[Surgical site infection after total knee arthroplasty: a monocenter analysis of 923 first-intention implantations]

We report the results of a retrospective analysis of 923 cases of first-intention total knee arthroplasties. The objective was to determine retrospectively the rate of surgical site infections, including all infections diagnosed during the first year, and to search for risk factors. We also wanted to present our surveillance system planned for a 10-year period.

From January 1994 to January 2004, first-intention total knee arthroplasty (TKA) was performed on 999 knees. HLS prostheses were implanted. At minimum 12 months, follow-up data was complete for 923 implants which constituted the study group. Female gender predominated (72%). Mean patient was 71 years (range 26-93). Anterior surgery was performed for 25% of the knees. Etiologies were osteoarthritis (87.5%), and rheumatoid polyarthritis (6.9%). Cefazolin was used for systematic preoperative (one injection) and postoperative (48 hr) antibiotic proxphylaxis. Vancomycin was used for patients with a contraindication for cefazolin. Information was collected from two sources: computerized consultation charts for all follow-up visits completed prospectively since 1995 et data collected by the Hygiene and Epidemiology Unit during the year following implantation. Data on surgical site infections was collected from the hospitalization files, outpatient files and control visits. Each case of infection was validated at an annual interdisciplinary meeting. We retained for analysis deep infections requiring revision surgery with identification of the causal agent on the intraoperative samples. We identified a subgroup of infections occurring during the first postoperative year, the delay generally retained for surgical site infections.

Twenty surgical site infections after TKA were identified during the 10-year surveillance period (2.1%). Mean follow-up was 43 months (range 12-123 months, median 37 months). The rate of surgical site infections occurring during the first postoperative year was 1.4%. Eighty-percent of the infections (n=16) occurred within the first two postoperative months. Two infections were diagnosed two to five years after surgery and two others more than five years after surgery due to hematogenous contamination. All of the observed infections involved a single causal germ. Agents identified were: Gram+ (90%) and Gram- (10%), with a clear predominance for Staphylococcus aureus (n=9). Infections developed 2.1-fold more often in patients with an inflammatory disease (rheumatoid polyarthritis). Age and body mass index did not differ between patients with and without surgical site infection.

The analysis of our series demonstrated the difficulties in conducting long-term surveillance.