Introduction

Staphylococcus aureus is one of the most common causes of community and hospital-acquired bacteremias [1]. Due to its complexity, a holistic approach incorporating diagnostic workup (follow-up blood cultures, echocardiography, metastatic foci identification) and management (antimicrobial treatment and source control) is needed to improve outcome [2,3,4]. Despite such an approach, mortality remains high, ranging from 21 to 42% [5,6,7,8,9,10].

Several factors have been associated with worst outcome among patients with S. aureus bacteremia, such as age, comorbidities [5, 6, 9,10,11], presence of sepsis or septic shock [6, 9, 12, 13], immunosuppression [14, 15], and specific foci of infection, such as pneumonia, endocarditis or bacteremia of unknown origin [5, 8, 9, 11]. Although, aforementioned factors are unmodifiable, management of bacteremia can also impact outcome; appropriate antimicrobial treatment was repeatedly shown to improve outcome [6, 7, 16]. Source control is also a key step in early management of infected patients; however, controversy exists concerning the rapidity of source control achievement, with some studies showing an improved survival [6, 7], while in others early source control did not confer significant survival benefit [2, 16].

The aim of the present study was to identify predictors of mortality in patients with S. aureus bacteremia and evaluate the role of source control in a Swiss tertiary university hospital.

Materials and methods

We conducted a retrospective study at the Lausanne University Hospital, Switzerland during a seven-year period (2015–2021). The Lausanne University Hospital is a 1100-bed primary and tertiary care hospital with 35 intensive care units (ICU) beds. The study was approved by the ethic committee of the Canton of Vaud (CER-VD 2021–02,516) that waived the need for informed consent allowing the inclusion of all hospitalized patients except those who refused the use of their clinical and laboratory data.

Inclusion criteria were adult patients (≥ 18 years old) and presence of at least one blood culture for S. aureus (database of the microbiology laboratory). Exclusion criteria were patients’ written refusal of the use of their data and incomplete medical files (patients transferred to other hospital upon infection onset without follow-up information).

Blood cultures were incubated the BacT/ALERT System (bioMerieux, Marcy l'Etoile, France). Matrix-assisted laser desorption-ionization time of flight mass spectrometry (MALDI-TOF MS; Bruker Daltonics, Bremen, Germany) was used for the identification to the species level. Susceptibility results were collected from the microbiology laboratory database and evaluated according the EUCAST criteria [17].

Twenty-eight-day mortality was the primary outcome. Data regarding demographics (age, sex), comorbidities, Charlson Comorbidity Index [18], laboratory results (white blood cells, platelets, C-reactive protein, procalcitonin) on the day of first positive blood culture, Sequential Organ Failure Assessment (SOFA) score [19], antimicrobial treatment, source control, presence of sepsis or septic shock, infection site were retrieved from patients’ electronic health records. All data were collected, stored and managed using REDCap by an infectious diseases specialist. REDCap electronic data capture tools is hosted at Lausanne University Hospital. REDCap (Research Electronic Data Capture) is a secure, web-based software platform designed to support data capture for research studies [20, 21].

The date of collection of the first positive blood culture was defined as infection onset. A new episode was included if more than 30 days had elapsed since the first positive blood culture. Since 2007, an infectious diseases consultation was performed on a mandatory basis within the same day of S. aureus blood culture positivity [3].

Bacteremia was characterized as community if the first positive blood culture was drawn upon hospital admission or within 48 h after hospital admission and nosocomial if the first positive blood cultures were drawn after 48 h from hospital admission. Sepsis or septic shock was defined according to definition proposed by the Sepsis-3 International Consensus [22]. Complicated bacteremia was defined as presence of endocarditis, metastatic infection, implanted prostheses or persistent bacteremia for more than 48 h. Infectious endocarditis was defined according to the modified Duke criteria [23]. Cardiac predisposing factors for endocarditis were defined as cardiac conditions at high or moderate risk for infectious endocarditis [24]. Infection site was defined by the infectious diseases consultant responsible of the case on the basis of clinical, radiological, microbiological, and operative findings. Appropriate antimicrobial treatment was defined as one that included an antimicrobial agent with in vitro activity against the infecting isolate, initiated within 24 h from the infection onset, at an adequate dosage. Source control considered as warranted was (1) removal of venous catheter in patients with bacteremia of unknown origin in the presence of vascular catheter or catheter-related bacteremia; (2) surgical or imaging-guided drainage of infected collections (abscess, peritonitis, and empyema); (3) joint fluid drainage (arthrotomy or arthroscopy); (4) cardiac surgery in endocarditis patients when indicated [23]; (5) correction of urinary-tract obstruction. Early source control was defined if performed within 48 h from infection onset.

SPSS version 26.0 (SPSS, Chicago, IL, USA) and R version 4.1.3 (2022, Vienna, Austria) statistical soft wares were used for data analysis. Categorical variables were analyzed using the chi-square or Fisher exact test and continuous variables with Mann–Whitney U test. Univariate logistic regression models were assessed with 28-mortality as dependent variable. Covariates were tested for multi-collinearity through variance inflation factor assessment: those not collinear and clinically relevant were used in multivariate analysis. After checking Cox assumptions, two multivariate Cox proportional hazards regression models were performed with 28-day mortality as the time-to-event: (i) first including all patients, (ii) second assessing only patients for whom a source control was needed based on the type of the infection. Odds ratios (ORs) and 95% confidence intervals (CIs) were calculated to evaluate the strength of any association. All statistic tests were 2-tailed and P < 0.05 was considered statistically significant. We finally performed Kaplan–Meier curves of the survival probability of patients with S. aureus bacteremia according to appropriate source control with 48 h from infection onset and presence of septic shock. Since it was previously suggested that source control could be influenced by care withdrawal [25], Kaplan–Meier curve was performed among patients that were alive and in maximal care for 7 days after infection onset in order to assess the role of early source control on survival.

Results

A total of 1156 episodes of S. aureus bacteremia were identified; 839 episodes in 779 patients were included (Fig. 1). Forty-seven patients had multiple episodes (41, 5, and 1 patients had 2, 3, and 4 episodes, respectively). The 60 subsequent episodes of bacteremia, occurred at a median of 8 months from the previous episode (range 1–73 months). Overall, 66 (7.9%) isolates were resistant to methicillin. Seventy-seven (9.2%) episodes were polymicrobial. Most bacteremias were related to bone and joint infections (268; 31.9%), followed by bacteremia of unknown origin (158; 18.8%), proven endocarditis (118; 14.1%), lower-respiratory tract (79; 9.4%), and central catheter (78; 9.3%). Among episodes with endocarditis, 105 had valvular infection (cardiac surgery in 36 patients among 56 with indication) and 21 lead infections of cardiovascular implantable electronic devices (CIEDs; CIED removal in 20 patients).

Fig. 1
figure 1

Flowchart of patients’ selection

Full size image

Overall 28-day mortality rate was 14.5% (122 episodes). Results of univariate analysis for predictors of 28-day mortality are shown in Table 1. Sepsis occurred in 352 (42.0%) episodes. Antimicrobial treatment was initiated within 24 h in 801 (95.5%) episodes and was appropriate in 761 (90.7%) episodes. Infectious diseases consultation was provided in 727 (86.7%) cases within 48 h from infection onset. Results from Cox multivariate regression model showed that Charlson comorbidity index > 5 (P < 0.001; OR 4.98, CI 2.61–9.48), nosocomial bacteremia (P 0.019; OR 1.57, CI 1.08–2.29), time to blood culture positivity ≤ 13 h (P 0.004; OR 1.85, CI 1.22–2.81), persistent bacteremia for ≥ 48 h (P 0.004; OR 1.83, CI 1.22–2.76), sepsis (P < 0.001; OR 3.39, CI 1.97–5.83), bacteremia of unknown origin (P 0.036; OR 1.64, CI 1.03–2.60) and lower respiratory tract infection (P < 0.001; OR 2.96, CI 1.77–4.95) were associated with 28-day mortality, while infectious diseases consultation within 48 h from infection onset (P < 0.001; OR 0.45, CI 0.30–0.69) was associated with better survival.

Table 1 Predictors of 28-day mortality of S. aureus bacteremia episodes
Full size table

Source control was warranted in 575 (68.5%) episodes and performed in 533 (92.7%); early source control was performed in 345 (60.0%) episodes. Table 2 shows the source control procedures warranted and those performed depending on infection site. Among the 575 episodes, 28-day mortality was 12.3%. Results from a second Cox multivariate regression model (Table 3) confirmed that Charlson comorbidity index > 5 (P < 0.001; OR 5.55, CI 2.65–11.62), nosocomial bacteremia (P < 0.001; OR 3.00, CI 1.75–5.14) and sepsis (P < 0.001; OR 5.46, CI 3.06–9.71) were associated with increased 28-day mortality, while infectious diseases consultation within 48 h from infection onset (P 0.002; OR 0.39, CI 0.22–0.71) and early source control (P < 0.001; OR 0.35, CI 0.20–0.60) were associated with better survival.

Table 2 Source control procedures warranted and performed depending on infection site
Full size table
Table 3 Predictors of 28-day mortality of S. aureus bacteremia in 575 episodes for which source control was indicated
Full size table

Figure 2 shows Kaplan–Meier curves of the survival probability of episodes with S. aureus according to early source control (A) in 575 episodes for which source control was warranted, (B) in episodes without septic shock, (C) with septic shock. Early source control was associated with better outcome in all episodes (P < 0.001) and in the subgroups of patients without (P 0.009) and with septic shock (P 0.035). Figure 2D shows the Kaplan–Meier curve among 555 episodes in patients that were alive and in maximal care for 7 days after infection onset; early source control was associated with better outcome (P 0.007).

Fig. 2
figure 2

Kaplan–Meier curves of the survival probability of patients with S. aureus according to early source control A in 575 episodes for which source control was warranted, B in episodes without septic shock, C with septic shock. D The Kaplan–Meier curve among 555 patients that were alive and in maximal care for 7 days after infection onset. Red line: no early source control, green line: early source control

Full size image

Discussion

The present study assessing factors associated with mortality among patients with S. aureus bacteremia highlights the crucial role of early interventions, such as source control and infectious diseases consultation on the management of such bacteremic patients.

The reported 28-day mortality rate was 14.5% which is lower than that reported in the literature (21–42%) [5,6,7,8,9,10]. One hypothesis of the increased mortality among aforementioned studies could be due to the higher rate of MRSA as compared to the present study (7.9%) [6,7,8,9,10]; the rate of MRSA in the present study is comparable to that reported from S. aureus surgical site infections from a multicenter Swiss study [26].

Charlson comorbidity index, as expected, was independently associated with a worse outcome [5, 6, 9,10,11]. Both immunosuppression and septic shock are known to impact mortality of S. aureus bacteremia [9, 14, 15]. Since, the vast majority of patients received appropriate empiric antibiotic treatment (90.7%), we could not assess its impact on survival; in previous studies, a smaller percentage of patients received appropriate antimicrobial treatment during the first 24 h (52.8–74.7%), and its administration was associated with better outcome [6, 7, 16].

In accordance to the literature, persistent bacteremia was found to independently predict mortality [5, 27]. In a previous study focusing on duration of persistent bacteremia, mortality increased by 16% for every day of delay in clearance of bacteremia [27]. In that report, delay or absence of source control was associated with persistence of bacteremia [27].

As previously shown, focus of infection plays an important role on outcome, with bacteremias of unknown origin [5, 8, 9, 11] or due to pulmonary infections [8, 11,12,13] being associated with worse outcomes. In contrast to prior studies, endocarditis was not associated with increased mortality [5, 8, 9]. An explanation for the absence of such association might be the high rates of cardiac surgery (34.3%) among patients with valvular endocarditis in our setting as compared to previous studies (15–26%) [28, 29]. Both previous studies showed that absence of cardiac surgery among patients with endocarditis was associated with worse outcome [28, 29].

The role of early source control on survival among patients with S. aureus bacteremia remains debated, since some studies have shown that source control improved survival [6, 7], while in others source control did not confer significant survival benefit [2, 9, 16, 30]. Achieving adequate source control accelerates S. aureus bacteremia’s clearance and reduces associated mortality [27]. This essential element of management was not included in the analysis of mortality of many previous studies [8, 10, 11, 14, 15, 31]. There are many confounders for the association of infection source eradication and improved survival, such as decision of care limitation or withdrawal and desistance of surgeon or interventional radiologist to perform the intervention [25]. To account for such a bias, a secondary analysis was performed by including patients that were on maximal care for the first 7 days from infection onset; early source control was associated with better outcome in that subgroup, underlining the importance of prompt control of infection focus on the management of S. aureus bacteremia. Our results are in accordance to studies focusing on early source control procedures among other types of infections such us intra-abdominal infections, necrotizing fasciitis and sepsis [32,33,34]. The timing of source control was different in the present study depending on site of infection and the complexity of each intervention, with catheter removal among patients with unknown origin or catheter-related bacteremia being performed in the majority of patients within 48 h from infection onset, while drainage of abscesses or replacement of prosthetic material being less commonly performed in the same timepoint.

The management of patients with S. aureus bacteremia is complex and our results show that infectious diseases consultation within 48 h played an important role in reducing mortality. As it was shown in a previous study from our institution (2001–2010), after the implementation of mandatory infectious diseases consultation for MRSA bacteremia, rates of source control increased leading to improved survival [3]. Infectious diseases consultation was shown to increase adherence to guidelines (follow-up blood culture, echocardiography) and improve management (appropriate antimicrobial treatment, source control) and outcome [12, 31].

The present study has several limitations. First, it is a single center retrospective study, which can make the identification of the source of S. aureus bacteremia difficult for some patients. It was conducted in a setting of low MRSA prevalence meaning that results cannot be extrapolated to settings with much higher prevalence. Our cohort represents the most complex cases, since we collected data from patients requiring hospitalization in a tertiary hospital; thus, our epidemiology may not reflect non-tertiary hospitals. Second, no molecular investigation, such exotoxin genes or clonal types, was performed; this area remains scarcely investigated and more research is needed to ascertain the role of superantigens, cytotoxic exotoxins, or various clones on mortality. Additionally, since no clonal types investigation was performed, we could not ascertain if the 60 recurring episodes were from the same or a new S. aureus strain.

In conclusion, we found that in patients with S. aureus bacteremia delaying source control interventions was associated with worse outcome. Moreover, this study underscores the importance of infectious diseases consultation by guiding antimicrobial treatment, diagnostic investigations and proposing source control interventions.