|Home | About | Journals | Submit | Contact Us | Français|
Identifying risk factors for zoonotic influenza transmission may aid public health officials in pandemic influenza planning.
We sought to evaluate rural Iowan agriculture workers exposed to poultry for previous evidence of avian influenza virus infection.
In 2004 we enrolled 803 rural adult Iowans in a 2-year prospective study of zoonotic influenza transmission. Their enrollment data and sera were compared with those from 66 adult controls enrolled at the University of Iowa in 2006 by proportional odds modeling.
The 803 participants were 58.8% male with a mean age of 55.6 yrs. Forty-eight percent reported previous poultry exposure. Sera were studied by microneutralization techniques for antibodies against avian H4, H5, H6, H7, and H9 viruses. Touching live birds was associated (OR = 1.2; 95% CI 1.02–1.8) with increased antibody titer against H5 virus. Similarly, participants who reported hunting wild birds had increased antibody titers against H7 virus (OR = 2.8; 95%CI = 1.2–6.5) and subjects who reported recent work with poultry had increased antibody titers against H6 (OR = 3.4; 95% CI 1.4–8.5) and H7 viruses (OR = 2.5, 95% CI = 1.1–5.7). There was no evidence of elevated antibody against avian H4 or H9 viruses.
These data suggest that hunting and exposure to poultry may be important risk factors for avian influenza virus infection among rural US populations. Agriculture workers should be included in influenza pandemic plans.
Studies of avian influenza virus transmission among the poultry-exposed have been technically difficult to conduct due to the poor performance and complexity of serological assays [1–3]. Despite other epidemiological data suggesting that subclinical or mild disease is more common than detected , serologic studies of humans exposed to avian influenza diseased poultry have often been negative [5–7]. However, a limited number of serological studies demonstrate that infections do occur. Retrospective seroprevalence studies among Hong Kong bird market workers in 1997 and 1998 showed that 10% had evidence of H5N1 infection . In addition, following the 2003 Netherlands outbreak, 49% of 508 poultry cullers, as well as 64% of 63 persons exposed to H7N7 infected humans, had serological evidence of H7N7 infection following the 2003 Netherlands poultry outbreak . A recent serological study of US duck hunters and wildlife biologists exposed to ducks and geese identified several subjects with elevated antibody titers against H11 viruses . A controlled, 2002 cross-sectional study of US poultry-exposed veterinarians revealed serological evidence of previous infections with avian H5, H6, and H7 viruses . Puzelli found evidence of low pathogenic avian influenza infection among 3.8% of Italian poultry workers in 2003 . Considering the recently emergent highly-pathogenic H5N1 viruses, the exposure most commonly implicated has been free-ranging poultry and small poultry flocks . In this study we sought to examine evidence for avian influenza virus transmission among poultry workers in Iowa, the leading US egg producing state.
Per our recent report  the study population consisted of 803 rural adults living in 29 counties in Iowa during 2004 selected from the 89,658-person Agricultural Health Study (AHS) cohort  for their nonimmunocompromised health status and their likely exposure or nonexposure to swine and poultry. The study was approved by the University of Iowa’s institutional review board. After informed consent was gained, participants completed a questionnaire and permitted serum specimen collection. Questionnaires and sera were again obtained at 12 and 24 months. At the enrollment and 12-month encounters, participants were given a first class US Postal Service-ready kit with detailed instructions to complete another questionnaire and self-collect gargle and nasal swab specimens within 96 hrs of symptom onset should they meet a case definition of influenza-like illness (fever ≥38°C and a cough or sore throat).
Data and sera from non-Agricultural Health Study controls from a concurrent cross-sectional study  were included for population comparisons at enrollment. These study subjects were generally healthy University of Iowa students, staff, and faculty who denied having swine or poultry exposures.
Gargle and swab specimens were transported to the University of Iowa via the US Postal Service in Micro Test M4RT Viral Transport Media (Remel, Inc., Lenexa, KS) and preserved at −80°C. These specimens were studied with both culture in MDCK cells and R-Mix FreshCells™ (Diagnostic Hybrids, Inc., Athens, Ohio) and with molecular techniques.
Per our previous reports [9, 10, 13, 15, 16] serum samples were tested using the Centers for Disease Control and Prevention (CDC) HI assay protocol against 3 human influenza A viruses: A/New Caledonia/20/99 (H1N1), A/Nanchang/933/95 (H3N2), and A/Panama/2007/99 (H3N2). The human influenza virus strains were grown in fertilized eggs. Sera were pre-treated with receptor destroying enzyme and hemabsorbed with guinea pig erythrocytes. Titer results are reported as the reciprocal of the highest dilution of serum that inhibited virus-induced hemagglutination of a 0.65% (guinea pig) or 0.50% (turkey) solution of erythrocytes.
Avian influenza viruses and antisera were kindly provided Dr. Richard Webby of St. Jude Children’s Research Hospital, Memphis, Tennessee; Alexander Klimov from CDC; and Dennis Senne of the National Veterinary Services Laboratories, Ames, Iowa. Per our recent reports [9, 10], a microneutralization assay, adapted from that of Rowe  was used to detect antibodies to avian strains thought be representative of those recently circulating in the United States: A/Duck/Cz/1/56 (H4N8), A/Chucker/MN/14591–7/98 (H5N2), A/Turkey/MA/65 (H6N2), A/Turkey/VA/4529/02 (H7N2), and A/Turkey/MN/38391–6/95 (H9N2). Avian influenza virus strains were grown in fertilized eggs.
Since prevalence was expected to be low, sera were first screened at a dilution of 1:10. Positive specimens were then titered out in duplicate by examining 2-fold serial dilutions from 1:10 to 1:1280 in virus diluent [85.8% minimum essential medium (Invitrogen, Carlsbad, CA), 0.56% BSA, 25mM HEPES buffer (Invitrogen), 100 mg/l streptomycin (Invitrogen), and 100,000 units/l penicillin (Invitrogen]. Virus neutralization was performed by adding 100 TCID50 of virus to the sera. The Reed Muench method was used to determine the TCID50/100μL. MDCK cells in log phase growth were adjusted to 2.0 × 105 cells/mL with diluent. One hundred microliters of cells were added to each well and the plate was incubated at 37°C with 5% CO2 for 24 hours. Plates were washed twice with PBS, fixed with cold 80% acetone, and incubated at room temperature for 10 minutes. The ELISA endpoint titer was expressed as the reciprocal of the highest dilution of serum with optical density (OD) less than X, where X = [(average OD of virus control wells) + (average OD of cell control wells)]/2. Test cells with an OD > 2 times the cell control OD mean were considered positive for virus growth. The back titer was run in duplicate and was only accepted when both replicates had matching results.
These procedures have been reported previously . Briefly RNA was extracted from 140 μl of each nasal swab and gargle sample using a QIAamp viral RNA extraction kit (Qiagen Inc., Valencia, CA) and screened via a proprietary real-time RT-PCR (rRT-PCR) protocol developed and kindly provided by the CDC. The protocol was designed to first screen for influenza A, and then through separate reactions, to rapidly determine influenza HA subtype. Samples positive by rRT-PCR for influenza A were further studied with RT-PCR and cDNA sequencing for phylogenetic analyses to confirm their subtype and, in some cases, for further genotypic analyses.
Realizing that serological cross-reactivity may occur between avian and human viral strains of the same hemagglutinin types, as per our previous seroepidemiological studies [10, 13, 15, 16], we adjusted for this potential confounding in each of the risk factor analyses by including human serological results in the multivariable models. They were included in the final models when statistically significant.
We examined a number of potential risk factors for association with avian influenza virus infection outcomes: age, gender, influenza vaccination (human) history, meat processing work, years in poultry production, recent poultry exposure and exposure during follow-up, touching live poultry or game birds, hunting wild birds and hunt times/year, smoking tobacco in the last year, frequency of touching poultry, exposure to poultry vaccine, type of domestic bird exposure, use of personal protective equipment, number of birds on the farm, chronic medical conditions, medications, military service, and seropositivity for human influenza viruses.
The distribution and geometric mean titers of MN assay results from enrollment sera were first compared between the exposure groups. Next, MN results for each avian influenza virus were compared to potential risk factors using a proportional odd modeling approach  or, in case of very sparse data, an exact logistic modeling approach. Ninety-five percent confidence intervals were computed about odds ratios. Final multivariable models were designed using a saturated model and manual backwards elimination.
We used bivariate and multivariable logistic regression to examine risk factors for evidence of influenza virus infection in two ways. First, using the classical approach, we examined risk factor associations for any 4-fold rise in MN titer (enrollment to 12 months, 12 to 24 months, or enrollment to 24 months) against the avian influenza viruses in a binary logistic regression model. Next, we examined risk factors for any increase in MN titer (using the participants’ greatest increase in titers during the periods: enrollment to 12 months, 12 to 24 months, and enrollment to 24 months) to the avian viruses through examining the entire spectrum of HI titer increase (e.g. no increase, 2-fold rise, 4-fold rise, 6-fold rise and 8-fold rise) through proportional odds modeling .
Among the 3259 AHS subjects contacted by telephone or mailing, 1274 (39.1%) were considered eligible and were willing to participate. Among these, 803 (63.0%) attended enrollment sessions, granted informed consent, and were enrolled. Of the subjects that attended enrollment, 385 participants were classified as poultry-exposed (AHS poultry-exposed) and 418 as non-poultry exposed (AHS non-poultry exposed, Table 1). Their enrollment data were compared to 66 non-poultry exposed University of Iowa controls (university controls, Table 1). Demographically, university controls were younger and more likely to be female.
During the first 12 months of follow-up, 3 of the enrolled subjects died and 2 withdrew from participation. Among the remaining 798 subjects, 372 of the AHS poultry-exposed and 368 AHS non-poultry exposed participated in the scheduled 12-month and/or 24-month follow-up encounters. An additional 33 farmers, who missed the 12-month and/or 24-month follow-up sessions, completed and submitted the follow-up questionnaire via mail, which increased participation in at least one follow-up to 97%.
More than 50% of the participants reported receiving influenza vaccines during the 4 years before enrollment (Table 1). Relatively few participants ever worked in the meat processing industry and few were recent tobacco smokers. While many AHS poultry-exposed had lived for >10 years on a poultry farm, relatively few continued to have frequent contact with poultry.
The distribution of MN titers from enrollment sera against avian H4, H5, H6, H7, and H9 viruses demonstrated modest serological reactivity among the 2 AHS groups and less activity among the university controls (Table 2). No differences were observed in enrollment sera MN assay geometric mean titers assays against the avian viruses.
In multivariate proportional odds modeling, the ordinal variable frequency of contact with poultry (assigned score of 0 for “never”, 1 for “rarely”, 2 for “monthly”, 3 for “weekly”, and 4 for “every day”) was statistically associated with an elevated MN assay titer against avian H5 virus (Table 3). However, the magnitude of this odds ratio (OR = 1.2; 95% CI 1.02–15) was meager suggesting this finding might be explained by chance alone. Considering avian H6 virus, working with poultry from the year 2000 to present (OR = 3.4; 95% CI 1.4–8.5) and having a chronic medical condition (OR = 5.2; 95% CI 1.9–13.9) were both associated with elevated antibodies titers. Considering avian H7 virus, hunting wild birds (OR = 2.8; 95% CI 1.2–6.5) and working with poultry from the year 2000 to present (OR = 2.5; 95% CI 1.1–5.7) were both associated with elevated antibodies titers. Age was not important in each of the 3 avian models mentioned above. Elevated antibody against human H1N1 influenza virus (HI assays ≥ 1:40) was important to the avian H5 and H7 models. No important risk factors were identified through the H4 and H9 modeling.
As indicated in our previous report , during the 24 months of follow-up, 66 participants developed an influenza-like illness and submitted 74 sets of self-collected nasal and gargle swab specimens. Two of the study participants were culture positive for influenza B virus and 22 were RT-PCR and culture positive for influenza A virus. One isolate was a “triple reassortant” swine H1N1 virus (GenBank accession numbers DQ889682-DQ889689) and the remaining 21 influenza A isolates were very similar to those from circulating human H3N2 viruses. No avian viruses were detected among the influenza-like illness specimens.
Like the enrollment sera, the 12-month and 24-month follow-up sera, revealed no geometric mean titer difference between the AHS poultry-exposed and the AHS non-poultry exposed participants for the avian influenza viruses (Table 2). Considering the 740 participants who donated sera at least twice and examining each sera pair (enrollment to 12 months, 12 to 24 months, and enrollment to 24 months), there was sparse evidence of incident avian influenza virus infection. Among the subjects with available MN results, 6 out of 740 (0.8%) and 2 out of 737 (0.3%) experienced a ≥4-fold rise in antibodies against avian H5 and H9 viruses, respectively (Table 4). Modeling for risk factors for these sparse incident infections was unfruitful (data not shown).
In this report we document serological evidence that Iowans who self-reported hunting birds or recent poultry work had elevated antibodies against low pathogenic avian H5, H6, and H7 viruses representing strains recently circulating in the United States. These data are consistent with previous investigations identifying hunting  and poultry agriculture [1, 11, 20] as risk factors for avian influenza infections in man.
These data are unique in that we avoided using a cutpoint in titer (binary outcome) approach which we have previously shown severely limits analytical power . Instead we compared the entire distribution of antibody titers against exposures with the proportional odds model. We conjecture that had a proportional odds modeling method been used in other studies that used binary outcome methodology [6, 21–23], their results would have likely been very different. Our study is also unique in that we had a nonexposed control group. Without such a control group it is difficult to evaluate titer activity among the exposed.
One might ask “What do the findings mean?” While these data do not show the magnitude of risk (odds radios) that our study of US veterinarians who work with poultry demonstrated , these study data support the position that US hunters and poultry workers are at increased risk of recreational or occupational avian influenza virus infections. We posit here as we have detailed before [24, 25], that their increased risk merits special public health attention. They should be educated about their increased risk, encouraged to use personal protective equipment, and to seek medical attention whenever they develop an influenza-like-illness. They should also receive priority access to annual and pandemic influenza vaccines so that they do not facilitate the reassortment of novel strains of influenza virus [24, 25] and do not accelerate human or avian influenza epidemics .
Our cross-sectional data are limited in that we cannot discern if the increase in antibody reflects infection or simply antigen exposure. However, other reports do seem to shed light on these questions. Hayden and Croisier  considered similar findings among Italian poultry workers and concluded that the low prevalence of antibody and temporal association with avian influenza epizootics argued for human infection with avian viruses as an explanation. We agree and further argue that as vaccine-generated immunity to influenza viruses wanes over time and as inactivated avian virus immunization may require large doses of unadjuvanted antigen  to cause an increase in antibody, one might point out that positive serological findings are more likely to represent true infection with avian viruses and their replication in tissue. Regarding the question of clinical illness we can only speculate as relatively few prospective studies of humans exposed to ill birds have been conducted. However, the available data suggest that subclinical avian influenza virus infections may be more common than expected [1, 5, 8, 11]. Comprehensive, prospective studies of large numbers of poultry-exposed individuals and their contacts are required to better understand the spectrum of illnesses associated with clinical avian influenza virus infections.
Our study data have a number of other limitations. A number of potential biases may have influenced results: voluntary participation, self-reporting exposure data, a younger mean age of the university controls, potential mismatching between study and circulating viruses, possible cross-reacting antibodies against avian viruses, and passive surveillance for humans with acute avian influenza virus infections. However, as previously described  adjustments have been made to examine or reduce these limitations, and study findings are biologically plausible and consistent with previous reports.
In summary, these data suggest that US bird hunters and poultry workers are at increased risk of avian H5, H6, and H7 influenza virus infections. Efforts should be made to include these citizens in influenza pandemic preparedness plans.
Supported by a grant (R21 AI059214-01) from the National Institute of Allergy and Infectious Diseases (Dr. Gray).
We thank Mark G. LeBeck, Kendall Myers, M.S., Whitney S. Baker, M.P.H., M. Ghazi Kayali, M.P.H., and Kerry Leedom, D.V.M., M.P.H. for their technical contributions to this research while working at the University of Iowa’s Emerging Pathogens Laboratory; Darcy Roseblum, R.N. and Tammy Pearson, R.N. for their assistance in enrolling and following study subjects; Ellen M. Heywood, Daniel Scaffinger, and Patricia A. Gillette, M.P.H. of the University of Iowa’s Agricultural Health Study for their assistance in recruiting study participants; Mike Mueller of University of Iowa College of Public Health for his survey scanning and interpretation assistance; Drs. Jackie Katz and Alexander Klimov (Centers for Disease Control and Prevention, Atlanta, GA), Dr. Richard Webby (St Jude Children’s Research Hospital, Memphis, Tennessee) and Dennis Senne (Diagnostic Virology Laboratory at the U.S. Department of Agriculture’s National Veterinary Services Laboratories in Ames, Iowa) for their advice and assistance with viruses, antisera, and the adaptation of serologic procedures; Carolyn B. Bridges, M.D. (National Immunization Program), Lucy E. DesJardin, Ph.D. (University of Iowa Hygienic Laboratory), Claudine Samanic, M.S. (National Cancer Institute) Kelley J. Donham, D.V.M., Ph.D. (University of Iowa College of Public Health), and Eileen L. Thatcher, D.V.M., Ph.D. (Iowa State University College of Veterinary Medicine) for their early suggestions regarding this research.
Potential conflicts of interest: Dr. Gray has helped to conduct vaccine trials for GlaxoSmithKline Biologicals and has served as a Scientific Advisory Board member for CSL Biotherapies.