Sjogren’s Syndrome is associated with higher healthcare utilization after primary hip, but not primary knee arthroplasty: A U.S. cohort study

Objective: To assess whether Sjogren’s Syndrome (SS) is associated with outcomes after total knee or hip arthroplasty (TKA/THA). Methods: We used the 1998-2014 U.S. National Inpatient Sample data. We performed multivariable-adjusted logistic regression analyses to assess the association of SS with healthcare utilization (hospital charges, length of hospital stay, discharge to non-home setting), and in-hospital complications (implant infection, revision, transfusion, mortality), controlling for important covariates and confounders. In sensitivity analyses, we additionally adjusted the main models for hospital location/teaching status, bed size, and region . Results: We examined 4,116,485 primary THAs and 8,127,282 primary TKAs performed from 1998-2014; 12,772 (0.2%) primary TKAs and 6,222 (0.2%) primary THAs were done in people with SS. In multivariable-adjusted models, SS was associated with a statistically signicant higher odds ratio (OR; 95% condence interval (CI)) of discharge to a rehabilitation/inpatient facility post-THA, 1.13 (1.00, 1.28), but not post-TKA, 0.93 (0.86, 1.02). We noted no differences in the length of hospital stay or hospital charges. SS was associated with signicantly higher adjusted odds of in-hospital transfusion post-THA, 1.37 (1.22, 1.55) and post-TKA, 1.21 (1.10, 1.34). No signicant differences by SS diagnosis were seen in hospital stay, hospital charges implant infection, implant revision or mortality rates. Conclusions: People with SS had higher transfusion rate post-TKA/THA, and higher rate of discharge to non-home setting post-THA. The lack of association of SS with post-arthroplasty complications should reassure patients, surgeons and policy-makers about the utility of TKA/THA in people with SS undergoing these procedures. Sensitivity examined the multivariable-adjusted logistic regression analyses in two SS subgroups: (1) Primary SS: no concomitant disease diagnoses; (2) Secondary SS: presence of one or more concomitant rheumatic disease diagnoses (systemic erythematosus, 710.0; systemic sclerosis, 710.1; sicca syndrome, 710.2; 710.3; polymyositis, 710.4; mixed connective tissue disease, 710.9; antiphospholipid syndrome, 289.81; rheumatoid arthritis,


Introduction
Sjogren's Syndrome (SS) is a systemic autoimmune disease, primarily of middle-aged women, characterized by dry eyes, dry mouth and systemic symptoms (1). Its incidence and prevalence in the general population are 7/100,000 and 60/100,000 people, respectively (1). SS can be primary, or secondary to diseases such as rheumatoid arthritis (RA), lupus, scleroderma, myositis etc. SS is associated with systemic in ammation with an over-expression of pro-in ammatory cytokines (2,3).
With the aging of the population, an increasing number of people with SS and the general population are undergoing joint arthroplasty. Total knee arthroplasty (TKA) or hip arthroplasty (THA) are the two most common arthroplasty surgeries performed for end-stage arthritis (4). Complications and healthcare utilization after arthroplasty are increased in other systemic in ammatory rheumatic disease such as RA, lupus and spondyloarthritis (5). Whether SS, which is also a systemic autoimmune disease, has a similar impact on post-arthroplasty outcomes is unknown. Therefore, we assessed whether SS was associated with higher complication, mortality and healthcare utilization rates after primary TKA or THA.

Data source, Study Cohort and Study Outcomes
We used the 1998-2014 U.S. National Inpatient Sample (NIS) data, a 20% strati ed sample of discharges from U.S. community hospitals (6). NIS is the largest publicly available, de-identi ed all-payer inpatient health care database in the U.S. The University of Alabama at Birmingham's Institutional Review Board approved this study and waived the need for informed consent (X120207004).
We identi ed a cohort of all hospitalizations with primary TKR or THA as the primary procedure by using validated International Classi cation of Disease, ninth revision, common modi cation (ICD-9-CM) codes for primary TKA (81.54) or primary THA (81.51) (7). Among these, we identi ed those with and without Sjogren's Syndrome (ICD-9-CM, 710.2), a valid approach with sensitivity of 95% and speci city of 96% (8), in non-primary position.
Our study outcomes of interest post-primary TKA/THA were: (1) three health care utilization: the discharge disposition to home vs. a rehabilitation/inpatient facility, length of hospital stay above the median (>3 days), and the hospital charges above the median; (2) three in-hospital complications identi ed by respective ICD-9-CM codes for transfusion, implant infection and implant revision; and (3) inhospital mortality.

Statistical Analysis
We performed separate multivariable-adjusted logistic regression analyses for each outcome controlling for clinically important variables, which are also potential and/or known confounders of TKA/THA outcomes (9-13), including age, race, sex, income, underlying diagnosis (listed in the primary diagnosis position), Deyo-Charlson comorbidity index, and insurance payer. Deyo-Charlson comorbidity index (14), is a validated measure of medical comorbidity that includes 17 comorbidities with score ranging 0-25, higher score indicating more comorbidity load. Sensitivity analyses additionally adjusted for hospital bed size, location/teaching status, and region (15,16
SS was associated with a statistically signi cant higher odds ratio (OR; 95% con dence interval (CI)) of discharge to a rehabilitation/inpatient facility post-THA, 1.13 (1.00, 1.28), but not post-TKA, 0.93 (0.86, 1.02) after multivariable-adjustment (Table 2). We noted no differences in the length of hospital stay or hospital charges. We found signi cantly higher adjusted odds of in-hospital transfusion post-THA, 1.37 (1.22, 1.55) and post-TKA, 1.21 (1.10, 1.34) associated with SS ( Table 2). No differences by SS diagnosis were seen in implant infection, implant revision or mortality rates. Sensitivity analyses adjusting for additional hospital characteristics con rmed the ndings from the main analyses with minimal attenuation of ORs (Table 2). Sensitivity analyses were performed for primary vs. secondary SS (Appendix 1). Main study ndings were reproduced transfusion for both primary and secondary SS. Association of SS with discharge to a rehabilitation/inpatient facility post-THA were no longer signi cant, when subgroups of primary and secondary SS were examined. Additionally, primary SS was associated with higher odds of revision post-TKA; and secondary SS with higher odds of length of hospital stay >3 days and lower odds of death post-THA (Appendix 1).

Discussion
In a nationally representative U.S. sample, we found that SS, a multi-system autoimmune disease, was associated with a signi cantly higher odds of discharge to a rehabilitation/inpatient facility post-THA with 1.13-fold higher odds, but not post-TKA. Fatigue (17) and reduced functional ability (18), frequent in SS, may interfere with optimal in-hospital rehabilitation and recovery. This can potentially increase the risk of discharge to a rehabilitation or an inpatient facility. The reason for increased risk of discharge to a rehabilitation facility only after THA, but not TKA, may warrant further study. The strength of association was not very high, therefore the absolute impact of this signi cant association in people with SS may be small.
SS was associated with a higher risk of transfusion after THA and TKA, at 1.2-1.4 fold higher. This association was reproduced in subgroups of people with primary or secondary SS. Systemic in ammation (2,3) in SS with associated anemia (19) and cytopenia (17) can lead to a higher transfusion risk post-THA/TKA. We found that SS was not associated with complications including implant infection, revision or mortality after primary THA or TKA. Our nding contrasts with previously noted higher post-TKA/THA complication rates in people with other systemic in ammatory conditions, such as RA, SpA or lupus (5). Limited joint or organ involvement in SS compared to RA, SpA or lupus (20) and/or the assessment of in-hospital complications in our study (vs. all post-operative complications) may explain these differences. An absence of an association of SS with post-arthroplasty infection, revision or mortality should reassure patients with SS and surgeons that these risks are not increased post-arthroplasty. The implications of possibly increased infection in primary SS post-TKA and decreased mortality in secondary SS post-THA are unclear; we believe that these ndings need further study and replication.
Study limitations include residual confounding bias (direction of bias unclear), the lack of data on SS severity and laboratory measures, misclassi cation bias which likely biased ndings towards the null, and the lack of longitudinal data. Bilateral simultaneous THA/TKA can not be distinguished from unilateral procedures in the NIS; however, these constitute <1% THA and <3% TKA, and therefore the bias is likely small. We were unable to separately examine the associations for primary versus secondary Sjogren's syndrome due to the lack of a separate ICD-9 code for these. However, since the analyses are adjusted for underlying reason for THA/TKA which includes all rheumatic conditions associated with THA/TKA, the associations are for the presence versus absence of Sjogren's Syndrome. These ndings should alert the clinician to a higher transfusion risk in SS patients undergoing THA or TKA, and reassure the policy-makers that early THA/TKA outcomes are only minimally impacted by SS.

Conclusions
In conclusion, SS was associated with a higher risk of transfusion after THA and TKA, higher odds of discharge to a rehabilitation/inpatient facility post-THA, but not post-TKA, but was not associated with post-arthroplasty infection, revision or mortality. These ndings can inform patients, providers and policymakers regarding the minimal impact of SS on post-primary TKA/THA outcomes.