Atrial fibrillation (AF) is associated with an increased risk of stroke or systemic thromboembolism. Approximately 5% of AF patients develop stroke or other embolic events each year.1 Anticoagulation with warfarin, a vitamin K antagonist, reduces stroke risk in AF patients by 64%.2 It has been the drug of choice for many years in both primary and secondary stroke prevention in AF. Unfortunately, anticoagulation increases the risk of bleeding, and intracerebral haemorrhage (ICH) has been the most life-threatening bleeding complication of concern. This study aimed to compare the clinical features and neurological outcomes of ICH in warfarinised AF patients with those in non-warfarinised patients.

This was a retrospective matched case series of consecutive patients with first acute ICH admitted to the medical unit of three tertiary hospitals in Hong Kong—Princess Margaret Hospital (PMH) and Caritas Medical Centre (CMC) in Kowloon West, and Queen Elizabeth Hospital (QEH) in Kowloon Central—from 1 January 2006 to 31 December 2011. The three hospitals cover about one quarter of the 7 million population in Hong Kong. It is a general practice in Hong Kong that patients with acute stroke symptoms are admitted to the medical unit (to acute stroke unit first, and to general medical ward if acute stroke unit is full) for further management, with computed tomography (CT) brain scans done within 24 hours of admission. If ICH is identified, a neurosurgeon will be consulted for assessment. Therefore ICH patients in the medical unit are a good indication of the general ICH population.

We searched our electronic database for all patients aged 18 years or above who developed first ICH in the presence of anticoagulation with warfarin for non-valvular AF (ICH-W group) from the three hospitals, and matched them with a comparison group (ICH-C group) without taking warfarin at a 1:1 ratio for age (±1 year), gender, and admission year. The comparison group comprised patients from the medical unit of PMH (principal study centre) who had a first episode of ICH without anticoagulation, regardless of any AF. Patients with isolated subdural, subarachnoid, or intraventricular haemorrhage were excluded. We retrieved and compared the data regarding neurological impairment and investigation findings, estimated the ICH volume on CT through the ABC/2 method, and calculated the ICH score.3 4 Hospital mortality and 6-month modified Rankin Scale score (mRS, 0-6) were selected as primary and secondary outcomes, respectively. We used independent sample t test and Mann-Whitney U test for univariate comparisons of continuous variables (Glasgow Coma Scale score, ICH score, ICH volume, mRS), and Chi squared test for categorical variables. Descriptive summary statistics, where appropriate, are presented as mean (range) or median (interquartile range [IQR]). All calculations were two-end conducted at a 0.05 level of significance. The study was approved by the local ethics committee (KW/EX-12-057[52-06]), with the requirement of patient informed consent waived because of its retrospective nature.

Overall, 114 patients were identified and recruited in the ICH-W group (37 from PMH, 8 from CMC, and 69 from QEH; Table 1) with a mean age of 75 years (range, 47-92 years) and a slight male predominance (56.1%). The same number of patients matched for age (±1 year) and gender were grouped for comparison (ICH-C). Both ICH-W and ICH-C groups had a median ICH score of 2, but there was a trend for higher ICH volume in ICH-W patients than in ICH-C patients (12.9 mL vs 10.5 mL) although it did not reach statistical significance. A total of 59 patients in ICH-W died during the same admission, and a further 10 patients had died by 6 months. The ICH-W patients had significantly higher hospital mortality (51.8% vs 36.0%; Chi squared test, P=0.02) and 6-month mortality (60.5% vs 43.0%; Chi squared test, P=0.01) than the ICH-C patients. The median mRS and the proportion with good recovery (mRS ≤3) at 6 months were comparable for the two groups (Table 1).

Two patients in the ICH-W group died before determination of international normalised ratio (INR). For the remaining 112, the admission INR was <2.0 in 16 (14.3%) patients, 2.0-3.0 in 66 (58.9%) patients, and >3.0 in 30 (26.8%) patients. The INR was re-checked in 85 patients after 12 hours and in 72 patients after 24 hours. The median INR was corrected from 2.6 (IQR, 2.1-3.1) to 1.4 (IQR, 1.2-1.7) at 12 hours post-event, and 1.3 (IQR, 1.1-1.5) at 24 hours post-event. The corresponding percentage of INR >1.5 was 96.4% (108/112) on admission, 35.3% (30/85) at 12 hours, and 22.2% (16/72) at 24 hours. No association was found between admission INR and ICH volume, hospital mortality, or 6-month mRS (Spearman’s rank correlation coefficient).

Regarding the warfarin reversal strategies, QEH had an in-patient protocol whereas PMH and CMC did not. Nevertheless, there was no significant difference in mortality rate among the three hospitals—hospital mortality/6-month mortality in PMH 18 (48.6%)/20 (54.1%), in CMC five (62.5%)/six (75.0%), and in QEH 36 (52.2%)/43 (62.3%). Prothrombin complex concentrate (PCC), fresh frozen plasma (FFP), vitamin K1 (VitK1), factor VII, and transamin were used alone or in combination in our ICH-W patients. Their frequency was as follows: FFP alone (69, 60.5%), FFP + VitK1 (17, 14.9%), FFP + PCC (9, 7.9%), PCC alone (3, 2.6%), transamin alone (3, 2.6%), VitK1 alone (2, 1.8%), FFP + transamin (1, 0.9%), and FFP + factor VII (1, 0.9%). Nine (7.9%) patients received no treatment—two patients died before admission INR was available, three patients died soon after admission INR was available, three patients had admission INR of <1.5, and one patient was on warfarin and aspirin for ischaemic heart disease. The high variation in anticoagulation reversal strategy made it impossible for any valid comparison.

Cardioembolic stroke related to AF carries significant morbidity, yet anticoagulation is not without risk. How to prevent stroke and at the same time minimise bleeding complications has always been a dilemma for physicians and patients. Risk assessment tools (eg CHA₂DS₂-VASc and HAS-BLED score, for assessment of thromboembolic and bleeding risk, respectively) have been developed to assist the decision making.5 6 It is important for clinicians to exercise individualised medical practice, however. In a recently published observation study on a large hospital cohort of Chinese AF patients from Hong Kong, the investigators found that the stroke risk in Chinese AF patients was higher than that in Caucasians at a given CHA₂DS₂-VASc score, and that excessive risk was more prominent in the low-risk group from CHA₂DS₂-VASc.7 At the same time, the ICH incidence in Chinese AF patients taking warfarin was also higher than that in Caucasians.8 These observations really drive clinicians towards a better anticoagulation strategy for Chinese AF patients.

Unfortunately ICH remains a significant threat even with very careful patient selection and treatment monitoring. Up to 25% of ICH can be associated with anticoagulant usage, and the rate is still increasing given the higher utilisation of anticoagulation in ageing populations with rising AF prevalence.9 10 Several features of warfarin-associated ICH in our patients deserve further elaboration, as listed below.

First, we found a large proportion (73.2%) of warfarin-associated ICH occurred when the INR was within or even below the therapeutic range (INR ≤3.0). Our findings were supported by a similar observation from another prospective study in which 68% of warfarin-associated ICH occurred at INR of ≤3.0.11 This shows that bleeding is increased even with an INR within the therapeutic range. Moreover, those ICH-W patients with a subtherapeutic INR reaffirm the high intrinsic ICH risk in AF patients who are old with multiple co-morbidities.

Second, we did not find any correlation between initial ICH volume and admission INR. The initial ICH volume in warfarinised and non-warfarinised patients was not significantly different. This is consistent with the results from some other studies.9 12 13 In contrast, Cucchiara et al14 reported larger haematoma volume in anticoagulant-associated ICH than spontaneous ICH, whereas Flaherty et al15 reported larger initial haematoma volume if admission INR was >3.0. It is reasonable to suppose that coagulopathy would have an impact on haematoma volume, but is difficult to confirm in clinical studies because it is hard to control other covariates that affect haematoma size. The relatively small number of subjects in the warfarin-associated ICH group is often underpowered to reach any solid conclusion. Nonetheless, other studies did show that warfarin use was a known predictor of haematoma expansion, and haematoma expansion an independent determinant of neurological outcomes in spontaneous ICH.12 14 16 All the consensus guidelines agree that coagulopathy should be reversed as soon as possible in warfarin-associated ICH.

Third, both the hospital and 6-month mortalities in our ICH-W patients were significantly higher than those in ICH-C patients (Table 1). This observation is highly consistent with other reports in which warfarin-associated ICH had higher mortality than patients without taking warfarin (Table 2).9 11 12 13 14 15 17 18 19 20 Our study adopted a matched case series design in order to eliminate the effect of age, and to minimise the age-related co-morbidities that are known predictors of poor neurological outcome from ICH. As a result, our selected ICH-C patients were older and probably had more co-morbidities than the general ICH patients. This explained the higher-than-usual mortality and poor neurological outcomes in our ICH-C group compared with other studies. Nonetheless, AF-related co-morbidity could not be controlled in this design, and we believe our observed difference in mortality and neurological outcomes is a compound effect from AF and warfarin use in the ICH-W patients.

Currently, the optimal therapy for coagulopathy reversal in warfarin-associated major bleeding is unclear and recommendations of international guidelines are mainly derived from consensus opinions. Treatment options include FFP, VitK1, PCC, and recombinant factor VIIa, in different combinations.21 One of the three study hospitals (QEH) had an in-house protocol for warfarin reversal. The other two hospitals provide guidelines only. Because of the incomplete data collection from this retrospective study design, we were unable to analyse the haematoma growth and warfarin reversals. The mortality, however, did not differ among the three hospitals.

Since 2010, newer oral anticoagulants have become available and provide an alternative to warfarin in AF stroke prevention. The three new oral anticoagulants—dabigatran, rivaroxaban, and apixaban22 23 24—might differ in efficacy, but the lower ICH rate compared with warfarin was a universal finding from their clinical trials, and is a class effect of these new agents. They should be considered in Chinese AF patients at a high risk of warfarin-associated ICH.

As a retrospective study, this study has several limitations. First, the methods and timing of warfarin reversal were not standardised and could have affected the neurological outcome. Second, there was no protocol for serial CT brain and serial INR monitoring to investigate the effect of coagulopathy reversal on haematoma expansion, and the effect of haematoma expansion on neurological outcome. Third, we were uncertain about compliance with the coagulopathy reversal protocol in QEH, and were unable to comment on whether the standardised protocol could improve ICH outcome. Moreover, ICH-W patients were AF patients who had additional vascular risk factors (to justify the use of warfarin) that may not have been present in ICH-C patients. This difference could potentially affect the ICH outcome. Finally, other possible confounders such as smoking or alcohol use were not adjusted for in this retrospective study.

Warfarin-associated ICH in AF patients may be associated with higher stroke mortality. It could be a serious problem in Chinese AF patients who are known to have more warfarin-associated ICH.

The authors declared no conflicts of interest in this study.

1. Wolf PA, Abbott RD, Kannel WB. Atrial fibrillation as an independent risk factor for stroke: the Framingham Study. Stroke 1991;22:983-8. Crossref

2. Hart RG, Pearce LA, Aguilar MI. Meta-analysis: antithrombotic therapy to prevent stroke in patients who have nonvalvular atrial fibrillation. Ann Intern Med 2007;146:857-67. Crossref

3. Kothari RU, Brott T, Broderick JP, et al. The ABCs of measuring intracerebral hemorrhage volumes. Stroke 1996;27:1304-5. Crossref

4. Hemphill JC 3rd, Bonovich DC, Besmertis L, Manley GT, Johnston SC. The ICH score: a simple, reliable grading scale for intracerebral hemorrhage. Stroke 2001;32:891-7. Crossref

5. Lip GY, Nieuwlaat R, Pisters R, Lane DA, Crijns HJ. Refining clinical risk stratification for predicting stroke and thromboembolism in atrial fibrillation using a novel risk factor-based approach: the Euro heart survey on atrial fibrillation. Chest 2010;137:263-72. Crossref

6. Pisters R, Lane DA, Nieuwlaat R, de Vos CB, Crijns HJ, Lip GY. A novel user-friendly score (HAS-BLED) to assess 1-year risk of major bleeding in patients with atrial fibrillation: the Euro heart survey. Chest 2010;138:1093-100. Crossref

7. Siu CW. One more “C” for CHA2DS2-VASc score? J Am Coll Cardiol 2015;65:1602-3. Crossref

8. Siu CW, Lip GY, Lam KF, Tse HF. Risk of stroke and intracranial hemorrhage in 9727 Chinese with atrial fibrillation in Hong Kong. Heart Rhythm 2014;11:1401-8. Crossref

9. Horstmann S, Rizos T, Lauseker M, et al. Intracerebral hemorrhage during anticoagulation with vitamin K antagonists: a consecutive observational study. J Neurol 2013;260:2046-51. Crossref

10. Flaherty ML, Kissela B, Woo D, et al. The increasing incidence of anticoagulant-associated intracerebral hemorrhage. Neurology 2007;68:116-21. Crossref

11. Rosand J, Eckman MH, Knudsen KA, Singer DE, Greenberg SM. The effect of warfarin and intensity of anticoagulation on outcome of intracerebral hemorrhage. Arch Intern Med 2004;164:880-4. Crossref

12. Flibotte JJ, Hagan N, O’Donnell J, Greenberg SM, Rosand J. Warfarin, hematoma expansion, and outcome of intracerebral hemorrhage. Neurology 2004;63:1059-64. Crossref

13. Sheth KN, Cushing TA, Wendell L, et al. Comparison of hematoma shape and volume estimates in warfarin versus non-warfarin-related intracerebral hemorrhage. Neurocrit Care 2010;12:30-4. CrossRef

14. Cucchiara B, Messe S, Sansing L, Kasner S, Lyden P, CHANT Investigators. Hematoma growth in oral anticoagulant related intracerebral hemorrhage. Stroke 2008;39:2993-6. Crossref

15. Flaherty ML, Tao H, Haverbusch M, et al. Warfarin use leads to larger intracerebral hematomas. Neurology 2008;71:1084-9. Crossref

16. Davis SM, Broderick J, Hennerici M, et al. Hematoma growth is a determinant of mortality and poor outcome after intracerebral hemorrhage. Neurology 2006;66:1175-81. Crossref

17. Flaherty ML, Haverbusch M, Sekar P, et al. Location and outcome of anticoagulant-associated intracerebral hemorrhage. Neurocrit Care 2006;5:197-201. Crossref

18. Zubkov AY, Mandrekar JN, Claassen DO, Manno EM, Wijdicks EF, Rabinstein AA. Predictors of outcome in warfarin-related intracerebral hemorrhage. Arch Neurol 2008;65:1320-5. Crossref

19. Huhtakangas J, Tetri S, Juvela S, Saloheimo P, Bode MK, Hillbom M. Effect of increased warfarin use on warfarin-related cerebral hemorrhage: a longitudinal population-based study. Stroke 2011;42:2431-5. Crossref

20. Seçil Y, Ciftçi Y, Tokuçoğlu F, Beckmann Y. Intracranial hemorrhages related with warfarin use and comparison of warfarin and acetylsalicylic acid. J Stroke Cerebrovasc Dis 2014;23:321-6. Crossref

21. Aguilar MI, Hart RG, Kase CS, et al. Treatment of warfarin-associated intracerebral hemorrhage: literature review and expert opinion. Mayo Clin Proc 2007;82:82-92. Crossref

22. Connolly SJ, Ezekowitz MD, Yusuf S, et al. Dabigatran versus warfarin in patients with atrial fibrillation. N Engl J Med 2009;361:1139-51. Crossref

23. Patel MR, Mahaffey KW, Garg J, et al. Rivaroxaban versus warfarin in nonvalvular atrial fibrillation. N Engl J Med 2011;365:883-91. Crossref

24. Granger CB, Alexander JH, McMurray JJ, et al. Apixaban versus warfarin in patients with atrial fibrillation. N Engl J Med 2011;365:981-92. Crossref