This review was based on the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (34). The review was registered by the International prospective register of systematic reviews with identification CRD42021287540.

Both published and unpublished studies were searched from MEDLINE/PubMed, CINAHL (EBSCO), Embase, Cochrane Library, SCOPUS, Web of Sciences databases, and Google Scholar by using a combination of Boolean logic operators (AND, OR, NOT), Medical Subject Headings (MeSH) as well as keywords. An additional search was also conducted using Google scholar and Mednar. The search strategy includes: “Recto-vaginal colonization” [Title/Abstract] or “Vertical transmissions'“[Title/Abstract] or “Antimicrobial susceptibility””[Title/Abstract] ‘AND “Ethiopia” [MeSH Terms]), (“Recto-vaginal colonization, vertical transmission and antimicrobial susceptibility” [exactly on title] AND “Ethiopia” [any field]), (“Recto-vaginal colonization, vertical transmission and antimicrobial susceptibility ”[all field] AND “Ethiopia” [all filed]) (Supplementary File 1). In addition, the investigators searched manually for gray literature and other relevant data sources such as unpublished thesis and articles with planned dates of coverage.

All observational studies conducted in Ethiopia, which reported the prevalence of recto-vaginal colonization, and/or vertical transmission, and/or antimicrobial susceptibility of GBS among pregnant women, were eligible. The eligibility criteria also included English language articles and facility-based studies conducted from 12 December 2012 to October 2021. Articles, theses, and articles fulfilling the aforementioned inclusion criteria were considered for the systematic review and meta-analysis.

All the searched and found articles were exported to the EndNote X8 citation manager, and duplicates were deleted. These were then screened carefully by reading the titles and abstracts. The review authors (AE, TG, AD, and MA) independently screened the titles and abstracts of the identified articles using the eligibility criteria. Full-text articles were further evaluated for eligibility, as needed. Any disagreement among the authors was resolved via discussion and critical reflection. The overall study selection process is presented using the PRISMA statement flow diagram (35) (Figure 1).

The review authors (AE, TG, AD, and MA) independently extracted the data. A pre-defined Microsoft excel 2016 format was used to extract the data from the included articles using author(s), publication year, region of study, study setting, study design, sample size, data collection technique, and the primary outcome of interest (Table 1). The accuracy of the data extraction was verified by comparing the results of the independently extracted data.

The major outcome of interest of this review was the pooled proportion of GBS colonization of pregnant women, vertical transmission of GBS, and antibiotic resistance profiles reported from different studies across Ethiopia. Sub-group analysis was done based on Ethiopia's current political-geographic regions. Measurement for recto-vaginal colonization was made using swabs collected and inoculated into 1 ml Todd Hewitt Broth supplemented with gentamicin and nalidixic acid to avoid the growth of contaminants. After 18–24 h of incubation, all broths were sub-cultured on 5% sheep blood agar for isolation of GBS. The proportion of susceptibility of GBS to the 9 different antibiotics was calculated by dividing the numbers of susceptible isolates by the total number of GBS isolated from pregnant women.

The authors critically evaluated the risk of bias from individual studies using the Joanna Briggs Institute Quality Assessment Tool for observational studies. To minimize the risk of bias, a comprehensive search (electronic/database search and manual search) for published, unpublished, institutional, or community-based studies was carried. Cooperative work among the authors was also considered seriously to reduce bias. All worked together in setting a schedule for the selection of articles based on the clear objectives and eligibility criteria, in deciding the quality of the articles, in regularly evaluating the review process, and also extracting and compiling the data.

The methodological reputability and quality of the findings of the included studies were critically evaluated using the quality assessment tool for observational studies (cross-sectional, case-control, and cohort studies) developed by the Joanna Briggs Institute (JBI) (52) (Supplementary File 3). The two groups of the authors, Group 1 (AD and MA) and Group 2 (AE, KS, and TG), independently evaluated the quality of the studies. The mean score of the two groups was taken for a final decision. The differences in the inclusion of the studies were resolved by consensus. The included studies were evaluated against each indicator of the tool and categorized as high-, moderate-, and low-quality. A high-quality score is above 80%, moderate-quality is 60–80%, and low-quality is below 60%. Studies with a score greater than or equal to 60% were included. This critical appraisal was conducted in order to assess the internal validity (systematic error) and external validity (generalizability) of the studies and to reduce the risk of biases.

Data synthesis and statistical analysis were conducted using STATA version 14.1 software. The random-effect model of analysis was adopted as a method of meta-analysis because it reduced the heterogeneity of included studies. A meta-analysis of observational studies was carried out based on the recommendations of the I² statistic described by Higgins et al. (I² >75/100% suggesting considerable heterogeneity). The p-value of < 0.05 for I² statistics was used to determine the presence of heterogeneity. Similarly, low, moderate, and high heterogeneity were assigned to I² test statistics results of 25, 50, and 75%, respectively. The results were reported using the pooled prevalence with a 95% confidence interval (CI) and forest plot.

Also, sub-group analyses were conducted using publication years and regions in Ethiopia to identify the sources of heterogeneity among the studies. The investigators checked for potential publication bias using visual inspection of a funnel plot (53) and Egger's regression test. Publication bias was assumed for p > 0.10. The results of the review were reported according to the PRISMA guidelines. The findings of the included studies were first presented using a narrative synthesis and secondly by a meta-analysis chart.

A total of 201 article were identified through the electronic databases and other relevant sources. From all identified studies, 31 articles were removed due to duplication while 170 studies were reserved for further screening. Of these, 137 were excluded after being screened according to titles and abstracts. Of the 33 remaining articles, 14 studies were excluded due to studies did not present the outcome of interest and other relevant issues related to outcome variables. Finally, 19 studies that fulfilled the eligibility criteria were included for the systematic review and meta-analysis.

The review included studies from regions in Ethiopia, though the majority of the studies were from Addis Ababa and Oromia. Eighteen cross-sectional and one case-control studies (i.e., reported magnitude and total at-risk population) were included in the meta-analysis (23–25, 36, 37, 39, 40, 42–51, 54, 55). The sample size of the included studies ranged from 126 to 1,688. A total of 5,924 study samples were included, among which 927 pregnant women were with recto-vaginal colonization of GBS (Table 1).

The pooled effect size for recto-vaginal colonization of GBS was 15% (95% CI: 11, 19) (Figure 2). Sub-group analysis showed that the highest pooled prevalence was from Addis Ababa city, i.e., 17% (95% CI: 9, 25) (Figure 3).

The overall substantial heterogeneity across the studies on recto-vaginal colonization of GBS was I² = 95.08%, p < 0.001. Sensitivity analysis was performed for the outcome variable to observe a significant change in risk ratio and CI. The meta-analysis result showed that there was no substantial difference in the overall risk ratio during the sequential removal of each study from the analysis.

The Egger's regression test revealed evidence of publication bias among the included studies (p-value = 0.001). Nevertheless, a visual inspection of the funnel plot revealed symmetry (Figure 4).

A meta-regression analysis was conducted with assumptions of statistically significant heterogeneity of I-square test statistics <0.05. However, the meta-regression analysis found no significant variable to indicate there was any heterogeneity. What this means is that there was no statistically significant level covariation among the sample size, as well as across the publication years of the included studies. Therefore, if significantly present, heterogeneity might be explained by other factors not included in this review (Table 2).

The pooled effect size for vertical transmission of GBS was 51% (95% CI: 45, 58). The overall heterogeneity across the studies on vertical transmission of GBS from mother to child was moderate (I² = 57.11%, p < 0.001).

The Egger's regression test revealed that there was no evidence of publication bias among the included studies (p = 0.491). In addition, a visual inspection of the funnel plot also confirmed the absence of asymmetry (Figure 5).

Among the 19 studies included in the analysis, 11 studies reported susceptibility of GBS to penicillin and erythromycin and 10 studies reported susceptibility of GBS to ampicillin and clindamycin. The highest pooled proportion of antibiotic susceptibility to GBS was observed in vancomycin, i.e., 99% (95% CI: 98, 100), followed by ampicillin, i.e., 94% (95% CI: 92, 97) (Table 2). Nonetheless, tetracycline is the lowest susceptible antibiotics, i.e., 23% (95% CI: 9, 36) (Table 3) (38).

The investigators used extensive and comprehensive search strategies from multiple databases. Published as well as unpublished studies and gray literature were included. Studies were evaluated for methodological quality using a standardized tool. Although the literature search was systematic and assessed all related studies within the desired scope, some relevant publications, e.g., publications reported in non-English language and local languages, may have been missed. Studies with abstracts were the only ones included. This could affect the finding's inclusiveness.