Transmission dynamics of avian influenza viruses along poultry value chains in Bangladesh : a controlled field experiment

Background:
Multiple avian influenza viruses (AIVs), including H9N2 and H5N1 subtypes, are endemic in the Bangladeshi poultry population. Previous studies have shown that AIV prevalence in chickens increased more than tenfold along the value chain and was highest in live bird markets (LBMs). With the aim of identifying at which stage viral amplification takes place, a controlled field experiment was designed to assess whether reducing the risk of infection for chickens during transport and trade reduces their likelihood of shedding AIVs in the LBM.
Methods:
Sixty-four batches of 10 chickens (340 backyard and 300 broiler chickens) were caged together in a market stall in Chattogram, Bangladesh, and followed over a period of 84 hours. For each batch, 5 chickens were either recruited in rural households or purchased from commercial farms. Those chickens were then transferred to a poultry shed and kept for 2.5 days before being moved to the market stall. High biosecurity standards on the way and during storage were emphasised (intervention group). For each batch, another 5 chickens were directly purchased from traders after delivery at the LBM to ensure that they had experienced usual transport and trading practices (control group). Oropharyngeal swabs were collected from individual chickens upon arrival at the LBM, as well as 12, 36, and 84 hours later. Those samples were tested for the M, H9, and H5 AIV genes by RT-qPCR. Discrete-time proportional hazard models were developed for each of those genes with the first positive test result being the event of interest.
Results:
Of all chickens that survived and showed a negative test result upon arrival at the LBM, 45% (36.2%/0.85%) were positive for the M (H9/H5) gene after 12 hours, 80.9% (75.5%/1.7%) after 36 hours, and 91.4% (87.2%/4.4%) after 84 hours. Backyard, compared with broiler chickens, displayed lower hazard rates for both M (P<.001) and H9 genes (P<.001). For the H9 gene, hazard rates were also lower for the intervention than for the control group (P=.045), suggesting that stringent biosecurity measures delay viral shedding in the LBM.
Conclusions:
Our results indicate that a substantial proportion of viral shedding observed in LBMs results from infection that occurred upstream, at the farm gate or during trade. To reduce viral amplification in Bangladesh, focus should be placed on this stage of the value chain and comprise all relevant stakeholders involved in those processes. Our study furthermore demonstrates that LBMs can serve as viral reservoirs and pose a significant threat if poultry remain in market stalls for a prolonged period of time. This has implications for human health regarding reassortment and zoonotic spillover. Genomic analysis and mathematical modelling will be used to further uncover the circumstances of AIV circulation and maintenance along poultry value chains.