Prevalence, antimicrobial susceptibility and genotypic characteristics of Staphylococcus aureus in Tanzania: a systematic review

Data on the prevalence, genotypes and antibiotic resistance patterns of colonizing and infection-associated Staphylococcus aureus (S. aureus) strains both in humans and animals in Tanzania are scarce. Given the wide range of infections caused by S. aureus and the rise of methicillin-resistant S. aureus (MRSA) globally, this review aims at collecting published data on S. aureus bacterium to improve our understanding of its epidemiology in Tanzania. We carried out a systematic review of scientific studies reporting on prevalence, antibiotic resistance and genotyping data pertaining to S. aureus in human and animal infection and colonization. The literature extracted from electronic databases such as PubMed and Google Scholar was screened for eligibility and relevant articles were included. The review is limited to manuscripts published in English language between the years 2010 and 2020. A total of 45 studies conducted in 7 of the 9 administrative zones in Tanzania were reviewed to gather data on S. aureus prevalence in humans and animals. Prevalence in humans ranged from 1 to 60%. Antibiotic resistance patterns of S. aureus isolated from colonized humans showed high resistance rates against co-trimoxazole (46%) and erythromycin (41%) as compared to reports from studies conducted outside Africa. The review suggests an increased MRSA prevalence of up to 26% as compared to 6–16% reported in previous years. Genotypic data reviewed suggested that MRSA predominantly belonged to ST88. The prevalence of S. aureus in animal studies ranged from 33 to 49%, with 4 to 35% of MRSA isolates. Most studies reported low antibiotic resistance levels, with the exception of penicillin (85%) and ampicillin (73%). The prevalence of S. aureus and MRSA in Tanzania is rising, although clear variations between different geographic areas could be observed. Non-susceptibility to commonly prescribed antibiotics in community-associated S. aureus is of concern. Research strategies to ameliorate our knowledge on S. aureus epidemiology should employ regular antibiotic resistance surveillance, antimicrobial stewardship as well as genotypic characterization.


Background
Antimicrobial resistance (AMR) is a global concern estimated to account for approximately 700,000 deaths each year (O'Neill 2016). If no appropriate measures are taken to slowdown the progression of this epidemic it is estimated that by 2050 it will cost the world around 10 million lives per annum (O'Neill 2016). The lack of development of new antimicrobial agents in the pipeline further emphasizes the necessity to reduce dependency on antibiotics by implementing infection control strategies (O'Neill 2016;WHO 2014).
Page 2 of 21 Mzee et al. Bull Natl Res Cent (2021) 45:162 According to the 2014 World Health Organization (WHO) report, most regions stated over 50% bacterial resistance against third-generation antibiotics, particularly methicillin-resistant S. aureus (MRSA) clones in hospitals (WHO 2014). Furthermore in 2017, the WHO listed antibiotic-resistant bacteria including Staphylococcus aureus (S. aureus) as a priority bacterium to guide research, discovery and development of new antibiotics (WHO 2017). The report further recommended special attention to be directed towards the identified resistant bacteria due to their ability to rapidly develop resistance against multiple antibiotic classes hence limiting therapeutic options (WHO 2017).
Staphylococcus aureus is an old bacterium discovered in the eighteenth century, antibiotic resistance in the bacterium against penicillin was described in the 1950's, the penicillin-resistant staphylococci inactivates penicillin function by an enzyme called penicillinase or beta-lactamase which degrades the β-lactam ring in penicillin, thus altering the shape of penicillin and preventing its binding to the penicillin binding proteins (Haddadin et al. 2002). Following penicillin failure in treating S. aureus infections, methicillin was introduced. Methicillin was particularly effective upon its introduction into clinical therapy due to its ability to resist the action of β-lactamase. However, methicillin resistance in S. aureus was also reported only a year after its introduction. Strains of S. aureus resistant to methicillin or oxacillin or other β-lactam compounds are still termed MRSA, only to honour the historic role of methicillin that used to effectively treat staphylococcal infections, and it is to methicillin that the resistance was first described (Jevons 1961). Several terms are used to describe different MRSA strains associated with outbreaks in different settings based on strains' genomic background and level of virulence (Zetola et al. 2005). MRSA strains associated with hospital-acquired infections are commonly abbreviated to HA-MRSA, while strains causing community-associated infection are abbreviated to CA-MRSA and those associated with livestock are abbreviated as LA-MRSA.
Staphylococcus aureus is both a commensal and potentially harmful pathogen in humans and animals. The bacterium can give rise to a variety of infections ranging from mild skin and soft tissue infections to more serious and complex diseases such as pneumonia, septicaemia, infective endocarditis and other deep-seated infections (e.g. osteomyelitis) in humans as well as mastitis and necrotic infections in a variety of animal hosts. Staphylococcus aureus can also colonize the skin and approximately 30% of the human population is found to be transiently colonized if nasal swabs are examined microbiologically.
High levels of MRSA have been reported across Africa, ranging between 43 and 72%, as reported in Cameroon, South Africa and Ethiopia (Founou et al. 2017). The vast majority of clinical S. aureus/MRSA data in Africa is associated with hospital acquired infection (HAI) affecting individuals with healthcare-related risk such as hospitalization, surgery and underlying chronic diseases (David 2010). Information on community-acquired S. aureus/MRSA infections (CAI) causing diseases in people with no healthcare-associated risk is also available on the continent even though to a lesser extent. Moreover, very limited data on livestock acquired S. aureus/ MRSA infections (LAI) whereby S. aureus clones of animal origins colonize or cause infections in humans have been published in Africa (Founou et al. 2017). Further information on the antibiotic-resistant clones occurring in both human and animal are very limited, hindering comprehensive understanding of the epidemiology of the bacterium (Lozano et al. 2016).
Tanzania, a developing country located along the East African coast, housing a population of about 57 million people, of which 36% is engaged in livestock keeping as the major source of livelihood. The potential for humananimal contact is very high, especially in tribes that cohabit with their animals such as Maasai people. The aim of this review is to summarize literature reporting on the prevalence, antibiotic resistance of S. aureus in Tanzania. Furthermore, reported genotypic characterization of S. aureus will be reviewed to provide a more nuanced profile of S. aureus genetic diversity in colonization and infections both in animal and human hosts as well as their possible relationship.

Eligibility criteria
A systematic review of Tanzanian scientific studies reporting on prevalence, antimicrobial resistance and genotypic characterization of S. aureus from human and animal sources published between 2010 and 2020 was performed. Information sought for in the publications included commensal and clinical S. aureus recovered from different infections including invasive and noninvasive. Animal studies reporting on S. aureus recovered from animals and their products were also considered. All studies that recovered less than five S. aureus isolates as well as publications that were written in a language other than English were excluded from this review.

Information sources and search strategies
PubMed database was researched in November 2018 and February 2021 by a librarian using Boolean operators "AND" and "OR" to identify studies fitting our inclusion Page 3 of 21 Mzee et al. Bull Natl Res Cent (2021) 45:162 criteria. The following search terms were used "Staphylococcus aureus OR S. aureus AND antimicrobial susceptibility OR antibiotic resistance AND prevalence AND Tanzania. " Additionally search words such as "Molecular typing AND S. aureus AND antibiotic resistance AND Tanzania" and "Staphylococcus aureus OR S. aureus AND antimicrobial susceptibility OR antibiotic resistance AND prevalence AND Tanzania" were used in Google scholar to identify eligible articles. The bibliographies of all eligible documents were hand-searched for additional publications eligible for review.

Data extraction and appraisal process
The Joana Briggs Institute checklist was applied to appraise and review the quality of each study accessed by two independent reviewers. A data extraction form was designed to capture required information such as author, year of publication, study period, methodology, MRSA identification strategies, S. aureus and MRSA prevalence, antibiotic resistance patterns as well as genotypic information. Two reviewers were involved in the process whereby the first reviewer (TM) extracted the information and the second reviewer (TK) double checked the information to eliminate possible bias. All disagreement raised during the critical appraisal process were resolved through reviewers discussions. All extracted literature was analysed using reference manager ENDNOTEX7. This study followed the standardized scientific writing format of the Preferred Reporting Items for systematic reviews and meta-analyses (PRISMA) guidelines. The study has not taken on any meta-analysis due to the heterogeneity of the studies under review; nevertheless, mean resistance rates and prevalence were calculated to help present the results better.

Scope of the study
This review is limited to prevalence, antibiotic resistance patterns and genetic typing information [i.e. specified resistance and virulence genes, Staphylococcus protein A (spa) typing and multi-locus sequence typing] of S. aureus retrieved from human and animal hosts.

Results
PubMed search resulted in 18 eligible articles followed by an additional 13 articles included from the Google scholar search. Rigorous reference list review supplemented another 14 eligible publications making a total of 45 ( Fig. 1). Data concerning S. aureus prevalence, antimicrobial resistance and genotyping were extracted from studies performed in 7 zones of the 9 administrative zones in Tanzania. Table 1 describes the distribution of the published articles review in their consecutive zones. The majority of the publications reviewed were from regions in the eastern and lake zones (i.e. 40% and 33%, respectively), showing lack of S. aureus researched data in the other parts of the country.
The majority of the publications were reported in the Eastern and Lake Zones. S. aureus data of animal origin have been poorly represented throughout the country.

Antibiotic resistance in human S. aureus isolates
The primary method for establishing S. aureus antibiogram in the reviewed work was done phenotypically using the Kirby-Bauer disc diffusion test along with Clinical Laboratory Standard Institute (CLSI) guidelines. Kirby-Bauer is an antimicrobial susceptibility test based on the size of inhibition zones of microbial growth in a lawn culture around discs impregnated with the antimicrobial drug (Hudzicki 2009).
Twenty-four studies contributed information on antibiotic susceptibility of S. aureus bacteria. MRSA in most cases was identified by resistance against cefoxitin; however, one study used methicillin as their identification disc. As summarized in Table 3, extremely high resistance rates against β-lactams, i.e. penicillin (87-99%) and ampicillin (67-92%), were observed.
Apart from high resistance to the reported β-lactams, trimethoprim/sulphamethoxazole (co-trimoxazole) and erythromycin showed average resistance of 54% and 47%, respectively. Notably individual analysis of antibiotics susceptibility in isolates collected from colonization showed lower rates of resistance compared to S. aureus isolated from clinical infections, as summarized in Table 3.
MRSA detection in all reported reviewed studies had an average prevalence of 21%, nevertheless 7 studies reported on specific resistance patterns of MRSA isolates sighting higher rates against other antibiotics (clindamycin, erythromycin and co-trimoxazole) ranging from 50 to 100% (Joachim et al. 2018100% (Joachim et al. , 2017Geofrey et al. 2015;Moyo et al. 2014;Moremi et al. 2014) apart from the expected β-lactams antibiotic resistance.
Even though most articles reported 100% susceptibility to vancomycin, Geofrey et al. 2015;Kayange et al. 2010;Seni et al. 2019a observed vancomycin resistance of above 10% in their respective studies (not demonstrated in Table 3).

Genotypic characterization of human S. aureus isolates
Five human-related studies were reviewed (summarized in Table 4). MSSA genotypic characterization by Staphylococcal protein A (spa) typing in one of the studies revealed 13 different spa types including one new spa type t10779. The most common spa types were t714 and t148 (associated with ST72) followed by t084 (ST15 and ST18) and t223 (ST 22). Furthermore, spa types t314 (ST121), t084 (ST15 and ST18) and t223 were reported to be shared by human and animal isolates in this study (Katakweba et al. 2016). In a more recent study whereby patients were swabbed for S. aureus carriage on admission, after their hospital stay as well as wound swabs for those who had SSTIs, the study further swabbed HCW attending the patients in question. Taking these groups into consideration 60 S. aureus were characterized by MLST, as in the other studies ST distribution was diverse. Eight STs were detected in the 17 isolates from admission of which ST8 (4/17) and ST5 (4/17) were dominant. Nine STs were detected among 13 acquired isolates typed of which ST152 (3/13) and ST5 (2/13) were predominant. Moreover in the 12 S. aureus SSTI isolates, eight STs were detected predominated by ST152 (3). Subsequently in a study which used the whole genome sequencing and multi-locus sequence typing (MLST) in their analysis reported 13 different sequence types predominated by ST8 (23%) followed by ST1 (13.3%) and ST152 (10%) (Kumburu et al. 2017). Unfortunately, the limited number of publications on genotypic MSSA data could not reveal dominance in any particular identified strain.
MRSA characterization was reported in three of the five publications (Moremi et al. 2019(Moremi et al. , 2012Kumburu et al. 2017;Moremi et al. 2019) screened for the mecA resistance conferring gene as well as the Panton-Valentine leukocidin (pvl) virulence gene by conventional PCR technique. All analysed 24 MRSA isolates harboured the mecA gene hence concordantly agreeing 100% with the phenotypic results. Of note 16.7% of the isolates also harboured the pvl gene. Further characterization of the isolates by MLST and spa typing was done. These typing methods categorized the isolates         Page 13 of 21 Mzee et al. Bull Natl Res Cent (2021) 45:162 in four sequence types. ST88 predominated by 52.2% (n = 13), followed by ST1719 at 29.2% (n = 7), ST8 and ST 1820 were assigned to 3 and 1 isolates, respectively. Of the 4 pvl positive isolates three belonged to ST88 and one to ST1820 (a single locus variant of ST 88). All isolates were ultimately divided into two clonal complexes, i.e. CC8 and CC88 (Moremi et al. 2012). In a subsequent study by Moremi et al. (2019) whereby conventional PCR methods were used so screen for mecA and pvl an account on the characterization of some of the MRSA analysed was given. The study reported that one MRSA from a healthcare worker nasal carriage belongs to ST88 and was pvl positive; furthermore, they reported that 2 SSTI-related MRSA were characterized to belong to ST612 (Moremi et al. 2019). Kumburu et al. (2018) on the other hand characterized MRSA using whole genome sequencing method. Of the 30 isolates analysed, 33.3% (n = 10) were confirmed to harbour the mecA resistance conferring gene. Among the identified 10 MRSA, 6 belonged to ST8 and 2 belonged to ST239, the remaining two had unknown sequence type. Furthermore, none of the MRSA strains harboured the pvl or toxin shock syndrome (tst) virulence genes . Nurjadi et al. 2014 while researching on trimethoprim resistance genes in S. aureus isolated from Sub-Saharan Africa determined that 100% of the Tanzanian S. aureus in the study which were phenotypically trimethoprim resistant in fact harboured trimethoprim conferring resistance genes (Nurjadi et al. 2014).

Prevalence, antibiotic resistance and genotyping of S. aureus isolated from animals
As indicated in Table 5, few studies (n = 5) have reported on prevalence, antimicrobial resistance and genetic characteristics of S. aureus in animals published between 2010 and 2018 in Tanzania.

Prevalence of S. aureus in animals
Five publication reported on prevalence of S. aureus in animals whereby the most examined sample in animals was milk (n = 4). Prevalence of S. aureus varied in different geographic areas. Kashoma et al. (2018) and Mohammed et al. (2018) reported the highest prevalence at 49% and 41% in Morogoro, respectively. Suleiman and Mdegela (2018) did not fall far back when observing a prevalence of 37% describing subclinical mastitis in cows on Unguja Island. The lowest prevalence of 33% was recorded in a study aiming at assessing raw milk quality in Arusha and Meru District (Ngasala et al.   MRSA detection in milk-associated studies varied tremendously ranging from 4 to 35% (Mohammed et al. 2018;Kashoma et al. 2018), even though both studies were conducted in Morogoro township. Furthermore, MRSA prevalence in isolates recovered from cow nasal cavities was repowered at 16% (Suleiman and Mdegela 2018), whereas no MRSA was detected in pigs and dogs nasal swabs (Kashoma et al. 2018).

Antibiotic resistance in animal originated S. aureus isolates
Four studies described antibiotic resistance in S. aureus using different antibiotic pallets which included cephalexin, gentamicin, kanamycin, neomycin, tetracycline, streptomycin, amoxicillin, cephalexin, clindamycin, vancomycin, ampicillin, co-trimoxazole, oxacillin and cefoxitin. This review, however, focused on the most reported antibiotics in this area as indicated in Table 6.
Most of the reviewed studies reported low antibiotic resistance levels with exception of penicillin and ampicillin which showed mean resistances rates of 85% and 73%, respectively. Suleiman and Mdegela (2018) also further observed resistance against amoxicillin and cephalexin at a rate of 47% and 30%, which was unfortunately not reported by any of the other reviewed studies. Moreover, Tetracycline, co-trimoxazole and cephalexin were reported to have a resistance proportion of 30%, 32% and 30%, respectively.
MRSA detection in all studies was confirmed by resistance against cefoxitin and/or oxacillin, whereby mean resistance rates of 10% and 17% were observed, respectively. An average of 8% resistance against vancomycin was observed, reported in 3/5 studies as described in (Table 6).

Genotypic characterization of animal originated S. aureus isolates
Genotypic data retrieved from 2 of the 5 studies were reviewed. In both, real-time-PCR (qPCR) was used to confirm S. aureus as well as screening for mecA resistance conferring gene. Katakweba et al. (2016) additionally screened for mecC, a recently discovered mecA homologue said to have the potential to be mis-categorized as methicillin-sensitive S. aureus (MSSA). None of the two resistance conferring genes were detected in this study; however, S. aureus was confirmed in 4 pigs and 11 dog samples, respectively. Further spa typing of the pig originating MSSA characterized them Page 16 of 21 Mzee et al. Bull Natl Res Cent (2021) 45:162 as spa type t131 associated with ST80. Additionally isolates of dog origin were characterized as spa types t314 (ST121), t084 (ST15 and ST18) and t223 (ST22) all of which were also identified in human nasal isolates reported in the same study. In the study by Mohammed et al. (2018), both coagulase positive (n = 46) and coagulase negative (n = 2) S. aureus were detected in their milk samples. mecA gene was detected in two coagulase positive and one coagulase negative S. aureus isolate making this study the first in Tanzania to report on coagulase negative S. aureus harbouring the mecA gene. Of the 3 MRSA isolates phenotypically detected in the study, one was genotypically coagulase negative S. aureus characterized as spa type t2603, whereas the other two MRSA's were spa un-typable.

Discussion
The publications under review showed that S. aureus was isolated in a number of infections including SSTI's, bloodstream, otitis media, respiratory tract and urinary tract infections staying true to characteristically causing a wide range of infections.
This review observed a S. aureus average prevalence ranging from 1 to 45% in human (Fig. 2). However, particularly higher infection rates were typically observed in SSTI's. Furthermore, it was notably observed that S. aureus colonization ranged between 28 and 45% (Fig. 2) indicative of considerable circulation of both hospital-and community-associated S. aureus in the country. Considering the wide range of infections S. aureus can cause especially in various vulnerable populations such as children, elderly and the immune compromised individuals which was also observed in this review, there is need to emphasize the necessity to establish a reliable and sustainable surveillance system to monitor the S. aureus bacterium countrywide.
This review discovered that the antibiotic resistance patterns in colonization strains recorded uncharacteristically high proportions. Mean resistance rates of 41%, 27% and 46% against erythromycin, co-trimoxazole and tetracycline could be observed, respectively, whereas resistance rates of the same antibiotics reported in colonization strains in Europe and some parts of Asia did not exceed 20% (Heijer et al. 2013;Lestari et al. 2008). Contrary to resistance pattern in clinical S. aureus observed in this review were in accordance with other reports around Africa (Shittu and Lin 2006;Onwubiko and Sadiq 2011). As in many parts recorded resistance against β lactams was high. Resistance rates ranging from 30 to 65% against erythromycin, gentamicin, co-trimoxazole and clindamycin were also described in this review. The affected antibiotics are readily available in the community and also commonly used for empirical treatment of different cases of bacterial suspected infections. A notable presence of resistance against these antibiotics is an indication of their failure in treatment and a need for broader class antibiotics which are expensive and not accessible for most Tanzanians.
The two zones that reported on MRSA prevalence showed immense variation in their studies; however, on average prevalence was reported to be around 26% (Fig. 2). More than a third of the included studies reported prevalence above 25%. This shows that MRSA prevalence has risen in the last 10 years. Abdulgader's review on MRSA prevalence on the African continent categorized Tanzania as belonging to countries with low MRSA prevalence ranging from 6 to 16% between the years 2001and 2009(Abdulgader et al. 2015. This observed abrupt increase should be taken seriously by employing active antibiotic stewardship as well as directing more research efforts towards understanding and preventing the spread of these strains. A wide spectrum of MSSA spa types and sequence types (ST) were identified in the reviewed studies. With the limited information gathered no dominance could be reported. Nevertheless, most pvl positive MSSA clones associate with skin infections identified in this review belonged to ST152 which concurs with the findings from  (2016) was the only study that screened for mecC gene; a mecA homologue which has been identified in other studies to be associated with causing infection in humans with animal contact (Petersen et al. 2013). This homologue can easily be misdiagnosed as MSSA hence causing consequences on patient management as well as on antimicrobial resistance surveillance strategies. In order to make informed decisions about disease prevention and management, there is a need for extended surveillance as well as genotypic based research in Tanzania. Studies reporting on prevalence, antimicrobial resistance and genetic characteristics of S. aureus in animals published between 2010 and 2020 in Tanzania were few; only 5 studies were included in this review, scantly representing a small area in the country. As in human, animal S. aureus isolates seem to harbour highly resistance rates against to penicillin and ampicillin antibiotics. Mean resistance rates against oxacillin (17%) and tetracycline (30%) were further able to describe the effects of overusing oxytetracycline previously stated as the most used antibiotic in the livestock keeping business in Tanzania (Caudell et al. 2017).
Staphylococcus aureus is one of the leading causes of bovine mastitis which explains the fact that the majority of publications included in this review concerning S. aureus in animals addressed the pathogen in clinical and sub-clinical mastitis or in regards to milk production quality. This has been instrumental in the reviews failure to make a link in describing either animals or humans acting as potential S. aureus reservoir for each other, as well as the effect of such in both public and animal health. Very few studies on genomic S. aureus characterization were available for review and did not suffice in showing such linkage. Even so one publication reported on the genetic characteristics managed to confirm S. aureus strains belonging to the same spa type found to colonize both dogs and humans. Unfortunately, the humans and animals involved in this study were epidemiologically unrelated hence making it impossible to link the two. These findings nevertheless present a clear possibility for the S. aureus strain to infect across host species.
The primary method for establishing S. aureus antibiogram in the reviewed work was done phenotypically by using the Kirby-Bauer disc diffusion test along with CLSI guidelines. This method has been reported to have inherent shortcomings such as being highly dependent on experimental conditions that may affect end results (Reller et al. 2009). With this fact in mind the Kirby Bauer antibiogram results would have to be confirmed by another test of a different principle to prevent over reporting of resistance, particularly in the confirmation of MRSA and Vancomycin resistant S. aureus strains. In this review, four studies had MRSA confirmed genotypically which showed 100% agreement with the phenotypic Kirby-Bauer disc diffusion test. This agreement between the methods should encourage the use of the Kirby-Bauer disc diffusion test as an antibiotic sensitivity monitoring strategy in the country, since it is the most available and affordable method in the Tanzanian setting.
Information about prevalence, antibiotic susceptibility and genotypic characteristics of S. aureus originating from different hosts and sources in Tanzania still remains scarce. It is evident that the majority of the publications included in this review are from research institutions or tertiary hospitals affiliated to universities with health-related focuses hence most information could be derived from three focal points MNH in Dar es Salaam, BMC Mwanza and, KCMC Moshi, respectively (i.e. Northern, Lake and Eastern zones). According to the Tanzania National Bureau of Statistics the focal regions represented in this review earn higher income per capital compared to the zones whereby no information could be gathered. This implies that most of the data obtained for review are based in relatively well to do regions (Tanzania National Bureau of Statistics 2018). No published data between 2010 and 2020 on the epidemiology of S. aureus were available for the southern and western parts of the country, which are home to some of the most impoverished regions in Tanzania (refer to Fig. 3). It is well known that poverty struck areas also face other challenges as poor healthcare facilities and access, lack of basic needs such as food, proper housing and sanitation, which in turn leaves the population ridden by different infectious diseases (Alvarez-Uria et al. 2016). Furthermore, since the income gained by the poor is mainly for subsistence, Page 18 of 21 Mzee et al. Bull Natl Res Cent (2021)

Limitations
The data from this review cannot be generalized as true prevalence, antibiotic resistance patterns or genotypic characterization of S. aureus bacterium in Tanzania, there is lack of S. aureus studies representation from other zones. High inter-study variations such as type of specimen analysed, objective of studies, time frame as well as methods employed including criteria set for resistance/MRSA confirmation may have influenced the outcome.