key: cord-0994412-1sustjvd authors: Ajayi, T.; Dara, R.; Misener, M.; Pasma, T.; Moser, L.; Poljak, Z. title: Herd‐level prevalence and incidence of porcine epidemic diarrhoea virus (PEDV) and porcine deltacoronavirus (PDCoV) in swine herds in Ontario, Canada date: 2018-04-01 journal: Transbound Emerg Dis DOI: 10.1111/tbed.12858 sha: f9898c959e124f7234179de547c5f35d39456336 doc_id: 994412 cord_uid: 1sustjvd Porcine epidemic diarrhoea virus (PEDV) and porcine deltacoronavirus (PDCoV) were first identified in Canada in 2014. Surveillance efforts have been instrumental in controlling both diseases. In this study, we provide an overview of surveillance components for the two diseases in Ontario (Canada), as well as PEDV and PDCoV incidence and prevalence measures. Swine herds located in the Province of Ontario, of any type, whose owners agreed to participate in a voluntary industry‐led disease control programme (DCP) and with associated diagnostic or epidemiological information about the two swine coronaviruses, were eligible to be included for calculation of disease frequency at the provincial level. PEDV and PDCoV data stored in the industry DCP database were imported into the R statistical software and analysed to produce weekly frequency of incidence counts and prevalence counts, in addition to yearly herd‐level incidence risk and prevalence between 2014 and 2016. The yearly herd‐level incidence risk of PEDV, based on industry data, was 13.5%, 3.0% and 1.4% (95% CI: 11.1–16.2, 2.0–4.2, 0.8–2.3), while the yearly herd‐level incidence risk of PDCoV was 1.1%, 0.3%, and 0.1% (95% CI: 0.5–2.2, 0.1–0.9, 0.0–0.5), for 2014, 2015 and 2016, respectively. Herd‐level prevalence estimates for PEDV in the last week of 2014, 2015 and 2016 were 4.4%, 2.3% and 1.4%, respectively (95% CI: 3.1–6.0, 1.5–3.3, 0.8–2.2), while herd‐level prevalence estimates for PDCoV in the last week of 2014, 2015 and 2016 were 0.5%, 0.2% and 0.2%, respectively (95% CI: 0.1–1.2, 0.0–0.6, 0.0–0.6). Collectively, our results point to low and decreasing incidence risk and prevalence for PEDV and PDCoV in Ontario, making both diseases possible candidates for disease elimination at the provincial level. Porcine epidemic diarrhoea (PED) was first described in England in 1971, and its causative agent, porcine epidemic diarrhoea virus (PEDV), was identified in 1978 (Chen et al., 2014) . PEDV causes anorexia, vomiting, diarrhoea and dehydration in pigs, resulting in near 100% mortality for piglets during the first few days of life (Hill et al., 2014) and low mortality in older pigs. The virus spreads via the faecal-oral route, either through direct contact with an infected pig or through indirect contact with contaminated fomites. Widespread outbreaks were reported in Europe during the 1970s and 1990s, while epidemics in Asia have caused significant disruption to Asian pig production since 2008 (Williamson et al., 2013) . Porcine deltacoronavirus (PDCoV, also known as swine deltacoronavirus, SDCV) was first identified in Hong Kong in 2012. The transmission modes and clinical signs due to PDCoV infections are similar to PEDV; however, the mortality rate is generally lower after PDCoV infections (Carvajal et al., 2015) . PEDV emerged in North America in May 2013, while PDCoV was first confirmed in February 2014, both in the United States (Chen et al., 2014; Ma et al., 2015) . These novel viruses rapidly disseminated throughout the US swine population, resulting in the mortality of an estimated seven million animals by May 2014 (Jung & Saif, 2015) . Due to the faecal-oral route of transmission, the infection spread through various mechanisms, including contaminated transportation vehicles (Lowe et al., 2014) . In Canada, PEDV emerged in January 2014 when a swine herd in Ontario tested positive for the virus (Kochhar, 2014) . Imported spraydried porcine plasma contaminated with PEDV was the likely pathway of introduction, as established through descriptive studies (Pasma, Furness, Alves, & Aubry, 2016) , analytical epidemiological studies (Aubry, Thompson, Pasma, Furness, & Tataryn, 2017; O'Sullivan, 2015) and experimental investigations (Pasick et al., 2014) . By July 2014, only 62 cases of PEDV had been reported in Ontario and the outbreak was largely under control (Pasma et al., 2016) . Initial emergence of the two novel porcine coronaviruses was followed by their successful elimination from several initial case farms (Misener, 2015) . The high rate of successful PEDV elimination from individual herds, and effective measures that seemed to have minimized widespread viral dissemination, resulted in the current position of industry organizations that both infectious agents can and should be eliminated at the provincial level (Ontario Swine Health Advisory Board, 2017). A disease control programme (DCP) involves disease monitoring, surveillance, intervention and control strategies (Salman, 2003) . The DCP considered here has been voluntary in nature as defined elsewhere (Christensen, 2003) . Furthermore, an important component of any disease control programme is measuring trends in incidence and prevalence, particularly when the disease of interest moves into the phase of possible elimination (Salman, 2003) . With the infrastructure built for management of endemic diseases in Ontario, the data to support estimation of disease trends are available. Thus, the primary objective of this study was to estimate herd-level incidence and prevalence measures for PEDV and PDCoV in swine herds in Ontario (Canada) between January 2014 and December 2016, based on industry data. The secondary objective was to describe relevant surveillance components that were used for identification of new PEDV cases. The Ontario Swine Health Advisory Board (OSHAB) maintains a database which contains premises information and PEDV/PDCoV herd status for producers enrolled in its voluntary regional disease control programmes. The program, and a database originally designed for management of porcine reproductive and respiratory syndrome virus (PRRSV) (Arruda, Poljak, Friendship, Carpenter, & Hand, 2015) , was expanded to PEDV and PDCoV after their emergence. The data relevant for this work included unique identifiers, herd type, date of herd enrolment into the database, PEDV and PDCoV status of individual premises, and the date that individual premises changed their PEDV and PDCoV status. For inclusion into the study, swine herds could be of any herd type, but had to meet the following criteria: (i) be located in the Province of Ontario, (ii) participate in the voluntary industry-led disease control programme and (iii) have diagnostic or epidemiological information about the infection status of porcine epidemic diarrhoea virus (PEDV) or porcine deltacoronavirus (PDCoV). The industry organization (i.e., OSHAB) provided relevant data to researchers for calculation of disease frequencies under a separate data transfer agreement. Due to the voluntary nature of the DCP, the enrolment of herds into the database has been an ongoing process. This could have resulted in the date of enrolment being later than the date the infection was originally detected in a specific herd. In rare instances where a herd's enrolment date was not specified or occurred later than the first reported case of PEDV or PDCoV for that herd, the disease status date was entered as the herd enrolment date for the purposes of this report. This was done so that herd-level prevalence on a weekly basis could be properly calculated. An Open Database Connectivity (ODBC) connection to this database was established, and relevant tables were imported into R (R Core Team, 2016) using the rodbc library (Ripley & Lapsley, 2016 ). . For a premise to be classified as confirmed positive, it had to have an associated diagnostic submission number that includes at least one positive test for PEDV/PDCoV using RT-PCR, regardless of the number of specimens that were submitted. An AHL reference number (also known as a G Number) was not available for some confirmed positive cases in the database, as the attending veterinarian obtained test results, but did not provide OSHAB with the AHL reference number; (ii) presumed positive-premises which housed animals that were moved from positive sites at a prior stage in the production system (i.e., defined as positive due to pig movement). This information was obtained from attending veterinarians based on their knowledge of pig flow and movement, and was not based on diagnostic testing conducted on the premises of interest. For premises to be classified as presumed positive, the herd veterinarian simply needed to indicate that a specific site received pigs from PEDV/PDCoV-positive sites; (iii) presumed negative-previously positive premises, either confirmed or presumed, that were tested using PCR tests according to industry guidelines and had all test results negative. The sampling requirements for declaring premises to be PEDV/PDCoV presumed negative varied based on the combination of herd type and the type of animal flow (i.e., all-in/all-out by barn, or continuous flow nursery and finisher herds). Complete criteria were, at the time of publishing, available on the website of the industry organization (Ontario Swine Health Advisory Board, 2015). Briefly, sampling strategy for farrow-wean, nursery and finisher sites aimed to detect prevalence of virus-positive animals of at least 10%, with expected herd sensitivity of 95%, assumed test sensitivity of 98%, test specificity of 100%. In instances where the sampling material was oral fluid collected through cotton ropes, the assumption was that five pigs contributed oral fluids to one rope, and such fluids were considered a pooled sample. For farrow-finish or farrow-feeder sow sites, the same assumptions were made, except that the sampling strategy was required to detect prevalence of 5% with 95% herd sensitivity. Because of alternative strategies, the required sample size varied but a minimum sample size was four oral fluids. In addition, the testing time in sow herds was prescribed to be a minimum of 10 weeks post-infection and was required to be repeated three times in the case of farrow-wean (FW) sites, or two times in the case of farrowfinish (FF) or farrow-feeder (FG) sites. The recommended specimen type could be swab, Swiffer (for covering larger areas in a pen), or oral fluids, depending on the target age group. In FW herds, individual farrowing crates were the target population for each individual sampling occasion, in particular if diarrhoea was evident. The minimum recommendation for one sampling occasion in FW herds was to sample four Swiffer samples, at least eight farrowing crates per one sample. Alternatively, individual swabs of 30 farrowing crates were deemed as acceptable sample after pooling 5:1. In FG farms, the recommended sample type was oral fluid, with recommendation to collect 12 oral fluids from nursery pigs. Similarly, in FF farms, six oral fluids were recommended for collection from nursery pigs and six for collection from finisher pigs. For all-in/all-out nursery and finisher farms, the recommendation was to sample six oral fluids, with added requirement that these herds should be supplied from sow herds with a confirmed negative status. All testing has been assumed to be performed using RT-PCR tests. Full description is available elsewhere (Ontario Swine Health Advisory Board (OSHAB), 2015); (iv) confirmed negative-premises which have had no clinical signs or diagnostic evidence of PEDV/PDCoV for at least 6 months after the presumed negative status date. In addition, herds that were part of the Ontario voluntary DCP but were not tested for emerging porcine coronaviruses-due to lack of clinical or other types of diagnostic or epidemiological triggers-had assigned status of NA (not available). For each week where a herd's PEDV/PDCoV status was not reported, the status was set to the last-reported status using the zoo package (Zeileis & Grothendieck, 2005) . For example, if a status is not reported for the current week, and "confirmed positive" was reported for the prior week, then the current week's status is "confirmed positive." The individual premises data were then aggregated to counts of premises on a weekly basis. Based on the former time series, prevalence area plots were generated, providing a visual assessment of "confirmed positive," "presumed positive," "presumed negative" and "confirmed negative" herds over time. The prevalence numerator was the positive herd count (sum of confirmed positive and presumed positive herds) for a specific week, while the denominator was the herd count in the premises table for that week (calculated previously). Subsequently, the following disease status changes were tracked by week for each premise: (1) "Not Available" to "Presumed Positive," (2) "Not Available" to "Confirmed Positive," (3) "Presumed Positive" to "Presumed Negative," (4) "Presumed Positive" to "Confirmed Negative," Any status change leading to new presumed or confirmed positive status (i.e., status changes 1, 2, 7, 8, 9 and 10) was classified as new positive. Similarly, status changes 3, 4, 5 and 6 were classified as new negative. The number of new positives and new negatives was then aggregated to the weekly level throughout the study period. The latter time series were then used to construct a chart of cumulative incidence counts for each year, and epidemic curves were constructed for positive herds and herds which became negative. In addition, for each week, the incidence risk was calculated by dividing the number of cases that occurred in a specific week and by the number of herds that were eligible to become cases at the beginning of the week. For the yearly incidence rate, we calculated total number of herd-years at risk for each year from the number of herds under risk in each week and used this as a denominator. The incidence rate was then expressed as number of cases per one herd-year. Exact 95% confidence intervals on the incidence rate were obtained via the poisson.test function in R. (Pasma et al., 2016) , and is the first such case in the production system. In other words, the PEDV case count pertains to primary cases only; subsequent secondary cases due to animal movement in the production system are not included in OMAFRA reporting, although they could be confirmed as PEDV- The cumulative number of new PEDV and PDCoV cases per week for the 3 years is provided in Figures 2 and 3 , respectively. Notably, there were many new PEDV cases detected in the winter of 2014 when the disease was first introduced to Ontario and Canada. herds from the source population were eligible to be listed as a case, at least during the initial phase of the outbreak when PEDV could be considered an emerging hazard. However, only the first case in a given production system was counted as a case and secondary cases due to planned animal movement to other T A B L E 1 Herd-level incidence risk and rate of two novel porcine coronaviruses (PEDV and PDCoV) (Brisson, 2014) . Thus, the OSHAB surveillance coverage was lower than for the OMAFRA surveillance, which had 100% coverage-by law, PEDV-infected herds were reported to OMAFRA during the phase when the hazard was still considered as emerging. The OSHAB case definition also included secondary sites that were confirmed positive due to animal movement and had The prevalence figures for PDCoV are quite different from PEDV-rather than a cyclical pattern as previously observed, the numbers are more erratic, with sudden peaks followed by constants over prolonged periods, and then declines. The relatively static and low prevalence (0.1%-0.2%) observed between July 2015 and December 2016 suggests that PDCoV is a candidate for disease elimination. There is also a possibility that PDCoV cases are underreported by the industry, either because of potentially lower clinical impact in swine herds or because of perceived lower importance than PEDV. Nonetheless, when any of the three porcine coronaviruses (PEDV, PDCoV, TGEV-transmissible gastroenteritis virus) is suspected in a herd and diagnostic material is submitted to the AHL, the diagnostic testing is automatically conducted for all three viruses. Some limitations of the current study include the fact that PEDV and PDCoV cases are actively pursued for inclusion into the voluntary DCP by industry organizations. Such strategy is likely to result in estimates of incidence and prevalence measures that are higher than in the source population. Because of the inherently open nature of the voluntary DCP, we had to modify the formulae for calculation of incidence risk. It could be argued that presumed cases are not diagnostically confirmed and are therefore subject to misclassification. However, the reality of a voluntary DCP for production-limiting diseases is that resources to conduct large-scale testing are scarce and, as such, need to be carefully deployed. Furthermore, in order to confirm premises as presumed negative, diagnostic testing to confirm absence of infection at the design prevalence level is still required. Also, the criteria to declare confirmed negative status is arguably open-ended and could be further improved. In conclusion, this study provides estimates of incidence and prevalence measures in Ontario based on industry data collected through voluntary disease control programmes. The data suggest that annual incidence risk and prevalence estimates are low and have been steadily decreasing between 2014 and 2016 for PEDV and PDCoV. Current estimates of disease frequency support planning of disease elimination at the provincial level, but much information should be available about factors that led to time to elimination in individual herds. In addition, our evaluation of surveillance components indicates that the two surveillance components were complementary and focused on different aspects of surveillance. OMAFRA surveillance was mostly focused on identification of primary cases aimed at quick disease investigations and traceability in the face of the outbreak, whereas OSHAB surveillance has the added benefit of having sufficient data that allow long-term evaluation of disease trends, long-term disease management and tracing disease status of individual herds over time. The OSHAB voluntary DCP database also provides a good tool for calculating weekly prevalence and incidence measures, which is a valuable statistic for producers and animal health experts during all phases of disease outbreak and control. Descriptive analysis and spatial epidemiology of porcine reproductive and respiratory syndrome (PRRS) for swine sites participating in area regional control and elimination programs from 3 regions of Ontario Weight of the evidence linking feed to an outbreak of porcine epidemic diarrhea in Canadian swine herds The changing face of the Canadian hog industry Porcine Epidemic Diarrhea (PED) situation in Canada -Animals Canadian Food Inspection Agency Porcine epidemic diarrhoea: New insights into an old disease Isolation and characterization of porcine epidemic diarrhea viruses associated with the 2013 disease outbreak among swine in the United States Danish swine salmonellosis control program R: A language and environment for statistical computing Veterinary epidemiologic research Pub. L. No. 09a31. Government of Ontario Emergence of porcine epidemic diarrhoea in North America Porcine epidemic diarrhea virus infection: Etiology, epidemiology, pathogenesis and immunoprophylaxis. Veterinary Canada: Porcine epidemic diarrhea in Canada: an emerging disease case study Role of transportation in spread of porcine epidemic diarrhea virus infection, United States Origin, evolution, and virulence of porcine deltacoronaviruses in the United States PED -A Canadian Update. Paper presented at the 15th London Swine Conference Ontario Ministry of Agriculture Food and Rural Affairs Herd Health: Porcine Epidemic Diarrhea Virus (PED) On-Farm Testing -Total Confirmed Cases Ontario Swine Health Advisory Board (OSHAB) Porcine Epidemic Diarrhea Virus (PEDV) and Porcine Delta Coronavirus (PDCoV) Position Statement Investigation of Factors That Led to Emergence of PEDV through Feed During Early Phase of Canadian Outbreak Investigation into the role of potentially contaminated feed as a source of the first-detected outbreaks of porcine epidemic diarrhea in Canada Outbreak investigation of porcine epidemic diarrhea in swine in Ontario RODBC: ODBC Database Access. R package version 1.3-14 Surveillance and monitoring systems for animal health programs and disease surveys Emergence of severe porcine epidemic diarrhoea in pigs in the USA zoo: S3 infrastructure for regular and irregular time series Program. Authors are grateful to participating industry organizations and producers for providing access to data. http://orcid.org/0000-0003-1034-810XR. Dara http://orcid.org/0000-0002-3728-0275 Additional Supporting Information may be found online in the supporting information tab for this article.