Although this study found a relatively low incidence of MRSA in Fiji in the community setting, the overall incidence of S. aureus bacteraemia was very high. There was considerable diversity of CC types seen amongst the CA-MRSA and the MSSA isolates, and this diversity was seen in both the community and hospital settings. While CC1 was the most common lineage observed and found both in the community and hospital studies, the CC239 clone was found to be well-established amongst hospital patients similar to more developed nations.
CC1 has also been found to represent 26% of MRSA strains in the Pacific Island nation of Samoa . Further sub-typing of CC1 strains demonstrated that there was no over-representation of any particular community-based sub-lineage as a cause of hospital infections. CC239 is the typical HA-MRSA lineage found in Australia and Asia , and is evidently present in Fiji as a cause of nosocomial infections. With regards to CC30 MRSA, although the name SWP lineage would suggest an origin from this region of the world, it only comprised five of the 22 nmMRSA of hospital isolates and is most likely not a dominant lineage in either Fiji or Samoa .
Perhaps most surprising was the presence of S. argenteus strains. This divergent and early-branching lineage of S. aureus is prevalent in Indigenous populations in northern Australia, [8, 27] and in an Amazonian tribe in French Guinea . S. argenteus typically lacks the genes for production of the carotenoid pigment staphyloxanthin  and is hypothesised to be well adapted to causing skin infections and to be less virulent than other S. aureus lineages . Findings from our current study extend the geographical distribution of this unusual lineage and confirm its non-pigmented phenotype. In addition, we provide supportive evidence regarding its virulence and epidemiological niche in that it represented 8.2% of community based impetigo isolates but only 1.6% of hospital based clinical isolates.
Our study found a prevalence of CA-MRSA in impetigo lesions among Fiji children of 6.7%, which is similar to a reported prevalence of 9% in Samoa . This is in striking contrast to reports of high prevalence of CA-MRSA among Pacific Island peoples and Australian Aboriginal populations living in developed country settings [9, 30, 31]. The reason for the lower prevalence in Fiji is unclear, particularly as many of the conditions thought to contribute to the emergence of CA-MRSA in Indigenous populations in the developed world setting  are also present in Fiji. These factors are likely to include overcrowding, inadequate health-hardware for washing of clothes and people, poor skin hygiene and high rates of antibiotic use and further research specific to Fiji to investigate these is required. The prevalence of MRSA among hospital isolates of S. aureus in Fiji is also low; in the year prior to this study the proportion of all S.aureus isolates from the CWMH laboratory that were methicillin-resistant was 5% (personal communication, S. Pravin).
The incidence of SAB of 50.1 per 100,000 overall and 66.2 per 100,000 for the i-Taukei (Indigenous Fijian) population is considerably higher than that reported from developed countries  and is similar to that seen in Indigenous populations in both northern Australia (65 per 100,000)  and for the whole of Australia (62.5 per 100,000) . The age distribution is also similar to that in Indigenous Australian populations with a peak incidence in the 45–65 age group. This is in stark contrast to that seen in developed countries, including the non-Indigenous population of Australia, where incidence increases with age [32, 33]. Although we cannot infer a causative link, there is likely to be an association between high rates of skin and soft tissue infections and the high incidence of SAB in both Fiji and Indigenous Australian populations. Community-based efforts to improve skin health among these populations [34, 35] will hopefully result in reduced rates of skin and soft tissue infections, and it will be important to determine if there is a concurrent reduction in incidence rates of community-onset SAB.
The hospital study identified consecutive patients that had MRSA isolated from clinical specimens sent to the hospital laboratory for culture, and in this small sample MRSA was primarily isolated from wounds (post-surgical or community acquired injuries). From the molecular typing data it appears that CC239 is commonly observed in hospital in Fiji and given the experience worldwide it would be expected that this is a nosocomial isolate. The nmMRSA CC1 isolates that were identified in the hospital setting however were most likely acquired in the community prior to admission, as the only MRSA detected in community study (in children that had likely not been to hospital) was CC1. However, these data come from a very brief snapshot of S. aureus infections at CWMH. It has been noted elsewhere that CA-MRSA isolates most frequently associated with community-associated infections can become established as the aetiology of nosocomial infections,  and therefore we cannot rule out this phenomenon occurring at CWMH.
Notably, of the hospital-based MSSA isolates, 59% carried pvl genes. While this represented only 16 of 27 isolates it is both surprising and concerning but is similar to findings in Africa  and Indigenous populations in northern Australia  where PVL was present in 57% and 40% of MSSA isolates respectively. Although we did not collect clinical information on the infections caused by these isolates, it nonetheless suggests that there is a large burden of PVL + MSSA disease given that MSSA represents >95% of S. aureus isolates at CWMH.
There are a number of limitations to our study. The community portion of our study investigated one type of clinical infection (impetigo) in one demographic group (school children) in two closely related regions of Fiji. A broader investigation of staphylococcal infections that includes older members of the community and a wider geographic region would be valuable. Within this school group there may also be may be an under-estimation of staphylococcal disease in this population as the surveys took place at two-monthly intervals it is likely that some episodes of impetigo occurred between visits and therefore were missed. The hospital portion of our study had a narrow time frame and a small numbers of cases (especially those where clinical information was available) and so the conclusions from this part of the study are limited.