Patients’ selection and ethical approval
This retrospective cohort-study was conducted from July 2005 to April 2016 in the 34-bed infectious diseases department at Nice University Hospital (France). The medical dashboard of our ward records prospectively 28 characteristics of each hospitalization including hospitalization motive, final diagnosis, comorbidities, microbiological data including blood culture and all antibiotics prescribed [10]. Patients are classified regarding the site of infection; in case of bacteremia without any organ infection detected they are included in the BUO group. The dashboard classifies infective endocarditis or spondylodiscitis in other groups. In France no ethical approval is required for non-interventional study. The medical dashboard of the Infectious Diseases Department of Nice University Hospital is authorized by the French National Commission on Informatics and Liberty (Number of Registration: 1430722). A signed consent form is used in our hospital for each patient in order to enable the use of the clinical data recorded during current care for medical research.
Definition of community-acquired BUO used for the study
The origin of the bacteremia was considered as unknown when clinical and paraclinical data failed to identify any infectious focus and when no other microbiological samples retrieved the bacteria isolated from blood cultures.
Inclusion and non-inclusion criteria
All patients classified in the BUO group were selected. After reviewing all the medical files, patients meeting the definition of community-acquired BUO used for the study were included.
Patients with coagulase-negative Staphylococci bacteremia and fungemia were excluded. Because of the difficulty to strictly rule out a cutaneous primary infection in case of chronic ulcers, pressure ulcers or in intravenous drug users, those patients were excluded. Health-care associated infections were also excluded.
Variable of interest
Symptoms leading to the emergency department before hospitalization were collected from the emergency department medical record. Fever was defined as a body temperature ≥38.3 °C. Inappropriate empirical treatment referred to an antibiotic administered before the blood cultures results, to which at least one bacteria isolate was resistant. The number of positive blood cultures bottles for each patient was recorded.
Results of the following exams were collected in the medical record: level of C-reactive protein (CRP) and/or procalcitonin (PCT) at day one, thoracic and abdominal CT-Scan, transthoracic and trans oesophageal echocardiography (TTE and TOE) and colonoscopy.
Duration of follow-up was determined by the date of the last visit in our hospital. Follow-up phone call was performed for the patients who did not visit our hospital after the treatment of the BUO in order to identify any recurrence of BUO. Unfavourable outcome included ICU transfer or death during in-hospital care.
Antibiotic combination evaluation
Based on the antimicrobial susceptibility testing, we retrospectively evaluated the efficiency of two antibiotic combinations for the treatment of BUO: amoxicillin–clavulanic acid + gentamicin (AMC/GM) and 3rd generation cephalosporin (cefotaxime or ceftriaxone) + gentamicin (3GC/GM). These associations were considered as active if the causative bacteria were susceptible to at least one of the two drugs.
Severe sepsis and organ dysfunction
Sever sepsis was defined by the Surviving Sepsis Campaign: systolic blood pressure <90 mmHg or mean arterial pressure <70 mmHg, lactate above upper laboratory limits, PaO2 < 60 mmHg or pulse oximetry <90% in air, creatinine >176.8 μmol/L, platelet count <100 G/L or INR > 1.5, hyperbilirubinemia >70 μmol/L and also altered mental status.
Microbiological findings
Blood samples were collected in a set of an aerobic and an anaerobic bottles which were incubated in a BacT/ALERT 3D (bioMérieux, Marcy l’étoile, France) automated blood culture system for 5 days. Bottles that showed a positive signal in the BacT/ALERT 3D system were routinely subjected to Gram staining and subcultured at least on blood agar plates and upon results of Gram on Drigalski agar or on chocolate agar. Colonies were identified using the API system (bioMérieux) and, since 2013, MALDI-TOF MS Microflex LT (Bruker Daltonics GmbH, Bremen, Germany) according to the manufacturer’s recommendation. Antimicrobial susceptibility testing was performed in accordance with the EUCAST disk diffusion test methodology, as recommended [11].
Statistical analysis
The analysis was performed using StatView®F-4.5. The relationship between variables were assessed with the Chi2 test for categorical variables; Fisher’s exact test was used for number of variables <5. Continuous variables were compared using the Mann–Whitney non-parametric test. Logistic regression was used for the multivariate analysis of risk factors associated to in-hospital mortality or ICU transfer during BUO and results are presented as adjusted odds ratios (AORs) with their 95% confidence intervals (CIs). Variables were selected as candidates for the multivariate analysis on the basis of the level of significance of the univariate (p < 0.1). Models were built up sequentially, starting with the variable most strongly associated with the outcome and continuing until no other variable reached significance or altered the odds ratios of variables already in the model. When the final model was reached, each variable was dropped in turn to assess its effect.