Liver Cirrhosis in Patients With Atrial Fibrillation: Would Oral Anticoagulation Have a Net Clinical Benefit for Stroke Prevention?

BACKGROUND: Patients with liver cirrhosis have been excluded from randomized clinical trials of oral anticoagulation therapy for stroke prevention in atrial fibrillation. We hypothesized that patients with liver cirrhosis would have a positive net clinical benefit for oral anticoagulation when used for stroke prevention in atrial fibrillation. METHODS AND
RESULTS: This study used the National Health Insurance Research Database in Taiwan. Among 289 559 atrial fibrillation patients aged ≥20 years, there were 10 336 with liver cirrhosis, and 9056 of them having a CHA2DS2-VASc score ≥2 were divided into 3 groups, that is, no treatment, antiplatelet therapy, and warfarin. Patients with liver cirrhosis had a higher risk of ischemic stroke (hazard ratio=1.10, P=0.046) and intracranial hemorrhage (hazard ratio=1.20, P=0.043) compared with those without. Among patients with liver cirrhosis, patients taking antiplatelet therapy had a similar risk of ischemic stroke (hazard ratio=1.02, 95%CI=0.88-1.18) compared to those without antithrombotic therapies, but the risk was significantly lowered among warfarin users (hazard ratio=0.76, 95%CI=0.58-0.99). For intracranial hemorrhage, there were no significant differences between those untreated and those taking antiplatelet therapy or warfarin. The use of warfarin was associated with a positive net clinical benefit compared with being untreated or receiving only antiplatelet therapy.
CONCLUSIONS: For atrial fibrillation patients with liver cirrhosis in the current analysis of an observational study, warfarin use was associated with a lower risk of ischemic stroke and a positive net clinical benefit compared with nontreatment, and thus, thromboprophylaxis should be considered for such patients.

Clinical Perspective

What Is New?

Patients with atrial fibrillation associated with liver cirrhosis have been excluded from randomized trials of oral anticoagulation therapy for stroke prevention.

In this nationwide registry study, atrial fibrillation patients with liver cirrhosis had a higher risk of ischemic stroke and intracranial hemorrhage compared with those without.

Among atrial fibrillation patients with liver cirrhosis, the risk of ischemic stroke was lowered and the risk of intracranial hemorrhage was similar among warfarin users compared with those without antithrombotic therapies.

The use of warfarin was associated with a positive net clinical benefit compared with being untreated or receiving only antiplatelet therapy.

What Are the Clinical Implications?

For atrial fibrillation patients with liver cirrhosis, warfarin use was associated with a lower risk of ischemic stroke and positive net clinical benefit compared with nontreatment, and thus, thromboprophylaxis should be considered for such patients.

Introduction

Liver cirrhosis may be associated with a coagulopathy, and such patients have been excluded from randomized clinical trials of oral anticoagulation (OAC) therapy for stroke prevention in atrial fibrillation (AF). This may be of concern, as alcohol is a common predisposition to liver cirrhosis as well as AF. Nonetheless, there are limited data on the epidemiology and stroke or bleeding risks associated if liver cirrhosis is concomitantly present with AF. In Asian countries hepatitis carrier status and hepatitis‐related liver cirrhosis are commonly encountered,1 and a major clinical dilemma is how to decide on thromboprophylaxis in such patients.

Stroke and bleeding risks in AF are not homogeneous and are dependent on the presence of established risk factors incorporated within established risk scores, such as the CHA2DS2‐VASc and HAS‐BLED scores.2, 3 These scores have not been validated in patients with liver cirrhosis, and the impact of OAC use in such patients is uncertain. Indeed, many such patients are perceived as being at too high risk for OAC and are often prescribed aspirin instead. This is despite the latter having minimal efficacy for stroke prevention and having a negative net clinical benefit (NCB) once ischemic stroke reduction is balanced against serious bleeds.4

We hypothesized that patients with chronic liver cirrhosis would have a positive NCB for OAC used for stroke prevention in AF. We tested this hypothesis in a nationwide cohort based on the Taiwan national insurance database.

Methods

This study used the National Health Insurance Research Database (NHIRD) released by the Taiwan National Health Research Institutes. The National Health Insurance system is a universal health insurance program that offers comprehensive medical care coverage to all Taiwanese residents. NHIRD consists of detailed healthcare data from >23 million enrollees, representing >99% of Taiwan's population. In this cohort data set the patients' original identification numbers have been encrypted to protect their privacy, but the encrypting procedure was consistent, so that a linkage of the claims belonging to the same patient was feasible within the National Health Insurance database and can be followed continuously. The large sample size of this database provided a good opportunity to study the risk of increased intracranial hemorrhage (ICH) and benefits of stroke risk reduction with warfarin use in AF patients with liver cirrhosis.

Study Population

The study protocol of the present study was similar to those of our previous studies.5, 6, 7, 8, 9, 10, 11 From January 1, 2000 to December 31, 2011, a total of 289 559 AF patients aged ≥20 years (10 336 with liver cirrhosis) were identified from the NHIRD. AF was diagnosed using the International Classification of Diseases, 9th Revision, Clinical Modification codes (427.31) registered by the physicians responsible for the treatment of patients. To ensure the accuracy of diagnosis, we defined patients with AF only when it was a hospital discharge diagnosis or confirmed at least 2 times in the outpatient department. The diagnostic accuracy of AF using this definition in NHIRD has been validated previously.12, 13 The risk of ischemic stroke and ICH for AF patients with (n=10 336) or without liver cirrhosis (n=279 223) stratified based on the strategies for stroke prevention was compared (Figure 1).

Figure 1

A flowchart of the enrollment of the study cohort. From January 1, 2000 to December 31, 2011, a total of 289 559 AF patients aged ≥20 years (10 336 with liver cirrhosis) were identified from the NHIRD. The risks of ischemic stroke and ICH were compared for patients with and without liver cirrhosis. Among 10 336 patients with liver cirrhosis, 9056 had a CHA 2 DS 2‐VASc score ≥2. These were divided into 3 groups, that is, no treatment (61.1%), antiplatelet therapy (30.6%), and warfarin (8.3%). The risk of ICH and benefit of stroke risk reduction were analyzed between patients without use of any anti‐thrombotic agent and those with anti‐platelet agents or warfarin use. AF indicates atrial fibrillation; ICH, intracranial hemorrhage; NHIRD, National Health Insurance Research Database.

Among 10 336 patients with liver cirrhosis, 9056 of them having a CHA2DS2‐VASc score ≥2 were divided into 3 groups, that is, no treatment (n=5532, 61.1%), antiplatelet therapy (n=2770, 30.6%), and warfarin (n=754, 8.3%). The risk of ICH and benefit of stroke risk reduction were analyzed between patients without use of any antithrombotic agent and those with antiplatelet agents or warfarin use. The flowchart of study design and patient enrollment is shown in Figure 1.

Calculation of Score and Definition of Clinical End Point

The CHA2DS2‐VASc score was calculated for each patient by assigning 1 point each for age between 65 and 74 years, history of hypertension, diabetes mellitus, recent cardiac failure, vascular disease (myocardial infarction or peripheral artery disease), and female sex, and 2 points each for a history of a stroke, TIA, or age ≥75 years.2

The clinical end point was the occurrence of ischemic stroke with concomitant imaging studies of the brain, including computed tomography or magnetic resonance imaging. The accuracy of diagnosis of ischemic stroke in Taiwan's NHIRD has been reported to be around 94%.14 Another validation study also demonstrated that the diagnostic accuracy of ischemic stroke in NHIRD was high, with positive predictive value and sensitivity of 88.4% and 97.3%, respectively.15 The safety end point was the occurrence of ICH (International Classification of Diseases‐9 code 430.x‐432.x), which necessitated admissions to intensive care units.

Analysis of Net Clinical Benefit

The NCB for the use of warfarin or antiplatelet therapy compared with no treatment was calculated using the formula:

The weighting factor reflects the relative impact, in terms of death and disability, of an ICH while receiving warfarin or antiplatelet agents versus experiencing an ischemic stroke while on no treatment.4, 16, 17 The NCB with 95%CI were calculated from rate differences of ischemic stroke and ICH of the present study based on the weights previously produced and reported in the studies by Singer et al,16 Connolly et al,17 and Lip et al.4 A positive NCB favors treatment (ie, warfarin) over no treatment.

Propensity Match Analysis

We performed propensity score–matched analyses for 2 kinds of comparisons among patients with liver cirrhosis: antiplatelet agents versus no antithrombotic therapy and warfarin versus no antithrombotic therapy. We calculated propensity scores for the likelihoods of receiving antiplatelet agents and warfarin compared to no antithrombotic therapy by multivariate logistic regression analyses, conditional on all baseline covariates listed in Table 1. After that, we matched patients in the antiplatelet‐agent group to those in the no‐antithrombotic‐therapy group with a 1:1 ratio on the basis of the closest propensity score for the use of antiplatelet agents within a threshold of ±0.01. If more than 1 patient in the no‐antithrombotic‐therapy group could be matched to the corresponding subject in the antiplatelet‐agent group, 1 patient from the no antithrombotic therapy group was selected randomly without repeat sampling. Similar matching processes were performed for the comparisons of warfarin versus no‐antithrombotic therapy based on the propensity scores for the use of warfarin.

Table 1

Baseline Characteristics of Patients

Variables	AF Patients With Liver Cirrhosis Having a CHA2DS2‐VASc Score ≥2 (n=9056)				P Valuea
	All	No Antithrombotic Therapy (n=5532)	Antiplatelet Agents (n=2770)	Warfarin (n=754)
Age, y	73.1±11.2	73.5±11.7	73.4±9.9	68.9±11.4	<0.001
Sex (male), n (%)	5506 (60.8)	3264 (59.0)	1771 (63.9)	471 (62.5)	<0.001
CHA2DS2‐VASc score	4.7±1.8	4.6±1.8	4.9±1.8	4.6±1.8	<0.001
Medical history (components of the CHA2DS2‐VASc score), n (%)
Hypertension	7554 (83.4)	4503 (81.4)	2454 (88.6)	597 (79.2)	<0.001
Diabetes mellitus	4096 (45.2)	2443 (44.2)	1327 (47.9)	326 (43.2)	0.003
Congestive heart failure	4995 (55.2)	2929 (52.9)	1582 (57.1)	484 (64.2)	<0.001
Previous stroke/TIA	3812 (42.1)	2194 (39.7)	1272 (45.9)	346 (45.9)	<0.001
Previous vascular disease	2628 (29.0)	1251 (22.6)	1098 (39.6)	279 (37.0)	<0.001
Medical history (other than the components of the CHA2DS2‐VASc score), n (%)
COPD	4675 (51.6)	2918 (52.7)	1448 (47.7)	309 (41.0)	<0.001
Hyperlipidemia	2682 (29.6)	1435 (25.9)	1014 (36.6)	233 (30.9)	<0.001
Malignancy	1333 (14.7)	912 (16.5)	342 (12.3)	79 (10.5)	<0.001
Autoimmune diseases	834 (9.2)	507 (9.2)	271 (9.8)	56 (7.4)	0.137
End‐stage renal disease	504 (5.6)	358 (6.5)	132 (4.8)	14 (1.9)	<0.001
HBV infection	1362 (15.0)	848 (15.3)	384 (13.9)	130 (17.2)	0.044
HCV infection	1981 (21.9)	1278 (23.1)	558 (20.1)	145 (19.2)	0.002
Hepatic encephalopathy	628 (6.9)	514 (9.3)	89 (3.2)	25 (3.3)	<0.001
EV with bleeding	421 (4.6)	339 (6.1)	64 (2.3)	18 (2.4)	<0.001
Degree of urbanization, n (%)
Urban	4062 (44.9)	2408 (43.5)	1291 (46.6)	363 (48.1)	0.030
Suburban	3181 (35.1)	1983 (35.8)	950 (34.3)	248 (32.9)
Rural	1813 (20.0)	1141 (20.6)	529 (19.1)	143 (19.0)
Income level, n (%)
Low	4940 (54.5)	3107 (56.2)	1443 (52.1)	390 (51.7)	0.004
Median	3199 (35.3)	1879 (34.0)	1037 (37.4)	283 (37.5)
High	917 (10.1)	543 (9.9)	290 (10.5)	81 (10.7)

AF indicates atrial fibrillation; COPD, chronic obstructive pulmonary disease; EV, esophageal varices; HBV, hepatitis B virus; HCV, hepatitis C virus; TIA, transient ischemic attack.

P value between groups with different strategies for stroke prevention (no antithrombotic therapy, antiplatelet agents, and warfarin).

Statistical Analysis

Data were presented as the mean value and standard deviation for normally distributed continuous variables and proportions for categorical variables. Differences between continuous values were assessed using an unpaired 2‐tailed t test or 1‐way ANOVA for the comparisons of 3 groups. Differences between nominal variables were compared by the chi‐squared test. The incidence of ischemic stroke and ICH were calculated from dividing the number of events by person‐time at risk, with the 95%CI estimated by exact binomial probabilities. The risk of ischemic stroke and ICH was assessed using the Cox regression analysis. For the comparisons of the risk of ischemic stroke and ICH among patients with or without liver cirrhosis, the analysis was adjusted for age, sex, CHA2DS2‐VASc score, chronic obstructive pulmonary disease, hyperlipidemia, malignancy, autoimmune diseases, end‐stage renal disease, degree of urbanization, and income level. Among patients with liver cirrhosis without the propensity match, the comparisons of the risk of ischemic stroke and ICH between different treatment groups were adjusted for age, sex, CHA2DS2‐VASc score, chronic obstructive pulmonary disease, hyperlipidemia, malignancy, autoimmune diseases, end‐stage renal disease, hepatitis B virus infection, hepatitis C virus infection, hepatic encephalopathy, esophageal varices with bleeding, degree of urbanization, and income level. Statistical significance was set at a P<0.05.

The present study was approved by the Institutional Review Board at Taipei Veterans General Hospital, Taipei, Taiwan, and the informed consent of study subjects was waived.

Results

Risk of Ischemic Stroke and ICH for AF Patients With or Without Liver Cirrhosis

Baseline characteristics of patients with or without liver cirrhosis are shown in Table S1. Figure 2 shows the risk of ischemic stroke and ICH for AF patients with liver cirrhosis compared to those without liver cirrhosis, stratified based on the strategies for stroke prevention. For patients who did not receive antithrombotic therapies, the risk of ischemic stroke and ICH was higher for AF patients with liver cirrhosis compared with those without after the adjustment for age, sex, CHA2DS2‐VASc score, chronic obstructive pulmonary disease, hyperlipidemia, malignancy, autoimmune diseases, end‐stage renal disease, degree of urbanization, and income level, with an adjusted hazard ratio (HR) of 1.10 (95%CI=1.00‐1.20, P=0.046) and 1.20 (95%CI=1.01‐1.43, P=0.043), respectively (Figure 2). For patients treated with warfarin, the adjusted risk of ischemic stroke and ICH was similar between patients with and without liver cirrhosis (Figure 2). Among patients treated with antiplatelet agents, patients with liver cirrhosis had a similar risk of ischemic stroke but a higher risk of ICH compared with those without (Figure 2).

Figure 2

Risk of ischemic stroke and ICH for AF patients with or without liver cirrhosis, stratified based on the strategies for stroke prevention. For patients who did not receive antithrombotic therapies, the risk of ischemic stroke and ICH was higher for AF patients with liver cirrhosis compared with those without. For patients treated with warfarin, the risk of ischemic stroke and ICH was similar between patients with or without liver cirrhosis. The hazard ratio was adjusted for age, sex, CHA 2 DS 2‐VASc score, COPD, hyperlipidemia, malignancy, autoimmune diseases, end‐stage renal disease, degree of urbanization, and income level. CI indicates confidence interval; ICH, intracranial hemorrhage.

Ischemic Stroke, ICH, and NCB on Antiplatelet Therapy and Warfarin Among Patients With Liver Cirrhosis Having a CHA2DS2‐VASc Score ≥2

Clinical and demographic characteristics of patients with liver cirrhosis having a CHA2DS2‐VASc score ≥2 are summarized in Table 1. There were significant differences in age with warfarin users being significantly lower than non–warfarin users, and mean CHA2DS2‐VASc score was higher in antiplatelet therapy users. Of associated comorbidities, warfarin users tended to have less hypertension, diabetes mellitus, and heart failure, wherefore 45.9% of those on warfarin (or antiplatelet therapy) had had a prior stroke/TIA. Warfarin users tended to have fewer comorbidities with CHA2DS2‐VASc score components and fewer complications of liver cirrhosis such as hepatic encephalopathy and esophageal varices with bleeding. Minor differences in degree of urbanization and income level were evident.

The adjusted risks for ischemic stroke and ICH are shown in Table 2. Compared to those on no antithrombotic therapy (references), patients taking antiplatelet therapy had a similar risk of ischemic stroke (HR=1.02, 95%CI=0.88‐1.18), but the risk was significantly lowered among warfarin users (HR=0.76, 95%CI=0.58‐0.99) (Table 2). For ICH, there were no significant differences between those untreated and those taking antiplatelet therapy or warfarin (Table 2). The effect sizes (95%CI) of these analyses comparing the risk of ischemic stroke and ICH of different treatment groups are shown in Table S2. When assessing the NCB, we found that that use of warfarin had a positive NCB when compared with being untreated or using antiplatelet therapy (Table 3).

Table 2

Risk of Ischemic Stroke and ICH Stratified Based on the Strategies for Stroke Prevention

Stroke Prevention Strategy	n	Ischemic Stroke				ICH
		No. of Event	Incidence (95% CI)a	Adjusted HRb (95% CI)	P Value	No. of Event	Incidence (95% CI)a	Adjusted HRb (95% CI)	P Value
No antithrombotic therapy (reference group)	5532	447	4.09 (3.72‐4.46)	Reference	···	107	0.92 (0.75‐1.09)	Reference	···
Antiplatelet agents	2770	338	4.13 (3.70‐4.56)	1.02 (0.88‐1.18)	0.784	77	0.87 (0.68‐1.06)	0.96 (0.71‐1.30)	0.811
Warfarin	754	65	2.79 (2.12‐3.46)	0.76 (0.58‐0.99)	0.040	27	1.11 (0.69‐1.53)	1.27 (0.82‐1.95)	0.284

AF indicates atrial fibrillation; COPD, chronic obstructive pulmonary disease; EV, esophageal varices; HBV, hepatitis B virus; HCV, hepatitis C virus; HR, hazard ratio; ICH, intracranial hemorrhage.

Per 100 person‐years of follow‐up.

Adjusted for age, sex, CHA2DS2‐VASc score, COPD, hyperlipidemia, malignancy, autoimmune diseases, end‐stage renal disease, HBV infection, HCV infection, hepatic encephalopathy, EV with bleeding, degree of urbanization, and income level.

Table 3

The Net Clinical Benefit Analyses for Each Treatment According to Different Weight Models

Stroke Prevention Strategy	NCB Based on Different Weight Models, % Per Year (95%CI)
	Relative Weight of ICH Compared to Ischemic Stroke According to Singer et al16 Weight=1.5	Relative Weight of ICH Compared to Ischemic Stroke According to Connolly et al17 Weight=3.08	Relative Weight of ICH Compared to Ischemic Stroke According to Lip et al4 Weight=2.44
Compared to no antithrombotic therapy (reference group)	···	···	···
Warfarin	1.02 (0.98‐1.05)	0.71 (0.63‐0.80)	0.84 (0.77‐0.90)
Compared to antiplatelet drugs (reference group)	···	···	···
Warfarin	0.98 (0.93‐1.03)	0.60 (0.49‐0.71)	0.75 (0.70‐0.84)

ICH indicates intracranial hemorrhage; NCB, net clinical benefit.

Propensity‐Matched Analysis

For the propensity‐matched cohorts of no antithrombotic therapy versus antiplatelet therapy, and no antithrombotic therapy versus warfarin, patient clinical and demographic characteristics are summarized in Table 4. Propensity scores between 2 groups in each comparison were similar. Age, sex, comorbidities, degree of urbanization, and income level were not significantly different between the groups in each comparison.

Table 4

Baseline Characteristics of AF Patients After the Propensity Match

Variables	Antiplatelet Agents Vs No Antithrombotic Therapy			Warfarin Vs No Antithrombotic Therapy
	No Antithrombotic Therapy (n=2770)	Antiplatelet Agents (n=2770)	P Value	No Antithrombotic Therapy (n=754)	Warfarin (n=754)	P Value
Age, y	73.3±11.8	73.4±9.9	0.696	68.3±12.9	68.9±11.4	0.309
Sex (male), n (%)	1780 (64.3)	1771 (63.9)	0.801	465 (61.7)	471 (62.5)	0.750
CHA2DS2‐VASc score	4.93±1.8	4.95±1.8	0.638	4.51±1.9	4.58±1.8	0.454
Medical history (components of the CHA2DS2‐VASc score), n (%)
Hypertension	2445 (88.3)	2454 (88.6)	0.705	607 (80.5)	597 (79.2)	0.521
Diabetes mellitus	1322 (47.7)	1327 (47.9)	0.893	320 (42.4)	326 (43.2)	0.755
Congestive heart failure	1616 (58.3)	1582 (57.1)	0.355	486 (64.5)	484 (64.2)	0.914
Previous stroke/TIA	1250 (45.1)	1272 (45.9)	0.553	338 (44.8)	346 (45.9)	0.679
Previous vascular disease	1077 (38.9)	1098 (39.6)	0.564	264 (35.0)	279 (37.0)	0.421
Medical history (other than the components of the CHA2DS2‐VASc score), n (%)
COPD	1485 (53.6)	1448 (52.3)	0.319	318 (42.2)	309 (41.0)	0.638
Hyperlipidemia	966 (34.9)	1014 (36.6)	0.178	242 (32.1)	233 (30.9)	0.618
Malignancy	347 (12.5)	342 (12.3)	0.839	75 (9.9)	79 (10.5)	0.734
Autoimmune diseases	278 (10.0)	271 (9.8)	0.753	56 (7.4)	56 (7.4)	1.000
End‐stage renal disease	150 (4.7)	132 (4.8)	0.271	16 (2.1)	14 (1.9)	0.712
HBV infection	380 (13.7)	384 (13.9)	0.876	120 (15.9)	130 (17.2)	0.489
HCV infection	560 (20.2)	558 (20.1)	0.947	143 (19.0)	145 (19.2)	0.896
Hepatic encephalopathy	98 (3.5)	89 (3.2)	0.503	21 (2.8)	25 (3.3)	0.549
EV with bleeding	71 (2.6)	64 (2.3)	0.542	16 (2.1)	18 (2.4)	0.729
Degree of urbanization, n (%)
Urban	1281 (46.2)	1291 (46.6)	0.788	355 (47.1)	363 (48.1)	0.680
Suburban	960 (34.7)	950 (34.3)	0.777	262 (34.7)	248 (32.9)	0.446
Rural	529 (19.1)	529 (19.1)	1.000	137 (18.2)	143 (19.0)	0.691
Income level, n (%)
Low	1475 (53.2)	1443 (52.1)	0.389	406 (53.8)	390 (51.7)	0.410
Median	985 (35.6)	1037 (37.4)	0.147	268 (35.5)	283 (37.5)	0.423
High	310 (11.2)	290 (10.5)	0.387	80 (10.6)	81 (10.7)	0.934
Mean propensity score	0.37±0.1	0.37±0.1	0.064	0.18±0.1	0.18±0.1	0.985

AF indicates atrial fibrillation; COPD, chronic obstructive pulmonary disease; EV, esophageal varices; HBV, hepatitis B virus; HCV, hepatitis C virus; TIA, transient ischemic attack.

The risks for ischemic stroke and ICH for the 2 propensity‐matched cohorts are shown in Table 5. Compared to those on no antithrombotic therapy (references), patients taking antiplatelet therapy had a similar risk of ischemic stroke (HR=1.00, 95%CI=0.85‐1.18, P=0.970), but the risk was significantly lowered among warfarin users (HR=0.71, 95%CI=0.51‐0.99, P=0.047) (Table 5). For ICH, there were no significant differences between those untreated and those taking antiplatelet therapy or warfarin (Table 5). The effect sizes (95%CI) of these analyses comparing the risk of ischemic stroke and ICH of different treatment groups are shown in Table S3.

Table 5

Risk of Ischemic Stroke and ICH With Different Strategies for Stroke Prevention After the Propensity Match

Stroke Prevention Strategy	n	Ischemic Stroke				ICH
		No. of Event	Incidence (95%CI)a	HR (95%CI)	P Value	No. of Event	Incidence (95%CI)a	HR (95%CI)	P Value
Antiplatelet agents vs no antithrombotic therapy
No antithrombotic therapy (reference group)	2770	238	4.20 (3.68‐4.72)	Reference	···	56	0.93 (0.69‐1.17)	Reference	···
Antiplatelet agents	2770	338	4.13 (3.70‐4.56)	1.00 (0.85‐1.18)	0.970	77	0.87 (0.68‐1.06)	0.99 (0.70‐1.39)	0.942
Warfarin vs no antithrombotic therapy
No antithrombotic therapy (reference group)	754	74	4.03 (3.13‐4.93)	Reference	···	17	1.08 (0.58‐1.58)	Reference	···
Warfarin	754	65	2.79 (2.12‐3.46)	0.71 (0.51‐0.99)	0.047	27	1.11 (0.69‐1.53)	1.10 (0.62‐1.94)	0.756

COPD indicates chronic obstructive pulmonary disease; HR, hazard ratio; ICH, intracranial hemorrhage.

Per 100 person‐years of follow‐up.

Discussion

There are limited data on the stroke and ICH risks in AF patients with associated liver cirrhosis, and in this analysis we clearly show that, when compared to those on no antithrombotic therapy, patients taking antiplatelet therapy had a similar risk of ischemic stroke, but the risk was significantly lowered among warfarin users. For ICH, there were no significant differences between those untreated and those taking antiplatelet therapy or warfarin. Importantly, the NCB with warfarin was positive when compared to being left untreated or if antiplatelet therapy was used.

One previous study has demonstrated that the incidence of ICH was higher among patients with liver cirrhosis due to thrombocytopenia or prolonged international normalized ratio.18 Indeed, abnormal liver function and cirrhotic liver disease are categorized as potentially and nonmodifiable bleeding risk factors, respectively, in the 2016 AF guidelines of the European Society of Cardiology and are important components of bleeding risk assessment, such as the HAS‐BLED score.3, 19

Interestingly, liver cirrhosis is associated not only with a bleeding tendency but also with a hypercoagulation status due to the decreased synthesis of anticoagulant factors or impaired degradation of prothrombotic factors.20 We are not aware of any specific data showing that liver cirrhosis independently contributes to a higher risk of ischemic stroke in AF, but as our population profile shows, such patients are at high risk given the associated comorbidities and high CHA2DS2VASc scores. Transient liver function test abnormalities (eg, γ‐glutamyl transferase) have been noted in stroke patients, but these would not necessarily reflect underlying liver cirrhosis.21

In the present study we clearly showed that AF patients with liver cirrhosis did have a higher risk of ischemic stroke and ICH compared with those without liver cirrhosis who did not receive antithrombotic therapies (Figure 2). Given the higher risks of both ischemic stroke and ICH, how to determine the optimal stroke prevention strategy for AF patients with liver cirrhosis is a clinically difficult scenario. Our data provide evidence that thromboprophylaxis should be considered for AF patients with liver cirrhosis to avoid the risk of AF‐related stroke given the positive NCB with OAC compared to being left untreated or if antiplatelet therapy was used in such patients, as shown in Table 3. The results of the present study showed that patients taking antiplatelet therapy had a similar risk of ischemic stroke as those not treated, and therefore, antiplatelet agents should not be used for stroke prevention among AF patients with liver cirrhosis. On the contrary, the risk of ischemic stroke was significantly lowered among warfarin users. For ICH, there were no significant differences between those untreated and those taking antiplatelet therapy or warfarin, and these findings may further support the use of OACs for AF patients with liver cirrhosis.

Study Limitations

Our data were based on warfarin, and whether the findings would apply to patients taking non–vitamin K antagonist oral anticoagulants is uncertain. In keeping with registry design, we did not have data on quality of anticoagulation control (as reflected by time in therapeutic range) given the close relationship between time in therapeutic range and thromboembolism or bleeding. Also, we did not have laboratory data to provide information on degree of liver function derangement, and prognostic scores of liver cirrhosis, such as model for end‐stage liver disease and Child‐Pugh scores, were not available. We were only able to regard a history of hepatic encephalopathy and esophageal varices with bleeding as the proxies of disease severity of liver cirrhosis. It should be noted that although we have adjusted for baseline differences between different treatment groups in multivariable regression and propensity‐matching analyses, other unmeasured confounders may still exist that could confound the analyses. Also, we did not adjust for multiple testing. Furthermore, the number of patients who received warfarin treatment was small, and therefore, the further analysis of NCB in different age strata or subgroups was not feasible. Besides, the NCB model only included ischemic stroke and ICH, the most devastating bleeding complications, and did not consider other bleeding events because the severity of other bleeding varied greatly and is difficult to be ascertained in the registry database. Last, the present study only enrolled Taiwanese patients, and whether the results can be extrapolated to other populations remains uncertain. Due to these limitations mentioned above, our data should be regarded as hypothesis generating, and further prospective studies are needed.

Conclusion

AF patients with liver cirrhosis had a higher risk of ischemic stroke and ICH compared with those without. For AF patients with liver cirrhosis in the current analysis of an observational study, warfarin use was associated with a lower risk of ischemic stroke and positive NCB compared with nontreatment, and thus, thromboprophylaxis should be considered for such patients.

Sources of Funding

This work was supported in part by grants from the Ministry of Science and Technology (MOST 104‐2314‐B‐075‐024‐MY3), and Taipei Veterans General Hospital (V105B‐023).

Disclosures

None.

Table S1. Baseline Characteristics of Patients With or Without Liver Cirrhosis

Table S2. Effect Size (95%CI) of the Analysis Comparing the Risk of Ischemic Stroke and ICH of Different Treatment Groups

Table S3. Effect Size (95%CI) of the Analysis Comparing the Risk of Ischemic Stroke and ICH of Different Treatment Groups After the Propensity Match

Click here for additional data file.