A cross-sectional study of environmental, dog, and human-related risk factors for positive canine leptospirosis PCR test results in the United States, 2009 to 2016

Leptospirosis is a worldwide bacterial disease that affects humans and animals and is regarded as re-emerging in dogs in North America. Leptospira spp. are tightly coiled spirochetes shed in urine and transmitted by maintenance hosts, such as rats, mice, and raccoons [1]. There are multiple pathogenic Leptospira spp. serovars circulating in the United States, each with different wild and domestic animal reservoir species and geographic spread [1]. Serovars identified in dogs in the United States include Canicola, Icterohaemorrhagiae, Autumnalis, Pomona, Grippotyphosa, and Bratislava [1]. Dogs acquire the bacteria from direct contact of mucous membranes or broken skin with urine of infected individuals or contact with food, water, or soil contaminated with infectious urine [2]. Leptospirosis can cause a wide spectrum of clinical illness in canines, ranging from subclinical to severe hepatic or renal failure, with the possibility of death [3]. Previous studies have identified spatial and temporal trends in the prevalence of canine leptospirosis cases in the United States [4–6]. Increased prevalence has commonly been identified from the west coast, the upper midwest, the northeast and the southeast [4, 5]. A seasonal increase in cases has been previously reported in late summer and fall [6, 7]. Additionally, environmental, dog owner, and dog-level risk factors have been linked to canine leptospirosis cases in previous studies [6, 8–13]. Despite these previous studies, the epidemiology of canine leptospirosis remains poorly defined, with studies generally limited to a single geographical area or a single subset of factors (e.g. environmental factors). The current study collectively examines factors for a positive polymerase chain reaction (PCR) test from many predictor subsets (temporal, geographical, environmental, animal) on a national scale to highlight risk factors and explore current canine leptospirosis epidemiology in the United States. Reliance on serologic testing, with varying criteria for defining a positive test result, has contributed to the challenge of studying this bacterium in dogs. The microscopic agglutination test (MAT) has been widely used to diagnose leptospirosis in dogs, and previous canine risk factors and prevalence studies have used MAT data [3, 5, 12, 14]. In recent years, PCR leptospirosis testing has become increasingly used in clinical veterinary medicine and may reduce the interpretation challenges commonly encountered with MAT (e.g. a single acute MAT has been found to have lower sensitivity for predicting urine shedding compared to a urine PCR assay, is not as influenced by recent vaccination, and since it does not identify serovars, there are no concerns about cross reaction among serovars) [3, 15]. The PCR test detects both viable and non-viable Leptospira spp. nucleic acid and not the organism itself. Since it can detect nonviable nucleic acid, it may take extensive antimicrobial use for a false negative to occur [3]. Nationwide analysis of the spatial and temporal distribution of canine leptospirosis and evaluation of associated risk factors using PCR test data is important to provide additional insights into this complex disease. The objectives of this study were: (1) to describe temporal trends of canine leptospirosis PCR test results in the United States, (2) descriptively map canine leptospirosis PCR test results in the United States, and (3) investigate the association between environmental, seasonal, dog- and human-level factors, using a large dataset of canine leptospirosis PCR-test results. Results Over the 8-year study period, a total of 41,370 test results were provided. Duplicates consisted of 1017 entries for which at least one exact matching entry (concordant test results) was present. The test entry with the most recent date for each dog was left in the dataset, resulting in the removal of 1152 test entries (range 1–116 duplicate entries removed per dog, median 1). Forty-one dogs had duplicate entries with discordant test outcomes. All entries for these dogs were removed from the data, resulting in the removal of 100 test entries (range 2–10 duplicate entries removed per dog, median 2). A total of 1252 duplicates were removed from the dataset due to duplicate entries, leaving 40,118 test results to be included in the analysis. The population of dogs tested was 51% female and had a mean age of 6.9 years (SD 0.02; range 0–21). As 38.5% (15,446/40,118) of the entries had breed missing or listed as mixed with no predominant breed designation for the mix, breed group was not examined. PCR test results were available for every state (Fig. 1), with a median of 357 [mean 787; range: 18 (Alaska) - 5405 (California)] test results per state over the study period. At least one PCR test result was available for 44.5% (1401/3143) of counties, with a median of 6 (mean 29; range: 1–1652) tests per county over the study period. Over 18% (262/1401) of counties with test results available only had one test result. A data governance agreement precluded presentation of county-level data in maps. Over the study period, canine leptospirosis PCR testpositive proportion in the United States was 5.4% (2176/ 40,118) (95% Confidence Interval (95% CI) = 5.2, 5.7%). Test-positive proportion ranged from 2.6% (2010 and 2011) to 8.3% (2013) and varied by month and region. When all years were pooled, the test-positive proportion steadily increased through the spring and summer seasons, and it peaked in the fall in the midwest, and northeast regions (Fig. 2). Test-positive proportion was lowest in February and March in these two regions. Testpositive proportion remained relatively constant in the Smith et al. BMC Veterinary Research (2019) 15:412 Page 2 of 12 southeast region, yet it was slightly higher in the fall season, with a dip in June (Fig. 2). Test-positive proportion remained relatively constant across seasons in the west and southwest, with the highest test-positive proportion observed in the winter months (Feb and Dec, respectively) (Fig. 2). All states had at least one positive test result except for Utah (total number of tests = 41), North Dakota (26) and Alaska (18) (Fig. 1). Of the counties with at least one test result, 33.9% (475/1401) had at least one positive test result. Test-positive proportion point estimates were highest in the midwest (7.0%) and southwest (8.1%) regions as Michigan (9.6%), Texas (9.1%), Iowa (9.1%), Nebraska (9.0%), and Illinois (8.7%) had the highest state-level test-positive proportion over the study period (Fig. 3, Additional file 1: Figure S1). Test-positive proportion in the other regions was highest for states bordering the midwest and southwest; Kentucky (6.9%) and Colorado (5.8%) had the highest test-positive proportion for their respective regions (Fig. 3, Additional file 1: Figure S1). Risk-factor analysis Of the 13 variables analyzed, 10 were significant predictors of a canine PCR-positive test result in univariable models (Table 1). No variables eligible for the