Much has changed since the release of the first edition of
The Computer-Based Patient Record: An Essential Technology for Health
Care. The current environment in which health care is practiced and the information
technology available to its practitioners are significantly different from that which
existed when the study was completed in 1991. Changes in the health care
environment produced fundamental shifts in the delivery of health care, favoring
outpatient care over inpatient care, primary care over specialty care, and
guidelines-driven care over autonomous decisionmaking. Technological advances have
overcome some barriers to computer-based patient records (CPRs) (e.g., World Wide
Web, applications that operate across distances on many different computers)
and heightened the visibility of others (e.g., confidentiality policies and
legislation). In this commentary, we describe some of the environmental and
technological changes that have occurred since publication of the first edition and highlight
the challenges that remain to be addressed. Probably the most significant
nontechnological change that occurred since 1991 was the change in the health care
practice environment. We begin by addressing the new environment.
THE CHANGING HEALTH CARE LANDSCAPE:
REINFORCING THE NEED FOR CPRs
As health care costs in the United States approached $1 trillion in the early 1990s, businesses, consumers, and payers began to call for mechanisms to manage the escalating costs (IOM, 1993). This issue was so much at the forefront of U.S. business and political priorities, that the 1992 presidential election was dominated by discussions of health care costs and debate on effective ways to control them.
Although efforts at full-scale health care reform led by the federal
government gained momentum, peaked, and then disintegrated, market-driven
changes in health care delivery were already under way and steadily moving
forward. Managed care plans continue to gain increasing shares of the market
(Hoechst Marion Rousell, Inc., 1995). The need to examine and manage the health needs
of a population has dramatically increased the demand for information systems
that capture clinical data. The health care information systems industry has shifted
its attention from financial systems to clinical systems, particularly CPR
systems. Two core tenets of managed care that impact the demand for CPR systems
have been the central role of primary care and the emergence of integrated
delivery systems.
The Evolving Role of Primary Care
Managed care has redefined the primary care provider as the principal provider of care and, in many instances, a gatekeeper for access to certain diagnostic tests and specialty care. To efficiently carry out these roles, the clinician needs ready access to both clinical and administrative data. At the same time, guidelines are being promoted to reduce the variances among clinical practices. When guidelines are integrated into a CPR system, they can streamline the steps necessary to adhere to them and simultaneously document that compliance. CPR systems that integrate clinical guidelines in the order-entry process have the best acceptance among providers (Sittig and Stead, 1994; Sullivan and Mitchell, 1995; CPRI, 1996c, 1997). However, despite the abundance of guidelines, few can be implemented in a computer as written (Tierney et al., 1995). Software tools are needed to help author guidelines that are internally consistent, include precise definitions of eligibility criteria, and accommodate a variety of patient conditions.
One of the major goals of managed care is to provide more of the care
outside of the hospital. However, the inpatient and the outpatient settings differ
substantially. There are differences in the temporal nature of information, the
responsibilities of each member of the health care team, the need for a
communications infrastructure to facilitate coordination of care, and other logistical
concerns which impact the detailed design of information systems. Consequently,
vendors of information system products for hospitals find that there is a steep
learning curve to understanding the information needs of physicians in the
ambulatory care setting. Deliberate analysis of the information needs and work flow
requirements in ambulatory care will help system developers design information
systems that increase the efficiency and effectiveness of clinical practice
throughout the continuum.
The Integrated Delivery System
Another fundamental concept that heightens the need for CPR systems is that of an integrated delivery system (IDS). An IDS is composed of health care providers, service providers, and facilities organized to provide a continuum of health care services to a defined population. These systems of health care were created in response to payers' desire to contract with single entities that provide comprehensive health care services for their clients. To manage the delivery of care in an IDS, a health system must have efficient and accurate ways of capturing, managing, and analyzing clinical data collected at all the different sites where care is provided.
In addition, payers and regulators are requesting "report cards" on quality, outcomes, and costs of care provided by the integrated delivery system. For example, the National Committee for Quality Assurance (NCQA) developed the Health Plan Employer Data and Information Set (HEDIS) as a standard report card to help employers evaluate different health plans. Initially focused more on administrative data, the evolving HEDIS criteria are increasingly targeting clinical processes and outcomes. Gathering the data to prepare these reports can be immensely time-consuming and costly when they are manually abstracted from paper records, but with a CPR, reporting on aggregate data can be a byproduct of capturing data electronically.
NCQA advised health plans to "move to fully implement the information framework, including the automated patient record" in order to meet the clinical reporting requirements of forthcoming regulations (NCQA, 1997). As outcomes reporting requirements become more sophisticated and deal with complex, multifaceted diseases, it will be essential to have electronic access to the record and tools to efficiently analyze practice patterns and patient outcomes. NCQA will develop HEDIS measures that assume health plans and provider organizations use CPRs by 2002.
In short, since the release of the first edition, the demand for clinical data
has become a business imperative. The organizational complexities of these
large, diverse, and geographically dispersed health systems add new
challenges—and new opportunities—for developers and implementers of CPR systems.
Technological Advances
Computer processing power doubles in performance and halves in cost
about every 2 years. These past 6 years have been no exception. Although it is not
our intent to review all the developments in information technology (because
technology has not been the critical impediment to adoption of CPR systems),
three striking technological shifts have occurred that favorably impact the
foundation for CPR systems. We briefly describe these developments below.
The Internet and the World Wide Web: Widespread Connectivity
Perhaps the single biggest technological change since the first edition of the report is the rising importance of the Internet. Although the Internet dates back to 1969 when the first node of ARPANET was installed at the University of California at Los Angeles, several changes have coalesced to bring together a paradigm shift that now touches all segments of society. The World Wide Web (WWW) has transformed a research network into the fabric of a new information age. Internet service providers have made access convenient and reasonably inexpensive. As much as 50 percent of the U.S. population will have access to the Internet by early in the next century (National Research Council, 1996). It is precisely this kind of ubiquitous connectivity that enables IDS organizations to share data throughout their geographically dispersed clinical delivery sites, and even to reach the consumer or patient at home. To the extent that health care becomes dependent on access to computer networks, however, policymakers need to pay special attention to the needs of the medically underserved population to ensure that lack of network access does not further impede their access to care.
Connectivity is not the only requirement for transmitting patient data to
remote sites. Confidentiality and security safeguards need to be developed and
enforced. Fortunately, the requirements of business to protect electronic
commerce over the Internet will drive technical solutions and policy standards, which
health care applications can leverage. Technology dedicated solely to health care
applications risk being orphaned due to the lack of a mass market. The market drive
of consumerism must pave the information infrastructure for health care
applications.
World Wide Web Browsers: A Universal View on the Internet
Probably the most important tool that led to the domestication of the Internet was the development of software that made it easy to connect to, search, browse, and download information from anywhere on the network as if it were located on the user's personal computer. Commonly called browser software (e.g., Netscape NavigatorTM, Microsoft Internet ExplorerTM), these programs give a graphical, intuitive, and common interface to functions that locate and interact with remote data on the Internet without the user having any technical knowledge of how it is done. Browser user interfaces have become so commonplace that they are being adopted as the interface to desktop computers.
Another fundamental breakthrough associated with WWW browsers is that the software runs on almost any computer (Cimino, 1995). One of the critical problems that had been plaguing computer users since the invention of electronic computing has been the general inability of programs written for one machine to run on another machine or to use data generated by another program. Efficient and cost-effective use of computers suffered due to the incompatibilities caused by a lack of standards. The WWW defined standards for document formats (hypertext markup language, HTML) and transfer protocols (hypertext transfer protocol, HTTP). Compliance with these standards permitted independent developers to write programs for heterogeneous computers and operating systems, yet have them all seamlessly access information on the WWW.
The success of the WWW demonstrates the market-expanding potential
of adopted standards. It also demonstrates the remarkable leverage provided
through natural entrepreneurial forces complying with industry standards. The benefits
to the consumer, including clinicians, are tremendous.
The Intranet
A variant, but powerful spin-off, of the Internet is the Intranet, a
controlled-access version of the Internet. Like the Internet, Intranets use
industry-standard document formats, data exchange protocols, and browsers. The prolific
development of new tools and products for the Internet can be directly applied to
Intranets. An estimated 80 percent of the Internet products are purchased for use on
an Intranet. The Intranet, however, typically has better bandwidth, security, and
administration because it is controlled and operated by the private enterprise.
These advantages are particularly important for health care applications, and most
health care organizations are exploring the use of Intranets as a component of their
information strategy.
Network-Centric Computing
Industry is at the cusp of another major change—network-centric computing. The predominant paradigm for computing today uses personal computers (PCs) to perform most of the computational work using software that resides on local disks. The hardware, software, and support costs for this type of operation require large budgets and staffs.
Network-centric computers are diskless computers that attach to a network and load software stored on a server computer. The capability to manage hardware and software more centrally reduces the acquisition and operating costs for the enterprise. Although this new architecture has not been used extensively in production, it has the potential to revamp distributed computing.
Having noted the significant changes in the health care environment
and advances in information technology, we next discuss the current state of the CPR.
THE CURRENT STATE OF CPRs
As discussed in Chapter 4 of the report, the development and diffusion of
a new technology are separable but interrelated events. In discussing the state
of CPR development and implementation in the United States today, it would
be desirable to examine two questions. First, to what extent have CPRs, as
defined by the study committee, been developed? Second, to what extent have the
available CPRs been adopted? Unfortunately, a comprehensive review of the
industry does not exist, and it would soon be out of date if it did. Alternatively, one
could rephrase the questions from the perspectives of providers and vendors. From
the perspective of system purchasers, do the CPR systems that are available in
the market meet the needs of health care institutions? And from the perspective
of vendors, is the market ready to buy CPR systems? Although in neither case is
the answer a resounding "yes," promising signs can be reported.
The CPR Market
One must be careful when describing a "CPR market" not to tether the concept to a single, static idea whose incarnation can be purchased "off-the-shelf." In some sense, it is precisely the static nature of the paper-based record that has been such a great burden to the practice of medicine. Instead, a CPR system is a constantly evolving concept whose value and function is expected to grow with the constantly changing demands of the health care environment and the improving technology upon which the system is built. The level of CPR development activity has definitely increased significantly since 1991. The number of commercial systems addressing various attributes of CPRs has increased and most major health care information technology vendors now offer CPR-related products. However, comprehensive information system products that seamlessly integrate data and coordinate processes across the entire continuum of health care services do not exist. Most health care information system vendors, whether their products were formerly based in the inpatient or the outpatient side, are working to extend their products to cover the needs of integrated delivery systems. Developers generally start from the data end of the system and work toward the human side where the clinician interacts directly with the system. Hence, one way to trace the evolving functionality is to look at the transformation of data into information that clinicians use to make decisions. Although the path is not necessarily sequential, five hallmarks of this transformation are enumerated below:
- 1. Integrated view of patient data,
2. Access to knowledge resources,
3. Physician order entry and clinician data entry,
4. Integrated communications support, and
5. Clinical decision support.
Integrated View of Patient Data
This is one of the earliest benefits of CPRs—improving access to all
patient data whenever and wherever clinical decisions are made independent of
where the data was originally acquired. Observational studies of clinicians'
information needs provide details on the kinds of information physicians require to
make decisions regarding the care of their patients (Tang et al., 1994). Up to 81
percent of the time, physicians could not find all the available patient information
desired to make patient care decisions during an outpatient encounter. Limited by a
format that has not substantively changed in close to a century, the paper record
is ill-suited to the information demands of modern clinical practice. Most CPR
system vendors offer products that combine data from various sources and
present an integrated view to clinicians.
Access to Knowledge Resources
Providers often need other information in addition to patient data and
their own personal knowledge (Covell et al., 1985). The guidelines, rules, and
regulations in the managed care environment have intensified this need to access
medical and administrative knowledge at the time decisions are made. CPR
systems sometimes provide methods for organizations to incorporate access to
local knowledge resources, but generally, this knowledge access is passive. That is,
the user searches for the needed information electronically but has to abstract
the pertinent content and enter any relevant orders manually into the clinical
system. Ideally, access to knowledge resources should be integrated with clinical
decision support in ways that directly influence physicians' ordering behavior, as
described below.
Physician Order Entry and Clinician Data Entry
Physician orders initiate clinical interventions. Proactively influencing
physicians' orders is the most efficient way to influence patient outcomes
(McDonald, 1976, 1984). Systems that physicians use routinely to enter orders, whether in
the inpatient setting or outpatient setting, can produce significant effects on
quality and costs of care (Barnett, 1984; Tierney et al., 1987, 1990, 1993; Sittig
and Stead, 1994). In addition, when clinical data are entered and maintained by
the clinicians responsible for care, the accuracy and quality of data are high.
Very few commercial systems, however, are used by physicians to write all their
orders. Human-computer interface issues and perceived benefit substantially
affect the success of this function. Some of the reasons are discussed later under
remaining barriers.
Integrated Communications Support
With an increasing emphasis on outpatient care, coordinating the activities
of health care professionals from multiple organizations at different sites,
including the home, becomes more important. No longer confined to an acute care
facility where proximity between the health care team members and the patient was
the rule, the patient and care team in the outpatient setting meet more by
appointment than by clinical demand. Relying on paper-based mail is inefficient and
fallible. Clinicians need integrated communications support for effective functioning
of the multidisciplinary outpatient health care team (Tang et al., 1996). A
communications infrastructure that is linked to the shared patient record facilitates
overall coordination of care and timely response to changing patient conditions.
Clinical Decision Support
Few, if any, commercial systems provide a high level of proactive decision support. Drug interaction checking and simple abnormal laboratory-test result alerts are available, but the richness of applying a broad range of knowledge to influence physicians' orders is still under development (Johnson, 1994).
Only when both patient data and clinical knowledge reside in the system in machine-understandable format can the system provide additional support to the clinician making decisions. For example, encoded medical knowledge about the meaning and significance of changing laboratory-test results would allow a system to provide alerts, an active function, in addition to the passive data retrieval function. Similarly, if the system could match the patient context with relevant clinical guidelines, it could present ordering options consistent with the appropriate guidelines. The clinician is responsible for the definitive decision, but the system can actively provide options and explanations that improve the clinician's efficiency and compliance with accepted guidelines of practice.
In summary, since the first edition, there has been a significant increase
in development efforts on CPR systems. We remain optimistic that the
remaining years of the decade-long challenge will deliver on the promises of
computer-supported decisionmaking.
At the Fork
In 1991, the Committee on Improving the Patient Record reported that the most advanced CPR systems were found in several academic medical centers or teaching hospitals affiliated with universities as well as in the Department of Veterans Affairs and the Department of Defense. Not surprisingly, today the most advanced CPR systems implemented in the United States can be found in the same places. Of the seven institutions that have been recognized as part of the Computer-based Patient Record Institute Davies CPR Recognition Program (described below), four are academic medical centers and one is the Department of Veterans Affairs. One of the distinguishing factors of institutions that have been recognized as having advanced CPRs is the clear organizational leadership and commitment to CPR system implementation that helped make their efforts successful (CPRI, 1995c, 1996c, 1997).
A comic saying attributed to Yogi Berra states: "When you come to the
fork in the road, take it." In contrast to the early CPR system adopters, most
health care institutions are standing at the fork, trying to decide whether or not to
begin implementing a CPR system. Those who have made the decision to invest in
a CPR system are grappling with the complicated issue of how to do so.
Many organizations implement the CPR system in some, but not all areas. Others
implement a partial system and depend on a combination of paper and electronic
documentation. All adopters of CPR systems must address how to integrate the
components of the CPR and how to integrate the CPR with other
institutional information systems. The challenge of implementing such an expansive,
robust system is daunting, but the option of continuing to manage the clinical and
administrative data of an IDS on paper is increasingly becoming a nonviable
alternative.
Recent Activity to Advance CPRs in the United States
Against the backdrop of technical and nontechnical changes, interest in and incentives to develop CPR systems have increased.
The National Library of Medicine (NLM) has been at the forefront in stimulating research on the effective use of CPR systems and networked access to shared data (Lindberg, 1995). NLM conducts intramural research that is directly applicable to technological and infrastructural needs of CPR system development and effectively uses its extramural research grants and contracts to apply the results of academic research to health care. Through its High-Performance Computing and Communications contracts, NLM has been a leader in facilitating the use of information technology by health care professionals of rural, urban, community, and statewide networks around the country. NLM's extensive work on the Unified Medical Language System (UMLS) has been a major contribution in the medical terminology arena. NLM and the Agency for Health Care Policy Research (AHCPR) are sponsoring a large-scale vocabulary test to assess the "extent to which a combination of existing health-related classifications and vocabularies covers vocabulary needed in information systems supporting health care, public health, and health services research" (Humphreys et al., 1996). NLM is further extending the reach of shared computer-based patient data and systems through its telemedicine program.
Due to AHCPR's role in developing scientifically based clinical guidelines, it has long recognized the importance of standard data definitions and capturing clinical data in structured form. AHCPR has also played an active role in facilitating standards development. Widespread use of CPR systems would not only facilitate the collection of aggregate data in support of guidelines development, but also make effective the dissemination and use of clinical guidelines in clinical practice.
In addition to the efforts undertaken by NLM and AHCPR, other federal agencies, including the Department of Defense, and the National Institute of Standards and Testing have internal and extramural activities in CPR systems. However, a cohesive federal policy to speed the development of a health information infrastructure and the diffusion of CPRs has not emerged in the United States (Shortliffe et al., 1996).
One focal point for accelerating the development of the infrastructure for CPR systems (e.g., confidentiality policies, standards, evaluation criteria) in the private sector has been the Computer-based Patient Records Institute (CPRI). CPRI was formed in response to the recommendation of the Institute of Medicine (IOM) study committee to promote and facilitate the development, implementation, and dissemination of the CPR. Led by many of the initial supporters of the IOM committee, CPRI was incorporated in January 1992, as an association of organizations representing the various stakeholders in health care. In its first year, 22 organizations became members of CPRI. Over the subsequent years, the organization has grown to more than 70 members. CPRI was charged by the IOM committee with the following objectives:
- 1. Support the effective, efficient use of computer-based patient records.
2. Educate change agents and stakeholders about the value of CPRs in improving patient care.
3. Foster the CPR as the primary vehicle for collecting patient data.
4. Promote the development and use of standards for CPR security and data content, structures, and vocabulary.
Despite efforts in the public and private sectors, however, significant
barriers impede the development and use of CPR systems in the United States. Many
of the remaining critical barriers to CPR system development and routine use
concern problems that are most effectively dealt with by cooperative, focused
activity. We describe some of these barriers below.
Remaining Barriers or Challenges
Technology has continued to move forward at a rapid pace. By
comparison, the human and organizational sides of the challenges have remained
relatively stagnant. In 1991, the committee stated that informational, organizational,
and behavioral barriers must be addressed to advance CPR systems, and that
these barriers overshadowed the technical barriers. Below, we elaborate on some
of the critical barriers to CPR development and diffusion.
Definition of the CPR
Although work on a common definition of a CPR and CPR system is under way by various groups, a universal understanding of the concepts embodied in a CPR does not exist. Without a clear understanding, users have a difficult time selecting systems that will meet their needs and vendors have difficulty supplying such systems. CPRI described a CPR as "electronically stored information about an individual's lifetime health status and health care." It replaces the paper medical record as the primary record of care, meeting all clinical, legal, and administrative requirements. A CPR system provides reminders and alerts, linkages with knowledge sources for decision support, and data for outcomes research and improved management of health care delivery. It is worth repeating a point made earlier that a CPR system is an evolving concept that responds to the dynamic nature of the health care environment and takes advantage of technological advances.
Beyond the first set of definitions, however, few details have been worked out and agreed upon. For example, there is no common data model for the CPR, no common set of data elements, no common vocabulary, and no common set of scenarios that are supported. These requirements are fundamental if developers are to create a person-centered CPR that links care across different sites, specialties, and circumstances.
Many systems still follow the traditional organization and characteristics
of the paper-based system and have simply automated that system. Narrative
documentation, for example, is far more prevalent than structured text. Even
though most new CPRs support a multimedia record, new data forms have not
been smoothly integrated into the record, and little has been done to evaluate their
true worth. Finally, the concept of incorporating patient-derived information
(e.g., health status) as part of the patient record has not been implemented to any
significant degree.
Meeting User Needs
In order for clinicians to rely on data in the system, they must be the direct users of the system. The emphasis on clinical data is a fundamental change from the previous era of hospital information systems, where clerks were the main users because the primary motivation was to capture charges and generate bills.
Analysis of the common questions concerning patient information that physicians ask (e.g., what evidence supports the diagnosis, has a patient ever had a specific test, and has there been any follow-up because of a particular laboratory test result?) provides insight into the difficulties clinicians have had finding the answers in the paper-based chart. Although current computer-based tools installed in health care institutions can typically help clinicians retrieve laboratory test results, they are not designed to answer many of the common questions clinicians ask about patient data. CPR system developers will need to address these needs to satisfy the new clinician users. The maker of the Swiss Army Knife describes the key to inventions as follows: "Make it useful. Very useful. Conveniently useful. . . . " Likewise, key to gaining clinician user acceptance is providing efficient tools that help clinicians retrieve and understand data relevant to their decision-making tasks.
Clinicians also need answers to many questions regarding medical knowledge (Covell et al., 1985). Diagnostic decision support tools are available as stand-alone microcomputer programs (e.g., QMR, Illiad, DXplain) or are available over the Internet (e.g., DXplain; Miller et al., 1986; Barnett et al., 1987; Feldman and Barnett, 1991; Berner et al., 1996). However, their greatest use would occur when proven diagnostic decision support tools are integrated with CPR systems.
Among the more challenging issues confronting CPR system developers is the issue of effective user-computer interfaces. Physicians must be the users of the system, performing data entry (e.g., orders, progress notes) as well as information retrieval, if they are to realize the benefits of interactive, on-line decision support. Progress must be made in understanding the "cognitive processes involved in human-computer interactions in order to design interfaces that are more intuitive and more acceptable" (Tang and Patel, 1994). Cognitive issues are also relevant to designing the presentation of clinical information in ways that facilitate rapid assimilation and analysis. Ultimately, good solutions to the human-computer interface will require changes not only in how the system looks but also in how humans interact with the system. What information the provider needs and what tasks the provider is performing will influence what is presented as well as how it is presented. Templates that reflect providers' work flows will aid them in using the system efficiently. Defaults that represent common, desired selections, for example, not only improve efficiency but also help increase compliance with practice-defined guidelines (CPRI, 1996c, 1997).
Although user acceptance was a major barrier in the past, more and more users are demanding that organizations implement computer-based means for accessing and managing patient data. This is not to say that the cultural and organizational challenges associated with major change are not substantial barriers to overcome, but the users' awareness of the potential benefits will become an increasing catalyst for change.
The change in users and uses has a dramatic impact on the applications
desired and the kinds of user interfaces required. The industry must now focus on
a completely different set of users and must define and address their
information needs. Proactive and informed user participation will be necessary for
implementations to succeed. The attention to clinical data and clinical users described
in this report is even more important today than it was 6 years ago.
Standards
Rather than existing as a monolithic system, contemporary health care information systems are made up of multiple-component systems manufactured by multiple vendors, owned by multiple entities. To share data, which is required in an IDS, industry-adopted standards must be defined for interfaces between components (AMIA, 1994; Hammond, 1994).
Health systems must have a unique health identifier (UHI) to accurately and reliably link all the data on a single individual. Until last year, striving for a national UHI had been an elusive goal. Public Law 104-191 (The Insurance Portability and Accountability Act of 1996) calls for the secretary of Health and Human Services "to adopt standards for unique health identifiers, confidentiality policies, and terminology." To build support for addressing these pivotal issues, in November 1996, CPRI convened a national Summit of over 80 senior executives representing health care providers, health systems, government agencies, information systems vendors, businesses, regulators, and quality assurance professionals. The organizations participating in the Summit developed consensus recommendations regarding a unique health identifier, confidentiality legislation and policies, and standard health terminology. CPRI was charged with follow-up actions to work with government agencies and industry groups to implement the recommendations.
The need for standards governing the content, vocabulary, and format of data remains a high priority. So far, most of the progress has been made in the development of messaging standards. Standards for data exchange have continued to be expanded and are increasingly being adopted by provider organizations and vendors. Standards now exist for exchanging clinical data (Health Level Seven [HL-7], 1994), images (ACR/NEMA), clinical observations (ASTM Committee E-31), bedside instrument data (IEEE, 1995), prescription data (NCPDP, 1992), and administrative data associated with claims (Accredited Standards Committee X12N). It is important to note, however, that even with standards that define message formats, implementation of messaging standards may differ among vendors. The next step is to call on industry to provide standard implementations for messaging standards such as HL-7. That would be a major step toward "plug-and-play" capability.
Although there has been progress in developing individual coding standards for data elements, none has emerged as a comprehensive standard. More than 150 different coding sets defining terms for use with the medical record have been created. Among these coding sets are ICD9-CM, ICD10, SNOMED III, CPT, NANDA, Read Classification, LOINC, and MEDRA, to name a few. Not only findings, but also results of laboratory tests must have standard terms (e.g., results of a Pap smear or mammogram). Additional coding sets are being defined by specialty groups that recognize the need for a common vocabulary. The NLM has expanded its UMLS to map many of these coding sets into a common set. Names for every element that must be exchanged among systems should be standardized. For the CPR, a clinically rich vocabulary that accurately describes patient problems and findings is mandatory. In addition, physicians must feel comfortable with the standard vocabulary if they are to use it for entering data. CPRI has published an evaluation of existing comprehensive codes (Chute et al., 1996).
In addition to standards for patient data, medical knowledge must be encoded in CPR systems with decision support capability. The Arden syntax has been used to exchange medical knowledge encoded as simple rules (Hripcsak et al., 1990). Further work must be undertaken to represent medical knowledge in standard, transferable form.
Leadership at the federal level is required to ensure that standards
necessary to preserve and enhance health care in the United States are developed.
Until standards exist for uniquely identifying individuals and coding and
exchanging health data, the value from capturing and aggregating data will go unrealized
and each organization will be its own pioneer.
Legal and Social Issues
Security, privacy, and confidentiality concerns have become major barriers to widespread implementation of CPR systems and sharing data. There is, as yet, no agreement on what must be done to establish the balance between appropriate use of health care data and the individual patient's rights to privacy (Detmer and Steen, 1996). The issue of who owns the data in a CPR is still being debated. Of equal importance to preserving patient privacy and confidentiality is the necessity of preserving institutional privacy. No institution will be willing to share data if those data can be used to provide a business advantage for a competitor. Again, the human factors outweigh the technical solutions in dealing with this issue (Barrows and Clayton, 1996).
Privacy and confidentiality are concepts that involve people, policies, and legislation. Information security technology plays an enabling and facilitating role by helping organizations prevent unauthorized access to confidential information. In addition, properly designed and monitored audit trails can enhance user accountability by detecting and recording unauthorized access to confidential information. CPRI has produced position papers on user authentication and access to patient data and provided substantive guidelines on security policies, security education programs, job descriptions for information security managers, model confidentiality policies, and security functionality requirements for CPR systems (CPRI, 1995a,b, 1996b,d). CPRI and the American Health Information Management Association have been instrumental in developing model policies and legislation regarding confidentiality and privacy. Public Law 104-191 establishes legal sanctions for wrongful disclosure of individually identifiable health information. It also calls on the secretary of Health and Human Services to provide detailed recommendations on privacy of health data and procedures and rules for authorized disclosure of such information. The recently revitalized National Committee on Vital and Health Statistics (NCVHS) advises the secretary on this and other standards related to health information. Federal legislation is necessary to overcome many of the inadequacies and inconsistencies between the state regulations and laws that are described in this report.
Whereas stringent security measures should be applied to protect the
confidentiality of patient information, it is also in the patient's best interest for
the CPR to be accessible for appropriate, legitimate uses by authorized users.
The measures used to protect the confidentiality of patient data must not be so
onerous that clinicians taking care of a patient do not have ready access to the
patient's health data. This balance must be carefully crafted in the enabling legislation
and policy standards, and enforceable through system security functions. In
addition to clinicians, researchers, quality assessment professionals, and health care
managers need access to aggregate data to continuously improve health and the
delivery of health care. Most of the time, these secondary uses of data can be
satisfied without access to individually identifiable information.
Costs and Benefits
Certainly, the cost of developing a CPR system remains a significant barrier, but this barrier has been offset at least in part by three trends. First, as CPR-related technologies continue to advance, greater performance can be obtained for equivalent costs. Second, as integrated delivery systems become more prevalent, the demand for CPR and related information systems will increase, thus reducing the risk to developers. Third, some federal funds have been made available (e.g., through NLM and AHCPR) to address specific development challenges associated with CPRs. The stability and level of federal funding in the future, however, are uncertain.
Much of the enthusiasm for the CPR is based on the belief that a CPR system will reduce the cost and improve the quality of care through the existence of better-informed health care providers and patients, the elimination of duplicate testing, and better coordination of treatment by more than one health care provider. Since the first edition, additional data have been published on the positive impact of CPR systems on the cost and quality of health care. As existing clinical systems continued to accumulate data, new studies have demonstrated improvements in the quality of care or reductions in the cost of care (Classen et al., 1991, 1992; Evans et al., 1992; Schoenbaum and Barnett, 1992; Tierney et al., 1993; Grandia et al., 1995). However, most of the carefully done empirical studies presented in the literature have examined the benefits of systems that are not commercially available. Like other enabling technologies (e.g., electronic mail, cellular telephones, computers), it may be difficult to identify and reliably quantify the direct and indirect benefits attributable to the CPR system. It is hard to quantify what is better, when so much may not even have been possible without the enabling technology.
Although discrete benefits of CPR system features have been quantified in selected environments, the comprehensive capital and operating costs have not been fully articulated. It is known, however, that the costs are substantial. One estimate for a capital budget (including capitalized labor costs) for an integrated health care information system with $1 billion to $2 billion in revenue is $75 million to $275 million (Council on Competitiveness, 1996). At a time when health care organizations need to reduce their costs, allocating capital to information systems is still a challenge. Although it is tempting to propose that large empirical studies be conducted to produce definitive cost-benefit data, it may be unrealistic to expect that such studies can be undertaken, or that they could truly represent the diverse health systems in the United States.
Many health care organizations are aware of the benefits of CPR systems, but have questions on how to implement and use this new technology. Recognizing that experience with these systems is not widespread, CPRI developed a program to help disseminate the knowledge of those who have previous experience with CPR systems and to provide a tool to help organizations undertake new projects and initiatives. Modeled after the Malcolm Baldrige Quality Award Program, CPRI developed a program to promote quality in CPR system projects. The program is named after Nicholas E. Davies, a member of the IOM Patient Record committee who was tragically killed in an airplane crash just as the committee was concluding its work. The Nicholas E. Davies Annual CPR Recognition Symposia provide recognition for organizations demonstrating exemplary measurable impact of CPR systems on health care at their organizations (CPRI, 1995c, 1996c, 1997).
The objectives of the Davies CPR Systems Recognition Program are to:
(a) promote the vision of CPR systems through concrete examples, (b) provide
visibility and recognition for excellence in CPR system implementation and
demonstrated results, (c) provide a forum for discussing critical success factors
and lessons learned, and (d) provide criteria for evaluating CPR system
projects (CPRI, 1996a). A comprehensive set of evaluation criteria was developed in
an iterative fashion by CPRI and external experts. Emphasizing the fact that
successful CPR system projects require a comprehensive, multidisciplinary
team approach led by senior management, the evaluation categories are divided
into four categories: (1) management, (2) system functionality, (3) technology,
and (4) impact on quality, costs, and access. These categories reflect the notion that
a CPR system implementation project is more than a selection or development
of an information system product, it is the conceptualization of a new
recordkeeping system that efficiently and effectively supports the delivery of health care
and health promotion by the health care team. Existing organizational structures,
departmental cultures, and an inward focus may have to be transformed to
efficiently deliver high-quality care and effectively compete for managed care
contracts. The CPR system is an essential tool to accomplish these broader goals.
Leadership
Due to the barriers described above, leadership of and commitment to
CPR system projects must come from the top of an organization. The
organization's Board of Directors must fund and support the implementation and use of
CPR systems as a business imperative. Often, chief executive officers must act
on personal conviction, existing evidence of benefits, and the desire to capitalize
on a strategic differentiator—the value of data. In the new managed care market,
a CPR system is not only an essential tool for health professionals, it is a
business imperative. Time will tell whether investing in clinical systems provides a
strategic advantage or is a minimum requirement to "stay in the game." Neither
outcome favors continuing the status quo. It takes a combination of leadership
conviction, organizational fiscal strength, and medical informatics and
systems expertise to implement CPR systems effectively. Where such leadership
exists, CPR systems are more likely to be acquired and successfully implemented.
CONCLUSION
A careful reading of The Computer-Based Patient Record by all stakeholders in health care is still fruitful. We believe that the original vision described by the IOM Committee on Improving the Patient Record was correct and remains timely and essential to the success of the new health care delivery system. Widespread use of CPRs would serve both private- and public-sector objectives to transform health care delivery in the United States. Equally important, early evidence indicates that the introduction and use of robust CPRs will enhance the health of citizens and reduce the costs of care, and in so doing, the use of CPRs will strengthen the nation's productivity.
Technological advances aside, progress toward CPRs as envisioned in this report has been slower than anticipated. The IOM committee expressed its strong belief that the early phase of CPRI's activities should be federally initiated and funded. A major coordinated national effort with federal funding and strong advisory support from the private sector is needed to accelerate the pace of change in the United States. Health care is a public good and many of the barriers to widespread implementation of CPR systems require national mandates, policy changes, or, in some cases, new legislation. Leadership in government and the private sector must be galvanized to make sweeping changes where possible (e.g., a national UHI, confidentiality legislation) and to instigate, motivate, and provide incentives to accelerate development of solutions to other impediments (e.g., terminology standards).
Data are the currency of quality management and the endowment for
continuous quality improvement of patient care. Only by capturing primary
clinical data from health care providers in a way that they can be applied to health
care decisions for individuals and to policy decisions for populations can the
United States achieve its goal of providing high-quality, affordable health care for all.
A computer-based patient record is essential to accomplishing that goal.
ACKNOWLEDGMENTS
We are deeply indebted to Octo Barnett who refocused our attention on
the important points we were trying to communicate. The message is far
stronger because of his constructive comments. We thank Don Detmer and Elaine
Steen for their gracious invitation to contribute to this second edition.
REFERENCES
Accredited Standards Committee X12N. Series of EDI standards available from Data Interchange Standards Association, Inc., 1800 Diagonal Road, Suite 200, Alexandria, VA 22314-2852.
ACR/NEMA Standards Publication. Digital Imaging and Communications in Medicine (DICOM). PS 3.1-3.13. National Electrical Manufactuers Assocation, 2102 L Street N.W., Washington, D.C. 1992-1995.
AMIA (American Medical Informatics Association) Board of Directors. 1994. Standards for Medical Identifiers, Codes, and Messages Needed to Create an Efficient Computer-Stored Medical Record. Journal of the American Medical Informatics Association 1:1-7.
ASTM Committee E-31 on Healthcare Informatics. Group of standards available from ASTM, 100 Barr Harbor Drive, West Conshohocken, PA 19428-2959.
Barnett, G. O. 1984. The application of computer-based medical-record systems in ambulatory practice. New England Journal of Medicine 310:1643-1650.
Barnett, G. O., J. J. Cimino, J. A. Hupp, and E. P. Hoffer. 1987. DXplain: An evolving diagnostic decision-support system. Journal of the American Medical Association 258:67-74.
Barrows, Jr., R. C., and P. D. Clayton. 1996. Privacy, confidentiality, and electronic medical records. Journal of the American Medical Informatics Association 3:139-148.
Berner, E. S., J. R. Jackson, and J. Algina. 1996. Relationships among performance scores of four diagnostic decision support systems. Journal of the American Medical Informatics Association 3:208-215.
Chute, C. G., S. P. Cohen, K. E. Campbell, D. E. Oliver, and J. R. Campbell. 1996. The content coverage of clinical classifications. Journal of the American Medical Informatics Association 3:224-231.
Cimino, J. J., S. A. Socratous, and P. D. Clayton. 1995. Internet as clinical information system: Application development using the World Wide Web. Journal of the American Medical Association 2:273-284.
Classen, D. C., R. S. Evans, S. L. Pestotnik, S. D. Horn, R. L. Menlove, and J. P. Burke. 1992. The timing of prophylactic administration of antibiotics and the risk of surgical-wound infection. New England Journal of Medicine 326:281-286.
Classen, D. C., S. L. Pestotnik, R. S. Evans, and J. P. Burke. 1991. Computerized surveillance of adverse drug events in hospital patients [published erratum appears in Journal of the American Medical Association 267(14):1922, 1992] [see comments]. Journal of the American Medical Association 266:2847-2851.
Council on Competitiveness. 1996. Highway to Health: Transforming U.S. Health Care in the Information Age. Washington, D.C.: Council on Competitiveness.
Covell, D. G., G. C. Uman, and P. R. Manning. 1985. Information needs in office practice: Are they being met? Annals of Internal Medicine 103:596-599.
CPRI (Computer-based Patient Record Institute). 1995a. Guidelines for Establishing Information Security Policies at Organizations Using Computer-Based Patient Record Systems. Schaumburg, Ill.: CPRI.
CPRI. 1995b. Guidelines for Information Security Education Programs at Organizations Using Computer-Based Patient Record Systems. Schaumburg, Ill.: CPRI.
CPRI. 1995c. Proceedings of the First Annual Nicholas E. Davies CPR Recognition Symposium. Ed. E. B. Steen. Schaumburg, Ill.: CPRI.
CPRI. 1996a. CPR Systems Evaluation Work Group: CPR Project Evaluation Criteria (version 2.1). The Nicholas E. Davies Recognition Program. Schaumburg, Ill.: CPRI.
CPRI. 1996b. Guidelines for Managing Information Security Programs at Organizations Using Computer-Based Patient Record Systems. Schaumburg, Ill.: CPRI.
CPRI. 1996c. Proceedings of the Second Annual Nicholas E. Davies CPR Recognition Symposium. Ed. E. B. Steen. Schaumburg, Ill.: CPRI.
CPRI. 1996d. Sample Confidentiality Statements and Agreements for Organizations. Schaumburg, Ill.: CPRI.
CPRI. 1997. Proceedings of the Third Annual Nicholas E. Davies CPR Recognition Symposium. Ed. J. J. Teich. Schaumburg, Ill.: CPRI.
Detmer, D. E., and E. B. Steen. 1996. Shoring up protection of personal health data. Issues in Science and Technology 12:73-78.
Evans, R. S., J. P. Burke, D. C. Classen, R. M. Gardner, R. L. Menlove, K. M. Goodrich, L. E. Stevens, and S. L. Pestotnik. 1992. Computerized identification of patients at high risk for hospital-acquired infections. American Journal of Infectious Control 20:4-10.
Feldman, M. J., and G. O. Barnett. 1991. An approach to evaluating the accuracy of DXplain. Computational Methods and Programs in Biomedicine 35:261-266.
Grandia, L. D., T. A. Pryor, D. F. Willson, R. M. Gardner, P. J. Haug, S. M. Huff, B. R. Farr, and S. H. Lam. 1995. Building a computer-based patient record in an evolving integrated health system. Pp. 5-33 in Proceedings of the First Annual Nicholas E. Davies CPR Recognition Symposium, ed. E. B. Steen. Schaumburg, Ill.: CPRI.
Hammond, W. E. 1984. The Status of Healthcare Standards in the United States. Pp. 87-92 in Information Systems with Fading Boundaries, ed. W. E. Hammond, A. R. Bakker, and M. J. Ball. New York: Elsevier Science.
Health Level Seven (HL-7), Version 2.2. 1994. Health Level Seven, 3300 Washtenaw Ave., Suite 227, Ann Arbor, MI 48104-4250. 1994.
Hoechst Marion Roussel, Inc. 1995. HMO-PPO Digest.
Hripcsak, G., P. D. Clayton, T. A. Pryor, P. Haug, O. B. Wigertz, and J. Van der Lei. 1990. The Arden syntax for medical logic modules. Pp. 200-204 in Proceedings of the Fourteenth Symposium on Computer Applications in Medical Care. Washington, D.C.: IEEE Computer Society Press.
Humphreys, B. L., W. T. Hole, A. T. McCray, and J. M. Fitzmaurice. 1996. Planned NLM/AHCPR large-scale vocabulary test: Using UMLS technology to determine the extent to which controlled vocabularies cover terminology needed for health care and public health. Journal of the American Medical Informatics Association 3:281-287.
IEEE (Institute of Electrical and Electronics Engineers). 1995. Standards available from IEEE Standards Department, 445 Hoes Lane, PO Box 1331, Piscataway, NJ 08855-1331.
IOM (Institute of Medicine). 1993. Assessing Health Care Reform. Washington, D.C.: National Academy Press.
Johnson, M. E., K. B. Langton, R. B. Haynes, and A. Mathieu. 1994. Effects of computer-based clinical decision support systems on clinician performance and patient outcome: A critical appraisal. Annals of Internal Medicine 120:135-142.
Lindberg, D. A. B. 1995. The High-Performance Computing and Communications Program, the national information infrastructure, and health care. Journal of the American Medical Informatics Association 2:156-159.
McDonald, C. J. 1976. Protocol-based computer reminders, the quality of care, and the non-perfectibility of man. New England Journal of Medicine 295:1351-1355.
McDonald, C. J., S. L. Hui, D. M. Smith, et al. 1984. Reminders to physicians from an introspective computer medical record: A two-year randomized trial. Annals of Internal Medicine 100:130-138.
Miller, R., F. E. Masarie, and J. D. Myers. 1986. Quick medical reference (QMR) for diagnostic assistance. MD Computing 3:34-48.
NCPDP (National Council for Prescription Drug Programs). 1992. Standards available from NCPDP, 4201 North 24th St, Suite 365, Phoenix, AZ 85016.
NRC (National Research Council). 1996. The Unpredictable Uncertainty: Information Infrastructure Through 2000. Washington, D.C.: National Academy Press.
National Committee for Quality Assurance. 1997. A Road Map for Information Systems: Evolving Systems to Support Performance Measurement. National Committee for Quality Assurance.
Schoenbaum, S. C., and G. O. Barnett. 1992. Automated ambulatory medical records systems: An orphan technology. International Journal of Technical Assessment in Health Care 8:598-609.
Shortliffe, E. H., H. L. Bleich, C. G. Caine, D. R. Masys, and D. W. Simborg. 1996. The federal role in the health information infrastructure: A debate of the pros and cons of government intervention. Journal of the American Medical Informatics Association 3:249-257.
Sittig, D. F., and W. W. Stead. 1994. Computer-based physician order entry: The state of the art. Journal of the American Medical Informatics Association 1:108-123.
Sullivan, F., and E. Mitchell. 1995. Has general practitioner computing made a difference to patient care? A systematic review of published reports. British Medical Journal 311:848-852.
Tang, P. C., and V. L. Patel. 1994. Major issues in user interface design for health professional workstations: Summary and recommendations. International Journal of Biomedical Computing 34:139-148.
Tang, P. C., D. Fafchamps, and E. H. Shortliffe. 1994. Traditional medical records as a source of clinical data in the outpatient setting. Pp. 575-579 in Proceedings of the Eighteenth Symposium on Computer Applications in Medical Care, ed. J. G. Ozbolt. Washington, D.C.
Tang, P. C., M. A. Jaworski, C. A. Fellencer, N. Kreider, M. P. LaRosa, and W. C. Marquardt. 1996. Clinician information activities in diverse ambulatory care practices. Pp. 12-16 in Proceedings of the 1996 AMIA Annual Fall Symposium, ed. J. J. Cimino. Washington, D.C.: American Medical Informatics Association.
Tierney, W. M., C. J. McDonald, D. K. Martin, and M. P. Rogers. 1987. Computerized display of past test results. Effect on outpatient testing. Annals of Internal Medicine 107:569-574.
Tierney, W. M., M. E. Miller, and C. J. McDonald. 1990. The effect on test ordering of informing physicians of the charges for outpatient diagnostic tests [see comments]. New England Journal of Medicine 322:1499-1504.
Tierney, W. M., M. E. Miller, J. M. Overhage, and C. J. McDonald. 1993. Physician inpatient order writing on microcomputer workstations: Effects on resource utilization. Journal of the American Medical Association 269:379-383.
Tierney, W. M., J. M. Overhage, B. Y. Takesue, L. E. Harris, M. D. Murray, D. L. Vargo, and C. J. McDonald. 1995. Computerizing guidelines to improve care and patient outcomes: The example of heart failure. Journal of the American Medical Informatics Association 2:316-322.
A Progress Report on Computer-Based
Patient Records in Europe
Jan H. van Bemmel, Astrid M. van Ginneken, and Johan van der Lei
After more than two decades of pioneering efforts, in both the United States and Europe, a new era in the use of computer-stored patient data has clearly begun. The proliferation of inexpensive personal computers and workstations, the development of more user-friendly interfaces, the standardization of network technology, the original release of this report 6 years ago, and the stimulation of health care telematics (i.e., network communications) in Europe have contributed significantly to this growing interest. Equally important, developers of systems have acquired much experience since they first attempted to develop electronic patient records, and their subsequent attempts have become more successful and gained momentum.
In this commentary we discuss some recent developments in
computer-based patient records (CPRs) in Europe and review some CPR developments in
Europe that fall within the Framework Programs on Research and Development of
the European Union. Specifically, we review European developments in CPRs
for primary care, clinical care (i.e., hospital care), and shared care. We also
address the use of CPR data beyond patient care and two barriers to CPR
development: protection of patient confidentiality and standards development for data.
THE CURRENT STATE OF CPRs IN EUROPE
Computer-based information systems are now abundant in a large percentage of European hospitals as well as in primary care settings. Systems support general tasks such as patient administration, scheduling, and billing, as well as specific tasks such as in the clinical laboratory, radiology, or the catheterization laboratory. Hospital information systems are primarily administration oriented and seldom contain patient record data other than, for example, laboratory results, drug prescriptions, or diagnostic codes. Some systems store patient data, but none of them fully replaces the paper-based patient record. Many existing systems are not connected or are only loosely interconnected in networks. CPRs are being developed as part of a hospitalwide network (Degoulet et al., 1991; Van Mulligen et al., 1994; Scherrer et al., 1995) or as stand-alone or interconnected applications, such as in primary care (Westerhof et al., 1987; Branger et al., 1992; van der Lei et al., 1993). Increasingly, systems are being interconnected by electronic data interchange.
The field of hospital information systems in Europe consists, on the one hand, of mainly homemade systems (e.g., Bakker, 1984; Scherrer et al., 1995) sometimes supporting a group of hospitals and, on the other, of systems offered by industry, mainly of U.S. origin. A wide variety of information systems are operational on different platforms. Information systems in Europe reflect the differences in health care organizations which arise from the different historical developments, legislation, education, and languages among European countries. Therefore, systems developed in one European country are rarely installed in another country. This is also why the European hospital information system industry, with few exceptions, is virtually absent from the global market.
The differences in health care structures of different European countries might pose less of a problem for patient-centered approaches, such as CPRs. In all countries, patient care itself is quite similar. Europe has succeeded early at its attempts at CPR development and has since made considerable progress in the use of CPRs, especially in the primary care setting (van der Lei et al., 1991a; Millman et al., 1995). On the basis of these successes, several collaborative projects were started in health care; these were financially subsidized in part by the European Union (EU, 1994). Three developments related to CPRs in Europe deserve attention: the use of CPRs in the primary care setting, in hospitals, and for shared care (i.e., collaboration between care providers). In addition, in the area of standardization in health care, the European Union plays a pivotal role on a global scale as well (De Moor, 1993).
Network technology and communication (in Europe the latter is called telematics) are now prominent developments in information technology and have a large impact on health care (Branger et al., 1992; EU, 1994). By using standard communication networks and standard software (De Moor, 1993), data interchange between the four levels of health care delivery (i.e., the region, the institution, the clinical department or outpatient clinic, and the individual physician, nurse, or patient) is more efficient. A fully operational exchange of patient data between systems, with proper authorization, is one of the present challenges in European health care.
The data in CPRs are not begging to be used for electronic data exchange, research, and shared care. The use of CPRs for purposes other than direct patient care has been associated with several problems (van der Lei et al., 1991a; Vlug and van der Lei, 1995). These limitations are, for instance, the lack of a common terminology for patient data collection, or the use of different modes for the CPR systems. As the knowledge gained from experience accumulates, CPR developers learn important lessons, particularly the need to develop structured patient records based on a clear conceptual model. If CPR systems are not based on a conceptual model, and if the data in CPRs are not well structured, it will not be possible to use CPR data for different goals, nor will it be possible for such data to be exchanged between health care providers to support shared care. Therefore, if CPR systems are to be used for these other goals, care providers should realize that it is not sufficient that the systems merely support their own practice or clinic, but they should take care that patient data can be exchanged over the entire health care domain and be used, for instance, to support clinical research, decision-making, the assessment of the quality of care, or management and planning.
A variety of CPR applications and projects have been completed or are ongoing including those supported through Framework Programs on Research and Development of the European Union. They concern computer-based records for primary care, in hospitals, and for shared care, as well as the use of data in CPRs for research, assessment of the quality of care, and decision support. Other projects focus on what the Institute of Medicine (IOM) Committee on Improving the Patient Record classified as barriers to CPR development: privacy and confidentiality and standardization of terminologies and data-exchange formats.
Several developments related to CPRs in Europe are discussed below,
including the status of CPRs in the primary care settings, in hospitals, and in
shared care (i.e., collaboration between care providers). The final sections consider
more specific uses of CPR data and remaining barriers to wider use of the
computer-based patient record.
CPR SYSTEMS IN PRIMARY CARE
In some European countries, especially The Netherlands (van der Lei et al., 1993) and the United Kingdom (Millman et al., 1995), there has been a rapid increase in the use of information systems by general practitioners (GPs) in primary care (Figure 1). CPRs are incorporated into these information systems and are quickly replacing existing paper-based records. The use of CPRs by GPs in particular is rapidly increasing. At the beginning of 1996 it was estimated that more than 90 percent of all GPs in The Netherlands and the United Kingdom used information systems in their practices. About 50 percent of Dutch GPs had implemented a CPR system for the support of patient care and about 25 percent of GPs had a paperless office.
Similar figures apply to the United Kingdom (Millman et al., 1995). This
is directly related to the fact that in both the United Kingdom and The
Netherlands, GPs are the "gatekeepers" in health care. Most GPs run a practice by
themselves or with a small number of colleagues. In contrast, specialists work in large
institutions in more complex settings, involving many more departments and
personnel. Furthermore, GPs keep less extensive records so that data entry is less
time-consuming and less detailed.
FIGURE 1 Growth of installed information systems and computer-based patient records in primary care in The Netherlands (100% represents 6,500 general practitioners [GPs]) and England and Wales (where 100% represents 9,000 GPs) as a function of time. The values for 1996 are extrapolations. Not indicated is the growing number of primary care systems that are electronically connected to other health care systems (see also the text and Figure 3).
The actual situation in The Netherlands offers a good example of both
the success and the potential use of CPRs (van der Lei et al., 1993). This can
be attributed to four factors: (1) the role of GPs, (2) physician training, (3) the
structure of health care, and (4) population-based care.
Role of GPs
From the outset, professional organizations of GPs have played an
active role in setting guidelines for information systems in general practice and in
assessing the systems available in the market. This active role was started in
the early 1980s, when they identified the basic needs of GPs for using systems
in their practices and set the first broad guidelines for systems tailored
specifically for primary care. Furthermore, these organizations announced to industry
that they would recommend to their members that they use only those systems
that would meet their requirements. In this way, a de facto standard was
established. Since then, systems requirements are refined and extended every 2 to 3 years,
and vendors must submit their systems for reevaluation by professional
organizations. This iterative way of refining guidelines and assessing systems gives
potential buyers of systems a basis on which to judge the available information systems.
It also gives vendors clear-cut standards for their products. During the last 15
years, the definition of standards started with requirements for patient
administration and evolved into standards for diagnostic coding (in all systems the
International Classification for Primary Care has been implemented [Lamberts and
Wood, 1987]), drug prescriptions (all systems use the same national thesaurus for
drugs), and the type of data to be contained in a basic CPR and requirements for
electronically interchanging patient data between different systems in the
primary care setting, that is, from one GP's system to that of another.
Training
The success of information systems is largely determined by physicians' expectations. Education at medical faculties of universities prepares physicians in the use of computer-based information systems and patient records. Early on, students get acquainted with how to use computers for the documentation of patient data in CPRs. Professional organizations have also ensured that postgraduate training of GPs conveys a realistic set of expectations. These expectations must be realistic, because if they are too high they lead to disillusionment, and if they are too low they may impede the introduction of systems. The training emphasizes that the proper use of systems and CPRs reduces administrative work-load, increases the accuracy of billing (no accountable patient data are lost), and enables the production of referral letters, the use of electronic data interchange, and the tracking of certain categories of patients.
Potential users should realize that, certainly at the beginning,
investments must be made. The investments must be expressed not only in financial terms
but also in time and personal efforts. For instance, for a typical practice of
about 2,300 patients, it takes about 2 years for all patient data to be entered into
the system's CPR. This data transcription from paper to computer is most often
done at the time that patients come for a consultation, since the majority of patients
pay a visit to their GP once every 2 years. A GP who is just starting to use a
CPR system needs an extra 2 to 3 minutes during the consultation. The GP's
assistant enters data related to all administrative operations, such as patient admission
and patient scheduling, but the GP enters the patient care-related data into the CPR.
Structure of Health Care
The factors described above are related to the role and training of GPs
in using CPRs and may be useful for other countries. A third factor, however,
is strongly related to the structure of health care in The Netherlands, which in
many ways is comparable to that in the United Kingdom and Scandinavian
countries. Patients in The Netherlands and the United Kingdom have a single GP who
coordinates their health care, acting as a gatekeeper to other specialists; the
latter report back to the GP. Several projects are now under way to use electronic
data interchange for this GP-specialist communication (see the section on "CPR
Systems for Shared Care," page 32). Thus, in principle, the GP coordinates the
health care-related data for a single patient over the patient's lifetime and is
requested by other specialists to share that information. The GP's role as coordinator
eases the introduction of CPR systems because the demands of care are easier to
meet when the data are available in electronic form. This factor may be
increasingly relevant in the United States as managed care and capitated reimbursement
continue to penetrate the health care delivery market.
Population-Based Care
A fourth reason for the success of the acceptance of CPRs in the primary
care setting is the fact that Dutch general practice is population-based care, in
contrast to institution- or physician-based care. The Dutch primary care system can
be characterized as providing services that are responsive to the needs both of
the individual patients and the population at large. Such a population-based
orientation creates demands that are easier to satisfy when CPRs are introduced.
For instance, CPRs with properly coded data greatly facilitate the creation of a
gender-age register to report on statistics in primary care or active case finding
(e.g., for periodic cervical smears or cardiovascular risk assessments). The
availability of CPRs facilitates studies on the use and effects of drugs in the population
and postmarketing drug surveillance (see the section on "Use of CPR Data,"
page 34).
CPR SYSTEMS IN HOSPITALS
Electronic information technology has been introduced in all hospitals in Europe. This is no longer restricted to special areas of care, such as radiology or the laboratory. In principle, the available technology allows central and monolithic hospital information systems to be transformed into hospitalwide networks with "intelligent" workstations or personal computers throughout the hospital, from the consultation room to close to the bedside. This transformation is just beginning, however. At present, graphical workstations and network technology are the latest developments in the computer industry. Client-server protocols ease the integration of systems and distribution of processing tasks (Degoulet et al., 1991; Van Mulligen et al., 1994; Nowlan et al., 1994). In principle, workstations offer an environment in which, from the user's perspective, all applications running either locally or through a network act as a single integrated system. Several institutions are developing health care workstations that enable the integration of patient data scattered throughout different systems (Degoulet et al., 1991; Van Mulligen et al., 1994; van Bemmel et al., 1996a). In principle, such systems offer to the clinician on one computer screen patient data, images, and biological signals, regardless of the computer on which they are stored or the system on which they are processed.
Yet, although the technical issues involved in the integration of various
systems were slowly being solved, the absence of a conceptual model for a CPR
was increasingly becoming a stumbling block, particularly as the amount of
available data increased. Thus emerged a trend in which a technical infrastructure
(the platform for the CPR) takes care of the communication with the numerous
systems available in the hospital, including legacy systems, such as a
laboratory system, a picture archiving and communication system (PACS), or an
electrocardiogram (ECG) interpretation system. CPRs are developed at the core of
this integration platform. This distinction is not trivial. When developing the
platforms, questions such as how an existing legacy system should be embedded
are addressed on a technical level. On the level of CPR development, questions
relate, for example, to how the data will be presented to the user in a
consistent manner or what additional data the physician must record are addressed.
Platforms for CPRs
Both the Helios (Degoulet at al., 1991) and the Hermes (Van Mulligen et al., 1994) integrated clinical workstations offer an environment for systems integration. Communication between applications is automatically performed by the workstation and follows ISO1 standards (ISO, 1987). Instead of having one large central database of patient data, all existing databases and applications are left unchanged and patient data are combined by the workstation on demand. This has two advantages over one large integrated database. First, data can remain distributed and stored at places that are most convenient (i.e., close to where the data are collected and used). Second, commercially available applications can be used for data processing without modifying them.
The Hermes workstation was initially developed to support clinical data analysis, but its architecture has also been extended to support patient care. It is integrated with the generic CPR system ORCA2 (van Ginneken, 1996). Hermes is also able to integrate with legacy systems by encapsulation.
The Helios workstation, similar to Hermes in many respects, also
integrates alphanumeric data (text), images, and biological signals. The Helios platform
is being used as one of the building blocks of a new generation of hospital
information systems. The Hermes approach, originally developed on a Unix
platform, has now migrated to a Windows NT/95 environment, forming the basis for
the European project I4C (van Bemmel et al., 1996), as explained below.
Integrated CPRs
As an example of the introduction of a CPR in a clinical department, we briefly describe the European project I4C (Integration and Communication for the Continuity of Cardiac Care). Although I4C is intended to support cardiology, it uses a generic CPR, ORCA, which is also applicable to other departments. In this way, the CPR used in the I4C project can be used by many other clinical departments (van Ginneken et al., 1995; van Ginnekin and Stam, 1995), because it contains a variety of patient data: patient history, physical examination findings, laboratory test results, images, biological signals, and so forth. Real-time data, such as from the catheterization laboratory, are also intended to be incorporated into the CPR. Data are meant to be used by more than one physician (e.g., the cardiologist often must coordinate his or her treatment with that of other specialists in the hospital or with GPs in primary care).
Figure 2 presents the CPR structure that forms the basis for the I4C project, together with its branches to other systems in the hospital or elsewhere. Patient record data, as well as images such as echocardiograms and coronary angiograms, or signals such as ECGs or blood pressure curves are collected, stored, and presented by using the workstation.
The I4C project in principle is intended to support cardiac care, from prevention and treatment to follow-up and rehabilitation. This is accomplished by: (a) consistently recording patient data in the CPR; (b) accessing diagnostic and therapeutic information; (c) accessing patient data, images, and biological signals wherever they are stored and irrespective of where the inquiry originates; and (d) collecting well-defined reference sets of patient data.
Data are to be used not only for patient care, research, and education but
also for the assessment of the quality of care and for management and planning.
The fourth issue is similar to the data collection that resulted from the European
CSE3 project that preceded I4C (Willems et al., 1991). The project is carried out
by centers in The Netherlands—which provides coordination—the United
Kingdom, France, Germany, Italy, and Greece.
CPR Data Entry
For a CPR to have benefits, the data in that record need to be structured and coded, at least to a certain degree. Obtaining structured and coded data, however, has proved to be a significant hurdle. Physicians are accustomed to paper records. How physicians use these records has been the subject of several European projects (e.g., Nygren and Henriksson, 1992). From such investigations it became apparent that physicians are far more likely to use computers to consult the data in present patient records than to enter patient data. This is not surprising: Data entry requires more effort from the physician than browsing through the record. Some researchers have argued that direct input by physicians will have to wait for radical improvements of the human-machine interface, such as speech input. Experience in a number of countries where GPs are particularly using CPRs (e.g., The Netherlands, the United Kingdom, and Sweden) has shown, however, that physicians will accept the limitations of the current human-machine interaction.
Views of CPR data for patient care, decision support, scientific data
analysis, or assessment of quality of care require patient data to be highly structured
and unambiguous. It is difficult to fulfill these requirements with narrative
(textual) data (e.g., Moorman et al., 1994). Ideally, therefore, patient data should be
acquired directly from the physician in a structured format and, preferably,
should also be entered by the physician. To accomplish this, different approaches to
data entry have received attention: form-driven data entry, natural language
processing, and structured data entry. The last two methods are also the focus of
several research projects in Europe (Nowlan and Rector, 1991; Baud et al.,
1992; Moorman et al., 1994).
Natural Language Processing
Natural language processing intends to automatically extract coded
medical data from free text. European research in this area has mainly been done by
Baud and colleagues in Geneva (1992). The basic advantage of natural language
processing is that physicians do not have to alter the way in which they express
their findings or document their decisions. Only when its application domain is
strictly confined can natural language processing offer advantages. A fundamental
disadvantage of natural language processing is that the data-capturing process
itself cannot be influenced and improved. Data that the physician has not written
or spoken remain unknown. Freedom of expression is inherent to free text
(Moorman et al., 1994). Therefore, it is difficult to impose structure on data that are used
as input for natural language processing.
FIGURE 2 Example of the integration of a computer-based patient record (CPR) (ORCA; van Ginneken, 1996) in a clinical network of information systems being developed in the European project I4C (van Bemmel et al., 1996), in which centers from six countries collaborate. At the core of the CPR (called the client here) are three generic core components: the state transition manager, the context manager, and knowledge-driven data entry. The first two deal with all data acquisition, storage, and presentation and communication with the application components of the CPR system and the legacy systems elsewhere in the hospital. Communication is controlled by a brokerage layer and a communication layer, which are part of a group of servers. The broker follows protocols such as those agreed upon by the Common Object Request Broker Architecture (CORBA) Group (OMG, 1996); the communication layer provides data exchange via Electronic Data Interchange for Finance, Administration, Commerce, and Trade (EDIFACT) or HL-7. In the data repository, CPR data and knowledge bases are stored, together with coding systems and thesauruses for diagnoses and drugs. The six systems at the bottom are examples of "legacy systems" from which data (e.g., from a departmental or hospital information system), images (e.g., from an angiogram server), or biological signals (e.g., from an ECG management system) are to be transported to the CPR for presentation or processing. Integration with the legacy systems is provided by encapsulation (Van Mulligen et al., 1994).
Structured Data Entry
Another way of data entry is context sensitive and is adaptable to different clinical domains. It is called structured data entry and consists of forms whose content is knowledge driven. In a structured data entry-oriented European project, the forms are structured, but their contents can continuously be adapted to accommodate the user's requirements and personal preferences (Moorman et al., 1994). In addition to a knowledge model, it contains a predefined vocabulary and specifications on how the terms of the vocabulary may be combined into meaningful expressions. The knowledge base keeps track of the individual physician's definitions of expressions with a nontrivial meaning, such as "the heart is normal" or "gastroenteritis complaints." When such expressions are used for the first time, the system will prompt the physician-user for an explicit description. Upon later use of such expressions, the system will store its explicit substitute. In this way, efficiency and completeness can be reconciled.
Structured data entry is also an efficient tool for enhancing data
completeness and testing for data reliability (Whiting-O'Keefe et al., 1985; Wyatt,
1994). Graphics and voice input may also serve to enhance the capabilities of
structured data entry.
CPR SYSTEMS FOR SHARED CARE
An environment in which patient record data can be electronically interchanged regionally (Figure 3) is a necessary condition for a shared care environment (Branger et al., 1992, 1994; Branger and Duisterhout, 1995). Local and regional networks are increasingly used for the exchange of patient data between health care institutions. They are also used for communication with, for example, health insurance bodies, governmental organizations, libraries, and research institutions. Orders and payments are exchanged by electronic data interchange, databases are consulted remotely, and laboratory data or referral letters are mailed electronically. Communication between specialists and GPs is also increasingly taking place via telecommunication. In some instances specialists and general practitioners communicate directly with a patient in the patient's home, for example, patients with pacemakers, undergoing hemodialysis, or receiving palliative care. For some purposes, remote medical consultation (telemedicine) is also envisaged. Several regional shared-care networks are under development in Europe within the Framework Programs for Research and Development in health care telematics, as will be discussed below. Such regional networks are a logical extension of networks within hospitals.
With the existence of a computer-based shared care environment, new applications for patient-centered care have arisen, such as:
- 1. GPs are able to send a patient's data to the hospital to obtain results
or diagnostic advice instead of referring the patient.
2. Patients' hospital stays are shorter and specialists may offer teleconsultation services.
3. Outpatient clinics increasingly function as diagnostic screening and ambulatory care centers. Patients are referred to these clinics only when specific diagnostic tests or interventions must be performed.
4. Patient admission and scheduling are greatly facilitated, resulting in better planning and decreased waiting times for examinations.
In the next two sections we provide two examples of European projects
involving the use of CPR data for shared care. The first describes the project
CoCo, concerning the coordination and continuity of health care (Bernstein and
Jensen, 1994). The second describes a project that deals with the exchange of data
in health care by smart cards, in this case, Diabcard (Schaefer and Sembritzki, 1996).
FIGURE 3 Regional electronic data-interchange network to support shared care. General practitioners (GPs) in primary care are interconnected with other health care providers, such as the clinical laboratory, pharmacies, pathology laboratories, radiology departments, and departments in the hospital. Usually, such an electronic data-interchange network makes use of public telephone lines or special lines (e.g., ISDN). In Europe, growing numbers of such networks are in development as part of European projects, such as the CoCo project (see text).
CoCo
The intention of the European project CoCo (Coordination and Continuity in Health Care) is to establish large-scale interconnected regional health care information networks, with the purpose of improving the coordination and continuity of health care (Bernstein and Jensen, 1994; van Bemmel et al., 1996). It uses available CPR-based systems and telecommunication infrastructures in the participating countries, European EDIFACT4 standards, and electronic mail techniques. The project will not interfere with regional regulations and will not impose new rules on the users of the networks. A major effort is dedicated to ensuring that the project will be implemented in a coordinated and standard way. The practical guidelines and the experiences with the collaboration within CoCo should also be usable for other applications in Europe. CoCo has two goals:
- 1. Establishment of networks for the regional exchange of health care
information. Each region will establish running networks that will allow for
communication between all the major actors in health care: GPs, laboratories,
radiology departments, hospitals, pharmacies, regional and national authorities, and
social services and home care providers.
2. Establishment of links between the regions through CoCo Link. Coco Link ensures coordination between the regions to facilitate a common methodology, to start concerted developments, and to use standardized communication in all regions. For instance, it will provide guidelines, it is responsible for developing an electronic test bed that can be used to test EDIFACT solutions, and it will also establish networks between networks.
Many countries within the European Union, such as Denmark
(coordination), The Netherlands, the United Kingdom, Ireland, Spain, and Greece,
participate in CoCo. There is also an extension to Canada. A typical application
within CoCo is shared care provided to patients with breast cancer. In this setting, GPs
in primary care, internists, surgeons, and radiotherapists in secondary care
collaborate with nurses in home care, covering the entire period from the first
patient visit to the GP up to chemotherapy or surgery and, eventually, palliative care.
All care providers have access to the same CPR; nurses in the home care setting
use portable computers to exchange data with the system in the clinic.
Diabcard
In Germany, health insurance cards are regulated by law. During 1993
and 1994, more than 70 million smart cards (chip cards) were issued to German
citizens, and about 165,000 read-write stations were installed in, for instance,
practices, hospitals, and emergency rooms. The insurance cards are primarily used
for administrative purposes, but debate is ongoing about the use of these smart
cards as carriers of patient data. The German project Diabcard, which has also
been extended to other European countries, uses smart cards for the storage and
exchange of data for patients suffering from diabetes (Schaefer and
Sembritzki, 1996). Other projects, such as for patients with cancer, are also being
introduced in the German health care system. Standardization of the health care data on
such cards is of utmost importance.
USE OF CPR DATA
Once patient data are available from CPRs in primary care centers and
hospitals, the data can be harvested. We briefly discuss here a project dealing
with critiquing a physician's care (van der Lei et al., 1991; van der Lei and
Musen, 1991; van Bemmel, 1993). Another European project intends to integrate
treatment protocols with CPR systems. It is oriented toward the support of
clinical oncology. The third project deals with the assessment of care by using large
collections of CPR data harvested from operational CPR systems in the primary
care setting. This project is directed toward the postmarketing surveillance of
drugs (Vlug and van der Lei, 1995).
Critiquing a Physician's Care
Automated review of CPRs is expected to limit errors in the delivery of care and to control costs. Two projects, carried out with GPs in The Netherlands, have investigated whether CPRs contain sufficient information to support direct care by assessing CPR data and generating critiques and decision support. In the first project (HyperCritic [van der Lei et al., 1991; van der Lei and Musen, 1991]), a system of critiquing the treatment of hypertension was developed. The second project (AsthmaCritic [Kuilboer and van der Lei, 1997]) covers the area of chronic obstructive respiratory diseases. Both systems rely on CPRs for patient data. HyperCritic was assessed in a research project, and we summarize here some of the main outcomes (van der Lei et al., 1991).
The critiquing systems generate comments in two stages. First, CPR data are automatically interpreted to review the actions of the GP at a given visit (e.g., starting a new drug, continuing treatment with a drug, or replacing one drug with another). Second, each action is assessed. The system searches the CPR for conditions that contraindicate that action (e.g., contraindications to specific drugs), determines whether the preparations required for the action have been performed, determines whether the GP has performed the routine monitoring required by the action, and searches for any undesirable condition that might have resulted from the action. Review of the CPR requires detailed knowledge of drugs, such as customary dosages, contraindications, side effects, interactions, workup requirements, and criteria for judging the efficacy of the treatment. Computer-based critiquing of patient care was compared with peer review-based critiquing in the area of mild hypertension. The comparisons revealed the following:
- 1. Automated assessment of CPR data by the computer-based critiquing
system could compete successfully with peer review.
2. The system was not able to reproduce some comments of experts. This was the result of insufficient CPR data, the lack of sufficient medical consensus, or omissions in the knowledge base of the critiquing system.
On the basis of the outcomes of this and other studies reported in the literature, the following can be concluded:
1. Systems for the assessment of patient care should be able to acquire patient data automatically from the CPR.
2. Ideally, the care assessment system should be functionally integrated with the CPR system.
INTEGRATION OF PROTOCOLS WITH CPR SYSTEMS
In many patient care areas, protocols are used to increase both the efficiency and the quality of care. In principle, such protocols are not limited to certain hospitals, regions, or countries. This was the reason for starting a large European project on the use of protocols in health care (both in the clinic and in the primary care setting) called Prompt (Lagouarde et al., 1996; van Bemmel et al., 1996), which was based on the earlier European project Dilemma (Thomson, 1995). The project has been established by centers from the United Kingdom (which performs the coordination), France, and The Netherlands and intends to integrate these protocols with CPRs and to deliver and evaluate a set of tools for supporting the use of protocols. The initial focus of the project is oncology; this is now followed by a focus on the primary care and shared care settings.
The clinical objective is to install systems for supporting the treatment of cancer in specialty hospitals in the participating countries. Applications will be based on state-of-the-art therapeutic protocols. By the end of the project it is anticipated that the technology should be ready to be installed in centers in other countries as well. The following are other objectives of Prompt:
- 1. extensions to each hospital system, to support GPs, interconnected by
regional networks for shared care;
2. initiation of comprehensive studies of the effects of the protocols on clinical outcomes; and
3. generic versions of the protocols for other clinical disciplines.
Review of CPR Databases
It is expected that CPRs will increasingly be used to improve the quality of care in practices, departments, and institutions as well as, for instance, to support postmarketing surveillance studies. Pressure from law enforcement bodies, third-party payers, peer-review organizations, hospitals, and physicians and patients themselves may lead to this use of automated review of CPRs. We briefly discuss a large ongoing project on the assessment of drug prescriptions in the population (Vlug and van der Lei, 1995; Visser et al., 1996). The project that is reported runs in The Netherlands, but similar projects have been started in other European countries (e.g., Germany and the United Kingdom).
Postmarketing surveillance examines the beneficial and adverse side effects of drugs on human health from the time that these drugs are marketed. Post-marketing surveillance consists of two stages: a hypothesis generation stage in which an effect or side effect is suspected and a hypothesis verification stage in which the hypothesis is tested. Hypothesis generation is typically based on the spontaneous reporting of potential side effects by physicians. Studies have indicated, however, that this spontaneous reporting leaves much to be desired. For instance, Classen and colleagues (1991) studied the prescription of drugs in cases in which the drug prescription was changed, increased, or stopped. It was shown that computerized monitoring of adverse drug events in hospitals by using computer-stored patient data offers many advantages over the voluntary reporting of such events.
An extensive project that analyzes prescriptions in primary care to
support the postmarketing surveillance of drugs is under way in The Netherlands. To
test a hypothesis, large populations need to be monitored. Reports of patient
contacts with GPs are anonymously sent to a research database. The data in that
database are subsequently used to conduct postmarketing surveillance studies (Vlug
and van der Lei, 1995). The Dutch authorities responsible for drug safety are
investigating the potential use of this type of database. Preliminary results indicate
that the CPRs can be used as a way to track large numbers of patients and that use
of CPRs will allow researchers to build longitudinal databases (Visser et al.,
1996). At the end of 1996, about 250,000 patients were followed longitudinally;
this number will increase to about 1.5 million in the next few years.
ADDRESSING BARRIERS
Privacy and Confidentiality
Now that CPR data are becoming available for the different goals mentioned in the introduction to this commentary, it is of great importance to protect these data, to guard the privacy of patients, and to protect the professional interests of health care providers (van Bemmel, 1992; Barber et al., 1996). Many parties (employers, insurance companies, etc.) are interested in permanent patient data,5 and these data should be extremely well protected in CPR systems. In Europe there may be more sensitivity toward the improper use of patient data than elsewhere; this sensitivity could even impede the introduction of CPRs in some countries, and for some purposes could hamper the use of CPR data for goals other than patient care. In the different European countries, privacy laws control health care data, but privacy laws also exist in Europe as a whole. European law supersedes the laws of the individual countries. The right to privacy has been anchored in the Treaty for the Protection of Human Rights and Fundamental Freedom (Treaty of Rome, 1954) and in European law (EU, 1995). Yet, it will be difficult to totally prevent the improper use of data stored in CPRs.
Because modern health care often requires shared care instead of care by a single physician only, the individual physician is no longer capable of personally guaranteeing the privacy of a patient's data. For that reason, after implementing regulations related to ensure the secrecy of patient data, modern societies must ensure the right to privacy. This means that in the different countries written regulations related to requests for personal data, including patient data, are required. These regulations should be supervised by a privacy committee and should contain descriptions of the purpose of the data registration that is maintained. It also should describe when data can be provided to third parties, and the right of all persons concerned to inspect, alter, and destroy their own data. In principle, these regulations do not concern data that have been made anonymous.
It is of utmost importance that in all (future) CPR systems proper
measures be taken to protect the privacy of patients and their data. This issue is also
the subject of discussions at the EU level, for example, in the standardization
committees established for health care informatics and telematics. Some of the
standardization activities related to CPRs are summarized in the following section.
Standardization
As discussed above, it is essential that standards for the acquisition, storage, and exchange of patient data be available. In health care this pertains to both administrative data and patient data. On the basis of the experiences over the past decade, a technical committee (TC 251 [De Moor and van Maele, 1996]) falling under CEN (Comité Europeén de Normalisation, the "European Committee for Standardization") was established. This committee defines preliminary standards and compiles reports on the standards to be introduced in the different countries of the European Union (De Moor, 1993; De Moor and van Maele, 1996). For instance, the ability to link systems via standard interfaces is widely recognized as a necessity. Especially in Europe, where data cross regional and national boundaries, agreement on how to exchange patient data is urgent. Directly and indirectly, such standards will contribute to the quality and efficiency of patient care.
From the long list of preliminary standards that have been or are being
elaborated by Technical Committee 251, a few examples specifically dealing with
data in CPRs are provided in the following sections (De Moor, 1996).
Electronic Health Care Record Architecture
Work on defining the basic architecture for data contained in an
electronic health care record has begun. Standardization is necessary if patient data are to
be exchanged between different health care professionals. It is also necessary so
that patients can travel or move from one region or country in Europe to another.
The architecture should enable physicians to use CPRs that are adapted to their
own requirements and should therefore support a variety of formats. The aim of
this standardization effort is not to specify the CPR system as such but to bring
structure to the data stored in such records.
Coding Systems for Drugs
Standardization of coding systems for drugs concerns prescribing drugs
and the effects of drugs on patients. A conceptual model for the correct labeling
of drugs is foreseen within Technical Committee 251. The standard covers the
identification of generic drug products as well as the drugs themselves. A
coding system itself, however, is not intended.
Syntax for Health Care Data Interchange
Electronic messages in health care can be coded in many different
ways. Therefore, consensus is required on the exchange formats for health care
data between different systems. This has many advantages, because the systems'
vendors can then handle the different data formats. The standard intends to
define and adopt syntaxes for this exchange of information.
Exchange of Medical Images and Related Data
The medical image exchange standard concerns the exchange of
high-volume data such as images or similar data. The Technical Committee 251
standard has adopted the industrial
DICOM6 3.0 standard for medical images,
enabling the possibility of standardized storage and global exchange of such images,
which is important for supporting, for example, teleradiology or similar services.
User Identification
As open systems are gradually installed in hospitals and regional
networks become operational, it will be important to prevent patient data from being
used illegally. To ensure the proper use of patient data, it is necessary to
positively identify authorized users. Software systems should therefore make use of
automatic procedures for user identification, for example, by the use of
passwords and different technical measures. Smart cards are other means of positive
user identification (e.g., fingerprints or face recognition are possible techniques
that could be standardized and that may eventually replace the more vulnerable
passwords).
CONCLUSION
Significant progress has been made in Europe toward the primary recommendation of the IOM Committee on Improving the Patient Record's report over the past 6 years. Several factors account for the success to date. First, in those countries where CPRs are most prevalent, the health care delivery system or structure relies on a gatekeeper model and emphasizes primary health care and the health of the population. Thus, there is a recognized need for individual and aggregate patient data and a willingness to invest at both macro- (i.e., government) and microlevels (i.e., a practitioner) to make such data available. Second, CPR development projects are ideal for the EU insofar as they offer benefits both within and across countries and collaboration in their development may well speed up visible success. The evidence of political support and support by the medical professional organizations for CPRs sends a strong message to system developers and users, as well as to the general public.
Third, the emerging focus on conceptual models that build upon rather than are driven by the technology is a significant breakthrough in CPR development and will likely result in systems that can be adopted throughout Europe. Ultimately, this conceptual focus, which is also evident in the United States, may increase the ability of generic CPR systems to be transported globally. Fourth, collaboration between end-users and developers in system specification is a critical element of a sound CPR development process, aiding both the developers and users of the systems. Finally, the readiness of users with realistic expectations eases the implementation and increases the likelihood that CPRs will be used to their full potential. Thus, the education of health professionals is an important strategy in speeding CPR diffusion.
Like in the United States, important challenges remain to be addressed
for CPR developers. First, continued support of CPR development efforts by
supra-regional and national organizations is absolutely essential. Second, the issues
surrounding the operational exchange of patient data must be fully explored
and addressed. Third, protecting the confidentiality of patient data in the
information age is a global issue that must be a priority in all countries and will
ultimately require greater collaboration among countries as the technologies that
support CPRs continue to evolve and are exploited in ways that we have not yet
identified. Fourth, flexible, user-adaptable generic models for the generation of
CPRs are needed to overcome the great variety in specialized care, the differences
in culture among various clinics, and particularly in Europe, among countries.
REFERENCES
ACR-NEMA (American College of Radiology and National Electrical Manufacturers Association). 1985. Digital Imaging and Communications Standard 300-1985. Washington, D.C.: NEMA Standards Publications.
Baud, R.H., A.M. Rassinoux, and J.-R. Scherrer. 1992. Natural langugage processing and semantical representation of medical texts. Methods of Information in Medicine 31(2):117-125.
Bakker, A. R. 1984. The development of an integrated and cooperative hospital information system. Medical Informatics 9:135-142.
Barber, B., A. Treacher, and K. Louwerse. 1996. Toward Security in Medical Telematics. Studies in Health Technology and Informatics, No. 27. Amsterdam: IOS Press.
Bernstein, K., and H. B. Jensen. 1994. Managing change in developing communication systems for integrated care; the Funen model. Pp. 266-269 in Proceedings of Medical Informatics Europe 1994, eds. P. Barahona, M. Veloso, and J. Bryant. Lisbon: Medical Informatics Europe.
Branger, P. J., J. C. van der Wouden, B. R. Schudel, E. Verboog, J. S. Duisterhout, J. van der Lei, and J. H. van Bemmel. 1992. Electronic communication between providers of primary and secondary care. British Medical Journal 305:1068-1070.
Branger, P. J., A. van't Hooft, J. S. Duisterhout, and J. van der Lei. 1994. A standardized message for supporting shared care. Pp. 473-477 in Proceedings of the 18th Annual Symposium for Computer Applications in Medical Care, ed. J. G. Ozbolt. (Special issue of the Journal of the American Medical Informatics Association 1994:473-477.)
Branger, P. J., and J .S. Duisterhout. 1995. Communication in health care. Methods of Information in Medicine 34:244-252.
Classen, D. C., S. L. Pestotnik, R. S. Evans, and J. P. Burke. 1991. Computerized surveillance of adverse drug events in hospital patients. Journal of the American Medical Association 266:2847-2851.
Degoulet, P., J. Coignard, F. C. Jean, et al. 1991. The Helios European project on software engineering. Pp. 125-137 in Software Engineering in Medical Informatics, eds. T. Timmers and B. I. Blum. Amsterdam: Elsevier Science.
De Moor, G. J. E. 1993. Standardization in medical informatics. Pp. 61-66 in 1993 Yearbook of Medical Informatics, eds. J. H. van Bemmel and A. T. McCray. Stuttgart: Schattauer Verlag.
De Moor, G. J. E., and G. van Maele, eds. 1996. Standards in Healthcare Informatics and Telematics. Brussels: CEN/TC 251.
EU (European Union). 1994. European Union DG XIII Work Programme on Health Telematics. Brussels: Commission of the European Communities.
EU. 1995. European Union Directive 95/46/EC of the European Parliament and of the Council of 24 October 1995 "On the Protection of Individuals with Regard to the Processing of Personal Data and on the Free Movement of such Data." Journal of the European Communities 38:L281/31-50.
ISO (International Standards Organization). 1987. Electronic Data Interchange for Administration, Commerce and Transport (EDIFACT), Application Level Syntax Rules, IS 9735.
Kuilboer, M. M., and J. van der Lei. 1997. Exploring the role of a critiquing system: A simulation. Journal of the American Medical Association in press.
Lagouarde, P., R. Thomson, J.-L. Renaud-Salis, et al. 1996. The PROMPT electronic health care record. In Proceedings of Medical Informatics in Europe 1996, ed. P. McNair. Amsterdam: IOS Press.
Lamberts, H., and M. Wood, eds. 1987. International Classification of Primary Care. Oxford, England: Oxford University Press.
Millman, A., N. Lee, and A. Brooke. 1995. Computers in General Practice. Parts I, II, and III. British Medical Journal 311:800-802, 864-867, and 938-941.
Moorman, P. W., A. M. van Ginneken, J. van der Lei, and J. H. van Bemmel. 1994. A model for structured data entry based on explicit descriptional knowledge. Methods of Information in Medicine 33(5):454-463.
Nowlan, W.A, and A.L. Rector. 1991. Medical knowledge representation and predictive data entry. Pp. 105-116 in Proceedings of Symposium on Artificial Intelligence in Medicine Europe, ed. M. Stefanelli. Heidelberg: Springer-Verlag.
Nowlan, W. A., A. L. Rector, T. W. Rush, and W. D. Solomon. 1994. From terminology to terminology servers. Pp. 150-154 in Proceedings of the 18th Annual Symposium for Computer Applications in Medical Care, ed. J. G. Ozbolt. (Special issue of the Journal of the American Medical Informatics Association 1994:150-154.)
Nygren A., and P. Henriksson. 1992. Reading the medical record. I. Analysis of physicians' ways of reading the medical record. Computer Methods and Programs in Biomedicine 39:1-12.
OMG. 1996. The Common Object Request Broker: Architecture and Specification (V2.0). X/Open Publ.
Schaefer, O. P., and J. Sembritzki. 1996. German patient cards and the impact on interoperability. Pp. 137-139 in Proceedings Toward an Electronic Patient Record '96, ed. P. Waegemann. Newton, Mass.: Medical Records Institute.
Scherrer, J.-R., C. Louis, and F. Borst. 1995. Diogene 2, a distributed hospital information system with an emphasis on its medical information content. Pp. 86-97 in 1995 IMIA Yearbook of Medical Informatics, eds. J. H. van Bemmel and A. T. McCray. Stuttgart: Schattauer Verlag.
Thomson, R. 1995. DILEMMA, decision support in primary are, oncology and shared care. Pp. 208-217 in Health in the New Communication Age, eds. M. F. Laires, M. J. Ladeira, and J. P. Christensen. Amsterdam: IOS Press.
Treaty for the Protection of Human Rights and Fundamental Freedom. Rome, 1954.
van Bemmel, J. H. 1992. Protection of medical data. Pp. 51-59 in Cryptography and Data Protection. Proceedings of the Royal Netherlands Academy of Arts and Sciences, Vol. 38, eds. J. H. van Lint and R. Tijdeman. Amsterdam: North-Holland.
van Bemmel, J. H. 1993. Criteria for the acceptance of decision-support systems by clinicians—lessons from ECG interpretation systems. Pp. 7-10 in Proceedings of Symposium on Artificial Intelligence in Medicine Europe, eds. S. Andreassen and J. Wyatt. Amsterdam: IOS Press.
van Bemmel, J. H., J. Fox, and K. Bernstein. 1996. Multimedia shared patient records—European experiences. Pp. 164-168 in Proceedings Toward an Electronic Patient Record '96, ed. P. Waegemann. Newton, Mass.: Medical Records Institute.
van der Lei, J., M. A. Musen, E. van der Does, A. Man in 't Veld, and J. H. van Bemmel. 1991a. Review of physician decision making using data from computer-stored medical records. Lancet 338:1504-1508.
van der Lei, J., and M. A. Musen. 1991b. A model for critiquing based on automated medical records. Computers and Biomedical Research 24:344-378.
van der Lei, J., J. S. Duisterhout, H. Westerhof, W. Boon, P. V. M. Cromme, and J. H. van Bemmel. 1993. The introduction of computer-based patient records in the Netherlands. Annals of Internal Medicine 119:1036-1041.
van Ginneken, A. M., J. van der Lei, and P. W. Moorman. 1993. Toward unambiguous representation of patient data. Pp. 69-73 in Sixteenth Annual Symposium on Computer Applications in Medical Care, ed. M. E. Frisse. New York: McGraw-Hill.
van Ginneken, A. M., H. Stam, and P. W. Moorman. 1995. A multi-strategy approach for medical records of specialists. Pp. 290-294 in Proceedings of MEDINFO 95, eds. R. A. Greenes, H. E. Peterson, and D. J. Protti. Vancouver, B.C.: Coach Publishing.
van Ginneken, A. M., and H. Stam. 1995. Can one patient record accommodate the diversity of specialized care? Pp. 406-410 in Proceedings of the 19th Symposium on Computer Applications in Medical Care, ed. R. M. Gardner. Philadelphia: Hanley & Belfus.
van Ginneken, A. M. 1996. Structured data entry: ORCA combines the strength of two models. In Proceedings of the 20th Symposium on Computer Applications in Medical Care. Philadelphia: Hanley & Belfus.
van Mulligen, E. M., T. Timmers, J. Brand, et al. 1994. Hermes: A health care workstation integration architecture. International Journal of Biomedical Computing 34:267-275.
Visser, L. E., A. E. Vlug, J. van der Lei, and B. H. Ch. Stricker. 1996. Cough due to ace inhibitors: A case-control study using automated general practice data. European Journal of Clinical Pharmacology 49:439-444.
Vlug, A. E., and J. van der Lei. 1995. Postmarketing surveillance with computer-based patient records. Pp 327-330 in Proceedings of the 8th World Congress on Medical Informatics, MEDINFO 95, eds. R. A. Greenes, H. E. Peterson, and D. J. Protti. Vancouver, B.C.: Coach Publishing.
Westerhof, H. P., W. M. Boon, P. V. M. Cromme, and J. H. van Bemmel. 1987. ELIAS: Support of the Dutch general practitioner. Pp 1-10 in Present Status of Computer Support in Ambulatory Care, eds. P. L. Reichertz, R. Engelbrecht, and U. Picollo. New York: Springer-Verlag.
Whiting-O'Keefe, Q. E., D. W. Simborg, W. V. Epstein, and A. Warger. 1985. A computerized summary medical record system can provide more information than the standard medical record. Journal of the American Medical Association 254(9):1185-1192.
Willems, J. L., C. Abreu-Lima, P. Arnaud, J. H. van Bemmel, et al. 1991. The diagnostic performance of computer programs for the interpretation of electrocardiograms. New England Journal of Medicine 325:1767-1773.
Wyatt, J. C. 1994. Clinical data systems. Part 3: Development and
evaluation. Lancet 344:1682-1688.
NOTES
1ISO: International Standards Organization.
2ORCA (Open Record for patient CAre) is a generic CPR system in which the knowledge model consists of a predefined vocabulary and knowledge about how the terms of that vocabulary may be combined into meaningful expressions (van Ginneken, 1996). ORCA is used in the framework of different (European) projects (van Bemmel et al., 1996). ORCA contains a powerful model for structuring a CPR and a knowledge model for capturing patient data. Moreover, because it is generic, ORCA can be used for different areas of health care and is user-adaptable. The relationships between the terms in the knowledge model are semantically defined (van Ginneken et al., 1993, Moorman et al., 1994), for instance, in an acyclic directed graph.
3CSE: Common standards for quantitative electrocardiography.
4EDIFACT: Electronic Data Interchange for Finance, Administration, Commerce, and Trade.
5Medical data can be categorized into permanent data (e.g., one's genetic profile) and variable data (e.g., a blood pressure). The privacy aspect of permanent data is perhaps the most crucial. Generally, these two types of data are used in different ways: variable medical data (including alphanumeric data, biological signals, and pictures) are primarily used for the diagnosis and treatment of transient diseases, whereas permanent data are often strongly related to an individual's life and may possibly predict his or her future health. The latter category is also of interest for one's next of kin: parents, brothers, sisters, and children. Genetic data, for instance, do not change or age; they are valid for an entire lifetime.
6DICOM: Digital Imaging and Communications standard (ACR-NEMA, 1985).