Introduction

Zoonotic pathogens maintained by ruminant reservoirs, such as Brucella species and Coxiella burnetii, are of concern in agricultural communities, particularly in low-income countries (LICs) where animals and humans frequently interact, infection status of livestock holdings is often unknown and controls are lacking. Workers in the livestock industry, including slaughterhouse workers, farmers of ruminants, animal healthcare workers and veterinarians, are particularly vulnerable to such zoonoses which not only impact health and wellbeing but also livelihoods through worker’s reduced capacity to do labour and through livestock productivity losses [1].

Brucellosis is transmitted through consumption of dairy products, via contact with infected animals and their bodily fluids, and inhalation of contaminated aerosols [2]. The main causative organisms are Brucella abortus and Brucella melitensis, the main reservoirs of which are cattle, and small ruminants respectively [3]. While commonly presenting clinically as a non-specific febrile illness, infection may result in more serious, chronic sequelae such as osteoarticular disease including spondylitis and osteomyelitis, neurological disease and endocarditis [4–6]. A recent study looking at peri-urban dairy production in West and Central Africa found a Brucella herd prevalence of 62% in dairy herds from the surroundings of the capital city of Togo, Lomé [7]. Brucella melitensis was isolated from cattle hygroma samples in this region (personal communication).

Coxiella burnetii is most commonly transmitted through inhalation of aerosolized particles, though dairy product consumption and direct contact with bodily fluids of infected animals are also implicated [8,9]. The disease caused, Q fever, may also present as acute fever, with a proportion of infections progressing to hepatitis, pneumonia, endocarditis and chronic fatigue [8,10,11]. A livestock study, in Northern Togo, found an individual C. burnetii seroprevalence of 16.1% in cattle, 16.2% in sheep, and 8.8% in goats [12].

More than half of people in Togo work in agriculture, despite a rapidly urbanising population [13]. Local livestock production systems are largely informal and may magnify risk of exposure to these pathogens due to unrestricted grazing and transhumance, leading to a high-level of mixing between species and across large areas, and manual milking and slaughter of animals with limited hygiene measures [14].

Due to the non-specific clinical presentation, and lack of awareness and laboratory tests for these diseases, they are often misdiagnosed as other febrile illness such as malaria and typhoid fever [8,10]. There is a scarcity of good quality data on the level of human exposure to Brucella species and C. burnetii in Togo [10,15]. An increased focus on assessing the risk in these populations is required to control the burden of such preventable, and often poverty-related diseases.

To our knowledge, no previous studies to ascertain the prevalence of Brucella or C. burnetii seropositivity have been undertaken in people who work with dairy cattle in Southern Togo. In this study we aim to focus on known at-risk populations, given the high prevalence of brucellosis in large ruminants in the maritime region of Togo, to ascertain the prevalence of past infection with Brucella, and C. burnetii. We used a cross-sectional serosurvey in abattoir and farm workers in the peri-urban area of Lomé, Togo. We also assessed risk factors for seropositivity to these pathogens.

Methods

Statistical analysis

Data were analysed using Stata 17 (Statcorp, College Station, TX, USA). Cross-tabulations were used to describe participants. The outcomes were positive Brucella or C. burnetii serology, defined as any individual who tested positive to either RBT or Brucella IgG ELISA for Brucella, and either Phase 1 or Phase 2 ELISA for C. burnetii. Borderline results were considered seronegative. Prevalence was calculated for both Brucella and C. burnetii seropositivity and 95% confidence intervals were computed using cluster-robust standard errors, with clustering at the site (individual farm/abattoir) level, and chi-square tests were used to compare clinical symptoms by serological status.

A hierarchical conceptual framework (Fig 1) was developed to guide the analysis, grouping potential risk factors from distal to proximal. Age and site of work (abattoir or farm) were considered a priori confounders. The association between seropositivity and farm-related exposure variables in farmworkers, slaughter-related exposure variables in abattoir workers, and animal products consumption in all participants, were respectively investigated using logistic regression, with robust standard errors to account for site-level clustering. The likelihood ratio test was used to check for interaction between site of work (abattoir or farm) and the exposure variables. Due to data sparsity, multivariable analysis only examined the association between consumption variables and Coxiella seropositivity. The Wald test was used to assess evidence of association between the exposure variables and the outcome in the multivariable model. Models were assessed for multicollinearity and if detected the collinear variable which was least associated with the outcome was removed. Confounding was assessed by a 20% or more change in main exposure odds ratios when potential confounders were included in the model. In this paper, a p-value between 0.1 and 0.05 is interpreted as weak evidence, 0.001 < p < 0.05 is interpreted as good evidence and p < 0.001 as strong evidence.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Hierarchical conceptual framework of the risk factors for seropositivity for Brucella and Coxiella burnettii.

Legend: Shading becomes darker the more proximal a variable is to the outcome. Solid shading: variables on the pathway from site of work (abattoir or farm). Diagonal lines: variables on the pathway from age. Diamonds: variables on both pathways (site of work, and age).

https://doi.org/10.1371/journal.pntd.0012657.g001

Results

Risk factors for Brucella seropositivity

In a logistic regression model adjusted for age only, no variable was associated with Brucella seropositivity in the abattoir group (Table 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Age-adjusted association^a between animal-contact risk factors and Brucella and Coxiella seropositivity in abattoir workers.

https://doi.org/10.1371/journal.pntd.0012657.t001

There was some evidence that animal ill health on farms was associated with Brucella seropositivity in farmworkers in a model adjusted for age only (abortion in goats OR 5.22, 95% CI: 1.01-26.99, p = 0.05; death of young cattle OR 6.42 95% CI: 0.72-57.40, p = 0.1; death of young goats OR 5.81 95%CI: 1.0-33.81, p = 0.05) (Table 2). The odds of human Brucella seropositivity were five times higher on farms with a positive milk test for antibodies against Brucella species than those with a negative test in the same model (OR 5.15, 95% CI: 1.21-21.97, p = 0.03).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Age-adjusted association^a between farm-level and animal husbandry risk factors and Brucella and Coxiella seropositivity in farmworkers.

https://doi.org/10.1371/journal.pntd.0012657.t002

There was weak evidence that participating in small ruminant husbandry was associated with nearly four times higher odds of Brucella seropositivity (age adjusted OR 3.9 95% CI 0.81-19.13, p = 0.09), as was assisting with cattle abortions (age adjusted OR 3.66, 95%CI: 0.85-15.84, p = 0.08).

In the combined farm and abattoir worker analysis farmworkers had nearly twice the odds of being seropositive compared to abattoir workers (age adjusted OR 1.93, 95% CI: 0.94-3.93, p = 0.07). Consuming raw cow’s milk on a monthly basis or more was associated with higher odds of Brucella seropositivity (OR 3.79, 95% CI: 2.34-6.13, p < 0.001) (Table 3) adjusted for age and site of work.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Association between socio-demographic and dairy consumption factors and seropositivity to Brucella species in all participants^a.

https://doi.org/10.1371/journal.pntd.0012657.t003

Risk factors for Coxiella seropositivity

Adjusting for age and considering farm-level clustering, there was no evidence of an association with any livestock husbandry or livestock health variable in farmworkers (Table 2).

After adjusting for age only, there was no evidence that abattoir activities were associated with seropositivity, other than weak evidence of an association with butchery/skinning of small ruminants (p = 0.09) or in participating in any husbandry with sheep/goats (OR 2.47, 95% CI 0.88-6.96, p = 0.09) (Table 1).

The odds of Coxiella seropositivity in those who consumed fermented milk (OR 1.59, 95% CI: 1.26-2.00, p < 0.001) and in those who consumed cheese (OR 1.70, 95% CI: 0.97-2.98, p = 0.07) were greater than in those who did not when looking at all participants in a multivariable model considering dairy product consumption (Table 4).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Baseline^a and Adjusted^b Odds Ratios for associations between dairy consumption factors and seropositivity to Coxiella burnetii in all participants.

https://doi.org/10.1371/journal.pntd.0012657.t004

Discussion

We found, through a cross-sectional study of livestock workers, evidence of Brucella and Coxiella burnetii exposure in people with occupational exposure in peri-urban Lomé, with overall human seroprevalences of 9.5% and 28% respectively. Seroprevalence varied with site type, with a higher seroprevalence to Brucella on farms (13.6% compared to 6.5% in abattoirs) and to C. burnetii in abattoir workers (31.5% compared to 23.5% on farms). Bulk milk samples from associated cattle herds found that 39% (95% CI 27–53%) of farms also had Brucella-positive milk samples, lower than a previous study in Lomé’s peri-urban herds in 2017(62%, 95%CI 55–69%) [7]. Seropositivity to either pathogen was associated with the consumption of dairy products (raw milk in the case of Brucella and fermented milk and cheese in the case of Q fever), and ruminant husbandry (small ruminant husbandry and assisting with cattle abortion in farmworkers for Brucella, and small ruminant husbandry in abattoir workers for Q fever). The odds of brucellosis seropositivity in farmworkers was also increased with morbidity/mortality of owned ruminants (death of young cattle and goats, and abortion in goats). Considering this, the widespread ill-health in livestock including death of young animals and abortions which are consistent with brucellosis/coxiellosis, consumption of raw milk and dairy products, and lack of PPE (personal protective equipment) usage is of concern. However, due to the small sample size, analyses, other than that examining the association between C. burnetii seropositivity and dairy consumption, were limited to minimally adjusted models and as such results should be interpreted with caution.

Incidence of both infections varies greatly between contexts, including in different parts of the same country, and over time [10,15]. For this reason, it is difficult to make assumptions based on previous studies to inform policy. Our findings on Brucella seroprevalence were in contrast to a 2013 study in Northern Togo which found a seroprevalence of 2.4% in Fulani villagers, a comparable population to the farm workers in this study who had a seroprevalence of 13.6% [12]. This may be due to a lower prevalence in animals (9.2% in village cattle, 7.3% in transhumant cattle and 0% in small ruminants) though these results are from individual serology rather than bulk milk sampling [12]. The higher seroprevalence to Coxiella burnetii we note is aligned with the same study which found more than ten times the prevalence against Coxiella compared to Brucella in both Fulani and non-Fulani groups [12]. Similar, though not as marked, findings were seen in studies in Ethiopia, and Kenya [16,17,19]. Despite a higher seroprevalence of Coxiella burnetii in many African contexts, brucellosis is often given greater attention by health agencies and research funders [20,21]. Q fever can have severe health sequelae, being responsible for up to 5% of endocarditis cases globally, and if this disparity in seroprevalence is reflected in clinical burden then it is important to reassess Q fever as a priority [22].

There was a C.burnetii seroprevalence of 31.5% in the abattoir workers, despite the lack of abattoir-specific activities being associated with seropositivity. This was in line with de Boni et al who found no specific abattoir activity was a risk factor and hypothesised that workers were exposed when the bacterium was aerosolized during slaughterhouse activities [23]. Aerosols are a major transmission mode for Coxiella burnetii and this abattoir with its lack of ventilation provides ideal conditions for exposure of employees regardless of role [11]. The finding that Dutch cull workers were at greater risk of seroconverting if they worked indoors rather than outdoors supports this [24].

For both diseases infected livestock are the main reservoir, shedding bacteria in birth and abortion materials, milk, faeces, and urine [6,11]. Close contact with ruminants has been shown to be associated with seropositivity to Brucella and C. burnetii, as was seen for Brucella in our study in farmworkers who participated in small ruminant husbandry and assisting with cattle abortions, and for C. burnetii in abattoir workers who participated in small ruminant husbandry [25,26].

Abortion, birth of weak offspring and death of neonatal animals are symptoms of ruminant brucellosis and coxiellosis [27,28]. We demonstrated associations between human Brucella seropositivity and livestock ill-health markers (abortion in goats, and the death of young cattle on farm). Literature on this is limited and largely confined to animal abortion [29]. Animal-health variables such as abortion, and mortality in neonatal animals, may be indicators of herd/flock infection, and could be monitored using syndromic surveillance to mitigate risk to workers and livelihood losses. The association between Brucella seropositivity and presence of abortion in goats, and with participating in small ruminant husbandry is of particular interest as we know B. melitensis is circulating in ruminants in this region (personal communication). This highlights the complexity of transmission amongst species and to humans, and the need for multi-species studies.

This population consumed a high-level of bovine dairy products, including those sourced from raw milk. We found both consuming fermented milk and cheese were associated with Coxiella burnetii seropositivity. A 2019 study of hard sheep’s cheese purchased in Spanish supermarkets demonstrated the presence of viable C. burnetii in cheese samples using experimental inoculation of mice and culture in Vero cells, showing dairy products may be viable transmission vehicles [30]. However, there was no evidence that consumption of any cow’s milk or milk/dairy products from sheep/goats were associated with Coxiella burnetii seropositivity. Previously dairy products were considered to rarely contribute to C. burnetii transmission but the evidence is mixed with some studies finding an association [8,12,31].

Regular raw cow’s milk consumption increased the odds of Brucella seropositivity. Multiple studies have found consumption of unpasteurised milk to be a risk factor [32–34]. Consumption of small ruminant dairy products has also been shown to be a risk factor for brucellosis [35,36]. However, consumption of small ruminant dairy products isn’t culturally common in Togo, despite large numbers of these species, and those who did consume these weren’t found to have increased odds of Brucella seropositivity.

As sampling frames for agricultural occupational groups in Togo do not pre-exist, without a comprehensive census it is a challenge to sample a representative group. There are also practical limitations in carrying out a questionnaire-based survey in occupational groups with limited time for participation. Dean et al regarded a lack of detailed information on exposures as a limitation in thoroughly assessing risk [12]. While our questionnaire allowed more nuanced examination of risk factors it also limited the sample size. For this reason we could not perform certain analyses such as adjusting for other exposures of interest, including other animal husbandry and dairy consumption variables, in multivariable analysis for Brucella seropositivity.

There is a paucity of studies looking at linked animal and human data. Here we were able to link human data on brucellosis to milk sourced from associated cattle to show an association between presence of a positive bulk milk sample and human seropositivity. Bulk milk sampling is relatively affordable and easy and could be utilised in regular herd surveillance. To expand on this study, future multidisciplinary studies which conduct direct multi-species sampling, including small ruminants, would aid in elucidating transmission dynamics where human, animal and environmental health are highly interconnected. In the case of Brucella, this should include species identification and pathogen characterisation of any isolated bacteria. Such One Health studies could be expanded upon by examining other relevant populations and settings, and the economics of infection impacts and control.

There were low levels of PPE use in both groups. Many studies, though not all, have found PPE to be beneficial in protecting against both infections [37,38]. PPE use should be recommended, particularly adequate masks (P2/N95) in abattoir areas where aerosols are produced [39].

This population is affected by both zoonoses, and there is a need for preventative measures. Implementation should involve at-risk workers, trusted community leaders, health practitioners and policymakers using culturally sensitive, pragmatic and economically feasible approaches. Our study highlights the need for increased clinician awareness about zoonoses, and for relevant history taking including occupation, which might raise clinical suspicion, particularly as malaria prevalence declines [40]. Seropositive participants commonly experienced febrile disease though no symptoms were found to be associated, and this non-specific clinical picture hinders diagnosis by clinicians. Many seropositive participants were only tested for malaria and treated with antimalarials and non-specific antibiotics, similar to other studies [4,41]. In order to improve diagnosis healthworkers require resources to test and treat non-malarial illnesses, both to improve clinical outcomes and to reduce inappropriate prescribing which contributes to selection for antimicrobial resistance. In parallel to this, sensitisation of at-risk populations to improve awareness of these infections and mitigation strategies that could be utilised should be undertaken.

Due to the complicated interactions at the human-animal-environment interface in many zoonoses more complex interventions involving interdisciplinary teams tend to be required [20]. Integrated control of multiple zoonotic diseases, such as Q fever and brucellosis, through interventions targeting both the animal and human populations could have a dual benefit through improving animal health and livelihoods, while simultaneously decreasing the risk of human disease, to alleviate the cycle of poverty. Such integrated approaches would be more cost-effective by optimising resource use, especially when working with marginalised communities who are isolated from public services [21]. The findings of our study provide evidence which can be used to tailor such public health interventions against both Brucella and Q fever in occupational groups working with livestock. A number of countries in the region, including Côte d’Ivoire and Ghana, have recently undertaken a One Health Zoonotic Disease Prioritization exercise using the tool developed by the U.S. Centers for Disease Control and Prevention which aims to not only help prioritize disease of greatest public health impact but strengthen multisectoral collaboration and focus resource allocation, and Togo would benefit from a similar process [42,43].

Supporting information

Acknowledgments

We thank all the livestock workers and abattoir workers from the municipal abattoir in Lomé who participated in this study, and community leaders who facilitated this study. This study was part of a larger UKRI Zoonosis in Emerging Livestock Systems project and we thank Professor Ayayi Akakpo from the Ecole Inter-Etats des Sciences et Médecine Vétérinaires de Dakar (EISMV) for facilitating the study. We thank Wemboo Afiwavi Halatoko, Koffi Akolly, Zoulkarneiri Issa, Adjaho Komla Koba and Afeignidou Telou at the INH and Emilie Go-Maro, Rufine Yempab, and Komlan Mawufemo Adjabli at the Department de l’Elevage who were involved in processing, testing and storage of samples. Our appreciation goes to the zootechnicians of the Department de l’Elevage Koffivi Amewou, Kom Agbéssi Adjessoklou and Dossou Zanou who facilitated all visits to dairy farms. Thanks also go to the field team; Kodjo Afangbegnon, Robertine Amoussou and Kocou Jean-Paul Amoussou.