INTRODUCTION
Staphylococcus aureus is an important human pathogen associated with a variety of clinical presentations ranging from mild to life threatening [Reference Ruhe1, Reference Tong2]. Worldwide, S. aureus is the leading cause of skin and soft tissue infection (SSTI) [Reference Gould3]. Severe SSTI including extensive carbuncles are common in Central Australia, particularly within the Indigenous population [Reference Tong4]. Infections may progress to involve deep structures (e.g. bone and joints), despite prompt surgical debridement and antibiotic therapy.
Alice Springs Hospital (ASH) is the only referral hospital servicing Central Australia and has one of the highest rates of S. aureus blood stream infection in the world, with a mean annual incidence rate in Indigenous patients of 160·7/1 00 000 [Reference Hewagama, Spelman and Einsiedel5]. Tong et al. reported a higher rate of S. aureus blood stream infections in Indigenous Australians compared with non-Indigenous Australians, but a lower 30-day mortality for the Indigenous patients – likely attributable to their younger age at the time of infection [Reference Tong6].
Recently, rising rates of MRSA have been reported locally [Reference Tong7], Australia-wide [Reference Coombs8] and globally [Reference Miller9, Reference Chua10]. This epidemiological shift has reflected the emergence and dissemination of non-multi-resistant strains of MRSA (nmMRSA), which are genetically and phenotypically distinct from multi-resistant MRSA isolates (mMRSA) [Reference Ng11]. While MRSA rates are increasing, currently most community onset staphylococcal disease in Australia is due to methicillin-susceptible S. aureus (MSSA) [Reference Coombs8].
The original multi-resistant MRSA phenotype (mMRSA) was associated with hospital acquisition (HA-MRSA); affecting frail, older patients. This contrasts with reports on nmMRSA that affects younger and often otherwise healthy patients, typically acquired in the community (community-associated methicillin-resistant S. aureus (CA-MRSA)) [Reference Chua10]. Worldwide, Indigenous populations appear to have higher rates of CA-MRSA and in fact the first report of CA-MRSA within Australia was in an Indigenous population in the Kimberley region of Western Australia [Reference Udo, Pearman and Grubb12]. It is now recognised that HA-MRSA strains occasionally circulate within the community; while CA-MRSA is increasingly associated with nosocomial transmission [Reference Seybold13].
Incision and drainage is well recognised to be the mainstay of treatment for purulent skin infections [Reference Chua10, Reference Williams14], while the role of adjuvant antibiotics remains unclear [Reference Stryjewski and Chambers15]. Some researchers have reported lack of active antibiotic therapy (i.e. to which the organism tests sensitive in vitro) increases the risk of treatment failure [Reference Ruhe1, Reference Frei16] while others have found no such association [Reference Miller9, Reference Chen17]. There is no consensus regarding the optimum treatment duration for staphylococcal SSTI, although emerging evidence suggests that short courses (⩽7 days) are appropriate in the majority of cases [Reference Schuler18].
Very few studies have addressed staphylococcal skin infection in Central Australia [Reference Einsiedel19, Reference Stevens20]. Our study aimed to characterise the epidemiology, clinical presentation, antibiotic susceptibility profile and outcomes of community-onset staphylococcal skin infection in our region. We were particularly interested to observe whether or not failure to receive active antibiotic therapy would impact on clinical outcome.
METHODS
Setting
ASH is a 186-bed teaching hospital in Central Australia. It is the sole referral hospital for an area of approximately 1·6 million km2, serving the urban region of Alice Springs, as well as numerous remote communities in the Northern Territory, Western Australia and South Australia [21]. Indigenous patients account for more than 80% of separations for ASH [22], although 2011 census data indicate that only 26·8% of the population of the Northern Territory is of Aboriginal or Torres Strait Islander origin [23].
Patient population
Adult and paediatric patients with community onset, purulent skin infections (i.e. abscesses, including furuncles and carbuncles) were prospectively identified over the 6-month study period (May–October 2014). Enrolment was restricted to those patients with infections requiring hospital admission, who were planned to have surgical debridement, and did not include patients with less severe, smaller abscesses managed within the Emergency Department. Consecutive patients were enrolled in the study if culture from either a superficial swab or a theatre sample yielded S. aureus. Infections involving deep structures (e.g. tendon, bone) at baseline were excluded; as were secondarily infected wounds; and polymicrobial infections such as pilonidal abscesses and diabetic foot infections.
Data collection
A standardised data collection tool was used to extract information from paper-based hospital records, as well as from an electronic health network linking the hospital system with community medical clinics. Not all patients were registered on this electronic system, which accounts for some of the missing follow-up data in our study.
Variables of interest included patient demographics, co-morbidities, clinical presentation, medical and surgical management, culture and susceptibility results, and outcome data at 30 days post-discharge. Systemic inflammatory response syndrome (SIRS) scores were calculated for adult patients according to observations from the time of admission, according to previously validated criteria [24].
Laboratory methods and definitions
Cultures and antibiotic susceptibility testing were performed by the Microbiology Department of ASH, in accordance with Clinical Laboratory and Standards Institute (CLSI) guidelines. Antibiotic susceptibility profiles were determined by Vitek testing (bioMerieux, version 7.01).
Oxacillin susceptible isolates were considered to be MSSA, and oxacillin-resistant isolates were considered to be MRSA. As per a previously described classification scheme, MRSA isolates were defined as non-multi-resistant (nmMRSA) if their phenotype demonstrated resistance to ⩽2 non-beta lactam antibiotics, and multi-resistant (mMRSA) if resistant to ⩾3 classes of non-beta lactam antibiotics (including clindamycin, erythromycin, trimethoprim-sulfamethoxazole, tetracycline, ciprofloxacin and gentamicin); however no examples of this latter phenotype were identified [Reference Hewagama, Spelman and Einsiedel5]. Clindamycin resistance was detected via the Vitek GP card, which incorporates testing for inducible clindamycin resistance.
Antibiotic therapy was determined by the treating clinicians. Therapy was defined as ‘active’ if the isolate was reported as susceptible to the prescribed drug on in vitro testing.
Outcomes
The outcomes of interest were: (1) progression of infection to involve deep structures; (2) need for further surgical debridement; (3) unplanned re-presentation to hospital; (4) unplanned presentation to a General Practitioner (GP); (5) a requirement for an additional or unplanned course of antibiotics (all within 30 days of discharge) and (6) length of hospital stay. We also examined a combined end-point of any adverse outcome – comparing patients with no adverse outcome, to those with one or more of outcomes (1)–(5). Outcomes were correlated with selected variables, including patient demographics (ethnicity, remote residence); age (child or adult); infecting strain phenotype (MSSA or nmMRSA); and antibiotic regimen (active or non-active).
Statistical methods
All analyses were conducted in R version 3.1.2 (R Core Team 2015). On univariate analysis, P-values for the differences between MSSA and nmMRSA were calculated using the Wilcoxon rank-sum test for continuous data, and Fisher's exact test for count data. On multivariate analysis, multiple imputation was performed via predictive mean matching. Adjusted odds ratios were estimated using logistic regression. Imputed estimates were combined using Rubin's rules. Statistical significance was defined by a P value of <0·05.
Ethics
Ethics approval was granted by the Central Australian Human Research Ethics Committee (HREC-14-223). Individual patient consent was not required by the committee since the study was observational in nature and had no impact on clinical management.
RESULTS
Baseline characteristics
Baseline characteristics are summarised in Table 1. A total of 160 eligible patients were enrolled during the 6-month study period. The median age was 32 years (IQR 18·0, 46·2); 78% of patients were Indigenous; 42% lived in a remote community; and 60% were female. Thirty-eight of the patients (24%) were children aged ⩽16 years.
MSSA, methicillin-susceptible Staphylococcus aureus; CA-MRSA, community associated methicillin-resistant S. aureus; OR, odds ratio; CI, confidence interval.
Data are No. (%) unless otherwise indicated.
The majority of S. aureus isolates (96/160, 60%) were methicillin-resistant and all these were CA-MRSA. The remainder of cases were MSSA (64/160, 40%).
Predictors for CA-MRSA vs. MSSA
There was no significant difference in rates of CA-MRSA among adults compared with children ⩽16 years (59% vs. 63%, P = 0·707). On univariate analysis, predictors of CA-MRSA infection were prior isolation of CA-MRSA (OR 2·22, P = 0·031); hospitalisation within the last 6 months (OR 2·32, P = 0·018) and haemodialysis dependence (12 patients with CA-MRSA, 0 patients with MSSA; P = 0·002). On multivariate analysis, none of these factors reached statistical significance as predictors of CA-MRSA. There was a non-significant trend towards increased risk of CA-MRSA infection related to Indigenous status (OR 2·08, P = 0·058).
Clinical presentation and severity
The median duration of symptoms prior to presentation was 5·5 days (IQR 3·0, 7·0), with a median abscess diameter of 5 cm (IQR 3·0, 6·2). Fifteen adult patients (12·5%) met the definition for SIRS. The decision about hospitalisation was based primarily on the local extensiveness of the abscess, and apparent need for formal surgical debridement under general anaesthetic. Clinical presentation did not differ with resistance phenotype (see Table 2).
MSSA, methicillin-susceptible Staphylococcus aureus; CA-MRSA, community associated methicillin-resistant S. aureus; SIRS, systemic inflammatory response syndrome; TMP–SMX, trimethoprim–sulfamethoxazole.
Data are No. (%) unless otherwise indicated.
Antimicrobial susceptibility profile
Differences in non-beta lactam antimicrobial susceptibilities were apparent between CA-MRSA and MSSA cases. The majority (90·6%) of CA-MRSA isolates were clindamycin sensitive, compared with only 40·6% of MSSA isolates. Trimethoprim–sulfamethoxazole (TMP–SMX) sensitivity was 100% for MSSA isolates, while 83 of 96 (86%) of CA-MRSA isolates were TMP–SMX sensitive.
Treatment details
A surgical drainage procedure was performed in 99% of patients; with a median time to surgery of 1 day from time of admission.
All patients except one were commenced on empirical intravenous antibiotics. In 55·6% of patients, the antibiotics prescribed in the first 48 h post-admission were active against the subsequently isolated organism. Patients with CA-MRSA infection were less likely to receive an active antibiotic in the first 48 h compared with MSSA cases (23·6% vs. 100%, P < 0·001). After changing to oral therapy, treatment was active in 80·4% of patients; however, CA-MRSA cases were still less likely to receive an active agent compared with MSSA cases (69·0% vs. 98·2%, P < 0·001). Overall, the median planned oral antibiotic duration was 5 days.
Outcomes
Within 30 days of discharge, 13·5% of patients had an un-planned representation to hospital with the same complaint, while 19·7% had an unplanned GP presentation. The most serious adverse outcomes were considered to be a requirement for further debridement in theatre (9·6% of patients) and progression of infection to involve deep structures (5·2% of patients). It was common for additional courses of antibiotics to be prescribed (27·8% of patients). There was one death within 30 days of discharge (believed to be from an unrelated cause). Overall, 39% of patients experienced at least one adverse outcome. More adults than children experienced at least one adverse outcome (41·8% vs. 28·9%), however this difference was not statistically significant (P = 0·184) and did not differ with receipt or not of active antibiotic therapy (35·9% vs. 42·7%, P = 0·415).
There was a high rate of missing data regarding progression to deep structures, unplanned GP presentation and repeat antibiotics (~22%) but only a minimal amount (~2%) for representation to hospital; repeat debridement in theatre; and length of stay. The median length of hospital stay overall was 3 days (IQR 2·0, 5·5) with CA-MRSA patients having a shorter length of stay compared with MSSA patients (rate ratio 0·78, P < 0·001); however, this result was not significant on multivariate analysis (P = 0·116). Indigenous patients had a longer length of stay relative to non-Indigenous patients, as did those living remotely (see Table 3). On multivariate analysis, increased length of hospital stay also correlated with the presence of diabetes (rate ratio 1·57, P = 0·001); congestive cardiac failure (rate ratio 3·12, P < 0·001); and chronic respiratory disease (1·78, P = 0·018).
MSSA, methicillin-susceptible Staphylococcus aureus; CA-MRSA, community associated methicillin-resistant S. aureus; OR, odds ratio; CI, confidence interval; GP, General Practitioner.
DISCUSSION
This is the first time that CA-MRSA has been reported as the dominant phenotype in an Australian population. Causing 60% of cases, CA-MRSA has now superseded MSSA as a cause of community onset purulent SSTI in Central Australia.
More broadly, it is conceivable that the ascendency of CA-MRSA observed here may foreshadow a more widespread ecological shift across Australia. Tong et al. previously raised the concern that Australian Indigenous communities may act as foci for the emergence of CA-MRSA, in a manner analogous to the US prison populations that were thought to represent core transmitters of the USA300 clone [Reference Tong25]. When considering the epidemiology of S. aureus in Central Australia, it is helpful to compare with the experiences of the Northern Territory's ‘Top End’, since both these regions contain urban and remote populations with high rates of social disadvantage [Reference Tong4, Reference Einsiedel19]. A recent Top End study described a threefold rise in community-acquired MRSA infections, from 7% to 24% over the course of the last two decades [Reference Tong7]. This compares with a 2012 nationwide survey in Australia that reported CA-MRSA accounting for 12·5% of all S. aureus isolates (SSTI specimens comprised 90·5% of all specimens) [Reference Coombs8]. The rate of 60% CA-MRSA in our study is of great concern, and appears to be on an upward trajectory.
The current Central Australian experience resembles the emergence of CA-MRSA skin infection observed in the USA over recent years [Reference Miller9, Reference Talan26]. While the US epidemic is largely attributable to the rapid rise of USA300 as a single virulent clone, the molecular epidemiology of S. aureus within Australia is more complex. The increase in CA-MRSA infections observed throughout Australia reflects the emergence of multiple clones [Reference Tong4, Reference Chua10, Reference Stinear27]. An analysis of S. aureus SSTI isolates from Central Australia revealed a high proportion of panton valentine leucocidin (PVL) across both MSSA and CA-MRSA. Whilst ST93 CA-MRSA and CC121 MSSA predominated, there was wide variety of circulating clones [Reference Harch28]. It has been postulated that CA-MRSA clones in the Top End have arisen de novo in the community, as a consequence of circulating MSSA strains independently acquiring SCC mecIV from sources such as coagulase negative staphylococci [Reference Tong4, Reference Tong25].
The process of SCC mecIV acquisition may be facilitated by the high organism burdens that arise when living conditions are crowded and access to the basics required for hygiene is limited, as is common in remote Indigenous communities [Reference Tong4, Reference Tong7, Reference Tong25, Reference McDonald29]. The frequent use of beta lactam antibiotics in these communities may contribute towards persistence of the methicillin-resistant phenotype [Reference Tong4, Reference Tong7, Reference Baggett30].
We found a disproportionate burden of staphylococcal skin disease among Indigenous patients. Indigenous people comprise approximately one quarter of the population in this region [23], but accounted for more than three quarters of infections in this study. A similar trend was previously reported with S. aureus bacteraemia in Central Australia [Reference Hewagama, Spelman and Einsiedel5]. A trend towards higher risk of CA-MRSA amongst Indigenous patients (univariate OR 2·08, 95%; CI 0·98–4·50; P = 0·058) was also reported in a recent local study of blood stream infections [Reference Einsiedel19] and a multi-centre prospective Australian study [Reference Tong6]. However, it can be difficult to distinguish the impact of ethnicity from potentially confounding variables such as the high chronic disease prevalence (including diabetes and haemodialysis dependence) within Indigenous populations, which may be independently associated with increased infection risk [Reference Einsiedel19].
A notable finding of this study was that both hospitalisation within the previous 6 months and haemodialysis dependence appeared to be predictors for CA-MRSA infection relative to MSSA infection, suggesting possible health-care associated transmission. Within Australia, CA-MRSA has not typically been considered a nosocomial pathogen, however, a study from the Royal Darwin Hospital reported that inpatients were 14 times more likely to be colonised with CA-MRSA compared with their status at the time of admission [Reference Brennan31]. Internationally, the epidemiology also appears to be changing in this regard, with an increasing number of studies reporting that CA-MRSA is now being transmitted from the community back into hospitals [Reference Stryjewski and Chambers15, Reference O'Brien32]. It is apparent from the literature that dialysis patients are at increased risk of S. aureus blood stream infections; as well as being at increased risk for MRSA colonisation and infection [Reference Zacharioudakis33] although much of this data refers to mMRSA rather than the nmMRSA seen in our study.
Some international reports have suggested more severe skin disease associated with CA-MRSA, particularly in relation to the methicillin resistant USA300 clone [Reference Jahamy34, Reference Young35]. However our study did not demonstrate any correlation of S. aureus resistance profile with either clinical presentation or with study outcomes, and aligns with reports from the neighbouring ‘Top End’ [Reference Tong4] of Australia.
In our study, 39% of patients overall experienced at least one adverse outcome, which is higher than has previously described in primary care based studies that reported failure rates of between 10% [Reference May36] and 21% [Reference Lee37]. This discrepancy may be explainable by our chosen study population (hospitalised patients, rather than those managed in the primary care setting), but our study does raise questions about whether particular virulence characteristics might exist among the S. aureus clones circulating in this region. The association we found between the presence of certain co-morbidities (e.g. diabetes) and increased length of hospital stay could potentially be a reflection of impaired healing; or the need for expanded care while in hospital.
Receipt of an active antibiotic was associated with a longer length of hospital stay, perhaps due to reverse causality – additional time for drug susceptibility results to become available and be acted upon appropriately by clinicians. Otherwise, antibiotic choice did not affect outcomes in our study, probably because almost all patients underwent formal surgical debridement in theatre. Lack of a drainage procedure is strongly associated with increased risk of clinical failure [Reference Miller9, Reference Frei16] and there is some evidence to indicate that drainage alone is often sufficient to achieve cure, irrespective of concurrent antibiotic therapy [Reference Chua10, Reference Jahamy34, Reference O'Brien38]. However, contrary results have been reported by other authors who suggest concurrent antibiotics are required [Reference Frei16, Reference Talan39].
An unexpected finding of our study was a high rate of lincosamide resistance – 59·4% of MSSA and 9·4% of CA-MRSA isolates were resistant to clindamycin. Molecular typing suggests that differences correlate with clonal complex type, with CC121 (primarily MSSA) commonly clindamycin resistant (60%); and ST93 (primarily CA-MRSA) infrequently clindamycin resistant (12·5%) [Reference Harch28].
This degree of clindamycin resistance limits its utility as an empiric antibiotic choice. Given the shared resistance mechanism for lincosamides and macrolides [Reference Leclercq40] this phenomenon may be linked to the widespread use of macrolides in our region [Reference Tong7]. TMP–SMX appears to be a more reliable choice as an empiric oral treatment option for staphylococcal SSTI in this population.
Our study has several limitations. Its small sample size limits its power to detect differences among the outcome subgroups. We have only included strains and data from a subgroup of SSTI, namely those presenting with abscesses requiring surgical drainage. Given the limited sampling time frame, these data provide a time limited snapshot and may not be generalisable to a broader time period, nor do they provide a longitudinal view. Most notably, the clinicians’ choice of whether or not to include an MRSA active agent as part of empiric therapy is likely to have been a weighted decision. It may be that those patients with more severe infections had a greater likelihood of receiving active therapy. Consequently, the negative effect of more severe disease at baseline may have diluted any potential benefit to be gained by active empiric therapy.
CONCLUSION
To our knowledge, this is the first time that rates of CA-MRSA infection have exceeded MSSA in the Australian literature, and this may well be a signal of future changes in the epidemiology of S. aureus in Australia. Locally, in light of such extreme rates of CA-MRSA, empiric antibiotic selection for SSTI in this population should routinely include CA-MRSA cover, particularly in the context of sepsis.
We were not able to demonstrate an association between antimicrobial therapy (active vs. non-active) and clinical outcome. Unlike many other reports within the literature, our study cohort underwent formal surgical debridement, rather than incision and drainage in the Emergency Department or out-patient setting. The contribution of antibiotic therapy to cure is likely to be relatively less important when definitive source control has been achieved by surgery.
Public health strategies which address the issues of inadequate housing and poverty are likely to have the greatest impact in terms of reducing the burden of staphylococcal skin disease; as well as limiting the progression of CA-MRSA in Australia.
ACKNOWLEDGEMENTS
James McLeod and the Microbiology Department, Alice Springs Hospital (ASH). Mr Jacob Jacobs and the Department of Surgery, ASH.
DECLARATION OF INTEREST
None.