The basis for plague diagnosis has changed over the last century. For purposes of this summary, a case of plague was defined as a clinically compatible human illness and at least 1 of the following: 1) Y. pestis isolated from or detected in a clinical specimen, 2) elevated antibody titer to Y. pestis F1 antigen in >1 serum specimen (22), or 3) supportive epidemiologic and other laboratory evidence (e.g., visualization of typical Y. pestis morphology on a stained slide). The clinical form of plague (e.g., bubonic, pneumonic, septicemic) was determined on the basis of explicit notations in the case records or from available clinical details; only the primary clinical form was considered. For example, patients who had primary bubonic plague and secondary pneumonic plague were classified as having bubonic plague.

Data from cases occurring during 1900–1981 were collected from lists maintained by the United States Public Health Service and later the US Centers for Disease Control and Prevention (CDC), and enhanced with additional data sources including state reports and publications in the peer-reviewed literature. Supplementary detailed information on clinical course, exposure, and treatment was collected and maintained for most plague patients beginning in 1956, continuing through the present time.

Human cases acquired within the continental United States were included in this summary. Case-patients were geographically represented by state of residence or state of exposure, as indicated. Antibiotics considered effective for plague for the purpose of this analysis were: streptomycin, gentamicin, tetracycline/doxycycline, chloramphenicol, fluoroquinolones, or co-trimoxazole (22). Categorical variables are described as counts and proportions, and statistically compared by using χ² tests with α = 0.05. Continuous variables are described by median and range.

Invasion of a geographical location by exotic plant and animal species has been described as a multistep process involving transport, introduction, establishment, spread, and progressive impact on the ecology and human population (23–25). Multiple introductions are often necessary for an invading species to become established, and once established, a “lag period” of years to decades often ensues during which the invading species remains relatively localized. This is typically followed by a period of rapid geographic spread, ultimately resulting in increased ecologic and human effects.

Seen from the perspective of human infection, the emergence of plague in the United States is analogous to the invasion process of exotic plant and animal species. Available evidence indicates that Y. pestis was introduced on multiple occasions into various port cities; however, establishment appears to have been successful only in Pacific port cities (3,7). In San Francisco, infection spread from a cycle involving urban rats and their fleas into native wild ground squirrels (Citellus spp.) on the outskirts of the city by 1908 (3). In contrast, Y. pestis apparently never successfully established in wild rodent populations outside of Gulf coast port cities, likely a result of inhospitable ecology and early and extensive urban rat control efforts (3,5,8).

For the next 2 decades, the disease remained localized to port cities and surrounding areas, causing intermittent but large outbreaks. Following this lag period of minimal spread, the pathogen began dispersing rapidly eastward from the Pacific Coast (7). In 1935, Y. pestis was documented among wild rodents outside of California. By the early 1950s, plague had reached its current geographic distribution, and since 1965 has caused consistent human illness, suggesting that Y. pestis has become fully entrenched in enzootic cycles throughout the West (3). Together with their respective flea species, multiple rodents contribute to the current ecology of plague in United States, including ground squirrels, prairie dogs, wood rats, chipmunks, deer mice, and voles (26).

Shifts in epidemiologic features of human plague have occurred during the years following its introduction. The demographic characteristics of plague patients are inextricably linked to the geographic distribution of the organism over time and the populations at risk in those areas. For example, the disproportionate excess of men among Asian case-patients in the early years may solely represent the male-dominated Asian immigrant population in the port cities of the Pacific at the beginning of the 20th century (27,28). The demographic shift in burden of illness from persons of Asian heritage to those of American Indian heritage over time reflects change in the geographic distribution of plague from overcrowded urban neighborhoods dominated by immigrant populations along the Pacific Coast to the rural areas in the Southwest, including tribal lands. The proportional increase over time in primary septicemic cases was likely the result of better recognition of a clinically indistinct form of infection coincident with improved laboratory diagnostic capability (i.e., blood cultures), or possibly a increase in number of infections associated with direct animal contact.

The overall plague mortality rate decreased with availability of effective treatment. Nevertheless, the risk of death from plague infection is still substantial, particularly for patients with primary septicemic, pneumonic, and pharyngeal manifestations. Although exposure information was limited, infection acquired through animal contact was associated with higher mortality than was infection transmitted by flea bite. Receipt of effective antibiotics was similar between case-patients who had animal contact and those bitten by fleas. The difference in mortality rates may be related to higher or more direct initial inoculum of bacteria or differences in protein expression between mammals and fleas (29–31).

This summary of human plague in the United States may underrepresent some infections. Although Y. pestis is among the most pathogenic bacteria known, and mild, self-limiting infection is not considered commonplace, this summary could exclude mild and undiagnosed infections (32). Unrecorded plague infections may also have occurred with unrecognized septicemic disease or among racial or ethnic minorities, whose populations typically have less access to medical care. Additionally, minorities may not have sought medical care because of fear of racial prejudice associated with the initial introduction of plague (3,33). Unrecorded plague infections would more likely have occurred in the early 20th century, however, and are unlikely to dramatically affect the overall trends observed in this analysis.

Now an endemic zoonosis in the United States, plague is likely to continue causing rare but severe human illness in western states. Historically, plague was often linked to poor sanitation that resulted in rodent infestations. However, plague in New Mexico has increasingly occurred in more affluent areas, a result of continued suburban and exurban development in enzootic plague foci (11,36). Regardless of a person’s race, ethnicity, or socioeconomic status, the primary risk factors for plague infection in the United States are behaviors and conditions that increase both direct and indirect human contact with rodents and their fleas. In recent years, few patients have reported clear exposures to infected animals or rodent fleas; most were likely infected while performing common outdoor peridomestic work (e.g., cutting brush or chopping wood) or as a result of contact with infected fleas that were brought into the home by indoor/outdoor pets (18,37; CDC, unpub. data). Clinical suspicion remains critical to early and appropriate treatment. Recommended treatment is with aminoglycosides and tetracyclines, but fluoroquinolones may also be effective (34). The US Food and Drug Administration recently approved levofloxacin for treatment of patients with plague, based on in vitro and animal studies (35). Additional antibiotics considered effective for the purposes of this summary should not be considered first-line treatment for plague. Accurate plague diagnosis is further challenged by reliance on automated identification systems that frequently misidentify Y. pestis (38). A few travel-associated cases have been diagnosed in areas without endemic plague (e.g., South Carolina, New York City, Connecticut), highlighting the importance of plague in the differential diagnosis of ill persons, even those lacking apparent lymphadenopathy, with recent travel history anywhere in the western United States (39,40; CDC, unpub. data).