Swiss Medical Weekly (Aug 2012)
Prevention and control of surgical site infections: review of the Basel Cohort Study
Abstract
Introduction: Surgical site infections (SSI) are the most common hospital-acquired infections among surgical patients, with significant impact on patient morbidity and health care costs. The Basel SSI Cohort Study was performed to evaluate risk factors and validate current preventive measures for SSI. The objective of the present article was to review the main results of this study and its implications for clinical practice and future research. Summary of methods of the Basel SSI Cohort Study: The prospective observational cohort study included 6,283 consecutive general surgery procedures closely monitored for evidence of SSI up to 1 year after surgery. The dataset was analysed for the influence of various potential SSI risk factors, including timing of surgical antimicrobial prophylaxis (SAP), glove perforation, anaemia, transfusion and tutorial assistance, using multiple logistic regression analyses. In addition, post hoc analyses were performed to assess the economic burden of SSI, the efficiency of the clinical SSI surveillance system, and the spectrum of SSI-causing pathogens. Review of main results of the Basel SSI Cohort Study: The overall SSI rate was 4.7% (293/6,283). While SAP was administered in most patients between 44 and 0 minutes before surgical incision, the lowest risk of SSI was recorded when the antibiotics were administered between 74 and 30 minutes before surgery. Glove perforation in the absence of SAP increased the risk of SSI (OR 2.0; CI 1.4–2.8; p 1 year after surgery. More than eighty patient and procedure variables were recorded, including age, sex, American Society of Anaesthesiologists (ASA) score, type of procedure, surgical team members, division of surgical speciality, timing of SAP, wound class, duration of surgery, compromised asepsis, and many other known and suspected SSI risk factors. We used an electronically readable form created by Cardiff TELEForm Software (Cardiff TELEForm Desktop V 8.0, 2002, Verity Inc., Sunnyvale, CA) to export the data into an Excel file (Windows Microsoft Excel 2003, Microsoft Corporation). The primary predictor variable of the first analysis was SAP timing. We divided time into intervals, and chose the cutoffs of 120 and 0 minutes before surgery on the basis of the findings of Classen et al. [7]. SAP was administered intravenously by the anaesthetists via single-shot infusion of 1.5 g cefuroxime in 20 ml sodium chloride solution over a few minutes in combination with metronidazole (500 mg, intravenous, 5 minutes) in colorectal patients. The exact time in minutes when the infusion of SAP ended was prospectively recorded by the anaesthesia team. The main predictor variable of the second analysis was compromised asepsis due to glove perforation. The operating room nurse was responsible for the detection and registration, either directly when glove perforation itself was visible or indirectly when liquid was detected inside a glove. When double gloves were used, leakage of the inner glove was noted. The main predictor variables of the third analysis were perioperative ABT and preoperative anaemia. Perioperative ABT was performed by the anesthesia team in the operating room. The number of allogeneic blood units was monitored. All blood donations were leukocyte-depleted (leukocytes <1.0 × 106/unit) by specific filtration (Optipure RZ 2000 filters, Baxter) and plasma was removed. Packed red cells were stored in saline-adenine-glucose-mannitol at 4 °C. Preoperative anaemia was defined as <120 g/l haemoglobin. For analysis of the impact of surgical training on the incidence of SSI, surgeries were divided into two groups: tutorial assistance or autonomous surgery. Tutorial assistance was defined as surgical training of a resident by a board-certified surgeon, or operations conducted by an inexperienced general surgeon supervised by a board-certified surgeon with extensive expertise in the field. For analysis of the economic burden of in-hospital SSI, the outcome variables were total duration of hospitalisation, duration of hospitalisation after surgery, duration of intensive care stay after surgery, duration of in-hospital use of antibiotics, patient charges and hospital costs. Since the collection of cost data was not part of the observational cohort study design, this information was derived from the hospital’s finance department. Their computerised internal cost and activity accounting database directly linked internal hospital costs with patient charges. Control patients had no SSI and were matched to case patients by age, procedure code, and National Nosocomial Infection Surveillance (NNIS) risk index. To assess the quality of in-house SSI surveillance by surgeons during the study period, an infection control team (ICT) evaluated all general surgery patients by full chart review and by gathering additional clinical information. Data from the surgeons’ surveillance system were available for the ICT. The ICT consisted of a board certified infectious diseases specialist and a hospital epidemiologist. Microbiological evaluation consisted of microscopic direct preparation and cultures with antibiotic resistance testing. As superficial wound swabbing is difficult to interpret, the decision to treat superficial SSI with antibiotics was made clinically rather than on the basis of the results of a wound swab. All analyses were conducted at the Institute of Social and Preventive Medicine in Bern using Stata version 9.2 as well as version 10 (Stata Statistical Software; Stata Corp., College Station, TX). In the first step, univariable analyses were performed to assess the association between the primary predictor variable and the unadjusted likelihood of SSI occurrence. Then multivariable logistic regression models were used to examine the risk-adjusted association between the primary predictor variable and the odds of contracting SSI. The decision on what variables to include in the regression model was based on their potential role as SSI risk factors or on indications of differences in distribution. Crude and adjusted odds ratios (OR), 95% confidence intervals (CI), and P values were used to express the strength of this association. All P values were two-sided and statistical significance was set at the 0.05 level.
Keywords
