Resistance of Staphylococcus aureus to antimicrobial agents in Ethiopia: a meta-analysis

Background Emergence of antimicrobial resistance by Staphylococcus aureus has limited treatment options against its infections. The purpose of this study was to determine the pooled prevalence of resistance to different antimicrobial agents by S. aureus in Ethiopia. Methods Web-based search was conducted in the databases of PubMed, Google Scholar, Hinari, Scopus and the Directory of Open Access Journals (DOAJ) to identify potentially eligible published studies. Required data were extracted and entered into Excel spread sheet. Statistical analyses were performed using Stata version 13.0. The metaprop Stata command was used to pool prevalence values. Twenty-one separate meta-analysis were done to estimate the pooled prevalence of the resistance of S. aureus to twenty-one different antimicrobial agents. Heterogeneity among the studies was assessed using the I2 statistic and chi-square test. Publication bias was assessed using Egger’s test. Because of significant heterogeneity amongst the studies, the random effects model was used to pool prevalence values. Results The electronic database search yielded 1317 studies among which 45 studies met our inclusion criteria. Our analyses demonstrated very high level of resistance to amoxicillin (77% [95% confidence interval (CI): 68%, 0.85%]), penicillin (76% [95% CI: 67%, 84%]), ampicillin (75% [95% CI: 65%, 85%]), tetracycline (62% [95% CI: 55%, 68%]), methicillin (47% [95% CI: 33%, 61%]), cotrimoxaziole (47% [95% CI: 40%, 55%]), doxycycline (43% [95% CI: 26%, 60%]), and erythromycin (41% [95% CI: 29%, 54%]). Relatively low prevalence of resistance was observed with kanamycin (14% [95% CI: 5%, 25%]) and ciprofloxacin (19% [95% CI: 13%, 26%]). The resistance level to vancomycin is 11% 995% CI: (4%, 20%). High heterogeneity was observed for each of the meta-analysis performed (I2 ranging from 79.36% to 95.93%; all p-values ≤0.01). Eggers’ test did not show a significant publication bias for all antimicrobial agents except for erythromycin and ampicillin. Conclusions S. aureus in Ethiopia has gotten notoriously resistant to almost to all of antimicrobial agents in use including, penicillin, cephalosporins, tetracyclines, chloramphenicol, methicillin, vancomycin and sulphonamides. The resistance level to vancomycin is bothersome and requires a due attention. Continued and multidimensional efforts of antimicrobial stewardship program promoting rational use of antibiotics, infection prevention and containment of AMR are urgently needed. Electronic supplementary material The online version of this article (doi:10.1186/s13756-017-0243-7) contains supplementary material, which is available to authorized users.


Background
Staphylococcus aureus (S. aureus) infection is a major cause of skin, soft tissue, respiratory, bone, joint, and cardiovascular disorders [1]. S. aureus remains a versatile and dangerous pathogen in humans. The frequencies of both community-acquired and hospital-acquired staphylococcal infections have increased steadily. Treatment of these infections has become more difficult because of the emergence of multidrug-resistant strains [2].
Various mechanisms are responsible for S. aureus antimicrobial resistance (AMR). Penicillin is inactivated by β-lactamase. AMR to methicillin confers resistance to all β-lactamase-resistant penicillin's and cephalosporins which require the presence of the mec gene that encodes penicillin-binding protein [3]. The enterococcal plasmidbearing gene for resistance to vancomycin has been transferred by conjugation to S. aureus in vitro [4]. Both increased cell-wall synthesis and alterations in the cell wall that prevent vancomycin from reaching sites of cellwall synthesis have been suggested as mechanisms [4]. Increase in vancomycin use has led to the emergence of two types of glycopeptide-resistant S. aureus. The first one, designated vancomycin intermediate-resistant S. aureus (VISA), is associated with a thickened and poorly cross-linked cell wall is due to continuous exposure to glycopeptide. The second type, vancomycin-resistant S. aureus (VRSA), is due to acquisition from Enterococcus species of the vanA operon resulting in high-level resistance and is a rare phenomenon [5].
In Ethiopia the first published antimicrobial preliminary report on AMR was published by Plorde et al. in 1970 for different microbial agents [6]. Beginning from that time AMR report were made by different antimicrobial surveillances and studies, it showed rapid rise and spread of resistant strains.
Facilitating more appropriate choices of treatment, minimizing the morbidity and mortality due to resistant infections, and preserving the effectiveness of antimicrobials requires summarization and synthesis of the evidence regarding AMR in a country. Appropriately summarized and synthesized evidence is mandatory for updating national treatment guidelines. To our knowledge, no previous meta-analysis or systematic review has been conducted on S. aureus AMR to all antimicrobial commonly in use in Ethiopia. The purpose of this study was, therefore, to determine pooled prevalence of S. aureus resistance to common antimicrobial agents in Ethiopia based on the best available studies.

Study design
This study did a meta-analysis of prevalence of S. aures resistance to different antimicrobial agents in Ethiopia using the best available studies.

Literature search strategy
Web-based search using PubMed, Google Scholar, Hinari, Scopus and the Directory of Open Access Journals (DOAJ) was conducted in June 2016. Google search was used for unpublished works and government documents. Two of the authors (SD and SF) independently searched for relevant studies to be included in this meta-analysis. The PubMed search was carried out via the EndNote software. Relevant search results from Google scholar, Embase, Scopus and the DOAJ were individually downloaded and manually entered into EndNote. The reference lists of the identified studies were used to identify other relevant studies.
The search was done using various key words: Staphylococcus, antimicrobial resistance, antibiotic resistance, drug resistance, drug susceptibility, antibacterial resistance, Ethiopia. These key terms were used in various combinations using Boolean search technique. We did not limit the search by year or language of publication.

Study selection procedures and criteria
Study selection was performed in two stages independently by two of the authors (SD and SF). First, the titles and abstracts of all retrieved articles were reviewed and then grouped as "eligible for inclusion" if they did address the study question and "ineligible for inclusion" if they did not. Second, articles which were grouped under "eligible for inclusion" were reviewed in full detail for decision.
All available studies and data were included based on the following predefined inclusion criteria. 1) Studies that were original journal articles, short communications, or unpublished works; 2) Studies that did the antimicrobial susceptibility test according to the criteria of the Clinical Laboratory Standards Institute (CLSI) and defined antimicrobial resistance range according to CLSI manual [7], 3) Studies which used human infection sample.
Studies that 1) were duplicates, 2) were based on small number of isolates (1-10), 3) were conducted on nonhuman samples like on foods, food handlers' belongings, health workers belongings or health workers carriage and 4) which were based on non-infectious carriage were excluded from this meta-analysis.

Data extraction
Required data were extracted from eligible studies using Excel spreadsheet format prepared for this purpose by AA and SD. The data extracted from eligible studies include name of author(s), year of publication, place where the study was conducted, study design, total number of S. aureus isolate tested in the study, number of resistant S. aureus isolates, and isolate source. If the proportion of drug sensitive isolates (q) was reported, the number of resistant isolates was calculated by multiplying the number of isolates (n) by one minus the proportion of drug sensitive isolates (1-q) and if the proportion of drug resistant isolates was given the number of resistant isolates was found by multiplying the proportion (p) with total number of isolates (n).

Statistical analysis and reporting
Statistical analyses were performed using Stata version 13.0 (Statacorp, LP, college station, TX). The prevalence values from the different studies were pooled using the metaprop command in Stata [8]. We did twenty-one separate meta-analyses to estimate the pooled prevalence of the resistance of S. aureus to twenty one different antimicrobial agents. The number of studies included in each of the meta-analyses ranged from 4 to 39. Heterogeneity amongst the studies was assessed using the I 2 statistic. Because of significant heterogeneity amongst the studies the random-effects model (REM) was used to estimate the pooled prevalence and 95% CIs using the DerSimonian and Laird method [9]. The Freeman-Tukey double arcsine transformation was used so that studies reporting proportions near or at 0 and 1 would not be excluded from the meta-analysis. The possible presence of publication bias was checked using Egger's test [10].
For studies that appeared to report unusually higher prevalence of resistance compared to others, we did sensitivity analysis after dropping the study which we suspected of reporting a higher-than-usual result. If the point estimate of pooled prevalence after dropping a study lies within the 95% CI of the overall pooled estimate for all studies combined, we considered the given study as having non-significant influence on the overall pooled estimate. Otherwise, the study was considered as having significantly influencing the overall estimate.
Results of the current meta-analysis are reported as per the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guideline. The PRISMA checklist was used to ensure inclusion of relevant information (the filled checklist is included as Additional file 1: S1) [11].

Publication bias and heterogeneity
Evidence of high heterogeneity was observed for each of the meta-analyses performed (I 2 ranging from 79.36% to 95.93%; all p-values ≤ 0.01). Eggers' test did not suggest any significant publication bias except for erythromycin and ampicillin (see Additional file 2: S2).

Prevalence of S. aureus resistance to different antimicrobial agents
Summary of the pooled prevalence of S. aureus AMR prevalence for twenty-one different antimicrobial agents and the number of studies included in the meta-analysis for each agent are presented in Table 2. Prevalence of S. aureus resistance for each antimicrobial agent based on pharmacological classification of the agents is given below. As new anti-MRSA agents such as linezolid, daptomycin, tigecycline, telavancin and ceftaroline are rarely available in Ethiopia and no published studies available on resistance to this agents, our results do not cover such agents.

Prevalence of resistance to glycopeptides (vancomycin)
Nineteen studies were included for meta-analysis of vancomycin resistance prevalence. The pooled prevalence for S. aureus resistance to vancomycin in Ethiopia is 11% (95% confidence interval [CI]: 4%, 20%). The forest plot for vancomycin resistance is presented in Fig. 2. The results of sensitivity analysis after exclusion of the two studies that appeared to report outlier prevalence values separately and both together showed nonsignificant influence of the two studies on the overall estimate. The pooled prevalence of vancomycin resistance when Guta

Prevalence of resistance to floroquinolones
Two antimicrobial agents were tested from the floroquinolones: ciprofloxacin and norfloxacilin. Thirty one studies were used to estimate the prevalence of ciprofloxacin resistance and eleven studies were included for the estimation of norfloxacilin resistance. The pooled prevalence of S. aureus resistance to ciprofloxacin was 19% (95% CI: 13%, 26%) and to norfloxacillin 25% (95% CI: 14%, 38%). The forest plot for ciprofloxacin resistance is presented in Fig. 6 while the forest plot for norfloxacilin included as Additional file 9: S9.

Prevalence of resistance to antimetabolites
Thirty five studies were included for estimation of pooled prevalence of S. aureus resistance to sulphametaxozoletrimethoprim and found to be 47% (95% CI: 40%, 55%). The forest plot for sulphametaxozole-trimethoprim resistance is presented in Fig. 8. Comparison of the prevalence of S. aures resistance to different antimicrobial agents addressed by this metaanalysis is given in Fig. 9. It is found that the magnitude of S.aureus resistance to the different antimicrobial agents ranges from 11% to vancomycin to 77% to amoxicillin. Accordingly, invitro antimicrobial effectiveness in decreasing order believed to be vancomycin, kanamycin, ciprofloxacilin, amikacin, clindamycin, amoxacilin-clavulanic acid, cephalothin, carbencilin, ceftriaxone, cefoxitine, chloramphenicol, erythromycin, doxycycline, methicillin, cotrimoxazole, tetracycline, ampicillin, pencilin, and amoxacilin.

Discussion
In this meta-analysis, we estimated the pooled prevalence of S. aureus resistance to 21 different antimicrobial agents commonly used in Ethiopia. Generally 45 studies were included for the meta-analysis, however the number of studies included in each meta-analyses ranged from 4 to 39. Overall, the 45 studies provided evidence regarding the level of S. aureus resistance to different antimicrobial agents based on 4530 isolates. It was found that S. aureus resistance to commonly available antimicrobial agents in Ethiopia was alarmingly high ranging from 11% to vancomycin to 77% to amoxicillin.
The pooled estimate of the prevalence of S. aureus resistance particularly to methicillin (MRSA) in Ethiopia is similar to 2014 global surveillance reports of the World Health Organization (WHO) 2014 [88], which showed MRSA prevalence between 33% to 95% in Africa. The pooled prevalence of MRSA in Ethiopia 47% (95% CI: 33%-61%) is within the range of the global WHO report for Africa.
The pooled estimate in study for MRSA is in agreement with the pooled estimate of community acquired-MRSA prevalence in Asia, Europe, and North America which ranges from 23.1% to 47.4% [89]. However, the pooled estimate 47% (95% CI: 33%-61%) MRSA prevalence in Ethiopia is higher than pooled estimate of community acquired MRSA prevalence 30.2% based on 27 retrospective studies and 37.3% based on 5 prospective studies [90]. The higher prevalence in our study may be due to the inclusion of both community acquired and nosocomial infection in the original studies. Nosocomial infection are believed to have higher rate of resistance due to larger exposure rate to antimicrobial agents. Increasing resistance to antimicrobial agents in hospitals is Fig. 6 Forest plot of the prevalence of S. aureus resistance to ciprofloxacin caused by transmission of resistant strains within hospitals by cross colonization of patients via hands of healthcare staff and direct patient to patient contact and subsequent spread [91].
Global pattern of AMR shows variation among different geographic, socioeconomic strata and among studies [49,88,92]. Variation may be to differences in time, place, design, and population involved in the study. This may be due to healthcare facilities conditions like implementation and monitoring of infection prevention policies and rational antibiotic usage which varies in different facilities. The most important reason is due to character of the study. Studies are conducted within a specified time and locality. It is reasonable to assume population under study might be infected by the same strains of agent at specified period of time and location. This could be a good reason why heterogeneity tests showed significant variability (p-value ≤0.01) among studies included in this meta-analysis for 2 l antimicrobial agents.
S. aureus acquires resistance by various mechanisms: formation of alternative pathways for sulphonamides [93,94], production of β-lactamase to β-lactam-sensitive antibiotics, increased efflux to tetracycline [95,96], presence of acetyltransferase to chloramphenicol, decrease in accumulation to macrolide antibiotics [97], aminoglycosidemodifying enzymes production to aminoglycosides, altered topoisomerase IV and DNA gyrase expression for fluoroquinolones, and expression of mec gene altering penicillin binding protein to β-lactam antibiotics [98]. Since the AMR for β-lactam sensitive β-lactam antibiotics is very high, it can be speculated that most strains of S. aureus found in Ethiopia produce the β-lactamase enzyme. However, there is no molecular study conducted to identify the type of resistant strains and mechanism responsible for resistance in Ethiopia.
Lower rate of resistance was seen with β-lactamaseresistant antibiotics (amoxicillin-clavulanic acid, methicillin, ceftriaxone, cefoxtine, and cephalothin) compared to β-lactamase-sensitive penicillins. Unlike β-lactamase sensitive penicillin's, resistance to carbeniciln is significantly lower. The lower rate of resistance observed with carbencilin and clindamycin may be due to their infrequent use in Ethiopia [99].
Resistance to methicillin confers resistance to all βlactamase-resistant penicillins and cephalosporins. This high level of resistance requires the presence of the mec gene that encodes penicillin-binding protein [98]. The implication of high prevalence of MRSA for suspected or verified S. aureus infections such as common skin and wound infections and surgical prophylaxis is that there is a need for better alternatives drugs. Alternative drugs needed to treat or prevent S. aureus infections are more expensive and, because of their adverse effects, monitoring during treatment is advisable which increases the costs even further.
The prevalence of resistance S. aureus to vancomycin 11% 995% CI: (4%, 20%) in this study is bothersome and higher compared to global prevalence estimate [100]. The prevalence of VISA was 2.05% before, 2.63% in 2006-2009, and 7.93% in 2010-2014. Vancomycin resistance is erasing all possible treatment options in Ethiopia for MRSA. The higher prevalence of vancomycin in Ethiopia compared to global estimate may be due to larger and irrational use of antimicrobial agents in Ethiopia, The prevalence estimates of glycopeptides/vancomycin resistance from Guta et al., and Desalegn et al., were unusually high, however sensitivity analysis showed nonsignificant influence on the overall pooled prevalence estimate. The prevalence estimates from Guta et al. and Desalegn et al. were unusually high, however sensitivity analysis showed non-significant influence on the overall pooled prevalence estimate. Larger exposure probability to resistant strains due to larger use of vancomycin in hospital settings might have resulted in a relatively higher prevalence of vancomycin resistance in the two studies [73,77].
In four of the twenty studies (published in 2014 and after) [48,51,73,77], the prevalence of S. aureus resistance to vancomycin is higher than 40%. In contrast, in studies published before 2014 the prevalence of S. aureus resistance to vancomycin in Ethiopia is much lower (0% to 16%). This may indicate a rapid rise and spread of vancomycin resistant S. aureus strains in Ethiopia as the rate of vancomycin use and exposure in Ethiopia increases. This calls for inclusion for effective new anti-MRSA antimicrobial agents for treatment of staphylococcal infections in the national medicine list and effective antimicrobial stewardship programs for prevention and containment of antimicrobial resistance.
Staphylococcal infection in Ethiopia can be better treated by vancomycin, floroquinolones, and aminoglycosides based on the finding of our invitro finding. However, clinical effectiveness study had not yet proved it. Resistance to vancomycin, the only choice for MRSA in Ethiopia, is of a great concern. It is bothersome due to lack of alternative agents in Ethiopia for the treatment of S. aureus infections. Making things worse, alternative new anti-MRSA agents (like linezolid, daptomycin, tigecycline, telavancin, and ceftaroline are rarely available in Ethiopia for treatment of vancomycin resistant S. aureus.
Many factors contribute to AMR. First, lack of infection prevention contributes to recurrent infection then to spread of resistant strains. Second, misuse of antimicrobials from prescription-dispensing-to patient use [101]. In Ethiopia, it is a common practice that antibiotics can be purchased without prescription, which leads to misuse of antibiotics by the public [102]. Third factor could be misuse of antibiotics by health professionals and non-standardized practice [101]. The fourth factor could be poor hospital hygienic conditions [103]. A last contributing factor could be lack of routine antimicrobial susceptibility testing which diverts to empiric therapy [49]. In line to strategies for prevention and containment of S. aureus there is a need for innovative way of halting AMR. Combination therapy and availability of new anti-MRSA agents will play vital role in fighting against AMR to S. aureus.
However, interpretation of the findings of this metaanalysis requires considering the limitations thereof. The limitations arise from the inherent characteristics of the included individual studies. First, this is invitro antimicrobial resistance testing and its direct translation to clinical effectiveness requires caution. Second, many studies involved very limited localities and were done mainly in teaching hospitals in bigger cities where patients with advanced, severe stages, recurrent infections are treated. Hence, the resistance level could have overestimated.

Conclusions
This meta-analysis demonstrates that S. aureus has gotten alarmingly resistant to many of common antimicrobials used in Ethiopia. It is highly resistant to penicillin, cephalosporin, tetracyclines, chloramphenicol, methicillin, sulphonamides, and vancomycin. Resistance to vancomycin is of a great concern and bothersome due to unavailability of treatment options for S. aureus infections in Ethiopia.
Continued and multidimensional efforts of antimicrobial stewardship programme promoting rational use of antimicrobials, infection prevention and containment of AMR are urgently needed. It is deemed necessary to include new anti-MRSA agents in national medicine list to Fig. 9 Comparison of the prevalence of S. aureus resistance to different antimicrobial agents in Ethiopia