(148) Association between incident interstitial lung disease with direct oral anticoagulants compared to warfarin users: Multi-database cohort study

Background: Interstitial lung disease (ILD) describes a heterogenous group of respiratory disorders affecting the interstitium of the lungs. Cases of interstitial lung disease (ILD) have been recently described in people using direct oral anticoagulants (DOACs). In addition, a recently published observational study suggests an increased risk of ILD in patients treated with factor Xa inhibitors (edoxaban, apixaban, rivaroxaban) as compared to warfarin users among subjects with non-valvular atrial fibrillation.

Objectives: To determine whether there is an increased risk of ILD in users of DOACs compared to warfarin users.

Methods:

Design: New user propensity score matched comparative cohort study.

Setting: Patients with a diagnostic code for atrial fibrillation (AF) or venous thromboembolism (VT) identified in primary care data sources from IMRD UK and IQVIA Germany between January 2010 and December 2019 were included. Patients with a prior history of ILD were excluded from the analysis.

Exposures: New users of DOACs (edoxaban, apixaban, dabigatran and rivaroxaban) and new users of warfarin defined as no prior use within 365 days of index date.

Main outcome measures: Incident diagnosis of ILD.

Statistical analysis: New DOAC users (target cohort) were propensity score matched to new warfarin users (comparator cohort) with propensity score diagnostics checked to ensure balance between included covariates associated with the outcome or outcome and exposure, and calendar year. Cause-specific Cox regression models were used to calculate the hazard of an incident ILD diagnosis in DOAC users vs warfarin users within each database separately. An intention-to-treat (ITT) and on-treatment (OT) treatment strategies were applied.

Results: For AF cohorts, 3603 DOAC users were matched in IMRD UK and 6140 in IQVIA Germany. For VT cohorts, 1606 DOAC users were matched in IMRD UK and 1606 in IQVIA Germany. ILD incidence rates in DOAC cohorts ranged from 11.1 to 23.0 per 10,000 person years depending upon the indication and treatment strategy. In AF cohorts, hazard ratios (HR) for ILD in DOAC users compared to warfarin users in: IMRD UK were HR 0.88 (95%CI 0.51 to 1.51) in the ITT analysis and HR 1.09 (95%CI 0.56 to 2.19) in the OT analysis; IQVIA Germany were HR 1.01 (95%CI 0.58 to 1.75) for ITT and HR 1.01 (95%CI 0.58 to 1.75) for OT analysis. Corresponding HRs for the VT cohorts in: IMRD UK were HR 1.44 (95%CI 0.60 to 3.62) and HR 1.36 (95%CI 0.30 to 7.80) respectively; IQVIQ Germany were HR 0.56 (95%CI 0.28 to 1.08) and HR 0.82 (95%CI 0.14 to 5.37) respectively.

Conclusions: When examining DOAC users as a group, we observed no strong evidence to support an association between DOAC use and increased development of ILD. However, estimates from individual database analyses may be less precise when analysed by indication. An important influencer in PS matching was calendar time suggesting temporal trends in comparative studies of DOACs with warfarin are important to account for in study designs.