DOI: 10.1161/CIRCULATIONAHA.113.004777

Warfarin Use and the Risk for Stroke and Bleeding in Patients with Atrial Fibrillation Undergoing Dialysis Running title: Shah et al.; Warfarin Use in AF Patients Undergoing Dialysis Mitesh Shah, MBBS, MSc1; Meytal Avgil Tsadok, PhD1; Cynthia A. Jackevicius, PharmD, MSc2; Vidal Essebag, MD, PhD3; Mark J. Eisenberg, MD, MPH4; Elham Rahme, PhD1; Karin H. Humphries, DSc5; Jack V. Tu, MD, PhD6; Hassan Behlouli, PhD1; Helen Guo, MSc7; Louise Pilote, MD, PhD1 1

Divisions of Clinical Epidemiology and General Internal Medicine, McGill Un University niv iver ersi er sity si ty H Health ealt ea lthh lt 2 Center, Montreal, Quebec, Canada; Dept of Pharmacy Practice and Administration, College of Pharmacy, Western University off Health Sciences, Pomona, CA; 3Division of Cardiology, McGill MccGi Gill ll University Uni n veers rsit ity Health Center, Montreal, Quebec; Que uebec; 4Divisions off C Cardiology a diology and Clinical ar Epid Ep Epidemiology, idem mio ologyy, Jewish Jewi Je wish sh General Gener e al a Hospital/McGill Hosspita tal/McG cGil ill University, Univ ver ersi s ty, Mo Montre Montreal, real al,, Queb Quebec; ebec e ; 5Di Division D v siion of vi 6 Carrdiology rd y, Un Univ iveersi iv sity si ty ooff Br Brit itis it ishh Co is Colu lumb lu m ia mb ia, Va V ncouve veer, B riti ri tish sh C o um ol umbbia;; In Institute Inst stit st itut it u e for ut for Cl Clin Clinical inic ni al Cardiology, University British Columbia, Vancouver, British Columbia; Eval Ev Evaluative a uative Scie Sciences, encces, T Toronto, oron nto o, On Onta Ontario; tari rio; o; Inst Institute tittute ooff He Health eal alth th P Policy, oliccy, Ma Management anageeme ement an and nd E Evaluation, vaaluaatioon, Facu Fa Faculty c lt cu lty off M Medicine, ed diccin inee, U University niivers vers rsit ityy of o T Toronto, oron ontto,, To Toro Toronto; ro ont ntoo; o; D Division iv vissio ionn ooff Ca Cardiology, ard dio iollogy, gy, S Schulich chullich hH Heart eart ea 7 Centre, Sunnybrook Health Sciences Centre, University Toronto, Toronto; Cent Ce ntre nt re,, Sunn re S unn nnyb ybro yb rook ro ok H e ltth Sc ea Scie ienc ie nces es C enttre, en re, Un Univ i er iv ersi sity si ty ooff To oro ront nto, nt o, T oron or on nto to;; Inst IInstitute nst stit itut it utee fo ut forr Clinical Clin Cl inic in ical ic al Evaluative Eva valu luat lu ativ at ivee Sciences, iv Scie Sc iennce ie cess, T Toronto, oron or onto on to, On to Onta Ontario, tari ta rioo, C ri Canada anad an adaa ad

Address for Correspondence: Louise Pilote, MD, MPH, PhD McGill University Health Centr 687 Pine Avenue West, V Building Montreal, Quebec, H3A 1A1 Canada Tel: 514-934-1934 ext. 44722 Fax: 514-934-8293 E-mail: [email protected] Journal Subject Codes: Etiology:[5] Arrhythmias, clinical electrophysiology, drugs, Etiology:[8] Epidemiology, Anticoagulants:[184] Coumarins, Stroke treatment - medical:[70] Anticoagulants, Treatment:[118] Cardiovascular pharmacology 1 Downloaded from http://circ.ahajournals.org/ by guest on January 2, 2015

DOI: 10.1161/CIRCULATIONAHA.113.004777

Abstract

Background—Current observational studies on warfarin use and the risk for stroke and bleeding in patients with atrial fibrillation (AF) undergoing dialysis found conflicting results. Methods and Results—We conducted a population-based retrospective cohort study of patients aged 65 years and older admitted to a hospital with a primary or secondary diagnosis of AF, in Quebec and Ontario, Canada from 1998 to 2007. The AF cohort was grouped into dialysis (hemodialysis and peritoneal dialysis) and non-dialysis patients and into warfarin and nowarfarin users according to the first prescription filled for warfarin within 30 days after AF hospital discharge. We determined the association between warfarin use and thee rrisk issk fo forr st stro roke ro ke stroke and bleeding in dialysis and non-dialysis patients. The cohort was comprised of 1,626 dialysis pa atiien ents ts aand nd 2204,210 04 4,210 ,2 210 non-dialysis patients. Among Amon ong dialysis patients, on patient n s, 46% nt 46% (756/1,626) patients patients werre re prescribed prescribeed warfarin. warf wa rffarin in n. Among Amoong Am ong dialysis dial di alys yssis patients, pattientts, wa arffar ariin uusers sers se rs hhad ad dm oree cong or ccongestive ongges e ti t vee hheart earrt ea were warfarin more failure faail ilur uree an ur andd di diab diabetes abeete etes es bbut ut lless ess pr es pprior iorr bl io blee bleeding eedi d ng eevent di ven entt co en comp compared mpar mp arred too th tthee no no-w no-warfarin -w warrfa fari rinn us ri uusers. erss. A er Among mongg mong ent nts, s, w a fa ar fari r n us ri uuse, e co e, comp mpar mp ared ed tto o no no-w -w warrfa fari r n us use, e, w as nnot o aassociated ot ssoc ss ocia oc i teed wi ia with th a lower riskk dialysis patie patients, warfarin compared no-warfarin was for stroke (adjusted hazard ratio (HR): 1.14, 95% confidence interval (CI): 0.78 to 1.67) but was associated with a 44% higher risk for bleeding (adjusted HR: 1.44, 95% CI: 1.13 to 1.85) after adjusting for potential confounders. Propensity score adjusted analyses yielded similar results. Conclusions—Our results suggest that warfarin use is not beneficial in reducing stroke risk but is associated with a higher bleeding risk in patients with AF undergoing dialysis.

Key words: atrial fibrillation, dialysis, warfarin, stroke, bleeding

2 Downloaded from http://circ.ahajournals.org/ by guest on January 2, 2015

DOI: 10.1161/CIRCULATIONAHA.113.004777

Introduction Patients with atrial fibrillation (AF) suffering from severe chronic kidney disease (CKD) have a higher risk for stroke and bleeding.1, 2 AF is the most common cardiac arrhythmia and is an independent risk factor for a new stroke.3, 4 Patients with AF suffering from severe CKD, which requires treatment with dialysis, have a five-fold higher risk for a new stroke.3, 4 AF is becoming increasingly prevalent among patients with severe CKD predisposing a patient to a much higher risk for a new stroke.1, 5, 6 Historically, warfarin, a vitamin K antagonist, has been considered the preferred anticoagulant for reducing the risk of stroke in most patients with AF.7 However, warfarin use has been shown to accelerate vascular calcification in CKD patients, which eventually may further increase the risk for ischemic stroke.5, 8-10 Therefore, unce uncertainty ceert rtai a nty st ai stil still illl il exists regarding whether warfarin confers similar protection to reduce the risk for stroke in 14 pa patients atiien ents ts w with ithh AF ssuffering it uf uffering from severe CKD.11-14

CKD is aalso lsso co considered onssid ider ered er ed aass an iindependent nddepeenddentt rrisk iskk ffactor acto acto or fo forr bbleeding lee eedi ding di g aand nd therefore, theere r fo fore ree, warfarin war arin warf suffering CKD use in use n patients pat atie ieent ntss su ufffer erin in ng fr from om m ssevere ever ev eree C er KD ccould ouuld l iincrease ncreeasse th thee ri rrisk skk ffor o bbleeding. or leeed edin ing. in g2M g. Moreover, oreo oveer, in in patients withh AF undergoing und nder ergo er g in go ng hemodialysis, hemo he m di mo diallys y is is,, it is is routine rout ro utin ut i e practice in prac pr acti ac t ce to t administer adm dmin inis in iste is t r heparin te hepa he pari pa rinn which ri could also increase the risk for bleeding.15 Current observational studies on warfarin use and the risk for stroke and bleeding in patients with AF undergoing dialysis present conflicting results.1, 8, 16, 17 Globally, due to lack of evidence from randomized controlled trials (RCTs), AF management guidelines have yet to make strong recommendations regarding anticoagulation management for patients with AF undergoing dialysis.18-23 Due to recognized limitations of warfarin use such as frequent blood monitoring for a therapeutic international normalized ratio (INR), numerous food and drug interactions, uncertainly regarding benefit for reducing stroke risk, and possible augmentation of

3 Downloaded from http://circ.ahajournals.org/ by guest on January 2, 2015

DOI: 10.1161/CIRCULATIONAHA.113.004777

bleeding risk, clinicians often raise concern about warfarin’s safety and effectiveness in patients with AF undergoing dialysis.7, 11-15 To enhance knowledge on this issue, we determined the association between warfarin use and the risk for stroke and bleeding in patients with AF undergoing dialysis in Quebec and Ontario, Canada.

Methods Study design We conducted a population-based retrospective cohort study of patients aged 65 years and older 200077, inn admitted to a hospital with a primary or secondary diagnosis of AF from 1998 too 2007, Quebec and Ontario, Canada. Residents in Quebec and Ontario have universal access to hospital caaree aand nd physician phy hyssiciian sservices ervices and those 65 years of of aage ge and older ha ave uuniversal nivversal prescription drug ni care have co overage. ve th his i study, stu tudy dy, we obtained dy obtai btaiine nedd in ins stituutiionaal rreview eviiew w bboard oaard ap pprovval ppro val from frrom M cGil cG illl il coverage. Forr this institutional approval McGill Unniv iver ersi er sity ty yF acullty acul y ooff Me M edi d ciine ne, Mo Mont ntre r al (Q re (Que uebe ue beec)) and nd ffrom rom ro m Su unnnyb y roook ok H eaalt lthh Scie S c encces University Faculty Medicine, Montreal (Quebec) Sunnybrook Health Sciences ont ntoo (O Ont n ar a io i ).. Centre, Toro Toronto (Ontario). Study Population and Data Sources Cohort formation has been described in detail elsewhere.24, 25 In brief, we identified patients with a primary or secondary diagnosis of AF according to the International Classification of Diseases – 9th/10th revision codes (427.3, 427.31, or 427.32 / I48) using the following hospital discharge abstract databases in Quebec and Ontario: Maintenance et Exploitation des Données pour l’Étude de la Clientèle Hospitalière and the Canadian Institute for Health Information Discharge Abstract Database, respectively. The primary (principal) diagnosis code is the main condition treated or investigated during the admission. However, up to seven diagnosis codes may be recorded by the

4 Downloaded from http://circ.ahajournals.org/ by guest on January 2, 2015

DOI: 10.1161/CIRCULATIONAHA.113.004777

hospital. The remaining diagnoses (secondary) are considered to be the subsidiary diagnoses. For patients with more than one eligible admission with an AF diagnosis, the date of the first admission with an AF diagnosis was considered the index date of entry into the study cohort. We determined patients’ baseline characteristics, outcome data, and drug prescriptions from linkage between hospital discharge, physician claims, prescription drug claims, and vital status databases in Quebec and Ontario (Supplemental Table 1). For the stroke and bleeding outcomes, we used data from emergency room (ER) visits in addition to the information from the hospital discharge databases. We used validated database codes (whenever possible) to determine stroke and bleeding outcomes.26-30 We used the physician claims databases maintained by la Régie de l’assur l’assurance urran nce m maladie alad al adie ad ie du Québec (RAMQ) and the Ontario Health Insurance Plan, which contain information on inand an nd ou out out-patient t-paati t-pa tiennt di dia diagnostic ag agnostic and therapeutic proce procedures. edur dures. We also uused s d th se the he R RAMQ AMQ and the Ontario Drug databases, which information D ruug ug Benefit Plan Pla lann drug druug claims dr cla laim imss da im data t ba ta b ses se s, w hich con hi ccontain ontai aiin in info f rm fo rmaatiion oon n ddispensed ispe is pens pe nsed ed d ooutpatient utppati ut pati tien en nt medications patients Drug me edi dica cati ca tion onns for for pa pati tien en ntss aaged gedd 65 yyears ears ea rs aand nd oolder. ld der e .D rugg pprescriptions resscri re ript ri ptio pt ions ns were werre identified id den nti tifi fied fi ed d from from rom these th hesse using drug These databases provide databases us sin ng dr rug iidentification denttif de ific icattio ic ionn nu nnumbers. m er mb e s. s T hesse pr he pprescription esscr crip ipti ip t on cclaims ti laaim imss da data taba ta baase sess pr prov o ide highly ov y accurate information on dispensed outpatient medications.31-33 We grouped the selected AF cohort into dialysis and non-dialysis patients according to the presence of three or more dialysis procedural codes (same or different codes for hemodialysis and peritoneal dialysis) within 12 months prior to AF hospitalization (database codes in Supplemental Table 2). Our three-code rule attempted to select patients undergoing maintenance dialysis. For all patients, we assessed demographic characteristics and comorbidities at and within one year prior to AF hospitalization using validated codes, whenever possible. We obtained information on the first prescription filled for warfarin, rate control drugs (ȕ-blockers,

5 Downloaded from http://circ.ahajournals.org/ by guest on January 2, 2015

DOI: 10.1161/CIRCULATIONAHA.113.004777

calcium channel blockers, and digoxin), rhythm control drugs (class Ia, Ic, and III antiarrhythmics), aspirin, clopidogrel, and NSAIDs (non-steroidal anti-inflammatory drugs) within 30 days after AF hospital discharge. We grouped dialysis and non-dialysis patients into warfarin and no-warfarin users. We selected a 30-day period to capture the majority of patients with the first prescription for warfarin after AF hospital discharge, while minimizing the potential for survival bias.34 Our follow-up period was started 30 days after AF hospital discharge (from the first day after the 30-day period). The outcomes of interest were the first hospital admission or ER visit for (i) stroke; or (ii) bleeding, at any point during follow-up period. We defined stroke as ischemic cerebrovascular disease including transient ischemic attack (TIA) and retinal infarct. infa faarcct. t We We did did not n t no nclude intracerebral hemorrhages in the stroke outcome because intracerebral hemorrhages include co oul uldd be a complication compllic icat a ion of warfarin use. We defined deffin ineed bleeding as intracerebral int n raace cerrebral re bleeding, could ggastrointestinal asttro r intestinal al bleeding, bleeed din i g, g, intraocular int ntra raoocu ra ula larr bbleeding, leeedinng, ng he ematu uri ria, a, aand nd uunspecified nspeci nspe ciifi f ed ed llocation ocat oc atio io on of bbleeding. leeed edin din i g. hematuria, Wee ccalculated alcu al cu ulaate tedd C HAD DS2 sc score scor oree by assigning or ass ssig gni n ng ng oone nee ppoint oinnt eeach oi acch fo forr con ccongestive onges nges esti tiive v hheart eartt fa failure, ailurre,, ailu CHADS hypertension n, ag agee • 775 5 ye yyears, ars, ar s, and nd ddiabetes, iabeete ia tes, s aand s, nd ttwo wo ppoints oint oi ntss fo nt forr hi hist s or oryy of sstroke/TIA; trok tr oke/ ok e/TI e/ TIA; TI A; tthe he hypertension, history CHADS2 scores ranged between 0-6.35 The CHADS2 score is a widely used clinical prediction score for estimating the risk for stroke and serves to guide clinicians in determining suitable usage of warfarin in AF.35 We also calculated the HAS-BLED risk stratification score, the clinical prediction score for estimating the risk for bleeding.36, 37 The HAS-BLED score is calculated assigning one point each for hypertension, abnormal renal function, abnormal liver function, history of stroke/TIA, history of bleeding, labile INR, age • 65 years, drug therapy (antiplatelet agents, NSAIDs), and alcohol intake.36, 37 Since our databases do not provide information on labile INR and alcohol intake, we calculated a modified HAS-BLED score, with

6 Downloaded from http://circ.ahajournals.org/ by guest on January 2, 2015

DOI: 10.1161/CIRCULATIONAHA.113.004777

a maximum score of seven rather than nine.36, 37 Statistical Analyses Descriptive analyses were used to compare demographic characteristics, comorbidities, and prescription for medications between warfarin and no-warfarin users in the group of dialysis and non-dialysis patients. We presented continuous variables as mean ± standard deviation and dichotomous variables as number (%). We calculated crude stroke and bleeding incidence rate (per 100 person-years) for the group of dialysis and non-dialysis patients. We also stratified crude stroke and bleeding incidence according to warfarin use, CHADS2 score (for stroke incidence rate), and HAS-BLED score (for bleeding incidence rate). Due to restrictions to access and merge databases, we did separate analyses in Quebec and Ontario, and then combined com mbine bine nedd study stud st udyy results esults from both the provinces. Results for descriptive analyses and incidence rate are weighted averages av ver erag agees ag es for for o results resul ultts ul ts from Quebec and Ontario. To determine detter ermi minee association mi asssoccia iati tion ti on between bet etw ween dialysis dialysiss status status us and and nd warfarin warfaari rinn filled f lled fi d prescription, pre resc sccri r ptio ptio ion,, w wee conducted multivariable logistic analysis. co ond nduc ucte uc tedd a mul m ultiv var ariiable ble lo ogi g st stic ic rregression egrress eg ress ssio ionn an io anal allyssis.. To ddetermine etter e mine mine association assoocia ociaati t on o between bet etweeen n warfarin usee and and the the risk ris i k for fo or stroke s ro st oke k and and d bleeding ble leed edin ed ingg inn the in the h group gro roup up of of dialysis d al di alys ysis ys is and and non-dialysis non on-d -d dia ialy lysis patients, we conducted multivariable Cox proportional regression analyses. In multivariable Cox proportional hazards models, we considered warfarin use vs. no-warfarin use as a time-fixed binary variable, where we assumed that patients who were prescribed warfarin within 30 days after AF hospital discharge remained on the same prescription throughout the follow up period. This approach is akin to an intention to treat analysis in RCTs.38 To account for the effect of potential confounders in the warfarin and stroke risk analyses, we adjusted for age (years), sex, and specific components of CHADS2 score (congestive heart failure, hypertension, diabetes, and history of stroke/TIA). In the warfarin and

7 Downloaded from http://circ.ahajournals.org/ by guest on January 2, 2015

DOI: 10.1161/CIRCULATIONAHA.113.004777

bleeding risk analyses, we adjusted for age (years), sex, and specific components of HAS-BLED score (liver disease, hypertension, history of stroke/TIA, history of bleeding, and use of aspirin, clopidogrel, or NSAIDs). For each patient in the warfarin or no-warfarin users, we derived a propensity score for receiving warfarin treatment from the following variables: age • 75 (years), sex, type of AF admission (primary diagnosis vs. secondary diagnosis), CHADS2 scores (1 and • 2), liver disease, congestive heart failure, hypertension, diabetes, history of stroke/TIA, history of bleeding, use of rate control drug, rhythm control drug, aspirin, clopidogrel, and NSAIDs. To verify the results of stroke and bleeding risk analyses, we performed Cox proportional regression analyses adjusted for a propensity score covariate.39 The propensity score is a good alternative to reduce educe bias when there is a riskk of statistical overfitting overffitting due to a low number of even eevents ven ntss pper er potential confounder (i.e. a low number of stroke and bleeding events in the dialysis group).40 Th he propensity proopen pr open ensi sitty score sco core r indicated the likelihood off receiving r ceiving warfarin re warfarrin gi ive venn that a particular patient The given related elaate ted charac characteristic cte terristticc iiss pres ppresent. res esen entt. en t.399 W Wee us use used ed m multivariable ultivvarriab blee llogistic ogisti ogi ticc rregression egr g essiionn m models odel od elss to el t dderive eriive er i ve individual ndi divi vidu vi d al propensity du pro rope peenssit ityy scores scor sc ores e for forr th thee gr grou group oupp of ou of dialysis dia i ly lyssis and a d non-dialysis an nonnonn-di diial alys ysis ys is patients, paatie atient n s, rrespectively. nt esppect es pecttiv vel elyy. y. Results Resu ultts ar aaree ex eexpressed pres pr e se sedd ass oodds ddss ra dd ratio ati t o (O (OR) R) ffor o llogistic or ogis og isti is tiic regression regr re gres gr e si sion on analysis ana naly lyssis ly i or or hazard haza ha z rd ratios (HRs) for Cox regression analyses with 95% confidence intervals (CIs). To combine results from Quebec and Ontario, we pooled the OR (or HR) for each predictor using a fixed-effects model, weighted for the inverse of the variance of the province-specific parameter estimate, ln (OR) [or ln (HR)].41 We performed all statistical analyses using SAS 9.2 (SAS Institute, Cary, NC, USA).

Results Baseline Characteristics The AF cohort includes 1,626 dialysis patients and 204,210 non-dialysis patients. Dialysis

8 Downloaded from http://circ.ahajournals.org/ by guest on January 2, 2015

DOI: 10.1161/CIRCULATIONAHA.113.004777

patients were younger, more likely to be men, and had more congestive heart failure, hypertension, diabetes, coronary artery disease, and past history of bleeding event, compared to non-dialysis patients (Table 1). A larger proportion of dialysis patients, compared to non-dialysis patients, had high risk score for stroke (CHADS2 score • 2: 72% (1,176/1,626) vs. 55% (112,049/204,210)) and bleeding (HAS-BLED score • 3: 85% (1,381/1,626) vs. 25% (50,203/204,210)). Prescription Pattern of Warfarin Comparable proportions of the dialysis patients and the non-dialysis patients filled a prescription for warfarin within 30 days after AF hospital discharge (46% (756/1,626) vs. 51% 103,473/204,210)). In the multivariable logistic regression model, dialysis statu us wa w asso soci so ciat ci ated (103,473/204,210)). status wass as associated with a lower proportion of filled prescriptions for warfarin (adjusted OR: 0.83, 95% CI: 0.74 to 0. .92 2). 0.92). Amon ng ddialysis ialys ysis i ppatients, atie at ient ie nts, nt s, tthose hose ho se w ho ffilled illled a pprescription reesc escrip crip ptiion ffor or w arfari rinn had had mo more re co ongeesttive onge Among who warfarin congestive hear he artt fa ar fail ilur urre an andd di iab abeetees es bbut utt le ess pprior riorr bbleeding rior leeedin le edin ng eve eevent ventt compared co omp mpar ared ar ed d to to the t e no-warfarin th no-w no-w warrfa fari rinn users. ri usser ers. s s. heart failure diabetes less Patients whoo filled fil ille l d a prescription le p es pr e crrip ipti t on for for w a fa ar fari rinn ha ri hhadd higher high hi g er proportion gh pro ropo port po rtio rt ionn of ppatients atie at ient ie ntts with w th the wi the high ris sk warfarin risk score for stroke (CHADS2 • 2: 77% (580/756) vs. 69% (596/870)) compared to the no-warfarin users but the proportion of high risk score for bleeding (HAS-BLED • 3: 84% (637/756) vs. 86% (744/870)) was similar between both the groups. Stroke Outcome Among dialysis patients, warfarin users did not have a lower crude incidence rate for stroke compared to the no-warfarin users (unadjusted incidence rate: 3.37 vs. 2.91 / 100 person-years; P = .44) (Table 2). On the contrary, among the non-dialysis patients, warfarin users did exhibit a lower crude incidence rate for stroke compared to the no-warfarin users (unadjusted incidence

9 Downloaded from http://circ.ahajournals.org/ by guest on January 2, 2015

DOI: 10.1161/CIRCULATIONAHA.113.004777

rate: 2.19 vs. 2.51 / 100 person-years; P < .001). After adjusting for potential confounders, warfarin use, compared to no-warfarin use, was not associated with a lower risk for stroke in dialysis patients (adjusted HR: 1.14, 95% CI: 0.78 to 1.67), however, it was associated with a 13% lower risk for stroke in non-dialysis patients (adjusted HR: 0.87, 95% CI: 0.85 to 0.90) (Table 3). We observed similar results when we performed propensity score adjusted Cox proportional regression analyses (dialysis patients – adjusted HR: 1.17, 95% CI: 0.79 to 1.75; non-dialysis patients – adjusted HR: 0.89, 95% CI: 0.87 to 0.92). Bleeding Outcome bleeding Among dialysis patients, warfarin users had a higher crude incidence rate for blee eeediing g eevent vent ve nt compared to the no-warfarin users (unadjusted incidence rate: 10.88 vs. 7.31 / 100 person-years; warfarin P < .001) .00 001) 00 1) (Table (Ta Tablee 2). 2). Similarly, among non-dialysis non-dialy ysis si patients, patients, warf far a in uusers sers se r had a higher crude incidence (unadjusted incidence rate: nciidence ratee for for bleeding bleeed e in ingg event eveent ev ent compared comp co mpaared to mp to thee no-warfarin no-w warfa arfaari rinn users users (u (una nadj na djuust usted ted in inci cide ci dencce ra de atee: 4.64 100 person-years; 4. .64 vvs. s 44.00 s. .0 00 / 10 00 pe per rsoonon-yeears earss; P < .001). .00 001) 1).. 1) no-warfarin was Afterr adjusting adju ad just ju stin st i g for in f r potential fo pote po t nt ntia i l confounders, ia c nf co n ou ound nder nd ers, er s, warfarin war a faari rinn use, use se,, compared c mp co mpar ared ar ed tto o no o-w war arfa farin use, wa fa as associated with a 44% and a 19% higher risk for bleeding event in dialysis patients (adjusted HR: 1.44, 95% CI: 1.13 to 1.85) and non-dialysis patients (adjusted HR: 1.19, 95% CI: 1.16 to 1.22), respectively (Table 3). We observed similar results when we performed propensity score adjusted Cox proportional regression analyses (dialysis patients – adjusted HR: 1.41, 95% CI: 1.09 to 1.81; non-dialysis patients – adjusted HR: 1.20, 95% CI: 1.17 to 1.23).

Discussion Our study indicates that in dialysis patients with AF, warfarin use, compared to no-warfarin use,

10 Downloaded from http://circ.ahajournals.org/ by guest on January 2, 2015

DOI: 10.1161/CIRCULATIONAHA.113.004777

did not reduce the risk for stroke but was associated with a 44% higher risk for bleeding event, while warfarin use in non-dialysis patients with AF was associated with a 13% lower risk for stroke and only a 19% higher risk for bleeding event. Thus, the risk-benefit profile does not appear to be favourable to support a recommendation of routine warfarin use for stroke reduction in dialysis patients with AF. Dialysis patients have several platelet and coagulation abnormalities and also have associated comorbidities such as uncontrolled hypertension and diabetes, which all contribute to an increase in the risk for stroke and bleeding.8, 42 Further, dialysis patients routinely receive heparin during dialysis procedure, which also increases the risk for bleeding.8, 42 Moreover, warfarin use in dialysis patients, through the inhibition of Matrix Gla protein and nd dG Gas-6, a -6 as -6,, ca cann accelerate vascular calcification, which eventually might increase the risk for ischemic stroke.5, 8810 0

These T hesse he se ffactors acto ac tors ccould to ou explain why, in our study ould study, y, w warfarin arfarin was not ott ass associated soc ociiated ia with a lower risk

ffor or ischemic is stroke str trok o e in dialysis dia iallysi lysiis patients, pati pa tien ti en nts ts, but but was wa rather raath her associated assoc ssocciaate tedd with with h an an increased in ncr crea ease seed risk risk for for or bleeding. bl lee eedi ding di ng.. ng We summarized sum umma mari ma rize ri zeed the th he results resuult re ltss of o our our current cur urre rent re nt study stu udy and and evidence evi vide denc de ncce fr from om pprevious reevi viou ouss pu ou ppublished blished studies of warfarin use and the risk for stroke and bleeding in patients with AF undergoing dialysis in Figure 1 and Figure 2, respectively. In a retrospective cohort study of 1,671 AF patients undergoing haemodialysis, Chan et al. observed a 1.9 fold higher risk for the composite stroke/death outcome with warfarin use.8 In another observational study analysing data from the international Dialysis Outcomes and Practice Patterns Study (DOPPS), Wizemann et al. stratified patients with AF undergoing haemodialysis according to age categories, ” 65, 65-75, and > 75 years.17 The authors reported that warfarin use in patients > 75 years (n=1,107) was associated with a 2.2 fold higher risk for the composite stroke/death outcome.17 In the younger

11 Downloaded from http://circ.ahajournals.org/ by guest on January 2, 2015

DOI: 10.1161/CIRCULATIONAHA.113.004777

age groups, the authors noticed that warfarin use did not reduce the risk for the composite stroke/death outcome.17 Winkelmayer et al. conducted a retrospective cohort study in hemodialysis patients with incidental AF and performed propensity score matched analyses (warfarin user: 237; matched nonusers: 948).16 The authors found that warfarin use did not reduce the risk for ischemic stroke but was associated with a 2.4 fold higher risk for hemorrhagic stroke.16 Olesen et al., in 901 dialysis patients, was the only one to observe that warfarin use was associated with a 56% decrease in the risk for the composite stroke/death outcome.1 However, there were several limitations to this study.1, 43-45 A larger proportion of dialysis patients had unusually low HAS-BLED score (HAS-BLED score: 2 – 35% (312/901); 0 or 1 – 43% (390/901)). 390/901)).1, 44 Contrary to our study and DOPPS, dialysis patients had low prevalence prev valen en ncee of of diabetes (14% (129/901)) and hypertension (54% (486/901)).1, 43 It is possible that a selection biiass of of healthier heal he alth al thierr patients th pati pa t ents undergoing dialysis co ould uld explain the rea eaasonn fo forr the decreased risk of bias could reason 43 stroke tro oke with warfarin waarf rfar arin in uuse see iin n the th he study stud st udyy by ud by Olesen Oleeseen ett aal. l.1, 43

Cont Co Contrary ntrrary nt rary y too our ouur results result res ltts of increased inc ncre reas ased as ed bleeding bleeed e ing g risk risk in dialysis dia ialy ly ysiis patients, pattieents pa entss, Winkelmayer Wink Wink nkeelm elmay mayer yer eett aal. l l. and Olesen et aal. l. observed obse ob serv se rved rv d nno o as association sso soci ciat ci a io ionn be betw between tw wee eenn wa w warfarin rffar arin in us usee an aand d th the he ri risk sk ffor o ggastrointestinal or astr as trroi o ntestinal bleeding and the composite bleeding/death outcome, respectively.1, 16 A major consideration when comparing our study results for stroke and bleeding risk with previous studies is the heterogeneity in stroke and bleeding definitions across the different studies.1, 8, 16, 17 In our study, we included ischemic stroke, TIA, and retinal infarct in stroke definition and excluded intracerebral hemorrhages. Contrary to the composite stroke/death and bleeding/death outcomes in previous studies,1, 8, 17 we did not include death in our stroke and bleeding definitions. Our study has a number of strengths. Our large sample size allowed us to study the

12 Downloaded from http://circ.ahajournals.org/ by guest on January 2, 2015

DOI: 10.1161/CIRCULATIONAHA.113.004777

association between warfarin use and the risk for stroke and bleeding in the AF cohort. Unlike other studies, we attempted to reduce concerns about statistical overfitting46 and included only the most relevant covariates in the adjusted analyses. Finally, the information available within the large Quebec and Ontario health care databases reflects routine clinical practice in Canada, and may be less prone to participation biases that can arise in other types of studies.47 There are some limitations to our study. First, biases due to residual confounding from unknown or unmeasured confounders and also confounding by indication are well described in observational studies on drug effects.48, 49 To overcome confounding bias, we adjusted for most appropriate covariates that may confound the association between warfarin use and the study outcomes, and also performed sensitivity analyses using the propensity score approach. appproa proaach h.39, 50 However, we still cannot rule out residual confounding.48 Second, our health administrative da ataaba basses ses do not nott contain contain on leve velss or heparin use ve usse during duuriing dialysis procedure and databases information on INR levels therefore herref e ore we ccould ou uld d nnot ot aaccount cccou ount nt ffor orr tthese hese he se var variables riaabless iin n the he aadjusted djuuste dj ustedd an anal analyses. allyses es.. Th Thir Third, ird, d tthe d, he aaccuracy cccur uraacy a cy off database dat atab a as ab asee codes code co des for f r patients’ fo paati tieents ts’ related ts rel elat ateed at ed health hea ealt lthh information lt in nfo form rmat rm atio ionn is a known known concern con once cern rn in in observational obsserv ob vattio ionnal na studies tudies basedd on hhealth ealt ea lthh ad admi administrative mini mi n st stra rati ra tiive v ddatabases. attab abas ases as e . IIn es n at aattempt teemp mptt to llimit imit im it tthis h s co hi conc concern, n er nc e n, n w wee us uused ed database codes with the best validation whenever possible. In summary, current and previous observational studies on warfarin use and the risk for stroke and bleeding in patients with AF undergoing dialysis failed to provide much evidence in favor of warfarin use, yet there was a signal for an increased bleeding risk.1, 8, 16, 17 Recently, the Canadian Cardiovascular Society (CCS) AF guidelines (2012) made conditional recommendation (on low quality of evidence) that patients with AF undergoing dialysis should not routinely receive anticoagulation treatment for primary prevention of stroke.23 This is consistent with the recommendation from the Kidney Disease: Improving Global Outcomes

13 Downloaded from http://circ.ahajournals.org/ by guest on January 2, 2015

DOI: 10.1161/CIRCULATIONAHA.113.004777

(KDIGO).22 Nevertheless, with no evidence from RCTs, there is a lack of strong recommendations for anticoagulation management guidelines for this patient population.18-23 Due to the observational nature of our and previous studies, the study results may not be conclusive. We propose that a large multi-centre RCT should be undertaken to clarify this issue and to guide AF management guideline bodies around the world.

Acknowledgments: Dr. Pilote had full access to all of the data in Quebec and Dr. Jackevicius had full access to all of the data in Ontario and take responsibility for the integrity of the data and the accuracy of the data analysis. The opinions, results, and conclusions reported in this paper are those of the authors and are independent from the funding sources. No endorsement by Canadian nstitutes of Health Research (CIHR), Institute for Clinical Evaluative Sciences (I ICE CES) S),, or tthe S) he Institutes (ICES), shoul uldd be iinferred. nfer nf erre er red ed. Ontario Ministry of Health and Long-term Care (MOHLTC) is intended or should We thank la Régie de l’assurance maladie du Québec (RAMQ) and Brogan Inc., Ottawa for use its Drug Dru rugg Product Prodduc Pr uctt and Therapeutic Class Database. Databa basee. ba off its Funding Fun nd So our u cees:: Th his stu udy dy was a ssupported u poorted up d bby y th he CIHR CIIHR operating operaatinng grant grrannt nt MOP-84304. MOP OP-88433044. Funding Sources: This study the Inte In tere rest st Disclosures: Dissclos osur u es es:: All A l th Al thee au auth th hor o s de decl c arre no ccompeting cl ompe om peeti t ng iinterest. nter nt e es er estt. Dr Dr.. Sh S ah Conflict of Interest authors declare Shah reported receiving a fellowship award from FRQS (Fonds de recherche du Québec - Santé). Dr. Avgil Tsadok reported receiving a fellowship award from the CIHR. Dr. Essebag is the recipient of a Clinician Scientist award from the CIHR. Dr. Pilote reported serving as a James McGill Chair at McGill University and receiving a national investigator award from the Fonds de recherche en sante´ du Québec.

References: 1. Olesen JB, Lip GY, Kamper AL, Hommel K, Kober L, Lane DA, Lindhardsen J, Gislason GH, Torp-Pedersen C. Stroke and bleeding in atrial fibrillation with chronic kidney disease. N Engl J Med. 2012;367:625-635. 2. Poli D, Antonucci E, Zanazzi M, Grifoni E, Testa S, Ageno W, Palareti G. Impact of glomerular filtration estimate on bleeding risk in very old patients treated with vitamin K 14 Downloaded from http://circ.ahajournals.org/ by guest on January 2, 2015

DOI: 10.1161/CIRCULATIONAHA.113.004777

antagonists. Results of EPICA study on the behalf of FCSA (Italian Federation of Anticoagulation Clinics). Thromb Haemost. 2012;107:1100-1106. 3. Seliger SL, Gillen DL, Tirschwell D, Wasse H, Kestenbaum BR, Stehman-Breen CO. Risk factors for incident stroke among patients with end-stage renal disease. J Am Soc Nephrol. 2003;14:2623-2631. 4. Wolf PA, Abbott RD, Kannel WB. Atrial fibrillation as an independent risk factor for stroke: the Framingham Study. Stroke. 1991;22:983-988. 5. Clase CM, Holden RM, Sood MM, Rigatto C, Moist LM, Thomson BK, Mann JF, Zimmerman DL. Should patients with advanced chronic kidney disease and atrial fibrillation receive chronic anticoagulation? Nephrol Dial Transplant. 2012;27:3719-3724. 6. Wetmore JB, Ellerbeck EF, Mahnken JD, Phadnis M, Rigler SK, Mukhopadhyay P, Spertus JA, Zhou X, Hou Q, Shireman TI. Atrial fibrillation and risk of stroke in dialysis patients. Ann Epidemiol. 2013;23:112-118. 2011;365:10527. Mega JL. A new era for anticoagulation in atrial fibrillation. N Engl J Med. 20 011;;3665: 5 10 1052 52-52 1054. 8. Chan KE,, Lazarus JM, Thadhani R, Hakim RM. Warfarin use associates with increased risk for stroke fibrillation. Nephrol. fo or st stro roke ro ke iinn hemodialysis hemo modi mo dialysis patients with atrial fib brilllation. J Am SSoc o N oc ephhrol. ep hr 2009;20:22232233. 22 2333. Chan Lazarus Thadhani R,, H Hakim RM. Anticoagulant antiplatelet usage 9.. C h n KE,, La ha L zarus JM zar JM, Th hadhaanii R akiim RM M. An nti tico coaagu co agulan nt and an nti tipplaatel ate et us sage sag associates with mortality among hemodialysis Nephrol. as sso soci ciat ci a es w at ithh mo it mort rtal allit ityy am amon o g he on hemo modi mo dial di alys al ysiss ppatients. ys attieent n s. J Am Am SSoc occ N e hrool. ep ol 22009;20:872-881. 00 09;;20 20::8772--881 -881 1. 10. Reynoldss JL, JL, Joannides Joa oann nnid nn i ess AJ, AJ,, Skepper Ske kepp p er JN, pp JN, McNair McNa Nair irr R, R, Schurgers Schu Sc hurg hu rger rg e s LJ, er LJ, Proudfoot Prou Pr oudf ou dfoo oott D, Jahnenoo Jah a nenDechent W, W Weissberg PL, Shanahan CM. Hu Human muscle Dech De chen entt W eiss ei ssbe berg rg P L S hana ha naha hann CM Huma mann vascular vasc va scul ular ar ssmooth moot mo othh mu musc scle le ccells ells el ls uundergo nder nd ergo go vesicle-mediated calcification in response to changes in extracellular calcium and phosphate concentrations: a potential mechanism for accelerated vascular calcification in ESRD. J Am Soc Nephrol. 2004;15:2857-2867. 11. Ng KP, Edwards NC, Lip GY, Townend JN, Ferro CJ. Atrial Fibrillation in CKD: Balancing the Risks and Benefits of Anticoagulation. Am J Kidney Dis. 2013;62:615-632. 12. Reinecke H, Brand E, Mesters R, Schabitz WR, Fisher M, Pavenstadt H, Breithardt G. Dilemmas in the management of atrial fibrillation in chronic kidney disease. J Am Soc Nephrol. 2009;20:705-711. 13. Shen JI, Turakhia MP, Winkelmayer WC. Anticoagulation for atrial fibrillation in patients on dialysis: are the benefits worth the risks? Curr Opin Nephrol Hypertens. 2012;21:600-606. 14. Sood MM, Komenda P, Sood AR, Rigatto C, Bueti J. The intersection of risk and benefit: is warfarin anticoagulation suitable for atrial fibrillation in patients on hemodialysis? Chest.

15 Downloaded from http://circ.ahajournals.org/ by guest on January 2, 2015

DOI: 10.1161/CIRCULATIONAHA.113.004777

2009;136:1128-1133. 15. To AC, Yehia M, Collins JF. Atrial fibrillation in haemodialysis patients: do the guidelines for anticoagulation apply? Nephrology (Carlton). 2007;12:441-447. 16. Winkelmayer WC, Liu J, Setoguchi S, Choudhry NK. Effectiveness and safety of warfarin initiation in older hemodialysis patients with incident atrial fibrillation. Clin J Am Soc Nephrol. 2011;6:2662-2668. 17. Wizemann V, Tong L, Satayathum S, Disney A, Akiba T, Fissell RB, Kerr PG, Young EW, Robinson BM. Atrial fibrillation in hemodialysis patients: clinical features and associations with anticoagulant therapy. Kidney Int. 2010;77:1098-1106. 18. Anderson JL, Halperin JL, Albert NM, Bozkurt B, Brindis RG, Curtis LH, Demets D, Guyton RA, Hochman JS, Kovacs RJ, Ohman EM, Pressler SJ, Sellke FW, Shen WK, Wann LS, Curtis AB, Ellenbogen KA, Estes NA, III, Ezekowitz MD, Jackman WM, January CT, Lowe JE, Page RL, Slotwiner DJ, Stevenson WG, Tracy CM, Fuster V, Ryden LE, Cannom DS, Crijns HJ, Curtis AB, Ellenbogen KA, Le Heuzey JY, Kay GN, Olsson SB, Prystowsky EN,, Tamargo Tama Ta marg rg go JL, JL ACCF/AHA/ESC Wann S. Management of patients with atrial fibrillation (compilation of 2006 A CCF AH CCF/ AHA/ A/ES A/ ESC and 2011 ACCF/AHA/HRS recommendations): a report of the American College ge ooff Cardiology/American Heart Association Task Force on Practice Guidelines. J Am Coll Cardiol. 2013;61:1935-1944. 2013;6 ;61:1935-194 9 4. 19. AJ, Kirchhof Lip GY, Schotten Savelieva Ernst S,, Va Gelder Al-Attar N, 19 9. Camm Ca A J, K irch ir chho hoof P, L ip pG Y S Y, chot ch otte tenn U, S aveliiev evaa I, E rnst rn stt S Vann Ge Geld lder ld er IIC, C, A l At lAtta tarr N ta Hindricks G, Prendergast P,, De C CR, H inndricks nd Preend der erggast st B, B, Heidbuchel H id He dbu buch ch hell H, Alfieri Alfierri O, Angelini Anggel e in i i A, A, Atar Ataar D, D, Colonna Colon olonna n P na R, Dee S SJ, Goette Gorenek B,, H Heldal M,, Hoh Hohloser SH, Kolh P,, L Lee H Heuzey D J, Goett ttee A, G oreeneek B eld l all M hlooserr S H, K olhh P ol euuzey ey JJY, Y, Ponikowski Y, Ponnik kow wski P, P, Rutten FH. Guidelines the management fibrillation: Task Force for Ru utt t en F H.. G uiddel ui deline liness for for th he ma mana nage na geme ment me nt ooff at aatrial riiall fib ib bri rill llat ll atio on: n tthe he T askk Fo orc rcee fo or tthe hee Management Europace. Ma ana nage g me ment n of of Atrial Attri rial al Fibrillation Fib ibrrilllat atio ionn of thee European Euro Eu ropeean a Society Soc ocie iety ty of of Cardiology Card Ca rdio iolo logyy ((ESC). ESC) ES C). Eu uro ropaace ce. 2010;12:1360-1420. 2010;12:1360 60 0-1 142 20. 20. Camm AJ, Lip GY, De CR, Savelieva I, Atar D, Hohnloser SH, Hindricks G, Kirchhof P. 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;14:1385-1413. 21. Fuster V, Ryden LE, Cannom DS, Crijns HJ, Curtis AB, Ellenbogen KA, Halperin JL, Le Heuzey JY, Kay GN, Lowe JE, Olsson SB, Prystowsky EN, Tamargo JL, Wann S, Smith SC, Jr., Jacobs AK, Adams CD, Anderson JL, Antman EM, Halperin JL, Hunt SA, Nishimura R, Ornato JP, Page RL, Riegel B, Priori SG, Blanc JJ, Budaj A, Camm AJ, Dean V, Deckers JW, Despres C, Dickstein K, Lekakis J, McGregor K, Metra M, Morais J, Osterspey A, Tamargo JL, Zamorano JL. ACC/AHA/ESC 2006 Guidelines 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 European Society of Cardiology Committee for Practice Guidelines (Writing Committee to Revise the 2001 Guidelines for the Management of Patients With Atrial Fibrillation): developed in collaboration with the European Heart Rhythm Association and the Heart Rhythm Society. Circulation. 2006;114:e257-e354.

16 Downloaded from http://circ.ahajournals.org/ by guest on January 2, 2015

DOI: 10.1161/CIRCULATIONAHA.113.004777

22. Herzog CA, Asinger RW, Berger AK, Charytan DM, Diez J, Hart RG, Eckardt KU, Kasiske BL, McCullough PA, Passman RS, DeLoach SS, Pun PH, Ritz E. Cardiovascular disease in chronic kidney disease. A clinical update from Kidney Disease: Improving Global Outcomes (KDIGO). Kidney Int. 2011;80:572-586. 23. Skanes AC, Healey JS, Cairns JA, Dorian P, Gillis AM, McMurtry MS, Mitchell LB, Verma A, Nattel S. Focused 2012 update of the Canadian Cardiovascular Society atrial fibrillation guidelines: recommendations for stroke prevention and rate/rhythm control. Can J Cardiol. 2012;28:125-136. 24. Avgil TM, Jackevicius CA, Rahme E, Humphries KH, Behlouli H, Pilote L. Sex differences in stroke risk among older patients with recently diagnosed atrial fibrillation. JAMA. 2012;307:1952-1958. 25. Pilote L, Eisenberg MJ, Essebag V, Tu JV, Humphries KH, Leung Yinko SS, Behlouli H, Guo H, Jackevicius CA. Temporal Trends in Medication Use and Outcomes in Atrial Fibrillation. Can J Cardiol. 2013;29:1241-1248. 26. Cattaruzzi C, Troncon MG, Agostinis L, Garcia Rodriguez LA. Positive pred predictive ed dic icti tivee vvalue ti alue al ue ooff ICD-9th Informativo CD-9th codes for upper gastrointestinal bleeding and perforation in the Sistemaa In nfo form rm mat ativ ivoo Sanitario Regionale database. J Clin Epidemiol. 1999;52:499-502. 27. Goldstein LB. Accuracy coding identification 27 7. Go Gold ldst ld stei st einn LB ei B. A ccuracy of ICD-9-CM codin ng ffor or the identifi ica c tion onn ooff patients with acute ischemic stroke: modifier codes. Stroke. 1998;29:1602-1604. sch cheemic str trok oke: ok e: eeffect f ec ff ectt of m odif od iffie ierr co code des. s. St Stro roke ke. 19 998; 8;;29 29:1 :160 :1 6 22-16 1 04 16 04. 28. Henderson Shepheard Sundararajan V.. Qual Quality diagnosis 28 8. H enderso sonn T, T, S heph phheard rd d JJ,, Su Sund nddarrarrajan nV Q ualit itty of di iagnnosis aand nd pprocedure roce roce c durre ccoding oding iin n ICD-10 CDD-10 10 aadministrative dm min inis isttraativ ativee data ddata. ata. Med Med Care. Care Ca ree. 2006;44:1011-1019. 200 006; 6;44 44:1 44 :1 1011--10 019 19. AD. The 29. Kirkmann MA MA,, Mahattanakul M ha Ma hattan anak an akul ak ul W, W, Gregson Grreg egso sonn BA, so BA, Mendelow M nd Me del elow ow A D T D. he aaccuracy ccur cc u accy of hhospital ur o pital os discharge coding Belg. disc di scha harg rgee co codi ding ng ffor or hhemorrhagic emor em orrh rhag agic ic sstroke. trok tr okee Acta Acta Neurol Neu euro roll Be Belg lg 22009;109:114-119. 009; 00 9;10 109: 9:11 1144-11 1199 30. Liu L, Reeder B, Shuaib A, Mazagri R. Validity of stroke diagnosis on hospital discharge records in Saskatchewan, Canada: implications for stroke surveillance. Cerebrovasc Dis. 1999;9:224-230. 31. Jackevicius CA, Paterson JM, Naglie G. Concordance between discharge prescriptions and insurance claims in post-myocardial infarction patients. Pharmacoepidemiol Drug Saf. 2007;16:207-215. 32. Levy AR, O'Brien BJ, Sellors C, Grootendorst P, Willison D. Coding accuracy of administrative drug claims in the Ontario Drug Benefit database. Can J Clin Pharmacol. 2003;10:67-71. 33. Tamblyn R, Lavoie G, Petrella L, Monette J. The use of prescription claims databases in pharmacoepidemiological research: the accuracy and comprehensiveness of the prescription claims database in Quebec. J Clin Epidemiol. 1995;48:999-1009.

17 Downloaded from http://circ.ahajournals.org/ by guest on January 2, 2015

DOI: 10.1161/CIRCULATIONAHA.113.004777

34. Zhou Z, Rahme E, Abrahamowicz M, Pilote L. Survival bias associated with time-totreatment initiation in drug effectiveness evaluation: a comparison of methods. Am J Epidemiol. 2005;162:1016-1023. 35. Gage BF, Waterman AD, Shannon W, Boechler M, Rich MW, Radford MJ. Validation of clinical classification schemes for predicting stroke: results from the National Registry of Atrial Fibrillation. JAMA. 2001;285:2864-2870. 36. Friberg L, Rosenqvist M, Lip GY. Evaluation of risk stratification schemes for ischaemic stroke and bleeding in 182 678 patients with atrial fibrillation: the Swedish Atrial Fibrillation cohort study. Eur Heart J. 2012;33:1500-1510. 37. Olesen JB, Lip GY, Hansen PR, Lindhardsen J, Ahlehoff O, Andersson C, Weeke P, Hansen ML, Gislason GH, Torp-Pedersen C. Bleeding risk in 'real world' patients with atrial fibrillation: comparison of two established bleeding prediction schemes in a nationwide cohort. J Thromb Haemost. 2011;9:1460-1467. Qualitative 38. Newell DJ. Intention-to-Treat Analysis: Implications m for Quantitative and Qual alit itat it a iv at ivee Research. Int J Epidemiol. 1992;21:837-841. 39. D'Agostino RB, Jr. Propensity scores in cardiovascular research. Circulation. 2007;115:2340-2343. 2007;1 ; 15:2340-2343. 40. MS, logistic versus 40 0. Cepeda Ce MS, Boston Bos osto tonn R, Farrar Far arra rarr JT, ra JT, Strom Stro St rom m BL. BL Comparison Comp Co par aris ison is o of on of lo logi g st gi stic icc rregression eg gre ress ssio ss ionn ve io vers rssus u propensity multiple prop pensity score sco ore when whe henn the th he number numb nu mber err of of events eventss iss low eve w andd ther tthere herre ar aree mult m ultip iplle cconfounders. ip onfo on foun unnde ders rs. Am J Epidemiol. Epid Ep demiol. 22003;158:280-287. 0 3;;158:28 00 280-28 28 87. 7. Examining Heterogeneity 41. Deeks Deeks ks JJ, J , Altman JJ Altm Al man DG, G, Bradburn Bra radb dbuurn MJ. MJJ. Statistical Stat St atis isti tica c l Methods Meth Me thod odss fo for Ex Exam aminin ng He Hete teroge gene n it ityy and and Results Several Studies Meta-Analysis. Care. Combining Re esuultts fr ffrom om S ever ev e al S er tudiies tu e iin n Me Meta taa-A Ana n ly ysi sis. s Systematic s. Syst Sy stem st emat em a ic Reviews at Rev evie i ws in ie in Health He BMJJ Publishing BM Publ Pu blis ishi hing ng Group; Gro roup up;; 2001:285-312. 2001 20 01:2 :285 85-312 312 42. Marinigh R, Lane DA, Lip GY. Severe renal impairment and stroke prevention in atrial fibrillation: implications for thromboprophylaxis and bleeding risk. J Am Coll Cardiol. 2011;57:1339-1348. 43. Baumann M, Seifert CL, Poppert H. Atrial fibrillation and chronic kidney disease. N Engl J Med. 2012;367:2157-2158. 44. Schlieper G, Kruger T, Floege J. Atrial fibrillation and chronic kidney disease. N Engl J Med. 2012;367:2157-2159. 45. Sood MM, Tangri N. Atrial fibrillation and chronic kidney disease. N Engl J Med. 2012;367:2158-2159. 46. Babyak MA. What you see may not be what you get: a brief, nontechnical introduction to overfitting in regression-type models. Psychosom Med. 2004;66:411-421.

18 Downloaded from http://circ.ahajournals.org/ by guest on January 2, 2015

DOI: 10.1161/CIRCULATIONAHA.113.004777

47. Ionescu-Ittu R, Abrahamowicz M, Jackevicius CA, Essebag V, Eisenberg MJ, Wynant W, Richard H, Pilote L. Comparative effectiveness of rhythm control vs rate control drug treatment effect on mortality in patients with atrial fibrillation. Arch Intern Med. 2012;172:997-1004. 48. Fewell Z, Davey SG, Sterne JA. The impact of residual and unmeasured confounding in epidemiologic studies: a simulation study. Am J Epidemiol. 2007;166:646-655. 49. Walker AM. Confounding by indication. Epidemiology. 1996;7:335-336. 50. Normand SL, Sykora K, Li P, Mamdani M, Rochon PA, Anderson GM. Readers guide to critical appraisal of cohort studies: 3. Analytical strategies to reduce confounding. BMJ. 2005;330:1021-1023.



19 Downloaded from http://circ.ahajournals.org/ by guest on January 2, 2015

DOI: 10.1161/CIRCULATIONAHA.113.004777

Table 1. Baseline Characteristics of Patients with Atrial Fibrillation Dialysis Patients, Non-dialysis Patients, N = 1,626 N = 204,210 Warfarin users, No-warfarin Warfarin users, No-warfarin N = 756 users, N = 870 N = 103,473 users, N = 100,737 Patients diagnosed with AF: AF as a main diagnosis, n (%) 150 (20) 125 (14) Age at the index AF 75.3 ± 8.1 75.1 ± 8.5 admission in years, mean ± SD* Male sex, n (%) 459 (61) 533 (61) Length of hospitalization in 19.4 ± 37.2 21.1 ± 44.5 days, mean ± SD* Co-morbidities, n (%): Coronary artery disease 470 (62) 517 (59) Acute myocardial infarction 201 (27) 249 (29) Valvular heart disease 131 (17) 115 (13) Liver disease 28 (4) 33 (4) History of bleeding event 65 (9) 139 (16) Specific components of CHADS2 score†, n (%): 312 (41) 299 (34) Congestive heart failure Hyype pert rten rt ensi en sion si on Hypertension 582 (77) 655 (75) Age • 75 years year ye ars 386 (51) (51 51)) 415 5 (4 ((48) 8)) Age 386 D Dia iabetes 330 (44) (4 44) 3400 (39) (39 9) Diabetes 330 H istory st str trokee/T e/TIA (6 6) (5) History of stroke/TIA 42 (6) 444 (5) † CH HAD ADS S2 sc CHADS score scor oree , n (%): or (%): (%) Lo ow risk riiskk (0) (0) 0 (3 3) (7) Low 23 (3) 59 (7) rissk (1) (1 1) 15 (20) ( 0) (2 215 (25) (25 25)) Moderate risk 153 215 High Hi gh rrisk iskk ((• is • 2) 2) 580 (77) (77) 5596 96 (69) (69 69)) 580 HAS-BLED score‡, n (%): Low and moderate risk§ (1-2) 119 (16) 126 (14) High risk (• 3) 637 (84) 744 (86) First filled prescription within 30 days after AF discharge, n (%): Rate control drugs 519 (69) 462 (53) Rhythm control drugs 203 (27) 172 (20) Aspirin 166 (22) 241 (28) Clopidogrel 30 (4) 58 (7) NSAIDs 8 (1) 20 (2)

34,710 (34)

19,802 (20)

77.9 ± 9.5 49,133 (48)

78.8 ± 10.6 50,425 (50)

10.4 ± 20.9

10.5 ± 24.0

40,163 (39) 15,489 (15) 15,633 (15) 1,792 (2) 4,680 (5)

40,199 (40) 16,413 (16) 10,140 (10) 2,41 2, 413 3 (2 (2) 2,413 99,042 9, 0422 (9) 04 (9

33,659 (33) 47,9 ,99722 (46) 47,972 65, 5,33 333 3 (63) 3) 65,333 21,5 21 ,574 ,5 4 ((21) 21)) 21 21,574 9,4 464 4 (9 9) 9,464 (9)

27,494 (27) 41,738 (41) 665,814 5,81 5, 8 4 (6 (65) 119,756 19 ,756 ,7 56 ((20) 20) 5,2833 (5 5) (5)

11, 1,87 870 0 (11) (11) 11,870 31,5 31 ,533 ,5 33 (30) (30 3 ) 31,533 60 070 ((58) 58)) 58 60,070

14, 4,30 3088 (1 (14) 4) 14,308 34,450 (34)) 51 979 ((52) 52)) 52 51,979

80,747 (78) 22,726 (22)

73,260 (73) 27,477 (27)

75,656 (73) 23,512 (23) 11,814 (11) 1,843 (2) 2,109 (2)

55,167 (55) 16,508 (16) 24,544 (24) 4,158 (4) 3,336 (3)

Abbreviations: AF, atrial fibrillation; SD, standard deviation; TIA, transient ischemic attack; NSAIDs, non-steroidal antiinflammatory drugs; Results are weighted average for results from Quebec and Ontario. We presented continuous variables as mean ± SD and dichotomous variables as number (%). * We used following formula to combine standard deviation: SDcombined = ¥(SD1)2+(SD2)2. † CHADS2 score is a clinical prediction score for estimating the risk for stroke. ‡ HAS-BLED score is a clinical prediction score for estimating the risk for bleeding. § HAS-BLED score has minimum score of one and two for non-dialysis patients and dialysis patients, respectively. In this study, all AF patients are age • 65 years, which accounts for one point. In dialysis group, all patients have abnormal renal function, which also accounts for one point.

20 Downloaded from http://circ.ahajournals.org/ by guest on January 2, 2015

DOI: 10.1161/CIRCULATIONAHA.113.004777

Table 2. Crude Incidence Rate for Stroke and Bleeding Events

Stroke† According to warfarin prescription (within 30 days post-discharge): Yes No According to CHADS2 score‡: Low risk (0) Moderate risk (1) High risk (• 2) Bleeding§ According to warfarin prescription within 30 days post-discharge): (within Yes No Acco According co ord rdin ingg to HAS in HAS-BLED AS-B AS -BLED score: L Lo Low ow w an aand d mo mode moderate era rate te rrisk issk# ((1-2) 1-2) 12) High Hi igh risk (• 33))

Dialysis Patients Non-dialysis Patients N = 1,626 N = 204,210 * No. of Incidence rate per No. of Incidence* rate per events 100 person-years events 100 person-years 107 3.12 19,489 2.35

52 55

3.37 2.91

9,241 10,248

2.19 2.51

4 23 80 275

1.99 2.35 3.55 8.89

2,270 6,078 11,141 34,035

1.49 2.06 2.91 4.32

149 126

10.88 7.31

18,340 15,695

4.64 4.00

443 3 2232 32

8.00 8.00 9. 9.08 .08 8

226,129 6,,12 1 9 77,906 ,9 906 6

4.07 4.07 55.45 .4 45

Abbreviations: A Ab brrev e iations: AF, AF, aatrial trrial fibr ffibrillation; i rillaatiion n; TIA, TIA IA,, tr tra transient ansient iischemic scchem micc attac attack; acck; Incidence In nci c de denc nce ra rate rates tees we wer were re ccalculated alcu al cu ulaateed with wi h tthe he following fol ollo owi wing ng formula: forrmu mula laa: no. no. off events eveentts / to total otaal follow-up f ll fo llow ow w-u -upp time tiime m ((100 100 person-years). 100 peersson on-y -yeaars). Incidence weighted average Quebec Ontario. nci c de denc ncee ra nc rate te aare re w eigh ei ghte gh tedd av te aver erag ge fo forr resu rresults esultts fr from om Q uebe uebe becc an andd On Onta tari ta rioo. ri o. † Stroke Stro St roke ro ke w was as defined def efin ined in ed as as the the first firs fi rstt ho rs hosp hospital spit sp ital it al aadmission dmis dm issi is sion si on oorr em emer emergency erge er genc ge ncyy room nc room visit vis isit it ffor or ischemic isc sche hemi he micc cerebrovascular mi cere ce rebr re brov br ovas ov ascu as cula cu larr di la dise disease, seas se asee, as e, infarct follow-up period. TIA, or retinall in infa farc fa r t at any rc any point poiint during durrin ingg fo foll l ow ow-u -u up pe peri r od ri od.. ‡ CHADS CHAD CH ADS S2 sscore core co re iiss a cl clin clinical inic ical al pprediction redi re dict ctio ionn sc scor score oree fo forr es esti estimating tima mati ting ng the the risk ris iskk for for stroke. stro st roke ke § Bleeding was defined as the first hospital admission or emergency room visit for intracerebral bleeding, gastrointestinal bleeding, intraocular bleeding, hematuria, and unspecified location of bleeding at any point during follow-up period. HAS-BLED score is a clinical prediction score for estimating the risk for bleeding. # HAS-BLED score has minimum score of one and two for non-dialysis patients and dialysis patients, respectively. In this study, all AF patients are age • 65 years, which accounts for one point. In dialysis group, all patients have abnormal renal function, which also accounts for one point.

*

21 Downloaded from http://circ.ahajournals.org/ by guest on January 2, 2015

DOI: 10.1161/CIRCULATIONAHA.113.004777

Table 3. Association between Warfarin Use and the Risk for Stroke and Bleeding in Patients with Atrial Fibrillation Patients with AF

Outcomes

Dialysis (n = 1,626)

Stroke‡ Bleeding§ Stroke‡ Bleeding§

Non-dialysis (n = 204,210)

Adjusted* HR (95% CI) 1.14 (0.78, 1.67) 1.44 (1.13, 1.85) 0.87 (0.85, 0.90) 1.19 (1.16, 1.22)

Propensity Score† Adjusted HR (95% CI) 1.17 (0.79, 1.75) 1.41 (1.09, 1.81) 0.89 (0.87, 0.92) 1.20 (1.17, 1.23)

Abbreviations: AF, atrial fibrillation; HR, Hazard Ratio; CI, Confidence Interval; TIA, transient ischemic attack; NSAIDs, non-steroidal anti-inflammatory drugs; * Stroke outcome was adjusted for: age (years), sex, specific components of CHADS2 stroke prediction score (congestive heart failure, hypertension, diabetes, and history of stroke/TIA). * Bleeding outcome was adjusted for: age (years), sex, specific components of HAS-BLED bleeding prediction score (liver disease, hypertension, history of stroke/TIA, history of bleeding, and use of aspirin, clopidogrel, or NSAIDs). † Propensity score was derived from the following variables: age • 75 years, sex, type of AF (primary vs. secondary), CHADS2 scores (1 and • 2), liver disease, congestive heart failure, hypertension, diabetes, history of stroke/TIA, historyy of bleeding, g, use of rate control drug, g, rhythm y control drug, g, aspirin, p , clopidogrel, p g , and NSAIDs. ‡ Stroke was defined as the first hospital admission or emergency room visit for ischemic cerebrovascular cerebrov vas ascu cula cu larr di la dise disease, seas se ase, as e e, TIA, or retinal infarct at any point during follow-up period. § Bleeding was defined as the first hospital admission or emergency room visit for intracerebral bbleeding, leed le edin ed ing, in g gastrointestinal bleeding, intraocular bleeding, hematuria, and unspecified location of bleeding at any point during follow-up ollow-up period.

Figure F igu gure Legends: Legen :

Figure 1. Wa Warfarin Warf rfar rf arin ar i U in Use se aand nd tthe h R he Risk iskk fo is forr St Stro Stroke roke ke iin n Pa Pati Patients tien ents ts w with itth At Atrial tri rial al F Fibrillation ibri ib rill llat attio ionn Un Underg Undergoing goingg Dialysis. Abbreviations: HR, Hazard Ratio; LCL, Lower Confidence Limit; UCL, Upper Confidence Limit; TIA, transient ischemic attack; Chan et al. defined stroke outcome as hospitalization or death from ischemic stroke, hemorrhagic stroke, or TIA.8 45% (747/1,671) patients were receiving warfarin.8 Wizemann et al. defined stroke outcome as hospitalization or death from stroke or cerebrovascular event.17 15% (146/1,001) patients, 17% (192/1,137) patients, and 15% (171/1,107) patients in age group ” 65 years, 66 to 75 years, and > 75 years were receiving warfarin, respectively.17 Winkelmayer et al. defined stroke outcome as ischemic or hemorrhagic stroke.16 11% (249/2,313) patients were receiving warfarin.16 237 warfarin users

22 Downloaded from http://circ.ahajournals.org/ by guest on January 2, 2015

DOI: 10.1161/CIRCULATIONAHA.113.004777

were matched to 948 non-users.16 Olesen et al. defined stroke outcome as hospitalization or death from stroke or systemic thromboembolism (ischemic stroke, peripheral artery embolism, and TIA).1 20% (178 / 901) patients were receiving warfarin only.1 Our study defined stroke as the first hospital admission or emergency room visit for ischemic cerebrovascular disease, TIA, or retinal infarct at any point during follow-up period. 46% (756/1,626) patients were receiving warfarin.

Figure 2. Warfarin Use and the Risk for Bleeding in Patients with Atrial Fibrillation Undergoing Dialysis. Abbreviations: HR, Hazard Ratio; LCL, Lower Confidence Limit; UCL, Upper Confidence Limit; GI, Gastrointestinal; Winkelmayer et al. defined bleeding outc outcome tccom me as G GII bleeding.16 11% (249/2,313) patients were receiving warfarin.16 237 warfarin users were matched to o 9948 48 nnon-users. on n-u -usserss.166 Ol Olesen O esen et al. defined bleedin bleeding ng ooutcome utcome as hos hospitalization spi p taaliiza zati t on or death from GI, intracranial, ntrracranial, urinary urin ur inarry tract, in trracct, and and air-way air ir-w -way -w ay bleeding. bleed din ng.1 220% 0% ((178/901) 178/9901) 178 901)) ppatients atie at ienntss we were ere re rreceiving eccei eivi vinng vi ng w warfarin arfa arfa farin only. on nly ly.1 Ou Ourr st O stud study udyy de defined efi finnedd bl bbleeding eeediing ooutcome utccome ut come aass th thee fi first irs r t ho hosp hospital spittal aadmission sp dm mis i sionn oorr em emergency merrge genccy ro room om m visit for intrac accer ereb eb bra rall bl blee e di ding ng, GI bbleeding, ng l ed le edin in ng, iintraocular ntrrao nt aocu cula cu l r bl blee eedi ee ding di ng,, he ng hema matu ma turi tu ria, ri a aand a, nd uunspecified nspe ns p cified intracerebral bleeding, bleeding, hematuria, location of bleeding at any point during follow-up period. 46% (756/1,626) patients were receiving warfarin.

23 Downloaded from http://circ.ahajournals.org/ by guest on January 2, 2015

Figure 1

Downloaded from http://circ.ahajournals.org/ by guest on January 2, 2015

Figure 2

Downloaded from http://circ.ahajournals.org/ by guest on January 2, 2015

SUPPLEMENTAL MATERIAL Supplemental Table 1. Databases in Quebec and Ontario Data Type

Quebec, Canada

Ontario, Canada

Hospital

Maintenance et Exploitation des

Canadian Institute for Health

Discharges

Données pour l’Étude de la Clientèle

Information (CIHI)

Hospitalière (Med-Echo) Physician

la Régie de l’assurance maladie du

Ontario Health Insurance Plan

Claims

Québec (RAMQ)

Prescription

RAMQ

Ontario Drug Benefit Plan

Med-Echo & RAMQ

Ontario Registered Persons and

Claims Vital Status

CIHI Databases

1

Supplemental Table 2. Characteristics and Database Codes Characteristics

Database Codes

Atrial fibrillation

ICD-9/10 codes: 427.3, 427.31, 427.32 / I48

Dialysis

ICD-9/10 codes: V451, V560, V561, V568, E8742, E8702, E8712, E8722, E8791 / Z992, Z490, Z4932, T824, Y602, Y612, Y622, Y841; CCI: 1OT53DATS, 1OT53HATS, 1OT53LATS, 1PZ21XX, 1PZ21HP, 1PZ21HQ, 1SY55LAFT, 7SC59QD, 1KG76MZ, 1KY76; CCP: 5127, 5142, 5143; MED-ECHO: 3927, 3995, 3942, 3943, 5127, 5142, 5143, 5195, 6698; RAMQ: 15035,15040-15048, 15050, 15051, 09216-09219, 09259-09264, 09274, 09275, 09279, 09291, 00147, 00283-00290; OHIP: R850, G324, G336, G327, G862, G865, G099, R825, R826, R827, R833, R840, R841, R841, R843, R848, R851, Z450, Z451, Z452, G864, R852, R853, R854, R885, G333, H540, H740 ICD9: 434, 435, 436, 362.3 or ICD-10 codes: I63, I64, G45 (excluding

Stroke G45.4), H34.1 Bleeding event

2

Intracerebral hemorrhage

ICD-9/10 codes: 431, 432 / I61 ICD-9/10 codes: 578 (main code), 530.7, 531.0, 531.2, 531.4, 531.6, 532.0, 532.2, 532.4, 532.6, 533.0, 533.2, 533.4, 533.6, 534.0, 534.2, 534.4, 534.6 / K92.0, K92.1, K92.2, (1st 3 are main codes) K25.0, K25.2,K25.4, K25.6 , K26.0, K26.2, K26.4, K26.6 , K27.0, K27.2, K27.4, K27.6, K28.0, K28.2,

Gastrointestinal hemorrhage

K28.4, K28.6, K29.0

Intraocular hemorrhage

ICD-9/10 codes: 362.8, 379.2 / H43.1, H35.6

Hematuria

ICD-9/10 codes: 599.7 / N02, R31

Hemorrhage not other specified (NOS)

ICD-9/10 codes: 459 / na

3

Warfarin Use and the Risk for Stroke and Bleeding in Patients with Atrial Fibrillation Undergoing Dialysis Mitesh Shah, Meytal Avgil Tsadok, Cynthia A. Jackevicius, Vidal Essebag, Mark J. Eisenberg, Elham Rahme, Karin H. Humphries, Jack V. Tu, Hassan Behlouli, Helen Guo and Louise Pilote Circulation. published online January 22, 2014; Circulation is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231 Copyright © 2014 American Heart Association, Inc. All rights reserved. Print ISSN: 0009-7322. Online ISSN: 1524-4539

The online version of this article, along with updated information and services, is located on the World Wide Web at: http://circ.ahajournals.org/content/early/2014/01/22/CIRCULATIONAHA.113.004777

Data Supplement (unedited) at: http://circ.ahajournals.org/content/suppl/2014/01/22/CIRCULATIONAHA.113.004777.DC1.html

Permissions: Requests for permissions to reproduce figures, tables, or portions of articles originally published in Circulation can be obtained via RightsLink, a service of the Copyright Clearance Center, not the Editorial Office. Once the online version of the published article for which permission is being requested is located, click Request Permissions in the middle column of the Web page under Services. Further information about this process is available in the Permissions and Rights Question and Answer document. Reprints: Information about reprints can be found online at: http://www.lww.com/reprints Subscriptions: Information about subscribing to Circulation is online at: http://circ.ahajournals.org//subscriptions/

Downloaded from http://circ.ahajournals.org/ by guest on January 2, 2015

Warfarin use and the risk for stroke and bleeding in patients with atrial fibrillation undergoing dialysis.

Current observational studies on warfarin use and the risk for stroke and bleeding in patients with atrial fibrillation (AF) undergoing dialysis found...
1MB Sizes 0 Downloads 0 Views