Overview and Introduction
The Internet is rapidly and radically transforming many aspects of society, reshaping industries from aircraft manufacturing to retailing by enabling the widespread sharing of information and creating new relationships between buyers and sellers of goods and services. Businesses now sell goods and services over the Internet, often dealing directly with customers rather than working through traditional distribution channels and intermediaries, tailoring products to match more closely the preferences of individual customers. Governments disseminate public information on World Wide Web sites, and consumers use the Internet to find information, communicate with friends and family, plan trips, shop, and pursue hobbies. Both the scope of applications and the number of Internet users will undoubtedly continue to grow as technologies improve and innovators continue to experiment with new online applications.
Health-related activities stand to benefit enormously from the Internet. As a highly information-intensive set of functions characterized by complex interactions among a large number of stakeholdersprimary care physicians, specialists, nurses, patients, health plan administrators, public health officials, medical librarians, researchers, and othershealth-related activities can take advantage of the nearly ubiquitous reach of the Internet and its capability to support communication between users who may not have interacted with each other before. Already the Internet is beginning to influence the health sector by forging new relationships among stakeholders and improving access to health information. Its application in the delivery of health care, maintenance of public health,continue
payment for health care services, education of health professionals, and conduct of health sciences research could improve the quality of care and access to it as well as reduce its cost.
Despite its promise, the Internet's future in supporting health and health care is far from assured. A number of technical, organizational, and policy barriers stand in the way of its adoption by health organizations and consumers. Furthermore, although much can be done with the Internet in its present form, some health applications demand greater technical capabilities than the Internet can now provide, especially in the areas of security, reliability, and timely transmission of information. As a result, some health applications cannot be implemented across the Internet and used in operational settings without potentially threatening the privacy and optimal care of patients.
Health applications have helped motivate a number of efforts to improve the nation's information infrastructure.1 Ongoing research and development (R&D) efforts, such as those being pursued under the federal government's Next Generation Internet (NGI) initiative and the private sector's Internet 2 initiative, also hope to foster technologies that could enhance the Internet's ability to meet the needs of the health sector. These efforts will also provide testbeds for improved evaluations of the benefits of different health applications of the Internet and their technical and nontechnical requirements. But these testbedsand ultimately the Internet itselfwill not adequately support health applications unless a better understanding is developed of the technical capabilities that these applications demand.
This report explores the use of the Internet in health-related applications and attempts to delineate the technical capabilities that such applications demand. Taking a broad view of health applications, it considers uses of the Internet in consumer health, clinical care, public health, medical education, health care financing and administration, and biomedical research.2 It does not, however, attempt to predict which applications are most likely to catch on or to estimate levels of use; rather, it attempts to illustrate the types of applications that are possible and to assess the technical capabilities required for their safe, effective deployment in an operational setting.
The report also addresses organizational and policy issues that stand in the way of broader adoption of Internet technologies for health applications.3 It became increasingly apparent during the course of the study that health applications of the Internet involve systems that combine network infrastructure with other computing technologies (both hardware and software) and with end users who operate in multiple organizational contexts and are influenced by the policy environment. The close coupling among these levels makes it impossible to focus on any one level to thecontinue
exclusion of the others. Trade-offs are often made between the capabilities embedded in different levels of the system,4 and networking can make issues associated with other levels more important. Security, for example, takes on wholly new dimensions in a networked environment in which information can be readily transferred among entities and stored in computers that are attached to a public network. Yet, many of the mechanisms for addressing security concerns will be implemented not in the network itself but in the devices or computers attached to the network. An individual's access to health information in such an environment, and the circumstances under which such access is allowed, will be determined by a confluence of organizational and national policies for protecting health information.
The strong interrelationships between the network, other technology, and organizational and national policy introduce great uncertainties into the evolutionary path of the Internet with respect to health applications. For example, although many would agree that the Internet will enhance the role of the consumer in health care, the future of specific applications, such as remote medical consultations or online access to patients' medical records, is more difficult to discern because of the range of technical, organizational, and policy issues to be resolved (as detailed in later chapters of this report). Further research and experimentation are needed to understand these issues more fully and develop workable solutions. Consistent with the charge to the committee, this report does not attempt to resolve these policy issues, but by highlighting their significance in enabling effective and safe applications of the Internet for health care it may hasten their resolution. In the end, the report recommends ways of helping the Internet better serve a range of health interests. It identifies both long-term needs that will require R&D and steps that must quickly be taken to help people and organizations adopt and adapt to the next generation of Internet technologies. This chapter provides a broad overview of past and present uses of the Internet in health care; technical terms and considerations; and current R&D efforts that may advance the applications of the Internet and so improve health care.
A Systems Perspective
An example may help to demonstrate both the potential value of the Internet in health care and the close linkages between networking technology, other information technology, and nontechnical issues. Consider the following hypothetical scenario:
Alice and Bob are recovering from a particularly virulent flu that kept them both out of work for the past week. They awakencontinue
one snowy February night to hear their 6-year-old daughter, Charlotte, coughing, wheezing, and crying. She seems warm and will not be comforted. Alice and Bob are worried, but they have recently joined a plan that offers them the option of an in-home consultation. Because packing up their daughter and driving to the emergency room of the nearest hospital would take at least half an hour, they telephone the on-call pediatrician. After hearing the symptoms, the pediatrician decides to ask for basic measurements and have a quick look at Charlotte right away to decide whether she needs to be brought to the emergency room.
Alice turns on their Internet access device (a set-top box) and their television, while Bob sets up the home health assessment pack, including a digital thermometer, heart rate monitor, stethoscope, and video camera. Alice uses the keyboard to navigate to the health plan's Web site and inserts a smart card into the box that authenticates them to the health plan server. While they wait a few moments, their access device exchanges digital certificates authenticating both the server and their device and establishes an encrypted session with the server. Because videoconferencing will be used, the device also reserves a suitable level of bandwidth from Bob and Alice's Internet service provider to carry the quality of video needed for the consultation (a few hundred kilobits per second).
Once connected to the health plan Web site, a menu of options appears, and the couple make a video call to the pediatrician. A live image of the pediatrician appears in a video window. Alice transmits an authorization code to the pediatrician enabling her to access Charlotte's medical record from the online repository in which Alice and Bob maintain all their family medical records. The pediatrician asks them to take Charlotte's temperature and pulse and to position the microphone so that she can hear the child's breathing. Alice first uses the thermometer and heart rate monitor, which transmit results to the set-top box over wireless links. Guided by the pediatrician, Alice then places the stethoscope around various landmarks on Charlotte's chest and back to listen to the child's respirations. The pediatrician can see an image of Charlotte beamed to the set-top box from Bob's video camera. Alice and Bob can see a split-screen image on their television showing the pediatrician on one side and the image from their video camera on the other.
The pediatrician determines that Charlotte's condition does not require her to come in to the emergency room. From her remote observations, she concludes that the most likely diagnosiscontinue
is acute asthma. Charlotte has had two previous episodes of asthma during the past year, and in both cases she responded well to inhalants. The pediatrician asks the parents to administer a dose of the inhalant. Because it is possible to determine within 10 minutes whether the inhalant will work, the pediatrician opts to keep the video call running. Bob makes Charlotte comfortable, seating her within range of the video camera. During the ensuing 10 minutes, the pediatrician engages the parents in a brief review of the events leading up to the evening, exploring such things as exposure to dust and toxins as well as stress events in the family. Recalling that Charlotte's school has some major renovations under way, Alice asks the pediatrician about a possible connection between dust from the renovation and Charlotte's asthma flare-up.
The pediatrician guides Alice to the American Lung Association's Web site, and together they review the information about asthma in children. A checklist of environmental risk factors appears simultaneously on the screen, and the pediatrician and Alice review these together. Next they listen to an audio clip of various breath sounds, with the pediatrician coaching Alice on how to identify the distinctive sound of wheezing.
The pediatrician notes that Charlotte's breathing is easing, and the little girl is no longer crying. The pediatrician asks to speak to Charlotte and asks a few questions about how she feels. Charlotte points to her chest and says it feels tight. Noting that she is able to pronounce common words and that the audible wheezing has stopped, the pediatrician judges the situation to be under control and advises the family that Charlotte should be helped back to sleep.
The on-call pediatrician also recommends that an appointment be made for Charlotte to be seen by her own pediatrician the following afternoon. Bob navigates to the health plan's scheduling program and sets up the appointment. The site provides a map to the clinic that can be printed. The next day, as soon as she arrives at the clinic, Charlotte is welcomed and escorted into the examination room. While her doctor is finishing up another appointment, the nurse takes Charlotte's vital signs and adds the information to her electronic medical record, which is accessed from the computer in the examination room. Shortly thereafter, the doctor enters the room, reviews Charlotte's vital signs, examines her, and provides a diagnosis. Once the diagnosis and a prescription for a new inhaler are entered into the electronic record, a claim for payment is automatically filed with Charlotte'scontinue
health plan and an electronic prescription is sent to the pharmacy near her house. The medication will be waiting when Bob and Charlotte stop by on their way home.
This scenario identifies a number of benefits that Internet-based communications could bring to health care and related activities. It allows the patient (and her family) to avoid a potentially hazardous auto trip on a cold and snowy night and it eliminates waiting in an emergency room, during which time Charlotte could have been exposed to other infectious patients. In addition, the remote consultation allows rapid examination of the patient and preliminary evaluation (or triaging) of needs using several data sources (e.g., sound, vision, and instrumented sensors). Had Charlotte's condition been more serious, her parents could have been directed to take her directly to an emergency room; had her condition been less serious, the system could have enabled the family to avoid an office visit altogether. Although telephone-based services can produce similar benefits, they do not enable the clinician to examine the patient visually or with medical devices. Similarly, they are not as effective at allowing care providers to teach patients and their families to distinguish among various symptoms and at providing expert educational materials for understanding a particular condition. The electronic system also supports paperless billing, which could speed payment for services and reduce error and loss as information proceeds through the system of reviews and approvals. The system also allows easy, but protected, access to the patient's medical record to give the care provider more complete information when making a diagnosis and plan of treatment. The record can be updated easily in real time as new information is collected and can be made available to any care provider who needs it.
Of course, considerable effort would be required to transform such a scenario into a reality on a broad scale. A number of technical advances, related to both the networking infrastructure itself and the devices attached to it, would be required. For example, communication links into and out of homes would be needed that are sufficient to support color video of adequate resolution, and there must be suitable assurance that the video service will be available without significant interruption for the duration of the call. Smart cards would need to be issued to consumers to authenticate them to a health care site and support encryption for a session. This type of health care would also depend on electronic patient records, to which patients can grant providers access as needed and which can be updated during the course of a consultation. The equipment used would have to be reliable enough to create and sustain a connection between a family and a care provider for the duration of a consultation and to provide valid measurements of vital signs. Internet-compatiblecontinue
medical devices would be needed to capture vital signs and transmit them to a remote physician. Nontechnical issues would need to be addressed as well. Families would have to be trained to properly use the system and the home medical equipment, such that care providers could be assured of receiving valid information remotely. Health plans would need policies on payment for remote consultations and on care providers' access to the electronic patient record. If any one of these capabilities was lacking, the system would fail.
The Internet and Health
The health sector has a three-decade-long history of linking computers together to improve heath care and administration. The National Library of Medicine (NLM) made its Medical Literature Analysis and Retrieval System (MEDLARS) available online to regional libraries over a time-shared network in the early 1970s. The resulting MEDLINE (for MEDLARS onLINE) system made the library's repository of biomedical references more widely available to support clinical decision making.5 Shortly thereafter, the first local area networks (LANs) were introduced at the University of Vermont Hospital to support clinical and administrative processes (Box 1.1).
Since these beginnings, the health care industry has gradually come to rely heavily on information technology (IT). In 1996, IT constituted 56 percent of the industry's total net capital stockthe fourth highest percentage out of 53 industries examined by the U.S. Department of Commerce (1999). Only the telephone and telegraph, radio and television, and securities and commodities brokerage industries were more IT-intensive. Nevertheless, health care expenditures on IT are relatively small in relation to the size of its labor force. The industry overall spent just $543 per worker on IT in 1996, compared to $12,666 for securities brokers and $29,236 for telephone and telegraph industry workers; on this scale, health care ranked only 38 out of the 53 industries in the Commerce Department sample.6
Health is already a bustling area of activity on the Internet. Recent surveys indicate that more than 22 million Americans used the Internet to retrieve health-related information in 1998a figure that was expected to grow to 33 million in 1999 (Davis and Miller, 1999). Other estimates place the number as high at 70 million (Morrison, 1999). Since it was made available to the public via the Internet in 1997, NLM's MEDLINE database, which contains more than 15 million abstracts and references from more than 3,900 medical journals, has experienced a surge in activity to 300,000 searches per day (Benton Foundation, 1999). Health is one of the more popular topics on the Internet, with estimates of the number ofcontinue
health-related Web sites running as high as 10,000 or more (Benton Foundation, 1999). Health-related Web sites allow consumers to search for information on specific diseases or treatments, pose questions to care providers, manage chronic diseases, participate in discussion groups, assess existing health risks, and purchase health-related products. By one estimate, the online consumer market will grow to $1.7 billion by 2003,continue
fueled largely by online sales of products such as prescription and nonprescription medicines and vitamin supplements (Nash, 1999).
Beyond its popularity with consumers, the Internet is also used by health care professionals, biomedical researchers, and health care administrators. Web sites geared to health care professionals allow them to access the professional literature, consult with colleagues electronically, order medical supplies, or communicate with insurance companies.7 Biomedical researchers use the Internet to access online databases of journal articles and scientific information. Organizations involved in the provision of health care, whether individual hospitals, managed care plans,8 or integrated delivery networks (IDNs),9 have begun to use the Internet to reach out to consumers. Their Web sites provide information on available services and may allow consumers to change their enrollment status, select physicians, and schedule appointments electronically.
Drivers of Internet Applications in Health
The health applications available on the Internet today take advantage of the Internet's expansive reach to enable health care organizations to interact with a growing number of online consumers (Miller and Reents, 1998). Whereas just 17 percent of U.S. households had Internet access in 1997, roughly one-third did by 1998 (NTIA, 1999), and analysts predict that 90 percent of U.S. households will have Internet access by 2005 to 2010 (Rosenberg, 1999). As people become accustomed to using the Internet for routine activities, from electronic commerce (e-commerce) to homework, they are likely to use the Internet for health-related activities. Consumer experiences in other areas of Internet activity, such as e-commerce and electronic mail (e-mail), will influence the expectations they bring to online health applications (Mittman and Cain, 1999).
Care provider organizations face a number of pressures to integrate the Internet more effectively into their operations. Recent trends toward consolidation in the health care industry and the expansion of managed care have erased some of the impediments to sharing information among competing organizations. As they attempt to link individual practices, clinics, and hospitals into single entities, IDNs have a greater need to share information with affiliated institutions. As purchasers, accrediting bodies, and the general public increasingly hold managed care plans accountable for the quality of health care, plans have developed schemes for the electronic sharing of data on facilities' utilization rates and health-related outcomes. With such data, managed care plans can compile statistics on quality-of-care indicators and monitor the quality and costs of the individual care providers. As care provider networks grow and consumers become more mobile, the electronic transmission of patient informationcontinue
among providers could improve care and reduce costs to the provider, the patient, and the managed care plan.10
Impediments to Broader Adoption of the Internet
Despite the flurry of Internet activity within and around health care, many potential applications have yet to be realized. Many organizations in the health sector continue to rely on private networks (e.g., leased lines) rather than the Internet for many data communications tasks, and some health-related applications have not yet been deployed across any type of communications network, public or private (Box 1.2). Few health care organizations, for example, have integrated the Internet directly into the provision of care. Remote medical consultations remain a novelty practiced by a few institutions, typically over dedicated networks, for a small subset of their patients and with support from external financial grants. Most public health offices remain unconnected to the Internet and there-soft
fore are unable to accept electronic reports from testing laboratories or communicate health information over the Internet to neighboring jurisdictions. Private insurers have in general not adopted the Internet for financial and administrative transactions but instead continue to seek payment through paper-based claims or electronic data sent over direct connections via modems.
The reasons for the limited adoption of the Internet in health-related activities are manifold, but the underlying reason is a lack of demonstrated value in different applications. The Internet has been widely adopted by the public as a tool for gaining insight into issues of illness and health because it is perceived to deliver value. Many (but not all) care providers use the Web frequently for searching online databases (such as MEDLINE), also because it is perceived to deliver value. A small, but growing, number of care providers engage in e-mail discussions with their patients about health problems. Care providers do not use the Internet more broadly in the process of treating patients because the valuable, usable, affordable, and practical Internet-based solution has yet to be built. The process of determining which applications add value in health applicationsand which specific capabilities and attributes provide that valuerequires continued experimentation and analysis of data on the benefits and costs of the Internet relative to those of other media.
To date, little information is available with which to gauge the contributions of the Internet to the provision of health carenot to mention its potential to improve public health, biomedical research, and professional education. Emerging evidence of the benefits to health care of information systems generally bodes well for the Internet; a growing number of studies demonstrate, for example, reductions in adverse drug interactions and improved diagnoses stemming from the use of computer-based decision support tools in clinical environments.11 Research has also demonstrated the positive effect of information technology applications in several other areas of health care.12 However, the ability of the Internet (as opposed to private networks) to improve the quality of health care or expand access to it has not been demonstrated. On the contrary, there has been considerable concern about the quality of health information available on the Internet and its potential to harm consumers (Mittman and Cain, 1999; SCIPICH, 1999). In an industry already facing serious fiscal and organizational upheaval, health care organizations may remain skeptical of a range of Internet applications until there is greater evidence of their benefits, along with more information about the policies and procedures needed to avoid the potential harms.
The benefits of the Internet in health applications may prove difficult to measure because the most notable benefits may be indirect and may vary across segments of the health sector. For example, the advantages ofcontinue
consumer-oriented Web sites to care provider organizations such as hospitals might include marketing, possibly advertising, and the collection of valuable data about interested consumers. The direct and indirect revenues from many of these activities, however, may be insufficient to support the development and maintenance of the applications themselves. Furthermore, the use of the Internet could stimulate changes in industry structure that are difficult to foresee at present. For example, the Internet could enable large provider organizations to extend their reach more directly into local communities, working with local care providers to provide greater continuity and consistency of care. Or, it could allow consumers to better triage their own health needs using online modules created by their health plans. These changes could improve health and disease management among local populations, but the benefits may accrue most directly to consumers. The benefits to care providers may be more difficult to measure, especially if healthy patients demand fewer health services in the long run.
Further slowing adoption of the Internet by health organizations are uncertainties about the technical capabilities needed to support health applications. Managers of many health organizations say that security concerns prevent them from using the Internet to transfer patient medical records among affiliated organizations or from allowing care providers to access such records remotely (Siwicki, 1999). At the same time, they are not certain what types of security technology are needed to adequately protect patient information in such applications. With respect to other applications, such as remote medical consultations, practitioners note that they cannot obtain sustained access to the band width they might need for real-time video. It is appropriate to question whether today's Internet provides a sufficiently strong infrastructure to support applications such as critical-care monitoring and automated delivery of medication. If deployed in health care settings without proper attention to these capabilities, the Internet could have an adverse effect by eroding patient privacy and preventing care providers from accessing needed information.
Whether the Internet will become more widely adopted in health care will depend, in part, on the technical capabilities it can provide and how these capabilities compare with those provided by other networking alternatives available to health organizations and consumers. A number of technical factors need to be considered in such evaluations. The five primary factors considered in this report are bandwidth, latency, availability, security, and ubiquity.break
1. Bandwidth is the rate at which information is transmitted through a network, measured in bits (or kilobits or megabits) per second. The bandwidth a network can provide is a property of the transmission medium (e.g., fiber optics, coaxial cable, telephone wire, radio waves), the network topology, and the switching or routing devices used to guide traffic through the network. The amount of bandwidth a particular application demands is determined by the amount of data to be transmitted and the time in which that transmission must be completed. Applications that must transfer large amounts of data quickly demand much greater bandwidth than do applications that transfer smaller amounts of data (such as e-mail) or transfer data more slowly (e.g., if a response is not needed quickly). From the point of view of an individual user (e.g., a consumer, a doctor, or a nurse), the demand for bandwidth is a demand not so much for a uniform increase in the bandwidth of the entire network but for access to sufficient bandwidth when needed.
2. Latency is the time required to transmit data across the network (i.e., the delay between a sender transmitting a message and a recipient receiving it). The minimum latency a network can provide is influenced by the speed of its switches and routers and the physical distance across which the message is sent. Data communications traverse different media at the speed of light, which places a lower limit on the time it takes for a message to travel between two points on the network. The latency an application demands can vary tremendously. Real-time, interactive applications demand low latency so that users can interact with each other easily. Many interactions, such as telephone conversations or control of remote devices, become unwieldy if round-trip latencies (i.e., across the network and back) exceed 300 milliseconds. Applications that do not demand real-time interactions between usersso-called asynchronous applications such as e-mail and store-and-forward messaging systemshave only weak demands on latency requirements. Closely related to latency is jitter, the variation in latency over time. High levels of jitter imply unpredictable degrees of latency across the network, although several techniques, including temporary buffering of information, can remove jitter at the receiver's end of the network. In some applications, the related notion of response time is more important than latency per se. Response time refers to the length of time needed to transmit a full message (such as a service request or an image) rather than an individual packet across the network and receive a response. Messages can consist of many packets, and successful transmission can depend far more on network reliability on a packet-by-packet basis than on the actual latency. The Transmission Control Protocol (TCP) used on the Internet, for example, may transmit more than one packet before it receives an acknowledgement that the first packet has been received, but if a packet is not received,continue
TCP has to wait for a certain period of time before transmitting that packet again. Thus, both response time and latency can be affected by the load on a network at a given point in time.
3. Availability refers to the continuous availability of the network, the individual links of which it is composed, and the services it offers. Availability can be measured in terms of the percentage of the time the network (or a particular link) is operational or by the average time between failures. A number of factors can render networks unavailable, including physical damage to network links or nodes, hardware or software failures (i.e., component reliability problems), operator error, software errors, and deliberate malicious attacks against the system. Steps can be taken to harden systems against these sorts of failures, and procedures can be developed for restoring some level of network services in the event of failures. Nevertheless, some applications cannot function properly if the performance of the network is degraded; and many time-critical applications cannot tolerate network failures, even if very brief.
4. Security, in the computer science community, generally encompasses three elements: system availability, confidentiality, and integrity. The first of these is addressed above. Confidentiality refers to the ability to prevent communications from being disclosed to unauthorized parties in violation of disclosure rules. Integrity refers to the ability to prevent malicious or accidental alteration of data. These two capabilities can be provided by a variety of technical mechanisms that support authentication of user identities, encryption of communications, and different forms of access control. The need for confidentiality and integrity varies greatly across applications. Both are important concerns in applications involving exchanges of personal health information. Integrity is of paramount importance for some applications, such as setting levels on dosimetry equipment that delivers drugs to patients and preserving the authenticity of medical images.
5. Ubiquity refers to the relative accessibility of a network. The ubiquity of a network is influenced by the network's geographic scope (i.e., whether it can be accessed from many places) and by rules regulating participation (i.e., whether it is open to the general public or to members only). The telephone network, for example, is highly accessible because roughly 94 percent of U.S. homes have telephone connections, and anyone is allowed to subscribe to the service. Cable television is almost as ubiquitous because it passes most homes and is also available to all who pay for service. However, cable modem service (to support data communications as opposed to television) is not yet available to all locations in the United States. Such systems stand in contrast to private networks, such as those used by financial institutions, which may have broad geographic reach but strict rules regarding membership. Applications thatcontinue
serve the general public usually demand high levels of accessibility (ubiquity), whereas those that serve limited populations do not, although there may be interest in allowing access by members from multiple locations.
Another term that is important in describing network performance is ''quality of service" (QOS). Network engineers use this term to refer to the ability of a network to provide a range of assured levels of performance. Performance is characterized by metrics such as the bandwidth obtained between two points in the network (which may be dramatically less than the bandwidths of the individual communications links involved, either because of other traffic on the network or because of the need to retransmit packets of information dropped during transmission); latency and jitter (defined above); and the packet loss rate, the percentage of transmitted packets that are dropped inside the network and not delivered to their intended destinations.13
The need for QOS stems from a design characteristic of networks whereby resources (e.g., especially backbone links) are generally shared among many users who are running applications simultaneously. Thus, even if the bandwidth of the system is measured in megabits or gigabits per second, any single user might gain access to only a small fraction of that bandwidth. How much bandwidth a single user obtains depends on the level of activity of the other users. When using the Internet, for example, a user often obtains high bandwidth and low latency over certain paths and at certain times of day, but it is not currently possible to ensure that such conditions will be available on a given path at a given time. Hence, although some users may succeed in making telephone calls of acceptable quality over the Internet, they cannot depend on this medium to make calls to any location at any time. Quality of service guarantees require mechanisms that enable applications with specific requirements to negotiate to receive appropriate treatment in the network. Those mechanisms must be able to deal with requests from huge numbers of applications running simultaneously.14
The Internet differs from other communications networks in all the dimensions outlined above. Many of these differences stem from the public character of the Internet (Box 1.3), which carries aggregated traffic from numerous parties, whereas private networks interconnect a limited number of sites using dedicated transmission links that are not shared with any other users. Two private networks are completely isolated from each other in the sense that no data can "leak" from one into the other, a user of one network cannot access resources on the other, and the level ofcontinue
usage on one network has no effect on the availability of resources in the other.15 Thus, users have greater control over QOS (i.e., the bandwidth that will be available at a given time) and confront fewer security risks than do users of the Internet. For these reasons, the Defense and Energy Departments, the banking industry, and other sectors have developed communications networks that are separate from the Internet.
And yet, the distinction between public networks such as the Internet and private networks is blurred by a number of factors. For example, many private networks used for internal communications among elements of a single corporation can be connected to the Internet in a controlled way that limits the messaging traffic that can transit the interface (indeed, many of the networks attached to the Internet are private). Similarly, many organizations develop private intranets that are based on Internet technologies and protocols to facilitate data sharing within an enterprise. Other networks, called extranets, use Internet standards and technologies to support secure exchanges of information among trading partners. Many of these private networkswhether intranets or extranetscould be made accessible to valid users over the public Internet through the use of appropriate security and authentication technologies. The distinction between private and public networks is further blurred by technologies such as virtual private networks (VPNs), which provide many of the same properties of private networks while using shared network facilities (including possibly the Internet itself) to provide connectivity. In this way, VPNs (discussed in greater detail in Chapter 3) create the illusion of private, dedicated point-to-point connections, but many VPNs may be supported on the same physical network. They provide almost the same level of security as true private networks because there is no connectivity among the VPNs, but messages on one virtual circuit must contend with those on other circuits for network resources (e.g., bandwidth).
Whether a network is truly or virtually private, it differs in a very important way from the Internet. In a private network, connectivity is deliberately constrained to a limited number of sites, and these sites are known a priori. Every time a new site is added to the network, some administrative overhead is involved, whether ordering and awaiting the installation of a new circuit or provisioning a new virtual circuit. Thus, private networks are ideally suited to an environment in which the necessary connectivity is known and remains stable over time. A classic application is interconnection of the different geographic locations of a single corporation. Another application is the connection of a number of companies that have a long-standing business relationship, such as a large manufacturer and its parts suppliers. However, private networks are fundamentally unsuited to environments in which arbitrary connectivitycontinue
is required. The Internet, by contrast, enables users to connect to each other without a prior arrangement. In a business setting, this enables consumers to find suppliers readily, and vice versa.
Because of the range of technical capabilities available across different types of networks, organizations can tailor their network architectures to their specific needs. When security and QOS are important and the relationships between communicating parties are sufficiently well known in advance, private networks are likely to be chosen. Accordingly, health organizations continue to use dedicated networks to transmit sensitive patient information, share large image files, and submit claims for reimbursement. When the overriding goal is maximum connectivity without a priori knowledge of the communicating parties, as in the case of making a product or service available to a wide set of consumers, the Internet is likely to be chosen. In some cases, the most judicious choice might be a network that uses a combination of private or dedicated lines that are connected in appropriate ways to the Internet to allow broader access, but with suitable security capabilities in place. What may be most important is the use of consistent, interoperable protocols for all communications so that various networks can be connected as needed with appropriate gateways. The architectures chosen by various organizations depend on the requirements and cost-benefit analyses of technical alternatives.
The value of open (i.e., public) networks in health care is rooted in the nature of the industry, which remains highly decentralized and involves a range of individuals and organizations in providing care, paying bills, analyzing health data, conducting health services research, and monitoring public health. As recently as 1995, the United States had 1.2 million health care providershalf of whom work in private practicesand more than 3,000 private insurance payers.16 The patterns of data sharing among these organizations are complex and can change frequently. Internet use could enable IDNs, for example, not only to exchange patient records among affiliated hospitals and clinics, but also to send records to other hospitals to assist in the treatment of patients who are injured or become ill while traveling. It could enable any rural care provider to arrange remote medical consultations with any remote specialist connected to the Internet who has the expertise needed to handle a particular patient's case. It could lead to much more rapid reporting of diseases, enabling state public health officials to accept reports directly from physicians and testing laboratories throughout the state.
Furthermore, Internet connections promise to be less expensive to install and maintain than private networks. Individual lines do not need to be leased from telephone companies to connect the various partners in the network. Internet connections entail some costs, especially if high-bandwidth connections are needed, but they tend to be lower than thecontinue
costs of leased lines and can be spread among a wider range of applications and users. Finally, the Internet has been the catalyst for the integration of many applications and systems of applications. The existence of a communication infrastructure built on a common set of protocols makes it more difficult to justify, both economically and technically, the use of separate, special-purpose networks for different applications; the common infrastructure also facilitates interactions between separately developed systems. When separate information systems are integrated on the foundation provided by the Internet, the investments frequently are justified on the basis of both savings in communication costs and improved functions. With improved security and QOS, the Internet might become preferable to private networks in almost all cases.
The Internet may prove to be an ideal technology for use by willing health care organizations to simplify and standardize processes and collaborate more effectively with one another. The value of this common communications infrastructure in other sectors, from entertainment to banking to retail sales, could extend to health care if technical obstacles, organizational uncertainties, and policy barriers can be overcome. Organizations that adopt the Internet could reap significant benefits, including cost savings from support of a wide range of applications and the ability to leverage the technology investments of other communities. An analogy to the telephone system may be illuminating. There is no separate U.S. telephone system for health applications. Despite the distinct set of priorities and trade-offs associated with health applications, the health community uses "plain old telephone service" (POTS) for most of its voice communications, leveraging the telephone companies' investments in R&D as well as infrastructure deployment and accepting inconveniences such as busy signals that might not be desirable in emergencies. Features such as 911 have been added to POTS to support emergency needs (whether related to health or public safety), but even this feature leverages the existing network. Some specialized voice communications networks, such as communications between emergency rescue vehicles and hospital emergency rooms, have also been established for health needs not well served by POTS.
Enhancing the Internet
A number of efforts are under way to improve the ability of the Internet to provide QOS, security, and availability, which would enable its broader use within the health domain. These efforts include attempts by individual companies to increase the capabilities of Internet routers and deploy high-speed data services on demand, as well as attempts by the Internet Engineering Task Force (IETF) to develop new standards andcontinue
protocols for improved services (see Chapter 3). In addition, two major collaborative efforts are under way to develop and demonstrate advanced networking technologies that promise to improve the QOS, availability, and security across the Internet. The government's Next Generation Internet initiative and projects sponsored by the private-sector University Consortium for Advanced Internet Development (UCAID) are attempting to develop advanced networking technologies and applications and deploy them in testbed networks that link a limited number of sites and allow early experimentation with advanced applications. The technologies and applications to be developed under these programs, which are described below, could diffuse onto the Internet as they are demonstrated and proven.
The Next Generation Internet Initiative
Formally initiated in October 1997, the NGI initiative is a multiyear program, funded at approximately $100 million per year, that involves a number of federal agencies: the Defense Advanced Research Projects Agency (DARPA), the National Science Foundation (NSF), the Department of Energy (DOE), the National Aeronautics and Space Administration (NASA), the National Institute of Standards and Technology (NIST), and the National Institutes of Health (NIH) (Table 1.1). The initiative is managed by individual agencies, with coordination provided by the large-scale networking working group of the White House National Science and Technology Council's Committee on Technology, Subcommittee on Computing, Information, and Communications R&D.17
The NGI initiative has three components: R&D on advanced networking technologies for improved performance and functionality; the deployment of high-speed testbed networks that emphasize end-to-end performance; and the development and demonstration of revolutionary applications that demand advanced networking and are not possible on today's Internet.18 The first component involves R&D projects in areas such as high-speed routing, security, QOS, and network management and modeling. This work will be funded primarily by DARPA but also by NSF, NASA, and NIST.
The second component will be achieved by developing and demonstrating applications of two types: (1) discipline-specific applications of interest to participating agencies, including health care, basic science, education, and environment and (2) their enabling technologies, including collaboration technologies, digital libraries, distributed computing, privacy and security, and remote operation and simulation (National Science and Technology Council, 1999). The NIH is actively involved in this effort through the NLM, which awarded 24 contracts totaling $2.3 millioncontinue
in October 1998 to investigate and develop health care applications of the NGI. These projects make up the first phase of a three-phase program. Several of the projects received phase II awards in late 1999 and early 2000 to allow their implementation in local testbed settings (see Appendix B for a list of all NLM project awards as of January 2000). Phase III will support scale-up to the regional or national level of successful phase II testbed projects. These projects are intended to improve the health community's understanding of the ways in which the NGI can affect health care, health education, and health research systems with respect to cost, quality, usability, efficacy, and security. Supported projects include efforts to (1) build a virtual human cadaver for educational purposes, (2) develop telemedicine technologies to support health care in rural areas,19 (3) demonstrate the feasibility of a national breast imaging archive and networking infrastructure to support telemammography, and (4) create a personal health record that can be integrated with more traditional sources of clinical information for patient use in the home, at work, or at school (see Box 1.4 for examples of these projects and Appendix A for a complete listing of NLM project awards).
Other federal agencies, including NASA and the NSF, have also funded projects that will demonstrate health-related applications of thecontinue
NGI. Researchers at NASA's Ames Research Center, for example, are developing a system for sharing high-resolution, three-dimensional medical images in real time for purposes of collaborative diagnosis and surgical planning.20 NSF is supporting work to provide psychological services over a distance to deaf patients, to develop digital video resources for teaching and learning the life sciences (using materials that reside atcontinue
the NLM), and to allow Web-based control of a remote electron microscope for biological research, among other projects. Such efforts reflect the importance of health-related applications in motivating large-scale information infrastructure programs such as the NGI.
The third component of the NGI initiative will be carried out by constructing two types of testbed networks, one of which will link approximately 130 participating universities and federal agencies at speeds 100 times faster than those available across the Internet in 199721 and the other of which will link about 10 sites at speeds 1,000 times faster than the 1997 Internet. The first testbed will be built on several existing federal networks: the NSF's very-high-performance Backbone Network Service (vBNS),22 NASA's Research and Education Network, DOD's Defense Research and Education Network, and DOE's Energy Sciences network. The vBNS, for example, operated at 622 megabits per second (Mbps) in 1998 but is expected to be upgraded to 2.4 gigabits per second (Gbps) by the year 2000. Universities connecting to the vBNS at 45 Mbps will be upgraded to 155 Mbps to help them take greater advantage of the increased backbone capacity.
The NGI initiative's other testbed will be built on DARPA's SUPERNET, a network composed of a variety of high-speed technologies and testbeds, enabling researchers to collaborate and experiment with advanced networking technologies and applications in a diverse, high-capacity, wide-area environment. It will use wave-division multiplexing technology (WDM) to allow multiple frequencies of light (and hence multiple communications channels) to share a single fiber-optic cable (see Chapter 3). DARPA demonstrated a 5-node network at 2.5 Gbps per channel in 1999 and plans to establish a 10-node network with 160 Gbps facilities in 2002. The NSF, NASA, and DOD networks will connect to this network.
The goal of these networks is to provide a cutting-edge but stable network that will support the development of revolutionary applications and serve as a testbed for new technologies and protocols. According to the 1998 NGI implementation plan, the testbed networks will be initially deployed with best-effort services using IP version 4 (IPv4) (Large Scale Networking Next Generation Implementation Team, 1998). New versions of IP (including IPv6), QOS technologies, multicast protocols (for facilitating group interactions), security protocols, and network management tools will be deployed in the networks as soon as they become stable. Feedback from application developers to network researchers, operators, and implementers will help ensure that the testbeds evolve in a manner suitable to the types of applications that are expected to be run on them.break
Private-Sector Efforts: Internet 2 and Abilene
The UCAID, which was incorporated in 1998, has two related networking projects under way that promise to enhance the capabilities of the Internet. The first is the Internet 2 project, which will link more than 100 member universities and partners to an advanced academic network. Research supported by Internet 2 is attempting to enable applications that are not possible with the technology underlying today's Internet (some examples are telemedicine, digital libraries, and virtual laboratories). The program is intended to demonstrate new applications for improving research and enhancing the delivery of education and other services, including health care. It will facilitate the development, deployment, and operation of an affordable communications infrastructure capable of supporting differentiated QOS based on the applications requirements of the research and education community, and it will promote experimentation with the next generation of communications technologies.23
Biomedical applications play a significant role in the Internet 2 initiative. The first demonstration of the network, in October 1999, consisted of an online broadcast of a gall bladder operation. The surgery team inserted light, camera lenses, and surgical tools inside the patient's body, creating internal views of the operation. Audio and video were transmitted over the network in real time, requiring network bandwith that would support a consistent data transmission rate of 2 Mbps. Only a small audience was able to view the demonstration, but it enabled a doctor based in Washington, D.C., to assist in the surgery, which took place at Ohio State University.24
A related UCAID project, Abilene, is seen as a second Internet 2 backbone. Abilene is based on a partnership with Qwest, Cisco Systems, Nortel, and Indiana University. The goals are to provide a high-availability backbone network to support the demands of the advanced research applications being developed by UCAID members; a separate network to enable the testing of advanced network capabilities (for example, QOS, multicasting, and security and authentication protocols) prior to their introduction into the application development network; and a separate network capability to conduct networking research, including the design of an alternative network capable of advancing both the Abilene network and the general state of the art.25 Internet 2 member universities have committed more than $70 million per year in new investment on their own campuses for the Internet 2 project, and corporate members have committed more than $30 million over the life of the project.
Although programmatically distinct from the NGI initiative, the UCAID's efforts are related to federal networking activities. More than 90 Internet 2 universities have received grants under NSF's High Perfor-soft
mance Connections program to support links to advanced backbone networks such as Abilene and the vBNS. Internet 2 is also participating in the NGI Joint Engineering Task Force to ensure the cohesiveness and interoperability of the technologies that Internet 2 is developing. Internet 2 member institutions may receive funding in the form of competitively awarded grants from the NSF and other federal agencies participating in the federal NGI initiative. Additional cooperative relationships are being planned as part of NGI implementation.
Deploying Enhanced Internet Technologies
Although they are structured as programs with a limited number of participants, the NGI and Internet 2 initiatives are intended to serve as launching points for enhancement of the public Internet. Both programs have a stated interest in transferring new technical capabilities to the public Internet once the technologies are developed and demonstrated to be robust. Just as early DARPA support for the ARPANET and subsequent NSF support for NSFNET laid the groundwork for today's Internet by funding networking research and applications development and deploying network infrastructure,26 so too, it is hoped, will the NGI and Internet 2 initiatives plant the seeds for an improved Internet that can serve the public at large. They intend to accomplish this by developing and demonstrating technologies that can later be deployed in networks maintained and operated by private companies.
Whether the public Internet will evolve into a network capable of supporting a full range of health applications will depend on many factors other than technology. Of particular importance will be economic incentives for network providers to deploy the levels of bandwidth, QOS, security, availability, and ubiquity that health applications demand. These incentives will be derived from the combined demands of many applications in different sectors, including health. The history of Internet development is one of innovation and experimentation, not planned development. The forces that drive its continuing evolution are increasingly economic, and these forces alone may not yield an infrastructure that can support the integration of critical and noncritical functions of the health community. In the end, some capabilities may prove too expensive to deploy throughout the Internet, leaving health organizations to operate with a mixture of different networking infrastructures to meet their various needs.
Only by making its needs explicit and working with organizations involved in the deployment of Internet capabilities can the health community hope to ensure that an enhanced Internet infrastructure meets health needs. This report respresents the first step in that effort. Bycontinue
evaluating the technical capabilities that the Internet must provide to support different health applications, the report offers the health community information that it can use to shape the networks being deployed as part of the NGI and Internet 2 initiatives and, ultimately, as part of the Internet. Clearly, ongoing evaluation and experimentation will be needed. The many uncertainties inherent in the process of developing and deploying Internet-based applications make any attempt to predict the long-term evolution of the Internet within the health community foolhardy. Sustained interaction will be needed to ensure that the emerging needs of the health community continue to be met by the evolving capabilities of the Internet.
Organization of This Report
The remainder of this report outlines the technical and nontechnical challenges that must be overcome if the Internet is to support a widening range of health applications. Chapter 2 examines specific applications of the Internet across this domain. The first part of the chapter focuses on applications of the Internet in the provision of health care, addressing topics such as consumer health, remote consultation, and the transfer of medical images for diagnostic purposes. The next parts of the chapter explore Internet applications in areas such as public health, health care finance and administration, and biomedical research. The chapter draws on a series of site visits by the committee that provided insight into the types of Internet applications being developed today and the networking challenges that cannot currently be ported to the Internet. The chapter reviews the technical capabilities that each application demands in terms of QOS (combining bandwidth and latency requirements), security, availability, and ubiquity. The applications examined are intended to illustrate the range of ways in which the Internet might be used rather than to identify them as likely paths.
Chapter 3 reviews the technical challenges posed by applications of the Internet in health, health care, and biomedical research. It examines ongoing efforts to enhance the capabilities of the Internet and identifies areas in which health care needs might not be addressed if they are not explicitly considered during the research process. Chapter 4 examines organizational barriers to the deployment of the Internet for health and health care. It describes ways in which the Internet can serve the strategic interests of health care organizations and identifies the range of uncertainties surrounding the Internet's use that hamper efforts to deploy it more broadly in such organizations. Chapter 5 discusses elements of public policy that stand in the way of greater use of the Internet in the health community. These barriers range from issues of payment for services andcontinue
licensure that have stymied previous attempts at telemedicine, to broad issues of intellectual property protection and privacy that have special significance in the health domain.
Finally, Chapter 6 summarizes the committee's conclusions and offers a series of recommendations for facilitating the more widespread use of Internet technologies in health care and biomedical research. The recommendations suggest ways in which technical and nontechnical barriers can be overcome to enable the design of an Internet that will more fully support the needs of the health sector.
AT&T. 1999. ''AT&T and M.D. On-Line, Inc. to Promote AT&T WorldNet Internet Connectivity to Healthcare Providers," News release. February 8. Available online at <http://www.att.com/press/item/0,1193,339,00.html>.
Barry, M.J., J.J. Fowler, Jr., A.G. Mulley, Jr., J.V. Henderson, Jr., and J.E. Wennberg. 1995. "Patient Reactions to a Program Designed to Facilitate Patient Participation in Treatment Decisions for Benign Prostatic Hyperplasia," Medical Care 33:771-782.
Benton Foundation. 1999. Networking for Better Care: Health Care in the Information Age. Benton Foundation, Washington, D.C., March.
Computer Science and Telecommunications Board (CSTB), National Research Council. 1999. Funding a Revolution: Government Support for Computing Research. National Academy Press, Washington, D.C.
Davis, Robert, and Leslie Miller. 1999. "Millions Comb the Web for Medical Info," USA Today, July 15. Available online at <http://www.usatoday.com/life/health/online/lhon1012.htm>.
Gillespie, Greg. 2000. "Online Clinical Guidelines Help Trim Costs," Health Data Management 8(1):39-45.
Greenfield, S., S. Kaplan, and J. Ware, Jr. 1985. "Expanding Patient Involvement in Care," Annals of Internal Medicine 102:520-528.
Gustafson, David H., R.P. Hawkins, E.W. Boberg, and E. Bricker. 1992. "CHESS: A Computer-Based System for Providing Information, Referrals, Decision Support, and Social Support to People Facing Medical and Other Health-Related Crises," pp. 161-165 in Proceedings of the Annual Symposium on Computer Applications in Medical Care. McGraw-Hill, Health Professional Division, New York.
Gustafson, David H., M. Wise, F. McTavish, J.O. Taylor, W. Wolberg, and J. Stewart. 1993. "Development and Pilot Evaluation of a Computer-Based Support System for Women with Breast Cancer," Journal of Psychosocial Oncology 11:69-93.
Gustafson, David H., R.P. Hawkins, E.W. Boberg, and E. Bricker. 1994. The Impact of Computer Support on HIV Infected Individuals. Final Report to the Agency for Health Care Policy and Research, Washington, D.C.
Halamka, J., and M. Hughes. 1998. "A Paradigm Shift in Health Care Information Systems: Clinical Infrastructures for the 21st Century," pp. 401-405 in Proceedings of the American Medical Informatics Association Fall Symposium, C. Chute, ed. Hanley & Belfus, Philadelphia.
Information Infrastructure Task Force (IITF), Committee on Applications and Technology. 1994. Putting the Information Infrastructure to Work. National Institute of Standards and Technology, Gaithersburg, Md.break
Institute of Medicine. 1996. Telemedicine: A Guide to Assessing Telecommunications in Health Care, Marilyn J. Field, ed. National Academy Press, Washington, D.C.
Large Scale Networking Next Generation Implementation Team. 1998. Next Generation Internet Implementation Plan, Second Printing. National Coordination Office for Computing, Information, and Communications, Arlington, Va., February.
Lindberg, Donald A.B., and Betsy L. Humphreys. 1998. "Medicine and Health on the Internet: The Good, the Bad, and the Ugly," Journal of the American Medical Association 280(15):1303-1304.
Miller, Thomas E., and Scott Reents. 1998. The Health Care Industry in Transition: The Online Mandate to Change. Internet Strategies Group, Cyber Dialog, Inc., New York.
Mittman, Robert, and Mary Cain. 1999. The Future of the Internet in Health Care: Five Year Forecast. Institute for the Future, Menlo Park, Calif., January.
Morgan, M.W., R.B. Deber, H.A. Llewellyn-Thomas, P. Gladstone, R.J. Cusimano, and K. O'Rourke. 1997. "A Randomized Trial of the Ischemic Heart Disease Shared Decision-Making Program: An Evaluation of a Decision Aid," Journal of General Internal Medicine 12(1):62.
Morrison, J. Ian. 1999. "Healthcare in the New Millenium: The Promise of the Internet," Presentation at Internet Health Day II: Health Care in Transition-Preparing for an Interactive Future. New York, October 12.
Nash, Sharon. 1999. "The Doctor Is Online," PC Magazine Online, July 14. Available online at <http://www.zdnet.com>.
National Science and Technology Council, Committee on Technology, Subcommittee on Computing, Information, and Communications R&D. 1999. Information Technology Frontiers for a New Millennium. Supplement to the President's FY 2000 Budget, National Coordination Office for Computing, Information, and Communications, Arlington, Va., April.
National Telecommunications and Information Administration (NTIA). 1999. Falling Through the Net: Defining the Digital Divide. U.S. Department of Commerce, Washington, D.C.
Rosenberg, Matt. 1999. "Popularity of Internet Won't Peak for Years," Puget Sound Business Journal, May 24. Available online at <http://www.amcity.com/seattle/stories/1999/05/24/focus.html>.
Science Panel on Interactive Communication and Health (SCIPICH). 1999. Wired for Health and Well-Being: The Emergence of Interactive Health Communication, Thomas R. Eng and David H. Gustafson, eds. Office of Disease Prevention and Health Promotion, U.S. Department of Health and Human Services, Washington, D.C., April. Available online at <http://www.scipich.org>.
Siwicki, Bill. 1999. "Applying the Internet in Health Care," Health Data Management 6(3):38-48.
Smith, K.A., and R.B. Mehnert. 1986. "The National Library of Medicine: From MEDLARS to the Sesquicentennial and Beyond," Bulletin of the Medical Libraries Association 74(4):325-32.
U.S. Department of Commerce, Economic Statistics Administration. 1999. The Emerging Digital Economy II. Washington, D.C., June.
Vickery, D.M., T.J. Golaszewski, E.C. Wright, and H. Kalmer. 1988. "The Effect of Self-Care Interventions on the Use of Medical Service Within a Medicare Population," Medical Care 26:580-588.break
1. For an illustration of the role of health applications in motivating federal programs to develop national information infrastructure, see IITF (1994).
2. The Internet also has applications in support of clinical research (e.g., clinical trials), but these applications are not investigated in great detail in the report.
3. Others have also noted the importance of organizational and policy issues in influencing the rate of adoption of Internet technologies in health applications. For example, see Lindberg and Humphreys (1998).
4. For example, systems to transfer large medical image files between sites can be designed in different ways. Some systems demand high network bandwidth because there is little preprocessing of images and little attention to representative workflows; others rely more on preprocessing, which reduces network bandwidth requirements.
5. For a more detailed history of MEDLARS and MEDLINE, see Smith and Mehnert (1986).
6. These data are from the Bureau of Economic Analysis as presented in U.S. Department of Commerce (1999), the Appendix to Chapter III.
7. For an example of a Web site enabling communications with insurers, see AT&T (1999).
8. Managed care plans integrate insurance and delivery of carefunctions otherwise provided by separate entities. Most managed care plans now pay care providers some form of discounted fee for services rendered, although some still pay a fixed fee based on the number of patients enrolled in their care.
9. Integrated delivery systems combine entities related to the provision of health care and may have relationships with health insurance plans. Such organizations typically include a range of different facilities, from major hospitals to local clinics, so they can provide a continuum of care.
10. For an example of Internet-based quality indicators and managed care data exchange, see Halamka and Hughes, 1998.
11. For example, researchers at Intermountain Health Care in Salt Lake City, Utah, have developed a system that provides clinical guidelines in real time to physicians who use the electronic medical record system. One study indicated that use of the system improved from 30 to 70 percent the percentage of diabetic patients with safe blood-sugar levels. It is estimated that the clinical guidelines have saved the organization $10 million, or $2,000 per patient, through improved clinical decision making (see Gillespie, 2000).
12. As noted by the Science Panel on Interactive Communications and Health (1999), self-care books provided to members of health maintenance organizations and Medicare beneficiaries have been shown to reduce office visits and specialty referrals (Vickery et al., 1988); systems to help patients prepare for office visits have been shown to improve treatment outcomes for chronic diseases (Greenfield et al., 1985); computer access to support groups and decision guidance has been shown to help women with breast cancer and patients with AIDS (Gufstafson et al., 1992, 1993, 1994); and shared decision-making tools have been shown to improve health outcomes while reducing the use of surgery and other high-cost medical procedures (Barry et al., 1995; Morgan et al., 1997).
13. These metrics are not independent of each other. For example, a high packet loss rate is likely to lead to low throughput because lost packets must be retransmitted, and the complete message cannot be reassembled until all packets are received.
14. Quality of service is distinct from reliability, which refers to the likelihood that a service remains available at all times. A network may be highly reliable in the sense that it is always possible to obtain connectivity to a given destination, but the same network may lack any assurance of performance (QOS, as defined here).
15. This level of isolation can be achieved even if there is some physical sharing at thecontinue
very lowest layer of the protocol stack; for example, the transmission links of the two different networks might share a physical fiber. At the same time, the separation is only as good as the trust of the user in the service provider. A simple misconfiguration of a router could connect a third-party link to a private network. In addition, the service provider has full access to the data carried over a private network.
16. As a result of recent consolidation in the insurance industry, for example, care providers now work with policies established in large corporate headquarters that are greater distances away, and standards for reducing the administrative burden on providers can no longer be set at the state level.
17. For more information, see <http://www.ccic.gov>.
18. The formal specification of the NGI program reverses the second and third items in the list above. The order of presentation is changed herein for stylistic purposes and to highlight that the development of testbed networks is just one element of a much broader-based program.
19. The term "telemedicine" refers to the delivery of health services when distance separates the care provider and patient (see Institute of Medicine, 1996). This construction recognizes that a range of different interactions are possible, from videoconferencing at the one extreme to the use of the telephone or text e-mail at the other. Indeed, the most prevalent uses of telemedicine today are not video-based but involve the use of asynchronous store-and-forward systems to exchange still images across networks. Other applications include telephone- or Internet-based systems for monitoring patients in their homes.
20. The study committee visited with the researchers at NASA Ames Research Center as part of this project. A summary of that visit is contained in Appendix A of this report.
21. It is expected that 25 more sites will be added to this testbed in FY00.
22. The vBNS is a nationwide network that supports high-performance, high-bandwidth research applications. Launched in 1995, it is the product of a 5-year cooperative agreement between NSF and MCI WorldCom. Approximately 100 research institutions, chosen through a peer-review process, will be connected to the network. It currently connects 92 institutions.
23. For additional information on Internet 2 and UCAID, see <http://www.ucaid.org>.
24. Belfast Telegraph Online 10/26/99 as summarized in "Internet 2 Gets Ready to Operate," Edupage, November 1, 1999.
25. This information was obtained from the Abilene Web site at <http://www.ucaid.edu/abilene>.
26. For additional information on these networks and the evolution of the Internet more generally, see Chapter 7 in Computer Science and Telecommunications Board (1999).break