Outbreak of Severe Vomiting in Dogs Associated with a Canine Enteric Coronavirus, United Kingdom

The lack of population health surveillance for companion animal populations leaves them vulnerable to the effects of novel diseases without means of early detection. We present evidence on the effectiveness of a system that enabled early detection and rapid response a canine gastroenteritis outbreak in the United Kingdom. In January 2020, prolific vomiting among dogs was sporadically reported in the United Kingdom. Electronic health records from a nationwide sentinel network of veterinary practices confirmed a significant increase in dogs with signs of gastroenteric disease. Male dogs and dogs living with other vomiting dogs were more likely to be affected. Diet and vaccination status were not associated with the disease; however, a canine enteric coronavirus was significantly associated with illness. The system we describe potentially fills a gap in surveillance in neglected populations and could provide a blueprint for other countries.

using secondary data, particularly from pet insurance providers (4). More recently, researchers have exploited the rapid digitization of electronic health records (EHRs) for passive surveillance. Data can be collected at great scale and analyzed in nearreal time. EHR data are now routinely used in human heath efforts (5)(6)(7)(8), in which their timeliness, simplicity, and breadth of coverage complements surveillance based on diagnostic data (9,10). Such approaches are beginning to find healthcare value in veterinary species, especially among companion animals (4,(11)(12)(13), a high proportion of which visit veterinarians (14).
In January 2020, one of the authors of this article (D.G.), a primary care veterinarian in northwest England, contacted the other authors about seeing an unusually high number of cases (≈40) of severe vomiting in dogs; responses to a social media post suggested other veterinarians might have been experiencing similar events. Vomiting is a common complaint among dogs whose owners seek treatment for them (15,16). However, documented outbreaks are rare because established vaccines are available for most common known pathogens (17). In the absence of robust populationwide data, such sporadic reports frequently do not raise awareness of outbreaks.
For the response we describe, we obtained data from syndromic surveillance and text mining of EHRs collected from sentinel veterinary practices and diagnostic laboratories, which we then linked with data from field epidemiology and enhanced genomic testing. In 8 weeks, using this approach, we described the temporal and spatial epidemiology, identified a possible causative agent, and provided targeted advice to control the outbreak. Ethics approval was given by Liverpool University Research Ethics Committees (Liverpool, UK; VREC922/RETH000964).

Data Sources Veterinary Practices
During March 17, 2014-February 29, 2020, we collected data from 7,094,397 consultation records (4,685,732 from dogs and 1,846,493 from cats) from EHRs from the Small Animal Veterinary Surveillance Network (SAVSNET), a volunteer network of 301 veterinary practices (663 sites) in the United Kingdom, recruited based on convenience (11). In brief, EHRs included data collected during individual consultations on species, breed, sex, neuter status, age, owners' postcodes, and vaccination status. Each EHR is also compulsorily annotated by the veterinary clinician with a main presenting complaint (MPC) at time of visit, using a questionnaire window embedded in the practice management system. Options for reasons for visit included gastroenteric, respiratory, pruritus, tumor, kidney disease, other unwell, post-op check, vaccination, or other healthy.
Given that severe vomiting was a key outbreak feature, we undertook 2 complementary analyses. First, we used regular expressions to identify clinical narratives describing frequent vomiting, but excluded common false positive search results (Appendix Table 1, https://wwwnc.cdc.gov/EID/article/27/2/20-2452-App1.pdf). Second, we used data on prescriptions to describe the frequency of all veterinary-authorized products containing the antiemetic maropitant (18). We calculated trend lines using Bayesian binomial generalized linear modeling trained on weekly prevalence during 2014-2019 (19), which allowed us to identify extreme (>99% credible interval [CrI]) or moderate (>95% CrI) observations. Laboratories SAVSNET also collects EHRs from participating diagnostic laboratories on samples submitted from more than half of UK veterinary practices. Canine diagnostic test results from January 2017 through February 2020 were queried from 6 laboratories for 6 gastroenteric pathogens. Test numbers, percentage of positive results, and associated 95% CIs were summarized ( Table 1). The number of sites was surmised from the submitting practices' postcodes.

Questionnaires
Online questionnaires to enable case reporting were made available to both veterinarians and owners beginning January 29, 2020. The required case definition of >5 vomiting episodes in a 12-hour period was based on clinical observations of early cases. Veterinarians were also asked to complete control questionnaires. Initially, we requested only controls matched to veterinary practices contributing case data; however, to increase recruitment, a nonmatched control questionnaire open to any veterinarian was deployed on February 5. The questionnaires (Appendix) requested a range of information including owner postcode, animal signalment, vaccination status, clinical signs, treatment and diagnostic testing, animal contacts, diet, and recovery status.
We performed all statistical analyses using R version 3.6.1 (https://cran.r-project.org). Case details were described for both veterinarian-and ownerreported data. We calculated proportions and 95% CIs for categorical variables and median and range for continuous variables. We constructed univariable and multivariable mixed-effects logistic regression models using data submitted by veterinarians using R package lme4. Explanatory variables from univariable logistic regression were considered in multivariable models for likelihood ratios of p≤0.20, which underwent manual stepwise backward elimination to reduce Akaike's and Bayesian information criteria. Practice was included as a random effect. We assessed confounding by the effect on model fit with sequential removal of variables and assessed 2-way interaction terms for improved model fit. We defined final statistical significance as p <0.05.

Spatiotemporal Analysis of Cases
We obtained records of consults weekly during November 4, 2019-March 21, 2020; cases were geolocated by pet owners' postcodes. We considered records of gastroenteric MPC as a binary outcome (i.e., 1 for gastroenteric consult, 0 for nongastroenteric consult). We used a logistic geostatistical model to investigate spatial clustering of cases for each week. We defined a spatial hotspot as a location having 95% posterior probability of prevalence exceeding the national mean prevalence over any 1-week period. With no discernible epidemic wave apparent over successive weeks, we aggregated weekly measures across the study period to show the number of weeks each location was a hotspot (Appendix).

Sample Collection, PCR, and Phylogenetic Analyses
Veterinarians submitting questionnaires were also asked to submit samples for microbiological testing including mouth swabs, fecal samples, and for gastrointestinal cases, vomit. In brief, we extracted nucleic acids using a QIAGEN QIAamp viral RNA kit (https://www.qiagen.com), reverse transcribed samples using ThermoFisher Superscript III (https:// www.thermofisher.com), and tested for canine enteric coronavirus (CeCoV) by M-gene PCR (20). To expedite results and reduce contamination risks, the PCR was run as a single-stage PCR rather than as the published nested reaction. We purified positive samples using QIAquick (QIAGEN) and sequenced them bidirectionally (Sanger sequencing; Source Biosciences, https://www.sourcebioscience.com) to produce consensus sequences (ChromasPro 2.1.8, http:// technelysium.com.au).
To rapidly explore the potential involvement of other viruses, we extracted nucleic acid from 19 random cases and 5 controls for deep sequencing. RNA was amplified by sequence-independent, single-primer-amplification (21), multiplexed libraries were prepared using 30 ng of cDNA with an Oxford Nanopore SQK-LSK109 ligation sequencing kit (Oxford Nanopore, https://nanoporetech.com) and sequenced using an Oxford Nanopore MinION Mk1B device for 48 hours. To perform real-time fast basecalling, we used the Oxford Nanopore MinKNOW Guppy toolkit and FASTQ files uploaded to an Oxford Nanopore EPI2ME data analysis platform for identification.

Syndromic Surveillance
On the basis of MPCs identified in the EHRs, we found a specific and significant increase in the number of dogs recorded as exhibiting gastroenteric signs; the final 10 weeks, during December 2019-March 2020, were outside the 99% CrI (extreme outliers; Figure 1, panel A). A similar trend was observed in maropitant therapy for dogs ( Figure 1, panel B). Both measures, peaked in the week ending February 2, 2020, at approximately double the preceding baseline. We observed no similar trends for respiratory disease in dogs, for gastroenteric MPCs, for maropitant treatment in cats ( Figure 1, panels C-E), or for antibiotic use in dogs (data not shown), together suggesting the signal was specific to canine gastroenteric disease, a finding supported by similar increases in the regular expression identifying vomiting dogs (Figure 1, panel F).
Spatiotemporal mapping of weekly cases of gastroenteric MPC showed prevalence was spatially clustered ( Figure 2). In particular, locations in northwest and southwest England and in Edinburgh, Scotland, had strong evidence of many weeks of prevalence higher than the national mean.

Diagnostic Tests
The patterns of test results for different PCR tests, generally carried out concurrently, were broadly similar ( Figure 3, panels A-C). The same was true for results based on cultured samples ( Figure 3, panels D, E). Of particular interest, CeCoV showed strong seasonality, positive tests peaking during the winter months ( Figure 3, panel A). However, similar peaks seen in previous years suggested the observed peak in February 2020 could not itself explain this outbreak.

Questionnaire
By March 1, 2020, a total of 1,258 case questionnaires had been received. After excluding 59 questionnaires missing key data, we used data from 165 veterinaryreported cases, 1,034 owner-reported cases (Table 2), and 60 veterinary-reported controls (Appendix Table  2) for analyses.
Most cases were from households in England (Table 2). Median case age at examination was 4.0 years (range 0.3-15.0 years) based on veterinary reports and 4.8 years (range 0.2-15.5 years) based on owner reports. Most animals had been vaccinated against core pathogens (17) and leptospirosis within the preceding 3 years and dewormed within the previous 3 months. A range of breeds (data not presented) were observed, broadly corresponding to previous studies (6). Most cases were fed dog food, but  Table 3). Approximately half of cases reported diarrhea, most without blood. Diagnostic testing was performed in 32.1% of veterinary-reported cases, most (78.9%) using hematology or biochemistry assays, or both.
Dogs in >90% of veterinary-reported cases were treated, compared with in 61.7% of owner-reported cases. In both, antiemetics were most often prescribed: in 89.1% (CrI 84.3%-93.9%) of veterinary-reported cases and in 48.1% (CrI 45.0%-51.1%) of owner-reported cases. The most common recovery time was Consults were geolocated to owners' postcodes, with gastroenteric main presenting complaint as a binary outcome (1 for gastroenteric consult, 0 for a nongastroenteric consult). Colored areas represent the number of weeks a given location had a 95% posterior probability of prevalence exceeding the national mean prevalence in any week. The geostatistical modeling approach used is further detailed in the Appendix (https://wwwnc. cdc.gov/EID/article/27/2/ 20-2452-App1.pdf).

3-7 days
; the dogs died in 0.6% of veterinary-reported and 1.0% of owner-reported cases.
Descriptive data about the control population, submitted by veterinarians, and univariable findings from analyses of the veterinary case controls are presented in Appendix Tables 2 and 3; multivariable findings are shown in Table 4. Both neutered and nonneutered male dogs were at significantly increased odds of contracting the illness, compared with neutered females, as were dogs living in the same household as another dog that had also been vomiting compared to those in households where other dogs were healthy. However, dogs living in a single-dog household were at increased odds of contracting the illness compared with dogs living in the same household as another dog that had not recently vomited. Dogs that had been in recent contact with another animal species (including humans) that had recently vomited were at reduced odds of vomiting, compared with those who had not. Other potential causes considered early in the outbreak, including foodborne etiologies, vaccine preventable diseases, or the possibility of interspecies transmission, were not significantly associated (Appendix Table 3).
We gathered useable M-gene sequences from 21 samples (16 dogs). When we sequenced 2 samples from the same animal, the sequences were identical and subsequently represented only once in analyses ( Figure 4). All sequences clustered with previously reported type II CeCoVs (22)   matched a canine rotavirus (1 case, 1 control; data not presented). Consistent with M-gene sequencing, 5 of the CeCoV genomes clustered together (>99% similarity), distinct from the genome from dog 15 ( Figure  4). The outbreak strain was most similar to a virus from Taiwan isolated in 2008 from a young dog with diarrhea (94.5% similarity; L. Chueh, pers. comm. [email] Apr. 27, 2020) and did not show any obvious sequence differences to published strains that might explain the unusual pattern of disease observed in the outbreak. Based on spike gene analyses, the outbreak strain clustered with IIb, having a TGEV-like N-terminal spike domain (23). Sequences were submitted to GenBank (accession nos. MT877072, MT906864, and MT906865).

Discussion
Using EHRs annotated with syndromic information by veterinarians, we rapidly identified an outbreak of canine gastroenteric disease that had started in November 2019. This finding was corroborated by parallel increases in relevant prescriptions and records of frequent vomiting. Those data were augmented by data from responses to a questionnaire, diagnostic laboratories, and enhanced microbiological analyses.
This system enabled us to determine case definitions and outcomes and to identify risk factors as well as a potential viral cause, within a 3-month period; findings were rapidly disseminated to veterinarians (24,25) and owners. This combined approach represents an efficient system that can fill a previously neglected national population health surveillance need for companion animals.
The first indication of an outbreak came from time-series analyses of syndromic data. Such syndromic surveillance is increasingly being used to monitor the impact of national events like natural disasters and bioterrorism on human population health, as well as changes in gastroenteric and influenza-like illness (6)(7)(8)(9). Such data can be simple to collect, provide real-time wide geographic coverage, and be flexibly applied to different conditions (10,11). Although in some cases these data can identify outbreaks earlier than more active surveillance, their predictive value can sometimes be low, particularly where there is a low signal to noise complaint ratio. In our case, the outbreak was large compared with background levels, associated with near doubling of the gastroenteric syndrome, and had many weeks in which the syndrome statistically exceeded the baseline.
The richness of data within EHRs enabled us to validate this outbreak using numbers of antiemetic prescriptions and text mining. Prescription data have been used to understand, for example, human health inequalities (26), and the use of critical antimicrobials in both humans (27) and animals (28,29). We used  these data to identify and track an outbreak, benefitting from a clear link between the syndrome (vomiting) and its therapy (antiemetic). It will be useful to identify other disease-therapy associations that could be used for similar surveillance.
We used text mining to identify records of frequent vomiting in clinical narratives. Such approaches can circumvent the need for practitioner-derived annotation and be flexibly and rapidly adapted to emerging syndromes as soon as case-definitions are determined. Similar approaches have been described in human health for conditions such as fever (30-32) but can suffer low sensitivity (31). Indeed, the outbreak peak based on text mining was ≈20% of that based on MPC analysis. However, it is also likely the outbreak as defined by the MPC included a considerable number of animals with milder signs that would not be detected by data mining using the regular expression developed here. Although data from text mining are unlikely to give an accurate estimate of the true prevalence of a given condition, they can still be used to track outbreaks.
To compliment syndromic surveillance, we implemented a rapid case-control study, collecting >1,200 responses from veterinarians and owners in 4.5 weeks. There was no evidence for similar disease in people or other species. The timing of the outbreak as shown by case data was in broad agreement with our syndromic surveillance. Questionnaires from owners and veterinarians were in broad agreement on date of onset, geographic density, clinical signs, and recovery. These data informed targeted health messages posted online and on social media on February 28, 2020, 4 weeks after we first became aware of the outbreak.
Clearly, evidence of transmission driving the outbreak was vital to providing disease control advice. Dogs in multidog households were more likely to vomit if other dogs in the household were also affected, suggesting either transmission between dogs or a common environmental source; these observations informed advice to the public around isolating affected dogs. Of note, dogs in single-dog households were also at increased odds of being affected compared to multidog households where only a single dog was vomiting. Some authors have shown that dogs from single-dog households are walked more and therefore could be at greater risk for infection (33). Factors affecting dog walking are clearly likely to be important for control of infectious disease transmission and should be explored further.
In addition to collecting epidemiologic data, we collected microbiological samples from cases and controls. Based on its known (34) and observed seasonality ( Figure 3, panel A), we tested all samples for CeCoV. Cases were significantly more likely to show positive results both when all samples (oral swabs, feces and vomit) were considered or when just fecal samples were considered, suggesting a possible role for CeCoV in the outbreak. However, many case samples tested negative: 33 of 50 overall, 6 of 16 dogs for which feces samples were submitted, and 7 of 13 dogs for which vomit samples were submitted. There are several potential reasons for these negative findings, including the sensitivity of the PCR, the high numbers of oral swabs (although simpler to collect, oral swabs were more likely to test negative), the timing of samples in relation to viral shedding, and the storage and transport of samples. In addition, it is important to note that our case definition, based as it was on a syndrome and lacking more specific confirmatory testing, is likely to include some animals that were not part of the outbreak. Indeed, at its peak, the outbreak only doubled the background level of gastroenteric disease seen at other times of the year; therefore, we might expect only half of our cases to be truly associated with the outbreak.
Sequencing results identified a predominant CeCoV strain in outbreak cases across the United Kingdom, in contrast with earlier studies showing that CeCoV strains tend to cluster in households, veterinary practices, or local areas (35). This finding lends further support to the role of this strain in the observed outbreak. In Sweden, a single strain was also implicated in several small wintertime canine vomiting outbreaks (36); genetically, however, the virus strain we identified was distinct from the strain from Sweden (data not shown). Ultimately, it will be necessary to perform a challenge study to confirm or refute the role of this CeCoV strain as the cause of this outbreak, as well as to explore the range of clinical signs associated with infection.
If this strain is proven to be the cause of the outbreak, several features mark the observed pattern of disease as unusual, including the outbreak scale, its geographic distribution, the severity of signs in some animals, a lack of notable viral co-infections, and the involvement of adult dogs. CeCoV is generally associated with mild gastroenteritis (37). Although sporadic outbreaks of more severe hemorrhagic diseases with high mortality (38-40), as well as systemic diseases (41,42), have been reported, these typically affect individual households, and are often associated with mixed infections (43). Such observations suggest that the genetic variability of CeCoVs may affect virulence and are supported by experimental infections recreating more severe disease (38). The genetic mechanism underlying such shifts in virulence in CeCoV have not been defined. However, mutations impacting virulence are described in closely related alphacoronaviruses (44-47).
In conclusion, this multidisciplinary approach enabled a rapid response to a newly described outbreak of canine gastroenteritis and identified a CeCoV as a potential cause. Previous CeCoV seasonality suggests further outbreaks may occur. Having such an efficient surveillance system provides the ideal platform to inform and target population health messaging. Several challenges remain for addressing the lack of national population health structures for companion animals: to systematically capture discussions of disease in social and mainstream media; to sustainably fund these activities, which currently are largely resourced by research grants; to understand and broaden the representativeness of such sentinel networks; and to link surveillance information with agencies empowered to act (12).

About the Author
Dr. Radford is a professor of veterinary health informatics at the University of Liverpool. His primary research interests are the molecular epidemiology of viral pathogens, particularly those of veterinary importance, and combining this subject with electronic health data to study animal diseases at a population level and their impact on people.

Supplementary Information on Geostatistical Modelling
The geostatistical model used to investigate spatial clustering for severe vomiting in dogs makes use of owner-geolocated prevalence data based on total consults recorded in SAVSNet.
Below, we first describe the geostatistical model setup, before describing how the results were presented using geographical information systems methods. ( ) is a spatial Gaussian process such that

Geostatistical Model for Prevalence
2 is a covariance matrix defined by a Matérn correlation function: where � − � is the Euclidean distance between locations and , 2 is the sill variance of the spatial Gaussian process, and is the length scale (1).
The computation of the log posterior probability density for this model involves the inversion of 2 which becomes computationally prohibitive beyond a few hundred points. Since in a typical week ≈ 24000, we use the inducing point approximation of Banerjee et al. (2).
Here, we choose a set of knot points ⋆ , = 1, . . . , and let where ⋆ is a realisation of the Gaussian process at knots ⋆ . In practice, we find that 300 knot points positioned using K-means clustering on gives satisfactory computational performance with negligible information loss compared to 600 and 900 knot points positioned similarly.
Finally, we investigated the requirement for a "nugget", or uncorrelated, random effect by adding a variance component to the diagonal of 2 , i.e. 2 = 2 + 2 . However, this did not improve the model fit and was removed for the sake of parsimony.
This model was fitted to the consulting data in a Bayesian framework. The following prior distributions were chosen to reflect relative a priori ignorance about parameters:

GIS Presentation of Results
Using Equation (1)

Potential Outbreak Investigation: Prolific Vomiting in Dogs
You are being invited to participate in an outbreak investigation study, following reports of an outbreak of prolific vomiting in dogs. Before you decide whether to participate, it is important for you to understand why the survey is being conducted and what it will involve if you do choose to take part. Please consider the following information. Epidemiologist contact details are listed below should you have any further questions.
Reading this information sheet and completing the survey will be considered as consent to participate in this survey.

What is the purpose of the survey?
This survey has been created in order to collect more detailed case information, following veterinary surgeon and social media reports of a potential outbreak of prolific, acute vomiting in dogs during December 2019 and January 2020.

Why am I being invited to take part and what will happen if I take part?
You are being invited to take part because you are a veterinary surgeon or owner currently working in a companion animal-treating veterinary practice or an owner, in the United Kingdom, who has potentially identified a case fitting the case definition of "dog with acute onset of prolific vomiting, with 5 or more episodes of vomiting within a 12 hour period".
If you decide to take part you will need to complete the online survey, which will take around 10 minutes.
Participation is voluntary and you do not have to take part in this study. You are free to withdraw at any time until you have selected the 'finish' button on the final page of the questionnaire. You do not have to give a reason if you do not wish to take part.
If you are willing, we will also request your postcode, name and email address so that we can ask for further case details if this becomes necessary during the potential outbreak investigation. We will only use your name and email for the purpose of seeking further information, and will destroy data containing these personal identifiers on conclusion of the survey.

Are there any benefits or risks in taking part?
There are no direct benefits or risks to you or your practice associated with taking part in this survey, but we will use the data to further characterise this potential outbreak, and if necessary assist in controlling the potential outbreak.

What will happen if I want to stop taking part?
If you want to stop taking part in this survey you can withdraw at any time until completion and submission of the online survey.

How will my data be used?
The data you provide will be stored securely for up to 7 years in line with data protection requirements at the University of Liverpool and GDPR. All data is strictly confidential and only researchers involved in the study will have access to it. Fully anonymised data may be archived for use in other research projects in the future. Under UK data protection legislation, the University acts as the Data Controller for personal data collected as part of the University's research. The Principal Investigator acts as the Data Processor for this study.

What will happen to the results of the survey?
The data will be used to further characterise the potential outbreak of prolific vomiting in dogs, potentially assisting in identifying causative factors and informing attempts (if necessary) to control this potential outbreak. Anonymised results may also be published -you and your clients (if relevant) will never be identifiable.

What if I am unhappy or if there is a problem?
If you are unhappy, or if there is a problem, please feel free to contact the epidemiologists listed below and we will try to help. 1. Please confirm that you have read and understood the above information and confirm your consent for data to be used for these purposes, as the owner or on behalf of the owner.
-I confirm that I have consent from the owner to collect and submit these data, and I understand that anonymised data may be used in publications -I confirm that I am the dog's owner, give consent for collection and submission of these data, and I understand that anonymised data may be used in publications

Basic Case information
We are firstly going to ask some basic information pertaining to the case of canine prolific vomiting you would like to report.

Are you describing a current or retrospective vomiting case?
-Current (dog vomited 12 hours or less before completion of survey) -Retrospective (dog last vomited over 12 hours ago) -Don't know 2a. If describing a retrospective case, please state date of onset of vomiting: -Free text response box Current cases:

In the last 12 hours before completion of this survey, how many times has the dog vomited?
-Less than five times -Five times or more* Retrospective cases:

When the dog was vomiting most frequently, approximately how many times did the dog vomit over a 12 hour period?
-Less than five times -Five times or more* * Only participants who selected 'five times or more' in questions 3 or 4 (hence describing a case fitting the case definition) were able to proceed with answering the remaining questions in this survey.

Which of the following statements best describes yourself:
-I am a veterinary surgeon wishing to report a potential case of prolific vomiting in a dog under my care (1) -I am an employee of a veterinary practice wishing to report a potential case of prolific vomiting in a dog (2) -I am a dog owner / main keeper wishing to report a potential prolific vomiting case in my own dog (3) -Other (4)* 5a. If you selected Other, please specify: -Free text response box * Only participants selecting 'Other' in Question 5 were able to answer Question 5a.
Participants selecting options (1) and (2) on question 5 were routed towards the 'Veterinary Professional Questionnaire', whereas those selecting (3) and (4) were routed towards the 'Owner Questionnaire'. These two sub-questionnaires are outlined on the following pages.

Practice: Case Details
This section will ask more details about the dog and veterinary practice under which (s)he is registered.

Please provide the name of the veterinary practice under which the dog is registered:
-Free text response box

Please provide the postcode of the veterinary practice under which the dog is registered:
-Free text response box

Please provide the phone number of the veterinary practice under which the dog is registered:
-Free text response box

Please provide the email address of the veterinary practice under which the dog is registered:
-Free text response box

Does the veterinary practice in which the dog is registered currently participate in the Small Animal Veterinary Surveillance Network (SAVSNET)?
-Yes -Free text response box

Practice: Control Cases
When investigating a potential disease outbreak, it is important to collect information relating to a population of animals NOT exhibiting clinical signs associated with the outbreak under investigation (the 'control population'). If possible, please complete some further questions relating to a randomly selected dog NOT exhibiting vomiting clinical signs that presented at your veterinary practice on the same day the affected animal presented e.g. the next nonvomiting dog you see where the owner is happy to participate.
1. Please confirm that you are able, and willing, to provide information regarding a control dog that has not reported to the veterinary practice with vomiting clinical signs within the last month.
-I am willing and able to provide information on a non-vomiting control dog* -I am NOT willing or able to provide information on a non-vomiting control dog * Only participants who selected 'I am willing and able to provide information on a nonvomiting control dog' in Question 1 were able to proceed with answering the questions pertaining to a control animal in this survey. Those who were not willing were directed to submit the case questionnaire details they had provided alone.   This section will ask more details about the dog and veterinary practice under which (s)he is registered.

Please provide the name of the veterinary practice under which the dog is registered:
-Free text response box

Please provide the postcode of the veterinary practice under which the dog is registered:
-Free text response box

Does the veterinary practice in which the dog is registered currently participate in the Small Animal Veterinary Surveillance Network (SAVSNET)?
-Yes  Reading this information sheet and completing the survey will be considered as consent to participate in this survey.

What is the purpose of the survey?
This survey has been created in order to collect more detailed CONTROL information, following veterinary surgeon and social media reports of a potential outbreak of prolific, acute vomiting in dogs during December 2019 and January 2020.

Why am I being invited to take part and what will happen if I take part?
You are being invited to take part because you are a veterinary surgeon or owner currently working in a companion animal-treating veterinary practice or an owner, in the United Kingdom, who is willing to provide information on CONTROL dogs, as part of an ongoing investigation concerning dogs with acute onset of prolific vomiting, with 5 or more episodes of vomiting within a 12 hour period". If you would like to submit information about a CASE, please click here.
If you decide to take part you will need to complete the online survey, which will take around 10 minutes.
Participation is voluntary and you do not have to take part in this study. You are free to withdraw at any time until you have selected the 'finish' button on the final page of the questionnaire. You do not have to give a reason if you do not wish to take part.
If you are willing, we will also request your postcode, name and email address so that we can ask for further CONTROL details if this becomes necessary during the potential outbreak investigation. We will only use your name and email for the purpose of seeking further information, and will destroy data containing these personal identifiers on conclusion of the survey.

Are there any benefits or risks in taking part?
There are no direct benefits or risks to you or your practice associated with taking part in this survey, but we will use the data to further characterise this potential outbreak, and if necessary assist in controlling the potential outbreak.

What will happen if I want to stop taking part?
If you want to stop taking part in this survey you can withdraw at any time until completion and submission of the online survey.

How will my data be used?
The data you provide will be stored securely for up to 7 years in line with data protection requirements at the University of Liverpool and GDPR. All data is strictly confidential and only researchers involved in the study will have access to it. Fully anonymised data may be archived for use in other research projects in the future. Under UK data protection legislation, the University acts as the Data Controller for personal data collected as part of the University's research. The Principal Investigator acts as the Data Processor for this study.
What will happen to the results of the survey?
The data will be used to further characterise the potential outbreak of prolific vomiting in dogs, potentially assisting in identifying causative factors and informing attempts (if necessary) to control this potential outbreak. Anonymised results may also be published -you and your clients (if relevant) will never be identifiable.

What if I am unhappy or if there is a problem?
If you are unhappy, or if there is a problem, please feel free to contact the epidemiologists listed below and we will try to help. If you remain unhappy or have a complaint which you feel you cannot communicate directly to the researcher then you should contact the Research Ethics and Integrity Office on 0151 794 8290 (ethics@liv.ac.uk). When contacting the Research Governance Officer, please provide details of the name or description of the study (so that it can be identified), the researcher involved, and the details of the complaint you wish to make. Please confirm that you have read and understood the above information and confirm your consent for data to be used for these purposes, as the owner or on behalf of the owner.
-I confirm that I have consent from the owner to collect and submit these data, and I understand that anonymised data may be used in publications -I confirm that I am the dog's owner, give consent for collection and submission of these data, and I understand that anonymised data may be used in publications Basic CONTROL information 1. Which of the following statements best describes yourself: -I am a veterinary surgeon wishing to provide information about a control dog -I am an employee of a veterinary practice wishing to provide information about a control dog -I am a dog owner / main keeper wishing to provide information about a control dog -Other * Only participants selecting 'Other' on Question 1 were able to answer questions 1a.
1a. If you selected Other, please specify: -Free text response box

Control Cases
When investigating a potential disease outbreak, it is important to collect information relating to a population of animals NOT exhibiting clinical signs associated with the outbreak under investigation (the 'control population'). If possible, please complete some further questions relating to a randomly selected dog NOT exhibiting vomiting clinical signs.
2. Please confirm that you are able, and willing, to provide information regarding a CONTROL dog that has NOT exhibited vomiting clinical signs within the last month.
-I am willing and able to provide information on a non-vomiting control dog* Page 28 of 31 -I am NOT willing or able to provide information on a non-vomiting control dog * Only participants who selected 'I am willing and able to provide information on a nonvomiting control dog' in Question 2 were able to proceed with answering the questions pertaining to a control animal in this survey.