Study design and setting
This is a cross-sectional cohort study using data acquired prospectively from October 2013 through November 2018. Black, White, and Hispanic patients with RA undergoing elective THA and TKA at a single, high-volume tertiary care center for musculoskeletal diseases were recruited prior to surgery and underwent a comprehensive evaluation, including patient- and physician-reported measures and laboratory evaluations performed pre-operatively. Confirmed cases were included in the study after informed consent was obtained.
Patients undergoing elective THA or TKA who were ≥ 18 were screened via electronic medical record to identify patients with RA. Patients were included if they met American College of Rheumatology/European League Against Rheumatism (ACR/EULAR) 2010 or 1987 criteria  or if the diagnosis was confirmed by the principal investigator (SG). We excluded patients with crystalline arthropathy or other rheumatic diseases including systemic lupus erythematosus, ankylosing spondylitis, and psoriatic arthritis, those who were unable to understand or read English, or those deemed unreliable to follow the study protocol.
Exposure and outcomes
The exposure was minority status (self identified as Black or Hispanic) among patients with RA undergoing arthroplasty. The outcomes of interest in our study were pre-operative function [the Multidimensional Health Assessment Questionnaire (MDHAQ)], pain [the Visual Analogue Scale (VAS)], and disease activity [the Disease Activity Score (DAS28)]. The exposure and outcomes were measured pre-operatively.
Baseline data collected included age, sex, education level, and medications. We grouped patients into minority and non-minority cohorts based on self-identification. The minority group included Hispanic, Black or African American, and mixed race patients; the non-minority group were White patients. Poverty at the community level was determined using census tract data, by linking individual patients’ addresses via geocoding to their respective census tracts. We obtained census tract-level socioeconomic variables from the American Community Survey/United States Census using the Arc Geographic Information Systems. We used 20% of the population living below poverty as the point below which health impacts are reported [12, 13]. The census tract variable of interest was “percentage of families and people whose income in the last 12 months is below the poverty level - All families”.
We assessed patients using patient and provider global assessments as well as tender and swollen joint counts. Inflammatory marker measurement including erythrocyte sedimentation rate (ESR) and C-reactive protein (CRP), rheumatoid factor (RF) and anti-citrullinated protein antibody (ACPA) were obtained within two weeks prior to surgery or on the day of surgery, prior to the surgical incision.
Medication use was per standard of care and was recorded. Biologic usage at the time of surgery was captured and analyzed as a binary categorical variable. Patients who discontinued their biologics for the surgery were considered to be taking biologics. Insurance status included Medicaid, Medicare, or commercial. Medicaid is a public assistance program providing insurance to those in need based on income, and Medicare is an age-based federal insurance program provided to people above age 65 as well as those with disabilities. Medicaid insurance status was used as a proxy for income at the individual level, given the inaccuracies of self-reported income assessments . Those presenting with two types of insurance were categorized into one of the existing insurance status categories according to the following priority: Medicaid followed by Medicare and finally commercial insurances. Patients were marked as having one type of insurance in these cases. For example, a patient who had both Medicare and commercial insurance was marked as having Medicare. Education was recorded as a binary categorical variable with those achieving some college or greater differentiated from those who achieved high school or less.
Function was assessed using the validated MDHAQ, which provides patients with specific questions (example: ability to get in and out of bed) in an effort to assess their overall functional status. The score is comprised of 10 activities of daily living which are scored from 0 to 3 with a score of 0 implying no difficulty performing the activity and 3 implying an inability to perform the activity. The score total is 0–30; a higher number indicates a worse functional status . Pain was measured using the visual analogue scale (VAS) pain score, which is a user friendly scoring system where patients indicate their level of pain on a spectrum from 1 to 10, with 10 being unbearable pain . Both MDHAQ and VAS were recorded as continuous variables. The DAS-28  is a composite measure of disease activity that includes swollen and tender joint counts performed by the physician, plus the ESR and the patient’s global assessment of their health (PGA) . A higher DAS-28 score implies a more active disease state. DAS28-ESR was chosen as the primary outcome of interest and as the measure of disease activity; it was binarized to patients with moderate or high disease activity (DAS28 ≥ 3.2) and those with low disease activity or remission of disease (DAS28 < 3.2).
The distribution of continuous baseline patient characteristics was assessed for normality using the Shapiro-Wilk test. Normally distributed values were summarized as mean ± SD and compared using t-tests. If non-normally distributed, values were reported as median [25th percentile, 75th percentile] and compared using the Wilcoxon rank-sum test. Categorical variables were summarized as frequency and percent and compared across minority status using Fisher’s exact tests. Student t-test or the Wilcoxon rank-sum test were used, as appropriate, to compare continuous variables across minority status.
Univariate and multivariable linear regressions were performed to assess the impact of minority status on preoperative pain (VAS score), and function (MDHAQ), while univariate and multivariable logistic regressions were performed to assess odds of moderate/severe RA disease activity (DAS28-ESR). Results of linear models are summarized with slope(β) ± SE, while logistic models are summarized with odds ratios (OR) and 95% confidence intervals (CI). Mean and Standard Deviation (SD) were used for values that were normally distributed and median and Interquartile Range (IQR) were used for values that were not normally distributed. The following variables were forced into all multivariable models as they have been recognized as predictors of total joint replacement outcomes in previous studies and are relevant to the research interest: education, insurance status, age, sex, and biologics .
For missing data, we retained all observations that recorded race and insurance status in the univariate analyses. In the multivariable model, the maximum number of patients that had data for each variable included in the model were retained i.e. minority status, education, insurance type, age, sex, use of biologics and MDHAQ score were all required in order to be included in the multivariable model for odds of moderate/high DAS-28 ESR (Table 3). We excluded patients with missing values for race and insurance status. One White patient opted not to disclose gender but was included in the analysis. All analyses were performed using SAS version 9.4. A two-tailed p < 0.05 was considered statistically significant.
This study is reported per the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) checklist for cohort studies. This study was approved by the hospital institutional review board and all included patients provided informed consent.