[Skip to Content]
Access to paid content on this site is currently suspended due to excessive activity being detected from your IP address 34.204.194.190. Please contact the publisher to request reinstatement.
[Skip to Content Landing]
Figure 1.
Flowchart of the Study Population
Flowchart of the Study Population

AF indicates atrial fibrillation; ICH, intracranial hemorrhage; and NOAC, non–vitamin K antagonist oral anticoagulant.

Figure 2.
Forest Plots of Studied Outcomes and Associations With Resumption or No Resumption of Warfarin Treatment
Forest Plots of Studied Outcomes and Associations With Resumption or No Resumption of Warfarin Treatment

No resumption was the reference condition. The analyses were stratified by index ICH event. Crude analyses are depicted by lines with blue boxes; lines with orange boxes for adjusted analyses. AHR indicates adjusted hazard ratio; ICH, intracranial hemorrhage; and SE, systemic embolism.

Table 1.  
Patient Characteristicsa
Patient Characteristicsa
Table 2.  
Studied Outcomes During 1 Year of Follow-up
Studied Outcomes During 1 Year of Follow-up
1.
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(12):857-867.PubMedGoogle ScholarCrossref
2.
January  CT, Wann  LS, Alpert  JS,  et al; American College of Cardiology/American Heart Association Task Force on Practice Guidelines.  2014 AHA/ACC/HRS guideline for the management of patients with atrial fibrillation: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines and the Heart Rhythm Society.  J Am Coll Cardiol. 2014;64(21):e1-e76.PubMedGoogle ScholarCrossref
3.
Kirchhof  P, Benussi  S, Kotecha  D,  et al.  2016 ESC Guidelines for the management of atrial fibrillation developed in collaboration with EACTS.  Eur Heart J. 2016;37(38):2893-2962.PubMedGoogle ScholarCrossref
4.
Camm  AJ, Lip  GYH, De Caterina  R,  et al; ESC Committee for Practice Guidelines-CPG; Document Reviewers.  2012 focused update of the ESC guidelines for the management of atrial fibrillation: an update of the 2010 ESC guidelines for the management of atrial fibrillation—developed with the special contribution of the European Heart Rhythm Association.  Europace. 2012;14(10):1385-1413.PubMedGoogle ScholarCrossref
5.
Fang  MC, Go  AS, Chang  Y,  et al.  Thirty-day mortality after ischemic stroke and intracranial hemorrhage in patients with atrial fibrillation on and off anticoagulants.  Stroke. 2012;43(7):1795-1799.PubMedGoogle ScholarCrossref
6.
Wolff  A, Shantsila  E, Lip  GYH, Lane  DA.  Impact of advanced age on management and prognosis in atrial fibrillation: insights from a population-based study in general practice.  Age Ageing. 2015;44(5):874-878.PubMedGoogle ScholarCrossref
7.
van Asch  CJ, Luitse  MJ, Rinkel  GJ, van der Tweel  I, Algra  A, Klijn  CJ.  Incidence, case fatality, and functional outcome of intracerebral haemorrhage over time, according to age, sex, and ethnic origin: a systematic review and meta-analysis.  Lancet Neurol. 2010;9(2):167-176.PubMedGoogle ScholarCrossref
8.
Pugh  D, Pugh  J, Mead  GE.  Attitudes of physicians regarding anticoagulation for atrial fibrillation: a systematic review.  Age Ageing. 2011;40(6):675-683.PubMedGoogle ScholarCrossref
9.
Chao  T-F, Liu  C-J, Liao  J-N,  et al.  Use of oral anticoagulants for stroke prevention in patients with atrial fibrillation who have a history of intracranial hemorrhage.  Circulation. 2016;133(16):1540-1547.PubMedGoogle ScholarCrossref
10.
Nielsen  PB, Larsen  TB, Skjøth  F, Gorst-Rasmussen  A, Rasmussen  LH, Lip  GYH.  Restarting anticoagulant treatment after intracranial hemorrhage in patients with atrial fibrillation and the impact on recurrent stroke, mortality, and bleeding: a nationwide cohort study.  Circulation. 2015;132(6):517-525.PubMedGoogle ScholarCrossref
11.
Lynge  E, Sandegaard  JL, Rebolj  M.  The Danish National Patient Register.  Scand J Public Health. 2011;39(7)(suppl):30-33.PubMedGoogle ScholarCrossref
12.
Kildemoes  HW, Sørensen  HT, Hallas  J.  The Danish National Prescription Registry.  Scand J Public Health. 2011;39(7)(suppl):38-41.PubMedGoogle ScholarCrossref
13.
Pedersen  CB.  The Danish Civil Registration System.  Scand J Public Health. 2011;39(7)(suppl):22-25.PubMedGoogle ScholarCrossref
14.
Krarup  L-H, Boysen  G, Janjua  H, Prescott  E, Truelsen  T.  Validity of stroke diagnoses in a national register of patients.  Neuroepidemiology. 2007;28(3):150-154.PubMedGoogle ScholarCrossref
15.
Wildenschild  C, Mehnert  F, Thomsen  RW,  et al.  Registration of acute stroke: validity in the Danish Stroke Registry and the Danish National Registry of Patients.  Clin Epidemiol. 2013;6:27-36.PubMedGoogle ScholarCrossref
16.
Stoker  TB, Evans  NR.  Managing risk after intracerebral hemorrhage in concomitant atrial fibrillation and cerebral amyloid angiopathy.  Stroke. 2016;47(7):e190-e192.PubMedGoogle ScholarCrossref
17.
Collins  CE, Witkowski  ER, Flahive  JM, Anderson  FA  Jr, Santry  HP.  Effect of preinjury warfarin use on outcomes after head trauma in Medicare beneficiaries.  Am J Surg. 2014;208(4):544-549.e1.PubMedGoogle ScholarCrossref
18.
Albrecht  JS, Liu  X, Baumgarten  M,  et al.  Benefits and risks of anticoagulation resumption following traumatic brain injury.  JAMA Intern Med. 2014;174(8):1244-1251.PubMedGoogle ScholarCrossref
19.
McGrath  ER, Kapral  MK, Fang  J,  et al; Investigators of the Registry of the Canadian Stroke Network.  Which risk factors are more associated with ischemic stroke than intracerebral hemorrhage in patients with atrial fibrillation?  Stroke. 2012;43(8):2048-2054.PubMedGoogle ScholarCrossref
20.
Nielsen  PB, Larsen  TB, Skjøth  F, Overvad  TF, Lip  GYH.  Stroke and thromboembolic event rates in atrial fibrillation according to different guideline treatment thresholds: a nationwide cohort study.  Sci Rep. 2016;6:27410.PubMedGoogle ScholarCrossref
21.
Lip  GYH, Nielsen  PB.  Should patients with atrial fibrillation and 1 stroke risk factor (CHA2DS2-VASc score 1 in men, 2 in women) be anticoagulated? yes: even 1 stroke risk factor confers a real risk of stroke.  Circulation. 2016;133(15):1498-1503.PubMedGoogle ScholarCrossref
22.
Charidimou  A, Wilson  D, Shakeshaft  C,  et al.  The Clinical Relevance of Microbleeds in Stroke study (CROMIS-2): rationale, design, and methods.  Int J Stroke. 2015;10(Suppl A100):155-161.PubMedGoogle ScholarCrossref
23.
Al-Shahi Salman  R, Bell  S.  The contemporary conundrum of antithrombotic drugs after intracerebral haemorrhage.  ACNR. 2015;15(3):12-15.Google Scholar
24.
van Nieuwenhuizen  KM, van der Worp  HB, Algra  A,  et al; APACHE-AF Investigators.  Apixaban versus antiplatelet drugs or no antithrombotic drugs after anticoagulation—Associated Intracerebral Haemorrhage in Patients With Atrial Fibrillation (APACHE-AF): study protocol for a randomised controlled trial.  Trials. 2015;16:393.PubMedGoogle ScholarCrossref
25.
Glynn  RJ, Knight  EL, Levin  R, Avorn  J.  Paradoxical relations of drug treatment with mortality in older persons.  Epidemiology. 2001;12(6):682-689.PubMedGoogle ScholarCrossref
26.
Lovelock  CE, Cordonnier  C, Naka  H,  et al; Edinburgh Stroke Study Group.  Antithrombotic drug use, cerebral microbleeds, and intracerebral hemorrhage: a systematic review of published and unpublished studies.  Stroke. 2010;41(6):1222-1228.PubMedGoogle ScholarCrossref
27.
Yates  PA, Villemagne  VL, Ellis  KA, Desmond  PM, Masters  CL, Rowe  CC.  Cerebral microbleeds: a review of clinical, genetic, and neuroimaging associations.  Front Neurol. 2014;4:205.PubMedGoogle ScholarCrossref
Original Investigation
April 2017

Outcomes Associated With Resuming Warfarin Treatment After Hemorrhagic Stroke or Traumatic Intracranial Hemorrhage in Patients With Atrial Fibrillation

Author Affiliations
  • 1Aalborg Thrombosis Research Unit, Department of Clinical Medicine, Faculty of Health, Aalborg University, Aalborg, Denmark
  • 2Department of Cardiology, Atrial Fibrillation Study Group, Aalborg University Hospital, Aalborg, Denmark
  • 3Unit of Clinical Biostatistics, Aalborg University Hospital, Aalborg, Denmark
  • 4Institute for Cardiovascular Sciences, University of Birmingham, City Hospital, Birmingham, England
JAMA Intern Med. 2017;177(4):563-570. doi:10.1001/jamainternmed.2016.9369
Key Points

Question  What is the prognosis associated with resuming oral anticoagulant treatment in patients with atrial fibrillation who sustain an intracranial hemorrhage?

Findings  In this observational study of 2415 patients who experienced a traumatic intracranial hemorrhage or a hemorrhagic stroke, the associated risk of resuming oral anticoagulant therapy suggested a reduction in thromboembolism in both patient subgroups. Patients who experiences a hemorrhagic stroke had a higher rate of intracranial hemorrhage recurrence after treatment resumption, which was not present for patients who had sustained a traumatic intracranial hemorrhage.

Meaning  Patients with atrial fibrillation who sustain a hemorrhagic stroke or traumatic intracranial hemorrhage have different prognoses, and recommendations for resuming oral anticoagulant treatment should consider this difference.

Abstract

Importance  The increase in the risk for bleeding associated with antithrombotic therapy causes a dilemma in patients with atrial fibrillation (AF) who sustain an intracranial hemorrhage (ICH). A thrombotic risk is present; however, a risk for serious harm associated with resumption of anticoagulation therapy also exists.

Objective  To investigate the prognosis associated with resuming warfarin treatment stratified by the type of ICH (hemorrhagic stroke or traumatic ICH).

Design, Setting, and Participants  This nationwide observational cohort study included patients with AF who sustained an incident ICH event during warfarin treatment from January 1, 1998, through February 28, 2016. Follow-up was completed April 30, 2016. Resumption of warfarin treatment was evaluated after hospital discharge.

Exposures  No oral anticoagulant treatment or resumption of warfarin treatment, included as a time-dependent exposure.

Main Outcomes and Measures  One-year observed event rates per 100 person-years were calculated, and treatment strategies were compared using time-dependent Cox proportional hazards regression models with adjustment for age, sex, length of hospital stay, comorbidities, and concomitant medication use.

Results  A total of 2415 patients with AF in this cohort (1481 men [61.3%] and 934 women [38.7%]; mean [SD] age, 77.1 years [9.1 years]) sustained an ICH event. Of these events, 1325 were attributable to hemorrhagic stroke and 1090 were secondary to trauma. During the first year, 305 patients with a hemorrhagic stroke (23.0%) died, whereas 210 in the traumatic ICH group (19.3%) died. Among patients with hemorrhagic stroke, resuming warfarin therapy was associated with a lower rate of ischemic stroke or systemic embolism (SE) (adjusted hazard ratio [AHR], 0.49; 95% CI, 0.24-1.02) and an increased rate of recurrent ICH (AHR, 1.31; 95% CI, 0.68-2.50) compared with not resuming warfarin therapy, but these differences did not reach statistical significance. For patients with traumatic ICH, resuming warfarin therapy also was associated with a lower rate of ischemic stroke or SE (AHR, 0.40; 95% CI, 0.15-1.11); however, in contrast to patients with hemorrhagic stroke, therapy resumption was associated with a significantly lower rate of recurrent ICH (AHR, 0.45; 95% CI, 0.26-0.76). A reduction in mortality was associated with resuming warfarin therapy among patients with hemorrhagic stroke (AHR, 0.51; 95% CI, 0.37-0.71) and those with traumatic ICH (AHR, 0.35; 95% CI, 0.23-0.52).

Conclusions and Relevance  Resumption of warfarin therapy after spontaneous hemorrhagic stroke in patients with AF was associated with a lower rate of ischemic events and a higher rate of recurrent ICH. Among patients with a traumatic ICH, a similar lower rate of ischemic events was found; however, a lower relative risk for recurrent ICH despite resuming warfarin treatment was also revealed.

Introduction

Treatment with oral anticoagulants (OACs) reduces the incidence of thromboembolic events and death among patients with atrial fibrillation (AF).1 Thus, in contemporary guidelines,2,3 anticoagulation is indicated for most patients with AF and at least 1 additional risk factor based on the CHA2DS2-VASc score (cardiac failure or dysfunction, hypertension, age 65-74 [1 point] or ≥75 years [2 points], diabetes, and stroke, transient ischemic attack or thromboembolism [2 points]–vascular disease, and sex category [female]). Given the close association of advanced age, AF, and subsequent increased risk for stroke, anticoagulation treatment is the mainstay of reducing the risk for cardioembolic stroke and all-cause mortality among these patients.2,4 However, the treatment may come at the cost of an increased risk for bleeding, including bleeding in the brain, which is the most feared clinical situation in patients receiving antithrombotic therapy. In patients with AF who encounter a warfarin-associated intracranial hemorrhage (ICH), the risk for 30-day mortality approaches 50%.5

Despite evidence of benefit, warfarin treatment has been reported to be underused in such patients of advanced age.6 This underuse may be related to a perceived risk for falls, but also an increase in the risk for ICH that has been shown to be associated with age.7,8 Patients with AF who survive an ICH event are still at risk for thromboembolism, but this risk has to be balanced against the treatment-related risk for recurrent intracranial bleeding. This serious treatment conundrum has been investigated in previous studies,9,10 which show a significant reduction in all-cause mortality among those who resume warfarin treatment. Nevertheless, evidence of benefit or harm from resuming warfarin treatment is scarce in this group of patients. We sought to investigate the prognosis in survivors of ICH with preexisting AF who receive warfarin treatment, stratified by whether or not warfarin treatment was resumed after the hospitalization and by the cause of the bleeding as a traumatic event, such as a fall or accident, or spontaneous ICH, such as hemorrhagic stroke.

Methods

This observational cohort study using historical data was based on 3 Danish nationwide databases: (1) the Danish National Patient Register11 (established in 1977), which includes admission and discharge dates and the discharge diagnoses for hospital admissions from the International Classification of Diseases and Health Related Problems, Tenth Revision (ICD-10); (2) the Danish National Prescription Registry,12 which holds information on purchase date, Anatomical Therapeutic Chemical classification code, and package size for every prescription withdrawal since 1994; and (3) the Danish Civil Registration System,13 which includes information on sex, date of birth, vital status and emigration status. Every resident in Denmark is assigned a unique identification number allowing linkage on an individual level between databases. No ethical approval or informed consent was obtained, because these are not mandated for registry studies in Denmark.

Study Population and Exposure

Persons with an incident diagnosis of AF from January 1, 1998, through February 28, 2016, were initially identified. All such patients with a subsequent ICH event underwent screening for inclusion based on primary and secondary diagnoses. For patients with a spontaneous hemorrhagic stroke defined by ICD-10 codes, we included nontraumatic subarachnoid hemorrhage (code I60); nontraumatic intracerebral hemorrhage, including bleeding in hemisphere, brain stem, cerebellum, intraventricular, or multiple locations or unspecified bleeds (code I61); and nontraumatic subdural hemorrhage or nontraumatic extradural hemorrhage (code I62). For patients with a trauma-induced ICH defined by ICD-10 codes we included focal traumatic intracranial bleeding (code SC063C), traumatic epidural hemorrhage (code SC064), traumatic subdural hemorrhage (code SC065), and traumatic subarachnoid hemorrhage (code SC065). Patients with a prior ICH or sequela from ICH (ICD-10 codes I690-I692) were excluded. To investigate warfarin-associated ICH events, we required patients to be treated with warfarin as their anticoagulant; thus, patients who did not claim a prescription of warfarin 6 months before their index ICH event were excluded.

We used information from the National Prescription Registry to categorize patients into groups of warfarin exposure after hospital discharge. All patients were initially assigned to the no treatment group when discharged from the hospital. Patients who claimed a prescription for warfarin were assigned to the warfarin treatment group on the date of prescription claim, and their grouping remained unchanged throughout the follow-up period. Thus, treatment exposure to warfarin was considered as a time-dependent variable.

Comorbidities and Outcomes

Baseline information was acquired on the date of hospital discharge. Comorbidities were based on preceding hospital diagnoses until discharge after the incident ICH event. Prescription claims within the last year before baseline defined concomitant medications. The CHA2DS2-VASc score was calculated to assess cardiovascular comorbidity and the risk for stroke at baseline. We calculated the HAS-BLED score (hypertension, abnormal renal or liver function, stroke or thromboembolism, bleeding history, labile international normalized ratio [not included because of data unavailability], elderly [age >65 years], drug consumption/alcohol excess [aspirin was not included owing to status of exposure]) for each patient to assess the bleeding risk. The eTable in the Supplement provides a detailed description of outcomes and concomitant medication. Owing to lack of clinically adjudicated ICH severity, the length of hospital stay (calculated as the date from the ICH event until the day of discharge) was used as a proxy of ICH severity; longer hospital stays reflected more complicated (and severe) conditions.

Patients were followed up in the National Patient Registry from 14 days after hospital discharge (index date) to ensure ascertained outcomes were not immediately related to the inclusion event. The primary study outcomes were (1) a composite outcome of ischemic stroke or systemic embolism (ICD-10 codes I63, I64, and I74) and (2) recurrent ICH defined as a composite outcome of all types of bleeding in the brain (ICD-10 codes I60, I61, I62, S063C, S064, S065, and S066). We also investigated a secondary outcome of all-cause mortality. To allow for a thorough prognostic evaluation, a principal outcome was defined as all stroke by a composite of ischemic stroke or intracerebral hemorrhage (ICD-10 codes I61, I63, and I64). Because thromboembolism may be closely correlated with a terminal outcome and no postmortem data were available, an additional outcome analysis was performed using a composite outcome of ischemic stroke or systemic embolism (SE) and death. Given the severity of the studied outcomes, we only considered coded diagnoses as an outcome event if the patient was admitted to the hospital for at least 24 hours and the outcome was coded as the primary cause of hospital admission. Because of poor validity, emergency department–coded diagnoses were not included in this study.14 The outcomes of ischemic stroke and ICH have previously been validated and found to be accurate for epidemiologic research.15

Statistical Analysis

We used time-to-event analysis to contrast the relative risk of an end point between groups, measuring time at risk from 14 days after hospital discharge until the relevant event; emigration; death; end of follow-up, defined as a maximum of 1 year; or the end of follow-up (April 30, 2016), whichever came first. Owing to lack of international normalized ratio values, we disregarded any treatment cessation; however, person-time was censored if patients initiated or shifted treatment to a non–vitamin K antagonist oral anticoagulation therapy (OAC). Exposure to warfarin treatment was regarded as a time-dependent variable with a single irreversible transition (0 for no treatment to 1 for OAC treatment). Crude 1-year incidence rates were calculated as the number of events divided by 100 person-years, where person-years were split in accordance with the time-dependent exposure variable. Adjusted Cox proportional hazards regression analyses were conducted to investigate relative risk between the treatment groups, with no warfarin treatment being the reference. The adjustment model included age (continuous, cubic spline), CHA2DS2-VASc score (continuous), HAS-BLED score (continuous), sex, previous thromboembolism, vascular disease, hypertension, diabetes, aspirin, β-blocker, nonsteroidal anti-inflammatory drugs, statins, and days in hospital from the index event (continuous, cubic spline). The adjustment model also included information on ischemic stroke or SE and recurrent ICH events occurring during the first 14 days after hospital discharge (binary) because such events could affect the choice of warfarin treatment before the observation time commenced.

To assess whether the analyses were confounded by indication (ie, selective prescribing due to complications of the index event and other comorbidities), we performed propensity-matched analyses in each stratum of hemorrhagic stroke and traumatic ICH. Specifically, we included the variables applied in the multivariable Cox proportional hazards regression model to calculate the propensity score of receiving OAC treatment or not. We attempted to match nonusers to users in a 2:1 ratio. Details on quality of matching are provided in the eMethods and eFigures 1 and 2 in the Supplement. Expert opinion suggests a maximum of 10 weeks before resuming warfarin treatment in patients with AF at high risk for stroke if no evidence suggests cerebral amyloid angiopathy.16 Therefore, we imposed a 10-week period to allow sufficient time for warfarin treatment resumption in terms of time to claim a prescription before the observation time began. The treatment exposure groups were subsequently defined within this period and analyzed according to those who claimed a warfarin prescription (warfarin treatment group) and those who did not (no treatment group), with follow-up starting at week 10. To further elaborate on the consequences of the chosen landmark, we calculated the mean time to warfarin prescription claim (if any) from the day of hospital discharge to 3 months; this time was subsequently used as the landmark (ie, start of observation time) for the repeated analyses.

We used STATA/MP (version 14; StataCorp) and R software (version 3.1.1; https://www.r-project.org) for the statistical analysis and graphical illustrations. A 2-sided P < .05 was considered to be statistically significant.

Results

A total of 6749 patients with incident ICH subsequent to their AF diagnosis were identified from January 1, 1998, through February 28, 2016; of patients who received prior warfarin treatment, 1372 died during hospital admission (Figure 1). In the cohort of patients discharged from the hospital, 436 died within 14 days. The study population consisted of 2415 patients with AF who received warfarin treatment at the first-time ICH event; of these, 1325 had a hemorrhagic stroke and 1090 had a traumatic ICH (Table 1). The mean (SD) age for the study population was 77.1 (9.1) years; 934 (38.7%) were women, and 1481 (61.3%) were men.

The most common comorbidities were hypertension (1552 [64.3%]) and heart failure or left ventricular dysfunction (708 [29.3%]). Among patients who presented with a hemorrhagic stroke, 550 (41.5%) had a prior thromboembolic event compared with 320 (29.4%) who presented with a traumatic ICH. The mean (SD) CHA2DS2-VASc score was 4.1 (1.7) for patients with a hemorrhagic stroke and 3.8 (1.7) for patients who sustained a traumatic ICH. The mean (SD) HAS-BLED score for all patients was 3.6 (1.1) and was similar for both subgroups. The hospital mean (SD) length of stay from the ICH event was higher for the hemorrhagic stroke cohort at 15.9 (18.4) vs 8.2 (12.0) days.

The cohort accrued a median follow-up period of 279 days. A total of 96 primary outcomes of ischemic stroke or SE and 139 recurrent intracranial bleeding episodes were observed. The overall annualized all-cause mortality was high (514 events [25.4%]) and was higher among those who did not resume warfarin treatment (418 events [34.9%]) compared with those who did (96 events [15.5%]).

Prognosis for Patients With Hemorrhagic Stroke

The absolute number of events and event rates are presented in Table 2. The primary event of ischemic stroke or SE was observed in 69 patients, and intracranial bleeding was observed in 50. Patients who resumed warfarin therapy had a lower rate of ischemic stroke or SE (3.3 per 100 person-years) than did patients who did not resume treatment (8.9 per 100 person-years; adjusted hazard ratio [AHR], 0.49; 95% CI, 0.24-1.02) (Figure 2); however, statistical significance was not reached. For recurrent ICH, the rates were 5.8 per 100 person-years vs 5.3 per 100 person-years (AHR, 1.31; 95% CI, 0.68-2.50). The rates for the composite end point of all stroke (ischemic and hemorrhagic strokes) were 7.3 vs 12.1 per 100 person-years (AHR, 0.74; 95% CI, 0.44-1.27). For the secondary end point of all-cause mortality, the associated risk with resumption of warfarin therapy was an AHR of 0.51 (95% CI, 0.37-0.71).

Prognosis for Patients With Traumatic ICH

During the first year of follow-up, 27 patients experienced an ischemic stroke or SE, whereas 89 experienced a recurrent ICH almost solely driven by diagnoses of recurrent traumatic ICH (87 [97.8%]). Patients who resumed warfarin therapy had a lower rate of ischemic stroke or SE (2.1 per 100 person-years vs 4.1 per 100 person-year; AHR, 0.40; 95% CI, 0.15-1.11) (Figure 2), but the difference was not statistically significant. The adjusted relative risk for recurrent bleeding with resumption of warfarin treatment was 0.45 (95% CI, 0.26-0.76), and resumption of treatment was associated with a lower adjusted relative risk for all stroke (AHR, 0.36; 95% CI, 0.16-0.84) and death (AHR, 0.35; 95% CI, 0.23-0.52).

Sensitivity and Additional Analyses

To investigate whether the main analyses were biased (eg, confounding by indication), 2 propensity-matched populations were constructed according to cause of ICH. For patients with an index event of hemorrhagic stroke, the population was confined to 675 patients, including 331 with no warfarin treatment and 203 who resumed warfarin treatment within 10 weeks after hospital discharge. The propensity-matched HR for ischemic stroke or SE was 0.58 (95% CI, 0.33-0.99) for those who resumed warfarin treatment compared with no resumption as the reference. For recurrent ICH, the HR was 2.24 (95% CI, 1.10-4.58). Thus, resumption of warfarin treatment entailed a lower relative risk for ischemic events at the expense of an increased risk for recurrent ICH.

For patients with an index event of traumatic ICH, the propensity-matched population included 347 with no resumption of warfarin treatment and 203 who resumed warfarin treatment. The propensity-matched HRs were 0.60 (95% CI, 0.28-1.30) for ischemic stroke or SE and 1.05 (95% CI, 0.49-2.23) for recurrent ICH. In general, these results are in line with the main analyses; however, owing to the smaller population (due to matching criteria), the CIs were wide and may preclude statistically significant conclusions.

The choice of a 10-week period for allocation of treatment groups was additionally investigated, using the mean time to prescription claim within the first 3 months after hospital discharge (mean of 31 days); this sensitivity analysis did not materially affect the results. When the composite outcome of stroke or SE and death was examined, the AHRs with resumption of warfarin treatment for patients with prior hemorrhagic stroke was 0.62 (95% CI, 0.47-0.82) and with traumatic ICH was 0.37 (95% CI, 0.26-0.52). These associations were mainly driven by the outcome of all-cause mortality because this event occurred more than 5 times more frequently than thromboembolic events.

Discussion

In this observational cohort study of patients with AF who sustained an ICH while receiving warfarin treatment, our principal findings were 3-fold. First, patients with a first-time hemorrhagic stroke had a poor prognosis, but resumption of warfarin treatment was associated with a lower rate of ischemic stroke or SE, a higher rate of recurrent ICH, and significantly lower mortality. Second, patients with an incident traumatic ICH had a similarly poor prognosis and lower rate of ischemic stroke or SE with resumption of warfarin treatment but lower rates of recurrent ICH and mortality after resumption of warfarin treatment. Third, the relative risk for recurrent ICH associated with resumption of warfarin treatment was higher among patients with a hemorrhagic stroke than observed among patients with a traumatic ICH event in this cohort.

Patients with AF and a high thromboembolic risk who sustain an ICH (due to either cause) present a clinical dilemma and treatment conundrum. We found that patients presenting with an ICH carry a different prognosis in terms of associated outcomes from resuming warfarin treatment according to the cause of the bleeding. In a previous investigation of the prognosis of head trauma,17 preinjury use of warfarin was associated with worse outcomes in comparison with no use of warfarin. For those receiving warfarin, regardless of cause, resumption of treatment resulted in a favorable prognosis in terms of a reduction in rates of ischemic and hemorrhagic stroke.18 Although the finding of an attenuated risk for hemorrhagic stroke among patients resuming warfarin treatment could be attributable to some residual confounding, our study confirms prior observations by Albrecht et al18 and extends the findings to a population of patients with AF. A very similar direction of associations can be observed with an adjusted odds ratio of 0.70 (95% CI, 0.52-0.95) vs an AHR of 0.45 (95% CI, 0.26-0.76).

In a cohort of patients with AF who shared risk factors for ischemic and hemorrhagic stroke, McGrath et al19 found that none of these risk factors was associated with hemorrhagic stroke or ischemic stroke. They observed that age was a stronger predictor of ischemic stroke than hemorrhagic stroke, which indicates a more favorable risk profile if patients with AF are treated with warfarin according to guideline recommendations.20,21 Whether these observations can be generalized to all patients with AF who sustain an ICH event remains to be fully investigated, but our results signal that these findings might apply to those who experience a traumatic ICH event. Ongoing prospective registries and randomized clinical trials will add to the current knowledge on whether to resume warfarin treatment after specific subtypes of ICH events.22-24

Limitations

The observational nature of the study should be considered when interpreting our results. The data are based on administrative databases primarily used for reimbursement, and thus not all clinically relevant and important variables were available. The result of a 50% reduction in mortality associated with resumption of warfarin treatment requires confirmation in other studies. Indeed, all-cause mortality associated with treatment exposure in observational studies requires careful consideration.25 Nonetheless, some deaths in real-world observational studies could also be from undiagnosed (fatal) strokes because postmortem autopsy or cerebral imaging is not mandated.

We had no information on intensity of warfarin treatment and could not access the location and volume of the hematoma. These variables are crucial to capture the severity of the ICH event and could bias our analysis in terms of associations with resumption of OAC treatment (ie, residual confounding by indication bias). Clearly, spontaneous intracerebral hemorrhage is a heterogeneous disease ranging from macrovascular lesions to small vessel disease; we were unable to capture these variations, and thus our associations in relation to warfarin treatment effects may reflect these discrepancies in prognosis from the original bleeding event. Aside from OAC treatment, left atrial appendage closure may be an option for stroke prevention in this frail population; however, randomized clinical trials are warranted before this treatment alternative may be recommended.

Ascertainment on medication use and stratification to resumption of warfarin treatment was based on filled prescriptions. Thus, we risked including patients not receiving warfarin treatment at baseline because our inclusion criterion was chosen as a prescription claim within 180 days before the index event. In addition, we did not assess the proportion of days covered by warfarin treatment remaining from before the index event. This process may have misclassified some patients as not resuming warfarin treatment while they indeed did resume treatment before claiming a new prescription of warfarin. The propensity-matched analysis allowing 10 weeks to resume treatment did not lead to materially changed effect sizes, although we found marginal statistically significant results for patients sustaining a hemorrhagic stroke.

We did not have access to imaging data in this study; thus, the analyses did not include information on specific subtypes of ICH. In particular, information on small bleeds (obtained from imaging data) could have strengthened our analyses, because physicians treating stroke could be cautious about recommending resumption of warfarin treatment in patients with evidence of clinically important small bleeds.26,27 We only considered primary coded diagnoses as events and required patients to be admitted to the hospital with a minimum stay of 24 hours. Although these criteria may not equate to the severity of the index diagnosis, they inform the adjustment model, ensuring that this information was not completely left out of the relative risk assessments.

Conclusions

Spontaneous hemorrhagic stroke and trauma-induced ICH confer different prognoses in patients with AF, and recommendations on resumption of warfarin treatment should consider this difference. Warfarin treatment resumption after a spontaneous hemorrhagic stroke event was associated with a lower rate for subsequent ischemic events, whereas the relative risk for recurrent ICH was increased; however, statistical uncertainty precludes firm conclusions of excess harm associated with treatment. Resumption of OAC therapy in patients with traumatic ICH was associated with a lower rate of ischemic events and a lower relative risk for recurrent ICH despite resumption of warfarin treatment. In both groups, warfarin resumption was associated with a lower risk for death within the first year after the event.

Back to top
Article Information

Corresponding Author: Peter Brønnum Nielsen, PhD, Aalborg Thrombosis Research Unit, Department of Clinical Medicine, Faculty of Health, Aalborg University, Søndre Skovvej 15, 9000 Aalborg, Denmark (pbn@rn.dk).

Accepted for Publication: November 27, 2016.

Published Online: February 20, 2017. doi:10.1001/jamainternmed.2016.9369

Author Contributions: Dr Nielsen had full access to all the data in this study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Nielsen, Larsen, Lip.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: All authors.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Nielsen, Skjøth.

Obtained funding: Larsen.

Administrative, technical, or material support: Nielsen, Larsen.

Study supervision: Larsen, Lip.

Conflict of Interest Disclosures: Dr Nielsen reports serving as a speaker for Boehringer Ingelheim, consulting for Bayer Pharma AG, and receiving an unrestricted research grant from BMS/Pfizer. Dr Skjøth reports consulting for Bayer Pharma AG. Dr Lip reports membership in and/or reviewing for various guidelines and position statements from the European Society of Cardiology, European Heart Rhythm Association, and National Institute for Health and Care Excellence, among others; serving on steering committees for various phase 2 and 3 studies, health economics, and outcomes research; serving as an investigator in various clinical trials in cardiovascular disease, including antithrombotic therapies in atrial fibrillation, acute coronary syndrome, and lipids; consulting for Bayer/Janssen, BMS/Pfizer, Biotronik, Medtronic, Boehringer Ingelheim, Microlife, and Daiichi-Sankyo; and serving as a speaker for Bayer, BMS/Pfizer, Medtronic, Boehringer Ingelheim, Microlife, Roche, and Daiichi-Sankyo. Dr Larsen reports serving as an investigator for Janssen Scientific Affairs, LLC, and Boehringer Ingelheim and on the speaker bureaus for Bayer, BMS/Pfizer, Roche Diagnostics, Boehringer Ingelheim, and Takeda Pharma. No other disclosures were reported.

Funding/Support: This study was supported in part by an unrestricted grant from the Obel Family Foundation.

Role of the Funder/Sponsor: The sponsor had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Additional Contributions: The Danish Health Data Authority provided the data material.

References
1.
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(12):857-867.PubMedGoogle ScholarCrossref
2.
January  CT, Wann  LS, Alpert  JS,  et al; American College of Cardiology/American Heart Association Task Force on Practice Guidelines.  2014 AHA/ACC/HRS guideline for the management of patients with atrial fibrillation: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines and the Heart Rhythm Society.  J Am Coll Cardiol. 2014;64(21):e1-e76.PubMedGoogle ScholarCrossref
3.
Kirchhof  P, Benussi  S, Kotecha  D,  et al.  2016 ESC Guidelines for the management of atrial fibrillation developed in collaboration with EACTS.  Eur Heart J. 2016;37(38):2893-2962.PubMedGoogle ScholarCrossref
4.
Camm  AJ, Lip  GYH, De Caterina  R,  et al; ESC Committee for Practice Guidelines-CPG; Document Reviewers.  2012 focused update of the ESC guidelines for the management of atrial fibrillation: an update of the 2010 ESC guidelines for the management of atrial fibrillation—developed with the special contribution of the European Heart Rhythm Association.  Europace. 2012;14(10):1385-1413.PubMedGoogle ScholarCrossref
5.
Fang  MC, Go  AS, Chang  Y,  et al.  Thirty-day mortality after ischemic stroke and intracranial hemorrhage in patients with atrial fibrillation on and off anticoagulants.  Stroke. 2012;43(7):1795-1799.PubMedGoogle ScholarCrossref
6.
Wolff  A, Shantsila  E, Lip  GYH, Lane  DA.  Impact of advanced age on management and prognosis in atrial fibrillation: insights from a population-based study in general practice.  Age Ageing. 2015;44(5):874-878.PubMedGoogle ScholarCrossref
7.
van Asch  CJ, Luitse  MJ, Rinkel  GJ, van der Tweel  I, Algra  A, Klijn  CJ.  Incidence, case fatality, and functional outcome of intracerebral haemorrhage over time, according to age, sex, and ethnic origin: a systematic review and meta-analysis.  Lancet Neurol. 2010;9(2):167-176.PubMedGoogle ScholarCrossref
8.
Pugh  D, Pugh  J, Mead  GE.  Attitudes of physicians regarding anticoagulation for atrial fibrillation: a systematic review.  Age Ageing. 2011;40(6):675-683.PubMedGoogle ScholarCrossref
9.
Chao  T-F, Liu  C-J, Liao  J-N,  et al.  Use of oral anticoagulants for stroke prevention in patients with atrial fibrillation who have a history of intracranial hemorrhage.  Circulation. 2016;133(16):1540-1547.PubMedGoogle ScholarCrossref
10.
Nielsen  PB, Larsen  TB, Skjøth  F, Gorst-Rasmussen  A, Rasmussen  LH, Lip  GYH.  Restarting anticoagulant treatment after intracranial hemorrhage in patients with atrial fibrillation and the impact on recurrent stroke, mortality, and bleeding: a nationwide cohort study.  Circulation. 2015;132(6):517-525.PubMedGoogle ScholarCrossref
11.
Lynge  E, Sandegaard  JL, Rebolj  M.  The Danish National Patient Register.  Scand J Public Health. 2011;39(7)(suppl):30-33.PubMedGoogle ScholarCrossref
12.
Kildemoes  HW, Sørensen  HT, Hallas  J.  The Danish National Prescription Registry.  Scand J Public Health. 2011;39(7)(suppl):38-41.PubMedGoogle ScholarCrossref
13.
Pedersen  CB.  The Danish Civil Registration System.  Scand J Public Health. 2011;39(7)(suppl):22-25.PubMedGoogle ScholarCrossref
14.
Krarup  L-H, Boysen  G, Janjua  H, Prescott  E, Truelsen  T.  Validity of stroke diagnoses in a national register of patients.  Neuroepidemiology. 2007;28(3):150-154.PubMedGoogle ScholarCrossref
15.
Wildenschild  C, Mehnert  F, Thomsen  RW,  et al.  Registration of acute stroke: validity in the Danish Stroke Registry and the Danish National Registry of Patients.  Clin Epidemiol. 2013;6:27-36.PubMedGoogle ScholarCrossref
16.
Stoker  TB, Evans  NR.  Managing risk after intracerebral hemorrhage in concomitant atrial fibrillation and cerebral amyloid angiopathy.  Stroke. 2016;47(7):e190-e192.PubMedGoogle ScholarCrossref
17.
Collins  CE, Witkowski  ER, Flahive  JM, Anderson  FA  Jr, Santry  HP.  Effect of preinjury warfarin use on outcomes after head trauma in Medicare beneficiaries.  Am J Surg. 2014;208(4):544-549.e1.PubMedGoogle ScholarCrossref
18.
Albrecht  JS, Liu  X, Baumgarten  M,  et al.  Benefits and risks of anticoagulation resumption following traumatic brain injury.  JAMA Intern Med. 2014;174(8):1244-1251.PubMedGoogle ScholarCrossref
19.
McGrath  ER, Kapral  MK, Fang  J,  et al; Investigators of the Registry of the Canadian Stroke Network.  Which risk factors are more associated with ischemic stroke than intracerebral hemorrhage in patients with atrial fibrillation?  Stroke. 2012;43(8):2048-2054.PubMedGoogle ScholarCrossref
20.
Nielsen  PB, Larsen  TB, Skjøth  F, Overvad  TF, Lip  GYH.  Stroke and thromboembolic event rates in atrial fibrillation according to different guideline treatment thresholds: a nationwide cohort study.  Sci Rep. 2016;6:27410.PubMedGoogle ScholarCrossref
21.
Lip  GYH, Nielsen  PB.  Should patients with atrial fibrillation and 1 stroke risk factor (CHA2DS2-VASc score 1 in men, 2 in women) be anticoagulated? yes: even 1 stroke risk factor confers a real risk of stroke.  Circulation. 2016;133(15):1498-1503.PubMedGoogle ScholarCrossref
22.
Charidimou  A, Wilson  D, Shakeshaft  C,  et al.  The Clinical Relevance of Microbleeds in Stroke study (CROMIS-2): rationale, design, and methods.  Int J Stroke. 2015;10(Suppl A100):155-161.PubMedGoogle ScholarCrossref
23.
Al-Shahi Salman  R, Bell  S.  The contemporary conundrum of antithrombotic drugs after intracerebral haemorrhage.  ACNR. 2015;15(3):12-15.Google Scholar
24.
van Nieuwenhuizen  KM, van der Worp  HB, Algra  A,  et al; APACHE-AF Investigators.  Apixaban versus antiplatelet drugs or no antithrombotic drugs after anticoagulation—Associated Intracerebral Haemorrhage in Patients With Atrial Fibrillation (APACHE-AF): study protocol for a randomised controlled trial.  Trials. 2015;16:393.PubMedGoogle ScholarCrossref
25.
Glynn  RJ, Knight  EL, Levin  R, Avorn  J.  Paradoxical relations of drug treatment with mortality in older persons.  Epidemiology. 2001;12(6):682-689.PubMedGoogle ScholarCrossref
26.
Lovelock  CE, Cordonnier  C, Naka  H,  et al; Edinburgh Stroke Study Group.  Antithrombotic drug use, cerebral microbleeds, and intracerebral hemorrhage: a systematic review of published and unpublished studies.  Stroke. 2010;41(6):1222-1228.PubMedGoogle ScholarCrossref
27.
Yates  PA, Villemagne  VL, Ellis  KA, Desmond  PM, Masters  CL, Rowe  CC.  Cerebral microbleeds: a review of clinical, genetic, and neuroimaging associations.  Front Neurol. 2014;4:205.PubMedGoogle ScholarCrossref
×