Table 4.

Effect of Warfarin and Direct-Acting Oral Anticoagulants on the Secondary Outcomes of Ischemic Stroke, Major Adverse Cardiovascular Events, and Bleeding Compared to No Anticoagulation

Model Specification	Ischemic Stroke				MACE				Bleeding
	Warfarin versus no AC		DOAC versus no AC		Warfarin versus no AC		DOAC versus no AC		Warfarin versus no AC		DOAC versus no AC
	n	HR (95% CI)	n	HR (95% CI)	n	HR (95% CI)	n	HR (95% CI)	n	HR (95% CI)	n	HR (95% CI)
ITT PS-matched cohorts:	1,694	1.40 (0.82-2.39)	704	0.47 (0.16-1.36)	1,694	0.86 (0.71-1.04)	704	0.45 (0.18-1.09)	1,694	1.50┼ (1.10-2.06)		0.77 (0.40-1.48)
Marginal structural models:*	2,694	0.29┼ (0.09-0.90)	2,694	0.23┼ (0.07-0.79)	2,694	0.25┼ (0.07-0.87)	2,694	0.38 (0.09-1.65)	2,694	1.29 (0.74-2.26)		0.37 (0.13-1.07)

Abbreviations: AC-anticoagulation, CI-confidence interval, DOAC-direct-acting anticoagulant, HR-hazard ratio, ITT-intention to treat, MACE-major adverse cardiovascular event, PS-propensity matched

Baseline covariates used for adjustment, weighting and/or matching: age, race, etiology of liver disease, diabetes, hypertension, congestive heart failure, obesity, serum total bilirubin, serum albumin, platelet count, eGFR, Child Turcotte Pugh score, alcohol misuse, history of substance abuse, tobacco use, cirrhosis comorbidity index, geographic region, hospital complexity Bleeding defined as: ≥ 2 gram per deciliter hemoglobin drop on 2 two consecutive laboratory tests, any ICD-9-CM/ICD-10-CM code for intracranial bleeding including hemorrhagic stroke, admission requiring any packed red blood cell transfusion, or a discharge diagnosis codes of gastrointestinal bleeding, hemarthrosis, other bleeding in the first position MACE defined as: ischemic stroke, acute myocardial infarction or revascularization

^*Adjusted for/weighted with time-dependent total serum bilirubin, serum albumin, platelet count, Child-Turcotte Pugh (CTP) scores, eGFR, platelet count, and bleeding complications in addition to baseline covariates

^┼p<.05,

^╪p<.001