Dr. Bogoch has received consulting fees, speaking fees, and/or honoraria from Eli Lilly, Procter & Gamble, Merck Frosst Canada, Merck Sharpe & Dohme, Novartis Canada Ltd., and the Alliance for Better Bone Health (less than $10,000 each) and has received unrestricted research grants from Amgen Canada, Novartis Canada Ltd., Warner-Chilcott, and the Alliance for Better Bone Health.
Most of the evidence regarding complications following total hip arthroplasty (THA) and total knee arthroplasty (TKA) is based on studies of patients with osteoarthritis (OA), with little being known about outcomes in patients with rheumatoid arthritis (RA). The objective of the present study was to review the current evidence regarding rates of THA/TKA complications in RA versus OA.
Data sources used were Medline, EMBase, Cinahl, Web of Science, and reference lists of articles. We included reports published between 1990 and 2011 that described studies of primary total joint arthroplasty of the hip or knee and contained information on outcomes in ≥200 RA and OA joints. Outcomes of interest included revision, hip dislocation, infection, 90-day mortality, and venous thromboembolic events. Two reviewers independently assessed each study for quality and extracted data. Where appropriate, meta-analysis was performed; if this was not possible, the level of evidence was assessed qualitatively.
Forty studies were included in this review. The results indicated that patients with RA are at increased risk of dislocation following THA (adjusted odds ratio 2.16 [95% confidence interval 1.52–3.07]). There was fair evidence to support the notion that risk of infection and risk of early revision following TKA are increased in RA versus OA. There was no evidence of any differences in rates of revision at later time points, 90-day mortality, or rates of venous thromboembolic events following THA or TKA in patients with RA versus OA. RA was explicitly defined in only 3 studies (7.5%), and only 11 studies (27.5%) included adjustment for covariates (e.g., age, sex, and comorbidity).
The findings of this literature review and meta-analysis indicate that, compared to patients with OA, patients with RA are at higher risk of dislocation following THA and higher risk of infection following TKA.
Total joint arthroplasty (TJA) is considered one of the most successful health care interventions for end-stage arthritis of the hip or knee (1–3). Among medical and surgical interventions, estimates of cost utility consistently rank TJA at or near the top for cost-effectiveness and patient satisfaction (4–6). This success is reflected in the increasing rates of total hip arthroplasty (THA) and total knee arthroplasty (TKA). Outcomes following THA and TKA are generally excellent, with low complication rates. However, some complications have significant consequences, including early revision, infection or dislocation, venous thromboembolism, and death (7–11).
The vast majority of THA and TKA procedures are performed for osteoarthritis (OA), which is the most common form of arthritis (12, 13). Thus, most of the literature regarding outcomes of TJA and their predictors is based on the experience in patients with OA. Among the inflammatory arthritides, rheumatoid arthritis (RA) is the most common. RA affects ∼0.8% of the population of North America (0.3–2.1%); in 80% of cases, RA develops between the ages of 35 and 50 years (14, 15). As in OA, TJA is indicated for the management of end-stage hip and knee arthritis in RA. Estimates of the prevalence of RA among TJA recipients vary considerably, in part because of the difficulty in accurately establishing this diagnosis using arthroplasty registries. However, a recent study that evaluated the medical history of TJA recipients in Ontario, Canada showed that ∼13% of these recipients had RA (10), corresponding to ∼170,000 TJAs from 2002 to 2010.
As RA is fundamentally different from OA in terms of pathogenesis, prognosis, and medical management, systematic differences in TJA outcomes would be expected (16). However, few studies have examined outcomes of TJA or their predictors in patients with RA, or have investigated whether there are differences in outcomes for patients with RA versus OA. Those that have examined this have yielded conflicting results. For example, Furnes et al (17) and Rud-Sorensen et al (18) found no difference in risk of revision following THA for RA versus OA, whereas Stea and colleagues reported a higher risk in RA (19). This lack of clarity regarding TJA outcomes, and their determinants, in patients with RA impedes patient-physician decision-making regarding when, and in which RA patients, TJA should be considered. We undertook the present study to compare the odds of complications following THA and TKA in patients with RA versus OA, using meta-analysis or systematic literature review.
Protocol, criteria for patient and study inclusion, and outcomes.
This review was conducted using a predefined protocol and in accordance with guidelines suggested by the Meta-analyses of Observational Studies in Epidemiology group (20).
Ambulatory adult patients (age ≥18 years) with RA or OA were included in this review. We included studies reporting data on RA patients in comparison with OA patients; the criteria used to establish the RA diagnosis were recorded when available. For studies that utilized diagnostic codes for RA from administrative databases or arthroplasty registries, we noted whether any information on the validity of these diagnostic codes was provided. We excluded TJA performed secondary to fracture, malignancy, “juvenile RA,” or posttraumatic arthritis.
We included peer-reviewed cohort, case–control, or case series studies published from 1990 (to more closely reflect current clinical practice) through December 2011 that examined primary TJA of the hip or knee. We excluded studies on partial knee arthroplasty, hip hemiarthroplasty, and hip resurfacing. We limited our selection to studies that examined outcomes in both OA and RA patients, with results in at least 200 joints described. We chose this sample size to have sufficient statistical power to evaluate the effect of multiple factors on rare outcomes, such as death or revision. Studies examining revision rates and those examining perioperative complications were required to have at least 1 year and 90 days of followup, respectively. We did not include annual reports from arthroplasty registries unless they were published in a peer-reviewed journal. We did not include meeting abstracts, as they did not contain enough information to assess for bias. We excluded editorials, commentaries, letters to the editor, and reviews, but they were read to identify any potential articles. We e-mailed the corresponding author of each report selected for inclusion, to clarify any details and to request access to patient-level data. The criteria for study eligibility/inclusion are summarized in Supplementary Table 1, on the Arthritis & Rheumatism web site at http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)1529-0131.
Studies of various TJA complications in patients with RA in comparison to patients with OA were evaluated. The following complications were included in the assessment: 1) revision, defined as exchange of any or all of the components due to any cause, 2) infection of the arthroplasty requiring therapy with antibiotics (any route) or surgery, 3) dislocation following THA, 4) mortality from any cause within 90 days of surgery, and 5) venous thromboembolic event within 90 days of surgery.
Information sources and search strategy.
With the aid of an experienced librarian, we searched 4 bibliographic databases (Medline, EMBase, Cinahl, Web of Science) without language restriction for reports published between January 1990 and December 2011. MeSH (Medical Subject Headings) terms and key words used for the search are shown in Supplementary Table 2 (http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)1529-0131). We also reviewed the bibliographies of included studies.
Two of the authors (BR, BE) reviewed the titles of all citations generated by the literature search and removed any that did not address THA or TKA. We reviewed the abstracts of the remaining studies and removed those that did not address RA or inflammatory arthritis. Thirty-eight abstracts were not available online or at the University of Toronto libraries; all were from non-English journals and were excluded from the review. Two additional authors (RJ, JC) independently reviewed all excluded citations to determine appropriateness of exclusion. Three citations were found to be inappropriately excluded and were restored. The abstract of each remaining citation was assessed (BR, BE) for its primary outcome and sample size. We excluded reports of studies that did not examine at least one of our outcomes of interest, or if they did not meet our sample size requirements.
Data collection process and data items.
The complete articles from the 217 eligible citations were assessed by 3 of the authors (BR, BE, RJ) for data abstraction. At least 2 authors, who were not blinded with regard to citation identifiers, independently abstracted data from each article. Briefly, for each study, we determined the following: number of replaced joints (RA and OA), number of centers, outcome measure, mean followup time, criteria used to establish arthritis diagnosis, and the type of implant(s) used. Discrepancies were resolved by consensus. Data from each article are summarized in Table 1 and in Supplementary Table 3 (http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)1529-0131). We excluded 173 studies in which the outcome measure was not stratified by arthritis diagnosis. For all selected studies, we attempted to contact the corresponding author for additional details.
Table 1. Characteristics of the 40 included studies*
Author, year (ref.)
Primary outcome measure
Additional details on each study, including study years, arthritis diagnosis (rheumatoid arthritis or osteoarthritis) and how the diagnosis was determined, outcomes, total hip arthroplasty (THA) and total knee arthroplasty (TKA) approach, means of outcome assessment and analysis, and bias score, are available in Supplementary Table 3, on the Arthritis & Rheumatism web site at http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)1529-0131.
For each included study, risk of bias was evaluated using published validity criteria (21). The domains included case definition (i.e., RA/OA classification), patient selection, followup, outcome assessment, and analyses. An additional criterion assessed data validation for database studies (2) (Supplementary Table 4). Articles with scores of ≥10 were classified as having a low risk of bias, scores of 7–9 indicated moderate risk of bias, and scores of ≤6 indicated high risk of bias.
Studies were stratified by the joint replaced (hip versus knee), outcome of interest, and duration of followup. For revision, we stratified followup into 3 periods: early (≤5 years), middle (6–10 years), and late (>10 years). When possible, studies were also stratified by the type of prosthesis and the use of bone cement. Studies were evaluated within each group to determine if meta-analysis was feasible and appropriate. If appropriate, meta-analyses were performed using random-effects models with Review Manager software (version 5.1). Meta-analytic estimates of proportion, unadjusted odds ratio (OR), adjusted OR, unadjusted relative risk (RR), and adjusted RR with 95% confidence intervals (95% CIs) were reported. For study reports that provided adjusted estimates but also provided information on the number of patients with OA and RA for both the treatment and the outcome, unadjusted estimates were calculated for pooling with other unadjusted estimates. The relative weight of each individual study in the meta-analysis was calculated based on the inverse of variance. Clinical heterogeneity among studies was assessed based on the clinical criteria described above. Statistical heterogeneity was assessed using I-square statistics. Where meta-analysis was not appropriate, a systematic review of relevant studies was conducted and the overall direction of evidence summarized qualitatively.
Synthesis of results.
Evidence based on meta-analysis was assumed to be “good” if there was adjustment for potential confounders, and “fair” if adjustment was not performed. For outcomes where meta-analysis was not possible, 2 reviewers (BR and BE) independently graded the overall strength of the evidence as good, fair, inconsistent, or insufficient (Table 2). A third reviewer (JC) resolved any differences. Grades were assigned using 3 criteria: quality, quantity, and consistency of findings. Quality was assessed based on the study's risk of bias, as defined above. Quantity was assessed based on the number of studies that evaluated each risk factor. Consistency was assessed based on similarity of findings reported across a range of study populations and study designs. The findings are reported using the guidelines included in the Preferred Reporting Items for Systematic Reviews and Meta-Analyses Statement (22).
Table 2. Grading the strength of the evidence
There is good evidence for or against an association between the complication and rheumatoid arthritis (RA) Determined by: consistent results across studies; >3 studies; at least 1 study graded as “low” bias
There is fair evidence for or against an association between the complication and RA Determined by: consistent results across studies but limited by quantity (3 studies) or quality (no studies graded as “low” bias)
There is inconsistent evidence for or against an association between the complication and RA Determined by: studies had conflicting results
There is insufficient evidence for or against an association between the complication and RA Determined by: inadequate number of studies evaluating the risk factor (<3 studies)
Risk of bias across studies.
It was decided a priori that for meta-analyses that included ≥10 studies, publication bias would be assessed using funnel plot asymmetry (23).
Study selection and quality of included studies.
The results of the search, the study selection log, and the number of studies are shown in Figure 1. Forty studies were included in this review. The results of the assessment of the overall risk of bias in the included studies are reported in Supplementary Table 3 (on the Arthritis & Rheumatism web site at http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)1529-0131). Seventeen studies had low risk of bias, 17 had moderate risk of bias, and 6 had high risk of bias. We were unable to obtain patient-level data for any of the studies selected for inclusion. Funnel plots were not assessed for asymmetry as none of the meta-analyses included 10 or more studies.
Five studies compared the rates of hip dislocation in patients with RA versus OA within 5 years of THA (risk of bias moderate in 4, high in 1) (Supplementary Table 5). Meta-analysis of these 5 studies, all of which reported unadjusted comparative data, revealed an increased risk of hip dislocation within 5 years of THA in patients with RA relative to those with OA (unadjusted OR 2.74 [95% CI 1.73–4.34]; I2 = 27% [n = 2,842 RA patients and 61,861 OA patients]) (Figure 2A). This increased risk was also found after meta-analysis of 4 studies that reported comparative data, adjusted for several variables (including age, sex, surgical approach, and surgeon volume) (adjusted OR 2.16 [95% CI 1.52–3.07]; I2 = 0% [n = 1,637 RA patients and 61,810 OA patients]) (Figure 2B).
Fifteen studies addressed THA revision (risk of bias low in 9, moderate in 5, high in 1) (Supplementary Table 6). For our analysis, these studies were grouped according to the amount of time between initial arthroplasty and revision (early, middle, or late, as described above).
Hip revision at ≤5 years.
Meta-analysis of 4 studies that reported unadjusted comparative data revealed increased odds of early revision among RA patients versus OA patients (unadjusted OR 1.33 [95% CI 1.03–1.71]; I2 = 6% [n = 3,913 RA patients and 76,221 OA patients]) (Figure 3A). However, a study that adjusted for age, sex, and comorbidity did not show increased odds of revision within 1 year of THA in patients with RA (adjusted OR 1.11 [95% CI 0.82–1.51] [n = 3,805 RA patients and 23,412 OA patients]) (10).
Hip revision at 6–10 years.
Meta-analysis of 7 studies revealed no difference in the unadjusted odds of revision at 6–10 years in RA patients versus OA patients overall (unadjusted OR 1.16 [95% CI 0.94–1.43]; I2 = 46% [n = 9,118 RA patients and 210,674 OA patients]) (Figure 3B). Allami et al also found no difference in the unadjusted risk of revision at 10 years in RA patients versus OA patients (unadjusted hazard ratio [HR] 3.33 [95% CI 0.4–24.5]) (24). Similarly, meta-analysis of 2 studies in which the investigators adjusted for age and sex (among other covariates) revealed no difference in the risk of revision between RA patients and OA patients following THA (adjusted RR 0.91 [95% CI 0.74–1.11]; I2 = 86% [n = 2,110 RA patients and 88,103 OA patients]). Additionally, Furnes et al found no difference in the odds of revision after adjustment for age, sex, and type of prosthesis (adjusted OR 1.10 [95% CI 0.90–1.35]) (17).
Hip revision at >10 years.
Meta-analysis of 2 studies revealed lower unadjusted odds of late revision in RA patients versus OA patients for cemented implants (unadjusted OR 0.28 [95% CI 0.17–0.47]; I2 = 0% [n = 229 RA patients and 1,710 OA patients]) (Figure 3C).
Meta-analysis of 3 studies indicated slightly increased odds of early revision in patients with RA versus OA (unadjusted OR 1.24 [95% CI 1.10–1.40]; I2 = 0% [n = 8,479 RA patients and 35,274 OA patients]) (Figure 4A). Himanen et al found no difference in the likelihood of revision at 4 years (unadjusted RR 1.25 [95% CI 0.87–1.8]) (25). Paterson and colleagues controlled for potential confounders (age, sex, comorbidity, provider volume) and found no significant difference in revision rates at 1 year (adjusted OR 1.08 [95% CI 0.78–1.50]) (10).
Knee revision at 6–10 years.
Meta-analysis of 6 studies revealed no difference between RA and OA patients in the odds of revision at 6–10 years (unadjusted OR 2.02 [95% CI 0.96–4.28]; I2 = 69% [n = 3,116 RA patients and 29,670 OA patients]) (Figure 4B). This was not affected by stratification by prosthesis type, i.e., cruciate-retaining prostheses (unadjusted OR 3.90 [95% CI 0.46–33.17] [n = 235 RA patients and 1,171 OA patients]) or posterior-stabilized prostheses (unadjusted OR 1.11 [95% CI 0.56–2.18] [n = 298 RA patients and 6,743 OA patients]). Schrama and colleagues specifically assessed revision due to infection after 6 years of followup, and demonstrated an unadjusted RR of 1.6 (95% CI 1.06–2.38) (26).
Knee revision at >10 years.
Meta-analysis of 2 studies that reported unadjusted comparative data on cemented implants revealed no difference between RA and OA patients in the odds of late revision (unadjusted OR 2.46 [95% CI 0.70–8.70]; I2 = 50% [n = 297 RA patients and 256 OA patients]) (Figure 4C). No studies that provided adjusted estimates were identified.
Five studies compared the rates of index joint infection in patients with RA versus OA following TJA (risk of bias low in 2, moderate in 2, high in 1) (Table 1 and Supplementary Table 3, http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)1529-0131). Meta-analysis was not possible due to variable definitions of infection and preoperative antibiotic protocols (with some centers forgoing antibiotics altogether), and pooling of primary and revision arthroplasty cases. In a study controlling for age, sex, prosthesis type, and antibiotic cement, Jamsen and colleagues found that patients with RA had an adjusted HR of 1.86 (95% CI 1.31–2.63) for infection following TKA, relative to OA patients (27). In a study by Wymenga et al the risk of infection following TKA was increased in patients with RA compared to those with OA (unadjusted RR 4.8 [95% CI 1.2–19]) (28). Bongartz and colleagues matched 402 RA patients (primary and revision THA/TKA) to OA patients by age, time of surgery, sex, and site of surgery (29). Compared with matched OA controls, RA patients had higher odds of infection in the first year (OR 10.30 [95% CI 1.31–80.26]). Among those with RA, increased infection risk was associated with prior infection in the replaced joint, prior infection in any joint, and longer duration of operating time (HR 1.36 per 60-minute increase [95% CI 1.02–1.81]). That study revealed no association with perioperative systemic corticosteroid use (HR 1.28 [95% CI 0.46–3.60]) or with withdrawal of biologic treatment prior to surgery (HR 0.65 [95% CI 0.09–4.95]) (29).
Mortality within 90 days of TJA.
Four studies reported on mortality following TJA (risk of bias low in 1, moderate in 3) (Table 1 and Supplementary Table 3). Meta-analysis of 2 studies revealed no difference between RA and OA patients in the odds of mortality within 90 days of THA (adjusted OR 1.40 [95% CI 0.82–2.39]; I2 = 68%). Similarly, meta-analysis of 2 studies examining rates of death within 90 days of TKA showed no difference between RA and OA patients (adjusted OR 0.86 [95% CI 0.66–1.12]; I2 = 0%). In a study of 3,048 TKAs, Gill and colleagues reported 14 deaths occurring within 90 days of surgery (2,871 in OA patients, 177 in RA patients); all of the deaths occurred in patients with OA (9).
Venous thromboembolism within 90 days of TJA.
Five studies reported on venous thromboembolic events following TJA (risk of bias low in 3, moderate in 2) (Table 1 and Supplementary Table 3). Meta-analysis of 2 studies revealed no difference in the odds of venous thromboembolic events within 90 days of THA in patients with RA versus those with OA (adjusted OR 0.84 [95% CI 0.28–2.54]; I2 = 85%). Similarly, White et al found that the unadjusted rates of in-hospital venous thromboembolic events were similar among RA and OA patients (P = 0.07) (30). Niki and colleagues found no difference in the rates of proximal deep vein thrombosis on routine ultrasound performed 7 days after TKA in patients with RA versus those with OA (unadjusted OR 0.81 [95% CI 0.26–2.52]) (31).
To our knowledge, this is the first systematic review to assess rates of complications following THA and TKA in patients with RA versus OA. In this review of 40 studies, we found strong evidence for increased risk of hip dislocation following THA, and fair evidence for increased risk of infection following TKA, in patients with RA versus OA. We also found fair evidence for increased risk of early revision following TKA, but no evidence to support any differences in later revision rates, 90-day mortality, or rates of venous thromboembolic events following THA or TKA in patients with RA versus OA.
The increased risk of dislocation following THA in patients with RA versus OA was an unexpected finding which, to our knowledge, has not been well recognized previously, possibly due to the lack of meta-analysis. Susceptibility to dislocation could be the result of poorer soft tissue quality in RA relative to OA, resulting in suboptimal hip abductor strength postoperatively. Other potential explanations include systematic differences in surgical approach, head size, or use of bone cement between groups. However, Khatod et al adjusted for surgical approach as well as head size, and still noted an increased risk of dislocation among RA patients (32). Further research is warranted to confirm this finding and elucidate potential modifiable risk factors such as head size, surgical approach, or use of cement.
It makes intuitive sense that risk of infection following TJA would be increased in patients with RA versus OA due to differences in the pathogenesis and medical management of these conditions. As a systemic autoimmune disease, RA is typically treated with immunosuppressive agents, including systemic corticosteroids, methotrexate, antimalarial drugs, and more recently, biologic agents, e.g., adalimumab and etanercept (33–35). Evidence regarding the effect of these medications on the rate of postoperative infection is inconsistent (34, 36).
Our review revealed fair evidence to support the notion of an increased risk of infection following TKA in patients with RA versus OA. Three of four studies showed increased infection risk following TKA in patients with RA, and 2 of these studies had a low risk of bias. This is consistent with our finding that RA patients had higher odds of revision within 5 years of TKA (although this did not include adjustment for confounders). Only 1 study showed an increased risk of infection following THA in patients with RA, and this study had a low risk of bias (29). In that study it was also found that perioperative systemic corticosteroid use and withdrawal of biologic treatment prior to surgery did not have any effect on the rate of infection following THA or TKA in patients with RA (29). However, most of the study reports included no comment on the role of drug use in the rate of infection following arthroplasty. Our review has identified the need for larger studies to confirm the present findings regarding infection risk after TKA in RA patients and if they are confirmed, to determine the effect of specific medical therapies and other factors on risk of infection in RA.
Although dislocation rates following THA were higher in RA patients than in OA patients, this was not reflected in an increased rate of early revision following THA. Potential explanations for this disparity include the following: 1) there may be systematic differences in the management of dislocations between RA and OA, with the former being preferentially managed with closed reduction and activity modification due to either surgeon or patient preference (37), or 2) it may be that an increased rate of early revision in RA does exist, but was not observed in our study due to lack of power.
We also did not find a difference between RA and OA patients in terms of 90-day mortality or venous thromboembolic events following THA or TKA. However, as the reported risk of these complications is low, the studies reviewed may have been inadequately powered to detect significant differences in rates between the 2 patient groups.
RA can be a challenging diagnosis to establish clinically, particularly early in the disease (38). Reports of only 3 of the 40 studies (7.5%) discussed the process by which the diagnosis of RA was established. One of these used the 1987 American College of Rheumatology criteria (39), and the other 2 used a referral diagnosis from an internist or rheumatologist. Sixteen studies (40%) utilized administrative databases or arthroplasty registries to establish the diagnosis of RA; of these, only 1 report provided information on the validity of the codes. The data in administrative databases are typically collected for non-research purposes, and without appropriate reabstraction and validation, the diagnosis may be inaccurate (40). Even a diagnosis recorded by the surgeon intraoperatively may be inaccurate, as the index joint of a patient with RA may have an appearance that is more consistent with OA, and be recorded as the latter (23). While these patients may have joint and/or bone characteristics more similar to those found in OA, they will still have the other risk factors for complications associated with RA, and should be classified as such. The misclassification of RA patients as having OA would reduce our ability to identify differences between these groups. Thus, we may have underestimated differences in TJA outcomes between OA and RA patients, where they exist.
Despite the large numbers of patients included in the studies reviewed, variable adjustment for confounders limited our ability to pool results, and ultimately lowered our statistical power. Most of the studies reviewed (29 of 40 [72.5%]) did not, or were inadequately powered to, adjust for potential confounders, including age, sex, and comorbidity.
It has been established that younger patients (<70 years old) have a higher revision rate following TJA, which is hypothesized to be due to increased activity leading to wearing out of the prosthesis (41). While this assumption may not apply to younger patients with RA, who may have more activity-limiting polyarticular joint involvement than patients with OA at a comparable age, lack of controlling for the effect of age is a significant limitation (42). Increased comorbidity, including specific conditions such as diabetes, is associated with early infection and subsequent revision (43), while cardiovascular disease may increase the risk of mortality following TJA (44). RA is a systemic disease characterized by an increased risk of cardiovascular disease (45), while OA is associated with a high prevalence of concomitant obesity, diabetes, hypertension, and heart disease (46, 47). Although our focus was on elective primary TJA procedures, in which patients are typically medically fit for surgery, the presence of comorbid conditions contributes to differential risk of TJA complications in patients with RA versus OA. Adequate consideration of and control for comorbidity is warranted in future studies examining the rates of revision and other complications in RA versus OA.
Since none of the meta-analyses that were included contained pooled results on 10 or more studies, we did not assess funnel plots for asymmetry. However, we did assess differences between studies via I-square values, which describe the percentage of total variation across studies that is due to heterogeneity rather than chance (48). I-square values of 25%, 50%, and 75% are considered low, moderate, and high, respectively. Our analyses yielded a broad range of I-square values (0–85%), although values were ≤50% in most of our meta-analyses (8 of 11). This degree of heterogeneity is consistent with our pooling of studies that utilized diverse types of implants, variable surgical approaches and techniques, and differing diagnostic criteria to define RA versus OA.
In summary, compared with TJA recipients with OA, we found that those with RA were at higher risk of dislocation following THA and higher risk of infection following TKA. However, after adjustment for covariates, we found no difference regarding risk of revision, 90-day-mortality, or venous thromboembolic events within 90 days of either THA or TKA. Adequately powered studies, which incorporate validated definitions for RA and OA diagnosis and control for appropriate confounders and other covariates, are needed to confirm these findings. Where differences are confirmed, further research is warranted to elucidate potential explanations, including the role of prosthesis type, medication use (e.g., biologic therapies), comorbidity, and bone quality. The results of such studies would be useful to guide decision-making regarding TJA in the setting of RA.
All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Ravi 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 conception and design. Ravi, Shah, Jenkinson, Bogoch, Kreder, Hawker.
Acquisition of data. Ravi, Escott, Jenkinson, Chahal.
Analysis and interpretation of data. Ravi, Shah, Hawker.