Incidence and risk factors of prosthetic joint infection after total hip or knee replacement in patients with rheumatoid arthritis.код для вставкиСкачать
Arthritis & Rheumatism (Arthritis Care & Research) Vol. 59, No. 12, December 15, 2008, pp 1713–1720 DOI 10.1002/art.24060 © 2008, American College of Rheumatology ORIGINAL ARTICLE Incidence and Risk Factors of Prosthetic Joint Infection After Total Hip or Knee Replacement in Patients With Rheumatoid Arthritis TIM BONGARTZ, CHRISTINE S. HALLIGAN, DOUGLAS R. OSMON, MEGAN S. REINALDA, WILLIAM R. BAMLET, CYNTHIA S. CROWSON, ARLEN D. HANSSEN, AND ERIC L. MATTESON Objective. Prosthetic joint infection is one of the most dreaded complications after total joint arthroplasty, a common procedure in patients with rheumatoid arthritis (RA). We conducted a study to evaluate potential risk factors of prosthetic joint infection and to clarify if RA is an independent predictor of this complication. Methods. This study included all patients with RA who underwent total hip or knee replacement at the Mayo Clinic Rochester between January 1996 and June 2004. The association of potential risk factors with prosthetic joint infection was examined using Cox models. A matched cohort of patients with osteoarthritis (OA) was assembled to determine whether RA is an independent risk factor for prosthetic joint infection. Results. We identiﬁed 462 patients with RA who underwent a total of 657 hip or knee replacements. Overall, 23 (3.7%) joint arthroplasties were complicated by an infection during a mean ⴞ SD followup of 4.3 ⴞ 2.4 years. Revision arthroplasty (hazard ratio [HR] 2.99, 95% conﬁdence interval [95% CI] 1.02– 8.75) and a previous prosthetic joint infection of the replaced joint (HR 5.49, 95% CI 1.87–16.14) were signiﬁcant predictors of postoperative prosthetic joint infection. Comparison of RA patients with a matched cohort of OA patients identiﬁed an increased risk of prosthetic joint infections (HR 4.08, 95% CI 1.35–12.33) in patients with RA. Conclusion. Patients with RA who undergo total hip or knee replacement are at increased risk of prosthetic joint infection, which is further increased in the setting of revision arthroplasty and a previous prosthetic joint infection. These ﬁndings highlight the importance of perioperative prophylactic measures and vigilance during the postoperative period. INTRODUCTION Joint replacement surgery is used extensively to alleviate pain and improve mobility in patients who develop severe Supported by the NIH (grant 1-K24-AR-47578-01A1), the David A. Oreck and L. A. Eaton Research Awards, and the Mayo Foundation. Tim Bongartz, MD, Christine S. Halligan, MD (current address: Orthopedic Institute, Sioux Falls, South Dakota), Douglas R. Osmon, MD, Megan S. Reinalda, BS, William R. Bamlet, MS, Cynthia S. Crowson, MS, Arlen D. Hanssen, MD, Eric L. Matteson, MD, MPH: Mayo Clinic College of Medicine, Rochester, Minnesota. Dr. Bongartz has received consultant fees, speaking fees, and/or honoraria (more than $10,000) from Wyeth, and educational grants (more than $10,000) from Amgen. Dr. Matteson has received consultant fees (less than $10,000 each) from Abbott and Amgen, and grants (more than $10,000 each) from Wyeth and Amgen. Address correspondence to Tim Bongartz, MD, Mayo Clinic College of Medicine, Division of Rheumatology, 200 1st Street SW, Rochester, MN 55905. E-mail: bongartz. firstname.lastname@example.org. Submitted for publication April 4, 2008; accepted in revised form July 11, 2008. destructive changes of their joints as a result of inﬂammatory or degenerative musculoskeletal diseases. Among the most dreaded complications of these procedures is prosthetic joint infection. This complication causes signiﬁcant morbidity and accounts for a substantial amount of health care expenditures for patients. Although the use of rigorous preventive measures such as perioperative antimicrobial prophylaxis has reduced the rate of postoperative prosthesis infections to less than 2%, the frequent use of this procedure in more than 600,000 Americans per year suggests a projected infection rate of 12,000 implants per year (1). Prosthetic joint infection often requires removal of the infected prosthesis and prolonged intravenous antimicrobial therapy and has a mortality rate of 2.7–18%. The cost of each infection episode is estimated to be in excess of $50,000 (2,3). Total joint replacement is a common procedure in patients with rheumatoid arthritis (RA). This patient population has been identiﬁed to have a higher baseline risk of infectious diseases compared with the general population (4,5). In addition, the immunosuppressive drugs used in the treatment of RA may further increase the risk of infec1713 1714 Bongartz et al tion (5,6). Identiﬁcation of patients with RA as being at higher risk for prosthetic joint infection compared with non-RA patients would be an important ﬁrst step toward improved preoperative assessment and development of preventive strategies. Several studies have explored the risk of prosthetic joint infection in patients with RA (7–18), but most of these studies were conducted between 1969 and 1993. Patients who undergo total joint replacement at present may differ signiﬁcantly from these early cohorts because the utilization of preventive measures, such as perioperative antibiotics, has changed over time. New treatment approaches for patients with RA, including earlier use of diseasemodifying antirheumatic drugs (DMARDs) and the advent of biologic drugs such as anti–tumor necrosis factor (antiTNF) inhibitors, may have signiﬁcantly altered the risk of perioperative complications in these patients. More importantly, the validity of many existing trials is limited by methodologic problems such as no application of acknowledged criteria for a diagnosis of RA by a rheumatologist (7–18), lack of an explicit deﬁnition of prosthetic joint infection (8,9,11–13,15), and no matching or adjustment for known risk factors such as age, site, or primary versus revision arthroplasty (8 –12,15). Not surprisingly, conclusions reached by these studies differ widely: while some studies reported an increased risk of prosthetic joint infection in patients with RA (7,8,10,11,13,15,17), others did not support this conclusion (9,14,16,18). Assessments of potential risk factors such as maintenance of DMARD therapy were inconclusive, with some trials arguing in favor (19,20) and some against (21,22) an association between DMARD therapy and prosthesis infection. In light of conﬂicting results presented in these studies and based on the identiﬁcation of their potential sources of bias, we formulated the following methodologic standards for a study of infection risk in patients with RA undergoing joint arthroplasty: modern cohort of patients with RA after Food and Drug Administration licensing and approval of anti-TNF treatment, veriﬁed diagnosis of exposure status (RA/osteoarthritis [OA]), documented followup for at least 1 year, reproducible criteria for outcome (prosthesis infection), and adjustment/matching for potential confounders when comparing patients with OA and those with RA. Based on these standards, we aimed to determine the cumulative and 1-year incidence of prosthetic joint infection in total hip arthroplasty (THA) and total knee arthroplasty (TKA) in a modern cohort of patients with RA and to further explore potential risk factors for these infections. In addition, we intended to evaluate whether this complication is more frequent in patients with RA as compared with patients with OA who undergo the same type of surgery. Board approval was obtained for the study prior to its commencement. All surgeries were performed by trained Mayo Clinic staff orthopedists in a single facility. By medical record review, patients were followed for at least 1 year after surgery. Cases of THA or TKA infection were recorded to calculate a cumulative incidence of this complication and to evaluate potential associations with clinical variables. To compare the infection risks of patients with RA versus patients with OA, a second, matched cohort was assembled consisting of patients with OA who underwent a THA or TKA within the same time period. PATIENTS AND METHODS Veriﬁcation of diagnosis in the OA cohort. ACR criteria for hip and knee OA (25,26) proved to be not useful for retrospective assessment because these criteria contain several clinical, radiographic, and laboratory data that were not documented in the medical records in a majority of cases. Therefore, the diagnosis of OA was veriﬁed during chart review using the following alternative criteria: Overview of the study design. We conducted a retrospective, single-center, double-cohort study that included all patients with a diagnosis of RA who underwent a THA or TKA at the Mayo Clinic Rochester between January 1996 and June 2004. Mayo Clinic Institutional Review Patients. The Mayo Clinic Total Joint Registry includes all patients who have undergone total joint surgery at the Mayo Clinic Rochester since 1969 (23). In addition to patient demographic data, the underlying disease diagnosis leading to surgery, site of joint replacement, year of surgery, and perioperative complications are recorded in this registry. All patients are followed up by surgeon examination at least twice in the ﬁrst postsurgical year and then at least every 5 years thereafter. If followup examination at the Mayo Clinic was not possible, patients were contacted by letter and/or telephone and asked to complete a standardized data collection form. The records of all patients with a registry diagnosis of RA who underwent a THA or TKA at the Mayo Clinic between January 1, 1996 and June 30, 2004 were retrieved. Veriﬁcation of RA diagnosis. A medical record review of all patients with a diagnosis of RA in the total joint registry was performed. Each patient’s diagnosis of RA was veriﬁed using validated RA classiﬁcation criteria. This step was mandatory because some patients were expected to be incorrectly coded as having RA in the database. Patients were only enrolled in the RA cohort if they met the American College of Rheumatology (ACR; formerly the American Rheumatism Association) classiﬁcation criteria (24). If the information available through chart review was not sufﬁcient for the veriﬁcation of diagnosis, the treating rheumatologist/primary physician was contacted to complete the clinical information assessment. Matching of the control cohort of patients with OA. Patients with RA were matched to patients with OA. The records of patients with a diagnosis of OA and no diagnosis of RA were retrieved from the total joint registry. Using a computer matching algorithm, a 1:1 matching was performed according to patient age (⫾ 5 years), site (hip or knee), type (revision/primary arthroplasty), and time point of ﬁrst Mayo surgery. Risk Factors of Prosthetic Joint Infection in RA Patients Table 1. Time deﬁnitions for perioperative diseasemodifying antirheumatic drug therapy to be considered as stopped Medication No. days medication not administered Methotrexate Leﬂunomide Oral gold Intramuscular gold Sulfasalazine Hydroxychloroquine Azathioprine Cyclosporine Cyclophosphamide D-penicillamine Etanercept Adalimumab Inﬂiximab Anakinra 8 85 or 14* 8 29 8 85 8 8 8 15 8 15 57 8 * With cholestyramine wash-out. physician’s diagnosis of hip and/or knee OA and absence of a physician’s diagnosis of an inﬂammatory joint disease. These criteria were validated against patients with a diagnosis of RA and a diagnosis of OA according to the following reference standard: radiograph consistent with OA, no previous diagnosis of RA or other inﬂammatory joint disease, and at least 1 year of followup information after the initial OA diagnosis from the primary physician of each patient by telephone interview. Based on 2 random samples of 60 patients for each group drawn from the total joint registry, we calculated the following test characteristics for our OA classiﬁcation criteria: sensitivity 95.08% (95% conﬁdence interval [95% CI] 86.08 –98.96%) and speciﬁcity 96.67% (95% CI 88.47–99.59%). Data collection. Data in both cohorts were collected by medical record review for each patient. In case of missing information or followup ⬍1 year after ﬁrst surgery, direct telephone interviews with primary care providers and/or treating rheumatologists were performed. Data were abstracted during the review process by 3 abstractors. Multiple variables were ascertained through this process: date of birth, sex, time of prosthesis implantation, site of surgery, type of surgery (primary or revision arthroplasty), comorbidities (according to physician’s diagnosis), Steinbrocker functional class criteria (27), and use of steroids at time of surgery. If DMARD therapy was withheld around the time of surgery, the stop and start dates were abstracted. Based on the pharmacokinetic half-life and/or data on the biologic activity of each DMARD, criteria for a strict deﬁnition of withheld versus maintained were developed (Table 1). According to these criteria, perioperative DMARD use was judged as either withheld or maintained. Outcome variable. An established set of criteria (28,29) was used to detect and verify the diagnosis of prosthetic 1715 joint infection. At least 1 of the following had to be present: isolation of the same microorganism from ⱖ2 cultures of joint aspirates or intraoperative tissue specimens, acute inﬂammation consistent with infection on histopathologic examination (as determined by the pathologist), cutaneous sinus tract communicating with the joint prosthesis, and/or purulence in the joint space (as determined by the surgeon). The Total Joint Registry provides a thorough followup of patients who underwent total joint replacement at the Mayo Clinic, ensuring almost complete ascertainment of adverse events related to the joint replacement. Again, for outstanding or unclear issues as well as for a followup ⬍1 year, primary physicians were contacted. Statistical analysis. The proportion of surgeries complicated by a prosthetic joint infection was estimated using Kaplan-Meier techniques (30). Cox proportional hazards models were used to examine the association between the risk of prosthetic joint infection for several explanatory variables including age at surgery date, sex, type of surgery (revision versus primary), site of surgery (hip versus knee), presence of comorbidities, RA disease duration, patient functional status, and DMARD therapy at time of surgery (withheld/maintained). Cox models were also used to compare the risk of infection between the RA and OA cohorts. For these models, only the ﬁrst surgery at our institution was used for each patient, and followup was censored at the time of a subsequent surgery or last followup. RESULTS Patient results. During the period from January 1, 1996 to June 30, 2004, 661 patients with a diagnosis of RA in the Total Joint Registry underwent a THA or TKA at the Mayo Clinic Rochester. After strict application of the 1987 ACR consensus criteria for the classiﬁcation of RA (24) during chart review by a rheumatologist, 461 (69.7%) of these patients could be classiﬁed as having RA. In 19 cases, medical records did not provide sufﬁcient information to either reject or conﬁrm a diagnosis of RA. Primary physicians and/or rheumatologists of these patients were contacted by telephone to request information on classiﬁcation criteria. In 9 of these 19 patients, a diagnosis of RA could be conﬁrmed. As a result, 470 patients with a reliable diagnosis of RA were included in our RA cohort. Followup. Medical charts provided sufﬁcient information for at least 1 year of followup in the vast majority of patients. However, in 18 patients, followup information was not sufﬁcient. In 10 of these patients, followup data could be obtained through telephone interviews with treating rheumatologists or primary physicians. In 8 cases, followup data could not be obtained and patients were excluded from our cohort, for a total of 462 patients. The ﬂow of patients into our RA cohort is depicted in Figure 1. Patient characteristics. The cohort comprised 462 patients with RA with a mean age of 64 years at their ﬁrst 1716 Bongartz et al The most common organism isolated from infected prosthetic joints was Staphylococcus aureus in 15 (65.2%) cases, followed by 4 (17.4%) cases of streptococcal infection, 2 (8.7%) gram-negative bacilli infections, and 1 fungal infection. Sixteen patients were treated with prosthesis removal, and 6 of these patients had a subsequent prosthesis reimplantation. Seven patients were treated with debridement and long-term antibiotic suppression therapy. The details of prosthetic joint infections that occurred in our cohort are summarized in Table 3. Figure 1. Flow of patients into the rheumatoid arthritis (RA) cohort. SpA ⫽ spondylarthritis; SLE ⫽ systemic lupus erythematosus; GCA ⫽ giant cell arteritis; ACR ⫽ American College of Rheumatology. total joint surgery between 1996 and 2004 at the Mayo Clinic Rochester. The mean disease duration was 21.1 years, with a mean followup of 4.3 years after the ﬁrst surgery. The majority of patients were female (78.6%). Surgery characteristics. The 462 patients with RA underwent a total of 657 THA or TKA surgeries during the study period. Of these surgeries, 61.2% were primary and 38.8% were revision surgeries. There were almost equal numbers of THA (328 [50.0%]) and TKA (329 [50.0%]). Approximately one-third of the surgeries were performed with antibiotic impregnated cement. The functional status of the majority of patients at the time of surgery was classiﬁed as class 2 (40.3%) or class 3 (39.3%). Surgery characteristics are summarized in Table 2. Immunosuppressive medication use at the time of surgery. Patients were receiving DMARDs prior to surgery in 429 cases. During 222 (33.8%) surgeries, at least 1 of these DMARDs was withheld perioperatively according to the deﬁnitions given above. In 341 (51.9%) procedures, patients received steroids on the day of surgery in a median dosage of 10 mg/day. The differing mean value of 29 mg/ day reﬂects the application of a steroid bolus at the time of surgery in some patients and the subsequent skewedness of steroid doses to higher values. Incidence of prosthesis infections. Overall, 23 (3.7%) of 657 surgeries in patients with RA were complicated by a prosthetic joint infection. This included 14 (2.2%) prostheses that became infected during the ﬁrst postoperative year. The majority of prosthetic joint infections occurred after revision arthroplasties: 255 revision arthroplasties were complicated by 15 (5.9%) infections as compared with 8 (2.0%) infections in 402 primary procedures. Seven (30.4%) of the 23 infections were observed in joints that had been affected by a previous prosthetic joint infection. Twelve (52.2%) infections were in the hip and 11 (47.8%) infections were in the knee. Predictors associated with prosthesis infections in patients with RA. Considering the ﬁrst surgery at our institution for each patient, univariate analysis revealed 3 statistically signiﬁcant predictors of an increased prosthesis infection risk: revision arthroplasty (hazard ratio [HR] 2.99, 95% CI 1.02– 8.75), previous infection of the replaced joint (HR 5.49, 95% CI 1.87–16.14), and operation time (HR 1.36 per 60-minute increase, 95% CI 1.02–1.81). Other potential risk factors such as age, disease duration, site of surgery, and functional class were not signiﬁcantly associated with the occurrence of prosthesis infections (Table 4). Perioperative management of immunosuppressive therapies was highly heterogeneous and did not show any apparent association with patient age, disease severity, or comorbidities. In 336 of 657 surgeries, patients were receiving at least 1 DMARD within 3 months prior to the procedure. DMARD therapy was withheld perioperatively in 57% of these procedures according to the deﬁnitions given in the Patients and Methods section. Stopping DMARD therapy at the time of surgery lowered the risk of subsequent prosthesis infection (HR 0.65, 95% CI 0.09 – 4.95), but this was statistically not signiﬁcant. There were 3 prosthesis infections in 38 patients who were Table 2. Characteristics of 657 surgeries in 462 patients with rheumatoid arthritis* Characteristic Type of surgery Primary Revision Joint replaced Primary THA Revision THA Primary TKA Revision TKA Perioperative systemic prophylactic antibiotics Antibiotic impregnated cement used Steinbrocker functional classiﬁcation at time of surgery Class 1 (patient normally active) Class 2 (patient able to maintain occupation but less active) Class 3 (unable to maintain occupation) Class 4 (largely or wholly incapacitated) Unknown Value 402 (61.2) 255 (38.8) 164 (25.0) 164 (25.0) 238 (36.2) 91 (13.8) 656 (99.9) 209 (31.8) 6 (1.0) 265 (40.3) 258 (39.3) 121 (18.4) 7 (1.0) * Values are the number (percentage). THA ⫽ total hip arthroplasty; TKA ⫽ total knee arthroplasty. Risk Factors of Prosthetic Joint Infection in RA Patients Table 3. Characteristics of 23 infections following 657 surgeries among 462 patients with rheumatoid arthritis* Characteristic Time from surgery to infection, years Mean ⫾ SD Median (interquartile range) Joint replaced Primary THA Revision THA Primary TKA Revision TKA Type of surgery Primary Revision Infection site Hip Knee Prosthesis age at time of infection diagnosis ⱕ30 days 31–365 days ⱖ366 days No. of revisions of the index joint 0 1 2 3 4 Prior prosthetic joint infection in index joint Surgery performed on infected joint No surgery Debridement only Removal only Removal and reimplantation Organism Staphylococci Streptococci Gram-negative bacilli Other Value 1.1 ⫾ 1.53 0.3 (0.1–0.8) 2 (8.7) 10 (43.5) 6 (26.1) 5 (22.7) 8 (34.8) 15 (65.2) 12 (52.2) 11 (47.8) 8 (34.8) 6 (26.1) 9 (39.1) 8 (34.8) 6 (26.1) 6 (26.1) 1 (4.3) 2 (8.7) 7 (30.4) 3 (13.0) 4 (17.4) 10 (43.5) 6 (26.1) 15 (65.2) 4 (17.4) 2 (8.7) 5 (22.7) * Values are the number (percentage) unless otherwise indicated. THA ⫽ total hip arthroplasty; TKA ⫽ total knee arthroplasty. treated with anti-TNF agents at the time of surgery as compared with no infection in 12 patients who stopped their anti-TNF therapy prior to surgery. This difference was not statistically signiﬁcant. Perioperative corticosteroid use was not associated with an increased risk of prosthesis infections. Is RA an independent risk factor for prosthesis infection? Each of the 462 patients with RA was matched to a patient with OA according to age, sex, site of surgery, and type of arthroplasty (primary versus revision). Characteristics of both cohorts are shown in Table 5. Compared with patients with OA, those with RA had a higher prevalence of comorbidities and impaired physical functioning. Patients in both cohorts were censored at last followup or second surgery for these analyses. The mean ⫾ SD followup time was 3.1 ⫾ 2.4 years for patients with RA and 3.8 ⫾ 2.6 years for patients with OA, corresponding to 1,427 and 1,760 person-years of followup, respectively. 1717 During followup, a total of 15 patients with RA (4.2% at 5 years) and 4 patients with OA (1.4% at 5 years) developed a prosthesis infection (log rank P ⫽ 0.005). The higher risk in patients with RA was statistically signiﬁcant (HR 4.08, 95% CI 1.35–12.33), even after adjusting for previous infection in the index joint (HR 3.74, 95% CI 1.23–11.33). Age, sex, functional class, and comorbidity were not signiﬁcant predictors of infection. When restricting the analysis to infections diagnosed within the ﬁrst year after surgery, 10 patients with RA (2.3% at 1 year) and 1 patient with OA developed an infection. Again, the excess risk for patients with RA was statistically signiﬁcant (odds ratio 10.30, 95% CI 1.31– 80.26). DISCUSSION Although advances in preventive strategies have contributed to a decline in prosthetic joint infection following total joint replacements, this serious complication was observed in 4.2% of total hip or knee replacements in RA patients by 5 years after the surgery. This estimate of risk for hip and knee prosthesis infections is consistent with previous studies. Fitzgerald et al reported a deep infection after hip replacement in 3.1% of 223 RA patients (17). Wilson et al (8) observed a prosthesis infection in 2.2% of 2,076 knee replacements. However, a direct comparison is hampered due to nonreporting of a mean followup and losses to followup in these trials. Elbow, ﬁnger, and shoulder arthroplasties were not ac- Table 4. Univariate analysis of potential predictors of prosthetic joint infection in 462 patients with rheumatoid arthritis* Risk factor Age (per 10-year increase) Revision arthroplasty Male sex Disease duration, years ⬍10 10–20 20–30 ⬎30 TKA vs. THA DMARD therapy withheld Corticosteroid use Operative time (per 60-minute increase) Functional class (Steinbrocker) 1 and 2 3 4 Comorbidities Extraarticular disease manifestations Previous infection at index joint Previous infection at any joint Hazard ratio (95% CI) 0.86 (0.61–1.22) 2.99 (1.02–8.75) 1.31 (0.42–4.12) 1.0 0.18 (0.02–1.59) 0.89 (0.24–3.32) 1.05 (0.28–3.91) 0.91 (0.33–2.50) 0.65 (0.09–4.95) 1.28 (0.46–3.60) 11.36 (1.02–1.81) 1.0 1.30 (0.40–4.27) 1.87 (0.50–6.96) 1.03 (0.33–3.24) 1.01 (0.36–2.78) 5.49 (1.87–16.14) 4.18 (1.43–12.25) * 95% CI ⫽ 95% conﬁdence interval; TKA ⫽ total knee arthroplasty; THA ⫽ total hip arthroplasty; DMARD ⫽ disease-modifying antirheumatic drug. 1718 Bongartz et al Table 5. Characteristics of 462 patients with OA and 462 patients with RA* Characteristic OA patients RA patients Female sex Age at ﬁrst Mayo surgery, years Mean ⫾ SD Median (IQR) Length of followup, years Mean ⫾ SD Median (IQR) Joint replaced Primary THA Revision THA Primary TKA Revision TKA Type of surgery Primary Revision Surgery on knee Functional class Class 1 Class 2 Class 3 Class 4 Unknown At least 1 comorbidity Previous prosthetic infection in index joint 338 (73.2) 363 (78.6) 67.2 ⫾ 10.8 67.8 (59.9–76.0) 63.6 ⫾ 13.3 66.2 (55.1–73.8) P 0.06 ⬍ 0.001 – 3.8 ⫾ 2.6 3.3 (1.8–5.7) 3.1 ⫾ 2.4 2.4 (1.1–5.0) 126 (27.3) 112 (24.2) 163 (35.3) 61 (13.2) 117 (25.3) 122 (26.4) 158 (34.2) 65 (14.1) 289 (62.6) 173 (37.4) 224 (48.5) 275 (59.5) 187 (40.5) 223 (48.3) 6 (1.3) 338 (73.2) 108 (23.4) 8 (1.7) 2 (0.4) 383 (83.0) 24 (5.2) 4 (0.9) 183 (39.6) 184 (39.6) 86 (18.6) 6 (1.3) 337 (73.0) 43 (9.3) 0.81 0.34 0.95 ⬍ 0.0001 ⬍ 0.001 0.02 * Values are the number (percentage) unless otherwise indicated. OA ⫽ osteoarthritis; RA ⫽ rheumatoid arthritis; IQR ⫽ interquartile range; THA ⫽ total hip arthroplasty; TKA ⫽ total knee arthroplasty. counted for in our analysis and would contribute an additional life-time risk for prosthetic joint infection in patients with RA. Previous studies of the frequency, risk factors, and other aspects of prosthetic joint infection in patients with RA have been derived from data that relied on an RA diagnosis not conﬁrmed by a rheumatologist (7–18). It has been demonstrated that the accuracy of such a diagnosis is low and the agreement between database and medical record diagnosis is poor (31). To prevent misclassiﬁcation of patients as having RA when in fact they do not, we applied ACR classiﬁcation criteria to each patient at medical record and electronic chart review. Our experience is consistent with the previously observed poor performance of electronic database diagnosis of RA: 31% of patients who were listed as having RA in the total joint registry did not meet ACR criteria for the diagnosis of this disease. This high rate of misclassiﬁcation represents a serious threat to any estimate derived from an RA cohort that uses the database diagnosis as the sole source of diagnostic classiﬁcation. Although we did not systematically explore the reasons for the high rate of misclassiﬁcation in our database, it appears unlikely that this discrepancy was due to a low accuracy of ACR criteria when applied during retrospective chart review. All records were reviewed by experienced rheumatologists who were able to assign an alternative diagnosis to the majority of misclassiﬁed patients. Revision arthroplasty has been identiﬁed previously as a predictor of prosthetic joint infection in mixed cohorts (including patients with OA and RA) (7,11,17). This observation underlines the importance of adjusting for revision arthroplasty when comparing the frequency of infectious complications between cohorts that have a different distribution of prosthetic joint revisions. Consistent with earlier ﬁndings (13), our study identiﬁed previous prosthetic joint infection of the replaced joint as a strong risk factor for a subsequent infection after implantation of a new prosthesis. Of 24 patients with RA who had a previous prosthetic joint infection and underwent revision arthroplasty, 7 (29.2%) experienced a subsequent prosthetic joint infection. In all of these patients, multiple methods such as indium scan, intraoperative histology/gram stain, and/or joint aspirations with culture were used to conﬁrm clearance of the previous infection prior to implantation of new hardware. Reimplantation procedures after removal of an infected prosthesis carry a considerable risk of reinfection. In a study of 160 patients with RA, a reinfection risk of 21% and 39% for 2-stage exchange and resection arthroplasty, respectively, has been reported (32). A reinfection risk of 9 –14% was observed in a mixed cohort of patients who underwent second-stage reimplantation after a ﬁrst infection (33). The high rate of infection in the replaced prosthetic joint raises the question of whether these reinfections are due to persistence of bacteria in the replaced joint and a low sensitivity of the diagnostic methods used to identify their presence prior to reimplantation, or if the previous infec- Risk Factors of Prosthetic Joint Infection in RA Patients tion is simply an indicator of a host with higher infection susceptibility. Patients in our study who had a previous prosthetic joint infection and then subsequent replacement of joints other than a previously infected joint had a risk of another infection of only 3.5%, as compared with 29.2% after revision of the previously infected joint. This ﬁnding suggests that these infections are due to undetected persistence of bacteria and not to a general susceptibility to infections. Our evaluation of potential predictors of prosthetic joint infections after hip or knee replacement in patients with RA also included examination of perioperative DMARD treatment. Although we found a lower risk of subsequent prosthesis infections in patients who stopped their DMARD therapy prior to surgery, our results were statistically not signiﬁcant. To avoid dilution of a potentially increased risk with immunosuppression, we excluded agents that are of lower immunosuppressive potency (hydroxychloroquine and sulfasalazine) and focused our analysis on methotrexate, azathioprine, leﬂunomide, and biologic response modiﬁers, including anti-TNF agents and anakinra. This strategy did not change our estimate of risk. Our comparison between patients with RA and OA showed a signiﬁcantly higher rate of infections in those with RA after accounting for age, sex, surgery site, and surgery type by matching. Although several studies support this ﬁnding (7,8,10,11,13,14,17), others did not detect an association between RA and increased risk of prosthesis infection (9,14,16,18). However, none of these studies made an attempt to verify a database diagnosis of RA through application of ACR criteria by a rheumatologist, some did not apply formal statistical tests (12,15,17), and only a few adjusted for some potential confounders (7,14,16 –18) or offered data on completeness of followup (7,17,18). Although our 2 cohorts were matched according to several potential confounders such as age, sex, site of surgery, and type of surgery, we cannot exclude residual confounding when comparing the incidence of prosthetic joint infections between patients with RA and those with OA. For example, insurance and socioeconomic status were not abstracted during chart review and may act as unmeasured confounders. In conclusion, we are able to provide an estimate for the incidence of prosthesis infections after hip and knee arthroplasty in a modern cohort of patients with RA. The identiﬁcation of patients with RA as being at signiﬁcantly higher risk of prosthetic joint infections emphasizes the importance of utilizing all possible pre- and postoperative prophylactic measures in this high-risk group. Especially in the setting of a revision arthroplasty or a preceding infection, physicians should maintain a high index of suspicion for postoperative infections. Our results are based on a methodologic approach that fulﬁlls several major criteria of methodologic quality in cohort studies: exposure deﬁnition (RA versus OA) according to validated criteria, minimal loss to followup, strict deﬁnition of outcomes (prosthetic joint infection), and accounting for important confounders by a conclusive 1719 matching algorithm contribute to the methodologic robustness of this study. AUTHOR CONTRIBUTIONS Dr. Bongartz had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study design. Bongartz, Osmon, Bamlet, Hanssen, Matteson. Acquisition of data. Bongartz, Halligan, Bamlet, Matteson. Analysis and interpretation of data. Bongartz, Osmon, Reinalda, Bamlet, Crowson, Hanssen, Matteson. Manuscript preparation. Bongartz, Osmon, Reinalda, Bamlet, Hanssen, Matteson. Statistical analysis. Bongartz, Reinalda, Bamlet, Crowson. REFERENCES 1. Darouiche RO. Treatment of infections associated with surgical implants. N Engl J Med 2004;350:1422–9. 2. Ahlberg A, Carlsson AS, Lindberg L. Hematogenous infection in total joint replacement. Clin Orthop Relat Res 1978;137: 69 –75. 3. Powers KA, Terpenning MS, Voice RA, Kauffman CA. Prosthetic joint infections in the elderly. Am J Med 1990;88:9N– 13N. 4. Doran MF, Crowson CS, Pond GR, O’Fallon WM, Gabriel SE. Frequency of infection in patients with rheumatoid arthritis compared with controls: a population-based study. Arthritis Rheum 2002;46:2287–93. 5. Bernatsky S, Hudson M, Suissa S. Anti-rheumatic drug use and risk of serious infections in rheumatoid arthritis. Rheumatology (Oxford) 2007;46:1157– 60. 6. Bongartz T, Sutton AJ, Sweeting MJ, Buchan I, Matteson EL, Montori V. Anti-TNF antibody therapy in rheumatoid arthritis and the risk of serious infections and malignancies: systematic review and meta-analysis of rare harmful effects in randomized controlled trials [published erratum appears in JAMA 2006;295:2482]. JAMA 2006;295:2275– 85. 7. Wymenga AB, van Horn JR, Theeuwes A, Muytjens HL, Slooff TJ. Perioperative factors associated with septic arthritis after arthroplasty: prospective multicenter study of 362 knee and 2,651 hip operations. Acta Orthop Scand 1992;63:665–71. 8. Wilson MG, Kelley K, Thornhill TS. Infection as a complication of total knee-replacement arthroplasty: risk factors and treatment in sixty-seven cases. J Bone Joint Surg Am 1990;72: 878 – 83. 9. Syahrizal AB, Kareem BA, Anbanadan S, Harwant S. Risk factors for infection in total knee replacement surgery at hospital Kuala Lumpur. Med J Malaysia 2001;56 Suppl D:5– 8. 10. Salvati EA, Robinson RP, Zeno SM, Koslin BL, Brause BD, Wilson PD Jr. Infection rates after 3175 total hip and total knee replacements performed with and without a horizontal unidirectional ﬁltered air-ﬂow system. J Bone Joint Surg Am 1982;64:525–35. 11. Poss R, Thornhill TS, Ewald FC, Thomas WH, Batte NJ, Sledge CB. Factors inﬂuencing the incidence and outcome of infection following total joint arthroplasty. Clin Orthop Relat Res 1984;182:117–26. 12. Petty W, Bryan RS, Coventry MB, Peterson LF. Infection after total knee arthroplasty. Orthop Clin North Am 1975;6: 1005–14. 13. Luessenhop CP, Higgins LD, Brause BD, Ranawat CS. Multiple prosthetic infections after total joint arthroplasty: risk factor analysis. J Arthroplasty 1996;11:862– 8. 14. Lidwell OM, Lowbury EJ, Whyte W, Blowers R, Stanley SJ, Lowe D. Infection and sepsis after operations for total hip or knee-joint replacement: inﬂuence of ultraclean air, prophylactic antibiotics and other factors. J Hyg (Lond) 1984;93: 505–29. 15. Lakatos J, Csakanyi L. Comparison of complications of total 1720 16. 17. 18. 19. 20. 21. 22. 23. 24. hip arthroplasty in rheumatoid arthritis, ankylosing spondylitis, and osteoarthritis. Orthopedics 1991;14:55–7. Gordon SM, Culver DH, Simmons BP, Jarvis WR. Risk factors for wound infections after total knee arthroplasty. Am J Epidemiol 1990;131:905–16. Fitzgerald RH Jr, Nolan DR, Ilstrup DM, Van Scoy RE, Washington JA 2nd, Coventry MB. Deep wound sepsis following total hip arthroplasty. J Bone Joint Surg Am 1977;59:847–55. Berbari EF, Hanssen AD, Duffy MC, Steckelberg JM, Ilstrup DM, Harmsen WS, et al. Risk factors for prosthetic joint infection: case-control study. Clin Infect Dis 1998;27: 1247–54. Carpenter MT, West SG, Vogelgesang SA, Casey Jones DE. Postoperative joint infections in rheumatoid arthritis patients on methotrexate therapy. Orthopedics 1996;19:207–10. Bridges SL Jr, Lopez-Mendez A, Han KH, Tracy IC, Alarcon GS. Should methotrexate be discontinued before elective orthopedic surgery in patients with rheumatoid arthritis? J Rheumatol 1991;18:984 – 8. Escalante A, Beardmore TD. Risk factors for early wound complications after orthopedic surgery for rheumatoid arthritis. J Rheumatol 1995;22:1844 –51. Grennan DM, Gray J, Loudon J, Fear S. Methotrexate and early postoperative complications in patients with rheumatoid arthritis undergoing elective orthopaedic surgery. Ann Rheum Dis 2001;60:214 –7. Berry DJ, Kessler M, Morrey BF. Maintaining a hip registry for 25 years: Mayo Clinic experience. Clin Orthop Relat Res 1997; 344:61– 8. Arnett FC, Edworthy SM, Bloch DA, McShane DJ, Fries JF, Cooper NS, et al. The American Rheumatism Association 1987 revised criteria for the classiﬁcation of rheumatoid arthritis. Arthritis Rheum 1988;31:315–24. Bongartz et al 25. Altman R, Asch E, Bloch D, Bole G, Borenstein D, Brandt K, et al. Development of criteria for the classiﬁcation and reporting of osteoarthritis: classiﬁcation of osteoarthritis of the knee. Arthritis Rheum 1986;29:1039 – 49. 26. Altman R, Alarcon G, Appelrouth D, Bloch D, Borenstein D, Brandt K, et al. The American College of Rheumatology criteria for the classiﬁcation and reporting of osteoarthritis of the hip. Arthritis Rheum 1991;34:505–14. 27. Steinbrocker O, Traeger CH, Batterman RC. Therapeutic criteria in rheumatoid arthritis. J Am Med Assoc 1949;140: 659 – 62. 28. Brandt CM, Duffy MC, Berbari EF, Hanssen AD, Steckelberg JM, Osmon DR. Staphylococcus aureus prosthetic joint infection treated with prosthesis removal and delayed reimplantation arthroplasty. Mayo Clin Proc 1999;74:553– 8. 29. Meehan AM, Osmon DR, Duffy MC, Hanssen AD, Keating MR. Outcome of penicillin-susceptible streptococcal prosthetic joint infection treated with debridement and retention of the prosthesis. Clin Infect Dis 2003;36:845–9. 30. Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. J Am Stat Assoc 1958;53:457– 81. 31. Gabriel SE. The sensitivity and speciﬁcity of computerized databases for the diagnosis of rheumatoid arthritis. Arthritis Rheum 1994;37:821–3. 32. Berbari EF, Osmon DR, Duffy MC, Harmssen RN, Mandrekar JN, Hanssen AD, et al. Outcome of prosthetic joint infection in patients with rheumatoid arthritis: the impact of medical and surgical therapy in 200 episodes. Clin Infect Dis 2006;42: 216 –23. 33. Mont MA, Waldman BJ, Hungerford DS. Evaluation of preoperative cultures before second-stage reimplantation of a total knee prosthesis complicated by infection: a comparisongroup study. J Bone Joint Surg Am 2000;82-A:1552–7.