View Article PDF

Liver cirrhosis results from hepatic cell death, nodule formation and irreversible liver scarring.[[1]] With liver cirrhosis, there is a reduction in the synthesis of procoagulant and coagulant factors. The parallel decrease in both factors rebalances the coagulation system; however, this new balance is fragile and may be tilted towards bleeding or thrombosis when exposed to minimal stimuli.[[2]] Traditionally, cirrhotic patients needing anticoagulant therapy have been treated with warfarin, a vitamin K antagonist that inhibits the synthesis of vitamin K-dependent coagulation factors, including thrombin. Dabigatran is a novel oral anticoagulant (NOAC) that acts as a direct thrombin inhibitor, blocking only a single step in the coagulation pathway.[[3]] Dabigatran has become available in New Zealand and overseas in the last decade, and can be used as an alternative to warfarin in several thrombotic conditions, excluding mechanical heart valve replacement.[[4]] Advantages of dabigatran over warfarin are a faster onset of action, fewer drug and food interactions and less laboratory monitoring required.[[5]] These advantages make dabigatran more convenient to use in some patients. However, landmark NOAC trials have excluded patients with cirrhosis.[[4,6]]

In New Zealand, dabigatran was approved for use in 2008 and became fully funded on prescription in 2011.[[7]] Dabigatran requires minimal hepatic metabolism and is primarily eliminated through the kidneys.[[8]] Renal dysfunction may occur secondary to late-stage liver disease, leading to accumulation and increased drug concentrations.[[9]] Extended periods of time with high dabigatran concentrations may potentially lead to a higher bleeding risk in patients.[[10]] The pharmacokinetic changes of medication in cirrhosis were cited as a key reason for exclusion from landmark studies.[[11]] However, there is a paucity of data comparing the safety of dabigatran to warfarin in cirrhotic patients.[[12]] These data are necessary as there is a growing need for an alternative to warfarin. Our study's primary aim was to assess the rate of bleeding in patients with liver cirrhosis taking warfarin compared to those taking dabigatran.

Method

Study design and patient selection

This was a retrospective cohort study conducted in patients admitted to three tertiary care centres (district health boards/DHBs) in Auckland, New Zealand, from 1 January 2008 (the year that dabigatran became available in New Zealand) to 31 December 2020. Each DHB has one tertiary hospital centre. These are Auckland City Hospital (Auckland District Health Board, ADHB), with 1,124 beds, North Shore Hospital (Waitematā District Health Board, WDHB) with 600 beds, and Middlemore Hospital (Counties Manukau District Health Board, CMDHB) with 800 beds. The study has been approved by the ethics board of Auckland Health Research Ethics Committee (AH1163).

Within the three hospitals, any adult patient (≥18 years) diagnosed with liver disease or cirrhosis was identified using each hospital’s respective data information service and cirrhosis registries. Liver disease was defined by having one of the following International Classifications of Diseases 10[[th]] revision Australian Modified version (ICD-10 AM) discharge diagnosis codes (K701, K703, K704, K717, K721, K740-K746, K754, K758, K758, K760). Only the first hospital admission with a related diagnosis of any of the above codes was included. This admission was linked to community pharmacy dispensing records (Testsafe CareConnect) to identify patients with an anticoagulant dispensed from an outpatient pharmacy during the study period. This dispensing database, started in 2010, includes all records of prescribed medications dispensed to patients by community pharmacies. Patients were included if they had a dispensing record of an anticoagulant while having a confirmed cirrhosis diagnosis. Medication use for patient admissions prior to 2010 was identified using available electronic clinical notes. The index date was defined as the date of the first recorded dispensing of dabigatran or warfarin following cirrhosis diagnosis. The follow-up for each patient commenced from the index date until the occurrence of one of the following events: death, a bleeding event, liver transplant, diagnosis of advanced hepatocellular carcinoma, discontinuation or switch of anticoagulation therapy or the end of the study period (31 December 2020), whichever came first. Patients were excluded if they had a prescription record for other NOACs besides dabigatran on the index date. Patients with familial coagulopathy (such as haemophilia as diagnosed by haematologist) or advanced hepatocellular carcinoma (HCC) recorded any time before the index date were also excluded to reduce potential residual confounding effects.

Baseline characteristics such as age, ethnicity, sex, renal function (estimated glomerular filtration rate, eGFR), serum total bilirubin concentration and haemostasis status were collected. We also collected comorbidities such as polypharmacy, alcohol misuse, diabetes mellitus (type one and two), chronic kidney disease, hypertension, peripheral vascular disease (PVD), previous history of bleed, cerebrovascular disease, cancer, hepatocellular carcinoma and history of peptic ulcers. These characteristics were collected using electronic clinical notes and biochemistry data. We collected data on baseline medication use if they fell into the following categories. These were identified as potentially interacting medicines that may affect bleed risk: antidepressants, antibiotics, antiplatelets, non-steroidal anti-inflammatories and corticosteroids.

Cirrhotic diagnosis and severity

Liver cirrhosis was confirmed by written clinician diagnosis in the clinical notes along with supporting medical imaging evidence such as a FibroScan or computerised tomography (CT).

The severity of liver cirrhosis was defined by the Sodium Model of End-Stage Liver Disease (MELD-Na) score. Variables used to calculate MELD-Na were collected within 90 days of the index date.

Anticoagulation use

Computerised inpatient notes confirmed the date of initiation of anticoagulation, electronic discharge summaries and any outpatient pharmacy dispensing records. The first and last dispensing dates for each anticoagulant were collected along with clinical indication and dosage. If patients switched their anticoagulant (e.g. warfarin to dabigatran, vice versa, or other NOACs), they were censored from further follow-up.

Different criteria were used for warfarin and dabigatran to estimate the end of supply due to their different monitoring requirements and data availability. Dabigatran discontinuation was established if there was a gap between dispensing records of 90 days unless otherwise stated in clinic letters. The 90-day period was used as it is the maximum legal period of supply that a prescription for NOACs can be dispensed in New Zealand. The discontinuation of warfarin was identified if there was a gap between the supply of 56 days with absent international normalised ratio (INR) monitoring. A 56-day gap has been used in previous studies to indicate a lack of monitoring, and it is a period across which time in the therapeutic range is not interpolated.[[13]]

Outcome measures

Bleeding events were identified by reviewing the electronic clinical notes, discharge summaries, previous imaging reports and biochemistry results during their period of anticoagulation therapy. Events were identified as a primary or secondary diagnosis in the clinical notes. Prior published criteria were used, and the bleeds were characterised as “major” or “minor”.[[14]] At the time of the bleeding event, we collected MELD-Na scores, haemoglobin, platelet count and total bilirubin serum concentration. All bleeding events were independently assessed by three physician investigators (MH, CS, HW) using the Naranjo scale to evaluate the likelihood of the bleeding event being attributed to the anticoagulant. The Naranjo scale consists of 10 items with points being added or subtracted depending on the response, with a minimum score of negative four to a maximum score of 13.[[15]] The cause of bleed due to anticoagulation is considered definite if the score is nine or higher, probable if five to eight, possible if one to four, and doubtful if zero or fewer.[[15]]

Statistical analyses

Patient characteristics were presented as means and standard deviations for normally distributed variables and as medians and interquartile ranges (IQR) for variables with a skewed distribution. The number of patients (n) and percentages were used to represent categorical variables. Between-group comparisons were performed using the Chi-squared and Fisher’s exact test for categorical variables. The Mann–Whitney U test was used to compare non-normally distributed continuous variables such as renal function, serum bilirubin and platelet count. ANOVA or Student's t-test was used to compare normally distributed variables, serum haemoglobin and MELD-Na score. The incidence rate of bleeding was calculated as the number of patients with any bleeding event during follow-up divided by total follow-up time in person-year for both groups. The Kaplan–Meier (KM) method was used to compare bleeding events between warfarin and dabigatran cohorts, and the groups were compared using the log-rank test. Patients with missing data were excluded from the relevant analysis. Due to a significant difference in the length of follow-up between patients, the bleeding event data was truncated at five years for survival analysis. All statistical tests were two-tailed, and statistical significance was set at a p-value <0.05. Data analyses were performed using SPSS v27, and KM curve was generated using STATA v14 software packages.

Results

Initially, 4,518 patients admitted with liver disease were identified over the study period. A total of 4,153 patients were excluded as they were not dispensed any oral anticoagulation between 2008–2020. Of the 365 patients on anticoagulation during this time, 265 were excluded (203 were not cirrhotic, eight had advanced hepatocellular carcinoma, and 54 were excluded for reasons defined as “other”). Baseline MELD-Na score was unavailable for six patients. Of those meeting inclusion criteria, 52 took warfarin and 48 took dabigatran. See Figure 1 for cohort selection.

Study cohort characteristics

At baseline, we did not observe any statistically significant differences in sex, age or ethnicity distribution between warfarin and dabigatran cohorts (Table 1). Furthermore, both groups were similar regarding the aetiology of liver disease, baseline medications and comorbidities. The study arms differed in two areas: indication for anticoagulation and MELD-Na score. A higher proportion of patients with valvular AF were in the warfarin compared to the dabigatran cohort (42.3% vs 14.6%). In contrast, the percentage of patients with non-valvular AF was lower in warfarin than dabigatran cohort (36.5% vs 64.6%). Regarding the MELD-Na score, a higher proportion of patients in the warfarin cohort had MELD-Na score >20 than the dabigatran cohort (25% vs 8.3%). There were no differences in specific biochemical markers between the two treatment cohorts, including estimated GFR (p=0.177), total serum bilirubin (p=0.458), platelet count (p=0.583) and serum haemoglobin(p=0.092).

Bleeding risk

A total of 24 bleeding events occurred in total, six being major and 18 being minor. Half (12/24) of the bleeding events were identified to be gastrointestinal in nature. Of these 13% (3/24) of the events were related to the head, urological and epistaxis, and a quarter (6/24) of the events were unspecific, with related causes including hematomas, dental and vaginal.

The overall incidence rate of bleeding was 14.4 (95% CI, 8.8–23.5) and 9.1 (95% CI, 4.5–18.1) per 100 person-years in warfarin and dabigatran users, respectively. The incident rate ratio comparing dabigatran to warfarin was 0.63 (95% CI, 0.23–1.60), suggesting that patients taking dabigatran may have less risk of bleeding than patients taking warfarin. However, when compared using KM curves, this difference was not statically significant (p=0.25) (see Figure 2). Furthermore, no statistical significance was observed when KM analyses were undertaken to compare both MELD-Na score (p=0.07) and disease aetiology (p=0.13) to evaluate whether bleeding risk differed between warfarin and dabigatran users.

Change in biochemistry between bleeding event and baseline

When compared to baseline, the serum haemoglobin and the MELD-Na scores at the time of bleeding events showed changes that were of both statistical and clinical significance. The serum haemoglobin showed a significant decrease from mean (SD) 126.6 g/L (18.86) at baseline to 105.5 g/L (27.1) at the event (p<0.001), and the MELD-Na score had a significant increase from mean (SD) 16.8 (5.9) at baseline to 23.12 (8.1) at the event (p<0.001). However, serum bilirubin and platelet count did not alter significantly between bleeding event and baseline (Table 2).

Causality assessment of bleeding event and anticoagulation

Of the 24 bleed events, nine were considered possible and 15 were probable in likelihood of the bleed being attributed to the anticoagulant. Five were due to warfarin, and four were due to dabigatran in the nine events considered possibly related to the anticoagulant. In the 15 events deemed to be probable, eight were due to warfarin and seven were due to dabigatran.

View Tables and Figures.

Discussion

This paper presents the first data to our knowledge in the literature directly comparing the bleeding rates between the oral anticoagulants warfarin and dabigatran in patients with liver cirrhosis. There has been growing interest in this field since the emergence of NOACs, yet data on bleeding risk in the context of liver cirrhosis is scarce. Patients diagnosed with liver cirrhosis have traditionally been excluded from landmark studies, but more real-world retrospective studies are emerging.[[16]] In our study, we demonstrated that the bleeding risk of dabigatran did not differ compared to warfarin. In our initial cohorts, warfarin tended to be used in patients with higher MELD-Na scores and for the indication of valvular AF, and this may increase the bleeding risk of warfarin. These differences are likely a reflection of prescribing practices, whereby dabigatran is contraindicated and off-license in both valvular AF, moderate and severe hepatic impairment (CTP categories B and C).[[17]]

Compared to other studies of similar design, our bleed rates are similar to a study recently published by Mort et al, which focusses specifically on NOACs and bleeding risk. Their study’s overall bleed rate was 21%, and ours is 24%.[[12]] Although Mort et al. did not compare their cohort directly with warfarin users, they state that their rate of bleed for NOACs was comparable to published rates of bleed for warfarin users.[[12]] Another larger retrospective cohort study was conducted in Taiwan, using national health administrative data. Over 2,428 non-valvular AF patients with cirrhosis were included in this study. The risk of major bleed (HR 0.51, 95% CI 0.32–0.74) and major gastrointestinal bleed (HR 0.51, 95% CI 0.32-0.79) was lower in NOACs users compared to Warfarin users.[[18]] Compared to our study, the study differs significantly in methodology and findings. However, Lee et al.'s Taiwan study only included Taiwanese patients and may not be applicable to other ethnicities. In addition, our study also collected multiple clinically important biomarkers that are not often available in the administrative database. Asian populations have been found to have a higher risk of bleeding when taking warfarin compared to non-Asian populations, and previous studies indicate that NOACs may be a safer option in Asian vs non-Asian populations.[[19]]

Our study had several strengths, including using the Naranjo scale. We were able to standardise the assessment of bleeding events in patients with cirrhosis on dabigatran and warfarin. By using three independent physicians to ascertain the Naranjo score and compare scores, we ensured a more robust assessment of each bleeding event being related to the anticoagulant of choice. The study had access to a wide range of clinical data across the three main hospitals in Auckland, New Zealand, by using computerised notes and paper notes and laboratory results. Thus, for each patient, we were able to assess their liver disease status comprehensively and to only include those with a robust diagnosis of liver cirrhosis. We also collected several important variables that may influence bleeding risk, e.g., baseline MELD-Na score, full blood count and renal function. A weakness for our study is its low sample size, being an exploratory study. The study lacked enough input from other ethnic groups to make any meaningful comparison in outcomes, with the majority of patients being NZ European.

Further investigation into the ethnic differences in bleeding rates on warfarin and NOACs in New Zealand’s population, particularly in Māori and Pasifika populations, would be an area for further research to potentially change clinical practice. We also opted to combine both minor and major bleed to all bleed for our outcome of interest due to the low number of bleeding events in total. Other limitations include that the study was retrospective, so we cannot control for all confounders at baseline. The data is also limited as it was based in one region—in Auckland, New Zealand—and the population characteristics, while representing a broad range of ethnicities and three different DHB sites, may not represent populations elsewhere. However, Auckland represents around one-third of New Zealand’s total population, and it is likely that New Zealand data is not sufficient for this study.[[20]] Our study was exploratory as we did not have any existing data on the rate of oral anticoagulated patients with liver cirrhosis, thus we did not have a predetermined sample size. It is evident from our research that the use of oral anticoagulation is a rare occurrence in patients with liver cirrhosis in New Zealand (prevalence of less than 0.1%). We did not observe a statistically significant association between warfarin vs dabigatran use and subsequent risk of bleeding but, due to the modest sample size, this study is likely underpowered to detect any such an association, if present. An adequately powered study with comparable methodology will likely need to be conducted in countries with larger centres such as those in Asia, the United States and the United Kingdom.

In conclusion, our study found no statistically significant differences in the bleeding rate in cirrhotic patients treated with warfarin versus those treated with dabigatran. Our results suggest dabigatran may be as safe to use as warfarin in patients with cirrhosis.

Summary

Abstract

Aim

The safety of dabigatran is poorly studied in patients with liver cirrhosis and has rarely been compared to warfarin in terms of bleeding risks.

Method

We undertook a retrospective cohort study across three tertiary centres in Auckland, New Zealand, between 2008 to 2020. Adults 18 years and over and those with a clinically confirmed diagnosis of cirrhosis were included. Data collected included demographic data and clinical characteristics, baseline medication and comorbidities. The primary outcome measure was the incidence of any bleeding event that resulted in hospital admission.

Results

Overall, 100 patients were included in this study. A total of 52 patients took warfarin, and 48 took dabigatran. Baseline characteristics for both groups were generally similar. The incidence rate of bleeds for patients taking warfarin was 14.4 per 100 person-years (95% CI, 8.8–23.5) compared to 9.1 per 100 person-years (95% CI, 4.5–18.1) for patients taking dabigatran. The incidence rate ratio comparing dabigatran to warfarin was 0.63 (95% CI, 0.23–1.60), p=0.25.

Conclusion

Our study found that patients on dabigatran may have a lower bleeding risk than patients taking warfarin, but this was not statistically significant.

Author Information

Ms Oriana Munevar Aquite: School of Pharmacy, Faculty of Medical and Health Science, The University of Auckland, New Zealand. Dr Michael Hayes: Department of Gastroenterology, Waitematā District Health Board, New Zealand. Dr Kebede Beyene: School of Pharmacy, Faculty of Medical and Health Science, The University of Auckland, New Zealand. Dr Amy Hai Yan Chan: School of Pharmacy, Faculty of Medical and Health Science, The University of Auckland, New Zealand. Dr Cameron Schauer: Department of Gastroenterology, Waitematā District Health Board, New Zealand. Dr Henry Wei: Department of Gastroenterology, Auckland District Health Board, New Zealand. Mr Jiayi Gong: New Zealand Liver Research Unit, Auckland District Health Board, New Zealand.

Acknowledgements

New Zealand Pharmacy Education and Research Foundation Summer Studentship, New Zealand Liver Transplant Unit and New Zealand Liver Research Trust. Professor Edward Gane and Miss Alice Yu for providing research support for this study.

Correspondence

Jay Gong: M&HS Building 502, 85 Park Rd, Grafton, Auckland, 1023, New Zealand.

Correspondence Email

Jay.gong@auckland.ac.nz

Competing Interests

Authors declarations of personal interest: OMA received funding as part of her summer studentship from the New Zealand Pharmacy Education and Research Foundation for this piece of work. MH declares no conflicts of interest. KB declares no conflicts of interest. AHYC reports consultancy fees from Spoonful of Sugar Ltd, grants from Health Research Council, Innovate UK, A+ charitable trust (Auckland District Health Board), Maurice and Phyllis Paykel trust, Universitas 21, NZPERF, Auckland Academic Health Alliance, Asthma UK, The University of Auckland; grants and consultancy fees from Janssen-Cilag; and is the recipient of the Robert Irwin Fellowship and Senior Research Fellowship from the Auckland Medical Research Foundation. None related to this work. CS declares no conflict of interest. HW no conflict of interest. JG research grants from the New Zealand Liver Research Trust for this piece of work, Health Research Council, Auckland Academic Health Alliance and A+ charitable trust (Auckland District Health Board). None are commercial entities.

1) McCormick PA, Jalan R. Hepatic Cirrhosis. Sherlock's Diseases of the Liver and Biliary System. 2018:107-26.

2) Intagliata NM, Caldwell SH. Coagulation in Cirrhosis. Sherlock's Diseases of the Liver and Biliary System. 2018:53-61.

3) Evans NS. Direct Oral Anticoagulants. In: Lau JF, Barnes GD, Streiff MB, eds. Anticoagulation Therapy. Cham: Springer International Publishing. 2018:87-103.

4) Connolly SJ, Ezekowitz MD, Yusuf S, et al. Dabigatran versus Warfarin in Patients with Atrial Fibrillation. N Engl J Med. 2009;361(12):1139-51. doi: 10.1056/NEJMoa0905561.

5) Mekaj YH, Mekaj AY, Duci SB, et al. New oral anticoagulants: their advantages and disadvantages compared with vitamin K antagonists in the prevention and treatment of patients with thromboembolic events. Ther Clin Risk Manag. 2015;11:967-77. doi: 10.2147/tcrm.S84210 [published online first: 2015/07/08].

6) Schulman S, Kearon C, Kakkar AK, et al. Dabigatran versus Warfarin in the Treatment of Acute Venous Thromboembolism. N Engl J Med. 2009;361(24):2342-52. doi: 10.1056/NEJMoa0906598.

7) Pharmac. Application Tracker 2021 [Available from: https://connect.pharmac.govt.nz/apptracker/s/. Accessed 18 February 2021].

8) Fawzy AM, Lip GYH. Pharmacokinetics and pharmacodynamics of oral anticoagulants used in atrial fibrillation. Expert Opin Drug Metab Toxicol. 2019;15(5):381-98. doi: 10.1080/17425255.2019.1604686.

9) Francoz C, Nadim MK, Baron A, et al. Glomerular filtration rate equations for liver-kidney transplantation in patients with cirrhosis: Validation of current recommendations. Hepatology. 2014;59(4):1514-21. doi: https://doi.org/10.1002/hep.26704.

10) Qamar A, Vaduganathan M, Greenberger NJ, et al. Oral Anticoagulation in Patients With Liver Disease. J Am Coll Cardiol. 2018;71(19):2162-75. doi: https://doi.org/10.1016/j.jacc.2018.03.023.

11) Steffel J, Verhamme P, Potpara TS, et al. The 2018 European Heart Rhythm Association Practical Guide on the use of non-vitamin K antagonist oral anticoagulants in patients with atrial fibrillation. Eur Heart J. 2018;39(16):1330-93. doi: 10.1093/eurheartj/ehy136.

12) Mort JF, Davis JPE, Mahoro G, et al. Rates of Bleeding and Discontinuation of Direct Oral Anticoagulants in Patients With Decompensated Cirrhosis. Clin Gastroenterol Hepatol. 2021;19(7):1436-42. doi: 10.1016/j.cgh.2020.08.007 [published online first: 2020/08/11].

13) Rose AJ, Miller DR, Ozonoff A, et al. Gaps in monitoring during oral anticoagulation: insights into care transitions, monitoring barriers, and medication nonadherence. Chest. 2013;143(3):751-57. doi: 10.1378/chest.12-1119 [published online first: 2012/11/29].

14) Ageno W, Riva N, Schulman S, et al. Long-term Clinical Outcomes of Splanchnic Vein Thrombosis: Results of an International Registry. JAMA Intern Med. 2015;175(9):1474-80. doi: 10.1001/jamainternmed.2015.3184.

15) Livertox. Adverse Drug Reaction Probability Scale (Naranjo) in Drug Induced Liver Injury. 2012 [updated 2019 May 4]. Available from: https://www.ncbi.nlm.nih.gov/books/NBK548069/. Accessed 31 August 2021.

16) Elhosseiny S, Al Moussawi H, Chalhoub JM, et al. Direct Oral Anticoagulants in Cirrhotic Patients: Current Evidence and Clinical Observations. Can J Gastroenterol Hepatol. 2019;2019:4383269. doi: 10.1155/2019/4383269.

17) New Zealand Datasheet for Pradaxa: Medsafe New Zealand; 2020 [cited 25 January 2022]. Available from: https://www.medsafe.govt.nz/profs/datasheet/p/Pradaxacap.pdf. Accessed 25 January 2022.

18) Lee HF, Chan YH, Chang SH, 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: 10.1161/jaha.118.011112 [published online first: 2019/03/06].

19) Wang CL, Wu VC, Kuo CF, et al. Efficacy and Safety of Non-Vitamin K Antagonist Oral Anticoagulants in Atrial Fibrillation Patients With Impaired Liver Function: A Retrospective Cohort Study. J Am Heart Assoc. 2018;7(15):e009263. doi: 10.1161/jaha.118.009263 [published online first: 2018/10/30].

20) Auckland's Population Auckland: Auckland Council; 2018 [cited 25 January 2022]. Available from: https://www.aucklandcouncil.govt.nz/. Accessed 25 January 2022.

For the PDF of this article,
contact nzmj@nzma.org.nz

View Article PDF

Liver cirrhosis results from hepatic cell death, nodule formation and irreversible liver scarring.[[1]] With liver cirrhosis, there is a reduction in the synthesis of procoagulant and coagulant factors. The parallel decrease in both factors rebalances the coagulation system; however, this new balance is fragile and may be tilted towards bleeding or thrombosis when exposed to minimal stimuli.[[2]] Traditionally, cirrhotic patients needing anticoagulant therapy have been treated with warfarin, a vitamin K antagonist that inhibits the synthesis of vitamin K-dependent coagulation factors, including thrombin. Dabigatran is a novel oral anticoagulant (NOAC) that acts as a direct thrombin inhibitor, blocking only a single step in the coagulation pathway.[[3]] Dabigatran has become available in New Zealand and overseas in the last decade, and can be used as an alternative to warfarin in several thrombotic conditions, excluding mechanical heart valve replacement.[[4]] Advantages of dabigatran over warfarin are a faster onset of action, fewer drug and food interactions and less laboratory monitoring required.[[5]] These advantages make dabigatran more convenient to use in some patients. However, landmark NOAC trials have excluded patients with cirrhosis.[[4,6]]

In New Zealand, dabigatran was approved for use in 2008 and became fully funded on prescription in 2011.[[7]] Dabigatran requires minimal hepatic metabolism and is primarily eliminated through the kidneys.[[8]] Renal dysfunction may occur secondary to late-stage liver disease, leading to accumulation and increased drug concentrations.[[9]] Extended periods of time with high dabigatran concentrations may potentially lead to a higher bleeding risk in patients.[[10]] The pharmacokinetic changes of medication in cirrhosis were cited as a key reason for exclusion from landmark studies.[[11]] However, there is a paucity of data comparing the safety of dabigatran to warfarin in cirrhotic patients.[[12]] These data are necessary as there is a growing need for an alternative to warfarin. Our study's primary aim was to assess the rate of bleeding in patients with liver cirrhosis taking warfarin compared to those taking dabigatran.

Method

Study design and patient selection

This was a retrospective cohort study conducted in patients admitted to three tertiary care centres (district health boards/DHBs) in Auckland, New Zealand, from 1 January 2008 (the year that dabigatran became available in New Zealand) to 31 December 2020. Each DHB has one tertiary hospital centre. These are Auckland City Hospital (Auckland District Health Board, ADHB), with 1,124 beds, North Shore Hospital (Waitematā District Health Board, WDHB) with 600 beds, and Middlemore Hospital (Counties Manukau District Health Board, CMDHB) with 800 beds. The study has been approved by the ethics board of Auckland Health Research Ethics Committee (AH1163).

Within the three hospitals, any adult patient (≥18 years) diagnosed with liver disease or cirrhosis was identified using each hospital’s respective data information service and cirrhosis registries. Liver disease was defined by having one of the following International Classifications of Diseases 10[[th]] revision Australian Modified version (ICD-10 AM) discharge diagnosis codes (K701, K703, K704, K717, K721, K740-K746, K754, K758, K758, K760). Only the first hospital admission with a related diagnosis of any of the above codes was included. This admission was linked to community pharmacy dispensing records (Testsafe CareConnect) to identify patients with an anticoagulant dispensed from an outpatient pharmacy during the study period. This dispensing database, started in 2010, includes all records of prescribed medications dispensed to patients by community pharmacies. Patients were included if they had a dispensing record of an anticoagulant while having a confirmed cirrhosis diagnosis. Medication use for patient admissions prior to 2010 was identified using available electronic clinical notes. The index date was defined as the date of the first recorded dispensing of dabigatran or warfarin following cirrhosis diagnosis. The follow-up for each patient commenced from the index date until the occurrence of one of the following events: death, a bleeding event, liver transplant, diagnosis of advanced hepatocellular carcinoma, discontinuation or switch of anticoagulation therapy or the end of the study period (31 December 2020), whichever came first. Patients were excluded if they had a prescription record for other NOACs besides dabigatran on the index date. Patients with familial coagulopathy (such as haemophilia as diagnosed by haematologist) or advanced hepatocellular carcinoma (HCC) recorded any time before the index date were also excluded to reduce potential residual confounding effects.

Baseline characteristics such as age, ethnicity, sex, renal function (estimated glomerular filtration rate, eGFR), serum total bilirubin concentration and haemostasis status were collected. We also collected comorbidities such as polypharmacy, alcohol misuse, diabetes mellitus (type one and two), chronic kidney disease, hypertension, peripheral vascular disease (PVD), previous history of bleed, cerebrovascular disease, cancer, hepatocellular carcinoma and history of peptic ulcers. These characteristics were collected using electronic clinical notes and biochemistry data. We collected data on baseline medication use if they fell into the following categories. These were identified as potentially interacting medicines that may affect bleed risk: antidepressants, antibiotics, antiplatelets, non-steroidal anti-inflammatories and corticosteroids.

Cirrhotic diagnosis and severity

Liver cirrhosis was confirmed by written clinician diagnosis in the clinical notes along with supporting medical imaging evidence such as a FibroScan or computerised tomography (CT).

The severity of liver cirrhosis was defined by the Sodium Model of End-Stage Liver Disease (MELD-Na) score. Variables used to calculate MELD-Na were collected within 90 days of the index date.

Anticoagulation use

Computerised inpatient notes confirmed the date of initiation of anticoagulation, electronic discharge summaries and any outpatient pharmacy dispensing records. The first and last dispensing dates for each anticoagulant were collected along with clinical indication and dosage. If patients switched their anticoagulant (e.g. warfarin to dabigatran, vice versa, or other NOACs), they were censored from further follow-up.

Different criteria were used for warfarin and dabigatran to estimate the end of supply due to their different monitoring requirements and data availability. Dabigatran discontinuation was established if there was a gap between dispensing records of 90 days unless otherwise stated in clinic letters. The 90-day period was used as it is the maximum legal period of supply that a prescription for NOACs can be dispensed in New Zealand. The discontinuation of warfarin was identified if there was a gap between the supply of 56 days with absent international normalised ratio (INR) monitoring. A 56-day gap has been used in previous studies to indicate a lack of monitoring, and it is a period across which time in the therapeutic range is not interpolated.[[13]]

Outcome measures

Bleeding events were identified by reviewing the electronic clinical notes, discharge summaries, previous imaging reports and biochemistry results during their period of anticoagulation therapy. Events were identified as a primary or secondary diagnosis in the clinical notes. Prior published criteria were used, and the bleeds were characterised as “major” or “minor”.[[14]] At the time of the bleeding event, we collected MELD-Na scores, haemoglobin, platelet count and total bilirubin serum concentration. All bleeding events were independently assessed by three physician investigators (MH, CS, HW) using the Naranjo scale to evaluate the likelihood of the bleeding event being attributed to the anticoagulant. The Naranjo scale consists of 10 items with points being added or subtracted depending on the response, with a minimum score of negative four to a maximum score of 13.[[15]] The cause of bleed due to anticoagulation is considered definite if the score is nine or higher, probable if five to eight, possible if one to four, and doubtful if zero or fewer.[[15]]

Statistical analyses

Patient characteristics were presented as means and standard deviations for normally distributed variables and as medians and interquartile ranges (IQR) for variables with a skewed distribution. The number of patients (n) and percentages were used to represent categorical variables. Between-group comparisons were performed using the Chi-squared and Fisher’s exact test for categorical variables. The Mann–Whitney U test was used to compare non-normally distributed continuous variables such as renal function, serum bilirubin and platelet count. ANOVA or Student's t-test was used to compare normally distributed variables, serum haemoglobin and MELD-Na score. The incidence rate of bleeding was calculated as the number of patients with any bleeding event during follow-up divided by total follow-up time in person-year for both groups. The Kaplan–Meier (KM) method was used to compare bleeding events between warfarin and dabigatran cohorts, and the groups were compared using the log-rank test. Patients with missing data were excluded from the relevant analysis. Due to a significant difference in the length of follow-up between patients, the bleeding event data was truncated at five years for survival analysis. All statistical tests were two-tailed, and statistical significance was set at a p-value <0.05. Data analyses were performed using SPSS v27, and KM curve was generated using STATA v14 software packages.

Results

Initially, 4,518 patients admitted with liver disease were identified over the study period. A total of 4,153 patients were excluded as they were not dispensed any oral anticoagulation between 2008–2020. Of the 365 patients on anticoagulation during this time, 265 were excluded (203 were not cirrhotic, eight had advanced hepatocellular carcinoma, and 54 were excluded for reasons defined as “other”). Baseline MELD-Na score was unavailable for six patients. Of those meeting inclusion criteria, 52 took warfarin and 48 took dabigatran. See Figure 1 for cohort selection.

Study cohort characteristics

At baseline, we did not observe any statistically significant differences in sex, age or ethnicity distribution between warfarin and dabigatran cohorts (Table 1). Furthermore, both groups were similar regarding the aetiology of liver disease, baseline medications and comorbidities. The study arms differed in two areas: indication for anticoagulation and MELD-Na score. A higher proportion of patients with valvular AF were in the warfarin compared to the dabigatran cohort (42.3% vs 14.6%). In contrast, the percentage of patients with non-valvular AF was lower in warfarin than dabigatran cohort (36.5% vs 64.6%). Regarding the MELD-Na score, a higher proportion of patients in the warfarin cohort had MELD-Na score >20 than the dabigatran cohort (25% vs 8.3%). There were no differences in specific biochemical markers between the two treatment cohorts, including estimated GFR (p=0.177), total serum bilirubin (p=0.458), platelet count (p=0.583) and serum haemoglobin(p=0.092).

Bleeding risk

A total of 24 bleeding events occurred in total, six being major and 18 being minor. Half (12/24) of the bleeding events were identified to be gastrointestinal in nature. Of these 13% (3/24) of the events were related to the head, urological and epistaxis, and a quarter (6/24) of the events were unspecific, with related causes including hematomas, dental and vaginal.

The overall incidence rate of bleeding was 14.4 (95% CI, 8.8–23.5) and 9.1 (95% CI, 4.5–18.1) per 100 person-years in warfarin and dabigatran users, respectively. The incident rate ratio comparing dabigatran to warfarin was 0.63 (95% CI, 0.23–1.60), suggesting that patients taking dabigatran may have less risk of bleeding than patients taking warfarin. However, when compared using KM curves, this difference was not statically significant (p=0.25) (see Figure 2). Furthermore, no statistical significance was observed when KM analyses were undertaken to compare both MELD-Na score (p=0.07) and disease aetiology (p=0.13) to evaluate whether bleeding risk differed between warfarin and dabigatran users.

Change in biochemistry between bleeding event and baseline

When compared to baseline, the serum haemoglobin and the MELD-Na scores at the time of bleeding events showed changes that were of both statistical and clinical significance. The serum haemoglobin showed a significant decrease from mean (SD) 126.6 g/L (18.86) at baseline to 105.5 g/L (27.1) at the event (p<0.001), and the MELD-Na score had a significant increase from mean (SD) 16.8 (5.9) at baseline to 23.12 (8.1) at the event (p<0.001). However, serum bilirubin and platelet count did not alter significantly between bleeding event and baseline (Table 2).

Causality assessment of bleeding event and anticoagulation

Of the 24 bleed events, nine were considered possible and 15 were probable in likelihood of the bleed being attributed to the anticoagulant. Five were due to warfarin, and four were due to dabigatran in the nine events considered possibly related to the anticoagulant. In the 15 events deemed to be probable, eight were due to warfarin and seven were due to dabigatran.

View Tables and Figures.

Discussion

This paper presents the first data to our knowledge in the literature directly comparing the bleeding rates between the oral anticoagulants warfarin and dabigatran in patients with liver cirrhosis. There has been growing interest in this field since the emergence of NOACs, yet data on bleeding risk in the context of liver cirrhosis is scarce. Patients diagnosed with liver cirrhosis have traditionally been excluded from landmark studies, but more real-world retrospective studies are emerging.[[16]] In our study, we demonstrated that the bleeding risk of dabigatran did not differ compared to warfarin. In our initial cohorts, warfarin tended to be used in patients with higher MELD-Na scores and for the indication of valvular AF, and this may increase the bleeding risk of warfarin. These differences are likely a reflection of prescribing practices, whereby dabigatran is contraindicated and off-license in both valvular AF, moderate and severe hepatic impairment (CTP categories B and C).[[17]]

Compared to other studies of similar design, our bleed rates are similar to a study recently published by Mort et al, which focusses specifically on NOACs and bleeding risk. Their study’s overall bleed rate was 21%, and ours is 24%.[[12]] Although Mort et al. did not compare their cohort directly with warfarin users, they state that their rate of bleed for NOACs was comparable to published rates of bleed for warfarin users.[[12]] Another larger retrospective cohort study was conducted in Taiwan, using national health administrative data. Over 2,428 non-valvular AF patients with cirrhosis were included in this study. The risk of major bleed (HR 0.51, 95% CI 0.32–0.74) and major gastrointestinal bleed (HR 0.51, 95% CI 0.32-0.79) was lower in NOACs users compared to Warfarin users.[[18]] Compared to our study, the study differs significantly in methodology and findings. However, Lee et al.'s Taiwan study only included Taiwanese patients and may not be applicable to other ethnicities. In addition, our study also collected multiple clinically important biomarkers that are not often available in the administrative database. Asian populations have been found to have a higher risk of bleeding when taking warfarin compared to non-Asian populations, and previous studies indicate that NOACs may be a safer option in Asian vs non-Asian populations.[[19]]

Our study had several strengths, including using the Naranjo scale. We were able to standardise the assessment of bleeding events in patients with cirrhosis on dabigatran and warfarin. By using three independent physicians to ascertain the Naranjo score and compare scores, we ensured a more robust assessment of each bleeding event being related to the anticoagulant of choice. The study had access to a wide range of clinical data across the three main hospitals in Auckland, New Zealand, by using computerised notes and paper notes and laboratory results. Thus, for each patient, we were able to assess their liver disease status comprehensively and to only include those with a robust diagnosis of liver cirrhosis. We also collected several important variables that may influence bleeding risk, e.g., baseline MELD-Na score, full blood count and renal function. A weakness for our study is its low sample size, being an exploratory study. The study lacked enough input from other ethnic groups to make any meaningful comparison in outcomes, with the majority of patients being NZ European.

Further investigation into the ethnic differences in bleeding rates on warfarin and NOACs in New Zealand’s population, particularly in Māori and Pasifika populations, would be an area for further research to potentially change clinical practice. We also opted to combine both minor and major bleed to all bleed for our outcome of interest due to the low number of bleeding events in total. Other limitations include that the study was retrospective, so we cannot control for all confounders at baseline. The data is also limited as it was based in one region—in Auckland, New Zealand—and the population characteristics, while representing a broad range of ethnicities and three different DHB sites, may not represent populations elsewhere. However, Auckland represents around one-third of New Zealand’s total population, and it is likely that New Zealand data is not sufficient for this study.[[20]] Our study was exploratory as we did not have any existing data on the rate of oral anticoagulated patients with liver cirrhosis, thus we did not have a predetermined sample size. It is evident from our research that the use of oral anticoagulation is a rare occurrence in patients with liver cirrhosis in New Zealand (prevalence of less than 0.1%). We did not observe a statistically significant association between warfarin vs dabigatran use and subsequent risk of bleeding but, due to the modest sample size, this study is likely underpowered to detect any such an association, if present. An adequately powered study with comparable methodology will likely need to be conducted in countries with larger centres such as those in Asia, the United States and the United Kingdom.

In conclusion, our study found no statistically significant differences in the bleeding rate in cirrhotic patients treated with warfarin versus those treated with dabigatran. Our results suggest dabigatran may be as safe to use as warfarin in patients with cirrhosis.

Summary

Abstract

Aim

The safety of dabigatran is poorly studied in patients with liver cirrhosis and has rarely been compared to warfarin in terms of bleeding risks.

Method

We undertook a retrospective cohort study across three tertiary centres in Auckland, New Zealand, between 2008 to 2020. Adults 18 years and over and those with a clinically confirmed diagnosis of cirrhosis were included. Data collected included demographic data and clinical characteristics, baseline medication and comorbidities. The primary outcome measure was the incidence of any bleeding event that resulted in hospital admission.

Results

Overall, 100 patients were included in this study. A total of 52 patients took warfarin, and 48 took dabigatran. Baseline characteristics for both groups were generally similar. The incidence rate of bleeds for patients taking warfarin was 14.4 per 100 person-years (95% CI, 8.8–23.5) compared to 9.1 per 100 person-years (95% CI, 4.5–18.1) for patients taking dabigatran. The incidence rate ratio comparing dabigatran to warfarin was 0.63 (95% CI, 0.23–1.60), p=0.25.

Conclusion

Our study found that patients on dabigatran may have a lower bleeding risk than patients taking warfarin, but this was not statistically significant.

Author Information

Ms Oriana Munevar Aquite: School of Pharmacy, Faculty of Medical and Health Science, The University of Auckland, New Zealand. Dr Michael Hayes: Department of Gastroenterology, Waitematā District Health Board, New Zealand. Dr Kebede Beyene: School of Pharmacy, Faculty of Medical and Health Science, The University of Auckland, New Zealand. Dr Amy Hai Yan Chan: School of Pharmacy, Faculty of Medical and Health Science, The University of Auckland, New Zealand. Dr Cameron Schauer: Department of Gastroenterology, Waitematā District Health Board, New Zealand. Dr Henry Wei: Department of Gastroenterology, Auckland District Health Board, New Zealand. Mr Jiayi Gong: New Zealand Liver Research Unit, Auckland District Health Board, New Zealand.

Acknowledgements

New Zealand Pharmacy Education and Research Foundation Summer Studentship, New Zealand Liver Transplant Unit and New Zealand Liver Research Trust. Professor Edward Gane and Miss Alice Yu for providing research support for this study.

Correspondence

Jay Gong: M&HS Building 502, 85 Park Rd, Grafton, Auckland, 1023, New Zealand.

Correspondence Email

Jay.gong@auckland.ac.nz

Competing Interests

Authors declarations of personal interest: OMA received funding as part of her summer studentship from the New Zealand Pharmacy Education and Research Foundation for this piece of work. MH declares no conflicts of interest. KB declares no conflicts of interest. AHYC reports consultancy fees from Spoonful of Sugar Ltd, grants from Health Research Council, Innovate UK, A+ charitable trust (Auckland District Health Board), Maurice and Phyllis Paykel trust, Universitas 21, NZPERF, Auckland Academic Health Alliance, Asthma UK, The University of Auckland; grants and consultancy fees from Janssen-Cilag; and is the recipient of the Robert Irwin Fellowship and Senior Research Fellowship from the Auckland Medical Research Foundation. None related to this work. CS declares no conflict of interest. HW no conflict of interest. JG research grants from the New Zealand Liver Research Trust for this piece of work, Health Research Council, Auckland Academic Health Alliance and A+ charitable trust (Auckland District Health Board). None are commercial entities.

1) McCormick PA, Jalan R. Hepatic Cirrhosis. Sherlock's Diseases of the Liver and Biliary System. 2018:107-26.

2) Intagliata NM, Caldwell SH. Coagulation in Cirrhosis. Sherlock's Diseases of the Liver and Biliary System. 2018:53-61.

3) Evans NS. Direct Oral Anticoagulants. In: Lau JF, Barnes GD, Streiff MB, eds. Anticoagulation Therapy. Cham: Springer International Publishing. 2018:87-103.

4) Connolly SJ, Ezekowitz MD, Yusuf S, et al. Dabigatran versus Warfarin in Patients with Atrial Fibrillation. N Engl J Med. 2009;361(12):1139-51. doi: 10.1056/NEJMoa0905561.

5) Mekaj YH, Mekaj AY, Duci SB, et al. New oral anticoagulants: their advantages and disadvantages compared with vitamin K antagonists in the prevention and treatment of patients with thromboembolic events. Ther Clin Risk Manag. 2015;11:967-77. doi: 10.2147/tcrm.S84210 [published online first: 2015/07/08].

6) Schulman S, Kearon C, Kakkar AK, et al. Dabigatran versus Warfarin in the Treatment of Acute Venous Thromboembolism. N Engl J Med. 2009;361(24):2342-52. doi: 10.1056/NEJMoa0906598.

7) Pharmac. Application Tracker 2021 [Available from: https://connect.pharmac.govt.nz/apptracker/s/. Accessed 18 February 2021].

8) Fawzy AM, Lip GYH. Pharmacokinetics and pharmacodynamics of oral anticoagulants used in atrial fibrillation. Expert Opin Drug Metab Toxicol. 2019;15(5):381-98. doi: 10.1080/17425255.2019.1604686.

9) Francoz C, Nadim MK, Baron A, et al. Glomerular filtration rate equations for liver-kidney transplantation in patients with cirrhosis: Validation of current recommendations. Hepatology. 2014;59(4):1514-21. doi: https://doi.org/10.1002/hep.26704.

10) Qamar A, Vaduganathan M, Greenberger NJ, et al. Oral Anticoagulation in Patients With Liver Disease. J Am Coll Cardiol. 2018;71(19):2162-75. doi: https://doi.org/10.1016/j.jacc.2018.03.023.

11) Steffel J, Verhamme P, Potpara TS, et al. The 2018 European Heart Rhythm Association Practical Guide on the use of non-vitamin K antagonist oral anticoagulants in patients with atrial fibrillation. Eur Heart J. 2018;39(16):1330-93. doi: 10.1093/eurheartj/ehy136.

12) Mort JF, Davis JPE, Mahoro G, et al. Rates of Bleeding and Discontinuation of Direct Oral Anticoagulants in Patients With Decompensated Cirrhosis. Clin Gastroenterol Hepatol. 2021;19(7):1436-42. doi: 10.1016/j.cgh.2020.08.007 [published online first: 2020/08/11].

13) Rose AJ, Miller DR, Ozonoff A, et al. Gaps in monitoring during oral anticoagulation: insights into care transitions, monitoring barriers, and medication nonadherence. Chest. 2013;143(3):751-57. doi: 10.1378/chest.12-1119 [published online first: 2012/11/29].

14) Ageno W, Riva N, Schulman S, et al. Long-term Clinical Outcomes of Splanchnic Vein Thrombosis: Results of an International Registry. JAMA Intern Med. 2015;175(9):1474-80. doi: 10.1001/jamainternmed.2015.3184.

15) Livertox. Adverse Drug Reaction Probability Scale (Naranjo) in Drug Induced Liver Injury. 2012 [updated 2019 May 4]. Available from: https://www.ncbi.nlm.nih.gov/books/NBK548069/. Accessed 31 August 2021.

16) Elhosseiny S, Al Moussawi H, Chalhoub JM, et al. Direct Oral Anticoagulants in Cirrhotic Patients: Current Evidence and Clinical Observations. Can J Gastroenterol Hepatol. 2019;2019:4383269. doi: 10.1155/2019/4383269.

17) New Zealand Datasheet for Pradaxa: Medsafe New Zealand; 2020 [cited 25 January 2022]. Available from: https://www.medsafe.govt.nz/profs/datasheet/p/Pradaxacap.pdf. Accessed 25 January 2022.

18) Lee HF, Chan YH, Chang SH, 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: 10.1161/jaha.118.011112 [published online first: 2019/03/06].

19) Wang CL, Wu VC, Kuo CF, et al. Efficacy and Safety of Non-Vitamin K Antagonist Oral Anticoagulants in Atrial Fibrillation Patients With Impaired Liver Function: A Retrospective Cohort Study. J Am Heart Assoc. 2018;7(15):e009263. doi: 10.1161/jaha.118.009263 [published online first: 2018/10/30].

20) Auckland's Population Auckland: Auckland Council; 2018 [cited 25 January 2022]. Available from: https://www.aucklandcouncil.govt.nz/. Accessed 25 January 2022.

For the PDF of this article,
contact nzmj@nzma.org.nz

View Article PDF

Liver cirrhosis results from hepatic cell death, nodule formation and irreversible liver scarring.[[1]] With liver cirrhosis, there is a reduction in the synthesis of procoagulant and coagulant factors. The parallel decrease in both factors rebalances the coagulation system; however, this new balance is fragile and may be tilted towards bleeding or thrombosis when exposed to minimal stimuli.[[2]] Traditionally, cirrhotic patients needing anticoagulant therapy have been treated with warfarin, a vitamin K antagonist that inhibits the synthesis of vitamin K-dependent coagulation factors, including thrombin. Dabigatran is a novel oral anticoagulant (NOAC) that acts as a direct thrombin inhibitor, blocking only a single step in the coagulation pathway.[[3]] Dabigatran has become available in New Zealand and overseas in the last decade, and can be used as an alternative to warfarin in several thrombotic conditions, excluding mechanical heart valve replacement.[[4]] Advantages of dabigatran over warfarin are a faster onset of action, fewer drug and food interactions and less laboratory monitoring required.[[5]] These advantages make dabigatran more convenient to use in some patients. However, landmark NOAC trials have excluded patients with cirrhosis.[[4,6]]

In New Zealand, dabigatran was approved for use in 2008 and became fully funded on prescription in 2011.[[7]] Dabigatran requires minimal hepatic metabolism and is primarily eliminated through the kidneys.[[8]] Renal dysfunction may occur secondary to late-stage liver disease, leading to accumulation and increased drug concentrations.[[9]] Extended periods of time with high dabigatran concentrations may potentially lead to a higher bleeding risk in patients.[[10]] The pharmacokinetic changes of medication in cirrhosis were cited as a key reason for exclusion from landmark studies.[[11]] However, there is a paucity of data comparing the safety of dabigatran to warfarin in cirrhotic patients.[[12]] These data are necessary as there is a growing need for an alternative to warfarin. Our study's primary aim was to assess the rate of bleeding in patients with liver cirrhosis taking warfarin compared to those taking dabigatran.

Method

Study design and patient selection

This was a retrospective cohort study conducted in patients admitted to three tertiary care centres (district health boards/DHBs) in Auckland, New Zealand, from 1 January 2008 (the year that dabigatran became available in New Zealand) to 31 December 2020. Each DHB has one tertiary hospital centre. These are Auckland City Hospital (Auckland District Health Board, ADHB), with 1,124 beds, North Shore Hospital (Waitematā District Health Board, WDHB) with 600 beds, and Middlemore Hospital (Counties Manukau District Health Board, CMDHB) with 800 beds. The study has been approved by the ethics board of Auckland Health Research Ethics Committee (AH1163).

Within the three hospitals, any adult patient (≥18 years) diagnosed with liver disease or cirrhosis was identified using each hospital’s respective data information service and cirrhosis registries. Liver disease was defined by having one of the following International Classifications of Diseases 10[[th]] revision Australian Modified version (ICD-10 AM) discharge diagnosis codes (K701, K703, K704, K717, K721, K740-K746, K754, K758, K758, K760). Only the first hospital admission with a related diagnosis of any of the above codes was included. This admission was linked to community pharmacy dispensing records (Testsafe CareConnect) to identify patients with an anticoagulant dispensed from an outpatient pharmacy during the study period. This dispensing database, started in 2010, includes all records of prescribed medications dispensed to patients by community pharmacies. Patients were included if they had a dispensing record of an anticoagulant while having a confirmed cirrhosis diagnosis. Medication use for patient admissions prior to 2010 was identified using available electronic clinical notes. The index date was defined as the date of the first recorded dispensing of dabigatran or warfarin following cirrhosis diagnosis. The follow-up for each patient commenced from the index date until the occurrence of one of the following events: death, a bleeding event, liver transplant, diagnosis of advanced hepatocellular carcinoma, discontinuation or switch of anticoagulation therapy or the end of the study period (31 December 2020), whichever came first. Patients were excluded if they had a prescription record for other NOACs besides dabigatran on the index date. Patients with familial coagulopathy (such as haemophilia as diagnosed by haematologist) or advanced hepatocellular carcinoma (HCC) recorded any time before the index date were also excluded to reduce potential residual confounding effects.

Baseline characteristics such as age, ethnicity, sex, renal function (estimated glomerular filtration rate, eGFR), serum total bilirubin concentration and haemostasis status were collected. We also collected comorbidities such as polypharmacy, alcohol misuse, diabetes mellitus (type one and two), chronic kidney disease, hypertension, peripheral vascular disease (PVD), previous history of bleed, cerebrovascular disease, cancer, hepatocellular carcinoma and history of peptic ulcers. These characteristics were collected using electronic clinical notes and biochemistry data. We collected data on baseline medication use if they fell into the following categories. These were identified as potentially interacting medicines that may affect bleed risk: antidepressants, antibiotics, antiplatelets, non-steroidal anti-inflammatories and corticosteroids.

Cirrhotic diagnosis and severity

Liver cirrhosis was confirmed by written clinician diagnosis in the clinical notes along with supporting medical imaging evidence such as a FibroScan or computerised tomography (CT).

The severity of liver cirrhosis was defined by the Sodium Model of End-Stage Liver Disease (MELD-Na) score. Variables used to calculate MELD-Na were collected within 90 days of the index date.

Anticoagulation use

Computerised inpatient notes confirmed the date of initiation of anticoagulation, electronic discharge summaries and any outpatient pharmacy dispensing records. The first and last dispensing dates for each anticoagulant were collected along with clinical indication and dosage. If patients switched their anticoagulant (e.g. warfarin to dabigatran, vice versa, or other NOACs), they were censored from further follow-up.

Different criteria were used for warfarin and dabigatran to estimate the end of supply due to their different monitoring requirements and data availability. Dabigatran discontinuation was established if there was a gap between dispensing records of 90 days unless otherwise stated in clinic letters. The 90-day period was used as it is the maximum legal period of supply that a prescription for NOACs can be dispensed in New Zealand. The discontinuation of warfarin was identified if there was a gap between the supply of 56 days with absent international normalised ratio (INR) monitoring. A 56-day gap has been used in previous studies to indicate a lack of monitoring, and it is a period across which time in the therapeutic range is not interpolated.[[13]]

Outcome measures

Bleeding events were identified by reviewing the electronic clinical notes, discharge summaries, previous imaging reports and biochemistry results during their period of anticoagulation therapy. Events were identified as a primary or secondary diagnosis in the clinical notes. Prior published criteria were used, and the bleeds were characterised as “major” or “minor”.[[14]] At the time of the bleeding event, we collected MELD-Na scores, haemoglobin, platelet count and total bilirubin serum concentration. All bleeding events were independently assessed by three physician investigators (MH, CS, HW) using the Naranjo scale to evaluate the likelihood of the bleeding event being attributed to the anticoagulant. The Naranjo scale consists of 10 items with points being added or subtracted depending on the response, with a minimum score of negative four to a maximum score of 13.[[15]] The cause of bleed due to anticoagulation is considered definite if the score is nine or higher, probable if five to eight, possible if one to four, and doubtful if zero or fewer.[[15]]

Statistical analyses

Patient characteristics were presented as means and standard deviations for normally distributed variables and as medians and interquartile ranges (IQR) for variables with a skewed distribution. The number of patients (n) and percentages were used to represent categorical variables. Between-group comparisons were performed using the Chi-squared and Fisher’s exact test for categorical variables. The Mann–Whitney U test was used to compare non-normally distributed continuous variables such as renal function, serum bilirubin and platelet count. ANOVA or Student's t-test was used to compare normally distributed variables, serum haemoglobin and MELD-Na score. The incidence rate of bleeding was calculated as the number of patients with any bleeding event during follow-up divided by total follow-up time in person-year for both groups. The Kaplan–Meier (KM) method was used to compare bleeding events between warfarin and dabigatran cohorts, and the groups were compared using the log-rank test. Patients with missing data were excluded from the relevant analysis. Due to a significant difference in the length of follow-up between patients, the bleeding event data was truncated at five years for survival analysis. All statistical tests were two-tailed, and statistical significance was set at a p-value <0.05. Data analyses were performed using SPSS v27, and KM curve was generated using STATA v14 software packages.

Results

Initially, 4,518 patients admitted with liver disease were identified over the study period. A total of 4,153 patients were excluded as they were not dispensed any oral anticoagulation between 2008–2020. Of the 365 patients on anticoagulation during this time, 265 were excluded (203 were not cirrhotic, eight had advanced hepatocellular carcinoma, and 54 were excluded for reasons defined as “other”). Baseline MELD-Na score was unavailable for six patients. Of those meeting inclusion criteria, 52 took warfarin and 48 took dabigatran. See Figure 1 for cohort selection.

Study cohort characteristics

At baseline, we did not observe any statistically significant differences in sex, age or ethnicity distribution between warfarin and dabigatran cohorts (Table 1). Furthermore, both groups were similar regarding the aetiology of liver disease, baseline medications and comorbidities. The study arms differed in two areas: indication for anticoagulation and MELD-Na score. A higher proportion of patients with valvular AF were in the warfarin compared to the dabigatran cohort (42.3% vs 14.6%). In contrast, the percentage of patients with non-valvular AF was lower in warfarin than dabigatran cohort (36.5% vs 64.6%). Regarding the MELD-Na score, a higher proportion of patients in the warfarin cohort had MELD-Na score >20 than the dabigatran cohort (25% vs 8.3%). There were no differences in specific biochemical markers between the two treatment cohorts, including estimated GFR (p=0.177), total serum bilirubin (p=0.458), platelet count (p=0.583) and serum haemoglobin(p=0.092).

Bleeding risk

A total of 24 bleeding events occurred in total, six being major and 18 being minor. Half (12/24) of the bleeding events were identified to be gastrointestinal in nature. Of these 13% (3/24) of the events were related to the head, urological and epistaxis, and a quarter (6/24) of the events were unspecific, with related causes including hematomas, dental and vaginal.

The overall incidence rate of bleeding was 14.4 (95% CI, 8.8–23.5) and 9.1 (95% CI, 4.5–18.1) per 100 person-years in warfarin and dabigatran users, respectively. The incident rate ratio comparing dabigatran to warfarin was 0.63 (95% CI, 0.23–1.60), suggesting that patients taking dabigatran may have less risk of bleeding than patients taking warfarin. However, when compared using KM curves, this difference was not statically significant (p=0.25) (see Figure 2). Furthermore, no statistical significance was observed when KM analyses were undertaken to compare both MELD-Na score (p=0.07) and disease aetiology (p=0.13) to evaluate whether bleeding risk differed between warfarin and dabigatran users.

Change in biochemistry between bleeding event and baseline

When compared to baseline, the serum haemoglobin and the MELD-Na scores at the time of bleeding events showed changes that were of both statistical and clinical significance. The serum haemoglobin showed a significant decrease from mean (SD) 126.6 g/L (18.86) at baseline to 105.5 g/L (27.1) at the event (p<0.001), and the MELD-Na score had a significant increase from mean (SD) 16.8 (5.9) at baseline to 23.12 (8.1) at the event (p<0.001). However, serum bilirubin and platelet count did not alter significantly between bleeding event and baseline (Table 2).

Causality assessment of bleeding event and anticoagulation

Of the 24 bleed events, nine were considered possible and 15 were probable in likelihood of the bleed being attributed to the anticoagulant. Five were due to warfarin, and four were due to dabigatran in the nine events considered possibly related to the anticoagulant. In the 15 events deemed to be probable, eight were due to warfarin and seven were due to dabigatran.

View Tables and Figures.

Discussion

This paper presents the first data to our knowledge in the literature directly comparing the bleeding rates between the oral anticoagulants warfarin and dabigatran in patients with liver cirrhosis. There has been growing interest in this field since the emergence of NOACs, yet data on bleeding risk in the context of liver cirrhosis is scarce. Patients diagnosed with liver cirrhosis have traditionally been excluded from landmark studies, but more real-world retrospective studies are emerging.[[16]] In our study, we demonstrated that the bleeding risk of dabigatran did not differ compared to warfarin. In our initial cohorts, warfarin tended to be used in patients with higher MELD-Na scores and for the indication of valvular AF, and this may increase the bleeding risk of warfarin. These differences are likely a reflection of prescribing practices, whereby dabigatran is contraindicated and off-license in both valvular AF, moderate and severe hepatic impairment (CTP categories B and C).[[17]]

Compared to other studies of similar design, our bleed rates are similar to a study recently published by Mort et al, which focusses specifically on NOACs and bleeding risk. Their study’s overall bleed rate was 21%, and ours is 24%.[[12]] Although Mort et al. did not compare their cohort directly with warfarin users, they state that their rate of bleed for NOACs was comparable to published rates of bleed for warfarin users.[[12]] Another larger retrospective cohort study was conducted in Taiwan, using national health administrative data. Over 2,428 non-valvular AF patients with cirrhosis were included in this study. The risk of major bleed (HR 0.51, 95% CI 0.32–0.74) and major gastrointestinal bleed (HR 0.51, 95% CI 0.32-0.79) was lower in NOACs users compared to Warfarin users.[[18]] Compared to our study, the study differs significantly in methodology and findings. However, Lee et al.'s Taiwan study only included Taiwanese patients and may not be applicable to other ethnicities. In addition, our study also collected multiple clinically important biomarkers that are not often available in the administrative database. Asian populations have been found to have a higher risk of bleeding when taking warfarin compared to non-Asian populations, and previous studies indicate that NOACs may be a safer option in Asian vs non-Asian populations.[[19]]

Our study had several strengths, including using the Naranjo scale. We were able to standardise the assessment of bleeding events in patients with cirrhosis on dabigatran and warfarin. By using three independent physicians to ascertain the Naranjo score and compare scores, we ensured a more robust assessment of each bleeding event being related to the anticoagulant of choice. The study had access to a wide range of clinical data across the three main hospitals in Auckland, New Zealand, by using computerised notes and paper notes and laboratory results. Thus, for each patient, we were able to assess their liver disease status comprehensively and to only include those with a robust diagnosis of liver cirrhosis. We also collected several important variables that may influence bleeding risk, e.g., baseline MELD-Na score, full blood count and renal function. A weakness for our study is its low sample size, being an exploratory study. The study lacked enough input from other ethnic groups to make any meaningful comparison in outcomes, with the majority of patients being NZ European.

Further investigation into the ethnic differences in bleeding rates on warfarin and NOACs in New Zealand’s population, particularly in Māori and Pasifika populations, would be an area for further research to potentially change clinical practice. We also opted to combine both minor and major bleed to all bleed for our outcome of interest due to the low number of bleeding events in total. Other limitations include that the study was retrospective, so we cannot control for all confounders at baseline. The data is also limited as it was based in one region—in Auckland, New Zealand—and the population characteristics, while representing a broad range of ethnicities and three different DHB sites, may not represent populations elsewhere. However, Auckland represents around one-third of New Zealand’s total population, and it is likely that New Zealand data is not sufficient for this study.[[20]] Our study was exploratory as we did not have any existing data on the rate of oral anticoagulated patients with liver cirrhosis, thus we did not have a predetermined sample size. It is evident from our research that the use of oral anticoagulation is a rare occurrence in patients with liver cirrhosis in New Zealand (prevalence of less than 0.1%). We did not observe a statistically significant association between warfarin vs dabigatran use and subsequent risk of bleeding but, due to the modest sample size, this study is likely underpowered to detect any such an association, if present. An adequately powered study with comparable methodology will likely need to be conducted in countries with larger centres such as those in Asia, the United States and the United Kingdom.

In conclusion, our study found no statistically significant differences in the bleeding rate in cirrhotic patients treated with warfarin versus those treated with dabigatran. Our results suggest dabigatran may be as safe to use as warfarin in patients with cirrhosis.

Summary

Abstract

Aim

The safety of dabigatran is poorly studied in patients with liver cirrhosis and has rarely been compared to warfarin in terms of bleeding risks.

Method

We undertook a retrospective cohort study across three tertiary centres in Auckland, New Zealand, between 2008 to 2020. Adults 18 years and over and those with a clinically confirmed diagnosis of cirrhosis were included. Data collected included demographic data and clinical characteristics, baseline medication and comorbidities. The primary outcome measure was the incidence of any bleeding event that resulted in hospital admission.

Results

Overall, 100 patients were included in this study. A total of 52 patients took warfarin, and 48 took dabigatran. Baseline characteristics for both groups were generally similar. The incidence rate of bleeds for patients taking warfarin was 14.4 per 100 person-years (95% CI, 8.8–23.5) compared to 9.1 per 100 person-years (95% CI, 4.5–18.1) for patients taking dabigatran. The incidence rate ratio comparing dabigatran to warfarin was 0.63 (95% CI, 0.23–1.60), p=0.25.

Conclusion

Our study found that patients on dabigatran may have a lower bleeding risk than patients taking warfarin, but this was not statistically significant.

Author Information

Ms Oriana Munevar Aquite: School of Pharmacy, Faculty of Medical and Health Science, The University of Auckland, New Zealand. Dr Michael Hayes: Department of Gastroenterology, Waitematā District Health Board, New Zealand. Dr Kebede Beyene: School of Pharmacy, Faculty of Medical and Health Science, The University of Auckland, New Zealand. Dr Amy Hai Yan Chan: School of Pharmacy, Faculty of Medical and Health Science, The University of Auckland, New Zealand. Dr Cameron Schauer: Department of Gastroenterology, Waitematā District Health Board, New Zealand. Dr Henry Wei: Department of Gastroenterology, Auckland District Health Board, New Zealand. Mr Jiayi Gong: New Zealand Liver Research Unit, Auckland District Health Board, New Zealand.

Acknowledgements

New Zealand Pharmacy Education and Research Foundation Summer Studentship, New Zealand Liver Transplant Unit and New Zealand Liver Research Trust. Professor Edward Gane and Miss Alice Yu for providing research support for this study.

Correspondence

Jay Gong: M&HS Building 502, 85 Park Rd, Grafton, Auckland, 1023, New Zealand.

Correspondence Email

Jay.gong@auckland.ac.nz

Competing Interests

Authors declarations of personal interest: OMA received funding as part of her summer studentship from the New Zealand Pharmacy Education and Research Foundation for this piece of work. MH declares no conflicts of interest. KB declares no conflicts of interest. AHYC reports consultancy fees from Spoonful of Sugar Ltd, grants from Health Research Council, Innovate UK, A+ charitable trust (Auckland District Health Board), Maurice and Phyllis Paykel trust, Universitas 21, NZPERF, Auckland Academic Health Alliance, Asthma UK, The University of Auckland; grants and consultancy fees from Janssen-Cilag; and is the recipient of the Robert Irwin Fellowship and Senior Research Fellowship from the Auckland Medical Research Foundation. None related to this work. CS declares no conflict of interest. HW no conflict of interest. JG research grants from the New Zealand Liver Research Trust for this piece of work, Health Research Council, Auckland Academic Health Alliance and A+ charitable trust (Auckland District Health Board). None are commercial entities.

1) McCormick PA, Jalan R. Hepatic Cirrhosis. Sherlock's Diseases of the Liver and Biliary System. 2018:107-26.

2) Intagliata NM, Caldwell SH. Coagulation in Cirrhosis. Sherlock's Diseases of the Liver and Biliary System. 2018:53-61.

3) Evans NS. Direct Oral Anticoagulants. In: Lau JF, Barnes GD, Streiff MB, eds. Anticoagulation Therapy. Cham: Springer International Publishing. 2018:87-103.

4) Connolly SJ, Ezekowitz MD, Yusuf S, et al. Dabigatran versus Warfarin in Patients with Atrial Fibrillation. N Engl J Med. 2009;361(12):1139-51. doi: 10.1056/NEJMoa0905561.

5) Mekaj YH, Mekaj AY, Duci SB, et al. New oral anticoagulants: their advantages and disadvantages compared with vitamin K antagonists in the prevention and treatment of patients with thromboembolic events. Ther Clin Risk Manag. 2015;11:967-77. doi: 10.2147/tcrm.S84210 [published online first: 2015/07/08].

6) Schulman S, Kearon C, Kakkar AK, et al. Dabigatran versus Warfarin in the Treatment of Acute Venous Thromboembolism. N Engl J Med. 2009;361(24):2342-52. doi: 10.1056/NEJMoa0906598.

7) Pharmac. Application Tracker 2021 [Available from: https://connect.pharmac.govt.nz/apptracker/s/. Accessed 18 February 2021].

8) Fawzy AM, Lip GYH. Pharmacokinetics and pharmacodynamics of oral anticoagulants used in atrial fibrillation. Expert Opin Drug Metab Toxicol. 2019;15(5):381-98. doi: 10.1080/17425255.2019.1604686.

9) Francoz C, Nadim MK, Baron A, et al. Glomerular filtration rate equations for liver-kidney transplantation in patients with cirrhosis: Validation of current recommendations. Hepatology. 2014;59(4):1514-21. doi: https://doi.org/10.1002/hep.26704.

10) Qamar A, Vaduganathan M, Greenberger NJ, et al. Oral Anticoagulation in Patients With Liver Disease. J Am Coll Cardiol. 2018;71(19):2162-75. doi: https://doi.org/10.1016/j.jacc.2018.03.023.

11) Steffel J, Verhamme P, Potpara TS, et al. The 2018 European Heart Rhythm Association Practical Guide on the use of non-vitamin K antagonist oral anticoagulants in patients with atrial fibrillation. Eur Heart J. 2018;39(16):1330-93. doi: 10.1093/eurheartj/ehy136.

12) Mort JF, Davis JPE, Mahoro G, et al. Rates of Bleeding and Discontinuation of Direct Oral Anticoagulants in Patients With Decompensated Cirrhosis. Clin Gastroenterol Hepatol. 2021;19(7):1436-42. doi: 10.1016/j.cgh.2020.08.007 [published online first: 2020/08/11].

13) Rose AJ, Miller DR, Ozonoff A, et al. Gaps in monitoring during oral anticoagulation: insights into care transitions, monitoring barriers, and medication nonadherence. Chest. 2013;143(3):751-57. doi: 10.1378/chest.12-1119 [published online first: 2012/11/29].

14) Ageno W, Riva N, Schulman S, et al. Long-term Clinical Outcomes of Splanchnic Vein Thrombosis: Results of an International Registry. JAMA Intern Med. 2015;175(9):1474-80. doi: 10.1001/jamainternmed.2015.3184.

15) Livertox. Adverse Drug Reaction Probability Scale (Naranjo) in Drug Induced Liver Injury. 2012 [updated 2019 May 4]. Available from: https://www.ncbi.nlm.nih.gov/books/NBK548069/. Accessed 31 August 2021.

16) Elhosseiny S, Al Moussawi H, Chalhoub JM, et al. Direct Oral Anticoagulants in Cirrhotic Patients: Current Evidence and Clinical Observations. Can J Gastroenterol Hepatol. 2019;2019:4383269. doi: 10.1155/2019/4383269.

17) New Zealand Datasheet for Pradaxa: Medsafe New Zealand; 2020 [cited 25 January 2022]. Available from: https://www.medsafe.govt.nz/profs/datasheet/p/Pradaxacap.pdf. Accessed 25 January 2022.

18) Lee HF, Chan YH, Chang SH, 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: 10.1161/jaha.118.011112 [published online first: 2019/03/06].

19) Wang CL, Wu VC, Kuo CF, et al. Efficacy and Safety of Non-Vitamin K Antagonist Oral Anticoagulants in Atrial Fibrillation Patients With Impaired Liver Function: A Retrospective Cohort Study. J Am Heart Assoc. 2018;7(15):e009263. doi: 10.1161/jaha.118.009263 [published online first: 2018/10/30].

20) Auckland's Population Auckland: Auckland Council; 2018 [cited 25 January 2022]. Available from: https://www.aucklandcouncil.govt.nz/. Accessed 25 January 2022.

Contact diana@nzma.org.nz
for the PDF of this article

Subscriber Content

The full contents of this pages only available to subscribers.
Login, subscribe or email nzmj@nzma.org.nz to purchase this article.

LOGINSUBSCRIBE