INTRODUCTION AND OVERVIEW
Health and health care are going digital. As multiple intersecting platforms evolve to form a novel operational foundation for health and health care—the nation’s digital health utility—the stage is set for fundamental and unprecedented transformation. Most changes will occur virtually out of sight, and the pace and profile of the transformation will be determined by stewardship that fosters alignment of technology, science, and culture in support of a continuously learning health system. In the context of growing concerns about the quality and costs of care, the nation’s health and economic security are interdependently linked to the success of that stewardship.
Progress in computational science, information technology (IT), and biomedical and health research methods have made it possible to foresee the emergence of a learning health system that enables both the seamless and efficient delivery of best care practices and the real-time generation and application of new knowledge. Increases in the complexity and costs of care compel such a system. With rapid advances in approaches to diagnosis (such as molecular diagnostics), therapeutics, genetic insights into individual variation, and emerging measurement modalities (such as within proteomics and imaging), clinicians and patients must sort through exponentially increasing numbers of factors with each clinical decision. At the same time, healthcare costs are draining the purchasing power of consumers and handicapping the competitiveness of U.S. businesses, yet health outcomes are falling far short of the possible.
Against this backdrop of opportunity and urgency, the Institute of Medicine (IOM) of the National Academies, sponsored by the Office of the National Coordinator for Health Information Technology (ONC), convened a series of expert meetings to explore strategies for accelerating the development of the digital infrastructure for the learning health system. Presentations and major elements of those discussions are summarized in this publication, Digital Infrastructure for the Learning Health System: The Foundation for Continuous Improvement in Health and Health Care.
The Learning Health System
In 2001, the IOM report Crossing the Quality Chasm called national attention to untenable deficiencies in health care, noting that every patient should expect care that is safe, effective, patient-centered, timely, efficient, and equitable (IOM, 2001). Based on the determination that health care is a complex adaptive system—one in which progress on its central purpose is shaped by tenets that are few, simple, and basic—the report identified several rules to guide health care. In particular, these rules underscore the importance of issues related to the locus of decisions, patient perspectives, evidence, transparency, and waste reduction. The report envisioned, in effect, engaging patients, providers, and policy makers alike to ensure that every healthcare decision is guided by timely, accurate, and comprehensive health information provided in real time to ensure constantly improving delivery of the right care to the right person for the right price.
The release of the IOM Chasm report stimulated broad activities related to clinical quality improvement and the effectiveness of health care, including the eventual creation by the IOM of the Roundtable on Value & Science-Driven Health Care. Begun in 2006 as the IOM Roundtable on Evidence-Based Medicine, it has explored ways to improve the evidence base for medical decisions and sought the development of a learning health system “designed to generate and apply the best evidence for collaborative health choices of each patient and provider; to drive the process of discovery as a natural outgrowth of patient care; and to ensure innovation, quality, safety, and value in health care.” From its inception, the Roundtable has conducted The Learning Health System Series of public meetings to consider the capture of emerging innovations—such as those occurring in IT, research methods, and care delivery—as building blocks in the foundation of a learning health system. Characteristics of such a system are noted in Box S-1 and in matrix form in Appendix A. In broad terms, they represent delivery of best practice guidance at the point of choice, continuous learning and feedback in both health and health care, and seamless, ongoing communication among participants, all facilitated through the application of IT.
Learning Health System Characteristics
SOURCE: Adapted from The Learning Healthcare System (IOM, 2007).
Because IT serves as the functional engine for the continuous learning system, this ONC-commissioned exploration was broadly conceived to consider the issues and strategies required for the emergence of a digital infrastructure that allows data collected during activities in various settings—clinical, research, and public health—to be integrated, analyzed, and broadly applied (“collect once, use for multiple purposes”) to inform and improve clinical care decisions, promote patient education and self-management, design public health strategies, and support research and knowledge development efforts in a timely manner.
The Digital Health Infrastructure
The digital infrastructure for the learning health system will not solely be the result of features designed and built de novo. Existing initiatives and
resources are actively in play at multiple levels—including electronic health records (EHRs); personal health records (PHRs); telehealth; health information portals; electronic monitoring devices; biobanks; health information databases maintained by large health systems, private insurers, and regulatory agencies; and advances in molecular diagnostics. Each adds important capacity for clinical care, clinical and health services research, public health surveillance and intervention, patient education and self-management, and safety and cost monitoring.
Still, these capacities are relatively early in their development, and progress depends on improvements on several dimensions. As of 2009, only about 12% of hospitals and 6% of clinician offices had an EHR in place (DesRoches et al., 2008; Jha et al., 2010) and only about 1 in 14 Americans had electronic access to any patient-oriented version of their health record (CHCF, 2010). On the other hand, since 2000, the number of Americans who have access to the internet has jumped from 46% to 74%, and the number of American adults who have looked online for health information has jumped from 25% to 61% (Fox, 2010). Wireless technology is quickening the pace of change. With 6 in 10 American adults using wireless capability with a laptop or mobile device (Smith, 2010), mobile applications are rapidly developing the potential for remote site access to health information, as well as diagnostic and even treatment services.
This developing potential presents opportunities and challenges for stewardship. Issues related to interoperability, governance, patient and public engagement, and privacy and security concerns, among others, will need to be better addressed for successful progress toward a learning health system. Approaches and lessons from sectors outside health include those from energy and the financial sector, two examples discussed in the meetings and summarized in this publication (see Appendix B). VISA used a minimalist approach, crafted on the combination of mutual self-interest and basic rules-of-play, to build its platform for a global credit card network. Consumer Energy’s work in the Smart Grid Initiative applied an analytically driven approach to accommodate and network a wide variety of legacy nodes in growing the electronic platform operating the nation’s energy system. Background on the Smart Grid Initiative is presented in Box S-2.
Regardless of the model, a key rationale for the workshop discussions was the reality that effective and efficient progress in the growth and development of our national and global digital health infrastructure requires active cooperation, collaboration, and role delineation among many organizations, companies, and agencies—private and public—at the cutting edge of using health IT to improve health and health care.
The striking, and accelerating, progress in the capacity and transformative influences of IT on society over the past three decades is a blended product of interrelated initiatives arising from within the commercial, in-
Case: The Smart Grid
The Smart Grid is a long-term, complex systems development project to grow the electronic platform operating the nation’s energy system using an engineering approach to accommodate a wide variety of legacy nodes that are organic—constantly growing and evolving, much like a biological system. This continuous evolution allows the Smart Grid’s architecture to preserve and encourage the capacity of each node to innovate locally and deal with complexity in a way that suits local and grid needs. As conceived, the Smart Grid will
- Enable active participation by consumers
- Accommodate all generation and storage options
- Enable new products, services, and markets
- Provide power equality for the digital economy
- Optimize asset utilization and operate efficiently
- Anticipate and respond to system disturbances (self-heal)
- Operate resiliently against attack and natural disaster
Because there is no need for consensus among the nodes on how they should operate within local boundaries, the Smart Grid development methodology is not based on comprehensive internal design and operating standards for each node on the Grid to follow. Instead, the approach accommodates highly diverse nodes connecting to the Smart Grid using open data translation protocols that standardize information management, rather than using the internal workings of each node. The Grid becomes a communications bus to which each node must be able to write, and from which each node must be able to read. This architecture preserves capacities for local operating autonomy and innovation throughout the Smart Grid. It also manages a standardized communications capacity among complex, and otherwise noninteroperable, legacy nodes on the Grid. These features are all characteristics of ultra-large-scale (ULS) software-intensive systems.
dependent, and public sectors. Leaps in the speed, power, and efficiency of information processing, the development of the Internet and World Wide Web, and its use to facilitate near-universally available real-time access to information, have spawned a new economy and new vehicles for progress.
Health information vendors, large and small, have emerged to meet the growing demand for capacity to manage the retrieval, storage, and delivery of information for agencies, institutions, professionals, and individuals in virtually every aspect of health and health care. The range of newly digitalized services—and the growth of vendors to provide them—is startling. Through technologies developed by companies such as Google, Microsoft, and Yahoo, the amount of web-based health information accessed daily
by individuals and clinicians is already transforming the care process. Beyond the publicly available digital resources, a vast array of specialized care management products have emerged for activities such as scheduling and billing; claims processing and payment; supply and equipment inventory maintenance; individual patient charting; medication prescribing and tracking; family and personal health records; clinician-patient communication; clinician and patient decision support; robotics-assisted procedures; telehealth for remote site diagnosis and treatment; disease surveillance; vital statistics reporting; postmarket product monitoring; safety and hazard exposure monitoring; clinical research protocols; disease and intervention registries; and data aggregation, analysis, and modeling.
Various large academic health centers and healthcare delivery organizations—Veterans Health Administration (VHA), Kaiser Permanente (see Box S-3), Geisinger Health System, Vanderbilt, MD Anderson, Palo Alto Medical Foundation, Group Health Cooperative, several Harvard facilities, Children’s Hospital of Philadelphia, Virginia Mason, and the Mayo Clinic, to name a few—have invested substantially in the creation of advanced digital resources for administrative, patient care, and research functions. Additionally, some related collaborative research networks have begun to develop. Nonetheless, the diversity and limited compatibility of the products, and the lack of economic incentives for their use have, to date, restrained the broader uptake, application, and functional utility of digital capacity across the system.
A number of public, private, and independent sector initiatives have emerged to accelerate stakeholder action on various dimensions important to progress. To supplement the relatively limited pre-2009 public investments, independent sector leadership has come from foundations such as the Markle Foundation, the Robert Wood Johnson Foundation, and the California HealthCare Foundation. Furthermore, in addition to the formation of capacity-building resources such as the Health Information Exchanges, a number of facilitative stakeholder groups have emerged—for example, the eHealth Initiative, the Clinical Data Interchange Standards Consortium (CDISC), and the National eHealth Collaborative. On the professional advancement dimension, the American Medical Informatics Association has developed as a growing resource for the contributions of biomedical and health informaticians working in activities to organize, manage, analyze, and use information in health care. An example of the coordinative potential of these groups is found in the development of integration profiles by Integrating the Healthcare Enterprise and CDISC to support the use of EHRs for clinical research, quality, and public health, and the testing and demonstration of these profiles by several vendors including Cerner, Allscripts, Greenway Medical, and General Electric Healthcare.
At the federal level, ONC was created in 2004 in the U.S. Department
Case: Kaiser Permanente
In 2003, Kaiser Permanente (KP) launched a $4 billion health information system called KP HealthConnect that links its facilities and clinicians throughout their delivery system and represents the largest civilian installation of electronic health records in the United States. The EHR at the heart of KP HealthConnect provides a reliably accessible longitudinal record of member encounters across clinical settings including laboratory, medication, and imaging data; as well as supporting:
- Electronic prescribing and test ordering (computerized physician-order entry) with standard order sets to promote evidence-based care
- Population and patient-panel management tools such as disease registries to track patients with chronic conditions
- Decision support tools such as medication-safety alerts, preventive-care reminders, and online clinical guidelines
- Electronic referrals that directly schedule patient appointments with specialty care physicians
- Personal health records providing patients with the ability to view their personal clinical information including lab results, plus linkage with pharmacy, physician scheduling, and secure and confidential e-mail messaging with clinicians.
- Performance monitoring and reporting capabilities
- Patient registration and billing functions
Physician leaders report that access to the EHR in the exam room is helping to promote compliance with evidence-based guidelines and treatment protocols, eliminate duplicate tests, and enable physicians to handle multiple complaints more efficiently within one visit. Ongoing evaluation by Kaiser indicates that patient satisfaction with outpatient physician encounters has increased and that the combination of computerized physician-order entry, medication bar coding, and electronic documentation tools is helping to reduce medication administration errors in hospital care.
Overall, Kaiser’s experience suggests that use of the EHR and online portal to support care management and new modes of patient encounters is having positive effects on utilization of services and patient engagement. For example, three-quarters or more of online users surveyed agreed that the portal enables them to manage their health care effectively and that it makes interacting with the healthcare team more convenient.
of Health and Human Services (HHS) to stimulate progress in the field. Since 2009, with the enactment of the Health Information Technology for Economic and Clinical Health Act (HITECH) as part of the American Recovery and Reinvestment Act, the federal government leadership profile has become especially prominent. This has included the commitment of
unprecedented resources for health information technology (HIT), administered through the leadership of ONC. Under HITECH, ONC was granted $2 billion to facilitate the adoption and meaningful use of HIT. In addition, an estimated $27 billion was designated for the Centers for Medicare & Medicaid Services (CMS) to distribute as incentive payments for physicians and hospitals to become meaningful users of HIT.
Designed as a set of staged requirements to qualify for CMS incentive payments, the first-stage elements of “meaningful use” were released by CMS on July 13, 2010. These established a core set of requirements for eligible professionals and hospitals, as well as a menu of additional choices, from which five are to be chosen. The stage 1 meaningful use target elements are listed in summary fashion in Box S-4, and details are contained in Appendix D. The subsequent stages of meaningful use are currently under development and are presented later in this summary, along with an indication of related issues flagged in workshop discussions.
In addition to the meaningful use requirements, ONC has funded a series of grant programs through HITECH, including the Beacon Community grants, aimed at demonstrating community-wide digital infrastructure capacity and use for health improvement, and the Strategic Health Information Technology Advanced Research Projects Program, to foster the capture of technological advances to improve system performance. At the broader level, ONC is pursuing a series of initiatives to foster health information exchange among stakeholders, including the regional health information exchanges and under the Nationwide Health Information Network (NWHIN).
Several additional HHS agencies have activities important to the development of the digital learning health system. CMS, in addition to establishing rules for meaningful use and requirements for uniform condition identifiers central to healthcare payment and research, recently created the Center for Medicare and Medicaid Innovation to test innovative payment and program service delivery methods. Within the National Institutes of Health (NIH), the National Library of Medicine serves as the central coordinating body for clinical terminology standards, and other NIH programs, such as the Clinical and Translational Science Awards Program, and the National Cancer Institute’s Enterprise Vocabulary Series and cancer Biomedical Informatics Grid (caBIG®, see Box S-5 and Appendix B for additional information) serve as key contributors to building the capacity to derive scientific discovery from patient care. Through its National Resource Center for Health IT and capacity initiatives on patient registries, the Agency for Healthcare Research and Quality (AHRQ) supports a number of programs to advance the digital utility for healthcare quality and safety.
At the Food and Drug Administration (FDA), the Sentinel Initiative (see Box S-6 and Appendix B) has been designed to build and implement a national electronic system for postmarket surveillance of approved drugs
Meaningful Use Requirement Categories
Core structured personal data (age, sex, ethnicity, smoking status)
Core list of active problems and diagnoses
Core structured clinical data (vital signs, meds, [labs])
Outpatient medications electronically prescribed
Automated medication safeguard/reconciliation
Clinical decision support
Care coordination support/interoperability
Visit-specific information to patients
Automated patient reminders
e-Record patient access (copy or patient portal)
Embedded measures for clinical quality reporting
Examples of optional elements:
Advance directives for ages >65
Condition-specific data retrieval capacity
Public health reporting (reportable conditions)
SOURCE: Adapted from Blumenthal and Tavenner (2010). See Appendix D for details.
and other medical products. The Centers for Disease Control and Prevention (CDC) supports several IT-based public health data collection and surveillance programs and serves as the primary agency responsible for these tracking efforts, response and public health links to domestic and international public health data systems, and the Health Resources and Services Administration (HRSA) has developed initiatives introducing HIT to improve care access and coordination in rural areas and for underserved populations.
Efforts to promote the development, implementation, and widespread adoption of HIT also build on a wide array of digital learning leadership efforts by other federal agencies. In particular, important contributions stem
Case: The National Cancer Institute’s caBIG® Initiative
The National Cancer Institute of the National Institutes of Health has developed an informatics program designed to improve patient care and accelerate scientific discoveries by enabling the collection and analysis of large amounts of biological and clinical information and facilitating connectivity and collaboration among biomedical researchers and organizations. More than 700 different organizations are actively engaged in caBIG®, including basic and clinical researchers, consumers, physicians, advocates, software architects and developers, bioinformatics specialists and executives from academe, medical centers, government, and commercial software, pharmaceutical, and biotechnology companies from the United States and in 15+ countries around the globe.
At the heart of the caBIG® program is caGrid, a model-driven, service-oriented architecture that provides standards-based core “services,” tools, and interfaces so the community can connect to share data and analyses efficiently and securely. More than 120 organizations are connected to caGrid. In partnership with the American Society of Clincal Oncology, caBIG® is developing specifications and services to support oncology-extended EHRs that are being deployed in community practice and hospital settings. caBIG® tools and technology are also being used by researchers working on cardiovascular health, arthritis, and AIDS. In addition, pilot projects have successfully connected caGrid to other networks, including the Nationwide Health Information Network, the CardioVascular Research Grid, and the computational network TeraGrid.
from responsibilities and activities of the VHA—for example, the highly regarded Veterans Health Information Systems and Technology Architecture system of IT supporting better care, as well as personal tools such as “My HealtheVet” and the Virtual Lifetime Electronic Record programs—the Department of Defense (DOD), the Federal Communications Commission (FCC), and the National Science Foundation (NSF). The VHA and the DOD have formed the Telemedicine and Advanced Technology Research Center as a joint program to advance research and applications in health informatics, telemedicine, and mobile health monitoring systems. Because of the deep and broad set of capabilities and initiatives collectively sponsored by federal agencies, their coordination and interface with private sector activities offers a vital strategic opportunity to accelerate the development of a learning health system.
Testament to the compelling priority of the prospects, in December 2010, the President’s Council of Advisors on Science and Technology (PCAST) issued its report, Realizing the Full Potential of Health Information Technology to Improve Healthcare for Americans: The Path Forward (PCAST, 2010). The PCAST report examines the opportunities and needs
Case: The FDA’s Sentinel Initiative
In 2008, the Department of Health and Human Services and the Food and Drug Administration (FDA) announced the launch of FDA’s Sentinel Initiative, a long-term program designed to build and implement a national electronic system—the Sentinel System—for monitoring the safety of FDA-approved drugs and other medical products. Data partners in the Sentinel System will include organizations such as academic medical centers, healthcare systems, and health insurance companies. As currently envisioned, participating data partners will access, maintain, and protect their respective data, functioning as part of a “distributed system.”
In a related pilot activity, FDA is working with Harvard Pilgrim Health Care, Inc. to develop a smaller working version of the future Sentinel System, dubbed “Mini-Sentinel.” Through this pilot, FDA will learn more about some of the barriers and challenges, both internal and external, to establishing a Sentinel System for medical product safety monitoring. The Mini-Sentinel Coordinating Center (MSCC) represents a consortium of more than 20 collaborating institutions, working with participating data partners to use a common data model as the basis for their approach. Data partners transform their data into a standardized format, based upon which the MSCC will write a single analytical software program for a given safety question and provide it to each of the data partners. Each partner will conduct analyses behind its existing, secure firewall and send only summary results to the MSCC for aggregation and further evaluation.
As this pilot is being implemented, a governance structure is being developed to ensure the activity encourages broad collaboration within appropriate guidelines for the conduct of public health surveillance activities. In order to accomplish that, the MSCC is developing a Statement of Principles and Policies that will include descriptions of the organizational structure and policies related to communication, privacy, confidentiality, data usage, conflicts of interest, and intellectual property.
for the use of HIT to improve healthcare quality and reduce cost, as well as the activities and aligment of current federal programs with relevant responsibilities. It sets out a series of recommendations intended to facilitate private, entrepreneurial initiatives through governmental action to speed development of a “universal exchange language” for health information, the application of which would maximize the ability to use existing and developing electronic record systems. Specifically, it recommends action by the federal government, especially ONC and CMS, in accelerating the identification of standards required for health information exchange using metadata-tagged data elements; mapping various existing semantic taxonomies onto the tagged elements; developing incentives for product use of tagged elements; fostering use of metadata for security and safety protocols;
bringing federal program capacity and policy leverage to bear in implementing and guiding the efforts; and developing metrics to assess progress. The PCAST recommendations are included as Appendix E.
About the Digital Infrastructure Meetings
It was in this general context of opportunity and challenge that the IOM workshops on the digital health infrastructure were organized. Since the inaugural workshop in 2006, the IOM has conducted 15 workshops in the Learning Health System Series, with 10 reports published and in production:
- The Learning Healthcare System
- Leadership Commitments to Improve Value in Health Care: Finding Common Ground
- Evidence-Based Medicine and the Changing Nature of Health Care
- Redesigning the Clinical Effectiveness Research Paradigm: Innovation and Practice-Based Approaches
- Clinical Data as the Basic Staple of Healthcare Learning: Creating and Protecting a Public Good
- Engineering a Learning Healthcare System: A Look at the Future
- Learning What Works: Infrastructure Required for Comparative Effectiveness Research
- Value in Health Care: Accounting for Cost, Quality, Safety, Outcomes, and Innovation
- The Healthcare Imperative: Lowering Costs and Improving Outcomes
- Patients Charting the Course: Citizen Engagement and the Learning Health System
This publication considers what has been variously described as the system’s nerve center, its circulation system, and the engine to drive the progress envisioned in the Learning Health System Series: the digital infrastructure. To explore the range of issues necessary to engage if that infrastructure is to develop as effectively and efficiently as possible, ONC requested that the IOM, through the Roundtable on Value & Science-Driven Health Care, organize the series of expert meetings summarized in this publication, Digital Infrastructure for the Learning Health System: The Foundation for Continuous Improvement in Health and Health Care.
As the title indicates, the primary intent of the meetings was to identify and explore strategic opportunities for accelerating the evolution of a digital infrastructure that will support and drive continuous improvement in health and health care. Three meetings were held in the summer and fall of 2010, bringing together researchers, computer scientists, privacy experts, clinicians, health care administrators, HIT professionals, representatives of
patient advocacy groups, healthcare policy makers, and other stakeholders. Building on the existing foundations of HIT, the main objectives were to foster a shared understanding of the vision for the digital infrastructure, explore the current state of the system, identify key priorities for future work, and consider strategy elements and priorities for accelerating progress on improving the infrastructure to build a more seamless learning enterprise that will improve health and health care in America.
Aims and Planning
A planning committee,1 composed of leading authorities on various aspects of the digital health learning process, shaped the workshop series around the following aims:
- Foster a shared understanding of the vision for the digital infrastructure for continuous learning and quality-driven health and healthcare programs.
- Explore current capacity, approaches, incentives, and policies; and identify key technological, organizational, policy, and implementation priorities for the development of the digital infrastructure.
- Discuss the characteristics of potentially disruptive breakthrough developments.
- Consider strategy elements and priorities for accelerating progress on the approach to the infrastructure and for moving to a more seamless learning enterprise.
Contextual considerations informing the Committee’s development of the agenda included
- Rapid developments in IT exponentially facilitating the potential of health data for knowledge generation and care improvement.
- Policy initiatives leading eventually to the digital capture and storage of virtually all clinical and related health data for use in performance improvement.
- Promising potential in federated and distributed research approaches allows data to remain local while enabling querying and virtual pooling across systems.
1 Institute of Medicine planning committees are solely responsible for organizing the workshop, identifying topics, and choosing speakers. The responsibility for the published workshop summary rests with the workshop rapporteurs and the institution.
- Ongoing innovation in search technologies with the potential to accelerate use of available data from multiple sources for new insights.
- Meaningful use criteria and health reform provisions that provide starting points, incentives, and guidance, while retaining the flexibility necessary to accommodate breakthrough capacities.
- Appreciation of the need to limit the burden for health data collection to the issues most important to patient care and knowledge generation.
- Requirement for governance policies that foster strengthening the data utility as a core resource to advance the common good; in particular by cultivating the trust fabric among stakeholders and accelerating collaborative progress.
- Developing standards that will facilitate distributed access to large datasets for comparative effectiveness research, biomarker validation, disease modeling, and research process improvement.
Structure and Thematic Arc
The three workshops in the series progressed from a broad exploration of the state of play and various stakeholder perspectives on a learning health system, to a more specific identification of strategic approaches to the challenges, and concluded with detailed discussions of strategic elements, stakeholder responsibilities, and key cross-cutting challenges. To maximize identification and sharing of perspectives, expert presentations were followed by open discussion among participants and separate small group discussion sessions were built into each of the meetings.
The first workshop, “Opportunities, Challenges, Priorities,” considered the overall vision of the digital infrastructure for the learning health system as well as some of the prominent issues and opportunities related to technical progress, ensuring commitment to population and patient needs, development of the necessary trust fabric, stewardship and governance, and the implications of the global character of the health data trust. The second meeting, “The System After Next,” went deeper into three crosscutting areas identified during the first workshop: engaging the patient and population, promoting technical advances, and fostering stewardship and governance structures. The third and final meeting of the series, “Strategy Scenarios,” reviewed the common themes and information from the previous workshops and extended into deeper consideration of strategy elements, opportunities, responsibilities, and next steps for progress on four key focus areas: technical progress, knowledge generation and use, patient and population engagement, and governance.
COMMON THEMES AND PRINCIPLES
Several common themes recurred throughout the rich and varied discussions. These themes, included in Box S-7 and summarized below, were reflected in discussions of each of the four focus areas (technical progress, knowledge generation and use, patient and population engagement, and governance), as well as the discussions around various strategic elements. They ranged from issues related to the culture and environment for learning, to the centrality of the patient and the importance of flexibility and trust.
- Build a shared learning environment. HIT provides an opportunity to change the current environment in which health decisions are made to one of shared input and active participation from patients, caregivers, and the population at large. Discussed approaches to developing this shared learning environment include the direct involvement and support of patient and population roles in the generation of knowledge through the incorporation of user-generated data; understanding the benefits of information use in patient care and population health improvement; and improving patient access to health information to allow for a more active role in care decisions.
- Engage health and health care, population and patient. Many participants reiterated that in order to improve health outcomes for the nation, thinking must extend beyond clinical encounters, and even beyond the individual patient, to the population as a whole. This shift of scope brought into clearer focus several issues discussed, including the opportunity to use HIT and its associated informa-
BOX S-7 Common Themes and Principles
- Build a shared learning environment
- Engage health and health care, population and patient
- Leverage existing programs and policies
- Embed services and research in a continuous learning loop
- Anchor in an ultra-large-scale systems approach
- Emphasize decentralization and specifications parsimony
- Keep use barriers low and complexity incremental
- Foster a sociotechnical perspective, focused on the population
- Weave a strong and secure trust fabric among stakeholders
- Provide continuous evaluation and improvement
tion to build a concept of health that is about more than medical care and draws on seamless interface with information from nonmedical health-related sources to generate knowledge that allows for a more inclusive view of population health improvement.
- Leverage existing programs and policies. A foundational assumption during the discussions was the advantage provided by building on and accelerating the substantial recent progress, both nationally and internationally, with an emphasis on the importance of fostering coordination among these efforts to capture efficiencies and prevent unnecessary duplication and waste going forward. Participants often noted that recent policies and legislation have laid a foundation for this work, and that the resulting investments and progress can be leveraged to move toward long-term system goals.
- Embed services and research in a continuous learning loop. Meeting participants often underscored that a digital infrastructure that supports both the generation and use of knowledge cannot be effective unless it is integrated seamlessly within the processes from which it draws and is meant to support care delivery, research, quality improvement, and population health monitoring. Ease of use for health system stakeholders, attention to the effects on workflow, and the delivery of useful decision support at point of care were often mentioned in discussions.
- Anchor in an ultra-large-scale systems approach. One of the most prominent features of the discussions was the notion that the health system is a complex, sociotechnical ecosystem, needing a unique conceptual approach. Grounding this approach to coordination and integration of the digital infrastructure for the learning health system in the principles of an ultra-large-scale (ULS) systems approach was suggested by several workshop participants from the computer science community (see Box S-8). The term “ultra-large-scale system” refers to the existence of a virtual system that has bearing on a social purpose—for example, improving health and health care—and in which a few key elements, such as interchange representation, may be standardized, but whose many participants have diverse and even conflicting goals, so adaptability is key. Institutions retain flexibility for innovation in their choices, and evolutionary functional change can be shaped by architectural precepts, incentives, and compliance assessment, but not by centralized control. ULS functionality is therefore facilitated by protocols that allow maximum practical flexibility for participants. Incorporating decentralization of data, development, and operational authority and control, this approach fosters local innovation, personaliza-
Ultra-Larg e-Scale (ULS) System Characteristics
The ULS approach can be best described by a set of characteristics that tend to arise as a result of the scale of the system (in this case health and health care) rather than a prescriptive set of required components. Previous work on the ULS concept has identified the following key characteristics of ULS systems:
Decentralization: The scale of ULS systems means that they will necessarily be decentralized in a variety of ways—decentralized data, development, evolution, and operational control.
Inherently conflicting, unknowable, and diverse requirements: ULS systems will be developed and used by a wide variety of stakeholders with unavoidably different, conflicting, complex, and changing needs.
Continuous evolution and deployment: There will be an increasing need to integrate new capabilities into a ULS system while it is operating. New and different capabilities will be deployed, and unused capabilities will be dropped; the system will be evolving not in phases, but continuously.
Heterogeneous, inconsistent, and changing elements: A ULS system will not be constructed from uniform parts: there will be some misfits, especially as the system is extended and repaired.
Erosion of the people/system boundary: People will not just be users of a ULS system; they will be elements of the system, affecting its overall emergent behavior.
Normal failures: Software and hardware failures will be the norm rather than the exception.
New paradigms for acquisition and policy: The acquisition of a ULS system will be simultaneous with the operation of the system and require new methods for control.
SOURCE: Northrop et al. (2006).
tion, and emergent behaviors. Participants felt that this approach was well suited to the complex adaptive characteristics of the health system, and that it could serve as an anchoring framework for approaching both the social and technical components of the overall infrastructure.
- Emphasize decentralization and specifications parsimony. In line with the complex adaptive qualities of the health system outlined
in the Quality Chasm (IOM, 2001) report and reiterated during the workshops, both the social and technical components of the digital health infrastructure require a framework that allows for tailoring to specific needs, local innovation, and evolvability. In this respect, the commonly repeated refrain was a call for the principle of parsimony and minimizing centralization that might constitute a barrier to entry: specify only the minimal set of standards or requirements necessary for key functional utility and push the maximum amount of control to the periphery. This approach is in line with strategies such as those suggested in the PCAST report for use of metadata for wrapping individual information packets to facilitate interoperability and health information exchange, in which a primary focus would be on development of the metadata standards.
- Keep use barriers low and complexity incremental. Similarly, incentives for broad participation in the digital infrastructure by all stakeholders was discussed as a crucial factor to its success. The proposal to keep the barriers for use of the infrastructure, such as deployment and operational complexity, low was articulated by workshop participants in order to allow for maximum participation at a baseline level, and allow for incremental complexity and sophistication where possible or necessary.
- Foster a sociotechnical perspective, focused on the population. From the outset of the discussions, participants pointed out that the major barriers to technical progress often lie in social and cultural domains. Acknowledging and engaging this fact were described as being crucial to success, with discussions centering on an approach that reorients future efforts to engage the patient more directly in the collection and use of information in a way that is most useful to them.
- Weave a strong trust fabric among stakeholders. Security and privacy concerns represent a strong threat to participation in, and therefore the success of, the sociotechnical ecosystem. Accordingly, they must be dealt with from both the social and technical perspectives. Participants emphasized the need for systems security to comply with all current requirements and regulations and retain an ability to evolve to meet future needs. In addition, continued honest communication to the public and other involved stakeholders about risks and benefits will be crucial to building a foundation of trust.
- Provide continuous evaluation and improvement. A learning system is one that assesses its own performance against a set of goals and uses the results of that evaluation to change future behaviors. Workshop participants articulated the importance that all compo-
nents of a digital infrastructure must themselves function as learning systems.
OPPORTUNITIES, CHALLENGES, AND PRIORITIES
During the first meeting, field authorities were invited to set the stage with overview perspectives summarized below on stakeholder views of the vision, data capture and use strategies, patient and population engagement, security and the trust fabric, stewardship and governance, and the global opportunities.
Visioning Perspectives on the Digital Health Utility
Building an effective learning health system requires arriving at a shared vision from sometimes highly varied perspectives. The initial discussion session brought out several of such perspectives, including those of the patient, clinician, quality and safety community, clinical research, and population health.
Informed and Empowered Patients: Moving Beyond a Bystander in Care
Adam Clark, formerly of the Lance Armstrong Foundation (now at FasterCures), shared his vision of a learning health system characterized by bidirectional exchange of health information (individuals as both donors and consumers). In order to support this vision, he described the need to develop appropriate interfaces to encourage and facilitate participation. This includes not only providing the most appropriate information to consumers in an accessible format, but accommodating the participation of family members and caregivers. Dr. Clark highlighted the value of including consumers as information donors in the learning health system, pointing to their ability to contribute types of information—such as accounts of fatigue or depression—and provide a level of context that would otherwise not be captured. He cited data from the Armstrong Foundation indicating that individuals want to share this information as long as their privacy concerns are addressed. Dr. Clark concluded by noting that the escalating complexity of medicine demands new kinds of relationships between patients, clinicians, and researchers, and that the digital infrastructure can serve as a platform for this going forward.
Building a Learning Health System Clinicians Will Use
The perspective of the healthcare team was explored by Jim Walker of Geisinger Health System. He defined a learning health system as one of
goal-oriented feedforward and feedback loops that create actionable information with the potential to effect marked improvements in population health and decreases in the cost of evidence-based care if implemented correctly. Dr. Walker described the steps to building a learning health system, including defining system learning needs and associated questions, identifying the right information to answer those questions and the best methods to collect that information. He noted that an effective learning health system must be useful and useable to all healthcare team members. Dr. Walker described his experiences with health IT implementation at Geisinger and highlighted the complex, sociotechnical nature of the challenge—requiring that as much attention be given to the social aspects as is currently being given to technical capacity. Citing examples of healthcare system learning needs—such as the proper second-line therapy for diabetes—Walker laid out the potential for a learning system to address these questions and feed that information back to healthcare team members. However, he noted, this goal will require fundamental health IT systems redesign in order to support healthcare team decision making.
Improving Quality and Safety
Janet Corrigan from the National Quality Forum (NQF) noted that little progress had been made to improve quality and safety since the publication of the Quality Chasm report (IOM, 2001), and that value has concurrently decreased. She stated that increases in the safety, quality, and effectiveness of health care will require investments in a digital infrastructure capable of collecting information across the longitudinal “patient-focused episode,” and feeding back performance results along with clinical decision support to patients and clinicians. Dr. Corrigan described the framework used by NQF to develop measures for reporting and value-based purchasing, and explored how a digital infrastructure could support capturing the relevant data. Finally, she stated that achieving better health outcomes will require collecting information from, and enabling communication with, individuals both within and outside of traditional healthcare settings.
Clinical Research in the Information Age
The growing information intensity of modern medicine and biomedical research, coupled with advances in computing capabilities, defined the clinical research perspective articulated by Christopher Chute from the Mayo Clinic. He observed that given these concurrent conditions, the technical requirements for information and knowledge management in health care should be high-priority issues. Drawing from examples of “big science” disciplines such as astronomy and physics, he suggested that the future of biol-
ogy and medicine will be characterized by collaborative efforts and shared data and knowledge. As such, he pointed to the need for standardization in order to allow for comparability and consistency in health information. Reviewing the historical state of standards uptake and development efforts, he suggested that meaningful use may be a transformative effort that moves health care in this direction.
Integrating the Public Health Perspective
Martin LaVenture, Sripriya Rajamani, and Jennifer Fritz from the Minnesota State Department of Health shared his account of the opportunities and challenges surrounding a digital platform that supports population health activities. Acknowledging that the learning health system holds great promise for improving health at the population level, he described the need to bolster the capacity and capabilities of population health services in order to realize this potential. The principal challenge, he noted, is the lack of an integrated, modernized digital health infrastructure that is used by a trained workforce and stewarded by public health leaders who understand the potential benefits for population health. Accordingly, he articulated the need for a more unified vision of a digital infrastructure for population health, including the development of a population health approach to data standards; aggregation and infrastructure; and intelligent, bidirectional messaging for patients and consumers.
Technical Issues for the Digital Health Infrastructure
IT constitutes the core of the digital learning health system, and technological innovation in several key areas will be crucial in meeting future needs for security, healthcare quality, and clinical and public health applications. Many of the issues center on interoperability, a feature of IT systems that allows for efficient and useful exchange of a core set of data among an array of systems. Ensuring that data collected in one system can be utilized by other systems for a variety of different uses (e.g., quality, research, public health) is necessary if clinical data are to be collected and analyzed across the entire learning health system to improve health and health care.
Building on the Foundation of Meaningful Use
Douglas Fridsma from the Office of Standards and Interoperability at ONC provided an update on the current standards and interoperability framework being developed. He reviewed several lessons learned in past standards development efforts that are currently informing their approach. Dr. Fridsma described the priorities shaping the work of the Office of
Standards and Interoperability, highlighting the need to manage the life-cycle of standards and interoperability activities by providing mechanisms for continuous refinement. He detailed the model being used in the development of the standards and interoperability framework, which consists of interplay between community engagement, harmonization of core concepts with other exchange models, development of implementation specifications, reference implementation, and incorporation into certification and testing initiatives. Dr. Fridsma emphasized the need to leverage existing work, coordinate capacity, and integrate successful initiatives into the framework.
Interoperability for the Learning Health System
Rebecca Kush from the CDISC suggested that one approach to defining interoperability within the digital infrastructure of the learning health system might be the exchange and aggregation of information upon which trustworthy healthcare decisions can be made. Dr. Kush cited existing enablers that will contribute to this goal, including the Coalition Against Major Diseases’s Alzheimer’s initiative to share and pool clinical trial data across pharmaceutical companies. Furthermore, she noted that a standardized core dataset of EHR information that could be repurposed for research, safety monitoring, quality reporting, and population health would help facilitate an interoperable digital platform for health. Dr. Kush shared several examples of existing standards initiatives that could be leveraged as a foundation for the learning health system—for example, increasing adverse drug event (ADE) reporting through the implementation of the ADE Spontaneous Triggered Events Recording.
Promoting Secure Data Liquidity
Building from the notion of health care as a complex adaptive system, Jonathan Silverstein, formerly of the University of Chicago (now at NorthShore University Health System), asserted that current technological failures of the healthcare system are a result of incompatibility between the technology employed and the nature of the system. He suggested that what is needed is secure data liquidity, supported by a functional architecture that enables ever-expanding secure uses of health data. According to Dr. Silverstein, this can be achieved by employing provable electronic policy enforcement in regard to access, provenance, and logging, as well, through scalable data transport mechanisms and transformations that make data unambiguous and computable. He predicted that the increasing scale and complexity of medicine and biology will lead to more collaborative endeavors and sharing of resources—both data and technical. As a result, he
noted, approaches to sharing technical resources through federated hosted services such as grids and clouds—which provide scalable ways to leverage existing distributed data, transport standards, and individual expertise—promise to be a crucial part of the digital infrastructure.
Innovative Approaches to Information Diversity
Drawing on his experiences with the Indiana Network for Patient Care, Shaun Grannis of the Regenstrief Institute shared his thoughts on what will be needed to mitigate data heterogeneity in a learning health system. Because information needed to support the functions of a learning health system must be compiled from a number of diverse data sources, integrating these data becomes a major barrier to learning. Dr. Grannis suggested that efforts to specify standards for vocabularies, messaging, and data transactions through interoperability specifications, standards, and use cases have not been sufficient to address this issue, and new approaches are needed. He noted that new strategies to deal with patient and provider identity management, vocabulary standardization, and value set maintenance by addressing elements, including patient-and provider-level aggregation, and health system metadata, should be prioritized.
Engaging Patient and Population Needs
The success of the digital infrastructure in improving health will require appreciation, support, enthusiasm, and active involvement from patients and the population. In this respect, measures were discussed on how the case can be best made on the value proposition for patients in terms that matter to them—for example, improved outcomes, enhanced efficiency, better satisfaction, more active participation, and greater equity.
Electronic Health Data for High-Value Health Care
Mark McClellan from the Brookings Institution detailed the essential components of a digital infrastructure that can more closely align quality measurement and improvement in order to achieve high-value health care. He stressed that patient-centered measures, repurposing data already being used to coordinate care for performance measurement, and alignment of these processes with other reform efforts—namely, value incentives—will be necessary to improve care and lower costs. Dr. McClellan used the example of diabetes care coordination to highlight ways in which information could be used to help providers improve care in a timely way, help patients obtain better care, and serve as the basis for driving value-based reforms. He noted that pilots such as accountable care organizations and ONC-funded Beacon
Communities will be instrumental in identifying best practices and aligning processes and incentives for systemwide improvement.
Engaging Individuals in Population Health Monitoring
Addressing the issue of engaging individuals in population health monitoring, Kenneth Mandl from Children’s Hospital Boston observed that harnessing the knowledge possessed by populations through longitudinal studies of large, distributed, consented populations will become the focus of work in population health over the next decade. Based on his experience developing Indivo—a patient-centered health record that places patients in control of their own health information—and recent federal incentive initiatives, he predicted a shift in the health information economy from institutional to individual control. This shift will likely change population health research in a way already being seen through forums such as PatientsLikeMe. Finally, Dr. Mandl noted that a critical research question that needs to be addressed is how to achieve sustained engagement of patients in research.
Optimizing Chronic Disease Care and Control
Sophia Chang from the California HealthCare Foundation noted that a digital infrastructure provides important opportunities for informing and improving the care of patients with chronic disease. She discussed the potential to actively engage patients in the management of their conditions, but observed that, currently, this is not possible as the locus of control lies solely with healthcare providers and not patients. Additionally, Dr. Chang pointed to the lack of common nomenclature, data formats, and protocols for incorporating patient-generated information as barriers to aggregating and translating health data into useful decision support. Pointing to Kaiser Permanente and VHA as examples of institutions that have successfully used EHRs for population health management, she acknowledged that smaller institutions or individual physicians might have less opportunity for exposure, and therefore be less aware of the value. She noted that in order to maximize the value of EHRs, research paradigms should shift to real-time knowledge development and feedback. Finally, Dr. Chang highlighted several steps to move toward the goals of recentering the system around the patient, such as providing useful support for chronic disease management, aligning EHR data elements with patient priorities, and developing better paradigms for learning from patient data.
Targeting Population Health Disparities
M. Christopher Gibbons of the Johns Hopkins Urban Health Institute discussed opportunities for using a digitally supported learning health system to better comprehend and combat health disparities. Noting that understanding and treating health disparities requires integrating knowledge spanning many sources and disciplines, he pointed to several demographic trends that make this challenge ever more pressing—rising prevalence of chronic disease, an aging population, and the growing racial and ethnic diversity of the U.S. population. Dr. Gibbons introduced the terms “populomics” and “populovigilance” to describe the integrative, systems-oriented, and informatics-intensive approaches to understanding and monitoring the complex causes and manifestations of diseases and disparities. He suggested that as more and more data from diverse sources are collected and available for analysis, it will be important to adopt these new perspectives in order to enable advances in treatment, public health, and healthcare disparities.
Weaving a Strong Trust Fabric
Building trust among all the stakeholders—in particular, patients and the public—is vital to progress. The various dimensions of this issue include building confidence in the security safeguards for clinical data, deepening the appreciation for personal and population health, the fundamental value of sharing data for research purposes to support better care decisions, and economic advantages that result from a well-developed digital health infrastructure and clinical data utility.
Demonstrating Value to Secure Trust
Edward Shortliffe of the American Medical Informatics Association addressed the need to build a strong fabric of trust among stakeholders by communicating and demonstrating value. He stated that in order for health IT to meet its full potential, patient and provider participation must be secure. This sense of security depends on an appreciation of the value presented by HIT use as well as creating and maintaining proper security and safeguards. Sharing a personal anecdote about a provider who admitted that only patient demand would motivate him to adopt an EHR system, Dr. Shortliffe observed that sufficient patient demand may even obviate the need for federal incentives. Using electronic banking as an example, he suggested that educational programs are necessary to inform stakeholders about the risks and benefits of EHRs, and predicted that with the establishment of an environment of trust the increased convenience and quality offered by EHRs and data sharing would overcome concerns about privacy.
Currently, however, the risks of adopting an EHR system are both better understood and more effectively communicated. As a result, he suggested that the focus of stakeholder engagement activities going forward should be on communicating the benefits of EHR use—most importantly, better care and lower costs—to providers and the public.
Policies and Practices to Build Public Trust
The implementation of fair information practices to ensure privacy and security was the focus of the Center for Democracy and Technology’s Deven McGraw’s remarks. Citing surveys that show individuals desire electronic access to their health information even though they have significant privacy concerns, she suggested that providing individuals with meaningful choices around privacy is an important approach to addressing these concerns. Ms. McGraw pointed to a comprehensive approach to patient privacy and data security based on the Markle Common Framework for Secure and Private Health Information Exchange. Key elements of the framework include an open and transparent process, specification of purpose, individual participation and control, and accountability and oversight. Closing with a warning that overreliance on consent leads to weak protection—shifting the burden of privacy protection from the institution to the individual—and that existing regulations are insufficient to cover the privacy issues inherent in a learning health system, she underscored the need for a trust fabric based on fair information practices.
HIPAA and a Learning Health System
Since its passage in 1996 the Health Insurance Portability and Accountability Act (HIPAA), has served as the legal and policy framework for health information privacy. Bradley Malin of Vanderbilt University described the current state of play around health data de-identification and highlighted some of the relevant learning health system–related issues posed by HIPAA. Included among these were identity resolution (while maintaining privacy) and concerns that de-identification could cause modifications to patient information to the extent that they influence the meaning of clinical evidence. Dr. Malin noted, however, that these challenges are not insurmountable, and that efforts to quantify risk are an important first step to mitigation. He suggested that use cases to better define health information utility and improved capabilities for distributed query-based research will be important in moving to a privacy-assured learning health system.
Building a Secure Learning Health System
Ian Foster of Argonne National Laboratory addressed the technical components surrounding trust in the digital infrastructure for the learning health system. He laid out a number of challenges facing the establishment of a secure digital platform, for example, the fact that a learning health system requires data sharing on an unprecedented scale, and that the purpose of this sharing needs to be extended beyond individual patient care support to include research and population health. Highlighting the challenge of a highly complex system with an unclear definition of security, Dr. Foster suggested some basic principles and technology solutions that can form a basis for progress: auditabililty (information can be mapped to an individual and data can be mapped to its origin); scalability; and transparency in terms of data usage, policies, and enforcement. Methods to achieve these principles include attribute-based authorization, distributed attribute management, and end-to end (scalable) security.
Stewardship and Governance in the Learning Health System
The growth and development of the digital infrastructure for health will be determined in part by the effectiveness of the stewardship and governance instruments designed to facilitate its appropriate structure and function, as well as enlist and channel the engagement and balance of stakeholder interests.
Governance Coordination, Needs, and Options
Laura Adams of the Rhode Island Quality Institute identified and addressed fundamental questions posed in contemplating the governance of the digital health infrastructure. Focusing on the source and scope of authority; mission, purpose, and primary goals; and theoretical foundations for a governance structure, she laid out several governance options for consideration. Ms. Adams suggested that all of these potential models of governance structure and stakeholder participation should be considered, and that the scope of the governing body’s authority should be succinctly communicated in a statement of purpose. This statement, she noted, should draw on guiding principles such as transparency and commitment to the common good, and that considering guiding theories—such as complexity theory—could aid in providing an ethical and legal framework. Pointing to some of the unique governance challenges posed by a learning health system, such as evolving privacy considerations and accommodating new sources of data, Ms. Adams suggested drawing on past successes and experiences while incorporating the widest array of viewpoints possible.
Consistency and Reliability in Reporting for Regulators
Theresa Mullin from FDA described ongoing efforts to implement a systematic strategy for data standards development and adoption. This process would address heterogeneity in new drug applications, improve regulatory efficiency, and contribute towards the agency’s public health mandate by facilitating exploration of safety and efficacy issues. Dr. Mullin suggested that, through the standardization of clinical data in EHRs, this effort presents an opportunity to facilitate information exchange and analysis for learning, reduce costs, and reduce burdens on providers for adverse event reporting. Dr. Mullin also highlighted some of the overarching governance principles driving this effort: an open, transparent, and inclusive process, as well as the need for resulting requirements to be practical, user-oriented, sensitive to costs, and sustainable.
Complying with Patient Expectations for Data De-Identification
Shawn Murphy from Partners HealthCare explained that meeting patient expectations for privacy and security is central to developing a learning health system. He detailed how current limitations to privacy through de-identification could be overcome by a comprehensive security and privacy approach that does a better job of addressing patients’ chief concerns around health information protection—avoiding embarrassment and economic risk. Citing an example of research program–based restrictions on physician access to data—whose risk to patient privacy is negligible given physicians’ otherwise broad access to patient information—Dr. Murphy suggested that the certified trustworthiness of the recipient should be a component of access control. He went on to note that this, coupled with appropriate de-identification and secure data storage, provides a balanced approach to security that better matches the expectations of the patient while facilitating access for approved data users.
Information Governance in the National Health Service (UK)
Guidance for approaches to governing the digital health infrastructure can be gleaned by drawing from examples of similar efforts. Harry Cayton of the National Information Governance Board for Health & Social Care (NIGB) in the United Kingdom described the approach they have taken in dealing with information governance issues facing the National Health Service. Cayton detailed the role played by the NIGB as an independent statutory committee to advise the government on the use of patient-identifiable data for clinical audit and research. He described their philosophy that information governance (or stewardship) is the responsibility of every orga-
nization involved and provided a list of principles developed by the committee to guide their work. Stating that the purpose of the NIGB is to deal with the “wicked questions” that arise around use of health information, Cayton affirmed that there is no right or wrong answer, only the best answer at the time. In conclusion, he suggested that all governance systems need the same things: mechanisms for agreeing and applying consistent principles, checks for the practicality of guidance given, consistent procedures, and credibility with stakeholders.
Perspectives on Innovation
Especially in a field as rapidly evolving a HIT, innovation is the lifeblood of progress. Observations on innovative approaches to current obstacles and challenges were invited from several field innovators.
Conceptualizing a U.S. Population Health Record
Drawing from the assertion that population health is more than the aggregation of individual disease and that therefore, an understanding of population health cannot simply be gleaned by aggregating patient care data, Population and Public Health Information Services’ Daniel Friedman advocated for the creation of a U.S. population health record. He emphasized that while the United States has large amounts of publicly accessible population-level disease-related data, challenges for population health include a lack of that same level of granularity for functional status and well-being as well as problems of data integration and integrity. In order to address these issues he proposed the establishment of a single source of population health data backed by an overarching data model and theoretical framework. In this model, data would be drawn from a number of different sources including those not typically integrated with clinical data such as environmental sampling and census data.
Accelerating Innovation Outside of the Private Sector
Molly Coye, formerly from the Public Health Institute (now with the University of California, Los Angeles), identified what she saw as three areas of opportunity for HIT innovation. Citing the need to improve the current state of clinical decision support, she suggested areas where innovation could help meet this goal: how to recognize and deal with incorrect or missing data, how to integrate a single patient’s data from multiple sources, and how to turn data into clinical guidance. Dr. Coye cited the need for integrating research into care processes and for evidence generated to be fed back in a continuous, seamless process that supports informed, shared deci-
sion making. Additionally, she noted the movement of healthcare delivery to integrated models—such as accountable care organizations—which increase the need for remote data collection, diagnosis, consultation, and treatment. Dr. Coye concluded by stressing that many of these challenges are social rather than technical in nature, and successful approaches, therefore, will need to take into account the complex character of these systems.
Combinatorial Innovation in Health Information Technology
The growing prevalence of personal information ecologies provided the context for remarks made by the Institute for the Future’s Michael Liebhold. He noted that these ecologies are composed of digital artifacts not only related to health and fitness, but also social activities, media use, and even civic life. Mr. Liebhold observed that citizens are ready and willing to collect and share their health information and, with the encouragement of industry and employers, to become more actively involved in their own health. However, he noted that effectively integrating information from all of these sources in a meaningful way presents a formidable challenge. Technologies such as those that underlie the semantic web hold much promise, but still face challenges, especially in the areas of privacy and security. Looking to the future, Mr. Leibhold stated the need for methods to curate web-based health information; interoperable health app stores; and the development of a web of linked, open healthcare information and knowledge interoperability.
Fostering the Global Dimension of the Health Data Trust
The ability to draw broadly from anywhere across the globe for lessons that can provide relevant insights for health and healthcare improvement is a long-term goal. Meanwhile, the ability to learn from the experiences of other countries, as well as to apply HIT for biosurveillance, can help facilitate progress. Several relevant activities were reviewed.
TRANSFoRm: Translational Medicine and Patient Safety in Europe
Brendan Delaney from Kings College London described the TRANSFoRm project, a European Union (EU) effort to develop a learning health system driven by HIT and aimed at improving patient safety as well as supporting and accelerating clinical research. Dr. Delaney outlined several of the challenges that have arisen, such as system interoperability, a need for advanced functionalities, and the support of knowledge translation. He also described several techniques being employed to address these challenges, including clinical research information models, service-based
approaches to semantic interoperability and data standards, detailed clinical data element representations built on archetypes, and an effort to prioritize EHR workflow integration in the development of clinical decision support systems capable of capturing and presenting fine-grained clinical diagnostic cues.
Healthgrids, the SHARE Project, and Beyond
Drawing from his involvement with SHARE, an EU-funded project to define the path toward greater implementation of grid computing approaches to health, Tony Solomonides from the University of the West of England discussed his current work to automate policy and regulatory compliance to allow health information sharing. He described an approach to the implementation of attribute-based access controls to ensure enforcement of privacy obligations which, becuase of variations in their interpretation between EU countries, require a logic-based computed approach.
Systematic Data Collection for Global Improvements in Care
Health IT holds great promise to increase quality and improve patient safety in developing and transitional countries. Harvard University’s Ashish Jha described how a dearth of reliable information has impeded efforts to both better understand and design solutions to higher rates of adverse event–associated morbidity in developing countries, as well as obtain an accurate calculation of global disease burden. Dr. Jha described an effort by the World Health Organization to maximize the impact of HIT in resource-poor settings through the development of a minimum dataset that would allow for systematic data collection of elements relating to safety issues.
Informatics and the Future of Infectious Disease Surveillance
David Buckeridge, from McGill University, described how HIT is enabling dramatic changes in domestic and international infectious disease surveillance. Detailing how the digital infrastructure can enhance existing systems through the use of automation and decision support, he explained novel approaches to surveillance enabled by recent informatics innovations. Using the DiSTRIBuTE project as an example of syndromic surveillance innovations that drastically improve coverage and speed, he called for a renewed science of disease surveillance that embraces IT—along with the potentially disruptive changes it brings—to improve disease control.
GROWING THE DIGITAL HEALTH INFRASTRUCTURE
Drawing on the collective expertise represented in the presentations and discussions of the first workshop, participants focused on four crosscutting priority domains in the two subsequent workshops: promoting technical advances and innovation, knowledge generation and use, engaging patients and the population, and fostering stewardship and governance. Encouraged to give due consideration to “out of the box” approaches and to use examples from health and nonhealth fields to illustrate and test key needs and opportunities through small group sessions, participants identified and presented for discussion a number of strategic elements important to progress in each domain. They are included in Box S-9 and described in more detail in the sections below.
A ULS system is complex, constantly growing, and evolving, much like an organic, biological ecosystem. Introduced to the digital health information conversation by colleagues from the computer science field, hallmarks of a ULS system were described earlier under “Common Themes and Principles” (see Box S-8), and include its decentralization of data, development, and operational authority to foster local innovation, personalization, and emergent behaviors without requiring consensus from all nodes. The complexity, constant evolution, and enormous scale of the digital health infrastructure is consistent with the ULS system framework and terminology. During the discussions focused on developing a set of strategic scenarios for technical progress, the ULS system approach emerged as an appropriate framework.
In discussing the implications and issues surrounding this approach, participants identified the relevance and appeal of the engineering approach to health care—systems analyses, design, implementation, and evaluation plans—inherent to the ULS system perspective. Specifically, they noted the potential of a collaborative effort between the computer science and HIT communities to develop a deliberate and systematic engineering analysis—characterized by iterative testing and development of prototypes—to set technical and sociotechnical system goals, requirements, specifications, and architecture. This could be supported by a multidisciplinary research community, armed with clarified terminology for ease of collaboration, and with participation from a wide array of both private and public stakeholders (clinical, public health, computer science, health informatics, law, policy, ethics, etc). Similarly, workshop participants stressed the need for technical policies that support experimentation and innovation and allow for the
TECHNICAL PROGRESS…activities that advance:
- Ultra-large-scale system perspective
- Functionality focus
- System specifications/interoperability
- Workflow and usability
- Security and privacy safeguards
- System innovation
KNOWLEDGE GENERATION AND USE…activities that advance:
- Shared learning environment
- Point of decision support and guidance
- Research-ready records for data reuse
- Patient-generated data
- Integration and use of data across sources
- Distributed data repositories
- Sentinel indicators
- Query capacity
- Analytic tools and methods innovation
PATIENT AND POPULATION ENGAGEMENT…activities that advance:
- Value proposition and patient confidence
- Shared learning culture
- Patient–clinician outcomes partnerships
- Person-centric, lay-oriented health information access
- Closing the disparity gap
- Continuous evaluation
GOVERNANCE…activities that advance:
- The vision
- Guiding principles
- Participant roles and responsibilities
- Process and protocol stewardship
- Implementation phasing
- Continuous evaluation
progressive adoption and evolution of system requirements, specifications, and architecture choices.
Participants pointed to a focus on functionalities consistent with ULS systems, and their application to the digital health system, as a potential starting point in advancing the ULS approach. Definition of the ULS principles and characteristics that support learning system functionalities,
including the feedback and feedforward nature of the learning engine, such as identification strategies, privacy controls, the availability of a complete longitudinal record at the point of care, inferential capacity, and research readiness, were highlighted as critical foundational steps in the development of this technical enterprise. Noted as similarly important to system functionality was the mechanism for developing and maintaining an approach to information structure, classification, and storage.
Promoting these targeted functionalities requires advancing parsimonious system specifications and interoperability. Discussions centered on the need to specify the minimum set of standards to allow for partial interoperability. Semantic comparability, maintenance of context and provenance, architectural consistency, and transportability were discussed as potential starting points. In congruence with the priorities laid out subsequently in the PCAST report (see Appendix E), particular attention was paid to the use of metadata to facilitate interoperability and information exchange—including to maintain data context and provenance, authentication, and privacy. This, in concert with a fast-prototyping component, can allow for incremental specification and system growth with the opportunity for functional enhancement, such as refinement of semantic interoperability, to meet specific requirements depending on use.
Part and parcel with the need to address the technical specifications of the digital utility for the learning health system is consideration for how these interface with users. Considerations for workflow integration were discussed by workshop participants as important to ensure that the technology is not only innovative and useful but also usable. To date, this disjuncture between established workflow patterns and an unfamiliar, often awkward, overlay of HIT tools has proved a substantial barrier to adoption.
Security and privacy safeguards were an important consideration in all areas of discussion. Participants often pointed to a lack of trust as being one of the major impediments to health information exchange. Therefore, attendance to the technical aspects of these issues was emphasized as a crucial part of building trust among stakeholders. Discussions and presentations described technical approaches such as attribute-based authorization and distributed identity management, and provided examples of how they could be deployed to address these concerns and achieve a state of secure data liquidity. Additionally, innovations around data security and privacy in alternative environments such as hosted, web-based systems were suggested in order to build capacity.
Finally, the need for continuous innovation was a recurring theme in technical discussions. Participants suggested strategies such as creating a test-bed network for assessment of innovative system functionalities, the use of challenge problems to test ULS system issues and opportunities, and
the cultivation of interdisciplinary research initiatives among academic, industry, and government stakeholders.
Knowledge Generation and Use
Discussions of the generation and use of knowledge fell into three areas: the availability and capture of reliable data, the tools to analyze the data, and seamless feedback of knowledge to the system. Research, quality improvement initiatives, and public health surveillance efforts are all examples of uses and drivers for these learning-associated processes.
A necessary precondition for successful progress on any of these dimensions is a shared learning environment. Technical advances and innovative research methods make it possible to bring clinical research and clinical practice much closer together. However, it was noted that the ability to take advantage of that opportunity depends on a healthcare culture in which both patients and clinicians are compelled by the prospects of clinical data to improve understanding, care delivery, and outcomes, as well as provide reliable, just-in-time information to assist decision making. For these reasons, participants highlighted the need for a learning environment that is supported, shared, and nurtured by both patients and clinicians.
Several tools and approaches currently exist to provide point of decision support and guidance. In the face of the number of interacting factors, competing priorities, and an ever-growing set of diagnostic and therapeutic options, “best practice” can only be a theoretical notion without the ability to bring the best available information to the decision process. On the other hand, it was noted that reminders and decision prompts not successfully engineered into natural workflow patterns will be little more than ignored distractions. Consequently, approaches are needed to better marshal reliable clinical information and guidelines in time, form, and content that is seamlessly accessed and used by clinicians and patients.
Participants identified a number of needs to be addressed in order for the digital health infrastructure to reach its full potential as a source of real-time clinical research insights. For example, clinical research activities require enlisting clinician support and involvement in research-ready clinical records on both quality and content dimensions for reuse in knowledge generation. The identification of a limited set of standardized core research-related components as basic elements across vendors and systems was one suggestion to facilitate individual and cooperative clinical research activities as well as sentinel event surveillance. Concerns over the reliability and heterogeneity of data in clinical records was underscored as an important rate-limiting factor for both quality of care and clinical research activities, again underscoring the importance of the mechanisms for information structure, classification, and storage. This is particularly important for repurposing
data collected for other uses, such as FDA clinical trial–associated data, in order to maximally leverage efforts and investments already in place.
Discussions on the increased utility of clinical records for research went hand in hand with those on the need to take advantage of information from patients and other sources. Patient-generated data can provide a level of context that is impossible to capture through more traditional data collection methods. Initiatives to better develop, test, and improve the capture and use of these data so that they can be used to support research, quality improvement, public reporting, and patient care were suggested as priorities.
Similarly, efforts to promote the integration and use of data across various sources—clinical, public health, commercial—were emphasized as central to effectively leveraging the full range of information for progress in improving efforts aimed at populations as well as individuals. Included in this, and considered with a longer term vision, were growing information sources outside of “mainstream” health care, such as online forums and communities. In order for such proposals to be successful, it was noted that protocols must be developed to build interoperability as a natural and seamless element of data sources.
Storage and aggregation of data for the purpose of analysis and knowledge generation have been problematic given the security and privacy issues they entail. Discussions of current and ongoing efforts in the creation of distributed data repositories, such as those being used in FDA’s Sentinel Initiative and the HMO Research Network, suggest a promising approach. Coordination between these ongoing efforts, additional support and incentives for their use for clinical research activities, and the support of coordinated intervention-specific patient registries were discussed as potential approaches moving forward. Prospects for the use of scalable, distributed, hosted, storage solutions—such as those used by Amazon—were also noted as promising future directions. These discussions, however, were often punctuated with caution around privacy and security, components that participants felt needed further exploration and development.
Finally, considerable attention was paid to the development of methods, tools, and query capacity for the generation of knowledge needed to sustain a digital learning health system. In line with the ULS system architecture approach, and the creation and support of distributed data repositories, the development of capacity for national, distributed query-based research— including the ability to identify and track sentinel events and indicators—was identified as a strategic priority. Challenges associated with the current state of public health IT infrastructure were cited as priorities for attention in order for these functionalities to be adequately sustained. To support this, and the continuing development and innovation around other analytical approaches, the importance of collaborative interdisciplinary networks of researchers was underscored. This was discussed
not only for cooperative studies, but for cooperative engagement of issues such as strategies on consistent identifiers for patients, the use of modeling and simulation for knowledge generation, evaluation of approaches for the use of diverse data types and varying data quality, and development of methods for the use of information from mobile consumer devices and patient-generated data.
Patient and Population Engagement
Discussions on the roles of patients and the public in growing the digital infrastructure for the learning health system were anchored strongly in the concept of reengineering the care culture to ensure the centrality of the individual patient in the care process—a concept underscored in the Quality Chasm report (IOM, 2001) that remains elusive. Signs of change are only beginning to appear as appreciation increases for the use of web-based information and the clinical and outcome advantages of a patient who is better informed and more involved. Often referenced in the discussions was the need for the establishment of a “new norm” around engaging patients and the public in health—both theirs and that of the population—through the use of the digital infrastructure. Basic to this “renorming” is a deepened appreciation for the personal and public benefits that are likely to occur, as well as a strong measure of confidence in the security of the system.
The value proposition must be apparent to the stakeholders. Communication of the value of a digital health infrastructure in the improvement of care coordination, quality, and, ultimately, the health of the population at large, was identified in workshop discussions as a fundamental priority. Furthermore, participants pointed out that, in order to be successful, the value proposition should be approached in the context of transparent conversations about privacy, security, and other impeding concerns. The use of case studies and quantitative assessments of the contribution of HIT to improved patient experiences and outcomes was discussed as potential starting points.
A common theme across several workshop discussions was the value in fostering a shared learning culture among system stakeholders—in particular, a culture that recognizes the unique contributions that patients and the general population can make to the learning system as collaborators, not subjects. Activities that foster patient involvement in and support of knowledge generation, including illustrating the importance of patient preference information to improving care, were discussed as potential approaches to this issue.
Following the theme of “renorming” participation of patients and the population in health improvement, and building on the framework established by previous IOM work in this area, the opportunity for strengthening patient–clinician outcome partnerships through the digital infrastructure
was discussed. The development of templates and protocols that support the use of HIT to engage patients in decision making as well as tools for more effective provider–patient communication were proposed. An important element in this respect is providing patients with secure access to and control of their health information. This includes further development of patient portals, building on technologies already widely accepted by consumers, and supporting efforts for increased information liquidity and control such as the VHA/CMS Blue Button initiative.
In concert with these efforts, participants discussed the need to increase the availability and access to lay-oriented, user-friendly clinical and nonmedical health information. Investing in templates for form and content of information for the lay consumer, as well as gathering patient-derived data for care and delivery improvement were suggested as areas of focus. Indeed, the “new norm” was discussed as involving a focus on improving patients’ health, not just health care, by emphasizing health maintenance as a lifelong process that includes a patient’s actions and decisions outside of the clinical care setting. To this end, many participants proposed providing individuals with useful information concerning their clinical encounters and the relevant state of evidence, as well as giving them more responsibility for utilizing this information in their own decision making.
HIT provides an opportunity for engaging populations not historically well served by the traditional healthcare community. For this reason, the potential of the digital health utility in the elimination of health disparities was discussed as a strategic priority for further attention and action. The impact of facilitating patient and population contribution to, and control of, their health information has the potential to address disparities in underserved populations.
The importance of a component of continuous evaluation and improvement in efforts for patient and population engagement in the digital health learning system was again emphasized. Areas of focus that were highlighted include ongoing assessment of patient preferences for use in tailoring of health plans, innovative approaches to confidentiality and privacy issues, and assessments of opportunities to use contemporary sociotechnological approaches (e.g., social networking and smart phones) for patient and population engagement.
Discussions of governance strategies for the digital infrastructure for the learning health system focused on facilitating activities to advance some very basic components and principles of the ULS digital health information system. Participants often struggled with the question “what are we proposing to govern?” and certainly the health information system as it
exists now does not easily fit into most established governance models. On the other hand, upon applying the ULS lens to this issue, and considering innovative governance approaches in cases outside of health (such as VISA and the Smart GRID; see Appendix B for more information), certain governance-related strategic elements emerged. Several participants pointed to the example of the Internet Engineering Task Force as one example of a governance approach that, while created under different circumstances, reflects many of the same governing principles.
Of principal concern is the issue of the vision. As a means of establishing a reference point for progress, workshop participants articulated the need for work to establish a shared vision of the digital health utility for the learning health system. Prospective components noted for this vision include expectations, guiding principles, modus operandi, and an appreciation for the global perspective. Considerations of the differences between a structure that governs versus one that provides guidance were included in these discussions.
Participants noted that a governance model in line with the ULS approach would be one that identified and depended on a minimal set of guiding principles with which all stakeholders must comport, maximizing local autonomy over all other decisions. Tolerance of change and adaptability were additional characteristics that participants felt were important to incorporate. Exploring the most decentralized level at which these standards might be delegated and focusing standards on major functional requirements were proposed as starting points. Additionally, the importance of tailoring the governance approach to the local situation and needs was emphasized. A focus on the ability to use an inclusive (both/and) rather than a deterministic (either/or) approach was discussed as a foundational principle that encapsulated this thinking. A related issue discussed was the broader context of the governance enterprise. Participants discussed the need to include societal values such as trust, privacy, and fairness; fair information practices such as transparency and collection and data use limitation; goals of the health sector to improve quality of care and enhance clinical knowledge; technical concepts such as innovation; and economic aspects such as promoting efficiency and reducing costs.
Identification of possible participant roles and responsibilities in the governance structure were identified as an important early step, and different approaches were considered. These included broad participation by all stakeholders, which was pointed out to be logistically very difficult; very narrow participation, which participants felt was unlikely to be successful; or a hybrid model that incorporated both broad and narrow participation depending on the needs at that particular level. Some participants noted that multiple layers of governance were likely to be required to address concerns at the appropriate level, whether local, regional, national, or international.
Several approaches to the establishment of a governance model were considered and discussed by workshop participants. Leveraging lessons through collaborative discussions among ongoing efforts—at both the national and local levels—and establishing a working group to begin collecting initial input were suggested as starting points. To enhance the efficiency of deliberative efforts, participants suggested coordinating these activities, potentially through the ONC HIT Policy Committee’s Governance Working Group; building upon and aligning existing policies, such as HIPAA, agency regulations, and informed consent processes to encourage learning health system activities; and nurturing the interfaces with the international community.
A potential responsibility discussed for the governance structure was the stewardship of processes and protocols associated with learning health system functionalities. Participants noted that developing processes for proposing, reviewing, and validating protocols on key elements including data gathering, security, and use is an integral part of this approach. Ongoing stewardship responsibilities for the governing entity will involve monitoring and maintaining protocols, managing variability across participants, and devising an approach to provide incentives to stakeholders to conform to stated goals and principles. A related element discussed as a governance challenge was that of implementation phasing, or sequencing protocol development activities so that barriers to progress in an entrepreneurial environment are not presented.
In the spirit of a continuously improving learning health system, a process for continuous evaluation and improvement of the governance entity and approach was emphasized as important. Areas highlighted included establishing an approach to ongoing assessment of progress and problems, systematic assessment of value realization for recognition and promotion of successful practices, and the support of research on governance and orchestration of the ULS digital health utility in the United States and globally.
Throughout the meetings—and especially at the third meeting—a number of specific cross-cutting action targets were identified as particularly pressing elements for attention. In several instances these involved seizing on the opportunities presented by ongoing efforts, and building upon them to include considerations or requirements specific to the learning capacity of the digital infrastructure. Those most frequently mentioned are presented in Box S-10 and described in more detail below.
Priority Action Targ ets Discussed
The case: Analyses to assess the potential returns on health and economic dimensions
Involvement: Initiative on citizens, patients, and clinicians as active learning stakeholders
Functionality standards: Consensus on standards for core functionalities—care, quality, public health, and research
Interoperability: Stakeholder vehicle to accelerate exchange and interoperability specifications
ULS system test bed: Identify opportunities, implications, and test beds for ULS system approach
Technical acceleration: Collaborative vehicle for computational scientists and HIT community
Quality measures: Consensus on embedded outcome-focused quality measures
Clinical research: Cooperative network to advance distributed research capacity and core measures
Identity resolution: Consortium to address patient identification across the system
Governance and coordination: Determination and implementation of governing principles, priorities, system specifications, and cooperative strategies
Priority Action Targets
The case: Analyses to assess the potential returns on health and economic dimensions. Because of the centrality of broad-based support to progress, and the “public good” nature of many of the activities, the need to demonstrate a value proposition or business case for participation by stakeholders in a digital learning health system was a topic of much discussion during the workshop series. This emphasis was reinforced by the approach taken by the PCAST report to encourage the development of a market around digital health information exchange. Support of methods that apply serious analytical rigor to these issues and generate both technical and policy suggestions were identified as being crucial to this effort. Researchers and organizations such as think tanks were discussed as likely being the best positioned to undertake the necessary analyses with support of a commissioning resource.
Involvement: Initiative on citizens, patients, and clinicians as active learning stakeholders. Many workshop discussions considered stakeholder investment to be a necessary component of any successful strategy. Participants identified the need to redefine the roles of citizens, patients, and clinicians in a way that activated their participation in their own health, and the health of the population at large, through the facilitative properties of the digital infrastructure. It was noted that patient and clinician groups can play a crucial role in this effort by helping to convey the value proposition and ensuring that the interests of their constituents are represented in the development and evolution of the system. Efforts that facilitate stakeholder participation—such as increased control of health information by patients and the use of patient-generated data in care plans and knowledge generating processes—were discussed as priority next steps in stakeholder engagement. Additionally, to attend to concerns around privacy, security, trust, and additional work burden, participants stressed the importance of honesty and transparency in facilitating support and understanding. Ultimately, discussions noted that demonstrating the value of a digital health infrastructure through the use of case studies that point to improved outcomes and efficiency was likely the most compelling strategy to appeal to stakeholders.
Functionality standards: Consensus on standards for core functionalities—care, quality, public health, and research. Progress on the technical standards necessary to support the core functionalities of the learning health system was continually referenced in workshop discussions. Participants focused on the standards necessary not only to improve, monitor, and guide care decisions but also to accelerate research, quality efforts, patient moni-
toring, and health surveillance. Related requirements include the ability to exchange information through the use of minimal standards (such as those to enable use of metadata-tagged information packets), query and analyze distributed repositories of data for research purposes, ensure care decision support, and enable quality improvement initiatives and public health surveillance and reporting. Discussions also touched on the need for the digital infrastructure to interface with next-generation systems including mobile health applications and the way in which these and other capacities could help engage patients and the public through improved information access. Participants also underscored the strategic importance of adhering to a minimal set of standards that support core functions but do not introduce unnecessary barriers to progress.
Interoperability: Stakeholder vehicle to accelerate exchange and interoperability specifications. System interoperability remains a major obstacle to realizing a digital learning health system. When applying the ULS system lens to this challenge, many participants stressed the need to develop a parsimonious set of standards—such as those for metadata—to allow for practical interoperability and information exchange across systems. Noting that this issue lies in the realm of both technical capacity and governance structure, several participants often compared this effort to the evolution and governance of the Internet. While the differences between the digital health infrastructure and the Internet were acknowledged, it was suggested that the establishment and work of the Internet Engineering Task Force might provide guidance for an industrial institution for the governance of interoperability-related standards. Additionally, leveraging and coordinating existing progress and ongoing efforts in the areas of standards development and facilitation were underscored as strategies to ensure that activities progress as efficiently as possible.
ULS system test bed: Identify opportunities, implications, and test beds for ULS system approach. As discussions focused on the characterization of the health system as a complex sociotechnical ecosystem, analysis was suggested on how the ULS approach might be applied to the health system in both the short and long term. Mapping of a key ULS system report (Northrop et al., 2006) to the learning health system through a collaborative effort between software engineers, computer scientists, medical informaticians, and clinicians was offered as a starting point for this effort. Furthermore, performing a rigorous engineering systems analysis leading to a concept paper was suggested to clarify further the opportunities and implications for the ULS system approach. Integral to the ULS approach is the need to support rapid prototyping for continuous innovation. It was suggested that test beds for the development, assessment, and dissemination
of these prototypes would be central to continual innovation. In this vein, several participants pointed to the opportunity presented by the creation of the Center for Medicare & Medicaid Innovation (CMMI). Certain communities of excellence already provide some capacity in this area, and participants often referenced ongoing activities at these institutions (see Appendix B).
Technical acceleration: Collaborative vehicle for computational scientists and HIT community. Much of the work in the development of a digital learning health system will necessitate interdisciplinary collaboration between academic, public, and private partners across the computer science, HIT, science, and engineering communities. Participants suggested establishing a collaborative forum where these efforts can be initiated and developed. This forum could catalyze the interdisciplinary research program necessary to develop the digital health infrastructure, and some participants suggested that funding for such a forum and its associated activities might best be served by collaborative efforts across relevant federal agencies (such as NIH and NSF), relevant private sector partners, or both.
Quality measures: Consensus on embedded outcome-focused quality measures. Participants noted that the first step in determining the usefulness of data collected by the digital health infrastructure is to identify the necessary elements to collect. It was stated several times that in order to support the quality improvement and research activities required for a learning system, consensus around useful outcome-based measures is needed. Participants suggested that this would motivate vendors and users to incorporate these measures into their systems, driving seamless integration of quality measurement and reporting into the digital infrastructure. Work at the NQF, through the ONC HIT Policy Committee, and at CMS has already begun addressing these needs.
Clinical research: Cooperative network to advance distributed research capacity and core measures. Discussions often highlighted the centrality of ongoing and continuous generation of knowledge from clinical data as a central feature of the learning health system. Efforts to do research on data held in distributed repositories, such as the HMO Research Network and FDA’s Mini-Sentinel program, were pointed to as important early-stage efforts in building systematic, larger scale capacity. Participants suggested that a multidisciplinary, cooperative network of the relevant stakeholders—principally computer scientists, clinical researchers, and data holders—could be a starting point in accelerating progress in this dimension. It was noted that this network would need to consider development of core datasets to facilitate research and quality efforts, fostering consensus on
levels of consent and de-identification strategies necessary for effective reuse of data, development of methodologies for query-based and automated research and signal detection across distributed systems, development of standards for distributed queries across the system, implications for a ULS approach to existing and future distributed networks, and implications for distributed research from possible advances in data structure and packaging strategies for data interoperability and exchange across systems.
Identity resolution: Consortium to address patient identification across the system. One of the major barriers discussed for several key system functions—care appropriateness, continuity, quality assessment, and research—relates to the current inability to reliably track and link individual patients with their associated information across the health system. This poses a problem for issues around care coordination, including the goal of being able to make care decisions based on comprehensive health information, as well as the development of a useful knowledge generation engine that can incorporate all relevant information and deliver useful, accurate support. Privacy and system security are paramount, but participants noted that approaches are available to address these issues responsibly and the barrier appears to be one of cultural hesitancy rather than a lack of technical capability. Targeting this issue through a consortium approach was proposed as a way to provide the opportunity for stakeholder representation and engagement in an honest, transparent conversation about the component value issues involved.
Governance and coordination: Determination and implementation of governing principles, priorities, system specifications, and cooperative strategies. Workshop participants articulated the idea that governance principles and priorities for a learning health system will require breaking new ground both organizationally and functionally. Discussions identified the need to improve coordination among key stakeholders to accelerate progress in identifying and sharing lessons, examining commonalities, and exploiting opportunities for efficiencies. It was noted that broad agreement will need to be cooperatively marshaled to attend to principles and priorities that support learning system functionalities such as data integrity, policies for data use, human subjects research issues, and proprietary interests. In addition, discussions highlighted the role of governance in planning for and mitigating system failures, an inevitable occurrence in all systems, but one particularly well tolerated within the ULS system. Such failures would, of course, be opportunities for learning, but are potentially alarming in the context of health- and healthcare-associated information. An interdisciplinary consortium of computer scientists and health infomaticians, such as the one mentioned above, was suggested as a suitable place to engage this
issue on a technical level. However, addressing system failures in the health system also has a deeply sociocultural component for which approaches that emphasize honesty and transparency with patients and the public were suggested. Education and outreach about this issue were identified as being crucial in preventing irreparable tears in the trust fabric necessary to support a digital learning health system. In this respect, participants noted the important contributions and potential of the Health IT Policy Committee’s Governance Working Group. Discussions also underscored the potential advantages of establishing a novel nongovernmental or public–private venture to foster the necessary governance capacity in this country and to work with similar efforts internationally.
Opportunities in the Next Stages of Meaningful Use
In line with these priorities, discussions often focused on the ongoing meaningful use requirement development process. Workshop participants discussed the “beyond meaningful use” issue as key to increasing the utility of digitally embedded clinical records in a learning health system. Specifically, since meaningful use is now such a well-established benchmark process, elements of particular importance to the development of a learning health system might not otherwise be addressed in the meaningful use process if they are not called out for explicit attention in the upcoming stages. Depicted in Box S-11 is a brief description of the meaningful use stages, the current expected focus of the requirements for stages 2 and 3, and bullets highlighting some key possibilities proposed by workshop participants.
Stage 2. Items that workshop participants felt were of particular importance in enhancing the impact that stage 2 of meaningful use could have on the progress of the digital learning health system cut across several dimensions. Flagged as especially key were actions to accelerate standards for semantic interoperability and exchange, as well as approaches for consistent identification of patients. In order to further the utility of EHRs in clinical research and population health, participants suggested core data elements for EHRs, and seamless access to information from immunization registries. Reflecting the extensive discussion on the opportunity for using the digital infrastructure to better engage patients in their health care, participants suggested the addition of lay-interpretable language for patient-accessible information, and incorporation of patient-generated data. Finally, discussions emphasized the need for clinical decision support to be seamlessly integrated into HIT systems to speed adoption.
Stage 3. Looking ahead to stage 3 of meaningful use, workshop participants suggested deepening the focus on requirements related to demonstrating
Meaningful Use and the Digital Learning Health System Infrastructure
STAGE 1: 2011–2012
Stage 1 of meaningful use established 14–15 (eligible hospitals or eligible professionals) required core functional components, focused on data capture and sharing, along with a menu set of 10 additional components, from which 5 are to be selected by the eligible hospitals or eligible professionals.
STAGE 2: 2013–2014
Stage 2 of meaningful use is under development by the HIT Policy Committee, including consideration of further focus on advanced clinical processes such as clinical decision support, disease management, patient access to health information, quality measurement, research, public health, and interoperability across IT systems. The following are items underscored in IOM discussions as being of particular and immediate importance to the impact of stage 2 enhancements on progress toward the Digital Infrastructure for the Learning Health System:
- Integration of semantic interoperability and exchange standards, including data provenance and context
- Elements fostering seamless integration of clinical decision support
- Use of lay-interpretable language for patient-accessible EHR information
- Incorporation of patient-generated data, including patient preferences
- Inclusion of core data elements that facilitate use of EHR data for clinical research.
- Strategy for seamless access to immunization history from immunization registries
- Strategy for consistent identification of patients
STAGE 3: 2015+
Stage 3 of meaningful use is expected to expand on requirements from stages 1 and 2, with more direct emphasis on improved patient outcomes through sharpened focus on quality, safety, efficiency, population health, and interoperability. Following are items, in addition to those noted above for stage 2, underscored in IOM discussions as being of particular and immediate importance to the impact of stage 3 enhancements on progress toward the Digital Infrastructure for the Learning Health System:
- Ability to access comprehensive, longitudinal patient record at point of care
- Incorporation of patient editing ability
- Demonstration of baseline semantic interoperability and exchange capacity among IT systems
- Integration of nonmedical, health-related information
- Seamless clinician–public health agency exchange on case-level information and alerts
semantic interoperability and exchange capacity among systems, the ability to access comprehensive patient records at the point of care, and seamless exchange of cases and alerts between clinicians and public health agencies. Participants also suggested strategies for including additional types of data—including nonmedical, health-related data—as well as providing patients with an annotated editing ability over their own records.
Stakeholder Responsibilities and Opportunities
Throughout each workshop, frequent reference was made to leadership responsibilities that fell naturally to individual stakeholders, or groups of stakeholders, to advance progress in the development of the digital infrastructure for the learning health system. In many cases, this involves leveraging ongoing efforts or building upon them with an orientation toward a continuous learning system. Summarized below are some of those most often noted.
Even though participants noted the decentralized manner in which localized innovation is likely to contribute to system progress, many of the central strategy elements and priority action targets discussed require strong leadership from federal agencies. Since a clear lead responsibility was given to ONC and the Secretary of HHS by the HITECH statute, many participants pointed to ONC as the natural leadership locus for activities needing coordination at the national level. Opportunities to build on the foundation laid by the HITECH requirements for work on standards, requirements, and certification criteria in meaningful use of EHRs include cooperation with other federal agencies in the development of a strategic plan for national HIT efforts; establishment of a governance mechanism for the NWHIN; accelerating, in cooperation with the National Institute for Standards and Technology, work on standards for exchange and interoperability; and work with FCC, FDA, and CMS to identify standards and reconcile regulations to facilitate wireless transmission of medical information. Participants noted that as the HITECH funds are used, the coordinating capacity of ONC will take on even greater importance, as coalitions will be needed to harmonize various key activities geared at developing the standards, policies, governance, and research projects necessary for effective progress toward a learning health system.
With respect to technical innovation, as the leading federal agency for funding computer science and engineering research, NSF was noted as a logical locus to work with ONC and NIH in the development of test beds for the rapid deployment and evaluation of innovative technological
approaches. This work would have the potential to transform the functionality and capacity of the digital health infrastructure, as well as to shepherd the establishment of collaborative vehicles for the ongoing partnerships between the HIT and computational science communities.
Similarly, it was noted that progress in the quality and knowledge generation dimensions of the digital platform will require leadership from federal health agencies. AHRQ, working with ONC, professional societies, and groups such as NQF and the National Committee for Quality Assurance, is a natural steward for initiatives that enhance the utility of the digital infrastructure for quality improvement and health services research.
The CDC’s focus on population health places it at the center of extending the scope of the digital infrastructure beyond health care. This carries implications for almost all elements of the system, but will be especially important for the support of public health processes and research as well as public engagement. To these ends, participants suggested development of templates and protocols for the integration of nonmedical population health and demographic information into the system.
As the nation’s largest healthcare financing organization, CMS currently serves as the principal vehicle for applying economic incentives and standards to accelerate application of the meaningful use requirements. Furthermore, much promise for future innovation in health IT to support a learning system resides in the CMMI which provides an opportunity for testing innovative approaches suggested by workshop participants. These approaches include test beds for ULS-associated programs and new approaches to integrating clinical decision support with care coordination and delivery models.
On the research front, both NIH and NSF have mandates and networks to develop and demonstrate methods of improving the functionality of the digital infrastructure for health research applications. NIH, VHA, DOD, FDA, and AHRQ all have active programs under way that can evolve into cooperative leadership efforts to expand the use of EHRs for research into the clinical effectiveness of health interventions.
To build support and engagement among patients and the general population, AHRQ, FDA, NIH, and ONC have each established links to patient communities that can serve as the building blocks for a collaborative initiative to better characterize and communicate the health and economic advantages of public involvement in a digital platform for health improvement.
Given this level of activity, and the number of central stakeholders, the importance of ONC’s coordination mandate was often underscored. Similarly emphasized was the need to cultivate strong counterpart capacity outside of government to partner in coordination and governance responsibilities.
State and Local Government Leadership
Given the regional emphasis of many of the ongoing efforts related to the digital learning health system—such as the establishment of regional health information exchanges—state and local governments and health departments have experience establishing governance structures and developing programs for engaging local stakeholders. As a result, participants noted, state and local bodies can function as resources and foundation stones for broader efforts. By collaborating with ONC, CMS, HRSA, and other federal initiatives, best practices and lessons learned can be leveraged from state and local efforts. Additionally, it was suggested that some of the more advanced local initiatives could serve as test beds for some of the innovative ULS-associated approaches suggested by participants.
Initiatives Outside Government
Outside of government, the entrepreneurial capacity of the commercial sector will certainly be a major driver of progress. Similarly, the full potential of the learning health system can only be achieved through the full engagement of patients and the public. Workshop discussants frequently underscored the roles of patient and clinician groups to facilitate dialogue between stakeholders and mediate public engagement. In particular, by using case studies to demonstrate the value of the digital infrastructure, participants felt these organizations could help develop the shared learning culture and trust necessary for the learning system to function. Many patient and clinician groups—such as the American College of Physicians, the American College of Cardiology, the Society of Thoracic Surgeons, and the National Partnership for Women and Families—are already involved in this type of work. Participants noted that these existing activities could be expanded to include issues of particular importance to a learning system.
Delivery systems, particularly those integrated across healthcare components, have been at the cutting edge of innovative EHR use, quality improvement, clinical data stewardship, patient engagement, quality initiatives, and distributed research efforts. Workshop conversations often pointed to these efforts, such as those at Kaiser Permanente and Geisinger Health System, suggesting that continued coordination between these delivery systems and relevant federal government agencies would be important in growing the digital health infrastructure.
As the stewards of the largest stores of clinical and transactional information outside of the federal government, insurers, payers, and product developers have an essential role to play in development of the digital infrastructure. Their use of transactional health data to assess utilization patterns, effectiveness, and efficiency is a foundational block on which
strategies for broader knowledge generation can build. Furthermore, companies such as UnitedHealthcare have begun engaging the public in the use of data in health. These efforts often were cited during discussions as crucial first steps in establishing a learning culture.
Research is a fundamental aspect of the learning health system. Consequently, participants noted the fundamental role researchers have in developing the infrastructure necessary for continuous knowledge generation and application. Formation of multidisciplinary research communities was often cited as a critical step in accelerating many of the strategies discussed. Funding for these communities was noted as a clear opportunity for collaboration between NSF and NIH. Additionally, discussions highlighted that much work remains to be done in order to maximize the knowledge generation capabilities of the digital infrastructure, and that clinical research and product development communities have an essential role in building this capacity.
As much of the progress to date is a result of initiatives from many independent organizations, their continued efforts as facilitators and innovators were noted as crucial to accelerating progress. Reference was often made to the importance of these organizations as the foundational elements for coordination and governance leadership from outside government.
Finally, and ultimately of paramount importance, is the global perspective. As highlighted during workshop discussions and presentations (see Chapter 8), meeting the goals of a learning health system will inevitably require drawing upon resources and leadership of similar efforts throughout the world. Some of this activity has begun in the limited arena of infectious disease surveillance and monitoring and offers a hint of the potential opportunities—and challenges—in developing a truly global clinical data utility for health progress.
Collectively, the discussions captured in this publication represent unprecedented promise for innovation and progress in health and health care. Yet, the discussions also underscored that without successful efforts to create the conditions necessary for seamless interoperability, to create the protocols for enhanced access and use of available information for knowledge generation, and to build the culture of engagement and support on behalf of the sort of information utility possible, the potential will go unmet. By thoroughly and candidly engaging in discussions on the vision, the current state of the system, the key priorities for future work, and the strategic elements for accelerating progress, participants have set in motion perspectives that can quicken the progress in building the digital infrastructure required for the continuously learning health system necessary—and possible—to ensure better health for all.
Blumenthal, D., and M. Tavenner. 2010. The “meaningful use” regulation for electronic health records. New England Journal of Medicine 363(6):501-504.
CHCF (California HealthCare Foundation). 2010. New national survey finds personal health records motivate consumers to improve their health. http://www.chcf.org/media/pressreleases/2010/new-national-survey-finds-personal-health-records-motivate-consumers-toimprove-their-health#ixzz12kT8FU00 (accessed October 18, 2010).
DesRoches, C. M., E. G. Campbell, S. R. Rao, K. Donelan, T. G. Ferris, A. Jha, R. Kaushal, D. E. Levy, S. Rosenbaum, A. E. Shields, and D. Blumenthal. 2008. Electronic health records in ambulatory care—a national survey of physicians. New England Journal of Medicine 359(1):50-60.
Fox, S. 2010. E-patients, cyberchondriacs, and why we should stop calling names. http://www.pewinternet.org/Commentary/2010/August/Epatients-Cyberchondriacs.aspx (accessed October 19, 2010).
IOM (Institute of Medicine). 2001. Crossing the quality chasm: A new health system for the 21st century. Washington, DC: National Academy Press.
______. 2007. The Learning Healthcare System: Workshop Summary. Washington, DC: The National Academies Press.
Jha, A. K., C. M. DesRoches, P. D. Kralovec, and M. S. Joshi. 2010. A progress report on electronic health records in U.S. hospitals. Health Affairs (Millwood) 29(10):1951-1957.
Northrop, L., P. H. Feiler, B. Pollak, and D. Pipitone. 2006. Ultra-large-scale systems: The software challenge of the future. Pittsburgh, PA: Software Engineering Institute, Carnegie Mellon University.
PCAST (President’s Council of Advisors on Science and Technology). 2010. Realizing the full potential of health information technology to improve healthcare for Americans: The path forward. http://www.whitehouse.gov/sites/default/files/microsites/ostp/pcast-healthit-report.pdf (accessed December 12, 2010).
Smith, A. 2010. Mobile access 2010. http://www.pewinternet.org/Reports/2010/Mobile-Access-2010.aspx (accessed October 19, 2010).