A cross‐sectional study of Q fever in Camels: Risk factors for infection, the role of small ruminants and public health implications for desert‐dwelling pastoral communities

Abstract Q fever represents an important ‘neglected zoonosis’, with high prevalences recorded across the Middle East region. Among rural desert‐dwelling communities in the region, camel milk is largely consumed raw, due to perceptions of dromedaries as a uniquely clean livestock species mentioned in the Qur'an and Islamic hadith, while milk from other livestock species is usually boiled. As a result, camels present a unique public health threat among such communities from milk‐borne pathogens, including Coxiella burnetii. In view of this, a cross‐sectional study was conducted among dromedary herds in southern Jordan between September 2017 and October 2018, including 404 camels from 121 randomly selected herds. In addition, 510 household members associated with these herds were interviewed regarding potential high‐risk practices for zoonotic transmission. Weight adjusted camel population seroprevalence for C. burnetii was 49.6% (95% CI: 44.7–54.5), with evidence of maternally derived immunity in calves ≤6 months old. Adjusted herd‐level prevalence was 76.0% (95% CI 72.7–80.2). It was estimated 30.4% (144/477) of individuals consumed raw milk from infected herds monthly or more. Following multivariable logistic regression analysis, seropositive status in camels was found to be associated with increasing age, high herd tick burdens, keeping the herd together throughout the year including when calving, and owning larger (>50) sheep and goat flocks, with goats presenting a higher risk than sheep. Racing camel status was found to be protective. Socioculturally appropriate interventions aimed at raising awareness of potential risks associated with drinking raw camel milk, alongside appropriate livestock management interventions, should be considered.


| INTRODUC TI ON
Q fever represents an important 'neglected zoonosis', which despite the presence of licensed vaccines, remains largely unrecognized and uncontrolled, particularly among lower and middle-income countries (LMIC) where seroprevalences are often high (Vanderburg et al., 2014). The causative agent, C. burnetii, is an obligate gramnegative intracellular bacterium of high tenacity, favouring hot dry conditions, with high infectivity (Maurin & Raoult, 1999). Human infections range from being asymptomatic to causing an acute nonspecific febrile illness, often with hepatitis and atypical pneumonia (van der Hoek et al., 2011). While most clinical infections are selflimiting, some individuals go on to develop chronic disease, which may include endocarditis and fatigue (Ayres et al., 1998;Brouqui et al., 1993). These non-specific and diverse signs and symptoms, compounded by a lack of awareness among many healthcare workers and lack of routine laboratory testing in many LMIC settings, mean that individuals presenting with clinical C. burnetii infection are frequently misdiagnosed (Buijs et al., 2021;Honarmand, 2012).
In ruminants, Q fever is an important production disease causing reproductive losses through abortions, stillbirths and infertility, alongside milk drop and chronic mastitis (Plummer, 2022). Bacteria are shed in high numbers through infected birth products, as well as in milk, faeces and urine (Canevari et al., 2018). Livestock and human infections occur largely via inhalation of contaminated dust particles, including infected tick faeces, as well as through contact with infected birthing products and from infected tick bites (Angelakis & Raoult, 2010). However, zoonotic transmission via consumption of infected raw dairy products is also known to occur (Signs et al., 2012). While the zoonotic impact of C. burnetii infection in small ruminant and cattle populations has been widely reported, the potential role of camels in zoonotic transmission of Q fever has until recently, remained largely unexamined, particularly in the Middle East region, where favourable conditions for the pathogen exist (Browne et al., 2017;Devaux et al., 2020;Hussien et al., 2017;Larson et al., 2019). The widespread consumption of raw camel milk across the Arab world, due to the perceptions of camels as uniquely clean livestock with mention in the Qur'an and Islamic hadith, means that camels present a unique public health threat (Al-Ghāshiyah; Qilaba; Galali & Al-Dmoor, 2019). A population-level seroprevalence of 24% was recently reported in Jordan, with seroprevalences of 52% and 35% reported among hospitalised patients with fever of unknown origin, in neighbouring Saudi Arabia and Egypt respectively, suggesting Q fever presents an important public health threat within the region. (Abbass et al., 2020;Almogren et al., 2013;Obaidat et al., 2019) To improve understanding of the epidemiology and potential zoonotic risks posed by Q fever in camels, we conducted a largescale epidemiological survey among camel herds in southern Jordan, largely owned by desert-dwelling Bedouin communities. This population is considered likely to be representative of analogous Bedouin and pastoral communities in the wider region, where larger flock or herd sizes are indicative of higher socioeconomic status. The objectives of the study were to: (i) estimate the prevalence of C. burnetii in the camel population in southern Jordan (ii) identify potential transmission pathways for C. burnetii infection in camels, particularly regarding the role of small ruminants, and (iii) assess the potential public health risk associated with these herds through consumption of raw milk and other activities.

| Study design and study population
A cross-sectional study was conducted between 28th October 2017 and 11th October to 2018, in Aqaba and Ma'an governorates of southern Jordan, an area of approximately 40,000 km 2 and 8000 camels (based on MoA data) ( Figure 1). Probabilistic sampling was conducted using camel owner lists supplied by the Ministry of Agriculture (MoA) according to four local administrative areas (Aqaba east, Aqaba west, Ma'an east and Ma'an west). The MoA records livestock numbers owned per registered individual in each administrative region, including camels, with these records updated annually. While it is possible that some individuals who own camels may have been omitted from this list, all individuals contacted from these lists owned camels, or had done so at the time of the list had been compiled, with numbers owned largely reflecting MoA records.
Based on an expected median herd size of 12, an expected prevalence of 35% and a confidence level of 95%, and in order to facilitate owner compliance, no more than 12 camels were sampled per herd, and in herds of <12 all camels were sampled, subject to accessibility and owner permissions. The formula used to detect at least one positive animal was: K = [1−(1−p1/d)][N−½(d−1)]. Two standardised structured questionnaires regarding potential risk factors for C. burnetii infection, in camels and humans respectively, were administered in the local dialect on Android tablets, using the application Open Impacts • C. burnetii, the causative agent of Q fever, is endemic at high levels among camels in Southern Jordan, which combined with widespread consumption of raw camel milk, suggest Q fever present an important public health risk among desert-dwelling pastoral communities.
• Controlling Q fever in small ruminants (sheep and goats) is likely to contribute to reducing C. burnetii prevalence in associated camel populations.
• Potential control strategies include husbandry practices such as livestock tick control and separation of birthing areas as well as socioculturally appropriate interventions to raise awareness of risks from drinking raw camel milk. Use of ruminant vaccines and trials to assess them in camels should be considered.
Data Kit (ODK), among herd owners and their household members.
All camels included in the study were clinically examined by a veterinary surgeon to assess general health and the presence of ticks (yes/no), prior to collection of a blood sample, from which serum was extracted.

| Statistical analysis
We calculated seroprevalence estimates, weighted according to sample size, relative to the estimated camel population, based on MoA data for each sub-region. Regression models were built for identification of risk factors, with camels ≤6 months of age excluded from analyses due to the potential influence of maternally derived antibodies.
Univariable analyses were conducted, using mixed-effects logistic regression to adjust for herd-level random effects, with camel serological status considered a binary outcome. All potential risk factors were analysed as categorical variables, with the exception of camel age, altitude of the holding, and small ruminant flock size which were analysed as continuous variables. Season was not considered for analysis due to the non-longitudinal nature of the study and likely correlation with sample location.
Variables associated with the outcome with a p-value <0.2 were considered for inclusion in the multivariable models, with the exception of any variables missing more than 10% of their values.
Collinearities between variables were examined using the Pearson R coefficient and a threshold of 0.4, with collinear variables excluded from the same multivariable model. Multivariable models were constructed using a backwards stepwise method, with the least significant variable removed at each step while p > 0.1, unless the variable was considered an a priori factor (sex and age) or the removal of the variable demonstrated a significant effect on the other variables (a change in log odds >20%), with model building then repeated using a forwards stepwise method.
The herd-level prevalence of C. burnetii (the proportion of herds containing at least one camel with C. burnetii antibodies) was estimated, taking into account the uncertainty arising from sampling only a proportion of each herd (the proportion being different in each herd). Based on the method described by Beauvais et al. (Beauvais et al., 2016), a frequency distribution of withinherd prevalence was calculated for each herd and then multiplied by the number of camels in the herd. This was used in a Bayesian computation with herd sample results, giving a discrete probability distribution for positives within a herd. Each herd was then simulated as being positive or negative using a random sample from a binomial distribution. This was repeated 10,000 times to create an uncertainty distribution, where the 2.5th and 97.5th percentiles gave a 95% credible interval and the 50th percentile gave the most likely herd-level prevalence. F I G U R E 1 Location of 121 camel herds sampled in southern Jordan October 2017 to October 2018 (due to local grazing movements there were three herds selected from the MoA list for Ma'an west that were sampled in the neighboring region, Tafilah. Results from these herds were attributed to Ma'an west) The number of individuals living in households with Q fever positive herds was compared against questionnaire data relating to potential pathways for C. burnetii zoonotic transmission, by calling a herd positive when there was at least a 50% probability (with 95% confidence) of the herd having at least one positive animal, using Bayesian probability. This figure was used to a calculate the percentage of the sample population likely to have been exposed to potential C. burnetii transmission via high-risk practices.
All statistical analyses were performed in R (version 3.5.6) with mixed-effects models generated using the glmer function of the package lme4 (version 1.1-23).

| RE SULTS
Blood samples were collected from 404 camels in 121 herds, with an average of 3.3 camels sampled per herd (median herd size 9, IQR 4-17). The questionnaire regarding potential risk factors for infection in camels was administered to all 121 herd owners, while the questionnaire regarding potential high-risk practices for human infection was administered to 510 members of camel-owning households (which included the 121 herd owners). Camel numbers sampled were: Ma'an east 90 (29 herds), Ma'an west 69 (21 herds (Ergas et al., 2006;Lafi et al., 2020). However, while the zoonotic risk of Q fever from small ruminants and cattle is well established, the potential risk from camels in the region is poorly understood. Deeply held religious and cultural beliefs regarding the healing benefits of camels' milk and urine -meaning that among rural communities in the Arab In desert-dwelling pastoral communities, camels and small ruminants are commonly kept together, with potential for pathogen transmission between species (Selmi et al., 2018 and 'Herd owner owns >50 goats', these variables were included in separate multivariable models. In these models, all variables listed continued to demonstrate significant association (p < 0.05) with C. burnetii seropositivity.
In the study population, almost a third of camel owners and their households were found to be frequently drinking raw camel milk from C. burnetii positive herds. This indicates a clear public health risk from Q fever in camels in the region -alongside other risk-associated camel-engagement activities such as cleaning pens, handling afterbirth, facilitating calving, slaughtering, handling raw meat or camel-tick bites (Devaux et al., 2020;Mohammadpour et al., 2020). In addition, high C. burnetii seroprevalences in camels suggest likely production losses through infertility, abortions, stillbirths, weak off-spring, milk drop or chronic mastitis, all with important economic impact (Plummer, 2022).
Where camels and small ruminants were owned together, simply owning sheep and/or goats was not found to be associated with a significant increase in C. burnetii seroprevalence in these camel herds. However, larger small ruminant flock sizes were significantly associated with higher C. burnetii seroprevalence in camel herds associated with these flocks. Small ruminant flock sizes of >50 sheep or goats were associated with increased C. burnetii seroprevalence in associated herds. In addition, the risk associated with large goat flocks was greater than that associated with large sheep flocks. This is consistent with previous findings identifying goats as posing a greater risk for C. burnetii transmission than sheep (Lafi et al., 2020;Vellema et al., 2021).
Our study findings suggest that controlling Q fever in small ru-  (Selim & Ali, 2020).
Due to the high shedding known to occur in small ruminant birthing and aborted materials, small ruminant flocks should be separated from camel herds during the lambing/kidding period where possible (Devaux et al., 2020). Separate birthing areas for small ruminants and camels are advisable, due to C. burnetii's ability to form tenacious spore-like cell variants, capable of remaining in the environment for more than a year, (Devaux et al., 2020;Plummer, 2022). In addition, because the pathogen can travel long distances via the wind, these areas should be as far apart as possible, and moved regularly. The practice of keeping herds together as a single group throughout the year was associated with significantly higher C. burnetii seroprevalences in studied herds. This is likely explained by the increased exposure to infected camel birthing products in these herds, compared to herds where pregnant females are removed prior to calving (Devaux et al., 2020). This suggests removal of pregnant females from the herd prior to calving to be a potentially important management intervention in reducing C. burnetii transmission between camels. The importance of herd owner hygiene through hand washing (and use of disinfectant foot baths where practical) after working with small ruminants at parturition time, should also be stressed (Bardenstein et al., 2021;Musallam et al., 2016;Signs et al., 2012).
Ticks are known to play an important role in C. burnetii transmission in livestock populations globally (Devaux et al., 2020). In TA B L E 3 Percentage of study population exposed to potential Coxiella burnetii transmission pathways, using a Bayesian method to predict positive herds with 95% confidence, among camel owning households in southern Jordan, October 2017 to October 2018 (due to the potential influence of maternal immunity, exposure to seropositive camels ≤6 m was not included)  (Getange et al., 2021;Mumcuoglu et al., 2022). This suggests that aggressive tick control, using frequent acaracide washes (for example monthly or every 2 months), plays an important role in Q fever control among camel populations in the region, with further research required (el-Azazy, 1996). In addition, parallel tick control in small ruminant populations associated with these herds, for example through quarterly dipping, could also be expected to offer a protective effect.
Racing animals are widely owned among camel-owning communities in southern Jordan, with seroprevalence significantly lower among these animals. This is likely explained by separation of racing camels from the main herd, and small ruminant flocks, for training and management purposes, rather than socioeconomic factors. However, parturition during racing lifetime is limited, and racing camels are not usually used in milk production, meaning lower Given the high percentage (over 30%) of individuals in camelowning households drinking raw camel milk from C. burnetii positive herds, educational efforts to promote boiling of camels' milk should be encouraged. However, in view of the profound cultural barriers likely to be encountered, detailed ethnographic studies to identify public health interventions that are culturally appropriate should first be conducted. In summary, Q fever represents an important zoonosis in the Middle East region and beyond, with high population seroprevalences previously recorded. High C. burnetii seroprevalences identified in camels, alongside widespread engagement in high-risk camel-associated practices, including consumption of raw milk, suggest camels likely present a high-risk species for human infection, with culturally appropriate veterinary and public health interventions urgently needed.