Occupational Risks during a Monkeypox Outbreak, Wisconsin, 2003

Veterinary staff were at high risk; standard veterinary infection-control guidelines should be followed.

D uring May-June 2003, an outbreak of monkeypox virus (MPXV) infections, initially detected in Wisconsin, occurred in the midwestern United States (1,2). These MPXV infections were the fi rst to be reported outside of Africa and involved a West African viral strain (1,3). African rodents imported from Ghana were implicated in virus introduction in the United States (2,(4)(5)(6)(7). The African rodents had been transported and housed with native prairie dogs that were subsequently distributed as household pets in Wisconsin (1). Veterinary and pet store staff are at risk for potentially serious occupationally related infections (8)(9)(10)(11)(12)(13)(14)(15)(16)(17)(18). Early links between MPXV infections and prairie dog exposures in veterinary facilities and pet stores (1) led us to investigate occupationally related exposures.
We conducted an outbreak investigation and a veterinary staff cohort study to quantify and characterize all cases that occurred during the 2003 Wisconsin MPXV outbreak, identify protective and risk factors for occupationally transmitted infections, and determine veterinary work practices amenable to infection-control guidelines. Because both investigations were urgent outbreak control measures, no institutional review board approval or written consent was required or obtained.

Outbreak Investigation
The Wisconsin outbreak case defi nition (online Appendix, available from www.cdc.gov/EID/content/13/8/1150app.htm) was similar to case defi nitions established by the Centers for Disease Control and Prevention (CDC) for human MPXV infection (19). Cases were classifi ed as confi rmed, probable, or suspected according to clinical, epidemiologic, and laboratory criteria. Case fi nding was done through electronic postings (email and website postings), faxes, and mass media. Active surveillance of persons in contact with infected persons or animals included self-recorded diaries of signs and symptoms for 21 days postexposure or daily telephone assessments by local health department personnel. Data were summarized at the Wisconsin Division of Public Health (WDPH).
Willing pet store employees were given a standardized questionnaire to assess prairie dog contact and were offered serologic testing. Affected animal distributors were interviewed about work roles and animal care.

Veterinary Staff Cohort Study
The eligible cohort was defi ned as all persons, regardless of work roles, employed at any Wisconsin veterinary facility where at least 1 outbreak-associated prairie dog was treated during May 13-27, 2003. Cohort members were defi ned as those facility employees who participated in the study. Cohort case-patients were defi ned as cohort members who had laboratory-confi rmed MPXV infections, regardless of the presence or absence of specifi c signs or symptoms. Tissue confi rmation required demonstration of MPXV by viral culture, PCR, immunohistochemistry, or electron microscopy. Although cases could not be serologically confi rmed by outbreak case defi nition criteria, cohort members with MPXV infections confi rmed by tissue or serologic testing were defi ned as cohort case-patients. Serologic confi rmation required the fi nding of elevated orthopox immunoglobulin M (IgM) titers in a specimen obtained within 56 days after rash onset or seroconversion in paired acute-and convalescent-phase specimens. The cohort study had no probable or suspected-case defi nitions and, hence, no probable or suspected cases.
Signs and symptoms surveyed were rash, fever, chills, sweats, headache, joint pain, or lymphadenopathy within 21 days of most recent exposure to an ill prairie dog. Cohort members with a history of vaccinia vaccination or unknown vaccination status and birth date before 1972 were defi ned as vaccinia-vaccinated.
A standardized questionnaire was used to determine exposure to prairie dogs, general work practices, demographic information, and medical history. Questions to assess contact with prairie dogs during the reception, initial examination, ongoing medical care, and discharge of the prairie dogs had possible answers of yes, no, unknown, or not applicable. Cohort members who did not work within 48 hours after the prairie dog's veterinary visit were excluded from the exposures analysis but included in the remainder of analyses. Questions about general work practices such as sanitizing, hand hygiene (handwashing or cleaning with alcohol gel), and animal bedding changing practices had possible answers of yes, no, unknown, or not applicable; or they used Likert-scale responses of always, usually, sometimes, rarely, never, or not applicable. WDPH or local health department personnel administered the confi dential questionnaire in person or by telephone. Data were entered into Microsoft Offi ce Access 2003 (Microsoft Corp., Redmond, WA, USA) and analyzed using Epi Info version 3.3 (CDC, Atlanta, GA, USA). Likertscale responses of always and usually were dichotomized from sometimes, rarely, and never. Responses of unknown or not applicable were excluded.
Willing participants provided acute-and convalescentphase serum specimens, which were tested for nonspecifi c orthopox virus IgM and IgG levels at the CDC poxvirus laboratory (20). Tissue testing was conducted as part of patients' clinical care.
Outbreak-associated prairie dogs treated in Wisconsin veterinary facilities were traced backward and forward. Information was obtained about their illnesses and treatments.
Signs and symptoms reported by >80% of case-patients were rash, headache, sweats, and fever. Those reported by 60%-70% of case-patients were chills, sore throat, cough, or lymphadenopathy. All other signs and symptoms were reported by <23% of case-patients. No statistically significant differences in signs and symptoms were reported between confi rmed and probable or suspected case-patients. Five (19%) patients were hospitalized; none died.
In terms of exposure settings, 12 (44%) cases, including 10 confi rmed, occurred in staff of veterinary facilities where ill prairie dogs had received care ( Figure 2). Other cases occurred in 6 members of households with prairie dogs, 4 pet store visitors, 2 pet store employees, 2 animal distributors, and 1 visitor to a household with prairie dogs. No known cases occurred in healthcare workers who treated patients or in laboratory workers who handled specimens.
Symptom diaries were completed by 258 persons, including 28 pet store employees and 7 veterinary staff. Lo-cal health department personnel monitored 243 other persons by telephone, including 77 veterinary staff. Among 501 persons followed up, 10 (2%) experienced illness; all 10 were veterinary staff.
Two pet stores had received outbreak-associated prairie dogs. Of 28 employees (26 from store A, 2 from store B), 19 (68%) completed the questionnaire: 11 had handled prairie dogs, 9 had fed prairie dogs, 2 had been bitten by a prairie dog, and 2 had been scratched by a prairie dog. One store A employee had a confi rmed case, and 1 store B employee had a suspected case; both had handled ill prairie dogs. The store A employee had a tissue diagnosis of MPXV confi rmed by viral culture and PCR and was positive for orthopox IgM and IgG antibodies in acute-and convalescentphase serum specimens. The store B employee had prior vaccinia vaccination; a convalescent-phase serum specimen was negative for orthopox IgM and positive for IgG. Approximately 2 months after the last prairie dog exposure, serum specimens were obtained from 12 noncase pet store employees, 4 of whom had handled ill prairie dogs. All 12 had negative orthopox IgM antibody results.
Two Wisconsin exotic animal distributors (distributors 2 and 3, a married couple) distributed outbreak-associated prairie dogs and housed animals in their home. Distributor 2 had a confi rmed case of MPXV infection, distributor 3 had a suspected case, and an immunocompromised household member who had no direct animal contact had a confi rmed case.

Veterinary Staff Cohort Study
Four veterinary facilities had treated 3 outbreak-associated prairie dogs. These facilities employed 81 (range 3-59) persons during the outbreak; 74 (91%) participated in the cohort study (cohort members), and 44 (54%) participated in the serosurvey. Table 1 summarizes demographic characteristics of cohort members, serosurvey participants, and cohort case-patients. At least 1 veterinarian from each facility was a cohort case-patient. Among 17 cohort case-patients, 8 (47%) had tissue and serologically confi rmed cases, 7 (41%) had serologic confi rmation only, and 2 (12%) had tissue confi rmation only. The 17 cohort casepatients included all 10 veterinary staff with confi rmed cases and 2 with probable or suspected cases (previously mentioned in the overall outbreak investigation). Five serologically confi rmed cohort case-patients did not meet the defi nition of an overall outbreak case-patient.
Fever, sweats, chills, rash, lymphadenopathy, and headache were each associated (p<0.001) with confi rmed MPXV infection (Table 2). Among cohort case-patients, 15 (88%) had multiple signs and symptoms and 2 (12%) had only 1 sign or symptom (headache and a nonvesicular, nonpustular rash of unknown onset date, respectively). Severe keratitis required corneal transplantation for 1 cohort casepatient, and a miscarriage occurred at 12 weeks of gestation. Two cohort case-patients did not experience a rash.
By using the number of cohort members from each facility as the denominator, we calculated veterinary facility attack rates as follows: facility 1, 25%; facility 2, 67%; facility 3, 7%; and facility 4, 25%. The attack rate among cohort members for all 4 facilities combined was 23%. All cohort case-patients had been in the veterinary facility within 48 hours of the prairie dog's visit. The only factor protective against MPXV infection ( Table 2) was never having handled an ill prairie dog (p = 0.004). Having a job involving direct animal care (e.g., veterinarian, veterinary technician, or veterinary assistant) was associated with having a confi rmed case (p = 0.002). Four types of exposures were associated with having a confi rmed case: participating in an initial (p = 0.004) or follow-up (p = 0.04) examination of an ill prairie dog, caring for an animal within 6 feet of the ill prairie dog (p = 0.03), and feeding the ill prairie dog (p = 0.002). Vaccinia vaccination status did not differ between those who performed at least 1 of these 4 high-risk activities and those who did not (p = 0.9). All cohort case-patients reported having practiced hand hygiene after examining or feeding the ill prairie dog. Gloves had been used by cohort case-patients during the following activities: 2 (40%) initial examination, 3 (60%) follow-up examination, and 3 (75%) feeding an ill prairie dog. No cohort case-patients reported having used surgical masks, goggles, or face shields during these high-risk activities. Four cohort case-patients had fed a prairie dog on 8 occasions: placed food in the cage without touching the prairie dog (1×), hand fed prairie dog by syringe (3×), placed food directly in the prairie dog's mouth (3×), or fed through gastric tube (1×). Although having spent >30 minutes handling the prairie dog approached signifi cance (p = 0.09), 7 (41%) cohort case-patients reported never having handled a prairie dog. All of these 7 cases resulting from indirect exposure occurred in employees of facility 4. Five of these 7 case-patients reported having been within 3 feet of prairie dog C, 1 reported having been in the same room as prairie dog C but not within 3 feet, and 1 reported not being in facility 4 while prairie dog C was there but being there within 48 hours of its death. Multivariate analysis was not possible because of the small number of cohort members with each type of exposure.
Using antihistamines during the prairie dog visit (p = 0.04) was associated with being a cohort case-patient (Table 2). Antihistamine use was considered a possible surrogate for hand-to-face contact because users of antihistamines generally have allergies or rhinorrhea and likely touch their eyes or nose frequently. No other personal medical history was associated with illness. Previous vaccinia vaccination (p = 0.95) was not protective against MPXV infection. Few cohort members reported immunosuppressive medication use (n = 3), immunosuppressive illness (n = 2), or being pregnant (n = 2).
No general work practice was a protective or risk factor for being a cohort case-patient. General work practices were not outbreak specifi c and were used to assess overall risk for communicable disease transmission. Several cohort members reported hand-to-mouth activities (eating, drinking, chewing gum, or applying lip products) in animal care areas (Table 3). Only 12% who cleaned ill animals' cages reported having used gloves during this task. Most (92%-93%) cohort members reported cleaning their hands before eating at work and after ill animal contact.
The 44 serosurvey participants included 9 of 10 patients with tissue-confi rmed cases and 35 of 64 persons without tissue-confi rmed cases (p = 0.04). Cohort members with direct animal care jobs were not more likely (p = 0.19) than those without such jobs to have participated in the serosurvey. MPXV infection was serologically confi rmed for 13 (65%) persons who provided paired serum specimens and 2 (10%) who provided only acute-phase serum specimens. No evidence of asymptomatic seroconversion was found. Among serosurvey participants, only feeding a prairie dog was statistically associated (p = 0.02) with having a confi rmed case of illness, and no personal medical history factors were associated with illness. A history of vaccinia vaccination was not protective against MPXV infection (p = 0.35). Nineteen (43%) serosurvey participants had been vaccinated. Of these, 6 (32%) had MPXV infection; 5 had multiple signs or symptoms; and 1 had only a nonpustular, nonvesicular rash. Four (67%) of the vaccinia-vaccinated serosurvey participants with confi rmed cases had serologic confi rmation only (no tissue confi rmation was attempted), and 1 (17%) had both tissue and serologic confi rmation of illness. Two serosurvey participants (A and B) with confi rmed cases and previous vaccinia vaccination had no acute elevation of IgM. Participant A had symptomatic illness confi rmed by IgG seroconversion and multiple high-risk exposures, including participation in 2 prairie dog examinations and having provided care to an animal within

Discussion
The 2003 outbreak of MPXV infections affected Wisconsin residents who had been exposed in multiple settings; however, 59% of cases occurred among occupationally exposed persons. Our cohort study demonstrates that veterinary staff were particularly at risk (23% attack rate). Pet store employees were at lower risk (7% attack rate). Infected prairie dogs were probably more ill and shedding more virus while in veterinary facilities than in pet stores, which would account for more observed infections among veterinary staff. Both Wisconsin distributors of ill prairie dogs became ill. The preponderance of occupationally acquired cases was unique to Wisconsin during this outbreak. Among other involved states, 1 veterinarian from Indiana had a suspected case, and cases occurred in 2 employees of distributor 1 (7,21,22).
Most outbreak cases (59%) and veterinary cohort cases (82%) were associated with exposure to prairie dog C. We found no intrinsic differences in the monkeypox infection of prairie dog C compared with that of prairie dogs A or B and no explanation for this association, although length and type of exposures in facility 4 were likely accountable. Facility 4 had many more employees than the other facilities, and prairie dog C was hospitalized there for a relatively prolonged period (3 days) and received extensive treatments there. In addition, the fact that 7 (54%) facility 4 cohort case-patients reported never having handled prairie dog C indicates that other transmission modes (e.g., fomites, aerosols) could have contributed to that facility's large number of cases. Nebulization treatments, which prairie dog C received ≈4× at facility 4, could have exposed employees to MPXV. Although nebulization was performed in an enclosed plastic chamber, the nature of the treatment would foster aerosolization, coughing, and possibly mobilization of respiratory secretions with which employees could have unknowingly come into contact. Although our Wisconsin investigation showed no defi nitive evidence of human-to-human transmission of MPXV, which concurs with other states' fi ndings during this multistate outbreak (6,7,21,23), such transmission remains a possibility. Because persons who had had no direct contact with ill animals became ill with MPXV infection, person-to-person transmission within veterinary facilities might have occurred. However, because of the lack of personal protective equipment use among the cohort members and the fi nding of MPXV in ill prairie dog's urine and feces (21), environmental exposure may well account for these cases.
The substantial amount of illness among veterinary staff underscores the importance of infection-control practices in veterinary settings. Cohort case-patients frequently did not use personal protective equipment during high-risk activities (e.g., examining or feeding ill prairie dogs). Furthermore, cohort members reported general work practices that foster hand-to-mouth activities in animal care areas. Few (12%) cohort members reported having used gloves when cleaning ill animals' cages, a task that can contaminate staff hands with animal dander, urine, and fecal matter. We cannot determine whether infection-control guidelines would have prevented MPXV infections among veterinary staff, but use of personal protective equipment might have limited viral transmission. The National Association of State Public Health Veterinarians recently released the Veterinary Standard Precautions Compendium (24), the fi rst guidelines to describe standard infection-control practices for veterinary facilities. Use of these guidelines should be encouraged.
In contrast with results from a previously published study (25), results of our cohort study do not support the conclusion that prior vaccinia vaccination protected against MPXV infection in this outbreak. Hammarlund et al. found that 3 (37%) of 8 vaccinia-vaccinated persons in this outbreak had asymptomatic MPXV infections and surmised that they had long-term immunity against MPXV infection (25). Our cohort study showed that previous vaccinia vaccination did not protect against MPXV infection; all previously vaccinated serosurvey participants with positive serologic results had at least 1 sign or symptom of MPXV infection. The more systematic inclusion and analysis of exposed persons within our cohort, compared with the cohort-series approach of Hammarlund et al., may account for the difference in this fi nding (26).
Our case defi nition for the cohort study differed from that of the overall outbreak investigation. Among veterinary staff, the 2 defi nitions resulted in different case numbers: the outbreak case defi nition resulted in 10 confi rmed and 2 probable or suspected cases; the cohort study case defi nition resulted in 17 confi rmed cases. Serologic results were not a confi rming criterion in the outbreak case definition because their results were not validated at the time and because serologic specimens were not systematically gathered throughout the multistate outbreak.
For unknown reasons, 2 cohort case-patients had no elevation of orthopox IgM. Although the IgM response might have been missed, this is unlikely given the timing of specimen collections. Also, previous vaccinia vaccination might have altered the immune response to the MPXV infection. It is also possible that participant B's past medical history might have affected the immune response to this infection.
Our study has several potential limitations. Although 91% of employees at the 4 affected veterinary facilities participated in the cohort study, only 54% participated in the serosurvey. Persons with tissue-confi rmed illness were more likely than persons without such illness to have participated in the serosurvey. These factors might have resulted in an underestimation of overall cases and limited the detection of asymptomatic seroconversion. Recall bias, which might have overestimated the association between prairie dog contact and illness, was likely limited by relatively brief intervals between exposures and data collection. Finally, statistical analysis beyond univariate analysis was limited because of the small number of cohort members involved in each of the high-risk prairie dog exposures. Our investigation and cohort study demonstrate that occupational exposure, especially among veterinary staff, was a critical factor during this outbreak. This outbreak highlights the importance of standard infection-control guidelines developed for veterinary settings and the need to encourage their use.