[ Original Research Cardiovascular Disease ] Intracranial Hemorrhage and Subsequent Ischemic Stroke in Patients With Atrial Fibrillation A Nationwide Cohort Study Peter Brønnum Nielsen , PhD ; Torben Bjerregaard Larsen , MD , PhD ; Anders Gorst-Rasmussen , PhD ; Flemming Skjøth , PhD ; Lars Hvilsted Rasmussen , MD , PhD ; and Gregory Y. H. Lip , MD BACKGROUND: The risk of ischemic stroke/thromboembolic events after an intracranial hemorrhage (ICH) in patients with atrial fi brillation (AF) who are on warfarin treatment is poorly characterized. Th e aim of this study was to assess the association between the risk of ischemic stroke/thromboembolic events and ICH. METHODS: Linkage of three Danish nationwide administrative registries in the period between 1999 and 2012 identified patients with AF on warfarin treatment. Event-rate ratios of stroke/thromboembolic events, major bleeding, and all-cause mortality stratifi ed by ICH were calculated, and Cox proportional hazard models were used to compare event rates among ICH survivors. A matched OR was calculated for ICH occurrences within 0 to 3 months relative to the 3 to 6 months prior to a stroke/thromboembolic event. A rate ratio of claimed warfarin prescriptions in a 3-month period pre- and post-ICH was also calculated. RESULTS: We studied 58,815 patients with AF (median age, 72.6 years; 60% men). When compared with the non-ICH group, the ICH group was at increased risk for stroke/systemic embolism/transient ischemic attack (TIA) (rate ratio, 3.67; 95% CI, 3.12-4.31) and mortality (rate ratio, 5.55; 95% CI, 5.20-5.92), but not for major bleeding (rate ratio, 1.06; 95% CI, 0.81-1.39). Th e matched OR of ICH occurrences prior to a stroke/systemic embolism/TIA was 4.33 (95% CI, 2.44-8.15). Th e rate ratio of claimed warfarin prescriptions post- and pre-ICH event was 0.28 (95% CI, 0.24-0.33). CONCLUSIONS: In this large-scale study of patients with AF treated with warfarin, fi rst-time ICH was associated with an increased rate of ischemic stroke/systemic embolism/TIA and mortality compared with the non-ICH group. Th ere was a decrease in the warfarin-prescription purchase rate in the post-ICH period compared with pre-ICH, which may partly explain the excess risk. CHEST 2015; 147(6):1651 - 1658 Manuscript received August 25, 2014; revision accepted October 31, Skjøth, and Rasmussen and Prof Lip), Department of Clinical Medicine, 2014; originally published Online First November 20, 2014. Faculty of Health, Aalborg University, Aalborg, Denmark; and University of Birmingham Centre for Cardiovascular Sciences (Prof Lip), City ABBREVIATIONS: AF 5 atrial fi brillation; CHA 2 DS2 -VASc 5 cardiac fail- ure or dysfunction, hypertension, age Ն 75 years (doubled), diabetes, Hospital, Birmingham, England . stroke (doubled)-vascular disease, age 65 to 74 years, and sex category FUNDING/SUPPORT : Th e Obel Family Foundation supported this study. (female); ICD-10 5 International Classifi cation of Diseases, 10th Revision ; CORRESPONDENCE TO: Gregory Y. H. Lip, MD, University of ICH 5 intracranial hemorrhage; INR 5 international normalized ratio; Birmingham Centre for Cardiovascular Sciences, City Hospital, Dudley NOAC 5 non-vitamin K oral anticoagulant; OAC 5 oral anticoagulant; Rd, Birmingham, B18 7QH, England; e-mail: [email protected] 5 5 TIA transient ischemic attack; VKA vitamin K antagonist © 2015 AMERICAN COLLEGE OF CHEST PHYSICIANS. Reproduction of AFFILIATIONS : From the Department of Cardiology (Drs Nielsen, this article is prohibited without written permission from the American Larsen, Gorst-Rasmussen, Skjøth, and Rasmussen), Atrial Fibrillation College of Chest Physicians. See online for more details. Study Group, Aalborg University Hospital, Aalborg, Denmark; Aalborg DOI: 10.1378/chest.14-2099 Thrombosis Research Unit (Drs Nielsen, Larsen, Gorst-Rasmussen, journal.publications.chestnet.org 1651 Downloaded From: http://journal.publications.chestnet.org/ by Veronica Wilson on 10/08/2015 Patients with atrial fi brillation (AF) have an increased In clinical practice, an ICH event necessitates temporary risk for ischemic stroke, and stroke-prophylactic treatment cessation of OAC therapy, requiring the physician to with oral anticoagulants (OACs) is recommended if one balance the risk of an additional bleeding event with the stroke risk factor is present. 1 Treatment with antithrom- risk of a thromboembolic event in the absence of OAC botic agents such as the vitamin K antagonists (VKAs) treatment.11 Th e preferred timing of OAC therapy resump- (eg, warfarin) reduces stroke by 64% and all-cause mor- tion aft er an ICH event has been much debated,12,13 tality by 26%, when compared with placebo or a control. 2 and suggestions diff er from 7 to 14 days to 30 weeks. 14-16 However, patients on warfarin therapy are also at a In practice, however, the timing of OAC therapy resump- higher risk of bleeding, the most severe being spontaneous tion is diffi cult, and attempts to err on the side of intracranial hemorrhage (ICH). Indeed, anticoagulation- caution may lead to general undertreatment (or non- associated intracerebral hemorrhage has been reported treatment), resulting in an increased risk for further to account for 20% of all ICH events. 3 fatal and disabling stroke among survivors of ICH. Th e anxiety of causing serious bleeding (particularly ICH) Th e objective of this study was to investigate outcomes can lead physicians to withhold anticoagulants in some of ischemic stroke and thromboembolic events associ- patients with AF, despite a greater benefi t from OAC ated with ICH among patients with AF. We hypothe- treatment (compared with no OAC or aspirin treatment) sized that patients experiencing an ICH event would be in terms of reduced risk of stroke. 4-6 More recently, the at increased risk for thromboembolism, as refl ected by a non-VKA OACs (NOACs, previously referred to as new composite of ischemic stroke, systemic embolism, and or novel OACs 7 ) have been shown to be associated with transient ischemic attack (TIA), and by all-cause mor- lower rates of ICH compared with VKAs, improving tality. Further, we hypothesized that this increase in safety, in addition to their effi cacy and convenience. 8-10 risk for thromboembolism among survivors of ICH Nonetheless, a history of ICH was an exclusion criterion would be accompanied by a decreased rate of warfarin- for patients included in recent trials of NOACs compared prescription purchase. We investigated these hypotheses with VKAs. in a large, nationwide cohort study. Materials and Methods was the composite of ischemic stroke/systemic embolism/TIA (ICD-10 diagnosis codes I63, I64, I74, G45). Secondary analyses were done on Registry Data Sources and Study Population major bleeding, as well as all-cause mortality. Th e defi nition of major All Danish residents are given a unique and permanent registration bleeding included acute posthemorrhagic anemia, hemothorax, con- number at the time of birth or immigration. Nationwide registries enable junctival hemorrhage, retinal hemorrhage, vitreous hemorrhage, recur- retrieval of individual patient data on (1) discharge diagnosis classifi ed rent and persistent hematuria, menopausal and other perimenopausal by the International Classifi cation of Diseases, 10th Revision (ICD-10) disorders, hemorrhage from the respiratory passages, unspecified and admission and discharge dates (the Danish National Patient hematuria, and unclassifi ed hemorrhage (ICD-10 diagnosis codes D62, Registry); (2) dispensed prescriptions identified by the Anatomic J942, H113, H356, H431, N02, N95, R04, R31, R58). 20 Therapeutic Chemical classification code, date of purchase, package size and volume of the medication since 1994 (the Danish National We imposed a dynamic stratifi cation on the cohort by considering each Prescription Registry); and (3) information on sex, date of birth, emi- patient’s ICH status during follow-up. At the index date, all patients gration, and vital status (the Danish Civil Registration System). were ICH free. If a patient sustained an ICH, they would move irrevers- ibly to the ICH group. Survivors of ICH were defi ned on the basis of the By linking the three databases, 17-19 we established a cohort of patients vital status from the Danish Civil Registration System (ie, no change in with incident nonvalvular AF. Nonvalvular AF was defi ned as presence the vital status at the day of an ICH diagnosis). of ICD-10 diagnosis code I48, and baseline absence of mitral stenosis and mechanical heart valve (ICD-10 diagnosis codes: I05 and Z952-Z954). The cardiovascular comorbidity and risk for stroke at baseline were Specifi cally, we identifi ed all patients discharged with a fi rst-time hos- assessed by the CHA 2 DS2 -VASc (cardiac failure or dysfunction, hyper- pital diagnosis (index date) of nonvalvular AF in the period January 1, tension, age Ն 75 years [doubled], diabetes, stroke [doubled]-vascular 1999, to December 31, 2012. e-Appendix 1 gives detailed defi nitions disease, age 65-74 years, and sex category [female]) score, with higher 21 on outcomes and concomitant pharmacotherapy. We excluded patients scores indicating greater risk for stroke. Th e CHA2 DS2 -VASc score has who died or experienced an ICH event (ICD-10 diagnosis codes: previously been applied in similar nationwide cohorts and is the pre- I60-I62, S063C, S064-S066, I690-I692) within the fi rst 7 days aft er the ferred risk stratifi cation tool for patients with AF, as recommended by index date. Patients who had claimed a prescription of heparin or the European Society of Cardiology. 22-25 We also calculated a (modifi ed) low-molecular-weight heparin or non-VKAs 1 year prior to the index HAS-BLED score (hypertension, abnormal renal and/or liver function, date were also excluded. We further restricted the cohort to patients stroke, bleeding history or predisposition, labile international normal- who initiated warfarin treatment, defi ned as a warfarin prescription ized ratio, elderly [ .