Dr. Bogoch has received consulting fees, speaking fees, and/or honoraria from Merck Frosst Canada and Merck Sharpe & Dohme (more than $10,000 each) and has received research support as a principal investigator from Amgen Canada, Novartis Canada, and Warner Chilcott.
University of Toronto and Institute for Clinical Evaluative Sciences, Toronto, Ontario, Canada
The opinions, results, and conclusions reported herein are those of the authors and are independent from the funding sources. No endorsement by the Institute for Clinical Evaluative Sciences or the Ontario Ministry of Health and Long-Term Care is intended or should be inferred.
Most of the evidence regarding complications following total hip arthroplasty (THA) and total knee arthroplasty (TKA) are based on patients with osteoarthritis (OA); less is known about outcomes in rheumatoid arthritis (RA). Using a validated algorithm for identifying patients with RA, we undertook this study to compare the rates of complications among THA and TKA recipients between those with RA and those without RA.
In patients who underwent a first primary elective THA or TKA between 2002 and 2009, those with RA were identified using a validated algorithm: a hospitalization with a diagnosis code for RA or 3 physician billing claims with a diagnosis code for RA, with at least 1 claim by a specialist (rheumatologist, orthopedic surgeon, or internist) in a 2-year period. Recipients with diagnostic codes suggesting an inflammatory arthritis, but not meeting RA criteria, were classified as having inflammatory arthritis. All remaining patients were deemed to have OA. Cox proportional hazards models, censored on death, were used to determine the relationship between the type of arthritis and the occurrence of specific complications, adjusting for potential confounders (age, sex, comorbidity, and provider volume).
We identified 43,997 eligible THA recipients (3% with RA) and 71,793 eligible TKA recipients (4% with RA). Total joint arthroplasty recipients with RA had higher age and sex–standardized rates of dislocation following THA (2.45%, compared with 1.21% for recipients with OA) and higher age and sex–standardized rates of infection following TKA (1.26%, compared with 0.84% for recipients with OA). Controlling for potential confounders, recipients with RA remained at increased risk of dislocation within 2 years of THA (adjusted hazard ratio [HR] 1.91, P = 0.001) and remained at increased risk of infection within 2 years of TKA (adjusted HR 1.47, P = 0.03) relative to recipients with OA.
Patients with RA are at higher risk of dislocation following THA and are at higher risk of infection following TKA relative to those with OA. Further research is warranted to elucidate explanations for these findings, including the roles of medication profile, implant choice, postoperative antibiotic protocol, and method of rehabilitation following joint replacement.
Total joint arthroplasty (TJA) is the surgical treatment for end-stage arthritis of the hip and knee, a stage defined as ongoing pain, limitation in function, and reduced quality of life resulting from joint disease, despite appropriate medical management ([1, 2]). Over the last decade, the age and sex–standardized rates of total hip arthroplasty (THA) and total knee arthroplasty (TKA) in North America have increased by approximately 25% and 65%, respectively (). This increase appears to be largely driven by patients with osteoarthritis (OA) (), with rates in patients with rheumatoid arthritis (RA) remaining largely static ([5-7]), although there is some evidence to suggest that the rates of TKA are increasing in patients with RA as well (). On average, patients report significant and sustained improvement in function and quality of life following THA and TKA (). However, there is a small risk of serious complications, including requirement for early revision surgery, infection, dislocation, venous thromboembolism, and death (all <2%) ([10-12]).
However, the estimated rates of these outcomes are largely based on the experiences of patients with OA, who are the majority of TJA recipients ([13, 14]). TJA is also used to treat patients with RA, a systemic autoimmune disease that differs fundamentally from OA in terms of pathogenesis, prognosis, and medical management. As such, the risk of complications following TJA, and their predictors, may differ between RA and OA. Therefore, it is not clear to what degree TJA outcomes based on the experience of patients with OA may be generalized to recipients with RA.
We performed a systematic review and meta-analysis to examine TJA complication rates for patients with RA and those with OA (). We found evidence of an increased risk of hip dislocation following THA and an increased risk of infection following TKA in patients with RA. However, these results were based on the findings of studies that had several limitations. The majority of these studies used different definitions of RA diagnosis, which raised concerns about misclassification bias. Some studies combined primary and revision procedures, which limited their generalizability and precluded meta-analysis. Finally, there was variable adjustment for confounders, with several studies not accounting for any of the systematic differences between RA and OA patients. Thus, it remains unclear whether the risks of dislocation following THA and infection following TKA are increased in patients with RA versus those with OA. Confirmation of these findings is important. In addition to the significant morbidity associated with these complications ([16, 17]), if joint replacement recipients with RA are found to be at greater risk of surgical complications, this may have implications for surgical decision-making. The current study sought to overcome the limitations of prior studies. Capitalizing on the availability of a recently validated algorithm for RA using administrative data with high sensitivity and specificity, we compared the rates of surgical complications following THA and TKA in recipients with RA and those with OA, controlling for potential confounders.
PATIENTS AND METHODS
The main data sources were hospital discharge abstracts from the Canadian Institute for Health Information Discharge Abstract Database (CIHI-DAD) and physician claims from the Ontario Health Insurance Plan (OHIP). Using specific procedure and diagnostic codes from the ICD-10-CA/CCI (International Statistical Classification of Diseases and Related Health Problems, Tenth Revision, Canada [ICD-10-CA] and Canadian Classification of Health Interventions [CCI]), we defined a cohort of patients who received their first primary elective THA or TKA between April 1, 2002 and March 31, 2009 (further information is available at http://www.womensresearch.ca/our-research/musculoskeletal-health/canadian-osteoarthritis-research-program/). The last date for followup was March 31, 2011. We excluded the records of those who underwent primary or revision TJA prior to April 1, 2002 (i.e., prebaseline), those for whom the first procedure was nonelective (e.g., for cancer, fracture, or external cause of injury) or revision surgery, and those with a history of joint infection, bilateral procedures at the index admission, or for whom the operating surgeon could not be identified (Figure 1).
Main effect—arthritis diagnosis
The diagnosis of RA was defined using the following validated algorithm: a hospitalization with a diagnosis code for RA or 3 physician billing claims with a diagnosis code for RA, with at least 1 claim by a specialist (rheumatologist, orthopedic surgeon, or internist) in a 2-year period (sensitivity 78%, specificity 100%) (). TJA recipients with at least 2 diagnostic codes for RA but not meeting our criteria for diagnosis were classified as having inflammatory arthritis. Patients with at least 1 diagnostic code for psoriasis or gout from a rheumatologist or internist, or with a diagnostic code for ankylosing spondylitis or systemic lupus erythematosus from any physician, were also classified as having inflammatory arthritis. All remaining patients were deemed to have OA.
Patient and provider characteristics
We controlled for patient and provider factors that have previously been linked with complications following joint replacement. Relevant patient demographic information was obtained from the OHIP Registered Persons Databases; this included age, sex, income quintile, and rurality ([9, 19, 20]). Comorbidities listed on hospital discharge abstracts in the 3 years before the index TJA admission were coded according to the Deyo adaptation of the Charlson Comorbidity Index ([21-23]). The presence of specific comorbidities (chronic obstructive pulmonary disease, congestive heart failure, baseline cardiovascular disease risk, diabetes, and hypertension) was identified by validated criteria using hospital discharge abstracts ([24-26]) and a look-back period of 2 years.
Additionally, Adjusted Clinical Groups, based on diagnosis codes from hospitalizations and physician visits in the 2 years before the index TJA admission, were used to classify recipients as frail (yes/no) (). Frailty is associated with a decline in physical activity and muscle strength, as well as immune system dysfunction ([28-31]). Frailty may therefore increase risk of falls leading to fracture or dislocation following TJA.
For each TKA, the volume of arthroplasties performed for each surgeon and each hospital was defined as the number of primary and revision knee replacement procedures performed in the 365 days prior to the index TKA procedure. This was repeated for each index THA procedure. We defined teaching hospitals as those that were members of the Council of Academic Hospitals of Ontario (www.caho-hospitals.com).
All patients were followed up from the date of their index TJA for the occurrence of venous thromboembolism and death within 90 days, and for infection of the replaced joint, dislocation of the replaced joint (only for THA), periprosthetic fracture, and revision of the index arthroplasty within 2 years. We also compared groups for the occurrence of a composite outcome encompassing all of the above complications (venous thromboembolism, death, infection, dislocation, fracture, and revision surgery). We limited followup to 2 years, as revision surgery during this period is considered a catastrophic complication ([10, 32, 33]). We identified death within 90 days of operation using the OHIP Registered Persons Databases (further information is available at http://www.womensresearch.ca/our-research/musculoskeletal-health/canadian-osteoarthritis-research-program/). Occurrence of a venous thromboembolism was defined as the presence of a diagnostic code for a deep vein thrombosis or pulmonary embolism in the CIHI-DAD or National Ambulatory Care Reporting System database.
Infections were identified using 3 methods (): 1) occurrence of an ICD-10-CA diagnostic code for intraarticular infection, with a confirmatory code for an irrigation and debridement; 2) occurrence of an OHIP code for a spacer insertion; and/or 3) occurrence of a procedure code for a peripheral intravenous central catheter line after the TJA, where the referring physician was an orthopedic surgeon. Dislocations were defined as the occurrence of a diagnostic code for dislocation or a procedure code for closed/open hip reduction (). Periprosthetic fractures were defined as the occurrence of a diagnostic code for fracture following insertion of an implant. Revision procedures were identified using ICD-10-CA/CCI procedure codes accompanied by the supplementary status attribute “R” ().
Age and sex–standardized rates of each complication according to the type of arthritis were determined using indirect standardization. Univariate analysis was used to compare demographics of TJA recipients at the time of their index surgical admission according to type of arthritis (RA versus OA). Wilcoxon's rank sum tests were used to compare non–normally distributed continuous variables. Chi-square and Fisher's exact tests were used to compare categorical variables. Cox proportional hazards models, censored on death and accounting for clustering of patients within surgeons, were used to estimate the hazard of each outcome for patients with RA and those with inflammatory arthritis, relative to those with OA. Age and sex were included in all multivariate models. Other variables were included if in the univariate analyses the covariate was associated with the outcome with a P value of ≤0.4. Other potential variables included income quintile, rurality, Charlson Comorbidity Index score, frailty, volume of arthroplasties performed for each surgeon and each hospital in the year preceding the surgery, and performance of the TJA at a teaching hospital.
We performed 2 sensitivity analyses: 1) repeating the multivariate analysis after excluding patients with a prior code for osteonecrosis, and 2) repeating the analyses after limiting the definition of infection to cases managed specifically by irrigation and debridement. Where appropriate, unadjusted hazard ratios (HRs) and adjusted HRs with 95% confidence intervals (95% CIs) are reported. All analyses were performed using SAS version 9.2 for UNIX (SAS Institute). The probability of a Type I error was set to 0.05 for all analyses.
Between April 1, 2002 and March 31, 2009, 60,305 THAs and 89,713 TKAs were performed in Ontario, Canada (Figure 1). Among patients undergoing these surgeries, 43,997 eligible THA recipients (37,881 with OA [86%], 4,953 with inflammatory arthritis [11%], 1,163 with RA [3%]) and 71,793 eligible TKA recipients (59,564 with OA [83%], 9,537 with inflammatory arthritis [13%], 2,692 with RA [4%]) were identified. THA and TKA recipients with RA were younger, more likely to be female, had greater comorbidity, and were more likely to be frail relative to recipients with OA (see Tables 1 and 2 for THA and TKA, respectively). Patients with inflammatory arthritis had a variety of inflammatory arthritides. Fifty-eight percent had diagnostic codes for RA but did not meet the criteria set by our algorithm. The remaining 42% met the described criteria for psoriatic arthritis, systemic lupus erythematosus, or ankylosing spondylitis. Patients classified as having inflammatory arthritis tended to have demographic characteristics intermediate between those of OA and RA patients (Tables 1 and 2).
Table 1. Characteristics of the eligible THA recipients by type of arthritis*
OA (n = 37,881)
Inflammatory arthritis (n = 4,953)
RA (n = 1,163)
Total (n = 43,997)
P, RA vs. OA
Except where indicated otherwise, values are the number (%) of patients. OA = osteoarthritis; RA = rheumatoid arthritis; IQR = interquartile range; CVD = cardiovascular disease; COPD = chronic obstructive pulmonary disease; CHF = congestive heart failure.
aNumber of primary and revision hip replacement procedures performed in the 365 days prior to the index total hip arthroplasty (THA) procedure.
Surgical complication rates following THA in patients with RA and those with OA
A total of 2,119 THA recipients (4.8%) experienced one or more surgical complications. These included 569 hip dislocations (1.3%), 612 venous thromboembolisms (1.4%), 487 joint infections (1.1%), 515 revision THAs (1.2%), 160 periprosthetic fractures (0.4%), and 220 postoperative deaths (0.5%) (Table 1). Compared with THA recipients with OA, those with RA were more likely to experience any surgical complication (5.7% versus 4.7%; P = 0.01) and hip dislocation (2.6% versus 1.2%; P < 0.001) and less likely to experience a venous thromboembolism (0.4% versus 1.4%; P = 0.02). Additionally, the age and sex–standardized rates of dislocation were higher in patients with RA, and the rates of venous thromboembolism were lower in patients with RA, relative to those with OA (Table 3). Controlling for other factors, RA diagnosis remained an independent and significant risk factor for dislocation following THA (adjusted HR 1.91 [95% CI 1.29–2.82], P = 0.001) (Table 4) and was protective against venous thromboembolism (adjusted HR 0.37 [95% CI 0.16–0.88], P = 0.02).
Table 3. Age and sex–standardized rates of complications*
Values are the rate per 100 arthroplasty recipients (95% confidence interval). OA = osteoarthritis; RA = rheumatoid arthritis.
Total hip arthroplasty
Venous thromboembolism within 90 days
Death within 90 days
Revision surgery within 2 years
Periprosthetic fracture within 2 years
Infection within 2 years
Dislocation within 2 years
Total knee arthroplasty
Venous thromboembolism within 90 days
Death within 90 days
Revision surgery within 2 years
Periprosthetic fracture within 2 years
Infection within 2 years
Table 4. Results from fully adjusted Cox proportional hazards models*
Dislocation within 2 years of THA
Venous thromboembolism within 90 days of THA
Infection within 2 years of TKA
Adjusted HR (95% CI)
Adjusted HR (95% CI)
Adjusted HR (95% CI)
Variables that did not meet selection criteria in univariate analyses were not included in the relevant multivariable model(s). OA = osteoarthritis; RA = rheumatoid arthritis.
aNumber of primary and revision hip or knee replacement procedures performed in the 365 days prior to the index total hip arthroplasty (THA) procedure or prior to the index total knee arthroplasty (TKA) procedure, respectively.
bValues are unadjusted hazard ratios (HRs) and 95% confidence intervals (95% CIs).
No differences were found between THA recipients with RA and those with OA with respect to risk of infection, revision surgery, periprosthetic fracture, or death following THA. Among those who had a dislocation in the 2 years following their THA, the median (interquartile range [IQR]) number of days to the dislocation was shortest for recipients with RA (OA patients, median 263 days [IQR 129–420 days]; patients with inflammatory arthritis, median 383 days [IQR 210–567 days]; RA patients, median 211 days [IQR 35–234 days]) (Figure 2).
Surgical complication rates following TKA in patients with RA and those with OA
A total of 2,577 TKA recipients (3.6%) experienced one or more surgical complications. These included 1,160 venous thromboembolisms (1.6%), 637 joint infections (0.9%), 850 revision TKAs (1.2%), 60 periprosthetic fractures (0.1%), and 299 postoperative deaths (0.4%) (Table 2). There was no difference between TKA recipients with RA and those with OA with respect to the occurrence of our composite outcome (3.6% versus 3.6% experienced ≥1 complication; P = 0.58) or with respect to the risk of venous thromboembolism, revision surgery, periprosthetic fracture, or death following TKA. When the values were unadjusted for other factors, TKA recipients with RA were at increased risk of infection relative to TKA recipients with OA (1.2% versus 0.8%; P = 0.02). The age and sex–standardized rate of infection was higher in patients with RA than in those with OA (Table 3). Controlling for potential confounders, RA diagnosis remained significantly and independently predictive of increased infection risk following TKA (adjusted HR 1.47 [95% CI 1.05–2.05], P = 0.03) (Table 4). Among patients who had an infection in the 2 years following their TKA, the median (IQR) number of days to the infection was least for recipients with RA (OA patients, median 468 days [IQR 199–732 days]; patients with inflammatory arthritis, median 429 days [IQR 229–717 days]; RA patients, median 196 days [IQR 98–403 days]) (Figure 2).
We repeated our analyses after excluding persons who had a diagnostic code for osteonecrosis, and persons with RA remained at an increased risk of dislocation following THA (adjusted HR 1.90 [95% CI 1.26–2.87], P = 0.002) and an increased risk of infection following TKA (adjusted HR 1.47 [95% CI 1.05–2.05], P = 0.03). We also repeated our analyses after limiting the definition of infection specifically to the occurrence of an irrigation and debridement procedure, and the elevated risk of infection following TKA in patients with RA persisted (adjusted HR 1.61 [95% CI 1.01–2.57], P = 0.04).
In a large population cohort, and controlling for previously identified predictors of TJA complications, we found that TJA recipients with RA were approximately twice as likely to experience a dislocation following THA, and one-and-one-half times as likely to experience an infection following TKA, relative to recipients with OA. THA recipients with RA were also one-third as likely to experience a venous thromboembolism as those with OA. No differences were found with respect to risk of infection following THA, risk of venous thromboembolism following TKA, or risk of fracture or death following either procedure.
The increased risk of dislocation is consistent with the findings of previous studies by Conroy et al and Khatod et al ([36, 37]), and this finding remained robust after controlling for systematic differences in health status between THA recipients with RA and those with OA (including a measure of frailty) that were associated with increased risk of dislocation. Dislocation following THA is a serious complication. In addition to being very painful and necessitating revision arthroplasty and/or aggressive rehabilitation following closed reduction (), dislocations are estimated to increase the hospital costs of a primary THA by >300% (). Potential explanations for our findings include systematic differences in the size of implants used for patients with RA and those used for patients with OA. On average, patients with RA have a lower body mass index (BMI) than patients with OA ([39, 40]); this smaller body size may result in a tendency for surgeons to use a smaller femoral head component, which may increase the risk of dislocation ([36, 41]). However, in their study examining dislocation following THA, Khatod et al controlled for femoral head size and still found an elevated risk in RA patients ().
Dislocation risk in RA may also be the result of anatomic differences at the level of the hip joint between patients with RA and those with OA. For example, studies have suggested that the prevalence of protrusio acetabuli is increased in patients with RA ([42, 43]). Acetabular protrusion may increase the risk of impingement and subsequent dislocation. Additionally, there is an increased prevalence of osteoporosis in patients with RA (), which may increase the risk of fractures and dislocations. There may also be a systematic difference in surgical approach between types of arthritis that accounts for our results ([45-47]). Finally, increased susceptibility to dislocation in patients with RA may result from poorer soft tissue quality than that in patients with OA (), resulting in suboptimal hip abductor strength and soft tissue laxity postoperatively (). Further research is warranted to elucidate the effect of these and other factors on the risk of dislocation in patients with RA.
The documented reduced risk of venous thromboembolism following THA in patients with RA compared with patients with OA was not expected. Previous studies have demonstrated an increased risk of deep vein thrombosis and pulmonary embolism in patients with RA relative to the age and sex–matched general population without RA, likely due to the effects of systemic inflammation and endothelial dysfunction ([50, 51]). Our findings may be the result of selection bias; it may be that healthier patients with RA are selected for THA and/or there may be an overall higher risk of venous thromboembolism among patients with advanced OA relative to the age and sex–matched general population without OA ().
The increased risk of infection following TKA in patients with RA compared with patients with OA persisted after adjusting for systematic differences between these groups in terms of health status (comorbidity and frailty) (). In addition to being a source of considerable morbidity, each infected arthroplasty is estimated to cost >$30,000, on average, to manage (). Foremost among the potential explanations for this increased risk is that RA is a systemic autoimmune disorder, and it is managed with immunomodulatory drugs that have been linked to increased risk of infection (disease-modifying antirheumatic drugs, systemic corticosteroids, antimalarials, and biologic therapies) ([54-57]). However, to our knowledge, the impact of biologic agents on the risk of infection following TJA has not been examined. It is also unclear why this increased risk of infection occurs following TKA, but not following THA. Future studies are required to determine whether use of specific medications contributes to the elevated risk of infection in TKA recipients with RA.
A major strength of our study was its use of a validated algorithm using administrative data for identification of RA patients. RA is a challenging diagnosis to establish clinically (). The algorithm used in this study has been shown to have very high specificity (100%), which allowed us to be confident that these patients had RA. Further, to increase the specificity of our OA diagnosis, we classified TJA recipients who had codes for RA but who did not meet our criteria, as well as those who had diagnostic codes suggestive of other inflammatory arthritides, into an inflammatory arthritis group. The results in our inflammatory arthritis group were intermediate between those in the other 2 groups. This suggests that pooling of this group with either RA patients or OA patients, as some studies have done, would lead to an underestimation of the surgical risk differences between RA and OA.
Although the risks of dislocation and infection were more common in recipients with RA after THA and TKA, respectively, we did not find a concomitant increased risk of revision surgery. It is possible that in patients with RA, surgeons may preferentially manage hip dislocations with closed reduction and activity modification (), and knee infections with antibiotics or irrigation and debridement, instead of revision arthroplasty for either complication.
Other strengths of our study include its large sample size and adjustment for potential confounders. However, there are also important limitations. First, we did not have information regarding patient-reported outcomes, and thus we cannot comment on whether the observed increased risks of dislocation and infection translated into worse patient-reported pain or disability following THA and TKA, respectively. Second, we did not have information on the types of implant used, and thus we could not determine whether systematic differences in the use of specific implants might explain the observed differences in complication risk. Finally, although we had access to information regarding several potentially important confounders, there are others for which we had no information, such as BMI. Thus, we cannot exclude the possibility that our results were due to lack of control for important unmeasured confounders.
In summary, compared to recipients with OA, we found that those with RA were at significantly greater risk of dislocation following THA and significantly greater risk of infection following TKA. As both dislocation and infection lead to significant morbidity and drastically increase health care costs, research is warranted to elucidate explanations for this increased risk among patients with RA. This information will be valuable to inform patient management decisions, including the development and implementation of strategies designed to target modifiable risk factors.
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, Hollands, Paterson, Kreder, Hawker.
Acquisition of data. Ravi, Hollands.
Analysis and interpretation of data. Ravi, Croxford, Paterson, Bogoch, Kreder, Hawker.