Seventy-nine swine farming participants were initially enrolled, including 20 farm owners (ages 28–59 years), 14 spouses of farm owners (34–57 years), 12 children of farm owners (7–21 years), 21 farm employees (19–43 years), 9 spouses of farm employees (21–39 years), and 3 veterinarians (29–54 years). In total, the farm participants included 44 males of median age 37 (range 13–59 years) and 35 females of median age 34 (range 7–57 years). Of these participants, two people moved out of the area during the study and did not participate further, and three people chose to withdraw from the study before their second blood samples were collected. The preseason samples from these five participants were not included in the analyses. Two participants (both with elevated titers to swine viruses) did not complete the questionnaire. Serologic data from these two participants were included in the comparison of preseason titers of farm participants versus controls but could not be used to assess variables associated with seropositivity to swine viruses. The control sera were from 54 males of median age 32 (range 2–58 years) and 60 females of median age 34 (range 4–54 years).

To interpret the HI titers as showing differences in exposure to swine versus human influenza viruses, we first had to demonstrate that no minimal serologic cross-reactivity in HI assays between the human and swine reference strains existed. We compared HI titers by using virus-specific sheep and ferret reference sera (Table 1) and clearly showed no serologic cross-reactivity in HI assays between the human H1N1, human H3N2, and swine H1N1 viruses. The HI titers to homologous viruses were 320–640, whereas titers to heterologous viruses (H1 vs. H3 or human H1 vs. swine H1) were only 5–10.

The preseason serum samples from 17/74 farm participants had HI titers >40 (titer range 40–160) against either A/NEB or Sw/IND; 15/17 were seropositive to both swine viruses. These participants included seven farm owners (range 41–55 years), seven family members of farm owners (range 7–54 years), a 33-year-old farm employee, a 38-year-old family member of a farm employee, and a 43-year-old veterinarian. In contrast, only 1/114 of the urban control serum samples (from a 41-year-old) had a positive HI titer against a swine virus (HI titer=40 against only A/NEB). The difference in the number of seropositive samples between the farm participant and urban control cohorts was statistically significant (p<0.001). Similarly, the GMTs of the preseason serum samples from the farm participants to both swine-lineage viruses (A/NEB and Sw/IND) were significantly higher (p<0.001) than the GMTs of the samples from the urban control participants (13.2 vs. 5.1 and 15.7 vs. 5.4, respectively). In contrast, the farm participants’ GMTs to the reference human H1 (A/JOH) and H3 (A/NAN) viruses were not significantly different from those of the urban control samples (Table 2).

Only three farm participants demonstrated fourfold rises in titer to either of the swine viruses. These rises were not associated with illness in either the human participants or the pigs on their farms.

Each of the variables on the questionnaire (Figure) was investigated for association with preseason sample seropositivity to Sw/IND at HI titers >40 or >80, A/NEB at HI titers >40 or >80, and either swine virus at HI titers >40 or >80. The variables associated with seropositivity to either swine virus at HI titers >40 or >80 and the statistical strength of those associations are shown in Table 3. (Results for seropositivity to each individual swine virus are not shown but were consistent with the summary statistics presented in Table 3.) Being a farm owner, being part of a farm family (a farm owner or a farm owner’s family member), living on a swine farm, and going into a swine barn >4 days/week were all associated with seropositivity to swine influenza viruses. Beyond these factors of pig contact, being >50 years of age (but not >36 years of age, the median age of the farm participants in the study) was associated with swine virus seropositivity; having received the swine flu vaccine in 1976–77 or having ever received any influenza virus vaccine was also associated with swine virus seropositivity. (All four persons who received the swine influenza vaccine also reported having received other influenza virus vaccines.)

Although zoonotic infections with swine influenza viruses have been documented previously (4,15), the results of the present study strongly support the hypothesis that people associated with swine production are infected with swine influenza viruses more regularly than the small number of zoonotic infections in the literature would suggest. Previous studies by Kluska et al. (18), Woods et al. (19), and Schnurrenberger et al. (20) in the 1960s suggested increased rates of infection among persons in contact with pigs or working with swine influenza viruses. In this study, we specifically associated factors related to a person’s degree of contact with pigs to seropositivity to swine viruses. The number of hours per day spent in the barn was not a factor of significance, suggesting that the overall frequency of pig contact is a more important consideration than the length of contact at any one time. This lack of significance is consistent with the fact that influenza virus infections in pigs occur sporadically, and pigs generally only shed virus for approximately 7 days after infection (21). During the course of this study, pigs on only one farm were reported to exhibit signs of influenza-like illness. Influenza viruses were not isolated from nasal swab samples collected from these pigs.

Two factors not directly related to swine contact were also statistically associated with seropositivity to swine viruses in our study. First, being >50 years of age was associated with swine virus seropositivity. In an earlier study, Schnurrenberger et al. (20) collected samples in 1966 from abattoir workers, pork producers, swine exhibitors at a state fair, and veterinarians; they also found an association between age and seropositivity to a classical H1N1 swine influenza virus. In that study, the major impact of age was apparent for people born before 1920, suggesting an effect from exposure to the swine-like 1918 pandemic influenza virus (22–24). We could not fully separate the effects of age and exposure over time to swine. All of the participants >50 years of age were farm owners or farm family members. However, several factors indicate that, although age may play a role in seropositivity to swine viruses, exposure to swine is a more dominant factor. Farmers and their family members were significantly more likely than employees and their family members to have elevated titers to swine viruses; farmers and their families were also more likely to have exposure to swine and to be exposed over a longer period of time. Specifically, 88% of the farm owners and their families lived on the swine farm, compared to 7% of the employees and their families. Furthermore, of the farm owners and their families who lived on the farm, 77% had lived there >11 years. Farm owners had significantly more years in swine production than their employees (Mantel-Haenszel chi-square test for trend, p<0.001). Among younger study participants (<50 years of age), 21% and 18% of the farm family members had titers of >40 and >80 to swine viruses, respectively, compared to only 7% and 4% for employees and their family members. Although these differences among the younger participants were not statistically significant (p=0.09 to 0.14), they are consistent with the pattern of elevated titers seen among those with a higher level of exposure to pigs. Finally, our control population was of the same overall age distribution as our farm participants, yet only a single 41-year-old person among these controls was seropositive to a swine virus.

A second factor unrelated to swine contact significantly associated with swine virus seropositivity was having received either the swine influenza virus human vaccine in 1976–77 or ever having received any human influenza virus vaccine. However, vaccination status alone most likely did not determine seropositivity to swine viruses among our farm participants. Vaccination was only associated with seropositivity at a titer >40, but not at >80. Although we do not have historical data for our urban control samples, we have no a priori reason to suspect that these people would have had substantially different vaccination histories. However, only 1/114 of these participants demonstrated a titer >40 to a swine virus. Likewise, the proportions of employees and their family members who received the swine influenza vaccine (4%) or other influenza vaccines (32%) were not significantly different from the proportions of farm family members who received the swine influenza vaccine (6%) or other influenza vaccine (22%), but farm family members were significantly more likely to have elevated titers to swine viruses. Regarding having received the 1976–77 swine influenza vaccine, antibodies produced against that vaccine would not likely be present at detectable levels 20 years later. However, studies have shown that vaccination with more recent human influenza A (H1N1) viruses can boost titers to swine-like viruses in those previously exposed (25). Therefore, the statistical association between seropositivity to swine viruses and vaccination likely reflects a vaccine-induced boosting of antibody titers in persons previously exposed to a swine influenza virus. Because of the overall low numbers of participants with elevated titers to swine viruses, we were not able to perform meaningful multivariate analysis of the data to definitively segregate the effect of vaccination history (or age) from other variables.

Because a relatively small number of zoonotic swine influenza virus infections have been documented by virus isolation, whether infections with swine influenza viruses are clinically different than infections with routine human influenza viruses remains unclear. Our data suggest that aggressively pursuing virus isolation when people involved in swine farming have influenza-like illnesses would be valuable. In this way, retrospective studies of the clinical appearance of a larger number of zoonotic swine influenza cases may be possible.

Our findings suggest a second issue. Pigs may serve as hosts for the adaptation of avian viruses to replication in mammalian species (26). In addition, pigs are clearly recognized as hosts in which genetic reassortment between human and avian viruses can produce novel strains of pandemic potential (7,11,12). While this concern has historically been thought to be most important in the “Asia epicenter” (12,27), avian H1N1 viruses have spread widely within the swine population of northern Europe since 1979 (7,28–31), avian H4N6 viruses were isolated from pigs in Canada in 1999 (32), and human/swine/avian reassortant H3N2 (33–35) and H1N2 (36,37) viruses have spread widely within the swine population of the United States since 1998. Given the apparent frequency with which swine farm workers in our study were exposed to influenza viruses from pigs, more closely monitoring such persons as potential sentinels for the emergence of novel influenza viruses from the swine populations of developed countries with extensive swine-raising industries may be prudent.