Bridging antimicrobial resistance knowledge gaps: The East African perspective on a global problem

Background There is worldwide concern of rapidly increasing antimicrobial resistance (AMR). However, there is paucity of resistance surveillance data and updated antibiograms in Africa in general. This study was undertaken in Kenyatta National Hospital (KNH) -the largest public tertiary referral centre in East & Central Africa—to help bridge existing AMR knowledge and practice gaps. Methods A retrospective review of VITEK 2 (bioMérieux) records capturing antimicrobial susceptibility data for the year 2015 was done and analysed using WHONET and SPSS. Results Analysis of 624 isolates revealed AMR rates higher than most recent local and international reports. 88% of isolates tested were multi-drug resistant (MDR) whereas 26% were extensively-drug resistant (XDR). E. coli and K. pneumoniae had poor susceptibility to penicillins (8–48%), cephalosporins (16–43%), monobactams (17–29%), fluoroquinolones (22–44%) and trimethoprim-sulfamethoxazole (7%). Pseudomonas aeruginosa and Acinetobacter baumanii were resistant to penicillins and cephalosporins, with reduced susceptibility to carbapenems (70% and 27% respectively). S aureus had poor susceptibility to penicillins (3%) and trimethoprim-sulfamethoxazole (29%) but showed excellent susceptibility to imipenem (90%), vancomycin (97%) and linezolid (99%). Conclusions The overwhelming resistance to commonly used antibiotics heralds a clarion call towards strengthening antimicrobial stewardship programmes and regular AMR regional surveillance.

Introduction is greatly affected by community usage as well as hospital usage. Therefore, the medical wards provided a pool of patients in which we were able to study a mix of community-acquired and healthcare-associated infections from this collective.

Data collection
This study reviewed twelve months' worth of data that had been previously collected from 1 st January to 31 st December 2015 in order to report 12-month antimicrobial susceptibility patterns, which is the recommended period of antibiogram reporting as per Clinical Laboratory and Standards Institute (CLSI) [12].

Data analysis
Data was retrieved from VITEK-2 Antibiotic Susceptibility Testing System (bioMérieux) and imported to WHONET 5.6 software through BACLINK (World Health Organization-WHO). Analysis was done using WHONET, SPSS and Microsoft Excel. All isolates were analysed using the Clinical & Laboratory Standards Institute (CLSI M100-S24) standards [13]. Only the first isolate per organism (species) per patient was included as per CLSI recommendations, and analysis done using CLSI breakpoints. Confidence intervals were calculated using the Agresti-Coull interval, as elaborated in the appendix H of the CLSI M39-A4 document which details analysis and presentation of cumulative antimicrobial susceptibility test data [12].

Ethics review
Study approval was granted Kenyatta National Hospital/University of Nairobi (KNH/UoN) Ethics and Research Committee. All the data in this study was analysed whilst maximizing patient confidentiality through de-identification and was fully anonymised. Personal identifiers such as patient name were not captured nor disseminated in any format. The Ethics and Research Committee waived the requirement for informed consent for the retrospective data.

Results
A total of 797 bacterial isolates from the medical wards were reported during the 12 months under study, and 173 of them were excluded due to reasons such as duplicate isolates or missing data, leaving 624 isolates for analysis. Hospital data indicated that the medical wards as a unit contributed the largest proportion of isolates cultured in the laboratory after the paediatrics department, as compared to other units in that year.

Spectrum of gram negative isolates and susceptibility patterns
There were twice as many gram-negative bacteria (419/624, 67%) as there were gram-positive bacteria (205/624, 33%) isolates included in the study. The most frequently isolated gram negative bacteria were Escherichia coli, Klebsiella pneumoniae, Proteus mirabilis, Pseudomonas aeruginosa and Acinetobacter baumanii in descending order. See Table 1 for isolate listing.

Spectrum of gram positive isolates and susceptibility patterns
The most frequently isolated gram positive bacteria were Staphylococcus aureus, Staphylococcus haemolyticus, Enterococcus faecalis, Staphylococcus epidermidis and Enterococcus faecium. Staphylococcus aureus (70 isolates) was the only species that met the threshold for antibiogram reporting (see Table 3). It is likely that a large number of Staphylococcus haemolyticus, Staphyloccus epidermidis and other coagulase-negative staphylococcus species isolated were skin contaminants, and thus their susceptibility rates should be interpreted with caution.

Antibiotic resistance
Nineteen (19) antibiotic class types were tested in total. Resistant bacteria were classified under various concentric categories of non-susceptibility, according to international expert consensus by the European Centre for Disease Prevention and Control (ECDC) and the Centers for Disease Control and Prevention (CDC) [12,14]. Drug resistant (DR) was defined as non-susceptibility to at least one antimicrobial agent. Multi-drug resistant (MDR) was defined as nonsusceptibility to at least one agent in three or more antimicrobial categories. Extensively drugresistant (XDR) was defined as non-susceptibility to at least one agent in all but two or fewer antimicrobial categories (i.e. bacterial isolates remain susceptible to only one or two categories). Possible pandrug-resistant (PDR) was defined as non-susceptibility to all agents in all antimicrobial categories tested. A total of 613 (98%) were drug resistant; 549 (88%) multidrug resistant; 163 (26%) extensively-drug resistant; and 51 (8%) possible pandrug-resistant.
To summarise the clinically important drug-resistant bacteria isolated in our study, the WHO Priority Pathogens List (PPL) released in February 2017 was used [15]. This list contains the 12 most significant antibiotic-resistant bacteria recognised worldwide and the following figures highlight their local prevalence in KNH medical wards as derived from this study. In the critical priority category, we found the following rates of carbapenem resistance: 73% A. baumanii, 30% P. aeruginosa, 13% E. coli [8-19%]

Discussion
The aim of this study was to describe the antimicrobial susceptibility patterns of bacterial isolates from KNH medical ward inpatients. The results presented in this analysis offers a wealth of microbiology and clinical data from the largest tertiary facility in Kenya thus giving the much needed insight into local resistance patterns. By the time of publication, this was the largest clinical study performed in this facility since a laboratory surveillance done from 1991 to 1995 [16].
The three bacterial agents of greatest concern in global antibiotic resistance (E. coli, K. pneumoniae and S. aureus) outlined by WHO [2] formed the majority of the bacteria isolated. E. coli and K. pneumoniae collectively contributed over 60% of the Gram negative isolates (34% and 26% respectively) whereas S. aureus formed the bulk of the Gram positive isolates (34%). This spectrum of isolates has also been demonstrated locally [5,7] and internationally [17] in other facilities where these three bacteria were the most common pathogens causing infection. Therefore, these organisms are of important consideration to a clinician when prescribing therapy to treat bacterial infections, especially those caused by gram negative organisms whose outer membrane confers additional resistance to antibiotics as compared to gram positive organisms which lack it [18]. We demonstrated significant rates of antimicrobial resistance to carbapenems, mostly in A. baumanii (73%) followed by P. aeruginosa (30%), K. pneumoniae (24%) and E. coli (13%). A local private tertiary hospital reported less rates of carbapenem resistance[6] among inpatients in 2014 for A. baumanii (55%), P. aeruginosa (15%), K. pneumoniae (8%) and E. coli (2%). Differences in resistance rates could be accounted for by differences in hospital infrastructure and patient demographics (higher sociodemographic status with fewer total inpatients) as well as the presence of stronger antibiotic stewardship programmes in the private facility. Meanwhile, the little global data available from WHO is from the region of the Americas and Europe, with some reports of more than 50% resistance to carbapenems in two WHO regions [2].
Cephalosporins, especially the third-generation such as Ceftriaxone and Ceftazidime, were among the most prescribed antibiotics in KNH by most cadres of clinicians and this is reflected as well in other local hospitals [5]. Consequently, there were alarming rates of Ceftriaxone and Ceftazidime resistance reported for E. coli (75% and 66%) and K. pneumoniae (82% and 83%) respectively. These rates surpass those seen in other private local facilities for instance one of which registered 49% E. coli and 61% K. pneumoniae resistance to Ceftriaxone among inpatients in 2014 [6]. The disparity in antibiotic resistance rates could be possibly attributed to the differences in patient characteristics, disease burden, infrastructure, clinician prescription practices and antibiotic policies between these facilities. Meanwhile, a systematic review of antimicrobial resistance among clinically relevant isolates in sub-Saharan Africa published in 2014 reported median prevalence of third-generation cephalosporin resistance ranging between 0% to 22% in East Africa, between 6% and 15.4% in central South Africa and between 0% and 46.5% in West Africa. [3] This is in contrast to the global estimates of 50% and 30-60% for E. coli and K. pneumoniae respectively [2]. Overall, the rates of cephalosporin resistance in this study surpass both regional and global estimates, and this could possibly be fuelled by the indiscriminate prescription of cephalosporins. This underscores the need to explore such and other aggressive drivers of antimicrobial resistance in our setup and their effective mitigation thereof through practices such as informed empirical therapy prescriptions. For instance, nitrofurantoin has been recommended in both local and regional reports to be a favourable option for uncomplicated urinary tract infections caused by E. coli [3,7].
Staphylococcus aureus has been known for the last half century to be "notorious" for its ability to rapidly develop antibiotic resistance, since it adapts very well to antibiotic pressure [19]. This was noted with concern in the KNH medical wards, which documented 50% methicillinresistance among 70 S. aureus isolates using cefoxitin screening. Reports of MRSA have been on the rise in past studies carried out in various parts of KNH since the 27.7% MRSA rate published in 2003 [20]. The overwhelming high resistance of MRSA in 3 other Kenyan public health facilities was documented in a 2013 publication which reported 84.1% MRSA prevalence through molecular characterisation of the mecA gene [21]. The presence of MRSA locally has been augmented by studies involving molecular gene typing of MRSA in both private and public healthcare setups, showing marked genetic diversity and significant presence of epidemic clones locally in Kenya [22]. Although data from Africa is scarce, the WHO 2014 AMR report mentioned national data from 9 African countries ranging between 12-80% [2]. A systematic review of MRSA in Africa published in 2013 documents prevalence as high as 82% in some countries [23].
On the other hand, we speculate that we encountered a possible risk of overestimation of MRSA, through confounding by methicillin-resistant coagulase-negative staphylococcus (CoNS) species misidentified as S. aureus. CoNS and S. aureus are frequently isolated together from the same clinical specimen [24]. Misidentification occurs even when using chromogenic agar plates [25] as well molecular Polymerase Chain Reaction (PCR) methods [24]. Since methicillin resistance gene mecA is detectable on resistant strains of CoNS as well as S. aureus, this presents a challenge in true MRSA reporting. False positives have been described [26]. Since molecular detection of mecA alone is insufficient for true identification of MRSA, additional S. aureus-specific gene markers such as nuc [24] and orfX [26] have to be included during testing. We were unable to carry out molecular testing due to financial constraints.
Overall, our study highlighted the burden of antimicrobial resistance in our setup, especially regarding the critical bacteria in the WHO 2017 priority pathogens list which pose the greatest threat to human health. The rising AMR rates are particularly alarming in the context of low-middle income regions where the new antibiotics developed from the dwindling pipeline in the West are either locally unavailable or unaffordable. Even so, with time these new antibiotics will not be spared from the continuous evolution of resistant bacteria. AMR is best fought by sticking to the first line as much as possible. Ultimately, there emerges a pertinent need for antimicrobial stewardship and continual surveillance locally and at a global scale to protect our antibiotic reserve currently in use.

Conclusions
This study addressed some key knowledge gaps as pertains to antibiotic sensitivity and resistance patterns in our region, making a significant contribution towards filling the global resistance map. There was overwhelming resistance noted to commonly used antibiotics such as penicillins and cephalosporins. Rising resistance to potent antibiotics such as carbapenems posed a cause of concern. Collaborative efforts involving clinicians with other key stakeholders are needed to strengthen antimicrobial stewardship efforts, and promote regular surveillance and further research towards combating antimicrobial resistance for the present and future generations to come.