Acute illness may increase bleed risk among oral anticoagulant users

Abstract

In the United States, oral anticoagulant use has increased by 54% since 2012, particularly with the use of newer direct oral anticoagulants (DOACs). Unfortunately, among clinicians in certain settings knowledge gaps persist concerning the safety of these therapies, including the risk of acute infection. A new study published in the BMJ December 21, 2021, suggests that acute respiratory tract infections (RTIs) may increase the rates of major and nonmajor bleeding in patients taking oral anticoagulants. Haroon and colleagues conducted a self-controlled case series that included over 1,200 participants on oral anticoagulant therapy who were exposed to an RTI and had major or non-major bleeding events. The analysis included longitudinal general practice EHR data from the UK’s Clinical Practice Research Datalink with over 3 million validated living patients registered at over 400 practices across the UK. The authors defined a specific “excess bleed risk” time window as a general practice consultation for an RTI for which immediate antibiotics were not prescribed. They then compared bleeding event outcomes within the excess risk window to an observation period in which the patients were unexposed to an RTI, allowing the patients to act as their own controls. This methodology had the advantage of allowing for reliable control of time invariant confounders. The researchers defined the RTI risk window as 90 days from general practice consultation, which was divided into smaller chunks of 0 to 14 days, 15 to 30 days, 30 to 60 days, and 60 to 90 days. The 0- to 14-day window was selected as the primary risk period. Additionally, a “pre-risk” window of 7 days prior to the risk period was established in which bleed events could not be captured as being “unexposed.” This was to compensate for RTI symptom onset before the day of general practice consultation as to not misclassify a bleeding event. To prevent confounding from antibiotic use and subsequent drug–drug interactions with a patient’s oral anticoagulant, a patient remained inside of the 90-day risk window as long as they weren’t prescribed antibiotics. Additionally, in order to control for experienced user bias, the total observation period began on the date of the first new prescription of warfarin or a DOAC. After adjusting for time confounders (e.g., age, calendar year, and season) the authors observed a statistically significant 268% relative increase in the rates of major bleeding—defined as either a GI bleed or an intracranial hemorrhage—in the 0- to 14-day risk window after an RTI consultation compared to the rates within the unexposed period. However, no increase in major bleeding was observed in the other aforementioned risk windows. Additionally, the authors observed a statistically significant 232% increase in the relative rates of clinically relevant non-major bleeding—defined as general practice consultation or hospital admission for haemoptysis, epistaxis, or haematuria—during the 0- to 14-day risk window. Despite 67% of the patients in the analysis being on warfarin, a sensitivity analyses demonstrated that there was little modification in the incidence rates of major and non-major bleeding during the 0- to 14-day risk window when conducting a subgroup analysis based on the specific oral anticoagulant used. A major limitation of this study was the inability for the authors to control or adjust for OTC treatments for RTI. Of particular concern are NSAIDs, which would have the potential to significantly bias the results of major bleeds away from the null given that 87% of the recorded bleed events seen were GI bleeds. The authors stated that they “explored the use of prescribed NSAIDs, but were unable to account for over-the-counter drugs” when discussing the limitations of their study. They went on to say that this confound was significant given that survey data from 1,000 adults in England with a recent RTI indicated that 60% of these patients sought OTC treatments. Other observational studies have shown that acute infection may increase international normalized ratio measure in warfarin users. As it stands, Haroon and colleagues’ study is the only one in the literature that has looked at the effect of acute infection on clinical events within the oral anticoagulant population as a whole. “These findings are important and timely given the increasing rates of oral anticoagulant use and the lack of knowledge and guidance on how to manage oral anticoagulant use during acute infection,” suggested the authors in one of their concluding remarks. “There are potential implications for how patients and clinicians manage oral anticoagulant use during an acute intercurrent illness, but further work is needed before any clinical recommendations are made.”