The VARA Registry
This retrospective, observational study was based on data obtained from US veterans with RA who were enrolled in the VARA registry. The VARA registry is an ongoing, longitudinal, multicenter registry that included patients from 12 VA medical centers (Birmingham, Alabama, USA; Brooklyn, New York, USA; Dallas, Texas, USA; Denver, Colorado, USA; Jackson, Mississippi, USA; Iowa City, Iowa, USA; Little Rock, Arkansas, USA; Omaha, Nebraska, USA; Portland, Oregon, USA; Philadelphia, Pennsylvania, USA; Salt Lake City, Utah, USA; and Washington, DC, USA). The VARA registry has been described previously [18, 19].
This study was approved by the University of Utah Institutional Review Board, the VA Research Service, and the Scientific and Ethical Advisory Board of the VARA registry for analysis of VARA and VA administrative data and was in compliance with the Helsinki Declaration for Ethical Principles for Medical Research Involving Human Subjects of 1964, as revised in 2013. All patients provided written consent upon enrollment in the VARA registry.
All patients met the rheumatologist-confirmed diagnosis of RA according to the 1987 American College of Rheumatology criteria . Patients were eligible to be included in the analysis if they had enrolled in the VARA registry before September 30, 2011, and their first TNFi therapy was initiated between March 17, 2003 (the date after which all three TNFi agents were available within the VA), and September 30, 2010, which allowed the potential for at least 1 year of observation, ending on September 30, 2011. To increase the likelihood of evaluating only patients initiating TNFi therapy within the VA, patients were required to have at least 6 months of treatment in the VA before their first TNFi prescription. Patients were excluded from the study if they had any TNFi exposure within the VA before March 17, 2003. Information on therapy prior to clinical care at the VA was not available.
Three administrative VA databases  were used in the analysis, including the Corporate Data Warehouse (CDW) , Decision Support System (DSS) National Pharmacy Extract , and Pharmacy Benefits Management (PBM) database . Descriptions of these databases have been previously presented . The Veterans Health Administration structured pharmacy data do not completely capture the dispensing of outpatient drug infusion data. Furthermore, the barcode medication administration data only capture medications administered to inpatients. Therefore, we conducted a chart review of VA electronic medical records over the study period to determine administration of infliximab through nurse infusion notes. Because some inconsistencies in the outpatient pharmacy dispensing data were observed for adalimumab and etanercept, we developed an algorithm that integrated information from the three databases (CDW, DSS, and PBM) and defined each treatment episode for each patient. The algorithm reconciled differences between databases by using information from all data sources as well as comparing treatment patterns before and after the discrepancies to estimate correct dosing . For patients with evidence of any infliximab infusion from CDW (outpatient pharmacy, inpatient unit dose, or IV package), DSS, or PBM, a complete individual patient medical record review was completed by trained chart abstractors over the study period to characterize infliximab administration events during the study period.
Each dispensing or administration event of a TNFi agent was used along with information on quantity dispensed and days supply to construct individual treatment episodes. The expected supply (in days) was determined based on dosing instructions and the number of syringes or vials that had been dispensed for injectable agents. For infliximab, each administration event was determined by chart abstraction as noted above. Our focus was limited to the first two drug courses with TNFi agents during the study period. A drug course was defined as a period of continuous TNFi treatment consisting of one or more treatment episodes without a gap of 90 or more days between the expected end of that episode based on days supply and the start of the subsequent treatment episode. Duration of treatment course was defined as the time from the date of first treatment until the date of the expected end of the last dispensing episode for the injectable TNFi agents or 8 weeks after the last infusion (based on the longest recommended dosing interval in the prescribing information) for infliximab, for each course of therapy.
For those patients who were naïve to prior biologic therapy at the VA, duration of treatment was calculated for the first and second course of therapy. Patients with single therapy had only a first course of treatment, while patients with interrupted and switched therapy had both first and second courses of treatment (Fig. 1). Patients who initiated a non-biologic TNFi agent were classified as having a TNFi discontinuation. Subsequent courses were not evaluated, thus each patient was classified as having either single (continuous), interrupted, or switched therapy. Patients began their second course of treatment after a gap in treatment of 90 or more days (interrupted therapy) or initiation of a second TNFi agent (switched therapy).
Disease activity was assessed by the Disease Activity Score based on 28 joints (DAS28) , using erythrocyte sedimentation rate as the laboratory measure of inflammation. DAS28 before starting TNFi therapy was defined as the mean of all DAS28 values from VARA enrollment until 30 days after TNFi therapy start date, with most of these values measured within the first 5 days of initiating therapy. Post-treatment DAS28 was the mean of all DAS28 values beginning 90 days after TNFi therapy start date to allow time for the medication to take effect. Changes in DAS28 represent the difference between mean DAS28 before TNFi therapy and mean DAS28 after TNFi therapy for patients with values at both time points.
Drug costs and associated administration costs were calculated using VA PBM prices, including a Blanket Purchase Agreement price for adalimumab and a Big 4 price—which is only available to VA, Department of Defense, Public Health Service (Indian Health Service), and US Coast Guard customers —for etanercept and infliximab . To approximate the most current drug costs, the January 1, 2013, to February 28, 2014, Federal Supply Schedule pricing was used to calculate adalimumab costs ($506.78 for 40 mg syringes); September 30, 2012, to September 29, 2017, pricing was used to calculate etanercept costs ($139.54 for 25 mg syringes and $279.08 for 50 mg syringes); and January 1, 2013, to February 29, 2016, pricing was used to calculate infliximab costs ($456.81 per 100 mg vial). The administrative cost for each treatment episode of etanercept and adalimumab was $25 (VA dispensing costs) and for each IV episode of infliximab was $169.09 (VA infusion costs). Total drug costs were the sum of the direct drug costs and the drug administration costs and are reported as the annualized cost of treatment by dividing the cost of the treatment course by the duration of the treatment course. Costs were calculated for the first course of TNFi treatment. Subsequent costs were then evaluated according to the initial drug assignment and calculated on an annualized basis. Second-course costs were based on the TNFi assignment for the agent that was selected for the second course of treatment.
Continuous data are presented as means and 95% confidence intervals (CIs), and dichotomous data are presented as proportions and 95% CIs. The focus on CIs instead of P values provides evidence for the stability of estimates along with statistical significance testing—when the CIs do not overlap between two groups then the P values are <0.05 and considered significantly different [27, 28]. The data analysis for this paper was generated using SAS software version 9.2 (SAS Institute Inc., Cary, NC, USA).