Coxiella burnetii (Q Fever) Seropositivity and

VECTOR-BORNE AND ZOONOTIC DISEASES Volume 16, Number 10, 2016 ª Mary Ann Liebert, Inc. DOI: 10.1089/vbz.2015.1909

Coxiella burnetii (Q Fever) Seropositivity and Associated Risk Factors in Sheep and Goat Farm Workers in Ontario, Canada Shannon Meadows,1 Andria Jones-Bitton,1 Scott A. McEwen,1 Jocelyn Jansen,2 Samir N. Patel,3,4 Catherine Filejski,5 and Paula Menzies1

Abstract

Coxiella burnetii is a zoonotic bacterium that causes Q fever, a potentially severe disease of humans. The objectives of this study were to determine the seroprevalence and risk factors for C. burnetii exposure in sheep and goat farm workers in Ontario, Canada. Between August 2010 and March 2012, 172 farm workers from 78 sheep and goat farms were surveyed regarding demographics, lifestyle, farm practices, and medical history. Sera from these people were collected and analyzed for Q fever titers using the immunofluorescence assay. A mixed multivariable logistic regression model was constructed to identify risk factors for seropositivity. Individual-level and farm-level seroprevalence for C. burnetii were 64.5% (111/172, 95% CI = 57.2–71.4) and 74.4% (58/78, 95% CI = 63.2–83.6), respectively. Farm worker seropositivity was positively associated with an increasing proportion of seropositivity of sheep/goats on farm (odds ratio [OR] = 1.04; 95% CI 1.02–1.07). A higher odds of seropositivity was also observed for people working on dairy goat farms compared to the odds on dairy sheep (OR = 0.04; 95% CI 0.003–0.53) or meat goat (OR = 0.09; 95% CI 0.01–0.67) farms. Coxiella burnetii seropositivity was common in workers on sheep and goat farms in Ontario. Given the significant risk of morbidity associated with this infection, early recognition and treatment of Q fever are important. The risk factors identified provide insight into disease transmission between animals and people, which is particularly important for farmers, researchers, medical doctors, veterinarians, and public health professionals. Physicians practicing in rural areas should consider Q fever infection when patients present with atypical pneumonia and suggestive risk factors. Keywords:

Coxiella burnetii, farm worker, Ontario, Q fever, risk factor, seroprevalence

Introduction

C

oxiella burnetii is an intracellular zoonotic bacterium (Maurin and Raoult 1999) that causes Q fever in humans, a febrile and often debilitating disease. C. burnetii is a common infection of many animal species (Marrie 1995), and sheep, goats and cattle are the species most frequently associated with human illness (Public Health Ontario 2012, van den Brom et al. 2015). Infected animals shed large numbers of organisms in their placenta, birth fluids, and milk (Agerholm 2013). Coxiella burnetii can also be excreted through vaginal mucous and feces postparturition (Roest et al. 2012). Parturition represents the time of highest transmission risk to humans

(Marrie 1990). The bacteria can remain infective for months in aerosols and contaminated dust (Woldehiwet 2004). Inhalation of contaminated aerosols is the major mechanism whereby C. burnetii is transmitted to humans (Williams and Thompson 1991). Consequently, people can become infected from direct contact with shedding animals (Fournier et al. 1998), contaminated materials such as animal bedding or manure, or through sharing an air space contaminated with C. burnetii (Rodolakis 2009). At higher risk for the disease are farm workers, veterinarians, and abattoir workers, particularly those working with sheep, goats, and cattle (Marrie and Fraser 1985). While *60% of Q fever cases are asymptomatic, 38% experience mild to moderate symptoms without the need for

1

Department of Population Medicine, University of Guelph, Ontario, Canada. Veterinary Science and Policy, Ontario Ministry of Agriculture, Food and Rural Affairs, Elora, Canada. 3 Public Health Ontario, Toronto, Canada. 4 Department of Laboratory Medicine and Pathobiology, University of Toronto, Toronto, Canada. 5 Ontario Ministry of Health and Long-Term Care, Toronto, Canada. 2

643

644

MEADOWS ET AL.

hospitalization, 1.8% are hospitalized with acute Q fever, and 0.2% develop chronic Q fever (Maurin and Raoult 1999). Symptomatic acute Q fever manifests primarily as a febrile illness associated with severe headaches, atypical pneumonia, and granulomatous hepatitis; endocarditis and chronic fatigue syndrome are common presentations of chronic Q fever (Maurin and Raoult 1999, Limonard et al. 2010). In Ontario, cases of Q fever are reportable to the local Medical Officer of Health under the Health Protection and Promotion Act (Ontario Ministry of Health and Long-Term Care 2003). From 2006 to 2011, 47 confirmed cases of Q fever were reported in Ontario (Ontario Public Health 2012); however, the nonspecific signs of Q fever likely contribute to underreporting (Marrie and de Carolis 2002, Public Health Ontario 2012, van der Hoek et al. 2012). A recent human Q fever epidemic occurred in the Netherlands in which 3522 acute Q fever cases were notified between 2007 and 2009 and corresponded to more than 44,000 infections in the same period (van der Hoek et al. 2012). This outbreak was attributed to infected dairy small ruminants (Roest et al. 2011) and demonstrates the potential public health importance of this disease. The immunofluorescence assay (IFA) is the reference serological test used in human diagnostics (Tissot Dupont Thirion and Raoult 1994, Maurin and Raoult 1999, Sidi-Boumedine et al. 2010) and is reported as antibodies to phase I and phase II C. burnetii antigens (Focus Diagnostics 2011). The ratio of phase I to phase II antibodies is used to determine if the exposure is acute (phase II > phase I) or chronic (phase I > phase II) (Ontario Ministry of Health and Long-Term Care 2003). C. burnetii exposure was recently determined to be common among sheep and goats in Ontario, as 48.6% (95% CI = 37.2– 60.1) of sheep farms and 63.2% (95% CI = 51.9–73.4) of goat farms had at least one seropositive animal (Meadows et al. 2015a, b). Studies of surveillance of C. burnetii infection in people in Canada are sparse and none were recently published. A survey of patients with community-acquired pneumonia in Ontario found a seroprevalence of 10.8% (21/193) (Marrie and de Carolis 2002), and a separate study from the same period found that 28.4% (23/81) of sheep farmers in the lower Saint-Lawrence River region of Quebec were seropositive to C. burnetii (Dolce´ et al. 2003). Given that exposure appears common in sheep producers and C. burnetii infection may be a significant cause of community-acquired pneumonia, it is important to investigate seropositivity and factors influencing seropositivity in people in contact with high-risk animal populations. In addition, since C. burnetii readily aerosolizes and the risk of infection is highest for those living within a 5 km radius from the source (Schimmer et al. 2010), the risk of exposure extends to those living nearby. Urbanization of rural areas contributes to increased risk for C. burnetii exposure among these populations (Hellenbrand et al. 2001). This study was performed to assess the seroprevalence of C. burnetii among farm workers who care for sheep and goats in Ontario, Canada, and investigate farm management practices, demographic characteristics, and lifestyle measures for association with exposure to C. burnetii in this population.

in Ontario, Canada (Meadows et al. 2015a, b). A total of 148 farms were enrolled for animal sampling (50 meat sheep farms, 22 dairy sheep farms, 34 meat goat farms, and 42 dairy goat farms) during the summer of 2010 until the autumn of 2011. Workers 14 years and older and who cared directly for the sheep and goats (e.g., owners, family members, employees) on these farms were asked to participate. Hence, a cluster sampling procedure was used for farm workers. Sampling of humans and questionnaire administration occurred between August 2010 and March 2012. The University of Guelph Research Ethics Board (Certification of Ethical Acceptability of Research Involving Human Participants–10JN005) and the University of Guelph Animal Care Committee (Animal Use Protocol–10R056) approved the study.

Materials and Methods

Questionnaire and serological data were entered into EpiData v2.2 for file management (EpiData Association, Odense, Denmark) and were manually checked against the original questionnaires and laboratory reports for errors. A mixed logistic regression model was constructed in Stata

Farm and farm worker selection

This research was part of a larger cross-sectional study to determine the C. burnetii seroprevalence of sheep and goats

Putative risk factor data collection

Questionnaires were administered on each participating farm by personal interview, or were given to participants on the farm and were subsequently completed and mailed/ emailed back to the researcher. A reasonably short recall period, usually assessing either current farm management practices or those conducted in the last year, was used to help minimize misclassification due to recall bias. Questionnaires used a mix of closed- and open-ended questions to gather information on the following: personal demographic characteristics; medical history, including illnesses that were possibly Q fever sequelae; prior Q fever diagnostic testing; practices for assisting animal births; individual biosecurity, hygiene, animal management practices; contact with animals; and lifestyle habits, including smoking, alcohol, and raw milk product consumption. Data obtained from sheep and goat serological testing on these farms (Meadows et al. 2015a, b), including the percentage of animals that tested seropositive at the farm level, were also used and tested for association with farm worker seropositivity. Blood collection and serological analysis

All blood samples were collected by a licensed healthcare professional. Participants either visited local participating human phlebotomy laboratories (CML Healthcare or Gamma Dynacare) or if necessary, a mobile phlebotomist attended the farm worker’s residence to collect blood samples. Blood samples were collected by venipuncture into 10 mL red top serum BD vacutainer tubes (Becton, Dickson and Company, Franklin Lakes, NJ). Vacutainer tube samples were centrifuged for 10 min at 1000g and 2 mL aliquots of the separated serum were then immediately pipetted into serum microtubes (Fisher Scientific, Ottawa, CA). Serological analysis was performed at the Public Health Ontario Laboratory Services using Focus Diagnostics IFA (Cypress, CA), according to manufacturer’s instructions. Participants were considered seropositive if either the phase I or phase II IgG titer was ‡ 1:16 (Focus Diagnostics 2011). Data management and statistical analysis

COXIELLA BURNETII SEROPOSITIVITY IN ONTARIO FARM WORKERS

Intercooled Version 10.1 (StataCorp, 2007) to assess putative risk factor associations with the outcome of farm worker seropositivity, while controlling for clustering by using farm as a random intercept. All covariates associated with farm worker seropositivity in univariable analyses (Wald and Likelihood Ratio tests for dichotomous and categorical variables, respectively) at p < 0.20 were initially included in the model. The model was built using a manual backward stepwise procedure and a final significance level of a < 0.05 (Dohoo et al. 2003) (Supplementary Material; Supplementary Data are available online at www.liebertpub.com/vbz). Results Study population and seroprevalence

Of the 148 farms on which sheep or goats were tested, 52.7% (78/148) had at least one person participate in human sampling (mean 2.2 people). The farm-level participation in human sampling varied by sector; 73.8% (31/42) of dairy goat farms, 55.9% (19/34) of meat goat farms, 42.0% (21/50) of meat sheep farms, and 22.7% (5/22) of dairy sheep farms participated in the human component of the study. Two farms had both sheep and goats; worker exposure on these farms was calculated using the percentage of seropositive sheep and goats combined. Complete questionnaires and serological results were available from 167 participants representing 78 farms. Participation was similar among genders; 49.4% (85/172) female and 50.6% (87/172) male. On average, participants had been working with sheep or goats for 10.8 years (–9.3) and any livestock for 25.1 years (–16.6), and spent 4.1 (–3.2) hours per day in contact with sheep or goats. The majority (91.0%, 152/167) of participants lived on the farm. Overall, one or more farm workers tested positive on 74.4% (58/78, 95% CI = 63.2–83.6) of farms (farm-level prevalence); 96.8% (30/31, 95% CI = 83.3–99.9) of dairy goat farms, 60.0% (3/5, 95% CI = 14.7–94.7) of dairy sheep farms, 71.4% (15/21,

645

95% CI = 47.8–88.7) of meat sheep farms, 47.4% (9/19, 95% CI = 24.5–71.4) of meat goat farms, and 50.0% (1/2, 95% CI = 1.3–98.7) of farms with animals from multiple sectors. Farm-level seroprevalence was significantly higher on dairy goat than meat goat farms ( p = 0.002) and dairy ( p = 0.028) or meat sheep farms ( p = 0.027), and was borderline significantly different from farms with both sheep and goats ( p = 0.051). Table 1 indicates the distribution of phase I and phase II IgG titres to C. burnetii antigens among the study populations, as determined by the IFA. Overall, 64.5% (111/172, 95% CI = 57.2–71.4) of sheep and goat farm workers (individual-level prevalence) were seropositive for C. burnetii. This was significantly higher on dairy goat farms (68/ 80, 85.0%, 95% CI = 75.9–91.6) than dairy sheep farms (4/ 13, 30.8%, 95% CI = 10.63–58.7, p = 0.005), meat goat farms (19/42, 45.2%, 95% CI = 30.8–60.4, p < 0.001), and meat sheep farms (19/34, 55.9%, 95% CI = 39.1–71.8, p = 0.007), but not for those on farms with both sheep and goats (1/3, 33.3%, 95% CI = 0.8–90.6, p = 0.112). Prior diagnostic testing and history of illness possibly related to Q fever

Twenty participants reported that they suspected they had been ill with Q fever at some point in their lives, 17 of whom were seropositive. Some sought medical attention for their symptoms (n = 9) and requested a test for Q fever (n = 7), which was performed in six cases, with five reported as positive. Of these positive cases, three were contacted by the local public health unit for routine follow-up of reportable diseases. A total of 46 of 167 people (27.5%) indicated that they had a physician diagnosis of one or more conditions in their lifetime, specifically the following: pneumonia (34), asthma (22), heart disease (9), cancer (7), low immune function (5), and emphysema (1). Of the (56.6%, 47/83) female participants of childbearing age (14–49 years), 14 had either been pregnant within the previous 2 years, or were currently pregnant. Almost all of these women (92.9%, 13/14) indicated that

Table 1. Number of Individual Sheep and Goat Farm Workers (Percentage of Total Samples) with Specific Immunoglobulin G Serum Titers to Phase I and Phase II Coxiella burnetii Antigens (n = 172) as Determined by the Immunofluorescence Assay (Focus Diagnostics) (August 2010–March 2012; Ontario, Canada) Phase I

Phase II

1:128

‡ 1:256

1 (0.6)

0 (0)

0 (0)

72 (41.9)

5 (2.9)

1 (0.6)

0 (0)

0 (0)

15 (8.7)

10 (5.8)

10 (5.8)

6 (3.5)

2 (1.2)

1 (0.6)

30 (17.4)

0 (0)

1 (0.6)

3 (1.7)

11 (6.4)

5 (2.9)

1 (0.6)

21 (12.2)

1:128

0 (0)

0 (0)

0 (0)

1 (0.6)

12 (7.0)

5 (2.9)

18 (10.5)

‡ 1:256

0 (0)

0 (0)

0 (0)

4 (2.3)

0 (0)

12 (7.0)

16 (9.3)

19 (11.0)

172 (100.0)

NR

1:16

1:32

1:64

NR

61 (35.5)

7 (4.0)

3 (1.7)

1:16

5 (2.9)

4 (2.3)

1:32

1 (0.6)

1:64

Total

67 (39.0)

22 (12.8)

21 (12.2)

24 (14.0)

19 (11.0)

Total

With the exception of nonreactive titers (NR), titers above the dotted line are suggestive of a chronic infection and those below suggestive of an acute infection. 35.5% (61/172) Unexposed (phase I and phase II IgG not reactive). 14.5% (25/172) Titers suggestive of an acute exposure (phase II titer > phase I titer). 50.0% (86/172) Titers suggestive of a chronic exposure (phase I titer ‡ phase II titer). NR, not reactive.

646

they had been in the barn during lambing or kidding when pregnant and nine also assisted with lambing or kidding while pregnant (64.3%, 9/14). Three of the 14 women (21.4%) reported an adverse pregnancy outcome (miscarriage or stillbirth), 8 of 14 pregnant women were seropositive (57.1%), and 1 of 3 women with adverse pregnancy outcomes were seropositive (33.3%). Risk factor analysis

The variables unconditionally associated ( p < 0.20) with farm worker seropositivity are presented in Table 2. The final mixed-effect multivariable model, which excluded variables possibly related to Q fever sequelae, is shown in Table 3. Dairy goat farm workers had 11 times higher odds of being seropositive, compared to meat goat farm workers (95% CI 1.5–83.3), and 25 times higher odds of being seropositive compared to dairy sheep farm workers (95% CI 1.88–333.3), but were not significantly different from meat sheep ( p = 0.18) or multiple species ( p = 0.54). The odds of farm worker seropositivity also increased with the proportion of seropositive animals on farm. To illustrate using Stata’s lincom command, an increase of 20% in the percentage of sampled animals that tested positive on the farm (e.g., from 30% to 50%) increased the odds of farm worker seropositivity by 2.4 times (95% CI 1.3–4.2). A history of having smoked tobacco (including current smokers and smokers who had quit) was marginally associated with seropositivity ( p = 0.072). This variable was also determined to confound the relationship between industry sector and seropositivity, and was therefore retained in the model. It should be noted that the likelihood ratio test statistic, comparing the multivariable model with and without the smoking history variable, was significant (LRT p = 0.038); however, the significance of the Wald’s test statistic was decided apriori as the test to determine inclusion of dichotomous variables in the multivariable model. As such, ever having smoked tobacco remained in the model only as a confounder. Interpretation

This is the first comprehensive study to investigate exposure to C. burnetii among people working on sheep and goat farms in Ontario, Canada, and the latest seroprevalence study in over a decade to examine C. burnetii exposure in Canadian populations (Marrie and de Carolis 2002, Dolce´ et al. 2003). Coxiella burnetii exposure was common; 64% of individuals overall were seropositive, and 76% of farms had one or more seropositive farm workers. We used the IFA cutpoint recommended by the test manufacturer ( ‡ 1:16) (Focus Diagnostics 2011); using a low titer cutoff minimizes false-negative results caused by waning titers (Teunis et al. 2013) and as the manufacturer reports a high specificity (100%), false-positive results are also likely low (Focus Diagnostics 2011). In the multivariable model, there was a strong positive association between the percentage of seropositive animals and farm worker seropositivity. Coxiella burnetii exposure among sheep and goats being cared for by these farm workers was common (Meadows et al. 2015a, b); therefore, it is likely that an extended opportunity for transmission to humans existed on these farms. Implementing measures to reduce or eliminate shedding in animals could reduce the likelihood of

MEADOWS ET AL.

Table 2. Covariates Associated (p < 0.2) with Farm Worker Seropositivity for C. burnetii, as Observed Through Univariable Mixed-Effects Logistic Regression of Data Collected from Sheep and Goat Farm Workers in Ontario, Canada (August 2010–March 2012) Risk factors

OR

95% CI

p

Industry sector (LRT v2 = 15.13, p = 0.0044) Dairy goats Ref — — Dairy sheep 0.03 0.003–0.34 0.005 Meat goats 0.04 0.006–0.24