Assessment of Dabigatran Utilization and Prescribing Patternsfor Atrial Fibrillation in a Physician Group Practice Setting

Blake Carley, PharmDa,*, Sara Griesbach, PharmDa, Tonja Larson, PharmDa, and Kori Krueger, MD,MBAb

For years, warfarin and aspirin have been standard therapies for prophylaxis of stroke in

aClinical Pharmfield, Wisconsin anMarshfield Clinic2013; revised man

Funding for thEducation Residen

See page 654*CorrespondinE-mail addres

0002-9149/13/$ -http://dx.doi.org/1

atrial fibrillation. In late 2010, dabigatran, an oral direct thrombin inhibitor, becameavailable for use in nonvalvular atrial fibrillation. We sought to evaluate utilization andprescribing patterns of dabigatran in a physician group practice setting. We retrospectivelycollected prescription data from October 2010 to December 2011 including indication ofuse, dose, renal function, drug interactions, history of warfarin therapy, and risk assessmentscores (CHADS2 and HAS-BLED). Off-label use (history of valve disease or no diagnosis ofatrial fibrillation) occurred in 20% (n [ 34) of 174 patients. Renal function assessed byCockcroft-Gault equation identified 1 case of contraindicated use and the need for initialrenal dose adjustment in approximately 1/2 of the patients with reduced renal function(15-30 ml/min). Review of anticoagulant use revealed 68% of patients (n [ 119) previouslyreceived warfarin and ultimately 20% of all patients on dabigatran resumed warfarintherapy. A significant increase in the use of permeability glycoprotein inhibitors and protonpump inhibitors after initiating dabigatran was observed. Nearly 10% of patients had aCHADS2 score of 0. For patients receiving novel oral anticoagulants, prospective inclusionin anticoagulation services and guidance from specific “place in therapy” statements havepotential to play a large role in maximizing safety while aiding in continuedresearch. � 2014 Elsevier Inc. All rights reserved. (Am J Cardiol 2014;113:650e654)

Use of the Marshfield Clinic Anticoagulation Service,which manages approximately 10,000 unique patientsannually or about 7,500 any given month, allows patientstaking warfarin to be closely followed and monitoredby specialists in anticoagulation. This provides improvedmanagement and prospective collection of patient data in alarge population. The addition of novel oral anticoagulantssuch as dabigatran, rivaroxaban, and apixaban to anti-coagulation monitoring services in any health-care settingmay allow these new agents to be tracked closely aswell, which could potentially offer data on prescribingpractices, adverse events, and efficacy. At the time of thisstudy, apixaban and rivaroxaban were still under review bythe Food and Drug Administration (FDA) for the preventionof stroke in patients with nonvalvular atrial fibrillation; theprimary objective of this quality improvement initiative wasto assess the utilization and prescribing patterns of dabiga-tran in a physician group practice setting. As a secondaryobjective, this initiative assessed the benefit versus the riskof anticoagulation therapy by comparison of CHADS2

1 andHAS-BLED2 scores.

acy Services Department of Marshfield Clinic, Marsh-d bInstitute for Quality, Innovation, and Patient Safety of, Marshfield, Wisconsin. Manuscript received July 30,uscript received and accepted November 7, 2013.is study was provided by Marshfield Clinic’s Division oft Research Program (Marshfield, Wisconsin).for disclosure information.g author: Tel: (715) 221-8755; fax: (715) 221-7880.s: carley.blake@marshfieldclinic.org (B. Carley).

see front matter � 2014 Elsevier Inc. All rights reserved.0.1016/j.amjcard.2013.11.008

Methods

This retrospective study included patients aged�18 yearswho had taken or were currently taking dabigatran fromOctober 19, 2010 (FDA approval date) to December 2, 2011.Patients who did not have a primary care provider within theMarshfield Clinic System were excluded, and for all patientsincluded, the specialty of the prescriber for dabigatranwas collected. The study was approved by the MarshfieldClinic Institutional Review Board. Most data were extractedelectronically, but consumption of alcohol and review ofinitial renal dose adjustments required manual chart review.Confirmation of atrialfibrillation diagnosiswas alsomanuallyreviewed to confirm the accuracy of the electronic data ex-traction. Drug interactions were assessed by observing ratesfor interacting drugs starting 7 days before therapy and duringdabigatran therapy in an attempt to see if dabigatran therapyinfluenced the interaction rates (95% confidence interval wasderived using the exact binomial distribution). Storage of datawas secured using computer files with restricted access.

Descriptive statistics includedmean, SD,median, and rangefor the continuous variables (age, height, weight, creatinineclearance, estimated glomerular filtration rate, CHADS2 score,and HAS-BLED score) and the frequency and percentage forthe categorical variables (gender, race, and diagnosis of atrialfibrillation [valvular vs nonvalvular]). All statistical analyseswere carried out using a commercially available statisticalsoftware package, SAS (version 9.2; English; Cary, NorthCarolina).

Results

Thebaseline characteristics of the 174 patients are displayedin Table 1. Nearly all patients were white and approximately

www.ajconline.org

Table 1Characteristics of population receiving dabigatran (n ¼ 174)

Characteristic n (%) Mean SD Median Range

Age (yrs) 174 (100) 70.8 12.2 72 27e94Men 97 (56) — — — —

Women 77 (44) — — — —

White 167 (96) — — — —

Asian 2 (1.1) — — — —

Black 1 (0.6) — — — —

Native Hawaiian/Pacific Islander 1 (0.6) — — — —

Not specified 3 (1.7) — — — —

Height (cm) 173 (99) 170.8 11 172.7 139.7e193Weight (kg) 174 (100) 92.6 25.9 90.4 44.5e178.1Renal functionCreatinine clearance (ml/min) 173 (99) 61.1 25.6 57 13.3e142.1Estimated glomerular filtration rate (ml/min) 146 (84) 62 16 61 20.5e89.3

CHADS2 174 (100) 2.0 1.3 2.0 0e6HAS-BLED 174 (100) 3.0 1.3 3.0 0e6

Table 2Reason for dabigatran use

Variable Atrial Fibrillation Total

Yes No

Nonvalvular 140 (80.5) 13 (7.5) 153 (88)Valvular disease 19 (10.9) 2 (1.1) 21 (12.1)Total 159 (91.4) 15 (8.6) 174 (100)

Data are presented as n (%).

Table 3Drug interactions

Drug Patients

Before PrescribingDabigatran

After PrescribingDabigatran

95% CI, %

Proton pumpinhibitors

46 (26) 59 (34) 26.9e41.5*

P-glycoproteininducers

1 (1) 1 (1) 0.0e3.2

P-glycoproteininhibitors

74 (43) 116 (67) 59.1e73.6*

Anticoagulants† 4 (2) 7 (4) 1.6e8.1Platelet aggregation

inhibitors16 (9) 10 (6) 2.8e10.3

Nonsteroidalantiinflammatorydrug (NSAID)

30 (17) 37 (21) 15.4e28.1

Note: Some drugs included in this data do not have documented orlabeled interactions with dabigatran, but based on P-glycoprotein inhibitionor inducement, pH altering properties, or propensity to cause bleeding maytheoretically interact with dabigatran (see Discussion section).Data are presented as n (%).CI ¼ confidence interval.* Statistically significant for the percentage after compared with the

percentage before.† Lepirudin and enoxaparin were the only anticoagulants identified, but it

was not determined if these were actually used simultaneously with dabi-gatran or used when transitioning on or off dabigatran therapy.

Arrhythmias and Conduction Disturbances/Dabigatran for AF in Physician Group Practice Setting 651

1/2 were men. The mean age was 70.8 years with a range from27 to 94. Notably, 40% (n ¼ 70) of the patients were aged�75 years.

Results for assessment of indication for dabigatran ther-apy are displayed in Table 2. Off-label use was identified ata rate of 20% and was defined as use in patients with ahistory of valve disease or no diagnosis of atrial fibrillationor atrial flutter. The primary prescribers of dabigatranincluded cardiology (38%) and internal medicine and/orfamily medicine (38%).

Use of the creatinine clearance identified 1 case of contra-indicated use (<15 ml/min) and 13 candidates for renal doseadjustments (15 to 30 ml/min). More than 1/2 of those patients(n¼ 8, 62%) did not receive an initial renal dose adjustment fordabigatran. In contrast, the estimated glomerular filtration rateidentified no cases of contraindicated use (<15 ml/min) andonly 7 patients requiring a renal dose adjustment. Of thesepatients, just under 1/2 (n ¼ 3, 43%) did not receive an initialrenal dose adjustment for dabigatran.

Assessing history of anticoagulant use showed that 57% ofpatients (n ¼ 99) had previously been using warfarin beforeusing dabigatran, and 12%of patients (n¼ 20)were transferredfrom warfarin to dabigatran and then back to warfarin. Addi-tionally, 24% of patients (n ¼ 42) with no history of warfarinusewere started on dabigatran therapy, and 8%of patients (n¼13) were initially started on dabigatran then later switched towarfarin. Reasons for the changes were not assessed. Mostpatients (98%) experienced “true switches,” whereas theremainder were “false switches.”An example of a false switchis a patient taking warfarin who is prescribed dabigatran but

decides not to take it and continues on warfarin therapy, whichwould document a change within the electronic medical re-cords but a change in therapy has not truly occurred.

The assessment of drug interactions is reported in Table 3.Statistically significant increases in the prescribing rates forproton pump inhibitors and permeability glycoprotein in-hibitors after initiation of dabigatran therapy were observed.The latter primarily consistedof cardiovascular drugs includingamiodarone, dronedarone, diltiazem, and carvedilol.

The percentage of patients in each category of theCHADS2

1 score (0 to 6) and HAS-BLED2 score (0 to 9) arepresented in Figures 1 and 2, respectively, and the results of

Figure 1. Percentage of patients in CHADS2 score categories (n ¼ 174).

Figure 2. Percentage of patients in HAS-BLED score categories (n ¼ 174).

652 The American Journal of Cardiology (www.ajconline.org)

the individual comparison of scores for each patient aredepicted in Figure 3.

Discussion

This study focused on the prescribing patterns of dabiga-tran therapy but did not assess adverse events. The internalprescribing rate for dabigatran during the study period, whichaccounted for 2% of the total patient population on oralanticoagulation, appears relatively low compared with the

number of patients taking warfarin. At the time of this study,apixaban and rivaroxaban were still under review by the FDAfor the prevention of stroke in patients with nonvalvular atrialfibrillation. Review of dabigatran prescribing patternsaccentuated strategies related to indication, drug interactions,and renal function, which helped to prepare the anti-coagulation service to provide safer care with dabigatran orother novel oral anticoagulants.

The FDA indication(s) for dabigatran and other novel oralanticoagulants varies greatly from warfarin. Assessment is

Figure 3. Individual comparison of CHADS2 score versus HAS-BLED score.

Arrhythmias and Conduction Disturbances/Dabigatran for AF in Physician Group Practice Setting 653

important if patients are switched from warfarin to a noveltherapy. Results showed that 2 of 3 patients (68%) takingdabigatran previously used warfarin, and ultimately 20% ofpatients transitioned from dabigatran to warfarin therapy.Reasons for altering therapy were not assessed. In 2012, theFDAwarned health care providers of increased risk for stroke,myocardial infarction, and clotting compared with warfarinwhen dabigatran is used in patients with mechanical heartvalves. This evidence from the Randomized, Phase II Studyto Evaluate the Safety and Pharmacokinetics of Oral Dabiga-tran Etexilate in Patients after Heart Valve Replacement(RE-ALIGN)3 trial also showed that dabigatran caused morebleeding after valve surgery compared with warfarin. Dabi-gatran may reduce some of the challenges faced when usingwarfarin therapy, but it should only be used for the FDA-approved indication. Rates of off-label use were not assessedamong the various prescriber specialties in this study, but thisquality improvement initiative has aided in supporting devel-opmental changes for prospective inclusion of all patientswithin anticoagulation services.

Development of novel oral anticoagulants has reduced butnot eliminated the arduous challenge of managing drug in-teractions. Drugs included in this study may not have anylabeled or documented interactions with dabigatran; however,based on permeability glycoprotein inhibition or inducement,pH altering properties, or propensity to cause bleeding, theymay theoretically interact.4e7 Considering interactions thatmay reduce drug levels, current prescribing information doesnot provide a specific recommendation on proton pumpinhibitors but recommends avoiding permeability glycoproteininducers while only identifying rifampin. When consideringinteractions that may increase drug levels, verapamil, amio-darone, quinidine, and clarithromycin are listed by the manu-facturer as permeability glycoprotein inhibitors that do notrequire dose adjustments, but included is a warning to avoid

extrapolation of this recommendation to other permeabilityglycoprotein inhibitors of which there are many. Consumptionof dabigatran at least 2 hours before a permeability glycopro-tein inhibitor allows sufficient time for conversion of the pro-drug to active drug or eliminationwhich helps reduce the effectof the interaction. Patient counseling on administration andadherence along with frequent monitoring of interactions andprescribing practices that avoid drugs that may reduce dabi-gatran levels or increase the risk for bleeding could mitigateassociated risks of drug interactions.4e7 The importance ofcounseling is underscored in this study by statistically signifi-cant increases in prescribing rates for permeability glycopro-tein inhibitors, which were primarily cardiovascular drugs, andproton pump inhibitors. The acute need for cardiovasculartherapies in these patients and the potential for increasedgastrointestinal side effects from dabigatran may explain theincreased rates of prescribing for these drugs. Additionalresearch is necessary to understand the full impact of thesepotential interactions with dabigatran as well as other in-teractions associated with new oral anticoagulants.

Use of the Cockcroft-Gault equation (creatinine clearance)compared with the Modification of Diet in Renal Diseaseequation (estimated glomerular filtration rate) identified twiceas many patients requiring renal adjustments and 1 case ofcontraindicated use. Assessment by either method demon-strated that approximately 1/2 of the patients requiring a renaladjustment did not initially receive one. Although dosing isavailable for creatinine clearance <30 ml/min,8 in ouropinion, avoiding use in this population is prudent; there isincreased risk for exposure and limited evidence of clinicaloutcomes in this population.9,10

Comparison of the CHADS21 and HAS-BLED2 scores

showed that more patients had a higher risk for bleeding thanfor stroke within a 1-year period while taking an anticoagu-lant; however, without end point data such as stroke or

654 The American Journal of Cardiology (www.ajconline.org)

bleeding rates, this comparison of scores is extremely limited.In the Euro Heart Survey on Atrial Fibrillation study,2 it wasdetermined retrospectively that a direct comparison of theCHADS2 and HAS-BLED scores to guide anticoagulationuse would have withheld oral anticoagulation use in 12% ofthe patients who suffered a major bleed within 1 year ontherapy and helped to initiate oral anticoagulation in 95% ofthe patients at high risk for stroke who were dischargedwithout therapy and experienced a stroke within 1 year. Ourstudy showed that 10% of patients on dabigatran had aCHADS2 score of 0 and 31% had a score of 1. In previousAmerican College of Chest Physicians guidelines (eighthedition), aspirin therapy was recommended for patients with ascore of 0 (grade 1B). Antithrombotic therapy with eitherwarfarin (grade 1A) or aspirin (grade 1B) could be used inpatients with a score of 1, with a preference for warfarin overaspirin (grade 2A), and in patients with a score of �2,antithrombotic therapy with warfarin was preferred (gra-de1A).11 The 2012 American College of Chest Physiciansguidelines (ninth edition) recommend using no therapy, overantithrombotic therapy, for patients with a CHADS2 score of0 (each grade 2B), but if patients choose to use antithrombotictherapy, aspirin is recommended over oral anticoagulation(grade 2B). Antithrombotic therapy with oral anticoagulantsfor patientswith aCHADS2 score of�1 is recommendedwithvarious levels of evidence over no therapy, aspirin, or com-bination aspirin and clopidogrel, and notably dabigatran isrecommended over warfarin (grade 2B).10

Limitations of the study included retrospective design,small sample size, and a limited study duration. Study dura-tion and proximity to dabigatran approval caused reluctancefor collection of data on outcomes (e.g., stroke or bleeding),which is a future direction. Additionally, comparison of theCHADS2

1 score and theHAS-BLED2 score is limited as thesehave not been proven to be more effective at predicting risk ofbleeding or stroke than clinical judgment.

According to the Institute for Safe Medication Practices,during 2011 dabigatran was the most frequently identifiedmedication involving direct safety-related reports to the FDA,which accounted for 3,781 domestic serious adverse events,including 542 patient deaths. Reports included hemorrhage(2,367 cases), acute renal failure (291), stroke (644), and sus-pected cases of liver failure (15). The Institute for Safe Medi-cationPractices annual report states “the approval of dabigatranis a prime example of a decision that should now be reassessedto determine whether additional measures can facilitate saferuse.”12Reports have also highlighted the risk of bleeding in theelderly and thosewith reduced renal function or bodyweight.13

TheMarshfield Clinic created “place in therapy” statementsbased on indication to help guide clinicians prescribing noveloral anticoagulants. None of these new oral anticoagulants arepreferred over each other or warfarin. For dabigatran, the rec-ommendations include use only for the FDA-approved indi-cation in nonvalvular atrial fibrillation under the followingconditions: age<75 years, estimated glomerular filtration rate>30 ml/min, no additional indications for warfarin, and diffi-culty maintaining a stable international normalized ratio onwarfarin. Additional recommendations are to avoid use in hy-percoagulable conditions, prosthetic heart valves, deep veinthrombosis, pulmonary embolism, or stroke. To maximizesafety, Marshfield Clinic is using a system-based approach of

prospective inclusion in anticoagulation services for all patientsreceiving dabigatran or other novel oral anticoagulants.

Acknowledgment: The authors thank Connie Folz, PharmD,BCPS, and Stuart Guenther, RPh, of the Marshfield Clinic’sInvestigationalDrugProgram;DebraKempf,BSN,MarshfieldClinic resident research facilitator; Po-Huang Chyou, PhD,director of biostatistics, Aaron Miller, PhD, biomedical infor-matics, Valerie McManus, MPH, research program analyst,and Carla Rottscheit, BSCS, senior research programmer an-alyst of the Marshfield Clinic Research Foundation’sBiomedical Informatics Research Center; and Marie Fleisner,AAS of theMarshfield Clinic Research Foundation’s Office ofScientific Writing and Publication.

Disclosures

None of the authors or those acknowledged has any rele-vant financial interest or other relationship(s) with a com-mercial entity producing health care related products and/orservices (i.e., no conflicts of interest to disclose).

1. Gage BF, Waterman AD, Shannon W, Boechler M, Rich MW, RadfordMJ. Validation of clinical classification schemes for predicting stroke:results from the National Registry of Atrial Fibrillation. JAMA 2001;285:2864e2870.

2. Pisters R, Lane DA, Nieuwlaat R, de Vos CB, Crijna HJ, Lip GY.A novel user-friendly score (HAS-BLED) to assess 1-year risk of majorbleeding in patients with atrial fibrillation. Chest 2010;138:1093e1100.

3. FDA Drug Safety Communication: Pradaxa (Dabigatran Etexilate Mesy-late) Should Not Be Used in Patients With Mechanical Prosthetic HeartValves. U.S: Food and Drug Administration. December 12, 2012. Avail-able at: http://www.fda.gov/Drugs/DrugSafety/ucm332912.htm. AccessedFebruary 21, 2013.

4. Pradaxa� (Dabigatran) [package Insert United States]. Ridgefield, CT:Boehringer-Ingelheim Pharmaceuticals, Inc; 2012.

5. Pradaxa� (Dabigatran) [package Insert Worldwide]. Rhein, Germany:Boehringer-Ingelheim; 2012.

6. Garnock-Jones KP. Dabigatran etexilate: a review of its use in pre-vention of stroke and systemic embolism in patients with atrial fibril-lation. Am J Cardiovasc Drugs 2011;11:57e72.

7. Horn JR, Hansten PD. Dabigatran: a new oral anticoagulant. Pharm Times2010;76: Online only. Available at: http://www.pharmacytimes.com/publications/issue/2010/December2010/DrugInteractions-1210. AccessedFebruary 21, 2013.

8. Hariharan S, Madabushi R. Clinical pharmacology basis of derivingdosing recommendations for dabigatran in patients with severe renalimpairment. J Clin Pharmacol 2012;52:119Se125S.

9. Mack DR, Kim JJ. Pharmacokinetic and clinical implications of dabi-gatran use in severe renal impairment for stroke prevention in non-valvular atrial fibrillation. Ann Pharmacother 2012;46:1105e1110.

10. Guyatt G, Akl EA, Crowther M, Gutterman DD, Schuunemann HJ,American College of Chest Physicians Antithrombotic Therapy andPrevention of Thrombosis Panel. Executive summary: antithrombotictherapy and prevention of thrombosis, 9th ed: American College ofChest Physicians evidence-based clinical practice guidelines. Chest2012;141:7Se47S.

11. Singer DE, Albers GW, Dalen JE, Fang MC, Go AS, Halperin JL, LipGY, Manning WJ, American College of Chest Physicians. Antith-rombotic therapy in atrial fibrillation: American College of Chest phy-sicians evidence-based clinical practice guidelines (8th edition). Chest2008;133:546Se592S.

12. Institute for Safe Medication Practices. Quarter Watch Monitoring FDAMedWatch Reports. Anticoagulants the Leading Reported Drug Risk in2011. Horsham, PA: ISMP;May 31, 2012. Available at: http://www.ismp.org/quarterwatch/pdfs/2011Q4.pdf. Accessed February 21, 2013.

13. Harper P, Young L, Merriman E. Bleeding risk with dabigatran in thefrail elderly. N Engl J Med 2012;366:864e866.