NCI's Cancer Information Systems-Bringing Medical Knowledge to Clinicians

Oncology, ONCOLOGY Vol 9 No 4, Volume 9, Issue 4

The National Cancer Institute's computerized information systems have been designed to help physicians cope with the information explosion by translating the medical literature into usable forms. Systems developed by the

The National Cancer Institute's computerized information systems have been designed to help physicians cope with the information explosion by translating the medical literature into usable forms. Systems developed by the NCI's International Cancer Information Center provide access to a comprehensive source of bibliographic citations on cancer research (the CANCERLIT database) and to current, peer-reviewed syntheses of state-of-the-art clinical information on cancer (the PDQ database). This article describes how the PDQ databases were developed and are kept up-to-date, and how physicians and other health professionals can access PDQ and other NCI information by computer and fax.


The information explosion in biomedical science is a fact of life. Physicians are bombarded with medical information in textbooks, journals, bulletins, direct mail advertising, audiotapes, videotapes, online systems, CD-ROM, and interactive compact discs. Patients learn about health and disease from many sources and ask their physicians sophisticated and pointed questions. One of the major challenges that clinicians face is how to keep up with what is important.

In a 1989 survey on information management, involving 520 primary care practitioners and opinion leaders, two thirds of the physicians surveyed felt that the volume of medical literature was unmanageable [1], while 78% reported difficulty in screening out irrelevant data when reviewing the medical literature.

Health professionals are not alone in their need for current medical information. Patients and their families are faced with the arduous task of interpreting information in order to make well-informed decisions about their care. Lastly, those who bear the responsibility of making policy decisions on health-care issues must have access to state-of-the-art medical information.

Computerized systems for improving access to medical information and accelerating the pace with which information in the medical literature is translated into usable knowledge have been developed by the International Cancer Information Center (ICIC) of the National Cancer Institute (NCI) [2]. These systems provide access to a comprehensive source of bibliographic citations on published cancer research (the CANCERLIT database) and current, peer-reviewed syntheses of state-of-the-art clinical information on cancer (the PDQ database). Both of these computerized information systems are designed to assist users in maximizing their effective use of current medical knowledge [3]. In this paper, we will describe how these resources are providing health professionals, patients, and policy-makers with state-of-the-art information on cancer. The article will focus on the PDQ database, explaining the process by which it is updated and the variety of ways in which it can be accessed.

Physician Data Query (PDQ)

Physician Data Query (PDQ) is the NCI's comprehensive cancer information database. PDQ first became available to health professionals in 1984. Since then, the scope of information it contains has expanded, not only for health professionals but also for patients and their families. It is comprised of three main types of information:

1. Full text statements based on the published literature that reflect the current state-of-the-art information on the treatment, supportive care, prevention, and screening of cancer, as well as information about selected investigational and newly approved anticancer drugs.

2. Summaries of research protocols under evaluation in clinical trials.

3. Directories of physicians and organizations that provide cancer care.

PDQ has been described as a knowledge base, as distinguished from a database. It incorporates expert opinion in its selection of literature and synthesizes the literature into concise, clear summaries and recommendations. The database also contains relevant abstracts from the CANCERLIT database. In this way, PDQ provides a practical way of keeping up with advances from the clinical literature.

Full Text Statements

Treatment Information--PDQ contains prognostic and treatment information on the major types of cancer in children and adults, including information on AIDS-related malignancies. For each major type of cancer, there is a detailed statement on prognosis, staging, and treatment directed to the information needs of health professionals. Key citations to the literature are referenced, and abstracts of these citations are available for review by the user. A limited number of brief statements on less common cancers are also included. In addition, PDQ provides treatment statements for patients and their families that contain similar information written in nontechnical language.

In August, 1994, the International Cancer Information Center (ICIC) made awards to patient educators at eight cancer centers who submitted proposals to perform demonstration projects designed to evaluate PDQ's patient information statements. The objectives of the project are to encourage the development of interventions in clinical settings that will increase the use of the PDQ patient information statements; encourage centers to use these statements as an established resource; increase knowledge and understanding of NCI educational and scientific resources; ensure that cancer patients are receiving information on state-of-the-art cancer treatment options to help them make informed decisions regarding their care; increase the efficiency and efficacy of cancer patient education efforts; and define the role of PDQ as one of several patient education resources.

Supportive Care Information--PDQ contains supportive care statements describing the pathophysiology and treatment of common complications of cancer and its treatment, such as pain, hypercalcemia, and nausea/vomiting. Each statement generally contains an overview, information on etiology, assessment and management, and references to the current literature.

Screening and Prevention Information--The screening information in PDQ includes statements on screening for nine cancers: breast, cervical, oral, skin, colorectal, prostate, testicular, ovarian, and gastric. Each statement contains a summary of the available data concerning screening for that particular disease site, the levels of evidence for that summary (Table 1), and the significance and evidence of benefit for the summary statement. The statements also include references to the current literature that support the information in the statement.

In early 1994, information on cancer prevention was added to PDQ. The first statements are on prevention of aerodigestive cancers, cervical cancer, colorectal cancer, and skin cancer. Similar in format to the PDQ information on screening, these statements will also contain a summary of data concerning prevention for that particular disease site, levels of evidence for that summary (Table 2), and the significance and evidence of benefit for the summary statement.

Investigational and Newly Approved Drug Information--Currently, there are 11 statements on investiga- tional and newly approved anticancer agents. The statements include a description of each drug, its mechanism of action, indications and contraindications, interactions, dose schedules and modifications, and pharmaceutical information.

The Review Process--Information in the PDQ statements is peer-reviewed by five core editorial boards, one for each type of information covered by the cancer information statements: Adult Treatment Editorial Board, Pediatric Treatment Editorial Board, Supportive Care Editorial Board, Screening and Prevention Editorial Board, and Drugs Under Clinical Investigation Editorial Board. Board members have the task of translating information culled from the medical literature into medical knowledge that can be effectively used by health professionals.

The core boards are comprised of 65 cancer specialists, the majority of whom are not government employees. Each core editorial board is supplemented by an external advisory board that reviews the statements at least once a year. The external advisory boards are comprised of over 100 physicians and other health professionals with special expertise in the diagnosis, prevention, treatment, and supportive care of cancer. Members of the core and external advisory boards oversee the development and maintenance of the cancer information in PDQ. The core boards meet regularly to discuss recent literature and to develop new state-of-the-art statements or revise existing statements. The boards base their recommendations on the scientific literature and the clinical expertise of its core and external advisory members.

To assist board members in reviewing the medical literature effectively and promptly, a process has been developed to provide them with appropriate information (see Figure). Each month, professionals review the tables of contents of more than 70 biomedical journals to identify articles of potential relevance. After reviewing these articles, NCI staff forward the articles of highest potential relevance and scientific validity to appropriate editorial board members. In some cases, the boards also review prepublication data. For example, treatment statements on breast and prostate cancer were revised based on recommendations made at the NIH Consensus Development Conferences while the reports were still in draft stage.

The board members suggest and the editor-in-chief chooses topics to be discussed at the next board meeting, including important changes to the state-of-the-art statements, such as the deletion or addition of a treatment to the list of standard treatment options. When there is disagreement about the interpretation of data, PDQ statements address the controversial nature of the topic.

The methods used for developing and maintaining the content of the PDQ cancer information statements vary from one editorial board to another. For example, the Adult Treatment Board members maintain the treatment statements through informal consensus development. Before each board meeting, members are asked to evaluate specific articles regarding their appropriateness for inclusion in existing cancer treatment statements. At the meeting, members discuss the positive and negative aspects of each article or report, and make decisions as to whether the statements should be modified based on this information. Treatment options deemed appropriate are included but are not usually accorded any particular weight. Pertinent articles are listed for each option so that users can make their own treatment decisions based on the current literature.

Board members, concerned with the growing volume of medical knowledge and the increased emphasis on physician accountability and cost-effective practice, are considering ways to assist physicians in understanding the available medical literature and in applying it when making treatment decisions. Toward this end, the Adult Treatment Board members are moving toward an evidence-based approach for evaluating new information, a method already used by members of the Screening and Prevention Board. This approach has proven to be well suited to the controversial nature of much of the scientific data on cancer screening and prevention. It allows the Screening and Prevention Board members to summarize the information currently available on a particular topic, and to indicate the strength of the available evidence. Although the summaries are very concisely written, they afford the user a good understanding of the particular method of cancer screening or prevention and the strength of the data supporting it.

The practice of developing evidence-based summaries of the existing literature often draws attention to significant gaps in medical evidence. Pointing out these gaps is not only important for physicians and patients making personal decisions about screening and prevention, but also helps to define areas that are most in need of additional research. Each screening or prevention statement also includes text discussing the background information from which the summaries were derived. The formal levels of evidence used by the Screening and Prevention Board to evaluate the strength of available information are shown in Tables 1 and 2.

The use of an electronic medium to disseminate current information enables the boards to revise the statements constantly as new information becomes available. PDQ's cancer information statements are quite dynamic. On average, 30 of the 210 statements were revised each month over the past year. While many of these changes represent the addition of new literature references, about 90% of the changes were revisions of the statements themselves. About 30% of the statements changed each month have undergone substantial revisions. It is a mandatory requirement to all PDQ vendors that all data carry the date of retrieval. This indicates to the user whether the version they have is current.

Research Protocols

PDQ contains more than 1,500 summaries of clinical trials that are open or approved for patient accrual, including protocols for cancer treatment, supportive care, and screening and prevention. For each trial, detailed summaries are prepared from the original protocol document, ensuring uniformity and accuracy of the content. PDQ protocols can be retrieved by diagnosis, treatment modality, phase, locality, and drug name, or any combination of these

All protocols supported by the NCI are listed in PDQ. Board members review protocols submitted by investigators who are not directly supported by the NCI prior to their inclusion in PDQ. The criteria for evaluating voluntary research protocol submissions are shown in Table 3. Foreign protocols and clinical trials that are not supported by the NCI are included after review and approval using a process sanctioned by the PDQ Editorial Board. In addition, there are more than 7,000 summaries of protocols that have been completed or are no longer accepting patients.

PDQ provides a source of information on previous and ongoing clinical cancer research whether the results are positive or negative. For physicians, the ability to easily retrieve these data is essential to responsibly planning the use of cancer treatment resources.

Accessing PDQ and Other NCI Databases

In 1994, more than 60,000 domestic and 7,000 foreign centers had access to the cancer databases included on the MEDLARS system of the National Library of Medicine. This includes 8,000 student access accounts. There are currently 15 principal foreign MEDLARS centers that offer access to PDQ for foreign medical institutions and physicians. An online service operated by the European Organization for Research and Treatment of Cancer (EORTC) makes PDQ available to its participating organizations. PDQ is also licensed to commercial vendors and academic and non- profit health-care organizations with computerized medical information systems.

Mechanisms for accessing PDQ, some developed by the NCI and others by private vendors, fall into two general categories: online time-sharing systems with dial-up or Internet access, and "local" implementations that reside on a single computer or a local area network for use by individuals or groups of individuals. Local access to PDQ is currently available as a licensed product from the NCI for installation on a personal computer or local area network. PDQ is also offered as a local system on CD-ROM by two commercial sources. (1. J.B. Lippincott; 2. SilverPlatter, Inc.; 3. Teton Data Systems) Subscriptions to the CD-ROM products are sold on an annual basis and are updated monthly. They can be purchased to run on a DOS-based personal computer or a local area network. In the spring of 1994, the entire PDQ database was made available as part of a new membership program on the Internet [4,5].

CancerFax--Advances in technology allow facsimile (fax) boards to be placed inside personal computers so that the computer can act as a fax machine, sending and receiving documents simultaneously. When coupled with digital voice technology and software that interprets user selections from a touch-tone telephone, information from a database like PDQ can be delivered to a caller without human intervention. In 1991, NCI staff began distributing the treatment information from PDQ using fax on-demand technology. The service CancerFax allows users to dial into one of the NCI's computers from their fax machine and retrieve a faxed image of any of PDQ's cancer information statements.

Additional information, including important news items such as the reanalyses of the National Surgical Adjuvant Breast and Bowel Project (NSABP) trial results, is also available through CancerFax. Much of the information is provided in English or Spanish. The service is available 24 hours each day, and there is no cost to the user other than the cost of the telephone call. More than 80,000 CancerFax requests were fulfilled in 1994. Currently, CancerFax fulfills over 7,000 requests each month.

CancerNet--In July, 1992, the NCI introduced CancerNet, an electronic service that enables computer users to obtain free access to selected PDQ and other cancer-related information, 7 days a week, via Internet electronic mail (e-mail). Access to CancerNet data was expanded in 1993 by implementing it on Internet Gopher Servers in the United States, Japan, and Singapore. Current use of CancerNet via e-mail is approximately 5,000 requests each month, and 30% of these requests come from outside the United States. Internet Gopher Servers fulfill 50,000 requests each month, 40% of which come from outside the United States.

In addition to PDQ information, CancerFax and CancerNet contain fact sheets on various cancer-related topics produced by the NCI's Office of Cancer Communications, selected news items, information about ordering patient information and publications from the NCI, citations and abstracts from CANCERLIT for selected types of cancer, and PDQ availability information.

FedWorld--The PDQ database is also available on FedWorld, an electronic bulletin board system produced by the National Technical Information Service that provides a gateway to more than 100 federal agency bulletin board systems. A bulletin board system is a dial-up computer service that supports e-mail and other forms of message exchange, access to current news, and the downloading or uploading of data files and software.

NCI'S Information Associates Program--The Information Associates Program (IAP) is a membership program open to health professionals and organizations worldwide. The program provides members with direct access to the scientific information services of the ICIC, including the Journal of the National Cancer Institute, Journal Monographs, the PDQ database, patient education materials from the Office of Cancer Communications, and NCI bulletins. The annual membership fee for the Information Associates Program is $100 for domestic members and $150 for international members.

Members can access this information through many electronic information vehicles: fax (CancerFax), a bulletin board system, and the Internet. Trained representatives are available via a toll-free customer service line to assist members in accessing the services. The bulletin board system developed by the ICIC staff contains the PDQ database, information on ICIC products and services, important news items from the NCI, the table of contents and abstracts from issues of the Journal of the National Cancer Institute, Fact Sheets from the Office of Cancer Communications, selected abstracts from CANCERLIT on a variety of clinical topics, e-mail capability, and an electronic conferencing capability.

New Methods of Dissemination

Currently, ICIC staff are developing a prototype for delivering interactive multimedia products that will be disseminated using a new software program called Mosaic [6]. Mosaic is a widely distributed hypermedia system designed for the presentation and retrieval of information over the Internet. Developed by the National Center for Supercomputing Applications (NCSA) at the University of Illinois, Mosaic uses a client/server model for information distribution. Servers sit on computers at various Internet sites fulfilling queries sent by Mosaic clients who may be located anywhere on the Internet. Units of information sent from servers to clients are called documents and may contain plain text, formatted text, graphics, sound, and other multimedia data, as well as links to other documents that may be located at any other server on the Internet.

At present, there is no cost for the server or client software. This means that most PC, Macintosh, and UNIX desktop computers can become Mosaic clients. To take advantage of Mosaic sound, graphics, and animation, the client computer should have a sound board and super VGA graphics. These capabilities are standard on many modern computers and low cost options on others. Servers are generally larger, more powerful computers.

As part of its challenge to develop novel communication technologies, the ICIC staff plan to take advantage of this technical breakthrough and provide interactive programs of up-to-date information via the Internet and Mosaic. These programs will utilize graphical images, sound, and possibly full motion animation. A totally digital format is being utilized so that products will be "platform independent," with text, images, and sound that can be transmitted interactively via any communications medium, including high-speed communication networks and interactive cable TV. This technology will permit the ICIC to develop programs to distribute the Journal of the National Cancer Institute; Journal Monographs; special reports from the NCI; and multimedia educational programs on cancer prevention, screening, and treatment that would be made available to Information Associates Program members free of charge. One program, an interactive compact disc entitled "Cervical Cancer: Screening for Life," is already being distributed to program members.

There is increasing interest in using computer technology to meet the practicing physician's need for up-to-date medical information and expert medical advice [7,8]. Such interest is due in part to the availability of increasingly powerful and sophisticated microcomputers; greatly improved graphical interfaces; and affordable, easily used software tools, coupled with the growing concern about the increase in medical knowledge, medicolegal accountability, and cost-effective practice [9,10]. A key requirement for physician acceptance of automated information and decision support systems appears to be the smooth integration of the computer programs into the routine patient-care environment. Narrowly defined systems that serve a single purpose and require users to make special and/or duplicate efforts to access them have met with limited success, even though the information they provide has been shown, in some cases, to be both accurate and valuable for patient care.

The ICIC staff are identifying emerging communication technologies that can facilitate the dissemination of information on cancer through the federal government's small business innovation research (SBIR) program. Areas of research include improving telecommunication access; optimizing the user interface; refining interfaces to facsimile; and developing other advanced personal computing technologies, such as portable digital storage media, handwriting and voice-recognition interfaces, and handheld, wireless communication devices for information dissemination applications. Other novel channels of information distribution will be explored to bring cancer information to those who require it, whether health professionals, patients, or policy-makers. Appropriate choices cannot be made unless the full range of options is available to these decision-makers.

It has become clear that computer and communication technologies have a major role to play in information management and decision support in medicine. Future systems should assist users not only in finding pertinent information but also in determining how the data apply to specific clinical situations. An improved understanding of the nature of medical knowledge, the way it should be used for optimal decision-making, and the logistical issues that serve as barriers to widespread use, all pose continuing challenges for system developers.


1. Williamson JW, German PS, Weiss R, et al: Health science information management and continuing education of physicians: A survey of US primary care practitioners and their opinion leaders. Ann Intern Med 110:151-160, 1989.

2. Hubbard SM: Information systems in oncology: Information retrieval systems, in DeVita VT, Hellman S, Rosenberg SA (eds): Principles and Practice of Oncology, 4th Ed, pp 2581-2590. Philadelphia, JB Lippincott, 1993.

3. Harlan WR: Creating an NIH clinical trials registry: A user-friendly approach to health care. JAMA 22:1729, 1994.

4. Glowniak JW, Bushway MK: Computer networks as a medical resource: Accessing and using the Internet. JAMA 271:1934-1939, 1994.

5. Lincoln TL: Traveling the new information highway. JAMA 271:1955-1956, 1994.

6. Andreessen M, Bina E: NCSA Mosaic: A global hypermedia system. Internet Res 14:7-17, 1994.

7. Schwartz WB, Patil RS, Szolovits P: Artificial intelligence in medicine: Where do we stand? N Engl J Med 316:685-688, 1987.

8. Perry CA: Knowledge bases in medicine: A review. Bull Med Libr Assoc 78:271-282, 1990.

9. Shortliffe EH: Computer programs to support clinical decision making. JAMA 258:67-74, 1987.

10. Shortliffe EH, Wulfman CE, Rindfleisch TC, et al: An Integrated Oncology Workstation. Bethesda, Md, National Cancer Institute, 1990.