Assessing Safety of Anticoagulation for Atrial Fibrillation in Patients with Cirrhosis: A Real-World Outcomes Study

Abstract

Aims

In patients with atrial fibrillation (AF) and stroke risk factors, randomized trials have demonstrated that anticoagulation decreases the risk of ischemic stroke. However, all trials to date have excluded patients with significant liver disease, leaving guidelines to extrapolate recommendations. We aim to evaluate the impact of anticoagulation on safety events in patients with AF and cirrhosis.

Methods and Results

In this retrospective cohort study, we obtained de-identified health record data to extract anticoagulation strategy, comorbidities, prescriptions, lab values, and procedures for a cohort of patients with cirrhosis who develop AF. After selecting a propensity matched population to match patients with various anticoagulation strategies, we tracked data on outcomes for death, transfusion requirements, hospital and ICU admissions. After propensity score weighting and multivariable adjustment, anticoagulation strategy was associated with increased hospital admission count (OR = 1.74 per admission, P < .001), binary risk of hospital admission (OR = 1.54, P = .010) and risk of ICU admission (OR = 1.41, P = .047). We detected no significant differences in mortality, transfusion of blood products, or average length of stay. Direct oral anticoagulant (DOAC) prescriptions were associated with increased binary risk of hospital admission compared to warfarin prescriptions. In a third comparison, DOAC strategy alone was associated with increased hospital admission count (OR = 1.41 per admission, P < .001) and binary risk of hospital admission (OR = 1.52, P = .038) compared to no anticoagulation strategy.

Conclusion

Anticoagulation strategy in patients with cirrhosis and AF was associated with increased rate of hospital admission and ICU admission but not associated with increased risk of mortality or transfusion requirement.

Keywords

atrial fibrillation cirrhosis anticoagulation safety outcomes

Introduction

Current guidelines recommend anticoagulation of individuals with atrial fibrillation (AF) who have a predicted yearly risk of stroke greater than two percent.^1,2 These recommendations are difficult to apply directly to patients with cirrhosis due to increased risk of bleeding and potential for decompensation. Clinicians lack evidence for appropriate decision-making in this area due to the exclusion of patients with significant liver disease from many seminal anticoagulation trials and all DOAC trials.^3–9 Specific guidelines for patients with cirrhosis feature conditional recommendations favoring anticoagulation in mild disease states but note the low level of evidence and indication for shared decision-making with patients.^1,10

Cirrhotic pathophysiology increases the risks of both bleeding and thrombosis.¹¹ Literature has shown an increased prevalence of AF in patients with cirrhosis.^12–15 AF is a prognostic risk factor associated with worse outcomes in end-stage liver disease, and cirrhosis has been associated with higher risk for hemorrhagic as well as ischemic stroke including from AF-related etiology.^16–18 Despite links between AF and cirrhosis, major safety and efficacy trials of anticoagulation (AC) have excluded patients with moderate and severe liver disease leaving clinicians with poor evidence to assess accurate risk of bleeding in this patient population. Prior observational studies demonstrated encouraging safety results for major bleeding and mortality despite variable definitions of major bleeding. Study populations were skewed towards mild-moderate cirrhosis and often utilized database or registry coding to identify patients with cirrhosis. Multiple studies and meta-analyses have noted no significant difference in risk of bleeding for patients with cirrhosis receiving oral AC compared to those not receiving AC^19–21 however other studies did demonstrate the increased risk of bleeding events.²² As global rates of cirrhosis and AF rise, it is critical to characterize the risks and benefits of AC in patients with cirrhosis to refine clinical guidelines. Prior research in the field has been limited by warfarin-dominant prescription landscapes and cohorts with primarily mild chronic liver disease.

This study is a retrospective cohort study of patients with cirrhosis who develop atrial fibrillation. Our aim was to characterize cirrhosis severity and compare AC safety outcomes, in particular, mortality, transfusion requirements, hospital and intensive care unit admissions and lengths of stays, among different groups of anticoagulants and compared to patients following a non-anticoagulation strategy. We studied a diverse cohort in a DOAC-prevalent landscape with significant numbers of patients with moderate and severe cirrhosis.

Methods

This retrospective cohort study was completed using the UCLA Data Discovery Repository (DDR). The DDR is a research database containing data pulled from de-identified electronic health record (EHR) information of all patients interacting with the UCLA Health System. Clinical data is structured and includes demographics, hospitalization data, International Classification of Diseases (ICD) −9 and −10 diagnostic codes, Current Procedural Terminology (CPT) codes, prescription and lab value information. Complete EHR data is available beginning in 2013 and this study used records up until December 2022.

We created a cohort consisting of all adult patients with at least one year of follow-up data after 2013 who carried one or more ICD codes for atrial fibrillation and cirrhosis (codes listed in Supplemental Tables S8-9). Timing of diagnosis was determined by date of first ICD code. ICD code diagnoses made in the first 6 months of follow-up were considered baseline or as having been diagnosed prior to establishing care. As stated above, only patients with atrial fibrillation diagnosed after cirrhosis or concurrently with cirrhosis were included in this analysis.

Clinical Characterization

Clinical data pulled from the DDR included baseline data such as demographics, comorbid conditions, relevant prescriptions, laboratory values as well as our safety outcomes of hospital and ICU days, transfusion requirements by CPT code, gastrointestinal or intracranial bleeding diagnoses and all-cause mortality. Pre-2016 Model for End-Stage Liver Disease (MELD) scores and CHA₂DS₂-VASc scores were calculated using the above data to stratify cirrhosis severity and risk of thromboembolism respectively. We used data from the six months preceding atrial fibrillation diagnosis to calculate these scores. We created an ad hoc cirrhosis severity qualification score by defining severe cirrhosis as the presence of two or more features suggesting decompensation: prescriptions for lactulose, non-selective beta blockers, or loop diuretics in addition to CPT procedure codes for paracentesis or diagnostic esophagogastroduodenoscopy (EGD) for varices from the time preceding atrial fibrillation diagnosis.

A patient's AC strategy was determined by the presence or absence of an outpatient AC prescription of warfarin, a DOAC (dabigatran, rivaroxaban, apixaban, edoxaban), or injectable AC (enoxaparin and fondaparinux) after the diagnosis of atrial fibrillation. AC strategy was determined by the first AC prescription. Patients prescribed multiple types of AC in follow-up were excluded from comparisons between AC groups (warfarin compared to DOAC for example). The cohort was subsequently followed for targeted outcomes from the time of the initial AC prescription.

Transfusion was determined by a procedural CPT code indicating transfusion of blood or blood components. The binary transfusion outcome indicates if a patient has ever received a transfusion whereas the ‘count’ transfusion outcome indicates the total number of transfusions a patient has experienced. All-cause inpatient hospital admission data does not distinguish between planned or unplanned admission. The binary hospital admission outcome indicates if a patient was ever admitted whereas the ‘count’ hospital admission outcome indicates how many inpatient admissions a patient experienced. Length of stay (LOS) per admission is the number of days patients stay in the hospital per admission encounter. This outcome data is reported for each individual admission not just per patient. Intensive care unit (ICU) admission is a binary variable indicating if a patient was ever admitted to an ICU. An ICU admission was counted if patients were admitted to an ICU bed for at least 16 hours.

Statistical Analysis

Multiple imputed chained equations (MICE) were used to impute baseline values of creatinine, total bilirubin, and International Normalized Ratio (INR) in order to create a baseline MELD score for the ∼29% of individuals missing this score. Minimum sodium, hemoglobin, and platelet levels were also similarly imputed. Imputations were carried out using 10 imputations and were based on linear regression models using all of the predictors listed in Supplemental Table S1. Within each imputation, a propensity score was created to predict anticoagulated and non-anticoagulated group membership at follow-up based on the predictors described in Table S1. This propensity score was then converted to an inverse probability weight to be used in all subsequent analysis in concert with regression adjustment for the variables used in creating the propensity score, termed inverse probability weighting with regression adjustment (IPW-RA). Bivariate and multivariable analyses examined differences between the AC and non-AC groups as well as between specific subsets of AC such as warfarin, DOAC, and injectable AC subgroups. Logistic regression models for the multiply imputed data using the IPW-RA approach were used to assess group differences in our binary safety outcomes which were: transfusion, hospitalization, all-cause mortality, and ICU admission.²³ Negative binomial regression models were used for count versions of these outcomes (number of transfusions, hospital stays) as well as for length of stay (LOS) analyses for the average number of days in the hospital per stay. Analyses were conducted using Stata version 18.0 (StataCorp LLC, College Station, Texas).

Results

Baseline Characteristics in Anticoagulated and Non-Anticoagulated Patients

We studied a cohort of 1063 patients with cirrhosis who subsequently or concurrently developed AF and received care in our health system from 2013 to 2022. In this cohort, 485 were prescribed outpatient AC (45.6%) (Table 1). At baseline prior to AF diagnosis, 20% of patients had heart failure, 9% had a history of myocardial infarction, 13% were taking cardio-selective beta blockers and no patients were on anti-arrhythmic medications. Additionally, 18% of patients had a history of esophageal or gastrointestinal varices, 36% had at least one feature of decompensated cirrhosis, and 20% had a history of blood product transfusion. Our institution's liver transplant committee had evaluated 30% of patients (324) prior to AF diagnosis though only two underwent transplant (Table 1). Prior to AF diagnosis, the mean CHA₂DS₂-VASc score was 2.7, median CHA₂DS₂-VASc score 3.0, and the mean MELD score was 19.6. The MELD score was imputed for 305 (∼29%) of these patients and the minimum sodium, hemoglobin, and platelet values were imputed for 111 (∼10%), 122 (∼11.5%), and 118 (11%) patients respectively. The median MELD was not reported due to imputation (Table 1). Prior to the development of AF, patients with non-AC strategy had higher mean MELD scores and were more likely to have 1 or more features of decompensated cirrhosis compared to patients with AC strategy. Patients prescribed AC had higher mean CHA₂DS₂-VASc scores. Patients with AC strategy were also more likely to have a longer follow-up time as well as a longer time in between the diagnosis of cirrhosis and AF in contrast to patients with non-AC strategy (Table 1). Inverse probability weighting eliminated significant differences in baseline covariates in the comparison between AC strategy and non-AC strategy groups (Supplemental Table S1). Distributions of propensity scores for each multiply imputed dataset between anticoagulated and non-anticoagulated patients demonstrate adequate overlap in the unweighted state and well-matched distributions after weighting was conducted (Supplemental Figure 1). At the end of the follow-up period, the following outcomes were observed: 276 patients (26%) deceased, 519 patients requiring transfusion (49%), 637 patients hospitalized for any cause (60%), 258 patients with ICU admissions (24%) (Supplemental Table S2).

Table 1.

Baseline Characteristics, Overall and by Anticoagulation Status.

Covariate	All patients	Anticoagulated	No Anticoagulation	P-value
N	1063	485 (46%)	578 (54%)
Anticoagulation Prior to Cirrhosis (N, %)	218 (21%)	145 (30%)	73 (13%)	<.001
Age at Cirrhosis Diagnosis (Mean, SD)	64.2 (13.7)	64.4 (14.5)	63.6 (12.1)	.611
Male Sex (N, %)	658 (62%)	298 (61%)	360 (62%)	.799
Race/Ethnicity				.474
Non-Hispanic Asian (N, %)	94 (9%)	38 (8%)	56 (10%)
Non-Hispanic Black (N, %)	67 (6%)	28 (6%)	39 (7%)
Hispanic (N, %)	306 (29%)	133 (27%)	173 (30%)
Other or Unknown Race/Ethnicity (N, %)	153 (14%)	82 (14%)	71 (15%)
Non-Hispanic White (N, %)	443 (42%)	215 (44%)	228 (39%)
Years Between AF and Cirrhosis Diagnoses (Mean, SD)	2.1 (2.2)	2.4 (2.3)	1.8 (2.1)	<.001
Follow-up Time in Years (Mean, SD)	4.5 (3.1)	5.2 (3.1)	3.9 (3.0)	<.001
Smoking Status Prior to AF and Cirrhosis Diagnoses				.845
Current (N, %)	56 (5%)	23 (5%)	33 (6%)
Former (N, %)	434 (41%)	203 (42%)	231 (40%)
Never (N, %)	567 (53%)	256 (53%)	311 (54%)
BMI at Diagnoses (Mean, SD)	27.4 (6.1)	27.7 (6.2)	27.2 (6.0)	.185
Prescriptions Prior to AF and Cirrhosis Diagnoses
Antiplatelet Therapy (Non-Aspirin) (N, %)	88 (8%)	43 (9%)	45 (8%)	.576
Aspirin (N, %)	339 (32%)	178 (37%)	161 (28%)	.002
Beta-Blocker (Cardioselective) (N, %)	138 (13%)	69 (14%)	69 (12%)	.273
Beta-Blocker (Non-selective) (N, %)	137 (13%)	50 (10%)	87 (15%)	.021
Lactulose (N, %)	270 (25%)	94 (19%)	176 (30%)	<.001
Loop Diuretic (N, %)	561 (53%)	244 (50%)	317 (55%)	.156
Proton Pump Inhibitor (N, %)	588 (55%)	252 (52%)	336 (58%)	.047
Thiamine (N, %)	56 (5%)	21 (4%)	35 (6%)	.218
Trimethoprim / Sulfamethoxazole (N, %)	177 (17%)	90 (19%)	87 (15%)	.137
Comorbidity Prior to AF and Cirrhosis Diagnoses
Esophageal / Gastrointestinal Varices (N, %)	187 (18%)	70 (14%)	117 (20%)	.015
Heart Failure (N, %)	209 (20%)	110 (23%)	99 (17%)	.024
Liver Transplant Committee Evaluation (N, %)	324 (30%)	109 (22%)	215 (37%)	<.001
Major Bleed (N, %)	192 (18%)	81 (17%)	111 (19%)	.299
Mitral Stenosis (N, %)	23 (2%)	19 (4%)	4 (1%)	<.001
Myocardial Infarction (N, %)	91 (9%)	44 (9%)	47 (8%)	.584
Venous Thromboembolism (N, %)	52 (5%)	36 (7%)	16 (3%)	<.001
Procedural History Prior to AF and Cirrhosis Diagnoses
Esophagogastroduodenoscopy (EGD) (N, %)	83 (8%)	34 (7%)	49 (8%)	.422
Paracentesis (N, %)	21 (2%)	9 (2%)	12 (2%)	.828
Transfusion (N, %)	212 (20%)	80 (16%)	132 (23%)	.010
Clinical Scores Prior to AF and Cirrhosis Diagnoses
CHA₂DS₂-VASc Score^a (Mean, SD)	2.7 (1.3)	2.8 (1.3)	2.6 (1.3)	.029
CHA₂DS₂-VASc Score^a (Median, IQR)	3.0 (2.0, 3.0)	3.0 (2.0, 4.0)	2.0 (2.0, 3.0)	.029
MELD Score^b (Mean, SD)	19.6 (11.1)	18.8 (11.4)	20.3 (10.1)	.026
Features of Decompensated Cirrhosis Score^c				.012
Decompensated Cirrhosis Score 0 (N, %)	672 (63%)	328 (68%)	344 (60%)
Decompensated Cirrhosis Score 1 (N, %)	267 (25%)	114 (24%)	153 (26%)
Decompensated Cirrhosis Score 2 (N, %)	111 (10%)	37 (8%)	74 (13%)
Decompensated Cirrhosis Score > 2 (N, %)	13 (1%)	6 (1%)	7 (1%)
Minimum Lab Values Prior to AF and Cirrhosis Diagnoses
Hemoglobin g/dl (Mean, SD)	10.4 (3.4)	10.9 (3.6)	10.1 (2.8)	<.001
Platelet Count in Thousands per mcL (Mean, SD)	126.1 (107.1)	141.7 (113.1)	113.0 (91.1)	<.001
Sodium mmol/L (Mean, SD)	134.7 (7.8)	135.5 (8.2)	134.0 (6.5)	<.001

Values are n (%), or mean SD.

Scoring tool for non-valvular AF. One point received for each of the following: heart failure, hypertension, diabetes mellitus, vascular disease (eg peripheral artery disease, myocardial infarction), female sex, age 65-74 years; two points for prior cerebrovascular accident or thromboembolism as well as age greater than or equal to 75 years.

Pre-2016 Model for End-Stage Liver Disease score calculated with the formula: (0.957*ln(Serum creatinine) + 0.378*ln(Serum bilirubin) + 1.120*ln(INR) + 0.643)*10.

Ad hoc cirrhosis severity qualification: severe cirrhosis defined as the presence of two or more features suggesting decompensation such as prescriptions for lactulose, non-selective beta blockers, or loop diuretics, history of paracentesis or diagnostic esophagogastroduodenoscopy.

Comparison of Anticoagulation and Non-anticoagulation Strategies

After propensity score weighting and multivariable adjustment, AC strategy was associated with increased hospital admission count (Odds Ratio [OR] = 1.74 per hospital admission, P < .001), and binary risk of hospital admission (OR = 1.54, P = .010) (Figure 1). AC strategy was also associated with increased risk of intensive care unit admission (OR = 1.41, P = .047) with a trend towards increased length of stay per hospital admission (OR = 1.24, P = .058). We detected no significant differences in mortality and risk of transfusion of blood products (Figure 1). Analysis stratifying by age or MELD score demonstrated no significant differences.

Figure 1.

Multivariate comparison of anticoagulation strategy versus no anticoagulation strategy. Odds ratios (ORs) and 95% confidence intervals (CI) from the model are shown in the figure. Multivariable Cox proportional hazard modeling was used to evaluate rate of outcomes. All variables were adjusted for after propensity weighting.

Baseline Characteristics and Comparisons Between Anticoagulation Strategies – DOAC and Warfarin

Of the 485 patients prescribed AC, 251 were initially prescribed DOACs (56%), 98 were initially prescribed Warfarin (26%), and 84 (17%) were initially prescribed injectable AC. Warfarin prescriptions were more common prior to 2015 in our dataset and patient cohort. Of DOAC prescriptions, 74.5% were for Apixaban, 7.2% for Dabigatran, 25.4% for Rivaroxaban, and 0.4% for Edoxaban. As referenced above, patients prescribed multiple types of AC in follow-up were excluded from comparisons between AC groups. Thus, 349 patients were prescribed only-DOAC or only-Warfarin with 251 in the only-DOAC group and 98 in the only-warfarin group. Patients initially prescribed DOACs compared to warfarin were more likely to have higher CHA₂DS₂-VASc scores whereas patients initially prescribed warfarin were more likely to have higher MELD scores and to have a history of major bleed and liver transplant committee evaluation. Patients initially prescribed DOACs were more likely to be older, have longer follow-up times, and have aspirin prescriptions. Patients initially prescribed warfarin were more likely to have preexisting AC prescriptions (Supplemental Table S3). Inverse probability weighting eliminated every significant difference in baseline covariates in the comparison between DOAC-prescription and warfarin-prescription groups except for smoking status (Supplemental Table S4). At the end of the follow-up period, the following outcomes were observed: 79 patients (24%) deceased, 154 patients requiring transfusion (44%), 219 patients hospitalized for any cause (63%), 111 patients with ICU admissions (32%) (Supplemental Table S5). After propensity score weighting and multivariable adjustments, we found that DOAC prescription was associated with increased binary risk of hospital admission (OR = 4.70, P < .001) without a similarly significant association with hospital admission count, length of stay, or intensive care unit admission. We found no significant differences among mortality or transfusion outcomes (Table 2). Analysis stratifying by age or MELD score demonstrated no significant differences.

Table 2.

Multivariate Comparisons of DOAC Versus Warfarin and DOAC Versus No Anticoagulation.

	DOAC versus Warfarin		DOAC versus No Anticoagulation
Outcome	OR (95% CI)	P-value	OR (95% CI)	P-value
All-cause Mortality	1.04 (0.49-2.18)	.92	0.84 (0.55-1.30)	.44
Transfusion (binary)	0.97 (0.45-2.09)	.94	1.04 (0.69-1.56)	.84
Transfusion (count)	1.34 (0.83-2.18)	.23	0.792 (0.59-1.06)	.11
Hospital Admission (binary)	4.70 (1.94-11.37)	<.001	1.52 (1.02-2.24)	.038
Hospital Admission (count)	1.45 (0.98-2.15)	.064	1.41 (1.16-1.72)	<.001
Length of Stay Per Admission (days)	0.97 (0.72-1.31)	.851	0.998 (0.80-1.25)	.99
Intensive Care Unit Admission (binary)	1.16 (0.48-2.77)	.746	1.48 (0.92-2.39)	.11

Odds ratios (ORs) and 95% confidence intervals (CI) from the model are shown in the table. Multivariable Cox proportional hazard modeling was used to evaluate rate of outcomes. All variables were adjusted for after propensity weighting.

Baseline Characteristics and Comparisons Between Anticoagulation Strategies – DOAC Alone and Non-Anticoagulated Patients

In the study cohort, 829 patients were either prescribed DOAC or not prescribed any AC. At baseline, patients with non-AC strategy had higher mean MELD scores and were more likely to have a history of transfusion and evaluation by liver transplant committee compared to patients initiated on DOACs. Patients with non-AC strategy were also more likely to have diagnosed varices and preexisting lactulose prescriptions. Patients initially prescribed DOACs had higher mean and median CHADS2-VASC scores and were more likely to have preexisting aspirin and AC prescriptions in contrast to patients with non-AC strategy. Patients initially prescribed DOACs were older, had longer follow-up times, and were more likely to carry heart failure diagnoses (Supplemental Table S6). Inverse probability weighting eliminated significant differences in baseline covariates in the comparison between DOAC-prescription and no-anticoagulation strategy groups (Supplemental Table S7). After propensity score weighting and multivariable adjustments, DOAC strategy was associated with increased hospital admission count (OR = 1.41 per hospital admission, P < .001) and binary risk of hospital admission (OR = 1.52, P = .038). We detected no significant differences in other outcomes including risks of mortality, transfusion, or ICU admission (Table 2). Analysis stratifying by age or MELD score demonstrated no significant differences.

Discussion

We present a retrospective analysis of safety outcomes for AC strategies in patients with cirrhosis who develop atrial fibrillation. We demonstrate that an AC strategy is associated with increased risk for hospitalization and ICU admission but not increased risk for mortality or transfusion when compared to propensity matched patients who do not receive outpatient AC. When compared to a non-AC strategy, DOAC prescription alone was associated with increased hospitalization risk. Warfarin and DOAC groups differed in hospital admission risk.

Our findings demonstrate that anticoagulating patients with cirrhosis is associated with increases in hospitalization but not increased mortality or transfusion risk. To our knowledge, no other studies have evaluated safety outcomes such as hospitalizations, average length of stay, and ICU admissions in patients with cirrhosis receiving AC for AF.

Possible explanations for lack of difference in transfusion rates include transfusion frequency and number being driven more by underlying cirrhosis rather than AC or bleeding episodes that did not require transfusion. Our sample size also limits our ability to detect smaller differences in transfusion frequency. Our mortality rate findings may reflect a balance between increased bleeding events and decreased thromboembolic events or insufficient power. Inherent differences between groups of patients receiving anticoagulation and patients following a non-AC strategy were addressed with propensity scoring analysis however some differences may remain unaccounted for.

Multiple investigations have demonstrated acceptable bleeding risk when anticoagulating patients with chronic liver disease and/or cirrhosis. Observational studies have reported rates ranging from 9-21%.^{19,21,22,24–30} Our study corroborates and extends these findings by including a population with DOAC-prevalent prescription patterns and significant percentages of patients with moderate and severe cirrhosis, many of whom were evaluated by a liver transplant committee.

However, we also presented significant associations between AC and increased safety events such as hospitalizations and ICU admissions. These results suggest that close monitoring and specific follow-up may be needed after AC initiation. Interventions such as AC clinics may assist in preventing safety events and help alleviate resource utilization in patients with cirrhosis and AF.

We also demonstrated warfarin and DOACs have similar rates of mortality, bleeding risk, and other safety events excluding hospital admission in agreement with the majority of prior studies that demonstrate no differences in bleeding or mortality in head-to-head comparison between DOACs and warfarin in patients without cirrhosis.^{21,28,31–33} Our study is the first to explicitly examine hospital and ICU admission count in patients with cirrhosis prescribed AC after an AF diagnosis. Several investigations in patients without cirrhosis demonstrated lower rates of inpatient admissions and unplanned readmissions in patients receiving DOACs compared to those receiving warfarin.^34–39 Other studies in general populations did not identify differences in rate of hospitalizations, LOS, ICU stay or overall costs.^34,36,40

Differences between our findings and prior literature could partially result from heterogeneous pharmacokinetics among DOACs although we did not perform intra-DOAC comparisons. For example, apixaban has increased hepatobiliary elimination compared with dabigatran and rivaroxaban.³¹ Guidelines currently recommend against rivaroxaban prescription in patients with moderate hepatic impairment due to significant increase in area-under-the-curve due to decreased total body clearance.^1,41 Given our de-identified data set, we were unable to assess time in therapeutic range of warfarin for patients who may have had difficult titration regimens especially with moderate or severe cirrhosis. Our study spanned the introduction of DOACs which may have impacted prescription patterns and decision-making around adverse events such as inpatient admission for minor bleeding. Clinicians may have been less familiar with DOACs and more wary of bleeding due to lack of access to reversal agents at the time compared to warfarin. When evaluating LOS and admission risk, warfarin's downsides include requiring titration prior to discharge, referral to an AC clinic or bridging during a hospitalization. Baseline differences between patients receiving warfarin and those receiving DOACs were adjust for with propensity scoring matching however unaccounted differences may have persisted.

Strengths of our study include a patient population with significant numbers of DOAC prescriptions and percentages of moderate to severe cirrhosis diagnoses. Our data on procedures, prescriptions, and hospitalizations were attached to individual documented provider encounters and verified by the billing department at a single institution, ensuring accuracy and consistency. Our sample size enabled evaluation of multiple covariates for weighting and adjustment.

Our study is limited by its retrospective design utilizing a single center population which may have lost patients to follow-up and missed encounters in other health systems. We did not evaluate ischemic stroke, the intended therapeutic target of AC, given our study was not adequately powered to detect differences in event rates. We only evaluated initial AC therapy and did not characterize intragroup differences among DOACs due to sample size limitations. Our patient cohort consisted of subjects treated at a large academic medical center with high liver transplant volumes and substantial experience managing cirrhosis patients with high risk for clinical decompensation. Given diagnosis code usage, misclassification or miscoding error is possible without identifiable patient charts. We were unable to confirm medication compliance given medication exposure data depends on filled prescription data. Our propensity matching was limited to baseline characteristics; however, this mimics information available at time of decision making and AC refills may prompt less thorough risk benefit assessment than the initial prescription.

In conclusion, oral AC prescription in patients with cirrhosis and AF was associated with increased rate of hospital admission and ICU admissions but not associated with increased risk of mortality or transfusion requirement in our retrospective cohort of 1063 patients. Elucidating safety risks of AC in this complex population may help inform development of surveillance and management guidelines. Future prospective studies with increased sample sizes are needed to evaluate efficacy as well as safety.

Supplemental Material

sj-docx-1-cpt-10.1177_10742484241256271 - Supplemental material for Assessing Safety of Anticoagulation for Atrial Fibrillation in Patients with Cirrhosis: A Real-World Outcomes Study

Supplemental material, sj-docx-1-cpt-10.1177_10742484241256271 for Assessing Safety of Anticoagulation for Atrial Fibrillation in Patients with Cirrhosis: A Real-World Outcomes Study by Justin J. Song, MD, Nicholas J. Jackson, PhD, MPH, Helen Shang, MD, Henry M. Honda, MD, and Kristin Boulier, MD in Journal of Cardiovascular Pharmacology and Therapeutics

Footnotes

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The research described was supported by NIH/National Center for Advancing Translational Science (NCATS) UCLA CTSI Grant Number UL1TR001881.

ORCID iDs

Justin J. Song

Kristin Boulier

Data Availability Statement

To protect patient confidentiality, individual-level data will not be made available. Please contact the corresponding author for aggregated data or specific analyses.

Supplemental Material

Supplemental material for this article is available online.

References

Joglar

Chung

Armbruster

, et al. 2023 ACC/AHA/ACCP/HRS guideline for the diagnosis and management of atrial fibrillation: a report of the American college of cardiology/American heart association joint committee on clinical practice guidelines. Circulation. 2024;149(1):e1-e156. doi:https://doi.org/10.1161/cir.0000000000001193.

Hindricks

Potpara

Dagres

, et al. 2020 ESC guidelines for the diagnosis and management of atrial fibrillation developed in collaboration with the European association for cardio-thoracic surgery (EACTS): the task force for the diagnosis and management of atrial fibrillation of the European society of cardiology (ESC) developed with the special contribution of the European heart rhythm association (EHRA) of the ESC. Eur Heart J. 2021;42(5):373-498. doi:https://doi.org/10.1093/eurheartj/ehaa612.

Singer

Hughes

Gress

, et al. The effect of low-dose warfarin on the risk of stroke in patients with nonrheumatic atrial fibrillation. N Engl J Med. 1990;323(22):1505-1511. doi:https://doi.org/10.1056/nejm199011293232201.

Petersen

Boysen

Godtfredsen

Andersen

. Placebo-controlled, randomised trial of warfarin and aspirin for prevention of thromboembolic complications in chronic atrial fibrillation. The Copenhagen AFASAK study. Lancet. 1989;1(8631):175-179. doi:https://doi.org/10.1016/s0140-6736(89)91200-2.

Ezekowitz

Bridgers

James

, et al. Warfarin in the prevention of stroke associated with nonrheumatic atrial fibrillation. Veterans affairs stroke prevention in nonrheumatic atrial fibrillation investigators. N Engl J Med. 1992;327(20):1406-1412. doi:https://doi.org/10.1056/nejm199211123272002.

Giugliano

Ruff

Braunwald

, et al. Edoxaban versus warfarin in patients with atrial fibrillation. N Engl J Med. 2013;369(22):2093-2104. doi:https://doi.org/10.1056/NEJMoa1310907.

Granger

Alexander

McMurray

, et al. Apixaban versus warfarin in patients with atrial fibrillation. N Engl J Med. 2011;365(11):981-992. doi:https://doi.org/10.1056/NEJMoa1107039.

Patel

Mahaffey

Garg

, et al. Rivaroxaban versus warfarin in nonvalvular atrial fibrillation. N Engl J Med. 2011;365(10):883-891. doi:https://doi.org/10.1056/NEJMoa1009638.

Connolly

Ezekowitz

Yusuf

, et al. Dabigatran versus warfarin in patients with atrial fibrillation. N Engl J Med. 2009;361(12):1139-1151. doi:https://doi.org/10.1056/NEJMoa0905561.

10.

O’Shea

Davitkov

, et al. AGA Clinical practice guideline on the management of coagulation disorders in patients with cirrhosis. Gastroenterology. 2021;161(5):1615-1627. e1. doi:https://doi.org/10.1053/j.gastro.2021.08.015

11.

Caldwell

Carlini

. Coagulation homeostasis in liver disease. Clin Liver Dis (Hoboken). 2020;16(4):137-141. doi:https://doi.org/10.1002/cld.935.

12.

Huang

Dunipace

Sorg

Vaseghi

. Liver disease as a predictor of new-onset atrial fibrillation. J Am Heart Assoc. 2018;7(15):e008703. doi:https://doi.org/10.1161/JAHA.118.008703.

13.

Lee

Chan

Chang

, et al. Effectiveness and safety of non-vitamin K antagonist oral anticoagulant and warfarin in cirrhotic patients with nonvalvular atrial fibrillation. J Am Heart Assoc. 2019;8(5):e011112. doi:https://doi.org/10.1161/JAHA.118.011112.

14.

Wijarnpreecha

Boonpheng

Thongprayoon

Jaruvongvanich

Ungprasert

. The association between non-alcoholic fatty liver disease and atrial fibrillation: a meta-analysis. Clin Res Hepatol Gastroenterol. 2017;41(5):525-532. doi:https://doi.org/10.1016/j.clinre.2017.08.001.

15.

Park

Lee

Choi

Kim

Chung

. The risk of atrial fibrillation in patients with non-alcoholic fatty liver disease and a high hepatic fibrosis index. Sci Rep. 2020;10(1):5023. doi:https://doi.org/10.1038/s41598-020-61750-4.

16.

Darrat

Smer

Elayi

, et al. Mortality and morbidity in patients with atrial fibrillation and liver cirrhosis. World J Cardiol. 2020;12(7):342-350. doi:https://doi.org/10.4330/wjc.v12.i7.342.

17.

Kuo

Chao

Liu

, et al.

Liver cirrhosis in patients with atrial fibrillation: would oral anticoagulation have a net clinical benefit for stroke prevention?

J Am Heart Assoc. 2017;6(6). doi:https://doi.org/10.1161/JAHA.116.005307.

18.

Lai

Chien

Chung

, et al. Atrial fibrillation, liver disease, antithrombotics and risk of cerebrovascular events: a population-based cohort study. Int J Cardiol. 2016;223:829-837. doi:https://doi.org/10.1016/j.ijcard.2016.08.297.

19.

Chokesuwattanaskul

Thongprayoon

Bathini

, et al. Efficacy and safety of anticoagulation for atrial fibrillation in patients with cirrhosis: a systematic review and meta-analysis. Dig Liver Dis. 2019;51(4):489-495. doi:https://doi.org/10.1016/j.dld.2018.12.001.

20.

Serper

Weinberg

Cohen

Reese

Taddei

Kaplan

. Mortality and hepatic decompensation in patients with cirrhosis and atrial fibrillation treated with anticoagulation. Hepatology. 2021;73(1):219-232. doi:https://doi.org/10.1002/hep.31264.

21.

Steensig

Pareek

Krarup

, et al. Thromboembolism and bleeding in patients with atrial fibrillation and liver disease – A nationwide register-based cohort study: thromboembolism and bleeding in liver disease. Clin Res Hepatol Gastroenterol. 2022;46(8):101952. doi:https://doi.org/10.1016/j.clinre.2022.101952.

22.

Qamar

Antman

Ruff

, et al. Edoxaban versus warfarin in patients with atrial fibrillation and history of liver disease. J Am Coll Cardiol. 2019;74(2):179-189. doi:https://doi.org/10.1016/j.jacc.2019.04.061.

23.

Seaman

White

Copas

. Combining multiple imputation and inverse-probability weighting. Biometrics. 2012;68(1):129-137. doi:https://doi.org/10.1111/j.1541-0420.2011.01666.x.

24.

Lee

Suah

Teo

, et al. Comparison of the efficacy and safety of direct oral anticoagulants and vitamin K antagonists in patients with atrial fibrillation and concomitant liver cirrhosis: a systematic review and meta-analysis. Am J Cardiovasc Drugs. 2022;22(2):157-165. doi:https://doi.org/10.1007/s40256-021-00482-w.

25.

Proietti

Marzona

Vannini

, et al. Impact of liver disease on oral anticoagulant prescription and major adverse events in patients with atrial fibrillation: analysis from a population-based cohort study. Eur Heart J Cardiovasc Pharmacother. 2021;7(Fi1):f84-f92. doi:https://doi.org/10.1093/ehjcvp/pvaa015.

26.

Chen

Pürerfellner

Meyer

, et al. Anticoagulation in atrial fibrillation and liver disease: a pooled-analysis of >20 000 patients. Eur Heart J Cardiovasc Pharmacother. 2022;8(4):336-345. doi:https://doi.org/10.1093/ehjcvp/pvab032.

27.

Choi. Risks Versus Benefits of Anticoagulation for Atrial Fibrillation in Cirrhotic Patients. 2017.

28.

Goriacko

Veltri

. Safety of direct oral anticoagulants vs warfarin in patients with chronic liver disease and atrial fibrillation. Eur J Haematol. 2018;100(5):488-493. doi:https://doi.org/10.1111/ejh.13045.

29.

Lee

Uhm

Kim

Pak

Lee

Joung

. The safety and efficacy of vitamin K antagonist in patients with atrial fibrillation and liver cirrhosis. Int J Cardiol. 2015;180:185-191. doi:https://doi.org/10.1016/j.ijcard.2014.11.183.

30.

Pastori

Lip

GYH

Farcomeni

, et al. Incidence of bleeding in patients with atrial fibrillation and advanced liver fibrosis on treatment with vitamin K or non-vitamin K antagonist oral anticoagulants. Int J Cardiol. 2018;264:58-63. doi:https://doi.org/10.1016/j.ijcard.2018.01.097.

31.

Lee

Choi

, et al. Direct oral anticoagulants in patients with atrial fibrillation and liver disease. J Am Coll Cardiol. 2019;73(25):3295-3308. doi:https://doi.org/10.1016/j.jacc.2019.04.052.

32.

Hum

Shatzel

Jou

Deloughery

. The efficacy and safety of direct oral anticoagulants vs traditional anticoagulants in cirrhosis. Eur J Haematol. 2017;98(4):393-397. doi:https://doi.org/10.1111/ejh.12844.

33.

Intagliata

Henry

Maitland

, et al. Direct oral anticoagulants in cirrhosis patients pose similar risks of bleeding when compared to traditional anticoagulation. Dig Dis Sci. 2016;61(6):1721-1727. doi:https://doi.org/10.1007/s10620-015-4012-2.

34.

Berger

Ashton

Laliberte

, et al. Healthcare resource utilization and costs of rivaroxaban versus warfarin among non-valvular atrial fibrillation (NVAF) patients with diabetes in a US population. Adv Ther. 2023;40(3):1224-1241. doi:https://doi.org/10.1007/s12325-022-02422-9.

35.

Weir

Chen

Bookhart

Campbell

Ashton

. Healthcare resource utilization and costs of rivaroxaban versus warfarin among nonvalvular atrial fibrillation patients with obesity and diabetes. Diabetes Ther. 2021;12(12):3167-3186. doi:https://doi.org/10.1007/s13300-021-01161-4.

36.

Gilligan

Franchino-Elder

Song

, et al. Comparison of all-cause costs and healthcare resource use among patients with newly-diagnosed non-valvular atrial fibrillation newly treated with oral anticoagulants. Curr Med Res Opin. 2018;34(2):285-295. doi:https://doi.org/10.1080/03007995.2017.1409425.

37.

Deitelzweig

Luo

Gupta

, et al. Effect of apixaban versus warfarin use on health care resource utilization and costs among elderly patients with nonvalvular atrial fibrillation. J Manag Care Spec Pharm. 2017;23(11):1191-1201. doi:https://doi.org/10.18553/jmcp.2017.17060.

38.

Chowdhury

Franchino-Elder

Wang

Yuce

Wang

Hartaigh

. Healthcare resource utilization and expenditures among newly-diagnosed elderly non-valvular atrial fibrillation patients initiating oral anticoagulants. J Med Econ. 2019;22(12):1338-1350. doi:https://doi.org/10.1080/13696998.2019.1672698.

39.

Charlton

Adeboyeje

Barron

Grady

Shin

Redberg

. Length of hospitalization and mortality for bleeding during treatment with warfarin, dabigatran, or rivaroxaban. PLoS One. 2018;13(3):e0193912. doi:https://doi.org/10.1371/journal.pone.0193912.

40.

Schulman

Dowlatshahi

, et al. Healthcare resource utilization and costs among patients with direct oral anticoagulant or warfarin-related major bleeding. Thromb Res. 2019;182:12-19. doi:https://doi.org/10.1016/j.thromres.2019.07.026.

41.

Kubitza

Roth

Becka

, et al. Effect of hepatic impairment on the pharmacokinetics and pharmacodynamics of a single dose of rivaroxaban, an oral, direct factor Xa inhibitor. Br J Clin Pharmacol. 2013;76(1):89-98. doi:https://doi.org/10.1111/bcp.12054.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.48 MB