The literature review was conducted from January 12–20, 2024, using the PRISMA checklist (44). This systematic evaluation of Streptococcus spp. in bovine mastitis and antimicrobial resistance involved seven key stages: suitability assessment, information sources, search strategy, outcome variables, data extraction, study quality evaluation, and data synthesis with statistical analysis. A comprehensive search was conducted using several databases, including PMC, SCOPUS, PubMed, HINARI, Web of Science, Google, and Google Scholar. Study selection was performed independently by two authors (M.D.F and A.S.M). The research question addressed the proportion, prevalence, and antimicrobial resistance of Streptococcus spp. causing bovine mastitis in Ethiopia. Meanwhile, the key words used were Streptococcus spp. OR Streptococcus infection, epidemiology OR prevalence OR infection proportion, cattle OR bovine OR animals, resistance proportion OR antimicrobial resistance AND (mastitis) AND Ethiopia. A restriction was placed on the language of publication as English. All identified studies were imported into EndNote 20 software to remove duplicates and citations of the references.

The meta-analysis was conducted in Ethiopia, which is located in the Horn of Africa between 3°00′–15°00′ N latitude and 32°30′–48°00′ E longitude. Covering 1.04 million square kilometers, Ethiopia is Africa’s second most populous country, with 123 million people. The country supports significant agricultural production, with approximately 70 million cattle, 52.5 million sheep, and 42.9 million goats (1). Its diverse topography includes highlands above 2,300 m, a. s.l. proportion transition zone between 1,500 and 2,300 m, and lowlands below 1,500 m.

To avoid reviewer bias, the search was carried out by three subject matter experts in veterinary clinical medicine, veterinary pharmacy, and veterinary public health and epidemiology. All of the primary descriptive studies that had been published in English and that showed the presence of Streptococcus spp. in dairy cattle were included in this meta-analysis. Articles that provided a precise estimate of the percentage of each bacterial isolate were required to meet the inclusion criteria. The research needed to come from observational studies, and the cause of bacterial mastitis in cows had to be determined from clinical or subclinical cases. The study animals were limited to domestic cattle, or cows, which are commonly raised for their milk. It was necessary to gather samples from animals that had not been exposed to an experimental infection. The geographical location of the bacterial isolates had to be Ethiopia, and the isolates were identified at least down to the genus level. The overall quantity of Streptococcus spp. investigated and the quantity of isolates that were resistant or sensitive may or may not have been disclosed. In cases where the scientific papers presented findings from identical sample times and methodologies but with varying Streptococcus spp., each occurrence was documented as an individual investigation within our database. Consequently, one scientific article could encompass multiple studies.

Studies focusing on milk from camels and other non-cattle species were excluded from the analysis. Additionally, studies that failed to provide clear and comprehensive estimates of the proportion of each bacterial species in relation to the affected host were not included. Review articles, duplicate studies, publications containing only abstracts, qualitative research, and studies based solely on KAP (knowledge, attitudes, and practices) questionnaires, book chapters, case reports, editorials, short communications, opinion pieces, and studies without original data were excluded. Furthermore, intervention studies that did not include baseline data on the association between animal exposure and disease were excluded from the meta-analysis.

In this review, we have two outcome variables: first, the pooled proportion/magnitude of Streptococcus spp. among the bacteria causing mastitis, and second, the antimicrobial resistance (AMR) profile of Streptococcus spp. In the first case, the proportion of mastitis-associated Streptococcus infection was estimated by considering the number of Streptococcus spp. isolates in the milk sample relative to the total number of bacterial isolates. In the second case, the resistance proportion of mastitis-associated Streptococcus isolates was calculated by determining the number of AMR isolates of Streptococcus spp. relative to the total number of isolates.

Two investigators (B.A.M and M.G.) extracted the data independently. Data extraction, both quantitative and qualitative, was performed via two tables and an Excel spread sheet from the included studies. The primary author’s name, the year the work was published, the region, the total number of bacterial isolates, the number of isolates of Streptococcus spp. (the main outcome of interest), diagnostic procedures, data collection, and ethical considerations were included in the extracted components. Information was extracted from each article and entered into a database, including the antimicrobial susceptibility testing methodology (disc diffusion or minimum inhibitory concentration (MIC) estimation), MIC methodology (broth dilution method, agar dilution method, or other), number of Streptococcus isolates analyzed, number of resistant isolates, and type of mastitis (clinical or subclinical). Conflicts were settled through discussion and advice from a third author.

To confirm the review’s methodological quality, a quality assessment was carried out by two independent authors (Y.D and A.S.M). The AXIS quality tool (45) was used to evaluate the included studies’ quality. The study design, sample size justification, sample representativeness, target population, use of validated measures, diagnosis of statistical methods, sample selection, sample frame, and discussion of nonresponse bias, funding reporting, and conflicts of interest are just a few of the items included in this quality assessment tool.

R software was used to perform a meta-analysis via the “metaprop” function from the “meta” package version 4.1. 3–0 (46) and “metafor” in R Studio (47). The pooled proportion and 95% confidence intervals (CIs) were estimated via a random effects model based on the restricted maximum likelihood (REML) method, which computes within-and between-study variability. It was applied to the resistance proportion, heterogeneity, overall effect size, and weight of each study in the meta-analysis. Furthermore, the pooled prevalence and resistance of bovine mastitis associated with streptococcal infection are illustrated via graphs and tables. The resulting variable is binary (i.e., the only parameter available to measure effect size for single groups (e.g., Streptococcus positive or negative); resistant/sensitive to the antimicrobial agent) was the raw proportion with 95% confidence intervals (48). In accordance with (49), a logistic-normal random-effect regression model was used to estimate pooled proportions via logit transformation, whereas a mixed effect logistic regression model was employed for subgroup analysis.

The sources of heterogeneity were evaluated via the Cochran’s Q test (reported as the p value), τ² (variance between studies), and the inverse variance index (I²), which indicates the proportion of total variation observed between studies as opposed to heterogeneity as a result of chance. According to (50), the I² index was calculated to correspond to values of 25, 50, and 75%, respectively, and was estimated to represent low, moderate and high heterogeneity. When the Q test produced a p value of less than 0.10 and the I² value was greater than 50%, heterogeneity was deemed statistically significant. A forest plot was used to evaluate the level of heterogeneity among the studies. Each study’s weights, effect sizes, and 95% confidence intervals are displayed in a forest plot diagram.

A subgroup analysis of the proportion of Streptococcus spp. in bovine mastitis was carried out on the basis of the study year, study location or region, species of bacteria, and level of mastitis (clinical and subclinical) to ascertain specific between-study variability.

Publication bias is typically assessed via Egger’s test, Begg’s rank test, and a funnel plot, which allows for the visual assessment of asymmetry (48). Therefore, Egger’s regression test and funnel plot diagrams were used to evaluate publication bias.

A sensitivity analysis of the studies was performed to evaluate the effect of each individual study (51). The results revealed that the studies were the prime determinants of the pooled result.

A comprehensive search was performed in several databases, yielding 4,151 articles, along with 6 additional records identified from other sources (Figure 1). After removing duplicates, 3,901 records remained. Of these, 2,896 records were screened, and 1,891 were excluded based on their title and/or abstract. A total of 1,005 articles were assessed for full-text eligibility, with 984 excluded for various reasons. In the end, twenty five studies (n = 25) were included in the meta-analysis.

The characteristics of studies on Streptococcus spp. isolates are detailed as follows. The study subjects were lactating dairy cattle. A total of 25 articles were analyzed for the proportion of Streptococcus spp. associated with mastitis, and 54 articles were reviewed on the basis of species isolates. We identified the following isolates: Str. agalactiae (n = 18; 33%), Str. dysgalactiae (n = 13; 24%), Str. uberis (n = 11; 20%), Str. faecalis (n = 4; 7%), and unidentified Streptococcus spp. (n = 8; 14.8%). The studies included were conducted in various regions of Ethiopia between 2008 and 2024, predominantly in the southern (Oromia) and central regions. Diagnostic methods included CMT and bacterial culture, following procedures described by (52). The regional distributions of studies were as follows: Oromia (11 studies, 44%), Addis Ababa (7 studies, 27%), SNNPR (2 studies, 8%), Amhara (4 studies, 16%), and Tigray (2 studies, 8%). The minimum sample size was 79 cattle, and the maximum sample size was 1,019 (53). In this review, 7,073 dairy cows were evaluated. The prevalence of bovine mastitis associated with Streptococcus infection ranges from 1 to 26% (24, 54), with Str. uberis and Str. agalactiae being the most prevalent. Of the 25 studies, 18 focused on subclinical mastitis, and 15 focused on clinical mastitis. Three studies (55–57) addressed only subclinical mastitis. The detailed characteristics of the included studies are presented in Table 1.

In this review, a spectrum of studies was evaluated with respect to their quality, which ranged from low to medium proportion. None of the 25 quantitative studies met the criteria set by the AXIS tool, which encompasses details pertaining to risk factors and outcome variables. A majority of the articles, specifically 22 out of 25 (88%), utilized the simple random method procedure outlined by (58). Moreover, 20 studies (80%) successfully obtained a sample frame from a suitable population that closely resembled the target or reference population being investigated. Among the total number of studies, 17 (approximately 68%) fulfilled the requirements for six out of the 20 questions, namely, aims/objectives, definition of target/reference population, internal consistency of results, authors’ justification of the results, sample size justification, analysis of appropriate techniques in the methods and conflicts of interest, and description of the statistical methods used.

The meta-analysis included 25 articles investigating Streptococcus species associated with mastitis. Importantly, some articles were referenced multiple times because of their relevance in similar years but involved investigations of different bacterial strains. The studies included in this analysis showed a substantial level of heterogeneity (I² = 87%, τ² = 0.203; p < 0.01). The estimated pooled proportion of Streptococcus species associated with mastitis among all bacterial isolates was 20% (95% CI: 17–23%; Figure 2). The variability between studies was statistically significant (Q = 386.5, DF = 25, p < 0.001). The funnel plots (Figure 3) and Egger’s regression asymmetry did not suggest the presence of publication bias (p > 0.05).

Because of high degree of heterogeneity, sub analyses were conducted on the basis of the study location or region, study year, degree of mastitis, and type of Streptococcus spp., as shown in Table 2 and Supplementary Figures 1–4. Significant heterogeneity between studies was found in the sub analysis by region-wise (p < 0.001). The subgroup analysis of Streptococcus bacteria associated to bovine mastitis by region revealed that SNNPR had the largest pooled proportion of mastitis-associated Streptococcus isolates (26%), followed by Amhara (24%), Oromia (19%), AA (18%), and Tigray (15%). However, Oromia region had the highest heterogeneity (I² = 91%; p < 0.01) across studies. A sub analysis was conducted regarding the study year (studies grouped into before 2013 and studies after 2013). The magnitude of heterogeneity were I² = 83% and I² = 86%, respectively. The highest sub pooled proportion of Streptococcus isolates associated to mastitis (20%) occurred prior to 2013. The subgroup difference test results (Q = 0.18; DF = 1; p = 0.743) indicated the absence of a statistically significant group effect.

A sub-analysis on the basis of the degree of mastitis also revealed that, with study weights of 40 and 59.5%, the percentage of Streptococcus isolates at the species level was greater in clinical mastitis at 24% (95% CI; 16–34%) than in subclinical bovine mastitis at 18% (95% CI; 14–24%). Both the clinical and subclinical mastitis categories experienced significant study variability across studies (I² = 89%; p < 0.01) and (I² = 87%; p < 0.01), respectively. Five groups were formed from the sub-analysis of the included studies on the basis of the types of Streptococcus species: Str. agalactiae (n = 18), Str. dysgalactiae (n = 13), Str. uberis (n = 11), Str. Faecalis (n = 4) and unknown Streptococcus spp. (n = 8). Several publications did not identify the species level (21, 55, 59, 60). In the sub-analysis of the proportion of Streptococcus associated with mastitis by species type, there were notable differences. According to the subgroup analysis proportion shown in Supplementary Figure 4, the pooled proportion of unidentified spp. was the next highest at 15% (95% CI: 10–22%) and (I² = 87%: τ² = 0.370; p < 0.01) for Str. agalactie. 13% (95% CI: 11–16%) and (I² = 79%) (τ² = 0.146; p < 0.01). In the present meta-analysis, the proportion of Str. dysgalactiae was found to be 6% (95% CI: 5–8%) with low heterogeneity (I² = 49%: τ² = 0.126; p = 0.02).

In all included studies, the Kirby-Baur disc diffusion method was used as an antimicrobial sensitivity test. The present studies on antimicrobial resistance in Streptococcus for the treatment of mastitis in cattle are depicted in Table 3. Only studies accurately proportionally identifying Streptococcus species to the species level were included in the meta-analysis. This approach helped to exclude studies that may have misclassified or grouped different species, which could have had varying resistance profiles.

The highest prevalence of resistant Streptococcus was against penicillin (pool estimate = 52, 95% CI = 38–67%), followed by streptomycin, tetracycline, and ampicillin (pool estimates = 42, 38, and 35%, respectively). Gentamycin and erythromycin presented the lowest overall prevalence of resistance (pool estimates = 16 and 19%, respectively) (Table 3). In general, the I² values were highest for the streptomycin antimicrobials tested (I² = 80%). The I² of amoxicillin, Co-trimazole and gentamycin were equal to zero since they were used in any of the included studies; therefore, there was no variability (Table 4).

Most of the articles describe the frequency of the isolates and the percentage of the resistance isolates however no articles tried to identify the resistance genes. Most of the articles used the convenience method of the sample selection so it may lead to selection biased. This systematic review has several limitations, which must be taken into consideration. First, the review is focused exclusively on a single genus, namely, Streptococcus. Second, few studies were included in the analysis of antimicrobial resistance. Third, no studies were included in some regions. Finally, the protocol was not registered in the PROSPERO database.