This study used data from the IBM® MarketScan® Research Commercial and the Medicare Supplemental Database, which contains patient-level information on clinical utilization, expenditures, and enrollment across inpatient, outpatient, prescription drug, and carve-out services from large US employers, health plans, government, and public organizations. The annual medical databases capture private sector health data from approximately 100 payers, with over 4.6 billion claim records available.
The database is compliant with the Health Insurance Portability and Accountability Act of 1996 and patient identifiers were removed to ensure patient privacy. The database is commercially available, therefore approval from an institutional review board was not required.
We included patients who were ≥ 18 years of age and had either ≥ 2 outpatient claims at least 30 or more days apart or ≥ 1 inpatient visit claim with an RA diagnosis between January 1, 2013 and March 31, 2017 (index period). The first RA claim, determined by International Classification of Diseases Ninth (ICD-9) or Tenth Revision (ICD-10) codes (714.0x, 714.1x, 714.2x, M05.x%, M06.0%, M06.8%) was set as the index date. Patients were also required to have ≥ 12 months of continuous enrollment with medical and prescription benefits and were allowed a maximum 29-day gap.
We excluded patients with one or more claims for the following diagnoses: juvenile idiopathic arthritis (714.3x, M08.%), ankylosing spondylitis (720.0x, M45.%), psoriatic arthritis (696.0x, L40.5%), systemic lupus erythematosus (710.0x, M32.%), and lupus nephritis (583.81, M32.14).
Prescription claims for drugs with potential DDIs (described below) were quantified from January 1, 2013 to March 31, 2018.
Frequency of Variables of Interest
To characterize the patient sample, we evaluated the number and percentage of the following variables: demographic characteristics, such as sex and age; insurance type (commercial or Medicare supplement); US geographic region (Midwest, West, Northeast, South); comorbidities as quantified by the Charlson Comorbidity Index (CCI)(12, 13); and claims for csDMARDs (i.e., hydroxychloroquine, leflunomide, methotrexate, or sulfasalazine), bDMARDS (i.e., abatacept, adalimumab, anakinra, certolizumab pegol, etanercept, golimumab, infliximab, rituximab, sarilumab, or tocilizumab) or tsDMARDs (i.e., tofacitinib; baricitinib and upadacitinib were not available during the study period).
We assessed the number and percentage of patients with claims at any time during the study period for strong OAT3 inhibitors, strong CYP3A4 inhibitors, and moderate or strong CYP3A4 inhibitors with strong CYP2C19 inhibitors in combination. Definitions of moderate and strong followed the classifications on DrugBank, an online, publicly available database, as described by Wishart et al. . The drugs evaluated for potential DDI are listed in Table 1. Two drugs were identified as strong OAT3 inhibitors, 47 as strong CYP3A4 inhibitors, 37 as moderate CYP3A4 inhibitors, and 16 as strong CYPC19 inhibitors.
Descriptive statistics were conducted. Sample selection and the creation of analytic variables were performed using the Instant Health Data (IHD) platform (BHE, Boston, MA). Statistical analyses were conducted with R, version 3.2.1 (R Foundation for Statistical Computing, Vienna, Austria).