Appropriateness of Rabies Postexposure Prophylaxis Treatment for Animal Exposures

Context Rabies postexposure prophylaxis (RPEP) treatments and associated costs have increased in the United States. The extent to which RPEP use is consistent with guidelines is not well understood.

Objective To characterize animal contacts and determine the frequency and factors associated with inappropriate RPEP use.

Design, Setting, and Patients Prospective case series study of patients presenting with an animal exposure–related complaint from July 1996 to September 1998 at 11 university-affiliated, urban emergency departments (the Emergency ID Net).

Main Outcome Measures Exposure type, circumstances, and RPEP use (appropriateness defined by local public health departments).

Results Of 2030 exposures, 1635 (81%) were to dogs; 268 (13%) to cats; 88 (4%) to rodents/rabbits; 10 (0.5%) to raccoons; 5 (0.2%) to bats; and 24 (1.2%) to other animals. Among those exposed, 136 (6.7%) received RPEP after dog (95), cat (21), raccoon (8), bat (4), or other animal (8) exposures. Use of RPEP varied by site (range, 0%-27.7% of exposures), with most frequent use reported at sites in the eastern United States. Management was considered appropriate in 1857 exposures (91.5%). Use of RPEP was considered inappropriate in 54 cases (40% of those in which it was given), owing to factors including animal availability for observation and exposure in a low-endemicity area. Rabies postexposure prophylaxis was considered inappropriately withheld from 119 cases (6.3% of those not receiving RPEP), often because a domestic animal was unavailable for observation or testing.

Conclusion These results suggest that use of RPEP is often inappropriate. Greater compliance with current guidelines would increase RPEP use. Physician education, improved coordination with public health officials, and clarification of RPEP guidelines could optimize use of this expensive resource.

Although human rabies infection is rare in the United States, animal bites are a frequently encountered problem in clinical practice. Several million US residents are victims of animal bites each year. Dog bites alone account for more than 300,000 emergency department (ED) visits, with total reimbursement of more than $100 million annually.¹ Rabies is invariably fatal but is preventable with proper treatment, including rabies postexposure prophylaxis (RPEP) for certain types of animal exposures. Control of rabies in domestic animal populations and use of RPEP has led to a decline in annual human rabies cases from more than 100 at the beginning of the 20th century to 1 to 3 per year at present. Only 32 cases of human rabies were diagnosed in the United States between 1980 and 1996.²

Since the 1970s, the expanding epizootic of rabies in raccoons in the eastern United States has caused increased concern about potential transmission to humans. In 1997, a total of 8509 animal rabies cases were reported in the United States—an increase of 19.4% from 1996.³ The epizootic of rabies in raccoons appears to be expanding, and reports of rabies in cats have increased in the same areas.⁴ Although no documented human rabies cases have occurred with the raccoon rabies virus variant, the use of RPEP has greatly increased.⁵ In 1987, it was estimated that 18,000 RPEP treatments were given in the United States; in 1997, an estimated 39,000 treatments were given.³ Although costs of RPEP can vary greatly, the cost in the United States is approximately $1500 per treatment course for biologics alone.⁶ Physician and follow-up clinic charges add even more to the cost of treatment.

General guidelines for proper administration of RPEP have been published by the Advisory Committee on Immunization Practices (ACIP), and many local public health agencies publish specific guidelines based on local rabies epidemiology.⁷ Guidelines indicate that unvaccinated persons potentially exposed to rabies should be treated with human rabies immune globulin and begin a 5-dose series of rabies vaccine. Although it is suspected that many RPEP treatments are not indicated based on guidelines, no prospective studies have investigated this practice.

Because RPEP is often initiated in the ED, we chose to prospectively study RPEP use in this setting. The objectives of this study were to characterize animal exposures of patients who presented to the ED and to determine the frequency of and factors associated with inappropriate RPEP use. Understanding the extent of and reasons for inappropriate RPEP use could facilitate development of strategies to reduce unnecessary RPEP while ensuring that RPEP is given when appropriate.

The current study was a prospective case series of patients presenting to a network of 11 geographically diverse, university-affiliated urban EDs (the Emergency ID Net). The approximate total combined annual visit census of these EDs is 900,000. A detailed description of the Emergency ID Net, including administrative structure, data collection methods, and data transfer, has been published.⁸ Institutional review board approval was obtained at all sites.

Patients presenting to these EDs between July 1996 and September 1998 with a primary complaint related to mammalian animal exposure by bite, scratch, body fluid exposure, handling, or proximity were enrolled in the study. Treating physicians (primarily emergency medicine residents) collected data prospectively at the bedside during the ED evaluation using standardized data entry forms. Data collected included demographic information, animal type, exposure type and circumstances, and RPEP use. Treatment decisions were left to the individual physicians. Patients were contacted by telephone approximately 1 month after enrollment to obtain follow-up data, including results of animal observation or animal testing for rabies and any subsequent treatment given.

To categorize RPEP treatments as appropriate or inappropriate, a standard algorithm for RPEP use was developed based on the animal type, exposure type and circumstances, animal vaccination status, and animal availability for observation or testing (Figure 1). The treatment algorithm was based on ACIP guidelines that were current at the time data were collected.⁹ Circumstances that were considered to indicate a provoked exposure included attempting to feed an animal, trespassing on an animal's territory, breaking up an animal fight, petting or playing with an animal, handling an animal in a veterinary office or laboratory, having contact with an injured animal, and walking, running, or riding past an animal. The treatment given was considered inappropriate if it deviated from the guideline algorithm.

The algorithm was reviewed with local public health officials at each of the study sites to ensure that the guidelines were consistent with local recommendations. Each of the study sites was identified by the local health department as a low-endemicity or high-endemicity area for wild or domestic animals, based on local rabies epidemiology. Sites considered to be high-endemicity areas for both wild and domestic animals were New York, NY; Philadelphia, Pa; Charlotte, NC; and Orlando, Fla. Sites that were considered high-endemicity areas for wild animals but low-endemicity areas for domestic animals were Kansas City, Mo; Atlanta, Ga; Albuquerque, NM; Phoenix, Ariz; Los Angeles, Calif; and New Orleans, La. The only site that was considered to be of low endemicity for both wild and domestic animals was Portland, Ore. Some sites (Atlanta, Kansas City, Albuquerque, and Philadelphia) considered a scratch from a terrestrial wild carnivore or larger animal (eg, raccoon, skunk, coyote) to be a potential exposure.

Case-finding audits were conducted by review of ED logs during two 1-month periods in January 1997 and January 1998 to determine the proportion of missed cases and to compare their demographic and clinical characteristics with identified cases. Case-finding sensitivity was defined as the total number of patients enrolled divided by the total number of patients who met entry criteria. Limited demographic and clinical data were also abstracted from charts of missed cases to compare with those of identified cases to evaluate selection biases.

Simple descriptive statistics were used to summarize characteristics of ED patients with animal exposures and the proportion of patients receiving RPEP, stratified by site and by animal type. Epi Info Version 6.04b (Centers for Disease Control and Prevention, Atlanta, Ga) was used to calculate univariate risk ratios (RRs) for factors related to inappropriate RPEP use. When appropriate, Mantel-Haenszel stratified analysis was used to adjust the RR estimates for confounding by other risk factors.

A total of 2030 patients with animal contact were enrolled. Their median age was 27 years (range, <1-91 years; interquartile range, 11-39 years), and 59% were male. The racial/ethnic distribution was 43% white non-Hispanic, 30% black, 22% Hispanic, 3.4% other, and 1.6% unknown. The distribution of animal types and RPEP use at the sites are summarized in Table 1. Bite exposures accounted for 1777 cases (87.5%). Other exposures included saliva contact with broken skin or mucous membrane (6.5%), scratch only (3.2%), touching only (3 cases), and proximity only (2 cases). The majority of exposures occurred in urban or residential areas; only 37 (2.0%) of 1873 exposures for which location information was available occurred in rural or wilderness areas.

Rabies postexposure prophylaxis was given to 136 (6.7%) of 2030 patients with animal exposures (Table 2). Although exposures to bats, raccoons, and other wild carnivores were uncommon, the majority of these patients received RPEP. Rabies postexposure prophylaxis was given for only 6.1% of dog and cat exposures, but because most exposures were to dogs and cats, these accounted for 86% of all RPEP treatments given.

Based on local guidelines, case management was considered inappropriate in 54 (40%) of 136 patients given RPEP and in 119 (6.3%) of 1894 patients who were not given RPEP. Overall compliance with guidelines was therefore 1857 (91.5%) of 2030. Inappropriate RPEP use occurred more frequently in low-endemicity areas (Table 3). Most exposures for which RPEP was given inappropriately were to dogs or cats (Table 4). The most common reason that RPEP was considered inappropriate was that the animal was available for observation or testing (28 of 54). Other reasons included exposure in a low-endemicity area due to a provoked attack with no unusual circumstances (22 of 54) and lack of bite or body fluid exposure (4 of 54). Inappropriate failure to give RPEP occurred more frequently in high-endemicity areas. The most common reason for inappropriate failure to give RPEP was exposure in a high-endemicity area from an animal unavailable for observation or testing (70 of 119). Other reasons included exposures in a low-endemicity area with an unprovoked attack or unusual circumstances for which the animal was not available (49 of 119).

Follow-up data were available for 985 cases (48.5%), including 69 (51%) of those who received RPEP and 916 (48%) of those who did not receive RPEP at the initial ED visit. Of the 69 receiving RPEP at the initial visit, 24 had not continued the full series at the time of follow-up approximately 1 month after the initial visit. In 7 of these cases, the animals had normal behavior during observation, and 1 was due to a rat exposure. Reasons for discontinuing treatment were unknown for the remainder. For 25 of the 69 patients receiving RPEP, the treatment was considered inappropriate, yet 10 (40%) had continued the full RPEP series at the time of follow-up. Of the 916 follow-up patients who did not receive RPEP at the initial visit, 45 should have received it according to local guidelines. For only 1 of these 45 patients, RPEP was started after the initial visit—a cat exposure in Florida for which the animal was not available.

No outside consultation for RPEP use was sought for 982 (87.1%) of 1127 cases for which this information was available. Consultation was sought from the local health department in 4.3% of these cases, from other public resources (eg, Centers for Disease Control and Prevention, poison center) in 3.2%, from another source in 4.1%, and from a written protocol in 2.5%. The proportion of inappropriate RPEP treatments did not appear to be related to whether consultation was used. Rabies postexposure prophylaxis was considered inappropriate for 16 (27%) of 60 cases for which it was reported that consultation was not obtained and was inappropriate for 2 (25%) of 8 cases for which the local health department was consulted.

Among the 1829 patients for whom RPEP was not considered indicated, there was no association between RPEP use and sex, age, race, insurance status, bite vs nonbite exposure, anatomic site, time elapsed since exposure, dog vs cat exposure, or class ification of site as high vs low endemicity. Rabies postexposure prophylaxis was more likely to be given inappropriately to those exposed to a wild vs domestic animal (RR, 9.5; 95% confidence interval [CI], 5.0-17.9, adjusted for availability of animal) or to an animal at a low-endemicity site that was not available for observation/testing (RR, 4.3; 95% CI, 2.7-7.1, adjusted for wild vs domestic animal).

Among the 201 patients for whom RPEP was considered indicated, there was no association between nonuse of RPEP and sex, age, race, insurance status, bite vs nonbite exposure, anatomic site, time elapsed since exposure, dog vs cat exposure, or animal availability. Rabies postexposure prophylaxis was more likely to be inappropriately withheld from those exposed to a domestic vs wild animal (RR, 5.0; 95% CI, 0.8-31.4) and from exposures associated with abnormal animal behavior occurring at sites classified as low vs high endemicity for rabies (RR, 1.4; 95% CI, 1.1-1.7).

Case-finding audits revealed that approximately 76% of eligible cases were enrolled. Compared with entered cases, missed cases were similar in terms of age (median, 37 years; range, 2-69 years), sex (63% male), race (59% white/Hispanic, 33% black, 17% other), type of animal (79% dog), and bite vs nonbite exposure (84% bite). Rabies postexposure prophylaxis was not given to any unenrolled patients reviewed in the audits.

The significant expansion of animal rabies in the United States has created a health care dilemma. Although RPEP is considered universally effective to prevent rabies, a disease considered to be invariably fatal, this treatment is expensive and may be overused based on current rates of human rabies. Past studies have retrospectively evaluated patients given RPEP and determined that it was frequently misused.^10-12 Consequently, a 50% reduction of RPEP use was a goal in the Healthy People 2000 objectives.¹³ To the best of our knowledge, this is the first prospective investigation of RPEP use and nonuse among patients presenting with an animal exposure. Our study demonstrated that RPEP was prescribed inappropriately in 40% of cases in which it was given, based on local guidelines. However, in more cases RPEP was not given when it would be recommended.

Many public health experts believe that improving physician compliance with guidelines for use of RPEP would reduce health care costs by reducing unnecessary use of RPEP. If all cases in our series were managed according to local guidelines, 54 cases in which RPEP should have been withheld would be offset by 119 cases for which RPEP should have been given, and the total number of RPEP treatments would actually increase 47% (from 137 to 201). Although our sites may not accurately reflect RPEP use in the entire United States, if this were extrapolated to the currently estimated 39,000 RPEP treatments given in the United States annually, more than 18,000 extra RPEP treatments would be given. At a cost of $1500 per treatment, this would result in additional expenditures of more than $27 million.

The sites at which RPEP was given most frequently are located in the region of the eastern US raccoon rabies epizootic, where the treatment strategies led to a more liberal use of prophylaxis. The greatest proportion of increased RPEP use at these sites appeared to be related to the strategy of giving it for exposures to domestic animals that are not available for observation. Since RPEP is considered appropriate in this setting, it would appear that the best way to reduce RPEP would be to increase the proportion of cases in which the animal is observed or tested. If physicians cannot be assured that proper animal observation or testing will take place, they may err on the side of giving RPEP, even for relatively low-risk exposures.

Although RPEP was given to a minority of patients with dog or cat exposures, these domestic animals accounted for the large majority of RPEP cases (including 85% of the inappropriately treated cases). In recent decades, more than 90% of animal rabies reported in the United States has occurred in wildlife.³ Because rabies is now rare among domestic animals in this country, it appears that RPEP use for domestic animal exposures could be targeted to reduce unnecessary RPEP. Common circumstances that should have precluded RPEP use for domestic animal exposures in our study included an animal being available for observation or testing, an exposure in a low-endemicity area in which the attack was provoked without unusual circumstances, and lack of exposure to animal saliva or body fluids. It may be useful to emphasize these common errors in written guidelines and educational programs for physicians.

Eight inappropriate RPEP treatments were for exposures to animals other than dogs or cats. Five of these exposures were to raccoons. Although some would recommend RPEP for raccoon exposures in high-risk areas pending results of brain testing, it is safe to withhold treatment if test results are available within 48 hours. Only 1 of our sites (Portland, Ore) stated that guidelines would not recommend treatment for a raccoon exposure if the animal were not available, yet 2 of 3 persons with raccoon exposures at this site received RPEP.

Rabies postexposure prophylaxis was given to 4 of 5 patients with bat exposures in this study. In the single bat exposure for which RPEP was not given, the bat tested negative for rabies. Bat-associated rabies virus variants have been identified from 17 (85%) of the 20 cases of human rabies acquired in the United States between 1980 and 1996.² Because a definite bite history was not elicited from many of these cases, the most recent ACIP guidelines emphasize that RPEP should be considered in all situations in which there is reasonable probability that contact with a bat may have occurred, unless prompt laboratory testing of the bat has ruled out rabies infection.⁷ The small proportion of bat exposures in our series would indicate that treatment of all bat exposures presenting to EDs would not significantly increase use of RPEP. However, a telephone survey in Oregon showed that 1.3% of respondents in the general population reported potential bat exposures and that treatment of all these individuals would represent an enormous burden on the US health care system.¹⁴

If the proportion of inappropriately withheld RPEP treatments is extrapolated to all animal exposures in the United States, it appears that a large number of potentially significant rabies exposures are going untreated. If this were the case, we would expect occasional cases of rabies among patients who were evaluated for animal exposures and not given RPEP. This has not been the case with any human rabies diagnosed in the United States in several decades. Although this is indirect evidence that RPEP is probably overused rather than underused, it is not known whether stricter RPEP guidelines could reduce RPEP use without withholding treatment for true rabies exposures.

Our study identified a large number of cases for which withholding RPEP was inconsistent with local guidelines, but this does not necessarily mean that it was an error to withhold RPEP. Good physician judgment in this setting may prevent the use of RPEP that might be considered indicated by a strict interpretation of local guidelines. All but 1 of these cases were due to dog or cat exposures and more than half involved exposures in a high-endemicity area for which the animal was not available for observation. Some areas that are currently considered as having high endemicity for rabies actually have very low reported rates of rabies in dogs and cats. Reclassification of these areas for domestic animal exposure risk could potentially reduce RPEP use safely. The other inappropriately untreated dog and cat exposures were due to unprovoked attacks or abnormally behaving animals that were not available for observation. Revision of guidelines to include very specific descriptions of the exposure circumstances that should prompt RPEP could reduce ambiguity. Although interpretation of animal behavior may be unreliable, perhaps future research could identify objective observations that are associated with a very low likelihood of rabies (eg, presence of a collar) such that RPEP could be safely withheld when the animal is not available for observation.

Several factors may limit the internal and external validity of this study. Not all patients with animal exposures were enrolled. However, our audits of all ED animal exposure cases indicated that enrolled cases were similar to nonenrolled cases. Physician behavior may have been influenced by the fact that a study of RPEP was being done. However, other than retrospective chart reviews for which data accuracy is seriously limited, no feasible and valid alternative study designs exist. We made no attempts to influence RPEP practices during orientation to the study. Rabies postexposure prophylaxis guidelines were not discussed, and physicians were instructed to treat patients as usual. Follow-up data were available on less than half of cases. We do not know how many patients who started RPEP treatments received the full series. Likewise, it is possible that we were not aware of some patients not given RPEP in the ED who received it subsequently. Practice patterns in large, urban, university-affiliated hospitals may not reflect practices in smaller community or rural hospitals. Because use of RPEP varies considerably by location, the proportion of patients with animal contacts who received RPEP in our study is sensitive to the selection of study sites, and may not be representative of the entire US ED population. However, we believe the geographic diversity of the network is adequate to describe RPEP use in US EDs with reasonable accuracy.

Analysis of the appropriateness of RPEP use is highly dependent on the algorithm used as the standard. It is possible that the local health department would agree on a general treatment algorithm but that the actual health department recommendation for a specific case might not strictly follow general guidelines. We do not know why physicians did not follow guidelines in individual cases. It is possible that there were specific details about the exposure circumstances that could not be adequately described using checklists on a data sheet. Although attacks were coded as "unprovoked" or specific types of animal behavior (eg, aggressive, lethargic) were coded as "abnormal," some subtle aspects of these cases that were not captured by our coding process may have influenced treatment decisions. Interpretation of subtleties is sometimes used as an argument against the use of written protocols or flow charts for RPEP. Written guidelines that are specific enough to address these concerns would seem to be the best way to achieve consistent and rational treatment of animal exposures based on the best scientific evidence.

Our data indicate that improved compliance with guidelines will not necessarily reduce the use of RPEP in the United States. For this to occur, the guidelines would need to be revised. Because the majority of RPEP use is for exposure to dogs and cats, and because rabies is now uncommon among these animals, it would appear that guidelines could be revised to reduce RPEP use for these exposures without significantly increasing the risk of human rabies. Health departments may wish to consider whether RPEP is really necessary for all dog bites for which the dog is not available—rabies in dogs is rare even in areas of the raccoon rabies epizootic. Perhaps treatments could be limited only to exposures involving specifically described high-risk circumstances.

The appropriateness of RPEP use in US EDs might be improved by developing and facilitating access to site-specific practice guidelines in conjunction with local public health agencies. Physicians should have a low threshold for consulting local public health agencies with questions regarding care and follow-up of persons with possible rabies exposure. Posting of the 24-hour public health department telephone number in the ED is recommended. Health departments could also post recommendations on Internet sites in an interactive algorithm format. Guidelines should be clear and provide examples regarding features that have been found to be associated with improper use of RPEP such as animal availability for observation, circumstances of exposure and provocation, and questionable exposure to animal body fluids. It is important that guidelines designed to reduce unnecessary RPEP do not lead to failure to give prophylaxis when truly indicated. As noted in a JAMA article more than 100 years ago, "With such a disorder prophylaxis is a necessity and any error had better be on the safe side."¹⁵ Future research should focus on development and evaluation of more specific practice guidelines and educational programs to reduce the need for and improve the appropriateness of RPEP use.