key: cord-0016341-i3hbcvkg authors: McBride, Dillon S; Perofsky, Amanda C; Nolting, Jacqueline M; Nelson, Martha I; Bowman, Andrew S title: Tracing the source of influenza A virus zoonoses in interconnected circuits of swine exhibitions date: 2021-03-04 journal: J Infect Dis DOI: 10.1093/infdis/jiab122 sha: ab0aa2513c691963bc3ba25a2ba791f0a61d63a3 doc_id: 16341 cord_uid: i3hbcvkg BACKGROUND: Since 2011, influenza A viruses circulating in US swine exhibited at county fairs are associated with over 460 zoonotic infections, presenting an ongoing pandemic risk. Swine ‘jackpot shows’ that occur before county fairs each summer intermix large numbers of exhibition swine from diverse geographic locations. We investigated the role of jackpot shows in influenza zoonoses. METHODS: We collected snout wipe or nasal swab samples from 17,009 pigs attending 350 national, state, and local swine exhibitions across eight states during 2016-2018. RESULTS: Influenza was detected in 13.9% of swine sampled at jackpot shows, and 76.3% of jackpot shows had at least one pig test positive. Jackpot shows had 4.3-fold higher odds of detecting at least one influenza-positive pig compared to county fairs. When influenza was detected at a county fair, almost half of pigs tested positive, clarifying why zoonotic infections occur primarily at county fairs. CONCLUSIONS: The earlier timing of jackpot shows and long-distance travel for repeated showing of individual pigs provide a pathway for the introduction of influenza into county fairs. Mitigation strategies aimed at curtailing influenza at jackpot shows are likely to have downstream effects on disease transmission at county fairs and zoonoses. A c c e p t e d M a n u s c r i p t 4 The diverse pool of influenza A viruses (IAV) in animals present an ongoing threat to human health [1] [2] [3] . Swine are mixing vessels for the evolution of genetically novel IAVs with genetic components from multiple host species [4] . The 2009 H1N1 influenza pandemic was of swine origin, highlighting the pathway for zoonotic IAV emergence provided by swine systems [3, 5, 6] . The 2009 pandemic furthered understanding of how global swine production facilitates the evolution of novel IAVs with zoonotic potential [7] [8] [9] , and how specific human-animal interfaces can be targeted to prevent interspecies spillover events. Over the last decade, the United States has recorded the highest number of swine-origin zoonotic IAV infections worldwide. Since 2011, novel reassortant IAVs of swine origin (H1N1v, H1N2v, and H3N2v subtypes) have been associated with more than 460 zoonotic infections in the United States [10] . While swine workers are at a higher risk for zoonotic IAV from swine [11, 12] , and zoonotic cases are sporadically liked tocommercial swine, the vast majority of these infections were in youth swine exhibitors at county fairs [13] [14] [15] . An estimated 150 million people attend fairs in North America annually [16] , providing a conduit for numerous people to interact with swine and their pathogens. Agricultural fairs, which include county fairs, facilitate commingling among hundreds of swine for up to a week, creating an environment conducive to the rapid spread of IAV between animals as well as an interface for zoonotic transmission. The pigs at fairs are raised primarily by youth exhibitors and their families for the purpose of being shown at fairs and other shows through agricultural education programs such as 4-H and FFA. In the Midwestern United States, most swine-origin zoonotic IAV infections have occurred, county fairs occur primarily during June-October. Although county fairs are typically restricted to pigs from that county, swine exhibitors report attending an average of 3.1 shows in a year and showing the same pig multiple times [17] . Many of the additional exhibitions belong to A c c e p t e d M a n u s c r i p t 5 circuits of ‗jackpot' shows that are held throughout the year. Swine are permitted to attend jackpot shows and return to their home farms afterwards, potentially facilitating long-range dissemination of IAVs ( Figure 1 ). Exhibitors with pigs attending multiple shows prior to their arrival at county fairs may be the source of the diverse IAVs detected in pigs that subsequently infect humans. Zoonotic infections have not been traced directly to jackpot shows. However, the occurrence of jackpot shows prior to the agricultural fair season positions them to serve as key upstream sources of the zoonotic viruses later isolated from humans. We conducted active IAV surveillance to determine IAV detection frequency at jackpot shows and county fairs and to elucidate the role of jackpots in disseminating IAVs in the exhibition network. During 2016-2018, we conducted IAV surveillance in exhibition swine at jackpot shows across eight US states (Arizona, Georgia, Illinois, Indiana, Iowa, Kentucky, Michigan, and Ohio) and at county fairs in Ohio, Indiana, and Michigan, due to the frequency of zoonotic IAV cases in these states [10] . Swine sampling was approved by The Ohio State University Institutional Animal Care and Use Committee (protocol 2009A0134-R2). State fairs differ from both county fairs and jackpot shows in their broader regionality, longer duration, and greater size. Many state fairs host multiple shows for different categories of exhibition swine. Due to the variability both within state fairs and between states regarding their structure and regulations, state fairs could not be appropriately grouped with jackpots or county fairs and were censored from our analyses. With permission from organizers overseeing each show, we collected 20 nasal swabs [16] or 24 snout wipes [18] at county fairs, 200 snout wipes at state jackpots, and 400 snout wipes at national jackpots, on the last day of each show. Samples were systematically A c c e p t e d M a n u s c r i p t 6 collected from pigs and spatially distributed thoughout the barnd without consideration for clinical signs of IAV. Samples were placed on dry ice for transportations immediately after sampling. The sample sizes for county fairs were chosen to ensure ≥95% probability of detecting of IAV at a prevalence of ≥0.15 at a typical sized fair. We targeted more samples at jackpot shows to account for shorter show durations (i.e. less viral amplification). The required number of shows at which we sampled was calculated to detect the expected difference in proportion of positive shows between jackpots and county fairs with 80% power. To evaluate the geographic distribution of swine at jackpot shows, we recorded US postal codes corresponding to the location of the farm on which each pig was raised. We did not record identifying information for pigs or exhibitors to preserve exhibitor anonymity. We screened viral RNA extracted from the samples with real-time reverse transcriptase PCR (rRT-PCR) using the VetMAX™-Gold SIV Detection Kit (Life Technologies) with the 7500 Fast Real-Time PCR System (Life Technologies) according to manufacturer's protocol. We conducted virus isolation on all rRT-PCR positive samples using MDCK cells [19] . We sequenced representative IAV isolates from each positive exhibition as previously described [20] [21] [22] and used full genome sequence data for antigenic subtype classification. All sequences are available in GenBank; accession numbers are listed in Supplementary Table S1 . All statistical analyses were conducted using R version 4.0.1 [23] . Exhibition-level IAV status was designated using the results of IAV rRT-PCR tests of nasal swab and snout wipe samples. Exhibitions were designated as IAV positive when ≥1 sample tested positive. We estimated exhibition-level IAV prevalence by dividing the number of individual swine that tested PCR-positive for IAV by the total number of swine sampled. In statistical analyses A c c e p t e d M a n u s c r i p t 7 with multiple comparisons, we used the Benjamini-Hochberg procedure to control the false discovery rate. The association of IAV status with exhibition type and location was measureded using Fisher's exact tests and posthoc tests with Monte Carlo resampling (10,000 replicates), in which the nominal variables were exhibition type or state, respectively, and IAV presence or absence. For spatial geographic analyses, we focused on state jackpots and county fairs in the three most-densely sampled states in our dataset: Ohio, Michigan, and Indiana. To determine if exhibition type or location was associated with influenza prevalence, we conducted general independence tests with Monte Carlo resampling (10,000 replicates) (coin package) [24] . For post-hoc comparisons of prevalence between exhibition types or between US states, we used pairwise permutation tests (rcompanion package) [25] . We recorded epidemiologically relevant metadata for each exhibition, including GPS coordinates, sampling date, date of entry for exhibitors, exhibition size (< 200 swine or > 200 swine), and whether other exhibitions had tested positive within the fourteen days prior and the corresponding distances of these exhibitions. We defined exhibition duration as the number of days between the required arrival date for swine exhibitors and the date pigs were sampled [26] . For each year, we calculated pairwise geographic distances and pairwise differences in entry date between all exhibitions. We obtained total swine population sizes for We used a Bayesian approach to model the probability of county fairs and state jackpots testing positive for IAV. We excluded national jackpots from this analysis because all were IAV-positive. Model covariates included show entry date (measured as the number of days from January 1st of each year), exhibition type (county fair or state jackpot), show size (< 200 swine or > 200 swine), and the number of swine residing in each show's respective county. In ten separate models, we included a covariate for whether IAV was detected at any exhibition within a 25-, 50-, 75-, 100-, or 200-mile radius within the prior two weeks (yes or no). Continuous variables were mean-centered and scaled prior to model fitting. We fit logistic generalized linear regression models (GLMs) using Markov Chain Monte Carlo (MCMC) with Stan [28] via the rstanarm package [29] . Weakly informative Student's-t priors (df=7, location=0, scale=2.5) were placed on intercept and coefficient effect sizes. For each model, we ran four MCMC chains, each for 2000 iterations (including a burn-in period of 1000 iterations that was discarded), producing a total posterior sample size of 4000. We verified convergence by inspecting trace plots and confirming that all parameters had sufficiently low values (all <1.1) and sufficiently large effective sample sizes (>15% of the total sample size). To evaluate each model's predictive performance, we computed approximate leave-one-out cross validation (LOOCV) using Pareto smoothed importance sampling (loo package) [30, 31] . We used LOO Information Criterion to estimate differences in model out-of-sample predictive accuracy and -stacking‖ of predictive distributions to compute model weights [32] . A c c e p t e d M a n u s c r i p t Across the three years, we collected nasal swab and snout wipe samples from over Table S2 ). All of the PB2 segments were TRIG lineage and selected for genetic distance analysis. Genetic distances between isolates were smaller within a single year compared to between study years (Supplementary Table S3 ), further supporting the notion that similar viruses propagate throughout shows in a given year, but patterns in dominant genotypes do not necessarily persist between show seasons. IAV first appeared in state jackpot shows in May, then in national jackpot shows beginning in June, and ultimately in county fairs from June to October (Figure 3 ). IAV was frequently detected at state jackpot shows that took place in late June and early July, and sporadic zoonotic infections coincided with the peak of county fair season in July and early August (Figure 3 ). We detected IAV in a higher proportion of state and national jackpots (76.3%) compared to county fairs (37.8%) (Figures 3 and 5) , and the probability of a swine exhibition having at least one IAV-positive animal was strongly associated with the exhibition being a jackpot show (Fisher's exact tests, p=0.0002 and p<0.00001). Among jackpot shows, we detected IAV in 68.9% of state shows and at every national show, even those held in regions with small local swine populations (Fisher's exact test, p = 0.03; Figure 2 ). The proportion of pigs testing positive for influenza varied across fair types and study years (Table 1, Figure 6A ), with national jackpots having the greatest overall prevalence when data from the three years were combined (Fisher exact tests: State Jackpot versus National Jackpot, P < 0.001; County Fair versus National Jackpot, P = 0.01). However, these numbers represent averages across hundreds of fairs, and a more detailed examination reveals that county fairs and jackpot shows have different dynamics. County fairs tend to be highly A c c e p t e d M a n u s c r i p t 11 bifurcated: either there are no IAV-positive animals (62.2% of fairs) or more than 75% of animals are IAV-positive (13.7% of fairs). In contrast, state and national jackpot shows are consistently infected but generally at lower frequencies (13.9% of animals). Therefore, if we limit our analysis to exhibitions with at least one IAV-positive pig, county fairs had significantly greater IAV prevalence than state and national jackpots, both in the three-year dataset (pairwise permutation tests, p=0.00008 and p=0.003) and when analyzing years separately (2016: p=0.00004, 2017: p=0.07, 2018: p=0.02; Figure 6B) . Among IAV-positive shows, the mean estimated prevalence was 16.8% for state jackpots, 16.2% for national jackpots, and 46.9% for county fairs. More than a third (36.3%) of IAV-positive county fairs had ≥75% pigs test positive, whereas only one IAV-positive jackpot show had ≥50% pigs test positive. For over a decade animal and public health officials have attempted to disrupt the transmission of IAVs from exhibition swine to youth exhibitors that occurs every summer. However, it is difficult to mitigate zoonotic transmission at county fairs where upwards of 70% of swine can be infected, and new preventive strategies are needed. Here, we demonstrate that IAVs are frequently detected in exhibition swine at jackpot shows that occur upstream to county fairs and draw pigs from wide geographic catchments, presenting a new target for mitigation efforts. Critically, IAVs detected at jackpot shows have been connected to zoonotic transmission events that occur at subsequent county fairs. For example, an emerging H3N2 subtype that was recovered from our jackpot swine samples in 2016 was subsequently detected one month later in pigs and people at a county fair [14] . This study highlighted the potential of jackpot shows to seed IAV diversity into county fairs where public health risk is high. Likewise, an early-season national jackpot show in 2018 appeared to serve as a superspreading event that propagated H1δ-2 (H1N2v) IAVs throughout the A c c e p t e d M a n u s c r i p t 13 exhibition circuit; this lineage caused the majority of zoonotic cases reported later that summer [34] . In each year of our study, genotype analyses indicate that IAVs detected at jackpot shows contribute to IAV outbreaks at county fairs later in the show season. Altering the timing of critical jackpot shows may be effective at reducing introduction of zoonotic potential IAV into county fairs. To date, zoonotic transmission occurs primarily at county fairs. Characterizing entire networks of swine exhibitions that drive the dispersal of zoonotic IAVs into county fairs has uncovered new opportunities for mitigating an ongoing pandemic risk. Although IAV prevalence is lower at jackpots, the likelihood of detecting at least one pig with IAV is higher at jackpot shows compared to county fairs. The structure of jackpot shows allows exhibition swine from widely-scattered locations to commingle, return home, and show again repeatedly throughout the year, which explains the simultaneous detection of highly related IAVs at geographically dispersed county fairs [35] . Swine can shed IAV for at least nine days post exposure and typically remain infectious for at least 5 days [36, 37] . This duration of infection enables swine to carry and transmit IAV from shows week-to-week and within home farms. Because the timing of state jackpot shows overlaps with both national jackpot shows and county fairs, IAVs introduced into jackpot shows have a clear route to seed county fairs. The introduction of IAV at a county fair by a few pigs can spread rapidly among participating swine [38] , which enables subsequent zoonotic transmission [14, 16] . Due to local control of show regulations, adoption of recommendations to mitigate zoonotic transmission of IAV is highly varied among the thousands of county fairs that occur annually in the United States. The jackpot show network offers public health control points that could be targeted to reduce IAV introductions at downstream county fairs. The lower estimated prevalence of IAV within IAV-positive jackpot shows compared to county fairs accounts for the observed absence of zoonotic infections tied to jackpot A c c e p t e d M a n u s c r i p t 14 shows. While county fairs are typically 5-7 days, many jackpot shows are short in duration (1-2 days), which does not likely afford time for sustained IAV amplification within shows. Beyond duration, lower IAV prevalence in jackpots shows could arise from differences in swine husbandry between practiced jackpot exhibitors and county fair exhibitors who only show pigs annually. For example, a higher proportion of jackpot exhibitor survey respondents report vaccinating their pigs against IAV [39] than county fair respondents in a different survey [17] . Throughout this study period, exhibitors likely had access to autogenous vaccines and commercially available live attenuated and whole inactivated influenza vaccines, but as of 2020, the only vaccine commercially available to exhibitors is a whole inactivated vaccine. IAVs are funneled down from national jackpots into state jackpot circuits and from state jackpots into the subsequent county fairs [34] . At a national jackpot show associated with IAV dissemination [34] , we recorded 11.2% Indiana zip codes across three years Exhibitor household members often have contact with swine other than their own, providing an interface for viral transmission between commercial and exhibition swine [17] . However, the commercial swine presence of a county may not be a reliable predictor of IAV infection at jackpot shows, which host a majority of swine exhibitors from outside of the county. Despite being a small niche in the swine industry, IAV surveillance efforts have detected substantial viral diversity in exhibition swine [16, 38] We thank all of the county fairs, jackpot shows, and youth exhibitors that participate in our surveillance program as well as lab staff for their technical expertise and assistance in collection and testing of surveillance samples. The conclusions of this study do not necessarily represent the views of the NIH or the US government. The origin of the 1918 pandemic influenza virus: a continuing enigma The importance of animal influenza for human disease Origins and evolutionary genomics of the 2009 swine-origin H1N1 influenza A epidemic The pig as a mixing vessel for influenza viruses: Human and veterinary implications Antigenic and Genetic Characteristics of Swine-Origin 2009 A(H1N1) Influenza Viruses Circulating in Humans Origins of the 2009 H1N1 influenza pandemic in swine in Mexico. Neher RA, editor. eLife. eLife Sciences Publications, Ltd Influenza A Virus in Swine Multiple Reassortment between Pandemic (H1N1) 2009 and Endemic Influenza Viruses in Pigs, United States When Pigs Fly: Pandemic influenza enters the 21st century Reported Infections with Variant Influenza Viruses in the United States | CDC Are Swine Workers in the United States at Increased Risk of Infection with Zoonotic Influenza Virus? Swine Workers and Swine Influenza Virus Infections Molecular evidence for interspecies transmission of H3N2pM/H3N2v influenza A viruses at an Ohio agricultural fair Influenza A(H3N2) Virus in Swine at Agricultural Fairs and Transmission to Humans Outbreak of Variant Influenza A(H3N2) Virus in the United States Subclinical Influenza Virus A Infections in Pigs Exhibited at Agricultural Fairs Movement patterns of exhibition swine and associations of influenza A virus infection with swine management practices Utility of snout wipe samples for influenza A virus surveillance in exhibition swine populations. Influenza Other Respir Viruses Comparative effectiveness of isolation techniques for contemporary Influenza A virus strains circulating in exhibition swine Viral deep sequencing needs an adaptive approach: IRMA, the iterative refinement meta-assembler Influenza A Virus Field Surveillance at a Swine-Human Interface. mSphere Avian Influenza Viruses in Wild Birds: Virus Evolution in a Multihost Ecosystem R: A Language and Environment for Statistical Computing Austria: R Foundation for Statistical Computing Implementing a Class of Permutation Tests: The coin Package Summary and Analysis of Extension Program Evaluation in R Exploration of risk factors contributing to the presence of influenza A virus in swine at agricultural fairs. Emerg Microbes Infect National Agricultural Statistics Service (NASS) A Probabilistic Programming Language rstanarm: Bayesian Applied Regression Modeling via Stan. R Package version Practical Bayesian model evaluation using leave-one-out cross-validation and WAIC Pareto Smoothed Importance Sampling. ArXiv150702646 Stat Using Stacking to Average Bayesian Predictive Distributions (with Discussion) Swine Influenza A Viruses and the Tangled Relationship with Humans. Cold Spring Harb Perspect Med A heterogenous swine show circuit drives zoonotic transmission of influenza A viruses in the United States Introduction, Evolution, and Dissemination of Influenza A Viruses in Exhibition Swine in the United States during 2009 to 2013 Efficacy in Pigs of Inactivated and Live Attenuated Influenza Virus Vaccines against Infection and Transmission of an Emerging H3N2 Similar to the 2011-2012 H3N2v Indirect Transmission of Influenza A Virus between Pig Populations under Two Different Biosecurity Settings Prevalence of Influenza A Virus in Exhibition Swine during Arrival at Agricultural Fairs Perceptions and attitudes of swine exhibitors towards recommendations for reducing zoonotic transmission of influenza A viruses The authors do not have any commercial or other association that might pose a conflict of interest.