doctors are going to take her and they want to do a stent. And essentially what
happened was a typical medical error that happens in very hospital in every
city in every part of the country, and probably every part of the world every
day, which is they missed a couple – the clinicians didn’t have good
information, when they did have the information they missed a couple of sort of
basic, well not basic, but it was obviously too hard for them to get into their
treatment program, missed some of that basic information, they probably didn’t
have really good solid EMR in that emergency room. And I forget the exact
details, they had taken her off one drug and put her on another, there was a
drug interaction, and if you go through the treatment guidelines there are
several things that should have been done that weren’t done.
By acting as the patient’s advocate, the CEO of this company was able in
fact to end up getting that elderly patient home, she didn’t need to have a
stent, and in fact it wasn’t anything to do with heart blockage, it was some
congestive heart failure caused as a byproduct of the treatment she was on, and
ended up not having her receive the stent, which not only saves Medicare a few
tens of thousands of dollars, but more importantly probably even saved her
life, because she might not have been able to survive that experience.
To me somebody somewhere in the system has to – we know healthcare is
too complex for any one clinician or any groups of clinicians to easily manage,
and especially with these very sick patients with multiple co-morbidities. I
think these kinds of tools are going to be used. And whether it comes more from
a consumer-based system or an EMR clinical-based system, it’s going to
integrate together. So I would – they’re becoming more widely used in most of
healthcare, it’s kind of as Charles Schwab puts a relatively well-paid
intelligent human being in front of a lot of computing power to talk about the
stock market, to adjust portfolios, but a lot of that sort of portfolio
adjustment and what have you goes on behind the scenes is done by computing
power, not done by the whim of the individual on the keyboard.
That is kind of where I see that going. I’m not sure it has much to do with
data collection, other than obviously the more we know about what’s going on
elsewhere we’re going to know more about that in terms of making those systems
MR. REYNOLDS: Justine, did I see yours? Paul, I saw Paul’s. Leslie, I’m
DR. TANG: Thanks, Matthew. I wonder if I can propose sort of an alternative
model, or at least some other framework and get your comments on it.
So rather than dogs and cats, I’d like to look at your continuum, the
arrows, and I didn’t see an exact role for instead of what I would call dogs
and cats I would call pilot and copilot, without assigning who has which seat.
The Health 2.0 or the Web 2.0 is sort of like a relationship anonymous with
anonymous. And even I think it’s called Medwell, the one in Hawaii, is sort of
anonymous MD with anonymous patient.
MR. HOLT: Actually their goal is to be a bit more like your organization, a
bit more of known patient with sort of licensed known approved doctor within a
health plan network format, it just may not be your doctor.
DR. TANG: Where I want to go is where does this fit? And do you see the
synergy of establishing a more continuous relationship between two known
parties? With in the case of this person you talked about in wherever he had
his chest pain between your PCP and standard terminology, and no matter where
you are and in no matter time.
Where we’re going, for example, you talked about let’s look at what happens
in the organization, the hospital or the clinic, where healthcare is delivered.
Well, ironically what we did is we went to the home where life is lived for a
diabetic, and tried to figure out what is it that we can do either to
instrument this what’s going on or provide the tools where life is lived that
would help this individual get more control of it and be able to manage their
The other thing that is described is oh by the way we can get a carbon copy
so that we can chime in where appropriate. But really this is all about the
individual and bringing the resources to the individual, that it is sort of
leveraging the relationship into underlying data of that relationship. How do
you fit that into the picture, and how do you leverage that?
MR. HOLT: Everything you said I couldn’t agree more with, and I think it’s
wonderful. And the problem here is not what you’re doing, the problem here is
that – is a wonderful organization which is not reproduced across the country.
We just know it’s the case that most Americans do not receive the healthcare in
an organized system of care with a strong medical group like that, or even at a
Partners, which Matthew will tell us is equally not quite as – across its
But most of them, I mean I hate to use personal anecdotes, but if it’s good
enough for Don Burlick(?), why not, I am a member of an august organization
called (?) Blue Cross, because it used to be, it changed its name for whatever
reason, but some of you who know Don is very smart, intelligent medical
director, they’re now paying people on the health, they’re actually paying
bonuses on their health or what have you. I’m also a member of their HMO, not
because my wife’s company’s – yet.
I’m also a member of a kind of quasi-concierge medical clinic in San
Francisco, where I went in last year and finally my wife bullied me and I
finally went in and had, I’m a typical 45 year old male who is never going to
die and never sees a doctor, she made me go and have the test and yes my
cholesterol should be lower, and I said I was going to go and get my nutrition
looked at, and all that kind of stuff. And really what I heard from either Blue
Cross or from my concierge medicine clinic, I was supposed to go back in
October and now it’s February, what have I heard about this been bullying me,
incenting me, getting me into place, I’ve heard nothing, despite the fact that
I’m in an HMO which is supposed to care and the concierge with champagne, $200
a year, or whatever the number is. And I always think I’m very, very typical.
So I am probably one of those people who are an accident waiting to happen in
ten years time, that the system can’t figure out, or fifteen years time or
twenty years time, and it’s kind of down to me. And even though I do this stuff
every day and I understand I should be eating a Mediterranean diet, and on and
on and on, to me most people are being left out of these systems.
I’m completely with you, I would love to see this integrated, and the Health
2.0 Web 2.0 community and services being integrated with everyday services in
American healthcare, but if we’re going to rely on American healthcare to
change I think we’ll be waiting a long time.
DR. TANG: So the question is just can you accommodate that in your mind so
that we can start latching on? Because I think there will –
MR. HOLT: Oh, no, no, no. Don’t get me wrong. I think Kaiser Permanente, VA,
all those are squarely in the second group here. And in fact, I don’t know
about your organization but I know Kaiser is looking more and more at going to
the left and trying to figure out how to use communities online, and should we
be the venue for that one? Sure. Similar using other things in this world as
well. So, yes, of course. And I think most people within the health people know
well from the people who had an event like the Heywood brothers and wanted to
start their own system, to the people who are the sort of Silicon Valley
entrepreneurs who are simply going to make a quick buck, all believe that the
healthcare system that they are not going to replace the healthcare system,
they’re going to work with the healthcare system. But having said that, I think
a lot of this stuff is running very fast head of the capacity that we’re seeing
each and every day in healthcare.
MR. REYNOLDS: We have Leslie and then Blackford.
DR. FRANCIS: I’m the co-chair of the Privacy and Security Subcommittee, and
I want to preface this by saying that first of all thank you. And as I
understood the goal of this it was to provide us with some of the kind of
enlightenment that will be helpful to us as we think about what are the kinds
of issues we ought to be thinking about in confronting, and indeed in making
maximal use of some of these wonderful new options that are out there.
With that in mind, obviously there are a number of kinds of privacy issues
that all of these sorts of interactive uses of the web raise, including one
just plain where the information goes and who has access to it, but also how
much people understand about what might happen if they log in with any kind of
personal information and so on. So there are understanding problems, and there
are questions about where it might go.
I know that with respect to MySpace or even Facebook there is some informal
understandings that have grown up, so people who are looking for jobs don’t put
stuff up on MySpace that they don’t want prospective employers to see anymore,
I mean if they’re smart, that’s kind of become known. Facebook just within the
last two weeks had a really interesting user corrective change in policy about
keeping information. And I can think of other things that have evolved, when
you pay with credit card now over the web only the last four digits get
recorded, you think, or questions about even Paypal emerging as an institution.
I guess I have a two-pronged question. What are the kinds of privacy issues
that you see as the ones that either should or could be addressed as we move
forward that would be helpful in encouraging more people to make use of this?
And second, are there any emerging either informal norms or formal norms or
formal mechanisms that you’re aware of that look like some of the examples I
MR. HOLT: Let me preface this by saying that while I have strong opinions
about the subject, and the strong opinions are weakly held because I’m not an
expert and within the privacy detail I don’t know the legislation about this
too deeply. But first off, there seems to be a major concern when you hear
about people concerned with privacy it’s not the fact that somebody may steal
their health information and know about it, but that it may be used against
them, kind of like in yours.
And I think a couple of things. One is that yes there is user pushback by
the Facebook example recently where the user said hey no you’re not going to
change the terms of service on us for your own gain. And that has happened in
several areas, and feedback is very fast in the web. There’s a case actually
where Microsoft initially overpaid some severance payments and then asked for
them back, and then the PR was not worth the hassle, we should have instead
done the right thing and just let them keep the money.
There is also this sort of bending the other way, which is I actually just
hired some people for my little startup, and in a recession I should get a
medal I think, but there were a couple in the application process who kind of
made big mistakes with Facebook. And actually one of them I missed an interview
I was going to have with them on the phone and they basically complained on
their Facebook page, and then invited me because they were trying to connect
with me on Facebook, but that meant that I could see their profile and their
ranting about me missing the interview. They didn’t get the job. That’s a
little unusual. But everybody who did apply, because they are all 27 – I
went and looked at them all, so they were drinking in a bar. Look, most people
go out when they’re 28 and drink in a bar. I’m past the age of saying oh you
should have nothing on that, we have to have complete professional stuff
online. But obviously people’s work life and social life online is – and
you’ve got to accept that, and you’ve got to have some more norms from that.
Now, to healthcare, there’s a couple of ways I think you can answer that.
One is the Patients Like Me approach, which is to say you stick it out there,
you put everything out there, and we want you to be as open as possible and
that’s part of the reason to join this community, but you can have one layer
back, you make sure that only other people in the community can see this. And
if you looked at this 325 people, I think only 122 were actually opened on that
To return to the big problem here, it’s not again what are the privacy
concerns of people about their information, if you have ALS people already know
you have ALS. There are obviously other conditions that are more sensitive and
stuff can be worked through that.
DR. FRANCIS: How about ones you can’t see, like MS?
MR. HOLT: Of course, a problem, I understand that. For me this is not a
healthcare issue or privacy issue, this is a regulatory issue around who can
buy insurance, and who can be discriminated against for that reason. I think
the way to fix that is we know a much bigger fix, which is outside the vestiges
of this committee and probably outside the vestiges of this building, but
hopefully will be fixed sometime within this administration.
Because I think that to me is the main fear, the people who are terrified
that their insurance company might find out, or their employer might find out,
and find some reason to get rid of them, because they won’t be able to buy
health insurance. And somebody in this room advised the health insurance in the
individual market or – very conscious of that.
DR. FRANCIS: Could I just follow up very quickly to ask if there are any
other mechanisms that you’re aware of other than the Patients Like Me, that if
you joined a community that are controller-based that we might want to know
MR. HOLT: There are absolutely. There are at least three or four companies
who are in various stages of developing their online privacy management
systems, whereby they will reveal a certain amount about you to people or
organizations that you want to for whatever reason. There is one called Private
Access, again very small company startup, which is intending to allow people to
sort of release information about themselves for potential clinical trials.
Essentially think of it as being a vault where you can release little
sequential dribbles of information, say you can put out there that yes you
might be interested in this trial, but go through these gradations until the
actual information, until the researcher finds out your name and email
essentially, where you can control exactly what people can see.
There’s another one called You Take Control, it’s similar. There’s a company
called Vdentity, which has got a sort of phone-based anti-phishing device
essentially where for release of patient information there’s like an
access-based phone-based system so that you could, you know, if an organization
wants to release information to another organization, like a doctor wanted to
release information to a clinic or what have you, they would get a call at home
and it would be a sort of identity that was already set up and put them through
So there are a number of companies like that who are trying to figure this
out online. I think it part of a wider societal issue about how much
information we’re going to release. And let’s face it, there are 20 percent of
Americans who don’t think that anybody should know anything about them, and
don’t believe in the big database, and we’ve heard enough from those people
recently I think.
MR. REYNOLDS: Blackford and then Don.
DR. MIDDLETON: Matthew, thanks again for coming, and great to hear you
speak, and I appreciate the way in which you have been not only chronicling and
observing the healthcare industry, but also pushing it I think in interesting
My question for you is to ask you to put on your philosopher’s cap for a
second. I guess there’s two, I have several questions if I may, but the first
one is really about do we know that community actually has a positive impact on
health? Certainly there is recent evidence to suggest that having fat friends
makes it more likely that you yourself are fat. Certainly drinking, smoking,
other untoward behaviors and a wide variety of things, may go cluster in ways
that are untoward for health, as opposed to the other kinds of benefits we see
with communities. So I think that question has to be put on the table, and it’s
really not for a response unless you care to.
MR. HOLT: Well, actually one quick response, is there’s an organization
called Center for Connected Health you may have heard of. They’re suggesting a
study about roles of people, the value of people – online community. I
think we can argue, there’s pretty clear evidence from David Sabell(?) and
others that on offline real life communities the patients improve the patient’s
experience and their outcomes. I think slowly emerging we’re seeing some of
this, more of this, we certainly have the anecdotes, yes, the proof is coming,
DR. MIDDLETON: That’s where the next question follows. I think we struggle a
little bit with the theory of evidence around these data coming from social
networks and, if you will, the epistemology of their truth or veracity and what
not. Several questions come to mind. When you think about the evidence, where
do we get the evidence for hyper-segmented care? That is my therapy, this came
up over lunch, may be unique to Blackford Middleton, God forbid, but it may be
that my gene sequence, my gene type, my behavior patterns, or what not,
suggests a particular type of therapy which no one else may ever actually need
to have. Thus there’s an evidence problem of how do we actually determine
what’s good and bad from a man of one perspective.
On the flip side of the same problem is how do we determine what’s
acceptable of population evidence from social networks, when there’s no
experimental design or control of the experiment, if you will, rather these are
association observations, association studies, suggesting that Lithium is
actually having some impact on ALS, and ask actually a change in language, it
was not a trial, it’s an observation, it’s an association.
Third, there may be a significant regression to the mean kind of problem.
That is that if we take a population and begin to look at it as only a
population, we’ll fail to identify the hyper-segmented individual
characteristics, and thus treatment and diagnosis may regress toward a mean.
Then last, two more actually, I’ll go back to them. If it is the patient who
is the source of these data, and I’m good with that, however we need to think
about data validity and we talked about this earlier today, accuracy,
reliability, precision, and non-repudiation for those data which you might
share with me or I might share with you if you were my caregiver, so I can’t
game the system.
Then lastly, how do we actually measure the values? The people who are going
to be paying for care which might include social networks, and the impact if
you will that they have on my outcomes, how do we actually measure the value?
MR. HOLT: Wow, okay, let’s start at the end and then remind me of the first
The value question to me is really tricky, but it’s almost – most of my
answers will be in this flavor. It is kind of we’re not really bothering about
this other stuff so much at the moment, so we should be putting it in a
universe in which we’re looking at the value of everything. Today – you
know, you get more evidence. But what we spend a lot of money on in Medicare
varies dramatically with no real reason for specifically increased value, in
fact may be decreased value from that extra spending.
Having said that, I think there are two things. One is that a patient choice
and patient experience in the absence of are there outcome measures and whether
people like it or not, I mean I think that’s got something – people’s
experience of their healthcare is somewhat valuable in and of itself if that
improves. So people who go to these online communities and say it helps them
and they have a better time dealing with their condition, I think that’s clear,
there’s evidence of that, and how you count it, that’s clearly one piece.
Then the other question, and it comes down to this, is are we replacing,
quote, unquote, “unnecessary or bad care,” not even bad care, are we
replacing care that could be – settings to these virtual settings, and is
it possible that you can measure the people who have some level of interaction
with others in communities or with the system online, are using resources and
doing as well if not better.
The only evidence I think so far is that one Kaiser study looking at the
number of office visits using, and you have probably been involved in similar
studies, using a number of office visits for those people using their online
email service, went down, satisfaction went up, and then overall care was as
good if not better. There has yet to be data showing similar evidence for use
of social networks.
But my guess is you could observationally go back and find similar evidence
using patient daily strength. We had one person come to a – meeting who
was on a diet actually and lost several hundred pounds, and really turned her
life around through a support group. And point one, could she have done without
daily strength, who knows, she says not. But I mean again, not reputable data,
which gets I think to your third or fourth question, which is how do you, if I
got the question right, it’s how do you know not regression to the mean, how do
you know these people wouldn’t have done it anyway, how do you know that you
just didn’t see the people who were self-selecting to this. That’s a huge
boring big research question, which that’s the reason I’m not still a
healthcare researcher, with that kind of a question.
I don’t quite know how you do it, I don’t have the answer.
MS. MCCALL: I actually do have a response to that. I think it is one of
those big hairy questions.
But I think what we have to do is prepare for it. These things are coming,
they will exist, people will join them, and they will work or they will not,
and they will work to a certain degree or they won’t. But the most important
thing to do is to understand that as that information and data and affect, as
the affect is out there, it will be in the data that is traditionally captured
and analyzed. And by God, if I were to do one thing I might want a doctor to
start learning to understand and ask the question that says are you part of any
health communities. Just that one thing, yes or no, can you tell me which one,
MR. HOLT: The one other thing I would add to that is I cannot believe this
problem is intrinsic just to health, with only healthcare has this. If you look
at people who have thrown millions of dollars in computing processing power at
these big data problems in the financial industry, they got what’s the impact
of chat rooms on the exchange in the stock trade. There has to be some kind of
analysis problem that people have looked at. But I don’t have the answer to
DR. MIDDLETON: Just one quick follow-up if I may. I agree with you that sort
of the BI world, the business intelligence world is taking mass computer power
and mass data and market analyses and financial analyses will tell you exactly
what is the right thing for you to do in your portfolio, blah, blah.
The problem is from the theory of evidence point of view, none of that is
based upon randomized control data as an input, if you will, to the modeling.
All the modeling is associative modeling. So I get concerned about rational
sort of allocation decision with scarce resources, or said another way, how do
we do comparative effectiveness research on social networking. I know it adds
value, I get it. But I think somewhere along the way I think the best thing we
have to prepare for figuring out how to arrive at the data and make sure the
data is there to do the studies.
MR. REYNOLDS: We got Don, Ed, and then Larry.
DR. STEINWACHS: I was just going to push you a little the other way.
MR. REYNOLDS: And if you just ask one you’re going to be boring as crap.
MR. HOLT: – I managed to get the first six that wipes it off.
DR. STEINWACH: It’s clear I cannot keep up with Blackford. I have already
lowered my IQ rating, and I know I’m not there.
In some ways what you’ve been talking about really are illness management
systems, and you talked a little bit about your own experience from the health
side. It seemed to me one of the advantages of this, Blackford points out that
all the different options there may not be ones we have any evidence about, but
still people choose, and maybe it’s just preference, which may not be all that
bad. Because many times in the health or illness encounters we don’t know as
patients what all the options are.
So this is partly I think sort of opening up, and much is the websites where
people report all the things they do in trying to manage their disease that the
clinician never knows about, so is there much on the health side of this?
Because this has huge potential it seems to me, and raises very interesting
issues on the illness side, but there are a lot of things many of us could do
if we knew all the options. I know I can lower my cholesterol by eating
oatmeal, is that true, I don’t know.
MR. HOLT: So let me answer that, because this is just to give you a sort of
range coming from the point of view of a guy who has to deal with people saying
can I speak can I speak to you in confidence, which is –
I think for the first one of these in 2007 there were 40 people presenting
different little demos and what have you, of those I think probably 35, 30 were
sort of viable, and the rest were a bit flaky. And when is say viable I mean
– I paid to come to that conference –
But for the one in 2008, which was only a year later, we had more than 170
applications of people who built something, not something I hadn’t heard about,
but there was pretty big growth in this. One of the areas had a lot of growth,
because the entrepreneurs think there’s money behind it, is wellness, because
they believe there’s a corporate wellness market that will pay for all this,
some of the interest in developments – one is of the online health sites
that make money by charging people to subscribe, consumers. The most visible
group is the diet sites. And that’s kind of health, because if you can fix your
diet. So I would say that’s already sort of on the way.
There is also growth in a few companies, one, Sensa, which is your company,
which has an interesting sort of we’ll put you on the Mediterranean diet, I
don’t think yet we’ll send you the food, I would imagine.
DR. STEINWACHS: Does that include they’ll take you to the Mediterranean?
MR. HOLT: I think that’s extra. Maybe if they save enough from heart
transplants, or whatever, you know. If you join Humana you get a free trip to
MR. HOLT: They don’t have a plan in California, you can visit local, yes.
But there is for example, Google or something, it’s like a new company called
Keers(?), which is essentially trying to like get to the people who have got
their lab tests, you’ve got their food, you could explain to them if you could
walk five miles instead of driving or whatever. There are a lot of people
thinking along those lines. There are a lot of wellness sites like that.
Again, the problem is to me how does that get integrated with the rest of
the healthcare system, because honestly before there was internet Web 2.0, yes,
it was always a good idea to not smoke and to exercise or whatever, and we
never integrated that with the healthcare system well before, your physician
may have bothered you about smoking, but it came from other sources. And I
think honestly yesterday I heard a number of comments at this integrated
summit, a medical summit, going back let’s talk about fast food, let’s talk
about smoking, all these kinds of things, which are not strictly under purview
of HHS and the healthcare system.
I think that’s a similar issue. The question is it is worth it? It’s
probably a question for Paul and for the Kaiser folks, does it make sense to go
back down the scale and try to find people who should be in this kind of
environment, and, you know, actuarially does it make sense to get them into
these kinds of programs, and do they work. It is early days, and my guess is it
probably has an impact.
MR. REYNOLDS: Matt, do you have a comment? And then, Carol, are you
following up on this?
MR. QUINN: I was going to say a great example, and I can speak of this from
personal experience, of where wellness is health is meeting sort of people’s
everyday lives. And it even gets into more medical sort of issues is something
like babycenter.com, which is if you have a baby, not me, my wife, then it
actually tracks along. And there are communities that range from everything
from, you know, the initial stages are actually having the baby and preparing
for the birth, but also following up, living your life after that. So it gets
into diet, it gets into post-partum depression, it gets into all these other
issues, so it sort of blends together. The means for doing is that it send you
this update on how old your kid is, or how far along in your pregnancy. Then
there are polls, without OMB approval you can poll hundreds of thousands of
But these self-segmenting very niched communities, and the same is probably
true of people of hang glide, there is a niche community for them that is how
do you –
MR. HOLT: Actually I’m in it, it’s just a boring list serve, by the way,
it’s just a list serve.
MR. QUINN: I think it’s even broader though, than just what we would call
Health 2.0. It’s people living their lives, and this blurs into health and
MR. REYNOLDS: Carol, did you have a comment on this?
MS. MCCALL: Yes. And I’m going to go back to what Don was really asking
about, the distinction between health and healthcare, and these models.
I don’t know if folks saw this, we actually did a press release on New
Year’s Eve, and it was obesity has a price. And the bottom line is that like
sugar and butter, I actually was the one who had the opportunity to do the
math, for every pound we are overweight the annual additional healthcare cost
is $19.39. And it starts at about $10.00 when you’re about 20, and it goes up
to about $30.00 when you’re age about 65, and when you do the math and you
scale it up through the United States, we’re 6.5 billion pounds overweight, and
that translates into $127 billion additional dollars that we pay every year for
that and how overweight we are, for overweight and obese.
So the question was does it pay actuarially to do this, I think research has
shown the evidence base is strong, not just the cost per pound, but that health
is vital, and the pursuit of health even more so, and that health is related to
healthcare but they’re not the same. And from an actuarial model what I would
say is that if we could do something to just change the business model, we’re
stuck in kind of a short-term ROI trap, that’s the main issue. So it’s not
about whether or not it’s valuable, it’s about whether or not there is enough
time for however it’s going to get funded and however that gets put together
for it to have a payback. No one would ever expect any other business or
industry to invest heavily in something whose payoff comes five to ten years
after their customers go away. And yet reform can turn those business models
inside out, upside down, fundamentally change the value proposition and the
equations, and it can all just be so different, just so fundamentally
I think that those are really the things to look forward to.
MR. REYNOLDS: I’ve got Ed, then Larry, and then we’ll try to get John.
DR. SONDIK: Thank you. This was really fascinating. I was kind of wondering
along the lines of Blackford, kind of on the other side of that, the market
side, so to speak. Given that this information is out there, is there either
sites sprung up that are critiquing what’s done, or is there scholarly work
that has been done on that fascinating one that you showed us on ALS or the
like, I which, you know, it’s just you know the whole thing about how to lie
with statistics and all of that, well it’s even easier to lie with graphics and
Then when you couple the two together you’ve got something that is
tremendously compelling. Boy, talk about caveat emptor, I mean it is
interesting to contrast that with the examples that Matt just gave. I was
thinking that when we had kids we had a community, and that community happened
to be within one square mile, and we didn’t live where our parents lived, so
the acquaintances we had. And you know when you’re in that mode everybody’s
pregnant it turns out, it’s quite incredible. So there was a tremendous amount
of experience in all of that that kind of gets shared. Here the community is
multiplied by a factor of several million I suppose.
But to come back to the other point, is the critiquing springing up? I mean
I would hope that it is, although I must say there are rules of evidence and
rules of evidence. And you look at the trials that go on in England, some of
the what do we call it, the low data, not low data, but the minimum data
clinical trials that are done and so forth, you have to look at those in a
particular way to understand the results. You have to sort of put on a
classical and a Bayesian hat at the same time.
I’m wondering, this may actually be a kind of a role for this Committee to
talk about at least a little bit where this ought to go. I must say I am both
tremendously impressed, and at the same time I’m really very wary of this
because the numbers make it seem to credible, the magnitude is what I’m saying
makes it seem so credible, and yet I would like to see some sort of feedback
that says that it is in terms of the information that is being given.
MR. HOLT: So there are two things which I am going to tell you, neither of
which I really understand more than 15 percent about. The first one, and I was
thinking this might come up today, so I went to work for the Institute for the
Future, having never heard of this thing called the “Delphi Method”
that apparently is a very popular thing invented in the 60s at Rand, and
there’s a science fiction novel book writer in which everyone bets on what the
Delphi pool is going to predict, and the idea of very large numbers. And it’s
the same thing with theme show Who Wants To Be A Millionaire, ask the
audience and the audience is always right.
So there has been sort of the pop, I wouldn’t say – there’s been a lot
of work, academic work done on this. James Surowiecki from New York has written
this book called The Wisdom of Crowds, which looks into this. And
basically there’s the argument for the side which is if you get the crowd
structured in the right way with certain prerequisites to how it’s led or how
it’s structured, how it’s asked a question, the crowd will tend to be smarter
than the smartest people in it, so you can expect that the answers you’re
getting out of something like that are likely to be correct. But then just as
well as with any other research you talked about you have to go and look at how
it was actually done, and whether it fulfilled those criteria.
So I will shut up about that because I’m the guy who read the book once and
barely understood it. But there is certainly some academic research behind the
concept of this whole wisdom of crowd approach.
The next question I think then is how does that equate to something where we
have very little information, because we haven’t really looked at it before. I
would say things like, my issue is traditional randomized clinical trial done
by a drug company it seems to me is the way it works in practice, and this is a
gross generalization but it has some grain of truth, is you figure out exactly
the few discreet numbers of people who this drug might work on, you do the
trial with those few individuals, and once the result comes out you generalize
it to all kinds of other people in everyday practice, and then it gets used on
or off label as well.
I think the classic example was the clinical trial around cabbages back in
the 80s or 90s, late 80s I think, which was only done on white males under 65,
and we do all kinds of heart procedures on all other people without having done
To me it is the same problem, and the academic research into the healthcare
side, this is really, I mean there’s Jerry Kane at Boston College who just
started looking into this, but it’s really early days. I think a great thing
for this organization to spend the $1.1 billion is to write some grants and go
look into this, and find the young academics who want to look at this in a
critical way, because it’s really just started. Most of the critiquing back and
forth has been about whether there’s a business model behind it.
Silicon Valley and the online investment communities is here really early,
this looks and smells like something where they can make some money, so they
have been looking at how do you make money out of this thing first, not what is
the social good out of this thing. I think many of the people doing these
communities or behind these communities and these online tools are actually
coming at it from a place where they’re thinking about the good health of their
brother, their family, their friend, and not looking at it from that way. So I
think they would welcome a lot more inquiry about this.
MR. REYNOLDS: Larry.
DR. GREEN: I really enjoyed the last 45 minutes, and appreciate you coming.
MR. HOLT: It must have been dry this morning then. Giving you a hard time.
DR. GREEN: Well, that’s another conversation. An observation, and then I
want to ask you a question about your fourth arrow.
The observation is I have sat here listening to each example of
phenomenology, something that now is happening. What strikes me, putting on my
hat as a physician, I did not detect human phenomenology that I have not
observed my entire clinical life. What is different is scale, scope, and
velocity, and the technology that is being used for the phenomenology to be
We can borrow physics, when you start changing the velocity, scale and scope
of things, they often turn into something that is quite different from what
they started as. That is my angiolytic. I can listen to all this sort of stuff
and feel very calm and sort of excited and very pleased with it all, because it
actually is anchored in a lot of human history, and it is expressing it in a
different sort of way that is just sort of a privilege to be alive while this
is happening in my point of view, because it’s so interesting.
My question about your fourth arrow is anchored in the Committee’s charge,
and mission, and what we are statutorily obligated to do. We’re supposed to be
helping figure out the national health information network and what it takes to
get it there and make it work, and particularly what it means for statistics
and the derivation of knowledge that’s of use for public good out of it.
So tell us, what must we be sure is an attribute or a component of the
national health information network to capitalize on this spontaneously going
MR. HOLT: My suggestion would be that, I don’t know if it’s in the $17
billion or where it is, it’s somewhere in the process that when we are talking
about rewarding/punishing, rewarding those who do and punishing those who don’t
financially, or trying to change information collection, that we should look at
two things. One is I think encouraging the integration of information from all
sources, including patient communities and some of the things I’ve discussed,
and some things I haven’t discussed like some of the early online attempts to
put GM investigation online, the 23 million – and others, including that
kind of stuff. That gets back to kind of Blackford’s remarks. I didn’t talk
about that, but that’s a part of this. They are very early stages, but they
have an interesting role.
But I also say that it’s one thing to encourage the stuff that’s out there
now, so whether it be one of these community groups or Google Health that is
collecting all the data or Microsoft is collecting a lot of data, to encourage
them to have a voluntary contribution from enough people so that their
anonymous data could be somehow integrated and used. I don’t know quite how
that works, but to get back information from these communities.
Then the second point, probably more important, is to encourage – I
think Paul is already doing it, Palo Alto Medical Foundation and many other
people are already doing it – is that if you’re going to be receiving
patient information, if you’re going to be receiving federal dollars, or even
anybody who is recording patient information in their clinic, their hospitals,
long-term care, wherever they’re doing it, should be trying to capture as much
of these other data sources as possible. And if it is looking at the totality
of the patient’s life and where they live. So one of the things that comes up,
and this was raised yesterday by Bonnie (?) in a conversation, it’s great all
about exercise and all the rest of it, but if you’re a 73 year old woman living
in a trailer in southern California and it’s 104 degrees outside you’re not
going to go and run five miles. We need to figure out information like that
about how people are living, things like do pets in the household lower stress
and blood pressure, and how many, do you have a cat or a dog, is it good or
bad, or fish. You’re sitting with one of the world’s greatest fish collectors,
she actually goes to South America to buy fish. It’s really the national
wildlife folks in the room – anyway. But that kind of stuff, I’m talking
about getting other data.
Because I think I would add into this the folks from Connected Health, or
the many different connected healths around the country and the planet, but the
folks in the continued health alliance, we’re talking about how do you get
information from other data sources like devices and what have you that
patients use at home, how do you include that in these databases. Then finally
it’s not very far along, but we’re getting smarter and smarter homes in terms
of putting information technology in our homes, collecting more and more
information. I think I would argue to include those. I don’t know how you
DR. GREEN: If I can, I listened as hard as I could, and I’m not sure I heard
what you said, but I think I did, and I’ll say back what I heard and then you
can fix it, okay? One is I heard you say include non-traditional patient
communities on the national health information network, need to have a spot on
it, need to have an input and a way to get the stuff into it. Right, something
MR. HOLT: Yes.
DR. GREEN: I heard a second thing, that there is a great need for
comparative based studies as the data and knowledge that comes from those
sources compared to the other sources that we might call traditional sources;
is that right?
MR. HOLT: Yes.
DR. GREEN: Then a third thing, you said it would be very nice if the people
providing healthcare could be incented to use the knowledge available from
these sources to the benefit of the public?
MR. HOLT: I also think collected, because my concern is that they will be
rewarded for furnishing information about what they do in their office in their
practice, and not looking at the wider patient experience.
DR. GREEN: Close enough?
MR. HOLT: Yes.
MR. REYNOLDS: Okay, I want to move onto – oh, Matthew, first, great.
DR. HALAMKA: [Via Telephone] The 2:30 testimony I was supposed to do?
MR. REYNOLDS: Yes, John, just one second.
DR. HALAMKA: Sure.
MR. REYNOLDS: Thank you. I’m thanking Matthew Holt for finishing up, but
great job, really appreciate what you did. You stimulated exactly what we
wanted to do, you were identified to us as a thought leader and you did not
fail us in any way.
MR. HOLT: Thank you very much.
MR. REYNOLDS: Thank you very much.
Now I would like to turn it over to John Halamka. John, how are you doing?
DR. HALAMKA: I’m doing very well. Always a pleasure to do testimonies for
NCVHS, and of course all my friends, they already know all this stuff.
So we have to chat about non-traditional sources of data, we’ll focus on
some user entered data. But I wanted to just mention a few other things too, if
MR. REYNOLDS: Yes, please.
DR. HALAMKA: First, I sent the slides back ahead, which I hope you guys got.
MR. REYNOLDS: Yes.
DR. HALAMKA: About the traditional role, what is traditional data is
important. And when I look at the fine work that Paul Tang is doing on high
tech, and a lot of the quality measurement I do regionally, the definition of
traditional data mostly is claims data that is aggregated from hospital
sources, so from Beth Israel Deaconess has a huge case mix database that –
claims that we have submitted for the last 30 years. And of course all of our
payers in the region have large claims warehouses.
Sometimes what we are starting to see regionally is that claims warehouses
are being combined together, and we try to normalize vocabularies so we can do
quality analysis on a population of Blue Cross, (?), Medicare Medicaid, and so
our physician organizations have aggregate claims warehouses.
And in the states we tended to also aggregate these so we can do benchmark
comparisons across health systems. That’s what we’ve done for a decade. So our
traditional data sources are all these administrative claims, and there are
some pharmacy claims that are of some valid use for clinical-like data. But
mostly these are things with ICD-9 diagnoses, and lab tests without lab
results, things that are very challenging to use for highly granular clinical
quality and outcomes analysis.
What we are starting to see, and this is onto slide 3, is we’re developing a
variety of much more clinically oriented data sources to help our quality
analysis, and we’ll get to in a moment some of the Google Health and Microsoft
Health Vault stuff.
But what we’re doing at least on the provider side is we’re creating
business intelligence datamarts, where we’re taking our clinical transactions,
and every night at 2:00 a.m. we move those from a transactional system into a
more longitudinal relational system that allows us to ask queries such as not
what was Mrs. Smith’s hemoglobin A1C, but across all patients with diabetes in
the last 20 years between the Joss(?) and Beth Israel Deaconess what are trends
in hemoglobin A1C, creatinine, and how does that correlate with age and weight,
these kinds of things are now over the last couple of years becoming a much
more granular way for us to do quality measurement. And datamarts for us we
have tended to invent, and we tend to evolve these things based on the nature
of the clinical questions that people are asking, and therefore every year we
get more and more granular data.
Partners is doing the same thing. Blackford I’m sure can talk about the
RPDR, that they have a very similar set of datamarts and repositories that are
really new and emerging sources of clinical data for granular quality analysis.
That is great if you’re in a Partners or a care group, but what do you do if
you want to look across a bunch of solo practices where you may not have a lot
of hospital affiliation.
So Micky Tripathi at the Massey Health Collaborative has created a quality
datamart for 600 practices through the Massey Health Collaborative. And what
those guys do is they run clinical works in Allscripts, NexGen, and GE
Centricity, and they are as part of participation in the Massey Health
Collaborative required to submit uniform datasets to a regional quality data
In fact, we feel so good about what Micky has done, I actually have a
meeting tomorrow, as to how Beth Israel Deaconess and its thousands of docs
will now contribute to quality data warehouse, and what I hope is working with
Paul Tang and his group that we can actually do a pilot for HITEP and some of
the quality indicators that are coming out of the NQF, how might we use HITSP
standards and Micky’s warehouse and Hightep’s measures and actually demonstrate
how a whole region can do quality measurements from EHRs and hospital
information systems to a regional quality warehouse.
Now, one of the problems we’re having with the payers in Massachusetts is
that everyone of the payers is starting to wakeup and say administrative claims
data is not good enough, I want lab data, so you hospitals provide me on a
monthly basis with an extract of 423 data elements, oh and by the way every
payer wants a totally different extract, and try to get these to the payers in
some uniform manner, and by the way every payer has a different designation for
what lab tests they want and what they’re called, there’s no standard
vocabularies. It’s a complete nightmare.
I have had meetings recently with the payers saying might we through our
regional health information organization mass share, figure out a way to submit
to the payers from the providers a uniform set of laboratory data so we do this
one time in one vocabulary constrained fashion, and this now enables the payers
to have more clinically oriented datasets for disease management, chronic care
management. It’s all RHIO-based transactions going from providers to payers.
Another novel data source we’re starting to see, Rx Hub and SureScripts have
nice data from pharmacies and PBM that gives us longitudinal medication lists
for patients. We’re using those for medication reconciliations, but it also
helps us understand a bit about as a community what medications are being used,
we can correlate that then with some of our own administrative databases and
look to see if best practices are being followed, so are people getting
beta-blockers post-MI, are they for diabetes getting appropriate medication.
It’s nice to go beyond a single hospital system and look actually more at a
national medication list, a community medication list, SureScripts Rx Hub.
Part of the HITSP work that we did over the last couple of years on
expanding and standardizing lab data transactions, is that HL7251 with LOINC is
the standard that we’ll use for a result, that by the way you’ll also have
certain elements of demographics that will be included with every transaction
so you can start using laboratory data flows from Quest, Lab Corp, hospitals,
and they can be used for actually secondary uses like quality measurement,
biosurveillance, public health reporting, etcetera. So the good news is that as
labs move to a nationally standard form not only will there be EHR lab
exchanges but the same data flows could be used for quality analysis.
The Social Security Administration spends about $500 million a year on paper
medical record retrieval for their disability claims adjudication, and recently
Beth Israel Deaconess worked with SSA and implemented HITSP standards to
exchange clinical summaries so the patient signs a consent, that consent
digitally is transferred to Beth Israel Deaconess, and then Beth Israel
Deaconess does a longitudinal lifetime medical record exchange back to the
Social Security Administration, all without human intervention. The whole
process takes about 30 seconds end to end, and everything is vocabulary
controlled so that the SSA can actually now run business rules against the data
stream and do near real time adjudication of a disability claim. So it saves
money, it saves time, everyone’s satisfied. It is a good model for how might
you use a nationally standard form between organizations and the federal
government that want to do something novel, whether that’s disability or
quality analysis in a hospital, and do it without a huge amount of burden on
the IT department.
We do something very similar. We do about 4,000 data elements to the CDC
every day from our clinical information systems, using standard formats. In
fact, we’re using the HITSP form for the biosurveillance transactions, and
we’re doing this also XML going from our clinical information systems to CDC.
Again, a model and a novel way of how might we transmit population data, and
their cases of biosurveillance that certainly could be used for quality.
Medical societies like the Mass Medical Society are starting to also
recognize the need for quality indicator, both aggregation and analysis
reporting. So they are working with Micky Tripathi and the Massey Health
Collaborative clinical data warehouse, and looking at novel ways to do
scorecards, benchmarks, risk adjustment, quality indicators. They are focusing
more on the reporting side. Micky’s doing the aggregation, they’ll do the
reporting, we’ll service the physicians of our region.
We have a lot of novel data exchanges with our local departments of public
health, more along the surveillance side and infectious disease reporting, but
similarly going in web-based XML structure from the hospital to the Department
of Public Health, enabling them to do very accurate clinical data analysis, not
administrative data analysis.
Back onto the user entered stuff, because I know that was a topic Matthew
covered, Continua Alliance, 200 organizations working together, to come up with
consumer healthcare devices that will get us data telemetry from the home, from
glucometer measurements, blood pressures, scaled, you can imagine a whole
variety of devices in the home that would send the data to PHRs and EHRs. This
is interesting because now we have the patient in control of the device and if
we actually want to discover not just what care was delivered, but actually the
outcomes of care as measured by patient body weight, as measured by their blood
pressure control, the glucometer readings, the Continua devices using standards
that have been jointly developed by HITSP and Continua give us some novel data
streams. Microsoft Health Vault has worked closely with Continua, and one of
their products you’ll see this year is how do you actually get into that Health
Vault, a standard form that can be used for a whole variety of purposes. You
might imagine the patient aggregating the data in the Health Vault, and then
using the third-party affiliations and the application program interfaces for
Health Vault, transmitting it all to quality measurement organizations or
others who may provide advice based on some of the telemetry gathered.
Done a lot of work on PHRs, and the three forms I want to talk about are,
one, we have a hospital provided PHR which is tethered to the EHR, and that has
both a function to display all data in the clinical records of the patient, but
to also enable the patient to add data, annotate the clinical record, put in
glucometer measurements, add medications, add allergies, add observations that
may have come from outside physicians.
I’ll tell you a real challenge, and I didn’t hear Matt’s presentation
entirely, how do I describe this, we’re not always sure how accurate the data
entered by the patient through this modality is. One example we had was a
patient had heard a diagnosis from a doctor, and that diagnosis was something
like spinal meningitis, and the patient put in “Smiling Mighty Jesus”
as the diagnosis. So okay the patient entered free text diagnoses may not be
the most helpful, whereas something more objective like coming from a
glucometer or coming from another data source like a pharmacy PBM being
transcribed by the patient may be more relevant.
I guess on the tethered EHR with patient entered data the quality of the
data has not been too trustworthy. And I think that does contrast ca bit to
what Google has done. We’ve done a full integration with the Google health
product, where Google Health has two interesting aspects. The first is it
allows the patient to aggregate their own data from multiple data sources,
including Beth Israel Deaconess, Cleveland Clinic, CVS, Longs, Walgreens,
Quest, and therefore when we look at that data stream it actually has the
non-repudiation qualities, that is it isn’t patient entered data per se, it’s
patient gathered data, and there is an indicator on each data element as to
where that data came from. And the patient can apply privacy flags to mask the
data, but they can’t change the data. So that is if we get a data stream from
Google Health and it says oh here are twelve medications and the source is
Walgreens, we’re going to actually take that as highly reliable data. So sure
there can be patient entered data that has the same problem as with our patient
PHR tethered to our EHR could be misspelled, could be inaccurate, but hey if I
look and I say it’s not patient entered specifically, it’s just patient
gathered from multiple data sources, that’s actually pretty interesting and
relevant and helps us in things like medication reconciliation. And yes it may
be incomplete, but getting aggregate data from the patient from multiple data
sources is a whole lot better than nothing.
Health Vault, you know, a similar model where the patient can aggregate data
from multiple sources, especially as I mentioned the Continua Alliance device
data, and then can transmit it to us, can share it with us. Both Google and
Microsoft are working on novel data sharing kinds of activities that I think
over the next year this notion of patient aggregated data, from multiple
trusted data sources, being then shared with hospitals, quality measurement
organizations, public health entities, with patient consent, is actually going
to be a really interesting novel way to gather information that certainly is
much richer than administrative data sources.
One other quick comment about Google, Google Trends, if you go to Google you
go to More and then you go to Google Labs and go to Google Trends, you’ll see
that you can actually look in really not so much a quantitative way, more of a
qualitative way at every search that’s going on on the web, so you can actually
put in the word “fever” and look at how many people over time are
searching on “fever” and where they are. This is an interesting way
if you want to use this as an early indicator for what the flu season may be
doing, outbreaks of certain types of diseases, Google Trends may be giving you
sort of a proxy and a kind of innovative data source to see what symptoms
people are searching on, and therefore it actually may be a kind of
Locally we have developed something called “Aegis,” developed by
Ken Mandl, he is probably going to chat about that, but it’s a mechanism for
gathering data from all of our emergency departments, and that being able to
look at influenza outbreaks. This is available on the web publicly at our
Catalyst site, at catalyst.harvard.edu.
Also one aspect of Catalyst, this is our clinical translational science
awards site all publicly available, is we’ve created some unique data mining
activities. This particular part of the Catalyst idea is password protected,
but if I move onto slide 5 where we have built XML filters on top of all the
clinical data sources in all the harbor hospitals, and with IRB approval for
some types of queries you can simply say hey Mass General, Brigham, Dana
Farber, Joslyn, Harvard Pilgrim, Beth Israel Deaconess, tell me some
interesting questions. I happen to show you one on this next screen, I want to
know what are the age distributions of all asthmatics and all obese people at
all hospitals associated with Harvard, go, and it doesn’t actually store all
this data centrally, it’s just a highly distributed federated query using XML
exchange to then get a nice result that in this particular case didn’t actually
require the use of patient identified data, it was all done within the
firewalls of the institutions, and then the results transmitted back.
If you look at the last slide what you can see is here is a real time
analysis across 17 different locations of the age distribution of these two
diseases, which show you for example that asthma starts young and then you see
peaks and valleys and then these people actually tend to live for a fairly long
time. But hey obesity that actually doesn’t start until you’re 15 to 20, and by
the way all these people die, and they’re rapidly starting to diminish in
numbers by age 65 and above. So this is example of how to use again existing
data sources in novel ways using a community federated construct.
In summary, I think we’re starting to see in our region quite a lot of
interesting non-traditional approaches to gathering data and measuring quality,
and I certainly do believe the PHR, especially the aggregation of data by the
patient and sharing of that data, and the use of device data are going to be
very important to emerging novel sources for quality measures.
Happy to answer questions.
MR. REYNOLDS: Thank you. As usual, very rich but concise. Thank you very
much for that, and again, congratulations on all the things you’re doing
nationally. You’re making a difference. Leslie.
DR. FRANCIS: I’m the co-chair of the Privacy and Security Subcommittee, and
I wanted to ask you about patient aggregation and also actually about patient
entered information. Part of our goal in this is to learn about what we ought
to be thinking about over the next while here at NCVHS. And I wonder if you
have any comments for us about what your experience has been, what novel
developments you see, what questions you see that we might want to be thinking
about with respect to privacy and either patient entered data or patient’s
ability to control aggregation. Any new things on the horizon that you think we
ought to know about?
DR. HALAMKA: Absolutely. I was on the Google advisory council, Paul was too,
and we put together a set of policies around Google Health that ensured that
since HIPAA doesn’t cover Google or many of these PHRs, right, they are not
covered entities, you want policies that are stronger than HIPAA. Of course,
with all the new Stimulus package and the new comments about privacy we might
actually start seeing extensions of these guys, but they’re not really business
associates, but maybe not, but we’ll see. But you definitely want to say oh
Microsoft, Google, you got a policy that the patient is in control, we’ll never
advertise based on this data, we’ll never share the data with any third party,
we’ll never be mined, and if you delete it it’s gone. You the patient are
totally in control.
So you build that level of trust and then if the patient decides hey they’re
feeling altruistic today and they would like to share their data in an
aggregate way with a quality measurement organization, they just like you can
on Google sign up for any third party service and say hey oh I see there’s a
quality measurement organization that is going to provide value to humanity,
and by the way may give me feedback on some things to talk to my doctor about,
and I control it, I’m going to sign up for it, oh and at any time I can un-sign
up for it, and absolutely ensure that not only does the primary site like
Google or Microsoft adhere to all these best practices, but all third party
value added providers or external partners adhere to the exact same policies.
Today what happens is Google actually goes out and validates before they
even allow a third party to link to their site that they adhere to all those
same policies and best practices, and if for some reason they ever violated
trust Google would dump them immediately. So it may be, given that we seem to
be the country expanding HIPAA beyond covered entities to now actually business
associates are going to conceivably have the same civil penalties, I think
there has been a kind of interesting movement to suggest maybe the Googles, the
Microsofts, these PHR vendors, should be held to some similar standard. They
are doing well today, they are self-policing, they’re coming up with policies
that are as good or better than HIPAA, but maybe we actually need to be a bit
more proscriptive about that.
MR. REYNOLDS: Justine.
DR. CARR: Hey John, great presentation, thank you. Nice to catch up on
what’s going on.
The uptake on the Google and Microsoft products, can you say a little bit of
that. I realize there are only about not quite a year old, but what percent of
people have been signing on for that?
DR. HALAMKA: Sure. The current statistics that I have, just to give you guys
a background, the patient site PHR that is tethered to our HER has been around
for about eight years, we have about 40,000 monthly users of that, and the
users come and go because there’s chronically ill people, they are on it for a
while, the people with acute disease, may say hey a PHR provides great value
for me while I’m getting acute care treatment but hey then I’m better, I don’t
use it very often, so I let my password lapse. So of the 40,000 monthly users,
about 5,000 currently share their data with Google Health, and about 1,500
share their data with Microsoft Health Vault.
The difference between Google and Microsoft is I think twofold. One, we have
been live with the Google site since May 19th of 2008, so we’re
coming on almost a year. The Microsoft I think only went live in fall, so it’s
not been up quite as long. Also the Google product tends to be very patient
friendly, it’s a user interface that is a full PHR, whereas the Microsoft
Health Vault is more of a storage container where the idea is you put all the
data there and then third parties will develop cool applications that you can
share your data with, but those cool applications are still in the process of
being developed, so I think some users are not having quite as rich an
experience with the Microsoft Health Vault yet as they do with Google, and that
may be why there’s just a little less uptake.
So far, just to tell the group, I have had a very positive experience with
both organizations. It took a matter of days to build the interfaces to Google
and Microsoft, they are very straightforward web-based, based on some stuff
that is pretty close to HITSP standards. In the case of Microsoft it is HITSP
standards, and in the case of Google it’s CCRG, a Google specific flavor of the
continuity of care record, but straightforward interface, very reliable, has
not had any issues with security, and have had no support issues whatsoever. Of
the thousands of people who are using them I had one support call, and that was
related to a gentleman who was running Windows 98 Second Edition with Internet
Explorer 4, the java script wouldn’t work. So that’s it. I mean it’s been a
very, very good.
MR. REYNOLDS: That person may be closer to reality than the rest of us.
DR. CARR: Just on follow-up, John. Of the 5,000 of patients that have signed
on, do you have an idea of how many have remained active after initially
DR. HALAMKA: I don’t have the exact statistic, because we look at to
determine active users is the number of data transfers that take place. So when
I actually said 5,000, actually that’s the number of data transfers I’m seeing
each month. The way that just to briefly technically the way that Microsoft and
the Google products work is there’s both pull and push, that is the patient can
sign up and pull their data in, and there is the technical capability then for
once the linkage between the hospital or the pharmacy or the lab is done those
data creators can push updates. We haven’t actually implemented that piece yet,
so anytime a patient wants to update their Google health record they re-pull
the data, so we’re seeing about 5,000 pulled a month, 5,000 users pulling their
data a month. There seems to be pretty consistent use, but I don’t have exact
statistics on who’s come and who’s gone.
DR. CARR: Thanks, John.
DR. SUAREZ: John, hi, this is Walter Suarez.
DR. HALAMKA: Hi.
DR. SUAREZ: A couple of points I wanted to make I guess on the privacy side.
HITSP has an actual harmonized standard for managing, capturing patient consent
directives, so I think that’s something that it would be helpful to bring up to
the attention of the Privacy and Security Committee.
The other thing is on PHR interfaces, John, that you have, I think you
mentioned this or might not have mentioned this but certainly you are one of
our poster boys for HITSP, and your organization is a poster organization for
HITSP, so you’ve been implementing those, but we do have also a harmonized
interoperable exchange standard with PHRs. Is that something you are
implementing in there as well?
DR. HALAMKA: So the Google folk have committed to adopt the continuity of
care document and all the HITSP standards around EHR PHR interfaces, but their
logic, which is reasonable, was they were over two years ago when they started
this project HITSP hadn’t yet had a recognized standard in that regard so they
went with the CCRG, their own flavor of XML they thought would do the job.
Microsoft was a little later, the work that HITSP had done was ready and
therefore the Microsoft folks have fully implemented the continuity of care
documents which we call the C32 clinical summary between EHRs and PHRs.
When we look at the way the data is going back and forth to Walter’s point,
we have a document called the TN900, which talks about a whole variety of
secure transport mechanisms to protect security, and the standards that are
being used by both Google and Microsoft are completely consistent with those,
although they haven’t gone as granular as we have with regard to some of the
things like consent access control, the way in which the data is going over
HGTPS, using soap envelopes, that’s very, very consistent with the HITSP stuff.
I think the HITSP work is getting some traction from PHR vendors, and you’ll
in the interoperability showcase at HIMSS this year quite a lot of
demonstration of exchanges of data using HITSP standards from Continua on the
device side into EHRs and PHRs, and EHR to PHR exchanges. So we’re getting it
MR. REYNOLDS: Yes, Matt.
MR. QUINN: Dr. Halamka, this is Matt Quinn from AHRQ. Thank you for your
I had a question about the Social Security Administration project. It seems
like the special sauce in that is in taking clinical data and turning it into
– and having that standard vocabulary and that allows the query ability
and the analytics. First, can you talk about the process for doing that
transformation, how applicable it is or necessary it would be to do it on a
broader basis, and then what that portends to doing for secondary use of data,
and analytics on a regional or national basis? . Thank you
DR. HALAMKA: Sure. We have spent a lot of time in all these various projects
that I’ve mentioned to you, trying to create some magically interoperable data
exchanges because the data is really only useful for quality measurements,
population health measurements. If you’re using for labs, Rx Norm for
medication, SNOMED CT to describe problems, the problem is that a lot of the
underlying data sources actually don’t use these nationally standard
vocabularies natively in their applications. We end up doing quite a lot of
What we have done, for example, in this Shrine example where I talk about
how we aggregate data across 17 Harvard hospitals is they’re actually mapping
tables at each local node that can map the local vocabularies into a common
vocabulary that is used across the Harvard system. And that has taken a fair
amount of work to develop.
We do try to leverage all the work in the National Library of Medicine, and
that source wherever we can, but it is still I would best describe it a work in
process because the vendor-based systems that many of us have implemented
aren’t natively using the vocabularies that will make the semantic
interoperability easy. So our best short-term activity is to just do the
mapping, knowing that that’s imperfect, and hopefully over time we’ll start
capturing the data in standard vocabularies natively.
MR. QUINN: Just a follow-up. It is extremely important for individual health
obviously to have a high degree or a perfect degree of accuracy, as high as
possible for that. But looking at population health especially on a bigger and
bigger basis, what is the applicability of natural language processing or less
precise methods where there is latitude for wiggle room on something like that?
DR. HALAMKA: We have had mixed results with natural language processing. Let
me give you an example. Suppose I have a discharge summary that says, “The
patient always uses alcohol swabs to clean the skin before they inject
insulin.” Well, the NLP system sees, “The patient always uses
alcohol” and immediately declares them oh this is a person that has a
substance abuse problem.
Certainly you can imagine mapping that is not totally accurate between ICD
and SNOMED, or representative NDC code because we don’t know the exact
medication that they’re on but hey we’ll give you an NDC code for that
medication, it’s close enough. That kind of thing seems to be okay. The NLP,
yes, there’s certain kinds of data mining you might do to help identify
patients of interest that then require manual review. At least in the work I’ve
done today the NLP has not been good enough for us to do – for example.
That’s definitely an area to keep pursuing.
MR. REYNOLDS: I want to thank you so much. Again, thanks for being very
prompt and for spending your time and continuing to drive this and helping us
understand the subject a little better.
DR. HALAMKA: Well, absolutely. And I live by email should any of you guys
have questions. And what I did to prep for this testimony is I wrote a blog
about it, so everything I’ve said is actually on my blog,
geekdoctor.blogspot.com. And I also included links to several of the systems
that I described, so you can actually look at them yourself.
MR. REYNOLDS: Great, John, thank you so much, have a good day.
DR. HALAMKA: You too, thank you, goodbye.
MR. REYNOLDS: Okay. That completes the discussion of that subject. Both
Matthews up here, thank you for enriching the subject quite a bit.
Do the gentlemen from the Populations Subcommittee have a letter? If we can
get that out so we can take a look at it. It’s true you guys have a hearing
tomorrow, so whatever we discuss today you’ll be able to work on some tomorrow.
What I would like the Committee to do as much as anything, we’re going to
look at it as a group today to see if there’s anything anybody recommends. They
can look at it again tomorrow as a group if they have to, because they’re going
to work on it. Then whether or not we send it to the Executive Committee and
what we do, or do we want it out today?
Okay, so this has to be voted on today.
MS. GREENBERG: That doesn’t mean it can’t be fine-tuned tomorrow.
MR. REYNOLDS: Wordsmithed, okay.
MS. GREENBERG: I think it probably should be sent on Monday.
MR. REYNOLDS: Okay.
DR. HORNBROOK: The first sentence the word “upcoming.” The
President isn’t upcoming; it’s the health reform plan that’s upcoming. English
is very strange is about where you put adjectives and how it could be written
lots of different ways.
MR. REYNOLDS: Who is going to deliver the letter from the Committee? Yes, we
need to read it, that’s the way we used to do it.
DR. SCANLON: This is addressed to the Acting Secretary. It begins:
“The American Recovery and Reinvestment Act of 2009
(ARRA) and President Obama’s Health Reform Plan recognize the critical
importance of healthcare to our national well-being and the economy. The ARRA
substantially increases our investment on health IT, comparative effectiveness,
and prevention and wellness initiatives.
To improve the population’s health and to increase the value and
efficiency of healthcare provision, it is essential that we have the capacity
to measure the return from these investments in terms of quality, safety,
efficiency, equity, accessibility, and longevity, both to assure accountability
as well as to guide future steps.
To achieve this capacity, NCVHS recommends that adequate funding and
resources be made available,”
And this is the first change I would suggest, since we were doing that on
the fly we started focusing on NCVHS and then we need to expand that focus, so
we’ll drop the two NCHS and instead it will read:
“To achieve this capacity, NCVHS recommends that
adequate funding and resources be made available to assure a robust vital
statistics and population health information infrastructure that can produce
rigorous, unbiased, statistical metrics of the nation’s health and
In particular the Committee recommends that the funding be allowed from
the Recovery Act to the following two areas:
Number 1, E-Vitals. An electronic vital records data collections
processing and analysis system is at the core of a fully functional Nationwide
Health Information Network. The National Vital Statistics System (NVSS) is the
oldest and most successful example of intergovernmental data sharing in public
health. Each year close to 7 million births and deaths are collected by 57
jurisdictions across the nation and provided to the National Center for Health
Statistics. Investing now in electronic birth and death record systems to
support the implementation of modern information systems and make them
interoperable with electronic health records in hospitals and clinics that are
part of the NHIN, will be an important way to achieve the goals of the Recovery
Act. Vital and health statistics projects are ‘shovel-ready’ and federal
contracts already exist with states and other entities that can be used to
ensure a speedy implementation startup.
Second. Population-Based Surveys. One of the most critical ways to assess
the health of our population and the effectiveness of the healthcare provided
is through the national health surveys being implemented by NCHS, AHRQ, and
others. Over the past five years we have seen a reduction in the capacity to
achieve these important efforts. It is not sufficient to reestablish our
previous capacity for population-based surveys. We believe that effective
monitoring requires an expansion of these efforts so that adequate information
on vulnerable populations and states and localities will be available for:
Evaluating the impact of health reform and the role of health IT on
communities, health disparities and the uninsured, for example, through the
Health Interview Survey.
Baseline and monitoring data for comparative effectiveness, for example,
through healthcare surveys and improved vital records data.
And finally, monitoring impact of EHR adoption on the overall health of
the nation and achieving health goals, for example, through the Health
Interview Survey, Health and Nutrition Examination Survey, the National Survey
on Ambulatory Care and Medical Expenditures Panel Survey.
Without the additional funding from the Recovery Act the nation is at
risk of being at the beginning of dismantling the national vital statistics
system, which has been the backbone of public health data for decades and
having inadequate population-based information.
While funding from the Recovery Act is essential, it should be seen as a
down payment to having the capacity to effectively monitor health and
healthcare in the nation. NCVHS recommends that stable adequate funding for
vital statistics and population-based information collections needs to be
We believe these recommendations are important to building the core
infrastructure for promoting and assessing progress in health reform.”
MR. REYNOLDS: Comments. Justine.
DR. CARR: Nice job, very nice job. The only observation I would have is
we’ve been talking about this funding for a long time, and now that we’re
putting it in this letter shouldn’t we have a little bit of background about
what we’ve been observing and what we’ve been talking about, so that it doesn’t
look like oh we heard there’s money, let’s make one of these. It’s really been
an ongoing deliberation over the last year.
DR. SCANLON: Right. I would agree, I think if it’s possible we could talk
about it, like we noticed the decline over time and we’ve been monitoring it.
DR. CARR: Right. We’ve had discussions, concerns, deliberations over it over
the last year.
MR. REYNOLDS: We have Marc Overhage, you had your hand up, right? And then
DR. OVERHAGE: I guess my first question or comment is that while this is
obviously an important opportunity to take advantage of it, it seems like the
Stimulus dollars are one-time dollars, they’re capital investment, and we have
to somehow I think weave into this how we’re going to leverage those as capital
investment not as operating sustainable funds, because I don’t see how that’s a
successful story to tell if we just say like you said, a one-time investment
and then it goes away and then we don’t know what to do with it afterwards.
So I’m not sure exactly what it is we could identify in here, what the work
is that is the equivalent of building a bridge, that we can use for the next 30
DR. SCANLON: This I would look to others to partially answer this. My
understanding though is there’s some opportunity for some operational
activities under the Stimulus Bill as well as for capital investment. Because I
think what we’re trying to do is to say for the short-term there can be
Stimulus money used, but the last paragraph to deal with the fact that this is
not going to be enough longer-term, that we do need to have the operational
funds in sufficient quantity to be able to make this an ongoing activity.
MR. REYNOLDS: Blackford.
DR. MIDDLETON: I just wanted to underscore Marc’s comment, which I had as
well. And secondarily suggest that it would be useful to try to ballpark what
exactly we’re asking for in terms of how much and from which source
specifically in the funding package we’re asking it to come from.
MR. REYNOLDS: Then it’s also because AHRQ has some of this money too.
DR. OVERHAGE: Just to finish my thought I guess I was specifically trying to
get to, I’m not sure, in the first paragraph on E-Vitals, I think you’re
suggesting there’s something to create here with Stimulus dollars. I think
underscoring that there’s an activity to create something that will then cost
less to operate and run over the long term, and potentially make some of the
other activities that we want to carry out less expensive over the long run,
making it a kind of rational investment. I don’t know if, for me anyway, that
didn’t come through in this paragraph.
The second one, I find it a lot harder to make that story in the surveys,
because I don’t know this field, you guys know this in and out, but that’s
people time, mailing costs. Is there anything there that we can think of or
develop that if we spend making an investment now, doing those surveys in the
long-term basis, a more sustainable, more achievable goal? I mean this feels
MR. REYNOLDS: You are talking about creating an infrastructure that would
allow better – no, I mean those are the kinds of –
DR. OVERHAGE: If there is such a thing, I think that’s easier –
DR. SCANLON: That could be an idea, but I think the point we’re trying to
make here is that it’s very short-sighted to make – not just through the
Stimulus Act, but what we’re talking about in terms of health reform, and not
to be monitoring what the impacts are, because we’ll have no idea then how huge
amounts are being spent in terms of their effectiveness.
MR. REYNOLDS: Walter.
DR. SUAREZ: You might want to point out too, and it’s been said a couple of
times, I think the Stimulus package is a very short-term geared two-years at
most investment. Then really there’s a need for ongoing support. And what we
try to do is to emphasize those two points was that the initial investment is
for infrastructure, for example, the E-Vitals will be to help states implement
this electronic birth and death records system. And we certainly do the link
and the importance, I like importance of being able to do that so that it can
be linked to the electronic health records systems that are going to be
implemented over time. But clearly there is a need and potentially a follow-up
with some more information from the Committee or Subcommittee, a need for an
ongoing funding, identifying ongoing funding for several of these activities.
MR. REYNOLDS: I know there’s a couple of other people, but let me make a
couple of things. When you get to the next to last paragraph, the stable, and
maybe add “long-term adequate funding.” And then
“dismantling” makes me quite nervous; “diminishing” might
be a little better.
MR. SCANLON: Or “deteriorating.”
MR. REYNOLDS: Yes, something like that. Deteriorating, yes. So I got Carol
and I got Leslie.
MS. MCCALL: In the second paragraph, I agree with the comment to have the
capacity to measure the return from the investments in health IT. I would
strengthen it personally to something that says to just completely realize the
return of that investment. It’s not just simply measuring it. That there’s
something vital that is missing, and it’s going to come from this enterprise
for vital and health statistics for the 21st century that we talked
about this morning. So it’s not just continuing the same store, there’s going
to be not just a difference in degree but a difference in kind of the type of
information that’s available and what it’s going to require to actually extract
I would want us to strengthen what we say we want it for.
MR. REYNOLDS: Leslie.
DR. FRANCIS: I’m going to put my brain on the side of a critic, and this is
just a thought, I’m not sure I can say it in the clearest way. People
distinguish between overall health status and the quality of care and access to
care that people are getting. Suppose somebody thought that if what we did was
billed all the health EHR and NHIN infrastructure maybe we don’t need our vital
statistics collected in the ordinary way at all, we can just go with the NHIN.
I know, shock, horror. Nobody wants to go there, but what I think this letter
needs to do so critics won’t go there is make it really clear that it’s
important to collect the traditional kinds of data that we’ve been collecting,
as well as being able to link it. And it wasn’t so clear to me that maybe in a
couple of places in this letter it just sort of blends the two, rather than
saying how important it is to have both.
MR. REYNOLDS: Did we do a letter on the funding last year? We talked about
MS. GREENBERG: Maybe a few years ago, I don’t know, it would be on the
MR. REYNOLDS: The only reason I’m saying that is again it’s looking like
we’re jumping onto something that just happened. If there was anything we’ve
said along, it would be great. In other words, we’ve been looking at this an
noted in our letter of such and such, and oh by the way now that there is
funding, because otherwise everybody’s running to the light right now. So if
there’s anything we can pull out from something we’ve done before that allows
us where we’ve talked about this subject, it shows a good trend, it shows this
isn’t like a brand new subject, that oh by the way it’s been on the back
burner, and oh by the way there’s now money, so we’re not bringing up a new
discussion we’re just saying fine this now makes it even more important. So I
think that would give a good trend line, rather than there’s money, hot dog,
there we go.
Jim and then Mark.
MR. SCANLON: I think my own advice to the Committee is that you don’t need a
lot of subtlety and distinctions here. The folks who understand it will know
it, and the ones who don’t won’t matter. The idea here is that at the very time
that there will be an increased need for monitoring or evaluating, we can say
what it is, but it’s the health status of the population and the functioning of
the health system, and the Committee’s become aware that the resources are at a
critical stage. You could say ironically and therefore the Committee urges and
recommends that, and you don’t have to identify specific parts of the Recovery
Act, just say appropriate provisions or so on.
Here you are just trying to get across the point that at the very point when
all these systems changes are envisioned the systems we use to monitor, and
again, this is not electronic health records, this is population-based data
collection, to evaluate those changes and health status and so on, are in
jeopardy because of resource constraints. Therefore, the Committee recommends
that. I think otherwise you have pretty much all of it here. And we can use it
within – it would just be another way of getting another voice in
perspective to this issue hopefully.
DR. HORNBROOK: I agree with Jim exactly. I was just going to add in the
notion of a transitory phenomenon. If you don’t do it now we’ll never see it,
so you’ve got to measure it now.
MR. REYNOLDS: Ed.
DR. SONDIK: I was going to make the point that Jim did.
MR. REYNOLDS: Any other comments from anyone?
DR. SONDIK: And I would just add to the question of can we use the one-time
money, there is absolutely no question that we can do that. This works around
on the population survey side that can be used to actually do surveys, and that
provides an information base even if the money went down, that base is a lot
more information on minority populations, for example, is there. So that is
just practical and really valuable.
The other on the E-Vital side, would be a major down payment for changing
MR. REYNOLDS: With all this discussion and not having rewritten the letter
is there anybody on the Committee that has, one, a concern about sending such a
MR. REYNOLDS: Two, about the fact that we agreed with the intent that we’ve
seen written down, plus the intent that Jim and others have added to it, along
with Ed, and so on?
DR. FRANCIS: I was going to move that we accept the letter subject to
MR. REYNOLDS: That’s where I was headed. Okay. I’ll stop my comments right
here and let that motion ride around. Okay, would there be a second? I see a
number of them, Larry. We’ll go with Larry.
DR. GREEN: [second]
MR. REYNOLDS: Judy, a comment?
DR. WARREN: From listening to what we’ve said I think it’s more than
wordsmithing we’re talking about on the letter. If we follow Jim’s comments on
MR. REYNOLDS: I guess the question is, Jim, with what you’ve said is the
requirement to point out E-Vitals and population-based surveys necessary? I
think that was the question. That’s kind of what you’re saying?
DR. WARREN: I guess I was asking Jim what he’s asking for, if you see that
as wordsmithing or as do you see it different than what’s in this letter.
MR. SCANLON: No. Again, I think there was a suggestion on the table that you
provide more detailing in the source of the funding, and I don’t think that’s
necessary for the purposes here. So, no, that’s fine. I think it’s fine the way
I would probably give one sentence to make the transition between I think
the third paragraph, then the resource concerns that we have, therefore the
Committee recommends that. As Ed said, that money can be used, obviously you
can’t depend on it for long-term operational purposes because it won’t be
there, but it certainly can be used for these kinds of purposes.
MR. REYNOLDS: So, Bill and Don, have you noted that? Ed, did you have a
DR. SONDIK: I was going to say the point that Jim made I thought could
actually come maybe at the beginning of the second paragraph, which is to link
to put at the beginning of that that there is this increased emphasis on
prevention wellness, understanding the health status of the population, and it
also could be a point made that this Committee has been continually concerned
about that, maybe that could come first. Now, at this point, with this
increased emphasis, and even throw in health reform, although the bill I don’t
believe mentions health reform.
MR. SCANLON: This morning the 2010 budget includes the health reform.
DR. SONDIK: Okay. So I think that is the point where it could be tied
together, that these provisional resources to meet the need that is clear, and
is expressed by this major initiatives.
MR. REYNOLDS: Is there any other discussion?
MR. REYNOLDS: All those in favor of moving the letter forward?
MR. REYNOLDS: Opposed?
MR. REYNOLDS: Okay. What I assume is that double-checking is that you guys
will clean this up tomorrow, we’ll get a copy of it. Then Jim will give us all
the right places to send it.
With that, is there any other discussion at today’s meeting?
MR. REYNOLDS: I would like to note that we are adjourning 30 minutes ahead
of time, and I think we got a lot done and a lot to learn.
So thank you very much, and everybody take care.
(Whereupon, the meeting adjourned.)