The persons in this study were participants in a survey conducted in 1994 by the Centers for Disease Control and Prevention (CDC). The primary objective of the survey was to assess the risk for hantavirus infections in persons whose occupations expose them to rodents. Participation in the survey was voluntary and entailed completion of a self-administered questionnaire and donation of a small volume of venous blood. Most of the 995 participants were enrolled at 1 of the following: American Society of Mammalogists meeting (Washington, DC, 1994), Wildlife Disease Association meeting (Pacific Grove, California, 1994), Southwestern Association of Naturalists meeting (Emporia, Kansas, 1994), Wildlife Society meeting (Wenatchee, Washington, 1994), 16th Vertebrate Pest Conference (Santa Clara, California, 1994), and Colorado Pest Control Meeting (Denver, Colorado, 1994). The other participants mailed their completed questionnaires and serum samples directly to CDC.

The questionnaire included detailed questions about previous exposure to rodents, use of personal protective equipment to minimize exposure to rodent excretions and secretions, and any previous occurrence of a severe febrile illness that included shortness of breath. The lifetime number of rodents handled by a person was measured categorically: I (1–99), II (100–499), III (500–999), IV (1,000–9,999), V (10,000–49,999), and VI (≥50,000). Use of gloves, protective masks equipped with high efficiency particulate air (HEPA) filters, and protective eyewear also was measured categorically: always (>90% of the time), usually (50%–90% of the time), sometimes (10%–49% of the time), seldom (<10% of the time), or never.

This study was restricted to the 757 participants in the CDC survey who had a history of exposure to rodents in North America and a history of occupational exposure to deer mice, white-footed mice, California mice, woodrats (Neotoma spp.), other neotomine rodents, cotton rats (Sigmodon spp.), oryzomyine rodents (Oryzomys spp. or Oligoryzomys spp.), or other sigmodontine rodents. Of the persons included in the study, 699 had worked with rodents only in North America. The 58 others had worked with rodents in North America and in South America. The geographic distribution of exposure to rodents in North America was Canada (n = 36), Alaska (n = 8), the contiguous United States or District of Columbia (n = 726, Table 1), Mexico (n = 91), Guatemala (n = 8), Belize (n = 3), Honduras (n = 3), Costa Rica (n = 21), Nicaragua (n = 4), and Panama (n = 8). Of the persons included in the study, 468 (61.8%) had worked with rodents in >1 state within the contiguous United States.

Persons included in the study had worked with rodents from 1 month to 65 years (mean 12.5 years). The total number of rodents handled by any 1 person ranged from category I (1–99) to category VI (≥50,000); the median was IV (1,000–9,999). Of the 757 persons in the study, 751 (99.2%) had handled deer mice, white-footed mice, cotton rats, oryzomymine rodents, California mice, or woodrats (Table 2).

HPS was first recognized as a clinical entity in 1993 in the southwestern United States (20). From March 1, 1993, through September 19, 2006, a total of 453 laboratory-confirmed HPS cases were reported to CDC from the contiguous United States (www.cdc.gov/ncidod/diseases/hanta/hps/noframes/epislides/episl7.htm). Of these, 259 (57.2%) were reported from 6 states in the southwestern United States: Colorado (n = 51), New Mexico (n = 71), Utah (n = 25), Arizona (n = 49), Nevada (n = 18), and California (n = 45). Sin Nombre virus is the only virus known to cause HPS in these 6 states. In this study, 387 (51.1%) persons had worked with rodents in Colorado (n = 124), New Mexico (n = 111), Utah (n = 65), Arizona (n = 90), Nevada (n = 33), or California (n = 169) and had handled deer mice. The total number of deer mice handled by persons in this group ranged from I (1–99) to VI (≥50,000); the median was II (100–499).

The geographic range of BAYV includes Georgia, Louisiana, and Texas (5,15–17), BCCV has been found only in Florida (4), and the geographic range of NYV includes New York, Pennsylvania, and Rhode Island (3,21,22). In this study, 22 persons had worked with rodents in Georgia (n = 1), Louisiana (n = 2), or Texas (n = 20) and handled oryzomyine rodents. The total number of oryzomyine rodents handled by persons in this group ranged from I (1–99) to IV (1,000–9,999); the median was I (1–99). Fourteen persons had worked with rodents in Florida and handled cotton rats. The total number of cotton rats handled by persons in this group ranged from I (1–99) to IV (1,000–9,999); the median was II (100–499). Eighty-one persons had worked with rodents in New York (n = 45), Pennsylvania (n = 42), or Rhode Island (n = 5) and handled white-footed mice. The total number of white-footed mice handled by persons in this group ranged from I (1–99) to V (10,000–49,999); the median was II (100–499).

BCNV virus has been found only in California (18) and TAMV only in Florida (14); the geographic range of WWAV and other arenaviruses naturally associated with woodrats (Neotoma spp.) includes Arizona, California, Colorado, New Mexico, Oklahoma, Utah, and Texas (19,23–27). In this study, 31 persons had worked with rodents in California and handled California mice (P. californicus). The total number of California mice handled by persons in this group ranged from I (1–99) to III (500–999); the median was I (1–99). As indicated previously, 14 persons had worked with rodents in Florida and handled cotton rats. Three hundred and thirty-three persons had worked with rodents in Arizona (n = 87), California (n = 130), Colorado (n = 76), New Mexico (n = 101), Oklahoma (n = 40), Utah (n = 59), or Texas (n = 101) and handled woodrats. The total number of woodrats handled by persons in this group ranged from I (1–99) to V (10,000–49,999); the median was I (1–99).

Lymphocytic choriomeningitis virus (LCMV) is the only Old World arenavirus species that is enzootic in North America. The house mouse (Mus musculus) is a member of the subfamily Murinae, family Muridae (11) and the principal host of LCMV. In this study, 526 (69.5%) persons had worked with house mice. The total number of house mice handled by persons in this group ranged from I (1–99) to VI (>50,000); the median was I (1–99).

Of the 757 persons in this study, 735 (97.1%) had worked with rodents before the discovery of HPS in 1993; during that time, 504 (68.6%) of them never or infrequently wore personal protective equipment (gloves, a protective mask equipped with HEPA filters, and protective eyewear) when handling rodents. In contrast, only 267 (36.3%) of these 735 persons never or infrequently wore personal protective equipment when handling rodents after the discovery of HPS. Use of personal protective equipment by the other persons in the study both before and after the discovery of HPS depended on the type of equipment.

All unique identifying information was removed from the serum samples before they were tested for antibodies. Furthermore, all unique identifying information was removed from the computer (electronic) records before analysis of the demographic and serologic data.

We tested the serum samples for immunoglobulin G (IgG) against SNV, WWAV, GTOV, and LCMV by using ELISA, as described (24,28). The SNV antigen was an Escherichia coli–expressed recombinant SNV nucleocapsid protein that is highly cross-reactive with other neotomine rodent-associated hantaviruses and with sigmodontine rodent-associated hantaviruses in the ELISA used in this study (T.G. Ksiazek, unpub. data). The control (comparison) antigen for the SNV IgG ELISA was an E. coli–expressed recombinant protein that is antigenically unrelated to the SNV nucleocapsid protein. The arenavirus antigens were detergent lysates of Vero E6 cells infected with WWAV strain AV 9310135, GTOV strain INH-95551, or LCMV strain Armstrong. WWAV is highly cross-reactive with BCNV and TAMV in the ELISA used in this study (M.L. Milazzo, unpub. data). Collectively, WWAV, GTOV, and LCMV represent the 3 major antigen groups in the family Arenaviridae, as defined by ELISA (19). The control antigens for the arenavirus IgG assays were detergent lysates of uninfected Vero E6 cells. The working concentrations of the SNV, GTOV, and LCMV antigens and the corresponding control antigens were determined by checkerboard titration against convalescent-phase serum samples from humans infected with SNV, GTOV, and LCMV, respectively. The working concentrations of the WWAV antigen and the corresponding control antigen were determined by checkerboard titration against a mouse ascitic fluid against WWAV strain AV 9310135. Serial 4-fold dilutions (from 1:100 through 1:6,400) of each serum sample were tested against the 4 test antigens and 4 control antigens. Antibody bound to antigen was detected by using a goat anti-human IgG (gamma chain–specific) peroxidase conjugate (Kirkegaard and Perry Laboratories, Gaithersburg, MD, USA). Optical densities (OD) at 405 nm (reference = 490 nm) were measured with a Dynex MRX II microplate reader (Dynatech Industries, Inc., McLean, VA, USA). The adjusted OD (AOD) of a serum-antigen reaction was the OD of the well coated with the test antigen minus the OD of the well coated with the control antigen. A sample was considered positive if the AOD at 1:100 was ≥0.200, the AOD at 1:400 was ≥0.200, and the sum of the AODs for the series of 4-fold dilutions (from 1:100 through 1:6,400) was ≥0.900. These criteria for positivity were based on the results of previous work with the test antigens and control antigens. The antibody titer of a positive sample was the reciprocal of the highest dilution of that sample for which the AOD was ≥0.200.