Typically, rabies is fatal once clinical signs develop. Although persistent infections occur regularly for other virus infections, they have not been documented unquestionably in rabies, mainly because of the added complexity of the disease’s relatively long incubation period. The carrier or asymptomatic rabies state was once considered to be important for public health, despite lack of adequate evidence that the phenomenon actually exists. This concern has been raised repeatedly from the early 1930s until recently (7). Reported carrier hosts have included vampire bats (8,9), cats (10), dogs (11–15), and hyenas (7). Because rabid dog bites are responsible for ≈99% of all human rabies cases in the world (16), the possibility of a carrier state or asymptomatic form of canine rabies deserves serious evaluation. Unfortunately, this possibility remains highly speculative. Although some investigators have questioned reports of a carrier state in dogs (17), an author reported RABV isolation from brains of healthy dogs (6). Carrier dog RABV isolates were even characterized at the molecular level in 1996 (15).

Similarly, reports of healthy-dog carriers were consecutively published in China in 1982 (18), 1999 (19), 2006 (20), and 2007 (21). Tang found that 5 (1.76%) of 283 healthy-looking dogs in Guangxi province were positive for RABV by reverse transcription–PCR (RT-PCR) and virus isolation (5). Another study showed that 24 of 42 RABV isolates were taken from dogs or cats classified as clinically normal (3). All animals from these studies were from areas experiencing rabies epidemics.

All cumulative convincing data for more than half a century from various parts of the world call into question either the concept of a rabies carrier state or the quality of research indicating such a state. Other persistent virus infections routinely occur in lymphocytes, monocytes, macrophages, and dendritic cells through the viruses’ curtailment of the host’s antivirus immune responses. Experimental RABV replication in murine bone marrow macrophages and in human macrophage–like cell lines suggests a mechanism of virus persistence (22). However, wild-type RABV is highly neurotropic. If RABV persistent infection occurs in neurons, these findings contradict the current understanding of RABV pathogenesis. Clearly, RABV infection causes minor morphologic changes in neurons but may result in neurophysiologic dysfunction. Also, virus persistence generally is life-long in infected hosts. The longest surviving presumed carrier dog lived for only 16 months; in this dog, the tonsil was found to be the only organ from which the virus was isolated (14). No data are available concerning how long such carrier dogs survive. In healthy vampire bats, the duration of salivary excretion of RABV was reported to be 690 days after infection by an extremely high dose of RABV (8). Generally, persistent infections are characterized by an excess of viruses or virus antigens; free antibodies, which circulate without binding to antigens, are difficult to detect. However, experimental carrier dogs and vampire bats presented high rates of virus-neutralizing antibodies in serum in these reports (8,14). From an evolutionary perspective, a carrier dog with normal behavior does not pose an advantage for virus survival because biting when an animal is controlled by an aggressive brain is the only major route for RABV to spread. Animal behavior change is fundamental for RABV survival/transmission. Therefore, if carrier dogs exist, they are inferior to rabid dogs for disease transmission.

In a report by Zhang (17), 15 dogs that were diagnosed as positive by ELISA were confirmed to be negative by the standard direct fluorescent antibody (DFA) method. Explanations for these contradictory findings include misidentification of infected dogs, detection of RABV early in the prodromal course once it has reached the central nervous system after infection and incubation, and inadequate diagnostics. These phenomena, rather than the existence of carrier hosts, may explain historical reports of asymptomatic rabies in enzootic areas.

A well-established surveillance system for infectious diseases depends on reliable, laboratory-based diagnostic methods. Human and animal rabies cases in China have been reported mainly on the basis of clinical presentations and retrospective epidemiologic surveys. Animal rabies is rarely diagnosed in China. Human rabies diagnosis based solely on clinical symptoms is unreliable because human rabies can be confused with Guillain-Barré syndrome, poliomyelitis, and other types of encephalitis (23). Similarly, rabies in animals is difficult to distinguish from canine distemper and other encephalitic conditions. Postmortem rabies diagnosis should be routinely performed on rabid animals, animals that have bitten victims, and human patients who die after an animal bite. Antemortem diagnosis of rabies in humans is challenging because of the disease’s long and variable incubation period. Also, distribution of virus antigens, virus nucleic acids, and antibodies is unpredictable at this stage (24). Consequently, all countries should establish standardized national rabies diagnostic protocols for postmortem examinations.

The DFA method, first introduced in the 1950s, is the global standard procedure for rabies diagnosis. It is simple, economical, and reliable (25). This method is approved by the World Health Organization (WHO) and the World Organisation for Animal Health (OIE) and has served as a cornerstone for rabies diagnosis for the past half century (26). All rabies diagnostic laboratories should follow a single standard protocol (27). Whether such a standard exists in China is unclear, and diagnostic reagents, equipment, and qualified diagnosticians are in short supply. The ELISA for RABV antigen detection is carried out in a few laboratories, but this method requires performance evaluation using the standard DFA method to assess specificity, sensitivity, and reproducibility. An ELISA-based rapid rabies enzyme immunodiagnosis (RREID) method that uses monoclonal antibodies against nucleocapsid and glycoprotein for rabies diagnosis has been developed at the Wuhan Institute of Biological Products (3). One analysis used the RREID method to detect potential virus antigens in dog saliva; when verified by the DFA method, all 15 samples showed false-negative results (17).

Much inconsistency exists when different methods are used for rabies diagnosis (6). Overall, of 76 positive samples examined by the DFA method (6, cited as an indirect immunofluorescent antibody method), only 36 were confirmed by RT-PCR. In a certified rabies reference laboratory, DFA-positive samples should be positive when confirmed by a sensitive RT-PCR method. Therefore, in China, either the DFA or the RT-PCR, or perhaps both protocols, have questionable validity. The high prevalence of rabies in China necessitates establishment of a standardized national DFA protocol in provincial Centers for Disease Controls and veterinary stations. Other methods should be compared against the established DFA standard. A direct rapid immunohistochemical test (dRIT) using low-cost light microscopy has been extensively investigated and shown to have excellent agreement with the DFA (28). The dRIT can be completed within 1 hour, and this method is a feasible alternative at the county level for confirmatory rabies diagnosis or enhanced field surveillance. The urgent need to establish an improved national standard DFA protocol in China should take precedence over current efforts to develop ELISA and RT-PCR methods.

From 1885, when the first human rabies vaccination occurred, to 1994, when the RV Street–Alabama-Dufferin (SAD) B19 strain was engineered with reverse genetics (29), methods for RABV manipulation have changed fundamentally from random attenuation to defined modifications. However, the basic concept for rabies vaccine development has not changed for more than a century. Several major modern human rabies vaccines include duck embryo vaccine, commercialized in 1957; human diploid cell vaccine, introduced in 1978; purified chicken embryo cell vaccine, developed in 1984; and a purified Vero cell rabies vaccine (PVRV), developed in the late 1980s.

In theory, all mammals are susceptible to RABV infection. In China, dogs play the dominant role in rabies transmission (4,5). Statistically, >95% of human rabies cases in China are due to rabid dog bites (16). In isolated serologic studies of rabies in wildlife such as badgers, raccoon dogs, rodents, and bats, no RABV was successfully isolated from these animals. Rodents do not serve as reservoirs for rabies. In China, all RABV isolates characterized by phylogeny using the N and G gene sequences are categorized into classic RABV genotype 1 (3,6). Homology among isolates is >90% at the amino acide level. Although some subgroups are suggested, the differences would be minor. In addition, RABV isolates from dogs on all continents are grouped into genotype 1. Phylogeny analyses reinforce the perspective that vaccine matching in China is redundant. In a rabies-epidemic region such as China, rabies in wildlife may result from spillover from dogs. Without proper investigation of animal population density and characterization of the RABV isolates, wildlife rabies in China can be elucidated only after dog rabies is well controlled.

The recent reemergence and severe incidence of rabies have attracted the attention of scientists and administrative authorities in China. However, efforts are distracted by and concerns misleadingly focused on healthy dog carriers, possible rabies in wildlife, vaccine matching, inferior or counterfeit vaccines, and seroconversion testing after vaccination. In the 1920s, long before the recognition of bat and other wildlife rabies and the availability of modern vaccines, rabies in Japan was successfully controlled through mass vaccination of dogs. At present, in large cities such as Beijing and Shanghai, although animal bites are fairly common, rabies cases are rare. In 2006, ≈140,000 animal bites were recorded in Beijing, but few human rabies cases were reported (16), mainly because of adequate dog vaccination. However, in rural areas of China, dog rabies vaccination coverage is <3% (4,32). WHO has determined that vaccination coverage >70% is needed to sufficiently control canine rabies, but the exact level of sufficient coverage varies according to demographic, behavioral, and spatial characteristics of dog populations. Historically, the 4 rabies epidemic waves in China reflect discontinuous efforts of dog control and vaccination. The periodic recess of rabies at 10-year intervals resulted from strict depopulation of dogs rather than mass vaccination. The density of 4.5 dogs/km² can lead to endemic rabies in vulnerable populations (36). The current estimated number of dogs in China is 80–200 million (5). If dogs are distributed evenly, the average density of dogs in China is 8–20/km². The estimated population density would be much higher in rabies-endemic provinces. With very low vaccination coverage in dog populations, rabies outbreaks are not surprising. The phenomenon of 4 rabies epidemic waves during the last half century in China corroborates our interpretation and shows that depopulation of dogs alone cannot efficiently control rabies. Mass vaccination of dogs has been demonstrated to be the most efficient way to control the disease (37,38).

Decisions to initiate or withhold PEP are based on postmortem diagnosis of the biting animal and use of a standardized method (27). Patients exposed to rabies in China were estimated to be between 1% and 10% of the population, according to disease prevalence in different areas. However, because of China’s weak national animal rabies diagnosis network, the number of PEP patients in China is probably arbitrary or only roughly estimated. To maximize efficiency of the limited resource of human vaccines and rabies immunoglobulin for patients at real risk, establishing a systemic rabies diagnosis network is imperative. At a minimum, several million persons are expected to require PEP each year.

RABV is highly neurotropic and functionally conservative. It is transmitted mainly through animal bites, although rare nonbite exposure routes have been reported. Because of this characteristic, the dissemination of rabies is relatively slow. The spread of the disease needs a minimum threshold support of host population density (36). Therefore, the frontline of waves of rabies infections garner little public attention. Once the density of dogs reaches a critical threshold, rabies spreads rapidly, and the spread can be accelerated through animal translocation. However, the animal population itself does not pose a rabies threat. An immunized dog population can be a solid barrier to prevent rabies from spreading to humans (37,38). The immediate challenge for rabies control in China is to stockpile enough vaccines for mass dog vaccination campaigns. A reliable systemic diagnostic network can support effective epidemiologic investigation, vaccine campaigns, and initiation of PEP by providing multiple opportunities for collaborations that work toward practical, humane, and economical rabies elimination in China.