Revision Total Hip Arthroplasty in Solid Organ Transplant Patients: A Propensity Score-Matched Cohort Study for Aseptic and Infected Revisions

Background Previous studies have demonstrated that solid organ transplant (SOT) patients undergoing primary total hip arthroplasty (THA) are at an increased risk of postoperative complications. The purpose of this study is to use a large, national database to investigate revision THA (rTHA) outcomes in SOT patients. Methods Nationwide Readmissions Database (NRD) from 2010-2018 was used, and ICD-9 and ICD-10 codes were used to identify all patients who underwent rTHA, including those with history of SOT. Propensity score matching (PSM) was used to analyze rTHA outcomes in SOT patients comparted to matched controls. Separate analysis performed for patients undergoing rTHA for prosthetic joint infection (PJI) vs other causes. Results A total of 414,756 rTHA, with 1837 of those being performed in SOT patients, were identified. Of these, 65,961 and 276 were performed for PJI in non-SOT and SOT patients, respectively. For non-PJI patients, SOT patients had higher 90-day all-cause readmission rates (24.0% vs 19.4%, P = .03) but lower rate for readmission related to rTHA (6.0% vs 9.2%, P = .03), but no difference readmission for specific rTHA complications, mortality (0.6% vs 1.3%, P = .20), or revision rTHA. Of PJI patients, SOT patients had no difference in overall 90-day readmission (38.6 vs 31.3%, P = .280), readmission for specific rTHA complications, re-revision, or mortality (4.7% vs 6.0%, P = .63). Conclusions SOT patients undergoing rTHA for aseptic reasons are higher risk of overall readmission but lower risk of readmission related to rTHA than appropriately matched controls. SOT PJI patients undergoing had similar rates of readmission, mortality, and revision surgery compared to matched non-SOT PJI patients.


Introduction
Total hip arthroplasty (THA) has long been established as a safe and effective treatment for hip arthritis with significant benefits for appropriately selected patients [1]. Given the advances in surgical technique, perioperative management, and low overall complications rates, the rate of THAs in medically complex patients is also expected to grow. Researchers predict the number of THA procedures performed annually in the United States to reach 900,000 by 2030, and 1.23 million by the year 2060. During this same time period, they also predict the number of rTHA performed to increase by 219% to 110,000 annually [2].
At the same time, survivorship in patients who have undergone solid organ transplant (SOT) has steadily improved in part because of improvements in surgical technique and postoperative No author associated with this paper has disclosed any potential or pertinent conflicts which may be perceived to have impending conflict with this work. For full disclosure statements refer to https://doi.org/10.1016/j.artd.2021. 10.007. immunosuppression and surveillance protocols [3]. As SOT patients live longer, the incidence of symptomatic arthritis is likely to increase, leading to an increased demand for both primary and revision joint replacement procedures [4,5]. SOT patients are also at higher risk to develop femoral head avascular necrosis, patients with renal transplants, in particular, undergo THA at 5-8 times the rate of the general population [6].
In primary joint replacement, patients with a history of SOT have been shown to have a higher rate of morbidity and mortality [7,8]. It is less clear how SOT patients do after revision THA (rTHA). A retrospective case series of 30 patients found an elevated risk for re-revision, particularly for prosthetic joint infection (PJI) in patients with a history of SOT [9]. More recently, Labaran et al. performed a study using a national administrative database and identified 661 rTHAs performed specifically in renal transplant recipients and found an increased risk of 90-day hospital readmission, 1-year septicemia, and 1-year mortality compared to matched controls [10]. Unfortunately, there are no currently available data in the literature which stratify outcomes based on aseptic vs infected revisions, which has been shown to influence the risk for 30-day complications after arthroplasty [11,12]. Furthermore, no studies have investigated outcomes more broadly for all types of SOT patients.
We present a propensity score matched cohort study using a large national administrative database to compare the outcomes of aseptic and infected revisions in patients with a history of SOT to matched controls. We hypothesize that SOT patients undergoing rTHA will have a higher length of stay (LOS) and increased rate perioperative complications.

Material and methods
The study cohort was identified from the Nationwide Readmissions Database (NRD) over a 9-year study period (2010e2018). The NRD is a nationally representative database developed and validated through a federal-state-industry partnership sponsored by the Agency for Healthcare Research and Quality. It is based on 22 state inpatient databases that track patients across multiple hospitals. Approximately 51.2% of the US population and 49.3% of all US hospitalizations were sampled in a stratified algorithm, designed to allow for estimation of nationally representative statistics. Available variables include demographic data, diagnoses, procedures, cost, LOS, and hospital characteristics. Because the NRD database has been sufficiently deidentified, this study was deemed exempt by the institutional review board at our institution.
Patients older than 18 years who were admitted for rTHA were considered for this study. Patients were identified using the International Classification of Diseases, ninth and tenth revision, (ICD-9 and ICD-10) procedure codes (Table 1). For ICD-10 procedure codes, patients required either a revision code or a relevant removal and replacement code as described [13]. Patients were separated into groups based on whether or not they had a diagnosis of SOT (Table 1). Indication for rTHA was determined based on associated ICD-9 and ICD-10 diagnostic codes (Appendix Table 1). All subsequent readmissions were considered for these groups. Baseline comorbidity was quantified using the Elixhauser Comorbidity Index (ECI), a composite score of 30 comorbid conditions using all admission diagnoses and the comorbidity package in R. [14] Higher ECI scores corresponded to greater burden of comorbid conditions. ECI score component variables were also extracted.
The primary outcomes of interest include 90-day and 180-day mortality, readmission rates, as well as readmission stratified by indication, and re-revision THA. Secondary outcomes included complications during index hospitalization. ICD-9 and ICD-10 codes were used to identify cardiac complications, myocardial infarction, cerebrovascular accident, respiratory complications, pneumonia (PNA), pulmonary embolism (PE), other pulmonary complications, deep vein thrombosis (DVT), acute kidney injury, wound complications, postoperative blood transfusions, or any in-hospital complications (Appendix Table 2).
Propensity score matching (PSM) was performed to compare relative risks (RR) of re-admissions and complications in transplant and non-transplant patients. [15] A propensity score multivariate logistic regression model was created using patient age, sex, ECI, hospital type, hospital size, insurance status, and zip code income quartile. Specific medical comorbidities were also in the model, including history of CHF, cardiac arrhythmia, pulmonary hypertension, chronic pulmonary disease, essential hypertension, diabetes, obesity, coagulopathy, solid tumor, and alcohol abuse. Patients undergoing rTHA for infection were analyzed separately from those undergoing rTHA for other reasons (ie, loosening, instability). Propensity scores were used to match transplant patients to nontransplant patients at a ratio of 1:3 with replacement improved balance using the MatchIt package in R [16]. Relative risk was estimated using weighted logistic regression.
All result sample sizes represented national estimates taking into account the NRD's stratified 2-stage cluster design incorporating individual discharge-level weights. Descriptive analysis was used to describe both baseline characteristics and outcome Table 1 Diagnostic and procedural codes used to identify patients with a history of SOT and rTHA procedure types. parameters within each comparison group. Categorical variables are compared using the chi-squared statistic, except when individual cell counts were less than 10, in which case the Fisher exact test was used. Continuous variables were reported using mean, 95% confidence interval (CI), and P values and were compared using the student t-test after ensuring normal distributions. For skewed distributions, continuous variables are presented as median (interquartile range) and the Wilcoxon rank-sum test. All tests were unpaired with a significance level defined as a 2-tailed P of 0.05. Statistical analyses were performed using R 3.6.0 (R Foundation for Statistical Computing, Vienna, Austria).

Infected vs noninfected patients
Univariate analysis of differences in outcomes of infected and noninfected rTHA patients showed a significant difference for SOT and non-SOT patients in the outcomes of interest. SOT patients undergoing rTHA for infection had a significantly longer LOS (10.2 vs 5.2 days, P < .001), a higher likelihood of readmission (35.3% vs 23.7%, P ¼ .03) and a higher risk of subsequent revision surgery (11.6% vs 3.2%, p0.002) compared to those revised for noninfectious indications. Non-SOT patients undergoing rTHA for infection had a significantly longer LOS (9.6 vs 4.5 days, P < .001), higher likelihood of readmission (36.3% vs 16.8%, P < .001) and a higher likelihood of subsequent revision (16.6% vs 5.4%, P < .001) compared to those revised for non-infectious indications.

PSM analysis of Non-PJI patients
For aseptic patients, SOT patients were more likely to be readmitted for any reason within 90 days of surgery (20.4% vs 19.4%, P ¼ .028) but less likely to be admitted with a diagnosis related to RTHA failure compared to matched controls (6.0% vs 9.2%, P ¼ .034, Table 4). SOT patients were less likely to be readmitted with a diagnosis of PJI compared to matched controls (1.4% vs 3.9%, P ¼ .01). There was no difference between SOT patients and matched controls regarding readmission for complications secondary to loosening, instability or fracture (Table 4). Compared to matched controls, SOT patients had a lower rate of 90-day overall re-revision THA with the difference approaching statistical significance (3.2% vs 5.5%, P ¼ .050). There was no difference in 90-day mortality between SOT and matched non-transplant patients undergoing rTHA for aseptic causes (0.6% vs 1.3%, P ¼ .20).
SOT patients had a lower rate of any complication (42.0% vs 48.3%, P ¼ .007) and a lower risk of receiving post-operative blood transfusions (28.3% vs 36.4%, P < .001, Appendix Table 3) comparted to controls. There were no differences between groups with regard to cardiac, pulmonary, renal, DVT, PE and wound complications, specifically. The rates of PE and DVT were also not significantly different between groups. The LOS was significantly shorter for SOT patients comparted to controls (5.0 vs 6.0 days, P < .001).

PSM analysis of PJI patients
For infected patients, there was no difference in overall 90-day readmission rate (36.8% vs 40.3%, P ¼ .54), readmission related to RTHA, or for complications secondary to loosening, infection, instability, and fracture ( Table 5). The rate of revision surgery was not different between groups at 90 days after RTHA (12.3% vs 17.9%, P ¼ .19). SOT patients had a similar rate of 90-day mortality compared to matched patients (4.7% vs 6.0%, P ¼ .63).
There was no difference in index hospitalization wound complications (9.1% vs 9.6%, P ¼ .87, Appendix Table 4) or postoperative blood transfusions (37.8% vs 45.0%, P ¼ . 15). There were no differences between groups in all complications or cardiac, pulmonary, or renal complications. The rates of PE and DVT were not significantly different between groups. The LOS was significantly shorter in the SOT group than that in the matched control group (10.0 vs 14.2 days, P ¼ .01).

Discussion
Over the past several decades, outcomes after SOT have continued to improve, and patients are living longer following transplantation [3]. It has been estimated that between 1987 and 2012, SOT resulted in 2.3 million life-years saved [17]. However, complications from immunosuppression, graft failure, and infection, among others, have been well documented in the transplant literature [18]. Arthroplasty surgery, and in particular revision arthroplasty, is also associated with risk, and the perioperative complications correlate in part with patient preoperative morbidity [19,20]. As patients with a history of both SOT and THA live longer, the need for rTHA will increase, highlighting the importance of anticipating risk to optimize outcomes. Furthermore, the indication for revision influences postoperative outcomes, with patients undergoing revision for infection generally faring worse. [12] For this reason, it is important to stratify rTHA outcomes by indication, specifically for aseptic causes vs infection.
Herein we present a propensity score matched cohort study using a large nationwide registry comparing outcomes of rTHA between SOT patients and matched controls for both aseptic and infected revisions. Rates of readmission and revision rTHA were significantly higher in PJI SOT and non-SOT patients than those in aseptic SOT and non-SOT patients, respectively. For patients undergoing aseptic rTHA, we found that patients with a history of SOT had an increased rate of overall readmission but a lower rate of readmission related to rTHA, including infection. Surprisingly, SOT patients had a lower rate of readmission for PJI than non-SOT patients. SOT patients also had lower rates of any complication or blood transfusion after rTHA compared to non-SOT patients. The lower rate of 90-day readmission related to rTHA and borderline lower rate of re-revision in the SOT cohort may reflect surgeon hesitancy to proceed with revision because of perceived increased risk in this population. The similar rate of hospital complications may reflect adequate matching between the cohorts based on demographics and comorbidities such as tobacco use, BMI, and diabetes mellitus status, which are known to be independent risk factors for post-operative medical complications, including SSIs. Labaran et al. reported no difference in 90-day major medical complications in patients with history of renal transplant undergoing revision arthroplasty compared to matched controls. [10] In a single institution review, Brown et al. also found no difference in peri-operative complications comparing SOT patients and matched controls who underwent primary total joint arthroplasty. [21] Prior studies on rTHA in SOT patients are limited. Labaran et al. investigated outcomes after revision hip and knee arthroplasty in renal transplant patients using a large Medicare database from 2005 to 2015 [10]. In a matched analysis they found that renal transplant patients (total 661 patients) had an increased 90-day readmission (27.8% vs 23.2%), 90-day septicemia and 1-year mortality (6.8% vs 2.3%) rates than matched non-transplant patients; they found no difference in LOS or 1-year infection rates between group. They did not analyze re-operations or re-admissions for specific rTHA complications. In comparison, we similarly found a higher rate of 90-day readmission in SOT patients undergoing aseptic rTHA but a lower rate of readmission for reasons related to rTHA, with no differences in readmission or revision surgery for PJI rTHA patients. We found no difference in 90-day mortality for SOT patients undergoing rTHA for PJI specifically (4.7% vs 6.0%) or for other causes (0.6% vs 1.3%). In a single institution case series of 9 rTKA patients, Ledford et al. described outcomes in 30 SOT patients undergoing rTHA, of which 3 underwent re-revision for PJI and 3 for instability (total of 6, or 20%) at a mean of 2.1 years [9]. This is roughly comparable to the 90-day rate of re-operation following rTHA for PJI (12.3%) in SOT patients in this study. We found the rate of re-operation following rTKA for other reasons to be substantially lower (3.2%). The LOS in their study (5.0 days) is similar to that in this study for non-PJI patients (5.6 days).
The roughly similar rates of complications between SOT patients and matched controls in this study may be surprising to some, especially with regard to subsequent infection and mortality. PJI is the most feared complication following arthroplasty in SOT patients given the hypothesized higher risk posed by the use of chronic immunosuppressive medications in these patients. However, across surgical specialties, in comparative studies of elective surgical procedures adjusting for underlying medical comorbidity, SOT patients have similar rates of post-operative infections and wound complications [22,23]. Prior studies on complications following TJA in SOT patients have also not conclusively shown an increased risk [5,[24][25][26][27]. The reasons for this may be threefold. First, immunosuppression is not unique to SOTdmany different types of patients are on immunosuppressive medications or immunosuppressed as a consequence of chronic disease [28]. Second, SOT patients often have higher levels of medical literacy and access to care than average [29][30][31][32]. Finally, SOT patients are more likely to be treated at specialized academic centers, and consequently, as we find in this study, are more likely to receive their arthroplasty care at these same institutions. These studies suggest that although SOT patients' immunosuppressive regimens may pose additional risk, some of that may be offset by increased access to specialized care and ease of navigating the healthcare system. Indeed, we find lower rates or readmission for infection in SOT patients undergoing aseptic rTHA, and no difference in readmission or revision surgery for SOT patients undergoing rTHA for PJI.
To our knowledge, this is the first report on rTHA in SOT patients which stratifies by all-cause aseptic indications and infection. Several authors have previously highlighted an increased mortality in septic vs aseptic revisions [21,[33][34][35]. For both SOT and nontransplant patients, we found worse outcomes in patients undergoing revision for infection, including longer LOS, as well as higher rates of 90-day readmission and 90-day re-revision. The primary advantages of this study are the large sample size and stratified outcome analysis based on indication for revision. To date, only one other report has used a national administrative database to analyze outcomes of revision arthroplasty in transplant patients [10], but the number of patients analyzed was smaller and aseptic and infected revisions were considered jointly. We believe this may obscure important differences when considering the postoperative risks faced by this unique group of patients. There are several limitations to this study. First, we recognize the inherent weaknesses in a large database study including potential for errors in coding and data entry. It is possible that some readmission events were missing from the NRD, biasing complication estimates downward in this study compared to single institution studies. Furthermore, the NRD allows for the analysis of short-term outcomes and therefore likely underestimates the true incidence of long-term complications after rTHA. Important clinical outcomes such as functional status, patient-reported outcome measures (PROMs), and pain scores are not recorded in the NRD. Despite controlling for demographic variables and comorbidities in our multivariate PSM analysis, there are some important confounding factors for which we were not able to control. Important surgical factors, including surgical complexity, were not available for analysis. Finally, information regarding surgical details such as implants used, procedure duration, intraoperative complications, and blood loss, was unavailable in the NRD.

Conclusion
In the present study, we find that SOT alone is not consistently associated with increased risks after rTHA despite the increased medical comorbidity associated with transplantation. We found a higher overall readmission rate but a lower readmission rate related to rTHA for SOT patients undergoing aseptic rTHA. Rates of 90-day mortality and revision rTHA for both aseptic and infected revisions were similar between SOT patients and matched controls. Index hospital complications were lower overall in SOT patients than those in matched controls for both aseptic and infected rTHA. This suggests that transplant patients undergoing rTHA have no greater risk for complications and/or mortality at short-term follow-up (90 days) when compared to matched nontransplant patients. Access to specialist academic medical centers for SOT patients may offset the increased risk associated with immunosuppressive medication.
Conflicts of interest Categorized rTHA diagnostic codes for ICD-9 and ICD-10.

Group
ICD-9 ICD-10 ICD10 Text Other specified complications of internal orthopedic prosthetic devices, implants and grafts T84.81 Embolism due to internal orthopedic prosthetic devices, implants and grafts T84.82 Fibrosis due to internal orthopedic prosthetic devices, implants and grafts T84.83 Hemorrhage due to internal orthopedic prosthetic devices, implants and grafts T84.84 Pain due to internal orthopedic prosthetic devices, implants and grafts T84.85 Stenosis due to internal orthopedic prosthetic devices, implants and grafts T84.86 Thrombosis due to internal orthopedic prosthetic devices, implants and grafts T84.89 Other specified complication of internal orthopedic prosthetic devices, implants and grafts T84. 9 Unspecified complication of internal orthopedic prosthetic device, implant and graft Propensity score matched analysis of index hospitalization complications for non-PJI patients.

SOT patients
Matched controls P-Value