Dermatological and non-dermatological risk factors associated with adverse clinical outcomes in patients with Staphylococcus aureus bacteraemia

INTRODUCTION Staphylococcus aureus bacteraemia (SAB) has a crude mortality of 39.0%[1] and excess resource consumption.[2] Describing epidemiological trends in clinical characteristics and outcomes is integral to establish treatment protocols and prevention measures. This study aimed to identify dermatological and non-dermatological risk factors in patients with SAB. We also examined if there were significant differences between those with methicillin-susceptible S. aureus (MSSA) and methicillin-resistant S. aureus (MRSA) bacteraemia. METHODS A retrospective cohort analysis of hospitalised patients aged ≥18 years in Singapore with ≥1 blood culture positive for S. aureus over a 2-year period between 2015 and 2017 was performed. Patients with S. aureus- positive blood cultures were identified from the microbiology laboratory database [Table 1]. Only the first (incident) episode of bacteremia was counted to avoid duplication of data. Dermatological conditions were assessed based on extent, depth and whether they were infectious or non-infectious [Table S1, Supplemental Digital Appendix at https://links.lww.com/SGMJ/A108]. Approval and waiver of consent were obtained from National University Health System institutional review board (reference number: 2019/00522). Primary endpoints were 30-day mortality and length of hospitalisation. Patients discharged before but eventually passed away within 30 days from the incident date of SAB were included for analysis. We used 30-day mortality as an endpoint as it is a benchmark in public reporting and comparing hospital quality. Secondary analysis was performed for significant primary endpoints to investigate the differences between MSSA and MRSA bacteraemia [Table S2, Supplemental Digital Appendix at https://links.lww.com/SGMJ/A108].Table 1: Characteristics of patients with Staphylococcus aureus bloodstream infections.Statistical analyses were performed using IBM SPSS Statistics version 10.0 (IBM Corp, Armonk, NY, USA). Dependent variables were all-cause mortality and length of hospitalisation. Logistic regression analysis was performed for variables associated with 30-day mortality, and linear regression analysis was performed for variables associated with length of hospitalisation. Tests were two tailed, with a significance level at a P value < 0.05. RESULTS A total of 355 cases were identified. Of them, 84.5% had MSSA bacteraemia and 14.5% had MRSA bacteraemia. Only incident episodes of bacteraemia were counted. Variables listed were used as dependent variables in multivariate analysis to assess significance against two primary outcomes, 30-day mortality and length of hospitalisation. Criteria used were the variables identified a priori based on existing literature and hypothesised variables with results reflected in terms of odds ratio with 95% confidence intervals. Only immunosuppression was significantly associated with increased 30-day mortality. Three variables were significant predictors of increased hospitalisation length: infectious dermatological conditions [Table S1, Supplemental Digital Appendix at https://links.lww.com/SGMJ/A108], immunosuppression and longer duration to culture clearance [Table 2]. Secondary analyses of these significant variables did not yield significant differences in outcomes between those with MSSA and MRSA bacteraemia.Table 2: Predictors of hospital outcomes (30-day mortality and length of hospitalisation).DISCUSSION This study examined a broad range of dermatological and non-dermatological conditions for clinically relevant outcomes in patients with SAB, with the former rarely described in detail. We found that immunosuppressed patients with SAB had a higher mortality rate and a longer hospitalisation. This observation in mortality is inconsistently reported in existing literature[3] and is related to heterogeneity in defining immunosuppression. In our cohort, immunosuppression was aligned to the United States' Center for Disease Control and Prevention's definition of immunosuppression. This includes patients with untreated human immunodeficiency virus infection with CD count <200/mm3, patients receiving chemotherapy with antimetabolites or systemic immunosuppressants with high-dose corticosteroids or biologic therapies. Causes included a critical state at presentation and a higher risk of receiving inappropriate initial empiric antibiotic therapy, which is associated with increased odds of mortality.[4] These reasons also contribute to a longer hospitalisation and indicate poor source control due to attenuated immune response. Patients with delayed bacterial culture clearance also required longer hospitalisation. In a seminal paper, Fowler et al.[5] established that persistent positive blood cultures beyond 48 h were associated with complicated SAB, mandating more extensive investigations and a longer duration of parenteral antibiotics. These patients may benefit from an earlier identification of infective foci and may be treated for complications under the guidance of specialists from the Infectious Diseases Unit. We studied cutaneous variables as they have not been described well in existing literature and found that patients with concurrent infectious dermatological conditions had longer hospitalisation. This may be explained by biofilm formation on the infected skin, inhibiting effective immune responses and restricting penetration of antibiotics as bacteria enter a non-growing state when depleted of nutrients.[6,7] Blood sterility is dependent on source control and removal of infective foci.[8] Suboptimal treatment of skin infections can lead to extended treatment or treatment failure. We originally hypothesised that more widespread inflammatory skin conditions are associated with adverse outcomes, but this was not found. In atopic dermatitis, deficiencies in endogenous antimicrobial peptides not confined to active eczematous lesions leads to an increased avidity of S. aureus on atopic skin.[9] This predisposes patients to cutaneous infections. It is possible that the extent of body surface involvement may not be significant as cutaneous immunodeficiency may be present intrinsically even in unaffected skin. The number of patients with relevant dermatological conditions in our study may also be too small (n = 90) to detect statistically significant differences. This presents an area worthy of further investigation. The 30-day mortality rate of 12.1% in our study is lower than that in larger studies. We attribute this to routine clinical guidance by infectious diseases consultation, which improves clinical outcomes.[10] No significant difference was found for the study's outcomes between patients with MRSA and MSSA bacteraemia. This emphasises that while invasive MRSA infections are regarded as more lethal, equal attention should be given to MSSA bacteraemia because of its higher prevalence and similarities in adverse hospital outcomes. We acknowledge some limitations. Firstly, we reported overall 30-day mortality, which overestimates the impact of SAB on adverse hospital outcomes. However, attempts to determine whether deaths were attributable to SAB lead to observer bias, so it was not performed. Secondly, the single-centre study design with a small cohort size limits the ability to draw conclusions. Information obtained retrospectively has inherent limitations including incomplete data. Efforts to design prevention measures and treatment protocols to improve hospital outcomes for patients with SAB require understanding and identification of patients with wide-ranging risk factors. Expert review and detailed documentation of a patient's dermatological condition in medical records are integral. The impact of immunosuppression on mortality rate remains variable, and larger studies are recommended to examine the impact of immunosuppression and skin conditions on clinically relevant outcomes. Financial support and sponsorship Nil. Conflicts of interest There are no conflicts of interest.