After review and approval of this study by our institutional review board, our joint registry database was utilized to identify all patients who underwent revision elbow arthroplasty at our institution between 2000 and 2007. The records for all of these elbows were reviewed to identify those with positive cultures of samples obtained at the time of revision surgery. Twenty-five of the 213 revision elbow arthroplasties performed at our institution during the study period had a positive culture of at least one sample obtained at the time of revision surgery.
Nine of these twenty-five elbows that underwent revision surgery had a previously documented history of infection and were excluded from the study: five elbows underwent two-stage reimplantation after resection arthroplasty and a full course of parenteral antibiotics, two elbows underwent single-stage exchange arthroplasty for a confirmed deep periprosthetic infection, one elbow had a history of subclinical periprosthetic joint infection diagnosed by a positive culture during the implantation of the primary elbow prosthesis and had been on lifelong suppressive antibiotic therapy until the time of the current index revision surgery, and the remaining elbow had a PROSTALAC (prosthesis with antibiotic-loaded acrylic cement)-like implant inserted during the current revision for a persistent recurring prosthetic joint infection.
The remaining sixteen patients with a positive intraoperative culture specimen had no prior documented history of infection in the elbow since the primary elbow arthroplasty had been performed. In addition, none appeared infected preoperatively, and none had overt clinical signs or symptoms of infection at the time of their index revision surgery. The positive intraoperative culture specimens discovered postoperatively were unexpected for these sixteen elbows, and these patients form the subjects of our study. Since this report is the first, to our knowledge, to discuss this topic, there are no preexisting standards to otherwise characterize or define this patient group.
We retrospectively reviewed the records and radiographs of the sixteen patients. Preoperative, intraoperative, and postoperative data pertaining to our study were obtained. Twelve of the sixteen patients had more than two years follow-up after the index revision, and the remaining four had less than two years of surveillance. Using the Mayo Elbow Performance Score, we determined the status and function of the involved elbow of these twelve patients during their most recent follow-up evaluation21.
Source of Funding
No source of external funding was used for any aspect of this study.
Patients
Intraoperative culture specimens that were unexpectedly positive were obtained from sixteen of the 213 revision elbow arthroplasties performed at our institution between 2000 and 2007, for a prevalence of 7.5%. There were twelve female and four male patients, with an average age of 60.3 years (range, forty-one to seventy-six years) at the time of revision surgery. Eight of the revisions were performed on a left elbow, while the other eight were on a right elbow.
The etiology leading to the initial primary elbow arthroplasty was rheumatoid arthritis in eight elbows, posttraumatic osteoarthritis in four elbows, failed fixation or reconstruction after elbow trauma in two elbows, primary osteoarthritis in one elbow, and familial osteochondromatosis in one elbow (Table I).
The average time between implantation of the primary arthroplasty components and the current index revision surgery was ninety-six months, with the earliest revision done just eighteen months after the primary elbow arthroplasty and the latest done after twenty-two years. In eleven of the sixteen elbows, the current index revision was the first revision procedure, whereas four had undergone one revision procedure prior to the current index revision surgery, with the duration between the first and current revision ranging from nineteen to 122 months. The remaining elbow had undergone two previous revision procedures, with the current revision procedure occurring forty-two months after the second revision procedure.
The indication for revision surgery was symptomatic loosening in fifteen elbows. Of these fifteen elbows, three had associated periprosthetic fractures, while one had an associated dislocated radial head prosthesis. In two, worn bushings were noted but were not the primary reason for revision. The indication for revision surgery in the remaining patient was a fracture of the ulnar component.
Preoperative Findings
None of the sixteen patients presented with overt clinical signs or symptoms of infection at the time of revision elbow arthroplasty. One elbow (Case 12) had a superficial granulating ulcer over the involved elbow; this elbow was aspirated to exclude infection and the culture of the aspirated fluid was negative. A second elbow (Case 5) had undergone implant resection because of the gross appearance of the tissues at the time of revision surgery. The intraoperative cultures were negative, and this elbow was reimplanted three days after the resection.
The majority of our patients had preoperative blood studies performed. Fifteen of the sixteen patients had white blood-cell counts and neutrophil counts available preoperatively, and only ten had erythrocyte sedimentation rate and C-reactive protein results available. Laboratory values that were considered to be elevated were a white blood-cell count of >11.0 × 109/L, a neutrophil count of >7.0 × 109/L, an erythrocyte sedimentation rate of >29 mm/hr, and a C-reactive protein level of >8.0 mg/L. On the basis of these threshold values, two patients had elevated white blood-cell counts, three had elevated neutrophil counts, three had an elevated erythrocyte sedimentation rate, and six had an elevated C-reactive protein level. Five of these six patients with elevated C-reactive protein levels had been diagnosed with rheumatoid arthritis. The average white blood-cell count was 7.41 × 109/L (range, 3.8 to 11.7 × 109/L), the average neutrophil count was 4.67 × 109/L (range, 2.01 to 9.05 × 109/L), the average erythrocyte sedimentation rate was 24.4 mm/hr (range, 3 to 68 mm/hr), and the average C-reactive protein level was 15.9 mg/L (range, 3 to 54.5 mg/L).
Five elbows underwent a preoperative aspiration of the elbow on the basis of the surgeon’s preference, despite the absence of overt symptoms or signs of infection. One of the five aspirations was a “dry” tap, and the four remaining successful aspirations yielded a negative finding on culture. Two patients also underwent a preoperative indium white blood-cell scan; one was negative, whereas the other showed increased tracer uptake.
Intraoperative Findings and Surgical Technique
All sixteen patients underwent revision elbow surgery by two experienced elbow surgeons at our institution, and all had preoperative prophylactic intravenous cefazolin administered. Eight patients had three to five doses of the intravenous antibiotics continued into the immediate postoperative period. On the basis of the surgeon’s intraoperative assessment, none of the sixteen patients had features highly suspicious of infection (Table II). Nevertheless, eleven had frozen-section histological assessment of an intraoperative tissue specimen, of which all were negative for acute inflammation, with evidence of a “postarthroplasty effect” characterized by a predominance of chronic inflammatory cells and an absence of a large number of polymorphonuclear neutrophils. Six patients underwent revision of both ulnar and humeral components, five required revision of only the humeral component, and four required revision of only the ulnar component. One to two grams of vancomycin per 40 g of bone cement has been used at our institution since 1989 and was used in all sixteen of these patients.
Culture Results
Fifteen elbows had intraoperative tissue specimens sent for culture, and the remaining elbow had the removed implant bushings sent for sonication and culture. The number of tissue specimens assessed was a function of the clinical concern. Our policy is to always send at least one sample for all patients having a revision. Aerobic cultures are reported at two days and finalized by three days. Anaerobic cultures are read at five days and finalized at seven days. In this sample, the number of specimens sent for culture varied among the patients, with an average of two specimens per elbow; three elbows had three intraoperative specimens sent for culture, ten had two specimens sent for culture, and the remaining three patients had one culture specimen sent. Fifteen elbows had one specimen yielding a positive culture, and one patient had two specimens yielding positive cultures, giving a total of seventeen positive culture specimens in sixteen elbows.
Ten of the organisms were identified on a solid medium, while the remaining seven revealed growth in broth (liquid medium) only. Eight specimens grew coagulase-negative Staphylococcus epidermidis (all but one cultured on solid medium), six specimens grew Propionibacterium acnes (all cultured in broth), while the remaining three yielded uncommon organisms: Burkholderia (Ralstonia) pickettii, Corynebacterium species, and a mixed growth of Streptococcus viridans and Lactobacillus species.
Postoperative Assessment and Antibiotic Therapy
On report of the positive cultures, consultation with an infectious disease specialist was obtained. A combined decision was made with regard to whether the culture result was likely spurious (false positive) because of contamination or that it represented a true infection. The decision to treat or not treat the patient was individualized on the basis of the surgeon’s interpretation of the clinical and laboratory data, and after consultation with an infectious disease physician specializing in musculoskeletal infections.
The positive cultures were considered to be spurious in thirteen of the sixteen elbows, largely on the basis of these factors: broth cultures, culture of an uncommon pathogen, or a nonsuggestive clinical presentation. None of these thirteen patients received any formal long-term antibiotic treatment, although three of the thirteen did have short courses (less than two weeks) of oral cephalexin.
Two elbows (Cases 1 and 16) were considered to have an infection, on the basis of the isolation of a known pathogen (Staphylococcus epidermidis in both), growth on a solid culture medium, and a history of unexplained early loosening after primary arthroplasty in both (see Appendix). Both were diagnosed with loosening at eighteen and twenty-four months, respectively, after the primary total elbow arthroplasty. Both patients were treated with long-term antibiotic suppression.
The remaining elbow (Case 15) presented on the fifth postoperative day after the index revision with acute postoperative periprosthetic infection, even prior to the report of the “unexpected” positive index culture. This patient underwent early surgical debridement, and multiple cultures during the debridement yielded Staphylococcus aureus, which was different from the organism isolated during the index revision surgery five days earlier (Staphylococcus epidermidis). He subsequently underwent multiple debridements followed by resection arthroplasty and staged reimplantation fifteen months later.
Final Outcome
Twelve of the sixteen patients had more than two years of follow-up, with an average follow-up period of sixty-five months (range, thirty-two to 121 months). Among these twelve patients, ten had culture results that were thought to be contaminants, one was thought to have an infection, and the remaining patient was the one who presented with an acute postoperative periprosthetic infection.
Among the ten patients within the group considered to have a contaminant, nine remained infection-free despite receiving no definitive long-term antibiotic therapy. The remaining patient (Case 8) presented with an infection three months after the index revision and underwent surgical debridement, bushing exchange, and component retention, followed by lifelong antibiotic therapy. Multiple culture samples taken during the debridement yielded Staphylococcus aureus, which was different from the organism isolated at the time of the index revision (Propionibacterium acnes).
The only patient (Case 16) among the twelve with adequate follow-up who was deemed to have an infection after the unexpected culture was reported, and who was treated with lifelong antibiotic suppression, had symptomatic loosening eight years after the index revision (see Appendix). Because of a strong suspicion of infection, she underwent resection arthroplasty and six weeks of intravenous antibiotics, followed by reimplantation and lifelong antibiotic therapy. Multiple culture samples obtained at the time of resection were positive for the same microorganism isolated during the index revision eight years earlier (Staphylococcus epidermidis).
Of the four patients with less than two years of follow-up, three were clinically determined to have a positive culture because of contaminants, whereas the remaining patient was thought to have an infection and was treated with long-term antibiotic suppression. None had recurrence during the short time of follow-up.
The most recent average Mayo Elbow Performance Score was 85.8 points. According to this score, seven had an excellent outcome, three had a good outcome, one had a fair outcome, and one had a poor outcome. The patient with the poor outcome was one of the three patients who had recurrent infection.
One of the most worrisome and unanswered questions in revision arthroplasty concerns the implications and management of an unexpected positive intraoperative culture result. The liberal use of intraoperative culture specimens during revision has led to a growing number of unexpected positive cultures16-20. The clinical scenario of discovering an unexpected positive result on intraoperative culture after revision arthroplasty in a joint with no other symptoms or signs of infection represents a management dilemma and has not been previously discussed, as far as we know, with regard to elbow surgery.
This issue has been discussed extensively with regard to hip and knee arthroplasty, and various diagnostic criteria have been proposed in an attempt to accurately diagnose a periprosthetic joint infection22-24. However, there is no consensus on how to manage an unexpected positive culture when infection was not clinically suspected. In our study, the prevalence of unexpected positive results on intraoperative cultures in the setting of a revision elbow arthroplasty was 7.5% (sixteen of 213 elbows). Similar positive culture results have been reported to occur during 5.9% to 30% of revision arthroplasties of the hip, knee, and shoulder16-20.
As alluded to by other authors, the so-called inflammatory markers were not predictive of a positive intraoperative culture18,20. The problem is compounded in our study by the fact that rheumatoid arthritis is a common underlying diagnosis in patients undergoing elbow arthroplasty4,5. In fact, five of our six patients with elevated C-reactive protein levels had underlying rheumatoid arthritis. We believe that these markers should not be used independently as a criteria for making the diagnosis of a periprosthetic joint infection without due consideration of the culture result and characteristics, as well as the clinical presentation of the patient.
Several authors have reiterated the fact that the positive predictive value of a positive culture of a preoperative aspirate is much higher if it is used as a confirmatory test in a patient with clinically suspicious findings rather than as a screening test for infection8,25. Any suggestion of an infection was an exclusion criterion in this study.
Intraoperative frozen-section histological analysis was also performed for the majority of our patients. All specimens were negative for histological evidence of acute inflammation. Despite having a high specificity rate, intraoperative histological analysis has a very low sensitivity rate and is thus not a very effective screening tool for the diagnosis of periprosthetic infection during revision elbow surgery26,27.
In our study, an average of two intraoperative culture samples was obtained. There is no consensus in the literature on the ideal number of culture specimens that should be taken during revision arthroplasty7,8,26,28. We know there is a tendency to submit fewer specimens for study when the concern of infection is low26. To decrease costs, we do not routinely send more than one or two tissue samples for assessment in reoperations when the cause of the failure is known or thought to be known.
Coagulase-negative Staphylococcus epidermidis and Propionibacterium acnes were the most common pathogens isolated in our study, as was the case in other published series in the literature16,18-20. It is of interest that the majority of the culture specimens with Staphylococcus epidermidis were grown on a solid medium, whereas all of the culture specimens with Propionibacterium acnes were identified in broth. The latter is often implicated as an infectious organism in the shoulder18,20.
The immediate and most vexing question to be answered is whether the identification of bacteria should dictate treatment for infection. In our practice, the answer is based on a few considerations. First, we consider the clinical presentation of the patient: infection is suspected when patients complain of pain at rest and/or at night, develop early implant loosening without explanation, or have noticed progressive loss of motion. Second, we consider the presence of any suggestive laboratory blood studies. Last, the culture medium (solid or broth) and the organism isolated (a classic pathogen or contaminant) are considered.
On the basis of these considerations, a diagnosis of “likely contaminant” was made for thirteen of the sixteen elbows, and all thirteen were managed expectantly and successfully without long-term antibiotic suppression. We believe it is not advisable to commit to long-term antibiotic treatment in those with little likelihood of having an infection. Hence, we do not simply treat on the basis of the culture report. Of interest, one of these patients presented with subacute infection three months after the index revision. Culture samples obtained during debridement yielded Staphylococcus aureus, whereas the isolate at the initial revision was Propionibacterium acnes. This suggests that, despite the development of infection, the initial unexpected culture from the index revision was likely spurious.
Two of the sixteen patients were classified as having infection on the basis of a few factors: unexplained early loosening after primary elbow arthroplasty (eighteen and twenty-four months postoperatively in our series), isolation of a known pathogen (coagulase-negative Staphylococcus epidermidis), and growth on solid culture medium. Unfortunately, only one of these two patients was available for follow-up. This patient (Case 16) presented with infection eight years later with the same pathogen, suggesting that the organism isolated during the index revision represented an infection. Of note, this patient was one of those on long-term antibiotic suppression.
Other authors have also found the majority of unexpected positive cultures in the revision setting to be false positives19,16,20. Despite this observation, unexpected positive cultures at the time of revision elbow surgery should not be disregarded without due consideration, as a minority of them do ultimately manifest themselves as infections. Perhaps it should be included as another category in the existing classification for infections after total elbow arthroplasty4,5.
Yet, simply adopting the “just to be sure” treatment mentality, as suggested by authors of smaller series17,18, and administering long-term antibiotic therapy for any patient with a positive culture without careful consideration of the clinical and microbiological findings, appear unfounded. Adopting such an approach is clearly not trivial, as it would invariably subject the majority of the patients with a false-positive result to an unnecessary prolonged and costly course of treatment therapy, which is also not without adverse effects. More importantly, it has the potential of breeding multidrug-resistant organisms.
Despite the fact that all broth cultures of Propionibacterium acnes were classified as contaminants in our series of patients, and taking into account the notorious reputation of this pathogen in the setting of revision shoulder arthroplasty18,20, we reemphasize the fact that a broth culture positive for Propionibacterium acnes should not be dismissed as a contaminant without due consideration.
Our study has several limitations. First, it was a retrospective study. Second, our absolute study population was small and had a short minimum duration of follow-up. Third, the preoperative evaluation of the patients did not follow a rigorous and consistent protocol, which makes it difficult to make any definitive statements about the use of these inflammatory tests in the types of cases that were studied. Last, we realize that the management protocol we adopted is largely based on the experience of the senior author (B.F.M.), and thus may not be universally accepted.
In conclusion, in our practice, there was a 7.5% chance of encountering an unexpected positive result on intraoperative culture when we performed a revision elbow arthroplasty in patients for whom there was a low suspicion of infection. The majority of these unexpected positive cultures were appropriately determined to be spurious. A minority of unexpected positive cultures will unfortunately be infections, which continues to challenge the decision-making process. On the basis of our experience, our current practice is to consider the following: unexplained early loosening after primary elbow arthroplasty with pain at rest or at night, isolation of a known pathogen, and identity on a solid culture medium. We withhold antibiotic therapy in the absence of these features, which will be the case in the majority of our patients with an unexpected positive result on intraoperative culture specimens during revision elbow arthroplasty.
Disclosure: None of the authors received payments or services, either directly or indirectly (i.e., via his or her institution), from a third party in support of any aspect of this work. One or more of the authors, or his or her institution, has had a financial relationship, in the thirty-six months prior to submission of this work, with an entity in the biomedical arena that could be perceived to influence or have the potential to influence what is written in this work. No author has had any other relationships, or has engaged in any other activities, that could be perceived to influence or have the potential to influence what is written in this work. The complete Disclosures of Potential Conflicts of Interest submitted by authors are always provided with the online version of the article.