- Email: [email protected]
Coxiella burnetii seropositivity and associated risk factors in sheep in Ontario, Canada
Accepted Manuscript Title: Coxiella burnetii seropositivity and associated risk factors in sheep in Ontario, Canada Author: S. Meadows A. Jones-Bitton S. McEwen J. Jansen P. Menzies PII: DOI: Reference:
S0167-5877(15)00258-5 http://dx.doi.org/doi:10.1016/j.prevetmed.2015.07.007 PREVET 3844
To appear in:
PREVET
Received date: Revised date: Accepted date:
3-2-2015 16-7-2015 21-7-2015
Please cite this article as: Meadows, S., Jones-Bitton, A., McEwen, S., Jansen, J., Menzies, P., Coxiella burnetii seropositivity and associated risk factors in sheep in Ontario, Canada.Preventive Veterinary Medicine http://dx.doi.org/10.1016/j.prevetmed.2015.07.007 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
1 2 3 4 5 6 7
Coxiella burnetii seropositivity and associated risk factors in sheep in Ontario, Canada
8
S. Meadows1, A. Jones-Bitton1, S. McEwen1, J. Jansen2, P. Menzies1
9 1
10
Department of Population Medicine, University of Guelph, 50 Stone Road East, Guelph,
11
Ontario, N1G2W1, Canada 2
12
Veterinary Science and Policy, Ontario Ministry of Agriculture and Food, and Rural Affairs,
13
6484 Wellington Rd 7, Unit 10, Elora, Ontario, N0B 1S0, Canada
14 15
Corresponding author: Shannon Meadows1
16
Phone: +1 519 760 5844
17
Email: [email protected]
18
Highlights
19
Determined the proportion of sheep exposed to the bacterium, Coxiella burnetii, in Ontario
20
Investigated risk factors and protective measures associated with Coxiella burnetii exposure
21
in Ontario sheep
22 23 1
Present address: 6 Forest Laneway Apt 2801, North York, Ontario M2N 5X9, Canada 1
24
Abstract
25
Coxiella burnetii is a zoonotic bacterium that can cause abortion in sheep in late
26
gestation, as well as the delivery of stillborn, and non-viable lambs (Rodolakis, 2006). A cross-
27
sectional study was conducted in Ontario, Canada, to investigate C. burnetii exposure in sheep.
28
Between August 2010 and January 2012, sera from 2363 reproductively active ewes from 72
29
farms were tested for C. burnetii specific antibodies using the CHEKIT Q fever ELISA Test kit
30
(IDEXX Laboratories). Overall, exposure was common; sheep-level seroprevalence was 14.7%
31
(347/2363, 95% CI: 13.3-16.2), and was higher in dairy sheep (24.3%, 181/744) than meat sheep
32
(10.2%, 166/1619) (p<0.0001). At the farm-level, 48.6% (35/72, 95% CI: 37.2-60.1) of farms
33
had at least one seropositive sheep. A mixed multivariable logistic model that controlled for
34
farm-level clustering, identified risk factors associated (p≤0.05) with sheep seropositivity.
35
Increasing female flock size (logarithmic scale) was associated with increased odds of
36
seropositivity. By way of illustration, increasing the female flock size from 100 to 200 increased
37
the odds of seropositivity by 2.26 times (95% CI: 1.5-3.5). Sheep that lambed in an airspace
38
separate from the flock had 11.3 times (95% CI: 2.9-43.6) the odds of seropositivity relative to
39
other sheep. The practice of loaning sheep that returned to the farm increased odds of
40
seropositivity by 8.1 times (95% CI: 1.8-33.6). Lambing pen hygiene practices also influenced
41
odds of seropositivity. Relative to sheep from farms where all lambing pen hygiene measures
42
were practiced after lambing (i.e. adding bedding, removing birth materials and disinfection),
43
sheep from farms that only added bedding, or those that just added bedding and removed birthing
44
materials had 5.9 times (95% CI: 1.1-32.1) and 9.0 times (95% CI: 2.2-36.9) the odds of
45
seropositivity, respectively. These results can be used to inform prevention and control strategies
46
with the aim of reducing C. burnetii exposure in sheep.
2
47
Keywords: Coxiella burnetii, sheep, Ontario, seroprevalence, risk factors, coxiellosis
48
Introduction
49
Coxiella burnetii is a zoonotic bacterium and infection can result in clinical disease with
50
stillbirth or abortions in sheep, termed coxiellosis (Maurin and Raoult, 1999). This intracellular
51
pathogen has been recovered from a vast array of other mammals, birds and arthropods
52
worldwide, except in New Zealand (Astobiza et al., 2011; Enright et al., 1971; Maurin and
53
Raoult, 1999; Thompson et al., 2012). Human infection has most frequently been attributed to
54
proximity or contact with shedding ruminants, primarily sheep, goats and cattle (Astobiza et al.,
55
2011; Lyytikäinen et al., 1998; Thompson et al., 2012; Tissot-Dupont et al., 1999).
56
Infection in both humans and animals is considered to be acquired primarily by the
57
inhalation of aerosols contaminated with C. burnetii spore-like organisms (de Bruin et al., 2012;
58
Maurin and Raoult, 1999). Sporadic cases attributed to the ingestion of infective materials have
59
also been reported, but this transmission route is considered less efficient than aerosol (Fishbein
60
and Raoult, 1992; Welsh et al., 1945). In animals, C. burnetii transmission via the oral route is
61
thought to likely occur by ingesting contaminated placentas, pasture, hay and straw (Angelakis
62
and Raoult, 2010; Woldehiwet, 2004).
63
Coxiella burnetii infection in sheep is often subclinical, but it can cause abortion in late
64
gestation, as well as delivery of stillborn, non-viable lambs (Rodolakis, 2006). Large quantities
65
of C. burnetii can be shed into the environment by infected ewes in the birth fluids, placenta and
66
fetal membranes during an abortion or normal delivery (Berri et al., 2001; Masala et al., 2004;
67
Berri et al., 2005). Consequently, during the lambing period, the risk of disease transmission is
68
high (Astobiza et al., 2010). However, the risk of transmission extends beyond the lambing
69
period, as intermittent shedding may occur in vaginal secretions, feces, urine and milk for several
3
70
weeks or months following lambing (Berri et al., 2000; Berri et al., 2001). Contributing to the
71
extended period of transmission is the ability of C. burnetii organisms to remain infective for
72
months in aerosols or contaminated dust that continue to be released as the shed material
73
desiccates (Woldehiwet, 2004).
74
Q fever in humans and coxiellosis in sheep and goats have been recognized as endemic in
75
Ontario since the 1980s (Palmer et al., 1983; Simor, 1987; Simor et al., 1984). While the level of
76
exposure in Ontario sheep had not been evaluated since, in the 1980s, Lang et al. (1991),
77
reported a flock-level and sheep-level seroprevalence of 21.4% (22/103) and 1.5% (58/3765),
78
respectively, among Ontario sheep. The passive surveillance system currently used for
79
coxiellosis in Ontario sheep relies on the reporting of positive diagnostic tests, primarily abortion
80
diagnoses (McEwen et al., 2012). However, this surveillance likely underestimates the true
81
prevalence of C. burnetii infection due to a high proportion of infections where no
82
abortions/stillbirths occur (Sidi-Boumedine et al., 2010), and because of under-utilization of
83
available diagnostic services (Hazlett, et al., 2012).
84
Risk factors for C. burnetii exposure in animals include contact with local infected
85
animal reservoirs and specific animal management practices (Enright et al., 1971; van der Hoek
86
et al., 2011a). In Canada, there is evidence that sheep, rodents, cats, cattle, and goats may be a
87
source of C. burnetii shedding (Lang, 1988; Marrie et al., 1988; Thompson et al., 2012), but a
88
clearer understanding of the risks of disease transmission between these species is needed.
89
Having livestock contact or being in the vicinity of livestock, have been identified as significant
90
risk factors for human exposure and disease, suggesting that it is important to understand
91
infection of sheep in order to develop measures to protect humans (van der Hoek et al., 2011b).
92
However, data on risk factors for sheep exposure to C. burnetii and disease in Canada is limited.
4
93
Determination of the seroprevalence and risks for sheep exposure to C. burnetii could aid the
94
development and implementation of practical prevention and control strategies.
95
The objectives of this study were to: i) determine the flock-level and within-flock
96
seroprevalence of C. burnetii exposure in Ontario dairy and meat sheep farms, and (ii) identify
97
demographic and management practices that are risk factors for C. burnetii exposure in Ontario
98
sheep.
99
Materials and Methods
100
The University of Guelph Research Ethics Board (Certification of Ethical Acceptability
101
of Research Involving Human Participants – 10JN005) and the University of Guelph Animal
102
Care Committee (Animal Use Protocol – 10R056) approved the study.
103
This study used a cross-sectional design. All producers of sheep and/or wool in Ontario
104
are required to register with the Ontario Sheep Marketing Agency (OSMA). Meat sheep farms
105
were selected from the OSMA database with a stratified random sampling procedure by OSMA
106
district, using an online random number generator (http://stattrek.com/statistics/random-number-
107
generator.aspx). A comprehensive sampling frame for dairy sheep farms in Ontario does not
108
exist; therefore, lists of farms that milked sheep were obtained from sheep milk processors and
109
added to those from an incomplete list from the Ontario Ministry of Agriculture, Food and Rural
110
Affairs (OMAFRA). All identified dairy sheep producers in Ontario that met inclusion criteria
111
were solicited for participation. Producers (meat and dairy) were solicited via letter or telephone
112
and asked to provide information regarding their farm’s industry sector and flock numbers,
113
regardless of their intent to participate, to allow for assessment of selection bias. Farm inclusion
114
criteria were: having at least 10 ewes that had given birth in the previous 12 months, and being
5
115
located within 800 km of the University of Guelph. Prior to farm selection, regions beyond 800
116
km of Guelph were excluded due to logistical concerns and low populations of sheep.
117
To estimate the flock-level prevalence using an a priori estimate of 24% with a 95%
118
confidence and 10% allowable error, 72 sheep farms were sampled; specifically, 22 dairy sheep
119
farms and 50 meat sheep farms. The number of farms sampled in each sector (i.e. meat/dairy)
120
was proportional to the total estimated number of farms in that sector in Ontario. To estimate
121
within-farm seroprevalence using an a priori estimate of 10% with a 95% confidence and 10%
122
allowable error, 35 ewes per farm that had lambed in the previous 12 months were randomly
123
selected and sampled. If a farm had less than 35 ewes that lambed in the previous 12 months,
124
samples were taken from all ewes that met this requirement.
125
Questionnaires were administered on each participating farm by one person on the
126
research team via a personal interview of the farm manager at the time of animal sampling, or
127
were returned via postal mail after the farm visit was completed. The questionnaire is available
128
from the corresponding author upon request. Sampling and questionnaire administration
129
occurred between August 2010 and January 2012. Measures of reproductive failure, such as
130
average risk of abortions in the preceding three years, and prior reproductive-related diagnostic
131
testing, were also collected for descriptive purposes, and to gain insight as to whether current
132
management practices were implemented in response to awareness of prior reproductive
133
challenges in their sheep. These measures of reproductive failure were considered a possible
134
outcome of exposure to C. burnetii, and were therefore not included in the putative risk factor
135
model.
136 137
Blood samples were collected by jugular venipuncture into10 ml red top serum BD vacutainer tubes® (Becton, Dickson and Company, Franklin Lakes, New Jersey, USA).
6
138
Vacutainer tube samples were chilled on ice at 4-8oC. Sample centrifugation occurred at the
139
University of Guelph if transport time was less than 4 hours or at local veterinary clinics if
140
transport time exceeded 4 hours. Samples were centrifuged for 10 minutes at 1000 x g and 2ml
141
aliquots of the separated serum were then immediately pipetted into serum micro tubes (Fisher
142
Scientific, Ottawa, CA). Serum micro tubes that were processed at the University of Guelph
143
during laboratory business hours were submitted immediately to the Animal Health Laboratory
144
(AHL), Laboratory Services Division, University of Guelph. Sera were frozen for up to 48 hours
145
at -20oC prior to submission if either: they were prepared at the University of Guelph after
146
regular business hours, or if they were processed off-site at local veterinary clinics.
147
Serology was performed by the AHL using the CHEKIT Q-Fever Antibody ELISA Test
148
Kit (IDEXX Laboratories, Broomfield, CO, USA), which uses the Nine Mile antigen. This kit
149
detects both phase I and phase II antibodies to provide a cumulative serological outcome. The
150
results were expressed as the ratio between the Optical Density (OD) of the sample (S) and
151
positive control (P) in the test kit. In accordance with the manufacturer’s instructions, a S:P ratio
152
≥40% was considered seropositive. Horigan et al. (2011) evaluated the IDEXX ELISA’s test
153
accuracy in the absence of a gold standard and reported a 100% sensitivity and 99.6% specificity
154
in sheep.
155
Questionnaire and serological data were entered into EpiData® v2.2 for file management
156
(EpiData Association®, Odense Denmark) and were manually checked against the original
157
questionnaire for errors. A mixed logistic regression model was constructed in Stata Intercooled
158
Version 10.1 (StataCorp®, 2007) to assess putative risk factor associations with the outcome of
159
sheep seropositivity, while controlling for farm-level clustering using a random intercept.
7
160
The initial step in model building was univariable screening of all covariates for
161
association with sheep seropositivity at a liberal 20% significance level using the likelihood ratio
162
test (LRT). The linearity of continuous variables was assessed graphically by plotting lowess
163
smoother curves and transforming the variable, if necessary (Dohoo et al., 2003). Significant
164
variables in the univariable analysis were screened for pair-wise collinearity using Spearman’s
165
rank correlation and were considered collinear when Spearman’s rho was > |0.8| (Mason and
166
Perreault, 2013). Collinear pairs were compared with respect to univariable significance, missing
167
variables, and biological relationship, and the most relevant variable was retained (Dohoo et al.,
168
2003). Variables with more than 25% missing values were excluded from analysis. This cut-off
169
was chosen based on the distribution of missing values, as it minimized the number of putative
170
risk factors excluded from analysis and maximized the number of observations used in the
171
multivariable model. Retained covariates were used to construct the multivariable model using a
172
manual backward stepwise model-building procedure. All eliminated variables were then tested
173
for confounding and retained if their inclusion caused a >20% change in the coefficients of one
174
or more of the significant variables (Dohoo et al., 2003). Two-way interaction terms were
175
generated for biologically relevant covariate pairs, and retained in the model if significant
176
(p≤0.05). The fit of the model was assessed by examining upper level residuals (by farm) using
177
best linear unbiased predictions (BLUPs) analysis to verify the model assumptions of normality
178
and homoscedasticity (Dohoo et al., 2003). Additionally, residuals were visualized in both the
179
fixed and random portions of the model to identify potential outliers that would require further
180
investigation.
181
Results
182
Study Population and Seroprevalence
8
183
Fifty-five percent (22/40) of dairy sheep farms and 30.7% (50/163) of meat sheep farms
184
that were solicited and met the inclusion criteria participated in the study. The mean adult female
185
flock size on participating farms was 171 sheep (SD + 157), and the median was 100
186
(interquartile range 52-400). None of the sheep farms studied reported use of a commercial C.
187
burnetii vaccine (Coxevac®, CEVA Animal Health, Libourne, France).
188
At the farm-level, 48.6% (35/72, 95% CI: 37.2-60.1) of farms had at least one
189
seropositive sheep, and the seroprevalence was not significantly different between dairy sheep
190
(63.6%, 14/22) and meat sheep farms (42.0%, 21/50) (p=0.09). Figures 1 and 2 demonstrate the
191
distribution of farm-level and within-farm seroprevalence in the dairy and meat sectors,
192
respectively. The average within farm seroprevalence on dairy sheep farms was 23.5%, and
193
ranged from 0-74.3%, as seen in Figure 1. On meat sheep farms, the average within farm
194
seroprevalence was 9.5%, and ranged from 0-65.7%, as seen in Figure 2. Overall, the sheep-level
195
seroprevalence was 14.7% (347/2363, 95% CI: 13.3-16.2), and was significantly higher in dairy
196
sheep (24.3%, 181/744) than meat sheep (10.2%, 166/1619) (p<0.0001).
197
Risk factor analysis
198
The factors associated with sheep seropositivity in a univariable analysis at p <0.20, are
199
presented in Table 1. The following factors were also investigated, but were non-significant
200
(p>0.20): number of lambing groups; use of artificial insemination; purchase of replacement
201
females; number and sources of replacement animals; returning of sheep from an agricultural
202
fair; having dogs, cattle, llamas, alpacas, goats, deer, horses, donkeys, pigs or fowl on farm;
203
rodent control measures; presence of wildlife other than birds in the barn; lambing outdoors;
204
replacement animals being exposed to adult sheep, their manure, or lambing areas; spreading
205
manure; shearing or crutching ewes prior to lambing; quarantining ewes after aborting; and
9
206
disposal practices for placentas and aborted fetuses. The variables ‘purchase of male sheep’ and
207
‘closed flocks’ were negatively correlated (Spearman’s rho= -1.0); ‘closed flocks’ was dropped
208
from the model as it was less significant than “purchase of male sheep” at the univariable level,
209
and had the same number of observations. The final multivariable model is presented in Table 2.
210
Visual assessment of the BLUPs residuals identified three outlier farms. If these three
211
farms were removed, the BLUPs residuals were normally distributed, and the interpretation of
212
the multivariable model did not change. There was no reason for their exclusion, hence; all
213
outliers were retained in the multivariable model. The latent variable technique (Goldstein et al.,
214
2002) was used to calculate the variance components at the farm and individual levels.
215
Discussion
216
Exposure to C. burnetii was common among Ontario sheep tested in this study.
217
Monitoring of C. burnetii infection may be of particular interest in the dairy sheep industry in
218
Ontario since a higher sheep-level C. burnetii seroprevalence was identified in dairy sheep
219
compared to meat sheep. Sheep had higher odds of being seropositive if they were from farms
220
that loaned sheep to other farms. Allowing a ram to be temporarily borrowed by another farm
221
could increase the odds of seropositivity if returning sheep were exposed while away, and more
222
so if the re-introduced sheep shed C. burnetii, exposing other flock-mates. The risk of
223
introducing C. burnetii to a previously negative flock is the basis of the restrictions on trade and
224
animal movement during the small ruminant-associated Q fever epidemic in the Netherlands
225
(EFSA, 2010). Since loaning sheep to other farms was associated with a significant risk of
226
exposure, animal movement should be voluntarily restricted (“closed” flock policy), where
227
possible, to mitigate the risk of C. burnetii introduction. Alternatively, the practice of loaning
10
228
sheep could be a surrogate measure for reduced biosecurity, which could increase the risk of C.
229
burnetii exposure through other pathways, compared to farms that practice higher biosecurity.
230
The odds of sheep seropositivity also increased with increasing female flock size
231
(logarithmic scale). By way of illustration, increasing the female flock size from 100 to 200
232
increased the odds of seropositivity by 2.26 times (95% CI:1.5-3.5). This finding is consistent
233
with research from Great Britain, which indicated a higher flock size and number of breeding
234
ewes in the flock increased the likelihood of seropositive sheep (Lambton, et al., 2015).
235
Additionally, goat and cattle research has demonstrated that when a farm was infected, intensive
236
operations with larger herd sizes would result in a larger degree of environmental contamination
237
(Capuano et al., 2001; Hogerwerf et al., 2013). Producers from larger flocks should be mindful
238
that there is an increasing risk of exposure as the female flock size increases. Therefore, more
239
stringent hygiene practices and biosecurity may be required on large farms to counterbalance this
240
effect.
241
The proportion of animals exposed at a given time may also depend on the distribution of
242
bacteria in the animal environment. Since the lambing area is a likely place for heavy C. burnetii
243
environmental contamination (EFSA, 2010), it was anticipated that lambing in an airspace
244
separate from the rest of the flock (i.e. in a separate barn, or a separate dedicated section in the
245
barn with no direct air exchange with the rest of the flock), would serve to minimize the odds of
246
sheep seropositivity. However, when farm managers indicated that ewes sometimes lambed in an
247
airspace separate from the rest of the flock, ewes had significantly higher odds of being
248
seropositive, compared to those from farms that never lambed in a separate airspace. The 14
249
farms that sometimes lambed in a separate airspace were not significantly different from the 36
250
farms that never lambed in a separate airspace with respect to C. burnetii-caused abortions
11
251
(p=0.69), or elevated abortion risk (>5%) over the past three years (p=0.49) (data not shown).
252
Therefore, the association between sometimes lambing in a separate airspace and seropositivity
253
was not suspected to be due to reverse causation (i.e. farm managers segregating lambing only
254
after becoming aware of Coxiella-induced reproductive challenges). Research has found the use
255
of windbreak curtains and windshields in barns increased the risk of goat seropositivity
256
(Schimmer et al., 2011). These authors reasoned that the restricted air-flow decreased ventilation
257
and facilitated the accumulation of C. burnetii (Schimmer et al., 2011). Therefore, segregating
258
lambing in a separate airspace may have implications for ventilation or air-flow, dependent on
259
the facilities available. If ventilation is more restricted in the lambing area, this could encourage
260
environmental contamination with C. burnetii and promote transmission within the segregated
261
group of breeding ewes. However, empirical evidence testing this hypothesis is required.
262
Sheep from farms that only added bedding after lambings, and from those that added
263
bedding and removed birth materials, had significantly higher odds of seropositivity compared to
264
sheep from farms that did these things in addition to disinfecting lambing pens. This suggests
265
disinfection may be an effective tool for reducing C. burnetii burden in the lambing environment.
266
However, the type of disinfectant used was not assessed, and the efficacy of specific
267
disinfectants for use in lambing pens should be considered for future research. The odds of sheep
268
seropositivity was not significantly different among sheep from farms where none of the lambing
269
pen hygiene practices were in place compared to those where all hygiene measures were taken;
270
however, the majority of the former farms practiced outdoor lambing, which could decrease the
271
odds of sheep becoming exposed.
272 273
Wildlife contact was hypothesized to be a possible risk of exposure to C. burnetii for sheep, especially considering the high prevalence of C. burnetii found in genital swabs of wild
12
274
rodents in a national park in Ontario (Thompson et al., 2012). No association between observed
275
wildlife in the barn and sheep seropositivity was found in the present study. However, simply
276
asking farm managers about wildlife contact may under-estimate its frequency, as it requires
277
some evidence of the presence of wildlife, either from seeing the wild animal itself or seeing
278
droppings or nests, which may not always occur. Also, the presence of wildlife inside the barn
279
may not be required for transmission to sheep to occur, due to the capacity for aerosolized C.
280
burnetii to travel in the wind (van der Hoek et al., 2011b). Therefore, the degree of contact
281
necessary for transmission to occur from wildlife to livestock remains unclear. Future research
282
can clarify if wildlife pose a risk of C. burnetii exposure to sheep by identifying if wildlife
283
species on or near sheep farms have C. burnetii infections, and if so, determining whether the
284
same C. burnetii strains are circulating in wildlife and sheep.
285
The farm-level seroprevalence obtained in our study (48.6%), is higher than a previous
286
serosurvey in 1988 in Ontario (21.4%, 22/103) (Lang et al., 1991). The apparent increase in
287
seroprevalence in the province over time may signify further spread of C. burnetii, but the
288
methodological heterogeneity between these studies hinders direct comparisons. The non-
289
commercial ELISA used by Lang et al. (1991) is no longer in use and the degree of formal
290
agreement with the IDEXX ELISA (used here) is unknown. The fluctuations seen in the size of
291
the Canadian sheep population over the past decade (Statistics Canada, 2012) may also affect
292
flock-level and within-flock seroprevalence due to associated changes in sheep loaning,
293
purchasing behaviours and flock sizes.
294
While long-term persistence of serum antibodies after natural infection has been
295
described in sheep, between 21% (7/34) to 63% (10/16) of sheep in small epidemic scenarios
296
have shed C. burnetii but remain seronegative (Berri et al., 2005a, 2002, 2001). The expected
13
297
proportion of seronegative shedders in non-epidemic scenarios remains unknown. Therefore, it is
298
possible that both the farm- and sheep-level seroprevalence reported in this study under-estimate
299
the true prevalence of exposure if seronegative shedders were captured in this study sample. If
300
sheep were sampled after a recent infection, but before an immune response had been generated,
301
it is also possible that they could be falsely classified as non-exposed to C. burnetii (Berri et al.,
302
2002). Most new infections typically occur during the lambing season (Astobiza et al., 2010),
303
and sero-conversion in sheep typically occurs 3 to 4 weeks after infection (Brooks, 1986). False
304
negatives caused by sampling in the latent period are therefore likely minimized in this study
305
sample, as farm managers had a strong preference to schedule farm visits during non-lambing
306
periods when they had more time available. In contrast, evidence suggests waning immunity may
307
limit the number of exposed sheep identified via serology during the non-lambing periods, as
308
sheep sampled during late pregnancy and in the periparturient period have been shown to have a
309
higher risk of seropositivity, compared to sheep sampled during early or non-pregnancy (van den
310
Brom, et al., 2012).
311
The findings from this work can also be used to direct future research and reinforce the
312
importance of hygiene and biosecurity measures in preventing or mitigating exposure of sheep to
313
C. burnetii. Moving forward, a better understanding of how C. burnetii spreads within and
314
between sheep flocks, as well as from other host species, would help improve the ability to
315
implement effective management interventions.
316
Acknowledgements
317
The authors would like to thank: the Ontario Ministry of Agriculture, Food and Rural
318
Affairs - University of Guelph Agreement through the Animal Health Strategic Investment fund
319
managed by the Animal Health Laboratory of the University of Guelph; Ontario Ministry of
14
320
Health and Long-Term Care; National Sciences and Engineering Research Council of Canada;
321
Public Health Ontario; and the Ontario Sheep Marketing Agency. The authors would also like to
322
acknowledge David Pearl and William Sears for statistical advice, and the cooperation of all
323
sheep producers that participated in the study.
324
15
325
References
326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366
Angelakis, E., and D. Raoult, 2010: Q fever. Vet. Microbiol. 140, 297–309. Astobiza, I., J. F. Barandika, F. Ruiz-Fons, A. Hurtado, I. Povedano, R. a Juste, and a L. GarcíaPérez, 2010: Coxiella burnetii shedding and environmental contamination at lambing in two highly naturally-infected dairy sheep flocks after vaccination. Res. Vet. Sci. 91. Elsevier Ltd, 58–63. Berri, M., D. Crochet, S. Santiago, and A. Rodolakis, 2005: Spread of Coxiella burnetii infection in a flock of sheep after an episode of Q fever. Vet. Rec. 157, 737–740. Berri, M., K. Laroucau, and A. Rodolakis, 2000: The detection of Coxiella burnetii from ovine genital swabs, milk and fecal samples by the use of a single touchdown polymerase chain reaction. Vet. Microbiol. 72, 285–93. Berri, M., A. Souriau, M. Crosby, D. Crochet, A. Rodolakis, and P. Lechopier, 2001: Relationships between the shedding of Coxiella burnetii, clinical signs and serological responses of 34 sheep. Vet. Rec. 148, 502–5. Berri, M., A. Souriau, M. Crosby, and A. Rodolakis, 2002: Shedding of Coxiella burnetii in ewes in two pregnancies following an episode of Coxiella abortion in a sheep flock. Vet. Microbiol. 85, 55–60. Brooks, D., 1986: Q fever vaccination of sheep: challenge of immunity in ewes. Am. J. Vet. Res. 47, 1235–1238. Capuano, F., M. C. Landolfi, and D. M. Monetti, 2001: Influence of three types of farm management on the seroprevalence of Q fever as assessed by an indirect immunofluorescence assay. Vet. Rec. 149, 669–672. De Bruin, A., R. Q. J. van der Plaats, L. de Heer, R. Paauwe, B. Schimmer, P. Vellema, B. J. Van Rotterdam, and Y. T. H. P. van Duynhoven, 2012: Detection of Coxiella burnetii DNA on small-ruminant farms during a Q fever outbreak in the Netherlands. Appl. Environ. Microbiol. 78, 1652–1657. Dohoo, I., W. Martin, and H. Stryhn, 2003: Veterinary Epidemiologic Research. Prince Edward Island, Canada: AVC Incorporated. EFSA, 2010: Scientific opinion on Q fever. EFSA J. 8, 1–114. Enright, J. B., C. E. E. Franti, D. E. E. Behymer, W. M. M. Longhurst, V. J. J. Dutson, and M. E. E. Wright, 1971: Coxiella burnetii in a wildlife-livestock environment: distribution of Q fever in wild mammals. Am. J. Epidemiol. 94, 79–90. Fishbein, D. B., and D. Raoult, 1992: A cluster of Coxiella burnetii infections associated with exposure to vaccinated goats and their unpasteurized dairy products. Am. J. Trop. Med. Hyg. 47, 35–40. Goldstein, H., W. Browne, and J. Rasbash, 2002: Partitioning variation in multilevel models. Underst. Stat. 1, 223–231. Hogerwerf, L., A. Courcoul, D. Klinkenberg, F. Beaudeau, E. Vergu, and M. Nielen, 2013: Dairy goat demography and Q fever infection dynamics. Vet. Res. 44, 28. Horigan, M., M. Bell, T. Pollard, 2011: Q fever diagnosis in domestic ruminants: comparison between compliment fixation and commercial enzyme-linked immunesorbent assays. J. Vet. Diagn. Invest. 23 (5), 924-931. 16
367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409
Lambton, S. L., R. P. Smith, K. Gillard, M. Horigan, C. Farren and G. C. Pritchard, D. 2015. Serological survey using ELISA to determine the prevalence of Coxiella burnetii infection (Q fever) in sheep and goats in Great Britain. Epidemiology and Infection, available on CJO2015. doi:10.1017/S0950268815000874. Lang, G. H., D. Waltner-Toews, and P. Menzies, 1991: The seroprevalence of coxiellosis (Q fever) in Ontario sheep flocks. Can. J. Vet. Res. 55, 139–42. Lang, G. H. 1988: Serosurvey of Coxiella burnetii infection in dairy goat herds in Ontario. a comparison of two methods of enzyme-linked immunosorbent assay. Can. J. Vet. Res. 52, 37– 41. Lyytikäinen, O., T. Ziese, B. Schwartländer, P. Matzdorff, C. Kuhnhen, C. Jäger, and L. Petersen, 1998: An outbreak of sheep-associated Q fever in a rural community in Germany. Eur. J. Epidemiol. 14, 193–9. Marrie, T. J., H. Durant, J. C. Williams, E. Mintz, and D. M. Waag, 1988: Exposure to parturient cats : a risk factor for acquisition of Q fever in Maritime Canada. Ann. N. Y. Acad. Sci. 158, 101–108. Masala, G., R. Porcu, G. Sanna, G. Chessa, G. Cillara, V. Chisu, and S. Tola, 2004: Occurrence, distribution, and role in abortion of Coxiella burnetii in sheep and goats in Sardinia, Italy. Vet. Microbiol. 99, 301–5. Mason, C. H., and W. D. Perreault, 2013: Collinearity, power and interpretation of multiple regression analysis. J. Mark. Res. 28, 268–280. Maurin, M., and D. Raoult, 1999: Q fever. Clin. Microbiol. Rev. 12, 518–553. McEwen, B., D. Slavic, D. Ojkic, J. Delay, H. Cai, M. Stalker, M. Hazlett, and M. Brash, 2012: Selected zoonotic pathogens and diseases identified at the Animal Health Laboratory, 2011. Anim. Heal. Lab. Newsl., March. Palmer, N. C., M. Kierstead, D. W. Key, J. C. Williams, M. G. Peacock, H. Vellend, and T. H. E. Editor, 1983: Placentitis and abortion in goats and sheep in Ontario caused by Coxiella burnetii. Can. Vet. J. 24, 60–61. Rodolakis, A., 2006: Q fever, state of art: epidemiology, diagnosis and prophylaxis. Small Rumin. Res. 62, 121–124. Schimmer, B., S. Luttikholt, J. L. A. Hautvast, E. A. M. Graat, P. Vellema, and Y. T. H. P. van Duynhoven, 2011: Seroprevalence and risk factors of Q fever in goats on commercial dairy goat farms in the Netherlands, 2009-2010. BMC Vet. Res. 7. BioMed Central Ltd, 81. Sidi-Boumedine, K., E. Rousset, K. Henning, M. Ziller, K. Niemczuck, H. I. J. Roest, and R. Thiéry, 2010: Development of harmonised schemes for the monitoring and reporting of Q fever in animals in the European union. Vet. Res. Parma, Italy. Simor, A. E., J. L. Brunton, I. E. Salit, H. Vellend, L. Ford-Jones, and L. P. Spence, 1984: Q fever: hazard from sheep used in research. Can. Med. Assoc. J. 130, 1013–1016. Simor, A. E., 1987: Q fever - human disease in Ontario. Can. Vet. J. 28, 264–266. Statistics Canada, 2012: Sheep and lambs on farm - Canada. Sheep Statistics. http://www5.statcan.gc.ca/bsolc/olc-cel/olc-cel?catno=23-011-x&lang=eng. Thompson, M., N. Mykytczuk, K. Gooderham, and A. Schulte-Hostedde, 2012: Prevalence of the bacterium Coxiella burnetii in wild rodents from a Canadian natural environment park. Zoonoses Public Health 59, 553–60. 17
410 411 412 413 414 415 416 417 418 419 420 421 422 423 424
Tissot-Dupont, H., S. Torres, M. Nezri, and D. Raoult, 1999: Hyperendemic focus of Q fever related to sheep and wind. Am. J. Epidemiol. 150, 67–74. van den Brom, R., L. Moll, G. van Schaik, P. Vellema, 2012. Demography of Q fever seroprevalence in sheep and goats in The Netherlands in 2008. Prev. Vet. Med. 109 (April 12), 76-82. http://dx.doi.org/10.1016/j.prevetmed.2012.09.002 van der Hoek, Wim, J. C. E. Meekelenkamp, F. Dijkstra, D. W. Notermans, B. Bom, P. Vellema, A. Rietveld, Y. van Duynhoven, and A. C. A. P. Leenders, 2011a: Proximity to goat farms and seroprevalence among pregnant women. Emerg. Infect. Dis. 17, 2360–2363. van der Hoek, Wim, J. Hunink, P. Vellema, and P. Droogers, 2011b: Q fever in the Netherlands : the role of local environmental conditions. Int. J. Environ. Health Res. 21, 441–451. Welsh, H. H., E. H. Lennette, F. R. Abinanti, and J. F. Winns, 1945: Air-borne transmission of Q fever: the role of parturition in the generation of infective aerosols. Ann. N. Y. Acad. Sci. 70, 528–540. Woldehiwet, Z., 2004: Q fever (coxiellosis): epidemiology and pathogenesis. Res. Vet. Sci. 77, 93–100.
18
425
426 Figure 1 Percentage of C. burnetii seropositive sheep per farm on 22 dairy sheep farms in Ontario, Canada (Aug 2010-Jan 2012)* * Farms have been sorted from lowest to highest within-herd seroprevalence
427
428
Figure 2 Percentage of C. burnetii seropositive sheep per farm on 50 randomly selected meat sheep farms in Ontario, Canada (Aug 2010-Jan 2012)* * Farms have been sorted from lowest to highest within-herd seroprevalence
429 430
19
Table 1 Farm management variables associated (p<0.2) with sheep seropositivity for C. burnetii, as observed through univariable mixed-effects logistic regression of data collected from 72 sheep farms in Ontario, Canada (Aug 2010-Jan 2012) Risk Factors OR OR 95% CI P-value Sector Meat Ref Dairy 4.87 1.07-22.24 0.041 Female flock size (Log10 scale) 1.89 1.19-2.99 0.007 Lambing pen cleaning practices (LRT χ2 = 387.23, P-value = 0.0491) Add bedding, remove birthing materials and Ref disinfecting pen Add bedding and remove birthing materials 6.53 1.15-37.03 0.034 Add bedding only 6.27 0.79-49.89 0.083 Do nothing 0.18 0.009-3.48 0.257 Loaning sheep that return to farm (yes/no) 8.25 1.25-54.27 0.028 Closed flock (yes/no) 0.15 0.009-2.34 0.175 Purchasing sheep from sales barns (yes/no) 7.89 0.49-126.90 0.145 Purchasing replacement males (yes/no) 6.82 0.69-4.22 0.101 Poison baits for rodent control (yes/no) 4.79 0.81-28.59 0.084 Barn cats on farm (yes/no) 6.28 0.51-77.99 0.153 Lambing ewes in separate area from flock (LRT χ2 = 362.96, P-value = 0.1310) Never Ref Not Applicable (do not lamb indoors) 0.18 -4.85-4.075 0.281 Sometimes 4.93 0.86-28.20 0.073 Always 2.87 0.34-24.44 0.336 Days replacement lambs spend with dam before 0.50 0.23-1.08 0.076 weaning (Log10 scale) Other sheep or goat farms within 5km (LRT χ2 = 411.79, P-value = 0.1774) No Ref Yes 5.59 0.86-36.13 0.071 Unknown 8.12 0.23-289.00 0.251 OR = odds ratio, Ref= Referent category LRT = Likelihood ratio test for categorized variable
431 432
20
Table 2 Farm management variables associated (p<0.05) with C. burnetii seropositivity in sheep, as observed through multivariable mixed-effects logistic regression of data collected from 72 sheep farms in Ontario, Canada (Aug 2010-Jan 2012) Risk Factors OR OR 95% CI P-value Lambing pen cleaning practices (LRT χ2 = 11.71, p = 0.0084) a Add bedding, remove birthing materials and Ref disinfect Add bedding and remove birthing materials 8.96 2.17-36.90 0.002 Add bedding only 5.94 1.10-32.12 0.038 Do nothing 0.97 0.04-23.29 0.986 Lambing ewes housed in separate airspace from flock (LRT χ2 = 11.77, p = 0.0082) a Always 2.08 0.43-10.10 0.362 Sometimes 11.26 2.91-43.56 <0.001 Never Ref Not Applicable (do not lamb indoors) 0.44 0.02-11.01 0.614 Loaning sheep that return to farm (yes/no) 8.14 1.81-36.61 0.006 Female flock size (Log10 scale) 3.24 1.71-6.15 <0.001 Random Intercept Parameter SD 95% CI P-value Farm b 1.59 1.11-2.26 <0.0001 OR = odds ratio, Ref= Referent category, SD= Standard deviation a LRT = Likelihood ratio test for categorized variable b Partitioning the variance, 43.3% occurred at the farm-level and 56.7% occurred at the sheeplevel
433 434
21
435 436
22
S0167-5877(15)00258-5 http://dx.doi.org/doi:10.1016/j.prevetmed.2015.07.007 PREVET 3844
To appear in:
PREVET
Received date: Revised date: Accepted date:
3-2-2015 16-7-2015 21-7-2015
Please cite this article as: Meadows, S., Jones-Bitton, A., McEwen, S., Jansen, J., Menzies, P., Coxiella burnetii seropositivity and associated risk factors in sheep in Ontario, Canada.Preventive Veterinary Medicine http://dx.doi.org/10.1016/j.prevetmed.2015.07.007 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
1 2 3 4 5 6 7
Coxiella burnetii seropositivity and associated risk factors in sheep in Ontario, Canada
8
S. Meadows1, A. Jones-Bitton1, S. McEwen1, J. Jansen2, P. Menzies1
9 1
10
Department of Population Medicine, University of Guelph, 50 Stone Road East, Guelph,
11
Ontario, N1G2W1, Canada 2
12
Veterinary Science and Policy, Ontario Ministry of Agriculture and Food, and Rural Affairs,
13
6484 Wellington Rd 7, Unit 10, Elora, Ontario, N0B 1S0, Canada
14 15
Corresponding author: Shannon Meadows1
16
Phone: +1 519 760 5844
17
Email: [email protected]
18
Highlights
19
Determined the proportion of sheep exposed to the bacterium, Coxiella burnetii, in Ontario
20
Investigated risk factors and protective measures associated with Coxiella burnetii exposure
21
in Ontario sheep
22 23 1
Present address: 6 Forest Laneway Apt 2801, North York, Ontario M2N 5X9, Canada 1
24
Abstract
25
Coxiella burnetii is a zoonotic bacterium that can cause abortion in sheep in late
26
gestation, as well as the delivery of stillborn, and non-viable lambs (Rodolakis, 2006). A cross-
27
sectional study was conducted in Ontario, Canada, to investigate C. burnetii exposure in sheep.
28
Between August 2010 and January 2012, sera from 2363 reproductively active ewes from 72
29
farms were tested for C. burnetii specific antibodies using the CHEKIT Q fever ELISA Test kit
30
(IDEXX Laboratories). Overall, exposure was common; sheep-level seroprevalence was 14.7%
31
(347/2363, 95% CI: 13.3-16.2), and was higher in dairy sheep (24.3%, 181/744) than meat sheep
32
(10.2%, 166/1619) (p<0.0001). At the farm-level, 48.6% (35/72, 95% CI: 37.2-60.1) of farms
33
had at least one seropositive sheep. A mixed multivariable logistic model that controlled for
34
farm-level clustering, identified risk factors associated (p≤0.05) with sheep seropositivity.
35
Increasing female flock size (logarithmic scale) was associated with increased odds of
36
seropositivity. By way of illustration, increasing the female flock size from 100 to 200 increased
37
the odds of seropositivity by 2.26 times (95% CI: 1.5-3.5). Sheep that lambed in an airspace
38
separate from the flock had 11.3 times (95% CI: 2.9-43.6) the odds of seropositivity relative to
39
other sheep. The practice of loaning sheep that returned to the farm increased odds of
40
seropositivity by 8.1 times (95% CI: 1.8-33.6). Lambing pen hygiene practices also influenced
41
odds of seropositivity. Relative to sheep from farms where all lambing pen hygiene measures
42
were practiced after lambing (i.e. adding bedding, removing birth materials and disinfection),
43
sheep from farms that only added bedding, or those that just added bedding and removed birthing
44
materials had 5.9 times (95% CI: 1.1-32.1) and 9.0 times (95% CI: 2.2-36.9) the odds of
45
seropositivity, respectively. These results can be used to inform prevention and control strategies
46
with the aim of reducing C. burnetii exposure in sheep.
2
47
Keywords: Coxiella burnetii, sheep, Ontario, seroprevalence, risk factors, coxiellosis
48
Introduction
49
Coxiella burnetii is a zoonotic bacterium and infection can result in clinical disease with
50
stillbirth or abortions in sheep, termed coxiellosis (Maurin and Raoult, 1999). This intracellular
51
pathogen has been recovered from a vast array of other mammals, birds and arthropods
52
worldwide, except in New Zealand (Astobiza et al., 2011; Enright et al., 1971; Maurin and
53
Raoult, 1999; Thompson et al., 2012). Human infection has most frequently been attributed to
54
proximity or contact with shedding ruminants, primarily sheep, goats and cattle (Astobiza et al.,
55
2011; Lyytikäinen et al., 1998; Thompson et al., 2012; Tissot-Dupont et al., 1999).
56
Infection in both humans and animals is considered to be acquired primarily by the
57
inhalation of aerosols contaminated with C. burnetii spore-like organisms (de Bruin et al., 2012;
58
Maurin and Raoult, 1999). Sporadic cases attributed to the ingestion of infective materials have
59
also been reported, but this transmission route is considered less efficient than aerosol (Fishbein
60
and Raoult, 1992; Welsh et al., 1945). In animals, C. burnetii transmission via the oral route is
61
thought to likely occur by ingesting contaminated placentas, pasture, hay and straw (Angelakis
62
and Raoult, 2010; Woldehiwet, 2004).
63
Coxiella burnetii infection in sheep is often subclinical, but it can cause abortion in late
64
gestation, as well as delivery of stillborn, non-viable lambs (Rodolakis, 2006). Large quantities
65
of C. burnetii can be shed into the environment by infected ewes in the birth fluids, placenta and
66
fetal membranes during an abortion or normal delivery (Berri et al., 2001; Masala et al., 2004;
67
Berri et al., 2005). Consequently, during the lambing period, the risk of disease transmission is
68
high (Astobiza et al., 2010). However, the risk of transmission extends beyond the lambing
69
period, as intermittent shedding may occur in vaginal secretions, feces, urine and milk for several
3
70
weeks or months following lambing (Berri et al., 2000; Berri et al., 2001). Contributing to the
71
extended period of transmission is the ability of C. burnetii organisms to remain infective for
72
months in aerosols or contaminated dust that continue to be released as the shed material
73
desiccates (Woldehiwet, 2004).
74
Q fever in humans and coxiellosis in sheep and goats have been recognized as endemic in
75
Ontario since the 1980s (Palmer et al., 1983; Simor, 1987; Simor et al., 1984). While the level of
76
exposure in Ontario sheep had not been evaluated since, in the 1980s, Lang et al. (1991),
77
reported a flock-level and sheep-level seroprevalence of 21.4% (22/103) and 1.5% (58/3765),
78
respectively, among Ontario sheep. The passive surveillance system currently used for
79
coxiellosis in Ontario sheep relies on the reporting of positive diagnostic tests, primarily abortion
80
diagnoses (McEwen et al., 2012). However, this surveillance likely underestimates the true
81
prevalence of C. burnetii infection due to a high proportion of infections where no
82
abortions/stillbirths occur (Sidi-Boumedine et al., 2010), and because of under-utilization of
83
available diagnostic services (Hazlett, et al., 2012).
84
Risk factors for C. burnetii exposure in animals include contact with local infected
85
animal reservoirs and specific animal management practices (Enright et al., 1971; van der Hoek
86
et al., 2011a). In Canada, there is evidence that sheep, rodents, cats, cattle, and goats may be a
87
source of C. burnetii shedding (Lang, 1988; Marrie et al., 1988; Thompson et al., 2012), but a
88
clearer understanding of the risks of disease transmission between these species is needed.
89
Having livestock contact or being in the vicinity of livestock, have been identified as significant
90
risk factors for human exposure and disease, suggesting that it is important to understand
91
infection of sheep in order to develop measures to protect humans (van der Hoek et al., 2011b).
92
However, data on risk factors for sheep exposure to C. burnetii and disease in Canada is limited.
4
93
Determination of the seroprevalence and risks for sheep exposure to C. burnetii could aid the
94
development and implementation of practical prevention and control strategies.
95
The objectives of this study were to: i) determine the flock-level and within-flock
96
seroprevalence of C. burnetii exposure in Ontario dairy and meat sheep farms, and (ii) identify
97
demographic and management practices that are risk factors for C. burnetii exposure in Ontario
98
sheep.
99
Materials and Methods
100
The University of Guelph Research Ethics Board (Certification of Ethical Acceptability
101
of Research Involving Human Participants – 10JN005) and the University of Guelph Animal
102
Care Committee (Animal Use Protocol – 10R056) approved the study.
103
This study used a cross-sectional design. All producers of sheep and/or wool in Ontario
104
are required to register with the Ontario Sheep Marketing Agency (OSMA). Meat sheep farms
105
were selected from the OSMA database with a stratified random sampling procedure by OSMA
106
district, using an online random number generator (http://stattrek.com/statistics/random-number-
107
generator.aspx). A comprehensive sampling frame for dairy sheep farms in Ontario does not
108
exist; therefore, lists of farms that milked sheep were obtained from sheep milk processors and
109
added to those from an incomplete list from the Ontario Ministry of Agriculture, Food and Rural
110
Affairs (OMAFRA). All identified dairy sheep producers in Ontario that met inclusion criteria
111
were solicited for participation. Producers (meat and dairy) were solicited via letter or telephone
112
and asked to provide information regarding their farm’s industry sector and flock numbers,
113
regardless of their intent to participate, to allow for assessment of selection bias. Farm inclusion
114
criteria were: having at least 10 ewes that had given birth in the previous 12 months, and being
5
115
located within 800 km of the University of Guelph. Prior to farm selection, regions beyond 800
116
km of Guelph were excluded due to logistical concerns and low populations of sheep.
117
To estimate the flock-level prevalence using an a priori estimate of 24% with a 95%
118
confidence and 10% allowable error, 72 sheep farms were sampled; specifically, 22 dairy sheep
119
farms and 50 meat sheep farms. The number of farms sampled in each sector (i.e. meat/dairy)
120
was proportional to the total estimated number of farms in that sector in Ontario. To estimate
121
within-farm seroprevalence using an a priori estimate of 10% with a 95% confidence and 10%
122
allowable error, 35 ewes per farm that had lambed in the previous 12 months were randomly
123
selected and sampled. If a farm had less than 35 ewes that lambed in the previous 12 months,
124
samples were taken from all ewes that met this requirement.
125
Questionnaires were administered on each participating farm by one person on the
126
research team via a personal interview of the farm manager at the time of animal sampling, or
127
were returned via postal mail after the farm visit was completed. The questionnaire is available
128
from the corresponding author upon request. Sampling and questionnaire administration
129
occurred between August 2010 and January 2012. Measures of reproductive failure, such as
130
average risk of abortions in the preceding three years, and prior reproductive-related diagnostic
131
testing, were also collected for descriptive purposes, and to gain insight as to whether current
132
management practices were implemented in response to awareness of prior reproductive
133
challenges in their sheep. These measures of reproductive failure were considered a possible
134
outcome of exposure to C. burnetii, and were therefore not included in the putative risk factor
135
model.
136 137
Blood samples were collected by jugular venipuncture into10 ml red top serum BD vacutainer tubes® (Becton, Dickson and Company, Franklin Lakes, New Jersey, USA).
6
138
Vacutainer tube samples were chilled on ice at 4-8oC. Sample centrifugation occurred at the
139
University of Guelph if transport time was less than 4 hours or at local veterinary clinics if
140
transport time exceeded 4 hours. Samples were centrifuged for 10 minutes at 1000 x g and 2ml
141
aliquots of the separated serum were then immediately pipetted into serum micro tubes (Fisher
142
Scientific, Ottawa, CA). Serum micro tubes that were processed at the University of Guelph
143
during laboratory business hours were submitted immediately to the Animal Health Laboratory
144
(AHL), Laboratory Services Division, University of Guelph. Sera were frozen for up to 48 hours
145
at -20oC prior to submission if either: they were prepared at the University of Guelph after
146
regular business hours, or if they were processed off-site at local veterinary clinics.
147
Serology was performed by the AHL using the CHEKIT Q-Fever Antibody ELISA Test
148
Kit (IDEXX Laboratories, Broomfield, CO, USA), which uses the Nine Mile antigen. This kit
149
detects both phase I and phase II antibodies to provide a cumulative serological outcome. The
150
results were expressed as the ratio between the Optical Density (OD) of the sample (S) and
151
positive control (P) in the test kit. In accordance with the manufacturer’s instructions, a S:P ratio
152
≥40% was considered seropositive. Horigan et al. (2011) evaluated the IDEXX ELISA’s test
153
accuracy in the absence of a gold standard and reported a 100% sensitivity and 99.6% specificity
154
in sheep.
155
Questionnaire and serological data were entered into EpiData® v2.2 for file management
156
(EpiData Association®, Odense Denmark) and were manually checked against the original
157
questionnaire for errors. A mixed logistic regression model was constructed in Stata Intercooled
158
Version 10.1 (StataCorp®, 2007) to assess putative risk factor associations with the outcome of
159
sheep seropositivity, while controlling for farm-level clustering using a random intercept.
7
160
The initial step in model building was univariable screening of all covariates for
161
association with sheep seropositivity at a liberal 20% significance level using the likelihood ratio
162
test (LRT). The linearity of continuous variables was assessed graphically by plotting lowess
163
smoother curves and transforming the variable, if necessary (Dohoo et al., 2003). Significant
164
variables in the univariable analysis were screened for pair-wise collinearity using Spearman’s
165
rank correlation and were considered collinear when Spearman’s rho was > |0.8| (Mason and
166
Perreault, 2013). Collinear pairs were compared with respect to univariable significance, missing
167
variables, and biological relationship, and the most relevant variable was retained (Dohoo et al.,
168
2003). Variables with more than 25% missing values were excluded from analysis. This cut-off
169
was chosen based on the distribution of missing values, as it minimized the number of putative
170
risk factors excluded from analysis and maximized the number of observations used in the
171
multivariable model. Retained covariates were used to construct the multivariable model using a
172
manual backward stepwise model-building procedure. All eliminated variables were then tested
173
for confounding and retained if their inclusion caused a >20% change in the coefficients of one
174
or more of the significant variables (Dohoo et al., 2003). Two-way interaction terms were
175
generated for biologically relevant covariate pairs, and retained in the model if significant
176
(p≤0.05). The fit of the model was assessed by examining upper level residuals (by farm) using
177
best linear unbiased predictions (BLUPs) analysis to verify the model assumptions of normality
178
and homoscedasticity (Dohoo et al., 2003). Additionally, residuals were visualized in both the
179
fixed and random portions of the model to identify potential outliers that would require further
180
investigation.
181
Results
182
Study Population and Seroprevalence
8
183
Fifty-five percent (22/40) of dairy sheep farms and 30.7% (50/163) of meat sheep farms
184
that were solicited and met the inclusion criteria participated in the study. The mean adult female
185
flock size on participating farms was 171 sheep (SD + 157), and the median was 100
186
(interquartile range 52-400). None of the sheep farms studied reported use of a commercial C.
187
burnetii vaccine (Coxevac®, CEVA Animal Health, Libourne, France).
188
At the farm-level, 48.6% (35/72, 95% CI: 37.2-60.1) of farms had at least one
189
seropositive sheep, and the seroprevalence was not significantly different between dairy sheep
190
(63.6%, 14/22) and meat sheep farms (42.0%, 21/50) (p=0.09). Figures 1 and 2 demonstrate the
191
distribution of farm-level and within-farm seroprevalence in the dairy and meat sectors,
192
respectively. The average within farm seroprevalence on dairy sheep farms was 23.5%, and
193
ranged from 0-74.3%, as seen in Figure 1. On meat sheep farms, the average within farm
194
seroprevalence was 9.5%, and ranged from 0-65.7%, as seen in Figure 2. Overall, the sheep-level
195
seroprevalence was 14.7% (347/2363, 95% CI: 13.3-16.2), and was significantly higher in dairy
196
sheep (24.3%, 181/744) than meat sheep (10.2%, 166/1619) (p<0.0001).
197
Risk factor analysis
198
The factors associated with sheep seropositivity in a univariable analysis at p <0.20, are
199
presented in Table 1. The following factors were also investigated, but were non-significant
200
(p>0.20): number of lambing groups; use of artificial insemination; purchase of replacement
201
females; number and sources of replacement animals; returning of sheep from an agricultural
202
fair; having dogs, cattle, llamas, alpacas, goats, deer, horses, donkeys, pigs or fowl on farm;
203
rodent control measures; presence of wildlife other than birds in the barn; lambing outdoors;
204
replacement animals being exposed to adult sheep, their manure, or lambing areas; spreading
205
manure; shearing or crutching ewes prior to lambing; quarantining ewes after aborting; and
9
206
disposal practices for placentas and aborted fetuses. The variables ‘purchase of male sheep’ and
207
‘closed flocks’ were negatively correlated (Spearman’s rho= -1.0); ‘closed flocks’ was dropped
208
from the model as it was less significant than “purchase of male sheep” at the univariable level,
209
and had the same number of observations. The final multivariable model is presented in Table 2.
210
Visual assessment of the BLUPs residuals identified three outlier farms. If these three
211
farms were removed, the BLUPs residuals were normally distributed, and the interpretation of
212
the multivariable model did not change. There was no reason for their exclusion, hence; all
213
outliers were retained in the multivariable model. The latent variable technique (Goldstein et al.,
214
2002) was used to calculate the variance components at the farm and individual levels.
215
Discussion
216
Exposure to C. burnetii was common among Ontario sheep tested in this study.
217
Monitoring of C. burnetii infection may be of particular interest in the dairy sheep industry in
218
Ontario since a higher sheep-level C. burnetii seroprevalence was identified in dairy sheep
219
compared to meat sheep. Sheep had higher odds of being seropositive if they were from farms
220
that loaned sheep to other farms. Allowing a ram to be temporarily borrowed by another farm
221
could increase the odds of seropositivity if returning sheep were exposed while away, and more
222
so if the re-introduced sheep shed C. burnetii, exposing other flock-mates. The risk of
223
introducing C. burnetii to a previously negative flock is the basis of the restrictions on trade and
224
animal movement during the small ruminant-associated Q fever epidemic in the Netherlands
225
(EFSA, 2010). Since loaning sheep to other farms was associated with a significant risk of
226
exposure, animal movement should be voluntarily restricted (“closed” flock policy), where
227
possible, to mitigate the risk of C. burnetii introduction. Alternatively, the practice of loaning
10
228
sheep could be a surrogate measure for reduced biosecurity, which could increase the risk of C.
229
burnetii exposure through other pathways, compared to farms that practice higher biosecurity.
230
The odds of sheep seropositivity also increased with increasing female flock size
231
(logarithmic scale). By way of illustration, increasing the female flock size from 100 to 200
232
increased the odds of seropositivity by 2.26 times (95% CI:1.5-3.5). This finding is consistent
233
with research from Great Britain, which indicated a higher flock size and number of breeding
234
ewes in the flock increased the likelihood of seropositive sheep (Lambton, et al., 2015).
235
Additionally, goat and cattle research has demonstrated that when a farm was infected, intensive
236
operations with larger herd sizes would result in a larger degree of environmental contamination
237
(Capuano et al., 2001; Hogerwerf et al., 2013). Producers from larger flocks should be mindful
238
that there is an increasing risk of exposure as the female flock size increases. Therefore, more
239
stringent hygiene practices and biosecurity may be required on large farms to counterbalance this
240
effect.
241
The proportion of animals exposed at a given time may also depend on the distribution of
242
bacteria in the animal environment. Since the lambing area is a likely place for heavy C. burnetii
243
environmental contamination (EFSA, 2010), it was anticipated that lambing in an airspace
244
separate from the rest of the flock (i.e. in a separate barn, or a separate dedicated section in the
245
barn with no direct air exchange with the rest of the flock), would serve to minimize the odds of
246
sheep seropositivity. However, when farm managers indicated that ewes sometimes lambed in an
247
airspace separate from the rest of the flock, ewes had significantly higher odds of being
248
seropositive, compared to those from farms that never lambed in a separate airspace. The 14
249
farms that sometimes lambed in a separate airspace were not significantly different from the 36
250
farms that never lambed in a separate airspace with respect to C. burnetii-caused abortions
11
251
(p=0.69), or elevated abortion risk (>5%) over the past three years (p=0.49) (data not shown).
252
Therefore, the association between sometimes lambing in a separate airspace and seropositivity
253
was not suspected to be due to reverse causation (i.e. farm managers segregating lambing only
254
after becoming aware of Coxiella-induced reproductive challenges). Research has found the use
255
of windbreak curtains and windshields in barns increased the risk of goat seropositivity
256
(Schimmer et al., 2011). These authors reasoned that the restricted air-flow decreased ventilation
257
and facilitated the accumulation of C. burnetii (Schimmer et al., 2011). Therefore, segregating
258
lambing in a separate airspace may have implications for ventilation or air-flow, dependent on
259
the facilities available. If ventilation is more restricted in the lambing area, this could encourage
260
environmental contamination with C. burnetii and promote transmission within the segregated
261
group of breeding ewes. However, empirical evidence testing this hypothesis is required.
262
Sheep from farms that only added bedding after lambings, and from those that added
263
bedding and removed birth materials, had significantly higher odds of seropositivity compared to
264
sheep from farms that did these things in addition to disinfecting lambing pens. This suggests
265
disinfection may be an effective tool for reducing C. burnetii burden in the lambing environment.
266
However, the type of disinfectant used was not assessed, and the efficacy of specific
267
disinfectants for use in lambing pens should be considered for future research. The odds of sheep
268
seropositivity was not significantly different among sheep from farms where none of the lambing
269
pen hygiene practices were in place compared to those where all hygiene measures were taken;
270
however, the majority of the former farms practiced outdoor lambing, which could decrease the
271
odds of sheep becoming exposed.
272 273
Wildlife contact was hypothesized to be a possible risk of exposure to C. burnetii for sheep, especially considering the high prevalence of C. burnetii found in genital swabs of wild
12
274
rodents in a national park in Ontario (Thompson et al., 2012). No association between observed
275
wildlife in the barn and sheep seropositivity was found in the present study. However, simply
276
asking farm managers about wildlife contact may under-estimate its frequency, as it requires
277
some evidence of the presence of wildlife, either from seeing the wild animal itself or seeing
278
droppings or nests, which may not always occur. Also, the presence of wildlife inside the barn
279
may not be required for transmission to sheep to occur, due to the capacity for aerosolized C.
280
burnetii to travel in the wind (van der Hoek et al., 2011b). Therefore, the degree of contact
281
necessary for transmission to occur from wildlife to livestock remains unclear. Future research
282
can clarify if wildlife pose a risk of C. burnetii exposure to sheep by identifying if wildlife
283
species on or near sheep farms have C. burnetii infections, and if so, determining whether the
284
same C. burnetii strains are circulating in wildlife and sheep.
285
The farm-level seroprevalence obtained in our study (48.6%), is higher than a previous
286
serosurvey in 1988 in Ontario (21.4%, 22/103) (Lang et al., 1991). The apparent increase in
287
seroprevalence in the province over time may signify further spread of C. burnetii, but the
288
methodological heterogeneity between these studies hinders direct comparisons. The non-
289
commercial ELISA used by Lang et al. (1991) is no longer in use and the degree of formal
290
agreement with the IDEXX ELISA (used here) is unknown. The fluctuations seen in the size of
291
the Canadian sheep population over the past decade (Statistics Canada, 2012) may also affect
292
flock-level and within-flock seroprevalence due to associated changes in sheep loaning,
293
purchasing behaviours and flock sizes.
294
While long-term persistence of serum antibodies after natural infection has been
295
described in sheep, between 21% (7/34) to 63% (10/16) of sheep in small epidemic scenarios
296
have shed C. burnetii but remain seronegative (Berri et al., 2005a, 2002, 2001). The expected
13
297
proportion of seronegative shedders in non-epidemic scenarios remains unknown. Therefore, it is
298
possible that both the farm- and sheep-level seroprevalence reported in this study under-estimate
299
the true prevalence of exposure if seronegative shedders were captured in this study sample. If
300
sheep were sampled after a recent infection, but before an immune response had been generated,
301
it is also possible that they could be falsely classified as non-exposed to C. burnetii (Berri et al.,
302
2002). Most new infections typically occur during the lambing season (Astobiza et al., 2010),
303
and sero-conversion in sheep typically occurs 3 to 4 weeks after infection (Brooks, 1986). False
304
negatives caused by sampling in the latent period are therefore likely minimized in this study
305
sample, as farm managers had a strong preference to schedule farm visits during non-lambing
306
periods when they had more time available. In contrast, evidence suggests waning immunity may
307
limit the number of exposed sheep identified via serology during the non-lambing periods, as
308
sheep sampled during late pregnancy and in the periparturient period have been shown to have a
309
higher risk of seropositivity, compared to sheep sampled during early or non-pregnancy (van den
310
Brom, et al., 2012).
311
The findings from this work can also be used to direct future research and reinforce the
312
importance of hygiene and biosecurity measures in preventing or mitigating exposure of sheep to
313
C. burnetii. Moving forward, a better understanding of how C. burnetii spreads within and
314
between sheep flocks, as well as from other host species, would help improve the ability to
315
implement effective management interventions.
316
Acknowledgements
317
The authors would like to thank: the Ontario Ministry of Agriculture, Food and Rural
318
Affairs - University of Guelph Agreement through the Animal Health Strategic Investment fund
319
managed by the Animal Health Laboratory of the University of Guelph; Ontario Ministry of
14
320
Health and Long-Term Care; National Sciences and Engineering Research Council of Canada;
321
Public Health Ontario; and the Ontario Sheep Marketing Agency. The authors would also like to
322
acknowledge David Pearl and William Sears for statistical advice, and the cooperation of all
323
sheep producers that participated in the study.
324
15
325
References
326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366
Angelakis, E., and D. Raoult, 2010: Q fever. Vet. Microbiol. 140, 297–309. Astobiza, I., J. F. Barandika, F. Ruiz-Fons, A. Hurtado, I. Povedano, R. a Juste, and a L. GarcíaPérez, 2010: Coxiella burnetii shedding and environmental contamination at lambing in two highly naturally-infected dairy sheep flocks after vaccination. Res. Vet. Sci. 91. Elsevier Ltd, 58–63. Berri, M., D. Crochet, S. Santiago, and A. Rodolakis, 2005: Spread of Coxiella burnetii infection in a flock of sheep after an episode of Q fever. Vet. Rec. 157, 737–740. Berri, M., K. Laroucau, and A. Rodolakis, 2000: The detection of Coxiella burnetii from ovine genital swabs, milk and fecal samples by the use of a single touchdown polymerase chain reaction. Vet. Microbiol. 72, 285–93. Berri, M., A. Souriau, M. Crosby, D. Crochet, A. Rodolakis, and P. Lechopier, 2001: Relationships between the shedding of Coxiella burnetii, clinical signs and serological responses of 34 sheep. Vet. Rec. 148, 502–5. Berri, M., A. Souriau, M. Crosby, and A. Rodolakis, 2002: Shedding of Coxiella burnetii in ewes in two pregnancies following an episode of Coxiella abortion in a sheep flock. Vet. Microbiol. 85, 55–60. Brooks, D., 1986: Q fever vaccination of sheep: challenge of immunity in ewes. Am. J. Vet. Res. 47, 1235–1238. Capuano, F., M. C. Landolfi, and D. M. Monetti, 2001: Influence of three types of farm management on the seroprevalence of Q fever as assessed by an indirect immunofluorescence assay. Vet. Rec. 149, 669–672. De Bruin, A., R. Q. J. van der Plaats, L. de Heer, R. Paauwe, B. Schimmer, P. Vellema, B. J. Van Rotterdam, and Y. T. H. P. van Duynhoven, 2012: Detection of Coxiella burnetii DNA on small-ruminant farms during a Q fever outbreak in the Netherlands. Appl. Environ. Microbiol. 78, 1652–1657. Dohoo, I., W. Martin, and H. Stryhn, 2003: Veterinary Epidemiologic Research. Prince Edward Island, Canada: AVC Incorporated. EFSA, 2010: Scientific opinion on Q fever. EFSA J. 8, 1–114. Enright, J. B., C. E. E. Franti, D. E. E. Behymer, W. M. M. Longhurst, V. J. J. Dutson, and M. E. E. Wright, 1971: Coxiella burnetii in a wildlife-livestock environment: distribution of Q fever in wild mammals. Am. J. Epidemiol. 94, 79–90. Fishbein, D. B., and D. Raoult, 1992: A cluster of Coxiella burnetii infections associated with exposure to vaccinated goats and their unpasteurized dairy products. Am. J. Trop. Med. Hyg. 47, 35–40. Goldstein, H., W. Browne, and J. Rasbash, 2002: Partitioning variation in multilevel models. Underst. Stat. 1, 223–231. Hogerwerf, L., A. Courcoul, D. Klinkenberg, F. Beaudeau, E. Vergu, and M. Nielen, 2013: Dairy goat demography and Q fever infection dynamics. Vet. Res. 44, 28. Horigan, M., M. Bell, T. Pollard, 2011: Q fever diagnosis in domestic ruminants: comparison between compliment fixation and commercial enzyme-linked immunesorbent assays. J. Vet. Diagn. Invest. 23 (5), 924-931. 16
367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409
Lambton, S. L., R. P. Smith, K. Gillard, M. Horigan, C. Farren and G. C. Pritchard, D. 2015. Serological survey using ELISA to determine the prevalence of Coxiella burnetii infection (Q fever) in sheep and goats in Great Britain. Epidemiology and Infection, available on CJO2015. doi:10.1017/S0950268815000874. Lang, G. H., D. Waltner-Toews, and P. Menzies, 1991: The seroprevalence of coxiellosis (Q fever) in Ontario sheep flocks. Can. J. Vet. Res. 55, 139–42. Lang, G. H. 1988: Serosurvey of Coxiella burnetii infection in dairy goat herds in Ontario. a comparison of two methods of enzyme-linked immunosorbent assay. Can. J. Vet. Res. 52, 37– 41. Lyytikäinen, O., T. Ziese, B. Schwartländer, P. Matzdorff, C. Kuhnhen, C. Jäger, and L. Petersen, 1998: An outbreak of sheep-associated Q fever in a rural community in Germany. Eur. J. Epidemiol. 14, 193–9. Marrie, T. J., H. Durant, J. C. Williams, E. Mintz, and D. M. Waag, 1988: Exposure to parturient cats : a risk factor for acquisition of Q fever in Maritime Canada. Ann. N. Y. Acad. Sci. 158, 101–108. Masala, G., R. Porcu, G. Sanna, G. Chessa, G. Cillara, V. Chisu, and S. Tola, 2004: Occurrence, distribution, and role in abortion of Coxiella burnetii in sheep and goats in Sardinia, Italy. Vet. Microbiol. 99, 301–5. Mason, C. H., and W. D. Perreault, 2013: Collinearity, power and interpretation of multiple regression analysis. J. Mark. Res. 28, 268–280. Maurin, M., and D. Raoult, 1999: Q fever. Clin. Microbiol. Rev. 12, 518–553. McEwen, B., D. Slavic, D. Ojkic, J. Delay, H. Cai, M. Stalker, M. Hazlett, and M. Brash, 2012: Selected zoonotic pathogens and diseases identified at the Animal Health Laboratory, 2011. Anim. Heal. Lab. Newsl., March. Palmer, N. C., M. Kierstead, D. W. Key, J. C. Williams, M. G. Peacock, H. Vellend, and T. H. E. Editor, 1983: Placentitis and abortion in goats and sheep in Ontario caused by Coxiella burnetii. Can. Vet. J. 24, 60–61. Rodolakis, A., 2006: Q fever, state of art: epidemiology, diagnosis and prophylaxis. Small Rumin. Res. 62, 121–124. Schimmer, B., S. Luttikholt, J. L. A. Hautvast, E. A. M. Graat, P. Vellema, and Y. T. H. P. van Duynhoven, 2011: Seroprevalence and risk factors of Q fever in goats on commercial dairy goat farms in the Netherlands, 2009-2010. BMC Vet. Res. 7. BioMed Central Ltd, 81. Sidi-Boumedine, K., E. Rousset, K. Henning, M. Ziller, K. Niemczuck, H. I. J. Roest, and R. Thiéry, 2010: Development of harmonised schemes for the monitoring and reporting of Q fever in animals in the European union. Vet. Res. Parma, Italy. Simor, A. E., J. L. Brunton, I. E. Salit, H. Vellend, L. Ford-Jones, and L. P. Spence, 1984: Q fever: hazard from sheep used in research. Can. Med. Assoc. J. 130, 1013–1016. Simor, A. E., 1987: Q fever - human disease in Ontario. Can. Vet. J. 28, 264–266. Statistics Canada, 2012: Sheep and lambs on farm - Canada. Sheep Statistics. http://www5.statcan.gc.ca/bsolc/olc-cel/olc-cel?catno=23-011-x&lang=eng. Thompson, M., N. Mykytczuk, K. Gooderham, and A. Schulte-Hostedde, 2012: Prevalence of the bacterium Coxiella burnetii in wild rodents from a Canadian natural environment park. Zoonoses Public Health 59, 553–60. 17
410 411 412 413 414 415 416 417 418 419 420 421 422 423 424
Tissot-Dupont, H., S. Torres, M. Nezri, and D. Raoult, 1999: Hyperendemic focus of Q fever related to sheep and wind. Am. J. Epidemiol. 150, 67–74. van den Brom, R., L. Moll, G. van Schaik, P. Vellema, 2012. Demography of Q fever seroprevalence in sheep and goats in The Netherlands in 2008. Prev. Vet. Med. 109 (April 12), 76-82. http://dx.doi.org/10.1016/j.prevetmed.2012.09.002 van der Hoek, Wim, J. C. E. Meekelenkamp, F. Dijkstra, D. W. Notermans, B. Bom, P. Vellema, A. Rietveld, Y. van Duynhoven, and A. C. A. P. Leenders, 2011a: Proximity to goat farms and seroprevalence among pregnant women. Emerg. Infect. Dis. 17, 2360–2363. van der Hoek, Wim, J. Hunink, P. Vellema, and P. Droogers, 2011b: Q fever in the Netherlands : the role of local environmental conditions. Int. J. Environ. Health Res. 21, 441–451. Welsh, H. H., E. H. Lennette, F. R. Abinanti, and J. F. Winns, 1945: Air-borne transmission of Q fever: the role of parturition in the generation of infective aerosols. Ann. N. Y. Acad. Sci. 70, 528–540. Woldehiwet, Z., 2004: Q fever (coxiellosis): epidemiology and pathogenesis. Res. Vet. Sci. 77, 93–100.
18
425
426 Figure 1 Percentage of C. burnetii seropositive sheep per farm on 22 dairy sheep farms in Ontario, Canada (Aug 2010-Jan 2012)* * Farms have been sorted from lowest to highest within-herd seroprevalence
427
428
Figure 2 Percentage of C. burnetii seropositive sheep per farm on 50 randomly selected meat sheep farms in Ontario, Canada (Aug 2010-Jan 2012)* * Farms have been sorted from lowest to highest within-herd seroprevalence
429 430
19
Table 1 Farm management variables associated (p<0.2) with sheep seropositivity for C. burnetii, as observed through univariable mixed-effects logistic regression of data collected from 72 sheep farms in Ontario, Canada (Aug 2010-Jan 2012) Risk Factors OR OR 95% CI P-value Sector Meat Ref Dairy 4.87 1.07-22.24 0.041 Female flock size (Log10 scale) 1.89 1.19-2.99 0.007 Lambing pen cleaning practices (LRT χ2 = 387.23, P-value = 0.0491) Add bedding, remove birthing materials and Ref disinfecting pen Add bedding and remove birthing materials 6.53 1.15-37.03 0.034 Add bedding only 6.27 0.79-49.89 0.083 Do nothing 0.18 0.009-3.48 0.257 Loaning sheep that return to farm (yes/no) 8.25 1.25-54.27 0.028 Closed flock (yes/no) 0.15 0.009-2.34 0.175 Purchasing sheep from sales barns (yes/no) 7.89 0.49-126.90 0.145 Purchasing replacement males (yes/no) 6.82 0.69-4.22 0.101 Poison baits for rodent control (yes/no) 4.79 0.81-28.59 0.084 Barn cats on farm (yes/no) 6.28 0.51-77.99 0.153 Lambing ewes in separate area from flock (LRT χ2 = 362.96, P-value = 0.1310) Never Ref Not Applicable (do not lamb indoors) 0.18 -4.85-4.075 0.281 Sometimes 4.93 0.86-28.20 0.073 Always 2.87 0.34-24.44 0.336 Days replacement lambs spend with dam before 0.50 0.23-1.08 0.076 weaning (Log10 scale) Other sheep or goat farms within 5km (LRT χ2 = 411.79, P-value = 0.1774) No Ref Yes 5.59 0.86-36.13 0.071 Unknown 8.12 0.23-289.00 0.251 OR = odds ratio, Ref= Referent category LRT = Likelihood ratio test for categorized variable
431 432
20
Table 2 Farm management variables associated (p<0.05) with C. burnetii seropositivity in sheep, as observed through multivariable mixed-effects logistic regression of data collected from 72 sheep farms in Ontario, Canada (Aug 2010-Jan 2012) Risk Factors OR OR 95% CI P-value Lambing pen cleaning practices (LRT χ2 = 11.71, p = 0.0084) a Add bedding, remove birthing materials and Ref disinfect Add bedding and remove birthing materials 8.96 2.17-36.90 0.002 Add bedding only 5.94 1.10-32.12 0.038 Do nothing 0.97 0.04-23.29 0.986 Lambing ewes housed in separate airspace from flock (LRT χ2 = 11.77, p = 0.0082) a Always 2.08 0.43-10.10 0.362 Sometimes 11.26 2.91-43.56 <0.001 Never Ref Not Applicable (do not lamb indoors) 0.44 0.02-11.01 0.614 Loaning sheep that return to farm (yes/no) 8.14 1.81-36.61 0.006 Female flock size (Log10 scale) 3.24 1.71-6.15 <0.001 Random Intercept Parameter SD 95% CI P-value Farm b 1.59 1.11-2.26 <0.0001 OR = odds ratio, Ref= Referent category, SD= Standard deviation a LRT = Likelihood ratio test for categorized variable b Partitioning the variance, 43.3% occurred at the farm-level and 56.7% occurred at the sheeplevel
433 434
21
435 436
22