This was a prospective study performed between November 2011 and September 2013. All patients provided informed written consent prior to inclusion in the study. Inclusion criteria were deep S. epidermidis infections after fracture fixation or prosthetic joint surgery. Fracture fixation infections includes fractures involving the long bones as well as the pelvis and the spine. Bacterial growth at the site of interest in combination with either pseudarthrosis, implant-loosening/failure or local and systemic signs of surgical site infection were required for inclusion. Exclusion criteria included infections involving external fixation pins, infections without any implanted hardware and culture positive patients not displaying any clinical sign of infection.
Upon entry into the study, the following surgical parameters were documented: affected bone or joint; type of implant; time between implantation of the device and onset of symptoms. Personal characteristics were also documented and included: gender; age; body mass index (BMI); smoker/non-smoker; overall medical condition (Charlson comorbidity index); and chronic immunosuppressive conditions (Diabetes mellitus, chronic alcoholism, Child's class C cirrhosis, neoplasia, transplantation, AIDS and steroid medication). Any revision surgeries involving the site of interest and all isolated pathogens were recorded throughout the course of treatment and follow-up.
A follow up examination was performed an average of 26 months after discharge. Primary outcome at follow up was cure. Cure was define by the authors as: missing local (at site of interest) or systemic signs of infection and terminated surgical and systemic therapy.
At the first surgical procedure after enrolment, at least four deep bone biopsies were taken from the interface between implant and affected bone. The tissue samples were placed in a sterile container with thioglycollate liquid medium (bioMérieux, Hazelwood, MO, USA). The samples were cultured for ten days at 37°C and examined each day macroscopically. Any growth was inoculated onto a blood agar plate (bioMérieux, Hazelwood, MO, USA) for further growth and subsequent identification. In all cases, additional swabs and soft tissue samples may have been taken as per clinical routine, however, only samples adjacent to the implant were used for diagnosis in this study.
Identification and antibiotic susceptibility testing of all growth was performed using a Vitek2 (bioMerieux Vitek Inc, Hazelwood, MO, USA). Multi-drug-resistance (MDR) was defined according to the definitions of the European Committee of Antimicrobial Susceptibility Testing (EUCAST). Oxacillin resistance was used as an indicator for methicillin resistant S. epidermidis (MRSE).
Biofilm formation was analysed and quantified in microtitre plate assays according to the well-established protocol of Stepanovic et al.(see references). This method traditionally assigns isolates to one of four categories: non-biofilm-formers and weak, moderate or strong biofilm-formers.
The key variables with regards to bacterial phenotype were: biofilm formation; methicillin resistance; and multidrug resistance. The primary outcome measure was whether the infection was "cured" or not. Primary outcome parameters were calculated as a function of all patients for whom data was complete. Statistical comparison was restricted to the lower extremity cohort since the other patients are not scored for many of the functional outcome measures and as such incomplete.
Univariate logistic regression models were used to determine the influence of each prognostic factor (obesity, smoking, diabetes mellitus, chronic immune suppression, open fracture (initially), early onset infection, biofilm formation, methicillin resistance, MDR and Charlson comorbidity index) on cure. P-values \<0.05 were considered significant. Statistical analyses were performed using SAS software (Version 9.2; Cary, NC, USA).
