Seroprevalence of Small Ruminant Brucellosis in Ethiopia: Systematic Review and Meta-Analysis

Abstract

The World Health Organization (WHO) has classified brucellosis as a neglected zoonotic bacterial disease and determined that it has the largest public health burden among all community segments. The purpose of this research is to perform a Meta analysis and systematic review on seroprevalence of small ruminant brucellosis in Ethiopia.The data searching journal like PubMed, Science Direct, Scopus, Embase and Google Scholar was used to search the articles. All articles are screened, which was reported seroprevalence of small ruminant brucellosis in Ethiopia to be included in the study. This systematic review and meta-analysis included 19 papers. Meta-analysis using random-effects models were made to calculate the pooled seroprevalence of brucellosis. The study determined that the estimated pooled seroprevalence of small ruminant brucellosis was 3.0% (95% CI: 0.02, 0.03). According to the subgroup analysis, a statistically significant difference was found between the disease and the study region, publication year, laboratory technique used and studies years. Subgroup analysis by study regions the highest prevalence was observed in the Oromia region, with a prevalence of 3%, whereas the lowest prevalence was reported in the Amhara region, with a prevalence of 1%. Additionally, there was some indication of publication bias in papers reporting the prevalence of small ruminant brucellosis in Ethiopia (Egger’s test, p = 0.001). This analysis demonstrates the high seroprevalence of small ruminant brucellosis in Ethiopia and the necessity of suitable intervention strategies, such as increased public awareness creations and vaccination campaigns, as well as ongoing surveillance to manage and prevent brucellosis in cattle husbandry methods.

Keywords: Brucellosis; Ethiopia; Meta-analysis; Seroprevalence; Small ruminant

Introduction

The World Health Organization (WHO) has classified brucellosis as a neglected zoonotic bacterial disease and determined that it has the largest public health burden among all community segments [18]. This is due to lack of effective control and proper disease surveillance [33,21]. Animal and public health issues related to brucellosis persist in many poor nations, such as Ethiopia, where the disease is still endemic [6].

The currently recognized species includes Brucella abortus, B. Melitensis, B. Suis, B. Ovis, B. Canis, B. Ceti, B. Pinnipedialis, B. Neotomae, B. Microti and B. Inopinata [27]. The two major Brucella species known to infect sheep and goats are B. melitensis and B. ovis; however, B. abortus has also been observed to periodically increase in sheep and goat populations [29,5]. In human, Brucellosis is always caused by B. melitensis (cause Undulant or Malta fever) followed by B. suis, B. abortus and B. canis [15].

The disease is transmitted to humans primarily from eating raw or undercooked animal products or through direct contact with infected animals. It causes a systemic infection with clinical manifestations such as fever, sweats, fatigue and joint pain [22]. The prevalence of brucellosis is affected by several risk factors such as production system, host and environmental factors [29]. In sheep and goats that are sexually mature, brucellosis is limited to the reproductive tract and typically causes placentitis and abortion in pregnant. Brucella melitensis and B. abortus are zoonotic pathogens that cause disease in humans [28,29].

Brucellosis causes significant financial losses, such as a trade barrier for animals and animal products and restricts to free animal movements [37]. In addition, it causes in losses from fetus abortion, breeding failure (culling), and decreased milk production in the affected animal population.

The disease is often prevalent in traditional pastoral communities both in animals and humans but, due to lack of awareness the disease is not diagnosed and treated [2].

Generally, poor hygiene, prevalence of the disease in animals that expose humans from infected animals or their products influence the occurrence of the disease in humans. Cattle producers, veterinarians, animal health professionals, workers in abattoirs, laboratory personnel, and members of the general public who consume animal products are among the occupational categories most at risk of infection [3]. The conventional way of living, prevailing attitudes and inadequate understanding of the illness provide conducive circumstances for the dissemination and exchange of Brucellosis. The dearth of readily available alternatives and straightforward, reasonably priced remedies makes it challenging to control the hazards connected to these behaviors. Cost-effectiveness in brucellosis control is likely to exist. To illustrate the advantages of intervention, accurate quantitative data on brucellosis in humans and livestock is crucial [35].

The prevention and control of brucellosis in small ruminants will contribute to reduce human brucellosis incidence, especially in the endemic regions of Ethiopia. Therefore, adequate knowledge of the epidemiology of Brucellosis is of great public health and economic importance, particularly amongst cattle owners, animal and animal product consumers, as this will greatly assist in controlling its infections. This study aimed to determine the pooled seroprevalence of small ruminant brucellosis by a systematic review and meta-analysis in Ethiopia.

Methods

The systematic review and Meta-analysis were performed according to the Preferred Reporting Items for Systematic review and Meta-Analysis (PRISMA) flow chart guideline [24]. The STROBE checklist was used to ensure the inclusion of relevant information from the selected articles in the analysis. The outcome of interest was the proportion for small ruminant brucellosis.

Literature Search Strategy and Eligibility Criteria

The purpose of this meta-analysis and comprehensive review was to calculate the weighted seroprevalence of small ruminant brucellosis. Literature was searched in Pub Med, Science Direct, Scopus, African Journal Online and Google Scholar databases until July 18, 2022 to September 27, 2022. During an online article search, a Boolean operator and/or was utilized by combination of keywords related to the issue.

The following were the main search terms: "brucellosis" or "Brucella"; "seroprevalence" or "prevalence" or "Seroepidemiology"; "risk factors" or "potential factors"; and "sheep and goat" or "ovine and Caprine" or “Ethiopia" AND "Small Ruminants".

We find studies with any of the keywords in their titles, abstracts, and complete texts using the Boolean operators "OR" and "AND." Moreover, unpublished thesis manuscripts were also accessed from University of Gondar library and College of Veterinary Medicine and Animal Sciences.

Inclusion and Exclusion Criteria

To ensure that the papers found during the search were eligible, we employed the subsequent inclusion criteria: 1) unique, peer-reviewed research papers and theses from Ethiopia; 2) cross-sectional studies that provided the seroprevalence of brucellosis in small ruminants; 3) articles with full-text studies; 4) research that included small ruminants in any type of management system as part of the targeted study population (intense or extensive). In this case, extensively managed small ruminants are kept on the grazing pasture and obtain their feed by grazing without supplementation; intensively managed small ruminants are those that are kept indoors the entire day or leave the house for leisure only a few hours each day; Studies reported the overall sample size and the outcome of interest (number of positive samples); 5) studies were conducted utilizing serological diagnostic tests such as, RBPT for screening and CFT or ELISA for confirmation; 6) studies provided the total sample size and the outcome of interest (number of positive samples); 7) studies published only in English language; and 8) studies published online between 2011 up to 2022. Papers which did not meet the above-mentioned criteria were excluded. Besides, the references of the selected papers will be check manually to find relevant papers that were not retrieved in the database search [36].

Study selection and Data Extraction Procedure

Records identified from various electronic databases, indexing services and directories would be exported to Endnote software version X7. We found, noted, and deleted duplicate records. Two independent researchers were extract full text data and evaluate the eligibility of them for final inclusion. In each case, the rest authors play a critical role in solving discrepancies arose between two authors to come up to consensus.

Similarly, data extraction format was prepared based on first author, publication year, study year, geographical location (region), study design, sampling method, sample size, diagnostic test, setting and number of positive samples among the study groups. Seroprevalence of small ruminant brucellosis would calculate by dividing the number of positive cases by the total number of individuals used for the study in a given population at a given period. The study effect size and their corresponding confidence intervals would be calculated from the extracted data. Microsoft Excel datasheet was used to code and manage all extracted information from all relevant studies.

Study Quality Assessment

Two independent researchers were evaluated the quality of the included papers using a quality assessment checklist (standard strengthening the Reporting of Observational Studies in Epidemiology checklist (STROBE) which includes 22 items to make up the quality assessment checklist, which covers the title, abstract, introduction, methodology, results, and discussion of the articles.

The checklist comprised items evaluating the objectives, includes various material and methods (e.g., sample size, study population, bias, statistical methods), outcomes and constraints of the research.

The checklist included items assessing objectives, different components of the methods (eg, study design, sample size, study population, bias, statistical methods), results, limitations, and funding of the studies. The article quality scores were ranged from 0 to 44. Following the checklist (STROBE), searched papers were classified into 3 groups: low quality score (<15.50), moderate quality score (15.50-29.50) and high-quality score (30.0-44.0) [17].

Meta-Analysis

Data on the seroprevalence and corresponding 95% Confidence Intervals (CIs) of the disease were calculated for each study. The pooled prevalence estimates would be compute using the formula given by [9]. Forest plot diagram was employed to present the heterogeneity among studies, outcomes of meta-analysis that display estimates of the seroprevalence, and their corresponding CIs of all included studies together with the pooled effect size. Similarly, subgroup analyses for the primary outcome (seroprevalence of brucellosis) would be done by study region, publication year, laboratory technique employed (CFT or ELISA) and sample size category.

Cochran’s Q-statistics and inverse variance index (I2) would be computed to determine the heterogeneity and inconsistency (true variation) among studies, respectively. Similarly, we considered the I2 values of 25%, 50% and 75% as low, medium and high heterogeneity respectively [19]. The tau statistics (t2) was used to assess the variance of the effect size estimates across the population of the study.

Based on the heterogeneity assessment result, we used DerSimonian and Laird’s random-effects method (if the p-value of the Q test is 5%) or Mantel-Haenszel’s fixed-effects method to pool the estimations [39]. Small study effects and publication bias presence were then visualized using funnel plot diagrams and, Egger’s and Begg’s asymmetry tests [11]. A funnel plot was computed using effect size and its corresponding standard error of the effect size. STATA software version 17 is used to do the meta-analysis.

Results

Descriptive Literature Search Results

A total of 187 potentially relevant studies were identified from several sources including PubMed, Science Direct, Scopus and Google scholar. From these, 27 duplicated articles were removed with the help of Endnote 7. The remaining 160 records were screened using their titles and abstracts and 134 of them were excluded. Full texts of 26 records were then evaluated for eligibility. From these, 7 articles were excluded due to the outcome of interest was found missing, insufficient and/or ambiguous.

A total of 19 articles were eligible for the final systematic review and Meta-analysis from all screened studies. All of the eligible studies have been used RPBT and ELISA or CFT for antibody detection. These selected eligible articles were conducted namely; Oromia, Tigray, Amhara, Somali and Oromia and Somali. From 19 published articles a total of 10,067 samples of small ruminant (both sheep and goats) were subjected to disease detection. The sample size of shoat ranges from 226 to 985 in each study area of Ethiopia. The seroprevalence of the disease in the 19 articles was ranges from 1.40% to 9.1%. The mean sample size from overall report was 528.94. Finally, a total of 19 articles fulfilled the eligibility criteria and quality assessment and thus included for systematic review and meta-analysis.

Descriptive Study Characteristics

The final 19 eligible studies which were considered for determining the seroprevalence of brucellosis in small ruminants are summarized for systematic review and Meta-analysis. The studies were published in the year between 2011 and 2022. All the selected studies were cross-sectional study design in nature.

Meta-Analysis

Pooled prevalence estimate: Due to the expected variation between studies, random-effects meta-analyses were employed using the total sample size and number of positives (effect size and standard error of the effect size). An overall pooled prevalence of the disease was estimated to be 3% (0.02 to 0.03 of 95% CI).

Summary of Meta-analysis: Random-effects meta-analyses were employed using the prevalence and standard error of prevalence for effect size and standard error of the effect size and using author and publication year for the study label of the Meta-analysis.

Forest Plot

Due to the expected variation between studies, random effects meta-analyses were carried out using the prevalence and standard error of prevalence (effect size and standard error of the effect size). (2 = 0.00; I2 = 85.65%, DF = 18, H2 =6.96, Q - test = 82.72 and P - value 0.00). Individual study prevalence estimates ranged from 1.40% to 9.1% with the overall random pooled prevalence of 3% (95% CI: 0.02, 0.03). Studies weighted approximately equal with weights on individual studies ranging from 2.94% to 6.19% due to high heterogeneity between studies.Due to the expected variation between studies, random effects meta-analyses were carried out using the prevalence and standard error of prevalence (effect size and standard error of the effect size). (2 = 0.00; I2 = 85.65%, DF = 18, H2 =6.96, Q - test = 82.72 and P - value 0.00). Individual study prevalence estimates ranged from 1.40% to 9.1% with the overall random pooled prevalence of 3% (95% CI: 0.02, 0.03). Studies weighted approximately equal with weights on individual studies ranging from 2.94% to 6.19% due to high heterogeneity between studies.

Subgroup Meta-Analysis

Subgroup Analysis by study Regions: Subgroup analyses were done for study Regions (Oromia, Somali Oromia and Somali, Amhara and Tigray regions of Ethiopia). Thus, high seroprevalence was observed in Oromia region 3% (95% CI: 0.02 - 0.05), whereas the same prevalence was observed in both Somali and Tigray region 2% (95% CI: 0.01– 0.02) and the lowest prevalence was observed in Amhara region 1% (95% CI: 0.00– 0.01).

Subgroup Analysis by publication year category: Subgroup analyses were done by articles publication year category. Thus, the same seroprevalence was observed the publication year category from 2011– 2014 and 2015 – 2018 with the prevalence 3 % (95% CI: 0.00 - 0.06 and 0.01 - 0.04) and the publication year category from 2019 - 2022 with prevalence of 2% (95% CI: 0.01 - 0.03) respectively.

Subgroup Analysis by Laboratory techniques: Subgroup analyses were done for laboratory techniques (RBPT, CFT and ELISA). Thus, high seroprevalence was observed in ELISA6% (95% CI: 0.03 - 0.09) followed by CFT and RBPT with both the same seroprevalence of 2% (95% CI: 0.00 - 0.02 and 0.00 – 0.03) respectively.

Subgroup analysis by sample size: Subgroup analyses were done for sample size which has been categorized into three parts like <300, 300 - 600 and >600. Thus, high seroprevalence was observed in both sample size category of <300 and 300 - 600 with the seroprevalence of 3% (95% CI: 0.00 - 0.06 and 0.01 - 0.05), whereas the least prevalence was observed in sample size category of >600 with 2% (95% CI: 0.01– 0.03) respectively.

Publication Bias

Funnel plot for visualizing publication bias: We assessed publication bias and small study effects by funnel plot observation and Egger’s test for small study effects. The funnel plot that visually observed there were asymmetry in which the result of effect estimates against its standard error showed that there was some evidence of publication bias and small study effect on studies reporting the seroprevalence of brucellosis in small ruminant in Ethiopia.

Egger test detecting publication bias: From Egger’s test statistics result there was publication bias and small study effect since the estimated bias coefficient 4.729 with standard error 0.666 and p - value 0.001.

Meta-Regression

Meta-regression analysis was done for each variable included in the study separately. The variables coded as categorical variables and those variables included were study regions, publication year, study year, laboratory techniques and sample size were employed. Those variables with p-values <0.05 were used in the multivariable Meta regression analysis. Only laboratory techniques had significant value and retained in the final multivariable Meta regression analysis.

Discussion

Brucellosis induces considerable human suffering and huge economic losses in animals [8,34]. It has a significant public health implication for a pastoral community in consequence of lifestyles, feeding habits, close contact with animals, low awareness, and poor hygienic conditions which favors infection [7]. Also, it can generally cause significant loss of productivity through abortion, prolonged calving, kidding, or lambing interval, low herd fertility, and comparatively low milk production in farm animals.