Preventive Veterinary Medicine 117 (2014) 610–614

Contents lists available at ScienceDirect

Preventive Veterinary Medicine journal homepage: www.elsevier.com/locate/prevetmed

Short communication

Risk factors for H7 and H9 infection in commercial poultry farm workers in provinces within Pakistan Abdul Ahad a,b,∗ , Ronald N. Thornton d , Masood Rabbani a , Tahir Yaqub a , Muhammad Younus a , Khushi Muhammad a , Altaf Mahmood a , Muhammad Zubair Shabbir a , Mohammad Abul Kashem b , Md. Zohorul Islam b , Punum Mangtani e , Graham William Burgess c , Hein Min Tun f,1 , Md. Ahasanul Hoque b,∗∗ a

University of Veterinary and Animal Sciences, Lahore, Pakistan Chittagong Veterinary and Animal Sciences University, Chittagong, Bangladesh c James Cook University, Townsville, QLD, Australia d AsureQuality, New Zealand e London School of Hygiene and Tropical Medicine, University of London, UK f The University of Hong Kong, Hong Kong Special Administrative Region b

a r t i c l e

i n f o

Article history: Received 10 October 2013 Received in revised form 15 September 2014 Accepted 7 October 2014 Keywords: Avian influenza Serological-survey Poultry farm workers H7 H9

a b s t r a c t A cross sectional survey was conducted involving 354 farm poultry workers on 85 randomly selected commercial poultry farms in high density poultry farm areas in Pakistan to estimate the sero-prevalence of H5, H7 and H9 and to identify the potential risk factors for infection with the avian influenza virus. A haemagglutination inhibition test titre at 1:160 dilution was considered positive, based on WHO guidelines. The estimated sero-prevalence was 0% for H5, 21.2% for H7 and 47.8% for H9. Based on a generalized linear mixed model, the significant risk factors for H7 infection were area, type of farm and age of poultry worker. Risk of infection increased with the age of poultry workers. Compared with broiler farms, breeder farms presented a greater risk of infection (odds ratio [OR] = 3.8, 95% confidence interval [CI]: 1.4, 10.1). Compared with the combined Khyber Pakhtunkhwa Province and Federal area, North Punjab had higher observed biosecurity measures and presented a lesser risk of infection (OR = 0.3, 95% CI 0.1, 0.9). Biosecurity should therefore be enhanced (especially in breeder farms) to reduce the occupational risks in poultry farm workers and to decrease the risk of emergent human-adapted strains of AI H7 and H9 viruses. © 2014 Elsevier B.V. All rights reserved.

∗ Corresponding author at: Faculty of Veterinary Medicine, Chittagong Veterinary and Animal Sciences University, Zakir Hossain Road, Khulshi, Chittagong 4225, Bangladesh. ∗∗ Corresponding author at: Faculty of Veterinary Medicine, Chittagong Veterinary and Animal Sciences University, Zakir Hossain Road, Khulshi, Chittagong 4225, Bangladesh. Tel.: +880 31 659093x105; fax: +880 31 659620. E-mail addresses: [email protected] (A. Ahad), [email protected] (M.A. Hoque). 1 Present address: Department of Animal Science, University of Manitoba, Winnipeg, MB, Canada. http://dx.doi.org/10.1016/j.prevetmed.2014.10.007 0167-5877/© 2014 Elsevier B.V. All rights reserved.

A. Ahad et al. / Preventive Veterinary Medicine 117 (2014) 610–614

1. Introduction Like many South East Asian nations, Pakistan has frequently experienced highly pathogenic avian influenza (HPAI) H5N1 outbreaks in commercial birds since 1998 (Naeem and Siddique, 2006). During 2003–2013 a total of 628 laboratory-confirmed influenza A H5N1 virus infections occurred, with around 60% human case-fatality. Most cases occurred in children and young adults who were exposed to live infected poultry (Shinde et al., 2011). Several studies have revealed that poultry workers are at high risk of infection with avian influenza viruses (AIVs), due to their frequent exposure to chickens (Wang et al., 2009; Zhou et al., 2009). HPAI H7N3 has caused outbreaks affecting poultry production across Pakistan since 1994–1995 (Naeem and Hussain, 1995) and is considered endemic in this country. HPAI H7 has also caused outbreaks in poultry in Australia (Westbury, 2003). A newly emerged avian origin H7N9 caused human cases and fatality in China (Bai et al., 2013). It is therefore relevant and timely to conduct seroepidemiological investigation in poultry farm workers in Pakistan. Low pathogenic H9N2 virus has been isolated in poultry in the northern Pakistan since 1998 (Naeem et al., 1999; Iqbal et al., 2009). Antibodies against H9N2 have also been detected in a variety of wild bird species in Pakistan (Khwaja et al., 2005). This virus was detected in humans, and poultry in live-bird markets in Bangladesh during 2008–2011 and H9N2 viruses were circulating around the year in poultry of live birds, whereas H5N1viruses were co-isolated with subtype H9N2 primarily during the winter months (Shanmuganatham et al., 2013). Transmission of H9N2 from poultry to humans has been evidenced in Hong Kong (Butt et al., 2005). A highly variable infection rate with H9 virus, with a variable titre of HI cut-off (1:10–1:160) has been reported in people from various countries: around 87% of poultry farm workers in Iran (Hadipour, 2010); 2.3–4.5% in poultry workers in China (Wang et al., 2009; Huang et al., 2013); 6.7% in people exposed to domestic and wild birds in the USA (Kayali et al., 2008). It is reasonable to assume therefore that poultry farm workers in Pakistan might also be infected. The present study was therefore conducted to estimate the sero-prevalence of H5, H7 and H9 and to identify risk factors associated with H7 and H9 infection in farm poultry workers in Pakistan. 2. Materials and methods A cross sectional study was conducted involving farm workers from broiler, layer and breeder chicken farms during 2010–2011 in regions of Pakistan with a high density of poultry farms: Khyber Pakhtunkhwa Province (KPK), the Federal Area (FA), North, Central and South Punjab (NP, CP and SP). A total of 96 poultry farms (N = 12,105) were required for the study assuming 50% expected farm prevalence of AI infection (as no published farm level AI prevalence), 95% confidence interval, ±10% precision and 1% design effect. Farms were selected at random but

611

included in the study only if all of their workers were willing to participate. A research permit was obtained from the ethics committee of the University of Veterinary and Animal Sciences, Lahore, Pakistan (UVAS). Participating farm workers were interviewed face-to-face using a questionnaire designed to collect information about putative risk factors for infection with the AIVs. Only six well recognized risk factors were explored. Factors investigated were geographical area (KPK/FA/NP/CP/SP), farm type (broiler/layer/breeder), age of worker, exposure time (up to 1 year/1.1–2.5/2.5+) and farm management system (open/controlled rearing). The participating farm workers were all male. Their responsibilities included feeding poultry, handling healthy, sick, and dying chickens, collecting eggs and cleaning poultry stalls. None of the participants had been vaccinated with seasonal flu vaccine. Chickens of breeder farms were reported to be vaccinated with the killed vaccine of H5, H7 and H9 against avian influenza. Blood samples were collected from all workers and analyzed for H5, H7 and H9 AIV antibodies. Serum samples were decanted soon after collection and submitted on ice to the diagnostic laboratory at UVAS where they were centrifuged and then stored at −20 ◦ C pending analysis. The horse red blood cell (RBC) haemagglutination inhibition test assay as published by OIE (2012) was used to evaluate serum samples obtained from poultry workers for AI sero-subtypes. Initially, the human serum samples were treated with receptor destroying enzyme with a ratio of 1:3 at 37 ◦ C overnight to destroy any inhibitory substance if present in the serum samples. The treated serum samples were further diluted with phosphate buffer saline at a ratio of 2:3. Finally, the serum samples were diluted at 1:10 for performing HI test using the standard OIE protocol. Each run of HI testing included specific positive antisera and negative control serum, together with back titration run. The antibody titre was expressed as the highest serum dilution that prevents haemagglutination. A titre of 1:160 measured in HI testing was considered an AI antibody reactive serum sample to the tested subtype (H5/H7/H9) (WHO, 2007). The referral antigens and antisera used in HI assay in this study were purchased from Veterinary Laboratory Agency, Surrey, UK. They were inactivated H5N1 antigen and antiserum (A/Ck/Scot/59), inactivated H7N7 antigen and antiserum (A/tky/Eng/647/77) and H9N2 antigen and antiserum (A/Turk/Wisc/66). Data were entered into Microsoft Excel 2007 for cleaning and then imported into STATA version 9.2 (StataCorp, 4905, Lakeway Drive, College station, Texas 77845, USA) or R version 3.0.1 (R: A Language and Environment for Statistical Computing; http://www.R-project.org) for analysis. Maps were produced using QUANTUM-1.8.0. Risk factors for human infection were screened for inclusion in a multivariable model using univariable logistic regression analysis with antibody status recorded as a binary outcome and a conservative acceptance criterion of p < 0.30. A likelihood ratio (LR) chi-square test significance level of p < 0.05 was used to select variables for inclusion in a generalized linear mixed model. This took into account clustering at the farm level (Cnaan et al., 1997). A generalized linear mixed model was used, with farm as a random

612

A. Ahad et al. / Preventive Veterinary Medicine 117 (2014) 610–614

Fig. 1. Study areas in Pakistan.

variable due to the clustering effect of having interviewed respondents on each of the farms surveyed (Dohoo et al., 2003). The assumption of linearity for age of worker, as the only numerical variable, was evaluated using a lowess curve created by plotting age against the log odds of the outcome. Backward stepwise selection was used to derive a final model from the initial full model. Population-averaged regression coefficients were calculated using a standard formula as an alternative to using generalized estimating equations to calculate odds ratios (Dohoo et al., 2003). Twoway interaction terms (e.g. between area, type of farm, or area and farm management system) were then evaluated. Variables and their interactions were retained in the full model if their LR chi square test was significant at p ≤ 0.05. Confounding was evaluated by comparing regression coefficients after removing each of the main fixed effects in turn. Prevalence of infection in humans was calculated as for a cluster sample of farms by province.

3. Results A total of 108 farms was randomly recruited (N = 12,105) for the study but we were only able to sample farm poultry workers from 85 farms where all workers participated in the study. A total of 354 male workers were interviewed on 85 farms. All of the workers had been in close contact with birds. Of the 85 farms, 46 raised broilers, 21 raised layers and 18 raised breeders. The number of workers interviewed on each farm varied widely (median: 2; range 1–31). The surveyed farms were in the following Provinces and districts (Fig. 1): KPK (Haripur); FA (Islamabad); NP (Chakwal, Gujranwala, Rawalpindi and Sialkot); CP

(Lahore, Faisalabad, Sheikhupura, Toba Tek Singh); SP (Bahawalnagar, Multan). Details about the number of poultry farms, the number of farms recruited and sampled, and the range of flock sizes by area are presented in Table 1 The prevalence of H7 antibodies ranged from 17% to 58% across the provinces (p < 0.001) with an overall prevalence of 34%. For H9 the prevalence ranged from 9% to 76% across the provinces (p < 0.001) with an overall prevalence of 43%. None of the serum samples were HI positive for H5. The only significant risk factor for H9 infection was area (p < 0.001). Four risk factors were selected for inclusion in the full model for infection with H7 (Table 2). These were: area (p < 0.001), farm type (p < 0.001), age of poultry worker (p = 0.015) and type of management (p < 0.001). None of these variables were highly correlated. The relationship between age of poultry worker and the log odds of a positive outcome was not linear. Age was therefore modelled as a categorical variable (12–20 yr, 21–25 yr and 25+ yr). The main fixed effects entered into the full generalized linear mixed model were area, farm type, categorized age of poultry worker and type of management. Farm identity was modelled as a random effect. Areas 1, 4 and 5 had no entries for farm types of layer, broiler and layer, respectively, so they had been re-grouped as: (1) KPK and Federal; (2) North Punjab; (3) Central and South Punjab. The final model derived after stepwise backward selection of the five risk factors is show in Table 2. There were no significant interaction terms. The final model identified three main explanatory variables for infection with H7 virus: area, farm type and age of poultry worker. Risk varied between the aggregated areas. Compared with the

A. Ahad et al. / Preventive Veterinary Medicine 117 (2014) 610–614

613

Table 1 Farms by type, area and size in a study conducted in Pakistan in 2010–2011. No of farms

Flock range

Area

Farm type

No of farms recruited/No of farms sampled

Khyber Pakhtunkhwa Province

Broiler Layer Breeder

730 17 5

5/0 3/0 5/5

3000–60000 3000–100000 5000–250000

Federal area

Broiler Layer Breeder

185 20 69

3/1 4/4 3/3

5000–50000 3000–100000 5000–250000

North Punjab

Broiler Layer Breeder

4053 700 137

10/10 10/10 10/10

2000–60000 3000–100000 5000–250000

Central Punjab

Broiler Layer Breeder

3263 1351 91

30/24 7/7 3/0

30000–500000 30000–500000 100000–250000

South Punjab

Broiler Layer Breeder

1330 151 2

11/11 2/0 2/0

30000–500000 100000–250000 100000–150000

12,104

All

combined KPK Province and FA, North Punjab presented a lesser risk of infection, whereas Central and South Punjab presented a higher risk. Compared with broiler farms, there was only weak evidence that layer farms presented a lower risk of infection (OR = 0.79; 95% CI: 0.25, 2.46), whereas there was more evidence that breeder farms posed a greater risk of infection (OR = 3.78, 95% CI: 1.42, 10.08). Risk of infection increased with the age of poultry workers. Workers older than 25 years had twice the risk of H7 compared with the youngest workers (2.18, 95% CI: 1.09–4.36). 4. Discussion Three important AI subtypes (H5, H7 and H9) were investigated in poultry farm workers in Pakistan in this study as they caused either human infection or human mortality (by H5N1) across the world and have pandemic potential in future. Both H7 and H9 were detected with a high prevalence in serum samples of poultry workers in this study. These results suggested that individuals with occupational exposure to commercial poultry were at increased risk of infection with the H7 and H9 viruses.

108/85

2000–500000

A high seroprevalence (H7-21.2%, H9-47.8%, and both16.7%) in poultry workers was not unexpected since these virus types were known to have been circulating in different sectors of poultry production in Pakistan since 1998 (Naeem et al., 1999; Muneer, 2008). Comparable results have also been found in neighbouring countries of Pakistan such as Iran (Hadipour, 2010). The overall prevalence for H9 antibodies was higher than that of H7 in serum samples obtained from poultry workers in this study. This is consistent with observation of the previous study (Ahad et al., 2013). No positive H5 antibody titres were noted at the usually accepted cut point of 1:160 in this study. The documented seroprevalence of H5 was 0.8% (based on HI titre of 1:160) in poultry retailers and poultry wholesalers each in southern China (Wang et al., 2009), and 1% in poultry collecting facility workers in Indonesia (Setiawaty et al., 2010). Cross reactions between different AI subtypes are potential problems. This risk was mitigated by using a reliable test protocol published by WHO (2007), using a HI high cut-off titre of 1:160. The use of horse rather than chicken or turkey erythrocytes also increased the accuracy

Table 2 Final model identifying risk factors for H7 serological responses in poultry farm workers. Variable Fixed effects Area

Farm type

Age of poultry worker categorized Random effect Farm identification

Category

Odds ratiosSS a

KPK Province and Federal Area North Punjab Central Punjab and South Punjab Broiler Layer Breeder 12–20 21–25 26+

1 0.30 1.80 1 0.79 3.78 1 1.65 2.18

95% confidence interval

ORPA b

0.11, 0.87 0.62, 5.2

0.35 1.67

0.25, 2.46 1.42, 10.08

0.81 3.23

0.79, 3.41 1.09, 4.36

1.55 1.99

Likelihood ratio test p value p < 0.001

p = 0.004

p = 0.023

1–85

SS is subject-specific; PA is population averaged. a Calculate as exponentials of subject-specific regression coefficients. b Calculated from population-averaged regression coefficients bPA ≈ bSS /(1 + 0.346 × sH2 [DR1])1/2 (Dohoo et al., 2003).

p < 0.001

614

A. Ahad et al. / Preventive Veterinary Medicine 117 (2014) 610–614

of the test (Meijer et al., 2006; Jia et al., 2008; Kayali et al., 2008). Compared with the combined KPK Province and FA, North Punjab presented a lesser risk of infection. This observation is supported by Ahad et al. (2013) who reported less H7 seroprevalence in poultry workers in north Punjab compared with KPK and FA. The lower risk in north Punjab could be due to better biosecurity measures observed. Biosecurity and housing considerations may reduce either poultry infection (‘entry to’/‘spread within’ farm) but there are other potential factors that could be protective for workers even if poultry were infected (e.g. hygiene/masks) that have not been recorded. The increased risk in breeder farms in our study might have been because the breeder farms were much larger in terms of total population and space, than the layer or broiler farms. Also, on breeding farms the birds were reared on the floor in a dusty environment with aerosols of faecal dust as observed in this study. Layer farms usually used barn or cage systems in the studied farms. Birds in the cages are much less likely to produce infective aerosols. The broilers are present on the farm for relatively short periods and then they are sold. The fact that the animals are on the farm for a relatively short time will probably contribute to a lower level of exposure of these birds in the situations where animal stay on the farm for much longer in the case of both layers and breeders (Wang et al., 2009). Increasing age might have reflected duration of exposure, but the exposure time variable was not in itself significant in the univariable analysis. The rate of infection with H9N2 was high (47.8%) but area was the only risk factor for infection identified in this study. A more general survey of similar size in Pakistan found a similarly high rate of infection with H9N1 and identified as risk factors area, being a poultry worker, age and season (Rasheed et al., 2013). In conclusion the prevalence of H7 and H9 in poultry farm workers was fairly high. The study shows greater risk of H7 for poultry farm workers who worked at breeder farms and older age group of workers. Acknowledgements The fund was provided by Ministry of Education, Government of Pakistan under the scheme of “Award of 100 scholarships to Bangladeshi students under Prime Minister’s Directives” (Grant No. F1-4/2006-FS-II, No. F1-2/2009FS-II). We are also grateful to all staff of the University Diagnostic Laboratory, University of Veterinary and Animal Sciences, Lahore, Pakistan who helped us in analysis of the human blood serum. We would like to give special thanks to participating poultry farm owners and the workers to allow us taking blood samples and recording information. We would like to give special thanks to Navneet Dhand for reviewing the Manuscript during preparation. References Ahad, A., Rabbani, M., Yaqub, T., Younus, M., Mahmood, A., Shabbir, M.Z., Fatima, Z., Khalid, R.K., Rasheed, M., 2013. Sero-surveillance to H9 and H7 avian influenza virus among poultry workers in Punjab Province, Pakistan. Pak. Vet. J. 33, 107–112.

Bai, T., Zhou, J., Shu, Y., 2013. Serologic study for influenza A (H7N9) among high-risk groups in China. N. Engl. J. Med. 368, 24. Butt, K.M., Smith, G.J.D., Chen, H., Zhang, L.J., Leung, Y.H.C., Xu, K.M., Lim, W., Webster, R.G., Yuen, K.Y., Peiris, J.S.M., Guan, Y., 2005. Human infection with an avian H9N2 influenza A virus in Hong Kong in 2003. J. Clin. Microbiol. 43, 5760–5767. Cnaan, A., Laird, N.M., Slasor, P., 1997. Tutorial in biostatistics: using the General Linear Mixed Model to analyse unbalanced repeated measures and longitudinal data. Stat. Med. 16, 2349–2380. Dohoo, I., Martin, W., Stryhn, H., 2003. Veterinary Epidemiologic Research. AVC Inc., Prince Edward Island, pp. 706. Hadipour, M.M., 2010. H9N2 avian influenza virus antibody titers in human population in Fars Province, Iran. Braz. J. Poult. Sci. 12, 161–164. Huang, R., Wang, A.-R., Liu, Z.-H., Liang, W., Li, X.-X., Tang, Y.-J., Miao, Z.M., Chai, T.J., 2013. Seroprevalence of avian influenza H9N2 among poultry workers in Shandong Province, China. Eur. J. Clin. Microbiol. Infect. Dis. 32, 1347–1351. Iqbal, M., Yaqub, T., Reddy, K., McCauley, J.W., 2009. Novel genotypes of H9N2 influenza A viruses isolated from poultry in Pakistan containing NS genes similar to highly pathogenic H7N3 and H5N1 viruses. PLoS ONE 4, e5788. Jia, N., Wang, S.X., Liu, Y.X., Zhang, P.H., Zuo, S.Q., Lin-Zhan Dang, R.L., Ma, Y.H., Zhang, C., Zhang, L., Lu, S., Cao, W.C., 2008. Increased sensitivity for detecting avian influenza-specific antibodies by a modified hemagglutination inhibition assay using horse erythrocytes. J. Virol. Methods 153, 43–48. Kayali, G., Setterquist, S.F., Capuano, A.W., Myers, K.P., Gill, J.S., Gray, G.C., 2008. Testing human sera for antibodies against avian influenza viruses: horse RBC hemagglutination inhibition vs microneutralization assays. J. Clin. Virol. 43, 73–78. Khwaja, J.Z., Naeem, K., Ahmed, Z., Ahmed, S., 2005. Surveillance of avian influenza viruses in wild birds in areas adjacent to epicenter of an outbreak in federal capital territory of Pakistan. Int. J. Poult. Sci. 4, 39–43. Meijer, A., Bosman, A., van de Kamp, E.E., Wilbrink, B., Du Ry van BeestHolle, M., Koopmans, M., 2006. Measurement of antibodies to avian influenza virus A (H7N7) in humans by hemagglutination inhibition test. J. Virol. Methods 132, 113–120. Muneer, M.A., 2008. Status of avian influenza in Pakistan Project Director. In: National Program for the Control and Prevention of Avian Influenza. NARC, Islamabad. Naeem, K., Hussain, M., 1995. An outbreak of avian influenza in poultry in Pakistan. Vet. Rec. 137, 439. Naeem, K., Siddique, N., 2006. Use of strategic vaccination for the control of avian influenza in Pakistan. Dev. Biol. (Basel) 124, 145–150. Naeem, K., Ullah, A., Manvell, R.J., Alexander, D.J., 1999. Avian influenza A subtype H9N2 in poultry in Pakistan. Vet. Rec. 145, 560. OIE, 2012. Manual of Diagnostic Tests and Vaccines for Terrestrial Animals 2012. OIE, Paris, France. Rasheed, M., Rehmani, S.F., Iqbal, M., Ahmad, A., Akhtar, F., Akhtar, R., Bibi, T., Aslam, H.B., Afridi, S.Q., Mukhtar, N., 2013. Seropositivity to avian influenza virus subtype H9N2 among human population of selected districts of Punjab, Pakistan. J. Infect. Mol. Biol. 1, 32–34. Setiawaty, V., Sedyaningsih, E.R., Sudiro, T.M., Holle, M.R.D.R.B., Pangesti, K.N.A., Ibrahim, F., 2010. Antibody anti-H5N1 detection in poultry farmers and workers in poultry collection facilities in Indonesia. Med. J. Indones. 19, 124–129, 2007. Shanmuganatham, K., Feeroz, M.M., Jones-Engel, L., Smith, G.J.D., Fourment, M., Walker, D., McClenaghan, L., Alam, S.M.R., Hasan, M.K., Seiler, P., Franks, J., Danner, A., Barman, S., McKenzie, P., Krauss, S., Webby, R.J., Webster, R.G., 2013. Antigenic and molecular characterization of avian influenza A (H9N2) viruses, Bangladesh. Emerg. Infect. Dis. 19, 1393–1402. Shinde, V., Hanshaoworakul, W., Simmerman, J.M., Narueponjirakul, U., Sanasuttipun, W., Kaewchana, S., Areechokechai, D., Ungchusak, K., Fry, A.M., 2011. A comparison of clinical and epidemiological characteristics of fatal human infections with H5N1 and human influenza viruses in Thailand, 2004–2006. PLoS ONE 6, e14809. Wang, M., Fu, C.X., Zheng, B.J., 2009. Antibodies against H5 and H9 avian influenza among poultry workers in China. N. Engl. J. Med. 360, 2583–2584. Westbury, H.A., 2003. History of highly pathogenic avian influenza in Australia. Avian Dis. 47, 23–30. WHO, 2007. Recommendations and laboratory procedures for detection of avian influenza A (H5N1) virus in specimens from suspected human. Zhou, L., Liao, Q., Dong, L., Huai, Y., Bai, T., Xiang, N., Shu, Y., Liu, W., Wang, S., Qin, P., Wang, M., Xing, X., Lv, J., Chen, R.Y., Feng, Z., Yang, W., Uyeki, T.M., Yu, H., 2009. Risk factors for human illness with avian influenza A (H5N1) virus infection in China. J. Infect. Dis. 199, 1726–1734.

Risk factors for H7 and H9 infection in commercial poultry farm workers in provinces within Pakistan.

A cross sectional survey was conducted involving 354 farm poultry workers on 85 randomly selected commercial poultry farms in high density poultry far...
540KB Sizes 0 Downloads 5 Views