Our epidemiologic and clinical findings support human-to-swine transmission of the pH1N1 virus at this swine research farm. No molecular adaptive changes occurred in the virus following transmission from humans to swine, verifying that almost identical viruses infected both species. Most humans and swine had mild or asymptomatic infection, and the ILI case definition showed low sensitivity for pH1N1 infection in humans.
The chronology of events supports human-to-swine transmission: humans became symptomatic and were positive for pH1N1 by RT-PCR before the swine, and the epidemic curve of clinical disease and positive nasal swab specimens in the herd over the next 6 weeks was also consistent with the introduction of an influenza virus by humans. Transmission from humans likely occurred early in their illness, probably before the appearance of any symptoms in individuals with ILI, because pH1N1 viral shedding occurs before the onset of symptoms and peaks on the second day of illness [14
]. The chronology of swine infection observed is similar to that seen in an experimental inoculation of swine with pH1N1-like influenza—swine were positive for pH1N1 within 1–4 days, and clinical symptoms developed 4–5 days after exposure [15
]. In this outbreak, it is unlikely that the disease originated in swine, because they had no respiratory illness in the previous 8 months, and no new animals had entered the herd in the previous 9 months. Finally, molecular characterization studies showed that no molecular adaptive changes occurred and almost identical viruses infected humans and swine. There were only minor differences between human and swine pH1N1 isolates, which is likely, given the natural mutation rate of influenza virus both in vitro and in vivo [16
The duration of viral detection and clinical illness among swine is consistent with other Canadian reports [18
]. The high rate (94%) of asymptomatic infections that we observed in swine may be attributed to the healthiness of the herd or some cross-protection rendered by previous vaccination. The A/Swine/Iowa/110600/00 H1N1 isolate in the vaccine offers some cross-protection against classical H1N1 viruses and reassortant H1N1 viruses [19
In our study, one-third of humans with positive serological test results had few symptoms, and one-third were asymptomatic, so the ILI case definition was not sensitive for pH1N1 infection. Asymptomatic pH1N1 infections have been documented in studies from the United States [20
], the United Kingdom [21
], and France [22
]. Our findings also correlate with results of seasonal influenza studies, in which 30%–50% of humans have few or no symptoms [23
Although swine-to-human transmission of other strains of influenza has been reported [26
], there was no epidemiologic evidence of swine-to-human transmission of pH1N1 in this outbreak. There was also no evidence that passage of the pandemic virus in swine led to biologic changes in the virus. Our genomic sequencing studies showed that, 10 days after the virus was introduced into the herd, there was virtually no change in the virus at the molecular level.
This study has several limitations. First, only 2 humans had throat swab specimens that were positive for pH1N1 virus. The TAG RVP, used to screen throat swab specimens, has a sensitivity of 90.2% and a specificity of 100% for pH1N1, so it is possible that some positive samples were not detected [11
]. Throat swab specimens, although acceptable, are not optimal to detect pH1N1 by RT-PCR [30
]. We performed throat swabs to comply with a public health directive explicitly forbidding nasopharyngeal swabbing and aspirates in community settings [31
]. Although the use of throat swab specimens for testing for the presence of virus may have contributed to the lack of positive RT-PCR results, it is more likely that humans with ILI were no longer shedding virus 2 weeks after their illness, when the swab specimens were taken [32
]. Secondly, some individuals may have been incorrectly classified as positive for pH1N1 on the basis of cross-reactive serological test results. For example, ~5% of staff may have had cross-reactive antibodies to pH1N1 from previous swine flu infections, seasonal vaccination, or vaccination for swine flu in 1976 [33
]. To minimize bias from cross reactive serological test results, we compared acute and convalescent phase serological test results, compared serological analysis of pH1N1 and seasonal influenza strains, and statistically ruled out correlation between positive pH1N1 serological test results and age or prior receipt of the seasonal influenza vaccine. Finally, there may have been recall bias when responding to the survey (eg, in remembering symptoms from the previous month). Although this would not affect our observation that humans were the source of this outbreak, it would affect the sensitivity and specificity calculations of the case definitions.
In summary, findings from this outbreak on a swine research farm in Alberta, Canada, support human-to-swine transmission. In both species, the virus caused mild illness without sequelae.