Seroprevalence and risk factors for bovine brucellosis in the Chittagong Metropolitan Area of Bangladesh

Abstract Brucellosis is a neglected endemic zoonosis in Bangladesh and has a significant impact on public health and animal welfare of dairy farming as well as dairy farm economics. A cross‐sectional study was conducted to evaluate the seroprevalence of and risk factors for brucellosis in dairy cattle in the Chittagong metropolitan area (CMA) of Chittagong, Bangladesh. We collected serum samples (n = 158) from six randomly selected dairy farms in the CMA between February and November, 2015. The Rose Bengal Plate Test (RBPT) and a competitive ELISA (cELISA) were used as the screening and confirmatory tests respectively. Farm level and animal level demographic and risk factor data were collected using a questionnaire. The risk factors were analysed using a multivariable logistic regression with random effects. The overall seroprevalences of antibodies against brucellosis in cattle were 21.5% (34/158) and 7.6% (12/158) based on parallel and serial interpretation of the two tests respectively. Our results revealed that 20.3% (32/158) samples were positive using the RBPT and 8.9% (14/158) were positive using the cELISA. The within‐herd seroprevalence ranged from 10% to 26.3% and 5 to 20.7% using the RBPT and cELISA tests respectively. The odds of seropositivity were significantly higher in lactating cows (OR: 2.59; 95% CI: 1.02–6.55), cows producing less than 2 litres of milk (OR: 29.6; 95% CI: 4.3–353.8), cow producing 2–12 litres of milk (OR: 4.8; 95% CI: 1.1–33.4) and cows with reproductive disorders (OR: 3.2; 95% CI: 1.2–10.1). About 7.6% (12/158) and 1.3% (2/158) of cattle were found to be infected with acute and chronic brucellosis respectively. Based on these results, we suggest that cows that have reproductive disorders and are producing little milk should be prioritized for brucellosis screening in CMA. The screening tests should be used to control brucellosis in cattle in order to protect animal welfare, human health and to minimize the economic losses.


| INTRODUC TI ON
Brucellosis is an important disease impacting veterinary and public health worldwide and caused by bacteria of the genus Brucella (Deka, Magnusson, Grace, & Lindahl, 2018). Globally, it is the second most frequently reported zoonotic disease to the World Organization for Animal Health (OIE) as it is regarded the most devastating trans-boundary animal diseases, which cause significant trade obstructions (OIE, 2020;WHO, 2015).
The diagnosis of brucellosis continues to be challenging in developing countries like Bangladesh. Rose Bengal Plate Test (RBPT) is the most commonly used conventional screening test for brucellosis in animals (Musallam, Abo-Shehada, Omar, & Guitian, 2015;Rahman, 2015). RBPT relies on the unique antigenic properties of lipopolysaccharides (LPS) that are present within the cell membrane of Brucella spp.; however, the LPS antigen is also present in a number of other gram negative bacteria, including Vibrio and Yersinia enterocolitica, which may cross-react on the brucellosis diagnostic assays (Munoz et al., 2005).
The OIE recommends the use of Competitive Enzyme Linked Immunosorbent Assay (cELISA) for international cattle trade as this assay is reported to be more specific than RBPT (OIE, 2016;Rahman et al., 2019;Wang et al., 2015).
Chittagong District is one of the most important intensive dairy production regions in Bangladesh with 109-175 cattle/ kmsq (Huque & Khan, 2017). However, only one previous study reported the seroprevalence and risk factors for brucellosis from Chittagong district (Sikder et al., 2012). Our objectives were to estimate the seroprevalence of brucellosis and to identify the risk factors for brucellosis in dairy cattle of Chittagong Metropolitan Area, Bangladesh.

| Ethics statement
Informed verbal consent was obtained from all animal owners for the collection of blood samples from their cattle (Appendix S1). This study was approved by the animal Ethics Committee of Chittagong Veterinary and Animal Sciences University, Bangladesh (Approval no: EC/2014/34-7).

| Study area, design and target population
This study was conducted in dairy cattle in the Chittagong metropolitan area (CMA), Bangladesh (22°22'N and 91°48'E) from February to November 2015. The study area is situated in the tropical zone and characterized by an annual average range temperature of 13-32°C, 5.6-727 mm of rainfall and 70%-85% humidity of (Anon, 2016). Chittagong metropolitan has 14 distinct areas in five sub-regions like Chandgoan

| Sample size determination
The required sample size (n) to estimate seroprevalence with 95% confidence was calculated based on the equation 1 (Charani et al., 2010).
where P is the expected seroprevalence (10%) and d is the desired precision (5%). We calculated a minimum required sample size of 138.

| Sampling strategy and distribution of sampled farms and cattle population
Within the farm, we could not collect samples from every animal due to some problem with technical and proper restraining of animals.
Moreover, for few animals, the owner did not allow to collect the samples, as they are in advance stage pregnancy. Thereafter, all eligible cattle (158) belonging to the selected farms were sampled and almost

| Collection of cattle blood samples and recording epidemiological data
To collect the blood samples, the cattle were run through a chute and held with a head gate. We conducted the venipuncture using the jugular vein and collected 5-10 ml of blood using serum tubes. The blood sample was labelled using the tag number assigned to each individual animal. The tubes were held vertically at room temperature for 1 hr and were then refrigerated at 4°C overnight before spinning at 3,000 rpm (2,555 g) for 10 min. The separated sera samples were

| Laboratory evaluation
Two serological tests were used to evaluate serum samples for brucellosis in cattle, the RBPT (Rahman, 2015) and cELISA (Chikweto et al., 2013).

| Rose Bengal plate test
The RBPT was conducted using the Atlas Brucella test ® (Atlas Medical, Cambridge), which is a rapid agglutination test, and was conducted per the manufacturer's instructions. The sensitivity and specificity of RBPT for the diagnosis of bovine brucellosis in Bangladesh were reported to be 87.4% and 99.4% respectively (Rahman et al., 2019). Briefly, equal volume (30 µl) of RBPT reagent and serum were mixed and rotated on a glass slide for 1 min. The result was considered positive if visible agglutination was identified positive and negative control sera samples used during RBPT were collected from Department of Medicine laboratory, Bangladesh Agricultural University.

| Enzyme linked immunosorbent assay
The SVANOVIR Brucella-Ab cELISA was used following manufac-

| Data entry and statistical analysis
Field and laboratory data were cleaned and coded in MS Excel 2007 before exporting to STATA-13 (StataCrop) for epidemiological analysis.

| Descriptive analysis
Categorical variables were summarized as frequency, percentages and 95% confidence intervals (95% CI); and continuous variables were summarized as mean ± standard deviation (SD

Univariable analysis
The associations between seropositivity and categorical risk factors were tested using univariable random effects (RE) logistic regression analysis considering farm as random effect.

Multivariable analysis
Independent variables that were significantly associated with seropositivity in the univariable analysis (p < .20) were included in to the multivariable RE logistic regression model. In the multivariable analysis, a backward elimination procedure was used applying the maximum likelihood estimation procedure and statistical significance of contribution of individual predictors (or group of predictors) to the models tested using the Wald's test and likelihood ratio test (Dohoo, Martin, & Stryhn, 2003
The herd sero-status based on cELISA result is presented in Figure 2. One farm had no cattle that tested seropositive on the cELISA (Panchlaish).

| Univariable analysis
In Fisher's exact test and univariable logistic regression analyses, lactation, anestrous, a history of reproductive disorders, milk yield, lactation number, trimester and abortion were significantly (p ≤ .2) associated with brucellosis seropositivity (Table 4). of goodness of fit was not significant (p = .14) and the area under the receiver operating curve (ROC) was 0.70, indicating that the model fitted the data well and had a high predictive ability to discriminate seropositive and seronegative animals (Figure 3).

| Farm management practice
All six farms followed regular deworming schedules, utilized artificial insemination (AI) for breeding and did not allow their cattle to graze.
Five farms had cement floors with no maternity pens and did not test the animals for brucellosis prior to introduction into the herd. Cattle on three of the farms consisted of animals that were both raised on the farm as well as animals from other sources (market, neighbour etc.).
Cattle from two farms and those from the last farm were born on the same farm and obtained from another source respectively (Table 6).

| Comparison of the serological test results
About 7.6% (12/158) and 1.3% (2/158) of cattle were found to be infected with acute and chronic brucellosis respectively. The relative sensitivity and specificity of the RBPT was found 85.7 and 60% respectively. The Kappa statistics value was 86% suggesting a very good agreement between the tests (p < .001) ( Table 7). Lactating animals were significantly associated with a higher risk of being seropositive to brucellosis both in univariable logistic regression and multivariable logistic regression analyses. Separate study from Ethiopia and Uganda reported, seropositivity was found in lactating and pregnant cows only (Adugna, Agga, & Zewde, 2013;Bugeza et al., 2019). In the non-lactating group, there were some heifers. Sexually mature and pregnant cows are thought to be more susceptible to brucellosis than sexually immature cattle of either sex (Adugna et al., 2013). This pattern might have been attributed to the affinity of this bacterial pathogen to the pregnant uterus, to erythritol in fetal tissue, and possibly to steroid hormones that are higher during pregnancy (Barbier et al., 2017).

| D ISCUSS I ON
In our study, the odds of seropositivity increased as milk production decreased. This finding was expected as one of the clinical signs of infection with brucellosis is that cows may have lower milk production. Decreased milk production is also associated with various diseases of the reproductive tract. Therefore, the seropositive cows might also have been suffering from different reproductive diseases like metritis or endometritis from the last parturition that led to physical problems, resulting in lower milk production (Patel et al., 2014). The last trimester of gestation when the cow reached the last stage of lactation (i.e. less milk production) is a likely period for Brucella to infect the host (Islam, Khatun, Werre, Sriranganathan, & Boyle, 2013;Xavier, Paixão, Poester, Lage, & Santos, 2009).
In our study, only female and cross breed animals were included; so no comparisons could be made between sexes and breeds. Higher seroprevalence of brucellosis in female and cross breed animals had also been reported by various studies (Joseph, Oluwatoyin, Comfort, Judy, & Babalola, 2015;Terefe, Girma, Mekonnen, & Asrade, 2017).
As cELISA is based on the specific epitopes of the (O-polysaccharide), it can therefore eliminate some of the cross-reaction and false negative problems seen in other serological tests.
We found that the RBPT and cELISA results agreed for 86% of the sampled. Previous studies that also compared RBPT with cELISA found agreement in 97% of samples (Ahasan et al., 2010;Rahman, 2015). The sensitivity was within the range (70.6%-97.7%) reported by Rahman (2015); however, the specificity was lower than the range (84.3%-99.9%) reported in that study for RBPT. The low sensitivity and specificity are not surprising (Kanani, 2007) as the RBPT cannot differentiate antibodies originating from infection with other gram negative organisms (Ducrotoy, Conde-Álvarez, Blasco, & Moriyón, 2016) and it is known to give false negative results in early stage of infection, or immediately after abortion (Mohammed, 2013).
In this study, 32 samples were positive by RBPT, 12 of them were positive by cELISA. In contrast, two samples that were negative by RBPT were positive by cELISA. RBPT is able to detect IgM and IgG, TA B L E 6 Existing farm management practices in dairy cattle in Chittagong Metropolitan City, Bangladesh (N = 6, n = 158)  We investigated the knowledge level among farmers regarding brucellosis infection, transmission, control and prevention. Our study suggests that most of the people that responded the questionnaire were not aware of brucellosis, which was also observed by Sikder et al. (2012). Knowledge of a disease is a crucial step in the development of prevention and control measures (Gumi et al., 2011). In the present study, majority of livestock keepers (83.3%) were not aware of brucellosis and its zoonotic potential. This lack of knowledge means that it is likely that farmers do not take required precautions when handling Brucella infected animals, animal products and animal by-products. Moreover, with these results, it is certainly that no precaution was taken to prevent spread of the disease to other herds within or outside the study area. The perception that brucellosis can be cured and the habit of selling diseased animals either to the market or other livestock keepers can lead to propagation of the disease to other areas or herds which are not infected (Holt et al., 2011).
The livestock keepers that participated did not separate animal(s) that abort from their other animals. They also were not aware of any potential modes of transmission of diseases from animals to people, except through direct contact with aborted calves and tissues. As a result of this lack of awareness, workers continue to participate in highrisk behaviours, including home slaughter of cattle and subsequent meat preparation (Lindahl, Sattorov, Boqvist, & Magnusson, 2015).
The majority of participants reported that they fed aborted fetuses to stray dogs or threw aborted materials into water canals used by small ruminants and other livestock for drinking or bathing. Dogs have been suggested to act as mechanical vectors while they drag aborted materials over the ground and increase the area where the bacteria is spread (Aparicio, 2013). The relationship of dogs and outbreak of brucellosis in cattle has also been demonstrated earlier (Wareth et al., 2017). Contamination of the water may increase the risk of disease transmission to people and other animal populations in that use those water sources (Wael, Tayel, Eltholth, & Guitian, 2010).
Strict biosecurity, restriction of animal movement and vaccination are suggested as effective control of brucellosis (Rahman, 2015).
Additionally, strain 19 and RB51 vaccines are commonly used to protect cattle against infection and abortion (Dorneles et al., 2015); however, in Bangladesh, vaccination is not recommended in cattle reared under small-scale dairy and subsistence/backyard management system due to very low seroprevalence (Rahman et al., 2019).
However, in high within-herd seroprevalence scenario, mass vaccination (avoiding pregnant animals in mid-gestation) could be the most effective and practical method for Bangladesh.
As the brucellosis is endemic in Bangladesh, we recommend that a national control strategy be developed. In preparation for this, further research should be done to assess the impact of brucellosis on the livestock economy, livestock health and human health.
Additionally, the national veterinary service must be strengthened to carry out the strategy, which includes increased collaboration between public health and veterinary services. Further, simulations of the costs associated with various control or eradication strategies must be evaluated to support this strategy.
Due to the small sample of farms in the study area, inter-farm transmission factors and farm-level variables (usual management practices in aborted cases, rearing other animals in to the farm etc.) were not investigated by statistical models. These are known to be important for brucellosis spread and maintenance in a farming system (Addis & Desalegn, 2018) and further research to characterize these risk factors is recommended. Other limitations of our study were that we only included female animals although male cattle can be infected with Brucella and play an active role in its transmission and not all cattle on a farm could be sampled due to difficulties in restraining and handling (lack of facilities at farms).

| CON CLUS ION
This study suggests that acute brucellosis is more frequent in the dairy cows of the study area. Cows that have reproductive disorders and are producing little milk should be prioritized for brucellosis screening in CMA. The screening tests should be used to control brucellosis in cattle in order to protect animal welfare, human health, and to minimize the economic losses. Moreover, culling of the acutely infected animals will decrease the spread of the disease in populations and thereby the risk of human brucellosis.

ACK N OWLED G EM ENTS
We thank the University Grants Commission (UGC), HEQEP

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.