This study demonstrated that the prevalence of MRSA among S.aureus BSIs was non-significantly decreasing over this 5-year period, from 33–15%. This finding is in contrast with our study conducted between 2007 and 2011, in which there was an increasing prevalence, doubling from 23–43%.  The SENTRY study reported a prevalence of 37.8% in the period 1997–2000, reaching a peak of 45.3% in 2005–2008, slightly declining to 40.0% from 2013–2016. The surveillance of antimicrobial resistance in Europe in 2018 showed a large difference in MRSA percentages ranging from 0–43%, giving an average rate of 16.4% in 2018. Almost one-third of European countries reported a significant decrease in the resistance rate from 2015–2018.  The decreasing trend in many countries may be attributed to the new technology enabling early detection of MRSA, leading to prompt initiation of appropriate antibiotics to reduce colonization pressure and improvement in infection control measures.
Demographic and clinical characteristics are comparable to the previous reports. The majority of patients had underlying diseases, had prior exposure to antibiotics within 3 months, and had medical devices inserted. Seventy percent of patients had concurrent other infections; 33% of patients had pneumonia and 19% of patients had skin and soft tissue infection. [6, 7, 24]. All clinical isolates were sensitive to vancomycin according to vancomycin MIC.
Vancomycin is a life-saving medication, and has been the mainstay of treatment for MRSA infections for several years. Although new medication i.e. linezolid, daptomycin, and ceftalorine is available,  they are not widely accessible for the majority of patients in Thailand, mainly due to high cost. Vancomycin has a narrow therapeutic index, high levels can cause toxicity and low levels can result in treatment failure.  Therefore, drug level must be monitored for patients receiving vancomycin. A level of 15–20 mg/L is required for treatment success of BSIs.[26, 27] The pharmacokinetics (Pk)/ pharmacodynamics (Pd) parameter of vancomycin, which is the area under the curve (AUC) above the minimal inhibitory concentration (MIC) of > 400 is correlated with good clinical outcome. However, in the era of vancomycin MIC creeps, higher drug levels may be required to achieve that Pk/Pd parameter. In the past, vancomycin MICs were mostly < 0.5 mg/L.[28, 29] however, many reports demonstrated that vancomycin MICs are increasing.[16, 28–30] A study in the US, reported a vancomycin MIC ≥ 1 mg/L from 2000–2008.  In China, the percentages of vancomycin MIC = 1 mg/L increased from 37.0% in 2006 to 75.7% in 2010. (25) In this current study, vancomycin MIC for each year was mostly ≥ 1 mg/dL. Sixteen patients (19.7%) had vancomycin MIC = 2 mg/L. There are conflicting data with regard to the association between treatment failure and an MIC of 1.5 mg/L or greater. [18, 19, 31, 32] This study demonstrated no association between mortality and an MIC of 1.5 mg/L or greater. Given a vancomycin MIC of ≤ 1.5 mg/L, vancomycin is still a mainstay of treatment for MRSA infection in our hospital. However, as some strains had an MIC of 2.0 mg/L, clinicians should be aware of MIC creep and monitor the position continuously.
A systematic review and meta-analysis reported mortality rates of MRSA BSI, ranging from 14.2–41.8% between 2006 and 2011. Our previous study which was conducted between 2007 and 2011 reported a mortality of 53.1%.  Mortality rate among those who had vancomycin MIC < 1.5 mg/L ranged from 9 and 39% and those who had vancomycin MIC ≥ 1.5 mg/L ranged from 5 and 66%.  Not all studies demonstrated a higher mortality rate among those who had MIC ≥ 1.5 mg/L.
Factors associated with death included older age, presence of alteration of consciousness and concurrent pneumonia. Aging is associated with low immunity,  alteration of consciousness may represent signs of sepsis, [34, 35] and concurrent pneumonia may be associated with prolonged hospitalization, frequent use of medical devices and broad-spectrum antimicrobials, which may directly or indirectly lead to fatal outcome. In addition to those factors, other studies also reported that those receiving corticosteroids, requiring mechanical ventilation, had sepsis or septic shock, and failure to receive anti-MRSA antibiotics within 24 hours of MRSA identification all had a higher incidence of fatality. [36–38] Chronic kidney disease reduced the risk of mortality as these patients, particularly those receiving renal replacement therapy, are acknowledged to be at risk of infection caused by MRSA or coagulase-negative staphylococci and are given early anti-MRSA therapy as a component of empirical antimicrobials.
What changed from the study carried out from 2007–2011? First, the number of cases declined and the prevalence of MRSA among S. aureus BSI decreased. Second, the clinical characteristics in both studies were somewhat similar i.e. majority of patients had underlying diseases, and medical devices insertion. Most commons concurrent infections were pneumonia and skin and soft tissue infections. Third, with data available, the means vancomyin MIC were ≥ 1 mg/L. Fifty-six percent of isolates had vancomycin MIC ≥ 1.5 mg/L. Fourth, the mortality rate was higher in the current study, which might be explained by the fact that more patients in the current study had underlying diseases (however, we could not directly test for statistically significant difference). Finally, risk factors of death were somewhat similar.
This study had several limitations. First, due to it being a retrospective study, some data may be missing and lead to misinterpretation of the results, e.g. the leading cause of death might be due to MRSA infection itself or from other causes. Second, the number of patients recruited was small; therefore, factors associated with death may not have been detected.