EMERGING INFECTIOUS DISEASES, VOLUME 1, NUMBER 4 Editors Editor Joseph E. McDade, Ph.D. National Center for Infectious Diseases Centers for Disease Control and Prevention (CDC) Atlanta, Georgia, USA Perspectives Editor Stephen S. Morse, Ph.D. The Rockefeller University New York, New York, USA Synopses Editor Phillip J. Baker, Ph.D. Division of Microbiology and Infectious Diseases National Institute of Allergy and Infectious Diseases National Institutes of Health (NIH) Bethesda, Maryland, USA Dispatches Editor Stephen Ostroff, M.D. National Center for Infectious Diseases Centers for Disease Control and Prevention (CDC) Atlanta, Georgia, USA Managing Editor Polyxeni Potter, M.A. National Center for Infectious Diseases Centers for Disease Control and Prevention (CDC) Atlanta, Georgia, USA Editorial and Computer Support Emerging Infectious Diseases receives editorial and computer support from the Office of Planning and Health Communication, National Center for Infectious Diseases. Editorial Support Maria T. Brito Anne D. Mather, M.A. Carol D. Snarey, M.A. Production Rita M. Furman, M.A. Electronic Distribution Carol Y. Crawford Cheryl G. Shapiro, M.S. Liaison Representatives Anthony I. Adams, M.D. Chief Medical Adviser Commonwealth Department of Human Services and Health Canberra, Australia David Brandling-Bennett, M.D. Deputy Director Pan American Health Organization World Health Organization Washington, D.C., USA Gail Cassell, Ph.D. Liaison to American Society for Microbiology University of Alabama at Birmingham Birmingham, Alabama, USA Richard A. Goodman, M.D., M.P.H. Editor, MMWR Centers for Disease Control and Prevention (CDC) Atlanta, Georgia, USA William Hueston, D.V.M., Ph.D Acting Leader, Center for Animal Health Monitoring Centers for Epidemiology and Animal Health Veterinary Services, Animal and Plant Health Inspection Service U.S. Department of Agriculture Fort Collins, Colorado, USA James LeDuc, Ph.D. Advisor for Arboviral Diseases Division of Communicable Diseases World Health Organization Geneva, Switzerland Joseph Losos, M.D. Director General Laboratory Center for Disease Control Ontario, Canada Gerald L. Mandell, M.D. Liaison to Infectious Diseases Society of America University of Virginia Medical Center Charlottesville, Virginia, USA Philip P. Mortimer, M.D. Director, Virus Reference Division Central Public Health Laboratory London, United Kingdom Robert Shope, M.D. Director, Yale Arbovirus Research Unit Yale University School of Medicine New Haven, Connecticut, USA Bonnie Smoak, M.D. Chief, Dept of Epidemiology Division of Preventive Medicine Walter Reed Army Institute of Research Washington, D.C., USA Robert Swanepoel, B.V.Sc., Ph.D. Head, Special Pathogens Unit National Institute for Virology Sandrinham 2131, South Africa Roberto Tapia, M.D. Director General de Epidemiolog¡a Direcci¢n General de Epidemiolog¡a Secretar¡a de Salud M‚xico Emerging Infectious Diseases Emerging Infectious Diseases is published four times a year by the National Center for Infectious Diseases, Centers for Disease Control and Prevention (CDC), 1600 Clifton Road., Mailstop C-12, Atlanta, GA 30333, USA. Telephone 404-639-3967, fax 404-639-3039, e-mail eideditor@cidod1.em.cdc.gov. The opinions expressed by authors contributing to this journal do not necessarily reflect the opinions of CDC or the institutions with which the authors are affiliated. All material published in Emerging Infectious Diseases is in the public domain and may be used and reprinted without special permission; proper citation, however, is appreciated. Use of trade names is for identification only and does not imply endorsement by the Public Health Service or by the U.S. Department of Health and Human Services. Electronic Access to Emerging Infectious Diseases If you have Internet access, you can retrieve the journal electronically through file transfer protocol (FTP), electronic mail, or World-Wide Web (WWW). The journal is available in three file formats: ASCII, Adobe Acrobat (.pdf), and PostScript (.ps). The ASCII version of the journal does not contain figures. Both the .pdf and .ps files, however, contain graphics and figures and are true representations of the hard copy of the journal. The Adobe Acrobat format requires an Adobe Reader. This reader is available in DOS, Windows, UNIX, and Macintosh versions. Installation instructions come with the Adobe software. Access Methods FTP: Download the journal through anonymous FTP at ftp.cdc.gov. The files can be found in the pub/Publications/EID directory in each of the file types listed above. WWW: Launch WWW browser for the Internet and connect to the following address: http://www.cdc.gov. Your WWW software will allow you to view, print, and retrieve journal articles. ListServer (e-mail lists): You may have the table of contents sent to your e-mail box by subscribing to the EID-TOC mailing list. When you subscribe to this list, you automatically receive the table of contents and will be able to receive individual journal articles by FTP or e-mail. If you choose to receive the entire journal, you may subscribe to one of three other lists. EID-ASCII sends the journal in ASCII format. EID-PDF sends the journal in Adobe Acrobat format. You can get the free Adobe Acrobat Reader by subscribing to the list. EID-PS sends the journal in PostScript format. However, because of the large file sizes and the complexity of sending the journal to different e-mail systems, it is strongly recommended that if you have FTP capabilities, you choose to access EID through FTP rather than by e-mail lists. To subscribe to a list, send an e-mail to lists@list.cdc.gov with the following in the body of your message: subscribe listname (e.g., subscribe EID-ASCII). Once you have requested a subscription, you will receive further instructions by e-mail. For more information about receiving Emerging Infectious Diseases electronically, send an e-mail to eidhelp@cidod1.em.cdc.gov. Instructions to Authors Editorial Policy and Call for Articles The goals of Emerging Infectious Diseases (EID) are to promote the recognition of new and reemerging infectious diseases and to improve the understanding of factors involved in disease emergence, prevention, and elimination. EID has an international scope and is intended for professionals in infectious diseases and related sciences. We welcome contributions from infectious disease specialists in academia, industry, clinical practice, and public health as well as from specialists in economics, demography, sociology, and other disciplines whose study elucidates the factors influencing the emergence of infectious diseases. EID will be published in English and will feature three types of articles: Perspectives, Synopses, and Dispatches. The purpose and requirements of each type of article are described in detail below. Instructions to Authors Editorial Material: Manuscripts should be prepared according to the Uniform Requirements for Manuscripts Submitted to Biomedical Journals (JAMA 1993:269[17]: 2282-6). Begin each of the following sections on a new page and in this order: title page, abstract, text, acknowledgments, references, each table, figure legends, and figures. On the title page, give complete information about each author (full names and highest degree). Give current mailing address for correspondence (include fax number and e-mail address). Follow Uniform Requirements style for references. Consult List of Journals Indexed in Index Medicus for accepted journal abbreviations. Tables and figures should be numbered separately (each beginning with 1) in the order of mention in the text. Double-space everything, including the title page, abstract, references, tables, and figure legends. Italicize scientific names of organisms from species name all the way up, except for vernacular names (viruses that have not really been speciated, such as coxsackievirus and hepatitis B; bacterial organisms, such as pseudomonads, salmonellae, and brucellae). Perspectives: Contributions to the Perspectives section are welcome from scientists and professionals in all disciplines and should address factors known to contribute to the emergence of infectious diseases, including microbial adaption and change; human demographics and behavior; technology and industry; economic development and land use; international travel and commerce; and the breakdown of public health measures. Articles should be approximately 3,500 words and should include references, not to exceed 40. The section should begin with an introduction outlining the relationship of the issues discussed in the paper to the emergence of infectious diseases. Use of additional subheadings in the main body of the text is recommended. If detailed methods are included, a separate section on experimental procedures should immediately follow the body of the text. Photographs and illustrations are optional. Provide a short abstract (150 words) and a brief biographical sketch. Synopses: Submit concise reviews of infectious diseases or closely related topics. Preference will be given to reviews of new and emerging diseases; however, timely updates of other diseases or topics are also welcome. Synopses should be approximately 3,500 words and should include references, not to exceed 40. The section should begin with an introduction outlining the relationship of the issues discussed in the paper to the emergence of infectious diseases. Use of additional subheadings in the main body of the text is recommended. If detailed methods are included, a separate section on experimental procedures should immediately follow the body of the text. Photographs and illustrations are encouraged. Provide a short abstract of no more than 150 words and a brief biographical sketch. Dispatches: Provide brief updates on trends in infectious diseases or infectious disease research. Dispatches (1,000 to 1,500 words of text) should be in a letter to the editor format and should not be divided into sections. Dispatches should begin with a brief introductory statement about the relationship of the topic to the emergence of infectious diseases. Provide references, not to exceed 10 and figures or illustrations, not to exceed two. All articles will be reviewed by independent reviewers. The Editor reserves the right to edit articles for clarity and to modify the format to fit the publication style of Emerging Infectious Diseases. Send documents in hardcopy (Courier 10-point font), on diskette, or by e-mail. Acceptable electronic formats for text are ASCII, WordPerfect, AmiPro, DisplayWrite, MS Word, MultiMate, Office Writer, WordStar, or Xywrite. Send graphics documents in Corel Draw, Harvard Graphics, Freelance, .TIF (TIFF), .GIF (CompuServe), .WMF (Windows Metafile), .EPS (Encapsulated Postscript), or .CGM (Computer Graphics Metafile). The preferred font for graphics files is Helvetica. If possible, convert Macintosh files into one of the suggested formats. Submit photographs in camera-ready hardcopy. Send all manuscripts and correspondence to the Editor, Emerging Infectious Diseases, National Center for Infectious Diseases, Centers for Disease Control and Prevention, 1600 Clifton Road, Mailstop C-12, Atlanta, GA 30333, USA, or by e-mail on the Internet to eideditor@cidod1.em.cdc.gov. CONTENTS, EMERGING INFECTIOUS DISEASES, VOL 1., NO. 4 Synopses The Ascension of Wildlife Rabies: A Cause for Public Health Concern or Intervention? Charles E. Rupprecht, Jean S. Smith, Makonnen Fekadu, and James E. Childs Diagnosis of Tuberculosis in Children: Increased Need for Better Methods, Ejaz A. Kahn and Jeffrey R. Starke Data Management Issues for Emerging Diseases and New Tools for Managing Surveillance and Laboratory Data, Stanley M. Martin and Nancy H. Bean Dispatches Helicobacter hepaticus, a Recently Recognized Bacterial Pathogen, Associated with Chronic Hepatitis and Hepatocellular Neoplasia in Laboratory Mice, Jerry M. Rice Hemolytic Uremic Syndrome Due to Shiga-like Toxin Producing Escherichia coli 048:H21 in South Australia, Paul N. Goldwater and Karl A. Bettelheim Does Treatment of Bloody Diarrhea due to Shigella dysenteriae Type 1 with Ampicillin Precipitate Hemolytic Uremic Syndrome? Abdulaziz A. A. Bin Saeed, Hassan E. El Bushra, and Nasser A. Al-Hamdan An Outbreak of Hemolytic Uremic Syndrome Associated with Antibiotic Treatment of Hospital Inpatients for Dysenstery, Sami Al-Qarawi, Robert E. Fontaine, and Mohammed-Saeed Al-Qahtani Epidemic Cholera in the New World: Translating Field Epidemiology into New Prevention Strategies, Robert V. Tauxe, Eric D. Mintz, and Robert E. Quick Are North American Bunyamwera Serogroup Viruses Etiologic Agents of Human Congenital Defects of the Central Nervous System?, Charles H. Calisher and John L. Sever Lymphocytic Choriomeningitis Virus: An Unrecognized Teratogenic Pathogen, Leslie L. Barton, C.J. Peters and T.G. Ksiazek Commentary Hemolytic Uremic Syndrome, Mary Beers and Scott Cameron News and Notes Guidelines on the Risk for Transmission of Infectious Agents During Xenotransplants, Louisa E. Chapman Emerging Infectious Diseases Featured at ICAAC/IDSA Meeting Building a Geographic Information System (GIS) Public Health Infrastructure for Research and Control of Tropical Diseases, Allen W. Hightower and Robert E. Klein APHA Session Features Emerging Infections, Martin S. Favero Southeast Asia Intercountry Consultative Meeting on Prevention and Control of New, Emerging, and Reemerging Infectious Diseases, Samlee Plianbangchang SYNOPSES (Page 107) The Ascension of Wildlife Rabies: A Cause for Public Health Concern or Intervention? Charles E. Rupprecht, V.M.D., Ph.D., Jean S. Smith, M.S. Makonnen Fekadu, D.V.M., Ph.D., and James E. Childs, Sc.D. Centers for Disease Control and Prevention, Atlanta, Georgia, USA ABSTRACT The epidemiology of rabies in the United States has changed substantially during the last half century, as the source of the disease has changed from domesticated animals to wildlife, principally raccoons, skunks, foxes, and bats. Moreover, the changes observed among affected wildlife populations have not occurred without human influence. Rather, human attraction to the recreational and economic resources provided by wildlife has contributed to the reemergence of rabies as a major zoonosis. Although human deaths caused by rabies have declined recently to an average of one or two per year, the estimated costs associated with the decrease in deaths amount to hundreds of millions of dollars annually. In future efforts to control rabies harbored by free-ranging animal reservoirs, public health professionals will have to apply imaginative, safe, and cost-effective solutions to this age-old malady in addition to using traditional measures. Introduction Rabies virus is the type species (serotype 1) of the Lyssavirus genus, a group of morphologically similar, antigenically and genetically related, negative-stranded RNA viruses, with a near global distribution (1). The lyssaviruses (Table 1) are well adapted to particular mammalian species (2) and rarely initiate panzootics. The public health threat of rabies as a preeminent zoonosis relates to the acute, incurable encephalitis that results from transmission of the virus by the bite of an infected animal. An estimated 40,000 to 100,000 human deaths are caused by rabies each year worldwide; in addition, millions of persons, primarily in developing countries of the subtropical and tropical regions (3), undergo costly postexposure treatment (PET). Although the number of human rabies cases has been significantly reduced in the United States, the total number of animal rabies cases approached historical limits in 1993. To appreciate the public health significance that lyssaviruses continue to play as persistent and emerging infectious agents, one must understand certain human activities, such as recent animal translocations (i.e., the natural or purposeful change by humans of the normal home range or geographic distribution of an animal) and animal ecology. Table 1. Recognized members of the genus Lyssavirus, family Rhabdoviridae Lyssavirus Rabies Reservoir Found worldwide, except for a few island nations, Australia, and Antarctica. Endemic and sometimes epidemic in a wide variety of mammalian species, including wild and domestic canids, mustelids, viverrids, and insectivorous and hematophagous bats; >25,000 human cases/year, almost all in areas of uncontrolled domestic dog rabies. History Descriptions of clinical disease in Greek and Roman documents. In the late 1800s, Pasteur attenuated the virus by serial passage and desiccation to vaccinate humans and animals. Pathognomonic inclusions in nerve cells described by Negri in 1903. An immunofluorescence test for rabies viral antigen developed in the 1950s. Lyssavirus Lagosbat Reservoir Unknown, but probably fruit bats. 10 cases identified to date, including 3 in domestic animals, in Nigeria, South Africa, Zimbabwe, Central African Republic, Senegal, and Ethiopia. No known human deaths. History Isolated in 1956 from brain of Nigerian fruit bats (Eidolon helvum) at Lagos Island, Nigeria, but not characterized until 1970; 3 cases in domestic animals initially diagnosed as rabies, but weak immunofluorescence led to suspicion of "rabies-related" virus, later confirmed by typing with monoclonal antibodies or nucleotide sequence analysis. Marginal cross-protection with rabies vaccines. Lyssavirus Mokola Reservoir Mokola, Unknown, but probably an insectivore or rodent species. Cases identified in Nigeria, South Africa, Cameroon, Zimbabwe, Central African Republic, and Ethiopia; 17 cases known, including 9 domestic animals and 2 human cases. History First isolated from Crocidura sp. shrews trapped in Mokola Forest near Ibadan, Nigeria, in 1968. Characterized in 1970. Like Lagos bat virus, evidence of infection with Mokola was recognized only by poor reaction with anti-rabies reagents. 7 domestic animal cases in Zimbabwe in 1981 and 1982 prompted serologic survey and identification of antibodies to Mokola in rodents, especially bushveld gerbils (Tatera leucogaster). No cross-protection with rabies vaccines. Lyssavirus Duvenhage Reservoir Unknown, but probably insectivorous bats. Cases identified in South Africa, Zimbabwe, and Senegal; 4 cases known, including 1 human death. No cases in domestic animals. History First identified in 1970 in rabies-like encephalitis in man bitten by an insectivorous bat near Pretoria, South Africa. Virus named after the victim. Although Negri bodies detected in histologic examination of brain tissue, negative immunofluorescence tests led to suspicion of rabies-related virus, subsequently confirmed by antigenic and genetic typing. Marginal cross-protection with rabies vaccines. Lyssavirus European bat Lyssavirus 1 (EBLV1) Reservoir European insectivorous bats (probably Eptesicus serotinus); >400 cases in bats. 1 confirmed human case in 1985 and a suspect case in 1977. No known domestic animal cases. History Although cases in European bats were reported as early as 1954, identification of the virus was not attempted until 1985, when the first of 100 infected bats was reported in Denmark and Germany. Almost all cases are in the common European house bat, E serotinus. Marginal cross-protection with rabies vaccines. Lyssavirus European bat Lyssavirus 2 (EBLV2) Reservoir European insectivorous bats (probably Myotis dasycneme); 5 cases identified, including 1 human death. No known domestic animal cases. First identified in isolate from Swiss bat biologist who died of rabies in Finland. Marginal cross- protection with rabies vaccines. Historical Perspectives The history of rabies in the New World reflects the interaction of chance, evolutionary constraint, ecologic opportunism, and human surveillance activities. Rabies may have existed in the United States before European colonization and the introduction of domestic animals incubating the disease. Various pathogens could have migrated during the exchanges of fauna and human populations over the Bering Strait some 50,000 years ago; folklore of a rabies-like malady among native people throughout the Pacific Northwest supports this notion (4). Records at the time of the Spanish conquest in Middle America associate vampire bats with human illness (5). If chiropteran rabies viruses were present and well established in the New World at the time of continental interchange, terrestrial virus counterparts also could have been present. Nonetheless, the first indication of terrestrial rabies did not surface until 1703 in what is now California (5). Dog and fox rabies outbreaks, reported commonly in the mid-Atlantic colonies throughout the late 1700s (4), were probably exacerbated by the introduction of dogs and red foxes (Vulpes vulpes), imported for British-style fox hunting, throughout New England in the 1800s; fox rabies epizootics ensued and spread to the eastern United States by the 1940s to 1950s (5,6). Skunk rabies reports were also frequent throughout the western states by the 19th century, and they were replete with cowboy tales of "phobey cats" (5). Although individual reports document a high incidence of dog rabies at the beginning of the last century, no national surveillance system existed. Human deaths from rabies in the United States were not commonly reported; the highest official record was of 143 cases, from a survey of death certificates in 1890. During 1938, when rabies in humans and other animals became a nationally reportable disease, the total number of rabies cases reported was 9,412 per year (mostly in domesticated species), with 47 human deaths. These numbers are certainly underestimates, since surveillance was limited, and sensitive diagnostic tests for human and animal rabies were not developed until the mid-1950s. An epizootiologic transition began in the United States in the 1920s, when rabies prevention efforts were no longer focused exclusively on human vaccination but began to include programs for the control of rabies in dogs. Domestic animal cases gradually declined, largely as a result of local dog rabies control programs that included vaccination, stray animal removal, and leash and muzzle ordinances. However, as such cases decreased, surveillance systems designed to track the source of infection for residual domestic animal foci detected increased cases in wild species. By 1960, rabies was diagnosed more frequently among wildlife than among domesticated animals. In 1971, rabies was reported for the first time from all 48 contiguous states and Alaska. Skunks (primarily the striped skunk, Mephitis mephitis) formed the major animal reservoir from 1961 to 1989, until they were unexpectedly supplanted by the raccoon (Procyon lotor) during the rabies outbreak in the mid-Atlantic and northeastern states (7). This epizootic is believed to have started during the late 1970s by the translocation of infected animals from a southeastern focus of the disease. The epidemiology of human rabies has also changed considerably over the last 50 years (8,9). From 1946 to 1965, 70% to 80% of human rabies cases occurred after a known exposure (most often a dog bite), and 50% of the cases before 1975 occurred after treatment with suboptimal vaccines. Over the last decade, 80% of rabies-related human deaths were among persons who had no definitive history of an animal bite (Table 2), and none resulted from postexposure prophylaxis failures. Almost all the recent human cases occurred after an animal exposure that was unrecognized by the patient as carrying a risk for rabies infection. The apparent source of human rabies has also changed: 14 of the 18 cases acquired in the United States since 1980 involved rabies variants associated with insectivorous bats (10). The latest report, in March 1995, typifies recent trends. A bat, subsequently found to be rabid, was found in the bedroom of a 4-year-old girl in Washington State. The child denied any contact with the bat, and no postexposure treatment was initiated. A bat-associated rabies virus variant was later identified in biopsy specimens from the child and from the bat's carcass (11). Despite the current prominence of raccoons as the largest wildlife reservoir in the United States (12), no documented human rabies cases have been associated with this ubiquitous carnivore. Table 2. Human rabies cases in the United States by exposure category, 1946-1995* Exposure source Years Domestic Wildlife Other Unknown (%) Case total 1946-1955 86 8 0 26 (22) 120 1956-1965 21 7 0 10 (26) 38 1966-1975 6 7 1 2 (13) 16 1976-1985 6 1 2 11 (55) 20 1986-1995* 2 2 0 14 (78) 18 * Through Oct. 1995. The Cost of Prevention Rabies prevention and control strategies in the United States have succeeded in lowering the number of human rabies deaths to an average of one to two per year. However, the reason for this low mortality level is a prevention program estimated to cost $230 million to $1 billion per year (13-15). This cost is shared by the private sector (primarily the vaccination of companion animals) and by the public (through animal control programs, maintenance of rabies laboratories, and subsidizing of rabies PET). Accurate estimates of these expenditures are not available. The number of PETs given annually in the United States is unknown, although the total must be substantially greater than the minimum of 20,000 estimated in 1980 to 1981 (16) when vaccine distribution was more tightly regulated. As rabies becomes epizootic or enzootic in a region, the number of PETs increases (17). Although the cost varies, a course of rabies immunoglobulin and five doses of vaccine given over a 4-week period typically exceeds $1,000. Potential exposure to a single rabid kitten in New Hampshire recently led to the treatment of more than 650 persons at an estimated cost of $1.5 million (18). Surveillance-related costs also rise as rabies becomes entrenched in wildlife. During 1993, the New York State rabies diagnostic laboratory received approximately 12,000 suspected animal submissions. This compares with approximately 3,000 submissions in 1989, before raccoon rabies became epizootic. In New Jersey, rabies prevention expenditures in two counties increased from $768,488 in 1988, before the raccoon epizootic, to $1,952,014 in 1990, the first full year of the epizootic (15); vaccination of pet animals accounted for 82% of this total. Vaccinated domestic animals are normally administered a booster vaccine dose after a known or suspected rabid animal exposure (19). This further increases costs, as wildlife rabies epizootics escalate. The cost per human life saved from rabies ranges from approximately $10,000 to $100 million, depending on the nature of the exposure and the probability of rabies in a region (20). What's more, most economic analyses do not take into account the psychological trauma caused by human exposure to rabies, the subsequent euthanasia of pets, or the loss of wildlife resources during rabies outbreaks. Rabies in wildlife has now reached historically high levels in the United States (12), and the costs of preventing human rabies are mounting. Human Influences and the Role of Translocation The colonization of the New World had a profound effect upon native fauna and consequent rabies epizootiology. Large-bodied carnivores, such as bears, cougars, wolves, and wolverines, were perceived as dangerous and killed outright. A few Carnivora have persisted and flourished. For example, the coyote (Canis latrans), a highly adaptable canid and the subject of many unsuccessful control programs, has been gradually expanding its range northward and eastward. Despite their widespread distribution and abundance (even in suburban neighborhoods), rabid coyotes have been reported rarely and sporadically, except for a brief period from 1915 to 1917, when an extensive outbreak occurred in portions of Utah, Nevada, California, and Oregon. While dog rabies has been largely controlled, a region of southern Texas that borders Mexico has persisted as a focus of both dog and coyote rabies. The number of cases of coyote rabies has gradually risen in this area since the late 1980s, accounting for 46 of the 50 cases of coyote rabies reported in the United States during 1991, 70 of 75 cases in 1992, and 71 of 74 cases in 1993 (12). The outbreak of coyote rabies has spread to the vicinity of San Antonio. One of the dangers of this outbreak is the continued spillover into the domestic dog population (21); at least 25 rabid dogs were reported from the area in 1991, 41 in 1992, and 54 in 1993 (12). Human rabies closely parallels the disease in domestic animals; at least two human deaths (in 1991 and 1994), probably due to coyote-dog interactions, have been associated with this canid outbreak in Texas (10,22). The translocation of infected coyotes from the south Texas focus is believed to be responsible for the transmission of this rabies variant to dogs in at least two other states: a single hunting dog in Alabama during 1993 (12) and at least seven cases of apparent dog-to-dog rabies transmission in Florida in 1994 (21). Expanded surveillance similar to that done in 1977 with the raccoon rabies focus in the mid-Atlantic region (7) is warranted for this canid virus. In this effort, state health departments should monitor unusual occurrences (such as the increased submission of canid specimens to the diagnostic facility), tracking of their time and location, and establishment of suitable public health interventions. These would include restricting further animal movements and enforcing mandatory companion animal rabies vaccination. Assessing control efforts is an important component of any intervention. In addition to the problems posed by the emergence of the coyote as a reservoir for rabies, the potential translocation of other species should be recognized. Since the transmission of rabies by a bat was first reported in 1953, rabid insectivorous bats have caused an average of 700 to 800 cases annually, and have been found throughout the United States, excluding Alaska and Hawaii (12). The discovery of these cases, coincident with the marked reduction of canine rabies cases, has afforded a certain epidemiologic luxury to enhance surveillance among wildlife. Similar to the Carnivora, the chiropteran families most important in rabies perpetuation (e.g., Vespertilionidae, Molossidae) have several species that are highly adaptable, abundant, and widespread. Rabies virus variants maintained by insectivorous bats appear to be exchanged largely independently from those in terrestrial mammalian reservoirs (23), despite documented spillovers. A similar epidemiologic situation exists among European bats, but with Lyssavirus genotypes (24) that can be readily differentiated from New World rabies isolates. The role of bats in Africa (25,26) in Lyssavirus maintenance is less clear (Table 1). Infections with non-rabies lyssaviruses have resulted in rabies vaccine failures (27). Such infections raise the specter of potentially serious public health consequences if introduced and subsequently established in susceptible bat populations. How probable is this scenario? The distances between Africa, Eurasia, Pacific Oceania, and the New World mitigate against the dispersal, migration, and introduction of healthy bats without human intervention (28). However, several recent events illustrate the opportunity for the transoceanic transfer of rabies-infected bats. In March 1986, researchers from Canada inadvertently shipped a big brown bat (Eptesicus fuscus) that was incubating rabies virus to colleagues in Tubingen, Germany. When the bat became ill and was euthanized, a diagnosis of rabies was made (29). A similar event occurred when Boston researchers collected a dozen wild big brown bats from Massachusetts during July 1994 and exported them to researchers in Denmark. By December 1994, six of the imported bats had died and were confirmed as positive for rabies virus by the Danish Veterinary Services, State Institute for Virus Research (L. Miller, pers. comm.). Commercial enterprises also serve as vehicles for the accidental translocation of animals infected with rabies virus. The first confirmed nonindigenous case of rabies in Hawaii resulted from the accidental introduction of a big brown bat (30). In March 1991, a bat was captured within a transport container unloaded from a ship in Honolulu harbor. The container held automobiles from Michigan loaded into the container ship in California. The local department of health laboratory diagnosed rabies; this was later confirmed, and the virus was characterized antigenically at the Centers for Disease Control and Prevention. The strain was a variant common to E. fuscus in the midwestern and western United States. None of the three instances cited above appear to have resulted in secondary cases or establishment of the virus in foreign animal populations. No unintended importations of non-rabies lyssaviruses to the United States have been documented. The likelihood of accidental introduction, escape, survival, and perpetuation of infected exotic bat species into the United States is remote. However, other more recent deliberate translocation activities may significantly enhance the probability of such introductions. During 1994, a number of improperly issued federal permits allowed as many as several thousand wild bats to be imported to the United States for sale in the commercial pet trade. These animals were primarily Egyptian tomb bats (Rousettus aegyptiacus), although several other bat species were imported as well. Sales of imported bats (and their offspring) to private collectors or as pets in the United States are prohibited, according to the Foreign Quarantine Regulations (42 CFR 71.54). Animals that may be vectors of diseases of public health concern are eligible for entrance only for restricted uses at accredited zoos or research institutions, where contact with the general public is limited. Imported bats that will be legally displayed normally undergo an extended quarantine period. Although no reports of lyssaviruses isolated from Egyptian fruit bats exist, active surveillance for such viruses has not been conducted. These bats are relatively common and widespread throughout the area that extends from Turkey and Cyprus to Pakistan, the Arabian peninsula, Egypt, and most of sub-Saharan Africa (31). Because they may roost by the thousands in a wide variety of habitats, there is ample opportunity for interaction with other Chiroptera, such as the widely distributed straw-colored (Eidolon helvum) or epauletted (Epomophorus wahlbergi) fruit bats; both of these species have been implicated in Lyssavirus epizootiology in Africa (25,26). The adaptability of Egyptian fruit bats should be a cause for concern because of the potential for survival and interaction among indigenous bat fauna, particularly in the southern United States. Additionally, beyond the obvious public health risks and foreign animal disease introduction, imported bat species should not be released into the wild because they may cause serious harm to local agriculture and may displace native species. Bats serve many critical ecologic functions worldwide and generally avoid contact with humans. However, they may be infected with many pathogens without demonstrating obvious clinical signs of infection. When bats are placed in a private household or pet shop, the hazard of disease transmission to humans is greatly increased. Persons currently possessing imported bats should be advised not to display them in settings where human contact can occur. Intervention Widespread, sustained population reduction of mammalian reservoirs to eliminate rabies is not justified (32) for ecologic, economic, and ethical reasons. Given the multispecies complexity and considerable geographic areas affected by wildlife rabies, and the opportunities for translocation, what alternative preventive strategies exist? Recent progress in implementing terrestrial wildlife rabies control programs elsewhere in the world has public health relevance for the United States. Oral rabies vaccination of the red fox with vaccine-laden baits is an integral aspect of rabies control throughout southeastern Canada and Europe, where more than 75 million doses of vaccine have been distributed over 5 million km2 during the past two decades (33). Consequently, rabies incidence among wild and domestic animals has fallen, as have PETS for human rabies. The raccoon rabies epizootic in the eastern United States provided renewed impetus for reconsidering oral vaccination technology, first conceived at the Center for Disease Control in the 1960s (34). The shift of the vaccination and baiting methods from a fox model to the raccoon involved extensive field and laboratory research during the 1980s. The existing attenuated rabies vaccines for foxes were shown to be less effective for raccoons and other carnivores (35,36). Additionally, studies of new candidate vaccines raised safety issues regarding vaccine-induced disease in wildlife (36). In 1983, a vaccinia-rabies glycoprotein (V-RG) recombinant virus vaccine was developed (37) that has proven to be an effective oral immunogen in raccoons and various other important reservoir species (38); vaccine advantages include improved thermostability and an inability to cause rabies. (Only the gene for the surface glycoprotein of a vaccine strain of rabies virus was included in the recombinant virus.) When vaccine-laden baits are offered under natural conditions, contact with them by nontarget wildlife species cannot be totally excluded. However, studies of V-RG virus have shown no vaccine-associated morbidity, mortality, or gross pathologic lesions in more than 40 warm-blooded vertebrate species examined. Moreover, with rare exceptions, there has been no contact-transfer of vaccine between vaccinated and control animals housed together (38); viral recovery has been limited to a few anatomical sites over a 48-h interval (39). While laboratory evaluations of target and nontarget species proceeded during 1987 to 1989, small-scale trials of V-RG were conducted in Belgium and France, with promising results (40). The first North American V-RG vaccine field trial began on August 20, 1990, on Parramore Island off the eastern shore of Virginia (41,42). This limited field trial demonstrated vaccine safety. Efficacy was also suggested: more than 80% of field-vaccinated raccoons survived a severe laboratory rabies challenge (7 months after V-RG release) to which more than 90% of control raccoons succumbed (43). A 1991 Pennsylvania study site closely approximated the ecologic communities of the eastern United States targeted for use of V-RG vaccine, while still maintaining relative biosecurity through its geographic barriers. The study at this site evaluated the rate of vaccine-laden bait contact and potential vaccine-related adverse effects among nonraccoon species, including rodents, carnivores, insectivores, and opossum. Raccoons and other furbearers demonstrated no adverse effects associated with vaccine contact. Examination of more than 750 nontarget individuals, representing 35 species, failed to demonstrate gross lesions suggestive of V-RG contact. After these safety trials, the first efficacy field experiments began in New Jersey during 1992 (44). Between spring 1992 and autumn 1994, more than 100,000 vaccine-laden fishmeal polymer baits were distributed by hand and helicopter over an area of 56,000 hectares. This trial attempted to create a population of immunized raccoons across the northern Cape May Peninsula to prevent the spread of epizootic raccoon rabies from affected portions of the state. Surveillance demonstrated a significant decrease in the rate of spread and overall rabies incidence in the target and other monitored areas (44), suggesting the potential effectiveness of this strategy. In the United States, oral vaccination of raccoons is now under way in Massachusetts (45), New York (46), and Florida, and an experimental extension of the program to coyotes is under way in south Texas. However, the future of such vaccination for wildlife in the United States may be seriously questioned. For oral vaccination to become an adjunct to traditional methods, the following major questions need to be answered: 1) What is the relationship between animal population density and the minimum density of vaccine/baits needed? 2) What level of herd immunity is necessary to eliminate rabies under various environmental circumstances? 3) What bait distribution techniques are optimal? 4) How can these methods be generalized from foxes and raccoons to other species, such as skunks, mongooses and dogs? 5) What long-term funding sources are available? 6) What are the various costs of rabies control and prevention methods? Given the problems inherent in wildlife control, the greater issue of extending these methods to the control of dog rabies in the developing world will be a challenge well into the next century. Dr. Rupprecht is chief of the Rabies Section, Ms. Smith is a research microbiologist, Dr. Fekadu is a research veterinary medical officer, and Dr. Childs is chief of the Epidemiology Section, Viral and Rickettsial Zoonoses Branch, Division of Viral and Rickettsial Diseases, National Center for Infectious Diseases, Centers for Disease Control and Prevention, Atlanta, Georgia. Acknowledgment The authors gratefully acknowledge the technical expertise of the staff of the Rabies and Epidemiology Sections, Viral and Rickettsial Zoonoses Branch, Division of Viral and Rickettsial Diseases, National Center for Infectious Diseases, Centers for Disease Control and Prevention, without whose assistance this work would not have been possible. Address for correspondence: Charles E. Rupprecht, Centers for Disease Control and Prevention, 1600 Clifton Rd., MS G33, Atlanta, GA 30333, USA; fax 404-639-1058; e-mail cyr5@ciddvd1.em.cdc.gov. References 1. World Health Organization. World survey of rabies 28 for the year 1992. Geneva: World Health Organization, 1994. 2. Wandeler A, Nadin-Davis SA, Tinline RR, Rupprecht CE. Rabies epizootiology: an ecological and evolutionary perspective. In: Rupprecht CE, Dietzschold B, Koprowski H, editors. Lyssaviruses. New York: Springer-Verlag, 1994:297-324. 3. Meslin FX, Fishbein DB, Matter HC. Rationale and prospects for rabies elimination in developing countries. In: Rupprecht CE, Dietzschold B, Koprowski H, editors. Lyssaviruses. New York: Springer-Verlag, 1994:1-26. 4. Winkler WG. Foxrabies. In: Baer GM, editor. The natural history of rabies. 1st ed. New York: Academic Press, 1975:3-22. 5. Baer GM. Rabies -- an historical perspective. Infectious Agents and Disease 1994;3:168-80. 6. Carey AB, Giles RH, McLean RG. The landscape epidemiology of rabies in Virginia. Am J Trop Med Hyg 1978;27:573-80. 7. Rupprecht CE, Smith JS. Raccoon rabies -- the re-emergence of an epizootic in a densely populated area. Semin Virol 1994;5:155-64. 8. Held JR, Tierkel ES, Steele JH. Rabies in man and animals in the United States, 1946-65. Public Health Rep 1967;82:1009-18. 9. Anderson LJ, Nicholson KG, Tauxe RV, Winkler WG. Human rabies in the United States, 1960 to 1979: epidemiology, diagnosis, and prevention. Ann Intern Med 1984;100:728-35. 10. Centers for Disease Control and Prevention. Human rabies -- Alabama, Tennessee, and Texas, 1994. MMWR 1995;44:269-72. 11. Centers for Disease Control and Prevention. Human rabies Washington state, 1995. MMWR 1995; 44:625-7. 12. Krebs JW, Strine TW, Smith JS, Rupprecht CE, Childs JE. Rabies surveillance in the United States during 1993. J Am Vet Med Assoc 1994;205:1695-709. 13. Stehr-Green JK, Schantz PM. The impact of zoonotic diseases transmitted by pets on human health and the economy. Vet Clin North Am Small Anim Pract 1987;17:1-15. 14. Fishbein DB, Arcangeli S. Rabies prevention in primary care: a four-step approach. Postgrad Med 1987;82:83-90. 15. Uhaa IJ, Dato VM, Sorhage FE, et al. Benefits and costs of using an orally absorbed vaccine to control rabies in raccoons. J Am Vet Med Assoc 1992;201:1873-82. 16. Helmick CG. The epidemiology of human rabies postexposure prophylaxis, 1980-1981. JAMA 1983;250:1990-6. 17. Centers for Disease Control and Prevention. Raccoon rabies epizootic: United States, 1993. MMWR 1994;43:269-73. 18. Centers for Disease Control and Prevention. Mass treatment of humans exposed to rabies -- New Hampshire, 1994. MMWR 1995;44:483-6. 19. Centers for Disease Control and Prevention. Compendium of animal rabies control, 1995. MMWR 1995;44:(RR-2):1-9. 20. Fishbein DB, Robinson LE. Rabies. N Engl J Med 1993;329:1632-8. 21. Centers for Disease Control and Prevention. Translocation of coyote rabies -- Florida, 1994. MMWR 1995;44:580-1, 7. 22. Centers for Disease Control and Prevention. Human rabies -- Texas, Arkansas, and Georgia, 1991. MMWR 1991;40:765-9. 23. Smith JS, Orciari LA, Yager PA. Molecular epidemiology of rabies in the United States. Semin Virol (in press). 24. Bourhy H, Kissi B, Lafon M, Sacramento D, Tordo N. Antigenic and molecular characterization of bat rabies virus in Europe. J Clin Microbiol 1992;30:2419-26. 25. Swanepoel R, Barnard BJH, Meredith CD, et al. Rabies in southern Africa. Onderstepoort J Vet Res 1993;60:325-46. 26. King AA, Meredith CD, Thomson GR. The biology of southern Africa lyssavirus variants. In: Rupprecht CE, Dietzschold B, Koprowski H, editors. Lyssaviruses. New York: Springer-Verlag, 1994:267-96. 27. Foggin CM. Mokola virus infection in cats and a dog in Zimbabwe. Vet Rec 1983;113:115. 28. Wiles GJ, Hill JE. Accidental aircraft transport of a bat to Guam. J Mamm (full title) 1986;67:600-1. 29. World Health Organization Collaborating Center for Rabies Surveillance and Research. Bat rabies cases in the Federal Republic of Germany. World Health Organization Rabies Bulletin Europe 1986;10:8-9. 30. Sasaki DM, Middleton CR, Sawa TR, Christensen CC, Kobayashi GY. Rabid bat diagnosed in Hawaii. Hawaii Med J 1992;51:181-5. 31. Nowak RM. Walkers mammals of the world. 5th ed. Baltimore: Johns Hopkins University Press, 1991:198. 32. Debbie JG. Rabies control of terrestrial wildlife by population reduction. In: Baer GM, editor. The natural history of rabies. 2nd ed. Boca Raton, FL: CRC Press, 1991:477-84. 33. World Health Organization. Oral immunization of foxes in Europe in 1994. Wkly Epidemiol Rec 1995;70:89-91. 34. Baer GM. Oral rabies vaccination: an overview. Rev Infect Dis 1988;10 (Suppl 4):S644-8. 35. Rupprecht CE, Dietzschold B, Cox JH, Schneider L. Oral vaccination of raccoons (Procyon lotor) with an attenuated (SAD-B19) rabies virus vaccine. J Wildl Dis 1989;25:548-54. 36. Rupprecht CE, Charlton KM, Artois M, et al. Ineffectiveness and comparative pathogenicity of attenuated rabies virus vaccines for the striped skunk (Mephitis mephitis). J Wildl Dis 1990;26:99-102. 37. Wiktor TJ, Macfarlan RI, Reagan KJ, et al. Protection from rabies by a vaccinia virus recombinant containing the rabies virus glycoprotein gene. Proc Natl Acad Sci USA 1984;81:7194-8. 38. Rupprecht CE, Hanlon CA, Hamir AN, Koprowski H. Oral wildlife rabies vaccination: development of a recombinant virus vaccine. Transactions of the North American Wildlife Natural Resources Conference 1992;57:439-52. 39. Rupprecht CE, Hamir AN, Johnston DH, Koprowski H. Efficacy of a vaccinia-rabies glycoprotein recombinant virus vaccine in raccoons (Procyon lotor). Rev Infect Dis 1988;10 (4 Suppl):S803-9. 40. Aubert MFA, Masson E, Artois M, Barrat J. Oral wildlife rabies vaccination field trials in Europe, with recent emphasis on France. In: Rupprecht CE, Dietzschold B, Koprowski H, editors. Lyssaviruses. New York: Springer-Verlag, 1995:219-44. 41. Hanlon CA, Hayes DE, Hamir AN, et al. Proposed field evaluation of a rabies recombinant vaccine for raccoons Procyon lotor: site selection target species characteristics and placebo baiting trials. J Wildl Dis 1989;4:555-67. 42. Hanlon CA, Buchanan JR, Nelson E, et al. A vaccinia-vectored rabies vaccine field trial:ante- and post-mortem biomarkers. Rev Sci Tech 1993;99-107. 43. Rupprecht CE, Hanlon CA, Niezgoda M, Buchanan JR, Diehl D, Koprowski H. Recombinant rabies vaccines: efficacy assessment in free-ranging animals. Onderstepoort J Vet Res 1993;60:463-8. 44. Roscoe DE, Holste W, Niezgoda M, Rupprecht CE. Efficacy of the V-RG oral rabies vaccine in blocking epizootic raccoon rabies. Presented at the 5th Annual International Meeting of Rabies in the Americas, Niagara Falls, Ontario, Canada, 1994, Abstract, p.33. 45. Robbins AH, Niezgoda M, Levine S, et al. Oral rabies vaccination of raccoons (Procyon lotor) on the Cape Cod Isthmus, Massachusetts. Presented at the 5th Annual International Meeting of Rabies in the Americas, Niagara Falls, Ontario, Canada, 1994; Abstract, p. 29. 46. Hanlon CA, Trimarchi C, Harris-Valente K, Debbie JG. Raccoon rabies in New York State: epizootiology, economics, and control. Presented at the 5th Annual International Meeting of Rabies in the Americas, Niagara Falls, Ontario, Canada, 1994; Abstract, p.16. (Pages 115-23) Diagnosis of Tuberculosis in Children: Increased Need for Better Methods Ejaz A. Khan, M.D., and Jeffrey R. Starke, M.D. Baylor College of Medicine, Houston, Texas, USA ABSTRACT In the last decade tuberculosis (TB) has reemerged as a major worldwide public health hazard with increasing incidence among adults and children. Although cases among children represent a small percentage of all TB cases, infected children are a reservoir from which many adult cases will arise. TB diagnosis in children usually follows discovery of a case in an adult, and relies on tuberculin skin testing, chest radiograph, and clinical signs and symptoms. However, clinical symptoms are nonspecific, skin testing and chest radiographs can be difficult to interpret, and routine laboratory tests are not helpful. Although more rapid and sensitive laboratory testing, which takes into account recent advances in molecular biology, immunology, and chromatography, is being developed, the results for children have been disappointing. Better techniques would especially benefit children and infants in whom early diagnosis is imperative for preventing progressive TB. Despite the availability of effective preventive measures and chemotherapy, the prevalence of tuberculosis (TB) is increasing in the developing world and in much of the industrialized world as well (1-4). According to World Health Organization (WHO) estimates, in 1990 there were 8 million new cases of TB and 3 million deaths due to the disease worldwide; 1.3 million new cases and 450,000 deaths were among children under 15 years of age (5). WHO projects that 90 million new cases and 30 million deaths -- including 4.5 million deaths among children -- will occur in the 1990s (6,7). In developing countries, the risk for TB infection and disease is relatively uniform in the population; annual rates of infection often exceed 2% (5,6). In industrialized countries, risk is more uneven and depends on the individual's past or present activities and exposure to persons at high risk for the disease (Table 1). From 1987 to 1991, the number of TB cases among children under 5 years of age in the United States increased by 49% from 674 cases to 1006 (8). Although cases among children represent a small percentage of all TB cases, infected children are a reservoir from which many adult cases will arise. The risk for infection by Mycobacterium tuberculosis among children depends primarily on the level of risk of developing infectious TB for the adults in their immediate environment, especially their household. Because most current diagnostic tests for TB infection and disease have low specificity and therefore low positive predictive values, epidemiologic investigation continues to be important in establishing the diagnosis of TB in children. In industrialized countries, clinicians and public health professionals in TB services must always ask: Has the child been exposed to an adult with infectious pulmonary TB? Table 1. Persons at high risk for Mycobacterium tuberculosis infection in industrialized countries Persons likely to be exposed to or become infected with M. tuberculosis Close contacts of a person with infectious tuberculosis (TB) Foreign-born persons from high-incidence areas (e.g., Asia, Africa, Latin America) The elderly Residents of long-term care facilities (e.g., correctional facilities and nursing homes) Persons who inject drugs Other groups identified locally as having increased prevalence of TB (e.g., migrant farm workers or homeless persons) Persons who may have occupational exposure to TB Persons at high risk of developing TB disease once infected Persons recently infected with M. tuberculosis (within the past 2 years) HIV-infected persons Persons with immunosuppressing conditions or medication use Persons with a history of inadequately treated TB Infants Natural History of TB in Children The natural history of TB in children follows a continuum; however, it is useful to consider three basic stages: exposure, infection, and disease (1). Exposure implies that the child has had recent and substantial contact with an adult or adolescent who has suspected or confirmed contagious pulmonary TB (a source case). Exposed children are usually identified during followup investigations for persons with suspected pulmonary TB by public health workers (9); the child's tuberculin skin test (TST) is nonreactive, the results of the chest radiograph are normal, and the child is free of physical signs or symptoms of TB. Some - exposed children are infected with M. tuberculosis. The clinician cannot know immediately which exposed children are infected because the development of delayed-type hypersensitivity to tuberculin may take up to 3 months. Unfortunately, in children under 5 years of age, severe TB -- especially meningeal and disseminated disease -- can occur in fewer than 3 months, before the TST becomes reactive (10). Young children in the exposure stage should receive chemotherapy, usually isoniazid, until infection can been excluded. TB infection is first signaled by a reactive Mantoux TST. In this stage, there are no signs or symptoms, and the results of the chest radiograph are either normal or show only fibrotic lesions or calcifications in the lung parenchyma or regional lymph nodes. In developing countries, TB infection is rarely discovered and almost never treated. In most industrialized countries, children with a positive TST receive isoniazid for 6 to 12 months. TB disease occurs when signs and symptoms or radiographic manifestations caused by M. tuberculosis appear. Radiographic abnormalities and clinical manifestations in infected children probably are influenced by the host inflammatory reaction more than by the number of organisms. Studies show that in 40% to 50% of infants with untreated TB infection disease develops within 1 to 2 years (11). The risk decreases to 15% among older children. In 25% to 35% of children TB is extrapulmonary and more difficult to confirm bacteriologically. In adults, the distinction between TB infection and disease is usually clear because most disease is caused by reactivation of dormant organisms years after infection. Disease in adults is usually accompanied by symptoms, and patients frequently are infectious. In children, who most often have primary disease, the interval between infection and disease can be several months to several years, and radiographic abnormalities often are not accompanied by symptoms; moreover, these children are rarely infectious. The major reason for separating infection from disease in children is that the perception affects the approach to treatment: infection is generally treated with a single anti-TB drug, whereas active disease is treated with two or more drugs. The rationale for the difference in treatment is that the likelihood of emergence of resistance to a drug increases as the bacillary population increases (3). This distinction is somewhat artificial in children since infection and primary disease are parts of a continuum. Because anti-TB medications are well tolerated by children and are relatively inexpensive in industrialized countries, the usual paradigm of infection and disease encourages overtreatment rather than undertreatment. Asymptomatic lymphadenopathy and mild lung parenchymal changes are labeled and treated as disease. When evaluating new diagnostic tests the basic differences between the pathophysiology of TB in adults and children should be considered. Among children with recent TB infection, active multiplication of mycobacteria occurs with or without the presence of radiographic abnormalities or clinical symptoms. For example, gastric aspirate cultures yield M. tuberculosis from a small proportion of recently infected children with normal chest radiographs. One can anticipate that most diagnostic tests designed to detect M. tuberculosis in adults with TB disease will be positive in some proportion of children who have what is usually called TB infection. It will take careful consideration and investigation to determine if and how the results of these new tests should influence the definitions and treatment of TB infection and disease in children. Established Diagnostic Methods Tuberculin Skin Test (TST) The Mantoux TST, which uses five tuberculin units of purified protein derivative, is the standard method for detecting infection by M. tuberculosis. The reaction is measured as millimeters of induration after 48 to 72 hours. Since TST is the only way to determine asymptomatic infection by M. tuberculosis, the false-negative rate cannot be calculated. A negative TST does not rule out TB disease in a child. Approximately 10% of otherwise normal children with culture-proven TB do not react to tuberculin initially (12,13). Most of these children have reactive skin tests during treatment, which suggests that TB disease contributed to the immunosuppression. In most cases, the anergy occurs to all antigens, but in some cases, reactions to tuberculin are negative but reactions to other antigens remain positive (13). The rate of false-negative TST is higher in those who are tested soon after becoming infected with severe TB; in children with debilitating or immunosuppressive illnesses, malnutrition, or viral and certain bacterial infections; in infants; and if poor technique is used (1,14). The rate of false-negative TST in children with TB who are infected with human immunodeficiency virus (HIV) is unknown, but it is certainly higher than 10%. False-positive reactions to TST are often attributed to asymptomatic infection by environmental nontuberculous mycobacteria (NTM). Vaccination with M. bovis can cause transient reactivity to a subsequent TST, but the association is weaker than commonly recognized. Most 80% to 90% in several studies -- children who received BCG as infants have a nonreactive TST at 5 years of age (15-17). Among older children or adolescents who receive BCG, most develop a reactive skin test initially; however, by 10 to 15 years postvaccination, 80% to 90% have lost tuberculin reactivity (18,19). Skin test reactivity can be boosted, probably by antigenic stimulation, by serial testing in many children and adults who received BCG (20). Various factors determine the TST reaction size after receipt of BCG (1). Many recipients of BCG have a reactive TST because they are infected with M. tuberculosis and are at risk for disease, especially if they have had recent contact with an infectious TB patient (18). In general, TST reaction should be interpreted in the same manner for persons who have received BCG (20) and for unvaccinated persons. The relatively low sensitivity and specificity of TST make the test very useful for persons at high risk for TB infection or disease but undesirable for use in persons at low risk (21,22). The predictive values of TST can be improved by varying the size of induration considered positive according to epidemiologic risk factors for infection (Table 2). However, most of even 15-mm reactions in children at low risk are false-positive results, and testing of persons at low risk should be discouraged. Although the scheme in Table 2 is scientifically and mathematically valid, it assumes that the clinician and family are willing and able to develop an accurate history for TB risk factors for children and adults in their environments. Table 2. Cut-off size of reactive area for a positive Mantoux tuberculin reaction ò5 mm ò10 mm ò15 mm Persons who had Foreign-born persons No risk contact with from high-prevalence factors infectious persons countries Persons with an Residents of prisons, abnormal chest nursing homes, radiograph institutions HIV-infected and Persons who inject other immuno- drugs suppressed persons Persons with other medical risk factors Health-care workers Locally identified populations at high risk Children in contact with adults at high risk Infants Clinical Signs and Symptoms Two scenarios lead the clinician to suspect that a child has TB disease. The first occurs when TB is considered during the differential diagnosis for an ill child. This is a common scenario in the developing world but is less common in developed countries. Infants are more likely to be symptomatic than older children with pulmonary TB. The most common symptoms are cough, fever, wheezing, and failure to gain weight (13). Clinical signs are surprisingly meager, but rales and wheezes over the affected lung field are most common. Signs and symptoms of extrapulmonary TB are referable to the involved organ. The sensitivity and specificity of signs and symptoms are extremely low and can lead to both overdiagnosis and underdiagnosis when radiographs and other tests are not available. The second scenario occurs when evaluating a child who has had significant contact with an adult with suspected or confirmed TB. Usually the TST is applied first and is reactive. A subsequent chest radiograph or physical examination leads to discovery of early disease. The child is usually relatively asymptomatic. In the United States, about 50% of childhood TB cases are discovered in this manner (13). Radiologic Studies Evidence of pulmonary TB in chest radiographs varies (23,24), but usually radiographs show enlargement of hilar, mediastinal, or subcarinal lymph nodes and lung parenchymal changes (Figure 1 cannot be viewed in ASCII format). Most of the radiographic abnormalities are caused by a combination of lung disease and the mechanical changes induced by partial or complete airway obstruction resulting from enlarging intrathoracic nodes. The most common findings are segmental hyperinflation, then atelectasis, alveolar consolidation, interstitial densities, pleural effusion, and, rarely, a focal mass. Cavitation is rare in young children but is more common in adolescents, who may develop reactivation disease similar to that seen in adults. The development of radiographic techniques, such as computed tomography (CT) scanning, illustrates some of the issues that arise when newer and more sensitive diagnostic tests become available (24). A CT scan may show enlarged or prominent mediastinal or hilar lymph nodes in some children with recent TB infection and a normal chest radiograph (25). In the absence of a CT scan, the child's disease stage would be called TB infection, and single drug therapy would be used. Many studies, involving thousands of children have shown this treatment to be successful. However, when the CT scan shows mild adenopathy, the clinician may consider this finding indicative of TB disease and treat with several drugs, although this probably is not necessary in the absence of drug resistance. These findings reinforce the idea that pediatric TB is a continuum, and the distinction between infectionand disease is somewhat artificial. There is no current role for the CT scan in the evaluation of the asymptomatic TB-infected child with a normal chest radiograph. This scan can be helpful in selected cases to demonstrate endobronchial disease, pericardial invasion, early cavitation, and bronchiectasis resulting from pulmonary TB when the chest radiograph is abnormal but the pathologic process is not clear. Mycobacterial Detection and Isolation Despite recent advances, early mycobacteriologic diagnosis of TB still relies primarily on examination of acid-fast-stained smears from clinical specimens. It is the easiest, least expensive, and most rapid procedure for obtaining preliminary information. However, children under 12 years of age with pulmonary TB rarely produce sputum and are usually unable to expectorate voluntarily. When sputum samples cannot be obtained, gastric aspirate samples are used for detection and isolation of M. tuberculosis. Even though an acid-fast bacilli (AFB) stain of sputum is positive in up to 75% of adults with pulmonary TB, fewer than 20% of children with TB have a positive AFB smear of sputum or gastric aspirate (26,27). The newer fluorochrome stains, such as auramine and rhodamine, are superior to classic carbolfuchsin stains (28). The rates of positive AFB stain from body fluids and tissues in children with extrapulmonary TB also are low, and false-positive results caused by NTM disease are common, especially in cervical lymph nodes. For most children with pulmonary TB, culture confirmation is not needed. Diagnosis is made on the basis of a positive TST, clinical and radiographic findings suggestive of TB, and history of contact with an adult source case. The drug-susceptibility test results from the source case isolate can be used to design the optimal treatment for the child. However, cultures should be obtained from the child if the source patient is unknown or has a drug-resistant organism and if the child is immunocompromised or has extrapulmonary TB. The best specimen for culture from children with suspected pulmonary TB is the early morning gastric aspirate obtained in the hospital by using a nasogastric tube before the child arises and peristalsis empties the stomach of the respiratory secretions swallowed overnight (29,30). Three consecutive morning gastric aspirates yield M. tuberculosis in only 30% to 50% of cases, although the yield from infants is as high as 70% (14). The culture yield from other body fluids or tissues from children with extrapulmonary TB is usually less than 50% (13). Gastric aspiration is inconvenient, expensive, and uncomfortable. The culture yield from random, outpatient gastric aspirates has not been determined recently. Therefore, this procedure cannot be recommended but should be studied. Bronchoscopy The role of bronchoscopy in evaluating children for TB is controversial. The culture yield is lower from bronchoscopy specimens than from properly obtained gastric aspirates (29,31). Most children do not need flexible fiberoptic bronchoscopy, but the procedure may be useful in diagnosing endobronchial TB and excluding other causes of pulmonary abnormality, particularly in immunocompromised children, such as those with HIV infection in whom other opportunistic infections may coexist with or mimic TB. In a recent study of 36 children with pulmonary TB, bronchoscopy showed endobronchial involvement in 42%; most (63%) of these children had no clinical or radiographic evidence of endobronchial TB (31). This technique may be used to determine if a child might benefit from corticosteroid therapy, but guidelines for making this decision have not been established. Clinical Scoring System TB is an enormous problem in developing countries, where about 95% of cases occur (6). Cost, technical difficulties, and lack of resources make TB diagnosis in children very difficult in these countries. Various clinical scoring systems have been proposed on the basis of available information and tests (32,33) (Table 3). Although helpful, many of these systems have low sensitivity and specificity. However, even in industrialized countries, the triad of a positive tuberculin skin test, an abnormal radiograph, and a history of exposure to an adult with TB remains the most effective method for diagnosing TB in children. Table 3. A set of criteria for the diagnosis of pulmonary tuberculosis (TB) in children when culture is not available A. Positive acid-fast stain of sputum or gastric aspirate or B. Two or more of the following: þ History of contact with a tuberculous adult þ Cough lasting longer than 2 weeks þ A reactive tuberculin skin test ò10 mm in children without prior BCG vaccination ò15 mm in children with prior BCG vaccination þ Radiographic findings compatible with TB þ Response to anti-TB therapy (increased body weight by 10% after 2 months, decrease in symptoms) Source: ref. 33. New Diagnostic Techniques Polymerase Chain Reaction (PCR) Diagnostic PCR is a technique of DNA amplification that uses specific DNA sequences as markers for microorganisms (34). In theory, this technique can detect a single organism in a specimen such as sputum, gastric aspirate, pleural fluid, cerebrospinal fluid, or blood. Recent publications show that various PCR techniques, most using the mycobacterial insertion element IS6110 as the DNA marker for M. tuberculosis-complex organisms, have a sensitivity and specificity greater than 90% for detecting pulmonary TB in adults (35,36). However, these tests are not performed correctly in all clinical laboratories (36) and may offer little advantage over high-quality microscopic examination of sputum (34). The cost involved and the need for sophisticated equipment and scrupulous technique to avoid cross-contamination of specimens preclude the use of PCR techniques in many developing countries. PCR may have a special role in the diagnosis of extrapulmonary TB and pulmonary TB in children since sputum smears are usually unrevealing in these cases. Use of PCR for detecting M. tuberculosis in children has not been evaluated extensively. Pierre et al. (37) used an IS6110-based PCR to detect M. tuberculosis in gastric aspirate samples from 22 children with pulmonary TB. They found that 15 (25%) of 59 samples were positive; however, testing multiple samples or testing samples at least twice improved the sensitivity. When three samples from the same patient were tested two times each, two or more positive results were obtained from 9 of 15 children with TB, but from 0 of 17 controls. However, 2 of 65 single samples from controls were positive by PCR. Using an IS6110-based PCR assay, Starke et al. (38) tested gastric aspirates from 35 hospitalized children with pulmonary TB and 30 controls to detect M. tuberculosis. When compared with the clinical diagnosis, PCR had a sensitivity of 40% and specificity of 80%. Six controls had false-positive PCR results; one had a recent TB infection, two had NTM disease, and three had conditions unrelated to mycobacterial infection. Delacourt et al. (39) studied 199 specimens from 68 children with suspected TB. An IS6110-based PCR identified M. tuberculosis in clinical samples from 83% of children with disease compared to the low yield from positive AFB smears (21%) and positive cultures (42%) (39). PCR identified 70% of children with clinical pulmonary TB but no other microbiologic proof of the infection. However, 39% of children with infection but no radiographic or clinical disease also had positive PCR results. These results again demonstrate the arbitrariness of the distinction between TB infection and disease in children. It appears that PCR may have a useful but limited place in evaluating children for TB. A negative PCR result never eliminates TB as a diagnostic possibility, and a positive result does not confirm it. PCR's major use will be in evaluating children with significant pulmonary disease, when the diagnosis is not easily established by clinical or epidemiologic grounds. PCR may be particularly helpful in evaluating immunocompromised children with pulmonary disease, although published reports of PCR performance in such children are lacking. PCR may also aid in establishing the diagnosis of extrapulmonary TB, though only rare case reports have been published. However, performing PCR on gastric aspirates is not a useful test to distinguish between TB infection and disease and should not be used for children with normal chest radiographs. Serology and Antigen Detection Despite dozens of studies published over the past several decades, serology has found little place in the routine diagnosis of TB in adults or children. Several recent studies have used the enzyme-linked immunosorbent assay (ELISA) to detect antibodies to various purified or complex antigens of M. tuberculosis in children. Rosen (40) used mycobacterial sonicates in an ELISA on samples from 31 children with clinical TB and found a sensitivity of 26% and a specificity of 40%. This ELISA was influenced by recent BCG vaccination in children under 5 years of age. Barrera et al. (41) used an ELISA that detects antibodies to purified protein derivative and found a sensitivity of 51% for culture-positive pulmonary TB cases in children, but the sensitivity was only 28% for the clinical cases. Hussey et al. (42) used an autoclaved suspension of M. tuberculosis to detect antibodies in serum from 132 children with clinical pulmonary TB; the test was 62% sensitive and 98% specific. Higher sensitivity was obtained among patients with positive culture results (69%, n = 35), miliary TB (100%, n = 6), tuberculous meningitis (80%, n = 15), and pleural effusion (78%, n = 16). No correlation was observed with the tuberculin skin-test result, BCG vaccination, or nutritional status whereas duration of therapy, increasing age and chronicity of infection were positively correlated. Delacourt et al. (43) used an ELISA to detect IgG and IgM antibodies directed against mycobacterial antigen A60 in children with TB. At a chosen specificity of 98%, IgG was detected in 68% of children with clinical disease when results were highly controlled for age and prior BCG vaccination. IgM detection had only a 19% sensitivity. However, using the same anti-A60 ELISA at a defined specificity of 95%, Turneer et al. (44) found the IgG sensitivity to be 26% for past TB, 6% for asymptomatic primary TB, 14% for symptomatic TB, and 9% for NTM adenitis. No available serodiagnostic test for TB has adequate sensitivity, specificity, or reproducibility under various clinical conditions to be useful for diagnosing TB in children. Mycobacterial antigen detection has been evaluated in clinical samples from adults, but rarely from children (45,46). Two recent assays detecting M. tuberculosis-specific antigens yielded high sensitivity and specificity in various clinical specimens from adults with TB (47,48). Measurement of tuberculostearic acid, a mycobacterial mycolic acid, has been used to detect M. tuberculosis in clinical specimens (49). Brooks et al. (50) demonstrated a sensitivity of 95% and specificity of 91% when chromatographic profile of carboxylic acids and detection of tuberculostearic acid were combined and compared with culture results and clinical findings in adults with pulmonary TB; however, these techniques require technically advanced equipment and expertise, which are not available where TB in children is most common. Their sensitivity and specificity in children are unknown. Implications for HIV Infection and Drug Resistance The resurgence of TB over the last decade has coincided with the HIV pandemic. HIV-infected infants and children are in close contact with their caregivers, who may be infected with HIV and M. tuberculosis and are at high risk of developing infectious TB as they become immunocompromised. None of the available diagnostic tests for TB infection or disease in children has been evaluated systematically in children with HIV infection and pulmonary disease or suspected TB. In adults with HIV infection and TB, the sensitivity of diagnostic tests that rely on the host immune response, such as the TST or serology, is much lower than in nonimmunocompromised TB patients. It is likely that the tests' sensitivity also will be lower in HIV-infected children with TB. Tests that directly detect M. tuberculosis, such as PCR or antigen detection assays, may be particularly important for HIV-infected children. The culture yield of M. tuberculosis from children with HIV infection and TB is unknown but appears to be similar to that from non-HIV-infected children. The most important diagnostic clue for detecting TB in HIV-infected children is a history of contact with an adult who has infectious TB. Since TB may not have yet been diagnosed in this adult, a rapid and aggressive evaluation for TB in adults who care for the child is a critical part of the evaluation of the child. The current prevalence of drug resistance among M. tuberculosis isolates in the United States is 8% to 14% (51, 52). Drug resistance is most common in patients who received treatment, are not responding to therapy, do not adhere to treatment, live in developing countries, are immunocompromised, are prisoners, are homeless, or are children exposed to adults at increased risk for drug resistance. Drug-resistant TB has increased significantly among children (52). Because of low culture yields from children with TB, the clinician must often rely on the antimicrobial susceptibility results for the M. tuberculosis isolate obtained from the adult source case who presumably infected the child. This again emphasizes the crucial need to identify and evaluate the source case for every child with TB. The rapid identification of drug-resistant organisms is necessary for control of drug-resistant TB. Various new methods, such as high-performance liquid chromatography or PCR and DNA sequence analysis, may help to identify and test for antimicrobial susceptibility within a few days of diagnosis, but these techniques remain experimental. Summary Most recently developed sensitive and specific diagnostic tests have not found a place in the routine evaluation of children with suspected TB. Clinical criteria, particularly skin-test results, radiographic changes, and documented exposure to an infectious adult remain standard diagnostic methods. In industrialized countries, the local public health entity is a crucial partner to the clinician in establishing the diagnosis in the child and determining if drug resistance is present. As new diagnostic tests are developed, they must be evaluated against clinical criteria. The basic differences in pathophysiology of TB in adults and children must be considered before new tests are applied in pediatrics. It will be crucial to study the new techniques in children and not simply extrapolate from results for adults with TB. Dr. Khan is a postgraduate fellow in pediatric infectious diseases at Baylor College of Medicine, Houston, Texas. Dr. Starke is an associate professor at Baylor College of Medicine and current chairman of CDC's Advisory Committee for the Elimination of Tuberculosis. Address for correspondence: Jeffrey R. Starke, Texas Children's Hospital, MC 3-2371, 1102 Bates Street, Houston, tX 77030, USA; fax 713-770-4347; e-mail jstark@msmailpo2.is5.tch.tmc.edu. References 1. Starke JR, Correa AG. Management of mycobacterial infection and disease in children. Pediatr Infect Dis J 1995;14:455-70. 2. Styblo K, Rouillon A. Tuberculosis in developing countries: burden, intervention and cost. Bull Int Union Against Tuber Lung Dis 1990;65:6-24. 3. Starke JR, Jacobs R, Jereb J. Resurgence of tuberculosis in children. J Pediatr 1992;120:839-55. 4. Cantwell M, Snider D Jr, Cauthen G, Onorato I. Epidemiology of tuberculosis in the United States, 1985 through 1992. JAMA 1994;272:535-9. 5. Kochi A. The global tuberculosis situation and the new control strategy of the World Health Organization. Tuber Lung Dis 1991;72:1-6. 6. Raviglione MC, Snider DE, Kochi A. Global epidemiology of tuberculosis. Morbidity and mortality of a worldwide epidemic. JAMA 1995;273:220-6. 7. Dolin P, Raviglione M, Kochi A. Global tuberculosis incidence and mortality during 1990-2000. Bull World Health Organ 1994;72:213-20. 8. Barnes PF, Borrows, SA. Tuberculosis in the 1990s. Ann Intern Med 1993;119:400-10. 9. Hsu KHK. Contact investigation: a practical approach to tuberculosis eradication. Am J Public Health 1963;53:1761-9. 10. Nolan RJ Jr. Childhood tuberculosis in North Carolina: a study of the opportunities for intervention in the transmission of tuberculosis to children. Am J Public Health 1986;76:26-30. 11. Brailey ME. Tuberculosis in white and negro children. II. The epidemiologic aspects of the Harriet Lane study. Cambridge, MA: Harvard University Press, 1958. 12. Steiner P, Rao M, Victoria MS, et al. Persistently negative tuberculin reactions: their presence among children culture positive for M. tuberculosis. Am J Dis Child 1980;134:747-50. 13. Starke JR, Taylor-Watts KT. Tuberculosis in the pediatric population of Houston, Texas. Pediatrics 1989;84:28-35. 14. Vallejo J, Ong LT, Starke JR. Clinical features, diagnosis and treatment of tuberculosis in infants. Pediatrics 1994;94:1-7. 15. Lifschitz M. The value of the tuberculin skin test as a screening test for tuberculosis among BCG-vaccinated children. Pediatrics 1965;36:624-7. 16. Landi S, Ashley MJ, Grzybowski S. Tuberculin sensitivity following the intradermal and puncture methods of BCG vaccination. Can Med Assoc J 1967;97:222-5. 17. Joncas JH, Robitaille R, Gauthier T. Interpretation of the PPD skin test in BCG-vaccinated children. Can Med Assoc J 1975;113:127-8. 18. Johnson H, Lee B, Kelly E, McDonnell T. Tuberculin sensitivity and the BCG scar in tuberculosis contacts. Tuber Lung Dis 1995;35:113-7. 19. Menzies R, Vissandjee B. Effect of bacille Calmette-Guerin vaccination on tuberculin reactivity. Am Rev Respir Dis 1992;141:621-5. 20. Sepulveda RL, Burr C, Ferrer X, Sorensen RU. Booster effect of tuberculin testing in healthy 6-year-old school children vaccinated with bacille Calmette-Guerin at birth in Santiago, Chile. Pediatr Infect Dis J 1988;7:578-82. 21. American Thoracic Society. Diagnostic standards and classification of tuberculosis. Am Rev Respir Dis 1990;142:725-35. 22. American Academy of Pediatrics Committee on Infectious Diseases. Screening for tuberculosis in infants and children. Pediatrics 1994;93:131-4. 23. Schaaf HS, Beyers N, Gie RP, et al. Respiratory tuberculosis in childhood: the diagnostic value of clinical features and special investigations. Pediatr Infect Dis J 1995;14:189-94. 24. Parisi MT, Jensen MC, Wood BP. Pictorial review of the usual and unusual roentgen manifestations of childhood tuberculosis. Clin Imag 1994;18:149-54. 25. Delacourt C, Mani TM, Bonnerot V, et al. Computed tomography with normal chest radiograph in tuberculous infection. Arch Dis Child 1993;69:430-2. 26. Strumpf IJ, Tsang AY, Syre JW. Reevaluation of sputum staining for the diagnosis of pulmonary tuberculosis. Am Rev Respir Dis 1979;119:599-602. 27. Lipsky BA, Bates J, Tenover FC, Plorde JJ. Factors affecting the clinical value of microscopy for acid-fast bacilli. Rev Infect Dis 1984;6:214-22. 28. Kent PT, Kubica GP. Public health mycobacteriology a guide for the level III laboratory. Atlanta, GA; Centers for Disease Control, 1985. 29. Abadco DL, Steiner P. Gastric lavage is better than bronchioalveolar lavage for isolation of Mycobacterium tuberculosis in childhood tuberculosis. Pediatr Infect Dis J 1992;11:735-8. 30. Carr DT, Karlson AG, Stillwell AA. A comparison of cultures of induced sputum and gastric washings in the diagnosis of tuberculosis. Mayo Clinic Proc 1967;42:23-5. 31. Chan S, Abadco DL, Steiner P. Role of flexible fiberoptic bronchoscopy in the diagnosis of childhood endobronchial tuberculosis. Pediatr Infect Dis J 1994;13:506-9. 32. Glidey Y, Hable D. Tuberculosis in childhood: an analysis of 412 cases. Ethiop Med J 1983;21:161-7. 33. Migliori AB, Borghesi A, Rossanigo P et al. Proposal for an improved score method for the diagnosis of pulmonary tuberculosis in childhood in developing countries. Tuber Lung Dis 1992;73:145-9. 34. Schluger NW, Rom WN. Current approaches to the diagnosis of active pulmonary tuberculosis. Am J Respir Crit Care Med 1994;149:264-7. 35. Eisenach KD, Sifford MD, Cane MD, Bates JH, Crawford JT. Detection of Mycobacterium tuberculosis in sputum samples using a polymerase chain reaction. Am Rev Respir Dis 1991;144:1160-3. 36. Noordhock A, Kolk A, Bjune G, et al. Sensitivity and specificity of polymerase chain reaction for detection of Mycobacterium tuberculosis: a blind comparison study among seven laboratories. J Clin Microbiol 1994;32:277-84. 37. Pierre C, Oliver C, Lecossier D, Bousssougant Y, Yemi P, Hance AJ. Diagnosis of primary tuberculosis in children by amplification and detection of mycobacterial DNA. Am Rev Respir Dis 1993;147:420-4. 38. Starke JR, Ong LT, Eisenach KD, et al. Detection of M. tuberculosis in gastric aspirate samples from children using polymerase chain reaction. Am Rev Resp Dis 1993;147(Suppl):A801. 39. Delacourt C, Poveda J-D, Churean C, et al. Use of polymerase chain reaction for improved diagnosis of tuberculosis in children. J Pediatr 1995;126:703-9. 40. Rosen EU. The diagnostic value of an enzyme-linked immunosorbent assay using adsorbed mycobacterial sonicates in children. Tubercle 1990;71:127-30. 41. Barrera L, Miceli I, Ritacco V, et al. Detection of circulating antibodies to purified protein derivative by enzyme-linked immunosorbent assay: its potential for the rapid diagnosis of tuberculosis. Pediatr Infect Dis J 1989;8:763-7. 42. Hussey G, Kibel M, Dempster W. The serodiagnosis of tuberculosis in children: an evaluation of an ELISA test using IgG antibodies to M. tuberculosis, strain H37RV. Ann Trop Paediatr 1991;11:113-8. 43. Delacourt C, Gobin J, Gaillard J-L, de Blic J, Veran M, Scheinmann P. Value of ELISA using antigen 60 for the diagnosis of tuberculosis in children. Chest 1993;104:393-8. 44. Turneer M, Nerom EV, Nyabenda J, Waelbroeck A, Duvivier A, Toppet M. Determination of humoral immunoglobulins M and G directed against mycobacterial antigen 60 failed to diagnose primary tuberculosis and mycobacterial adenitis in children. Am J Respir Crit Care Med 1994;150:1508-12. 45. Sada E, Ruiz-Palacios AM, Lopez-Vidal Y, et al. Detection of mycobacterial antigens in cerebrospinal fluid of patients with tuberculous meningitis by enzyme-linked immunosorbent assay. Lancet 1983;2:651-2. 46. Radhakrishnan VV, Sehgal S, Mathai A. Correlation between culture of Mycobacterium tuberculosis and detection of mycobacterial antigens in cerebrospinal fluid of patients with tuberculous meningitis. J Med Microbiol 1990;33:223-6. 47. Wadee AA, Boling L, Reddy SG. Antigen capture assay for detection of a 43-kilodalton Mycobacterium tuberculosis antigen. J Clin Microbiol 1990;28:2786-91. 48. Sada E, Aguilar D, Torres M, et al. Detection of lipoarabinomannan as a diagnostic test for tuberculosis. J Clin Microbiol 1992;30:2415-18. 49. Brooks JB, Daneshvar MI, Fast DM, et al. Selective procedures for detecting femtomole quantities of tuberculostearic acid in serum and cerebrospinal fluid by frequency-pulsed electron-capture gas-liquid chromatograph. J Clin Microbiol 1987;25:1201-6. 50. Brooks JB, Daneshvar MI, Harberger RL, et al. Rapid diagnosis of tuberculous meningitis by frequency-pulsed electron-captive gas-liquid chromatography detection of carboxylic acids in cerebrospinal fluid. J Clin Microbiol 1990;28:989-97. 51. Centers for Disease Control and Prevention. National action plan to combat multidrug-resistant tuberculosis. MMWR 1992;41:5-50. 52. Bloch AB, Cauthen GM, Onorato IM, et al. Nationwide survey of drug-resistant tuberculosis in the United States. JAMA 1994;271:665-71. (Pages 124-8) Data Management Issues for Emerging Diseases and New Tools for Managing Surveillance and Laboratory Data Stanley M. Martin, M.S., Nancy H. Bean, Ph.D. Centers for Disease Control and Prevention, Atlanta, Georgia, USA Data Management Issues for Emerging Diseases Since 1976, when Legionnaires' disease affected attendees at the American Legion Convention in Philadelphia (1), the scope of public health has expanded. During the 1976 outbreak investigation, public attention was drawn to news accounts of the increasing numbers of cases and deaths as well as to speculations about diseases causes and prevention. After the outbreak, public health officials contended with volumes of information, including clinical data, epidemiologic survey results, and records of specimens collected from patients and the environment. This information was managed on mainframe computers. In 1980, a cluster of cases of unrecognized illness, primarily affecting young women, created a data management situation similar to that surrounding the Legionnaires' disease outbreak. A major epidemiologic investigation, which included examining a multitude of laboratory specimens and analyzing volumes of data, was undertaken by a large team of federal, state, and local public health officials, as well as numerous academic institutions and private industries. The problems with establishing databases and implementing a data management system for toxic shock syndrome (2) were essentially the same as the data management problems of Legionnaires' disease, except that computer technology had crept forward slightly in public health offices. During the spring of 1993, a cluster of cases of another unknown illness, eventually attributed to hantavirus (3), occurred in the southwestern United States. The reaction to this unknown disease by public health officials reflected a startling fact: even though the epidemiologic and laboratory methods for curtailing the outbreak were in place, a consistent data management strategy had not been established. Ad hoc databases built by outbreak investigators for a multitude of purposes began to bog down the investigation. Cases were recorded in multiple databases that did not recognize duplicate reports of cases. Updates of data about cases were done in some, but not all, databases. Laboratory data about specimens from patients were not linked to other clinical and epidemiologic data about a patient. No single database was available with well-edited, complete data about all the cases. Parallel, fragmented data management efforts evolved in at least 15 locations, with no coordinated mechanism to integrate them into one system. Introducing a single system for data management in the midst of the hantavirus outbreak involved more than the data management issues encountered in the earlier outbreaks. Previously, computer technology was viewed as a solution that, although somewhat cumbersome, enabled officials to move from data management by hand to electronic management. However, during the hantavirus outbreak, computer technology became part of the problem; it initially prevented good data management and may have hindered some of the laboratory and epidemiologic efforts to control the outbreak. Data were essentially being locked into various databases and could not be adequately analyzed or merged with data in other databases. In some instances, this peculiar circumstance caused investigators to perform analyses by hand using printouts from electronic databases or entering data again into other systems. In recent years, legal considerations, such as the Privacy Act enacted in 1974 and the Freedom of Information Act enacted in 1966 (4,5), have also complicated data management. These acts, in their efforts to protect individual privacy and ensure availability of data, have in some cases, constrained public health responses to emergency situations and subsequent surveillance efforts by enforcing strict database design and handling requirements. Data Management Requirements In epidemiologic investigations, disease problems are generally characterized by person, place, and time, whether the problem concerns the emergence of a new disease, a change in the resistance pattern of a known pathogen, an emergency response to an outbreak, or a routine disease surveillance program. The principles of data gathering, management, and analysis are essentially the same for all these purposes. Computer systems developed to manage data associated with these problems should be regarded as tools for the epidemiologic characterization of pathogens, syndromes, cases, and risk factors. Therefore, laboratory data management and reporting systems must be able to handle data about all of these. The most stringent requirements for data management are imposed by data from laboratory testing of specimens from patients, human and nonhuman sources, and the environment. A system having a relational data model adequate to properly handle the laboratory data requirements will almost certainly be adequate to handle the clinical, exposure, and demographic data requirements. Two primary data management functions can satisfy the laboratory data demands with multiple requirements in each function. The first function, internal laboratory data management, consists of entering test results and tracking specimens. The second, surveillance, includes gathering data and moving data beyond the electronic files of the laboratory to appropriate sites for analysis. A data management system should be able to perform these functions not only during an outbreak but throughout the period of surveillance as well. The internal laboratory function, universally similar among most public health laboratories, includes data entry tailored for individual laboratories at the site; retrieval/query ability; and ability to add or delete tests, manage aliquots, share data input in different laboratories of the site, track the status of every specimen regardless of which laboratory tested it, develop reports for specimen submitters, and in some cases assign costs for laboratory tests performed and prepare invoices for submitters. Requirements for the surveillance function include, in addition to certain critical laboratory data, the following facilities: to record clinical, exposure/risk factor, and demographic data about patients; to include data about multiple specimens and aliquots related to the same person, regardless of the interval separating the specimen dates; and to change questions or test results that are recorded for each specimen. Although internal and surveillance functions are clearly separate, they are not independent. Data entered into databases for the internal function should be available without additional effort for the surveillance function. In fact, when the internal function is not electronic or when the internal electronic system is inadequate, the system performing electronic surveillance should also perform to some extent the internal functions. Good laboratory data management does not address the internal function at the exclusion of the surveillance function. If a laboratory data management system is to be useful for emergency situations, it must provide mechanisms for adapting quickly to the emergency situation. For example, it must provide a way to immediately create an electronic data collection instrument and to incorporate this new instrument into the system at all reporting sites electronically. For the surveillance function, these electronic features must include communications facilities to move data electronically from one location to another, mechanisms for sending messages or files, functions for simple analysis, and methods for preparing and printing reports. While some systems perform some of these functions, most systems do not provide all of them. With appropriate systems in hand, data management plans for both urgent and routine events can be approached in a sequential fashion. With consensus among all participating investigators, epidemiologists must decide what data (both laboratory and epidemiologic) are needed so that data field characteristics can be defined. Consensus should be reached in the early phase of the outbreak investigation; otherwise participants in the investigation will of necessity begin developing ad hoc data management systems. The more thoroughly and carefully this task is performed, the more stable the data will ultimately become. In a well-designed system, the initial definitions in an emergency situation can include projections about which data fields will be needed. However, for routine surveillance these can be more thoroughly planned. Thus, the data system should allow fields to be deleted if not needed and to be added if they become important. These modifications should 1) be handled without having to alter the system, 2) use simple menu-driven functions requiring no computer programmer intervention, 3) accomplish the changes immediately, 4) be distributed to all investigators without disrupting their other functions during the investigation, and 5) be incorporated automatically. Next, all known participants in the investigation must be identified. These should include local, state, and federal officials as well as academic or private participants who may provide reports to the central data repository. These participants must be identified to the system specifically by person and by site for system security. Appropriate state and federal offices should be informed concerning the computer system and the rules for its use well before an emergency occurs; therefore, sites will be on the system in advance of an urgent problem. However, the system must allow for additional sites to be added quickly. In an emergency, a temporary agreement must be drawn for all participants to cooperate with the demands of the situation, i.e., to use a particular software system and operate under a standard set of rules for collecting and reporting data for the emergency. This agreement may occasionally stipulate that participants share data temporarily in a common database for the sake of data integrity. Entering clinical, epidemiologic/risk factor, and laboratory data about the same cases into the same database, rather than merging separate databases after the data are collected, provides such great payoffs in time savings and data integrity that the effort to obtain cooperation for a common database during an urgent situation is worthwhile. Although merging multiple databases during routine surveillance is feasible, emergency situations do not lend themselves to this type of data management. Therefore, the system to be used for these situations must accommodate a common database and provide a means of connecting the reporting sites to the database. When the reporting system is activated and data begin arriving at a central location, the system should facilitate analysis at every reporting site and provide a mechanism to export data (e.g., ASCII or .dbf files) for external analysis. Emergency situations create unusual demands for epidemiologic and laboratory resources; therefore, data management should not disrupt or threaten to divert resources devoted to these other purposes. As the system is implemented, before emergencies occur, discussions of the resources required should be held with - participants. Participants must devote some resources to data management, but these should be minimized. This is consistent with implementing a single system in the beginning of the outbreak investigation and continuing with it into the routine surveillance follow-up. Incorporating data into a second system for surveillance could waste resources. Although internal data management does not need to change to accommodate an outbreak, laboratories must implement systems that can directly feed data into the master reporting system database, either through an import function contained in the master system or by a direct interface between the internal laboratory system and the surveillance reporting system. Data management considerations during outbreak investigations and surveillance in the United States include the political concerns of the participants. Political and legal constraints of all participants must be addressed before the need to deal with them arises. On a global scale, this consideration is equally important, especially in countries whose economies may be adversely affected by news of a dangerous disease situation. Individual country sovereignty must not be violated by data reporting, and the cooperation of each participating country or political entity (e.g., World Health Organization [WHO], Pan American Health Organization [PAHO]) must be obtained in an atmosphere of confidentiality. All attempts to obtain, share, or combine data on a regional or global basis must include a well-defined set of rules agreed upon by all participants. For example, data for scientific purposes might be received at an office of WHO or PAHO but not sent beyond these organizations. Most often, for the sake of surveillance on a regional or global scale, data management considerations must focus first on establishing in-country data management infrastructures. This means that regional or global surveillance will first translate into establishing a master system, or at least compatible systems in individual participating countries. In most cases, data management systems available to developing countries do not provide the relational model needed by the laboratory. Therefore, efforts should be initiated to introduce and establish systems that can meet these needs in countries desiring to use them. A plan for regional or global surveillance must include tools to respond to outbreaks and provide for the computing equipment and modems or other means of transmitting the data electronically. Today's environment demands that most data management be done on personal computers located at critical sites where data can be input. However, data volume may ultimately require that the system provide for archiving data onto another medium. This does not preclude the use of personal computers for data management but simply recognizes that current technology limits the volume of data that can practically be managed and analyzed on personal computers. The initial data management plan for a country should include a section on reporting procedures and the appropriate medium for archiving data. To handle an immediate, urgent situation the system should contain, at a minimum, a personal computer with large hard-disk capacity (at least 1-2 gigabytes at the central level and possibly 300-500 megabytes at each reporting site), large memory (at least 4 megabytes of RAM at every reporting site), adequate speed (at least 33 megahertz at every reporting site), and fast modems if appropriate. For sites located in areas with inadequate telephone lines, other provisions for electronic transmissions should be planned (e.g., diskettes). Until security can be assured on the Internet, we do not recommend using this medium for electronic transmission of laboratory clinical data for outbreak investigations and surveillance. New Tools for the Management of Surveillance and Laboratory Data The Public Health Laboratory Information System (PHLIS) To address the need for a data management system for outbreak investigations and surveillance, the National Center for Infectious Diseases, CDC, in cooperation with the Association of State and Territorial Public Health Laboratory Directors in the United States, developed PHLIS. With this system, data entry screens (modules) are created and distributed to all reporting sites electronically, and data are input and reported within hours, without involving computer programmers. PHLIS provides the capacity for a hierarchical reporting scheme involving reports to multiple, successively higher reporting levels; a database is created at every reporting level so that all data reported to a site or input at the site are included in the database at that site. The most recent version of PHLIS (Version 3.0), is a menu-driven system based on a relational data model sufficient for the needs outlined in the first part of this report. The system allows for a patient record to be input only one time and links multiple specimens for that patient record. This is true even if specimens for the same patient are entered in different disease modules, or if the patient's name is to be added into a module that contains only epidemiologic data (no laboratory specimens). PHLIS provides a core set of data to be collected on every patient. In addition, each disease module can be customized by adding additional fields to the core data if needed. The system can accommodate data for epidemiologic, laboratory, survey, and case-control studies, and for other public health needs. Field staff can rapidly add their own data fields to existing disease modules to customize the data entry for special needs at each data reporting site. During an outbreak, a new module can be rapidly developed and electronically transmitted to all participating reporting sites. The system, which includes data communication software, is configured so that data flow in a pyramid reporting structure: that is, data are reported from lower level reporting sites through higher level reporting sites and ultimately to a single central site. As data are passed to each successively higher level, they are automatically assimilated into that site's database. Thus, databases are built and updated at successively higher reporting sites. Additional information about a case or specimen may be added at any reporting site; if desired, these additional data are also transmitted to the next higher reporting site. To meet the need for feedback, PHLIS has a menu-driven option to transmit files or messages up and down the reporting chain, with these files and messages being transmitted automatically when connections are established for each data transmission. This facility is flexible enough to allow any valid user in the reporting chain to transmit files or messages to any other user in the reporting chain. For example, in the United States, a county health official who is included in the reporting system in one state can send messages or files to a participating county official in another state. The feedback system does not mimic electronic mail because these files and messages are sent along the reporting chain in the same communications configuration as data reporting. Therefore, successful arrival of these messages at their destination(s) depends upon each member of the reporting chain between the sender and the receiver to establish a connection for reporting purposes. However, the system provides an alternative mechanism for sending files and messages directly to any other reporter having the capacity to receive them without going through the reporting chain. PHLIS is used in all 50 state public health laboratories, as well as the District of Columbia and Guam. Disease modules included are animal rabies, Campylobacter, Escherichia coli O157:H7, Lyme disease, mycobacteria, respiratory and enteric viruses, human Salmonella, nonhuman Salmonella, Shigella, and drug-resistant Streptococcus pneumoniae. PHLIS can be implemented independently: organizations can develop their own PHLIS pyramid reporting system. For example, PHLIS is currently being implemented at the Caribbean Epidemiology Center (CAREC) in Trinidad and in its member countries for the reporting of HIV/STD infections with the expectation that the reporting system will be expanded to accommodate other diseases. CAREC can receive reports from the member countries as each country is added to the reporting structure. Laboratory Information Tracking System (LITS) The second system, LITS, is a PC local area network-based system for tracking laboratory specimens. The system allows specimen information to be entered at a central specimen receiving site; additional information about the specimen can be entered into the system in any of the laboratories performing tests on that specimen. Although modules are customized for each laboratory's needs, laboratorians can add additional tests or delete obsolete ones. Furthermore, users can examine all the data about a specimen, including data from all laboratories that performed tests on the specimen. Other features in the system include cost billing, user defined reports, user defined query, and specimen or patient tracking and security. For emerging diseases, LITS provides a mechanism to standardize laboratory protocol across organizations and a mechanism to share data about specimens within an organization. Acknowledgments We thank Tim Kuhn for leading the programming team; Bruce Wilson, Dana Crenshaw, Joe Bates, and Neil Jones for programming support; Tim Day for user support; Kathleen Maloney, Joy Goulding, Lori Hutwagner, and Cecile Ivey for evaluating program integrity; Brian Plikaytis for his early involvement with LITS; and Cheryl Shapiro for financial management. References 1. Fraser DW, Tsai TR, Orenstein W, Parkin WE, Beecham HJ, Sharrar RG. Legionnaires' disease: description of an epidemic of pneumonia. N Engl J Med 1977;297:1189-97. 2. Shands KN, Schlech WF III, Hargrett NT, Dan BB, Schmid GP, Bennett JV. Toxic shock syndrome: case-control studies at the Centers for Disease Control. Ann Intern Med 1982;96:895-8. 3. CDC. Outbreak of acute illness southwestern United States, 1993. MMWR 1993;42:421-4. 4. Administrative Conference of the U.S. Privacy Act. In: Federal Administrative Procedure Sourcebook, 2nd ed. Office of the Chairman, 1992:863-979. 5. Administrative Conference of the U.S. Freedom of Information Act. In: Federal Administrative Procedure Sourcebook, 2nd ed. Office of the Chairman, 1992:633-61. DISPATCHES (Pages 115-23) Helicobacter hepaticus, a Recently Recognized Bacterial Pathogen, Associated with Chronic Hepatitis and Hepatocellular Neoplasia in Laboratory Mice Gastric carcinoma, one of the most prevalent human cancers worldwide, is among the neoplasms for which epidemiologic evidence of environmental causes is strongest. The exact nature of these environmental causes was obscure until mounting evidence recently linked chronic infection of the gastric antrum mucosa by Helicobacter pylori (a microaerobic, gram-negative, spiral bacterium) with elevated cancer risk (1). It is now recognized that gastric B-cell lymphoma of mucosa-associated lymphoid tissue is also closely linked to gastric H. pylori infection, and eradication of the infection with antibiotics can result in regression of the lymphoma (2,3). This startling finding has stimulated intense interest in the genus Helicobacter and related organisms; as a result, additional species of Helicobacter are now frequently isolated and characterized from many non-human hosts. Until 1994, however, only H. pylori was known to be associated with tumor development, in humans or in any other animal species. In 1992, at the National Cancer Institute's Frederick Cancer Research and Development Center (FCRDC) in Frederick, Maryland, a high prevalence of liver disease was observed among certain strains of mice; these mice were untreated controls in long-term chemical carcinogenesis experiments. Affected strains, notably A/JCr, had been bred at FCRDC under pathogen-free conditions and were free of known serologically detectable murine viruses and parasites; moreover, they had no histologically demonstrable hepatic abnormalities, except for a very low incidence (1% to 2%) of hepatocellular tumors in mice 15 months of age or older. Over a very short period, the prevalence of a histologically distinctive form of hepatitis increased to virtually 100% in male mice at 1 year of age (Table 1). The earliest demonstrable lesions were small, undistinctive foci of hepatic necrosis seen in young mice aged 2 to 6 months. In older mice, aged 6 to 10 months, there was a highly distinctive pericholangitis, consisting of abundant mononuclear cell infiltrates around bile ducts within portal triads. The biliary epithelium within affected ducts was focally swollen, and the luminal surfaces of damaged epithelial cells were poorly defined in hematoxylin and eosin-stained sections (4,5). In livers with extensive lesions, bile ductular (oval cell) hyperplasia was also prominent. Moreover, mice with hepatitis usually had hepatocellular tumors, often multiple, that included both adenomas and carcinomas (4). Hepatocellular tumors in mice are one of the most common endpoints in bioassays for chemical carcinogens. They were not, at that time, known to be associated with infectious agents. Accordingly, initial efforts to identify the cause of the hepatitis/hepatocellular tumor syndrome were directed toward possible sources of chemical exposure. The possibility of accidental exposure to experimental substances within the research animal facilities was ruled out when liver disease was identified in mice that had never left the breeding areas which are located in separate buildings. Extensive chemical analyses of food, bedding, water, and other possible sources of toxic substances had negative results. Detailed pathologic examination by light microscopy of tissue sections from diseased livers was continued, and many special stains were used. One such stain, Steiner's silver impregnation procedure for spirochetes (6), revealed in hepatic tissue uniform bodies that were consistent in size and shape with bacteria. Homogenates of fresh liver tissue from diseased mice proved effective in transmitting hepatitis to A/J mice purchased from commercial sources outside FCRDC, when given by intraperitoneal injection (5). In addition, from these homogenates, a motile, spiral bacterium could be cultivated on blood agar plates incubated at 37 degrees C under anaerobic or microaerobic conditions. This organism was subsequently characterized by ultrastructural morphologic examination, biochemical characteristics, and 16S rRNA gene sequence. Determined to be a new species related to H. pylori, it was given the name H. hepaticus (7). The bacterium is motile and gram negative, 0.2 to 0.3 æm in diameter, 1.5 to 5.0 æm long, and curved to spiral in shape, with one to several spirals; it has bipolar sheathed flagella (one at each end) but lacks the periplasmic fibers that envelope the bacterial cells in other mouse Helicobacter species. H. hepaticus has strong urease activity, is oxidase and catalase positive, produces H2S, reduces nitrate to nitrite, and grows microaerobically at 37 degrees C but not at 25 degrees C or 42 degrees C. It is resistant to cephalothin and nalidixic acid but sensitive to metronidazole. Photographs illustrating its morphologic structure by light (4,5) and electron (4,5,7) microscopy have been published. The species-defining characteristic of the organism, the nucleotide sequence of its 16S rRNA gene, has been used to develop a diagnostic assay based on polymerase chain reaction (8). Systematic examination of rodents of all species and strains produced at FCRDC, especially retired breeders, showed that the characteristic hepatitis and associated bacteria were present in mice of several strains (A/JCr, DBA/2NCr, C3H/HeNCr) and that within these strains, the male mice were more severely affected than the female. Mice with severe combined immunodeficiencies were especially vulnerable. The precise location of organisms demonstrable by Steiner stain within infected liver parenchyma was shown by transmission electron microscopy to be invariably extracellular and characteristically within bile canaliculi (4,5). No liver disease was seen in some strains (e.g., C57BL/6NCr) or in F1 hybrids between sensitive and resistant strains (e.g., B6C3F1). Rodent species other than mice (e.g., rats, Syrian hamsters, and guinea pigs) were not affected. In infected mice with severe combined immunodeficiency, cecal inflammation was histologically demonstrable (5), and organisms were isolated from the mucosa of the large intestine (7), which may mean that the usual ecologic niche occupied by H. hepaticus is that of a commensal colonizer of the intestinal tract (8). Since mice are coprophagic, it appears highly likely that natural transmission of the organisms is the oral-fecal route. Why and how H. hepaticus invades the liver in mice of certain strains remain to be determined. Hepatitis is also characteristic of certain other enteric pathogenic bacteria, such as Campylobacter jejuni (9) that, unlike H. hepaticus, have not been associated with liver tumor development. The tissue damage that accompanies persistent infection by H. hepaticus, H. pylori, and certain other Helicobacter species may be due, at least in part, to a soluble, trypsin-sensitive cytotoxin of high molecular weight produced by these organisms (10). There is no precedent for any direct role of such a toxin in carcinogenesis. On the other hand, chronic infections by viruses, bacteria, or certain parasites are recognized risk factors for human cancers at various sites. The hypothesis that chemically reactive, potentially genotoxic, substances of low molecular weight (including nitric oxide and active oxygen species) generated by inflammatory cells at the site of chronic infection may initiate or enhance carcinogenesis has been examined (11). The hypothesis is under active investigation in the context of H. hepaticus-associated liver disease. H. hepaticus is susceptible to a number of antibiotics; treatment of susceptible, naturally infected 8- to 10-week-old strain A/JCr mice with single or combined antimicrobial agents has been evaluated for efficacy in eradicating established infections (12). Amoxicillin, metronidazole, and tetracycline administered singly failed to eradicate bacteria from the gastrointestinal tract, but either amoxicillin or tetracycline, in combination with metronidazole and bismuth, was effective in eradicating H. hepaticus from the liver, cecum, and colon when given by oral gavage for a period of 2 weeks (12). The effect of antibiotic therapy on the carcinogenic process, or in older animals, remains to be established. The importance of H. hepaticus to humans is not yet completely known. The organism clearly has the potential to confound bioassays for chemical carcinogens, but this potential has no direct effect on humans. Even though most Helicobacter species identified to date are characteristically associated with (and named after) specific mammalian host species in which they generally inhabit the gastrointestinal tract (with or without causing gastritis or other chronic inflammatory disease), the potential host range for some species is quite broad. H. pylori, originally isolated from humans, has recently been isolated also from the domestic cat; this raises the possibility that Helicobacter pylori may be a zoonotic pathogen that can be transmitted from companion animals to humans (13). Exploring the possibility of zoonotic transmission of H. pylori, H. hepaticus, or any other Helicobacter species would require isolation of the organism in question by culture methods. Serologic methods have not yet been refined to the level of species specificity. Humans infected with H. pylori mount a serum antibody response to the bacteria that is readily detected by enzyme-linked immunosorbent assays and is considered evidence of ongoing disease (1); mice infected with H. hepaticus similarly produce serum antibodies to that species that have been demonstrated by Western blotting (5). Antisera to H. pylori can be used to visualize H. hepaticus in mouse liver tissue sections stained by the avidin-biotin complex immunohistochemical procedure (5). The cross- reactivity between these two species precludes use of available serologic methods to establish whether H. hepaticus has infected humans. Regardless of whether H. hepaticus is itself capable of infecting humans, it serves to demonstrate that liver tissue can be persistently infected by at least one member of the genus Helicobacter, and that liver cancer can be a long-term consequence of such infection. This discovery raises questions about the existence of a comparable relationship between liver cancer in humans and unrecognized bacterial infections. Reviews are under way of tissue blocks from pathology archives in search of organisms demonstrable by the Steiner stain in liver sections from human populations at high risk for liver cancer. Jerry M. Rice Laboratory of Comparative Carcinogenesis, National Cancer Institute, Frederick, Maryland, USA Table 1. Increasing prevalence of hepatitis and hepatocellular neoplasia in control male A/JCr mice at the National Cancer Institute's Frederick Cancer Research and Development Center, 1989-1992(a) Number Age, Mice with Mice with Date killed of mice in weeks hepatitis (%) liver tumors (%) ___________________________________________________________________ Jan-Mar 1989 48 47+orñ3 0 0 May-Jul 1989 47 70+orñ60 0 1 (2) Jan-Apr 1992 6 36+orñ4 2 (33) 0 Jul 1992 16 54 16 (100) 1 (6) Aug-Oct 1992 6 64+orñ3 5 (83) 3 (50) Dec 1992 12 77 12 (100) 11 (92) _______________________________________________________________ (a)Adapted from ref. 4. References 1. Parsonnet J, Friedman GD, Vandersteen DP, Chang Y, Vogelman JH, Orentreich N, et al. Helicobacter pylori infection and the risk of gastric carcinoma. N Engl J Med 1991;325:1127-31. 2. Wotherspoon AC, Doglioni C, Diss TC, Pan L, Moschini A, de Boni M, et al. Regression of primary low-grade B-cell gastric lymphoma of mucosa-associated lymphoid tissue type after eradication of Helicobacter pylori. Lancet 1993;342:575-7. 3. Bayerd”rffer E, Neubauer A, Rudolph B, Thiede C, Lehn N, Eidt S, et al. Regression of primary gastric lymphoma of mucosa-associated lymphoid tissue type after cure of Helicobacter pylori infection. Lancet 1995;345:1591-4. 4. Ward JM, Fox JG, Anver MR, Haines DC, George CV, Collins MJ Jr, et al. Chronic active hepatitis and associated liver tumors in mice caused by a persistent bacterial infection with a novel Helicobacter species. J Natl Cancer Inst 1994;86:1222-7. 5. Ward JM, Anver MR, Haines DC, Benveniste RE. Chronic active hepatitis in mice caused by Helicobacter hepaticus. Am J Pathol 1994;145:959-68. 6. Garvey W, Fathi A, Bigelow F. Modified Steiner for the demonstration of spirochetes. J Histotechnology 1985;8:15-7. 7. Fox JG, Dewhirst FE, Tully JG, Paster BJ, Yan L, Taylor NS, et al. Helicobacter hepaticus sp. nov., a microaerophilic bacterium isolated from livers and intestinal mucosal scrapings from mice. J Clin Microbiol 1994;32:1238-45. 8. Battles JK, Williamson JC, Pike KM, Gorelick PL, Ward JM, Gonda MA. Diagnostic assay for Helicobacter hepaticus based on nucleotide sequence of its 16S rRNA gene. 1995;33:1344-7. 9. Kita E, Oku D, Hamuro A, Nishikawa F, Emoto M, Yagyu Y, et al. Hepatotoxic activity of Campylobacter jejuni. J Med Microbiol 1990;33:171-82. 10. Taylor NS, Fox JG, Yan L. In-vitro hepatotoxic factor in Helicobacter hepaticus, H. pylori and other Helicobacter species. J Med Microbiol 1995;42:48-52. 11. Ohshima H, Bartsch H. Chronic infections and inflammatory processes as cancer risk factors: possible role of nitric oxide in carcinogenesis. Mutat Res 1994;305:253-64. 12. Foltz CJ, Fox JG, Yan L, Shames B. Evaluation of antibiotic therapies for eradication of Helicobacter hepaticus. Antimicrob Agents Chemother 1995;39:1292-4. 13. Handt LK, Fox JG, Dewhirst FE, Fraser GJ, Paster BJ, Yan LL, et al. Helicobacter pylori isolated from the domestic cat: public health implications. Infect Immun 1994;62:2367-74. (Pages 132-3) Hemolytic Uremic Syndrome Due to Shiga-like Producing Escherichia coli O48:H21 in South Australia Enterohemorrhagic Escherichia coli (EHEC) other than serotypes O157:H7 are increasingly recognized in association with hemolytic uremic syndrome (HUS) (1) and have been reported in Australia (2). While detecting strains of O157:H7 has become easier over the years, identifying the expanding number of other serotypes of EHEC also associated with HUS, with other conditions, and with healthy domestic animals is still very difficult. Cases of HUS have been reported in Australia over a number of years. The most common serotype found was O111:H-, and Australia's recently reported first HUS outbreak (3) was caused by EHEC O111:H-. We wish to report a case of severe HUS due to serotype O48:H21, which, as far as we know, has not been previously reported as a cause of HUS. This case occurred in 1993, before surveillance of HUS had been initiated; after this case, between July and December 1994, 10 cases of HUS (from which four isolates were obtained; two were EHEC O111) were reported to the Australian Paediatric Surveillance Unit (E. Elliott, pers. comm.). The patient in the 1993 case was an 8-year-old girl, living in a rural setting in the outskirts of Adelaide, South Australia. Her home was adjacent to a farm on which cows, sheep, and ducks were kept. A kelpie/healer cross puppy was in the house in November 1993. Also kept were a pet galah (Australian cockatoo) and pet fish. She was well until 23 December 1993, when she had diarrhea described as very smelly and watery "like the juice of tinned crab." The diarrhea became bloody on 2 January 1994 and was associated with severe abdominal pains which made the patient draw up her legs. She was having bowel movements six times a day, had become very weak, and was unable to stand. She was admitted to Adelaide Children's Hospital on 3 January 1994, and her condition progressed to anuric renal failure over the next few days. Serum biochemistry on 7 January showed a urea level of 23.3 mmol/L and creatinine level of 539 æmol/L. Her hemoglobin level fell from 157 g/L on 3 January to 86 g/L on 10 January. Her hematocrit fell from 48% to 24%, and her platelet count fell from 463 x 109/L to 47 x 109/L on these dates, respectively. The blood film showed microangiopathic hemolytic anemia with fragmented red cells. She required hemodialysis for 3 weeks and was discharged from the hospital on 31 January 1994. Apart from the patient's 5-year-old brother, who had loose bowel movements for 1 day on 28 December 1993, no other family members were affected. An adequate dietary history was not obtained; however, no food had been eaten from commercial food outlets. Stool samples were collected on 4 and 5 January 1994. The samples were probed for Shiga-like toxin (SLT)-I and SLT-II genes by polymerase chain reaction (PCR), and the results were positive. Approximately 80% of lactose-fermenting colonies on MacConkey agar were also SLT positive. No sorbitol-negative colonies were observed on sorbitol-MacConkey agar. In addition to being cultured for E. coli, the stools were also routinely cultured for Shigella, Salmonella, Yersinia, Vibrio, and Clostridium. In addition, stained concentrates were examined for Giardia lamblia and Entamoeba histolytica with negative results. Four typical E. coli strains were subjected to further tests. They were typical E. coli, positive in the indole and ONPG tests, negative in the Voges-Proskauer, citrate, TDA, malonate, urease, gelatine, and H2S tests. The strains fermented glucose, lactose, mannitol, xylose, rhamnose, arabinose, sorbitol, sucrose, and melibiose. They did not ferment inositol, adonitol, salicin, raffinose, or amylose. They decarboxylated arginine, lysine, and ornithine. All the strains produced enterohaemolysin (4). The strains were O and H serotyped (5, 6) and found to be serotype O48:H21. Supernatant preparations were tested on Vero cells (7) and found to give typical verocytotoxic reactions in titers of 103 to 104. The supernatants were also tested by enzyme-linked immunosorbent assay (ELISA) by using monoclonal antibodies 13C4 and 11E10 directed against SLT-I and SLT-II, respectively, and strong reactions with both antibodies were noted, confirming the presence of both SLTs. Stool samples taken from the patient on 8 February 1994 were negative for SLT-I and SLT-II genes by PCR and were not cultured further. Stool samples from the patient's brother and local animals were not forthcoming. That all four E. coli isolates tested were of serotype O48:H21 and demonstrated identical toxigenicity by both PCR and ELISA and the fact that SLT-positive organisms were not found in the stools collected during the patient's convalescence strongly suggest tha