Respiratory viruses in individuals with a high frequency of animal exposure in southern and highland Vietnam

Abstract Active surveillance for zoonotic respiratory viruses is essential to inform the development of appropriate interventions and outbreak responses. Here we target individuals with a high frequency of animal exposure in Vietnam. Three‐year community‐based surveillance was conducted in Vietnam during 2013‐2016. We enrolled a total of 581 individuals (animal‐raising farmers, slaughterers, animal‐health workers, and rat traders), and utilized reverse transcription‐polymerase chain reaction to detect 15 common respiratory viruses in pooled nasal‐throat swabs collected at baseline or acute respiratory disease episodes. A respiratory virus was detected in 7.9% (58 of 732) of baseline samples, and 17.7% (136 of 770) of disease episode samples (P < .001), with enteroviruses (EVs), rhinoviruses and influenza A virus being the predominant viruses detected. There were temporal and spatial fluctuations in the frequencies of the detected viruses over the study period, for example, EVs and influenza A viruses were more often detected during rainy seasons. We reported the detection of common respiratory viruses in individuals with a high frequency of animal exposure in Vietnam, an emerging infectious disease hotspot. The results show the value of baseline/control sampling in delineating the causative relationships and have revealed important insights into the ecological aspects of EVs, rhinoviruses and influenza A and their contributions to the burden posed by respiratory infections in Vietnam.

(PIV)1-4, human CoV (including subtypes OC43 and NL63), human bocavirus (BoV) and parechovirus (PEV) are the most common viruses detected in respiratory samples worldwide. [3][4][5][6][7][8][9] Of these, influenza A virus, influenza B virus, and CoV have been reported as the most common viruses detected in people over 5 years old, [10][11][12][13] while RSV and PIVs have been regarded as the leading causes of respiratory infections in children under 5 years old in South East Asia. 3,14,15 Zoonotic infections are of global concern, and approximately 60% of known infectious diseases in humans are of zoonotic origin. 16 In addition, Southeast Asia, including Vietnam, is one of the hotspots of emerging infectious diseases. Indeed, many of the recent respiratory outbreaks are linked with zoonotic viruses as SARS-CoV, 17 avian influenza A virus H5N1, 18

pandemic influenza
A virus subtype H1N1, 19 and more recently MERS-CoV, 20  and community-based studies, was conducted across Vietnam. 21,22 Herein, we focus our analysis on a community-based study, which was designed to capture the cross-species transmission events of zoonotic viruses among individuals with a high risk of zoonotic infections in southern and highland Vietnam. In this study, our aim was to describe the frequency of common respiratory viruses in clinical samples collected from these individuals, later called cohort members, at baseline and when a respiratory disease episode was reported during the study period.

| Study design and inclusion criteria
This study was a part of the High-Risk Sentinel Cohort (HRSC) study which was a community-based component of the VIZIONS project. 21 The HRSC study was first commenced in June 2013 in Dong Thap and then in February 2014 in Dak Lak. These are provinces located in southern and highland of Vietnam, respectively, representing two different geographic areas in Vietnam.
Animal-raising farmers, animal health workers, and slaughterers were eligible to be enrolled in the study since these are common occupations in rural Vietnam with frequent occupational exposure to animals. Rat traders in Dong Thap were additionally recruited due to the commonality of this occupation in this locality. The animal-raising farmers accounted for about twothird of the population with occupational exposure to animals in these study provinces.
On the basis of the animal farm census, letters were sent out to invite potential participants to attend an introductory meeting. The consent forms were then obtained from those who were willing to join the HRSC study. For each farmer household, up to four members having the highest frequency of working with animals were recruited. The slaughterers were recruited from all local central abattoirs or slaughter points. The animalhealth workers and rat traders were selected by convenience. Consequently, a total of 581 individuals (median age in year, 38; range, 2-89), including 415 (71.4%) animal-raising farmers, 100 (11.7%) slaughterers, 61 (10.5%) animal-health workers, and 5 (1.8%) rat-traders, were recruited. Each cohort members were followed up annually for up 3 years since recruitment.

| Data collection
Annually, to establish the baseline data (ie, no disease episode reported), the cohort members were interviewed, and clinical specimens, including rectal, pooled nasal, and throat swabs and blood were also collected from each interviewee. These baseline data were collected from all cohort members, except for the farmers, for which only one person mostly working with animals per household was interviewed and sampled.
During the study period, whenever getting illness (diarrhea and respiratory infection) defined as any signs/symptoms of respiratory tract infections (eg, sneezing, coughing or sore throat), plus fever (≥38°C), the cohort members informed the local study teams. Within 48 hours, the site study doctors made a visit to the participant houses and collected information about animal exposures, associated symptoms, and medication. In addition, clinical specimens, including blood, and (when relevant) rectal-or pooled nasal and throat swabs were collected. All the specimens were stored at −80°C until analysis. Here, we focused on respiratory episodes. As such, only pooled nasal-throat swabs of each individual were analyzed.

| Respiratory virus detections by real-time polymerase chain reaction analysis
To detect common respiratory viruses in pooled nasal and throat swabs, we first isolated total nucleic acid (NA) from patient samples  Table S2). All the RT-PCR reactions were carried in a LightCycler 480 Instrument II (96-wells) (Roche Molecular Systems, Inc).

| Data analysis
The data were analyzed by STATA software, version 12.0. 27 The pairwise comparisons of categorical variables were calculated by Pearson's χ 2 test (or Fisher exact test when the sample size was less than five in any of the cells of a contingency table) or two-sample t test with equal variances.
The errors of multiple comparisons were corrected by the Bonferroni method. 28 P ≤ .05 was considered the significance. EpiTools 29 were used to calculate 95% confidence intervals for the odds ratio. The rat traders (n = 5) were excluded from these tests because of an insufficient sample size.    Table 2). In addition, mixed infections were recorded in 2 (0.3%) and 7 (0.9%) samples collected at baseline and disease episodes, respectively (Table 2).

| Ethics
Of the detected viruses, EVs, HRV and influenza A virus were the most common viruses detected in samples collected at both baseline and disease episodes, followed by ADV and CoV (   Table 3). In addition, gastrointestinal symptoms were recorded in 7.3% (56 of 770), but watery diarrhea was more often recorded in cohort members without a virus detected than in those with a positive finding, 52 of 634 (8.2%) vs 4 of 136 (2.9%), P = .029 (Table 3).

| Clinical signs/symptoms of cohort members in acute respiratory diseases with the detected viruses
Of the virus-positive cases, watery diarrhea was only recorded in those positive for EVs and HRV, whilst sore throat was predominantly recorded in those positive for influenza A virus.
Otherwise, there were considerable similarities in age and clinical presentations of cohort-member groups who were positive for different viruses (Table 3).

| The frequency of respiratory viruses detected by provinces
To assess the differences in the frequencies of respiratory viruses under investigation between Dong Thap and Dak Lak, which represent the two distinct geographic localities in Vietnam, we stratified the data for these two individual provinces ( were significantly more often detected in samples collected at disease episode than at baseline; P < .001 for both EVs and HRV. In Dak Lak, no significant differences were found (Table 4).

| Temporal and seasonal differences in the frequency of detection of respiratory viruses
There were some fluctuations in the detection of the most common viruses (especially EVs and HRV; In terms of seasonality, overall, there were some clear trends

| Animal exposure
Overall, the cohort members were exposed to a wide range of animals, including 11 types of exotic animals and 11 types of domestic animals, within ≤1 month prior to the disease episode (n = 770) (Table S1)  The antibiotic use of the patients from the first symptoms to the incidence interview/sampling. Note: Other viral pathogens were not showed as they were detected in less than 10 samples. P values were calculated by Pearson's   HRV and ADV, our report supports previous findings. [39][40][41][42][43] Our overall RT-PCR yield of 17.7% of viral agents in respiratory samples of the cohort members with the majority age from 16 years or above is in agreement with the diagnostic yields of previous studies. [44][45][46][47][48][49] The results suggest that it is probably because adults have acquired substantial immunity during their life, leading to the rapid clearance of the infecting viruses from their respiratory tract, thereby shortening the duration of viral shedding.
Our study has some limitations. First, no human subjects without animal exposure were recruited as controls. Therefore, we were unable to assess the effect (if any) of animal exposure on the The seasonal distribution of symptomatic EVs-, HRV-, influenza A virus-, ADV-, and CoV (subtype OC43 and NL63)-infected cases detected by RT-PCR assay. The bars show the proportion of the viruses detected among total samples tested (the line chart) each month. EVs and influenza A virus were more likely detected in the rainy season than in the dry season (P = .002 and P = .023, respectively), while the ADV detections were more frequent in the dry season as compared with the rainy one (P = .044). There was no significant difference in the detections of HRV and CoV (subtype OC43 and NL63) between dry and rainy seasons (P = .333 and .227, respectively). ADV, adenovirus; CoV, coronavirus; EV, enterovirus; HRV, human rhinovirus; RT-PCR, real-time polymerase chain reaction frequency of the respiratory disease incidence, as well as the observed viral patterns. Second, despite a holistic effort, nonviral agents as bacterial pathogens were not tested. Third, a slight decrease in sensitivity of the multiplex RT-PCR platforms used in the present study as compared with that of the corresponding monoplex RT-PCRs have previously been reported, 23