Monitoring temporal changes in prevalence and viral load of avian influenza viruses at live poultry markets in Guangzhou, China, 2015-2018

Abstract

Live poultry markets (LPMs) are the “hot bed” for the amplification and dissemination of avian influenza viruses (AIVs). Spillover human zoonotic infections by AIVs during winter seasons were reported after direct or indirect exposure to live poultry. However, the temporal pattern of AIVs at LPMs, including changes in prevalence and viral load in poultry across seasons have not been comprehensively recorded. Furthermore, risk factors associated with the detection of AIVs at LPMs leading to subsequent human infections have not been fully characterized. To assess the transmission dynamics of avian influenza viruses, a longitudinal surveillance was set up to monitor monthly changes in prevalence and viral load of H5, H7 and H9 viruses at retail and wholesale LPMs in Guangdong province in China from December 2015 to March 2018.

During the study period, significantly higher AIV positive rates and viral loads (copy number) were consistently detected from the poultry swabs or environmental sampled from the retail than the wholesale market, signifying the impact of poultry mixing practice at retail markets in amplifying AIVs At the retail market, the highest median monthly positive rates for H9, H7 and H5 subtypes were detected at 71.8% from environmental swabs, 7.4% from environmental swabs and 4.3% from environmental swabs respectively. Specifically, H5 positive rates and viral load detected at both the wholesale (poultry swabs and environmental samples) and retail (poultry swabs) markets were significantly correlated with H5N6 human infection risk, while H7 positive rate and viral load at the retail market (environmental samples) was significantly correlated with H7N9 human infection risk by Spearman’s rank correlation test. The results highlight the potential of setting up alarm threshold values based on the H5 and H7 detection frequency from longitudinal surveillance studies at the LPMs as indicators for human zoonotic infection risk.

Logistic regression model was applied to investigate the temporal pattern and risk factors associated with detection of the H5, H7, H9 subtypes from the environmental samples at the LPMs. Model predicted peak H9 activity in environmental swabs in December at both LPMs. At the wholesale market, the number of poultry sale may increase H9 detection risk, while market rest day may reduce H9 detection risk. However, at the retail market, market rest day has limited effect in reducing H9 positive rate. Model fitting identified annual peak of H5 activity in January and biannual peak of H7 activity in January and May at the retail market, which correlated with our surveillance data. Lower temperature and lower absolute humidity were associated with increased H5 and H7 detection. Adopting market rest days or closure was effective in reducing H5 detection rate but was less prominent in reducing H7 contamination. Introduction of updated H5/H7 vaccines in Summer 2017 significantly reduced H7 detection risk at the retail market. Although our models are limited by the parameters recorded during the study period, our findings suggest the existence of uncommon risk factors for the detection of H5, H7, H9 subtypes at LPMs. The information can be considered for future development of effective intervention strategies.

(Word counts: 500)