Factors Associated with Rabid Animals Since the Introduction of Raccoon Rabies Variant in Massachusetts,1992–2007

Introduction

S ince the 1950s, rabies has been more frequently found and reported among wildlife species than among domestic animals in the United States. A combination of animal control efforts and dog vaccination campaigns, initiated in the 1940s and 1950s, has eliminated canine rabies (Rupprecht et al. 1995, Blanton et al. 2007). Raccoons infected with a raccoon rabies variant (RRV) virus were first recognized in central Florida in the 1940s and were introduced into the mid-Atlantic region via translocation by hunters in 1970s (Jenkins and Winkler 1987). The virus spread among wild animals at approximately 40–48 km/year, eventually reaching New England area in the 1990s (Jenkins and Winkler 1987, Wilson et al. 1997, Smith et al. 2002). The RRV is now considered enzootic throughout the eastern United States. The epidemiology of raccoon rabies in the United States is distinctive from that of fox rabies in Europe. Fox rabies is largely restricted to the host species with minimal spillover to other mammals and rarely occurs in urban areas, largely due to the behavioral patterns of the reservoir host except for the reports from Canada (Rosatte 1988, Rosatte et al. 2007). The RRV has emerged as a unique threat to human and wild animal health, agricultural production, and public safety (Real et al. 2005). This variant is also frequently found in other terrestrial species, particularly skunks, and because of the successful adaptation of raccoons to urban and suburban environments, it occurs in areas heavily populated with humans and their domestic animals (Guerra et al. 2003).

Massachusetts confirmed its first case of the animal RRV infection in September 1992. The 10 most frequently tested animals (in order) in Massachusetts are cats (Felis catus), bats (multiple species, majorly big brown, Eptesicus fuscus), dogs (Canis familiaris), raccoons (Procyon lotor), skunks (Mephitis mephitis), squirrels (gray squirrel, Sciurus carolinensis; red sqirrels, Tamiasciurus hudsonicus; northern flying squirrel, Glaucomys sabrinus; and southern flying squirrel, Glaucomys volans), woodchucks (Marmota monax), opossums (Didelphis virginiana), foxes (gray fox, Urocyon cinereoargenteus, and red fox, Vulpes vulpes), and rabbits (multiple species, including Sylvilagus floridanus, Oryctolagus cuniculus, and Lupus americanus). The animals most often found positive are raccoons, skunks, bats, cats, foxes, woodchucks, cows, coyotes, and dogs (Wang et al. 2008).

Postexposure prophylaxis with passive rabies immune globulin and vaccination remains the only effective way to prevent human rabies (Corden and Kazmierczak 2000). A combination of land use and human population density has been shown to be predictive of risk of human exposure to RRV (Jones et al. 2003). The purpose of the present analysis was to identify factors associated with a positive rabies laboratory test for RRV. Although decisions regarding the need for postexposure prophylaxis have to be based on the laboratory test result if the suspect animals are available, our data may serve to guide judgment regarding risks in unusual animal species or with unusual behavior.

Materials and Methods

Results

The total number of specimens tested and the number confirmed rabid from each animal species are listed in Table 1. Among the animals, the 10 most commonly tested animals, plus cows and coyotes, comprised 95.9% (45,203/47,162) of the total specimens and 99.6% (4520/4538) of the rabid animals. The pooled reference group, LTALFRAS, contributed to 4.2% of the specimen volume and 0.4% of the rabid animals identified. Out of 46 animal species included in the pooled reference group, 9 were found to include at least one positive specimen, including bobcat (3/9), chinchilla (1/2), deer (1%, 1/98), fisher (7.4%, 2/27), goat (0.9%, 1/106), horse (2.4%, 3/124), otter (3/6), pig (10.3%, 3/29), and shrew (2%, 1/49). Wild animals were the most frequently tested type of animal (47.2%) and also the most frequently found to be rabid (96%). Pet/companion, stray, and undefined types of animals contributed 34.6%, 16%, and 2.2% of the specimens, respectively. The majority of stray animals were cats (94.2%).

Up-to-date on vaccination were reported in 13.3% (6242/47,162) of the specimens and in 38.2% (6242/16,335) of pet/companion animals, and eight rabid animals (six cats, one goat, and one cow) were found among them. There were eight rabid dogs that were infected with the RRV during the study period, and none of them were up-to-date on vaccination. The animas with at least one of the observed sign of disease were reported in 56.5% (26,638/47,162) of the specimens.

“Human exposure only” was listed as the reason for requesting rabies testing for 61.1% (28,832/47,162) of the specimens, and 2.4% (702/28,832) of them were positive for the rabies. About 21.2% (9989/47,162) of the specimens were submitted with “pet/companion animal exposure only” and had the highest positive proportion for rabies at 26.9% (2689/9989). By the end of 2002, the RRV had been identified in most of Massachusetts except the islands comprising Dukes and Nantucket Counties. About 2097 (4.5% of total) animals were submitted from counties before the arrival of RRV had been documented, and they contributed 0.6% (26/4538) of rabid animals (Table 2).

Factors that were significantly associated with a specimen testing positive for rabies in the model included presence of the RRV in the county where the animal was found; not up-to-date on vaccination; multiple exposures; reasons for rabies testing; and observed signs of aggression, ataxia, disorientation, paralysis, and unexplained wounds. These were all significant at α = 0.05 level after adjusting for animal species and type of animal. The observed signs of “found dead” and “seizures” were negatively associated with a positive rabies test. The odds ratios of each associated risk factor are listed in Table 3. The logistic regression model is significant (p <0.0001) and c-statistics is 0.94 (0.8% tied and 93.4% concordant).

A complete set of first- and second-order interaction terms were tested to determine whether there was sufficient improvement in the model to justify their inclusion. None of the interaction terms were included in the final model. The probability of a specimen testing positive was calculated directly from the estimates of the factors left in the final model (Table 3). The ROC of the final model was compared with the one containing only three variables: animal species, type of animal, and observed signs of disease (Wang et al. 2008). The additional factors increased the predictive value of the model and decreased percent tied from 3.2 to 0.8, as well as increased percent concordant from 90.9 to 93.3 (Fig. 1).

OPEN IN VIEWER

FIG. 1.

ROC of the final model versus a model with categorized animal species, type of animal, and syndromes/observations.

Discussion

This study demonstrated that introduction of RRV in the county where the animal was found, reason for rabies testing, animal vaccination status, occurrence of multiple exposures, and observed signs of disease and unexplained wounds were all significantly associated with the rabies test result after adjusting for animal species and type (p < 0.05). Type of animal was close to border line significance in the final model at α 0.05 level (p = 0.0694). The c-statistic, a measure of the discriminative power of the model, supports validity of the final model. The inclusion of the defined factors increased the predictive ability of the model over the base model with categorized animal species, type of animal, and observed signs of disease (Wang et al. 2008).

All testable specimens were included in the analysis to identify the association of observed signs of disease with positive rabies test results (Wang et al. 2008). Specimens from outside of Massachusetts were excluded, as were United States Department of Agriculture–submitted surveillance specimens (collected to monitor the spread of RRV and evaluate a wildlife rabies vaccination project). We obtained similar prediction results with signs of disease, including aggression, disorientation, paralysis, ataxia, and unexplained wounds. However, seizures and animals found dead were negatively associated with a diagnosis of rabies.

Compared to the reference group, the high-risk animals for rabies were raccoons, skunks, foxes, cows, coyotes, woodchucks, and bats, while dogs, opossums, and squirrels were low-risk animal species. Raccoons and skunks were 95 and 45 times more likely to be positive for rabies than the reference group, respectively. Cows, foxes, and coyotes were also more than 10 times as likely to be positive for rabies as compared to the reference group (OR = 16.7, 20.67, and 13.67, respectively). Although opossums, squirrels, and rabbits were all among the 10 most frequently tested animal species in Massachusetts, only one out of 1500 squirrels and 559 rabbits were rabies positive. No rabid opossums were found out of 789 testable specimens. There was no evidence that these animals would be more naturally resistant to the RRV than the other animal species. Most likely, the squirrels, the rabbits, and the opossums would not have the chance to survive an attack by an aggressive rabid raccoon. As testing resources become limited, some animals such as opossum, rabbit, and squirrels may not be routinely tested unless a human and/or pet exposure is documented. This may justify more discriminating submission of these animals for testing.

Although the pooled animal species reference group included only a small proportion of the total number of specimens submitted and an even smaller proportion of the total rabid animals, the group was not homogeneous. This group comprised rarely submitted animals, including bobcats, chinchillas, fishers, otters, and pigs, which had a very high proportion positive for rabies.

In a study from Colorado, the prevalence of rabies in bats that bit humans was 2.1 times higher than in bats that did not bite humans (Pape et al. 1999). This was consistent with our observation and other jurisdiction that aggressive animals posed a significantly higher risk of rabies than nonaggressive animals (Rosatte et al. 2006). However, the risk of rabies in aggressive animals versus nonaggressive animals (OR = 4.1) was almost twice the value calculated in the Colorado study. However, our study included wild and domestic animals in addition to bats. Specimens associated with multiple exposures, another factor highly associated with aggression, were 1.88 times more likely to result in a positive rabies test than from animals with single exposure or undefined exposure.

Animal species submitted from counties where the RRV had already been identified were 17.3 times as likely to be rabid than those from counties before identification of the RRV, and by late 2002, all counties in Massachusetts, except the islands comprising Dukes (including Martha's Vineyard) and Nantucket Counties, had confirmed the presence of the RRV. The result is consistent with our previous report that an extremely high proportion of non-bat rabid animals had been infected by RRV in Massachusetts since 1992 (Wang et al. 2009).

The results of this study reinforce the importance of maintaining up-to-date rabies vaccination in domestic pet/companion animals. None of the rabid dogs were reported to be up-to-date on their rabies vaccination. Among all animals, being current in vaccination status reduced their risk of being positive for rabies by 75%, after adjusting for all other factors. Unfortunately, 62% of submitted pet/companion animals, mostly cats, were reported not to be up-to-date for rabies vaccination.

One of the major strengths of this study is the extent of data, incorporating data on all testable specimens collected over a 16-year period. The time span includes the period during which the RRV was first introduced into and spread across Massachusetts, allowing for comparisons between the pre- and postintroduction periods. In addition, data were extracted from the rabies specimen submission forms, which are completed and recorded at the time of specimen collection and before laboratory testing.

Several limitations of the study must be acknowledged. The rabies specimen submission forms are designed to be straight forward and are largely completed by professionals from veterinary clinics and local boards of health, police departments, and local animal control personnel. However, the information used to complete the forms is gathered from reports from a variety of other sources, many times from individuals who have just experienced some type of direct or indirect contact with an animal that is possibly rabid, and this may abnormally weigh certain categories of information. For instance, members of the public are likely to interpret normal territorial defense by a wild animal as aggression. Another source of information error or incompleteness occurs when the individual completing the submission form is not the individual who responded to the inciting situation. For example, police officers called out during the night shift to respond to a potentially rabid animal may not complete the specimen submission form, but leave it for the day shift or animal control officer. Information could be conveyed inaccurately or lost during those transfers.

The data for this study were derived from existing records. There was no opportunity to verify the accuracy of the report retrospectively, although the laboratory staff routinely attempt to get complete information for submitted specimens before testing. For instance, vaccination status cannot always be validated. There did not appear to be a difference in completeness of the form related to the submitter, rather an inherent inconsistency simply due to the huge number of submitters involved.

A final limitation of the study is related to the fact that the SLI Rabies Laboratory encourages the submission of any mammal that has either bitten or scratched a human or domestic animal. This results in a large number of low suspect animals being submitted for testing. These animals, particularly vaccinated pets and small rodents and lagomorphs with neurological symptoms, may dilute the effect of observed signs as predictive. Some observed signs of diseases would be very difficult to detect in certain animal species (e.g., salivation in bats), which may also distort or mask inferences.