Canine infection with Dirofilaria immitis, Borrelia burgdorferi, Anaplasma spp., and Ehrlichia spp. in the United States, 2013–2019

Background Dogs in the US are commonly infected with vector-borne pathogens, including heartworm and tick-borne disease agents. The geographic distribution of both arthropod vectors and the pathogens they transmit continues to expand. Methods To describe the current geographic distribution and prevalence of antigen of Dirofilaria immitis and antibody to Borrelia burgdorferi, Ehrlichia spp., and Anaplasma spp. in dogs, we summarized over 144 million test results from 2013 to 2019, inclusive, by county, state, and region. Canine seroprevalence by state was compared to population-adjusted human reports of tick-borne diseases. Results Results varied regionally, with D. immitis antigen and Ehrlichia spp. antibodies more frequently detected in the Southeast (2.6% and 5.2%, respectively) and antibody to B. burgdorferi and Anaplasma spp. most common in the Northeast (12.1% and 7.3%, respectively). Overall, percent positive test results to D. immitis decreased in the Southeast by 33.3% when compared to earlier summaries using the same strategy (from 3.9 to 2.6%). Geographic expansion of areas where dogs commonly test positive for Ehrlichia spp. was evident, likely because of a change in the test made in 2012 to allow detection of antibodies to E. ewingii concomitant with expansion of vector tick populations. Percent positive test results to Ehrlichia spp. increased in every region; this shift was particularly pronounced in the Southeast, where percent positive test results increased fourfold (from 1.3 to 5.2%). Continued geographic expansion of B. burgdorferi and A. phagocytophilum was apparent in the Northeast, Midwest, and Upper South, although canine seroprevalence of antibody to B. burgdorferi was much lower than prior surveys in many Lyme-endemic areas. Annual reports of human cases of Lyme disease, ehrlichiosis, and anaplasmosis were associated with percent positive canine results by state for the three tick-borne disease agents (R2 = 0.812, 0.521, and 0.546, respectively). Within endemic areas, percent positive test results for all three tick-borne agents demonstrated evidence of geographic expansion. Conclusions Large scale analysis of results from screening dogs in practice for evidence of vector-borne infections, including those with zoonotic importance, continues to be a valuable strategy for understanding geographic trends in infection risk over time.

Background National summaries of canine vector-borne disease seroprevalence generated from testing individual dogs in practice aid understanding of pathogen distribution and provide insights into geographic and temporal changes [1,2]. To facilitate early diagnosis, treatment, and prevention, advisory boards recommend that all dogs receiving veterinary care be evaluated annually for vector-borne infections [3]. Accordingly, each year in the US, millions of dogs are tested for antigen of heartworm (Dirofilaria immitis) and antibody to tick-borne disease agents, most commonly Borrelia burgdorferi, agent of Lyme borreliosis; Anaplasma spp., which cause anaplasmosis in people and animals; and Ehrlichia spp., causative agents of human and canine ehrlichiosis [1,2,4,5]. In addition to benefiting canine patient health, analysis of the aggregated results captured using the same test method over many years can serve as a bellwether to identify areas where infection risk may be changing [6][7][8].
Despite the widespread availability of preventives, ticks are common on dogs across the USA, and evidence suggests the prevalence of heartworm infection and of antibodies to tick-borne disease agents is increasing in some regions [9][10][11][12]. These increases likely result from a combination of factors including increased vector populations, resulting in more intense transmission, geographic spread of natural maintenance cycles for infection, and translocation of infected dogs [11][12][13][14][15][16]. Canine serologic evidence of past or current tick-borne infection also correlates with human case reports on a county-and statewide basis [2,6,10,17], and the number of human cases is similarly increasing [18]. In the present article, we update our earlier publications by reporting the percent positive test results of dogs evaluated by veterinarians in the US from 2013 to 2019, documenting continued changes in both distribution of these infections and overall infection risk.

Source of data
Results for the present analysis and summary (2013-2019) were generated using USDA-licensed test kits (IDEXX Laboratories, Inc.) and included: SNAP ® 4Dx ® Plus Test kit, an in-clinic enzyme-linked immunosorbent assay (ELISA) for detection of D. immitis antigen and canine antibodies to B. burgdorferi, Ehrlichia spp. (E. canis, E. ewingii), and Anaplasma spp. (A. phagocytophilum and A. platys); SNAP ® HW RT Test kit, an in-clinic ELISA for the detection of D. immitis antigen in canine serum, plasma, or whole blood. In addition, results generated from microtiter plate ELISA tests for the detection of D. immitis antigen (e.g., PetChek ® Heartworm PF Test) and canine antibodies to B. burgdorferi, Ehrlichia spp. (E. canis, E. ewingii), and Anaplasma spp. (A. phagocytophilum and A. platys) in canine serum or plasma at a reference laboratory were included. The performance of each test and associated diagnostic reagents has been reported previously [1,2,19].
Testing results were collated from two sources: (i) directly from veterinary practices testing patients inclinic (SNAP ® HW RT Test and SNAP ® 4Dx ® Plus Test) and (ii) through the IDEXX Reference Laboratories network (PetChek ® Heartworm PF and Lab 4Dx ® Plus Test). Results from veterinary practices were recorded in IDEXX VetLab ® Instrumentation and Software and either automatically recorded by the IDEXX SnapShot Dx ® Instrument or SNAP Pro ® Analyzer or manually entered by clinic staff. To ensure data privacy, results were collected without owner information or canine patient identification and thus repeat testing events or translocated dogs (i.e. dogs with a travel history to another region) cannot be identified or omitted.

Data analysis
State and county of the veterinary hospital providing the test result or submitting the sample to the laboratory were used to assign results to region as previously described [1,2,20], with four primary regional groups (Midwest, Northeast, Southeast, and West) used to compare results to previous publications. Percent positive results were calculated by dividing the number of tests reported as positive for each agent by the total number of testing events recorded in each county, state, or region. For state-wide summary tables and comparison to human disease reports, all results collected from 2013 to 2019 were included. For construction of maps depicting percent positive test results by county, individual counties with fewer than 210 total test results, or fewer than an average of 30 test results per study year, were excluded [1,2] and then maps created with R (version 3.6.1) [21] using the albersusa [22] and tmap [23] packages. Large scale analysis of results from screening dogs in practice for evidence of vector-borne infections,  including those with zoonotic importance, continues to be a valuable strategy for understanding geographic trends  in infection risk over time

Summary
A total of 144,022,232 results were available from dogs tested in 2298 counties and in all 50 states in the US over the 7-year period summarized in the present paper (Tables 1, 2 , 3, 4). This represented results from nearly 50 million tests for antigen of D. immitis and more than 30 million tests for antibodies to B. burgdorferi, Ehrlichia spp., and Anaplasma spp. Evidence of all four agents was found in dogs from every state considered. Distribution of positive tests and relative percent positive values by county and state are shown in Figs. 1, 2, 3, 4.

Heartworm
Percent positive test results for D. immitis antigen were higher in the Southeast than in the other three regions and were higher in the West and Midwest than in the Northeast (Table 1, Fig. 1). National and regional prevalence of percent positive test results for D. immitis was largely unchanged from our previous reports [1,2] with the exception of the Southeast, where the overall prevalence was lower than in previous reports; this decrease in prevalence was evident in every state in the Southeast except Alabama (Table 1).

Lyme disease
Percent positive test results for antibody to B. burgdorferi were highest in the Northeast, followed by the Midwest, Southeast, and West (Table 2, Fig. 2). The overall, national prevalence of antibodies to B. burgdorferi was decreased from our previous report (from 7.2 to 5.9%) and was lower in every region except the Southeast where percent positive test results remained unchanged ( Table 2). Prevalence of antibodies to B. burgdorferi for most individual states remained unchanged or decreased; only New York, Iowa, Ohio, and West Virginia were significantly higher than the earlier report (Table 2).

Ehrlichiosis
Percent positive test results for antibody to Ehrlichia spp. were higher in the Southeast than in the other three regions ( Table 3, Fig. 3). National and regional canine seroprevalence for Ehrlichia spp. antibodies was also higher than in previous reports, including twofold higher in the Northeast and as much as threefold higher in the Southeast ( Table 3). Prevalence of Ehrlichia spp. antibody was also higher in every state in the Northeast and Southeast, every state in the Midwest except South Dakota, and all states in West except Hawaii, New Mexico, and Washington (Table 3).

Anaplasmosis
Percent positive test results for antibody to Anaplasma spp. were highest in the Northeast and lowest in the Southeast (Table 4). Overall, seroprevalence of antibodies to Anaplasma spp. decreased nationally (from 4.4 to 3.3%) and in every region except the Northeast, where it remained largely unchanged from the previous report (Table 4). By state, percent positive test results for Anaplasma spp. antibodies were lower in states throughout the Southeast, Midwest, and West. In the Northeast, statewide seroprevalence was higher in northern New England (Maine, New Hampshire, Vermont) but lower in several mid-Atlantic states (Connecticut, New Jersey, Rhode Island) ( Table 4).

Comparison to human disease reports
Seroprevalence of B. burgdorferi antibodies in dogs and reported cases of Lyme borreliosis in people by state were positively associated (R 2 = 0.812, F = 207.0). Reported human cases of Lyme borreliosis were lower than expected based on canine seroprevalence in Massachusetts, New York, Virginia, and West Virginia, and higher than expected in Delaware, Maine, New Hampshire, and Vermont. Seroprevalence of Ehrlichia spp. antibodies in dogs and reported cases of ehrlichiosis in people by state were positively associated (R 2 = 0.521, F = 52.2). In some states, reported human cases of ehrlichiosis were lower than expected (Mississippi) based on canine serology. Seroprevalence of Anaplasma spp. antibodies in dogs and reported cases of anaplasmosis in people by state were positively associated (R 2 = 0.546, F = 57.6). In some states, reported human cases of anaplasmosis were lower (Connecticut and Massachusetts) or higher (Vermont) than expected based on canine serology.

Discussion
The present study is the third in a series which taken together describe nearly 2 decades of results and constitutes the largest ever reported survey of dogs for multiple vector-borne infections, expanding on our previous publications by including data from the past 7 years and 4-to 30-times more canine test results [1,2]. These data represent findings from annual testing of approximately 7% to 12% of the estimated 60 million pet dogs that receive annual veterinary care in the US [26] and thus provide valuable insights into canine infection with heartworm, infestation with ticks, and past or current infection with several tick-borne disease agents. Positive results for all four canine vector-borne agents considered (heartworm, Borrelia burgdorferi, Ehrlichia spp., and Anaplasma spp.) were found in every state in the US (Tables 1, 2, 3, 4), the first time we have observed national evidence of all four types of infections in the 17 years of data included in these surveys [1,2]. Because data are anonymized, we cannot exclude the possibility of repeated testing events or of identifying evidence of infections acquired in a different geographic location. Nonetheless, a number of key trends are evident in the geographic distribution (Figs. 1, 2, 3, 4) and relative prevalence of percent positive results in different regions (Tables 1, 2, 3, 4).
For canine heartworm infection as determined by detection of antigen, the data in the present paper suggest that overall prevalence in well-cared-for pet dogs has remained largely unchanged since national surveys using similar sample sets were first reported [1] and that percent positive test results from pet dogs have decreased in southern states (  [11,12,27,28], although regional areas where prevalence was clearly decreasing were also identified [27]. Apparent discrepancies between the data in the present article and those reported by others may be due to differences in testing platforms, survey methods, or patient profile. The reports showing an apparent increase in percent positive test results included data from multiple diagnostic laboratories [29] whereas the present paper used only data from practices using well-validated IDEXX tests, which may be a source of selection bias. Surveys of the number of dogs treated for heartworm are useful but cannot be directly compared to infection prevalence trends, and artificial increases in percent positive results can be seen when testing is used primarily for diagnostic verification of a suspected infection rather than for routine screening [28,29]. For example, the similar percent positive test results in dogs in Alaska (1.4%) and Florida (1.4%) in the present study likely reflect targeted testing by Alaska veterinarians based on a travel history compared to routine screening of dogs protected from infection by preventives despite intense transmission pressure in Florida [30]. Nonetheless, heartworm is a preventable infection, and the finding that 1.4% of pet dogs receiving veterinary care in the USA, or approximately 840,000 pet dogs overall, are identified as infected with this potentially fatal parasite each year, is dispiriting.
Canine infection with B. burgdorferi remains widespread, with antibodies detected in 5.1% of dogs in the present study overall and 12.1% of dogs in the Northeast. Significant increases in percent positive test results were evident in some areas of the upper South, including West Virginia and North Carolina, consistent with other reports of geographic spread of the maintenance cycle for this pathogen [10,13,31]. At the same time, decreased statewide seroprevalence of antibodies to B. burgdorferi,  in some cases by more than 40%, was evident in several states where Lyme disease has long been endemic or hyperendemic, including Connecticut, Massachusetts, New Jersey, Delaware, and Maryland (Table 2), a trend that has been previously reported [10]. Canine infection with B. burgdorferi varies widely even in relatively focal areas [17] and can be prevented with a combination of tick control and vaccination [32]. Increased adherence to these recommendations, including the widespread use of systemic isoxazoline acaricides, which first became available in the USA in 2014, would be expected to result in decreased percent positive test results in dogs over time. However, B. burgdorferi infection remains common, particularly in areas of the Northeast where I. scapularis vector populations are intense or expanding, and canine seroprevalence to B. burgdorferi continues to increase in northern New England, western New York, and western Pennsylvania [10,33]. In other regions, such as in states along the southern border of the USA, autochthonous transmission of B. burgdorferi has not been documented, but antibodies to B. burgdorferi may occasionally be identified in dogs that move to that region, creating confusion when survey results are not carefully interpreted in context [34,35]. Results of the current serosurvey demonstrate an increase in the number of dogs with antibodies to Ehrlichia spp. in nearly all states within the US. Since the publication of the 2010-2012 seroprevalence results, the ELISA test for antibodies to Ehrlichia spp. was modified to include a new peptide for the detection of antibodies to E. ewingii [19]. Canine seroreactivity to this new target, as well as the increase in seropositive results observed particularly across the Southeast, is consistent with regional results from a study where canine serum samples were obtained from academic and private veterinary hospitals [6]. In that earlier study, antibodies to E. canis were more frequently detected in dogs in the southcentral region where the brown dog tick vectors predominate, while antibodies to both E. ewingii and E. chaffeensis were more prevalent in dogs from a band of states across the upper south, where A. americanum is most common [6]. Indeed, A. americanum, the tick responsible for transmitting E. ewingii, is highly prevalent across the Southeast region with evidence of geographic spread to states within the Midwest and Northeast [15,36]. Even before the test could detect antibodies to E. ewingii, the Midwest region had evidence of canine seroreactivity to Ehrlichia spp. [1,2]. It is not known if this previous and continued seroreactivity is associated with the novel Ehrlichia muris subsp. euclairensis reported there [37] since only one clinical infection with this pathogen has been documented in the dog [38]. Because the different Ehrlichia spp. antibodies are detected by reactivity on a single spot or microtiter well of the test, it is not possible to determine the infecting species of Ehrlichia. While geographic location and predominant tick species may provide some guidance, elucidating the causative Ehrlichia species is becoming increasingly difficult because of expanding tick ranges and the transport of dogs across the country for adoption.
The seroprevalence of Anaplasma spp., although decreased slightly in the Midwest and certain Northeastern states, remained largely unchanged relative to the previous report [2]. In much of the northern USA, antibodies to Anaplasma spp. (presumably to A. phagocytophilum) were detected less frequently than antibodies to B. burgdorferi in most states despite transmission by the same species of tick, Ixodes scapularis. Surveys   [42]. A greater proportion of ticks harboring B. burgdorferi could make it more likely that dogs would be exposed to this pathogen and perhaps contribute to the observed difference in seroprevalence observed with Anaplasma in these regions. In the southcentral and southwestern states, increased prevalence of Anaplasma spp. is more likely due to A. platys transmitted by brown dog ticks, which are commonly present in this region [43,44]. The use of systemic and topical acaracides may help to prevent anaplasmosis and Lyme borreliosis as both have been reported to reduce I. scapularis transmission of B. burgdorferi and A. phagocytophilum to dogs [45][46][47].
Key limitations of the present study include lack of clinical information about individual dogs that may have provided insight into veterinarians' reasons for testing. Selective testing, which may be more common when dogs present with clinical signs characteristic of a given vector-borne disease and owner-reported travel to an area where a suspected pathogen is endemic, would be expected to create a positive bias in percent positive test rates in non-endemic areas. We addressed this concern, in part, by omitting counties reporting few test results (less than 210 over 7 years) from the maps (Figs. 1, 2, 3, 4) as we have done in both previous reports [1,2]. Nonetheless, dogs testing positive in regions where pathogens are not known to be transmitted are most likely due to translocation of infected dogs. Large-scale relocation of dogs from heartworm-endemic areas to counties where infections are rare has been associated with an increase in local prevalence of dogs testing positive for D. immitis [12]. Smaller studies have documented past travel in dogs that test positive for B. burgdorferi in areas where disease is not known to be locally transmitted [48,49]. Canine serologic data alone cannot confirm pathogens are established in a given area, and multiple lines of evidence, including directly testing vector mosquitoes or ticks, are necessary to confirm apparent transmission of vectorborne pathogens in new regions [1,2,12,13,47,48]. Similarly, while the present study confirms that canine serology remains a useful predictor of where corresponding tick-borne diseases are most likely to be reported in people, limitations in both human and canine datasets suggest this finding must be interpreted with appropriate caution [2,[48][49][50]. Nonetheless, routine screening of dogs for vector-borne infections, as recommended by veterinary advisory groups to allow early detection of infection in individual patients [3,4,32], has a clear, albeit secondary, veterinary and public health benefit: careful analysis of the resulting aggregate data enhances our understanding of the changing pattern of the distribution of these vector-borne pathogens across the USA [1,2].

Conclusions
In this study-the third large-scale vector-borne disease serosurvey conducted since we began analyzing multi-year data sets in 2001-we have provided a comprehensive update on the frequency of positive test results for the most common canine vector-borne    decreases in Lyme-endemic areas, while increases in B. burgdorferi were observed in non-endemic geographies, suggesting continued spread of this infection into new areas. Finally, the data show a positive association between seroprevalence of vector-borne disease agents in dogs and reported human cases of disease. Regular screening of canine pets can not only improve healthy outcomes for dogs but can also provide insight into the risk of vector-borne disease for people.