Outbreak of Influenza A (H3N2) Variant Virus Infection among Attendees of an Agricultural Fair, Pennsylvania, USA, 2011

Avoiding or limiting contact with swine at agricultural events may help prevent A(H3N2)v virus infections in such settings.

T riple reassortant swine infl uenza A viruses have circulated in swine herds in North America since 1998 (1)(2)(3). On the rare occasions that these viruses infect humans, they are called infl uenza A variant viruses (4). Viruses resulting from reassortment of swine infl uenza A (H3N2) virus and infl uenza A(H1N1)pdm09 (pH1N1) virus have emerged among US swine (4)(5)(6), and similar viruses have been identifi ed among swine outside the United States (7,8). During August 2011, the fi rst known human infection with infl uenza A (H3N2) variant [A(H3N2)v] virus containing the pH1N1 matrix (M) gene was reported in the United States (9). The pH1N1 M gene is implicated in increasing infl uenza transmissibility in animal models (10,11), and there was concern that this new A(H3N2)v virus could be effi ciently transmitted among humans. Because these viruses contain a novel combination of genes, little is known about the epidemiologic and clinical characteristics of human infections.
During August 2011, a child who had attended an agricultural fair in Pennsylvania (Fair A) attended by ≈70,000 persons became ill; the Centers for Disease Control and Prevention (CDC) confi rmed infection with A(H3N2)v virus in the child 6 days after Fair A closed and immediately began an investigation with the Pennsylvania Department of Health (PA DOH), the Allegheny County Health Department, and the Pennsylvania Department of Agriculture (PDA) to determine the extent of A(H3N2)v virus transmission and to identify illness risk factors among Fair A attendees.

Case Finding
We identifi ed cases through multiple methods. First, 1-2 weeks after Fair A closed, PDA investigators conducted open-ended interviews with swine exhibitors to determine whether they or their household members had be-come ill; exhibitors were identifi ed though a list provided by fair organizers. Ill exhibitors or their surrogates were interviewed by CDC/PA DOH to assess whether their illness met suspected case criteria. Second, members of a national children's agricultural club who participated in activities in the county where Fair A occurred were interviewed about illness occurring in their households after attendance at Fair A and/or swine exposure. Third, investigators went to another fair (Fair B), which occurred 3 weeks after Fair A »20 miles away in the same county, to enroll a convenience sample of Fair B attendees. Investigators enrolled Fair B attendees as they visited the health department booth, food service areas, exhibit halls, rides, and games. Fair B enrollees were subsequently asked during a phone interview whether they or their household members had become ill after attending Fair A. Fourth, media sources, including newspapers, television, and websites, encouraged community members to contact PA DOH if they became ill with infl uenza-like symptoms after attending Fair A. Fifth, clinicians were encouraged to obtain respiratory specimens from patients with suspected infl uenza virus infection who had recent swine or agricultural fair exposure and to refer such patients to PA DOH for interview. Sixth, state and county infl uenza surveillance supported prospective detection of persons with test results positive for infl uenza. Seventh, ill contacts of case-patients were interviewed by investigators.
If close contacts of case-patients became ill, they were offered testing for infl uenza virus infection regardless of whether they attended Fair A. Testing was done only if specimens could be collected <7 days after symptom onset; nasopharyngeal swab samples were used for testing.

Retrospective Cohort Study
We conducted a retrospective cohort study among a systematic random sample of members of a children's agricultural club (Club X) who resided in the county where Fair A occurred and who attended Fair A. Club members were children who conducted projects, such as raising livestock, for exhibition at Fair A. From a list of 994 Club X members, every fourth name was selected, yielding a cohort of 247 children. Using a standard questionnaire, we queried the parents of Club X members about illness occurring since Fair A; animal exposures at Fair A, home, work, and/or school; infl uenza vaccination history; and underlying medical conditions. If no adult was reached after 3 telephone attempts, the household members were considered nonrespondents. Interviews were conducted 3-4 weeks after Fair A concluded. Levels of swine exposure were categorized as 1) no exposure (attending Fair A but not visiting a swine exhibit); 2) indirect exposure (visiting a swine exhibit but not touching swine at Fair A); and 3) direct exposure (touching swine at Fair A). Risk for illness was estimated from the beginning of Fair A through 7 days after its conclusion.

Case Defi nitions
A suspected case-patient was a person with >1 sign/ symptom of infl uenza virus infection from >2 of 4 categories occurring <7 days after attending Fair A. Categories were: 1) fever (temperature >38C°) or subjective fever; 2) respiratory (cough, sore throat, or runny nose); 3) gastrointestinal (vomiting or diarrhea); and 4) constitutional (fatigue, muscle aches, or joint pain). At least 1 category was required to be fever or respiratory.
A broad clinical case defi nition was used because the clinical characteristics of A(H3N2)v virus infections were not well understood. Suspected case-patients were reclassifi ed as noncase-patients if respiratory specimens obtained <7 days after symptom onset had real-time reverse transcription PCR (rRT-PCR) or genomic sequencing results negative for A(H3N2)v virus or if convalescent-phase serology results were negative for A(H3N2)v virus infection.
A probable case-patient was a person <4 years of age (explained below) who met suspected case-patient criteria and who was seropositive for A(H3N2)v virus. A confi rmed case-patient was a person who had rRT-PCR and genomic sequencing results positive for A(H3N2)v virus infection; RNA from a respiratory specimen was used for genomic sequencing (12).

Respiratory Specimens
Respiratory specimens were obtained <7 days after symptom onset. We used rRT-PCR with the Human Infl uenza Virus Real-Time RT-PCR Diagnostic Panel (CDC, Atlanta, GA, USA) to test specimens. Specimens positive for infl uenza A were subtyped, and amplifi ed RNA from specimens with results consistent with A(H3N2)v virus infection (positive for InfA, pdmInfA, and H3 markers) or with indeterminate results underwent partial genome sequencing as described (12,13).

Serologic Testing
We asked suspected case-patients <13 years of age to participate in serologic testing. We chose this age group for testing because it is assumed that children have limited prior exposure to viruses similar to A(H3N2)v virus and therefore fewer cross-reactive antibodies. A convalescent-phase serum sample was obtained from participating suspected case-patients 3-5 weeks after illness onset. Serum samples were tested by microneutralization and hemagglutination inhibition (HI) for antibodies to variant strains A/Minnesota/11/2010 (H3N2) and A/Indiana/08/2011 (H3N2). The outbreak strain could not be used as an antigen because vi-able virus was not isolated from any of the case-patients. Microneutralization and HI tests were performed as described (14).
Preliminary testing of serum samples collected in 2007-2008 and in 2010 indicates that no children <4 years of age have antibodies to A(H3N2)v virus, but some children 4-13 years of age have cross-reactive antibodies (15). Therefore, test results for children <4 years of age were considered seronegative if HI titers to the variant strains were <10, indeterminate if titers were 10 to <40, and seropositive if titers were >40. Test results for children 4-13 years of age were considered seronegative if titers to the variant strains were <10 and indeterminate if titers were >10. Children with seronegative test results were reclassifi ed as noncase-patients, and those with seropositive results were reclassifi ed as probable case-patients. Children with indeterminate results retained suspected case-patient status.

PDA veterinarians routinely inspected all swine at Fair
A. In addition, veterinarians called Fair A swine exhibitors 1-2 weeks after Fair A closed to ask whether signs of illness developed in any swine during or shortly after Fair A.

Data Analysis
We entered data into a Microsoft Access 2010 database (Microsoft, Redmond, WA, USA) and analyzed it by using SAS version 9.3 (SAS Institute, Cary, NC, USA). Relative risks and exact 95% CIs, determined by using the Farrington-Manning method (16), are reported for selected exposures.

Ethical Considerations
This investigation was determined to be a response to a public health threat; in accordance with Federal human subjects' protection regulations, it was not considered to be human subjects research. Participation in interviews was voluntary; parents or guardians were interviewed for subjects <18 years of age. Parents or guardians consented to collection of respiratory and serum samples from subjects <18 years of age. Minors >7 years of age assented to collection of respiratory and serum samples.

Case Finding
We identifi ed 3 confi rmed, 4 probable, and 82 suspected cases, including the index case. No A(H3N2)v virus infections were identifi ed by state or county infl uenza surveillance or by clinicians among persons who did not attend Fair A. Of the confi rmed, probable, and suspected cases, 19 (21%) were identifi ed from Fair A swine exhibitor households, 29 (33%) were identifi ed from Club X households, 4 (4%) were identifi ed among Fair B attendees who also attended Fair A, 34 (38%) were identifi ed among persons who called PA DOH to report illness, 10 (11%) were identifi ed by another case-patient, and 2 (2%) were detected by state infl uenza surveillance; persons could be identifi ed by >1 method. The median age of all case-patients was 12 years (range 6 months-60 years); 39 (44%) were male ( Table 1). Dates of illness onset ranged from day 0-13, where day 0 was the opening day for Fair A (Figure). Most case-patients had illness onset within 4 days after either the swine show or swine auction, and no cases were identifi ed >6 days after the fair ended. Of 87 case-patients for whom medical history was known, 18 (21%) reported at least 1 underlying medical condition. Case-patients reported spending a median of 6 days (range 1-10 days) at Fair A, and 29 (33%) of 89 reported that their household owned swine. Of 87 case-patients for whom swine exposure was known, 80 (92%) reported direct or indirect swine exposure at Fair A.
The fi rst confi rmed case occurred in a previously healthy girl <4 years of age who touched swine at Fair A (case-patient 1); fever, cough, and rhinorrhea developed 4 days after she had contact with swine. The second confi rmed case occurred in a previously healthy girl in the 4to 13-year-old age group who exhibited swine at Fair A and had subjective fever and vomiting without respiratory symptoms (case-patient 2). The third confi rmed case occurred in a girl in the 4-to 13-year-old age group who had a preexisting medical condition (case-patient 3); the girl had contact with swine at Fair A and was hospitalized for respiratory distress. Of 3 confi rmed and 4 probable case-patients, 2 (29%) were male and all were <13 years of age. All confi rmed and probable case-patients attended Fair A on or after day 3 of the fair ( Table 2). All except case-patient 2 had fever and respiratory symptoms, and all recovered. All 7 confi rmed and probable case-patients visited the swine exhibit at Fair A, and 6 (86%) touched swine.
Illness developed in contacts of 4 case-patients (3 suspected and 1 confi rmed case-patient); the contacts had not attended Fair A <7 days before illness onset. Respiratory specimens were obtained from 3 of these 4 contacts, including the contact of the confi rmed case-patient, <7 days after illness onset; all tested negative for infl uenza by rRT-PCR, and 1 contact tested positive for rhinovirus. One person declined testing for infl uenza.

Laboratory Results
Respiratory specimens from case-patients 1 and 3 were positive for InfA, H3, and pdmInfA markers by rRT-PCR, and the specimen from case-patient 2 was InfA positive. Phylogenetic analysis of the 3 specimens showed that the genome contained the M gene from pH1N1 and 7 gene segments (hemagglutinin, neuraminidase, polymerase PB1, polymerase PB2, polymerase PA, nucleocapsid protein, nonstructural protein) similar to those from North American swine H3N2 subtype viruses and variant viruses that previously caused infection in humans (13).
Of 127 Club X members, 15 initially met the suspected case defi nition. Serologic testing was performed for 3 members: 1 was seronegative for A(H3N2)v virus and was reclassifi ed as a noncase-patient, and 2 had indeterminate results. Thus, 14 (11%) of the 127 Club X members were suspected case-patients. Respiratory specimens were not obtained from any club members.
The risk for suspected case status increased as exposure to swine increased from no exposure (referent) to indirect exposure (relative risk [RR] 2.1; 95% CI 0.2-53.4) to direct exposure (RR 4.4; 95% CI 0.8-116.3; p = 0.07 by Cochran-Armitage trend test); however, these differences were not statistically signifi cant (Table 4). Exhibiting swine was not associated with suspected case-patient status (RR 1.1; 95% CI 0.2-3.2). Suspected case-patient status was more common, but not statistically signifi cantly so, among persons whose families owned swine and among persons who fed or bathed swine or who cleaned the swine pen during the fair (Table 4).

Animal Investigation
The PDA veterinarian inspected >150 swine on day 3 of Fair A. All swine were healthy-appearing at inspection, although fever had developed in 1 pig, and that pig had already been removed from the fair. The febrile pig was housed with other swine at the fair for »24 hours before removal; it was not tested for infl uenza. Another pig died after a seizure on the last day of the fair; the cause of death was unknown.
No other illness in swine was reported to PDA during the fair. After the fair ended, PDA veterinarians attempted to call 135 households of swine exhibitors and reached 80 (59%). Of those 80 households, 8 (10%) reported that the swine they exhibited had signs of respiratory illness during or shortly after the fair. Ill swine had recovered or had been slaughtered before these interviews. No swine were tested for infl uenza.

Discussion
We describe an outbreak of respiratory illness, including 3 confi rmed infections with a variant infl uenza A virus not identifi ed in humans before August 2011. The outbreak occurred during a large agricultural fair, where humans and animals had opportunities for repeated and/or prolonged contact.
Outbreaks of variant infl uenza A viruses at agricultural events have been described, and these events may be key settings for zoonotic infl uenza transmission (17)(18)(19)(20)(21)(22)(23)(24). Triple reassortant H3N2 subtype viruses containing the pH1N1 M gene were fi rst identifi ed among swine in the United States in 2009 and have been detected among swine in multiple states, including Pennsylvania (25,26). During July-November 2011, 13 human infections with A(H3N2)v virus containing the pH1N1 M gene were identifi ed, and 5 were linked to agricultural fairs (24). Although the frequency of zoonotic infl uenza transmission at agricultural events is unknown, these events provide opportunities for swine infl uenza viruses to infect humans who have contact with infected swine. Human and swine infl uenza viruses may circulate at these events, creating opportunities for virus reassortment and the emergence of novel strains.
This investigation suggests that swine contact during Fair A was a risk factor for illness. Persons reporting direct contact with swine were more likely to report illness. Most case-patients became ill within 4 days after the swine show  or auction, suggesting a temporal relationship between human-swine contact and onset of human illness within <4 days. The epidemic curve, which suggests that case-patients were exposed to a common infectious source that was present for several days, is consistent with the hypothesis that infected swine were present for the duration of the fair. Prior investigations of human variant infl uenza virus infections have documented contact with infected swine (17,21), and cases have also occurred after contact with apparently healthy swine (4,23). No swine were tested for infl uenza during this investigation because swine at Fair A had either been slaughtered or had recovered before the fi rst human case was reported; however, triple reassortant H3N2 subtype viruses containing genetic material from pH1N1 have been detected in swine (4-6). Because of limited diagnostic testing, the extent and distribution of illness caused by A(H3N2)v virus among Fair A attendees are unknown; however, two thirds of children <4 years of age who were tested were seropositive for A(H3N2)v virus. This fi nding suggests that illness in at least some suspected case-patients can be attributed to A(H3N2)v virus infection. Suspected case-patients had illness onset dates and symptoms similar to those for probable and confi rmed case-patients. Symptoms were similar to those of seasonal infl uenza (27), but no seasonal infl uenza was circulating at the time in Pennsylvania.
Although we cannot rule out human-to-human transmission of A(H3N2)v virus at or after Fair A, enhanced surveillance after Fair A through the beginning of the typical infl uenza season detected no additional cases of A(H3N2) v virus infection in the community; this suggests that the virus did not exhibit effi cient or sustained human-to-human transmission. However, A(H3N2)v virus infection has occurred with limited human-to-human transmission among persons who reported no swine contact (4).
This investigation is subject to a number of limitations. First, interviews occurred when media sources began reporting "swine fl u" linked to Fair A. Persons who became ill after attending Fair A may therefore have been more likely to report swine exposure, thus biasing toward an association between illness and swine exposure. Second, testing for infl uenza was not conducted for most case-patients. The timing of the investigation allowed for collection of few respiratory specimens and only convalescent-phase rather than paired serum samples. Serologic testing was further limited to young children because cross-reactive antibodies in older age groups made interpretation of test results for convalescent-phase serum samples diffi cult. Because only convalescent serum samples were obtained and baseline serologic studies for A(H3N2)v were conducted in a differ-  Novel infl uenza A viruses will continue to emerge sporadically, but steps can be taken to reduce risks to human and animal health. Our fi ndings suggests that swine contact increases risk for A(H3N2)v virus infection; therefore, advising fair attendees, especially those at high risk for complications from infl uenza, to avoid or limit swine contact may help prevent A(H3N2)v virus infections at agricultural events (28). Agricultural club members and others with prolonged swine exposure should also be educated about the risk of zoonotic infl uenza transmission and actions they can take to reduce transmission risk, such as using personal protective equipment when they or their animals are ill (29). We found simultaneous illness in humans and swine at the fair; this fi nding supports those from prior studies showing that transmission of infl uenza virus occurs from swine to humans and vice versa (30)(31)(32). Preventing seasonal infl uenza in humans who have contact with swine (e.g., through annual infl uenza vaccination) can reduce reassortment opportunities in swine that become co-infected with swine and human infl uenza viruses. Prompt and thorough investigations should be conducted of all novel infl uenza virus outbreaks among humans and animals. Investigations can be more timely if patients with infl uenza-like symptoms inform clinicians of recent swine exposure and if clinicians consider variant infl uenza virus infection in patients with infl uenza-like symptoms and recent swine or agricultural fair exposure. Clinicians should work with public health offi cials to test respiratory specimens by rRT-PCR when they suspect variant infl uenza virus infection. This investigation was limited by the lack of infl uenza testing in swine. Representative and timely infl uenza surveillance among swine, especially during fair season in states where swine are present at agricultural events, would facilitate future investigations.
This outbreak of A(H3N2)v virus infections among persons attending an agricultural fair was likely associated with swine contact. We did not identify sustained human-to-human transmission of A(H3N2)v virus during this investigation; however, the identifi cation of ≈300 human A(H3N2)v virus infections in multiple states during 2011 and 2012 and the occurrence of limited human-tohuman transmission in small clusters (33,34) demonstrate that variant infl uenza viruses remain a public health concern for animals and humans who may infect each other at venues such as agricultural fairs. Collaboration among public health offi cials with responsibilities for human and animal health is critical to determining the transmissibility and pandemic potential of variant infl uenza viruses, such as A(H3N2)v virus, and the epidemiologic features of illnesses caused by them.