Clinical and epidemiologic features of SARS-CoV-2 in dogs and cats compiled through national surveillance in the United States

Objective To characterize clinical and epidemiologic features of SARS-CoV-2 in companion animals detected through both passive and active surveillance in the U.S. –


Introduction
Pet ownership provides many documented positive impacts, including improvements to mental health 1,2 . In the United States, a 2021-2022 survey reported that approximately 70% or 90.5 million households owned at least one pet 3 , with around 23 million U.S. households acquiring a pet during the rst year of the COVID-19 pandemic (March 2020-May 2021) 4 . Owners and their pets commonly have close relationships, often eating, sleeping, snuggling, and recreating together 5 . While these close interactions have many bene ts, they also pose a risk for zoonotic disease transmission. However, the extent of surveillance efforts to detect zoonotic disease transmission in companion animals, including SARS-CoV-2, is limited at both the national and global level.
Similar to other coronaviruses, it is now evident that SARS-CoV-2 has a broad mammalian host range 6 .
As of July 12, 2022, 35 countries have reported SARS-CoV-2 infections in species from 14 mammalian families to the World Organisation for Animal Health (WOAH) 7 . Susceptible animals can be categorized into four groups by the nature of their interaction with people: companion animals, farmed animals (including mink 8 and cervids 9 ), free-ranging wildlife [10][11][12] , and exotic animals (including big cats and nonhuman primates) in zoos, sanctuaries, and aquaria 13 . Companion animals are the second-most commonly reported animal group to be infected with SARS-CoV-2 after farmed mink 8 , comprising 60% (n = 399; cats: 205, dogs: 191, hamsters: 11, ferrets: 3) of all animals reported globally to WOAH between February 29, 2020 and December 31, 2021 14 .
Here, we use the largest compilation of zoonotic SARS-CoV-2 surveillance data available globally to synthesize the epidemiological and clinical features of SARS-CoV-2 in companion animals, speci cally dogs and cats, residing in the United States.

Materials And Methods
Identifying companion animals con rmed positive for SARS-CoV-2 In the United States, animal cases of SARS-CoV-2 are identi ed by passive or active surveillance. Through passive surveillance, case identi cation is typically initiated when owners bring animals to veterinary clinics or hospitals, and samples are submitted to a variety of veterinary diagnostic laboratories (governmental, university, and private) for SARS-CoV-2 testing. Through active surveillance, animals with a known SARS-CoV-2 exposure or clinical signs compatible with SARS-CoV-2 infection are actively sought out by health o cials or researchers. These include collaborative One Health investigations of SARS-CoV-2 transmission among animals and people in households, animal shelters, animal rescues, animal rehabilitation centers, zoos, or veterinary clinics [15][16][17]  We categorized clinical signs into 3 categories based on body systems affected: respiratory (cough, di culty breathing or shortness of breath, sneeze, nasal discharge, ocular discharge), gastrointestinal (vomiting, diarrhea), and non-speci c (lethargy, inappetence, fever). Clinical signs were described in 4 subsets: 1 -clinical vs subclinical among all con rmed positive companion animals; 2 -clinical vs subclinical based on surveillance detection method (active or passive surveillance); 3 -detailed clinical presentation among companion animals presenting with clinical signs; and 4 -detailed clinical presentation among companion animals presenting with clinical signs by species.

Multi-pet Households
In some instances, there was more than one companion animal known to live in a household. To assess the likelihood that another animal in the household would become infected following the rst, we calculated conditional probability, subset to only animals detected through passive surveillance (where an index pet could be identi ed). Conditional probabilities were calculated based on whether the index pet was a cat, dog, or either, and whether the secondary animal was a cat, dog, or either.

Diagnostics
To understand timeline of infection and immune response of companion animals infected with SARS-CoV-2, we assessed Ct values (the number of cycles necessary for viral nucleic acid detection, with lower values indicating higher viral load) and VN titers (a measure of neutralizing antibody levels, indicating immune response) after exposure to a person with COVID-19 were assessed. Sampling days were calculated as the number of days between a person's symptom onset or date of positive SARS-CoV-2 test and the animal's sample collection date. In animals with clinical signs, the length of observable illness was de ned as the number of days between onset and resolution of clinical signs. While extremely rare, a small number of animals died while positive for SARS-CoV-2, documented in Carpenter et al. 2021 25 , and were therefore excluded from analyses. In analyses or visualizations that used Ct values, the lowest Ct value obtained from respiratory swabs (nasal & oral), collected on the same day was used. Nonrespiratory samples such as fecal and rectal samples were excluded, since the viral load in these sample types is commonly low (i.e., higher average Ct values) compared to respiratory samples 26 . Conjunctival swabs were infrequently collected, but since their diagnostic e cacy has not been evaluated, were excluded. Fur swabs were also omitted from analyses, since they are used as indicators of environmental contamination and not infection. SARS-CoV-2 RT-PCR results, measured by cycle threshold (Ct), and SARS-CoV-2 VN titers were compared to the presumed date of exposure, measured as the reported date of human symptom onset or positive human test. The lowest Ct values among respiratory swabs (oral and nasal) and geometric mean titer were calculated in 2-day intervals with 95% con dence intervals. Average Ct values and log-transformed geometric mean virus neutralizing antibody titers were calculated for con rmed animals, by species. Analysis was conducted for all observations, as well as a strati ed analysis by species. To account for the inherent rise and fall of viral nucleic acid and neutralizing antibody, a polynomial function was applied to detect trends in Ct values and VN titer over time since presumed exposure. As early sampling in the rst several days after exposure was rarely conducted, xed axis points were applied to the regression model to re ect a Ct of 40 (no viral nucleic acid present) and VN of 0 on the day of likely exposure (Day 0). Two animals with Ct values < 38 on the day of presumed exposure (Day 0) were excluded from analysis, as it is not biologically plausible to have measurable infection or immunity this early after exposure. It is likely that the date of presumed exposure occurred earlier than what was re ected in the epidemiologic investigation for these two animals.
Variant analysis was conducted using whole-genome sequencing results from USDA-NVSL. In late 2020, SARS-CoV-2 variants were identi ed and classi ed by CDC based on their impacts to human health, diagnostics, therapeutics, and vaccines 27 . In this study, strains sequenced prior to the identi cation of the rst variants were classi ed as early circulating strains.

Zoonotic Transmission
In analyses or visualizations that described the association between human SARS-CoV-2 infection and animal infection, a person's symptom onset date was assumed to represent most likely date of exposure in the animal. In instances where this date was not available, the date the person rst tested positive for COVID-19 was used.
To assess whether population increases in COVID-19 in people might cause subsequent increases in companion animal cases, a time series analysis using a cross-correlation function was performed to determine if there was a relationship between national human COVID-19 case reporting data and SARS-CoV-2 cases in companion animals. Daily national human COVID-19 case counts were downloaded from CDC's COVID Data Tracker and aggregated into monthly counts from March 2020 to December 2021 28 . SARS-CoV-2 infections in cats and dogs were aggregated into monthly counts. Sample collection date, which was available for 161 of 204 (79%) animals, was used as a proxy for date of infection. The crosscorrelation time series analysis was restricted to one year, March 2020 to March 2021, since all presumptive positive companion animal cases were forwarded to USDA-NVSL for con rmatory testing during this time.

Results
Overview From March 2020 to December 2021, 345 animals from 33 states in the U.S. were con rmed positive for SARS-CoV-2 ( Figure 1; Figure S1). Of these, 204 (59%), were companion animals including 109 cats and 95 dogs (Table 1; Table S1). SARS-CoV-2 was also detected by RT-PCR in one ferret; this animal was omitted from further analyses due to low sample size. In companion animal cases detected through passive surveillance, 94% were initially exposed to a person with COVID-19. In the remaining 6% of cases, the source of SARS-CoV-2 exposure was unknown (e.g., circumstances such as an animal tested positive upon arrival at a shelter, or owners declined investigation).
Multi-pet Households 36 households had more than one cat or dog. There was a 25% (Wilson CI  likelihood that if one cat or dog became infected in the household, a second cat or dog would also test positive for SARS-CoV-2 (data not shown). Probability was higher of a second cat or dog testing positive if the index pet was a cat (30%; Wilson CI 16-51), and lower if the index pet was a dog (15%; Wilson CI 4-42%).
The average Ct value from con rmatory RT-PCR was 28.6 for all con rmed positive companion animals with RT-PCR results (n = 69). VN titers (n= 107) ranged from 8 to 512, with a median titer of 64 (geometric mean titer of 1.8) for all con rmed positive companion animals. Titers from con rmed positive cats ranged from 32 to 512 with a median titer of 128 (geometric mean of 1.9), whereas results from dogs ranged from 8 to 128 with a median titer of 32 (geometric mean of 1.6). The highest titer of 512 was detected in a cat sampled 23 days after onset of symptoms in its owner.

Zoonotic Transmission
To estimate viral incubation period, the number of days between human symptom onset (or date of positive test) and onset of clinical signs in animals was calculated. Data for this analysis was restricted to animals with clinical signs only, and animals for which data were available to show that a person in the house had symptoms (n=32; Figure 4a). The median number of days between human symptom onset and onset of clinical signs in a companion animal was 10 (9.5, range: 0-24) days in cats (n=23), and 6 (8, range: 1-24) in dogs (n=9). We also assessed the length of active infection using clinical sign onset and resolution dates (Figure 4b). According to data from con rmed RT-PCR-positive animals, excluding deceased animals, with both onset and resolution dates collected (n=24), the median length of clinical infection was 10 (7.25, range: 3-36) days in cats (n=16) and 16.5 (10.75, range: 1-31) in dogs (n=8).
The likelihood of detecting an active infection by RT-PCR is largely dependent on the length of time between the animal's most-likely exposure to SARS-CoV-2 and the sample collection date. The median delay from presumed exposure date to animal sampling for a positive RT-PCR result was 10 (9, range: 0-35) days in cats (n = 35) and 6 (5.5, range: 0-24) days in dogs (n = 15; Figure 4c).
We also investigated whether patterns in human case counts were predictive of companion animal case counts. Given the available data in the restricted timeframe, a time series analysis using a crosscorrelation function determined that while there appeared to be an observable relationship between human and animal case counts, this relationship was not signi cant ( Figure 5).

Discussion
This study is the rst to summarize nationally compiled surveillance data on the epidemiological and clinical characteristics of natural SARS-CoV-2 infection in companion animals. While there are publications describing SARS-CoV-2 in companion animals in many countries including those in Europe [30][31][32][33][34][35][36] and Asia 37,38 , studies are often led by academic institutions conducting independent research; surveillance is not sustainable or systematic. In the United States, data on SARS-CoV-2 positive animals is collected through systematic One Health investigations and is shared voluntarily through collaborations with local, state, and federal and academic public health and animal health o cials [15][16][17]25 . This perhaps helps to explain why the majority (56%) of all companion animal cases reported globally are from the United States 14 .
Overall, our data show that among companion animals detected through passive surveillance, 94% had known exposure to a person with COVID-19 prior to the animal's infection. This provides strong evidence that people, most often owners, are the source of infection for their pets. These results corroborate ndings from a large-scale study in northern Italy that reported dogs living in households with COVID-19 positive people were more likely to have detectable antibodies than dogs living in households without 30 .
Other case studies from countries in Europe, Asia and South America have identi ed similar patterns 39 . These results support guidance developed by federal One Health partners that was released in January 2020 and continues to be updated, including recommendations to avoid animals just like you would other people when sick or have a suspected COVID-19 infection, and to wear a mask around both people and animals when ill with COVID-19 40 .
While the evidence for human-to-pet transmission is robust, less data are available to determine the likelihood and frequency of pet-to-pet or pet-to-person transmission within households. Our analysis of 36 households containing more than one pet indicate that any cat or dog in the household has a 25% probability of becoming infected with SARS-CoV-2 if there is a positive index pet. This probability was higher when cats were the index pet (30%), than dogs (15%), in line with experimental and challenge studies that suggest cats are more susceptible 41,42 and may be more infectious based on lower overall Ct values than dogs. While these data suggest pet-to-pet transmission may occur in households, we cannot determine whether subsequent pets in a multi-pet household were infected from a person or another animal. More One Health research to examine transmission dynamics among animals and among animals and people living in household environments is warranted.
To-date, evidence of cats or dogs transmitting SARS-CoV-2 to people is limited, although detecting and accurately attributing transmission from an animal source is challenging against a background of signi cant human-to-human transmission. This is further complicated since, like people, animals may not be tested, especially if they are subclinical, but may still be capable of shedding virus and infecting other individuals. Attempts to attribute directionality of transmission requires both human and animal samples that can be successfully sequenced and compared, in addition to a robust epidemiologic One Health investigation. One recent case study suggests cat-to-human transmission as likely from an infected pet cat in Thailand 43  Our study also estimated thresholds of diagnostic detection in companion animals. Data from 142 companion animals (74 cats, 68 dogs) sampled after presumed exposure to a person with SARS-CoV-2 suggests that viral nucleic acid detection by RT-PCR occurs shortly after presumed exposure, typically less than 5 days. For cats, our data suggest the ideal sampling window to detect viral nucleic acid is 3-17 days after exposure, and 3-10 days for dogs (Fig. 4). We also discovered that virus speci c neutralizing antibody is rapidly produced. Rapid and sustained titers of neutralizing antibody after infection may have contributed to the mild nature of disease observed in the majority of animals in this study (see Carpenter et al. 25 , Carvallo et al. 46 , and Rotstein et al. 47 for descriptions of companion animal mortalities that occurred while pets were positive for SARS-CoV-2).
Finally, samples from 34% of companion animals included in our data were successfully characterized by whole genome sequencing. These sequences were used to identify variants and their corresponding clinical presentation (Table S4). Among this subset, early circulating strains and four variants were detected: each corresponded temporally with variants circulating in the human populations in the same geographic area at the time. Processes to continue to generate and analyze sequence information from animal populations are essential to ensure that novel mutations, strains and variants arising from animal populations, including companion animals, are detected expediently, ideally before detrimental impacts to public health can be recognized 48 .
Some limitations exist with the nature of data collection and the analyses presented in this manuscript. First, no federal agency is currently mandated to oversee companion animal surveillance or response, including for emerging zoonotic diseases. Reporting for companion animal zoonoses to public or animal health o cials may also be jurisdiction or disease speci c. Given that surveillance varies by jurisdiction, the approach taken by One Health investigations and reporting are likely to have varied. Ongoing efforts to improve surveillance, including enhancing coordination and data sharing among One Health sectors and supporting data modernization initiatives, are underway 49 .
Second, since there is no standardized diagnostic method or sample validation criteria for SARS-CoV-2 in animals in the U.S., we opted to include only companion animals that were con rmed positive at USDA-NVSL to ensure consistency and comparability in diagnostic results. In doing so, however, we limited our sample size and perhaps skewed results toward animals that met standards of con rmation. Third, our analysis based on a human source of exposure does not account for households where multiple people Given that SARS-CoV-2 infections in animals are not currently nationally noti able in the United States, it is possible that unreported animal cases were missed within the timeframe of the data included in this manuscript. This is corroborated by published research which also suggests that current surveillance may be vastly underestimating the true burden of SARS-CoV-2 in animals [15][16][17]36 . Without continued One Health collaboration across sectors to pursue more extensive surveillance (both active and passive), many SARS-CoV-2 infections in companion animals will remain undetected.

Conclusion
Despite the known susceptibility of companion animals to SARS-CoV-2, testing and disease reporting of pet cases of SARS-CoV-2 infection has been limited in the U.S. Lack of mandatory reporting of companion animal cases of SARS-CoV-2 infection has continued to be a challenge throughout the COVID-19 pandemic. Relying on voluntary reporting of a novel, emerging zoonotic disease with unknown transmissibility and disease in animals is a hurdle for understanding the clinical and epidemiological features of a rapidly spreading zoonosis. This is especially apparent with companion animals, whose oversight falls in a government jurisdictional void, and where structures and systems to detect, monitor, and respond to companion animal zoonoses are typically not a standard component of public health or animal health programs. This manuscript provides support that systematic surveillance in animal populations can be established, sustained, and bene cial in a global public health emergency. In the instance of SARS-CoV-2, strong collaborations between public health and animal health sectors at the local, state, and federal level were able to circumvent some of these issues. However, formalized One Health collaboration mechanisms that institutionalize joint investigation and coordinated surveillance are necessary to best protect human and animal health and to most e ciently respond to future emerging zoonotic disease threats.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. CASupplementary.pdf