Initiation of warfarin is associated with decreased mortality in patients with infective endocarditis: A population-based cohort study

IMPORTANCE: The use of warfarin to prevent thromboembolism in patients with infective endocarditis (IE) remains controversial due to potentially increased bleeding risks.

DESIGN: Population-based retrospective cohort study.

PARTICIPANTS: Patients aged 18 or older and diagnosed with IE in Hong Kong between January 1st, 1997 and August 31st, 2020 were included. Patients with use of any anticoagulant 30 days before IE diagnosis were excluded. Patients initiated on warfarin within 14 days of IE diagnosis and patients without warfarin use were matched for baseline characteristics using 1:1 propensity score matching.

EXPOSURE: Warfarin use within 14 days of IE diagnosis.

MAIN OUTCOMES AND MEASURES: Patients were followed up to 90 days for the outcomes of ischemic stroke, all-cause mortality, intracranial hemorrhage, and gastrointestinal bleeding. Cox regression was used to determine hazard ratios (HRs) [95 % confidence intervals (CIs)] between treatment groups. Fine-Gray competing risk regression with all-cause mortality as the competing event was performed as a sensitivity analysis. In addition to 90-day analyses, landmark analyses were performed at 30 days of follow-up.

RESULTS: The matched cohort consisted of 675 warfarin users (57.0 % male, age 59 ± 16 years) and 675 warfarin non-users (53.5 % male, age 61 ± 19 years). Warfarin users had a 50 % decreased 90-day risk in all-cause mortality (HR:0.50 [0.39-0.65]), without significantly different 90-day risks of ischemic stroke (HR:1.04 [0.70-1.53]), intracranial hemorrhage (HR:1.25 [0.77-2.04]), and gastrointestinal bleeding (HR:1.04 [0.60-1.78]). Thirty-day landmark analysis showed similar results. Competing risk regression showed significantly higher 30-day cumulative incidence of intracranial hemorrhage in warfarin users (sub-HR:3.34 [1.34-8.31]), but not at 90-day (sub-HR:1.63 [0.95-2.81]). Results from Fine-Gray regression were otherwise congruent with those from Cox regression.

CONCLUSIONS AND RELEVANCE: Warfarin initiated within 14 days of IE diagnosis was associated with significantly decreased risks of mortality but higher risks of intracranial hemorrhage, with similar risks of ischemic stroke and gastrointestinal bleeding, compared with non-use of warfarin with 14 days of IE diagnosis.

KEY POINTS: Question: Is warfarin, initiated within 14 days of a diagnosis of infective endocarditis (IE), efficacious and safe?

FINDINGS: In this propensity score-matched, population-based, prospective cohort study from Hong Kong, warfarin use within 14 days of IE diagnosis was associated with a 50 % decrease in the risk of all-cause mortality, albeit with higher risk of intracranial hemorrhage, and without significant differences in the risk of ischaemic stroke and gastrointestinal bleeding. Meaning: In patients with IE, warfarin use within 14 days of diagnosis may have mortality benefits, despite increased risks of intracranial hemorrhage.