Cardiovascular, bleeding, and mortality risks in elderly Medicare patients treated with dabigatran or warfarin for nonvalvular atrial fibrillation.

Cardiovascular, Bleeding, and Mortality Risks in Elderly Medicare Patients Treated with Dabigatran or Warfarin for Non-Valvular Atrial Fibrillation David J. Graham, Marsha E. Reichman, Michael Wernecke, Rongmei Zhang, Mary Ross Southworth, Mark Levenson, Ting-Chang Sheu, Katrina Mott, Margie R. Goulding, Monika Houstoun, Thomas E. MaCurdy, Chris Worrall and Jeffrey A. Kelman Circulation. published online October 30, 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/2014/10/30/CIRCULATIONAHA.114.012061

Data Supplement (unedited) at: http://circ.ahajournals.org/content/suppl/2014/10/30/CIRCULATIONAHA.114.012061.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/ at Marquette University on November 4, 2014

DOI: 10.1161/CIRCULATIONAHA.114.012061

Cardiovascular, Bleeding, and Mortality Risks in Elderly Medicare Patients Treated with Dabigatran or Warfarin for Non-Valvular Atrial Fibrillation Running title: Graham et al.; Comparative safety of dabigatran and warfarin David J. Graham, MD, MPH1; Marsha E. Reichman, PhD1; Michael Wernecke, BA2; Rongmei Zhang, PhD3; Mary Ross Southworth, PharmD4; Mark Levenson, PhD3; Ting-Chang Sheu, MPH2; Katrina Mott, MHS1; Margie R. Goulding, PhD1; Monika Houstoun, PharmD, MPH1; Thomas E. MaCurdy, PhD2,5;Chris Worrall, Wor orra rall ra l , BS6; ll Jeffrey A. Kelman, MD, MMSc6 1

Office Of ffi fice ce ooff Su Surv Surveillance vei eill l an ll a ce and Epidemiology, Cen Center nte terr for Drug Evaluati Evaluation ion on aand nd Research, Food and D Dr ugg Administration, Adm min inis istr t at atio on, Silver Silv lver er Spring, Spr pring, g, MD; MD; 2Ac Acumen Acum men L LLC, LC,, Bu Burlin Burlingame, inga gam me,, CA CA;; 3Of Office ffice ce of Drug Bi Biostatistics, iostatistic i s, C Center en nteer fo forr Drug Drug E Evaluation valu va luat lu a ion and and Re Research, esearrch ch, Fo Food od an aand d Dr Drug ug A Administration, dmiinis dm inis istr trat tr attion, ion, S Silver i ver il Spring, Sp pring, MD; 4Of Office Off ficee of of New Neew Drugs, Drug ugs, s Center Centeer for for Drug Dr Evaluation Eval Ev aluaatioon and al and Research, Reseearrch, Food Foood and an nd Drug Druug Administration, Admi Ad miniist stra rati ra t on ti n, Si Silv Silver lv ver S Spring, prrin ng, M MD; D; 5Dept D; Dept ooff E Economics, conom onom omic ics, s, Stanford Sta tannfoord ord Un Univ University, iveersi ersiity ty,, St Sta Stanford, anffor ford, rd, CA CA; 6 Centers Cent Ce nter nt erss for er foor Medicare Meddicare Me di e & Medicaid Med edic dic icaaid Services, Ser ervi vice vi cees, s, Washington, Was ashi hing hi ngto ng ton, to n, DC DC

Addr Ad Address dres esss for for Correspondence: C rr Co rres espo pond nden ence ce:: David J. Graham, MD, MPH Office of Surveillance and Epidemiology 10903 New Hampshire Ave. Building 22, Room 4314 Silver Spring, MD, 20993-0002 Tel: 301-796-0163 Fax: 301-796-9832 E-mail: [email protected] Journal Subject Codes: Anticoagulants:[184] Coumarins, Anticoagulants:[185] Other anticoagulants, Etiology:[8] Epidemiology, Stroke:[53] Embolic stroke, Stroke:[43] Acute cerebral hemorrhage

1 Downloaded from http://circ.ahajournals.org/ at Marquette University on November 4, 2014


Abstract

Background—The comparative safety of dabigatran vs. warfarin for treatment of non-valvular atrial fibrillation (AF) in general practice settings has not been established. Methods and Results—We formed new-user cohorts of propensity score matched elderly patients enrolled in Medicare, who initiated dabigatran or warfarin for treatment of non-valvular AF between October 2010 and December 2012. Among 134,414 patients with 37,587 personyears of follow-up, there were 2,715 primary outcome events. The hazard ratios (95% confidence intervals) comparing dabigatran with warfarin (reference) were ischemic stroke: 0.80 (0.67-0.96); intracranial hemorrhage: 0.34 (0.26-0.46); major gastrointestinal bleeding: 1.28 (1.14-1.44); acute myocardial infarction: 0.92 (0.78-1.08); and death: 0.86 (0.77-0.96). In the g p treated with dabigatran g g twice daily, y, there was no difference in risk compared p subgroup 75 mg with warfarin for any outcome except intracranial hemorrhage, where dabigatrann risk isk s w as was educed. Most patients treated with dabigatran 75 mg twice daily appeared not to have severe reduced. enal impa p irment,, the intended population for this dose. In the dabigatran 150 mg twice daily renal impairment, ubg bgrroup rou , th he ma magn gnit itud de of eff ffec ect fo for ea eeach c out ch tco ome wass ggreater reaterr tthan re han in the the combined-dose com mbinedd do dose s subgroup, the magnitude effect outcome an nallys y is. analysis. Conc Co ncclu usionss—I — n general generaal practice gen prac pr a ti ac t ce settings, set e tiings, dabigatran dabig igaatrran ran was was associated asssocciaated with with ith reduced r du re d ce c d risk riskk of of Conclusions—In sch hem emiic sstroke, trokke, int tr ntra racr c an nia i l he hemo morrhagee, an andd de death, aand nd iincreased ncre nc reas a ed rrisk iskk off ma is majo jorr gast stro r in nte test stin nal ischemic intracranial hemorrhage, major gastrointestinal ge compared com ompa pare pa redd with re w th wi h warfarin warrfa fari rinn inn elderly eld ldeerl rlyy patients p ti pa tien en nts with wit i h non-valvular nonno n vaalv lvul ular ul ar AF. AF. F These Thes Th esee es hemorrhage associations were most pronounced in patients treated with dabigatran 150 mg twice daily, whereas the association of 75 mg twice daily with study outcomes was indistinguishable from warfarin except for a lower risk of intracranial hemorrhage with dabigatran.

Key words: anticoagulant, safety, pharmacoepidemiology, warfarin, thrombin inhibitor, mortality, atrial fibrillation arrhythmia



Background Warfarin, a vitamin K antagonist, is a commonly used anticoagulant in patients with atrial fibrillation (AF). Meta-analyses of randomized trials have shown that it reduces the risk of embolic stroke by 40-80% and of mortality by about 30%, but also doubles the risk of intracranial hemorrhage and increases the risk of extracranial bleeding by up to 66%.1-3 It can be difficult to maintain patients in the therapeutic range (international normalized ratio (INR) of 2 to 3). A meta-analysis found that warfarin-treated patients were in the therapeutic range 61% of the time.4 In this study, an INR below 2 was associated with a 5-fold increase in stroke risk and an INR above 3 was associated with a 3-fold increase in bleeding risk.4 Dabigatran is a competitive direct thrombin inhibitor approved in the US S to reduce red duc ucee the the risk isk of stroke and systemic embolization in patients with non-valvular AF.5 In the Randomized Ev val alua uati ua tion ti onn ooff Lo ong ng-T - erm Anticoagulation Ther rapy (RE-LY) trial all, pa atiien ents t with AF were Evaluation Long-Term Therapy trial, patients randomized and dom o ized too da dab dabigatran biga gattran an ddoses oses ose es ooff 11 1110 0m mg g oorr 1500 m mg g ttwice wicee ddaily wice ailly oorr ad ai adju adjusted-dose d ust sted ed-d ed -dos osee warf w warfarin. ar ar ariin. in.6 At the he dose dose s subsequently sub ubssequ sequuen ntl tlyy approved ap ppr provved d in in the thee US (150 th (150 150 mg mg twice twice wicee daily), daily ly), ly ), dabigatran dab abig ig gatrran ran reduced red educ ucced the thhe he risk ris iskk off stroke troke and int intracranial ntra racr ra c an cr ania i l he ia hhemorrhage morr mo r haage an rr aand d in increased ncr crea ease ea seed th thee ri risk sk ooff ma majo major jor ga jo gast gastrointestinal stro st roin ro inte in t st te stin in nal a hhemorrhage e orrhage em compared with warfarin.12 While dabigatran has a favorable net clinical benefit compared with warfarin,6 it lacks a validated assay by which to monitor anticoagulation intensity and there is no proven method of rapidly reversing its effect.5-7 Given that AF primarily affects the elderly and that the safety profile of dabigatran may be different in general practice versus controlled trial settings, we compared the risk of stroke, major gastrointestinal and intracranial bleeding, acute myocardial infarction (AMI), and mortality in elderly Medicare beneficiaries with non-valvular AF who initiated therapy with warfarin or dabigatran.



Methods Study population Medicare provides health insurance coverage to about 42 million persons age 65 years and older as well as nearly 9 million persons under age 65 with end-stage kidney disease or who are disabled.8,9 This study was restricted to the approximately 21 million beneficiaries aged 65 years or older enrolled in fee-for-service Medicare Part A (hospitalization), Part B (office-based medical care), and Part D (prescription drugs) since claims from these sources were necessary for research purposes. For each beneficiary, we linked claims from all settings of care to create a longitudinal record of their health encounters, diagnoses, and drug prescriptions. A new user retrospective cohort design was used d to compare patients ini itiiattin i g da dabi biga bi g tran ga initiating dabigatran or warfarin for the treatment of non-valvular AF.10 We identified all patients with any inpatient orr outpatient out utppati pati tien entt diagnoses en d ag di gnoses no of atrial fibrillation or flutter, fllut utteer, based on IC CDD 9 co cod ding who also filled at ICD-9 coding least easst one prescription prescr crip ipttion for ip forr either eithe ithe herr drug drrug from fro rom m October Octtobeer 19,, 2010 20100 (US (US S dabigatran dab abig igattra ig rann approval approoval appr oval date) datte) through hro oug ughh December Deece cemb mb berr 31, 31, 1, 2012, 201 012, the the study stu tudy dyy end end date. datte. e Patients Pati Pa t ents ents w were erre excl eexcluded xcl clud u ed iiff th ud they ey hhad ad le less ess th than an n6 months of enrollment en nro roll l me ll ment n in nt in Medicare Medi Me d caare prior di pri r or to to their thei th e r index ei in nde dexx di disp dispensing, spen sp ensi en siing ng, were were under und nder er age age 65, 65, received prior treatment with a study medication or rivaroxaban or apixaban (anticoagulants approved during the study), or were in a skilled nursing facility or nursing home, or were receiving hospice care on the date of their cohort qualifying prescription. Patients were also excluded if they had a hospitalization that extended beyond the index dispensing date. Patients discharged from hospital on the same day as their index dispensing were included. Patients undergoing dialysis and kidney transplant recipients were also excluded. Additionally, because warfarin is approved for indications other than AF, we excluded patients with diagnoses indicating the presence of mitral valve disease, heart valve repair or replacement, deep vein thrombosis, pulmonary



embolism, or joint replacement surgery in the preceding six months. Baseline covariates and cohort follow-up Claims data on chronic medical conditions, cardiovascular risk factors, risk factors for bleeding events, and health care utilization were collected for each patient during the 6 months preceding their cohort-qualifying prescription fill. We also collected data on prescriptions for medications used for treatment of cardiovascular disease and other chronic medical conditions, as well as potentially interacting medications that might alter warfarin or dabigatran pharmacokinetics. Finally, to the extent possible using claims data, we calculated the CHADS2 score,11 which predicts the risk of stroke in patients with AF, and the HAS-BLED score,12,13 which predicts the risk isk of bleeding in patients with AF treated with warfarin. To reduce confounding due to imbalance in study covariates, propensity score matching t 14-16 16 was was us used used. ed.14ed Unconditional Uncconditional Un co logistic regressionn wa was used d to estima estimate m tee the the predicted predicted probability of

patients pati ien e ts initiating initiattin ingg dabigatran dabi dabi b gaatran trran therapy ther erap apyy ggiven ap ivven th ttheir eir soc ssociodemographic ocioddemo emogra ograp aphic ccharacteristics, haraact ha cteeri eristi istics css, ba bas baseline seli line li ne medical comorbidities, medications months, me edi dica call co ca omo m rb rbiidiitie itiess, m e iccattio ed ions ns uused sed du during ng tthe he ppreceding rece cedding dingg 6 m onth on th hs,, pprescriber reesccri ribe berr be characteristics, characteristiccs, aand nd oother th her e ppotentially oten ot e ti tial a ly al y rel relevant elev el evan ev antt va an vari variables riab ri able ab l s ((Table le Tabl Ta blee 1 and bl and Su Supp Supplemental pple pp leme le ment me ntal nt al T Table able 1). Dabigatran users were propensity score-matched to warfarin users in a 1:1 ratio using a greedy matching algorithm. The balance of measured covariates between the matched cohorts was assessed using the standardized mean difference, a measure not influenced by sample size and thus useful for comparing cohorts in large observational studies.17 A standardized mean difference of 0.1 or less indicates a negligible difference in the measured variables between groups.17 Follow-up began on the day after the first qualifying anticoagulant prescription fill and continued until disenrollment from Medicare, occurrence of a study outcome, a gap in



anticoagulant days of supply exceeding 3 days, a prescription fill for a different anticoagulant, initiation of dialysis or kidney transplantation, admission to a skilled nursing facility or nursing home, transfer to hospice care, or the end of the study period, whichever came first. We chose a 3-day gap in anticoagulant therapy to increase the likelihood that patients were therapeutically anticoagulated given the short half-life of dabigatran, estimated at 14 hours.5 We censored patients for admission to a skilled nursing facility or nursing home due to concerns about incomplete capture of prescription fills and outcomes in these settings. We also censored patients transferred to hospice care because most deaths in these patients were expected and therefore unlikely to be related to anticoagulant use. Study outcomes The primary outcomes were ischemic stroke, major bleeding with specific focus on intracranial an nd ga gast stro st ro oin inte testin te in nal al bleeding, and AMI. Seconda dary da ry y outcomes we ere r al ll ho hospitalized bleeding and gastrointestinal Secondary were all events ev vents en n and mortality. mor orta taaliity ty. The The ICD-9-CM ICD CD-9 -9-C -CM -C M codes co odess used used d too define deefine fine these thhese hesee outcomes out utco co omes mes are are listed list li sted st ed in in Supp Su pple pp leme le ment ntal nt al Table Tab ble 2. 2. The The co code dess def de ddefining efini fini ning ng ischemic isc sche sc hemi he mic st stro rokke ro ke hhave avee a po av pos sittive tive ppredictive redi re diict ctiivee va alu luee Supplemental codes stroke positive value 18-20 20 PPV) of 88-95%. 88--95 95%. %.18 M Major ajor aj or bbleeding leed le ed din i g wa wass de defi defined f ne fi nedd as a a fa fata fatal tal bl ta blee bleeding e di ding ng eevent, vent ve n , a ho nt hosp hospitalized spit sp i alized it (PPV)

bleeding event requiring transfusion, or hospitalization with hemorrhage into a critical site (i.e., intracranial, intraspinal, intraarticular, intraocular, pericardial, retroperitoneal, or intramuscular with compartment syndrome).6 Intracranial hemorrhage was defined using codes for atraumatic hemorrhage, with a PPV of 89%-97%,18-20 and codes for hemorrhage with closed head trauma, which have not been validated. We included these codes to capture situations where a bleeding event preceded by a fall may have been coded as trauma-related. The codes for gastrointestinal hemorrhage have a PPV of 86-88%,21,22 and those defining all hospitalized bleeding have a PPV of 89%.21 The codes for AMI have a PPV between 89% and 97% in a variety of administrative



claims databases.22-27 Out-of-hospital death occurring within one day of an emergency department visit for acute ischemic heart disease was also classified as an AMI. Death was ascertained by linkage to the Social Security Master Beneficiary Record database, which provides the date, but not cause, of death and captures over 95% of deaths for US residents age 65 years or older.28 Our death outcome included deaths not preceded by a study outcome plus deaths within 30 days after hospitalization for an outcome event. Statistical analysis Analyses were performed on the propensity score-matched cohorts, thereby accounting for the potential confounding factors shown in Table 1 and the Supplemental Material. Incidence rates lo ots t w eree er were estimated using event counts and exposure follow-up time. Kaplan-Meier pplots were generated to characterize the contour of risk over time for each outcome. Cox proportional haaza zard rdss regression rd reegr greessiion was was used to compare time-to-event time-to o-ev e ent in dabigatran dabigattra ran compared co ompared mp hazards with warfarin reffer e ence) co oho hort r s. IIncidence ncid id den ence ce rrates atees an at andd C oxx m oddels w erre aalso lsso ggenerated ennera neratted ted to eexamine xami xami mine nee rrisk isk du duri r ng (reference) cohorts. Cox models were during prepr e-de edefi de f ne fi nedd intervals inte in terrvaals als of of time tim imee onn therapy the herrapy rapy (1-90 (11-990 90 days, day ays, s,, 911-18 1800 da 18 days yss, >1 >180 80 0 ddays) ays) ay s) bbecause ecau ec ause bbleeding au leeed ediing ing pre-defined 91-180 days, isks with war arrfa fari rinn ma mayy bee ggreatest r at re ates e t du es dduring ring ri ng tthe he ffirst irrst 3 m onth on thss af th afte terr in te init ittia iati tion ti on.29 St on Statistical Stat atis at isti is t cal ti risks warfarin months after initiation. significance was determined using 95% confidence intervals (CI) and 2-tailed p-values (p 0.05). Subgroup analyses were performed in categories defined by age, gender, hospitalization within 30-days prior to anticoagulant initiation, and chronic kidney disease. We also examined subgroups with concomitant use of SSRI antidepressants or prescription antiplatelet agents at anticoagulant initiation because they may increase bleeding risks with anticoagulants. Finally, we examined outcome risk by dabigatran dose (150 mg or 75 mg, twice daily). Lower-dose dabigatran (75 mg twice daily) was approved in the US for patients with severe renal impairment



(creatinine clearance 15-30 mL/min) on the basis of pharmacokinetic modeling.5 We performed sensitivity analyses using different definitions of cohort follow-up to assess whether the main analyses were affected by misclassification of exposed time. We did this by 1) restricting to patients with initial prescriptions of 30 days duration, 2) restricting to patients with at least two prescription fills of a study drug, and 3) increasing the gap allowance between anticoagulant prescriptions from 3 to 14 days. This study was performed as part of the SafeRx Project, a joint initiative of the Centers for Medicare & Medicaid Services (CMS) and the US Food and Drug Administration (FDA). It was approved by the Research in Human Subjects Committee of FDA’s Center for Drug Evaluation and Research. Analyses were performed f using R 3.0.2 (R Foundatio on fo or St Stat atis at isti is tica ti cal Foundation for Statistical Computing, Vienna, Austria) and SAS 9.2 (SAS Institute Inc., Cary, North Carolina).

R essults su Results A to tota tall of 667,494 ta 7,49 7, 4944 dabi ddabigatranabiigaatr tran an-- an nd 27 273, 3,92 9200 wa 92 w arf rfar rf arrin in--tre reeatted pat atie at ieent ntss we weree eeligible liigi g bl b e fo for st tuddy total and 273,920 warfarin-treated patients study nclusion. Da abi biga gatr ga tran tr an uusers sers rs ttended end nded e tto ed o be b yyounger, oung ou nger ng err, we w ree lless esss li es ike kely ly tto o ha have ve cchronic hrron onic ic kkidney i ney id inclusion. Dabigatran were likely disease, and were more likely to be treated by a cardiologist and to have received antiarrhythmics and prescription antiplatelet agents. A propensity score match was obtained for 67,207 (99.6%) new dabigatran users, resulting in cohorts closely balanced for all baseline covariates (Table 1 and Supplemental Table 1). Dabigatran patients contributed 18,205 person-years of on-therapy follow-up time and warfarin users contributed 19,382 person-years. Of note, 52.0% of dabigatran and 50.2% of warfarin users filled only a single prescription of their anticoagulant. During follow-up, there were 2,715 primary outcome events including 475 ischemic strokes, 1,628 major bleeding events, and 612 AMIs. Compared with warfarin, dabigatran use



was associated with reduced risk of ischemic stroke, intracranial hemorrhage, and mortality, and with an increased risk of major gastrointestinal bleeding (Table 2). There were no differences between cohorts in risk of AMI, or all hospitalized bleeding events. The absolute incidence of ischemic stroke, major gastrointestinal bleeding, intracranial hemorrhage, and death were substantially higher during the first 90 days of therapy than during later time periods for both dabigatran and warfarin (online Supplemental Figure 1). However, the point estimates of the hazard ratios did not vary substantively over intervals of 1-90, 91-180, or > 180 days of continuous anticoagulant use, although the confidence intervals widened due to lower numbers of events in the later intervals (Supplemental Table 3). Kaplan-Meier plots showed early separation eparation of survival curves for ischemic stroke with slightly later separation for orr iintracranial n raacr nt cran ania an iall ia hemorrhage, major gastrointestinal bleeding, and death (Figure 1). Subgroup that risk Subg Su bgrroup bg up p aanalyses n lyses stratified by age and ge na ggender nder showed tha hat ri ha isk ooff major a gastrointestinal dabigatran was women men bbleeding leeedi d ng with da dabi biigaatr tran nw as iincreased ncre nc reaase re ased ffor or w omeen agee 75 75 yyears ears aand ears nd oolder l er aand ld nd ffor orr m en n aage ge 885 ge 5 years older with warfarin Supplemental Tables and Below ye year arss an ar aandd ol olde derr comp ccompared omp mpaareed ed w ithh wa it warf rfar rf ariin in ((Table Tabl Ta blle 3 aand n Su nd upp ppleeme ment nttal a T ablees 4 an able nd 5). ).. Belo B elo ow these hese ages, gastrointestinal gast ga stro st r in ro inte t st te s in inall bleeding blee eedi ding di ng risk rissk was was comparable comp co mpar mp arab ar able le for for both both t anticoagulants. ant ntic icoa ic oagu oa gula gu laant nts. s. Th The point estimate for the risk of death with dabigatran compared with warfarin was reduced in all strata except women age 85 years and older, where it was increased compared with lower aged women (P (interaction) = 0.004). There were no interactions for ischemic stroke or intracranial hemorrhage. Results in other subgroups defined by chronic kidney disease, use of prescription antiplatelet agents or SSRI antidepressants, or in patients hospitalized in the 30 days before starting anticoagulant use were similar to the main analysis. The magnitude of effect for each outcome was greater in the subgroup treated with dabigatran 150 mg twice daily compared with the main analysis, which included patients treated



with the 150 mg and 75 mg doses (Table 4 and Supplemental Table 6). Approximately 16% of patients received dabigatran 75 mg twice daily and among these, none of the outcome comparisons were statistically significantly different from warfarin except for a lower risk of intracranial hemorrhage with dabigatran (Table 4). Only 33% of patients treated with the lower dose of dabigatran had a diagnosis of chronic kidney disease within the preceding six months, and of these, coding explicitly indicated severe renal impairment in only 20%. Lower-dose recipients were more likely to be older, to be receiving home health care or home oxygen, and to have higher CHADS2 and HAS-BLED scores (Supplemental Table 7). Sensitivity analyses yielded results similar to those of the primary analysis.

Discussion In n a large lar arge ge cohort coh ohortt of elderly Medicare beneficiari beneficiaries ies e w with ith non-valvular non-valv vul u ar A AF, F, risk of ischemic stroke, F, intracranial ntrracranial hem hemorrhage, emor orrh or rhag ag ge,, aand nd m mortality orta or t li ta l ty yw was as rreduced educed ed d and nd rrisk iskk of m is major ajor ggastrointestinal ajor ast stro ro oin inte test sttin inal al bbleeding leeed edin din i g was wa as increased incr in c ea cr ease sedd in patients se pat atie ieent ntss treated treaate tr tedd with witth wi th dabigatran dab abig igat ig atra rann compared ra comp co mpar mp arred with wit ithh warfarin. warf wa r arin rf ar n. The The levels lev eveels of rrisk iskk is were similarr in in direction d re di rect c io ct ionn and an nd magnitude maagn g it itud udde with w th those wi thos th osse observed o seerv ob rved ed in in the the randomized rand ra ndom nd omiz om i ed trial, iz tri rial a , RE al R RE-LY, -LY, where dabigatran 150 mg twice daily was compared with adjusted-dose warfarin therapy.6 The absolute incidence of outcome events for both dabigatran and warfarin was greatest during the first 90 days of treatment, although the hazard ratios for these outcomes were constant over time. Our results for gastrointestinal bleeding differed from those of a Mini-Sentinel Modular Program analysis that found a two-fold increase in incidence for warfarin compared with dabigatran.30 Modular Programs were unadjusted for any confounding factors and included substantial numbers of younger patients, in whom bleeding risks may be lower. Our results also differed from those of two small observational studies from Denmark. In the first, no difference in



thromboembolic or hospitalized bleeding risk was noted in765 new users of dabigatran compared with warfarin.31 In the second study, no difference in stroke or gastrointestinal bleeding risk was noted in 2239 patients initiating dabigatran compared with warfarin.32 In our study, the increased risk of major gastrointestinal bleeding with dabigatran appeared to be restricted to women age 75 years or older and to men age 85 years or older. The beneficial effect of dabigatran on mortality was not present in women age 85 years and older, where there was a trend for a higher risk of death with dabigatran compared with warfarin. This shift in hazard ratio between younger and older aged women represented a statistically significant interaction and suggests that the benefit-risk profile of dabigatran may be less favorable in women age 85 years and older than in other age-gender groups. Although Alt ltho houg u h many ug many subgroups ubgroups were examined in this analysis, it shouldd be noted that in RE-LY, an interaction was ob bse serv rv ved bbetween etw et ween en ttreatment reatment and age for major ga gast strrointestinal ble leeed e in ng bbut ut not stroke, or observed gastrointestinal bleeding 33,34 3,34 3,34 4 ntrracranial hem emor orrh or rhag ag ge,, ssimilar imil im ilar il ar to to our our findings fi inggs hhere. findi eree.33 er The nu The number umber mber off wo wom women men ag men agee 85 8 yyears eaarss or intracranial hemorrhage,

ol older lde derr wa w wass sm smal small alll in R RE-LY E-LY E-L LY an and nd aanalyses nd naly na lyse ly sees of aage gee aand nd m nd mortality orta tali ta litty li ty w were erre no nott re rep reported. port port rted ed d. Of note, not otte, the the 75 75 mg m dose dos o e of dabigatran dab a ig igat atra at raan was was approved appr ap p ov pr oved ed for for use usee in in patients pati pa tien ti e tss w en with i h se it ssevere vere renal impairment based on pharmacokinetic modeling rather than a randomized trial in patients. Our study represents the largest examination of the clinical effect of dabigatran 75 mg twice daily. Although we lacked laboratory data on creatinine clearance and are uncertain of the accuracy of kidney disease coding, our results suggest that many patients treated with this lower dose may not have had severe renal impairment, in which case, based on current product labeling,5 they should have been treated with the 150 mg dose. In the setting of moderate, mild, or no renal impairment, off-label use of the 75 mg dose might result in patients being under-dosed and could explain why we found no difference in risk of ischemic stroke, major gastrointestinal bleeding,



or mortality between warfarin and the lower dose of dabigatran. On the other hand, if most of the patients treated with the 75 mg dose actually had severe renal impairment, this would suggest that dabigatran dosing based on pharmacokinetic modeling was suboptimal. This raises the question whether patients treated off-label with the 75 mg dose would have experienced improved outcomes for ischemic stroke and mortality had they been treated with the 150 mg dose instead. We cannot answer this question with certainty because our comparison across dose levels was indirect and was not based on a head-to-head analysis. Of possible relevance here, the RE-LY trial suggested that a 110 mg twice daily dose of dabigatran was less effective than the 150 mg dose,6 and as a result, this lower dose was not approved for marketing in the US. This study had several limitations. It was observational and may be subje subject jeectt tto o confounding from factors not adjusted for in the analysis, such as over-the-counter aspirin or no non-steroidal onn-st steeroi st eroi oida dall anti-inflammatory da an nti ti--inf -in lammatory drug use. To reduce reeduc duce this possibility, possibil illit i y, w wee included an extensive variables agents, nnumber um mb of vari mber riab ablees in our ab ourr propensity pro rope pennsi nsity ity score sccore model, modeel,, including incclu udi ding ng prescription prescr crip ipti t on aantiplatelet ntip nt ipla laate t le lett ag age ents ents ts, an aand d close cl los osee balance b la ba lanc ncee for nc for these th hes esee factors factorrs was fact was achieved. achiev ach hiev eved ed. Nonetheless, ed Nonnetheeleess ss,, residual reesi s dual dual confounding conf onfoun founndiing by by unmeasured un nmeas meassur ured e factors cannot ott be be excluded. excl ex clud uded ud d. Medicare Med edic i ar ic are data d ta do da do not not capture capt ca p ur uree laboratory labo la bora bo rato ra tory ry results res esul u ts so ul so we had had a no basis upon which to assess the quality of warfarin anticoagulation. It is possible that the favorable effects of dabigatran on ischemic stroke and mortality and its adverse effect on major gastrointestinal bleeding in our study were at least partly due to low time in the therapeutic range with warfarin. However, this would not explain the reduced risk of intracranial hemorrhage with dabigatran. More importantly, whether warfarin management in our study was or was not adequate, it reflects the quality of anticoagulation likely to be experienced by patients treated with warfarin in the general practice setting in the US. In that context, our study results suggest that dabigatran is associated with generally better patient outcomes. Also, approximately 50% of



patients in each cohort received only a single prescription of their study anticoagulant. This represents the ambulatory care experience in Medicare. However, the constancy of hazard ratios in the time period beyond 180 days of use and the results of our sensitivity analysis restricted to patients receiving two or more prescriptions suggest that bias was not introduced by this limited persistency of use. In summary, in elderly Medicare beneficiaries with non-valvular AF, dabigatran was associated with a reduced risk of ischemic stroke, intracranial hemorrhage, and mortality and an increased risk of major gastrointestinal bleeding compared with warfarin. These associations were strongest for the 150 mg dabigatran dose, whereas the 75 mg dose was associated only with a reduced risk of intracranial hemorrhage.

Acknowledgments: Ackn kn now owle ledg le dgme dg ment nts: nt s Role of Sponsors: The authors auth hor orss are employees or ccontractors ontractors of the CMS or on the other officials CMS FDA the design he FDA; FD hhowever, ow wev ever er, ot er othe herr of he offi fici cial alss at tthe al he C MS aand nd d tthe he F DA hhad ad d nno o ro role le iin n th he de desi s gn aand si nd cconduct ondu on d c interpretation data; review, ooff the the study; the collection, co ollecctiion,, analysis, anal an alyssis, is, aand nd int terrprettat tation n ooff the the da ataa; or tthe hee ppreparation, reparattioon, re rep eview w, or approval appr ap prov pr oval ov al of of the the manuscript. manu ma nusscrript ript. The The manuscript manu anuscr uscr crip iptt was ip was subject suubj bjec e t to administrative ec adm dmin i is in istr t ativve tr ve review rev e ie i w prior pr or to prio to submission, content altered Disclaimer: The ubm bmis issi is sion si on, bu on butt th thee co cont nten nt entt wa en wass no nott al alte tere te redd by tthis re hiss rreview. hi evie ev iew ie w. D iscl is clai cl aime ai mer: me r: T he vviews iews ie ws eexpressed xpre xp ress re ssed ss ed aare re the authors necessarily those Department Health Human Services, he au auth thor orss an andd no nott ne nece cess ssar aril ilyy th thos osee of tthe he D epa part rtme ment nt ooff He Heal alth th aand nd H uman um an S ervi er vice cess tthe he CMS, or the FDA. Data access and integrity: Dr. MaCurdy had full access to all of the study data and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Funding Sources: This study was funded through an Intra-Agency Agreement between the Centers for Medicare & Medicaid Services (CMS) and the US Food and Drug Administration (FDA).

Conflict of Interest Disclosures: None.



References: 1. Aguilar MI, Hart R. Oral anticoagulants for preventing stroke in patients with non-valvular atrial fibrillation and no previous history of stroke or transient ischemic attacks. Cochrane Database Syst Rev. 2005:CD001927. 2. Lip GY, Edwards SJ. Stroke prevention with aspirin, warfarin and ximelagatran in patients with non-valvular atrial fi brillation: a systematic review and meta-analysis. Thromb Res. 2006;118: 321-333. 3. Hart RG, Pearce LA, Aguilar MI. Meta-analysis: antithrombotic therapy to prevent stroke in patients who have nonvalvular atrial fibrillation. Ann Intern Med. 2007;146:857-867. 4. Reynolds MW, Fahrbach K, Hauch O, et al. Warfarin anticoagulation and outcomes in patients with atrial fibrillation. A systematic review and meta-analysis. Chest. 2004;126:19381945. 5. Product label for Pradaxa (dabigatran etexilate mesylate). Available at: http://www.accessdata.fda.gov/drugsatfda_docs/label/2012/022512s011lbl.pdf. Accessed Acc ssed July Acce Jul ulyy 3, 2012. 6. Connolly SJ, Ezekowitz MD, Yusuf S, Eikelboom J, Oldgren J, Parekh A, Pogue J, Reilly PA, Themeles Th hem emel eles el es E, E, Varrone Varr Va rroone rr on J, Wang S, Alings M, Xavier Xavi vier vi er D, Zhu J, Diaz Diaaz R,, Lewis Lewis ew BS, Darius H, Diener RE-LY Steering Committee Diien ner H-C, H-C C, Joyner Joyn Joyn yner e CD, er CD, D Wallentin Wal alle lent ntin nt in L, L, and and the the RE RE-L -L LY St tee eeri r ng ri n C ommi mitt ttee tt e aand ndd IInvestigators. n es nv esti tiga gato ga tors rs.. rs Dabigatran versus warfarin patients with atrial fibrillation. Med. Dabbigatran bi ver ersu suus wa warf r arin rf arin iin n pa pati tiien nts w ith at atr rial fi ibrillla llatio atio on. N En Engl gl J M ed d. 2009 22009;161:11390099;1 161 61:1 : 13 1399 91151. 11151. 7. B Bauer KA. Reversal auer K au A R A. ever ersa sal off aantithrombotic nttit ithr hrom ombotiic agents. agen ag ents ts.. Am A J Hematol. Hem emat atol o . 2012;87:S119-S126. 201 012; 2;87 8 :S :S1199-S1 S1266. 8. Centers Medicare Medicaid Services. Medicare program general information. 8 C ente en ters rs ffor or M edic ed icar aree an andd Me Medi dica caid id S ervi er vice cess M edic ed icar aree pr prog ogra ram m ge gene nera rall in info form rmat atio ionn Available at: http://www.cms.hhs.gov/MedicareGenInfo/. Accessed June 26, 2012. 9. Centers for Medicare and Medicaid Services. Prescription drug coverage-general information. Available at: http://www.cms.hhs.gov/PrescriptionDrugCovGenIn/. Accessed June 26. 2012. 10. Ray WA. Evaluating medication effects outside of clinical trials: new-user designs. Am J Epidemiol. 2003;158:915-920. 11. 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. 12. Pisters R, Lane DA, Nieuwlaat R, de Vos CB, Crijns HJ, Lip GY. A novel user-friendly score (HAS-BLED) to assess 1-year risk of major bleeding in patients with atrial fibrillation: the Euro Heart Survey. Chest. 2010;138:1093–1100.



13. Lip GY, Frison L, Halperin JL, Lane DA. Comparative validation of a novel risk score for predicting bleeding risk in anticoagulated patients with atrial fibrillation: the HAS-BLED (Hypertension, Abnormal Renal/Liver Function, Stroke, Bleeding History or Predispostion, Labile INR, Elderly, Drugs/Alcohol Concomitantly) score. J Am Coll Cardiol. 2011;57:173-180. 14. Rosenbaum PR, Rubin DB. The central role of the propensity score in observational studies for causal effects. Biometrika. 1983;79:516-524. 15. Rubin DB, Thomas N. Matching using estimated propensity score: relating theory to practice. Biometrics. 1996;52:249-264. 16. Stuart EA. Matching methods for causal inference: a review and a look forward. Stat Sci. 2010;25:1-21. 17. Austin PC. Balance diagnostics for comparing the distribution of baseline covariates between treatment groups in propensity-score matched samples. Stat Med. 2009;28:3083-3107. 18. Tirschwell DL, Longstreth WT. Validating administrative data in stroke resear research. arch ch. Stroke. ch Stro St roke ke. 2002;33:2465-2470. 19. Roumie CL, Mitchel E, Gideon PS, Varas-Lorenzo C, Castellsague J, Griffin MR. Validation of ICD-9 IC CD-99 codes with a high positive predictive predictive value for incident stroke resulting in hospitalization using ho osp spit ital it aliz al izat iz atio at ion us io sin ingg Medicaid health data. Pharmacoepidemiol Pharrma macoepidemiol Drug D ugg Saf. Dr Saf. af 2008;17:20-26. 20. Kokotaila Hill MD. Coding factors 20. K okotaila RA RA,, Hi ill l M D. C D. odin od in ng ooff sstroke trokke andd sstroke trooke rrisk issk fa fac ctoorss using usin ng International Inttern In ternat a io at iona nall Classification Diseases, 10. Classsi s fication on off D iseas asees, revision revisiion 9 aand rev nd 10 0. Stroke. Strooke. 2005;36:1776-1781. 200 005; 5;36 6:17776 776-17 7811. 21. Cu Cunningham AW, Stein CM, Chung CP, Daugherty Smalley WE,, Ra Cunnin ingh g am A W, Ste tein i C M, C hungg C P, D augheert rtyy JR JR, Sm Smal alle leyy WE W Rayy WA WA. An A automated da database bleeding ata taba b se definition ba def efin i ittio ionn for for serious seeri r ou ouss bl blee eedi ee d ng ddue di u tto ue o or oral al aanticoagulant n iccoa nt oagu gula gu lant la nt uuse. se.. se Pharmacoepidemiol Phar Ph arma maco coep epid idem emio ioll Dr Drug ug SSaf. af 2011;20:560-566. 201 011; 1;20 20:5 :560 60-566 566 22. Wahl PM, Rodgers K, Schneeweiss S, Gage BF, Butler J, Wilmer C, Nash M, Esper G, Gitlin N, Osborn N, Short LJ, Bohn RL. Validation of claims-based diagnostic and procedure codes for cardiovascular and gastrointestinal serious adverse events in a commercially-insured population. Pharmacoepidemiol Drug Saf. 2010;19:596-603. 23. Petersen LA, Wright S, Normand ST, Daley J. Positive predictive value of the diagnosis of acute myocardial infarction in an administrative database. J Gen Intern Med. 1999;14:555-558. 24. Levy AR, Tamblyn RM, Fitchett D, McLeod PJ, Hanley JA. Coding accuracy of hospital discharge data for elderly survivors of myocardial infarction. Can J Cardiol. 1999;15:1277-1282. 25. Austin PC, Daly PA, Tu JV. A multicenter study of the coding accuracy of hospital discharge administrative data for patients admitted to cardiac care units in Ontario. Am Heart J. 2002; 144:290-296.



26. Kiyota Y, Schneeweiss S, Glynn RJ, Cannuscio CC, Avorn J, Solomon DH. Accuracy of Medicare claims-based diagnosis of acute myocardial infarction: estimating positive predictive value on the basis of review of hospital records. Am Heart J. 2004;148:99-104. 27. Choma NN, Griffin MR, Huang RL, Mitchel EF, Kaltenbach LA, Gideon P, Stratton SM, Roumie CL. An algorithm to identify incident myocardial infarction using Medicaid data. Pharmacoepidemiol Drug Saf. 2009;18:1064-1071. 28. Hill ME, Rosenwaike I. The Social Security Administration’s Death Master File: the completeness of death reporting at older ages. Social Security Bulletin 2001/2002; 64:45-51. Available at: http://www.socialsecurity.gov/policy/docs/ssb/v64n1/v64n1p45.pdf. Accessed June 26, 2012. 29. Hylek EM, Evans-Molina C, Shea C, Henault LE, Regan S. Major haemorrhage and tolerability of warfarin in the first year of therapy among elderly patients with atrial fibrillation. Circulation. 2007;115:2689-2696. reports 30. Southworth MR, Reichman ME, Unger EF. Dabigatran and postmarketing repor orts t of bleeding. N Engl J Med. 2013;368:1274-1278. 31. Sørensen R, Gislason G, Torp-Pedersen C, Olesen JB, Fosbøl EL, Hvidtfeldt MW, Karasoy D, Lamberts M, Charlot M, Weeke P, Lip GYH, Hansen ML. Dabigatran use in Danish atrial fibrillation 2013;3:e002758. fibr ril illa lati tion on patients pat a ient ntss in 2011: a nationwide study. BM BMJ Open. 2013; 3 3:e0 002 02758. 32. Larsen TB, Rasmussen Rasmu usssen n LH, LH, H Skjøth Skj kjøøthh F, F, Due ue KM, KM M, Callréus Caalllréuus T, Rosenzweig Rosenzzweig Ro g M, M, Lip Lip GYH. GYH. Efficacy safety etexilate warfarin “real-world” with Effi Ef ica c cy and safe etyy off dabigatran dabig gattra r n etex e ilaate aand ndd wa arffarin in iin n “r “real-world”” ppatients attien ents w ithh aatrial trrial fibrillation. Coll Cardiol. fibr bril illa lati t on n. J Am mC olll C ardio iol. l. 22013;61:2264-2273. 0133;661:2 01 :226 264--22 2 73 73. Thompson Medical review Application Number 33. Beasleyy N, T homp ho mpso son A. M edic ed ical a oofficer ffiice ff cerr re revi view ew off Ap Appl plic icat atio i n Nu Numb mber er 222-512 2 51 25122 (dabigatran). Research, Drug Administration, August, dab abig igat atra ran) n) Center Cent Ce nter er for for Drug Dru rugg Evaluation Eval Ev alua uati tion on and and Re Rese sear arch ch Fo Food od aand nd D rugg Ad ru Admi mini nist stra rati tion on Au Augu gust st 2010. Available at: http://www.accessdata.fda.gov/drugsatfda_docs/nda/2010/022512Orig1s000MedR.pdf. Accessed December 16, 2013. 34. Eikelboom JW, Wallentin L, Connolly SJ, Ezekowitz M, Healey JS, Oldgren J, Yang S, Alings M, Kaatz S, Hohnloser SH, Diener H-C, Franzosi MG, Huber K, Varrone J, Yusuf S. Risk of bleeding with 2 doses of dabigatran compared with warfarin in older and younger patients with atrial fibrillation. An analysis of the Randomized Evaluation of Long-Term Anticoagulant Therapy (RE-LY) Trial. Circulation. 2011;123:2363-2372.



Table 1. Sociodemographic factors, medical conditions, and medication use at baseline in propensity score matched Medicare beneficiaries initiating dabigatran or warfarin for atrial fibrillation from 2010-2012. Additional factors included in the propensity score model are shown in the Supplemental Material. Characteristic Age-group 65-74 75-84 85+ Female Race/ethnicity White Black Other Medical history General Diabetes Hypercholesterolemia Hypertension Kidney failure Acute Chronic Chr hroonic icc Obesity O es Ob esit itty Peptic uulcer disease lcer lc e dis er i ea is ease se Prior bleeding bleeddinng event bl eveent Hospitalized H ospi os pita taaliize zedd Not N ot hhospitalized o pi os pita itali lizedd li S mokin kin ng Smoking Car ardi diov ovas ascu cula larr di dise seas asee Cardiovascular disease Acute myocardial infarction Past 1-30 days Past 31-183 days Coronary revascularization Heart failure Hospitalized Outpatient Other ischemic heart disease Stroke Past 1-30 days Past 31-183 days Other cerebrovascular disease Transient ischemic attack Cardioablation Cardioversion Other medical conditions Falls Fractures

Dabigatran (%; n=67,207)

Warfarin (%; n=67,207)

Standardized mean difference

42 43 16 51

41 43 16 52

0.01 0.01 0.00 0.01

92 3 5

92 3 5

0.00 0.00 0.00

33 74 87

34 74 87

0.00 0 00 0. 00.00 .00 00 0.00 0.00

5 133 11

Dabigatran etexilate: An alternative to warfarin for patients with nonvalvular atrial fibrillation.

Dabigatran, Rivaroxaban, or Apixaban versus Warfarin in Patients with Nonvalvular Atrial Fibrillation: A Systematic Review and Meta-Analysis of Subgroups.

Management of Major Bleeding Events in Patients Treated With Dabigatran for Nonvalvular Atrial Fibrillation: A Retrospective, Multicenter Review.

Dabigatran exhibits low intensity of left atrial spontaneous echo contrast in patients with nonvalvular atrial fibrillation as compared with warfarin.

Warfarin or dabigatran for treatment of atrial fibrillation.

Effectiveness and Safety of Dabigatran, Rivaroxaban, and Apixaban Versus Warfarin in Nonvalvular Atrial Fibrillation.

A novel coagulation parameter monitoring bleeding tendency of Chinese nonvalvular atrial fibrillation patients prescribing dabigatran etexilate.

Cost-Effectiveness of Dabigatran (150 mg Twice Daily) and Warfarin in Patients ≥ 65 Years With Nonvalvular Atrial Fibrillation.

NOACs vs. Warfarin for Stroke Prevention in Nonvalvular Atrial Fibrillation.

Bleeding events among new starters and switchers to dabigatran compared with warfarin in atrial fibrillation.

Characteristics of patients with non-valvular atrial fibrillation using dabigatran or warfarin in the US.

Rivaroxaban versus warfarin or dabigatran in patients undergoing catheter ablation for atrial fibrillation: A meta-analysis.

Warfarin use, mortality, bleeding and stroke in haemodialysis patients with atrial fibrillation.

Cardiovascular and mortality risks in older Medicare patients treated with varenicline or bupropion for smoking cessation: an observational cohort study.

Dabigatran Versus Warfarin for Direct Current Cardioversion in Atrial Fibrillation.

Population-based cohort study of warfarin-treated patients with atrial fibrillation: incidence of cardiovascular and bleeding outcomes.

Appropriateness of prescribing dabigatran etexilate and rivaroxaban in patients with nonvalvular atrial fibrillation: a prospective study.

Bleeding outcomes associated with rivaroxaban and dabigatran in patients treated for atrial fibrillation: a systematic review and meta-analysis.

Risk of bleeding with dabigatran in atrial fibrillation.

The changing characteristics of atrial fibrillation patients treated with warfarin.

Dabigatran Versus Warfarin in Atrial Fibrillation: Multicenter Experience in Turkey.

Impaired renal function and bleeding in elderly treated with dabigatran.

International normalized ratio stability in warfarin-experienced patients with nonvalvular atrial fibrillation.

International normalized ratio stabilization in newly initiated warfarin patients with nonvalvular atrial fibrillation.