The epidemiology of human rabies is ultimately linked to cycles of rabies virus transmission in animals. With the interruption of dog-to-dog transmission in most regions, the incidence of human rabies in the United States has reached a level that cannot be further reduced without targeting wildlife. An understanding of epidemiologic patterns of rabies virus maintenance in natural populations has emerged in the past 20 years, largely because of advances in immunology and molecular biology. Monoclonal antibody and genetic sequence analyses of rabies virus variants permit detailed descriptions of enzootic maintenance cycles of specific virus variants in the United States [4, 5]. These analyses have led to an understanding of how variants of rabies virus are maintained in natural reservoirs within geographic regions and have provided information on variability of the virus itself.

Current epidemiologic patterns of rabies in the United States can be summarized as follows: The annual reports of rabies in wildlife exceed those of rabies in domestic animals [6]; rabies variants in bats are associated with a disproportionate number of infections in humans, although bats constitute only about 10% of all reported rabies cases in animals annually; most other cases of human rabies diagnosed in the United States can be attributed to infections acquired in areas of enzootic canine rabies outside of the United States; most persons with a case of rabies that originated in the United States have no history of an animal bite; and rabies is diagnosed after death in more than one third of the latter group.

The last published summary of cases of human rabies in the United States covered the period from 1960 to 1979 [3]. This review discusses the clinical and epidemiologic features of cases of human rabies in the United States from 1980 to 1996.

Methods

Top Methods Results Discussion Author & Article Info References

Case Definition

This report includes all laboratory-confirmed cases of human rabies in the United States or its territories from 1980 to 1996 [7-31]. All of the cases were reported to the Centers for Disease Control and Prevention (CDC) by health authorities as part of ongoing national surveillance.

Variable Definitions

Onset of illness was defined as either the first day of reported symptoms attributable to rabies or the date of initial presentation for medical care before confirmation of rabies. Clinical signs attributable to rabies included paresthesia, anxiety, agitation, confusion, disorientation, hydrophobia, aerophobia, hypersalivation, dysphagia, paresis, paralysis, and fluctuating levels of consciousness [32, 33]. The type of transmitting animal and the geographic location of exposure were listed if the case history included a definite animal bite. The reliability of information that linked rabies exposure to a human was assessed by subsequent laboratory typing of the rabies virus variant. All other exposures were defined as "unknown." The diagnosis of rabies was considered antemortem if it was tentatively made and samples were obtained specifically for rabies testing before the patient's death.

Laboratory Tests

The diagnosis of rabies was confirmed by using standard tests [34] conducted at the CDC or at a state laboratory.

Serology

Two tests were used to detect rabies antibody: the rapid fluorescent focus inhibition test and the indirect immunofluorescence assay. The rapid fluorescent focus inhibition test measures neutralizing antibody. An antibody titer of 1:5 or more, as defined by the reciprocal of the serum or cerebrospinal fluid dilution that reduces the challenge virus by 50%, was considered positive. An indirect immunofluorescence assay, using patient serum or cerebrospinal fluid diluted 1:4 or more, detects serum reactive with rabies antigen in infected cell cultures. The presence of antibody in serum was considered diagnostic if no vaccine or antirabies serum was given to the patient. Antibody in the cerebrospinal fluid, regardless of the rabies immunization history, was considered indicative of rabies virus infection.

Virus Isolation

Suspensions of brain or saliva specimens were added to mouse neuroblastoma cells and cultured for 24 and 48 hours. Culture slides were fixed and examined by direct immunofluorescence assay for antigen. Samples that were initially negative were maintained for an additional 3 to 4 days and retested. The negative result was considered definitive if it occurred both times.

Antigen Detection

Antigen detection was performed by direct immunofluorescence of assay serial frozen sections of nuchal skin biopsy specimens, touch impressions of corneal epithelial cells, or fresh brain matter. Paraffin-embedded fixed brain matter was sectioned and enzyme-digested before direct immunofluorescence.

RNA Detection

Standard extraction procedures and reagents were used to obtain nucleic acids from samples of undiluted saliva; from fresh or paraffin-embedded fixed samples of the brain; or, occasionally, from other tissues. Reverse transcription of RNA and complementary DNA amplification were performed by polymerase chain reaction (PCR) with primers derived from the sequence of the N protein gene. The nucleotide sequence of all PCR products was obtained by standard dideoxynucleotide sequencing methods. Rabies virus variants were identified by comparing samples of rabies virus obtained from all known reservoirs for rabies in the United States [5] with samples of rabies virus obtained from dogs in Asia, Africa, and Latin America [35].

Statistical Analysis

Data analyses were performed by using EPI INFO 6 (Centers for Disease Control and Prevention, Atlanta, Georgia) or SPSS 6.0 for Windows (SPSS Inc., Chicago, Illinois) [36, 37]. Specific tests are identified in the text. Some variables were dichotomized before statistical comparisons for determination of odds ratios and 95% CIs. All reported P values are for two-tailed tests of significance.

Results

Top Methods Results Discussion Author & Article Info References

Demographic Information

Thirty-two persons died of rabies in the United States from 1980 through 1996. Patients ranged in age from 4 to 82 years (median, 27 years) and 20 (63%) were male (Table 1). Cases were reported from 20 states; 7 cases (22%) were reported in California and 6 in Texas. Eleven patients were exposed to rabies in eight foreign countries on the basis of variant typing. The onset of illness occurred in all months and had no apparent seasonal pattern. Dates of exposure, based on the history of an animal bite, were obtained for 7 patients (22%) (Table 1).

Exposure History

A definite history of animal exposure was identified in 7 of the 32 patients (22%), and 25 remained unknown or indefinite (Table 1). Of the 7 cases of definite exposure, 6 resulted from a dog bite received in a foreign country and 1 was from a bat bite received in the United States. Although rabies was not diagnosed in any of the animals that inflicted bites, in each case the rabies virus variant identified in the human sample was consistent with that in the animal species implicated as the source of infection (Table 1).

Contact with an animal, thereby suggesting the source for infection, was identified in 12 persons (8 with a bat, 2 with a dog, 1 with a cow, and 1 with a cat). This human-animal contact, however, could not be linked to a bite or mucous membrane contact with the saliva of an animal potentially infected with rabies virus. The remaining 13 patients did not report animal contact; thus, a potential source of exposure was not identified. Histories were obtained before death from friends or relatives in 9 cases and from 4 children aged 11 to 13 years.

Prophylaxis

None of the 32 patients received a complete series of rabies prophylaxis after exposure; patient 7 reported receiving a single injection of an unknown type after a dog bite in Guatemala, and patients 15, 29, and 30 received human rabies immune globulin during the course of their clinical illness.

Clinical Presentation

For the 7 patients in which a definite animal bite occurred, the median incubation period was 85 days (range, 53 to 150 days). The first signs and symptoms of rabies were often nonspecific, including fever, sore throat, chills, malaise, anorexia, headache, nausea, vomiting, dyspnea, cough, and weakness. Specific symptoms, such as paresthesias at or near the presumed exposure site, were also reported early in the clinical course, and 19 of the 32 patients (59%) had three or more clinical findings suggestive of rabies during the course of their illness (Table 2). The 32 patients were seen by physicians on an outpatient basis a median of one time (range, 0 to 5 times) before hospitalization, and the median length of time from the onset of illness attributable to rabies to hospitalization was 4 days (range, 1 to 10 days).

On admission, 21 of the 32 patients (66%) were febrile (oral temperature > 37.8°C), including 12 patients with temperatures greater than 39.5°C. Of the 11 patients who were afebrile on admission, 5 reported being febrile before admission, 2 became febrile within 2 days of admission, and 4 had no additional temperatures recorded.

The antemortem diagnosis of rabies was first considered at the time of hospitalization in 5 patients, within 1 day of hospitalization in 5 patients, and after a median of 6 days of hospitalization (range, 2 to 12 days) in 10 patients. In 12 patients, rabies was diagnosed after death. The clinical findings significantly associated with antemortem diagnosis were the presence of hydrophobia or aerophobia or the presence of three or more signs typical of rabies (Table 2).

Treatment and Clinical Course

All 32 patients were hospitalized in the United States and received intensive supportive therapy. Because herpesvirus infection was frequently considered in the differential diagnosis, 10 of the 32 patients (31%) received antiviral therapy that included interferon, ribavirin, adenine arabinoside, or acyclovir. The median duration of illness was 19 days (range, 7 to 28 days) in patients receiving antiviral therapy and was similar for patients who did not receive antiviral therapy (median duration, 15 days [range, 7 to 32 days]). Patients 15, 29, and 30 received human rabies immune globulin during the course of their illness; the median duration of illness in these patients was also 19 days (range, 11 to 42 days).

Rabies Diagnostic Testing

Antemortem Test Results

Laboratory confirmation of rabies infection was done before death in 18 of the 20 cases for which a diagnosis of rabies was considered before death (Table 3 and Table 4). In one case, all antemortem samples were negative and in another, samples were not tested until after the patient's death. In cases of rabies diagnosed before death, the first sample that was analyzed for rabies was obtained a median of 7 days (range, 3 to 17 days) after the onset of clinical symptoms. Serum was the most common type of sample submitted (18 of 20 cases). Antibody to rabies virus was detected in serum in 10 of the patients and was the earliest positive diagnostic test result in 7 patients. In 3 cases, serum antibody to rabies was the only antemortem test that had positive results for rabies (Table 3 and Table 4). Serum antibody was present as early as day 5 of the clinical course but was absent as late as day 10 in 5 patients and as late as days 14 and 24 in 2 patients. Antibody in cerebrospinal fluid was detected in 2 of 13 patients for whom cerebrospinal fluid was submitted for antemortem diagnosis of rabies.

Isolation of rabies virus from saliva was attempted in 15 of the 20 cases of rabies diagnosed before death, and in 9 cases virus was found in 1 or more samples. In 7 cases, isolation of virus from saliva was the earliest diagnostic sign of rabies. Isolation of virus from saliva was more successful in antibody-negative patients (13 of 15 samples [87%]) than in antibody-positive patients (0 of 17 samples). Reverse transcription of RNA by PCR was positive for viral RNA in saliva in all 10 patients in whom this test was performed, including 3 patients in whom saliva was negative for virus isolation.

A nuchal skin biopsy specimen was submitted for testing in 15 of the 20 cases diagnosed before death. Rabies viral antigen was detected in 10 patients and was the first positive diagnostic test result in 9 patients. Antigen was detected in samples collected as early as day 5 and was unrelated to antibody status; 6 of 10 samples (60%) were positive when antibody was not present and 3 of 7 samples (43%) were positive when antibody was present (Table 3 and Table 4).

Corneal impression slides were tested in 7 of the 20 patients but were positive in only 2 patients. Samples taken as early as day 3 and as late as day 18 after the onset of clinical signs were negative for antigen. Too few positive samples were available to determine whether this test was successful in demonstrating rabies antigen in corneal epithelium on the basis of the presence of serum antibodies. In patient 23, rabies antigen in corneal epithelium was the only laboratory finding, but no other samples from this patient were submitted for testing.

Rabies antigen was detected in all three brain biopsy samples submitted for testing. In patients 5 and 22, the biopsy results were the first diagnostic findings that were positive for rabies, although retrospective tests of tissue samples obtained before the biopsy also yielded positive results.

Postmortem Test Results

The clinical history of 6 of the 12 patients in whom samples were not obtained specifically for rabies testing before the patient's death was sufficiently suggestive of rabies that fresh brain matter obtained during autopsy was submitted immediately for direct immunofluorescence testing. The results were positive for rabies in these 6 cases. Rabies was diagnosed in the remaining 6 cases from 3 weeks to 6 months after autopsy.

Identification of Rabies Virus Variants

For all 32 rabies cases, nucleotide sequences were obtained from rabies virus RNA extracted from samples and amplified by reverse transcription of RNA by PCR. Rabies variants associated with domestic dogs in areas outside the United States in which rabies is endemic were identified in 12 patients (Table 1). A history of travel and residence was obtained for all 12 patients; 6 had received an animal bite outside the United States from a dog or cat suspected of being rabid. Rabies virus variants indigenous to animal reservoirs in the United States were identified in the 20 remaining cases. Three cases were associated with virus variants found in terrestrial animals and 17 cases were associated with variants found in insectivorous bats. Twelve of these 17 cases had evidence of infection with a variant found primarily in the silver-haired bat (Lasionycteris noctivagans) or eastern pipistrelle (Pipistrellus subflavus).

Postexposure Prophylaxis in Contacts of Cases

In 30 patients, data on administration of postexposure prophylaxis to contacts were available; at least 1763 human contacts received postexposure prophylaxis (median number of persons per case, 54 [range, 4 to 179]) consisting of five doses of human diploid cell vaccine or rabies vaccine, adsorbed, and one does of human rabies immune globulin. No differences in the number of contacts receiving postexposure prophylaxis per patient were found between groups dichotomized by time of diagnosis (median interval between clinical onset and diagnosis, 7 days [range, 3 to 17 days]; Wilcoxon rank-sum test, P > 0.02), sex (Wilcoxon rank-sum test, P = 0.08), duration of illness (median, 15 days [range, 7 to 42 days]; Wilcoxon rank-sum test, P > 0.02), or chronologic date of death (dichotomized at the median, 11 July 1993; Wilcoxon rank-sum test, P = 0.11).

Additional Cases

In addition to the 32 patients discussed here, 3 other U.S. citizens died of rabies while living abroad. The first case occurred on 7 May 1981, when a 29-year-old woman died in Rwanda after being bitten by a stray dog [38]. The second case occurred on 27 August 1983, when a 23-year-old woman died in Kenya after being bitten by her puppy, which was suspected of being rabid [39]. The last case occurred on 21 April 1992, when a 44-year-old man died in Bangkok, Thailand, after having been bitten by a puppy in Bangladesh [40]. The patient who died in Rwanda had received five doses of human diploid cell vaccine but had not received the recommended immune globulin.

The patient who died in Kenya had received preexposure immunoprophylaxis against rabies that consisted of three 0.1-mL intradermal inoculations with human diploid cell vaccine but had not received postexposure immunoprophylaxis [41]. The patient who died in Thailand did not receive any form of immunoprophylaxis.

Discussion

Top Methods Results Discussion Author & Article Info References

The clinical presentation, duration of illness, and course of illness in the 32 cases presented here are similar to those of previous reports of human rabies [2, 3, 32, 33]. However, several changes in the epidemiologic patterns of human rabies in the United States are apparent. From 1960 to 1979, 16 of 38 patients (42%) receiving a diagnosis of rabies in the United States had received some form of rabies prophylaxis after exposure [3], whereas none of the 32 patients receiving a diagnosis of rabies in the United States between 1980 and 1996 had received postexposure prophylaxis. The circumstances surrounding 12 of the recent cases could have suggested the need for postexposure prophylaxis [41], including dog bites and contamination of open wounds or mucous membranes by saliva from dogs that died abruptly in rabies-endemic areas. Three cases (patients 4, 11, and 13) were associated with exposure to bats in situations that could have involved a bite. A factor that contributed to the lack of rabies postexposure prophylaxis in patients in the United States was the inability to recognize a potential rabies exposure by both the patients and their clinicians.

From 1980 to 1996, 85% (17 of 20) of the rabies virus variants obtained from cases of human rabies acquired in the United States were associated with insectivorous bats. Clear evidence of a bite was found in only one of the 17 bat-associated cases reported from 1980 to 1996. In 8 cases, contact with a bat was reported by the patient, a family member, or an acquaintance, but no bite or wound was recognized. Although the exact nature of human-bat contact resulting in rabies virus transmission remains uncertain in many of the recent human cases, an undetected or an unreported bat bite remains the most plausible hypothesis. Rabies virus transmission other than by bites is rare and occurs under exceptional circumstances, such as tissue (corneal) transplantation [42], accidental laboratory aerosolization of concentrated virus, or exposure to aerosolized virus in caves inhabited by millions of bats [43]. Identification in 12 of the 17 bat-associated cases (71%) of a rabies virus variant that is almost exclusively found in the silver-haired bat or eastern pipistrelle is a remarkable feature of this series. These species of bats are rarely submitted for rabies testing [44], although their geographic ranges are extensive [45].

Bites by terrestrial carnivores are obvious and usually lead to recognition of the potential rabies exposure and administration of postexposure prophylaxis. During 1960 to 1979, 13 of 28 indigenously acquired human rabies cases (46%) were caused by terrestrial mammal bites, most commonly from dogs [3]. These patients recognized the risk of exposure to rabies, sought medical care, and received post-exposure prophylaxis; unfortunately, treatment was not always successful. Since the late 1970s, human rabies immune globulin and potent cell-culture-derived vaccines have almost assured successful postexposure prophylaxis. Chloroquine chemoprophylaxis for malaria can interfere with the antibody response to preexposure immunization against rabies by the intradermal route [41]. This interference may have contributed to the death of the U.S. citizen in Kenya who was bitten by a rabid dog but did not receive postexposure prophylaxis [39].

Another emerging pattern in the epidemiology of human rabies is lack of clinical suspicion of rabies and a delay in diagnosis. Twelve of the 32 cases of rabies in this series were diagnosed after death of the patient; 6 cases were not diagnosed for several weeks after the death of the patient. This trend is strongly influenced by at least two factors. First, a history of animal contact relevant to a rabies exposure was either not recognized or not remembered by the patient or attending friends and family members or was not ascertained by medical personnel. Careful attention to this detail early in the clinical course may result in earlier diagnosis of rabies and clarification of the mystery surrounding the cause of many recent cases. A history of foreign travel to or immigration from areas in which canine rabies is endemic may facilitate a more timely diagnosis of rabies in persons suspected of having rabies, even when travel occurred more than a year previously. Incubation periods of 5 years or longer occurred in patients 5, 9, and 10 [46]. A second factor is a low level of suspicion of rabies because of the rarity of this disease in the United States. Any patient who presents with encephalopathy of unknown cause should be considered to potentially have rabies, even in the absence of known exposure to the virus through an animal bite.

Aspects of rabies pathogenesis limit the design and success of antemortem diagnostic tests. By the time clinical signs became apparent, virus was present in all of the brain biopsy specimens tested. However, the risks of performing a brain biopsy prohibit its use as a method for diagnosing rabies. Of the antigen-detection methods, testing of nuchal skin biopsy specimens was more sensitive than tests of corneal epithelial cells. In three of the four patients in whom both types of tissue were sampled on the same day, antigen was detected in cutaneous nerves but was not found in corneal epithelium. Both tests had negative results in the fourth patient. Because corneal abrasions or more permanent damage may occur when touch impressions are made incorrectly, corneal epithelium should not be considered a sample of choice. Nested PCR of RNA extracted from saliva yielded positive results in all cases tested, suggesting that this test may greatly enhance rapid diagnosis of human rabies. However, many of these tests were done after the virus variants responsible for the infections were known and appropriate primers could be identified. Further study on sensitivity and specificity is required before reverse transcription of RNA by PCR can replace virus isolation as a standard antemortem diagnostic test. Serum antibody to rabies virus was not present until several days after the onset of clinical signs and appeared even later in cerebrospinal fluid, as previously noted [34, 47]. Serum obtained early in the clinical course is unlikely to be useful for diagnosis except to provide a baseline value for subsequent samples. In this case series, rabies antibody was detected in 2 of 9 serum samples collected in the first week of clinical illness, in 6 of 11 samples collected in the second week, and in 5 of 6 samples collected in the third week. When paired samples were available, cerebrospinal fluid remained negative for 3 to 6 days after antibody appeared in the serum.

All rabies virus variants matched the species suspected as the source of rabies identified by the medical history, and these molecular tools provided inferences about otherwise-unknown exposure events. Because rabies virus variants associated with a reservoir can also be found in other species during spillover events [35], the identification of a variant indicates the ultimate reservoir, if not the proximate animal source of infection. Virus variants identified in 19 of the 25 patients who did not report an animal bite were associated with well-known reservoirs for rabies in the United States. In three of these cases, terrestrial animal reservoirs were implicated (striped skunks in the southwestern United States and coyotes or domestic dogs in southern Texas and Mexico), whereas in the remaining 16 cases, genetic typing implicated virus variants associated with different bat species (Tadarida brasiliensis, L. noctivagans, P. subflavus, and Myotis species). In one case, the bat that was implicated in the exposure was available for species identification and rabies testing. Analyses indicated that the bat was rabid, and the virus variant was identical to that obtained from the patient [26]. Although no potential animal exposure was identified in patients 6, 17, and 19, patients 17 and 19 had tended to a sick animal that disappeared or died of an unexplained illness 3 to 9 months earlier. Although transmission of rabies by a bite from an infected bat cannot be excluded in these patients, it is possible that these animals were infected with rabies virus after contact with a bat and subsequently transmitted the bat virus variant to the patients. Rabies virus variants can infect and kill members of other species [35], but typically the rabies virus variant is unlikely to propagate among members of the new species.

The lack of significant differences in the number of contacts who received rabies postexposure prophylaxis was an unexpected finding among the various groups examined. Fewer treatments would be expected in cases diagnosed earlier rather than later in the clinical course. One explanation might be the extreme concern that accompanies a suspected case of human rabies. In addition, different algorithms for prescribing postexposure prophylaxis were maintained in different hospitals. Postexposure prophylaxis is occasionally offered as reassurance to a concerned, but unexposed, person rather than as a preventive measure in a suspected exposure situation [48]. More than 20 000 postexposure prophylaxis regimens are given each year in the United States [49]. The number of treatments given to medical personnel who attend patients suspected of having rabies can be minimized by instituting isolation procedures that prevent contact with the patient's body fluids [42].

As suggested by this report, contact with bats may occur under unique circumstances. Patients 21, 22, 26, and 28 had handled dead or dying bats with their bare hands. Patients 25 and 27 may have slept in a room where a bat was present. Unlike bites from dogs, raccoons, or other terrestrial carnivores, bites inflicted by bats may not be noticed or reported. For example, some solitary bat species, although normally reclusive, may be accessible to humans when they use woodpiles and shrubs as day roosts or use human dwellings for temporary night roosts. Bats infected with the rabies virus may be unable to fly or find appropriate shelter, thus increasing the opportunity for human contact [44]. Moreover, the small, sharp teeth of some insectivorous bats may produce such an insignificant wound that the bite is not apparent if it is inflicted during some other activity, such as carrying wood, thus hindering recognition of the bite. The limited injury inflicted by a bat bite and an inaccurate recall of the exact exposure history may hinder the ability of health care providers to determine the risk for rabies from an encounter with a bat.

Rabies postexposure prophylaxis is recommended for all persons with a history of a bat bite or scratch or mucous membrane exposure to a bat, unless the bat is available for testing and is proven not to be rabid. For all potential human exposures involving bats, the bat in question should be safely collected, if possible, and submitted for rabies diagnosis. If the bat cannot be collected, postexposure prophylaxis may be appropriate (even in the absence of a demonstrable bite, scratch, or mucous membrane exposure) if exposure is likely to have occurred, such as when a person awakens to find a bat in the room or if a bat is in proximity to a previously unattended child. This recommendation, used in conjunction with current guidelines [41], should allow optimal intervention with rabies postexposure prophylaxis when an accurate exposure history is not obtainable while minimizing inappropriate postexposure prophylaxis.

Dr. Drenzek: Department of Human Resources, Suite 6-110, 2 Peach Tree Street, Atlanta, GA.

Drs. Fekadu, Olson, Rupprecht, and Childs, Ms. Smith, Mr. Krebs, Ms. Orciari, Mr. Shaddock, Mr. Sanderlin, and Ms. Whitfield: Viral and Rickettsial Zoonoses Branch, Centers for Disease Control and Prevention, 1600 Clifton Road, Mailstop G-13, Atlanta, GA 30333.