Protective measures and human antibody response during an avian influenza H7N3 outbreak in poultry in British Columbia, Canada

Methods

We sought voluntary participation from anyone involved in any aspect of outbreak control. Recruitment was by advertisements posted at the worker deployment site, published through the media and distributed by the poultry association in newsletters sent directly to farmers. The advertisements invited anyone involved in control of the outbreak to participate in a study being conducted by the BCCDC to test for possible human infections and provided information on how to enroll. To enable comparison of serologic results across risk categories, we accepted participation by CFIA workers, farmers and their family members with and without direct contact with poultry, infected poultry or other poultry workers since February 2004. A mobile trailer at worker locations served as the study clinic. Blood samples (7 mL each) and interview data were collected by trained nurses using a standard questionnaire. Epidemiologic data were collected to describe the study cohort, to evaluate compliance with recommended protective measures and to aid in the interpretation of serologic results; these data included baseline characteristics (age, sex, province of residence, occupation), history of possible exposure to avian influenza (poultry-related activities before and during the 2004 outbreak, direct contact with infected or non-infected birds or other workers, number of birds or farms visited), protective measures applied and consistency of their use (categorized as never, sometimes or always), and illness experience between February 2004 and the date of blood collection and within 48 hours of starting oseltamivir among those who received this drug. Direct contact was defined as handling poultry or poultry products or sharing the same confined airspace. Influenza-like illness was defined as fever and one or more of cough, rhinorrhea, sore throat, myalgia or headache.12 Univariate statistical comparisons were performed by χ² test. The Ethics Board of the University of British Columbia granted approval.

All sera were tested for antibody by microneutralization assay.24^,25 A subset of 16 sera was also tested by Western blot and by routine and modified hemagglutination inhibition.15 This subset consisted of sera from 10 participants randomly selected from the full list of participants and from an additional 6 participants who were known to have had unprotected exposure to infected birds. The latter group included the 2 participants from whom virus had been isolated and from whom convalescent serum was collected more than 21 days after onset of symptoms.

Microneutralization assay was performed as previously described24 using the H7N3 human isolate from British Columbia (A/Canada/444/04). Animal control serum against A/H7N3/Turkey/Minnesota/29206/83 was provided by Dr. J. Pasick, CFIA, Winnipeg. Neutralizing titer for positive control was 1:320 and for negative sera less than 1:8. Western blot was performed as previously described by Bastien and associates.26 Recombinant hemagglutinin protein was generated by infecting insect cells with recombinant baculovirus carrying the H7 gene from A/Canada/444/04. Routine hemagglutination inhibition assay with turkey erythrocytes and modified hemagglutination inhibition assay with horse erythrocytes were performed as previously described.15^,17^,24^,25

Results

Interpretation

The BC outbreak in 2004 was the largest poultry outbreak of avian influenza recorded in Canada and the first in this country due to an H7N3 subtype.33 The virus evolved rapidly from low pathogenicity to high pathogenicity on the first affected farm. This was also the first outbreak worldwide in which it was confirmed, through virus isolation, that H7N3 avian subtypes, both high and low pathogenicity, can cause human infection and illness.34^,35

Once introduced, avian influenza viruses flourish in the setting of highly susceptible and genetically monotonous populations of poultry housed by the tens of thousands in commercial barns. Accelerated evolution of the virus in such settings is thought to occur through abundant error-prone replication, facilitating, through the sheer number of mutating viruses, the chance emergence of variants with altered characteristics. Pandemics of the previous century are thought to have evolved not from highly pathogenic avian influenza, but from low-pathogenicity forms that adapted surreptitiously to humans either directly (in the pandemic of 1918) or indirectly through genetic reassortment (in the pandemics of 1957 and 1968).36^,37 In this context, all avian influenza viruses in domestic poultry may have presumed pandemic potential.36–38 Exposed farmers and cullers constitute a potential interface between novel avian influenza viruses and the communities to which the workers return. To prevent further spread between barns and to protect individual workers and their communities from a highly changeable virus, rigorous attention to biosecurity and to occupational and public health measures is important during all outbreaks of avian influenza in commercial poultry.

Participants in this survey were asked to describe any concerns they had about biosafety or personal risk while working on the outbreak; they most often cited concerns related to eye protection, including feathers and sawdust in the air, poor fit of goggles over regular glasses, frequent fogging and general interference with vision. Given that the only human infections in British Columbia followed direct contact with unprotected conjunctiva, these concerns should be addressed. Other investigators have also underscored the importance of eye protection and the particular ocular tropism of H7 subtypes.12–14^,39^,40 Human influenza viruses are thought to bond preferentially to α2,6 cellular receptors, which predominate in the upper respiratory tract.39^,40 Conversely, avian influenza viruses preferentially bond to α2,3 receptors, which are expressed with preponderance in the human eye and deep in the lung.39^,40 Avian influenza viruses that successfully replicate in the conjunctiva may gain access to the respiratory tract through the nasolacrimal duct. Given enough opportunities, adaptation to α2,6 receptors in the upper airway may occur, setting the stage for efficient human-to-human transmission.40 A combination of oseltamivir and PPE, including goggles, may be warranted as back-up precautions to guard against this possibility, since compliance with any one protective measure, notably PPE, appears to be unreliable.

Among participants in our survey, a greater percentage of those who had direct contact with infected birds than of those without such exposure reported red or watery eyes. These ocular symptoms were not accompanied by discharge. Those who reported exposure to infected birds also reported exposure to more birds overall, which would have increased opportunities for eye irritation. Most of those who reported eye symptoms received oseltamivir prophylaxis and would not have been considered at high risk of infection with avian influenza virus. Enhanced surveillance confirmed 2 cases of H7N3 conjunctivitis in unprotected workers with symptom onset and virus isolation several days after exposure. These findings, delayed by a compatible incubation period, are unlikely to have been the result of irritation, trauma or contamination. Lack of serologic response in these confirmed cases may raise doubts about the sensitivity of our assay. Microneutralization is the definitive method for antibody detection.24 It is possible that mild infections induced by H7 viruses, especially those involving the immunoprivileged eye, may not induce a strong systemic antibody response.8–10 In the Netherlands, routine hemagglutination inhibition failed to detect antibody to H7N7, but modified hemagglutination inhibition identified an unprecedented number of subclinical infections. These, however, were not confirmed by microneutralization assay.13^,15–17 We performed multiple retrospective assays, including modified hemagglutination inhibition, without further yield. Given the uncertainties in serologic assays for avian influenza, caution should be applied in the interpretation of serologic results. Definitive evidence for human infection is the real-time detection and isolation of virus, particularly from sites where symptoms are manifest.

Management of the BC outbreak benefited from prior experience in the Netherlands. Compliance with vaccination and prophylaxis was comparable, but reported compliance with PPE was generally higher among BC participants than among those surveyed in the Netherlands.13 In British Columbia, vaccine and antiviral drugs were supplied free of charge to both farmers and cullers hired by the CFIA. Since 2004, influenza vaccine has been provided free annually to residents of British Columbia who work with live poultry and/or swine. During the BC outbreak, the CFIA provided PPE to its workers, but a mechanism for providing PPE to farmers was not established. Our survey results may reflect the resulting differential in the use of barrier precautions. Reduced use of barrier precautions by farmers in particular was also noted during the outbreak in the Netherlands.13 Surveys conducted in both British Columbia and the Netherlands were voluntary, were based on self-reporting and recall, and included only a limited number of those involved in controlling the outbreak. Participants were engaged in multiple outbreak-related activities, so it was not possible to exclusively distinguish them on that basis. Unlike vaccine or antiviral use, PPE must be repeatedly donned and doffed, and compliance associated with each activity might be more difficult to recall or report. Given the extent of the outbreak and the number of short-term workers involved, the true denominator of those potentially exposed was difficult to quantify; associated biases should be kept in mind. Nonetheless, lessons from both British Columbia and the Netherlands are clear: recommended protective measures should be provided and readily accessible to any potentially exposed person during future outbreaks of avian influenza. These precautions should be simple and feasible and should enable safe and unobstructed work; evaluation of compliance, effectiveness and impact should be undertaken. Given predictions of the further inexorable spread of the Eurasian H5N1 virus and its possible entry into North America, these lessons should be collectively addressed now.