The Veterinary Journal 198 (2013) 649–655

Contents lists available at ScienceDirect

The Veterinary Journal

journal homepage: www.elsevier .com/ locate/ tv j l

Cluster randomised trial of the impact of biosecurity measureson poultry health in backyard flocks

1090-0233/$ - see front matter � 2013 Elsevier Ltd. All rights reserved.http://dx.doi.org/10.1016/j.tvjl.2013.09.010

⇑ Corresponding author. Tel.: +27 71 8007292.E-mail address: aconan@hotmail.fr (A. Conan).

Anne Conan a,⇑, Flavie Luce Goutard a,b, Davun Holl c, Sok Ra a, Aurélia Ponsich d, Arnaud Tarantola a,San Sorn c, Sirenda Vong a

a Epidemiology and Public Health, Institut Pasteur du Cambodge, Réseau International des Instituts Pasteur, Phnom Penh, Cambodiab Centre de Coopération Internationale en Recherche Agronomique pour le Développement (CIRAD), Département ES, UR AGIRs, Campus International deBaillarguet, 34398 Montpellier Cedex 5, Francec Department of Animal Health and Production, National Veterinary Research Institute (NaVRI), Ministry of Agriculture, Forestry and Fisheries, Phnom Penh, Cambodiad Agronomes et Vétérinaires Sans Frontières, Boeng Tompun, Meanchey, Phnom Penh, Cambodia

a r t i c l e i n f o a b s t r a c t

Article history:Accepted 8 September 2013

Keywords:Backyard poultryBiosecurityCluster randomised trialAvian influenza virusNewcastle disease virus

In Cambodia, most poultry are raised in backyard flocks with a low level of biosecurity, which increasesthe risk of spread of infectious diseases. The aim of this study was to evaluate the effectiveness of a prac-tical biosecurity intervention based on affordable basic measures. A cluster randomised trial was con-ducted in 18 villages in Cambodia from November 2009 to February 2011. Generalised estimatingequations were used to test the association between the intervention and mortality rates in flocks ofchickens and ducks. Mortality rates in chicken flocks in intervention villages (mean 6.3%, range 3.5–13.8%, per month) were significantly higher than in control villages (mean 4.5%, range 2.0–9.7%, permonth; P < 0.01). Mortality rates in duck flocks in intervention villages (mean 4.1%, range 1.9–7.9%, permonth) were significantly higher than in control villages (mean 2.8%, range 0.6–8.0%, per month;P < 0.01). Despite good compliance among poultry owners, the biosecurity intervention implementedin this study was not associated with improvements in poultry mortality rates. These findings suggestthat basic biosecurity measures may not suffice to limit the spread of infectious diseases in backyardpoultry flocks in Cambodia.

� 2013 Elsevier Ltd. All rights reserved.

Introduction

In developing countries, backyard poultry rearing contributes tohousehold incomes and home food consumption (VSF, 2005; Hen-ning et al., 2006; Liao et al., 2009; Conan et al., 2012). Infectiousdiseases occur relatively frequently in backyard poultry flocks,since few or no biosecurity measures are usually implemented(Conan et al., 2012). Several major poultry diseases, such as New-castle disease (ND), are enzootic in these flocks in Southeast Asia(Cameron et al., 1999), causing high mortality in poultry, with a di-rect impact on income generation (FAO, 2002). There is also a riskfor food security and public health, for example from H5N1 highlypathogenic avian influenza (HPAI) virus (Lay et al., 2011; Thearyet al., 2012; Chaka et al., 2013).

In Cambodia, backyard poultry represent 71% of the duck indus-try and 94% of the chicken industry (VSF, 2005). Mortality in back-yard poultry flocks has been estimated at 30% (Ear, 2005).Mortality of poultry, particularly in the case of outbreaks of

H5N1 HPAI, has a direct impact on the livelihood of villagers in rur-al Cambodia (Ear and Burgos Caceres, 2009).

The World Organization for Animal Health (OIE), the Food andAgriculture Organisation (FAO) and the World Health Organization(WHO) have suggested measures to control the spread of H5N1HPAI virus and, by extension, other infectious agents, consistingof changes to husbandry practices among smallholders to improvethe biosecurity of backyard poultry flocks (Chitnis, 2012). However,evidence of the feasibility and efficacy of these measures has notbeen demonstrated in extensive systems (Conan et al., 2012).

In 2006, Agronome et Vétérinaires Sans Frontières (AVSF), withthe support of the FAO, carried out a sanitary management projectin three villages in Cambodia. The findings suggested that theadoption of good management practices and hygiene improvementat the village level, without using vaccination, dramatically re-duced mortality and morbidity of poultry (VSF-CICDA, 2007). How-ever, the study was not randomised with a control group and thesample was not large enough to conclude that there had been areal impact. To confirm the latter finding, the effectiveness of lowcost biosecurity interventions in backyard poultry flocks in Cambo-dia was evaluated in the present study.

Table 1Morbidity, mortality and fatality rates of poultry flocks, Cambodia, November 2009–December 2010.

Chickens (All ages) Chicks (<1 month) Ducks (all ages) Ducklings (<1 month)

Number of households followed 732 (98%) 454 (61%) 655 (87%) 115 (15%)Mean number of birds per household ± SD 24.1 ± 18.8 13.2 ± 9.7 16.3 ± 56.1 17.7 ± 54.3Number of households with sick animals (%) 2344 (24.5%) 821 (13.8%) 724 (7.4%) 179 (12.9%)Mean number of sick birds (0 excluded) ± SD 7.4 ± 8.2 7.0 ± 8.3 8.9 ± 19.7 11.5 ± 22.2Mean morbidity rate ± SD 6.4 ± 15.9% 5.2 ± 16.4% 14.1 ± 3.6% 4.7 ± 17.1%Number of households with dead animals (%) 2144 (22.4%) 757 (12.7%) 697 (9.6%) 176 (12.7%)Mean number of dead birds (0 excluded) ± SD 6.6 ± 7.5 6.5 ± 7.9 8.5 ± 19.4 11.4 ± 22.3Mean mortality rate ± SD 5.3 ± 14.5% 4.4 ± 14.7% 3.4 ± 13.6% 4.5 ± 16.6%Mean fatality rate ± SD 82.2 ± 31.8% 85.3 ± 30.2% 92.6 ± 22.5% 95.9 ± 17.8%

SD, standard deviation.

Fig. 1. Trends of (a) chicken and (b) duck flock mean mortality rates per month in

650 A. Conan et al. / The Veterinary Journal 198 (2013) 649–655

Materials and methods

A cluster randomised trial was conducted using two sets of nine villages each inthe Tram Kak and Samraong Districts in Takeo Province, south-eastern Cambodia.The protocol was approved by the Pasteur Institute in Paris (June 2009) and the Na-tional Veterinary Research Institute (NaVRI) (August 2009). A cascade training ap-proach, using village teams to relay key biosecurity measures, was implementedin the nine intervention villages from December 2009 (Month 1, M1) to February2010 (M3). The intervention consisted of an educational package, which focussedon implementing biosecurity measures relating to poultry health and husbandry,including cleaning yards and equipment, quarantine of newly introduced and sickanimals and burning dead birds (Conan et al., 2013). The compliance of the villagerswith the interventions was considered to be good, since 68% of the poultry ownersin intervention villages changed poultry raising practices as a result of the cascadetraining vs. 22% in control villages (P < 0.001) (Conan et al., 2013). It was estimatedthat at least 39 households should be visited per village to observe a statisticallysignificant reduction of 50% in poultry mortality (Hayes and Bennett, 1999). The vil-lage team interviewed the randomly selected household owners each month fromNovember 2009 (M0) to December 2010 (M13) using a standardised questionnaire.

In the same 39 households, blood samples were collected from ducks on threeoccasions: November 2009 (M0, before the training), August 2010 (M9) and Febru-ary 2011 (M15). Assuming an infection rate of 40% in the case of H5N1 HPAI or NDviruses in a duck flock, it was estimated that a sample of at least six birds would beneeded to detect at least one seropositive animal (Conan et al., 2010). If the numberof ducks was <6, blood samples were collected from all ducks. Blood was taken fromthe wing veins using a 4 mL syringe, using a new syringe and needle for each duck.The samples were placed on ice and sent daily to the NaVRI. Serological testing wasperformed on pooled samples from each household using haemagglutination inhi-bition tests for ND and H5 avian influenza viruses (OIE, 2012).

The main end-point of the trial was household mortality rates, considering oneduck flock and one chicken flock per household. It was assumed that mortalities inchicken and duck flocks were most likely to be due to infectious agents (Halimaet al., 2007; Mete et al., 2013). Mortality rate was defined as the number of deadpoultry divided by the number of poultry per flock during a defined period. If thepoultry owners reported deaths due to accidents or predators, the mortality rateof these flocks was defined as 0. To account for the likely positive intra-cluster cor-relation among poultry flocks within a village, a linear generalised estimated equa-tion (GEE) with an exchangeable matrix structure was used. The influence of theintervention (taken as a binary variable) on trends of mortality rates was tested.Co-factors included in the model were chicken and duck flocks (categorical vari-ables defined by quartile and median of number of animals), purchase of chickensor ducks (binary variable), trading with intermediaries (‘middlemen’), visit to mar-ket, presence of fighting cocks or geese and observed changes in poultry raisingpractices, presumably due to the training (intervention). The factors ‘trading withmiddlemen’ and ‘visit to market’ were treated first as frequency variables. If thesefactors were not significant, they were introduced as binary variables. Except forthe intervention variable, only variables with a P value <0.05 were retained in themodel. Collinearity between retained variables was calculated.

For serology results, four different case definitions were tested for a flock re-cently infected by ND or H5N1 HPAI viruses using incremental criteria from onedefinition to another. Each new criterion brought additional evidence of the pres-ence of one of the two viruses in the flock. Definition 1 relied on evidence of sero-conversion, i.e. antibody titre >1:16 on the second sample, while the precedingpooled sample tested negative (OIE, 2012). Three other definitions were establishedto increase specificity. Definition 2 included Definition 1 plus the farmer not buyingany poultry during the preceding 6 months. Definition 3 included Definition 2 plusthe report of at least one dead chicken or duck during the preceding 6 months. Def-inition 4 included Definition 3 plus the report of neurological signs in chickens orducks during the preceding 6 months. A Poisson GEE model was performed to com-pare each definition between intervention and control flocks at M9 and M15. Allanalyses were performed using Stata 11 (Statacorp).

Results

The 18 study villages comprised a total of 2343 households,with an average of 130 households per village (range 92–176households per village), including an average of 132 householdsper intervention village and 128 households per control village(Wilcoxon test; P = 0.5). The initial census survey accounted for1890 households, since 453 heads of households were not presentduring the visit. Of these 1890 households, 271 (14%) were ex-cluded from the analysis because they did not raise poultry. Duringthe period from November 2009 (M0) to December 2010 (M13),9803 (99.7%) monthly survey questionnaires were collected outof the expected 9828. During the follow-up period, 21 interventionand 26 control households were excluded and replaced by

intervention and control villages, Cambodia, December 2009 to December 2010.

Table 2Monthly chicken flock mortality rate using generalised estimated equation linear regression model, Cambodia, November 2009 (M0) to December 2010 (M13).

Month of monitoring Number ofobservations

Impact of intervention Impact of cofactors

Pa Slope (if P < 0.05) Significant cofactors Pa Slope

0 685 0.2 Chicken population ]16;30] 0.1 4Reference: ]0;16]b ]30;50] 0.1 3.5

>50 0.001 11.2Presence of geese 0.002 �5.6Purchase of ducks <0.001 �6.9Visit to the market (binary variable) 0.02 �3

1 679 0.01 3.5 Chicken population ]16;30] 0.5 �0.8Reference: ]0;16] ]30;50] 0.9 0.1

>50 0.006 4.3Presence of geese <0.001 �3.3Number of visits to the market 0.001 �0.1

2 689 0.9 Chicken population ]16;30] 0.005 4Reference: ]0;16] ]30;50] 0.1 2.2

>50 <0.001 4.33 682 0.05 2.6 Duck population ]0;8] 0.2 1.5

Reference: 0 >8 0.02 2.14 678 0.2 Purchase of ducks 0.004 24.6

Purchase of chickens 0.001 �7.05 677 0.0546 681 0.4 Presence of fighting cocks 0.001 �2.2

Number of visits of the middlemen at the farm 0.003 1.7Number of visits to the market <0.001 �4.6Changed the way of husbandry linked with the intervention 0.001 �2.7

7 680 0.6 Duck population ]0;8] 0.01 1.7Reference: 0 >8 0.009 2.3Presence of fighting cocks <0.001 �3.0Purchase of chickens 0.003 �2.7

8 683 0.3 Duck population ]0;8] 0.05 2.1Reference: 0 >8 0.04 2.3Purchase of ducks 0.002 �3.7

9 685 0.1 Presence of geese 0.04 �1.9Presence of fighting cocks 0.02 �2.2Purchase of chickens 0.01 �2.4

10 682 0.611 682 0.01 2.8 Presence of geese 0.002 �3.2

Purchase of chickens 0.005 �2.712 679 0.1 Presence of geese <0.001 �2.213 687 0.2 Chicken population ]16;30] 0.05 �2.3

Reference: ]0;16] ]30;50] 0.1 �2.1>50 0.5 �1.3

Duck population ]0;8] 0.004 1.9Reference: 0 >8 0.1 1.5

a Linear logistic generalised estimation equations.b ]x;y], included between x and y; x excluded and y included.

A. Conan et al. / The Veterinary Journal 198 (2013) 649–655 651

randomisation. Households were followed for an average of13 months (range 1–14 months). Ninety-eight per cent of house-holds raised chickens and 87% of households raised ducks(Table 1).

Mortality rates in chicken flocks in intervention villages (mean6.3%, range 3.5–13.8%, per month) were significantly higher than incontrol villages (mean 4.5%, range 2.0–9.7%, per month; Student’s ttest P < 0.01) (Fig. 1a). Mortality rates in duck flocks in interventionvillages (mean 4.1%, range 1.9–7.9%, per month) were significantlyhigher than in control villages (mean 2.8%, range 0.6–8.0%, permonth; Student’s t test P < 0.01) (Fig. 1b). In multivariate analyses,intervention was not correlated with mortality trends in ducks orchickens as a protective factor, whereas intervention was corre-lated with an increase in mortality rates in several individual studymonths for both ducks and chickens (Tables 2 and 3). Interventionvillages B and I had higher mortality rates in chickens and ducks,even though they had the highest level of adherence to the inter-vention (Conan et al., 2013) (Fig. 2a and b).

Factors frequently linked to chicken mortality rates were thenumber of ducks (risk factor) and the presence of geese (protectivefactor). The mortality rate in ducks was mainly influenced by thepresence of geese and the number of chickens in the flock (riskfactor), visiting the market (frequency or binary variable),

purchasing ducks (risk or protective factor, depending on themonth) and the presence of fighting cocks (protective factor).

During the three surveys, 3366 blood samples were collectedfrom ducks in 884 households (411 in November 2009, 256 in Au-gust 2010 and 217 in February 2011). Using pooled sample testingby household, 116/884 (13.1%) households had a positive test (at athreshold titre of 1:16) for antibodies against ND virus. Serumsamples from 50/884 (5.7%) households tested positive for anti-bodies against H5 HPAI virus. Intervention was not associated withany of the four case definitions for an infected flock or household(Table 4).

Discussion

This study examined the outcomes of intervention based on hy-giene practices to reduce poultry mortality that were implementedin nine villages in Cambodia. Initial analysis showed that the com-munity-based cascade training programme was a success; themeasures were well implemented in intervention villages andthe knowledge and practices of farmers improved during the year(Conan et al., 2013). The impact of the study was evaluated bycomparison of mortality rates of ducks and chickens, along with

Table 3Monthly duck flock mortality rate using generalised estimated equation linear regression model, Cambodia, November 2009 (M0) to December 2010 (M13).

Month ofmonitoring

Number ofobservations

Impact of intervention Impact of cofactors

Pa Slope (ifP < 0.05)

Significant cofactors Pa Slope

0 626 0.8 Duck population ]4;9] 0.03 2.2Reference: ]0;4]b ]9;19] <0.001 4.7

>19 <0.001 11.91 624 0.1 Duck population ]4;9] 0.2 2.3

Reference: ]0;4] ]9;19] 0.07 2.3>19 0.02 1.7

Presence of geese <0.001 �2.7Presence of fighting cocks 0.002 �2.9Purchase of ducks 0.01 �2.4

2 571 0.6 Duck population ]4;9] 0.4 0.9Reference: ]0;4] ]9;19] 0.04 2.7

>19 0.001 6.7Chicken population ]0;26] 0.003 3.2Reference: 0 >26 <0.001 3.3Presence of geese <0.001 �3.9Presence of fighting cocks 0.04 �2.3Visit to the market (binary variable) <0.001 2.1

3 558 0.4 Purchase of ducks 0.001 �4.44 519 0.8 Number of visits to the market <0.001 �0.1

Number of visits of the middlemen at the farm 0.001 �0.2Interaction between purchase of chickens and ducks,visit of middlemen and visit at the market

5 484 0.7 Chicken population ]0;26] 0.008 2.5Reference: 0 >26 0.01 1.9Presence of geese 0.003 �1.9Presence of fighting cocks 0.006 �1.8

6 475 0.7 Chicken population ]0;26] 0.02 2.2Reference: 0 >26 0.01 1Presence of geese <0.001 �1.4Purchase of ducks 0.01 �1Interactions between purchase of ducksand visit at the market

7 476 0.7 Duck population ]4;9] 0.2 1Reference: ]0;4] ]9;19] 0.007 1.6

>19 0.007 3.6Chicken population ]0;26] 0.02 2.9Reference: 0 >26 0.03 2.2Presence of geese 0.01 �1.3Visit to the market (binary variable) 0.004 �5.6Purchase of chickens 0.04 �1.6The middlemen comes at least once 0.006 1.9Interaction between visit at the market and visit of themiddlemen

8 469 0.005 2.2 Presence of geese 0.02 �0.99 476 0.03 1.9 Ducks population ]4;9] 0.1 0.7

Reference: ]0;4] ]9;19] 0.09 1.3>19 0.03 4.1

Presence of geese 0.04 �1.2Number of visits to the market 0.002 �1.1Interactions between purchase of ducks and visit at the market

10 471 0.009 5 Changed the way of husbandry linked with the intervention <0.001 �7.4Interaction between the intervention and the change ofhusbandry

11 472 0.001 2.7 Chicken population ]0;26] 0.001 3.4Reference: 0 >26 0.006 2.4Number of visits to the market 0.02 �1.3Changed the way of husbandry linked with the intervention 0.03 �2.1Interaction between the intervention and the change ofhusbandry

12 474 0.054 Chicken population ]0;26] 0.05 1Reference: 0 >26 0.02 0.9Presence of fighting cocks 0.009 �1Purchase of ducks 0.01 2.9Changed the way of husbandry linked with the intervention <0.001 �1.3

13 464 0.4 Chicken population ]0;26] 0.002 2.2Reference: 0 >26 0.1 1.2Number of visits of the middlemen at the farm 0.006 �0.7Interaction between visit of the middlemen and visit at themarket

a Linear logistic generalised estimation equations.b ]x;y], included between x and y; x excluded and y included.

652 A. Conan et al. / The Veterinary Journal 198 (2013) 649–655

Fig. 2. Average mortality rates in backyard (a) chicken and (b) duck flocks in intervention (red bars) and control (blue bars) villages, Cambodia, November 2009 to December 2010.

Table 4Comparison of number of Newcastle disease (ND) and H5 avian influenza positive households in August 2010 (month 9) and February 2011 (month 15).

Survey August 2010 February 2011

ND H5 ND H5

Intervention Control Pa Intervention Control Pa Intervention Control Pa Intervention Control Pa

Case definition of a ‘positive’ household: Infected by ND or H5 virusesNumber of observations 107 62 109 62 94 78 95 80Positive sample + Seroconversion 33 (31%) 22 (35%) 0.6 22 (20%) 4 (6%) 0.051 3 (3%) 3 (4%) 0.8 2 (2%) 2 (2%) 0.8Positive sample + Seroconversion + No poultry

bought during the last 6 months20 (19%) 19 (31%) 0.2 12 (11%) 1 (2%) 0.1 3 (3%) 2 (3%) 0.7 2 (2%) 2 (2%) 0.8

Positive sample + Seroconversion + No poultrybought during the last 6 months + Mortality

19 (18%) 11 (18%) 0.9 10 (9%) 1 (2%) 0.2 1 (2%) 2 (3%) 0.5 2 (2%) 2 (2%) 0.8

Positive sample + Seroconversion + No poultrybought during the last6 months + Mortality + Neurological signs

9 (8%) 1 (2%) 0.1 5 (5%) 1 (2%) 0.6 0 1 1 (1%) 1 (1%) 0.9

a Logistic generalised estimation equations.

A. Conan et al. / The Veterinary Journal 198 (2013) 649–655 653

serology for ND and H5 HPAI viruses, between nine interventionand nine control villages.

The primary analysis showed two peaks of mortality in ducksand chickens (December–February and July–October), but failedto show a decrease in mortality in the intervention households(Fig. 1). On the basis of interviews with farmers in three interven-tion villages, it was suspected that outbreaks of ND and/or fowlcholera (due to Pasteurella multocida) could be responsible for thefirst mortality peak. During the second peak of mortality, an out-

break of ND or H5N1 HPAI was suspected in one intervention vil-lage, since the clinical signs reported by farmers wereconvulsions and swollen heads. These two mortality peaks arelikely to be related to increased poultry movement or trading re-lated to national celebrations (Chinese New Year at M3 and PchumBen at M9) (Minh et al., 2009).

The multivariate analysis provided us with an indication of fac-tors at the farm level which may have influenced mortality rates.The presence of geese and fighting cocks were protective factors.

654 A. Conan et al. / The Veterinary Journal 198 (2013) 649–655

The role of fighting cocks, especially in H5N1 HPAI epidemics, re-mains unclear (Gilbert and Pfeiffer, 2012). Since the loss of fightingcocks or geese has higher economic consequences than the loss ofa chicken, it is assumed that farmers with fighting cocks or geesewould be more inclined to set up biosecurity measures if a diseaseappeared in the locality. Other risk or protective factors identifiedin the present study, such as market visits, middlemen visits or pur-chase of new animals, have been reported to be risk factors for infec-tious diseases in backyard flocks in previous studies (Njagi et al.,2010; Desvaux et al., 2011; Paul et al., 2011).

The peaks of chicken mortality in February or March in the vil-lages with suspected H5N1 HPAI may have been related to a con-firmed outbreak of H5N1 HPAI detected in January in the south ofthe Province and which is likely to have spread to the north of theProvince, where the study villages are located (OIE, 2010).

The findings of this study need to be interpreted in light of somemajor limitations. The duration of the study (i.e. follow up and inter-vention) may have been too short for an impact of intervention to bemeasured. On the other hand, for measures to be adopted readily,there is a need for benefits, especially economic benefits, to be expe-rienced quickly by farmers (Wiegers and Curry, 2009). In addition,our strategy was to develop an intent-to-treat approach, whichwould not be feasible with a longer duration of training, making itaffordable should any national scale-up programmes be planned(Wiegers and Curry, 2009; Conan et al., 2012; Sultana et al., 2012).

Biosecurity in backyard flocks is a necessary, but not sufficient,condition to mitigate the spread of infectious diseases and to de-crease the mortality rate of poultry (Abdelqader et al., 2007; Cristalliand Capua, 2007; Bhandari et al., 2011). If basic hygiene and biosecu-rity measures failed to effectively prevent dissemination of infectionin poultry flocks in rural Cambodia, other complementary solutionsshould be explored. Biosecurity-based intervention could also targettrade actors (e.g. middlemen, market sellers) (Fournie et al., 2012). Acascade training approach, whereby some villagers train other vil-lagers, with regular monitoring, can be used effectively to reachfarmers and different poultry trading actors (Conan et al., 2013).

In an ideal situation, biosecurity measures would include build-ing hen houses and/or vaccination, although it is recognised thatthese measures involve additional costs that are beyond the reachof most Cambodian farmers (Alders and Pym, 2009); the cost of ahen house is about US$25.001 in Cambodia, while the monthly aver-age income of a Cambodian household is US$75.00 (data from AVSFand Institut Pasteur, Cambodia). However, vaccination may need tobe promoted despite numerous issues related to provision and costsfor poor farmers (Domenech et al., 2009).

Conclusions

During a 15 month study period, peaks of mortality and sero-conversion to ND and H5N1 HPAI viruses in backyard poultryflocks in Cambodia were observed, suggesting active spread ofthese viruses. However, mortality of chickens or ducks was notprevented by low cost interventions designed to improve flock bio-security. It is recommended that further studies should be con-ducted to determine cost-effective interventions in rural settingsthat can reduce poultry mortality and increase income to farmersfrom the sale of poultry.

Conflict of interest statement

None of the authors of this paper has a financial or personalrelationship with other people or organisations that could inappro-priately influence or bias the content of the paper.

1 US$1.00 = UK£0.66 = €0.77 as at 12 July 2013.

Acknowledgements

The study was financially supported by the WHO, Geneva, Swit-zerland, the Institut Pasteur, Cambodia, a grant from the AgenceFrançaise pour le Développement (AFD) through Surveillance andInvestigation of Epidemic Situations in South-East Asia (SISEA), aregional project coordinated by the International Network of Pas-teur Institutes, by the DGAl-funded FRIA-08-009 REVASIA projectand by the US Department of Health and Human Services. Weare grateful to Dr Monica Naughtin for reviewing and checkingthe English language of the manuscript.

References

Abdelqader, A., Wollny, C.B., Gauly, M., 2007. Characterization of local chickenproduction systems and their potential under different levels of managementpractice in Jordan. Tropical Animal Health Production 39, 155–164.

Alders, R.G., Pym, R.A.E., 2009. Village poultry: Still important to millions, eightthousand years after domestication. World’s Poultry Science Journal 65, 181–190.

Bhandari, D.P., Wollen, T.S., Lohani, M.N., 2011. Preventing highly pathogenic avianinfluenza (HPAI) at the rural community level: A case study from Cambodia.Tropical Animal Health and Production 43, 1071–1073.

Cameron, A.R., Baldock, C., Chamnanpood, P., 1999. Epidemiology and dynamics ofmajor livestock diseases in Southeast Asia. In: Understanding Animal Health inSoutheast Asia. Management and Use of Animal Health Information. AustralianCenter for International Agricultural Research (ACIAR), Canberra, pp. 15–31.

Chaka, H., Goutard, F., Roger, F., Bisschop, S.P., Thompson, P.N., 2013. Household-level risk factors for Newcastle disease seropositivity and incidence ofNewcastle disease virus exposure in backyard chicken flocks in EasternShewa zone, Ethiopia. Preventive Veterinary Medicine 109, 312–320.

Chitnis, K., 2012. Risk communication and emerging infectious diseases. In: TheHandbook of Global Health Communication. Wiley-Blackwell, Oxford, UK, pp.08–425.

Conan, A., Holl, D., Goutard, F., Buchy, P., San, S., Vong, S., 2010. Clinical definition ofhighly pathogenic avian influenza (H5N1) outbreaks in Cambodian backyardflocks. In: Proceedings of the International Conference on Options for theControl of Influenza VII, Hong Kong, China, 3–7 September 2010.

Conan, A., Goutard, F.L., Sorn, S., Vong, S., 2012. Biosecurity measures for backyardpoultry in developing countries: A systematic review. BMC Veterinary Research8, 240.

Conan, A., Ponsich, A., Goutard, F.L., Khiev, R., Tarantola, A., Sorn, S., Vong, S., 2013. Acommunity-based education trial to improve backyard poultry biosecurity inrural Cambodia. Acta Tropica 125, 294–302.

Cristalli, A., Capua, I., 2007. Practical problems in controlling H5N1 highpathogenicity avian influenza at village level in Vietnam and introduction ofbiosecurity measures. Avian Diseases 51, 461–462.

Desvaux, S., Grosbois, V., Pham, T.T., Fenwick, S., Tollis, S., Pham, N.H., Tran, A.,Roger, F., 2011. Risk factors of highly pathogenic avian influenza H5N1occurrence at the village and farm levels in the Red River Delta region inVietnam. Transboundary and Emerging Diseases 58, 492–502.

Domenech, J., Dauphin, G., Rushton, J., McGrane, J., Lubroth, J., Tripodi, A., Gilbert, J.,Sims, L.D., 2009. Experiences with vaccination in countries endemically infectedwith highly pathogenic avian influenza: The Food and Agriculture Organizationperspective. Revue Scientifique et Technique 28, 293–305.

Ear, S., 2005. The Political Economy of Pro-Poor Livestock Policy in Cambodia. Pro-Poor Livestock Policy Initiative Working Paper Number 26. Food and AgricultureOrganization (FAO), Rome, Italy, 59 pp. http://www.fao.org/ag/againfo/programmes/en/pplpi/docarc/wp26.pdf (accessed 12 July 2013).

Ear, S., Burgos Caceres, S., 2009. Livelihoods and highly pathogenic avian influenzain Cambodia. World’s Poultry Science Journal 65, 633–640.

FAO (Food and Agriculture Organization), 2002. Improved Animal Health forPoverty Reduction and Sustainable Livelihoods. FAO Animal Production andHealth Paper Number 153, Rome, Italy, 44 pp. ftp://ftp.fao.org/docrep/fao/005/y3542e/y3542e.pdf (accessed 12 July 2013).

Fournie, G., Guitian, J., Desvaux, S., Mangtani, P., Ly, S., Cong, V.C., San, S., Dung, D.H.,Holl, D., Pfeiffer, D.U., et al., 2012. Identifying live bird markets with thepotential to act as reservoirs of avian influenza A (H5N1) virus: A survey innorthern Viet Nam and Cambodia. PLoS ONE 7, e37986.

Gilbert, M., Pfeiffer, D.U., 2012. Risk factor modelling of the spatio-temporalpatterns of highly pathogenic avian influenza (HPAIV) H5N1: A review. Spatialand Spatio-Temporal Epidemiology 3, 173–183.

Halima, H., Neser, F.W.C., Van Merle-Koster, E., De Kock, A., 2007. Village-basedindigenous chicken production system in north-west Ethiopia. Tropical AnimalHealth Production 39, 189–197.

Hayes, R.J., Bennett, S., 1999. Simple sample size calculation for cluster-randomizedtrials. International Journal of Epidemiology 28, 319–326.

Henning, J., Khin, A., Hla, T., Meers, J., 2006. Husbandry and trade of indigenouschickens in Myanmar – Results of a participatory rural appraisal in the Yangonand the Mandalay divisions. Tropical Animal Health and Production 38, 611–618.

A. Conan et al. / The Veterinary Journal 198 (2013) 649–655 655

Lay, K.S., Vuthy, Y., Song, P., Phol, K., Sarthou, J.L., 2011. Prevalence, numbers andantimicrobial susceptibilities of Salmonella serovars and Campylobacter spp. inretail poultry in Phnom Penh. Cambodia. Journal of Veterinary Medical Science73, 325–329.

Liao, Q.Y., Lam, W.W., Dang, V.T., Jiang, C.Q., Udomprasertgul, V., Fielding, R., 2009.What causes H5N1 avian influenza? Lay perceptions of H5N1 aetiology in SouthEast and East Asia. Journal of Public Health 31, 573–581.

Mete, A., Giannitti, F., Barr, B., Woods, L., Anderson, M., 2013. Causes of mortality inbackyardchickensinnorthernCalifornia:2007–2011.AvianDiseases57,311–315.

Minh, P.Q., Morris, R.S., Schauer, B., Stevenson, M., Benschop, J., Nam, H.V., Jackson,R., 2009. Spatio-temporal epidemiology of highly pathogenic avian influenzaoutbreaks in the two deltas of Vietnam during 2003–2007. PreventiveVeterinary Medicine 89, 16–24.

Njagi, L.W., Nyaga, P.N., Mbuthia, P.G., Bebora, L.C., Michieka, J.N., Minga, U.M.,2010. A retrospective study of factors associated with Newcastle diseaseoutbreaks in village indigenous chickens. Bulletin of Animal Health andProduction in Africa 58, 22–23.

OIE (Office International des Epizooties), 2010. Update on Highly Pathogenic AvianInfluenza in Animals (Type H5 and H7). Immediate Notification Report:Cambodia. 5 February 2010. OIE (World Organization for Animal Health)Reference 8913. OIE, Paris, France. http://web.oie.int/wahis/reports/en_imm_0000008913_20100205_133840.pdf (accessed 12 July 2010).

OIE (Office International des Epizooties), 2012. Manual of Diagnostic Tests andVaccines for Terrestrial Animals, Seventh Ed., vols. 1 and 2. OIE (WorldOrganization for Animal Health), Paris, France, 1404 pp. http://www.oie.int/

international-standard-setting/terrestrial-manual/access-online/ (accessed 12July 2013).

Paul, M., Wongnarkpet, S., Gasqui, P., Poolkhet, C., Thongratsakul, S., Ducrot, C.,Roger, F., 2011. Risk factors for highly pathogenic avian influenza (HPAI)H5N1 infection in backyard chicken farms, Thailand. Acta Tropica 118, 209–216.

Sultana, R., Rimi, N.A., Azad, S., Islam, M.S., Khan, M.S., Gurley, E.S., Nahar, N., Luby,S.P., 2012. Bangladeshi backyard poultry raisers’ perceptions and practicesrelated to zoonotic transmission of avian influenza. Journal of Infection inDeveloping Countries 6, 156–165.

Theary, R., San, S., Davun, H., Allal, L., Lu, H., 2012. New outbreaks of H5N1 highlypathogenic avian influenza in domestic poultry and wild birds in Cambodia in2011. Avian Diseases 56, 861–864.

VSF (Vétérinaires Sans Frontières), 2005. Review of the Poultry Production andAssessment of the Socio-Economic Impact of the Highly Pathogenic AvianInfluenza Epidemic in Cambodia. Final Report. FAO (Food and AgricultureOrganization) TCP/RAS/3010. Emergency Regional Support for Post AvianInfluenza Rehabilitation. FAO, Rome, Italy, 89 pp. http://www.apeiresearch.net/document_file/document_20070706103902-1.pdf (accessed12 July 2013).

VSF-CICDA, 2007. Listing of Cattle Diseases and Sanitary Management of Villages.Final Report. The Smallholders Livestock Production Project. FAO, EU,Cambodia.

Wiegers, E., Curry, J., 2009. Understanding smallholder’s decisions towards adoptingHPAI prevention control measures. HPAI Research Brief 16, 4.