Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter.
Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 69
3
Clinical Data as a Public Good
for Discovery
INTRODUCTION
Clinical data have immense potential to drive progress in health care
by providing the means to measure and track care processes and outcomes,
to develop and refine best practices, and to enable rapid discovery and in-
novation. Over the past decade, the amount of data captured in electronic
form has increased exponentially. Given the federal government’s encour-
agement of the adoption of electronic health records (EHRs), these data
will increasingly be augmented by clinically rich course-of-care data. With
the potential to be easily stored, aggregated, and shared, these data can
enable rapid learning and continuous improvement in the efficiency and
effectiveness of care practices (Blumenthal, 2010; IOM, 2007). Harnessing
the power of these data, however, will require efforts to address barriers to
their access and use.
The papers in this chapter explore the potential of clinical data to im-
prove research and health care, strategies to enhance access to health data
and information, and key challenges to ensure data integrity (e.g., privacy,
security, and proprietary concerns). Additionally, this chapter discusses op-
portunities to better inform and engage patients and the public as advocates.
The first paper, by Farzad Mostashari of the Office of the National
Coordinator for Health Information Technology (ONC), argues that as
the nation works to develop a unified health information technology (HIT)
infrastructure, efforts will be needed to identify a limited set of core data
that can meet the basic needs of multiple functions. As ONC continues its
efforts to encourage and support data capture and use, it is considering how
69
OCR for page 70
70 PATIENTS CHARTING THE COURSE
to create data requirements that are relevant and not burdensome, how to
reward patients and providers for the creation and documentation of struc-
tured data, and the merits of distributed versus centralized approaches to
information exchange.
Todd Park of the Department of Health and Human Services (HHS)
highlights the significant amount of data currently held by the various agen-
cies of HHS and how these data could improve the value, science base, and
patient experience of health care. He describes HHS’s efforts to open access
to high-value data sets and to encourage public participation in the use of
these data for socially beneficial purposes.
The quality and accuracy of research results depend on the availability
and integrity of data. Don Detmer of the University of Virginia discusses
opportunities to increase the quantity and quality of data for health care
and research. To achieve this end, he proposes several options for national
policy that address security and privacy concerns while empowering citizens
to allow their health data to be used for learning and discovery.
INFORMATION NEEDS FOR A LEARNING HEALTH SYSTEM
Farzad Mostashari, M.D., Sc.M.
Office of the National Coordinator for Health Information Technology
The mission and goal of the Office of the National Coordinator for
Health Information Technology is to improve health and health care for all
Americans through the appropriate use of HIT. ONC is therefore focused
on characterizing key system needs and outcomes in addition to determin-
ing how technology can be a means to that end.
Meaningful Use and a Learning Healthcare System
“Meaningful use” is the term coined by Congress to connect technology
to desired outcomes, and ONC’s proposed regulations represent our best
guess as to how technology should be used to achieve these outcomes. If
providers—eligible professionals or hospitals—use HIT systems in a mean-
ingful way, they will be able to qualify for payments from CMS.
It is our goal, however, that meaningful use be applied for more than
just qualifying for payments. Rather, it should be used to ensure that mea-
surable improvements are made in health and in the quality, safety, and
efficiency of health care. A secondary goal is to move beyond improving
care for an individual patient at the point of care to creating a learning
healthcare system. This is an outcome to which ONC aspires and a worthy
endpoint toward which to build.
The next decade will witness a fundamental transformation of the
OCR for page 71
71
CLINICAL DATA AS A PUBLIC GOOD FOR DISCOVERY
healthcare delivery system, including changes that in many ways will be
more profound than those that transpired during the previous decades of
American medicine. HIT will certainly help providers take better care
of patients, but there is also the prospect that, in the aggregate, electronic
health systems will contribute to a learning healthcare system and enable
providers to understand and influence healthcare improvement—whether in
public health, quality improvement, drug discovery, or clinical effectiveness
research. The HIT infrastructure of the future should not just be the eyes
and ears but also be the action arm of population health.
Key Considerations for Developing a National Unified HIT Infrastructure
A number of projects now under way focus on developing HIT infra-
structure for many different functions, including clinical effectiveness re-
search, drug discovery, quality measurement, and public health surveillance.
Because these projects are taking different approaches and building differ-
ent architectures, it is temping to call for a halt to work on these siloed
activities. Indeed, if investment in HIT continues to create and support
a multitude of data islands, the nation will not achieve a unified HIT
infrastructure.
This is the critical challenge faced by ONC in its work to implement a
national system, to define meaningful use, and to develop grant programs:
How can we work to develop a common, national HIT infrastructure while
not creating additional network and system silos? Because projects now
under way will not be put on hold, this is a critical time to bring key stake-
holders and HHS together to begin a discussion on laying the groundwork
for a unified HIT infrastructure.
Clinical Data Needs and the 80/20 Rule
One approach to clarifying key data needs for a learning health sys-
tem is to agree on a core data set—sufficient, if not perfect, for a number
of information needs. Known as the 80/20 rule, this approach advocates
starting with something simple that can be achieved now and developing
clever approaches over time. The American Recovery and Reinvestment
Act (ARRA) presented a significant opportunity for HIT in this respect, as
it gave patients a right to their records in electronic format. Even if only
a small fraction of the patients in any given system choose to exercise this
right, every healthcare provider and EHR vendor must produce a patient
summary document in a common format. This could be a key opportunity
to use the clinical care summary—which includes the medications list, the
problem list, the allergies, the lab values, and the patient encounters—as
OCR for page 72
72 PATIENTS CHARTING THE COURSE
the foundational 80 percent, the common data core on which to build a
unified HIT infrastructure.
As data users have a wide variety of needs and will require additional
information based on these needs, some users are likely to believe that start-
ing with the clinical care summary is inadequate. For quality reporting, for
example, detailed information with which to calculate exclusions may be
needed; for public health case reporting, information on whether a certain
infection was central line associated may be required; for drug safety, the
first date of prescribing of the medication may be needed. Although this
may be true in each instance, the nation will never achieve a unified HIT
infrastructure if the starting point is to identify a data set that is perfect for
each user. An alternative approach is to leverage the 80/20 rule and supple-
ment it with targeted additional data collection where needed.
Creating Infrastructure for Targeted Data Collection
The example of clinical trials recruitment at a storefront doctor’s office
illustrates this approach. One cannot expect a provider to use an EHR in
the routine delivery of care while also collecting all the information needed
to determine patient eligibility for a multitude of clinical trials. As one
example, an individual’s occupation may be a data element necessary to
determine eligibility for a clinical trial. Although those with an interest in
this information may think that requiring it as a data element should be el-
ementary, a provider’s front-office staff is not going to determine and select
each patient’s occupation from among the hundreds of potential Census
Bureau categories as a routine part of delivering care.
A more feasible approach would be to collect a limited amount of infor-
mation in the routine delivery of care, which could also serve as an oppor-
tunity to screen for the need for additional data collection. Instead of asking
each provider to record each patient’s occupation and expecting these data
to be collected in a structured form, a system could be developed to trigger
deeper data collection when appropriate (i.e., manual rather than routine
data collection on a small subset of patients). If a patient is diagnosed with
hepatitis A, for example, it is entirely appropriate to prompt the clinician to
inquire about the patient’s occupation—specifically, whether the patient is a
healthcare worker, daycare worker, or food worker. This type of approach
makes sense to providers, as the information requested is relevant to the
person’s care and is limited in scope. For clinical trials, minimal informa-
tion can also serve as an initial screen, with follow-up questions about trial
participation asked only as appropriate.
This example clearly illustrates that despite the broad data needs of a
learning health system, the nation should start with something simple—
perhaps the humble clinical care summary. Additional work could expand
OCR for page 73
73
CLINICAL DATA AS A PUBLIC GOOD FOR DISCOVERY
on this data core to create infrastructure that supports additional and tar-
geted data collection only as needed.
Adding Value for Providers and Patients
To ensure data quality, it is also important to consider the value propo-
sition for data capture and use. The provider is doing work that involves
both the care of the patient and the creation and documentation of the
structured data needed for other purposes. A business case needs to be
made to ensure that the latter work is done well. For recruitment of clinical
trial participants, for example, it would be beneficial for all if recruiting
appropriate patients for trials was not just the right thing to do, but also a
source of income for the primary care provider. As the owner of the data,
this notion extends to the patient as well. Perhaps patients should also
benefit financially from the use of their health information.
Choosing the Best Approach for Critical Infrastructure
ONC continues to identify the key principles that should underlie the
creation of a national HIT infrastructure and develop supporting policies
and incentives. A particularly important principle to consider as we work
toward achieving our goals is to use a distributed rather than centralized
infrastructure. Although a centralized infrastructure works well for some
purposes, it is not particularly effective for supporting the participation
of individual physicians in advancing a wide range of population health
missions.
A centralized infrastructure requires separation of the data producer
from the data, centralization of the data, and central analysis. In today’s
fragmented and heterogeneous system, this approach typically does not
yield the answers people expect in a timely way. The costs are always
higher, the points of failure are always greater, the system is more frag-
ile, and the quality and cost of cleaning the data (once separated from
the source) become prohibitive. My view is that a decentralized system
whereby the questions go out to the data and the answers come back
is a much more resilient, feasible, cost-effective, and privacy-protective
approach.
Part of ONC’s mission is to think about how such distributed queries
can be expressed in a standardized way. Again, the 80/20 rule can be ap-
plied to these issues—whether to develop a standardized data model across
spheres of activity or a standardized way of expressing the question or
receiving responses. Formulating a strategy across fields, whether public
health surveillance, quality measurement, or effectiveness research, will
require simple case definitions tailored to the needs of the field—perhaps
OCR for page 74
74 PATIENTS CHARTING THE COURSE
one case definition for public health, two (numerator and denominator and
maybe exclusions) for quality reporting, and four (i.e., a 2 × 2 table) for
much of clinical effectiveness research and drug discovery.
Next Steps
Throughout its work, ONC needs to consider opportunities to stream-
line data needs and create limited but multifunctional shared elements of a
unified HIT infrastructure. These actions can include identification of over-
riding policies and principles, such as those discussed above (“architecture”);
establishment of basic, shared infrastructure (standards, services, directories,
terminologies, value sets); development and deployment of common tools
(e.g., CONNECT, National Health Information Network [NHIN] Direct,
popHealth); demonstrations (e.g., Beacon Community Program); research
(e.g., the Safety and Health Achievement Recognition Program [SHARP]);
and coordination across federal agencies. ONC looks forward to receiving
the input of all stakeholders and to making progress on these critical issues.
How much progress the nation will make depends on our collective ability
to start simple and work together.
OPENING ACCESS TO HIGH-VALUE DATA SETS
Todd Park
Department of Health and Human Services
Across many industries, opening access to and encouraging the use of
quality data has driven dramatic innovation and performance improve-
ment. Recognition of data’s transformative potential is reflected in the Open
Government initiative launched by the Obama Administration in 2009,
which seeks to enhance the efficiency and effectiveness of the public sector
through improved transparency, public participation, and collaboration.
The vast stores of data captured by HHS agencies, for example, are an im-
portant national resource for enhancing the value, science base, and patient
experience of health care. The creation within HHS of the position of
Chief Technology Officer—whose mission is to leverage the power of data,
technology, and innovation to improve health—and the development of an
HHS Open Government Plan signal HHS’s commitment to the principles
of open government and to playing a catalytic role in the development of
a learning health system.
OCR for page 75
75
CLINICAL DATA AS A PUBLIC GOOD FOR DISCOVERY
Open Government
Fostering the development of open government is an important means
of improving government efficiency and effectiveness and ensuring public
trust, and is an ongoing priority of the Obama Administration. Almost
immediately following his inauguration, the President issued a memo-
randum to heads of Executive departments and agencies outlining key
principles of open government—transparency, public participation, and
collaboration.1
• Transparency means publishing data with the intent of improving
government accountability and enabling citizens to access and use
data in ways that are socially beneficial.
• Public participation means welcoming and accessing widely dis-
persed national interests and expertise to help the government
better accomplish its mission of improving the well-being of the
American people.
• Collaboration means the federal government will work effectively
across agencies; with state and local governments; and with others
outside of government, such as nonprofits, advocacy organizations,
business, and academia.
Collectively these principles reflect the reality that the challenges and op-
portunities facing the United States are so complex that no one organization,
regardless of how capable, ingenious, or well funded, can possibly address
them alone. Instead, efforts are needed to improve the availability of data; to
engage a broad spectrum of talent and expertise in the use of these data;
and to support the work, across all sectors, needed for meaningful progress.
A Presidential memorandum led to an Open Government Directive
issued by the Office of Management and Budget (OMB),2 which required
all federal departments and agencies to take steps to better support and
implement these principles. Key opportunities outlined by the directive
include publishing government information online, improving the quality
of government information, creating and institutionalizing a culture of
open government, and establishing an enabling policy framework for open
government.
In response to the OMB directive, HHS developed an Open Govern-
ment Plan, which outlines immediate opportunities to improve access to
data and to provide a means for their use in accomplishing HHS’s mission
1 See http://www.whitehouse.gov/the_press_office/TransparencyandOpenGovernment/ (ac-
cessed October 11, 2010).
2 See http://www.whitehouse.gov/sites/default/files/microsites/ogi-directive.pdf (accessed Oc -
tober 11, 2010).
OCR for page 76
76 PATIENTS CHARTING THE COURSE
more effectively. The plan represents a sharp break from past approaches
to HHS data and is an initial step toward better leveraging these data to
accomplish the following objectives: “to help citizens understand what
(HHS) does and hold (HHS) accountable; help the public hold the private
sector accountable; increase awareness of health and human services issues;
generate insights into how to improve health and well-being; spark public
and private sector innovation and action; and provide the basis for new
products and services that can benefit the American people” (HHS, 2010).
Key elements of the plan are summarized in Box 3-1. Of particular
note are 14 new data sets and three tools that will be released by the end
of 2010.3 These new resources are considered “high value” because they
can be used to “increase agency accountability and responsiveness, improve
public knowledge of the agency and its operations, further the core mis-
sion of the agency, create economic opportunity, or respond to need and
demand as identified through public consultation” (HHS, 2010). The plan
also includes five Flagship Initiatives: the Centers for Medicare & Medic-
aid Services (CMS) Dashboard; the Food and Drug Administration (FDA)
Transparency Initiative; FDA Transparency, Results, Accountability, Cred-
ibility, and Knowledge-Sharing; Freedom of Information Act Excellence;
and the Community Health Data Initiative.
From Open Government to a Learning Healthcare System
Open government is a paradigm for how HHS can help catalyze a
learning health system. HHS has a fiscal year 2010 budget of $840 billion
and comprises 27 agencies and offices—including Agency for Healthcare
Research and Quality (AHRQ), the Centers for Disease Control and Pre-
vention, CMS, FDA, and National Institutes of Health (NIH). It has as-
sembled vast stores of data that have the power to yield insights into how
to improve health. These data stores would be a tremendous public good
even if HHS were to invent nothing new and simply make these data more
accessible. Yet more is needed to catalyze and foster the development of a
national learning healthcare system. In essence, publishing data is necessary,
but insufficient, to drive the transformative change needed without support
for data use and improvement.
Developing a Data Ecosystem
Developing a system that encourages access to and use of healthcare
data will make available important information and guidance for patients,
clinicians, and many other health system stakeholders. Perhaps most impor-
3 As of April 2010, HHS has 117 data sets and tools posted on Data.gov.
OCR for page 77
77
CLINICAL DATA AS A PUBLIC GOOD FOR DISCOVERY
BOX 3-1
HHS Open Government Plan
Key elements (Flagship Initiatives indicated in bold):
• multi-faceted new transparency push by the Centers for Medicare & edicaid
A M
Services (CMS) in 2010:
— A new interactive “CMS Dashboard,” debuting in beta mode with the launch
of this Open Government Plan, which allows the public to visualize and
analyze Medicare spending with unprecedented ease and clarity—starting
with inpatient hospital spending. You can visit the Dashboard at http://www.
cms.gov/Dashboard/
— Creation of 9 Medicare claim “basic files,” one for each major category of
healthcare service, to be released from September to December 2010 for
free public download on Data.gov. These files will contain a limited number
of variables and be de-identified and configured through a rigorous pro-
cess, in close consultation with privacy experts, re-identification experts,
researchers, and key stakeholders, to ensure the protection of beneficiary
privacy and confidentiality
— A significantly improved user interface and analytical tool for viewing ex-
isting CMS COMPARE data on quality performance of hospitals, nursing
homes, home health agencies, and dialysis centers. This interface and tool
is debuting at http://data.medicare.gov with the launch of the Open Govern-
ment Plan
— Online publication of detailed Medicaid State Plan documents and amend-
ments on the CMS website by 2011
— The release of new national, state, regional, and potentially county-level
data on Medicare prevalence of disease, quality, costs, and service utiliza-
tion, never previously published, as part of HHS’s Community Health Data
Initiative by the end of 2010
• Food and Drug Administration (FDA) Transparency Initiative is being
A
formulated with extensive public input, and focused on (1) providing the public
with information regarding how FDA works, (2) proactive disclosure of useful,
nonproprietary information in the possession of the agency, and (3) ways in
which FDA can become more transparent to regulated industry. FDA is also
launching FDA-TRACK (Transparency, Results, Accountability, Credibility, and
Knowledge-Sharing), a new agency-wide performance management system,
which debuts in beta mode with the launch of the Open Government Plan.
FDA-TRACK, when fully implemented, will allow the public to monitor more
than 300 performance measures and 80 key projects across 90+ FDA program
offices on an ongoing basis
• ther new data sets and tools to be published from across HHS
O
• mplementation of a new process for the proactive identification, prioritization,
I
publication, and monitoring of data releases—to be coordinated overall by
HHS’s Data Council. This process will include an “HHS Apps Challenge”—a
public competition for the best applications built utilizing our data
continued
OCR for page 78
78 PATIENTS CHARTING THE COURSE
BOX 3-1 Continued
• major push to assess current HHS operations in support of the Freedom of
A
Information Act (FOIA), identify and prioritize improvement opportunities, and
define a roadmap to implement the improvements [FOIA Excellence]
• xpanded opportunities for public participation in HHS activities and for col-
E
laboration across HHS and with the world outside HHS—especially via the use
of new information and communications technologies. Through a new HHS
“Community of Practice” for Participation and Collaboration, Open Government
innovators at HHS will be able to network with each other, share learnings and
best practices, compile these best practices into an HHS “workplace menu”
of participation and collaboration tools, compare the efficacy of different ap-
proaches, and work together on common issues. The Community of Practice
will focus in particular on the advancement of innovative mechanisms for
participation and collaboration at HHS—mechanisms that apply blogging,
“crowdsourcing,” group collaboration, idea generation, mobile, and on-line
challenge capabilities to key HHS activities:
— Medical research collaborations—e.g., via the application of “crowdsourcing”
and innovative patient engagement approaches to research on diabetes and
women’s health issues
— Collaboration among HHS employees—e.g., via work by the Office of
the Assistant Secretary for Planning and Evaluation to research and pilot
advanced collaboration tools
— Better health care through better information—e.g., via the community-
driven, highly collaborative
— “Nationwide Health Information Network–Direct” initiative being pursued by
the Office of the National Coordinator for Health Information Technology
tant, ongoing data use will drive improvements in data quality and appli-
cability. HHS therefore seeks to catalyze the development of an ecosystem
of data by stimulating both data supply and use.
Building on the foundation provided by increased availability of data,
HHS will encourage public participation and collaboration by engaging in
an ongoing dialogue with potential users of the data. Through such com-
munication, HHS seeks to catalyze learning by advancing understanding
of what types of data could be useful, the limitations of currently available
data, and means of better supporting data use. This approach is best illus-
trated by the Community Health Data Initiative, which seeks to help Ameri-
cans better understand and take action to improve healthcare performance
in their communities. The initiative’s initial short-term goal is to build a
network of data suppliers and users. In the longer term, this network will
OCR for page 79
79
CLINICAL DATA AS A PUBLIC GOOD FOR DISCOVERY
— Innovation in the workplace—e.g., via the piloting of an online employee
idea-generation tool and challenge program by the CMS
• ommunity Health Data Initiative. A major new public–private effort whose
C
goal is help Americans understand health and healthcare performance in their
communities relative to others—and to help spark and facilitate action to im-
prove performance. As a core enabler of this initiative, HHS will be providing to
the public, free of charge and any intellectual property constraint, a large-scale
Community Health Data Set harvested from across HHS—a wealth of easily
accessible, downloadable data on public health and health care performance
across communities, including a major contribution of Medicare-related data
from CMS (i.e., prevalence of disease, quality, cost, and service utilization
data at the national, state, regional, and potentially county levels). The initia-
tive is simultaneously working with a growing array of technology companies,
researchers, health advocates, consumer advocates, employers, providers,
media, etc. to identify and deploy uses of the data that would be most ef-
fective at
1. Raising awareness of community health performance,
2. Increasing pressure on decision makers to improve performance, and
3. Helping to facilitate and inform improvement efforts.
Such applications and programs could include interactive health maps, compe-
titions, social networking games that educate people about community health,
enhanced web search results for health searches, etc. By leveraging the power
of transparency, participation, and collaboration, the Community Health Data
Initiative seeks to significantly improve the health of our communities.
develop into a community health data ecosystem in which a broad array of
users take data supplied by HHS and others and develop means for their
display or application in ways that drive continuous improvement in the
health of communities across the United States (Figure 3-1).
HHS tested the potential of such a data ecosystem at a workshop,
convened with the help of the Institute of Medicine (IOM) on March 11,
2010, featuring leaders from two largely separate worlds—healthcare and
information technology applications. About 40 individuals with deep ex-
pertise in either healthcare data or how to use data to change people’s lives
came together at this meeting, and an amazing thing happened. The infor-
mation technologists became excited about data already captured by HHS
and began a conversation with healthcare experts in a burst of creativity.
By the end of the day, ideas had been developed for 19 new applications
OCR for page 84
84 PATIENTS CHARTING THE COURSE
BOX 3-2 Continued
complexity and revealing new insights about important healthcare issues. For
example, The County Health Map, created by Ben Fry, allows users to explore
County Health Rankings data using key indicators. The resulting maps unveil hot
spots and highlight the factors that influence the health of a community. The maps
are designed to allow users to see both broad U.S. patterns as well as individual
county metrics.
Pillbox—Pillbox, from the National Library of Medicine (NLM) and FDA, has trans-
formed FDA’s drug label data into a platform for innovation. Working with patient,
clinical, and HIT communities, the Pillbox team broke through barriers to the
creation of applications based on FDA’s drug label data. Opening Pillbox’s data,
images, and search function enabled the building of innovative applications that
improve the health of citizens. Applications currently in development include voice-
activated and instant-messaging pill identification systems, as well as iPhone
medication identification and information applications. A group of programmers
at the Great American Hackathon used Pillbox’s data to begin development of a
Facebook game, writing hundreds of lines of open source code in the process.
This fall, as part of the Community Health Data Initiative, the Pillbox team will
repeat this process with other government health data sets.
County Health Rankings—The County Health Rankings—a collaboration
between the University of Wisconsin and Robert Wood Johnson Foundation
(RWJF)—uses a model to describe the overall health of a community, as well as
the factors that affect health. This annual “check-up” for each county in the United
States allows everyone to compare the health of where they live to that of their
neighbors or other counties in their state. Following the February 2010 release of
the interactive website (www.countyhealthrankings.org), the Rankings have gar-
nered significant attention throughout the nation. The use of summary measures
of health and ranking within states engages the news media and policy makers,
leading to community-wide collaborations to “dig deeper” and better understand
the problems, as well as to seek solutions.
Quick Health Data Online (HHS Office on Women’s Health)—Quick Health
Data Online is an interactive database that provides state- and county-level data
for all 50 states, the District of Columbia, and U.S. territories and possessions.
Data are available by gender, race, and ethnicity and come from a variety of
national and state sources. The system is organized into eleven main categories,
including demographics, mortality, natality, reproductive health, violence, preven-
tion, disease, and mental health. Within each main category, there are numerous
OCR for page 85
85
CLINICAL DATA AS A PUBLIC GOOD FOR DISCOVERY
subcategories. Users have the flexibility of presenting results in tables, bar graphs,
and maps, or exporting results into other processing or display tools.
The MedWatcher Application: Mobile Phones and Crowdsourcing for Drug
Safety Surveillance (Children’s Hospital Boston)—The application is designed
to engage users (general public and healthcare practitioners) in issues of drug
safety. The system has two functions: alerting and reporting. First, the application
alerts users about new drug warnings through both official FDA channels (such as
MedWatch Alerts) and through informal channels (such as news media). Second,
using forms tailored separately to the public and clinicians, we provide a user-
friendly tool for reporting information about drug side effects. Serious adverse
events from clinicians are automatically submitted to FDA, while public discourse
between patients creates a sense of community, as well as a source of information
for hypothesis generation for researchers. We believe these early efforts represent
an important step in engaging the public as participants in the public health pro-
cess and empowering individuals to make informed health decisions.
HealthLandscape (Health Foundation of Greater Cincinnati)—HealthLandscape
is a web-based data mapping platform developed in collaboration by the Robert
Graham Center and the Health Foundation of Greater Cincinnati that helps Ameri-
cans understand health and healthcare system performance in their communities—
and to help spark and facilitate action to improve performance and value. It allows
users to create and display maps of community health and health resource data. It
also allows secure access to data for limited sharing and for secure uploading for
display in combination with the other public data. The goal is to help users add a
spatial context to data in order to promote understanding and improvement of health
and health care. Tools include
• ommunity HealthView: A home for data relevant to health in communities.
C
• rimary Care Atlas: A place to explore physician workforce data and
P
federally designated shortage area data.
• ealth Center Mapping Tool: A Health Insurance Portability and Account-
H
ability Act-compliant, secure home for uploading, geocoding, and mapping
patient data to understand service areas, neighborhood penetration, and
relationships to local and regional populations.
• y HealthLandscape: A secure environment for users to upload and
M
geocode their own health-relevant data.
SOURCE: IOM Workshop, Community Health Data Initiative, June 3, 2010.
OCR for page 86
86 PATIENTS CHARTING THE COURSE
number who will buy homes in the next 5 years. Such an application also
has the potential to create a new business case for investing in health at
the community level, as it transforms health improvement from something
that is a good thing to do into something that also affects property values,
a principal revenue source for communities.
This innovation—just one of the many possibilities with increased
availability of data—underscores the need for public participation and col-
laboration to develop creative applications that serve local needs. Because
creating applications requires users to gain familiarity with the data, such
work also helps HHS identify opportunities to improve the quality and
utility of the data to support learning and health system and health im-
provement. The ecosystem approach therefore solves two problems. First,
data that were previously collected but not used are being applied in a
timely way to generate a discernable social return and a means to improve
data quality. Second, because of these returns, a business case is developed
for HHS to produce and deliver more data as a vehicle to engender social
change.
At a follow-up meeting hosted by the IOM on June 3, 2010, the 11
volunteer teams showcased their results to an audience that included indi-
viduals interested in joining the work under way or starting new projects,
organizations that could supply data beyond those captured by HHS, and
community leaders who could use the applications to improve health.
As a result of these initial discussions, the potential impact of these
data has dramatically expanded. For example, Clay Johnson of the Sunlight
Foundation has proposed launching a great American hack-a-thon for
health, in which advocacy organizations and nonprofits would compete to
develop innovative ideas for applications that leverage community health
data to advance their mission. The best ideas would then be handed over
to hundreds of hackers in 30 different cities who would volunteer for a
weekend out of the year to build these applications. Other participants,
such as Ingenix, are considering contributing data to help make some of
the proposed applications more robust.
Concluding Thoughts
The spirit of commonwealth and of “all hands on deck” emerging from
efforts to foster open government is essential to a data ecosystem and to a
learning health system. Initial results of HHS’s Open Government Plan also
illustrate the important catalytic role government can play in health system
and health improvement. Instead of developing a centralized system to con-
trol data access and use, HHS can supply data to all and help consolidate
and catalog ecosystem information so that data use can be driven by and
serve the needs of the community.
OCR for page 87
87
CLINICAL DATA AS A PUBLIC GOOD FOR DISCOVERY
Opening access to health-related data from the public and private sec-
tors could be transformative for improving the health of Americans and
the effectiveness of the healthcare system. However, people of all talents,
capabilities, and resources must come together to realize this potential.
HHS needs guidance from the public about what data it should capture
and release and how the data can be used to advance social missions. As
good as the HHS data are now, they will become much better over time
with public input. The dialogue should be about how HHS and the public,
in the spirit of open government, can collaborate on an ongoing basis to
achieve the promise of a learning health system.
ENSURING DATA INTEGRITY: ADDRESSING PRIVACY
PROTECTION AND PROPRIETARY CONCERNS
Don E. Detmer, M.D., M.A.
University of Virginia
In the American context, a worthy society is exemplified by a healthy
citizenry pursuing happiness through the beneficence of sound governance
and positive expressions of personal freedom. Key to those positive expres-
sions of personal freedoms is the exploration of “the illimitable expressions
of the human mind.”4 The nation’s origins were suffused with the optimism
and courage characteristic of the Enlightenment. Discovery and support of
progress across all dimensions of human activity were central themes, as
characterized by efforts as diverse as the pursuit of copyright law and the
rapid spread of smallpox vaccination (Sheldon, 2010). As Ralph Waldo
Emerson noted, however, “The first wealth is health.”5
These early emphases on discovery and improvements in health and
health care were underscored 150 years later when, in the midst of a great
world war, Franklin Delano Roosevelt dedicated the NIH for the discovery
of cures for the maladies plaguing humanity. Over the years, this national
investment has benefited the care of Americans and people everywhere,
while also leading to major improvements in the health of domesticated
and even wild animals.
The Blue Ridge Academic Health Group has advocated for some years
for a value-driven healthcare system (Blue Ridge Academic Health Group,
1998, 2004) that continually improves the quality of care and practices
evidence-based management of illnesses. To secure such a system requires
4 See http://wiki.monticello.org/mediawiki/index.php/Illimitable_freedom_of_the_human_
mind_(Quotation) (accessed October 11, 2010).
5 See http://www.notable-quotes.com/e/emerson_ralph_waldo.html (accessed October 11,
2010).
OCR for page 88
88 PATIENTS CHARTING THE COURSE
that the entire enterprise focus on learning and become a learning health
system (IOM, 2007). Requirements for a learning health system include
(1) widespread access to patient information to better understand the com-
parative effectiveness of treatments and foster ongoing research into less
understood conditions; (2) policies and procedures that do not waste scarce
resources, including money and the valuable time of health professionals;
(3) active engagement of patients in their care to improve their under-
standing of and commitment to treatments of known effectiveness; and
(4) ongoing, credible systems that preserve privacy and manage patient
information securely to achieve data integrity for both healthcare opera-
tions and research of all types (e.g., biomedical, public health, and health
services research).
Recently, Secretary of Health and Human Services Kathleen Sebelius
noted: “The key to benefiting from change will be in ensuring that health-
care decision makers are guided by science and research” (Sebelius, 2010).
With the passage of the ACA, access to basic healthcare services was af-
firmed as worthy of the investment in terms of national health status and
social productivity. The absence of prior legislation providing all citizens
with access to basic healthcare services gave rise to fears that unauthorized
distribution of personal health data would lead to healthcare uninsurability.
With the passage of ARRA, the Genetic Information Nondiscrimination Act
of 2008, and the ACA, we have now entered a new era where this concern
is moot.
Secretary Sebelius wants decisions to be based on evidence and re-
search. To this end, public policy must support a learning health system.
Further, citizens should be actively engaged in the support of legitimate
healthcare research and institutional review boards (IRBs). The nation
now has stringent policies and regulations, including both civil and crimi-
nal policies, relating to the use and protection of person-specific health
information. Indeed, in 2009, ARRA added so many new protections that
many experts consider this law to have created a second incarnation of the
privacy provisions of the Health Insurance Portability and Accountability
Act (HIPAA) of 1997. Even before ARRA was passed, multiple expert
groups involved in the nation’s research enterprise had examined the im-
pact of HIPAA on a range of biomedical, public health, and health services
research. An IOM report reviewed all such studies and offered the conclu-
sion that improvements to HIPAA were needed to facilitate better privacy
protections (some of which are included in ARRA), as well as better access
to data for research (IOM, 2009).
Another component of ARRA is a mandate that the government assess
the comparative effectiveness of treatments to develop greater understand-
ing of which treatments are most effective. This mandate specifically calls
for health services research that will require access to person-specific health-
OCR for page 89
89
CLINICAL DATA AS A PUBLIC GOOD FOR DISCOVERY
care data. The time has come to put before the American people the op-
portunity to support the discovery of new treatments and help ensure that
only those resources needed to support research and protect data integrity
and security are expended. At the same time, the research enterprise should
be constructed so as to offer the greatest levels of patient safety and highest
quality of care for standard practices.
To address these needs, Congress should introduce and pass a law that
will facilitate the achievement of a number of important objectives with
respect to biomedical, public health, and health services research. This
law should include a number of features (Box 3-3). First, it should enable
Americans to have a unique personal health identifier, which will facilitate
the integrity of research data. The identifier should also be allowed to be
used for authentication of routine care unless the individual wishes to
formally opt out of such usage. Second, the law should mandate that all
research studies meeting HIPAA requirements and having been approved
by an IRB have access to nonanonymized health record data for legitimate
research purposes. Further, citizens should explicitly be given the opportu-
nity to opt out of the accessibility of any personal genetic information for
research that meets HIPAA requirements and has been approved by an IRB
to use nonanonymized health record data for legitimate research purposes.
Any breaches of data security relating to research use would fall within
BOX 3-3
Components for a “Health Research and Safe Care Act of 2011”
Citizens have the opportunity to
• pt out of selection of a unique personal health identifier for use in research
o
databases (including an option for the identifier to be used for routine personal
health care);
• pt out of consent to share their personal health data for IRB-approved re-
o
search that complies fully with HIPAA security regulations and requirements
without any data anonymization; and
• pt out of consent to share their personal genetic data, if available, for IRB-
o
approved research that complies fully with HIPAA security regulations and
requirements without any data anonymization.
In addition:
• se of anonymized data would continue to be allowed without explicit personal
u
consent; and
• ublic–private partnerships would be able to allow researchers to identify
p
potential research participants, who might then be approached to take part in
clinical studies for which consent is needed.
OCR for page 90
90 PATIENTS CHARTING THE COURSE
the current boundaries of HIPAA and ARRA as defined above. Finally, a
public–private collaboration should be established to develop a process by
which citizens interested in becoming involved in a research protocol that
may apply to them as a result of preliminary data screening could be located
and offered information on potential inclusion in the study.
Some further explanations are appropriate to better understand the
logic behind the above policy proposal. The above proposal flows from
the 2008 policy conference of the American Medical Informatics Associa-
tion (AMIA). AMIA held a conference on “Informatics, Evidence-based
Care, and Research: Implications for National Policy” (Bloomrosen and
Detmer, 2010). A diverse group including federal regulators, citizens,
researchers, and informaticians, concluded that national policy must be
revised to secure a learning health system, improve and ensure data integ-
rity, attend to data security and privacy, and facilitate research. The group
concluded that only through a refocusing of public policy with respect
to data access could these objectives be met. It concluded that a learning
health system requires broad access to personal health data. Further, citizen
support for legitimate biomedical and health research should be facilitated
because today the entire burden of engaging citizens in research falls not
on society but on researchers and IRBs, thereby creating significant disin-
centives for caregivers and researchers to pursue many promising lines of
inquiry. In short, at this time, federal policy adversely impacts legitimate
biomedical and health research by restricting data access, and there is no
evidence that this is the will of most Americans.
The perspective of clinicians is important to data integrity as well.
With the drive to make EHRs the norm for clinical practice settings, it is
important to motivate the clinical community to be highly committed to
data integrity. Payment for “meaningful use” holds promise to at least en-
courage recording data to enhance payment rather than spending the time
to capture the exact clinical situation. There is a real possibility that data
quality will benefit from clinicians learning that their data will be reviewed
for both research purposes and for assessing their clinical practice behavior
for quality assurance and maintenance of licensure. EHRs have the poten-
tial to reduce the quality of clinical notes as well as improve care through
better data access, and it is wise not to overlook the role that support for
professionalism can play in a contractual regulatory world.
At the broader policy level, instead of supporting research, current
policy focuses overwhelmingly on privacy protection of personal health in-
formation, and such federally enforced protections continue to mount with
neither the ability nor support by citizens to alter this policy course. Further,
with Americans strongly in favor of legitimate biomedical, public health,
and health services research, no policy exists to encourage the system to
OCR for page 91
91
CLINICAL DATA AS A PUBLIC GOOD FOR DISCOVERY
favor access to patient data for such research (Allcott and Mullainathan,
2010; Thaler and Sunstein, 2008).
Equally disturbing in the context of a learning healthcare system, cur-
rent policies and regulations limit data integrity through a variety of mecha-
nisms. The most prominent of these is the absence of unique personal health
identifiers, which is a threat to privacy and safety in day-to-day care and
a major threat to data integrity because data (especially those relating to
people with common names) can be misfiled quite easily. Another issue
with major implications in terms of cost and wasted time relates to the gap
between the use of health data for quality management and for clinical and
health services research. There is a substantial risk that ARRA will further
limit data access through the “minimum necessary” requirement reviewed
in summer 2010.
Other nations are ahead of the United States in addressing this problem.
For example, the Research Capability Program of the Department of Health
of England and Wales,6 passed in 2007, is explicitly establishing mecha-
nisms to allow researchers full access to 21 formal data sets created by the
Department. The goal is to facilitate access to such data by researchers,
including the pharmaceutical industry. England is convinced that this can be
good law while remaining consistent with patient privacy protections. The
National Health Service (NHS) data sets include NHS Care Records Service
(CRS) Personal Demographics Service, Death Registration, Birth Notifica-
tion Dataset, Commissioning Data Set/Hospital Episode Statistics, Cancer
Register, NHS CRS Electronic Prescribing Service and National Prescribing
Service, General Practice Research Database, Health Improvement Net-
work, EMIS Primary Care (QResearch), IMS-Mediplus, General Practice
Extraction Service, Cardiac Register; National Diabetes Audit, National
Congenital Anomaly System, UK Renal Registry, Townsend Material De-
privation Score, Jarman Underprivileged Area Score, Office for National
Statistics Demographic indexes, National Statistics Postcode Directory,
NHS HealthSpace, and Mental Health Minimum Data Set.
Specific recommendations of the United Kingdom Research Capability
Program include “safe havens” for population-based research in which the
protection of confidentiality is paramount; systems for approving and ac-
crediting researchers allowed to work in such environments; involvement
of academic and other partners in safe havens; and development of systems
to allow researchers to identify potential participants, who may then be ap-
proached to take part in clinical studies for which consent is needed.
The recommendation for an opt-out option relates to recent multiyear
experiences in Massachusetts, Minnesota, and Utah (Okwumabua, 2010).
6 See www.connectingforhealth.nhs.uk/systemsandservices/research (accessed October 11,
2010).
OCR for page 92
92 PATIENTS CHARTING THE COURSE
Recently, Massachusetts has used an opt-in option for gaining informed
consent to share personal data among regional caregivers using a unique
health identifier. Upward of 90 percent of patients opt in to data sharing,
but the monetary costs to maintain this approach are significant, and sub-
stantial time is required by clinicians as well. Meanwhile, in Minnesota, a
data sharing system has been in play for some years. It too uses an opt-in
option, and experience with this system mirrors that in Massachusetts (e.g.,
monetary costs and clinician and other administrative costs are substan-
tial). Finally, Utah has had an opt-out program for children’s health data
whose original focus was on minimizing under- and overimmunization.
Roughly 3 percent of parents opt their children out, and the system is much
less expensive than those of Massachusetts and Minnesota. Public accep-
tance is thus shown by these real-world experiences to approach or exceed
90 percent for sharing of data for care purposes.
With respect to such options and biomedical research, Daniel Masys
of Vanderbilt University offered to the author the following personal com-
munication relating to deidentified data:
Vanderbilt has an opt-out model for building a biobank of DNA samples
from blood left over from clinical testing linked to de-identified electronic
medical record (EMR) data that currently has more than 63,000 samples
in it, growing at about 800 per week. The predicted opt-out rate before
the project started, based on focus groups and a 5000 person patient sur-
vey was five percent. The observed opt-out rate over the past two years,
across about one million consent-for-treatment signing events, is about
3.8 percent. Surveys of those who did opt out suggest an “anti-science”
phenotype that is suspicious of all organizations, a belief that commercial
profit motives are behind any veneer of science altruism. Most are person-
ally convinced to a degree that no additional information would change
their minds. These data seem to be a useful validation of the estimated
size of an often vocal minority who figure so prominently in all of the
public debates about research, uses of clinical data and electronic data
systems. . . . This is not to say that such views are not valid and they
clearly must be accommodated in discussions of architecture and policy,
but they often seem to occupy a disproportionate amount of public dialog
and pedagogy. (Masys, 2009)
If the recommended opt-out option were signed into law, one might
predict that, based on available data and with sensible administration,
well over 75 percent of the American public would “sign up.” This would
benefit research immensely and save billions of dollars and thousands of
hours as well.
In conclusion, two quotations from James Madison relate directly
to this public policy issue and the challenge facing those seeking to help
OCR for page 93
93
CLINICAL DATA AS A PUBLIC GOOD FOR DISCOVERY
Americans find more cures sooner through the passage of legislation such as
that suggested here. Madison noted: “A people who mean to be their own
Government, must arm themselves with the power to which knowledge
gives.”7 Progress comes only when information flows. Madison observed
in a letter to Benjamin Rush that he was no stranger to the clash of deeply
felt values and how the intensity of belief can warp perception. He stated:
“Nothing is so contagious as opinion, especially on questions which, being
susceptible of very different glosses, beget in the mind a distrust of itself.”8
One cannot but reflect on this quotation in the context of the continuing
health data privacy wars of the past two decades. Within the Beltway, it is
commonly taken as fact that patients are more concerned about the privacy
of their data than anything else and that trust in health care and EHRs is
based overwhelmingly on assurance of this privacy. What facts exist reveal
that this is so much finely glossed opinion. According to Mechanic and
Meyer (Mechanic and Meyer, 2000), trust in health care is based on three
main pillars and a fourth that sometimes is not even mentioned. First and
foremost is the ability of the clinician to communicate and connect with
the patient. Second is the perceived competence of the clinician. Third is
whether the clinician is willing to advocate for the patient’s benefit if nec-
essary. If it is mentioned at all, maintaining privacy and confidentiality of
data is fourth.
Americans should be given the option to become full and open par-
ticipants in a learning health system. To do otherwise is to prevent them
from being either Americans in the finest sense of that word or adults with
respect to their care and concern for their fellow beings. The right of Ameri-
cans to be left alone must not be allowed to become a right to be unknown.
REFERENCES
Allcott, H., and S. Mullainathan. 2010. Behavior and energy policy. Science 327:1204-1205.
Bloomrosen, M., and D. Detmer. 2010. Informatics, evidence-based care, and research; impli-
cations for national policy: A report of the American Medical Informatics Association’s
health policy conference. Journal of the American Medical Informatics Association
17:115-123.
Blue Ridge Academic Health Group. 1998. Academic health centers: Good health is good
business. http://whsc.emory.edu/blueridge/_pdf/blue_ridge_report2.pdf (accessed Octo-
ber 11, 2010).
———. 2004. Converging on consensus: Planning the future of health and health care. http://
whsc.emory.edu/blueridge/_pdf/blue_ridge_report_8_2004oct.pdf (accessed October 11,
2010).
7 See http://presspubs.uchicago.edu/founders/documents/v1ch18s35.html (accessed Octo -
ber 11, 2010).
8 See http://www.revolutionary-war-and-beyond.com/james-madison-quotes-6.html (accessed
October 11, 2010).
OCR for page 94
94 PATIENTS CHARTING THE COURSE
Blumenthal, D. 2010. Launching HITECH. New England Journal of Medicine 262(5):382-385.
HHS (Department of Health and Human Services). 2010. HHS open government plan. http://
www.hhs.gov/open/plan/opengovernmentplan/index.html (accessed October 11, 2010).
IOM (Institute of Medicine). 2007. The learning healthcare system: Workshop summary.
Washington, DC: The National Academies Press.
———. 2009. Beyond the HIPAA Privacy Rule: Enhancing privacy, improving health through
research. Washington, DC: The National Academies Press.
Masys, D. 2009. Personal communication, August 14.
Mechanic, D., and S. Meyer. 2000. Concepts of trust among patients with serious illness.
Social Science & Medicine 51:657-658.
Okwumabua, E. 2010. Public Health, George Washington University.
Sebelius, K. 2010. Statement of Kathleen Sebelius Secretary U.S. Department of Health
and Human Services on the President’s fiscal year 2011 budget before the Subcommit-
tee on Labor, Health and Human Services, Education, and Related Agencies, Committee
on Appropriations, U.S. House of Representatives.
Sheldon, G. F. 2010. Hugh Williamson: Physician, poet and founding father. New York:
Humanity Books.
Thaler, R. H., and C. R. Sunstein. 2008. Nudge: Improving decisions about health, wealth
and happiness. New York: Penguin.
