Carriage of Staphylococcus aureus in Thika Level 5 Hospital, Kenya: a cross-sectional study

Staphylococcus aureus is both a human commensal and an important pathogen [1]. Methicillin-resistant S.aureus (MRSA) is a common cause of healthcare-associated infections in both developed and developing countries, though limited information is available from the latter [2]. In hospital wards in developing countries, the risk of nosocomial transmission is likely to be high due to close physical proximity of patients, inadequate staffing, unreliable water-supply for handwashing, lack of alcohol hand-rub, isolation facilities or expertise for infection control [3] - these could represent near-ideal conditions for nosocomial circulation of drug-resistant bacteria. In sub-Saharan Africa, screening for MRSA carriage during hospital admission is rarely considered a healthcare priority and is infrequently performed, so patients carrying MRSA in hospitals in this region are unlikely to be identified or isolated.

Carriage of S.aureus is most commonly detected in the anterior nares (nostrils), though other sites on the body are frequently colonised, most commonly the groin and the axillae [1]. Isolates from the nares are normally indistinguishable by molecular typing from those found on other body sites [4]. Most S.aureus infections originate from self-carried strains, although disease only develops in a small minority of carriers [4, 5]. Cross-sectional studies in adult populations have typically found nasal carriage rates of around 25%, although much of this work has been conducted in high-income settings [1]. Carriage of S.aureus (either MRSA or methicillin-sensitive S.aureus (MSSA)) often lasts from months to years, and tends to be longer amongst individuals with chronic skin disease [6].

MRSA expresses resistance to all β-lactam antibiotics (penicillins, cephalosporins, carbapenems) by producing a penicillin-binding protein (PBP2a) from the mecA gene in the staphylococcal cassette chromosome (SCCmec). MRSA was first described in 1961 in the UK, very soon after the introduction of anti-staphylococcal penicillins [7]. Some of the earliest reports of MRSA in sub-Saharan African came from various Kenyan Hospitals (including Thika Hospital) in the early 1990’s, where it was principally found amongst burns patients [8]. The prevalence of MRSA is the percentage of a collection of S.aureus isolates expressing resistance to methicillin – a marker for the presence of mecA. This has been estimated in a small number of studies in sub-Saharan Africa over the past 25 years, mostly based on clinical (disease-causing) isolates obtained at university teaching hospitals in major cities [9‐12]. However, there are relatively few studies examining the prevalence of S.aureus carriage and of MRSA amongst these carriage isolates from patient populations in sub-Saharan Africa [13‐18] – see Table 1. These studies show considerable variation in MRSA carriage prevalence amongst inpatient groups: from as high as 21% amongst TB inpatients in Tulega Ferry, South Africa to as low as 1.3% amongst a population comprising paediatric and surgical inpatients in Accra, Ghana - it is unclear what gives rise to this diversity.

Most MRSA strains in Africa appear to derive from a small number of common ancestors: a recent study examining 86 MRSA isolates from clinical isolates at five hospitals in West Africa and Madagascar found that 88% of these isolates were from one of three clonal strains [12]. By contrast, there was wide diversity amongst MSSA clones from the same institutions [11]. No similar studies have been reported from East Africa and the types of MRSA circulating in East Africa are largely unknown.

We aimed to describe the prevalence and diversity of MSSA and MRSA carriage amongst inpatients in a typical mid-sized public-service hospital in Kenya.

In this typical mid-sized Kenyan government hospital, carriage of S.aureus amongst inpatients was found to be relatively infrequent (8.9%; 95%CI 7.1-10.8) in individual screens. Due to the repeating cross-sectional sampling design, this result may underestimate the actual inpatient prevalence due to a time-dependant bias [24] introduced by only performing multiple rounds of screening in patients with longer admissions. The actual prevalence might be better estimated by the proportion of patients in the study who were ever detected to have S.aureus carriage (74/733; 10.1%; 95%CI 7.2-12.3) – though some of these patients probably acquired carriage during admission. MRSA constituted only 6.9% (95%CI 1.5-12.5%) of all S.aureus isolates obtained. Both the overall carriage rate and the proportion of MRSA were surprisingly low, given that nasal carriage of S.aureus in European populations is typically around 25% [1, 25] and the crowded conditions in Thika Hospital should have led to frequent transmission of MRSA. The small number of MRSA isolates obtained leaves us with a wide confidence interval for the true microbiological prevalence.

Almost all MRSA carriage was confined to patients with burns who constituted approximately 10% of adult inpatients admitted to the surgical wards and approximately 5% of inpatients overall. In sub-Saharan Africa, prolonged hospital admission after burns injuries are common [26], and as specialist burns units are rarely available, these patients often make up a sizeable proportion of the general inpatient population. Burns patients are at increased risk of invasive bacterial disease [27] and hence colonisation of these patients with a multi-drug resistance organism is of particular concern. Chloramphenicol is a widely used 2nd line antibiotic in sub-Saharan Africa and as all isolates in Thika Hospital were found to be susceptible to this drug, this might be a suitable empirical alternative to vancomycin for suspected MRSA infections in this setting.

These Thika MRSA isolates belong to the widespread ST239 strain, formerly known as the Hungarian/Brazilian epidemic strain which is typically associated with hospital-acquisition. This has been documented to occur in sub-Saharan Africa in Senegal, Niger [12], South Africa [28] and was noted to be common amongst hospitalised burns patients in Nigeria [10]. The pattern of multi-class antibiotic resistance that we found in Thika Hospital is typical of this strain in African settings [12].

The reason for the surprisingly low overall rate of S.aureus carriage (8.9%) in our study is unclear. We feel that is it unlikely that procedural errors with swab collection could account for this – all swab collection was supervised in person and regularly quality-controlled by one investigator (AA) and an experienced researcher (HG) made an on-site review of the collection and processing methods in the first month of the study. The anterior nares and axillae are widely accepted to be reliable sites for S.aureus detection [1] and we used an enrichment culture intermediate step to assist with detection of small numbers of organisms. However, we accept that swabs from the skin surface have inherently limited sensitivity to detect carriage organisms and we note that in our study, patients who had multiple screenings performed were more likely to ever have had S.aureus carriage detected. It is therefore possible that a single round of screening is not sufficient to exclude carriage in this setting. Alternatively, progressive acquisition of carriage during hospital admission, as evidenced by our molecular studies, could provide some explanation for this finding of increasing carriage with repeated screening. Genetic differences might also provide an explanation for a low carriage rate – human genetic factors are thought to be important determinants for persistent colonisation with S.aureus[29].

There were some intriguing findings revealed by the molecular analyses. In particular, we note that in comparison to isolate collections from other settings, there were high proportions of isolates with genes for the Panton-Valentine Leukocidin (PVL, 15/86, 19%) and toxic shock syndrome toxin 1 (tst1, 18/86, 23%). From the perspective of staphylococcal classification systems, many of the spa types (18%; 5/28) and MLST profiles (15%, 3/20) identified in this population were novel. This is reflective of the limited contribution of African sites to the relevant typing databases. The lack of diversity amongst the MRSA and MSSA isolates obtained in our institution over a four month period suggests that these were probably being transmitted by local (within-hospital) routes rather than repeated introduction from external sources, though without greater knowledge of the diversity of strains circulating in Kenya or East Africa as a whole, we cannot be certain of this. Due to overcrowding and lack of isolation facilities, eliminating nosocomial transmission of organisms such as MRSA in institutions like Thika Hospital will be challenging.

Antibiotic resistance is a growing concern in developing countries [30] and local studies are needed to understand the transmission of drug-resistant organisms and guide strategies for containment. MRSA is an important nosocomial pathogen in developed countries but is little-known in sub-Saharan Africa. This is the largest study examining inpatient carriage of S.aureus in sub-Saharan Africa and it was conducted in a typical mid-sized Kenyan Government Hospital. We found inpatient carriage of MRSA to be relatively infrequent and was mainly confined to patients with burns. The lack of genetic diversity amongst the isolates obtained suggests “within-hospital” transmission. The multiple drug resistance of MRSA found in Thika hospital coupled with the vulnerability of burns patients to invasive bacterial infection is a cause for concern. Thika Hospital faces many infection control challenges that are typical of those encountered in African healthcare facilities. Much work is needed in Kenya and throughout the region to formulate appropriate national and local policies to tackle issues such as MRSA and, more importantly, to put these into effective practice.