Factors associated with development of Canine Infectious Respiratory Disease Complex (CIRDC) in dogs in 5 Canadian small animal clinics

Abstract

This study investigated the association between presence of respiratory pathogens and development of Canine Infectious Respiratory Disease Complex (CIRDC) in dogs in 5 Canadian small animal clinics. In total, 86 dogs were tested using a commercial PCR respiratory panel; 64 dogs were considered as cases and 22 were control dogs matched by veterinary clinic. No control animals (0/22) were positive for canine parainfluenza virus (CPIV), whereas 27/64 (42%) CIRDC cases were positive. Furthermore, 81% of case dogs tested positive for Mycoplasma cynos, compared with 73% of control dogs. Canine respiratory corona virus (CRCoV) was detected in no control dogs compared with 9.4% of clinical dogs. No animals were positive for any influenza virus type A present in the diagnostic panel. Presence of CPIV was associated (P < 0.01) with the occurrence of CIRDC after adjustment for demographic factors and presence of CRCoV (P = 0.09).

Introduction

Canine infectious respiratory disease complex (CIRDC), commonly known as “kennel cough,” is an endemic respiratory syndrome that causes both sporadic disease and outbreaks (1,2). The disease is of particular concern in crowded populations and where there is extensive dog-to-dog contact, particularly in re-homing kennels (2). Sporadic and epidemic disease also occurs in dogs in the general population, and clusters of infection can occur in veterinary hospitals (3).

Canine infectious respiratory disease is a complex, multi-factorial disease, with similar clinical signs caused by a wide range of pathogens acting individually or as co-infections (1). Common clinical signs associated with many of these causative agents include nasal discharge, coughing, respiratory distress, and lethargy (3). Traditionally, canine parainfluenza virus (CPIV), canine adenovirus type-2 (CAV-2), and canine herpesvirus 1 (CHV-1) have been considered the main viral agents of CIRDC, with recent emergence of canine influenza virus (CIV) in some regions (1,2,4–6). Bordetella bronchiseptica is the most commonly identified bacterial pathogen, with Streptococcus zooepidemicus causing occasional (but often severe) disease (7). The role of Mycoplasma has been unclear, since potentially pathogenic Mycoplasma can also be found in many healthy dogs (2,8,9).

In recent years, the role of novel pathogens in the development of CIRDC has been questioned and researched (1,2). There is some debate as to whether apparently emerging pathogens such as canine respiratory coronavirus (CRCoV) (2,8,9) are truly new emerging pathogens or pre-existing pathogens that are now detectable using improved molecular diagnostic tools. Likely both situations apply, but data pertaining to CIRDC pathogens are often limited because of the failure to perform diagnostic testing in cases of sporadic disease. Hence, many questions remain about established pathogens, new pathogens, and the role of co-infection in the pathophysiology of disease. A better understanding of CIRDC is important for prevention and management of disease, including decision-making for vaccination (1–3,7,9). Thus, the objective of this study was to determine the association between CIRDC and the presence of potential pathogens in deep pharyngeal swabs in dogs.

Materials and methods

This was a case-control study with control dogs matched by the veterinary clinic for convenience. The source population consisted of dogs serviced by 5 small animal clinics from 3 Canadian provinces: Alberta (n = 1), British Columbia (n = 1), and Ontario (n = 3). One of the clinics was a full-time emergency clinic, 2 clinics were general practice and emergency clinics, and 2 clinics were general practices. For each veterinary clinic, the number of cases ranged between 3 and 20, and the number of control dogs ranged between 2 and 5. In total, 64 case dogs and 22 control dogs were included in the study. The study period was between June 2013 and February 2014.

Case definition and inclusion criteria

All canine patients of any age with an acute cough and other signs typical of CIRDC (nasal or ocular discharge, respiratory distress, or lethargy) were considered for inclusion in the study. Patients were excluded if they had a chronic cough, had been treated with antimicrobials in the last 2 mo, had co-morbidities (e.g., congestive heart failure, pneumonia) that could lead to coughing, or had received an intranasal or oral vaccine against B. bronchiseptica, CPIV, or CAV-2 in the last month. Clinical signs of CIRDC in patients were assessed by veterinarians upon presentation at each of the participating clinics. Cases were assessed based on history, physical examination, and thoracic radiographs when considered necessary according to the veterinarian’s opinion. Each client had to sign a consent form to allow their pet to be included in the study.

The following protocol was applied to case patients included in the study. Up to 3 swabs were collected from the areas of most predominant clinical signs (pharyngeal region, nasal discharge, ocular discharge), including 1 culture swab (Copan Diagnostics, Murrieta, California, USA). One to two dry sterile cotton-tipped wooden swabs were first used; swabs were broken off and submitted in a sterile “red top tube” with no medium. In addition, samples for bacteriological culture were collected with a sterile swab that was then placed into a sterile tube containing Amies transport medium (Becton Dickinson, Mississauga, Ontario). The most clinically relevant regions were cultured: nasal discharge, conjunctiva, or deep pharyngeal region. The majority of the samples were from the pharynx. Most case patients were presented in the first week of their illness. Samples were submitted to the Antech Laboratory (Antech Diagnostics Canada, Mississauga, Canada) for testing, which was conducted using CT995 — Fast Panel PCR Canine Respiratory Disease Profile. A predetermined total of 20 patients were selected from each of the 5 project sites. Each participating clinic was to enroll the first 20 eligible clinical cases over the study period. Demographic and vaccination data, where available, were then collected from the medical records for use in statistical analysis.

Control definition and inclusion criteria

To select controls, any patient which was presented at 1 of the 5 participating clinics for a regular health examination and was confirmed not to have CIRDC by regular physical examination by a veterinarian could be included in the study. For each clinic, 5 controls were to be enrolled starting September 2013. A total of 22 controls were recruited.

Diagnostic testing

Samples were submitted to the Antech Laboratory for test CT995 — Fast Panel PCR Canine Respiratory Disease Profile for the detection of B. bronchiseptica, CAV-2, canine distemper virus (CDV), CHV-1, canine influenza virus H3N8, influenza H1N1 virus, influenza H5N1virus, CPIV, CRCoV, Mycoplasma cynos, and Streptococcus equi ssp. zooepidemicus. Bacteriological swab samples were plated on Tryptic Soy agar and MacConkey agar for growth of B. bronchiseptica, and on CNA agar for growth of Streptococcus zooepidemicus.

Statistical analysis

Data for case and control dogs were described using frequencies and proportions. Simultaneous detection of multiple respiratory pathogens was tabulated using frequency tables. For the purposes of inferential statistical analysis, demographic data such as age, dog breed, and time since last vaccination were aggregated into binary variables on the basis of biologically plausible cut-off points. Exact logistic regression was used to evaluate associations between the CIRDC status and putative risk factors. This was done in 2 steps. In the first step, association between CIRDC and all diagnostic, management, and demographic variables was screened using univariable regression models. In the second step, manual forward building procedure was implemented to build the final multivariable regression model. For this purpose, only the variables that were univariably associated with CIRDC at P < 0.20 were considered in the model-building. Presence of pathogens was considered as the primary set of exposures, whereas demographic variables were considered as confounders and were a-priori determined to be forced into the model regardless of statistical significance. Management factors such as different sets of vaccination were not considered for the multivariable model building for 3 reasons. First, vaccination status of some dogs could not be accurately assessed. Second, specific vaccinations could be applied to high-risk dogs and could give biased estimates. Third, due to small numbers of study animals and variety of vaccination practices, time since vaccination could not be defined with required accuracy and evaluated.

Two types of final multivariable models were presented. The first model contained only variables in which estimates for each covariate could be produced. The second model was adjusted for additional demographic variables that could not be used as covariates in the final model because of computational requirements. Biologically plausible interactions were evaluated for statistical significance.

Results

Several respiratory pathogens were detected in both CIRDC case and control dogs, including M. cynos, CPIV, B. bronchiseptica, CRCoV, CHV-1, and CDV (Table 1). Complete results based on detection of individual pathogens, as well as unique combinations of pathogens further stratified by the CIRDC status are presented in Table 1. Interestingly, no study dogs tested positive for CAV-2, any influenza A virus (IAV) present in the diagnostic panel (including the CIV H3N8) or S. zooepidemicus.

Table 1.

Occurrence of respiratory pathogens in CIRD case dogs (Clinical = 1) versus control dogs (Clinical = 0) in a case-control study of 88 dogs in 5 Canadian veterinary clinics

Status	Mycoplasma cynos	CPIV	Bordetella bronchiseptica	CRCoV	CHV	CDV	N pathogens	N
Control	0	0	0	0	0	0	0	6
Control	+	0	0	0	0	0	1	13
Control	+	0	0	0	+	0	≥ 2	1
Control	+	0	+	0	0	0	≥ 2	2
N + ve	16	0	2	0	1	0		22
Mean	72.7%	0%	9.1%	0%	4.5%	0%
Case	0	0	0	0	0	0	0	5
Case	0	0	0	+	0	0	1	2
Case	0	0	+	0	0	0	1	1
Case	0	+	0	0	0	0	1	4
Case	+	0	0	0	0	0	1	20
Case	+	0	0	0	+	0	≥ 2	1
Case	+	0	0	+	0	0	≥ 2	3
Case	+	0	0	+	0	+	≥ 2	1
Case	+	0	+	0	0	0	≥ 2	4
Case	+	+	0	0	0	0	≥ 2	20
Case	+	+	+	0	0	0	≥ 2	3
N + ve	52	27	7	6	1	1		64
Mean (%)	81.2	42.2	10.9	9.4	1.6	1.6

CPIV — Canine parainfluenza virus; CRCoV — Canine respiratory coronavirus; CHV — Canine herpes virus; CDV — Canine distemper virus.

Eighty-one percent of case dogs and 73% of control dogs tested positive for M. cynos. No control dogs had CPIV, while 27/64 (42%) case dogs had the virus. No control animals had CRCoV detected compared with 6/64 (9.4%) case dogs. Little difference was found between the presence of B. bronchiseptica, CHV-1 and CDV among healthy and clinical animals (Table 1). In dogs with CIRDC, 32/64 (50.0%) had 2 or more pathogens present, while in dogs without CIRDC, only 3/22 (13.6%) had 2 or more pathogens (Table 1).

Table 2 contains basic demographic information on case and control dogs. Descriptively, a higher proportion of control dogs were large breed dogs, and > 5 y of age (Table 2). A lower proportion of control dogs were fertile (Table 2).

Table 2.

Demographics of study dogs with CIRDC from 5 Canadian veterinary clinics

	Case % (N)	Control % (N)
Age of dog
0 to 6 mo	14 (9)	5 (1)
> 6 mo to 1 year	20 (13)	5 (1)
> 1 y to 5 y	39 (25)	14 (3)
> 5 y	27 (17)	77 (17)
Breed (by size)
Unknown	0 (0)	5 (1)
Mixed	17 (11)	18 (4)
Toy	6 (4)	0 (0)
Small	30 (19)	18 (4)
Medium	9 (6)	5 (1)
Large	28 (18)	50 (11)
Giant	9 (6)	5 (1)
Gender
Unknown	0 (0)	5 (1)
Male	69 (44)	59 (13)
Female	31 (20)	36 (8)
Fertility status
Unknown	0 (0)	5 (1)
Infertile	73 (47)	86 (19)
Fertile	27 (17)	9 (2)
Core vaccination
None	14 (9)	9 (2)
Unknown	20 (13)	5 (1)
Vaccinated	66 (42)	86 (19)
Total N animals	64	22
CIRDCa vaccination
None	25 (16)	50 (11)
Unknown	19 (12)	5 (1)
Vaccinated	56 (36)	45 (10)
Total N animals	64	22

^aCIRDC vaccination defined as application of any vaccine for B. bronchiseptica through any route, with or without CPIV and CAV-2 component.

Table 3 contains variables that were univariably associated with the CIRDC status of dogs. All exposure variables (i.e., pathogens) regardless of statistical significance and demographic and management variables associated at the P < 0.20 level are shown. Vaccination status was not reported here.

Table 3.

Univariable associations between CIRD and diagnostic and demographic variables in the case-control study conducted in 5 Canadian small animal clinics

	Odds ratio	95% CI		P-valuea
Variable
CPIV (Canine parainfluenza virus)
Negative	Baseline	—	—	—
Positive	21.76b	3.53	∞	< 0.001
CRCoV (Canine respiratory corona virus)
Negative	Baseline	—	—	—
Positive	2.99b	0.41	∞	0.32
CDV (Canine distemper virus)
Negative	Baseline	—	—	—
Positive	0.34b	0.01	∞	1.00
CHV (Canine herpes virus)
Negative	Baseline	—	—	—
Positive	0.34	< 0.01	27.40	0.90
BB (Bordetella bronchiseptica)
Negative	Baseline	—	—	—
Positive	1.23	0.21	13.04	1.00
M. cynos
Negative	Baseline	—	—	—
Positive	1.62	0.43	5.62	0.57
Demographic variables
Age
> 12 mo	Baseline	—	—	—
0 to 12 mo	5.26	1.08	50.0	0.036
Fertility
Other	Baseline	—	—	—
Fertile	3.57	0.73	34.79	0.15
Gender
Other	Baseline	—	—	—
Female	0.80	0.26	2.57	0.85

^aP-values estimated using exact logistic regression.

^bEstimates represent “median unbiased regression estimates”.

95% CI — 95% confidence interval.

Results of the final multivariable regression models are reported in Table 4. Once adjusted for the effect of age and other demographic factors, dogs with CIRDC had higher odds of infection with CPIV (Table 4; OR = 20.07, 95% CI: 3.17, ∞, P < 0.001). Once adjusted for the effect of age, fertility status, and breed category, dogs with CIRDC had higher odds of infection with CRCoV (Table 4; OR = 6.11), although this association was not statistically significant at P < 0.05 level (Table 4, P = 0.087, 95% CI: 0.79, ∞). This association was based on a small number of dogs (Table 1) and requires care in interpretation. The variable representing age of dogs was forced into the model. In the model that was not adjusted for the effect of fertility and the breed category, the age of < 12 mo was a risk factor for development of CIRDC (OR = 4.54, P = 0.092, 95% CI: 0.83, 50). Once further adjusted for the fertility status and the breed category, the age of < 12 mo was not statistically significant (Table 4), but it was forced in the model as an important confounder. Spearman correlation coefficient indicated negative correlation between presence of CPIV and CRCoV when considering all data, but this was not statistically significant (r = −0.18, P = 0.088). When considering cases alone, the Spearman correlation coefficient indicated a negative correlation between the presence of CPIV and CRCoV at a level that was significant (r = −0.27, P = 0.028). No CIRDC-case dog had simultaneous detection of CPIV and CRCoV (Table 1).

Table 4.

Two final multivariable models of factors associated with CIRD in dogs in 5 Canadian small animals clinics based on exact logistic regression. Further adjustment was done with the fertility status and breed category, for which odds ratios were not estimated due to computational requirements

		Without further adjustment			With further adjustment
Variable	Level	Odds ratio	(95% CI)	P	Odds ratio	(95% CI)	P
CPIV	Negative	Baseline			Baseline
	Positive	22.89	(3.62, ∞)	< 0.001	20.07	(3.17, ∞)	< 0.001
CRCoV	Negative	Baseline			Baseline
	Positive	6.25	(0.82, ∞)	0.081	6.11	(0.79, ∞)	0.087
Age	< 12 months	Baseline			Baseline
	0 to 12 months	4.54	(0.83, 50)	0.092	5.55	(0.36, > 100)	0.33

95% CI — 95% confidence interval.

Discussion

The results of this study suggest that CPIV was associated with presence of CIRDC in this population of dogs, after adjustment for the effect of age, fertility, and dog size. Canine parainfluenza virus is a well-known canine respiratory pathogen that belongs to the Paramyxoviridae family of RNA viruses (3). The incubation period varies between 2 to 9 d (10,11) and clinical signs are typically present for 3 to 5 d (11). Usually, the animal is infectious between 2 and 10 d post infection (11) and viral shedding often precedes clinical signs (3). Clinical signs include a dry hacking cough, along with nasal discharge, pharyngitis, and tonsillitis. This virus is spread mainly through direct contact and infectious aerosols, and transmissibility and attack rates within susceptible populations are high. Our data, along with a recent outbreak case report (3), suggest that CPIV is an important contributor to CIRDC in the general dog population.

Data regarding the influence of CRCoV on CIRDC are less conclusive. The pathogen was first isolated from dogs with CIRDC in 2003 (2,12) and characterized as having importance in the development of CIRDC, especially in dogs entering kennels (1). CRCoV belongs to the Betacoronavirus genus of the Coronaviridae family and is distinct from the enteric canine coronavirus (CCoV), both genetically and serologically (2). It is not clear whether CRCoV alone can lead to clinical signs of CIRDC, or whether it plays a secondary role in disease and predisposes animals to become susceptible to infection by other respiratory pathogens (13). It has been shown recently that CRCoV likely plays a role in the early stages of CIRDC and, in combination with other pathogens, can lead to clinical signs like dry cough and nasal discharge (2). In the current study, infection with CRCoV approached significance in terms of association with the presence of CIRDC. As a pathogen that was only recently identified and for which testing is often limited, the role of CRCoV in CIRDC deserves further study (1,9). It is noteworthy that in the present study CIRDC cases were detected with CRCoV alone or in combination with M. cynos and that there were no dogs co-infected with CPIV and CRCoV.

The role of Mycoplasma in CIRDC has been difficult to discern, and data from this study provide no clarity. Mycoplasma cynos was found in 81% of affected dogs, but was also found in 73% of controls. There was no association between its presence and CIRDC, although previous work suggested that M. cynos could play a role in the development of the CIRDC (14–16). It is possible that diagnostic results based on pharyngeal swabs were not suitable for evaluating the role of M. cynos in CIRDC. In fact, Chalker et al (17) reported that M. cynos was the only canine Mycoplasma that showed significantly higher presence in CIRDC cases than controls, but only for samples taken from the lower respiratory tract by tracheal and bronchial lavage (BAL). Tracheal swabs and BAL were attempted for M. cynos detection in another study (14), but such sampling was beyond the scope of the current study.

The detection rate of B. bronchiseptica in this study was similar in CIRDC positive and negative dogs, despite a well-established role that this pathogen plays in the development of CIRDC (13). This finding is not surprising because B. bronchiseptica can be present in healthy dogs and in dogs affected by CIRDC, because of its ability to self-regulate its virulence (13). Overall, with the exception of B. bronchiseptica, the results of this study are in agreement with a study recently conducted in Germany (18).

The CIV was not detected in case or control animals, which is consistent with one Ontario surveillance study (19) and clinical observation that CIV is very rare to non-existent in the Canadian dog population. This contrasts with reports from parts of the US (20); however, it appears that CIV has a rather patchy distribution and is uncommonly implicated in endemic disease.

Co-infections in this study were common, largely because of the high prevalence of Mycoplasma. We did not identify statistically significant interactions among pathogens in this study. Overall, the role of co-infections in CIRDC is poorly understood (1,3,15) and is another area worth investigating.

Demographic variables were considered primarily as confounders. Nonetheless, when considered in isolation, an age of > 12 mo was protective and reproductive status was a risk factor. The protective effect of older age is likely a result of immunity against one or more pathogens through previous exposure or vaccination, and potentially a lower incidence of exposure to high risk situations (e.g., puppy classes).

This study has several limitations. Most importantly, selection of controls was based on convenience sampling and could lead to selection bias. Ideally, the control group should represent the source population from which the cases were recruited. Several dogs owned by veterinary staff at the clinics were sampled as a part of this group. If bias occurred because of that, we would argue that it could lead to an overestimate of associations due to viral pathogens, and underestimation of associations with bacterial pathogens. In future studies of this nature, more strict definition of the source population would help study validity. Recording of the response rate and comparison of demographics with the source population would also be of benefit. Misclassification bias could have occurred due to the nature of some measurements, including pharyngeal swabs. It is possible that with such measurement the true role of pathogens such as B. bronchiseptica and M. cynos could not be fully evaluated since both pathogens could be present in healthy animals (13,17). We were also not aware of diagnostic sensitivity and specificity of assays for pathogens used in the panel. Use of cotton-tipped wooden swabs could also affect diagnostic sensitivity of assays used in the study. The sample size calculation was not done before the study commenced and it is likely that the number of animals used in the study was sufficient to detect only strong associations. Vaccination against various respiratory pathogens could have affected the presence of pathogens. Vaccine type and vaccination protocols, time since their application, rationales for applying them, and a considerable amount of missing data made this task difficult. For future studies, we recommend that the effect of specific vaccines be investigated as the primary research question and that possible case control study be designed and conducted around this primary question.

In conclusion, we demonstrated that CPIV plays an important role in the development of CIRDC in dogs admitted to veterinary clinics, and under conditions of this study. The CRCoV was detected in cases only, but difference in the presence of this pathogen between cases and controls was not significant at P < 0.05. The frequency of M. cynos was similar in affected and control patients, so the significance of finding this organism in a pharyngeal swab is questionable. Sampling from the lower respiratory tract is likely necessary to confirm that M. cynos is associated with clinical respiratory disease. Practitioners should consider such possibility when interpreting diagnostic results obtained from predominantly pharyngeal swabs. No dogs with influenza virus were detected, which adds further support that influenza A virus has not been circulating in the Canadian dog population.