Incidence and Risk Factors for Infection When Teicoplanin Is Included for Prophylaxis in Patients with Hip Fracture

Patients and Methods

All patients who underwent open surgery for hip fracture in our institution from June 2012 to September 2013 were retrospectively reviewed. All types of hip fracture including intra-capsular, intertrochanteric and subtrochanteric were retrieved and studied. Relevant patient information was gathered: Demographics, comorbidities, pre-operative performance status (measured by use of the American Association of Anesthesiology [ASA] classification), laterality, type of fracture, type of surgery, delay from admission to surgical intervention, length of surgery, hemoglobin value at admission, urinary or respiratory tract infections during hospitalization, and the need for pre-operative, intra- or peri-operative and post-operative red blood cells (RBC) transfusion.

Antibiotic prophylaxis consisted of 1.5 g of cefuroxime and 800 mg of teicoplanin administered during the induction of anesthesia. Cefuroxime is infused over 5 to 10 min starting 30 min before the surgical incision and teicoplanin over 15 min after the administration of cefuroxime. When length of surgery goes beyond 2 h from induction of anesthesia, a second dose of 1.5 g of cefuroxime is administered without additional subsequent doses. Operations were performed in an operating room with a non-laminar airflow, and throughout the study period there were no variations in the pre-operative washing protocol, the method of skin preparation, the hand hygiene solutions used, the type of sterilization of surgical equipment, the surgical team, surgical techniques, or operating theatres. Screening for S. aureus carriers was not performed, and antibiotic-loaded cement was never used in these patients. The medical history of the patients was reviewed using the electronic records during hospital admission and outpatient clinic visits for 1 y. Superficial and deep infections were defined according to CDC criteria [6]. All patients with a deep infection were taken back to the operating room for debridement, and three to six deep samples of synovial fluid or peri-implant tissue were submitted to the Microbiology Laboratory. The Ethical Committee of our institution approved the study.

Results

A total of 657 consecutive patients were analyzed and the main characteristics of non-infected and infected patients are depicted in Table 1. The mean (SD) age of the cohort was 83.1 (10.4) y and 469 (71.4%) were female. An intra-medullary nail was used in 334 (51%) cases, a prosthesis in 211 (32%) (198 hemiarthroplasties and 13 total arthroplasties), and other internal fixation methods (such as dynamic hip screws or cannulated screws) in 112 (17%). The mean (SD) number of d from admission to surgery was 3.4 (1.9). Thirteen (2.0%) surgical site infections were identified. Six were superficial (0.9%), and seven deep (1.1%). The SSI rate among different surgical approaches was: 2.4% (8 out of 334) in intra-medullary nails, 1.4% (3 out of 211) in prostheses, and 1.8% (2 out of 112) in other synthesis. The isolated micro-organisms according to the type of infection (superficial or deep) are shown in Table 2. Right fractures (3% vs 0.9%, p = 0.04), patients with a hemoglobin ≤10 g/dL at admission (6% vs 1.1%, p = 0.003), patients operated >5 d after being admitted (8.1% vs 1.3%, p = 0.005), when the length of surgery was >120 min (7.1% vs 1.5%, p = 0.01), and intra-operative RBC transfusion (5% vs 0.45%, p = 0.0001) had a substantially greater rate of SSI in the univariate analysis. There was a trend toward a greater SSI rate among patients with diabetes mellitus (4.2% vs 1.7%, p = 0.15), an ASA score of 4 (6.7% vs 1.8%, p = 0.11) and with a urinary tract infection during hospitalization (5.1% vs 1.6%, p = 0.057).

Reviewing the data, patients that were operated beyond 5 d from admission had a rate of intra-operative RBC transfusion of 48.4% (30 out of 62), whereas in those operated within the first 5 d, the transfusion rate was 31.8% (p = 0.01). None of the 32 patients operated later than 5 d from admission and non-receiving intra-operative transfusion had a SSI. In contrast, five out of 30 of those receiving an intra-operative RBC transfusion developed an infection (0% vs 16.7%, p = 0.052). Because the entire risk was attributable to those receiving RBC transfusion, we performed a final model without including surgical delay. Variables independently associated with SSI were intra-operative RBC transfusion (OR: 11.6, CI95%: 2.47–54.54, p = 0.002), a length of surgery >120 min (OR: 4.5, CI95%: 1.25–16.12, p = 0.02), and having a urinary tract infection (OR: 4.28, CI95%: 1.18–15.44, p = 0.02).

Discussion

During the study period, the rate of SSI was 2% and the rate of deep infection was 1.1%. These results are lower than the rate of 4% to 20% reported by other authors in similar populations using the current recommended prophylaxis [3,4,7–9]. A potential explanation for this finding is that we used a broad prophylaxis with a cephalosporin plus teicoplanin that recently has demonstrated good results in primary arthroplasty [10].

The risk factors obtained from our cohort are similar to those previously reported in similar populations [3,8]; however, our analysis deserves some comments. As in previous series, transfusion of RBCs was a potent risk factor [9] but our study identified as a main risk factor transfusion performed intra-operatively or within the first 12h after surgery. Considering all patients transfused intra-operatively (n = 219), they had a SSI rate of 5%, whereas the SSI rate was 0.45% among those not transfused intra-operatively (p < 0.001). In contrast, pre-operative and post-operative transfusion was not associated with SSI. Indeed, a recent clinical trial has demonstrated that liberal post-operative RBC transfusion after cardiac surgery was not associated with greater sepsis or surgical site infection rates than with a restrictive policy [11]. The precise mechanism to explain the detrimental effect of intra-operative RBC transfusions is not clear. A potential explanation could be that intra-operative bleeding and the need of intra-operative RBC transfusion are associated with a reduction of antibiotic tissue concentrations [12] and this could lead to a greater risk of infection. Current guidelines recommend re-dosing the antibiotic when blood loss is excessive [13], but this recommendation is frequently not followed. An alternative explanation may be that storage time of transfused RBC packages could be associated with complications after transfusion [14]. Recently, we observed that transfusion of RBC packages with a storage time >14 d was associated with a greater rate of PJI after primary knee arthroplasty [15]. Unfortunately, for the present study, this information was not available. A recent clinical trial in cardiac surgery did not support this hypothesis; however, they did not specifically analyze intra-operative RBC transfusion and the risk of SSI [16]. In the future, it would be necessary to increase our knowledge on the influence of RBC transfusion on the SSI risk.

The impact of surgical delay in patients with a hip fracture is controversial. Although Westberg et al. [3] found a substantial increase in the risk of PJI when surgical delay was ≥72 h, in our cohort, a surgical delay of >5 d was associated to a greater SSI risk in the univariate analysis. However, the entire risk was among those patients receiving an intra-operative RBC transfusion. A possible explanation for this fact could be that patients under treatment with anti-platelet agents (aspirin, clopidogrel) or oral anti-coagulant drugs (acenocumarol or warfarin) need to wait several d for clearing this medication before being operated on. It is plausible that even after several d without these drugs, these patients have a greater risk of bleeding during surgery, and more frequently requiring intra-operative RBC transfusions. Unfortunately, this information was not available in our database and we cannot confirm this hypothesis.

The main drawback of our study is the retrospective nature of the analysis and the low number of patients with SSI that precludes definitive conclusions about the risk factors for infection. Another limitation is that the revision of medical charts was made according to the electronic system and we could have potentially missed patients transferred to other institutions. However, our hospital is the tertiary referral center for those with SSI, and it is mandatory to transfer them to our center. In conclusion, the SSI rate was 2%, of which 1.1% were deep infections requiring open debridement. Intra-operative transfusion, the length of the surgery, and urinary tract infections were the major predictors of infection. Reducing SSI is of utmost importance for the patients' quality of life and for avoiding additional costs derived from subsequent surgical procedures.