There are approximately 1 million total hip arthroplasties (THAs) performed worldwide per year and more than 250,000 total knee arthroplasties (TKAs).23 These numbers are expected to double from 1999 to 2025 as a result of an ageing society and THAs and TKAs being done at an increasingly early age.24 Infection is one of the major complications in THAs and TKAs. Infection occurs in 1% to 2% of THAs and in 2% to 4% of TKAs.2,3,10,11 Periprosthetic infections are associated with a substantial increase in morbidity, which increases hospital admittance time and adds substantial costs to the health-care system.12,17,28 Treating an infected prosthesis can cost as much as $80,000, which is 4.1 times the cost of a primary prosthesis.12 Periprosthetic infections also prolong total hospital stay by more than 6 weeks.12 Patients with postoperative orthopaedic infections have substantially greater physical limitations and a reduced quality of life.17,28
The presence of a superficial wound infection has been identified as a risk factor for having a periprosthetic infection develop, but the exact extent of the risk is unknown.1,7,22,25 Postoperative superficial wound infections occur far more often than periprosthetic infections and reportedly occur in 1.2% to 17.3% of all patients.1,7,22,26 The discrepancy in percentages is partly related to differing definitions. There are two commonly used definitions of superficial wound infection. The Surgical Infection Study Group defines superficial wound infection on clinical observations without microbiologic confirmation.20 However, The Centers for Disease Control and Prevention requires microbiologic confirmation before diagnosing a superficial wound infection.13 Both groups' guidelines state drain sites should be included, and there should be purulent discharge or a painful spreading erythema.13,24 Despite these definitions, diagnosing a superficial wound infection is subject to personal variations.25
Intraoperative contamination is common in every operating room.5,15 The main sources of intraoperative contamination are the patient's skin and airborne particles from surgical personnel.9,14 In 1982, Whyte et al suggested 2% of the bacterial wound contaminations were from the patient and 98% were caused by bacteria in the air of the operating room.29 In the latter case, 30% reaches the wound directly via the air and 70% reaches the wound via the hands of the surgical personnel or instruments.29
We asked whether bacterial contamination of the instruments and removed bone during primary THA were risk factors for prolonged wound discharge, accounting for other risk factors. Subsequently, we asked whether there was an association between intraoperative contamination, prolonged wound discharge, and periprosthetic infection.
MATERIALS AND METHODS
We prospectively analyzed primary THAs performed from August 2001 to August 2003 in the University of Groningen Medical Center, Groningen, The Netherlands. We obtained written permission from the hospital Ethical Committee. We obtained a representative sample by using a list of random numbers generated by computer to determine whether the protocol would or would not be used for the particular patient. A restriction in the number of patients was used to minimize the burden for the personnel involved, as the protocol was not yet part of standard practice during the study. We included 100 patients because approximately ⅓ of our patients had prolonged wound leakage, which allowed us to use five covariates in multivariate modeling (which was arbitrarily judged to be desirable) without great risk of overfitting.
Patients received antimicrobial prophylaxis (cefazoline, 1000 mg intravenously) 20 minutes preoperatively and postoperative anticoagulation (nadroparine, 0.3 mL subcutaneously combined with acenocoumarol orally). The surgeries were performed in an operating room with a conventional air flow, and the operating team wore disposable impervious drapes. At the end of surgery, drains were placed at the operation site. We documented preoperative parameters thought to influence postoperative wound discharge, which included age, gender, the existence of any immunocompromising disease (eg, rheumatoid arthritis [RA] or diabetes), and body mass index (BMI). We also recorded intraoperative blood loss greater than 400 mL, operating time exceeding 100 minutes, and the use of cement (Simplex® without antibiotics, Stryker Orthopaedics, Mahwah, NJ). There were 33 men and 67 women, with a mean age of 61.3 years (range, 28-87 years; standard deviation [SD], 12.8 years). Thirteen patients had RA and four had diabetes. The mean BMI was 27 (range, 18.5-37.2; SD, 3.7). The mean operating time was 106 minutes (range, 50-180 minutes; SD, 24.8 minutes), and in 48 (48%) patients, the duration was more than 100 minutes. The mean amount of blood loss was 424 mL (range, 40-2000 mL; SD, 269 mL) and exceeded 400 mL in 56 (56%) of the patients. Cement was used in 54 (54%) patients.
Intraoperatively, samples were taken at different stages of the procedure: two from the instruments used, two from the instruments not used, and two from removed bone. The first sample (Culture 1) represented the swab of the smallest unused acetabular broach. After sampling, the reaming procedure was started with this broach. The second sample (Culture 2) represented the swab of the largest unused acetabular broach after reaming. This broach was never used at the prosthesis site. The third sample (Culture 3) represented the swab of the smallest unused femoral broach. After sampling, the reaming procedure was started with this broach. The fourth sample (Culture 4) represented the swab of the largest unused femoral broach after the reaming procedure. This broach was never used at the prosthesis site.
The removed bone chips also were sampled for contamination. Culture I represented the acetabulum and Culture II represented the femur. During all procedures, a clean swab was taken out of the charcoal medium in the operating room, after which it was immediately put back to make sure no contamination occurred during transport and culturing.
The cotton swabs (Cultures 1-4 and the control swab) were transported in a transport medium (Transwab charcoal medium, Medical Wire & Equipment Co, Bath, UK). Removed bone material (Cultures I and II) was placed in sterile cups filled with Tryptone Soya Broth (TSB, Oxoid, Hampshire, UK). Within 2 to 4 hours after sampling, the cotton swabs (1-4) were smeared over blood agar and incubated together with the cups containing bone cultures (Cultures I and II) for 7 days at 37°C; aerobically and anaerobically. After 7 days, the contents of the cups also were smeared over the blood agar and incubated for 5 more days. Instrumentation or bone material was considered contaminated when any bacterial growth was observed. The control swab was negative at all times. The study was performed blind, without informing the orthopaedic surgeon of the test results to ensure all patients were treated regardless of the evaluation.
Wound discharge was recorded postoperatively by a specialized nurse (NB) from the local hospital infection committee who monitored the wound and drain site. The cut-off point was 5 days postoperatively. Patients with a leakage for 5 days or more were the patient group. Patients with a wound and drain site that closed within 4 days postoperatively were the control group. Postoperatively, the drain was removed after 2 days in all patients. The mean duration of wound discharge was 4.2 days (range, 1-28 days; SD, 3.5 days). Twenty-eight patients (28%) had wound discharge for 5 days or longer, with a median duration of 8 days (range, 5-28 days). The wound and drain site had closed within 4 days in the other patients.
To determine whether periprosthetic infection occurred in patients with and without intraoperative contamination and prolonged wound discharge, patients were followed at standard postoperative controls at 6 weeks, 3 months, 6 months, 1 year, and 2 years after index surgery, or if a patient came to the emergency room. At followup we evaluated patients' symptoms along with C-reactive protein, erythrocyte sedimentation rate, and leukocyte count. A prosthesis was considered infected if there was an increase of infection parameters caused by the prosthesis site. This was substantiated by culturing the aspirated joint fluid and/or culturing during revision of the prosthesis.
We performed univariate analyses to assess the associations between the different variables and prolonged wound discharge. We used a Student's t test for independent samples for the continuous variable BMI, and the Pearson chi square test for categorical variables when all cells of the contingency table contained at least five patients. Otherwise, we used Fisher's exact test. We compared the groups with and without prolonged wound discharge. The initial model was based on the results of the univariate analysis and covariates that were judged clinically to be possible confounders. Subsequently, we created a logistic regression model by omitting the most poorly associated covariates. The odds ratios (OR) were transformed to relative risks (RR) with the following formula, with Prev indicating the prevalence of the outcome in the baseline group. For instance, if the parameter ‘Gender (Female)’ is analyzed, the variable is coded 0 = Male and 1 = Female, and the Prev is the proportion of the outcome in the Male group.30
The associations between the different types of cultures and periprosthetic infections were investigated with Pearson chi square or Fisher's exact tests. We also calculated the positive predictive values. The same was performed to investigate the associations between intraoperative contamination, prolonged wound discharge, and periprosthetic infection. All statistical procedures were performed with SPSS version 12.0 statistical software (SPSS Inc, Chicago, IL).
In addition to a positive intraoperative culture, age, RA, use of cement, and increased blood loss were associated (p < 0.05) with prolonged wound discharge (Table 1). In the logistic regression model only a positive intraoperative culture, RA, and increased blood loss were factors (Table 1). Age as a covariate was removed because age was not as directly linked to infection risk as was the use of cement. The RR of intraoperative bacterial contamination was estimated as 2.5, whereas that for RA was 6.4, cement use was 1.6, and increased blood loss was 1.5. The positive predictive values of the instrument swabs for predicting prolonged wound discharge were fairly low (range, 17- 67%), whereas the positive predictive values for the bone chip cultures were much higher (range, 81-90%). We observed an association (p < 0.001) between positive bone chip cultures and the occurrence of prolonged wound discharge (Table 2). Patients with bacterial contamination had a 56% (positive predictive value) chance of having wound discharge develop, whereas only 13% of patients without bacterial contamination had prolonged wound discharge. Bacterial growth occurred in at least one intraoperative culture in 36 patients (36%). One patient had four positive cultures, 11 patients had two positive cultures, and 24 patients had one positive culture.
We found an association (p = 0.008) between intraoperative contamination and the occurrence of a periprosthetic infection, and an association (p = 0.002) between prolonged wound discharge and periprosthetic infection. The positive predictive value of intraoperative contamination and prolonged wound discharge for the occurrence of periprosthetic infection was 25%, whereas its negative predictive value was 98% (Table 3).
Periprosthetic infection occurred in six of the 36 patients with intraoperative contamination. The prolonged wound discharge was monitored postoperatively in 20 patients. Five of the 20 patients subsequently had periprosthetic infections develop (Fig 1). One (four positive cultures) patient had a periprosthetic infection develop within 1 month after the primary surgery. One of the 16 patients with intraoperative contamination in the absence of prolonged wound discharge had an infection develop. In the group of 64 hips without intraoperative contamination, only one hip (1.6%) became infected. The patient had prolonged wound discharge that was monitored postoperatively. In the 56 patients without intraoperative contamination or prolonged wound discharge, periprosthetic infection was not diagnosed during the first 2 years of followup.
Several studies on intraoperative culturing of equipment and bacterial analysis of air samples have yielded conflicting findings on relationships with postoperative infections.5,18,19,21,27 The relationship between prolonged wound discharge and postoperative wound infection and between postoperative wound infection and periprosthetic infection have been reported.1,3,22,25 We describe associations between intraoperative contamination of the operating site (instruments used and bone chips), the occurrence of prolonged wound discharge, and the development of periprosthetic infection. We found an association between intraoperative contamination and prolonged period of postoperative wound discharge, with a positive predicting value of 80% to 90%.
Our study has several limitations. Although we associated prolonged wound discharge with intraoperative contamination, it remains uncertain whether the wound was infected during surgery, in the postoperative period, or just was discharging because of limited ability of the local skin tissue to heal; the latter creating a risk for cross-infection. As another possible limitation, of all possibilities to sample in an operating room,5,18,19,21 we chose to take swabs from the used set of instruments and collected bone chips. We chose these sites as they are the most likely to represent possible contamination of the wound, as was confirmed in our study. The selection and removal of covariates in our model need explanation, as removing covariates does not mean they are (un)important from an etiologic or causal point of view. Age was omitted because of the strong correlation between age and the use of cement, despite the use of cement increasing the immuno- compromising zone surrounding prostheses and further compromising the immune system.8 Age as a covariate was removed because age was not as directly linked to infection risk as with the use of cement. Covariates that were confounders of other relations in this dataset were not deleted from the model. We did not determine the type of bacteria, although this would have highlighted the causal link between intraoperative contamination and later deep periprosthetic infection.
Binary logistic regression showed RA, the amount of intraoperative blood loss, and the use of cement were predictors for prolonged wound discharge after THA. Rheumatoid arthritis10 and extensive intraoperative blood loss25 have already been proven to be risk factors for postoperative wound infection and periprosthetic infection. On the other hand, this study shows that in addition to age, BMI and operating time drop-out were risk factors for prolonged wound discharge. Operating time did not predict prolonged wound discharge, although it is considered a risk factor for wound infection.4,6,10,16,22
Identifying cement as a risk factor might have excluded operating time as a risk factor because these factors are interrelated, like age and use of cement. Inserting an un- cemented prosthesis requires less time than for a cemented prosthesis, decreasing exposure to airborne bacteria in the operating room. It could be the use of cement alone is a more important risk factor than the increase in operating time. Because patients with a high BMI had RA and cemented prostheses, BMI was also a risk factor.
Prolonged wound discharge is important because it can be a risk factor on its own, as well as a potential marker for periprosthetic infection.22,25 Moreover, as measured in this study, if prolonged wound discharge is monitored together with intraoperative bacterial contamination a periprosthetic infection is likely to occur (Fig 1). Alternatively, if prolonged wound discharge is monitored in the absence of intraoperative bacterial contamination, it is important to identify whether one of the other risk factors for prolonged wound discharge exists. Prolonged wound discharge in patients with RA, with a cemented prosthesis, or with more than normal blood loss, does not require immediate additional antibiotic therapy. No periprosthetic infection occurred when prolonged wound discharge and intraoperative contamination were absent (n = 56).
We recommend taking cultures of the wound discharge on Day 5 postoperatively before administering antibiotics and recommend routine intraoperative cultures to provide guidance on whether it is appropriate to initiate immediate antibiotic treatment after prolonged wound discharge.
We thank Bertram The, MD, for assistance with the statistical analysis, and Natalie Boss for monitoring the wound discharge postoperatively.
1. Abudu A, Sivardeen KA, Grimer RJ, Pynsent PB, Noy M. The outcome of perioperative wound infection after total hip and knee arthroplasty. Int Orthop
2. An YH, Friedman RJ. Prevention of sepsis in total joint arthroplasty. J Hosp Infect
3. Berbari EF, Hanssen AD, Duffy MC, Steckelberg JM, Ilstrup DM, Harmsen WS, Osmon DR. Risk factors for prosthetic joint infection: case-control study. Clin Infect Dis
4. Charnley J. Postoperative infection after total hip replacement with special reference to air contamination in the operating room. Clin Orthop Relat Res
5. Davis N, Curry A, Gambhir AK, Panigrahi H, Walker CR, Wilkins EG, Worsley MA, Kay PR. Intraoperative bacterial contamination in operations for joint replacement. J Bone Joint Surg Br
. 1999;81: 886-889.
6. Fitzgerald RH Jr, Nolan DR, Ilstrup DM, Van Scoy RE, Washington JA, Coventry MB. Deep wound sepsis following total hip arthroplasty. J Bone Joint Surg Am
7. Gaine WJ, Ramamohan NA, Hussein NA, Hullin MG, McCreath SW. Wound infection in hip and knee arthroplasty. J Bone Joint Surg Br
8. Gristina AG. Implant failure and the immuno-incompetent fibro- inflammatory zone. Clin Orthop Relat Res
9. Ha'eri GB, Wiley AM. Total hip replacement in a laminar flow environment with special reference to deep infections. Clin Orthop Relat Res
10. Hanssen AD, Rand JA. Evaluation and treatment of infection at the site of a total hip or knee arthroplasty. Instr Course Lect
. 1999;48: 111-122.
11. Harris WH, Sledge CB. Total hip and total knee replacement (1). N Engl J Med
12. Hebert CK, Williams RE, Levy RS, Barrack RL. Cost of treating an infected total knee replacement. Clin Orthop Relat Res
. 1996;331: 140-145.
13. Horan TC, Gaynes RP, Martone WJ, Jarvis WR, Emori TG. CDC definitions of nosocomial surgical site infections, 1992: a modification of CDC definitions of surgical wound infections. Infect Control Hosp Epidemiol
14. Howorth FH. Prevention of airborne infection during surgery. Lancet
15. Hughes SP, Anderson FM. Infection in the operating room. J Bone Joint Surg Br
16. Ilstrup DM, Nolan DR, Beckenbaugh RD, Coventry MB. Factors influencing the results in 2,012 total hip arthroplasties. Clin Orthop Relat Res
17. Kirkland KB, Briggs JP, Trivette SL, Wilkinson WE, Sexton DJ. The impact of surgical-site infections in the 1990s: attributable mortality, excess length of hospitalization, and extra costs. Infect Control Hosp Epidemiol
18. Lawal OO, Adejuyigbe O, Oluwole SF. The predictive value of bacterial contamination at operation in post-operative wound sepsis. Afr J Med Med Sci
19. Lidwell OM, Lowbury EJ, Whyte W, Blowers R, Stanley SJ, Lowe D. Airborne contamination of wounds in joint replacement operations: the relationship to sepsis rates. J Hosp Infect
. 1983;4: 111-131.
20. Peel AL, Taylor EW. Proposed definitions for the audit of postoperative infection: a discussion paper. Surgical Infection Study Group. Ann R Coll Surg Engl
21. Robinson AH, Drew S, Anderson J, Bentley G, Ridgway GL. Suction tip contamination in the ultraclean-air operating theatre. Ann R Coll Surg Engl
22. Saleh K, Olson M, Resig S, Kuskowski M, Gioe T, Robinson H, Schmidt R, McElfresh E. Predictors of wound infection in hip and knee joint replacement: results from a 20 year surveillance program. J Orthop Res
23. Schierholz JM, Beuth J. Implant infections: a haven for opportunistic bacteria. J Hosp Infect
24. Simon LS. Osteoarthritis: a review. Clin Cornerstone
. 1999;2: 26-37.
25. Surin VV, Sundholm K, Backman L. Infection after total hip replacement: with special reference to a discharge from the wound. J Bone Joint Surg Br
26. Taylor GJ, Bannister GC, Calder S. Perioperative wound infection in elective orthopaedic surgery. J Hosp Infect
27. Taylor GJ, Leeming JP, Bannister GC. Assessment of airborne bacterial contamination of clean wounds: results in a tissue model. J Hosp Infect
28. Whitehouse JD, Friedman ND, Kirkland KB, Richardson WJ, Sexton DJ. The impact of surgical-site infections following orthopedic surgery at a community hospital and a university hospital: adverse quality of life, excess length of stay, and extra cost. Infect Control Hosp Epidemiol
29. Whyte W, Hodgson R, Tinkler J. The importance of airborne bacterial contamination of wounds. J Hosp Infect
30. Zhang J, Yu KF. What's the relative risk? A method of correcting the odds ratio in cohort studies of common outcomes. JAMA
. 1998; 280:1690-1691.