[This Transcript is Unedited]
NATIONAL COMMITTEE ON VITAL AND HEALTH STATISTICS
Joint Meeting of
Subcommittee on Population Health
Subcommittee on Privacy, Confidentiality, and Security
Subcommittee on Standards
Joint Roundtable on
Health Data Needs for Community Driven Change
May 1, 2013
National Center for Health Statistics
3311 Toledo Road
Hyattsville, MD 20782
CASET Associates, Ltd.
Fairfax, Virginia 22030
TABLE OF CONTENTS
- Review work plan for Day 2
- Panel 3: Using Data to Promote Community Driven Change
- Len Nichols, Population Health
- Carladenise Edwards, PHD, Senior eHealth Advisor
- Sharona Hoffman, JD, Professor of Law and Bioethics, Case Western Reserve
- Donald Malec, PhD, Mathematical Statistician – National Center for Health
- Report of Small Group Discussion 3 – Vickie Mays, Population Health and Privacy Confidentiality & Security
- Panel 4: Recommendations for Supporting Health Data Needs
for Community Driven Change
- Walter Suarez, Co-Chair Standards
- Denise Chrysler, JD, Director, Mid-States Regional Center, Network for
Public Health Law
- Christopher Fulcher, PhD, Co-Chair, Center for Applied Research and
Environmental Systems University of Missouri (Community Commons)
- Denise Love, BSN, MBA, Executive Director National Association of Health
- Report of Small Group Discussion 4
P R O C E E D I N G S (9:00 a.m.)
DR. FRANCIS: I think there are a couple of people here who were not here
yesterday, and I would like to make sure to welcome then. Jim Scanlon and
Vickie Mays. Is there anybody else here? Yes, Walter Suarez. This is Standards
coming to join – the third subcommittee that is partnering in this is
Standards. Without more, I am going to turn it over to Lee, who is going to
give a summary of where we got yesterday. That can start us off. Then I will
say about three things and then turn it over to Len who is going to run this
morning’s panel. Lee.
DR. CORNELIUS: Good morning. I am glad that we all made it through last
night, and for those who have moved on, we say prayers for them out there. I
promised you all that I would create an abstract.
Basically, my self-inherited was to capture what took place in both
plenaries, and to synthesize across the three breakout sessions. Then really,
we will ball it into a short abstract. The biggest challenge was that there
were so many wonderful things that took place yesterday. I think that when we
come back to the next steps, we will rely on the transcripts and the notes from
the breakout room to kind of filter back into this. Please do not feel that
this summary is the be-all and end-all.
The second point to think about is as you go through this. I was originally
was go do the summary and then step back and ask the group if there was
something in here that you think really needs to be added to the summary that
would help us move forward? As you look through the page and half, and it is on
both sides, just ask yourself was there something on the breakout groups where
I couldn’t quite physically be present. Where there something substantive that
you really think we need to keep in mind today.
There are a couple quick comments. I will talk about the breakout sessions.
What I did in addition to going through what came out of the groups, I asked
myself how do I think of a bold print title that would serve as a summary for
what the issues are? The lead-in, I put in bold. You also notice on the first
page at the bottom. Other issues, I have to admit, were really the richness
that came out of the red group whereas all of the other items was across all
Probably, one last quick thing just to kind of point us to the beginning
here, there was a lot of discussion about trying to understand communities as
an organic being. In essence here we are talking about data use for its
purposes. Yet, the discussions resonated on the fact that it is like the social
return. You start where the client is at.
With communities, the idea is it is nice for us to think about the planning
tools and the data systems, yet in order to get them on board, we have to
listen to the fact that the prior for the community at today’s town hall
happens to be this. If you really want to get to where you want to get to, you
have to build a trust around what their immediate priorities are. Part of our
discussions may be both thinking about what our role is in the federal;
advisory capacity and the real issue of community life.
Then the other piece I will quickly mention is when we filter through this,
the mantra of putting data to practical really resonated throughout today. That
means both in terms of actionable research and policy issues, and really what
technical assistance is needed so that the consumers can own the data and it is
part of what they do. I will step back, I promise to only give a summary of the
summary since it is all in front of you.
DR. FRANCIS: Thank you very much. So just for everyone, the work plan for
today is to continue as on the agenda. There will be a set of word documents
that represent all of the various report-outs that the groups will have for the
last discussion so that we will be in a position to really build by the end of
the day on what has been happening yesterday, and what will happen today. I
also want while everybody in here, to take the opportunity to thank the staff
for organizing so amazingly. Thank you guys. You made it work. Without more,
Len, you are on.
DR. NICHOLS: Thank you. It is my distinct honor and high privilege to
introduce what I am sure will be an even more interesting panel than what we
had yesterday, which was fantastic. There are three members; Carladenise
Edwards is a Senior eHealth Advisor for Zerodivide. Sharona Hoffman who is a
professor of Law and Bioethics at Case Western Reserve, and Donald Malec who is
a statistician at NCHS, who will talk to us about small area estimation. So,
MS. GREENBERG: Why don’t all the members just introduce themselves and
whether they have any conflicts.
DR. FRANCIS: I am Leslie Francis. I am a member of the Committee. I have no
DR. KLOSS: I am Linda Kloss, a member of the Committee, no conflicts.
MS. MILAM: I am Sally Milam, a member of the Committee, no conflicts.
DR. COHEN: I am Bruce Cohen, a member of the Committee, no conflicts.
DR. MAYS: I am Vickie Mays, a member of the committee, no conflicts. Good
morning, Vickie Mays, University of California, Los Angeles. I am member of the
Subcommittee on Population and a member of the Subcommittee on Privacy,
Confidentially and Security.
DR. CORNELIUS: I am Llewellyn Cornelius. I have no conflicts. I am the
member of the two committees, the major one and the minor one.
DR. GREEN: I am Larry Green, A member of the Committee, no conflicts.
DR. NICHOLS: I am Len Nichols, a member of the Committee, no conflicts.
DR. TANG: I am Paul Tang, a member of the Committee, no conflicts.
DR. SUAREZ: I am Walter Suarez. I am with Kaiser Permanente. I am a member
of Full Committee, co-chair of the Standards subcommittee and member of the
Populations, Health and Privacy and Security Subcommittee. I am not sure I have
any conflict across all of those, but I don’t have any conflicts.
MR. SCANLON: Good morning everyone. Jim Scanlon from HHS Planning and
Evaluation and the executive staff director for the Full Committee, and I have
DR. CARR: Justin Carr, chair of Working Group on HHS Data Use.
DR. BURKE: I am Jack Burke, member of the Committee, no conflicts.
DR. EDWARDS: So I am going to get started. It is really hard for me to stand
still, but I have an appreciation for the fact that this is being recorded and
so that I need to use the microphone. I really want to start with story before
we do anything else. Justine and I and were just in the ladies room. I was
chuckling because my story is about front page news. She was just saying what
we do isn’t front page news, but it is important. I picked up the paper this
morning, Wall Street Journal, for those of you who can’t see me holding up the
paper. The headline above the fold, “Housing Market Accelerates. Home
Prices Jump 9.3 Percent in Quickest Rise since 2006. Gains Seen Across the
Those of you who were here yesterday saw me texting and emailing and running
off for phone calls. My husband and I trying to get a house in California, not
buy one, just rent one. It has been two months of heck just trying to rent a
house in California. That is how hot the housing market is. Finally we were
approved after giving away our first born child and the dog and they approved
us for renting a house in California. Let me tell you, this article talks about
the home prices rising, the rate of price increasing 9.3 percent. Home prices
are changing in margins that have never been seen before. They are grass, they
are statistics. Below the fold, “Consumer Confidence Rebounded in April.
The Conferences Boards’ Index Rose 6.2 Points from March’s Level of 68.1,
Driven Mostly By Strong Expectations for Later this Year.”
The figure has been volatile in recent months amidst Washington budget
battles and higher gas prices. One worry is that consumers reported being less
optimistic about finding jobs. My dream and my vision is that above the fold,
it reads, “Healthcare Quality Increases at Exponential Rates. Obesity is
Down. Under-Served Communities no Longer Disadvantages by Health
Disparities.” Wouldn’t that be cool if we had a headline that looked like
that? Are we even measuring the quality of health and the rate of change
between and among our communities in the same manner and with the same
vigilance that are measuring the housing market? This is not happening.
What this committee means to me is that there is actually and interest in
making front page news things that we are interested in which is improving the
quality of health through the measurement of vital and health statistics. The
work that you have cut out for you if I were in charge is great. Who am I? Why
am I here? I am Tessie Guillermo who was actually invited to speak, but I am
going to try to show you a little You-Tube video of here. I am a senior eHealth
advisor for Zerodivide which is the organization that Tessie runs.
Tessie is a rock star in the field of technology for under-served
communities, advocacy for Asian-Pacific Islanders and other minorities. She is
a huge philanthropist and sounding board for the community. I think that is she
or her organization was invited to be represented because you guys wanted to
get down into the dirt and into the gritty of community work. I personally run
a consulting business and Zerodivide is my best and favorite client. Hopefully,
none of my other clients are on the phone.
I also serve as a professor at University of Miami and the Miller School of
Medicine in the Department of Primary Care and Family Medicine where I help
them think about health policy and strategy and business development as well. I
met a couple of people who know me in my past life, so if you were confused and
thought maybe I had a twin, I just thought I would disclose that yes I did
serve as the commissioner for Georgia’s Department of Community Health for a
short time. I was chief of staff under Rhonda Meadows there. I actually oversaw
public health and our state register for vital health statistics. Those people
reported to me. Then I actually served as California’s HIE lead under the
organization called Cal eConnect.
Some of you actually know from when I used to work in Florida when I ran the
Medicaid Policy Unit in the State of Florida, and so I have been all around, up
and down and backwards, but I landed in a place that I think is incredibly
important. It is with an organization that has the community at the center of
all of its work. What Zerodivide does is transformation through technology. We
have three main areas of work. One is in eOpportunity, the other is eDemocracy,
and most recently eHealth. Basically, we look at ways in which technology
transforms communities. How do we use that to get the community to the goal
that they have defined for themselves? That is what we do.
Why are we here today? The key questions I was asked to address are listed
here, and I think they are also in your agenda. One is how to best promote
community engagement and participation. The other thin that we have been
charged with talking about today, and I won’t cover all of this in my
presentation. I am just going to give you some fire starters for the breakout
sessions. What role does government play in promoting and supporting change at
the community level? What technical and analytical resources can government
provide? Most importantly, how can we collaborate? How can we work together? I
will admit in advance of the biostatistician and mathematician that I have no
regression analysis or statistics in my presentation. I will save that for
Why is this important to Zerodivide? I think the reason that the work that
this committee does is so important to Zerodivide is because we see the
connectivity between individuals and between communities and actually between
nations. That connectivity is increasing exponentially as we speak, as a result
of the technology that is available to us in our hands, in our pockets, on our
desk, through out phones. There is more connectivity today than there has ever
been. Most of you all have heard the statistics about the number of people who
have cell phones. There are more cell phones than there are cars, than there
are houses than there is anything else.
Even those are in the underserved communities have access to communication
tools in ways that they didn’t in the past. That actually exceeds their
capacity to have access to other things, like food and water. It is easier to
get a cell phone in some countries than it is to get clear water in some
countries. If that is the case, which we know that it is, how do we transform
the use of that technology for something as simple as making a phone call to
know improving the inequity that exists between us in the arena of health? How
do we use those devices to help us achieve eHealth Equity, is what we are
calling it. It is really health equality, but through the use of technology.
How do we get there?
One of the reasons we talk at Zerodivide about eHealth Equity is because we
think it is a stepping stone to health equity. If the disparities in
utilization of these tools for the benefit of improving health continue, then
the technology will actually exacerbate the problem. We continue to build
technology in tools that don’t meet the unique needs of the communities we
actually think should benefit from them. We are not actually solving the
problem; we are making the problem worse. These devices and these tools have to
be culturally sensitive and appropriate, a term that I learned from Paul Tang.
They need to be universally designed, so people who have hearing
disabilities or visual disabilities can actually use them. They have to be
built in languages and at literacy levels that make sense for the communities
that we want to impact. We all know about the Triple Aim. We want better
quality efficiency, lower cost, and who are the high cost drives that really,
really need to impact. They are the under-served. They are the folks from the
community who have less access to everything except for what, the cell phone,
and the technology. We have to make sure that the technology and the
information that comes from that technology because that is were you guys stand
is actually compatible with the needs of the community we are trying to impact.
I just pulled up a couple of statistics because we talked yesterday a little
bit about eHealth and EHRs, and I just thought I would share with you some of
the stuff that I found when I Googled last night. I found the rate and
utilization of social media and technology in Hospitals. Right now 25 percent
of U.S. hospitals use social media, and they do that in the form of blogging,
You-Tube, Twitter and Facebook, and 84 percent use Facebook. I don’t know if
you guys understand the engine and the financial engines behind these things,
but it is all about the data. People invest in Facebook and Google because they
are investing in their capacity to collect data. Then they will use that data
to make business decisions.
This is a little bit of a slippery slope, and I know it is a little bit
controversial, but those business decisions drive the housing market, the
car-buying market. The forced me to buy Dora the Explorer and Spiderman
underwear for my kids because they get emails. Those emails come to me. When
are we going to figure out that we need to be using that data to drive the
healthcare quality and reduce the disparity in access that we are all
experiencing underserved or not. At some point, we have to; I think Denise said
yesterday, make public health data sexy. Release the data. Doctors, 60 percent
use social media to improve the quality of care delivered to patients, and 40
percent believe that there is actually no value. There is still some work to do
here in terms of utilization among physicians because haven’t all drunk the
Kool-Aid in this regard.
I think there are some good reasons for that. I think there is some solid
justification for not using social media today. I don’t know if they will
withstand the test of time if we do this right. One out of three healthcare
consumers report using social media to do the following: Share information
about their symptoms, seek information about medical treatment, track of
monitor health conditions, share feelings about a doctor, a treatment or
medicine –- one out of three.
Typically what you will hear for all the reasons why we can’t share health
information between us and among us is because the patients don’t want it. They
are scared of their privacy being jeopardized or their confidence being
violated. People are putting their private information on Facebook, but the
doctor is saying I can’t give the other doctor you are seeing your information
because I am worried about your privacy. There is incongruence between what the
patient wants and needs and what the healthcare industry is willing and able to
provide. Honestly, it boils down to out ability or inability to interpret laws
appropriately or to implement laws that are in the best interest of the people
we are trying to serve. That is where you all come in.
There is a huge disparity or difference in the use of this media when you
actually get down to the nitty gritty. It is really easy to say one in three,
but if you really look at from a demographic standpoint, you are going to find
some differences. What I found to be incredibly interesting is that seniors
actually have the highest rate of utilization of social media and other tools
to help manage their health. I thought about this, and there is a little bit of
data to support it and provide some explanation.
One is they have more chronic and comorbid conditions than the other
population. They are more likely to need to manage their health. Two is they
have time on their hands. There is something social about that. You can connect
with other people in ways that you probably don’t want to connect personally.
There is a level of privacy that comes with connecting with people on the
internet that you don’t get when you have a face-to-face conversation. Young
people, of course, are probably the least likely to use it, but we are finding
that young people really like to share their information and use it almost in a
way of getting some level of anonymity, particularly for teenagers who might
have questions about STDs, questions about things that they don’t want to ask
their parents and other adults.
I think I have time to do this three minute video, but I am going to save it
until the end just in case we have technical difficulties. I wanted you to hear
the voice of a couple of people. One is Tessie Guillermo who wasn’t able to be
with us today. The other is Terhilda from Kaiser Permanente. In the little
video, she does this clip about the disparate use of KP.org. I know in
California since I moved here, if you don’t know what KP is, you probably don’t
live in California. It is Kaiser Permanente. They think they own the universe.
They definitely own California. In California, at least in that market, you
hear thrive, and you hear all about the EHR on the radio, on the TV, on the bus
stops. They have the most robust EHR. I am not betting paid to advertise for
One of the things that is interesting in Terhilda’s talk is that despite the
fact that we have this system, we are seeing disparate use of the system. We
need to figure out why that is. What is happening that we are having disparate
use? You know why they need this system, not only for people, not for my
benefit, but it is for their benefit to have a better understanding of what is
happening with the populations. I am going to share that with you. I am going
to skip through that.
I will talk a little bit about what we have to do. The background I gave you
is basically summarized in this slide and one more slide. We have a huge uphill
battle. If we are going to change the percentage of people who don’t have
access to the health and wealth of this country, it is a huge uphill battle.
It is only going to get worse if we don’t figure out ways to leverage the
resources that we have before us and to use them in meaningful ways. We have to
make a concerted effort to figure out how we use technology to advance our
goals. We have to make a concerted effort to figure out how we amend and change
the rules and regulations that were put in place before the technology has
taken the shape and form that it has today.
As we think about how we release the data? How do we make the data sexier?
We have to think about why we are doing that. It goes back to the questions you
raised which started the convening or resulting in the convening. What does the
community want and what does the community need? We had the opportunity to do
two things this spring to find out what the community wanted and needed. One
was a summit at the Whitehouse where we convened some key stakeholders around
issues of eHealth and technology. The other one was in Oakland with groups of
folks who actually work in the field. What they told us the community needs is
that we need to develop better tools. If we are actually going to get people to
use the technology to improve healthcare which then puts us in a position of
having better data, we need to have better tools. They also said we needed
greater awareness of the importance in value of information, health
People don’t really get why it is important to participate in a clinical
trial. They don’t get why it is really important to monitor their blood
pressure using electronic device that plugs into their computer. If we can get
folks to understand why that is important and how that doesn’t just help the
individual, but helps the community, we will greater buy-in. The other thing
they said we needed is evidence of effectiveness. Not a lot of research showing
that there is actually a correlation between let’s say the use of Blue Button
and improvement in healthcare outcomes for a particular community.
We need researchers to actually get out there and start looking at how these
devices and the use of those devices and how data actually results in better
outcomes, so we then convince people that this a good idea and increase the
demand fro greater utilization and devices that make sense. We need money. We
need financial incentives. Yesterday, as we were having the conversation about
the why, I kept saying it really all boils down to money. The reason why we
need to have demographic data and information from communities about what is
happening in the community is because that is how funders make decisions about
the allocation of resources.
If you don’t have good data about the number of diabetics in your community,
if you don’t have good data about the inability to get fresh food in your
neighborhood, you can’t convince people to invest in your community. The need
for that data really boils down to the need for use to create financial
incentives. Then the thing that I thought was interesting. This is in rank
order. Government intervention was not the top solution. It wasn’t at the top.
Paul was there. He can hold me accountable. Nobody was screaming that
government needs to do something. They were a partner. They were at the table,
but it wasn’t the answer to the problem.
This is a little bit about Zerodivide’s approach in trying to find out what
communities need. I am sharing this with you because I heard a little bit
yesterday about how do we go about creating community governance? How do we go
about ensuring we have voices of people and their concerns as we develop and
craft protocols that enable us to collect data? How do we make sure that
communities are not adversely impacted when we share information maybe for
intents that were not initially posted? What we typically do is we start by
asking the community what they need. Then we ask the community what they think
the right answer is. Then we provide assistance and support as they craft the
solution that they think is in their best interest. Then we work through, over
time, ways to create an advocacy agenda, how we increase provider or user
adoption of particular tools and the patient or the consumer’s engagement in a
particular practice and then integrate that around the community.
Then we land with a greater adoption strategy for technology. I have eHealth
on this slide, but we do this whether it is eDemocracy or civic engagement of
workforce to get us to the point that everybody wants to get is to a healthier
community. For this coming year, the goals that Zerodivide has setout for
itself are to do policy and advocacy work that results in establishing
partnerships, doing some technology demonstrations and increasing education and
I really wanted to take a minute. It is three minutes. I don’t know if I
should take it now or see if we have time left over this video clip? I will let
you all decide. Is there a preference? Then we can talk about it in the
breakout sessions or at some other point. I am going to do that, and I am just
going to say thank you very much for inviting me and having me here to be a
part of this today.
DR. NICHOLS: Thank you.
MS. HOFFMAN: My talk is going to be a little more theoretical. These are
just some ideas I have actually written extensively about relating to data
collection and contribution. I think I sent some of the articles to Maya. Are
they posted somewhere?
MS. BERNSTEIN: We have a collection of things that people sent us, which are
posted. They are on the public website as a background to this meeting.
MS. HOFFMAN: So you can look at the website for background material, and I
will maintain the illusion that somebody at some point reads my work. If you
can’t find it, I am happy to provide it. The first concept I wanted to talk
about is the concept of the common good. The idea is that when it comes to
non-interventional research or public health initiatives, perhaps the
prevailing value should be the idea of the common good rather than notions of
privacy and autonomy. One thing to think about is that we have a free rider
problem. Everyone benefits from medical advances and public health initiatives.
If people are allowed not contribute their own information, they are free
riders. They get the benefit without making a contribution.
We have this idea already imbedded in public health precedents. For public
health purposes, we don’t ask anyone for permission. We do the reporting. We
report infectious disease. We report during public health emergencies. This
isn’t a revolutionary concept. We will only get public buy-in to this notion of
the common good if the public is fairly confident that their privacy will be
protected. We do need to continue to improve data security. All it is going to
take is one major breech where peoples’ information is misused or posted
somewhere very public, and we will lose public confidence. We do have to use
identity concealment techniques, which I will talk about in a minute.
We may actually need some more regulatory interventions. We have the HIPAA
privacy and HIPAA security rules, and they have been amended as recently as
2009, but they still apply only to a limited set of covered entities do not
include all research data base operators. If you are not a provider, an
insurer, a clearing house or their business associates, you are untouched by
the HIPAA rules. We do need to keep that in mind as we build more and more big
data databases and use them.
There is a very good question about the degree to which the public really
cares about privacy and we shouldn’t get hysterical about it. The Facebook
generation does post their most intimate details and Twitter them and whatever.
We don’t want major catastrophic breeches either. What do we do? Well,
de-identification is of course one technique. We already have large databases
of de-identified record. We believe that de-identification protects the public.
How do you de-identify? HIPAA spells out 18 elements that can be stripped from
data, and you get de-identification. There is a question about how valuable the
data is if you strip away all 18 elements. They did have research and public
health initiatives in mind. It is valuable at least for some purposes.
A lot of people will talk about a concern relating to re-identification. We
do have studies that show if you have a lot of time on your hands and expertise
and grants to finance this, and certain publicly available information such as
voter registration record, even with the 18 identifiers removed, there is a
0.01 to 0.25 percent risk of re-identification by skilled experts.
That is a teeny, tiny percentage with the 18 identifiers removed, but if you
think about 300 million records, that is already a significant absolute number.
If you don’t have all 18 identifiers removed, there is a much, much larger
chance of re-identification, and it is much easier to do if you have birth
date, zip code and things like that, as we heard about yesterday.
There is also another technique, and that is to use federated systems. We do
have some initiatives that use that. That is where a large number of
organizations agree to cooperate and to enable secure statistical analysis of
distributed databases. Institutions maintain control of the records. They
remain in their place of origin. They are not transported to a database and
de-identified and so on.
You then have researchers or public health officials submit queries. Those
queries are processed by a trusted aggregator who looks at he queries, puts
everything together. Researchers receive only summary statistics. They do not
receive any identifiable records or any individual records. This is a fairly
good technique that is useful for a lot of types of queries, not for all
The FDA uses in the Sentinel Health Data Network, which is used for purposes
of post-marketing surveillance of drugs and devices. It works well for that
purpose. Nevertheless, you are going to have advocates who say if you are going
to use peoples’ information, you need to have informed consent. You need to ask
people for permission. It is important to keep in mind that historically
informed consent has been concerned with experimental interventions. The
doctrine arose in the context of the Nazi experimentation, the Japanese
experimentation in World War II and the Tuskegee Syphilis Trial. If you look at
the major ethnical codes, the Nuremberg Code, the Declaration Helsinki, the
Belmont Report, it actually explicitly refers to experimentation, to things
being done to peoples’ bodies and minds, not to record-based reviews. That is
not in the background of the concept of informed consent.
Do people own their data? Is that a reason to ask for permission, to ask for
informed consent in every case? In the abstract, people do own their health
data. The problem is that we have a lot of case law that tells us as soon as
that health data is put into a physical record that the people who created the
record own the record. It is not that meaningful to own your health data
because as soon something useful is done with it, the healthcare provider owns
the record. We even have Supreme Court casein 2011. The Supreme Court refused
to recognize a constitutional right to control medical records. Individuals do
not have that right. Why might we resist consent for record-based reviews?
First of all, it can lead to a significant problem called selection bias.
Selection bias means that the subgroup of records that you are looking at is
not representative of the target population in which you are interested. You
can’t generalize your results. Whatever results you get from looking at the
subgroup are not generalizable. They are skewed in some way. That is
particularly true if you have members of certain communities opting out
disproportionately if you allow opt out. You might have people of a particular
race or a particular socioeconomic group or people with a particular disease
opting out disproportionately because they don’t trust research or medicine.
Then you have selection bias.
Of course, there are also cost and burden problems. If you are trying to
build a database or a federated system with millions of records, you have to
track people down, ask them for permission, and get forms signed. That is very
costly and burdensome and may impede a lot of initiatives. I just want to
mention a couple of other data and analysis problems that can affect
communities. One very important one has to do with data quality problems. This
did emerge in some of the breakout groups yesterday.
If we are using EHRs, but this is also true for other databases, you have to
be aware that databases are deficient. There are a lot of errors. E-Chart
technology is very exciting, cutting-edge, but there are now emerging studies
that tell us about all of these problems. People type quickly. They reverse
numbers. They misspell words in ways that are misleading. They put information
in the wrong place so you can’t find it, or it means something different.
Records are fragmented or incomplete. That is largely because different EHR
systems are not interoperable. They don’t work together. These are major, major
problems. You have a piece of the patient’s record here. A piece of the
patient’s record there, but you can’t put it together into a hole that gives
you the full picture of what is going on with the patient.
We often don’t have outcomes data. The doctor gives you a medicine. Then you
don’t get to come back and tell him how you are doing. We don’t have the
outcome. We don’t know if the patient got better or if she got worse and went
to a specialist somewhere else, went to Kaiser Permanente. How can you do
research or other initiatives with no outcomes data?
Coding is also very problematic according to some studies. It is easy to
mess up the codes. It is easy to up-codes, so you can charge more. Sometimes
the codes aren’t sensitive enough, or in different EHRs, they mean different
things. We have data harmonization problems. A recommendation that we talked
about in the red group is that we need to establish incentives whether they be
regularity or otherwise for interoperability for data harmonization.
There is a little bit of that in the Meaningful Use Regulations, but
certainly not enough at this point. That is why a lot of people are saying EHRs
are not very useful for data collection. We also have to keep in mind that data
that is collected in the clinic is collected for clinical purposes and for
billing purposes just as importantly. There is a lot of temptation to up-code
or exaggerate a little bit about what happens during the patient encounter.
These records are not always a good fit for research and public health
This is my last slide. Even if you have very good data, it is extremely
challenging to make sense out of it. I mentioned yesterday that there is a
danger that it will be manipulated or misused. Even if there is no malice, it
is very easy to mess things up. Selection bias we already talked about. You
might be looking at a subgroup that you think is a great subgroup of records,
but it is skewed in some way, so you can’t generalize your results.
Measurement bias: You might have equipment that is not calibrated correctly
or inferiorly, especially in economically disadvantages area. You might need to
rely very often on patient’s own accounts of their medical histories or their
symptoms. Patients may not remember things that happened a long time ago.
Patients may have dementia or other memory problems. Believe it or not, I am
told patients lie to doctors when they are embarrassed about the truth. Again,
this will skew your results.
The most complicated issue is confounding bias where you have variables that
you should consider, but don’t consider. These variables might be confounding.
They might affect treatment choice and outcomes. If you are not considering
them, you will not be able to reach correct conclusions about causations.
A classic example is disease severity. Doctors might choose to give patients
who are very sick treatment A rather than treatment B more often. Of course,
patients who are very sick are also not going to do as well. They are less
likely to get better or get appreciably better. If you just look at the study,
it looks like treatment A is not very effective because these sick patients
that are getting are not getting better.
However, if you calculate into the formula the fact that you should be
considering severity of disease and that severity of disease affects the choice
of treatment A, the picture changes. Then you not conclude so easily that
treatment A is ineffective. You have to take into account these confounding
variables. You have to have experts that can carefully design studies and make
sure that the correct variables are taken into account. Often the variables are
things like genetic abnormalities, stress, socioeconomic considerations, which
can, in fact, be characteristics of particular communities. If you get this
wrong, you might have stigmatization or otherwise actually harm particular
Making sense out of the data is a very big challenge, and you need
tremendous expertise to do this right. I think I will stop here. I,
unfortunately, don’t have a video with which to entertain anyone, but we will
move on to the next speaker.
DR. NICHOLS: Thank you, Sharona. Donald
DR. MALEC: There is a lot of talk about using data here. It is nice to hear
people are interesting in making decisions with data and lots of types of data.
What I am going to be talking about is how to use data from sample surveys.
Specifically, using techniques where there is some data, but not quite enough
to be able to make precise, well informed decisions. I will be talking about
small area estimation and how that data can still be used to be informative.
I was going to kind of break it up into two parts. From my experience, there
are a lot of different definitions of small area estimation. People think it is
one thing and not another. I am going to kind of lay the ground work, the kind
of small area estimation I am talking about. The second part I am going to talk
about, the type of small area estimation that we are currently working on it at
National Center for Health Statistics. I am going to use the term statistical
small area estimation because we don’t really need to or want to provide
estimates for geographic small areas. By a small area estimation, we mean any
unit, geographical or a population domain where there is not enough sample to
make precise estimates. It is really a small data problem, sample size, but it
has been termed small area estimation.
What is a required element and statistical small area estimation is there is
a lot of data somewhere, maybe not completely relevant, but there is something
else. Generally in sample surveys, they are from national surveys or from large
state surveys, where one can provide estimates for outcomes for larger
aggregates, larger areas that are precise, but the sample runs out when one
looks at sub-domains.
What is optional, but usually makes a different from a successful small area
method is that there is that are related covariants, either through
administrative records, mostly through administrative records or even Census
data –- that is an administrative record, so generally through
administrative records to estimate relationships and apply those to the small
Here are a few examples of using this definition. With this definition, we
can have countries being small areas. For example, Columbia, Costa Rica,
Indonesia, one would not think of them as being small geographic areas, but
according to this definition, they are. According to this definition, a Census
block group can be a large area. There are about 200,000 of those in the U.S.
in Puerto Rico. They can be considered large areas. A demographic group,
for example making estimates for native Hawaiians or Pacific Islanders,
according to this definition, could be a small area, if there is not enough
sample. These are actually real examples.
Back in 1985, a smaller area method was devised and utilized by Wong and
Mason based on the World Fertility Survey and making estimates for
contraceptive use for countries. They had a lot of covariants. Within country
covariants, educational level, rural status. They also had between country
covariants, which they used in their model.
Our census block groups; what I am talking about here is data from the
American Community Survey. Many estimates can be made from them including some
health estimates for some health outcomes. If one aggregates American Community
Survey data over five years, there is enough data for each block group. There
is no modeling needed. They are précised. They are published.
Native Hawaiians and Pacific Islands: There is interest in making estimates
from the National Health Interview Survey. It does fall in a small area
requirement. For example, if one wants to estimate diabetes prevalence for this
group, one can estimate diabetes prevalence across demographic groups.
HIS has a well enough, large enough sample for that. Another difference
between domain base smaller estimation and geographic is a lot of times apriori
once one controls for covariants, there might not be a big difference across
geographic areas, but apriori across population sub-domains, demographic
groups, race ethnicity, there apriori might be a difference. That does change
To delay things out a little bit, what I am calling small area estimates?
The typical form is a weighted average. One has a direct estimate. That is
using data only from the small area. One has an estimate based on a model,
where one has used the data across small areas to relate to the particular
Another feature of small area estimation is weight. They are both Y-direct
and Y-model are estimating the same thing, the same target, but the weights are
also estimated from the data. They vary as to how much information one has for
the small area. As the sample size in the small area gets larger, that weight
goes to 1 and 1 relies more on the small area. It is not an either or
situation. It is let’s compromise and try to make the best out of the two
One of the earliest small area estimates that follows this model is a paper
by Fay-Herriot, a few decades ago. The direct estimates were logged per capita
income for country.
The model base used Census per capita income during a census year, which was
very precise, and just did a regression for the model-based estimate.
Why small area estimation? I think this group knows this very well, but I
will just repeat it that there are policy decisions, funding allocation and
interventions are often based on quantifiable needs. There might not be the
right kind of data for that.
Instead of one can make guesses or one can try to use the data, but often
the data that is even available for the direct estimates can be suppressed for
discloser avoidance reasons. Small area estimation can use that suppressed data
and combine it with models, therefore, avoiding disclosure avoidance. It is a
way of filling the gap and sort of making an informed decision based on data
instead of just making an informed decision without any data.
This is getting out of health, but part of the Census Bureau SAIPE Program,
where funds are allocated to school districts. It is based on a formula and
small area methods are used to help allocate that. Another reason is that there
is always more demand for data smaller areas, smaller domains, no matter how
much data one has. Small area estimation is sort of an inexpensive way of
exploring what we can find out. Instead of increasing the sample or creating
new samples, one can sort of use it to explore, to see whether there is
something going on in small areas, where it does require more data and a better
Another thing that has come up the past couple of days is uniform quality.
When one gets data from different sources, a lot of times it could be mode
effects, definitional differences the way small area techniques work is because
they are based on a larger sample. The sample is uniform and is usually
collected by the same agency, same rules. Covariants can be required to be
uniform, and that is generally the case. They are usually administrative
records that are from a national database with uniform requirements.
One model is used for all of the areas, so the assumptions, although they
may be hard to validate, they are applied across areas, and they are up for
inspection, and it is one set of assumptions that people can see and decide
whether they do fit or they are they are appropriate.
Speaking of assumptions, it is model-based. We are making estimates for
areas where we don’t have direct data to really assess how well they are. There
are assumptions needed. One can fit models, and that is what is done. The
models don’t work for all of the small areas. Model fits are generally sort of
a majority rules. Most of the data fits the model. There are always outliers
that can be left out. One has that danger when one makes small area estimates
and uses it.
Even to incorporate the direct estimate into the small area, one needs
further assumptions. Usually some normal distribution is assumed. The
variability of the small area also is needed to know how much to weight it
together. That also is subject to data limitations. If you only have a little
bit of data to make an estimate for a small area, you also just have a little
bit of the data to make an estimate for the variants. You sort of have a double
I look to think about small area estimations as more a way of filling in
missing data and more of a missing data problem. The idea is you would like to
have more direct data. The more direct data you have, the better. If you don’t
have it, this is a principle way of filling it in.
That is a little bit of summary of statistical small area estimation. I
would like to talk a little bit bout what we are doing here at NCHS and what we
have done. I would like to go through five projects kind of quickly. This is in
order of maturity. The first one is making county estimates for smoking
prevalence and cancer screening rates. Another problem is making state and
sub-state areas about 92 sub-state areas of people who use wireless phones only
and have no landline phones anymore. A project for fast-screening outcomes —
This is a little bit different from the usual small area where we would like to
find out which outcomes vary to begin with. This is sort of a screening method.
A brand new project which is in the works is making small area estimates
from the National Health Interview Survey with data linking to the American
Community Survey, which has a lot of data at the small area. This is the
potential source of a lot of covariants for health surveys. I am not going to
talk about the last one anymore, but there is some preliminary research work
looking at model-based estimates for the healthcare system.
I would like to briefly talk about the first four methods. The first one
there has been a lot of praise for the behavioral risk factor surveillance
system and the first method developed jointly with National Cancer Institute
and University of Michigan built on the behavioral risk factors surveillance
system, but attempts to make it better. The behavioral factor surveillance
system is a phone survey. HIS is both phone and no phone population. The method
combines those two surveys together. It fills in the people who don’t have a
phone, eliminating that sort of selection bias. It also overlaps those who do
have a phone in both surveys to eliminate a mode effect of being answered.
Patient’s lie if you are asked your weight over the phone, you might get a
different answer than if someone is at your doorstep, asking how much you
weight as the HIS has a personal interview survey.
This is well-developed. The estimates are on the web. It is for the National
Cancer Institute, and that is where they can be found. There were two periods
2000 to 2003 and 1997 to 1999. The current estimate is being developed. The
method is being modified because of the large increase in cell phone-only
population, which HIS also covers.
The method is being modified to account for that. This is a relatively new
method. It helps out our National Immunization Survey, which is dual frame
telephone survey. Dual frame means cell phone frames and landline frames. Its
combination is to use data through six month intervals. It is borrowing not
only across the United States. It is also borrowing across time from the same
place, which sort of makes sense. You would expect a big change, and it might
estimate a trend through time.
These estimates are for about 92 sub-state areas. These estimates are
combined with American Community Survey, socioeconomic variables at these
larger areas, and they are used as covariants. These are also available online
for 2011 and 2012. Like I said, they are used for the National Immunization
Survey, and I have heard second hand that they are very popular even for state
survey organizations because they provide the base population, the denominator
when one is doing dual frame surveys for people who have only wireless and
people who have landline.
This is relatively new, and we are looking at possible improvements. Right
now, that is the current research. Either we will have improvements or we will
have better defense that this method is developed and is fine the way that it
DR. NICHOLS: I think I have been remiss in being a steward of our time here.
I am afraid we are about to run out. Maybe you could just do one more and then
give us kind a big picture wrap up.
DR. MALEC: I will do my favorite project and then a wrap up. May favorite
project is current one. We are not looking at specific small area estimates. We
want to screen large numbers of outcomes because small area estimates are
intense, analyst intense. They take a lot of time. They may not be as expensive
or require as much organization, but we are looking at methods of going through
large number of outcomes quickly estimating which outcomes have more
variability across small areas because going through all of this trouble of
modeling for small area estimation when there isn’t much variability across
small areas to begin with.
We have had success at the state level. We are now trying to see at what
sample size does this start to break down. The story about the Census is there
is confidential data. There is a lot of work of sharing Title 13 data and
vice-versa of Census looking at our data. We are going through those barriers.
Other uses of small area are the methods are used for disclosure avoidance.
There are generally good ways of analyzing data from surveys. We are finding
out better ways of analyzing data when there are selection differences in
selection from the population. That is it. I hope this generates some
discussion amongst the group and also if anybody has any suggestions or there
are gripes or their favorite things or they would like to do small area
estimation, I would like to hear about that as well.
DR. NICHOLS: Thank you, Donald. Can you stay around today to answer
questions at least this morning for the breakout groups? I think people might
have specific questions about what you could do for them.
MS. KLOSS: I am going to remind everyone of what we are doing next. I think
we know the drill pretty well on our breakouts, but Kassie is going to pull
back up our round table slide just to remind us what we are going into. Next is
a breakout on how do we promote information enabled community driven change. We
had some different approaches in our groups to what we did with the more
detailed questions. I think it might be helpful as we go into this next one, to
have some consensus about how we are going to handle those.
My suggestion actually is because we are kind of converging now on beginning
to transition the thinking more about recommendations that you use the time
available, roughly 50 minutes, and take eight minutes or so on each question
and then the rest of it to wrap up and give yourself some time for just
stepping back and thinking big picture thoughts.
I think that if we all did it that way, maybe we would be a little more
comfortable that we were getting more consistent level of discussion out of
each group. We do ask you while you step through those four questions how best
to promote what role does government have, what specific analytic and technical
resources could be useful, and how can local healthcare organizations partner.
If you take eight minutes on each of those and then step back and think about
other high level thoughts that perhaps weren’t captured by the those questions
so that we have rounded up all of your thinking.
Then we ask you to really focus to keep the level of discussion up. If you
find you are getting too granular, we will ask the facilitators to say move
let’s move on. I hope that helps some learning from yesterday. Maya would like
us to condense the report out a little bit. I think if we look at these four
questions and then sort of wrap around there may be several thoughts per, so
keep it to a couple of slides. That is really the message.
DR. FRANCIS: We are going to start. Jeff.
MR. BURKE: So our group talked about a number of different concepts about
how to provide useful assistance to communities to help them drive their own
change. One of the principles that emerged quickly was the one of co-ownership,
where government is the helper, the community as the effecter arm would work
together to identify solutions as well as problems. This was driven off of an
observation that Carladenise offered us this morning about how it is more
important often to help communities by hearing what the problem is and then
asking them how they recommend a solution rather than government saying here is
how you should solve your problem.
Obviously, this requires a lot of partnership and a commitment to working
with people, but not just those who represent themselves as the leaders of a
particular group, but those who are in fact the most influential people in that
group. That may not always be the same. I was reminded of an observation from
an earlier session of this group before you step in, step back and look at who
the leaders are. They many not always be the people who are representing the
The opportunity government has to help communities develop the capacity to
be engaged in the process and the solution is very important. It is very real.
All communities are not at the same level of evolution, of maturity,
development. Their needs will be different, and it would be important to
recognize that. There is variation in how well communities are organized, and
how their skills are developed.
The group noticed that government’s responsibility might be clearly, to
redefine how public health is viewed and adopt the communities’ method of
defining problems and solutions. Public health can provide information about
the problems that the data show. The community can provide insight into how
those problems are seen and where they live. Entry points for government in
this equation, in this partnership vary. It can be an event, a solution; we
were given the example of the rats in Rochester as an event which mobilized a
community, or as Jim would say, “A spark” that gets the group
Assessment is always key. That is what government does well. Government can
help assess the dimensions of a problem, its depth, and its edges and assists
the community in solving it. Partnerships are noted particularly private sector
partnerships have become a very effective way of supporting communities in
achieving their improvement objectives.
A couple of possible recommendations, whether here or for the next section,
community health needs assessments. We need to be community-driven and co-owned
with a community. There was note of the pending ACA requirement for
not-for-profit institutions, hospitals, to perform community health
assessments. This is a point of potential leverage.
As community health assessments are being increasingly seen as a way to
understand the issues that communities are facing, government has an
opportunity to require that these assessments be community driven and not
simply driven by the institution who is under an IRS obligation to deliver. The
mechanisms to feed the information back to the community where it can be used
and it can be acted upon by community groups. There is a fragment of a sentence
here which says it would good to have profit hospitals. We already have profit
hospitals. The second half of that sentence is wouldn’t it be great if for
profit hospitals felt the same obligation. Right now it is only an obligation
for the non-for-profits.
Government can do a better job of providing and presenting data in ways that
show the gaps and make the information actionable. The notion of meeting the
community where it is goes back to working block by block, not simply shoveling
data over the wall. The possible recommendation of improving resources at the
community level including coaching and all of the other opportunities to assist
every community goes to this notion of variation and where particular groups
Some of those that have been more well-established, more well-resourced, or
longer-standing, are going to be more evolved and able to take advantage of
some of the tools that the government typically applies uniformly to all groups
no matter how well-developed they are. That is a quick summary of how we spent
our hour. Is there anyone else in the group who would like to add?
DR. MAYS: How about the blue group.
DR. CATLIN: Okay, Blue Group, help me out here. This was a very lively
discussion. I am probably not going to do the discussion justice in any way at
all, but here are the bullets anyway that at least some of us worked on right
at the end. We first talked about the need to shine a light on best practices,
about what is working in communities. Then we had a lot of healthy discussion
about promoting community engagement and things that funders could and should
do when they consider criteria for eligibility for grants. The Patient-Centered
Outcome Research Institute received kudos for their work, what they do, and the
contributions they can make for communities.
I am going to actually interpret these bullets. I think this is along the
lines of what we do at the county health rankings is putting data out there to
grab attention. You can’t just throw out a thick report. You have got to do
something to get peoples’ attention, to get them engaged, and just data alone
won’t do it without some hook in some way. There are a lot of interventions out
there that that will the health of communities that have been show to be
Communities need help in finding and getting to information about those and
the notion of a clearing house. There are lots of siloed evidence databases and
that is all part of this knowledge base that communities need in order to do
the work. Partnership is absolutely key, as has been mentioned.
Then the idea was promoted about taking the model that has been used for
years and years with the agricultural extension model where people in
communities have helped farmers do their work, access the information that they
need to do their work better. What we really need is something similar for
health. We need people on the ground in communities who can both help people
use and interpret data and then move that to action.
Then one of the members of the group suggested that maybe there was a model
that we could use to help guide this work. It is a blue model from the Blue
Group. We just put this together. Now you know why I am the one reporting back.
We just happened to have this model, which I renamed as the model for community
change. We call it take action model. It is part of the companion to the county
health rankings, as the county health road maps.
In this model, right at the top, it says work together. That is key to
anything that we do to improve health in our communities. As you move around
the outside cycle, the first step is assessing needs and resources. That is
where data comes into play and where government can play a huge role in helping
to collect and disseminate data. Then communities need help moving from that to
focusing in on what is important in setting priorities. Everybody wants to work
on everything. We do need to start somewhere. Giving assistance to communities
to help them focus, and then one of the most important things of the model is
not showing up on the slide. The bottom step says choose effective policies and
programs. That is where this notion of needing the evidence, needing databases
As an aside, we do happen to have a little evidence database ourselves
called what works for health that draws on the work of a number of different
federal government agencies and private entities that have evaluated the
effectiveness of different policies and programs. Communities need help there.
Most importantly, they need help on actually moving and acting on what is
important and implementing things. Then to close off of the loop, a really
important piece that ties in to building that evidence base is evaluating
actions, and so then there is another role again for data.
At the center of all of this is the community. Then you see that in the
light blue, in the middle, and then the other circles represent not some, but
not all of the different players who can work together to improve health. We
did try to follow the rules a little. We did talk about the role of government
and what government contributes to build a sustainable infrastructure,
convening in supplying data, building criteria for who can do the work. There
is a model where university PIs are valued more than communities, and so that
leads sometimes to the wrong incentives. Another area where incentives are
wrong is the often interventions are implemented and studied. Then that is it,
and then they are dropped. Government can help change some of those incentives.
Anything else from the Blue Group that anybody wants to add?
DR. MAYS: So we will have red.
DR. FULCHER: Thank you. We also had a really intense and very productive
discussion. I would like to thank our facilitators for really cranking through
and really going through the process of getting a lot of ideas on the board.
This is a report from the red room, and it is a play on words from a movie, but
I won’t go into that.
The key takeaways here, or insights, about government agencies working
together to avoid duplicativeness and maximize efficiency, coordination
alignment. What this means is basically how we work smarter together. How do we
really leverage these different capacities at the federal, coordination with
the state and local? It is around also alignment in funding. How do we really
recognize the roles of funders, like with CDC and the CTG grantees around the
country, with what RWJF is doing for example with their grantees, YUSA, et
cetera? The idea around alignment with data collection and also where are
investments going and are we making a difference?
The second area around messaging and marketing. This is key because as we
really want to get the word out to communities around the country, we need an
effective messaging channel. Right now, we have a lot of fragmentation. Many
different websites competing or vying for web users, and trying to get that
message out is really critical. How do we pull ourselves together to provide
that more cohesive messaging channel?
Regarding government providing technical assistance, this is around the area
that we talked about earlier, the presentation around small area estimates,
which was great and so very important to really get at that sub-county level,
community granularity that we need.
Also, the area of re-aggregation of data. There is a lot data that we have
that is point level data tied to individual addresses. Of course, we don’t want
that. We are all very sensitive to privacy, et cetera, but can we re-aggregate
that data to a stable population base, like a census track, rather than a zip
code, which we cannot use from year to year, because you can’t use trends over
times with zip code boundaries.
First of all, the post office doesn’t even recognize a zip code as a
boundary, so it is fraught with peril. The government can provide great
technical assistance in helping us think through the way we re-aggregate data
and the way that we provide methodologies for small area estimates.
Other insights including understand and be careful about defining community.
We had a lot of great discussion around transparency, around really engaging
communities on the issue rather than getting focused on data and technology at
the outset. What are the issues? We use the word governance or facilitation,
depending on the issues and sensitivities we face in communities around making
sure that there is buy-in from all parties.
The idea of community comes up time and again. I have been in this for 20
years. It is a pervasive question. What is a community? Is it geographic? Is it
issue-based? It will never be resolved completely because we all have our
mindsets around community. There are communities of interest. There are
communities with stakeholders that are interested in childhood obesity, but may
not be as interested in third grade reading proficiency levels.
We really need to be careful about defining community’s outset. They need to
be diverse, avoid marginalization and discrimination. Again, how do we pull
people together making sure all parties at the table reflect the issues that
are being addressed? Another key insight is around nesting, local
collaboratives and a national framework. One of the questions was how do local
health organizations and NGOs and government agencies partner? We can do all of
the partnering we want locally, but if we do it as a number of different
islands that are not connected, we lose the huge opportunity to provide a
comparative context across all of these communities and the types of
partnership they are doing.
There is discussion around nesting these collaboratives, organically grown
around the issues they are addressing, but have the under-pinning of a national
framework so we can ultimately get to that comparative context. Empower rather
than impose on communities. We talk about top-down and bottom-up, and it is
really important that how the communities engage being prosumers of content and
data and how they contribute to the process.
One note on the word empower. I was in Saint Louis in a meeting several
years ago. I was with a group of African-American women at my table. I talked
about the word empower. They said don’t use that word. If you can empower
somebody, you can de-empower them. Again, words mean something very important.
I have always used that word empower, but I thought about that group meeting
and how the word was being reflected in that community context. It was a really
ah ha for me.
I think that is about it from our key insights. Any other comments or
reflections from the group?
DR. MAYS: I think we have only a couple minutes left. I am gong to make only
a couple of comments, and then I am going to follow in your leadership of
actually trying to do a written summary.
I think what I want to leave you with in terms of the themes is ROI, return
on investment. When we think about the community and the energy that is put
into the community, reaching a level of good health, what they want to see at
the end of the day is that something happens. The federal government, for the
same reason, collects a lot of health data, engages in a lot of infrastructure
to give us better health. At the end of the day, they are attempting to do the
same thing, which is to get this data out and to make the health of the nation
In between is where we really need to focus, and that is how do we go
between what the community needs and what the federal government is going. How
do we build the kind of infrastructures that leverage the return of investments
for both groups? I think I am going to end this here. Any other comments? I
think we are done.
MS. KLOSS: Marieta is just handing out now something that will we will come
back to after our next panel but it is a recap of all of the reports from all
of the groups and suggestion from us co-chairs as to how we frame our
recommendations for breakout. Just set that aside fro now, and we will come
back to it when we do set up for breakout 4.
DR. SUAREZ: Well this is the last panel really, panel number four. One of
the goals of our panel was to try to provide some coalescence of messages and
ideas and try to bring the topics and the perspectives that all of the other
panels presented in the last day and half. We have a really distinguished panel
of presenters that will attempt to do that. They will be presenting a few
slides and making some additional statements about some aggregation thoughts,
some summary thoughts, and themes of what is heard during the other three
The main topic is what recommendations might NCVHS advance to support data
enabled community health? I think that is an opportunity to really highlight
some of the themes and some of the ideas that came across during the last day
and a half. We are going to start with Denise Chrysler from the Mid-States
Regional Center. Then we will go to Chris and then finish up with Denise.
MS. CHRYSLER: Hi everybody. Thank you so much for inviting me. I have gotten
a lot of out of being here. I hope I can contribute a little. When Walter said
you really can’t prepare because you need to listen and react, it sort of left
me in a situation of quite not knowing how to focus. I took my cue from Eve, I
believe, the confidentially officer. The reason I took my cue from Eve, that
could have been me with my agency. There is a lot of talk about freeing that
data. Freeing the data not only involves public health practitioners and them
seeing a need a way to free the data and how people use the data, they also
work with attorneys and they work with privacy officers.
Those folks can sometimes be seen as obstacles. I was thinking about how
could the federal government help support public health to be a data provider
and specifically looking at privacy officers and attorneys and what they bring
or how they can be seen as a barrier or what might be able to happen from a
federal level to help address issues that they bring.
I first want to mention where I work, which is the Network for Public Health
Law. Who we are and what do: We are an initiative funded by Robert Wood Johnson
Foundation. We have five regional centers, either in a law school or a school
of public health. I am in a school of public health. At no cost, we help
promote and support the use of law to solve public health problems. We provide
technical assistance doing research, answering questions. We can’t know
everything, so it is often hooking you up with experts. We provide resources
and various tools. If you look on our website, for example, you will see a
summary of the changes to the HIPAA Privacy Rule from December 2012.
We provide opportunities to build connections. I mentioned we are national
in scope, but we each serve regions so that you can build a connection with
your region. We serve public health practitioners at all levels of government,
policy makers, and advocates, anybody who is interested in using law to improve
We have various topics where we as regions specialize, and health
information data sharing is a topic of the mid states region where they provide
national leadership to the region. Like I said, there are a lot of things we
don’t know, a lot of things we do know. We help you get to the experts to get
the information you need.
A few of the health information data sharing activities: We field a lot of
requests for technical related to data sharing, everything from HIPAA and what
am I required to do to how can I as a public health agency access data to
improve what I do for my job. A real popular request these days is how do I
access or can I access prescription monitoring information, so that I can use
it for assessment and strategy regarding substance abuse.
We have a number of resources on our website, write blogs on data sharing
issues. At the 2012 Public Health and the Law Conference, we had a data sharing
track with five sessions. Sharona was one of our group leaders. We have a state
privacy officer listserv with every participant. The listserv provides peer
assistance to one another in resolving data sharing issues.
In March 2013, we had a public health in the Learning Health System National
Meeting, where we convened about 60 public health leaders from all over the
country. We talked about learning health systems and public health being part
of the dialogue. You all weren’t there. You know why, I did the invitations. I
had no idea that the committee was doing all of this work. I have been spending
my time here asking myself how this could have happened. Lacy was there.
Bridget was asked, but it is right before their rankings were due. Sharona was
I interviewed a lot of public health informatics experts and who should be
invite. Gail Horlick was there, but I didn’t ask Gail who we should invite. She
came and was a very important presence. Gail told people we just had this big
meeting, and you all aren’t connected. I talked with Linda a little bit about
this today. I offer a major, major apology. Our materials are on our website,
and we will be providing a summary report. We have generated a lot of ideas for
follow-up. These dots need to be connected as far as the public health presence
and the public health dialogue in learning health system is essential. When I
see the disconnect that has been there, I am very concerned.
After that, of course public health is rich with data. These are just some
of the many public health databases and lots of elements, everything from birth
records, disease, surveillance, immunization registries, birth defects
registry, newborn screening, WIC, and we go on to cancer registry, early
hearing detection and intervention, children’s special healthcare services,
Medicaid, childhood lead screening, death records, and this is just a fraction.
When I come here and I listen to you all, and liberating the data, this is
what it feels like –- too big and too hard, and it is like how do we not
get paralyzed in too big, too hard. What I hope you can all help us do is just
go to making it big and hard. That seems doable. There are so many legal policy
and technical issues to resolve in liberating the data and using it in ways
that lead to action to improve our communities. Since I have been here, I have
been thinking a lot about how the federal governments take us from too to too
hard, to simply big and hard.
Here are some challenges from my little narrow viewpoint. I spent 20 years
as an assistant attorney general. I spent seven years in House Council to
Health Department including as privacy officer. Remember the IRB coordinator.
We called ourselves Data Our Us, and you know what, we are really good at
saying no. First, one of the challenges I have heard people talk about here is
state law mostly governs with the exception of the areas of public health that
are covered by HIPAA. Depending on how you organize this, 46 states are hybrid
entities when it comes to HIPAA, meaning their public health functions are not
covered. We have four states that had their entire public health agency
covered. It does cause some headaches with how they handle data. Mostly state
law governs, and of course that law varies from state to state, as we are
talking about pulling data from multiple sources.
Even more of a challenge, every one of those databases that I just had on
the preceding slide is often some general state laws. Everyone is controlled by
its own laws. When I look at the birth defect registry, I go to certain laws.
When I look at the immunization registry, I go to a different set of laws. It
is not like I can make general statements. It depends on your data sources, and
data sources are siloed with regard to the law. They may be in other ways too.
The privacy officer’s are often in doubt because law is not clear, it is
gray. If it were clear, we wouldn’t need lawyers. When it doubt, they say no.
One thing about no, and how do you get to yes? No often comes out of
interpretation. How can you expand a mindset of interpreting the law? Think of
it this way. When we are lawyers, we think of risk in one way. If you do
something, you risk getting sued. That means that we are not going to do that.
There is another side of risk. If you hoard the data, you are going to shoot
yourself in the foot. You are not going to see the important work that your
agency does go forward or that the community can do with your data. It is like
how you, federal government, and committee can; help expand that mindset of
attorneys and privacy officers. How can they be part of the team, not to ignore
the law, but to help inform how does the law establish a public health agency?
What are their core functions? What are the 10 essential services? How can you
look at your job in keeping that in mind as you do your interpretation?
De-identification is a big, big issue and then the other factor
re-identification. HIPAA may not apply to a lot of functions of public health.
In the absence of clear standards, HIPAA is often the default. I mean the safe
harbor of HIPAA, remove 18 identifiers. I don’t have time to tell stories, but
I have some stories with regard to the absence of statistical experts to help
us understand, re-identification, identification, and be able to help us say
yes. They also can help us when needed to protect the data. We also comply with
the Freedom of Information Act. There is a story I want to tell. When you get a
request, you have to say yes unless you can set forth a legal reason to say no.
Not only can we have the experts to help us with the de-identification piece,
but how do we show the risk of re-identification?
I was just going to give you an example of interpretation, Michigan’s HIV
Confidentiality Law. This is the section passed when we had balance about how
we used data. Privacy ruled in the 80s when it came to HIV data. This law, when
it says all data pertaining to testing care, treatment reporting and then
research, the question was should this law be read to say you public health
cannot share publicly any information you derived from individual care reports
which is heaven where we get all of our information because you are violating
This actually was an issue that came to the Network for Public Health. We
thought through what does public health do? What makes sense here? Once we
figured out what makes sense and what does not make sense is to say public
health you had this data about HIV, but for you to tell the community what the
incidence, prevalence and mode of transmission would mean you have to go look
at these individual case reports. You can’t do that. That just didn’t make
sense. We found a way to conclude that the aggregate data on the website wasn’t
appropriate and allowable use of information.
Shining a light on transparency and that has a downside. We used to be able
to work a lot easier before communities got upset that the government has their
data, the government is tracking them. The government this and that, and how we
deal with that fear. I don’t know if anyone has seen a manuscript that is
entitled Mission Creep. Have you, Sharona? It is all about public health
agencies expanding beyond what was ever intended as far as their data
collecting abilities. Everywhere is pretty clear about public health having
broad and flexible authority to collect data and then what do with it after is
what we are talking about here.
Some people think that area of the law developed during rapid infectious
disease and now that public health is using those powers to look at lifestyle
chronic disease, some people call that into question. The one thing about
shining the light and engaging the community and people getting upset is public
health needs its functions of collecting data and the concern about
jeopardizing public health’s functions and services. There was some talk here
about consent models beyond individual consent, such as community consent.
Individual consent is difficult to work in a public health context. Sharona
talked about all of the different selection biases. It would really put public
health back tremendously.
With the community consent, the federal government and the committee may be
able to help with. What are consent models that engage the community, comfort
the community, but are workable from a practical level. Connection to fellow
travelers such as I had mentioned in our small group the health and all
policies movement in looking at policies whether it is transportation or
anything else. What are the health aspects of those policies and what data
brings to bear there might be a good connection to make. I want to mention that
a lot of my view points come from the work I have done with the development of
the Michigan BioTrust for Health.
I was legal counsel when the BioTrust was developed for a formal process for
using dried blood spots and linking it to data for public health and research
purposes. Michigan actually had an opt-out model for many, many years. We have
over 4 million blood spots representing the last generation of children since
1984. We went to an opt-in, an informed-consent model in 2010. As a legal
issue, we debated and debated whether this was legally required. We just threw
up our hands and said it is the right thing to do.
There are a lot of people who disagree with us on that, but legally, our
concern was when we collected blood spots for testing for metabolic and genetic
illness, we were not doing it for the purpose of research. Once we set up a
formal system to make those blood spots available and facilitate their use for
research, were we collecting them for research? That is where the whole game
changed. When we talk about public health as your agent or public health being
a leader in collecting data; I just want to point out this concern when it is
collecting data for research rather than public health purposes, it is a
I am on our Community Values Advisory Board now that I am not longer with a
department with a public health perspective. We do have a very robust community
group that spends a lot of time on acceptable uses, transparency, the kind of
things that are important here. I did learn very quickly from studies and from
input that choice matters, and it seems to matter even it is de-identified.
When we talk about social media and how can people who blab all about their
personal lives on social media say they want to have a say so. It is because
they say we control what we put out there. We want to have a say so about we do
not want to defer to you, but as Sharona mentions, we all benefit from data. It
is for the common good. It is like we all pay taxes so we can have roads,
education and the other things. How can you help us expand that mindset?
I did want to mention models. You might be able to help us with models of
consent. Lastly, reciprocity is important in the sense of how can we help the
community to see and make data reciprocal. I know we talked about returning
data to the community. Finally, I have learned from working with Minnesota
about privacy advocates and privacy absolutists. Privacy absolutists can be
pretty vocal. I know Lacy is very aware of the issues. We are not going to
please everybody. How can we move forward just acknowledging that there are
people who will not see the greater good and do believe it is government
tracking children being in an inappropriate power relationship to the people?
That is it. Thank you everybody.
DR. SUAREZ: I think we are going next to Chris.
DR. FULCHER: What I am going to first start out with is just focusing on
data and mapping and then get into my Power Point presentation. I was not going
to do this part originally, but I thought in the context of the discussion over
the past two days to give you a sense of what we are doing with Community
Commons. I am not going to do a detailed presentation of all of the mapping
functionality, the engine and all of that, but just suffice to say that this is
an area on Community Commons called starter maps. A lot people don’t know how
to pick all of these different datasets, so a jumpstart for them may look at
rate of obesity among adults, healthy food access, and access to affordable
When I click on the map here what pops up is an interactive GIS environment,
zooming in to Louisville, Kentucky. When that map pops up what you have here is
a description. A starter map shows USDA designated food deserts, et cetera. You
can look at the legend here. Locations of farmers markets, SNAP retail
locations, and the percent of students eligible for free or reduced lunch where
70 percent or more of the population is in the darker brown. There are food
deserts and underlying poverty down to the census track level. That is the
sense of the map. I may be interested in going to Atlanta. You can go anywhere
in the country with this engine here. You can bring up this same data. You are
zooming into Atlanta, and you are looking at this same data.
Let’s see how I want to look at some other data. This is where we have over
23,000 National Source GIS Data Layers that go down from the state, county,
census track, block group, down to a point level nationwide. Let’s bring up one
dataset, food environment, access to food, because there has been a lot of
criticism about USDA’s food desert boundaries. We were able to acquire this
modified retail food index core. I am going to update the map for Atlanta. Now,
the GIS are like a stack of pancakes. You got the top layer there. Let’s move
that on down so you can see those blueberries on top or in this case SNAP
retail locations, by manipulating this data.
This is publicly available to everybody. I am going to this SNAP retail
location and move it down to underneath here. This is a better proxy than food
deserts has been for communities. When they look at this data, they are looking
at what does this data layer show? This is where there are no healthy food
retail outlets in the darker brown followed by index score where index is high
access is in the darker blue.
This is a quick snapshot going anywhere in the country, a wealth of publicly
available data, and what I learned about eight years ago when I presented what
we have with this national data engine, the president of United Way came up and
said that is really fantastic what you have. It is absolutely amazing, but it
is not good enough for our communities.
What we need to do is integrate local and regional data because our data is
more current. We have people on the ground who understands the tacit knowledge
around the data that is being presented. That really started us on this journey
around collaborative management systems. This is the community comments mapping
environment that you see now. We have been working over the past year with
community comments beta because we have been getting feedback from many users
on what works and what does not work. In the next couple of weeks, we will be
launching Community Commons. This is another short video of the mapping
environment of Community Commons.
This is a new interactive GIS interface. It is a lot more intuitive than
what we had before, such as social and civic data, congressional districts,
legislative districts, food environment, and economics. I am going to click on
one data layer, which is predominant race ethnicity down to the census block
When I click on that, it brings up nationwide white, Hispanic, black, Asian.
I am going to zoom into Saint Louis Missouri, and as I zoom in closer, I type
in Saint Louis up there, you are seeing the block group level predominant race
data. It is predominant black, greater than 90 percent, predominant white, et
cetera. Overlaying schools, public schools, and their percent of student
populations eligible for free or reduced lunch. That was 88 percent of the
student body. Next to it is about 39 percent of the student body.
We have these ways of looking at all of this data. We can also split the
screens. This is another thing we are doing with our new mapping environment.
There is a lot of data out there. How can we look at not just predominant race
ethnicity, but look at poverty. We are seeing county level poverty data there.
We can actually go down to the track level to look at the sub-county spatial
variability around poverty, and you see that on this left side there. You can
also expand this view and also go to other places in the country. I believe
where I will go next is San Antonio, Texas to bring up that data. I am looking
at predominantly Hispanic populations. You can do this with many, many
different datasets. Let me go ahead and start my Power Point now.
What I wanted to do was show the data because what recognized at the
beginning of this meeting and throughout the last couple of days is we are all
on the same page when we say the word data. This is page 7, but I am reading
page 7 of a comic novel, because even though we are on the same page, we are
not reading the same book. That is the analogy that I am looking at. The type
of world that Bridget and I primarily work with is this population level data
of this built environment data, the locations of grocery stores, farmers
markets, that is data. I was having a conversation with someone about what do
you mean by data. I said grocery stores. They said that is not data. I said
that it is. How much of that grocery store is available in terms of fresh
fruits and vegetables. Where are the grocery stores, et cetera?
It is an interesting thing for me and it has been a fantastic conversation
because there are great complementary roles between what we are doing around
social determinants and data as we frame it with how we are talking about
health data, clinical data, and how we can do the crosswalk between the
contexts that lies in communities.
Our Center has been involved for many years in developing and implementing
different systems, working first with United Way. It was that United Way
President that came up and said we need to be able integrate local and regional
data, and by the way, we don’t want a data repository. We want to be able to
tell stories. We want to engage our stakeholders in a community around a
process whether it is third grade reading proficiency or obesity, we want a
site to do that. That was called Community Issues Management.
We worked with HRSA that is an ongoing project with Health for Rural
America and with Kaiser Permanente. We worked with them on building their
national CHNA platform or their platform for all of their regions. I really
want to acknowledge Kaiser Permanente because of this public good notion of the
commons and they gifted all of the code that was provided for the creation of
this website and gifted it to create a community common CHNA that benefits the
whole country. It has been a great leveraging opportunity working with CDC,
Kaiser and a great group of people and making that a reality, including Bridget
and their team.
Another group, CTG Communities, California for Health, Current
Community Action Partnership, working around needs assessment –- they have
a whole different idea of what needs assessments are from what we talk about
community health needs assessment. The whole human service dimension is around
needs assessment. What we need to be thinking about is an overarching
assessment framework for communities, not health, not human services, but an
assessment framework, not for assessments to be the endgame but where we
ultimately make investments and being able to track those investments over time
and how we are improving communities.
Another site, working with Robert Wood Johnson Foundation around their
childhood obesity, GIS, with YUSA on their national mapping website and with
the Kellogg Foundation and many more sites. You are seeing all of the different
pretty colors and they are all completely different. They are all drawing off
of this public engine that I showed you. We have thousands of GIS data layers
across these areas, environment, education, all of the USDA food atlas data,
poverty data, economic data, transportation, and a wealth of health data from
Providers of Service and CMS, et cetera, and political boundaries. What you are
seeing here is a holistic framework of community.
It is not a health-driven framework. It is not around education. People
living in those communities don’t focus on the sectors, the focus on the
issues. Often, the data transcends those different sectors in terms of making
meaning of what they are trying to address. An example I showed you is that
census plot group data with predominant race ethnicity. We continually add and
update data. For example, a couple of weeks ago we added in working with the
Administration for Children and Families that the Human Services Side of HHS.
They said we love the stuff you are doing with community health needs
assessment, but we want to integrate all of our Head Start locations nationwide
in the system.
We said give it to us, and we will do that. You are again building richness
in context around communities as we were adding data to the system. Being able
to drill down very closely and being able to bring up images. This is another
dataset. What we have here and what I have showed you so far is disconnected
consumer drawing data from the national engine for their own purposes.
Every foundation or agency has their mission and how they are trying to
accomplish what they are doing for communities. There is a lot of overlap
between everything that we do. They are all disconnected. That is the real
challenge for us. They are disconnected. There is so much money being put into
recreating wheel about we want to draw up that ACS data. We want this health
data, and we are putting humpty dumpty together in many, many places.
Let’s stop doing that. Let’s start leveraging off of what is already
available to take us further down the road. Another area in this disconnected
environment is this prosumer part where they are producing data, GIS data
layers, content images, videos. This idea of consumer and prosumer is
important. We have been using the web for many years’ consumers taking away
from the web. The real opportunity and you see this from big data in the
private sector is the prosumer, the producers of information, content, et
cetera. The private sector is doing this in a big way. We don’t have a public
good big data analytics presence yet.
Most importantly, there is a diffuse communications channel. Diffuse
communications, they are all using their websites to reach out to their
separate audiences, and it is all the same type of people who want variations
on the information provided. We are doing a disservice to communities around
the country by further fragmenting and pulling people apart and not having that
coherent conversation that needs to be had around the issues that we are
addressing. How do we think differently about this diffuse communications
What we have been doing over the past year is connecting the dots. We have
all of these different websites, and a lot of other people who have done great
work around the country with their websites, how do we think about more of a
collective intelligence approach to really pull in the pieces together.
Community commons is in the embryo stage of that process, being out the gate
for a year.
You saw the GIS mapping environment around social media, ability to add
prosumer content, and blogs. This assessment reporting; we have a CHNA tool
that I am not going to show now. Policy Center –- it is how do we
translate what we are doing around data, story telling, comparative context and
geo-tagging policies that have been implemented or at the early stage of that
policy spectrum. What is happening where and why? Feature stories and these
spaces or groups off the commons.
This is free, free access to tools and functionality. The key thing and like
with the CHNA tool is vendors are charging thousands and tens of thousands of
dollars just to access basic data. We are raising the bar. We are raising the
bar to engage the vendors to make meaning and go beyond just providing data, to
make meaning of CHNA process for example. The report is just the starting
point. It really gets stakeholders in the community to say what are we doing in
our community and how can we benefit our community more?
This is my one slide with a lot of words. Community commons is really an
interactive mapping network and learning utility for people across the country
doing place-based work. The CHNA tool is being able to go into different
parts of the country. You can click on multiple counties or one country, and
what pops up for example is a lot of data that comes up. This is a snapshot of
free and reduced lunch eligibility for this region of the country. This is the
dashboard. This is Wayne County, Michigan.
The key thing is getting at the sub-county level with the CHNA tool. All of
the CHNA report is county level based. What we need to do from at least a
nonprofit hospital standpoint to look at what area we need to focus on
community benefit. This is a tool. It is a simple tool with two key drivers.
Population is below poverty level, and population is less than high school. The
red areas are the overlap showing literally the hotspots that means different
things to different people. I will use that word. The communities able to take
that slider and change that slider to 50 percent or 40 percent as well to able
to work in a community process looking at where we need to focus our efforts.
From that, all of this is special data that is also available.
What does community commons do? The aspiration of what we are actually doing
now is integrating these disparate systems that I showed you earlier into a
community commons framework so they are connected systems that ultimately lead
to coherent conversation. There is two way content flow; it is not just about
data. It is about content. It is about the data, but it is also the stories,
the qualitative and the quantitative that is being pushed both ways in this
Co-branded custom groups: There are going to be organizations. Kaiser
Permanente has a custom system with tools and functionality that is not
relevant to CDC, but CDC sure liked what Kaiser put together, so there are
variations on that theme. These custom spaces are really around co-branding. I
think it is important in the public sector that we don’t really obsess about
branding and the me at the expense of the we. If we don’t talk about
communications and branding, it is a very private sector approach to drive
traffic to one organization at the expense of others. The notion of co-branding
is very, very important if we are collectively working together to address
these issues. An idea of having a common framework to work from is important.
Another thing is about public/private spaces. We are working in communities
in the breakout sessions we talked about. In communities, you need stakeholders
really vetting the data. Is it appropriate? What stories are we going to push
out? There are user roles. There are different controls where you can have a
private space, but then push content out publicly ultimately. It is not just
around the major funders and organizations that I showed you before, but it is
potentially thousands of groups whether they are national networks data
organizations, local organizations, regional, et cetera. It is the same type of
underlying framework to have this coherent way to pull the pieces together.
Why a unifying framework for what we are doing? It is coherent navigation
–- we are driving users crazy with all of these different systems and the
ways that we try to find out data, information and stories. Second, accelerated
learning: Ultimately, what we are about is helping to improve the lives of
people in communities around the country. How do we accelerate learning by
providing common navigation that still has custom features in these different
areas? A growing audience base, so let’s build an auditorium where there is
more than 150 people that show up. That is what is happening now with the
proliferation of websites and how we are really vying and competing for
attention of communities.
Prosumer-driven: This gets into the big data opportunities with smart phones
and being able to go out there. We talked about in the breakout session,
identifying three data points we want to collect across the whole country.
Let’s say it is around playgrounds, and this month is playground conditions
month. You push that out through a common media channel. You push it out and
say, go to you playground, download this app and just tell us four different
aspects around safety. Are they in good condition? Are there issues of crime?
Take a couple of photos to show us.
If you do something like a very simple survey approach, just three simple
things across the whole country is quite powerful. If you do three other things
the next month, and three other things the following month, you are able to
start to garner the attention of these communities because they are getting
feedback on the relevance of what they are inputting.
Three is important, three to five at the most, because people don’t have
time. They get overwhelmed with researchers constantly hitting them with
surveys, complex surveys, long surveys, and it is not just one research group,
it is again and again. System infrastructure stability provides coding, et
cetera, and lowering cost by leveraging resources. The whole idea is around
Thank you very much for inviting me to this meeting. It has been fantastic.
Talking about an NCVHS role is greatly complimentary to what we are doing
around built environment and a lot of data that I am thinking of in a different
way. You are thinking about clinical level data, how we aggregate that up, what
is the appropriate way that we can use this within this broader community
Clinical data without community context is not as meaningful. The other way
around is not as meaningful. How do we pull the pieces together so that
crosswalk is key? The issue of re-aggregation small area estimates that is
fantastic. I think that is something we could really hone in on that could
serve us all well, not just with health data, but with a lot of the data that
we are working with.
I think I just want to end with you hear these words like war on poverty and
war on obesity. If we are going to declare war, we better be prepared to fight
it. It is really important that we talk about what we can’t do and focus on
what we can do and be creative. I am thinking of the lawyers. It is very
important that you are there and you are pushing back. I really liked what you
said about expanding the way. It is not no, it is how do we reframe it so we
can move forward. Thank you very much again.
DR. SUAREZ: Now we have Denise.
MS. LOVE: It is really an honor to be here and be among these wonderful
speakers who have done so much with applications of data and connecting data
with the communities. I will hopefully not be a downer. I think I want to tell
a little story about infrastructure. I talk about data with many different
audiences and it is really interesting to do so because I think a lot people
assume that the data are just there, and they will always be there. In my
world, we see it a little differently. My world is the National Association of
Health Data Organizations.
In my world, we talk about the data. We worry about the data every day. One
of the things that happens and is happening more is even if we are blessed
enough to get a data collection going in a state, we often just barely can get
the collection funded. The analytics are so down the line, and they get
short-shrift. This is where partnerships and all of those things we talked
about for two days must come in and must be critical.
I think what I will do, and Walter can interrupt me anytime because he knows
the story and he knows what we have to do here. I will rely on that. I wanted
to just kind of use one data system as a story for other data systems, not just
mine, but I think it has implications for the electronic health records. It has
implications for the community assessment and some of the things that have to
Let’s first say, National Association of Health Data Organizations, we have
been around since 1986. We have been working with states as they implement
state-wide reporting initiatives. Those are taking on a more expansive role
than just hospital discharge data. I am very pleased to say that as of this
date, we have 48, almost statewide, in patient systems. I am working with a
couple of states right now. I think the number is always going to increase.
Thirty two AMSERD(?) systems in emergency department. Those numbers I think are
boosting as soon as I do my counts. They are based on national standards. They
are collected often, but not always under legislative mandate, and they used
We have a new kind of data system, and I won’t get into the nuts and bolts.
We don’t have time, but on apcdcouncil.org is everything you want to know about
all payer claims databases. This map is old. I just noticed today that we have
three dark blue. The dark blue are implementing states of all payer claims
We have some more light blues that have been in touch with us and who are
planning. The medium blues are going to be dark blue soon. They are moving in
that direction. These are claims database that put together eligibility,
medical pharmacy, everything. They are really working in tandem with hospital
discharge databases, and the vision is so exciting. The reality is a little
different, but the vision is so exciting.
We were talking models and roadmap. I love the community rankings model.
This is the model that is kind of a proven model when we are on the ground. I
won’t go into all of the steps of building a data system. These aren’t one-off
data systems. These are data systems that are longitudinal. They go on and on
and we hope in perpetuity. The two areas in the boxes, funding is a huge
struggle. Every time I hear from folks on the ground that discharge data
something isn’t working or an all-payer claims database isn’t giving them the
data as quickly or timely as possible. It often is linked to funding because
the funding decides governance. It decides sustainability and analytic
Again with all payer claims databases, we are largely not funded for
applications, so we have a lot of data warehousing going on, especially with
all payer claims. There is not a lot of experience or tools to get these
complex data systems out. Again, I still defend the collection, even though the
applications are a little weak.
The governance, we can talk about that all day, but most of them are falling
into a state-led with a few public/private initiatives going on. Typically, we
will put commercial and then move in. The center is the Holy Grail. If any of
us could figure out how to get Tricare, VA and Federal employee and IHS into
these data systems, a lot of states would be happy, but we are not putting that
on my 2013 work list. We have worked with CMS on Medicare. I think Nile Brennan
and his team, have done wonders to get states access to the Medicare date.
Realizing that time is short, this is a typical day at NAHDO where I
will get a call from a Governor’s staff for legislative research or someone in
a hospital in a state saying the state of X is going to do what. They are
creating an unfunded mandate. They are going to take proprietary data and make
it public. They are going to create a government repository of patient records,
and they want physician identifiers. Then they are going to make report cards.
Then it goes on and on. Basically, the short answer is yes. We are here to help
This is really quite a fun process, and I have been through it more times
than I can count. It really is a process, and it can work. They can come on
board. That is usually the first call that I get. Lessons have been learned. We
don’t have the analytic workforce, and maybe I should end with what the
recommendations are, industry pushback, and risk adjustment, all of these
Data quality, in the beginning, these data bases really are poor data
quality, hospital as an example, but now APCDs. As they get used, as more
people use them, as more people look at them, we find that data quality really
is not a static issue.
What we have learned with hospital data is that national standards are
essential. Nobody wants to pay for the data. They all want access, but nobody
wants to pay for the pipeline. Many of our discussions on the ground confuse,
purposely or not purposely, patient privacy issues versus proprietary privacy.
This is a very tricky issue. You cannot give short-shift to the patient privacy
issues, but you do have to put your laser on what the real concerns are and
work with those.
Data ownership, which is not a very good term, but it does crop up into
these discussions because of lack of trust of a feeling that the competitive
interests A weighs heavier than competitive interest B will kill a data system
as well. These are lovely sociologic and political conversations we have in
just about every state that has built a system.
We do have a vision we will merge the administrative data world and the EHR
world. I really look for this committee to set sort of a roadmap and a guide
post so we don’t have two separate worlds. Right now, I think we do. This is a
slide that I won’t get too far down in the weeds, but this is what we call APCD
Where you really start looking at the health insurance exchanges is with
HIEs and the need for shared services. We need shared directories. We need
shared patient directories and provider directories. No state data organization
is funded to do that on their own. If they could just tap in with their HIE. A
few states are doing this. I wish there were more. It is conceptually possible
to just take the best of what the EMR will eventually offer and hybridize the
data so we have hybrid measures.
Unresolved issues for this group today: I don’t know how we are going to
sustain these data systems. Funding is pretty rickety out there. I feel like
some days I am holding data systems together with bailing wire and duck tape,
and I am kind of not kidding. When we go to a state that is about to do an
all-payer claims database, I don’t have a good value proposition. I can’t say
it saved the State of New Hampshire, nine million dollars. It may not have at
all. We don’t know yet.
I have to say that the evidence is by having system-wide data, we all
benefit. The commons benefit. This is just based on past history, but I don’t
have too many use cases for APCDs that really resound with legislators, so they
still seem to want them.
Collection of identifiers. I mentioned this before, and I will mention it
again. We need all of these data systems to have identifiers. I need street
address in every hospital discharge database, and I don’t have it. The states
that we have gotten to put street address in, guess what, the Public Health
Department has done it. I am not mentioning states. Then they say well now we
have street address with hospital discharge data, but we are not giving it out,
not even to the epis and that surveillance program over there. This is going to
be locked down.
What are those workarounds? Even if we can get granular data, then it is
held pretty close and then I get the calls saying the state of X won’t give a
street address for spatial geographic surveillance. Data access will continue
to be a problem. Data sharing, I can’t tell you how many cross-boarder
initiatives that we are stuck on, where it is down to the lawyers.
We have two willing health departments, two willing hospital associations
and two willing states and two years. They are on the desk of the lawyers in
each of the state over wording. We have re-done the MOU like six times. Then
they say they will wait because they had some workforce turnover. We will wait
and see where we are at next year.
We have a pretty rickety analytic infrastructure. I do think that we need to
start thinking of public health databases including hospital and vital records
and others as data scientists. We need to train these workforce people into the
concepts of big data integration and data sharing. I think so many are just
silos of practices.
What is at state? Basically, in my world, if we start losing data systems,
which is a possibility, I worry about this, we lose trends on quality and cost
information. We lose some granular and longitudinal data at the community and
state levels on how people are accessing and using healthcare. The denominators
for morbidity, for public health surveillance are no longer there and basically
a common community dataset. I went through a lot. Walter, did I miss anything?
DR. SUAREZ: Thank you so much for that. I think we have about five minutes
for questions if anybody has any questions to any of our presenters today.
DR. GREEN: For Denise Chrysler, I would just ask you to say a little bit
more about your thinking and what you meant when you made the comment about
that it is important to distinguish data for surveillance and research. What
are the specifics about that?
MS. CHRYSLER: Legally, it is important to distinguish data for public health
and research because of the common role of federal protections or human
research subjects. For example, with newborn screening dried blood spots. It we
are collecting those for research, not just for clinical purposes or public
health purposes, then we have to meet those standards.
DR. GREEN: Is there conceptual importance in distinguishing? Intellectually,
what is the difference between for surveillance versus data for research?
MS. CHRYSLER: If it is feasible and we are collecting data for research,
then we need informed consent.
DR. GREEN: So there is not an intellectual basis that you are talking about.
MS. CHRYSLER: There is an intellectual basis. Public health has been granted
broader authority to collect the most intimate information about you because it
has a role in protecting people. Research is seen differently even though we
know it is very valuable.
DR. GREEN: For the other Denise can you just go back to the data access and
data sharing. Could you say more about the following: Our standard subcommittee
is quite busy in establishing standards for claims attachments? For data access
and data sharing, is there is an opportunity in the standards for claims
attachments or not?
DR. SUAREZ: Let me point out one thing. There is a standard for exchanging
clinical information, but between whom. There is clinical information exchanged
between providers and providers. That is what meaningful use is all about. That
is the standard between two providers. The standard is the standard electronic
message structure and content that a message that is sent from one provider to
another is going to be using. That is the standard. That is between two
Let’s talk about between a provider and a payer. The payer needs very
similar, but not identical clinical information that is being exchanged between
providers. That is what this attachment standard is about. It is allowing a
provider to send the same type of clinical information that it sends to another
provider for a referral for a specialty consultation to the health plan that
needs it for quality measurement for payment for operation.
It is the exact same attachment concept that is applied in two different
scenarios. In the end, I think the concept of convergence and coming together
with the same standard is what is going to hopefully prevail.
It is going to allow us now to get to the point that I think Denise at the
end was talking about, the two worlds, administrative and clinical are going to
begin to share the same standard, the same content, the same data. It is not
going to be like interpreted data or converted data into one code to another
and that kind of thing.
DR. GREEN: Any prosumer interest there.
DR. FULCHER: Well I think there is great prosumer interest, but it depends
on the type of data we are going to be asking for. It is not individually
identifiable and it is more about built environment and the conditions of our
streets or walkability and bikeability. There are a lot of folks who are using
big data and prosumer approaches to get very personal data. I think that is
something I really like to have a much deeper and longer discussion with this
group about. As we go out there and start doing the very simple survey, what is
appropriate, what is not, how far can we go? What can’t we do? There is a great
potential around prosumer input here.
DR. SUAREZ: Bruce and then Leslie.
DR. COHEN: I was going to just pose a question to each of the three
panelists. If you were the secretary and the feds could do one thing, what
would that one thing be? You don’t need to answer it now, but for each of these
panelists, I want to make sure that you answer it in your small work groups.
DR. SUAREZ: Thank you, Bruce, because I think that is the question that we
are going to address in the panel.
DR. FRANCIS: This is probably occasioned by the fact that I am lawyer, but I
would like to hear a little bit more. I don’t think we are going to be able to
do anything about the fact lawyers protect proprietary interests of those they
Is there any anything that you would suggest that we recommend to try to cut
through the two years kind of problem? Would it be cooperating with your group,
Denise? What would help you, Denise? This is probably a question for both of
MS. CHRYSLER: I would just say that when HIPAA first became effective April
14, 2003, I will never forget. There were no tools for us. We were just sort of
sink and swim on our own. Today, if you work in the field of privacy, HHS has
put, I get like once a week a new set of tools to help us in understanding how
to apply the public health laws. In fact, it is all of the information in HIPAA
and public health that was not available in 2003. Looking at what type of tools
are made available to lawyers by HHS and how can you facilitate freeing up the
data here would be one thing I would suggest.
MS. LOVE: One of my dreams is to someday have funding for a system of
sharing. These one-off sharing things just don’t work. It is clear. It works
sporadically. I look to my cousins in vital stats, and they have a thing called
STEVE. It works. It took a while. It took some planning, and it took a system.
We have talked to the vendor. It probably wouldn’t be that hard to adapt it for
other data systems. Say hospital discharge data for one.
The data collectors come around and say here are the business rules,
apriori. Then we agree on what that common exchange looks like and then it is
instituted. I think then it is not lawyer dependent. It is not person
dependent. Even if you data exchange in state A and B, once that one person
retires, you have to back and remind everyone why they were exchanging that
data between those two states. It is very important because we have overlapping
markets, the measures can’t be made. The comparisons can’t be made. STEVE is
working for the vital records. I think that is something we need to look at, at
a higher level and try to design systems and policies that sort of bridge all
of this noisiness down below. I think it is possible. You asked.
DR. SUAREZ: Thank you.
MS. KLOSS: We are at break time, but before we break, just a word about how
we will conduct ourselves in our final small group discussion. Showing up there
are the original set of discussion questions. When our design committee met, we
thought we don’t know enough yet to figure out how to frame these questions.
We put those four up as placeholders, thinking that we would listen
carefully yesterday and today and reshape the questions that we would like you
to walk through in your small groups. That is what the bolded first page of
this new handout is. There are six questions. Three are recommendations
pertaining to communities, government, community partners and that is a broad
list of potential partners. You probably won’t get to all of them, so decide
where your sweet spot is. Recommendations for variables, concept standards,
recommendations for security, privacy stewardship, and then we put in a six
which is the other transformative recommendations.
We just have an hour, and I know we are primed to get to recommendations
because everybody was prefacing their remarks that we should be doing this or
that. We are ready to get to recommendations. Don’t be frustrated if you don’t
feel like you have tapped all of these areas because when we come back instead
of a 30 minute quick report out, we have got virtually the rest of the
afternoon to do a report out.
As one of the groups reports out, we are going to have discussion about that
report-out before we go on to the next. We are building on recommendations.
Then tomorrow morning, we continue to massage recommendations. We have got a
lot of time for this next chunk. If we don’t get everything through the small
group, we will get it through as a group as a whole.
MS. KANAAN: This is perhaps obvious to everyone, but the first three I
noticed, are actors we are making recommendations for. Four and five are really
the word four could really be the word about. In a way it is a different
approach, different framing.
MS. KLOSS: Let’s take a short break and reconvene in small groups at 2:15.
MS. BERNSTEIN: I just want to make clear that the meeting continues
tomorrow. Everybody is invited and encouraged to come tomorrow morning as well
to the rest of the committee’s deliberation.
MS. KLOSS: If there is any reason you need to leave before, we are intending
to do a written report of this meeting and share it with you for your input.
You won’t wait until we get final work products done. We have committed to
doing a report of our experience here during these two days.
MS. KLOSS: So I think this is the time in the meeting when if we want to
stand up and roam around, that is all okay. We need your complete attention
because this is where the meat is. Who wants to begin, blue, red or yellow?
MS. HART: So in the red fashion, we didn’t do things following any rules. We
made up our own rules. We did come in tackling each of the different areas. We
are just cross-cutting themes no matter what we talked about. It just kept
coming up in each category. That is education, marketing and messaging at all
levels. Whether you are talking about data, whether you are talking about a
community, we really need to get the message out and how do we do that in an
appropriate way. At all levels, we need to align the incentives to get to the
endgame. We did the full exercise of going through what we have recommendations
for communities and community partners, but when we got done with it, there
really wasn’t any major difference in the recommendation, just the level of
granularity for the recommendations.
Meeting guidelines for defining what is a community and ensuring what are
the constituencies of that community. Is it a demographic community? Is it a
diseased community, or however you may define it? Then when you do define it,
make sure that it is fairly done. Then we have the next one of how do you have
inclusion and engagement process? How do you identify key stakeholders, getting
to your mission and leverage expertise that already exist?
What are some practices or tips for avoiding marginalization,
discrimination, or geostereotyping? Leveraging the expertise everywhere from
local to international and aligning the resources and investments. One of the
things that struck us is there is so much out there, let’s not recreate the
wheel. We said that a few times in our conversation. The recommendations for
government was enable all of the previous stuff I just said, but also again
make sure the funding from government. The government is a major source of
funding for most of these initiatives. Let’s ensure that it is in according
with our definitions of community in getting us to what we would like to do.
We talked about communities aligning and partnering. How do you incentivize
them to do that when right now a lot of the funding sources force them to
compete, or they encourage vendors to compete? Think about how that works when
you do funding. Of course, reexamine policies in light of a changing landscape.
Everything again from data privacy standards and account for intended versus
We were talking about misuse as well, accounting for misuse of information.
It strikes me that we always kind of talk about the negative, but we don’t
always talk about what we actually intended to happen and how do you measure to
get to that intension. Another one is to bring national strategic initiatives
down to a local level. A, what are they, and how does it relate to them as at a
local level and then B how do they start measuring collecting data and working
towards that common goal.
For privacy, definitely a framework for repurposing data; that was a big
issue. Doing the landscape as well as a gap analysis for data protections, so
what is already there? Where are there overlaps? Where are there are gaps?
Where could we have things harmonized? We talked about things like HIPAA and
FERPA. Standard definition of who is bound to privacy. We also talked about
that, so sometimes the privacy rules are such a narrow view of who all should
really be covered by privacy, and then how to do your harmonize those rules?
When looking at variables and data, develop models for community consent to
use data. We had the privacy really tackled individual consenting issues or
individual privacy, but how do you deal with community privacy and consent to
use. Incentivize the Standard Development Organizations, or the SDOs, to
collaborate and how they would then use common formats health data categories,
promote uniformity, interoperability, and harmonization. Again, don’t recreate
the wheel, get them to work together and incentivize then in some way to work
Consider what is sensitive data and its applicability to public health or
population health. In other transformational ideas, this could have been
cross-cutting too, but explore what public good may come from uses of big data,
emerging data, and how do we deal with the tolerance of ambiguity in this
landscape of today. Policy should be flexible and scalable to adjust to
advances in technology. Policy could take forever. The same research could take
forever. We don’t have time for that anymore, so how do we adapt? That is it.
Anything else, Red Group?
MS. HOFFMAN: You kind of mentioned it, but I just wanted to develop it a
little bit because it took some time in our discussion and that is we have been
talking a whole lot about community, but to me and some of the others, it is an
ambiguous and a little bit of a dangerous term. If we use it too loosely, we
might be defining communities in ways that disempower minority elements.
We might say the community is this town or this city, but that means
majority rules, and we are marginalizing or we are discriminating against
religious communities, communities of color, and communities of disability and
so on. I have been a little uncomfortable the whole time just assuming that
community is a clear term and we go from there because I don’t think it is.
DR. COHEN: This conversation has been an undertone probably for a variety of
conversations. What we thought we would stick to, and we should all consider,
is the definition of community that we used in the CHIP report, the Communities
of Learning System Report. I would like to read you that definition of
community and see if everybody is comfortable with using that right now.
“A community is an interdependent group of people who share a set of
characteristics and are joined over time by a sense that what happens to one
member affects many or all of the others.
This sense is sometimes combined with recognition of mutual
responsibility.” Then we go on to essentially say communities can be
geographic, they can be diseased, they can be subgroups, they can be defined by
many different conditions, but the essential definition of community is an
interdependent group of people with a set of characteristics who are joined for
a purpose. Is that too broad? Is that too narrow? Is that unacceptable?
MS. HOFFMAN: I don’t have a clear answer, but I still think there are risks.
I live in a little town, and they have a football field with a college. They
keep talking about how great it is for the community, so they are adding more
and more and more night games and weekend games that will benefit the
It is killing my quality of life because at 11 o’clock, there are loud
speakers tormenting me and lights in the room. You are saying it is great for
the community, but you are really disserving the people who don’t want football
games at 11 o’clock at night every night.
DR. TANG: I actually like the way it was written, and I still like it. It
has a lot of meaning, and it goes back to what Barbara says as well. The
question I have, and I don’t whether it is time to ask it or get an answer is
do the data that we currently have can they be applied to the communities as
defined like that? Do the data we have on populations, can they be applied at
DR. COHEN: I would say some yes, some no. One recommendation would be to
collect more and detailed data to address communities that are
under-represented or not represented. I am focusing, for instance, on sexual
orientation data or adding information about disability or more detailed
ethnicities. These are communities that are currently probably hidden in most
DR. TANG: When you say some and some, can you give me an estimate is it 40,
60 or 20?
DR. COHEN: Again it depends on what particular community or communities you
are defining. For many communities, there is very complete data. For some that
I identified, the data are mostly incomplete. Again, it is tough to give you a
MS. KLOSS: It was exactly the first question out of the box for the Yellow
Group too, and that is what forced us to go back and look at what we, as I
recall, suffered mightily over getting that definition that was in the CHIP
report. That was worked at very diligently by the committee.
DR. TANG: If we are focused on actionable recommendations, this will
probably come up a lot more tomorrow. We talk about measures that matter, I
think we have to talk about actions that matter. They actually matter to
micro-communities of people. I don’t think they apply to even a zip code. That
is just a subject thing.
MS. GREENBERG: Can I suggest we get all of the recommendations out.
DR. COHEN: I think having some sense of agreement or agreement to disagree
on fundamentally what we mean by community, maybe two or three more comments,
and then we can continue.
DR. CARR: I was simply wanted to point out how a community votes to do
something is different from what is a community. I like the definition. I
sympathize with the Friday Night Lights, but I don’t think that undermines the
definition of the community.
DR. FULCHER: It is really the first time I had heard that definition. I
would like to mull over that as well and just not really make any type of
decision now. I think it really is not necessarily geographic. Communities are
really issue based. You have stakeholders in the same community that would be
much more polarized or focused or emotionally drawn to some issue and not to
another. The issue framing really defines community as well. I am finding that
we have communities physically, but the virtual communities transcend our sense
of physical community. There is a lot more to it.
DR. COHEN: There is nothing in this definition that had a geographic
DR. FULCHER: It sounds good. I like it overall. I am not disagreeing. I
think I need to process it more.
MS. BERNSTEIN: When Bruce talked about it, and I have heard this definition
before. I thought I agreed with it until we started talking about it. It occurs
to me that the subgroups you are talking about might have a problem with what
the larger community is doing.
You referred to those, and maybe not intentionally as communities
themselves. It occurs to me that those minorities of various kinds whether it
is issues or diseases or ethnic or racial groups, or whatever it is, they are
not communities necessarily. They might or might not, be organized and
politicized or whatever it is that takes the gel to be a community. They might
just all happen to have cancer and know nothing about each other and not be a
community as such, just some subgroup that has no way to advance its interests.
That is the point. Their ox is more likely to get gored if they are not
organized and politicized and have a way of getting represented. They have to
be particularly careful.
DR. EDWARDS: I think the reason that it was important to have this
conversation now as Bruce stated as opposed to when the other groups met is
because we have all probably made recommendations that are juxtaposed against
our own notions of what community is. In order for the committee to execute or
move forward with those recommendations, some of them may need more contexts. I
like that definition of community; however, Sharona’s point is that in order to
effectuate change, you need to know where and how that changes needs to be
applied. It is not one shoe fits all.
One recommendation might be totally applicable to a historically
under-represented minority group. It may not be applicable to the LGBT
community in terms of access to health information. If we put it in the context
of driving improvement in health because this committee is focused on the
utilization of data about health, I think your community definition will be
variable. My recommendation to the committee is that you make sure you have a
context for these recommendations before they are applied.
MS. KLOSS: One of the thought that we had as we came out of this last
exercise is that is was going to be really difficult for us to look across the
three groups and what they said about community the way we have the report out
and that we need to do one more transitional piece this evening, which is to
produce a handout that brings all of the recommendations pertaining to
community together in one place. I think that suggests as we do that, we will
look at issues of definition. We will certainly be returning to that issue in
terms of context. So know that we are going to do that. As we kick off
tomorrow, we will have reorganized this. I wanted to see if there are any other
questions or comments on anything from the Red Group before we move on. That
was great. Thanks for bringing that up, Sharona. Anybody else heard anything in
this report-out that you want to delve a little bit more or you don’t think you
MS. ZAPPIA: So our group we also started out with talking about what
community meant. We added the recognition of the continuum of where communities
are at in using data for local wellbeing and understanding that communities
follow different points along the continuum. Our first recommendations four and
two communities engage local and state health departments to figure out where
the data area, for communities to identify potential partners including
healthcare organizations, identify the communities strengths and weakness. So
you try to self-identify where communities need assistance, where they are on
Can a mechanism for self-assessment be developed? Community should develop
the ability to collect primary local data, and that is both qualitative data
and quantitative to augment the data that is already available. Moving on to
recommendations to government: We should shine a light on best practices and
case studies. We should partner with the evolving NGO intermediary and
facilitation organizations such as community commons, Zerodivide and county
health rankings and make full use of the supports they are developing for
communities on a national level.
We can provide statistical small area estimates to make the data more
actionable at the local level. We can do further work to understand what
information is needed beyond that in the contemporary government surveys to
provide adequate insight into community health and wellbeing. An improved
coordination and collaboration of federal data related services and supports
for communities are recommended; for example, making use of the CHNA
Further discussion about the CHNA is that it needs to be community-driven
and co-owned. We talked about that issue of co-ownership a lot. The IRS CHA
requirement is a potential point of leverage, and government should require
these assessments to be community-driven.
You can develop mechanisms to feed the CHNA information and other usable
information back to the community where it can be used and see ways to hold for
profit hospitals to the same standards. We can seek ways to provide and present
data in ways that show the gaps and make the informational actionable and meet
the community where it is on the continuum.
We can improve resources at the community level including coaching, small
area estimates, and analytic tools to assist every community wherever it is.
Expand opportunities to collect qualitative data and provide guidance to
communities for collecting these data. Refer to NCVHS recommendations for
federal role in the CLS report, expanding them based on new information. For
example, continue to identify and encourage adoption of standardized community
Recommendations for community partners including universities and academics
encourage academic institutions and provider organizations to reevaluate
co-ownership of joint projects.
MS. KANAAN: Re-evaluate the extent to which their joint projects are truly
MS. ZAPPIA: We have recommendations on data stewardship and NCVHS should
reaffirm the stewardship model developed by NCVHS and urge its promulgation.
Federal and state governments should increase community awareness of
stewardship and privacy related issues in data. NCVHS should identify potential
partners to work with in pursuing the issues and stewardship recommendations to
address these concerns. Federal government should provide guidance to help
communities evaluate how to balance risk and public benefit with respect to
Finally, other transformative recommendations: We should create a mechanism
for connecting academics looking for community-based projects with communities
in need of this academic assistance. National public health organizations
should identify partners who can be conduits to the local level to help engage
local communities in neighborhoods, identify their needs and promote awareness
of available data resources.
MS. KLOSS: Question, discussions.
DR. FULCHER: A lot of work went into the CHNA effort on community commons. I
would encourage you to take a look at that and also provide feedback and input.
It is a large committee that made this a reality. Again, it is a public good
utility. It is making that available at no cost for all communities or all
hospital areas and nonprofits, et cetera. There is also a CHNA group that we
are forming on the commons to provide more of an ongoing dialogue about what
needs to be done in this area.
MS. KLOSS: Our group we realize that we didn’t know enough about it, what
kind of formats were being used, what data, what level of collaboration or if
it is too soon to know.
DR. COHEN: I guess some points of clarification. Are you talking about
community health needs assessment in general or the particular IRS Mandate for
DR. FULCHER: What we developed this CHNA tour for was basically to meet that
IRS requirement. CDC, working with Paul Tang and their team there and with
Kaiser Permanente and their work and ACHI and others really came out with that
tool in mind as far as meeting that short-term need. Paul kept cracking the
whip and saying you have got to get it out there to help these nonprofit
hospitals to meet that. That is where we are at now.
Once it is evolving in a community commons framework is people are using
that CHNA report a lot of other purposes than just the IRS requirement around
grant seeking, better understanding their context, their regional context. As
much input as we get, feedback, what is working, how we can improve on it,
indicators, there is an advisory group around this CHNA process. Did I answer
MS. KANAAN: There is an interesting opportunity here around that to get very
concrete about several of the recommendations where they come together around
CHNA because we talked about recommending that the federal government
collaborate with and use these national infrastructure organizations like
community commons. We also had a recommendation in our group that the
government mandate community-based needs assessment that it would be a
community driven process.
It is possible that we can consider bringing those two things together that
we cite your CHNA tool as specific resource. I don’t know that the government
is going to mandate the use of that. Also, we would encourage you, community
commons, to have some suggestions about process because we have talked a lot
about co-ownership and how the process is done. Maybe there is a way to bring
all of those strands together.
MS. HART: One of the things of the Community Needs Assessment. I think the
tool is absolutely fabulous to give hospitals and communities a leg up on how
they understand their communities. A Community Needs Assessment is by not for
profit hospitals. What do you do when there is more than one in a defined
community? The other component to that is each not for profit hospital has to
state how they are going to react to the community needs. In our particular
community one of the challenges was how do you report that someone else can
take care of that need in the community, and this is what my unique
contribution is? That is where the process became the most important once we
understood the data.
DR. FRANCIS: So this is the question for possible partners who were here.
One of the recommendations was partner with the evolving NGO intermediary and
facilitation organizations. There are three of you here. I guess you are all
still here. There are potentially others. Bridget had to leave. My question is
what do you envision that partnership looking like? Do you have some thoughts
for HHS that might be born in mind, and are their other NGO intermediaries that
you know of that we don’t that may be should have been on that list?
MS. KANAAN: That is not intended as an exhaustive list, by the way.
DR. FRANCIS: I know that. We modeled partnership here. We invited you here.
There is clearly much more that could be done by way of partnership. As Denise
indicated, it has been hard for us to get the word out about ourselves, about
NCVHS, but how should we become aware of potential community partners for HHS
DR. FULCHER: In terms of the type of partnerships advisory roles and
formalizing advisory groups around these specific areas, around CHNA. Another
advisory role for example with community commons is around data itself. Another
advisory role with community commons is around public policy. If we talk about
how we might formalize partnerships, I have not a governance person; I am more
on the technology side. We are very open because community commons is
all of ours. How do we best pull ourselves together to advance that public
utility? We are open into different types of partnership models or ideas.
I think that’s what I can say at this point. What we have been doing is this
emerging advisory role to community commons around specific areas. If it is
around CHNA, that is great. I think the committee would be tremendously hopeful
around this area of looking at that crosswalk between the clinical and health
data and what I presented earlier and how we best represent that. There are a
lot of opportunities. I am asking the same question. I don’t really know what
the opportunities are for types of partnerships, but we are wide open because
it is what we all want to do together. Did that help answer your question?
DR. EDWARDS: I would echo what Christopher said and add to that that a
meaningful partnership has some level of equality or equitability in it in that
it has to be mutually beneficial and so the value will come from having a goal
that both entities want to work together to achieve. Under the domain of health
data, vital and health statistics, I think what is going to be beneficial to
the community is the capacity for that information and that data to help drive
their own goals. To the extent that this committee they can work in partnership
to put the data in a form that is translatable and actionable is going to be
the benefit or the value to the community. We would hope that you have a mutual
goal of wanting to effectuate change through the use of information that you
are managing or collecting.
A meaningful partnership would involve asking communities directly what they
want and then what resource they can both bring to bear to get to that in
state. It is really hard to have those meaningful partnerships when they are
not mutually beneficial. I think the first step is identifying what the need
is. That has to come from the community. Listening session would be a
recommendation that I would have. I truly appreciate being invited to come
I think you would get a totally different picture or story if people from
the committee went “there” and actually were forced to listen to
them, whatever demographic it is. I almost want to say not community anymore,
but something else. It is like meaningful use; I don’t like to use that term
anymore, because it has no meaning. My contribution would be thinking about
listening sessions and approaching any partnership with the understanding that
it has to be mutually beneficial to both sides.
DR. FULCHER: I want to preface this by saying I am not speaking for Andrew
Bazemore or Bridget Caitlin as well. We have that common understanding around
public. We are all driven by our funders and the project work we do. Andrew
does great work with what his system has to offer. Bridget does great work with
county health rankings and beyond.
We recognize the great value that we can do something together, but I just
want to stress that we are pulled in ways and we are so busy meeting the funder
requirements based on what we are doing. We have that under-pinning. It has
been such a pleasure just meeting you as well. I feel from your presentation
that you are just right on that same page there as far as our under-pinning is
MS. BERNSTEIN: Maybe in the same book.
MS. HART: So Leslie, a very practical thing would be to look in the
government’s own back yard. Many of the things we are talking about and whether
or not you are going to have a partnership. Chris said it. It is actually
driven by funding incentive. When we talk about mutual value, I might say
mutual value, but if I am not funded to work with you, it makes it very
difficult. The other frustration because I do actually manage Cross
organizationally funded programs, and it strikes me every time when they ask me
for something and I ask why are you asking me for this information. I already
reported that to government agency X and government agency Y. Why do I have to
do that again? How come you are not talking to each other?
The same is said for even these recommendations. They have been made before
in other arenas. Some of these are not aha moments. How do we work together? I
think I said that in our little Red Group too. Government can work together
first, and then when you start looking at how you engage industry and what are
the incentives to engagement?
MS. KLOSS: Any other comment? Blue.
MS. GREENBERG: I am going to start by apologizing to the organizers because
I don’t think that I was supposed to be one of the people who reported out. I
got so energized that they recruited. Of course that was half an hour ago, and
my energy level has gone down quite a bit, but I will do my best. I do want to
thank all of you buy the way. It has been an incredible few days.
I am going to start with the recommendations for communities because they
kind of relate to what we have been hearing, and then I will go backwards. Our
first recommendation to the communities was that you stay engaged with NCVHS
and that you spread the word that we are working in this area and we are
interested and we are committed to it, et cetera. I think we were struck by one
of the many Denises who are here right now said that last month had a meeting
on this topic. That is out question, and we will do our best to stay in touch
with you and others. We have to think about how to do that best.
We had a recommendation that there needs to be a paradigm shift for
universal design in communities to reduce barriers to participation in all type
of community activities. We also recommended a universal design approach to the
government at all levels.
Also, there was a request that communities work a definition of emotional
wellbeing and quality of life for healthy people 20/20. This was specifically
addressed to communities as opposed to suggesting that the government do this.
I will start it with that. Our number one recommendation: What we did is we
went around the room. Everybody got three, and then we just kept going.
Then we voted. This was systematic. The one that got the most votes, and we
had talked about this in our third break-out session too, was support for the
agricultural extension model for data support analytics support and stewardship
for communities. There is something kind of like this in the ACA section 5402,
which I think is called the Primary Care Extension Model, but it is not really
just meant for primary care. It is a broad definition, or so we believe, of
primary care as partnering with public health and communities, et cetera.
Almost everything could have come under this. Being guided by Paul, at one
point, I did want to put everything under this, a lumping model, but we kept
going. The next one definitely comes under this.
A subtopic, and we haven’t really defined this, but is that there needs to
be a health statistics modernization act. The whole idea of all of everything
from collection to analysis to stewardship to use, all of that sort of needs to
come into this 21st century of communities as learning health
systems. That needs to be flushed out more, but I think you get the flavor of
MS. KLOSS: Would that be separate?
MS. GREENBERG: Well it could be separate. That is the one we started sort of
lumping. It could be a separate one. Clearly, if you were going to support this
agricultural extension model, and it was going to really cover the U.S., you
would need to build the scientific under-girding and workforce for the new
world of communities as a learning system and the community as a learning
system. Then you want the health statistics modernization act to reflect that
new world. That is how they are sort of all interconnected.
Right now, there really aren’t the resources to put someone in every
community. We heard that time and again about the limited resources even to
help people understand what it means, the disclosure issues, what statistical
requirements are there? Although it started slow, it really gained support in
the voting process and that was to keep mental health data in our health
databases with appropriate protections.
This is a little challenging doing all of this on the fly. Clarify the legal
protections for sensitive information. Really just stop marginalizing mental
health data. If you think about it, it does relate to the definition for
wellbeing and quality of life because you can’t separate that from a lot of
mental health issues. I, like Paul, like to connect the dots, but that is good
because then we are not having a lot of disparate recommendations.
They have told me in WHO Network that you can retire that is fine, but you
can’t retire from the WHO Network. I suppose it is the same with the national
committee, but you can’t have too many of those. Then we had a recommendation
of a short, and these two kind of go together. They could be separate, but we
will put them together, a short menu of health challenges that a community
could investigate. There are million things you could do. You can’t get health
challenges, use cases, whatever.
So much was what we heard and what talked about, but I think what we heard
in the big room and in the smaller rooms was the whole process of community
engagement starts something that can lead to something you never even thought
of because people are talking to each other and they are finding new partners,
and they are listening hopefully to each other, et cetera. If we had a
challenge, like the apps challenge. If every community was challenged to
address just one thing on this menu, and it would be a short menu, and we gave
them some tools, et cetera, you could start a process that maybe would
snowball. Of course, data would be part of it.
We also talked a lot about going from data to interventions to evaluation.
Develop and teach how to use a parsimonious set of valid measures that are
important to providers, consumers and the community, which builds on the
measures that matter. It is not just quality of care ones. The measures would
be connected to whatever your use case was in this menu. Reduce and remove
cross-jurisdictional challenges with data exchange and collection. One thing
that was suggested that maybe didn’t get here, but it fits in with this was to
develop a guide on how to do that. Some of that might actually require changes
in laws or at least in HIPAA regulations. It is certainly the messaging and
marketing that the first group or the second group talked about.
It fits in with a lot of those. It was a big problem that kept coming up.
Here was one that could have put it on the transformational one. It is to
create an office of community science innovation at the White House. We don’t
think small. Work with the White House Office of Science and Technology. We
have already got some precedent here. Integrate science in society for better
community health. We need this to be the year or the decade of the brain. We
want a decade of community health. We would like to crystallize and harmonize
policies for the three categories of population health data, which we had
already talked about several times, public use, limited and then very limited.
How to navigate among them and best practices? How do you make decisions as
to which one does it fit into? Funders both public and private: Need to accept
funding “indirect costs” of communities to become learning health
systems. That includes like the costs of collaboration, maintaining an
infrastructure, capacity to do the work, and you shouldn’t have to rely on
volunteer efforts to do that, which is mostly what is happening. There needs to
be a set of tools for sharing data at the level discussed in section 4302.
Those are the basic things that ACA required standards for in federal data,
which are that they could be disaggregated or displayed by age, race,
ethnicity, primary language, gender and I think disability and provide guidance
on small area estimates when that is the way you have to get them.
MS. KLOSS: Any discussion on what you just heard?
DR. FULCHER: I work in a land grant university, and I am housed in the
college of agriculture and work quite a bit with extension. I think it is
really important to know that we often thing of extension as an ag-model. In
fact, extension is working in health. It is working around nutrition. It is
working around entrepreneurship. If there is one recommendation it is really
trying to convince USDA funding. I think it is coming mainly through USDA
funding, but extension is really hurting. I am not speaking for extension.
I am not with it, but it is an observation by working closely with them. We
have a real opportunity with a great brick and mortar capacity across the
entire country that is withering away because we are not realizing the full
potential of an extension reinvented with what we are talking about. There is a
mindset and many people around ag, and even some people that work in extension
have that you talk across rural America, under-served populations, a real
strong recommendation is that somehow with HHS working with USDA really trying
to retain that presence. Once it is lost, it is lost. It is hard to get
something like that back.
Community development actually is another area we have extension specialists
around the country. Again, different states vary in terms of what level of
capacity they have around these different sectors. It is a real opportunity to
really help that whole infrastructure that is really eroding.
DR. MAYS: I guess I would like to broaden it because it is not just the USDA
budget that we are thinking about. Part of what the real struggle is, is land
grant institutions currently are struggling with the fact that they are no
longer state supported. They are state assisted. We would like that budget to
actually cut across several groups because we want to make sure that the
orientation includes things like housing and the social determinants model that
would cut across other funders.
DR. FULCHER: Thank you for clarifying that. State funding is just really
going down. That is what is really hurting.
DR. FRANCIS: This is a question especially for people who won’t be here
tomorrow, but just a general question to the group. The issue of the importance
of being able to use but also to protect sensitive information, mental health
information in particular because of its importance to communities. I wonder
whether anybody had any ideas. Would it be helpful to try to develop a use case
around mental health information? What kinds of ideas to people have about that
whole area and how to address it? Put some meat on the bones of that concept.
DR. EDWARDS: I think all of the violence that out nation has experienced
over the last several years is a strong enough use case that we need to start
thinking about the cause and not just treat the symptom of trying to ban guns.
If we can figure out how many people in a community are obese and then
create a vehicle for funding weight loss programs, walking trails and
addressing food deserts with the data that is available, we need to be able to
figure out how many people are depressed and then put the same amount of effort
into creating interventions that treat people’s depression. As a consequence of
a misconception of what can and cannot be reported around mental health and
substance abuse just don’t do anything at all.
I think part of it is not just a change of rule and regulation because
regulations actually are not as stringent as we perceive them. It is an
education about the importance of using the information getting it and then
using it in a way that can effectuate the change that we need. I think one of
the things this committee can do is bring some leadership to what you can and
cannot do with data that then enables us to determine whether or not we need to
change the rules.
As community, we need to advocate for the change in those rules and then
advocate for the intervention and the funding of those interventions based on
the data as it is made available.
MS. CHRYSLER: My topic is different than mental health.
MS. HART: I just want to echo that as well. I think when you talk about a
use case, but sometimes it gets you too narrow. The first step is to get people
used to using data for community issues and how that works. We talked about
that even understanding food data on obesity could be sensitive. We have kind
of gotten over that hurdle. Mental health is really no different. I would offer
though our stepping stone in our community. We are sharing mental health data
by the way. I didn’t mention that because that gets people all riled up. In our
community, we do, and we started with quality of life. We didn’t actually put
any stigma to any particular group, but really what is the overall quality of
life, and that was out stepping stone. That is how we were able to get into
that, getting over some of those barriers.
DR. CARR: I would echo what the Blue Group said yesterday. We need to step
back, get the landscape of policies that came into place at a different time,
different era, different availability of data, and look at the gap, look at
what we have, what we need, and what needs revision and almost start anew. I
think as we try to navigate the overlapping and the things that never
anticipated what we are dealing with today. I just underscore what you said. We
put it in our report as well.
DR. EDWARDS: I just want to shout out to my namesake, Denise Love, because
she is the one who put up the modernization act as a recommendation. It speaks
to what Justine just said. You could probably for Paul’s sake lump a lot of
these recommendations under modernization act of health and statistics
modernization act. I cam remember exactly what, but Denise knows, and she needs
credit for the record for that.
MS. KLOSS: Just a time check. We had time set aside for public comment. Do
we have any reason to believe that there is anyone who wants to comment?
MS. BERNSTEIN: We understand that there are no requests to
make public comment.
MS. KLOSS: We will do Vicky and then Denise, and then shall
we agree to call it a day.
DR. MAYS: One of the things about mental health data is there is a lot of
confusion. For example, the California Health Interview Survey collected
recently mental health data, and it decided to make it sensitive. It is very
difficult to access it. Yet, the federal government has datasets that go beyond
what was asked and will give you diagnosis. That is not sensitive.
I happen to be editing right now, a special issue which is on the
Integration of Behavioral Health into Primary Care. We have two papers that are
commissioned that are actually talking about how difficult it is to have a
continuum of care in terms of the sharing of data, both from the sense of the
providers as well as from the sense of individuals who don’t want notes about
their therapy, even when you say we are not going to say everything that went
on. There are a lot of problems around the interpretation of what results in
stigmatization that is making it difficult for there to be integrated care
between health and mental health. I think this is really an important
MS. CHRYSLER: The one thing today that resonated the most with me is the
whole idea of a health statistics modernization act. In our group I just
mentioned, do we need a federal law, and there were groans because people think
of HIPAA. They think of the protections of human subjects and advanced rule
making that would even cove the identified dataset. I know Sharona spoke well
about having lost site over the years of why that law was even developed. It
wasn’t about use of data. These laws have unintended consequences for public
Instead of focusing on how we address privacy, how do we address human
subjects, it is focusing on how do we affirm and progress towards the use of
health statistics to improve public health. We need to try to undo some of
those unintended consequences.
The thing I was racking my brain about last night was a lot of data issues
are state. It is state powers. HIPAA is because of interstate commerce, the
connection to health insurance and those sorts of things. How do se work on
health statistics modernization and be able to do it at a federal level and
have jurisdiction when it comes to the state. I assume the way the federal
government usually does that is through funding incentives. If you want an
upgrade in how you collect data and how you use data, then you need to meet
federal standards to get that.
DR. TANG: I think Denise is responsible for the making it sexy too. I would
just add make it safe sex. The reason is because a lot of harm can be caused by
misuse of data or use of bad data, and I think that it is the health statistics
modernization act is what is in the core. It is a lot of angst I have thought
these couple of days, but I think that recommendation is beautiful. It just
really would change the world if we just rethought our data plan for the
country under this strategy. That would include the HR, but they cannot do it
the way that it is now. That is where I have all of my angst.
DR. EDWARDS: You need to bring the Office of Civil Rights to the table. The
safe part is the Office of Civil Rights so that you can protect people from
discrimination and stigmatization.
DR. COHEN: There are models of state and federal partnership, the
Cooperative of Vital Health Statistics System, which essentially helped create
where we are today in the National Committee on Vital and Health Statistics.
This cooperative system leverages really a nice interaction between the states
and the federal government to create a synergy. I think this is a wonderful
model when we talk about health statistics modernization. Denise mentioned
There are a variety of inter-jurisdictional exchange agreements among the
state jurisdictions to provide data amongst the states that could be a model
for transfer across states and within states and to the federal government. I
think there are some existing models that we can reflect on as we think through
this really grand scheme.
MS. KLOSS: That is a really high note to adjourn the day on, Bruce. Did you
have any closing comments?
MS. MILAM: It relates to the mental health and sensitive data issue. I
really think it nests nicely with the overall recommendation that there be
guidance to states primarily, on how to do population health privacy well
within meeting current needs today. That includes mental health data. What I
see commonly is that people overreact and strip it from the database because
they don’t know how to handle it. It is such an important part of looking at
our health. I really think that that kind of guidance is needed everywhere to
be able to move ahead. I think people react in a very conservative protective
manner simply because they don’t know what to do.
MS. KLOSS: Well thank you all for working so hard today. Thanks to the
staff. Everybody, you just made all of this run and great day. We will be back
at it in the morning.
(Whereupon, the meeting adjourned at 5:00 p.m.)