Key findings of this study include the very large burden of SSTIs experienced in Central Australia and the vast disparity between Indigenous and non-Indigenous Australians. The incidence of SSTIs requiring hospitalisation within the Indigenous population of Central Australia is estimated to be 18.8 per 1000 people years. This is almost seven times the incidence for the local non-Indigenous population. More concerning, the incidence is over 30 times the estimated national Australian hospitalisation rate for cutaneous abscesses of 62 per 100, 000 people years . Moreover the incidence is significantly higher than other developed nations, with an American retrospective survey describing an inpatient admission SSTI incidence of 2.19 per 1, 000 person years .
There continues to be a rise in CA-MRSA particularly within Australian Indigenous communities  and our study reflects this trend. We have demonstrated that CA-MRSA is now the dominant S. aureus phenotype in carbuncle and furuncle related SSTIs in the desert regions of Central Australia. The prevalence of CA-MRSA in this SSTI subset (57%) well exceeds the prevalence rates of approximately 20% described in previous Australian S. aureus studies [6, 7, 17]. CA-MRSA is increasingly responsible for SSTIs in Australian Indigenous populations, and in particular the dominant ST93-MRSA clone has continued its spread within Australia into the desert regions of Central Australia. MSSA continues to contribute substantially to the burden of carbuncle and furuncle related SSTIs with CC121 the major MSSA clone. Notably, both ST93 and CC121 are pvl positive clones.
Despite the geographic isolation, high Indigenous population and substantial burden of disease, the molecular epidemiology of SSTI isolates in Central Australia had a number of similarities to other Australian regions. Clonal diversity is a unique feature of S. aureus in Australia, with a recent prevalence survey identifying more than 30 clones in CA-MRSA alone . There are six major MRSA clones including ST93-IV, ST30-IV, ST1-IV, ST45-V, ST78-IV and ST5-IV . ST93 is notable for causing abundant [9, 17, 18] and possibly more virulent disease in Australia  . We also observed significant clonal diversity with 8 clonal complexes identified within 50 isolates, including CC1, CC5, CC8, CC22, CC30, ST93, CC121 and CC88.
Like other regions in Australia, we observed ST93 to be the most frequent strain (48%) and strongly associated with CA-MRSA (96%). ST93 has been found to hyperexpress the exotoxin α-haemolysin which enhances the virulence of the clone, particularly in SSTIs . The high virulence and expression of α-haemolysin in ST93 may be a driving factor in the substantial burden of disease seen in Central Australia [19,20,21].
Following ST93, the next major clone was CC121 (30%), which was exclusively methicillin-susceptible. The association of CC121 with SSTI is consistent with the international literature and CC121 is phenotypically usually MSSA . The molecular epidemiology of MSSA infections has been studied in less detail than CA-MRSA , but CC121 is commonly associated with SSTIs in Australia. However, the high prevalence of CC121 in our study (30%) is greater than that reported elsewhere in the Northern Territory ‘Top End’ and metropolitan centers [9, 18], and may be a unique feature of the Central Australian region.
The absence of S. argenteus is an interesting finding of this study and key difference to other Australian regions. S. argenteus was initially reported in the Northern Territory ‘Top End’ [9, 24], but it is now clear it has a global distribution . The clinical niche of S. argenteus appears to differ from S. aureus . S. argenteus may have less of a predilection to cause abscesses and furuncles and further studies of non-complicated SSTI will be required to determine whether S. argenteus is truly absent from Central Australia.
The dominance of pvl positive S. aureus isolates (90%) may be contributing to the very high incidence of abscess related SSTIs in Central Australia  and is notable for two reasons. Firstly, it was widespread in both MSSA and CA-MRSA isolates. The association between PVL and abscess related SSTIs is well described , however the prevalence of pvl in our cohort was substantially higher than in other Australian urban settings (MSSA 15%, CA-MRSA 42%)  and the Northern Territory Top End (MSSA 40%, CA-MRSA 54%). Secondly, the presence of pvl appears to have limited impact on clinical outcomes following surgical intervention. Controversy continues regarding the pathogenic role of PVL, with laboratory and animal model data suggesting it is not a major virulence factor [26, 27]. In our study, detailed clinical outcome comparisons were limited by the small number of pvl negative cases (10%). Despite this, most patients achieved good clinical outcomes with only one episode of bacteraemia and a minority experiencing progression to deep tissue structures (12%) and secondary infection (2.5%). It was not possible to determine if these complications were related to the presence of pvl or a delay in presentation to health-services or poor wound care upon discharge. Therefore, we believe the typical rapid response to surgical intervention observed in our study provides further support to the findings of Shallcross et al. that provided ‘appropriate surgical treatment and the correct antibiotics’ are received, PVL appears to have little impact on clinical outcomes .
Patients with CA-MRSA frequently received inactive empiric antimicrobial therapy with antibiotic prescription often in accordance with the Australian Therapeutic Guidelines recommendation of di/flucloxacillin for empirical therapy . It follows then that infections due to ST93 were more likely to receive inactive antimicrobials compared to CC121 related infections (54.2% vs 100%, p < 0.001). More importantly, all cases requiring further debridement (n = 6) were ST93 (p = 0.039) and 50% of these received inactive empiric antimicrobials. The high rates of CA-MRSA and potential impact on clinical outcomes lead us now to recommend the empiric use of vancomycin for patients in Central Australia presenting with SSTIs requiring surgical management. Trimethoprim-sulfamethoxazole may be a preferred empiric oral agent given the high rates of observed clindamycin resistance (32%) and dosing ease.
Associations between clonal complexes and antibiograms were noteworthy. Clindamycin resistance occurred in 32% of isolates, substantially higher than previous studies . There was also a significant difference in clindamycin resistance between the major strains CC121 (60%) and ST93 (12.5%), p = 0.006. The impact on antimicrobial prescribing is attenuated by CC121 isolates being exclusively MSSA but this highlights the evolution to a more broadly resistance antibiogram. The rise in clindamycin resistance may be driven by high use of macrolide antibiotics in Indigenous communities of Central Australia, typically indicated for Chlamydia trachomatis or chronic suppurative lung disease.
There are now reports of resistance to trimethoprim-sulfamethoxazole in CC5 strains from Indigenous communities in the neighboring jurisdiction of Western Australia . We found two CC5 trimethoprim-sulfamethoxazole resistant MRSA strains that were also pvl positive. Therefore, ongoing surveillance for changes in resistance and clonal patterns of S. aureus are required to update prescribing recommendations.
This study is limited by its observational design with reliance on coding data to determine SSTI incidence rates and retrieval of clinical data from medical records. It was not possible to obtain ABS population data for the entire ASH catchment given it includes partial Local Government Areas of Western Australia, South Australia and Queensland. Consequently, the population incidence rate may be overestimated. Correlation of the estimated ASH catchment population of 60, 000 with 44% identifying as Indigenous determines an incidence of SSTIs within the Indigenous population as 14.5 per 1, 000 people years and within the non-Indigenous population as 2.4 per 1000 people years.
The small sample size of cases, microbiological isolates and patients with PVL negative disease is the major limitation of this study and prevented detailed molecular and clinical correlations. Similarly, our sampling occurred over 1 month and if seasonal variation was present, we would not have identified this. Review of the geographic distribution of clonal complexes was not possible due to use of broad residential categories and the highly mobile nature of the local Indigenous population. Given asymptomatic patients colonized with pvl positive S. aureus were not included, associations with pvl positivity and clinical disease are only epidemiological in nature and not causal. Other recognized S. aureus virulence factors such as α-haemolysin were not assessed.