In response to a request from the Department of Health and Human Services, the Institute of Medicine convened a committee to produce a detailed plan to facilitate the development of data standards applicable to the collection, coding, and classification of patient safety information.
Americans should be able to count on receiving health care that is safe. To achieve this, a new health care delivery system is needed—a system that both prevents errors and learns from them when they occur. The development of such a system requires, first, a commitment by all stakeholders to a culture of safety and, second, improved information systems.
A national health information infrastructure is needed (1) to provide immediate access to complete patient information and decision support tools for clinicians and their patients and (2) to capture patient safety information as a by-product of care and use this information to design even safer delivery systems. Health data standards are both a critical and time-sensitive building block of the national health information infrastructure. The Department of Health and Human Services should be given the lead role in establishing and maintaining a public–private partnership for the promulgation of standards for data that support patient safety. The committee considered the status of
current standards-setting activities in three key areas—health data interchange, terminologies, and medical knowledge representation. For each of these areas, the committee reviewed the future work needed and recommended a work plan.
To achieve an acceptable standard of patient safety, the committee recommends that all health care settings establish comprehensive patient safety programs operated by trained personnel within a culture of safety and involving adverse event and near-miss detection and analysis. In addition, the federal government should pursue a robust applied research agenda on patient safety, focused on enhancing knowledge, developing tools, and disseminating results to maximize the impact of patient safety systems. Finally, the committee recommends that a standardized format and terminology be developed for the capture and reporting of data related to medical errors.
Since the release of the Institute of Medicine (IOM) report To Err Is Human: Building a Safer Health System (Institute of Medicine, 2000), national attention has been focused on the need to reduce medical errors. The health care community and the public at large have come to realize that the nation’s health care system is not as safe as it should be.
Every day, tens if not hundreds of thousands of errors occur in the U.S. health care system. Fortunately, most of these errors result not in serious harm but in near misses. A near miss is defined as an act of commission or omission that could have harmed the patient but did not do so as a result of chance (e.g., the patient received a contraindicated drug but did not experience an adverse drug reaction), prevention (e.g., a potentially lethal overdose was prescribed, but a nurse identified the error before administering the medication), or mitigation (e.g., a lethal drug overdose was administered but discovered early and countered with an antidote). Sadly, however, a small proportion of errors do result in adverse events—that is, they cause harm to patients—exacting a sizable toll in terms of injury, disability, and death.
To Err Is Human focuses primarily on errors that occur in hospitals and is based on the evidence available at the time that report was written. Newly released research indicates the existence of serious safety issues in other settings as well, including ambulatory settings and nursing homes (Gurwitz et al., 2000, 2003). In fact, because the number of outpatient encounters far exceeds the number of inpatient admissions, the consequences of medical errors in the former settings—and the opportunities to improve—may dwarf those in hospitals.
Earlier research on patient safety also focused on errors of commission (e.g., prescribing a medication that has a potentially fatal interaction with another drug the patient is taking). However, errors of omission (e.g., failing to prescribe a medication from which the patient would likely have benefited) may pose an even greater threat to health. On average, patients receive only about 55 percent of those services from which they would likely have benefited (McGlynn et al., 2003).
It is not possible to quantify the full magnitude of the safety challenge with certainty. The health care sector does not routinely identify and collect information on errors. Experts have challenged the estimates of patient harm attributable to errors, as well as the methodologies used to derive them (Brennan, 2000; Hayward and Hofer, 2001; McDonald et al., 2000; Sox and Woloshin, 2000). As substantial evidence about adverse events continues to accumulate in the United States and other countries (Vincent et al., 2001; Wilson et al., 1995), however, there is no doubt that their occurrence is a serious matter warranting attention. The risks to public safety—and the opportunities for large-scale improvements—are sizable.
As concerns about patient safety have grown, the health care sector has looked to other industries that have confronted similar challenges, in particular the airline industry. This industry learned long ago that information and clear communication are critical to the safe navigation of an airplane. To perform their jobs well and guide their planes safely to their destinations, pilots must communicate with the air traffic controller concerning their destinations and current circumstances (e.g., mechanical or other problems), their flight plans, and environmental factors (e.g., weather conditions) that could necessitate a change in course. Information must also pass seamlessly from one controller to another to ensure a safe and smooth journey for planes flying long distances; provide notification of airport delays or closures due to weather conditions; and enable rapid alert and response to an extenuating circumstance, such as a terrorist attack.
Information is as critical to the provision of safe health care—care that is free of errors of both commission and omission—as it is to the safe operation of aircraft. To develop a treatment plan, a doctor must have access to complete patient information (e.g., diagnoses, medications, current test results, and available social supports) and to the most current science base. The doctor and the patient must also be aware of other environmental factors that affect the ability to implement a treatment plan, such as the availability of hospital beds, current waiting times to obtain specific services, and insurance requirements for prior authorization or use of specific providers. Doctors and nurses armed with information on infectious diseases (e.g., in-
fluenza, West Nile virus) can also better counsel their patients about preventive steps that may be beneficial. For patients with chronic conditions, it is critical that information flow freely among all authorized members of the care team (e.g., primary care providers, specialists, pharmacists, home health aides, patients, and lay caregivers).
On the other hand, a salient difference between the work environment of airline pilots and that of clinicians and patients is the level and sophistication of the information technology infrastructure that supports their work. Pilots have immediate access to the information they need to make informed decisions. In health care today, no such information technology infrastructure exists. Only a fraction of hospitals have implemented a comprehensive electronic health record (EHR) system,1 although many have made progress in certain areas, such as computerized reporting of laboratory results (Brailer, 2003). Rates of adoption of EHR systems are higher in ambulatory settings—probably about 5–10 percent of physician offices—but there is much variability in their content and functionality. Also, only a handful of communities have established a community-wide, secure Internet-based platform to facilitate access to clinical information by multiple providers, not just those within a given institution, such as a hospital or group practice (Institute of Medicine, 2002a).
Better management of health information is a prerequisite to achieving patient safety as a standard of care.
The airline industry also learned that for every tragic accident there are many near misses and that much can be learned from analysis of these events. Aviation reporting systems are in place to capture detailed information on near misses (e.g., time, place, individuals involved, nature of event, and circumstances that allowed harm to be averted). Accidents are meticulously documented and investigated by the National Transportation Safety Board.
In recent years, patient safety reporting systems have emerged in the health care arena. Many hospitals now routinely capture information on errors, both near misses and adverse events, as a part of their internal safety
improvement programs. The Joint Commission on Accreditation of Healthcare Organizations (JCAHO) requires hospitals to conduct root-cause analyses of adverse events as part of its accreditation program and encourages reporting of these events to JCAHO. Moreover, about one-half of states have reporting systems that focus on adverse events (Rosenthal and Booth, 2003). Various federal agencies maintain reporting systems as well, including those pertaining to drugs and medical devices, hospital-acquired infections, and blood products. There are also many examples of voluntary reporting systems in the private sector, including those for medication errors and adverse events occurring in hospitals.
The usefulness of these patient safety reporting systems, however, has been limited. As a result of the paucity of EHR systems, most patient safety reports cannot be generated automatically as a by-product of the patient care process. Nor can the lessons learned through analysis of patient safety reports easily be transferred back to the point of care. Without EHRs, reporting systems typically rely on special data collection mechanisms (both human- and computer-based), making reporting a cumbersome, costly, and sporadic exercise. The data collected in these systems are neither complete nor standardized, making it difficult to aggregate the data or identify trends or patterns. Liability concerns also impede participation in many reporting systems.
Patient safety is the prevention of harm to patients.
The development of an information technology infrastructure is essential to improve the safety of health care (Institute of Medicine, 2001). Computer-based reminder systems for patients and clinicians can improve compliance with preventive service protocols (Balas et al., 2000). The availability of complete patient health information at the point of care delivery, along with clinical decision support systems (e.g., for medication order entry), can prevent many errors from occurring (Bates et al., 1998, 1999; Evans et al., 1998). Computer-assisted diagnosis and chronic care management programs can improve clinical decision making and adherence to clinical guidelines (Durieux et al., 2000; Evans et al., 1998).
The committee strongly believes that patient safety is indistinguishable from the delivery of quality care. A new delivery system must be built to achieve substantial improvements in patient safety—a system that is capable of preventing errors from occurring in the first place, while at the same time incorporating lessons learned from any errors that do occur. To achieve such
a system capable of providing care that is safe will require a culture of safety and the active participation of all health care professionals, organizations, and patients themselves. A critical component of this new health care delivery system will be the development of an information technology infrastructure.
Recommendation 1. Americans expect and deserve safe care. Improved information and data systems are needed to support efforts to make patient safety a standard of care in hospitals, in doctors’ offices, in nursing homes, and in every other health care setting. All health care organizations should establish comprehensive patient safety systems that:
Provide immediate access to complete patient information and decision support tools (e.g., alerts, reminders) for clinicians and their patients.
Capture information on patient safety—including both adverse events and near misses—as a by-product of care, and use this information to design even safer care delivery systems.
To support the objectives of care delivery that is free of errors and the implementation of robust safety reporting systems, a broad range of patient data will be needed, including demographic information, signs and symptoms, medications, test results, diagnoses, therapies, and outcomes. In addition, to learn from near misses and adverse events, the system must capture such data as the individuals involved, where and when the event occurred, what happened, the likely severity of avoided or actual outcomes, contributing factors, and recovery procedures, as well as reporters’ narratives that will reveal the underlying system failure.
ELEMENTS OF A HEALTH INFORMATION INFRASTRUCTURE
As shortcomings in the current health care system have become increasingly apparent, the National Committee on Vital and Health Statistics and numerous expert panels have called for the establishment of a national health information infrastructure to meet many of the nation’s needs for safe and efficient care delivery, public health, homeland security, and health research (Institute of Medicine, 2001; National Committee on Vital and Health Statistics, 2001; President’s Information Technology Advisory Committee, 2001). It will not be enough for individual providers making independent decisions to invest in information technology because patients often receive
services from many different providers and in a variety of settings within and across communities. Components of a national health information infrastructure include EHR systems with decision support, a secure platform for the exchange of patient information across health care settings, and data standards to make that information understandable to all users. A partnership between the public and private sectors will be needed to build this infrastructure (Department of Health and Human Services, 2003a; Institute of Medicine, 2002a; Markle Foundation, 2003).
A previous IOM committee has spoken to the need for strong federal leadership and financial support for a national health information infrastructure (Institute of Medicine, 2002b). Private-sector investments will account for a good deal of the capital required to build this infrastructure, but the federal government also has an important role to play in providing financial support. To achieve the greatest gains, federal financial support should be targeted to three areas. First, federal financial investment should support the development of critical building blocks of the national health information infrastructure that are unlikely to receive adequate support through investment by private-sector stakeholders, including the establishment of a secure platform for the exchange of data across all providers and, as discussed below, maintenance of a process for the ongoing promulgation of national data standards. Second, the federal government should provide financial incentives to stimulate private-sector investments in EHR systems; this might be done through revolving loans, differential payments to providers with certain information technology capabilities, or other means. Third, federal government funding of safety net providers will be necessary to support their transition to a safer health care delivery system.
The Department of Health and Human Services (DHHS) recently commissioned the IOM to identify the key capabilities that an EHR system should possess to support patient safety and quality of care. This committee responded to that request in a letter report entitled Key Capabilities of an Electronic Health Record System (Institute of Medicine, 2003a) (see Appendix E). In structuring financial incentives, both public and private purchasers should consider linking provider incentives to the acquisition of EHRs that possess these important capabilities. The letter report also provides a framework that should prove useful to accreditation organizations in establishing standards for EHR systems, as well as to providers in selecting and venders in designing such systems.
National leadership will also be needed to establish and maintain standards for the collection, exchange, and reporting of data to support patient safety. Data standards are both a critical and time-sensitive building block of
the proposed national health information infrastructure. In the absence of national data standards, health care organizations will likely be slower to invest in information technology, and the systems that are built will be inadequate to make patient safety a standard of care.
Recommendation 2. A national health information infrastructure—a foundation of systems, technology, applications, standards, and policies—is required to make patient safety a standard of care.
The federal government should facilitate deployment of the national health information infrastructure through the provision of targeted financial support and the ongoing promulgation and maintenance of standards for data that support patient safety.
Health care providers should invest in electronic health record systems that possess the key capabilities necessary to provide safe and effective care and to enable the continuous redesign of care processes to improve patient safety.
Although the focus of this report is on patient safety, it is important to note that the proposed national health information infrastructure will yield many other benefits in terms of new opportunities for access to care, care delivery, public health, homeland security, and clinical and health services research. Through the use of telemedicine, critically ill patients in small rural hospitals will be able to benefit from round-the-clock remote monitoring by physicians with advanced training in intensive care. Like air traffic controllers, public health officials and clinicians with appropriate information and communication supports will be able to detect earlier and respond more rapidly to infectious disease outbreaks. Enhanced communication and information technologies will allow the health system to make a significant leap forward toward safer care.
A PUBLIC–PRIVATE PARTNERSHIP FOR SETTING STANDARDS
Efforts of both the public and private sectors to invest in information technology are hampered by the lack of national data standards for the collection, coding, classification, and exchange of clinical and administrative data. The establishment of such national data standards must be an ongoing process, including updates to reflect both advances in clinical knowledge and changes in safety and quality reporting requirements.
Establishing and maintaining data standards is integrally linked to the
advancement and diffusion of clinical knowledge. The discovery of new knowledge leads to the redefinition of what constitutes best practices in a specific clinical area. Overlooking or failing to adhere to best practices is an important source of errors of omission that lead to morbidity and mortality among patients. In the early 1980s, for example, new scientific evidence became available indicating that medications known as beta-blockers administered to patients at the time of a heart attack greatly reduce the likelihood of a subsequent heart attack (Beta-Blocker Heart Attack Trial, 1982). To be applied speedily and consistently in practice, such new evidence must be translated into a care guideline (e.g., absent contraindications, patients experiencing a heart attack should be prescribed beta-blockers). Hospitals, physicians, and other providers must modify their care processes to be consistent with the new best practice (e.g., the patient’s attending physician is responsible for prescribing a beta-blocker to the patient at the time of the heart attack). Information systems must be modified to capture information on the new practice (e.g., the pharmacy system must add this new drug to the formulary), and computerized decision support systems must be modified to issue an alert to the clinician and patient if the patient’s record does not include entries substantiating that beta-blockers were prescribed at the time of the heart attack, if appropriate.
Unfortunately, the current health care delivery system lacks well-defined processes for translating new knowledge into practice. Not surprisingly, then, a 1997 study showed that only 21 percent of eligible elderly patients suffering a heart attack had received beta-blockers, and there was a 75 percent higher mortality rate among those who did not receive the treatment than among those who did (Soumerai et al., 1997). Similar examples can be found in virtually every area of clinical practice (Institute of Medicine, 2001; McGlynn et al., 2003). Overall, the toll in terms of lost lives, pain and suffering, and wasted resources is staggering.
As a complement to the present study, DHHS asked the IOM to identify a limited number of clinical areas that might serve as a starting point for public- and private-sector efforts to improve care delivery. In fall 2002, the IOM released the report Priority Areas for National Action: Transforming Health Care Quality (Institute of Medicine, 2003b) identifying 20 areas—consisting primarily of leading chronic conditions—that account for a sizable proportion of health care services.
Through the efforts of the Agency for Healthcare Research and Quality (AHRQ) progress is being made on translating knowledge into practice in selected clinical areas, including the 20 priority areas identified by the IOM. As of October 2002, AHRQ had provided support to 13 evidence-based
practice centers to develop evidence reports and technology assessments and to work in partnership with other groups to develop practice guidelines and implementation tools (Agency for Healthcare Research and Quality, 2003). In September 2000, AHRQ also initiated the Integrated Delivery System Research Network, which encompasses nine partnership arrangements, each linking health care organizations, research institutions, and managed care organizations for the conduct of applied health services research and the dissemination of findings (Agency for Healthcare Research and Quality, 2002). Nearly all of the 20 priority areas identified by the IOM are being addressed in this ongoing work.
The health care sector also lacks standardized measurement and reporting mechanisms that can be used for routine monitoring of the extent to which health care is safe and effective. In designing and building information technology systems, it is helpful to know in advance the reporting specifications that must be satisfied. As noted earlier, there are many safety and quality measurement and improvement efforts sponsored by health care providers, public and private purchasers, federal and state agencies, and accreditors. Some focus on near misses or adverse events, while others assess compliance with best practices through medical care process and outcome measures. Some noteworthy efforts have been made to encourage standardization of reporting requirements. In late 2003, DHHS is expected to release the first National Healthcare Quality Report, in which an attempt has been made to address the IOM’s 20 priority areas. In the future, this report will likely extend to the state and community levels. Likewise, the Quality Interagency Coodinating Task Force, an interagency government committee, has made some progress toward establishing standardized safety and quality measures and tools, some of which have been incorporated into multiple government health care programs. And the National Quality Forum, a public–private partnership organization, has established standardized reporting requirements for various health care settings (e.g., nursing homes, hospitals) and for certain safety-related events, called serious reportable events (National Quality Forum, 2002). Much work remains to be done, however. The data requirements for clinical guidelines and for safety and quality reporting must feed into the process used to develop data standards for EHR systems if those systems are to serve as the primary source of information and decision support for providers seeking to follow best-practice guidelines and respond to patient safety reporting requirements.
The National Committee on Vital and Health Statistics (NCVHS), a public–private advisory committee established to provide advice to DHHS and Congress on national health information policy, has for many years rec-
ommended that the federal government assume a more active role in establishing national data standards (National Committee on Vital and Health Statistics, 2000). In 1996, Congress passed the Health Insurance Portability and Accountability Act (HIPAA, Public Law 104-191), which mandated standardization of administrative and financial transactions. In 2001, the Consolidated Health Informatics (CHI) initiative, an interagency effort, was established as part of the Office of Management and Budget’s eGOV initiative to streamline and consolidate government programs among like sectors (Office of Management and Budget, 2003). DHHS was designated the managing partner for the CHI initiative, with the Centers for Medicare and Medicaid Services taking the lead. The CHI initiative played a pivotal role in the recent decision by the federal government that programs of DHHS, the Veterans Administration, and the Department of Defense would incorporate certain data standards and terminologies (Department of Health and Human Services, 2003b).
The CHI initiative, although off to a very promising start, lacks a clear mandate to establish standards. In addition, the future of the initiative once initial standards and gaps have been identified is unclear. The initiative would also benefit from closer collaboration with NCVHS to ensure the active participation of private-sector stakeholders.
Recommendation 3. Congress should provide clear direction, enabling authority, and financial support for the establishment of national standards for data that support patient safety. Various government agencies will need to assume major new responsibilities, and additional support will be required. Specifically:
The Department of Health and Human Services (DHHS) should be given the lead role in establishing and maintaining a public–private partnership for the promulgation of standards for data that support patient safety.
The Consolidated Health Informatics (CHI) initiative, in collaboration with the National Committee on Vital and Health Statistics (NCVHS), should identify data standards appropriate for national adoption and gaps in existing standards that need to be addressed. The membership of NCVHS should continue to be broad and diverse, with adequate representation of all stakeholders, including consumers, state governments, professional groups, and standards-setting bodies.
The Agency for Healthcare Research and Quality (AHRQ) in collaboration with the National Library of Medicine and others should
(1) provide administrative and technical support for the CHI and NCVHS efforts; (2) ensure the development of implementation guides, certification procedures, and conformance testing for all data standards; (3) provide financial support and oversight for developmental activities to fill gaps in data standards; and (4) coordinate activities and maintain a clearinghouse of information in support of national data standards and their implementation to improve patient safety.
The National Library of Medicine should be designated as the responsible entity for distributing all national clinical terminologies that relate to patient safety and for ensuring the quality of terminology mappings.
Without federal leadership in the establishment of standards for data that support patient safety, information technology systems built over the coming decades will be inadequate to support the delivery of safe and effective care. The time to act is now.
Given the sizable purchasing power (over 40 percent of health care expenditures) and regulatory authority of the federal government, the incorporation of data standards into government programs is one approach to establishing national standards. After providing a reasonable time period for health care organizations to comply with national standards identified by CHI, the major government health care programs, including those operated or sponsored by DHHS, the Veterans Administration, and the Department of Defense, should immediately incorporate these data standards into their contractual and regulatory requirements (e.g., Medicare conditions for participation).
AN ACTION PLAN FOR SETTING DATA STANDARDS2
The standards-setting process, like any other major undertaking, needs a focus and specific objectives. This committee considered the need for standards and the status of current standards-setting activities in three key areas:
Data interchange formats—standard formats for electronically encoding the data elements (including sequencing and error handling). Interchange standards can also include document architectures for structuring
data elements as they are exchanged and information models that define the relationships among data elements in a message.
Terminologies—the medical terms and concepts used to describe, classify, and code the data elements, and data expression languages and syntax that describe the relationships among the terms/concepts.
Knowledge representation—standard methods for electronically representing medical literature, clinical guidelines, and the like for decision support (Hammond, 2002).
Following is a discussion of future work needed in each of these areas and a recommended work plan.
Data Interchange Standards
Because health care data are distributed across several locations (databases), standards for data interchange (i.e., rules for transmitting data from one database to another) are necessary. For messaging standards, a number of mature standards cover the required domains:
Administrative data (the X12 standard of the Accrediting Standards Committee, Subcommittee on Insurance, Working Group 12)
Clinical data (Health Level 7 [HL7])
Medical images (Digital Imaging and Communications in Medicine [DICOM])
Prescription data (National Council for Prescription Drug Programs [NCPDP] Script)
Medical device data (Institute for Electrical and Electronics Engineers [IEEE] standard 1073).
These standards3 were recently endorsed by the Secretary of Health and Human Services. However, there is an urgent need to accelerate the development of the next version of HL7 (version 3.0) to support increased interoperability of systems and comparability of clinical data, as well as patient safety initiatives. In addition to standards, implementation guides, conformance testing, and certification procedures must be developed to ensure consistent application of the standards in commercial systems.
Clinical information will continue to appear in textual clinical notes for many years to come. A document architecture standard is needed to enable the interchange of clinical notes and to facilitate the extraction of information using natural language processing techniques. DHHS and AHRQ should support the development of the HL7 Clinical Document Architecture for this purpose. An intuitive and efficient user interface is also an important part of clinical information systems (Shortliffe et al., 2001). User interface tools to facilitate data acquisition are still in the early stages of development, and a number of research projects are under way to resolve technological constraints on the widespread implementation of clinical information systems. Much is being learned from the ubiquity of Web interfaces, and continued research is necessary to design user interfaces that incorporate human factors into the engineering of applications.
When exchanging patient-specific data for clinical and patient-safety reasons, it is imperative that the data be linked to the correct patient accurately and reliably through a unique health identifier (UHI). Without a national UHI, fragmentation of patient data can lead to medical errors and adverse events. Although a UHI was mandated by HIPAA, Congress placed a hold on further action until privacy protection was enacted. Now that privacy and security rules to protect health data have been established under the provisions of HIPAA,4 Congress should authorize DHHS to take immediate steps to identify options for implementing a UHI system. Consideration should be given to implementing a voluntary UHI system in which patients can elect to participate.
If health professionals are to be able to send and receive data in an understandable and usable manner, both the sender and the receiver must have common clinical terminologies for describing, classifying, and coding medical terms and concepts. Use of standardized clinical terminologies facilitates electronic data collection at the point of care; retrieval of relevant
data, information, and knowledge; and reuse of data for multiple purposes (e.g., disease surveillance, clinical decision support, patient safety reporting).
No single terminology has the depth and breadth to represent the broad spectrum of medical knowledge; thus a core group of well-integrated, nonredundant clinical terminologies will be needed to serve as the backbone of clinical information and patient safety systems. Efforts are now under way within the National Library of Medicine and other key government organizations to evaluate existing terminologies and identify those that should be included in the core set. Patient safety is an important area in which significant gaps in terminology for concept representation exist and for which a new terminology and classification system needs to be developed. The new terminology should be fully integrated with the core set and made publicly available for widespread dissemination and use. The National Library of Medicine should be responsible for dissemination, mapping, and updating of the core terminology standards.
As noted above, to support patient safety, ongoing syntheses of the clinical literature should be conducted to determine best practices for clinical management in the 20 priority areas identified by the IOM. The National Institutes of Health and many private-sector academic and research centers play critical roles in the ongoing generation of clinical knowledge. Various professional associations and AHRQ, working through evidence-based practice centers, contribute to the development of practice guidelines and the identification of best practices. This information is critical to the development of decision support tools that can assist clinicians and patients in making evidence-based decisions. Standards are needed for the representation of clinical guidelines and the implementation of automated triggers.
A Work Plan
Accelerating the development and adoption of standards for data to support patient safety will require a concerted and sustained effort in both the public and private sectors. Leadership and support from the federal government will be necessary.
Recommendation 4. The lack of comprehensive standards for data to support patient safety impedes private-sector investment in informa-
tion technology and other efforts to improve patient safety. The federal government should accelerate the adoption of standards for such data by pursuing the following efforts:
Clinical data interchange standards. The federal government should set an aggressive agenda for the establishment of standards for the interchange of clinical data to support patient safety. Federal financial support should be provided to accomplish this agenda.
After ample time for provider compliance, federal government health care programs should incorporate into their contractual and regulatory requirements standards already approved by the secretaries of DHHS, the Veterans Administration, and the Department of Defense (i.e., the HL7 version 2.x series for clinical data messaging, DICOM for medical imaging, IEEE 1073 for medical devices, LOINC for laboratory test results, and NCPDP Script for prescription data).
AHRQ should provide support for (1) accelerated completion (within 2 years) of HL7 version 3.0; (2) specifications for the HL7 Clinical Document Architecture and implementation guides; and (3) analysis of alternative methods for addressing the need to support patient safety by instituting a unique health identifier for individuals, such as implementation of a voluntary unique health identifier program.
Clinical terminologies. The federal government should move expeditiously to identify a core set of well-integrated, nonredundant clinical terminologies for clinical care, quality improvement, and patient safety reporting. Revisions, extensions, and additions to the codes should be compatible with, yet go beyond, the federal government’s initiative to integrate all federal reporting systems.
AHRQ should undertake a study of the core terminologies, supplemental terminologies, and standards mandated by the Health Insurance Portability and Accountability Act to identify areas of overlap and gaps in the terminologies to address patient safety requirements. The study should begin by convening domain experts to develop a process for ensuring comprehensive coverage of the terminologies for the 20 IOM priority areas.
The National Library of Medicine should provide support for the accelerated completion of RxNORM5 for clinical drugs. The National Library of Medicine also should develop high-quality
mappings among the core terminologies and supplemental terminologies identified by the CHI and NCVHS.
Knowledge representation. The federal government should provide support for the accelerated development of knowledge representation standards to facilitate effective use of decision support in clinical information systems.
The National Library of Medicine should provide support for the development of standards for evidence-based knowledge representation.
AHRQ, in collaboration with the National Institutes of Health, the Food and Drug Administration, and other agencies, should provide support for the development of a generic guideline representation model for use in representing clinical guidelines in a computer-executable format that can be employed in decision support tools.
PATIENT SAFETY SYSTEMS IN HEALTH CARE SETTINGS
Since patient safety is an integral part of the delivery of quality care, achieving an acceptable standard of patient safety requires that all health care settings develop comprehensive patient safety systems, including both a culture of safety and organizational supports for safety processes. A key aspect of a patient safety system is a culture that encourages clinicians, patients, and others to be vigilant in (1) identifying potential or actual errors, (2) taking appropriate steps to prevent and mitigate harm, and (3) disclosing appropriate information on errors that do occur to facilitate learning and the redesign of care processes. As noted above, safe care settings are ones that have an adequate information infrastructure to provide clinicians and patients with immediate access to health information. But other organizational supports are needed as well, including trained professionals with expertise in safety and well-designed reporting systems for near misses and adverse events.
The establishment of patient safety systems is in a relatively early stage of development in most health care settings. Some aspects of a patient safety system can be found in all, or nearly all, institutional settings, but this is not the case for ambulatory settings, where the majority of health care is provided.
In general, patient safety systems have been evolving along three dimensions: (1) an expansion of the types of events that are analyzed to include both adverse events and near misses; (2) increased use of automated surveillance, as opposed to relying on clinicians or patients, to identify and report
cases; and (3) increased attention to the application of knowledge gleaned from reporting systems to the design of systems that can prevent errors. Traditionally, patient safety systems have detected events through individual reports (e.g., a clinician reports an adverse event to a hospital risk manager), document review (e.g., retrospective review of patient records and death certificates), or monitoring of patient progress. In the future, most events will likely be identified through automated surveillance of clinical data (e.g., identifying patients with unusual laboratory results) as more and more of the important components of the patient record become computer based. Automated surveillance, sometimes called data-driven triggers, offers many advantages, including (1) more immediate identification of events when there may still be an opportunity to mitigate patient harm, (2) identification of larger numbers of adverse events than is possible with methods that rely on individual reports or sampling techniques, and (3) a less labor-intensive approach than individual case finding.
To date, most patient safety efforts have focused on the detection and analysis of events, especially adverse events. Adverse events are certainly important, but as noted earlier, they occur infrequently and, by definition, after patients have been injured. Less attention has been focused on the detection and analysis of near misses, and this relative neglect represents a missed opportunity. Experts believe that for each serious adverse event there are probably dozens of near misses, which might best be described as warning signs. Because near misses occur more frequently, monitoring and analysis of these events provide quantitative insight into the distribution of factors that contribute to the occurrence of and recovery from errors (Billings, 1999). The monitoring of near misses may also contribute to a higher level of risk awareness in the working environment. This increased awareness can lead to proactive efforts and system changes that can increase the probability of preventing errors from occurring. Finally, multiple failures often contribute to a single adverse event, and early detection of the first such failure provides an opportunity to intervene and stop what could become a chain of failures leading up to a serious adverse event. However, none of these patient safety reporting systems for detecting and analyzing adverse events and near misses can function effectively in the absence of universally adopted standards for data to support patient safety.
Perhaps the most important dimension of the evolution of patient safety programs among the three cited at the beginning of this section is an increased emphasis on prevention. Progress along this dimension is closely related to progress along the other two. As reporting and analysis move upstream from adverse events to near misses, more knowledge is discovered
about high-risk conditions and patients, thus opening the door to preventive interventions. The committee believes that continued evolution along these three dimensions is critical and that steps should be taken to accelerate the pace of this evolution. All health care settings, not just hospitals, nursing homes, and large group practices, should have mature patient safety systems and cultures.
Recommendation 5. All health care settings should establish comprehensive patient safety programs operated by trained personnel within a culture of safety. These programs should encompass (1) case finding—identifying system failures, (2) analysis—understanding the factors that contribute to system failures, and (3) system redesign—making improvements in care processes to prevent errors in the future. Patient safety programs should invite the participation of patients and their families and be responsive to their inquiries.
Efforts should also be made to develop a rich portfolio of knowledge and tools that will be useful to all health care settings seeking to establish comprehensive patient safety systems. Research in this area should focus on the development of the full range of data-driven trigger systems for the detection and prevention of adverse events. Additional research is also needed to assist health care settings in establishing effective reporting systems for near misses. As noted above, the health care sector has far less experience with such systems than with those focusing on adverse events. The high volume and diversity of reports submitted to near-miss systems pose certain challenges.
Recommendation 6. The federal government should pursue a robust applied research agenda on patient safety, focused on enhancing knowledge, developing tools, and disseminating results to maximize the impact of patient safety systems. AHRQ should play a lead role in coordinating this research agenda among federal agencies (e.g., the National Library of Medicine) and the private sector. The research agenda should include the following:
High-risk patients—Identify patients at risk for medication errors, nosocomial infections, falls, and other high-risk events.
Near-miss incidents—Test the causal continuum assumption (that near misses and adverse events are causally related), develop and test a recovery taxonomy, and extend the current
individual human error/recovery models to team-based errors and recoveries.
Hazard analysis—Assess the validity and efficiency of integrating retrospective techniques (e.g., incident analysis) with prospective techniques.
High-yield activities—Study the cost/benefit of various approaches to patient safety, including analysis of reporting systems for near misses and adverse events.
Patient roles—Study the role of patients in the prevention, early detection, and mitigation of harm due to errors.
Early detection capabilities—Develop and evaluate various methods for employing data-driven triggers to detect adverse drug events, nosocomial infections, and other high-risk events (e.g., patient falls, decubitus ulcers, complications of blood product transfusions).
Prevention capabilities—Develop and evaluate point-of-care decision support to prevent errors of omission or commission.
Data mining techniques—Identify and develop data mining techniques to enhance learning from regional and national patient safety databases. Apply natural language processing techniques to facilitate the extraction of patient safety–related concepts from text documents and incident reports.
Dissemination—Deploy knowledge and tools to clinicians and patients.
PATIENT SAFETY REPORTING
As concerns about safety and quality have grown, so, too, have reporting requirements. Performance information can serve a range of purposes. At one end of the spectrum are applications used by public-sector legal and regulatory bodies, such as professional and institutional licensure and legal liability, that are intended to hold health care professionals and organizations accountable. At the other end of the spectrum are applications that focus on learning—both organizational and professional. The feedback of performance data to clinicians for continuing education purposes falls into this category, as does the redesign of care processes by health care organizations based on analysis of data collected by reporting systems for near misses and adverse events. Somewhere between these two ends of the spectrum are applications intended to encourage health care providers to strive for excellence by rewarding those who achieve the highest levels of performance with larger payments and greater demand for their services.
EHR systems should be capable of supporting the full range of applications outlined above. Ideally, performance reports, whether for external accountability or internal quality improvement purposes, should be generated automatically as a by-product of the EHR system. Achieving this objective will require a great deal of standardization of both the information reported and the patient and other data captured as part of the patient care process.
Some progress has been made in the standardization of certain types of performance reporting requirements. For example, there are standardized care process measures (e.g., immunization rates, proportion of diabetics who received an annual eye exam) for health plans, hospitals, and nursing homes (Department of Health and Human Services, 2002a, b; National Committee for Quality Assurance, 2003), and efforts are under way to encourage national adoption of various standardized measurement sets (Kizer, 2001). Far less attention has been focused on the standardization of reporting requirements applicable to reporting systems for near misses and adverse events. These types of reporting systems capture detailed information on specific events. Although sometimes used to produce error rates (e.g., adverse drug events per 1,000 admissions), such reporting systems focus more on the conduct of root-cause analyses to determine the factors that contributed to the event and identify ways of redesigning the care process to reduce the likelihood that similar events will occur in the future.
Most health care organizations must comply with a multitude of reporting requirements for errors. Many public and private purchasers, state governments, and private accrediting and certifying organizations require or encourage the reporting of errors. In addition, many health care organizations have their own internal reporting systems that play an integral role in the organization’s quality improvement programs. For the most part, each reporting system determines what types of events are reported, and many also have their own terminology to represent information. As a result, there is little if any ability to share and compare data, and the reporting burden on health care organizations is sizable.
The development of a standardized format and terminology for the capture and reporting of data related to patient safety events (i.e., adverse events and near misses) would improve the usefulness of the data and ease the reporting burden considerably. The standardized format should use the HL7 Clinical Document Architecture and include the reporter’s narrative; who was involved; what happened, where, and when; risk assessment of severity and probability of recurrence; preventability; causal analysis; recovery factors; and corrective actions if an adverse event. Standard taxonomies for the domain areas of the report format should be developed. Taxonomies for the
sets of contextual variables (who, what, where, outcome, etc.) should be tailored to each domain. The widely used Eindhoven Classification Model—Medical Version should be used as a standard taxonomy to classify root causes identified through analysis of near misses and adverse events. All new terms should be incorporated into the key reference terminology (Systemized Nomenclature of Human and Veterinary Medicine Clinical Terms [SNOMED CT]) of the NCVHS core terminology group, with mappings to higher-level classifications in supplemental terminologies, such as the International Classification of Diseases (ICD) 9/10 CM E-codes. The National Library of Medicine should be funded to maintain and distribute the patient safety taxonomies. Also, in light of the recently established patient safety initiative of the World Health Organization, additional work on the ICD 9/ 10 CM E-codes should be undertaken to enhance their capacity for representing adverse events and to facilitate international comparisons (World Health Organization, 2002).
An earlier IOM committee recommended that AHRQ establish a national patient safety reporting database containing standardized, deidentified patient data drawn from various public- and private-sector reporting systems (Institute of Medicine, 2001). Before acting on that recommendation, AHRQ will need to develop an event taxonomy and common report format for submission of data to the national patient safety database.
Recommendation 7. AHRQ should develop an event taxonomy and common report format for submission of data to the national patient safety database. Specifically:
The event taxonomy should address near misses and adverse events, cover errors of both omission and commission, allow for the designation of primary and secondary event types for cases in which more than one factor precipitated the adverse event, and be incorporated into SNOMED CT.
The standardized report format should include the following:
A standardized minimum set of data elements.
Data necessary to calculate a risk assessment index for determining prospectively the probability of an event and its severity.
A free-text narrative of the event.
Data necessary to support use of the Eindhoven Classification Model—Medical Version for classifying root causes, including expansions for (1) recovery factors associated with near-miss events, (2) corrective actions taken to recover from adverse events, and (3) patient outcome/functional status as a result of those corrective actions.
A free-text section for lessons learned as a result of the event.
Clinical documentation of the patient context.
The taxonomy and report format should be used by the federal reporting system integration project in the areas for basic domain, event type, risk assessment, and causal analysis but should provide for more extensive support for patient safety research and analysis (Department of Health and Human Services, 2002c).
The event taxonomy and standardized report format are intended to serve as a framework for federal, state, and private-sector reporting systems. AHRQ should also develop tools and guidelines to assist public- and private-sector reporting programs in implementing the common report format and data standards. Furthermore, the development of external data auditing criteria would provide assurance to all stakeholders that data used for reporting are valid and reliable.
THE JOURNEY AHEAD
The committee has laid out an ambitious agenda that has the potential to produce dramatic improvements in patient safety. This agenda is likely to yield considerable benefits in many other areas as well, including public health, homeland security, clinical and health services research, and health professions education.
At the heart of the agenda is the development of a national health information infrastructure, including EHR systems that adhere to national standards for data supporting patient safety in all health care settings. Although the committee recognizes that carrying out this agenda will require a sizable up-front capital investment, we believe its creation is essential not only to patient safety but also to the health of the American people more generally. The committee believes that establishing this information technology infrastructure should be the highest priority for all health care stakeholders.
This is a journey that will take a decade to accomplish and in which the federal government, working in partnership with the private sector, has a critical role to play. The federal government should act immediately to establish the national data standards called for in this report. Although modest financial resources will be required, the committee believes the return on this investment will be very high. Many if not most providers are investing in EHRs. National data standards are needed now to ensure that these systems possess the necessary capabilities to improve patient safety and are capable of exchanging information reliably.
There is little doubt that sizable financial investments will be needed to build the national health information infrastructure. It was beyond the scope of this study to develop estimates of the resources required to accomplish the agenda proposed herein or to evaluate alternatives for providing these resources. The conduct of such an analysis represents an important next step that should be pursued immediately.
Once the basic health information infrastructure has been built, the health care sector should be able to function at a far higher level of safety and efficiency. Many of the factors that lead to errors (e.g., illegible handwriting in clinical records, mistakes in calculating drug dosages, lack of access to information on a patient’s known drug allergies) will have been eliminated. Although human analysis of errors that do occur will still be necessary, the information technology infrastructure should greatly reduce the human effort currently required to identify and analyze most errors. Over time, the infrastructure, including health care data standards, will need to evolve to accommodate developments in medical knowledge, technological innovations, and social changes in the way patients and their families interact with the health care delivery system.
Although building the information technology infrastructure is critical to both error prevention and error reporting, the elegance of implementing an EHR system is that it is a single solution to both objectives. Investment in EHR systems is critical to applying much of the knowledge that already exists about error prevention. Robust internal and external reporting systems for near misses and adverse events provide new knowledge that makes it possible to design even safer delivery systems. In building their EHR systems, health care organizations may want to target initial investments to the establishment of key capabilities for which a sizable knowledge base already exists with regard to the prevention of errors (e.g., medication order entry systems significantly reduce medication errors) and in areas in which computerized data will be useful in detecting and analyzing errors. All health care providers should also derive benefits in the near future from AHRQ’s efforts to establish standardized error reporting requirements and to conduct applied research that will lead to enhanced knowledge and tools that can be used to improve patient safety.
The committee apologizes for the heavy use of acronyms in the Executive Summary and the report itself. Both spoken and written discussion on health care data standards are replete with acronyms. To help the reader, a
list of acronyms used in the Executive Summary is provided below; a fuller list is provided in Appendix B.
—Agency for Healthcare Research and Quality
—Consolidated Health Informatics (initiative)
—Department of Health and Human Services
—Digital Imaging and Communications in Medicine
—electronic health record
—Health Insurance Portability and Accountability Act
—Health Level Seven
—International Classification of Diseases
—Institute of Electrical and Electronics Engineers
—Joint Commission on Accreditation of Healthcare Organizations
—National Council on Prescription Drug Programs
—National Committee on Vital and Health Statistics
—Systemized Nomenclature of Human and Veterinary Medicine Clinical Terms
—unique health identifier
Agency for Healthcare Research and Quality. 2002. Integrated Delivery System Research Network: Fact Sheet. Online. Available: http://www.ahrq.gov/research/idsrn.htm [accessed August 14, 2003].
———. 2003. AHRQ: Evidence-Based Practice. Online. Available: http://www.ahrq.gov/clinic/epcix.htm [accessed April 25, 2003].
Balas, E. A., S. Weingarten, C. T. Garb, D. Blumenthal, S. A. Boren, and G. D. Brown. 2000. Improving preventive care by prompting physicians. Arch Intern Med 160 (3):301–308.
Bates, D. W., L. L. Leape, D. J. Cullen, N. Laird, L. A. Petersen, J. M. Teich, E. Burdick, M. Hickey, S. Kleefield, B. Shea, M. Vander Vliet, and D. L. Seger. 1998. Effect of computerized physician order entry and a team intervention on prevention of serious medication errors. JAMA 280 (15):1311–1316.
Bates, D. W., J. M. Teich, J. Lee, D. Seger, G. J. Kuperman, N. Ma’Luf, D. Boyle, and L. Leape. 1999. The impact of computerized physician order entry on medication error prevention. J Am Med Inform Assoc 6 (4):313–321.
Beta-Blocker Heart Attack Trial. 1982. A randomized trial of propranolol in patients with acute myocardial infarction. I. Mortality results. JAMA 247 (12):1707–1714.
Billings, C. E. 1999. The NASA aviation safety reporting system: Lessons learned from voluntary incident reporting. In: Enhancing Patient Safety and Reducing Errors. Chicago, IL: National Patient Safety Foundation.
Brailer, D. J. 2003. Use and Adoption of Computer-Based Patient Records in the United States: A Review and Update. PowerPoint Presentation to IOM Committee on Data Standards for Patient Safety on January 23, 2003. Online. Available: http://www.iom.edu/file.asp?id=10988 [accessed December 16, 2003].
Brennan, T. A. 2000. The Institute of Medicine report on medical errors: Could it do harm? N Eng J Med 342 (15):1123–1125.
Department of Health and Human Services. 2002a. HHS Launches National Nursing Home Quality Initiative: Broad Effort to Improve Quality in Nursing Homes Across the Country. Online. Available: http://www.dhhs.gov/news/press/2002pres/20021112.html [accessed August 19, 2003].
———. 2002b. Secretary Thompson Welcomes New Effort to Provide Hospital Quality of Care Information. Online. Available: http://www.dhhs.gov/news/press/2002pres/20021212.html [accessed August 19, 2002].
———. 2002c. HHS Moves Forward to Establish New System for Collecting Patient Safety Data. Online. Available: http://www.hhs.gov/news/press/2002pres/20021125.html [accessed August 18, 2003].
———. 2003a. National Health Information Infrastructure 2003: Developing a National Action Agenda for NHII. Online. Available: http://www.nhii-03.s-3.net/welcome.htm [accessed May 30, 2003].
———. 2003b. Federal Government Announces First Federal EGov Health Information Exchange Standards. Online. Available: http://www.dhhs.gov/news/press/2003pres/20030321a.html [accessed May 29, 2003].
Durieux, P., R. Nizard, P. Ravaud, N. Mounier, and E. Lepage. 2000. A clinical decision support system for prevention of venous thromboembolism: Effect on physician behavior. JAMA 283 (21):2816–2821.
Evans, R. S., S. L. Pestotnik, D. C. Classen, T. P. Clemmer, L. K. Weaver, J. F. Orme Jr., J. F. Lloyd, and J. P. Burke. 1998. A computer-assisted management program for antibiotics and other antiinfective agents. N Engl J Med 338 (4):232–238.
Gurwitz, J. H., T. S. Field, J. Avorn, D. McCormick, S. Jain, M. Eckler, M. Benser, A. C. Edmondson, and D. W. Bates. 2000. Incidence and preventability of adverse drug events in nursing homes. Am J Med 109 (2):87–94.
Gurwitz, J. H., T. S. Field, L. R. Harrold, J. Rothschild, K. Debellis, A. C. Seger, C. Cadoret, L. S. Fish, L. Garber, M. Kelleher, and D. W. Bates. 2003. Incidence and preventability of adverse drug events among older persons in the ambulatory setting. JAMA 289 (9):1107–1116.
Hammond, W. E. 2002. Patient Safety Data Standards: View from a Standards Perspective. PowerPoint Presentation to IOM Committee on Data Standards for Patient Safety on May 6, 2002. Online. Available: http://www.iom.edu/file.asp?id=9915 [accessed December 16, 2003].
Hayward, R. A., and T. P. Hofer. 2001. Estimating hospital deaths due to medical errors: Preventability is in the eye of the reviewer. JAMA 286 (4):415–420.
Institute of Medicine. 2000. To Err Is Human: Building a Safer Health System. Washington, DC: National Academy Press.
———. 2001. Crossing the Quality Chasm: A New Health System for the 21st Century. Washington, DC: National Academy Press.
———. 2002a. Fostering Rapid Advances in Health Care: Learning from System Demonstrations. Washington, DC: The National Academies Press.
———. 2002b. Leadership by Example: Coordinating Government Roles in Improving Health Care Quality. Washington, DC: The National Academies Press.
———. 2003a. Key Capabilities of an Electronic Health Record System: Letter Report. Washington, DC: The National Academies Press.
———. 2003b. Priority Areas for National Action: Transforming Health Care Quality. Washington, DC: The National Academies Press.
Kizer, K. W. 2001. Establishing health care performance standards in an era of consumerism. JAMA 286 (10):1213–1217.
Markle Foundation. 2003. Connecting for Health. A Public-Private Collaborative: Key Themes and Guiding Principles. New York, NY: Markle Foundation.
McDonald, C. J., M. Weiner, and S. L. Hui. 2000. Deaths due to medical errors are exaggerated in Institute of Medicine report. JAMA 284 (1):93–95.
McGlynn, E. A., S. M. Asch, J. Adams, J. Keesey, J. Hicks, A. DeCristofaro, and E. A. Kerr. 2003. The quality of health care delivered to adults in the United States. N Engl J Med 348 (26):2635–2645.
National Committee for Quality Assurance. 2003. HEDIS®—Health Plan Employer Data and Information Set. Online. Available: http://www.ncqa.org/communications/publications/hedispub.htm [accessed November 11, 2003].
National Committee on Vital and Health Statistics. 2000. Uniform Data Standards for Patient Medical Record Information. Online. Available: http://ncvhs.hhs.gov/hipaa000706.pdf [accessed April 15, 2002].
———. 2001. Information for Health: A Strategy for Building the National Health Information Infrastructure. Online. Available: http://ncvhs.hhs.gov/nhiilayo.pdf [accessed April 18, 2002].
National Quality Forum. 2002. Serious Reportable Events in Patient Safety: A National Quality Forum Consensus Report . Washington, DC: National Quality Forum.
Office of Management and Budget. 2003. Consolidated Health Informatics. Online. Available: http://www.whitehouse.gov/omb/egov/gtob/health_informatics.htm [accessed April 21, 2003].
President’s Information Technology Advisory Committee. 2001. Transforming Health Care Through Information Technology. Online. Available: http://www.hpcc.gov/pubs/pitac/pitac-hc-9feb01.pdf [accessed May 7, 2003].
Rosenthal, J., and M. Booth. 2003. Defining Reportable Events: A Guide for States Tracking Medical Errors. Portland, ME: National Academy for State Health Policy (NASHP).
Shortliffe, E. H., L. E. Perreault, G. Wiederhold, and L. M. Fagan. 2001. Medical Informatics: Computer Applications in Healthcare and Biomedicine. New York, NY: Springer-Verlag.
Soumerai, S. B., T. J. McLaughlin, D. Spiegelman, E. Hertzmark, G. Thibault, and L. Goldman. 1997. Adverse outcomes of underuse of beta-blockers in elderly survivors of acute myocardial infarction. JAMA 277 (2):115–121.
Sox, H. C., and S. Woloshin. 2000. How many deaths are due to medical error? Getting the number right. Eff Clin Pract 6:277–283.
Vincent, C., G. Neale, and M. Woloshynowych. 2001. Adverse events in British hospitals: Preliminary retrospective record review. BMJ 322 (7285):517–519.
Wilson, R. M., W. B. Runciman, R. W. Gibberd, B. T. Harrison, L. Newby, and J. D. Hamilton. 1995. The quality in Australian health care study. Med J Aust 163 (9):458–471.
World Health Organization. 2002. Quality of Care: Patient Safety. Online. Available: http://www.who.int/gb/EB_WHA/PDF/WHA55/ewha5518.pdf [accessed December 16, 2003].