Deep infection after total hip replacement can be one of the most devastating complications in orthopaedic surgery. Treatment of such infections is demanding and often involves multiple operations. Although the use of prophylactic antibiotics, laminar airflow operating rooms, and whole-body exhaust suits has been associated with a decreased prevalence of infection12,16,24, the risk of infection remains approximately 1 per cent over the lifetime of the prosthesis, as determined by a review of the results of 3051 total hip arthroplasties26.
In 1993, more than 128,000 primary total hip replacements and more than 26,000 revision procedures were performed in the United States, resulting in approximately 1500 new cases of deep infection19. Because the rate of deep infection is relatively low, most surgeons see few patients who have this complication. When a surgeon does treat an infected hip, he or she is likely to choose a method that seems the most conservative and the most likely to succeed. Historically, this method has been staged revision, which currently is the most commonly used technique in the United States for treatment of infection after a total hip replacement9, despite the associated cost and morbidity.
Recent reports of successful one-stage revisions to treat infection after total hip replacement have increased interest in condensing the operative treatment into a single procedure and a single hospitalization15,20,22,28. As methods of treatment and of patient selection have been refined, there has been a concomitant increase in the reported rates of success, although few reports have included true long-term follow-up15,22. The reported rates of success of one-stage exchanges that involve perioperative antibiotic treatment, meticulous operative technique, and use of antibiotic-impregnated cement have often been as good as those reported for staged procedures3,15,17,20,22,28. If the results of one-stage and two-stage revisions are comparable, the advantages of a one-stage procedure, such as decreased patient morbidity, less cost, improved mechanical stability of the limb, and a shorter duration of disability, weigh in its favor.
The purpose of the present study was to evaluate the long-term safety and efficacy of one-stage direct exchange for the treatment of infection after total hip replacement.
*No benefits in any form have been received or will be received from a commercial party related directly or indirectly to the subject of this article. Funds were received in total or partial support of the research or clinical study presented in this article. The funding source was the Los Angeles Orthopaedic Hospital Foundation.
†Asheville Orthopaedic Associates, 111 Victoria at Oakland Road, Asheville, North Carolina 28801.
‡Joint Replacement Institute, 2400 South Flower Street, Los Angeles, California 90007.
§28800 Ryan Road, Suite 120, Warren, Michigan 48092.
We prospectively followed twenty consecutive patients (sixteen men and four women) who had had a total of twenty direct-exchange reimplantations performed by one of us (H. C. A.), between October 1979 and July 1990, because of failure of a total hip replacement due to infection (Table I). The average age at the time of the revision operation was 61.4 years (range, thirty-eight to eighty-five years). The primary reason for the initial total hip replacement was osteoarthrosis in ten patients, posttraumatic osteoarthrosis in three, acute fracture in two, congenital dysplasia in two, avascular necrosis in two, and slipped capital femoral epiphysis in one. The average interval from the primary hip arthroplasty to the direct exchange was fifty-three months (range, 6.6 to 148 months).
Pain was the predominant clinical finding; it was present in all twenty patients. Three patients had a chronic draining sinus, and two had a persistent fever. The average erythrocyte sedimentation rate was thirty-six millimeters per hour (range, five to sixty-one millimeters per hour; normal, two to twenty millimeters per hour). Seventeen patients had an elevated erythrocyte sedimentation rate; one of the three patients who had a normal rate had been taking antibiotics orally for more than a year.
Loosening of the prosthesis was evident on plain anteroposterior and lateral radiographs of all patients. Preoperative bone scans also were made for fifteen patients: nine had a technetium-99 bone scan, five had an indium-111 scan, and one had both. The findings on eleven scans (three of the six indium scans and eight of the ten technetium scans) were interpreted as being consistent with infection. Five patients did not have a preoperative scan.
The presence of infection was confirmed by a positive culture of fluid obtained by aspiration of the joint preoperatively or by a positive culture of joint fluid obtained intraoperatively, or both, and by a finding of acute or chronic inflammation, as determined with use of the criterion of more than five polymorphonuclear leukocytes per high-power field established by Mirra et al., on histological examination of tissue obtained at the time of the operation. The most common infecting organism was Staphylococcus epidermidis (nine patients), followed by Streptococcus species and Staphylococcus aureus (five patients each) (Table II). No organism grew on the preoperative culture of specimens from one patient; however, Staphylococcus aureus and Staphylococcus epidermidis grew on culture of joint fluid obtained at the time of the operation and acute inflammation was demonstrated on examination of frozen and permanent sections.
The operative technique included meticulous, thorough débridement to remove all infected tissue and bone as well as acrylic cement. Biopsy specimens were obtained from the areas that appeared to have the most inflammation and were sent for immediate frozen-section analysis. A trochanteric osteotomy was performed in eighteen patients. The femoral canal was inspected carefully with the aid of a head-mounted fiberoptic light and was debrided with long curets and a high-speed burr. Copious pulsatile lavage with several liters of normal saline solution was carried out with use of a long femoral-canal irrigator; this was followed by irrigation with antibiotic solution (polymyxin-bacitracin in saline solution). The canal was sealed with an acrylic plug containing antibiotics and was dried. Antibiotic-impregnated cement was then packed into the canal, and the implant was inserted. The average duration of the anesthesia was 263 minutes (range, 190 to 300 minutes). Blood loss averaged 2200 milliliters (range, 825 to 5000 milliliters). Closed suction drainage was used, and an elastic bandage hip-spica wrap was applied to minimize the so-called dead space.
Antibiotics were added to the cement (Simplex P; Howmedica, Rutherford, New Jersey) for all patients. The type of antibiotic (Table II) was selected on the basis of the findings on preoperative culture. These findings were also used to select the antibiotics to be administered parenterally, beginning intraoperatively after tissue and fluid had been obtained for culture. If no organism had grown on preoperative culture, vancomycin was administered to provide broad-spectrum coverage. When the results of the intraoperative cultures (Table II) and the sensitivities had been determined, the parenteral antibiotic therapy was modified, when necessary, with the aid of an infectious-disease consultant. The antibiotics were administered parenterally for an average of five weeks (range, two to eighteen weeks) postoperatively and then were taken orally for an average of 4.7 months (range, 2.5 to 6.0 months). (The duration for which one patient took antibiotics orally was not known.)
Postoperatively, each patient was followed with serial radiographs, determinations of the erythrocyte sedimentation rate, complete blood-cell counts, and physical examinations. Pain, walking, function, and activity were assessed with use of the hip-rating system of the University of California at Los Angeles1. With this system, each parameter is evaluated on a scale of 0 to 10 points (with 0 being the worst rating and 10 being the best).
Fifteen patients were followed clinically for an average of 10.9 years (range, 5.8 to 17.1 years). One patient died 3.5 years postoperatively from causes unrelated to the hip procedures. There was no evidence of infection at the time of death. The minimum duration of follow-up for the remaining nineteen patients was five years, and the average duration was 10.3 years (range, 5.4 to 17.1 years). Four other patients died of causes unrelated to the direct-exchange operation. The average interval between the operation and the time of death of the five patients who died was 7.2 years (range, 3.5 to 13.6 years). Family members of these five patients reported that the hip replacement was functioning satisfactorily, without symptoms suggesting recurrent infection, at the time of death. The last clinical and radiographic follow-up for these patients was performed at an average of five years (range, 3.2 to 7.0 years) postoperatively.
For the past two decades, a two-stage approach has usually been the method of choice for the treatment of infection after a total hip replacement in the United States9. However, we are not aware of any conclusive evidence that the delay between stages is useful, and organisms often remain in the operative field despite the delay2. Indeed, the delay can often result in a more difficult second procedure because of extensive scarring, shortening of the limb, and loss of bone density due to disuse. In addition to the extensive morbidity associated with a delayed exchange, there is increasing concern regarding the cost.
In 1981, Buchholz et al. reported the results of one-stage revision with use of antibiotic-loaded cement to treat deep infection in 583 patients. The overall rate of success after the first exchange was 77 per cent, and the rate of success after additional exchange procedures was 90 per cent. This success, as well as that reported by others3,17,20, encouraged additional attempts at direct exchange.
The present study of direct-exchange arthroplasties was begun after favorable results were reported for an earlier series of thirty-three patients who had had a one or two-stage exchange for the treatment of active infection after open reduction with internal fixation, hemiarthroplasty, cup arthroplasty, or total hip replacement4. Despite the 100 per cent rate of success, with regard to eradication of the infection, in the present study, the 95 per cent confidence interval is 85 to 100 per cent because of the relatively small number of patients. However, even the rate at the low end of the confidence interval (85 per cent) compares very favorably with the success rates that have been reported after delayed exchange; many studies of delayed exchange have also involved a small number of patients and confidence intervals have not been given5,10,13,18,25. Meaningful comparison of the results of the present study with those of previous studies is complicated further by many uncontrolled variables, differences in the patient populations, differences in the antibiotics that were used (both in the cement and perioperatively), and lack of information. Several authors have emphasized that no study of infection around joint implants is completely valid until the patients have been followed for four or five years because of the possibility of latent infections6,14. We are aware of eight series10,13,15,18,20,22,25, including the present one, with a follow-up interval of at least four years, and these studies provide information with which comparisons can be made (Table IV).
Because of the poor symptomatic and functional results reported after resection arthroplasty and the increased difficulty associated with delayed reimplantation procedures, Duncan and Masri developed a temporary antibiotic-loaded facsimile of a hip prosthesis, which they called the PROSTALAC (prosthesis of antibiotic-loaded acrylic cement). In 1989, they introduced the prosthesis into clinical trials and later reported a 93 per cent rate of success in forty-six hips that had been followed for at least two years8. Those authors concluded that there is no obvious risk or disadvantage associated with this temporary prosthesis. We believe that the success reported by Duncan and Masri argues in favor of direct exchange, given that both procedures involve use of antibiotic-impregnated cement to coat all but the articular and periarticular surfaces of the prosthesis. The obvious advantages of direct exchange compared with the PROSTALAC system are that only one procedure is needed in most patients and the variable interval between stages, during which the function of the hip is less than optimum, is avoided. A potential advantage of the PROSTALAC system and other staged techniques is that they allow for revision with adjuvant bone-grafting and components inserted without cement, should they be necessary, at the time of the second exchange, with a chance of less risk.
Fitzgerald reported that the success of a direct exchange arthroplasty may be influenced by the infecting organism's ability to elaborate a glycocalyx or to form a biofilm layer. However, there is little available information with which to determine whether an organism's ability or inability to produce a glycocalyx notably influences the success of a one or two-stage exchange arthroplasty.
It is unrealistic to expect any technique or protocol for revision total hip arthroplasty to eradicate infection all of the time. Colyer and Capello reported the results of two-stage reimplantation in thirty-seven patients who had had an infection around a hip implant (the implant was an internal fixation device in four patients). The two-stage reimplantation involved a total of three operations: a second débridement was performed approximately one week after the first operation, followed by the reimplantation approximately one month later. Those authors reported an 84 per cent rate of success, which they said was "comparable to those in the literature using either one or two-stage revision."
It is important to note that when a patient is given a choice between one operation or at least two, with the expectation of roughly comparable results, not surprisingly he or she chooses to have the one operation. When patients are fully informed about the risk involved in a direct exchange that is performed because of infection around a total hip replacement, they prefer to take that risk rather than to face the possibility of being left with an unstable joint after a resection arthroplasty that, due to unforeseen developments, cannot be followed by reimplantation23. With proper selection of patients, thorough débridement, and use of antibiotic-impregnated cement and postoperative antibiotics, the rate of successful direct-exchange arthroplasties has been reported to be at least 89 per cent (fifty-five of sixty-two20)15,28. The satisfaction expressed by most patients who have a successful outcome after direct exchange is in marked contrast to the feelings expressed by a patient in whom the second-stage reimplantation cannot be performed after the resection arthroplasty.
We attribute our success with direct exchange to several factors. First, we performed extensive and meticulous débridement. It is impossible to excise all contaminated tissue and bone stock; however, it is unrealistic to expect resolution of an infection without reduction of the bacterial count to an absolute minimum through removal of as much infected tissue, cement, and dead bone as possible. While this factor seems obvious, its importance cannot be overstated. The removal of cement and infected tissue is tedious and time-consuming. In our series, the average duration of the anesthesia was approximately 263 minutes. A fiberoptic headlamp was used for all débridements, and a trochanteric osteotomy was performed in eighteen patients to provide adequate visualization and to allow access for the removal of cement and dead bone. The sciatic nerve was identified and tagged to minimize the risk of injury during extensive débridement of the surrounding soft tissues. The procedure also included copious pulsed lavage with several liters of saline solution, followed by irrigation with antibiotic (polymyxin-bacitracin) solution.
Second, we used antibiotic-impregnated polymethylmethacrylate cement, with the type of antibiotic based on the findings of the preoperative culture whenever possible.
Third, we administered antibiotics intraoperatively (beginning after specimens had been collected for culture) and for a long period postoperatively. Parenteral therapy, which lasted for an average of five weeks, was provided with the aid of an infectious-disease consultant and was followed by oral therapy, which continued for an average of 4.7 months.
Finally, we performed the procedure only in properly selected patients. Although this was a consecutive series of patients who met our inclusion criteria, we do not recommend direct exchange for patients who are immunocompromised; for those who have an infection with a known resistant gram-negative or methicillin-resistant organism; or for those who have a major skin, soft-tissue, or osseous defect that makes it impossible to obtain a closed wound or a stable implant.
Decreased morbidity, decreased cost27, and increased patient satisfaction all argue in favor of direct exchange if the rate of success is similar to that for delayed exchange. Our experience suggests that direct exchange can yield a rate of success comparable with that of delayed exchange if antibiotic-loaded cement and appropriate postoperative antibiotics are used; the organism is of low virulence; the surgeon is experienced; and meticulous débridement and extensive irrigation, including both pulsed lavage and irrigation with solution containing two antibiotics, are performed. Proper selection of patients, as detailed in this study, is important. Additional investigation is needed to determine whether gram-positive organisms that elaborate a glycocalyx cause a substantially higher risk of reinfection after exchange arthroplasty.