Background Abatacept is the first selective T-cell co-stimulation modulator approved as a biologic (b)DMARD for the treatment of RA. Real-world comparative safety studies of this treatment require that characteristics of patients initiating abatacept be similar to those for patients initiating other bDMARDs or non-bDMARDs.
Objectives To identify an appropriate comparison group for patients with RA initiating abatacept among patients treated with bDMARDs and non-bDMARDs enrolled in administrative claims databases.
Methods A retrospective cohort of RA patients treated with abatacept, bDMARDs or non-bDMARDs between 1 July 2006 and 30 June 2012 was formed from the MarketScan Commercial and Supplemental Medicare databases. The study cohort was formed hierarchically, first identifying all patients who had initiated abatacept treatment, irrespective of prior treatment with another study drug. The date of the first abatacept prescription was taken as cohort entry. Patients with less than 6 months of continuous health plan enrolment prior and 1 day post-cohort entry were excluded. Baseline data on patient characteristics, comorbidities and co-medications received during the 6-month period prior to cohort entry were identified. Subsequently, comparison patients initiating treatment with bDMARDs or non-bDMARDs during this period were identified, with the date of the first such prescription taken as cohort entry and 6-month baseline characteristics obtained as above. Patients with less than 6 months of continuous health plan enrolment prior to cohort entry were excluded. History of DMARD treatment during and prior to the 6-month baseline period were obtained to distinguish first-time users (no use in 6-month baseline or prior), new users (no use in 6-month baseline but prior use) and switchers (use of another DMARD in 6-month baseline).
Results There were 10,958, 37,593 and 139,290 patients in the abatacept, bDMARD and non-bDMARD cohorts, respectively. Patients in the abatacept cohort were more likely to be female and were older compared with the bDMARD and non-bDMARD cohorts. A major difference between the cohorts was in the prior use of bDMARDs during the baseline period, with prior switching from a bDMARD occurring in 47% of patients in the abatacept cohort compared with 14.4% in the bDMARD cohort and 11.1% in the non-bDMARD cohort. In addition, 34.5% of patients initiating abatacept were first-time users and 18.5% were new users. In contrast, 64.4% of other bDMARD initiators were first-time users and 21.2% were new users.
Conclusions When conducting or interpreting real-world comparative studies of newer DMARDs such as abatacept, the history of use of other bDMARDs prior to treatment initiation can be a major confounding factor and effect modifier. Teasing out the effects, whether on acute outcomes such as infections or latent outcomes such as cancer, of prior bDMARD use presents methodological challenges to the design and analysis of such studies of comparative effectiveness or safety of a new DMARD such as abatacept in RA.
Disclosure of Interest S. Suissa Consultant for: Bristol-Myers Squibb, Genentech, Roche, N. Baker Shareholder of: Bristol-Myers Squibb, Employee of: Bristol-Myers Squibb, A. Ravindran: None declared, H. Kawabata Shareholder of: Bristol-Myers Squibb, Employee of: Bristol-Myers Squibb, T. Simon Employee of: Bristol-Myers Squibb