Advertisement
Mayo Clinic Proceedings Home

Effectiveness and Safety of Oral Anticoagulants in Patients With Nonvalvular Atrial Fibrillation and Diabetes Mellitus

      Abstract

      Objective

      To address gaps in the data comparing non–vitamin K antagonist oral anticoagulants (NOACs) and warfarin among patients with nonvalvular atrial fibrillation (NVAF) and diabetes.

      Patients and Methods

      A retrospective study was conducted on patients with NVAF and diabetes newly initiating apixaban, dabigatran, rivaroxaban, or warfarin from January 1, 2013, through September 30, 2015, with Medicare data from the US Centers for Medicare & Medicaid Services and 4 other US commercial claims databases. One-to-one propensity score matching was completed between NOACs and warfarin and between NOACs in each database, and the results were pooled. Cox proportional hazards models were used to evaluate the risk of stroke/systemic embolism (SE) and major bleeding (MB).

      Results

      A total of 154,324 patients were included in the 6 matched cohorts, with a mean follow-up time of 6 to 8 months. Compared with warfarin, apixaban (hazard ratio [HR], 0.67; 95% CI, 0.57-0.77) and rivaroxaban (HR, 0.79; 95% CI, 0.71-0.89) were associated with a lower risk of stroke/SE; dabigatran (HR, 0.84; 95% CI, 0.67-1.07) was associated with a similar risk of stroke/SE. Apixaban (HR, 0.60; 95% CI, 0.56-0.65) and dabigatran (HR, 0.78; 95% CI, 0.69-0.88) were associated with a lower risk of MB; rivaroxaban (HR, 1.02; 95% CI, 0.94-1.10) was associated with a similar risk of MB compared with warfarin. Compared with dabigatran and rivaroxaban, apixaban was associated with a lower risk of MB. Compared with rivaroxaban, dabigatran was associated with a lower risk of MB.

      Conclusion

      This study—the largest observational study to date of patients with NVAF and diabetes taking anticoagulants—found that NOACs were associated with variable rates of stroke/SE and MB compared with warfarin.

      Trial Registration

      clinicaltrials.gov Identifier: NCT03087487

      Abbreviations and Acronyms:

      AF (atrial fibrillation), ARISTOPHANES (Anticoagulants for Reduction In STroke: Observational Pooled analysis on Health outcomes ANd Experience of patientS), ARISTOTLE (Apixaban for Reduction In STroke and Other ThromboemboLic Events in Atrial Fibrillation), CHA2DS2-VASc (congestive heart failure, hypertension, age ≥75 years, diabetes, stroke (previous), vascular disease, age 65-74 years, sex), ENGAGE-AF (Effective Anticoagulation with Factor Xa Next Generation in Atrial Fibrillation), GI (gastrointestinal), HAS-BLED (hypertension, abnormal (renal/liver function), stroke, bleeding, labile (international normalized ratio), elderly, drug/alcohol/medication (usage history)), HR (hazard ratio), ICD-9-CM (International Classification of Diseases, Ninth Revision, Clinical Modification), INR (international normalized ratio), MB (major bleeding), NOAC (non–vitamin K oral anticoagulant), OAC (oral anticoagulant), PSM (propensity score matching), RE-LY (Randomized Evaluation of Long-term anticoagulant therapy), ROCKET-AF (Rivaroxaban Once-daily Oral Direct Factor Xa Inhibition Compared with Vitamin K Antagonism for Prevention of Stroke and Embolism Trial in Atrial Fibrillation), SE (systemic embolism)
      There have been important developments in the management of atrial fibrillation (AF), including the evolution of approaches to stroke prevention and bleeding risk minimization, specifically through the emergence of oral anticoagulants (OACs).
      • Lip G.
      • Freedman B.
      • De Caterina R.
      • Potpara T.S.
      Stroke prevention in atrial fibrillation: past, present and future.
      Vitamin K antagonists, such as warfarin, have previously dominated the therapeutic market of AF, and non–vitamin K oral anticoagulants (NOACs) have had increasing presence since their approval and inclusion in AF clinical guidelines in recent years. Warfarin is associated with a higher risk of major bleeding (MB) as compared with no antithrombotic treatment.
      • Hart R.G.
      • Pearce L.A.
      • Aguilar M.I.
      Meta-analysis: antithrombotic therapy to prevent stroke in patients who have nonvalvular atrial fibrillation.
      Moreover, required periodic monitoring of the international normalized ratio (INR) and the frequent need for dose adjustment make warfarin inconvenient and burdensome to the treatment of AF.
      • De Caterina R.
      • Husted S.
      • Wallentin L.
      • et al.
      Vitamin K antagonists in heart disease: current status and perspectives (Section III). Position paper of the ESC Working Group on Thrombosis—Task Force on Anticoagulants in Heart Disease.
      In randomized controlled trials comparing NOACs with warfarin, the 4 NOACs—apixaban, dabigatran, edoxaban, and rivaroxaban—have exhibited noninferiority for safety and efficacy.
      • Ruff C.T.
      • Giugliano R.P.
      • Braunwald E.
      • et al.
      Comparison of the efficacy and safety of new oral anticoagulants with warfarin in patients with atrial fibrillation: a meta-analysis of randomised trials.
      Concomitant comorbidities, such as diabetes, can complicate AF management. Diabetes increases the risk of developing AF and is one of the most common comorbidities in patients with AF.
      • Nieuwlaat R.
      • Capucci A.
      • Camm A.J.
      • et al.
      European Heart Survey Investigators
      Atrial fibrillation management: a prospective survey in ESC member countries: the Euro Heart Survey on atrial fibrillation.
      • Huxley R.R.
      • Lopez F.L.
      • Folsom A.R.
      • et al.
      Absolute and attributable risks of atrial fibrillation in relation to optimal and borderline risk factors: The Atherosclerosis Risk in Communities (ARIC) study.
      • Huxley R.R.
      • Filion K.B.
      • Konety S.
      • Alonso A.
      Meta-analysis of cohort and case-control studies of type 2 diabetes mellitus and risk of atrial fibrillation.
      • Mozaffarian D.
      • Benjamin E.J.
      • Go A.S.
      • et al.
      American Heart Association Statistics Committee and Stroke Statistics Subcommittee
      Heart disease and stroke statistics—2015 update: a report from the American Heart Association [published corrections appear in Circulation. 2015 and Circulation. 2016].
      The Framingham Heart Study reported that male and female diabetic patients were, respectively, 4 and 6 times more likely to develop AF than did nondiabetic patients.
      • Benjamin E.J.
      • Levy D.
      • Vaziri S.M.
      • D’Agostino R.B.
      • Belanger A.J.
      • Wolf P.A.
      Independent risk factors for atrial fibrillation in a population-based cohort: the Framingham Heart Study.
      In addition, increasing levels of hemoglobin A1c and duration of diabetes have been reported to increase the risk of thromboembolism.
      • Overvad T.F.
      • Skjøth F.
      • Lip G.Y.
      • et al.
      Duration of diabetes mellitus and risk of thromboembolism and bleeding in atrial fibrillation: nationwide cohort study.
      • Fangel M.V.
      • Nielsen P.B.
      • Kristensen J.K.
      • et al.
      Glycemic status and thromboembolic risk in patients with atrial fibrillation and type 2 diabetes mellitus.
      Diabetes is also an independent risk factor for stroke, and diabetic patients have more disabilities and an increased risk of mortality compared with nondiabetic patients.
      • Tuomilehto J.
      • Rastenyte D.
      • Jousilahti J.
      • Sarti C.
      • Vartiainen E.
      Diabetes mellitus as a risk factor for death from stroke: prospective study of the middle-aged Finnish population.
      • Klem I.
      • Wehinger C.
      • Schneider B.
      • Hartl E.
      • Finsterer J.
      • Stöllberger C.
      Diabetic atrial fibrillation patients: mortality and risk for stroke or embolism during a 10-year follow-up.
      Furthermore, other risk factors for stroke in patients with AF, such as renal failure and peripheral vascular disease, are more prevalent in patients with diabetes.
      • Lip G.Y.
      • Nieuwlaat R.
      • Pisters R.
      • Lane D.A.
      • Crijns H.J.
      Refining clinical risk stratification for predicting stroke and thromboembolism in atrial fibrillation using a novel risk factor-based approach: the Euro Heart Survey on atrial fibrillation.
      • Singer D.E.
      • Chang Y.
      • Borowsky L.H.
      • et al.
      A new risk scheme to predict ischemic stroke and other thromboembolism in atrial fibrillation: the ATRIA study stroke risk score.
      The CHA2DS2-VASc (congestive heart failure, hypertension, age ≥75 years, diabetes, stroke [previous], vascular disease, age 65-74 years, sex) score takes diabetes history into consideration (contributing 1 point to the final calculation), which emphasizes the importance of diabetes in AF management.
      • Lip G.Y.
      • Nieuwlaat R.
      • Pisters R.
      • Lane D.A.
      • Crijns H.J.
      Refining clinical risk stratification for predicting stroke and thromboembolism in atrial fibrillation using a novel risk factor-based approach: the Euro Heart Survey on atrial fibrillation.
      Diabetes is, therefore, an important risk factor for disease progression and adverse outcomes in patients with AF, making patients with diabetes a high-risk subgroup.
      • Huxley R.R.
      • Alonso A.
      • Lopez F.L.
      • et al.
      Type 2 diabetes, glucose homeostasis and incident atrial fibrillation: the Atherosclerosis Risk in Communities study.
      • Benjamin E.J.
      • Levy D.
      • Vaziri S.M.
      • D’Agostino R.B.
      • Belanger A.J.
      • Wolf P.A.
      Independent risk factors for atrial fibrillation in a population-based cohort: the Framingham Heart Study.
      • Aksnes T.A.
      • Schmieder R.E.
      • Kjeldsen S.E.
      • Ghani S.
      • Hua T.A.
      • Julius S.
      Impact of new-onset diabetes mellitus on development of atrial fibrillation and heart failure in high-risk hypertension (from the VALUE trial).
      Because of the increased risk of stroke/systemic embolism (SE) in patients with diabetes, OACs are recommended for patients with AF and concomitant diabetes.
      • January C.T.
      • Wann L.S.
      • Alpert J.S.
      • et al.
      ACC/AHA Task Force Members. 2014 AHA/ACC/HRS guideline for the management of patients with atrial fibrillation: a report of the American College of Cardiology/American Heart Association Task Force on practice guidelines and the Heart Rhythm Society [published correction appears in Circulation. 2014;130(23):e272-e274].
      A meta-analysis of the 4 NOAC trials found no significant interaction between treatment (NOACs vs warfarin) and diabetes status for stroke/SE or MB.
      • Patti G.
      • Di Gioia G.
      • Cavallari I.
      • Nenna A.
      Safety and efficacy of nonvitamin K antagonist oral anticoagulants versus warfarin in diabetic patients with atrial fibrillation: a study-level meta-analysis of phase III randomized trials.
      • Plitt A.
      • McGuire D.K.
      • Giugliano R.P.
      Atrial fibrillation, type 2 diabetes, and non-vitamin K antagonist oral anticoagulants: a review.
      However, in a subgroup analysis of the ARISTOTLE (Apixaban for Reduction In Stroke and Other Thromboembolic Events in Atrial Fibrillation) trial, diabetes and treatment had a significant interaction for the risk of MB, although there is no good mechanistic hypotheses to explain the interaction.
      • Ezekowitz J.A.
      • Lewis B.S.
      • Lopes R.D.
      • et al.
      Clinical outcomes of patients with diabetes and atrial fibrillation treated with apixaban: results from the ARISTOTLE trial.
      In the controlled trials, 23% to 40% of patients had diabetes, so it is an important high-risk subgroup study to evaluate.
      • Ezekowitz J.A.
      • Lewis B.S.
      • Lopes R.D.
      • et al.
      Clinical outcomes of patients with diabetes and atrial fibrillation treated with apixaban: results from the ARISTOTLE trial.
      • Brambatti M.
      • Darius H.
      • Oldgren J.
      • et al.
      Comparison of dabigatran versus warfarin in diabetic patients with atrial fibrillation: results from the RE-LY trial.
      • Bansilal S.
      • Bloomgarden Z.
      • Halperin J.L.
      • et al.
      ROCKET AF Steering Committee and Investigators
      Efficacy and safety of rivaroxaban in patients with diabetes and nonvalvular atrial fibrillation: the Rivaroxaban Once-daily, Oral, Direct Factor Xa Inhibition Compared with Vitamin K Antagonism for Prevention of Stroke and Embolism Trial in Atrial Fibrillation (ROCKET AF Trial).
      There have been few observational studies comparing NOACs and warfarin in patients with NVAF and diabetes. To contribute real-world evidence from several data sources that may facilitate future research regarding this underrepresented population, this study analyzed the diabetes subgroup of the ARISTOPHANES (Anticoagulants for Reduction In Stroke: Observational Pooled Analysis on Health Outcomes and Experience of Patients; NCT03087487) study. The present study pooled NVAF patients with diabetes who were newly prescribed OACs and compared the risks of stroke/SE and MB associated with apixaban, dabigatran, rivaroxaban, and warfarin use.

      Patients and Methods

      Data Sources

      This study was a retrospective observational database analysis of a patient population of more than 180 million beneficiaries per year (∼56% of the US population) using fee-for-service Medicare data from the US Centers for Medicare & Medicaid Services and 4 other US commercial claims databases: the Truven MarketScan Commercial Claims and Encounter and Medicare Supplemental and Coordination of Benefits Database, the IMS PharMetrics Plus database, the Optum Clinformatics Data Mart, and the Humana Research database.
      The databases include patients with Medicare Fee-For-Service, Medicare Advantage, and commercial insurance. Database records include comprehensive demographic and clinical information and International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes, Healthcare Common Procedure Coding System codes, and National Drug Code.
      Previously published articles include detailed descriptions of the data sets, the rationale for the pooling process, and the approaches to minimizing potential patient record duplicates across data sources.
      • Li X.S.
      • Deitelzweig S.
      • Keshishian A.
      • et al.
      Effectiveness and safety of apixaban versus warfarin in non-valvular atrial fibrillation patients in “real-world” clinical practice.
      • Lip G.Y.H.
      • Keshishian A.
      • Li X.
      • et al.
      Effectiveness and safety of oral anticoagulants among nonvalvular atrial fibrillation patients: the ARISTOPHANES study.

      Patient Selection

      Patients with NVAF were selected if they had 1 or more pharmacy claim for apixaban, dabigatran, rivaroxaban, or warfarin from January 1, 2013, through September 30, 2015 (identification period). Edoxaban was evaluated but excluded because of a small sample size due to its recent Food and Drug Administration approval. The first NOAC prescription date was designated as the index date if patients had a NOAC claim. The first warfarin prescription date was designated as the index date for patients without any NOAC claim. Patients were required to have an AF diagnosis before the index date and have continuous medical and pharmacy health plan enrollment for 12 months or more before the index date (baseline period) (Supplemental Table 1, available online at http://www.mayoclinicproceedings.org).
      To evaluate new initiators, patients treated with an OAC within 12 months before the index date were excluded. Patients were also excluded if they had claims indicating any of the following: valvular heart disease (defined by the presence of International Classification of Diseases, Ninth Revision codes 394.xx, 396.xx, 424.0, and 745.xx), venous thromboembolism, transient AF (pericarditis, hyperthyroidism, and thyrotoxicosis), or heart valve replacement/transplant during the baseline period; pregnancy during the study period; or hip or knee replacement surgery within 6 weeks before the index date. Detailed selection criteria are presented in Figure 1. Among the resulting patients with NVAF prescribed OACs, patients with type 1 and 2 diabetes (ICD-9-CM code 250.xx) during the baseline period were selected.
      Figure thumbnail gr1
      Figure 1Selection criteria. Application of the selection criteria yielded more than 100,000 patients with nonvalvular atrial fibrillation and diabetes mellitus. These patients were matched in cohorts by propensity scores. AF = atrial fibrillation; ICD-10 = International Classification of Diseases, Tenth Revision; OAC = oral anticoagulant; VTE = venous thromboembolism.

      Outcome Measures

      Outcome measures were time to first stroke/SE, including ischemic stroke, hemorrhagic stroke, and SE, and time to first MB, including gastrointestinal (GI) bleeding, intracranial hemorrhage, and bleeding at other key sites (Supplemental Table 1).
      • Thigpen J.L.
      • Dillon C.
      • Forster K.B.
      • et al.
      Validity of International Classification of Disease codes to identify ischemic stroke and intracranial hemorrhage among individuals with associated diagnosis of atrial fibrillation.
      • Cunningham A.
      • Stein C.M.
      • Chung C.P.
      • Daugherty J.R.
      • Smalley W.E.
      • Ray W.A.
      An automated database case definition for serious bleeding related to oral anticoagulant use.
      Outcomes were based on hospitalizations with stroke/SE or MB as the principal or first-listed diagnosis. The follow-up period ranged from 1-day postindex date to 30 days after the discontinuation date, medication switch date, death (only inpatient death for the commercial databases and all-cause death for the Medicare database), end of continuous medical or pharmacy plan enrollment, or end of study (September 30, 2015), whichever occurred earliest.

      Statistical Methods

      Propensity score matching (PSM) was conducted between NOAC and warfarin cohorts (apixaban vs warfarin, dabigatran vs warfarin, and rivaroxaban vs warfarin) and between NOAC cohorts (apixaban vs dabigatran, apixaban vs rivaroxaban, and dabigatran vs rivaroxaban) within each data set. The variables used for PSM are related to key patient characteristics, including demographic characteristics, Charlson Comorbidity Index scores,
      • Charlson M.E.
      • Pompei P.
      • Ales K.L.
      • MacKenzie C.R.
      A new method of classifying prognostic comorbidity in longitudinal studies: development and validation.
      common comorbidities, diabetes complications, and comedications. (A complete list of PSM model covariates is given in Tables 1 and 2.) In each database, patients were matched using 1:1 nearest neighbor matching without replacement (with a caliper of 0.01). The covariate balance was checked using standardized differences, with a threshold of 10%.
      • Austin P.C.
      Balance diagnostics for comparing the distribution of baseline covariates between treatment groups in propensity-score matched samples.
      After ensuring that the cohorts were balanced in each database, study patients from the 5 data sets were pooled for the analysis.
      Table 1Baseline Characteristics of NOACs vs Warfarin After PSM
      ACEi/ARB = angiotensin converting enzyme inhibitors/angiotensin-receptor blocker; CHA2DS2-VASc = congestive heart failure, hypertension, aged ≥75 years, diabetes mellitus, prior stroke or transient ischemic attack or thromboembolism, vascular disease, aged 65-74 years, sex category; DPP-4 = dipeptidyl peptidase 4; GLP-1 = glucagon-like peptide; HAS-BLED = hypertension, abnormal (renal/liver function), stroke, bleeding, labile (international normalized ratio), elderly, drug/alcohol/medication (usage history); NOAC = non-vitamin K oral anticoagulant; NSAIDs = non-steroidal anti-inflammatory drugs; PSM = propensity score matching; SE = systemic embolism; SGLT-2 = sodium-glucose co-transporter 2.
      ,
      Data are presented as mean ± SD or as No. (percentage).
      CharacteristicApixaban cohortWarfarin cohortDabigatran cohortWarfarin cohortRivaroxaban cohortWarfarin cohort
      Sample size35,26935,26912,95412,95444,41244,412
      Age (y)75.8±9.075.8±8.973.7±9.173.9±9.375.2±8.975.3±8.9
       18-54635 (1.8)641 (1.8)388 (3.0)393 (3.0)827 (1.9)831 (1.9)
       55-642468 (7.0)2447 (6.9)1362 (10.5)1341 (10.4)3371 (7.6)3362 (7.6)
       65-7412,388 (35.1)12,306 (34.9)5129 (39.6)5100 (39.4)16,542 (37.2)16,467 (37.1)
       ≥7519,778 (56.1)19,875 (56.4)6075 (46.9)6120 (47.2)23,672 (53.3)23,752 (53.5)
      Sex
       Male18,963 (53.8)18,936 (53.7)7460 (57.6)7484 (57.8)24,511 (55.2)24,491 (55.1)
       Female16,306 (46.2)16,333 (46.3)5494 (42.4)5470 (42.2)19,901 (44.8)19,921 (44.9)
      US geographic region
       Northeast6402 (18.2)6362 (18.0)2613 (20.2)2676 (20.7)8488 (19.1)8373 (18.9)
       Midwest7756 (22.0)7869 (22.3)2772 (21.4)2761 (21.3)10,653 (24.0)10,686 (24.1)
       South15,884 (45.0)15,863 (45.0)5339 (41.2)5258 (40.6)17,883 (40.3)17,892 (40.3)
       West5139 (14.6)5087 (14.4)2174 (16.8)2209 (17.1)7225 (16.3)7310 (16.5)
       Other88 (0.2)88 (0.2)56 (0.4)50 (0.4)163 (0.4)151 (0.3)
      Baseline comorbidity
       Deyo-Charlson Comorbidity Index score4.7±2.74.7±2.84.1±2.64.1±2.64.5±2.74.5±2.7
       CHA2DS2-VASc score4.8±1.54.8±1.54.5±1.54.5±1.54.7±1.54.7±1.5
      1157 (0.5)142 (0.4)124 (1.0)105 (0.8)250 (0.6)235 (0.5)
      21515 (4.3)1486 (4.2)938 (7.2)921 (7.1)2080 (4.7)2005 (4.5)
      35092 (14.4)4893 (13.9)2430 (18.8)2316 (17.9)7066 (15.9)6820 (15.4)
      ≥428,505 (80.8)28,748 (81.5)9462 (73.0)9612 (74.2)35,016 (78.8)35,352 (79.6)
       HAS-BLED score
      As the international normalized ratio value is not available in the databases, a modified HAS-BLED score was calculated with a range of 0 to 8.
      3.5±1.33.5±1.33.2±1.33.2±1.33.4±1.33.4±1.3
      0134 (0.4)159 (0.5)101 (0.8)120 (0.9)229 (0.5)253 (0.6)
      11424 (4.0)1490 (4.2)855 (6.6)837 (6.5)2095 (4.7)2148 (4.8)
      26879 (19.5)6940 (19.7)3171 (24.5)3132 (24.2)9339 (21.0)9637 (21.7)
      ≥326,832 (76.1)26,680 (75.6)8827 (68.1)8865 (68.4)32,749 (73.7)32,374 (72.9)
       Bleeding history7836 (22.2)7811 (22.1)2517 (19.4)2529 (19.5)9819 (22.1)9802 (22.1)
       Congestive heart failure13,516 (38.3)13,603 (38.6)4197 (32.4)4275 (33.0)16,320 (36.7)16,352 (36.8)
       Type 1 diabetes
      Diabetes type was defined by the presence of International Classification of Disease, Ninth Revision, Clinical Modification diagnosis codes only and was not further validated.
      5146 (14.6)5322 (15.1)1920 (14.8)1786 (13.8)6474 (14.6)6477 (14.6)
       Type 2 diabetes
      Diabetes type was defined by the presence of International Classification of Disease, Ninth Revision, Clinical Modification diagnosis codes only and was not further validated.
      35,019 (99.3)35,048 (99.4)12,873 (99.4)12,871 (99.4)44,064 (99.2)44,132 (99.4)
       Hypertension33,642 (95.4)33,652 (95.4)12,220 (94.3)12,237 (94.5)41,984 (94.5)41,986 (94.5)
       Renal disease12,557 (35.6)12,639 (35.8)3337 (25.8)3365 (26.0)13,869 (31.2)13,726 (30.9)
       Liver disease2234 (6.3)2205 (6.3)753 (5.8)788 (6.1)2888 (6.5)2814 (6.3)
       Myocrdial infarction4372 (12.4)4385 (12.4)1276 (9.9)1294 (10.0)5365 (12.1)5395 (12.1)
       Dyspepsia or stomach discomfort8238 (23.4)8287 (23.5)2654 (20.5)2593 (20.0)10,187 (22.9)10,042 (22.6)
       Nonstroke/ SE peripheral vascular disease22,511 (63.8)22,479 (63.7)7574 (58.5)7553 (58.3)27,328 (61.5)27,382 (61.7)
       Stroke/SE5110 (14.5)5036 (14.3)1608 (12.4)1657 (12.8)6124 (13.8)6190 (13.9)
       Transient ischemic attack2816 (8.0)2861 (8.1)891 (6.9)897 (6.9)3350 (7.5)3365 (7.6)
       Anemia and coagulation defects12,969 (36.8)12,928 (36.7)3893 (30.1)3960 (30.6)15,663 (35.3)15,534 (35.0)
       Alcoholism612 (1.7)605 (1.7)265 (2.0)250 (1.9)918 (2.1)922 (2.1)
       Peripheral artery disease9366 (26.6)9687 (27.5)2943 (22.7)3089 (23.8)11,557 (26.0)11,599 (26.1)
       Coronary artery disease20,088 (57.0)19,903 (56.4)6709 (51.8)6630 (51.2)24,174 (54.4)24,144 (54.4)
       Obesity10,725 (30.4)10,692 (30.3)3779 (29.2)3750 (28.9)13,041 (29.4)13,002 (29.3)
       Hypoglycemia909 (2.6)990 (2.8)312 (2.4)283 (2.2)1178 (2.7)1156 (2.6)
       Dyslipidemia30,820 (87.4)30,793 (87.3)11,132 (85.9)11,132 (85.9)38,222 (86.1)38,099 (85.8)
       Diabetic nephropathy4004 (11.4)4032 (11.4)1073 (8.3)1079 (8.3)4221 (9.5)4140 (9.3)
       Diabetic neuropathy7375 (20.9)7348 (20.8)2438 (18.8)2439 (18.8)8759 (19.7)8892 (20.0)
       Diabetic retinopathy4262 (12.1)4278 (12.1)1458 (11.3)1510 (11.7)5022 (11.3)5072 (11.4)
      Baseline medication use
       ACEi/ARB25,712 (72.9)25,790 (73.1)9519 (73.5)9533 (73.6)32,172 (72.4)32,362 (72.9)
       Amiodarone4548 (12.9)4542 (12.9)1523 (11.8)1535 (11.8)5350 (12.0)5424 (12.2)
       β-blockers21,646 (61.4)21,569 (61.2)7751 (59.8)7744 (59.8)27,087 (61.0)27,059 (60.9)
       H2-receptor antagonists2889 (8.2)2881 (8.2)956 (7.4)925 (7.1)3598 (8.1)3585 (8.1)
       Proton pump inhibitors12,554 (35.6)12,621 (35.8)4195 (32.4)4203 (32.4)15,252 (34.3)15,229 (34.3)
       Statins25,459 (72.2)25,409 (72.0)8966 (69.2)8978 (69.3)31,263 (70.4)31,310 (70.5)
       Anti-platelets9206 (26.1)9132 (25.9)2860 (22.1)2852 (22.0)10,741 (24.2)10,692 (24.1)
       NSAIDS8940 (25.3)8888 (25.2)3445 (26.6)3456 (26.7)11,210 (25.2)11,158 (25.1)
       Diuretics22,484 (63.8)22,506 (63.8)7972 (61.5)8002 (61.8)27,610 (62.2)27,674 (62.3)
       Calcium channel blockers16,540 (46.9)16,584 (47.0)5921 (45.7)5932 (45.8)20,465 (46.1)20,501 (46.2)
      Baseline diabetes medications
       Biguanides14,764 (41.9)14,818 (42.0)6000 (46.3)6027 (46.5)18,927 (42.6)19,071 (42.9)
       Sulphonylureas9199 (26.1)9269 (26.3)3513 (27.1)3602 (27.8)11,702 (26.3)11,789 (26.5)
       Meglitinide405 (1.1)396 (1.1)147 (1.1)141 (1.1)529 (1.2)520 (1.2)
       Thiazolidinediones1540 (4.4)1583 (4.5)641 (4.9)608 (4.7)1934 (4.4)1930 (4.3)
       DPP-4 inhibitors3875 (11.0)3892 (11.0)1461 (11.3)1474 (11.4)4664 (10.5)4600 (10.4)
       Insulin8199 (23.2)8180 (23.2)2886 (22.3)2855 (22.0)10,109 (22.8)10,211 (23.0)
       α-glucosidase inhibitors166 (0.5)156 (0.4)58 (0.4)64 (0.5)195 (0.4)178 (0.4)
       SGLT-2 inhibitors170 (0.5)165 (0.5)67 (0.5)62 (0.5)180 (0.4)182 (0.4)
       GLP-1 agonists788 (2.2)797 (2.3)320 (2.5)299 (2.3)896 (2.0)899 (2.0)
      Dose of the index prescription
       Standard dose
      Standard dose: 5mg apixaban, 150 mg dabigatran, 20 mg rivaroxaban.
      26,383 (74.8)10,493 (81.0)30,215 (68.0)
       Lower dose
      Lower dose: 2.5 mg apixaban, 75 mg dabigatran, 10 or 15 mg rivaroxaban; 2,460 patients treated with rivaroxaban were prescribed 10 mg rivaroxaban.
      8886 (25.2)2461 (19.0)14,197 (32.0)
      Follow-up time (d)
       Mean183.3±166.7240.5±218.0224.5±223.5244.6±220.8225.4±213.7244.0±220.0
       Median121158122161142162
      a ACEi/ARB = angiotensin converting enzyme inhibitors/angiotensin-receptor blocker; CHA2DS2-VASc = congestive heart failure, hypertension, aged ≥75 years, diabetes mellitus, prior stroke or transient ischemic attack or thromboembolism, vascular disease, aged 65-74 years, sex category; DPP-4 = dipeptidyl peptidase 4; GLP-1 = glucagon-like peptide; HAS-BLED = hypertension, abnormal (renal/liver function), stroke, bleeding, labile (international normalized ratio), elderly, drug/alcohol/medication (usage history); NOAC = non-vitamin K oral anticoagulant; NSAIDs = non-steroidal anti-inflammatory drugs; PSM = propensity score matching; SE = systemic embolism; SGLT-2 = sodium-glucose co-transporter 2.
      b Data are presented as mean ± SD or as No. (percentage).
      c As the international normalized ratio value is not available in the databases, a modified HAS-BLED score was calculated with a range of 0 to 8.
      d Diabetes type was defined by the presence of International Classification of Disease, Ninth Revision, Clinical Modification diagnosis codes only and was not further validated.
      e Standard dose: 5mg apixaban, 150 mg dabigatran, 20 mg rivaroxaban.
      f Lower dose: 2.5 mg apixaban, 75 mg dabigatran, 10 or 15 mg rivaroxaban; 2,460 patients treated with rivaroxaban were prescribed 10 mg rivaroxaban.
      Table 2Baseline Characteristics of NOACs vs NOACs After PSM
      ACEi/ARB = angiotensin converting enzyme inhibitors/angiotensin-receptor blocker; CHA2DS2-VASc = congestive heart failure, hypertension, aged ≥75 years, diabetes mellitus, prior stroke or transient ischemic attack or thromboembolism, vascular disease, aged 65-74 years, sex category; DPP-4 = dipeptidyl peptidase 4; GLP-1 = glucagon-like peptide; HAS-BLED = hypertension, abnormal (renal/liver function), stroke, bleeding, labile (international normalized ratio), elderly, drug/alcohol/medication (usage history); NOAC = non-vitamin K oral anticoagulant; NSAIDs = non-steroidal anti-inflammatory drugs; PSM = propensity score matching; SE = systemic embolism; SGLT-2 = sodium-glucose co-transporter 2.
      ,
      Data are presented as mean ± SD or as No. (percentage).
      CharacteristicApixaban cohortDabigatran cohortApixaban cohortRivaroxaban cohortDabigatran cohortRivaroxaban cohort
      Sample size13,00613,00636,54936,54913,11513,115
      Age (y)73.8±9.473.6±9.275.3±9.375.2±9.273.5±9.273.5±9.4
       18-54426 (3.3)430 (3.3)881 (2.4)869 (2.4)444 (3.4)441 (3.4)
       55-641373 (10.6)1402 (10.8)3014 (8.2)2995 (8.2)1432 (10.9)1421 (10.8)
       65-745100 (39.2)5107 (39.3)12,866 (35.2)12,860 (35.2)5160 (39.3)5175 (39.5)
       ≥756107 (47.0)6067 (46.6)19,788 (54.1)19,825 (54.2)6079 (46.4)6078 (46.3)
      Sex
       Male7526 (57.9)7499 (57.7)19,761 (54.1)19,688 (53.9)7596 (57.9)7656 (58.4)
       Female5480 (42.1)5507 (42.3)16,788 (45.9)16,861 (46.1)5519 (42.1)5459 (41.6)
      US geographic region
       Northeast2560 (19.7)2604 (20.0)6423 (17.6)6427 (17.6)2645 (20.2)2674 (20.4)
       Midwest2758 (21.2)2766 (21.3)7737 (21.2)7682 (21.0)2779 (21.2)2750 (21.0)
       South5496 (42.3)5441 (41.8)17,130 (46.9)17,200 (47.1)5450 (41.6)5503 (42.0)
       West2142 (16.5)2142 (16.5)5174 (14.2)5160 (14.1)2181 (16.6)2133 (16.3)
       Other50 (0.4)53 (0.4)85 (0.2)80 (0.2)60 (0.5)55 (0.4)
      Baseline comorbidity
       Deyo-Charlson Comorbidity Index score4.1±2.64.1±2.64.5±2.74.5±2.74.1±2.64.1±2.6
       CHA2DS2-VASc score4.5±1.54.4±1.54.7±1.54.7±1.54.4±1.54.4±1.5
      1104 (0.8)121 (0.9)192 (0.5)193 (0.5)130 (1.0)127 (1.0)
      2967 (7.4)1007 (7.7)1941 (5.3)1950 (5.3)1040 (7.9)1072 (8.2)
      32445 (18.8)2432 (18.7)5588 (15.3)5515 (15.1)2470 (18.8)2383 (18.2)
      ≥49490 (73.0)9446 (72.6)28,828 (78.9)28,891 (79.0)9475 (72.2)9533 (72.7)
       HAS-BLED score
      As the international normalized ratio value is not available in the databases, a modified HAS-BLED score was calculated with a range of 0 to 8.
      3.2±1.33.2±1.33.4±1.33.4±1.33.2±1.33.2±1.3
      089 (0.7)102 (0.8)167 (0.5)157 (0.4)106 (0.8)110 (0.8)
      1849 (6.5)890 (6.8)1722 (4.7)1786 (4.9)916 (7.0)892 (6.8)
      23131 (24.1)3191 (24.5)7428 (20.3)7301 (20.0)3229 (24.6)3176 (24.2)
      ≥38937 (68.7)8823 (67.8)27,232 (74.5)27,305 (74.7)8864 (67.6)8937 (68.1)
       Bleeding history2529 (19.4)2508 (19.3)7798 (21.3)7869 (21.5)2521 (19.2)2544 (19.4)
       Congestive heart failure4229 (32.5)4182 (32.2)13,362 (36.6)13,373 (36.6)4202 (32.0)4308 (32.8)
       Type 1 diabetes
      Diabetes type was defined by the presence of International Classification of Disease, Ninth Revision, Clinical Modification diagnosis codes only and was not further validated.
      1802 (13.9)1926 (14.8)5151 (14.1)5397 (14.8)1938 (14.8)1874 (14.3)
       Type 2 diabetes
      Diabetes type was defined by the presence of International Classification of Disease, Ninth Revision, Clinical Modification diagnosis codes only and was not further validated.
      12,903 (99.2)12,924 (99.4)36,295 (99.3)36,290 (99.3)13,030 (99.4)13,006 (99.2)
       Hypertension12,288 (94.5)12,282 (94.4)34,855 (95.4)34,837 (95.3)12,374 (94.3)12,385 (94.4)
       Renal disease3336 (25.6)3331 (25.6)12,030 (32.9)12,108 (33.1)3338 (25.5)3314 (25.3)
       Liver disease766 (5.9)754 (5.8)2337 (6.4)2358 (6.5)758 (5.8)732 (5.6)
       Myocardial infarction1297 (10.0)1274 (9.8)4332 (11.9)4330 (11.8)1277 (9.7)1288 (9.8)
       Dyspepsia or stomach discomfort2665 (20.5)2650 (20.4)8479 (23.2)8570 (23.4)2666 (20.3)2693 (20.5)
       Nonstroke/SE peripheral vascular disease7548 (58.0)7572 (58.2)22,962 (62.8)23,016 (63.0)7619 (58.1)7543 (57.5)
       Stroke/SE1642 (12.6)1607 (12.4)5046 (13.8)5069 (13.9)1606 (12.2)1628 (12.4)
       Transient ischemic attack875 (6.7)895 (6.9)2866 (7.8)2903 (7.9)894 (6.8)893 (6.8)
       Anemia and coagulation defects3864 (29.7)3876 (29.8)12,768 (34.9)12,830 (35.1)3894 (29.7)3890 (29.7)
       Alcoholism287 (2.2)261 (2.0)638 (1.7)652 (1.8)269 (2.1)278 (2.1)
       Peripheral artery disease2979 (22.9)2938 (22.6)9342 (25.6)9780 (26.8)2952 (22.5)3002 (22.9)
       Coronary artery disease6672 (51.3)6711 (51.6)20,503 (56.1)20,347 (55.7)6754 (51.5)6632 (50.6)
       Obesity3849 (29.6)3827 (29.4)11,272 (30.8)11,355 (31.1)3835 (29.2)3821 (29.1)
       Hypoglycemia295 (2.3)310 (2.4)912 (2.5)976 (2.7)313 (2.4)304 (2.3)
       Dyslipidemia11,157 (85.8)11,183 (86.0)32,023 (87.6)32,092 (87.8)11,268 (85.9)11,286 (86.1)
       Diabetic nephropathy1089 (8.4)1073 (8.3)3685 (10.1)3681 (10.1)1076 (8.2)1023 (7.8)
       Diabetic neuropathy2479 (19.1)2440 (18.8)7357 (20.1)7340 (20.1)2455 (18.7)2466 (18.8)
       Diabetic retinopathy1486 (11.4)1459 (11.2)4248 (11.6)4297 (11.8)1474 (11.2)1472 (11.2)
      Baseline medication use
       ACEi/ARB9571 (73.6)9576 (73.6)26,897 (73.6)26,855 (73.5)9656 (73.6)9669 (73.7)
       Amiodarone1498 (11.5)1528 (11.7)4679 (12.8)4710 (12.9)1536 (11.7)1580 (12.0)
       β-blockers7743 (59.5)7792 (59.9)22,538 (61.7)22,578 (61.8)7842 (59.8)7772 (59.3)
       H2-receptor antagonists954 (7.3)949 (7.3)2935 (8.0)2974 (8.1)956 (7.3)1008 (7.7)
       Proton pump inhibitors4332 (33.3)4219 (32.4)13,034 (35.7)13,084 (35.8)4231 (32.3)4309 (32.9)
       Statins9035 (69.5)8995 (69.2)26,354 (72.1)26,379 (72.2)9057 (69.1)9023 (68.8)
       Anti-platelets2868 (22.1)2864 (22.0)9514 (26.0)9587 (26.2)2878 (21.9)2921 (22.3)
       NSAIDs3557 (27.3)3499 (26.9)9731 (26.6)9714 (26.6)3533 (26.9)3527 (26.9)
       Diuretics8041 (61.8)7998 (61.5)23,044 (63.0)23,067 (63.1)8057 (61.4)8095 (61.7)
       Calcium channel blockers5988 (46.0)5966 (45.9)17,188 (47.0)17,183 (47.0)6006 (45.8)5962 (45.5)
      Baseline diabetes medications
       Biguanides6091 (46.8)6066 (46.6)15,989 (43.7)15,921 (43.6)6130 (46.7)6089 (46.4)
       Sulphonylureas3565 (27.4)3505 (26.9)9416 (25.8)9378 (25.7)3550 (27.1)3574 (27.3)
       Meglitinide157 (1.2)151 (1.2)427 (1.2)438 (1.2)151 (1.2)150 (1.1)
       Thiazolidinediones674 (5.2)672 (5.2)1702 (4.7)1689 (4.6)681 (5.2)708 (5.4)
       DPP-4 inhibitors1537 (11.8)1503 (11.6)4321 (11.8)4282 (11.7)1519 (11.6)1559 (11.9)
       Insulin2875 (22.1)2890 (22.2)8252 (22.6)8244 (22.6)2919 (22.3)2949 (22.5)
       α-glucosidase inhibitors56 (0.4)57 (0.4)168 (0.5)168 (0.5)56 (0.4)60 (0.5)
       SGLT-2 inhibitors72 (0.6)78 (0.6)364 (1.0)357 (1.0)81 (0.6)83 (0.6)
       GLP-1 agonists402 (3.1)355 (2.7)1016 (2.8)1001 (2.7)356 (2.7)357 (2.7)
      Dose of the index prescription
       Standard dose
      Standard dose: 5 mg apixaban, 150 mg dabigatran, 20 mg rivaroxaban.
      10,545 (81.1)10,543 (81.1)27,852 (76.2)24,605 (67.3)10,648 (81.2)9509 (72.5)
       Lower dose
      Lower dose: 2.5 mg apixaban, 75 mg dabigatran, 10 or 15 mg rivaroxaban; 2,005 and 672 patients were prescribed 10 mg of rivaroxaban in the apixaban-rivaroxaban and dabigatran-rivaroxaban cohorts, respectively.
      2461 (18.9)2463 (18.9)8697 (23.8)11,944 (32.7)2467 (18.8)3606 (27.5)
      Follow-up time (d)
       Mean187.6±171.0225.0±223.5183.7±167.0225.5±213.7224.6±223.3230.0±216.9
       Median124123122142123147
      a ACEi/ARB = angiotensin converting enzyme inhibitors/angiotensin-receptor blocker; CHA2DS2-VASc = congestive heart failure, hypertension, aged ≥75 years, diabetes mellitus, prior stroke or transient ischemic attack or thromboembolism, vascular disease, aged 65-74 years, sex category; DPP-4 = dipeptidyl peptidase 4; GLP-1 = glucagon-like peptide; HAS-BLED = hypertension, abnormal (renal/liver function), stroke, bleeding, labile (international normalized ratio), elderly, drug/alcohol/medication (usage history); NOAC = non-vitamin K oral anticoagulant; NSAIDs = non-steroidal anti-inflammatory drugs; PSM = propensity score matching; SE = systemic embolism; SGLT-2 = sodium-glucose co-transporter 2.
      b Data are presented as mean ± SD or as No. (percentage).
      c As the international normalized ratio value is not available in the databases, a modified HAS-BLED score was calculated with a range of 0 to 8.
      d Diabetes type was defined by the presence of International Classification of Disease, Ninth Revision, Clinical Modification diagnosis codes only and was not further validated.
      e Standard dose: 5 mg apixaban, 150 mg dabigatran, 20 mg rivaroxaban.
      f Lower dose: 2.5 mg apixaban, 75 mg dabigatran, 10 or 15 mg rivaroxaban; 2,005 and 672 patients were prescribed 10 mg of rivaroxaban in the apixaban-rivaroxaban and dabigatran-rivaroxaban cohorts, respectively.
      The risk of stroke/SE and MB was evaluated using Cox proportional hazards models, with robust sandwich estimates.
      • Austin P.C.
      The use of propensity score methods with survival or time-to-event outcomes: reporting measures of effect similar to those used in randomized experiments.
      Oral anticoagulant treatment was included as the independent variable; as the cohorts were balanced, no other covariates were included in the model. P<.05 was considered statistically significant. No adjustments for multiple comparisons were made.

      Subgroup Analyses

      Propensity score matching was conducted again in subgroup patients on the basis of the index dose of the NOAC. Patients prescribed standard-dose (apixaban 5 mg, dabigatran 150 mg, and rivaroxaban 20 mg) and lower-dose (apixaban 2.5 mg, dabigatran 75 mg, and rivaroxaban 15 mg/10 mg) NOACs were matched on the basis of their index dose. Furthermore, Cox proportional hazards models were completed for the standard-dose and lower-dose subgroups separately.
      Institutional review board approval was not required because the study did not involve the collection, use, or transmittal of individual identifiable data. Both the data sets and the security of the offices in which the analysis was completed (and in which the data sets are kept) met the requirements of the Health Insurance Portability and Accountability Act of 1996.

      Results

      After applying the selection criteria, a total of 167,815 patients with NVAF and concomitant diabetes mellitus (35.9% of patients with NVAF [466,991]) were identified, including 37,558 patients prescribed apixaban, 13,128 dabigatran, 51,200 rivaroxaban, and 65,929 warfarin (Figure 1). Before PSM, patients prescribed warfarin were the oldest and had the highest CHA2DS2-VASc and hypertension, abnormal (renal/liver function), stroke, bleeding, labile (international normalized ratio), elderly, drug/alcohol/medication (usage history) (HAS-BLED) scores, followed by those prescribed apixaban, rivaroxaban, and dabigatran. The number of patients who were prescribed the lower dose in each cohort was 9180 (24%) of those prescribed apixaban (2.5 mg), 2467 (19%) of those prescribed dabigatran (75 mg), and 12,477 (24%) of those prescribed rivaroxaban (15 mg). In addition, 5% of patients treated with rivaroxaban were prescribed 10 mg of rivaroxaban (Supplemental Table 2, available online at http://www.mayoclinicproceedings.org).
      The unadjusted incidence rates of stroke/SE in the warfarin, apixaban, dabigatran, and rivaroxaban cohorts were 2.5, 1.7, 1.8, and 1.7 events per 100 person-years, respectively. The unadjusted incidence rates of MB in the warfarin, apixaban, dabigatran, and rivaroxaban cohorts were 8.2, 4.8, 4.8, and 6.9 events per 100 person-years, respectively (Supplemental Table 3, available online at http://www.mayoclinicproceedings.org).
      After PSM, a total of 154,324 unique patients were included. PSM produced 35,269, 12,954, and 44,412 patient pairs for the apixaban-warfarin, dabigatran-warfarin, and rivaroxaban-warfarin cohorts, respectively. PSM for NOAC comparisons included 13,006, 36,549, and 13,115 patient pairs for the apixaban-dabigatran, apixaban-rivaroxaban, and dabigatran-rivaroxaban cohorts, respectively (Figure 1). The baseline characteristics of the matched populations are listed in Tables 1 and 2. After matching, all demographic and clinical characteristics were well balanced. Across the matched cohorts during the baseline period, 22% to 23%, 42% to 47%, and 26% to 28% were prescribed insulin, biguanides, and sulfonylureas, respectively. The mean follow-up ranged between 6 and 8 months in all matched cohorts.
      The baseline characteristics of patients with NVAF prescribed standard- and lower-dosed NOACs are summarized in Supplemental Tables 4, 5, 6, and 7 (available online at http://www.mayoclinicproceedings.org).

      Non–Vitamin K Oral Anticoagulant and Warfarin Comparisons

      Compared with warfarin, apixaban (hazard ratio [HR], 0.67; 95% CI, 0.57-0.77) and rivaroxaban (HR, 0.79; 95% CI, 0.71-0.89) were associated with a lower risk of stroke/SE. There was no significant difference in the risk of stroke/SE (HR, 0.84; 95% CI, 0.67-1.07) between dabigatran and warfarin. Compared with those prescribed warfarin, patients prescribed apixaban had a 26% lower risk of ischemic stroke (HR, 0.74; 95% CI, 0.65-0.85) whereas those prescribed rivaroxaban had a 14% lower risk of ischemic stroke (HR, 0.86; 95% CI, 0.77-0.97) (Figure 2A). In addition, patients prescribed apixaban (HR, 0.48; 95% CI, 0.30-0.77), dabigatran (HR, 0.36; 95% CI, 0.21-0.60) and rivaroxaban (HR, 0.56; 95% CI, 0.45-0.69) had a lower risk of hemorrhagic stroke than did patients prescribed warfarin.
      Figure thumbnail gr2a
      Figure 2Incidence and hazard ratios of (A) NOACs vs warfarin and (B) NOACs vs NOACs. Incidence rates were measured per 100 person-years for matched NOAC cohorts. Hazard ratios were measured along with 95% CIs. aUpper limit of 95% was rounded from 0.997 to 1.00. GI = gastrointestinal; ICH = intracranial hemorrhage; NOAC = non–vitamin K oral anticoagulant; ref. = reference; SE = systemic embolism.
      Figure thumbnail gr2b
      Figure 2Incidence and hazard ratios of (A) NOACs vs warfarin and (B) NOACs vs NOACs. Incidence rates were measured per 100 person-years for matched NOAC cohorts. Hazard ratios were measured along with 95% CIs. aUpper limit of 95% was rounded from 0.997 to 1.00. GI = gastrointestinal; ICH = intracranial hemorrhage; NOAC = non–vitamin K oral anticoagulant; ref. = reference; SE = systemic embolism.
      Compared with warfarin, apixaban (HR, 0.60; 95% CI, 0.56-0.65) and dabigatran (HR, 0.78; 95% CI, 0.69-0.88) were associated with a lower risk of MB. Compared with warfarin, rivaroxaban was associated with a similar risk of MB (HR, 1.02; 95% CI, 0.94-1.10).
      Compared with those prescribed warfarin, patients prescribed apixaban had a lower risk of GI bleeding (HR, 0.58; 95% CI, 0.53-0.65), patients prescribed rivaroxaban had a higher risk of GI bleeding (HR, 1.19; 95% CI, 1.09-1.30), and patients prescribed dabigatran had a similar risk of GI bleeding (HR, 0.99; 95% CI, 0.84-1.17). All NOACs were associated with a lower risk of intracranial hemorrhage compared with warfarin (Figure 2A).

      Non–Vitamin K Oral Anticoagulant and NOAC Comparisons

      Apixaban was associated with a lower risk of stroke/SE compared with dabigatran (HR, 0.78; 95% CI, 0.64-0.94) and rivaroxaban (HR, 0.87; 95% CI, 0.75-1.00). Similarly, apixaban was associated with a lower risk of MB compared with dabigatran (HR, 0.73; 95% CI, 0.63-0.84) and rivaroxaban (HR, 0.59; 95% CI, 0.54-0.65), both driven by GI bleeding. Dabigatran was associated with a similar risk of stroke/SE (HR, 1.11; 95%, 0.85-1.46) and lower risk of MB (HR, 0.76; 95% CI, 0.66-0.86) compared with rivaroxaban (Figure 2B).
      The Kaplan-Meier curves for the cumulative incidence rates of stroke/SE and MB in the matched populations are shown in Supplemental Figure 1A and B (available online at http://www.mayoclinicproceedings.org).
      The results of the standard- and low-dose subgroup analysis were generally consistent with the main analysis (Figure 3) .
      Figure thumbnail gr3
      Figure 3Incidence and hazard ratios for dose subgroup analysis. Incidence rates were measured per 100 person-years for matched non–vitamin K oral anticoagulant cohorts. Hazard ratios were measured along with 95% CIs. comp. = comparator; ref. = reference; SE = systemic embolism. aUpper limit of 95% was rounded from 0.999 to 1.00.

      Discussion

      This ARISTOPHANES analysis of a high-risk subgroup of diabetic patients is the largest retrospective observational study to date that examines the risk of stroke/SE and MB patients with NVAF and diabetes who have initiated OAC treatment. The relevance of evaluating diabetic patients as a high-risk subgroup is paramount, as diabetes is one of the most common concomitant comorbid conditions in patients with AF.
      • Huxley R.R.
      • Lopez F.L.
      • Folsom A.R.
      • et al.
      Absolute and attributable risks of atrial fibrillation in relation to optimal and borderline risk factors: The Atherosclerosis Risk in Communities (ARIC) study.
      • Huxley R.R.
      • Filion K.B.
      • Konety S.
      • Alonso A.
      Meta-analysis of cohort and case-control studies of type 2 diabetes mellitus and risk of atrial fibrillation.
      With pooled data of US Centers for Medicare & Medicaid Services Medicare and 4 large US national claims databases, this study found that apixaban and rivaroxaban were associated with lower rates of stroke/SE compared with warfarin. In addition, apixaban and dabigatran were associated with lower rates of MB compared with warfarin. As a hypothesis-generating analysis, NOAC and NOAC comparisons suggested that there was a significantly lower risk of stroke/SE with apixaban compared to dabigatran and rivaroxaban. Apixaban was associated with a lower risk of MB compared with dabigatran and rivaroxaban and dabigatran was associated with a lower risk of MB compared with rivaroxaban.
      Subgroup analyses of the ARISTOTLE, Randomized Evaluation of Long-term Anticoagulation Therapy (RE-LY) and Rivaroxaban Once-daily Oral Direct Factor Xa Inhibition Compared with Vitamin K Antagonism for Prevention of Stroke and Embolism Trial in Atrial Fibrillation (ROCKET-AF) trials have revealed that apixaban, dabigatran, and rivaroxaban have no significant interaction with diabetes status for the reduction of stroke/SE.
      • Ezekowitz J.A.
      • Lewis B.S.
      • Lopes R.D.
      • et al.
      Clinical outcomes of patients with diabetes and atrial fibrillation treated with apixaban: results from the ARISTOTLE trial.
      • Brambatti M.
      • Darius H.
      • Oldgren J.
      • et al.
      Comparison of dabigatran versus warfarin in diabetic patients with atrial fibrillation: results from the RE-LY trial.
      • Bansilal S.
      • Bloomgarden Z.
      • Halperin J.L.
      • et al.
      ROCKET AF Steering Committee and Investigators
      Efficacy and safety of rivaroxaban in patients with diabetes and nonvalvular atrial fibrillation: the Rivaroxaban Once-daily, Oral, Direct Factor Xa Inhibition Compared with Vitamin K Antagonism for Prevention of Stroke and Embolism Trial in Atrial Fibrillation (ROCKET AF Trial).
      However, in the diabetes subgroup analysis of the ARISTOTLE trial, apixaban was associated with a reduction in MB in patients without diabetes (HR, 0.60; 95% CI, 0.51-0.72) but a similar risk of MB in patients with diabetes (HR, 0.96; 95% CI, 0.74-1.25; Pinteraction=.003) compared with warfarin.
      • Ezekowitz J.A.
      • Lewis B.S.
      • Lopes R.D.
      • et al.
      Clinical outcomes of patients with diabetes and atrial fibrillation treated with apixaban: results from the ARISTOTLE trial.
      The significant interaction may have been due to chance. There was no significant interaction for intracranial hemorrhage; apixaban was associated with reduced intracranial hemorrhage compared with warfarin in patients with and without diabetes. The ROCKET-AF and the RE-LY trials reported that the incidence of MB for dabigatran (150 mg) or rivaroxaban was similar to that for warfarin, regardless of diabetes status.
      • Brambatti M.
      • Darius H.
      • Oldgren J.
      • et al.
      Comparison of dabigatran versus warfarin in diabetic patients with atrial fibrillation: results from the RE-LY trial.
      • Bansilal S.
      • Bloomgarden Z.
      • Halperin J.L.
      • et al.
      ROCKET AF Steering Committee and Investigators
      Efficacy and safety of rivaroxaban in patients with diabetes and nonvalvular atrial fibrillation: the Rivaroxaban Once-daily, Oral, Direct Factor Xa Inhibition Compared with Vitamin K Antagonism for Prevention of Stroke and Embolism Trial in Atrial Fibrillation (ROCKET AF Trial).
      Meta-analyses of randomized controlled trials (including ARISTOTLE, RE-LY, ROCKET-AF, and Effective Anticoagulation with Factor Xa Next Generation in Atrial Fibrillation [ENGAGE-AF]) have revealed that diabetes status has no differential effect on the safety or effectiveness of all NOACs combined vs warfarin. Non–vitamin K oral anticoagulants were found to reduce the risk of stroke/SE and MB compared with warfarin in both diabetic and nondiabetic patients.
      • Patti G.
      • Di Gioia G.
      • Cavallari I.
      • Nenna A.
      Safety and efficacy of nonvitamin K antagonist oral anticoagulants versus warfarin in diabetic patients with atrial fibrillation: a study-level meta-analysis of phase III randomized trials.
      • Plitt A.
      • McGuire D.K.
      • Giugliano R.P.
      Atrial fibrillation, type 2 diabetes, and non-vitamin K antagonist oral anticoagulants: a review.
      Few retrospective observational studies have evaluated clinical outcomes in patients with NVAF and diabetes prescribed NOACs. One retrospective study using US commercial claims, leveraging data from the Truven MarketScan database, found no significant difference in the risk of stroke/SE and MB between rivaroxaban and warfarin therapy in diabetic patients with NVAF.
      • Coleman C.I.
      • Bunz T.J.
      • Eriksson D.
      • Meinecke A.K.
      • Sood N.A.
      Effectiveness and safety of rivaroxaban vs warfarin in people with non-valvular atrial fibrillation and diabetes: an administrative claims database analysis.
      A study of US Department of Defense records found that rivaroxaban was associated with a higher incidence of MB in diabetic patients than in nondiabetic patients (3.7 events per 100 person-years vs 2.5 events per 100 person-years).
      • Peacock W.F.
      • Tamayo S.
      • Sicignano N.
      • Hopf K.P.
      • Yuan Z.
      • Patel M.
      Comparison of the incidence of major bleeding with rivaroxaban use among nonvalvular atrial fibrillation patients with versus without diabetes mellitus.
      A study in Taiwan in diabetic patients with NVAF found that compared with those prescribed warfarin, patients prescribed dabigatran had a lower risk of all-cause mortality and GI bleeding whereas those prescribed rivaroxaban had a similar risk of mortality, stroke, and bleeding. Also, compared with those prescribed rivaroxaban, patients prescribed dabigatran had significantly lower rates of all-cause mortality.
      • Hsu C.C.
      • Hsu P.F.
      • Sung S.H.
      • et al.
      Is there a preferred stroke prevention strategy for diabetic patients with non-valvular atrial fibrillation? Comparing warfarin, dabigatran and rivaroxaban.
      An analysis of patients with and without diabetes from a retrospective observational study, leveraging data from US HealthCore claims, revealed that patients prescribed rivaroxaban had MB event rates similar to those of patients prescribed warfarin; however, compared with those prescribed warfarin, patients prescribed apixaban and dabigatran were associated with a reduction in MB events, regardless of diabetes status.
      • Adeboyeje G.
      • Sylwestrzak G.
      • Barron J.J.
      • et al.
      Major bleeding risk during anticoagulation with warfarin, dabigatran, apixaban, or rivaroxaban in patients with nonvalvular atrial fibrillation.
      Both patients prescribed dabigatran and those prescribed apixaban had a lower risk of MB than did those prescribed rivaroxaban among patients with and without diabetes. Compared with those prescribed dabigatran, patients prescribed apixaban had a similar risk of MB irrespective of diabetes status.
      • Adeboyeje G.
      • Sylwestrzak G.
      • Barron J.J.
      • et al.
      Major bleeding risk during anticoagulation with warfarin, dabigatran, apixaban, or rivaroxaban in patients with nonvalvular atrial fibrillation.
      Compared with previous studies, the ARISTOPHANES study included a larger sample of patients with NVAF and concomitant diabetes, providing this analysis with higher statistical power. Consistent with earlier publications, the present study reported that in this high-risk diabetes subgroup, apixaban and dabigatran were associated with a lower risk of MB compared with warfarin and rivaroxaban.
      • Lip G.Y.H.
      • Keshishian A.
      • Li X.
      • et al.
      Effectiveness and safety of oral anticoagulants among nonvalvular atrial fibrillation patients: the ARISTOPHANES study.
      • Adeboyeje G.
      • Sylwestrzak G.
      • Barron J.J.
      • et al.
      Major bleeding risk during anticoagulation with warfarin, dabigatran, apixaban, or rivaroxaban in patients with nonvalvular atrial fibrillation.
      • Amin A.
      • Keshishian A.
      • Trocio J.
      • et al.
      Risk of stroke/systemic embolism, major bleeding and associated costs in non-valvular atrial fibrillation patients who initiated apixaban, dabigatran or rivaroxaban compared with warfarin in the United States Medicare population.
      This retrospective observational study has several limitations. First, only statistical associations could be concluded, not causal relationships. Although cohorts were matched through PSM, there were potential residual confounders. This limitation is especially important for interpreting NOAC and NOAC comparison results, which are intended for hypothesis generation, given the lack of head-to-head trials. Second, because of the nature of claims studies, outcome measures could only be based on ICD-9-CM codes without further adjustment with precise clinical criteria. In addition, the dose of warfarin and laboratory values, such as INR measurements, are not available in the data set, so the time in therapeutic range for patients prescribed warfarin was indeterminable. Nonetheless, the inclusion of patients with potentially poorer quality control of warfarin therapy in everyday clinical practice may enable the study findings to better reflect real-world situations. In addition, a previous study found that PSM using claims-defined baseline characteristics was sufficient in balancing mean INR and INR categories across post-PSM warfarin cohorts matched to different NOACs (dabigatran, rivaroxaban, and apixaban).
      • Huybrechts K.F.
      • Gopalakrishnan C.
      • Franklin J.M.
      • et al.
      Claims data studies of direct oral anticoagulants can achieve balance in important clinical parameters only observable in electronic health records.
      Given our limited ability to clinically characterize diabetes type and severity (represented by diabetes medications and complications) because of our reliance on claims data, we could not further assess whether our findings would be different by type and severity of diabetes, such as duration of diabetes or hemoglobin A1c levels.
      • Overvad T.F.
      • Skjøth F.
      • Lip G.Y.
      • et al.
      Duration of diabetes mellitus and risk of thromboembolism and bleeding in atrial fibrillation: nationwide cohort study.
      • Fangel M.V.
      • Nielsen P.B.
      • Kristensen J.K.
      • et al.
      Glycemic status and thromboembolic risk in patients with atrial fibrillation and type 2 diabetes mellitus.
      Type 1 diabetes occurs at a younger age and type 2 diabetes occurs at an older age and has a higher prevalence of AF. Previous studies have found that stroke risk may be higher in type 1 diabetic patients, but a recent study of patients with NVAF did not find an association between type of diabetes and risk of thromboembolism.
      • Fuller J.H.
      • Stevens L.K.
      • Wang S.L.
      Risk factors for cardiovascular mortality and morbidity: the WHO Mutinational Study of Vascular Disease in Diabetes.
      • Janghorbani M.
      • Hu F.B.
      • Willett W.C.
      • et al.
      Prospective study of type 1 and type 2 diabetes and risk of stroke subtypes: the Nurses’ Health study.
      • Fangel M.V.
      • Nielsen P.B.
      • Larsen T.B.
      • et al.
      Type 1 versus type 2 diabetes and thromboembolic risk in patients with atrial fibrillation: a Danish nationwide cohort study.
      We are not able to assess the effect of the type of diabetes in our study. Moreover, unobserved heterogeneity may exist across the 5 data sources. Although some of the data sets contain information from different insurance plans that do not overlap at the plan level, others are employer-based claims data sets that may contain duplicate patient records when pooled together. However, the number of such duplicates is likely to be low—on the basis of a published estimate of 0.5%—and therefore unlikely to have any significant effect on the results.
      • Broder M.S.
      • Neary M.P.
      • Chang E.
      • Cherepanov D.
      • Katznelson L.
      Treatments, complications, and healthcare utilization associated with acromegaly: a study in two large United States databases.
      Finally, the results may not reflect the overall population with NVAF in the United States, because the study did not include uninsured patients and those solely covered by other public health insurance plans.

      Conclusion

      This study, the largest observational study of NVAF with concomitant diabetes, reports that NOACs were associated with variable rates of stroke/SE and MB compared with warfarin and compared with each other. These findings supplement the information from the NOAC randomized controlled trials and may support future studies on patients with NVAF and concomitant diabetes to provide clinicians with a better understanding of the real-world clinical outcomes of diabetic patients in routine clinical practice.

      Supplemental Online Material

      References

        • Lip G.
        • Freedman B.
        • De Caterina R.
        • Potpara T.S.
        Stroke prevention in atrial fibrillation: past, present and future.
        Thromb Haemost. 2017; 117: 1230-1239
        • Hart R.G.
        • Pearce L.A.
        • Aguilar M.I.
        Meta-analysis: antithrombotic therapy to prevent stroke in patients who have nonvalvular atrial fibrillation.
        Ann Intern Med. 2007; 146: 857-867
        • De Caterina R.
        • Husted S.
        • Wallentin L.
        • et al.
        Vitamin K antagonists in heart disease: current status and perspectives (Section III). Position paper of the ESC Working Group on Thrombosis—Task Force on Anticoagulants in Heart Disease.
        Thromb Haemost. 2013; 110: 1087-1107
        • Ruff C.T.
        • Giugliano R.P.
        • Braunwald E.
        • et al.
        Comparison of the efficacy and safety of new oral anticoagulants with warfarin in patients with atrial fibrillation: a meta-analysis of randomised trials.
        Lancet. 2014; 383: 955-962
        • Nieuwlaat R.
        • Capucci A.
        • Camm A.J.
        • et al.
        • European Heart Survey Investigators
        Atrial fibrillation management: a prospective survey in ESC member countries: the Euro Heart Survey on atrial fibrillation.
        Eur Heart J. 2005; 26: 2422-2434
        • Huxley R.R.
        • Lopez F.L.
        • Folsom A.R.
        • et al.
        Absolute and attributable risks of atrial fibrillation in relation to optimal and borderline risk factors: The Atherosclerosis Risk in Communities (ARIC) study.
        Circulation. 2011; 123: 1501-1508
        • Huxley R.R.
        • Filion K.B.
        • Konety S.
        • Alonso A.
        Meta-analysis of cohort and case-control studies of type 2 diabetes mellitus and risk of atrial fibrillation.
        Am J Cardiol. 2011; 108: 56-62
        • Mozaffarian D.
        • Benjamin E.J.
        • Go A.S.
        • et al.
        • American Heart Association Statistics Committee and Stroke Statistics Subcommittee
        Heart disease and stroke statistics—2015 update: a report from the American Heart Association [published corrections appear in Circulation. 2015 and Circulation. 2016].
        Circulation. 2015; 131: e29-e322
        • Benjamin E.J.
        • Levy D.
        • Vaziri S.M.
        • D’Agostino R.B.
        • Belanger A.J.
        • Wolf P.A.
        Independent risk factors for atrial fibrillation in a population-based cohort: the Framingham Heart Study.
        JAMA. 1994; 271: 840-844
        • Overvad T.F.
        • Skjøth F.
        • Lip G.Y.
        • et al.
        Duration of diabetes mellitus and risk of thromboembolism and bleeding in atrial fibrillation: nationwide cohort study.
        Stroke. 2015; 46: 2168-2174
        • Fangel M.V.
        • Nielsen P.B.
        • Kristensen J.K.
        • et al.
        Glycemic status and thromboembolic risk in patients with atrial fibrillation and type 2 diabetes mellitus.
        Circ Arrhythm Electrophysiol. 2019; 12: e007030
        • Tuomilehto J.
        • Rastenyte D.
        • Jousilahti J.
        • Sarti C.
        • Vartiainen E.
        Diabetes mellitus as a risk factor for death from stroke: prospective study of the middle-aged Finnish population.
        Stroke. 1996; 27: 210-215
        • Klem I.
        • Wehinger C.
        • Schneider B.
        • Hartl E.
        • Finsterer J.
        • Stöllberger C.
        Diabetic atrial fibrillation patients: mortality and risk for stroke or embolism during a 10-year follow-up.
        Diabetes Metab Res Rev. 2003; 19: 320-328
        • Lip G.Y.
        • Nieuwlaat R.
        • Pisters R.
        • Lane D.A.
        • Crijns H.J.
        Refining clinical risk stratification for predicting stroke and thromboembolism in atrial fibrillation using a novel risk factor-based approach: the Euro Heart Survey on atrial fibrillation.
        Chest. 2010; 137: 263-272
        • Singer D.E.
        • Chang Y.
        • Borowsky L.H.
        • et al.
        A new risk scheme to predict ischemic stroke and other thromboembolism in atrial fibrillation: the ATRIA study stroke risk score.
        J Am Heart Assoc. 2013; 2: e000250
        • Huxley R.R.
        • Alonso A.
        • Lopez F.L.
        • et al.
        Type 2 diabetes, glucose homeostasis and incident atrial fibrillation: the Atherosclerosis Risk in Communities study.
        Heart. 2012; 98: 133-138
        • Benjamin E.J.
        • Levy D.
        • Vaziri S.M.
        • D’Agostino R.B.
        • Belanger A.J.
        • Wolf P.A.
        Independent risk factors for atrial fibrillation in a population-based cohort: the Framingham Heart Study.
        JAMA. 1994; 271: 840-844
        • Aksnes T.A.
        • Schmieder R.E.
        • Kjeldsen S.E.
        • Ghani S.
        • Hua T.A.
        • Julius S.
        Impact of new-onset diabetes mellitus on development of atrial fibrillation and heart failure in high-risk hypertension (from the VALUE trial).
        Am J Cardiol. 2008; 101: 634-638
        • January C.T.
        • Wann L.S.
        • Alpert J.S.
        • et al.
        ACC/AHA Task Force Members. 2014 AHA/ACC/HRS guideline for the management of patients with atrial fibrillation: a report of the American College of Cardiology/American Heart Association Task Force on practice guidelines and the Heart Rhythm Society [published correction appears in Circulation. 2014;130(23):e272-e274].
        Circulation. 2014; 130: e199-e267
        • Patti G.
        • Di Gioia G.
        • Cavallari I.
        • Nenna A.
        Safety and efficacy of nonvitamin K antagonist oral anticoagulants versus warfarin in diabetic patients with atrial fibrillation: a study-level meta-analysis of phase III randomized trials.
        Diabetes Metab Res Rev. 2017; 33
        • Plitt A.
        • McGuire D.K.
        • Giugliano R.P.
        Atrial fibrillation, type 2 diabetes, and non-vitamin K antagonist oral anticoagulants: a review.
        JAMA Cardiol. 2017; 2: 442-448
        • Ezekowitz J.A.
        • Lewis B.S.
        • Lopes R.D.
        • et al.
        Clinical outcomes of patients with diabetes and atrial fibrillation treated with apixaban: results from the ARISTOTLE trial.
        Eur Heart J Cardiovasc Pharmacother. 2015; 1: 86-94
        • Brambatti M.
        • Darius H.
        • Oldgren J.
        • et al.
        Comparison of dabigatran versus warfarin in diabetic patients with atrial fibrillation: results from the RE-LY trial.
        Int J Cardiol. 2015; 196: 127-131
        • Bansilal S.
        • Bloomgarden Z.
        • Halperin J.L.
        • et al.
        • ROCKET AF Steering Committee and Investigators
        Efficacy and safety of rivaroxaban in patients with diabetes and nonvalvular atrial fibrillation: the Rivaroxaban Once-daily, Oral, Direct Factor Xa Inhibition Compared with Vitamin K Antagonism for Prevention of Stroke and Embolism Trial in Atrial Fibrillation (ROCKET AF Trial).
        Am Heart J. 2015; 170: 675-682
        • Li X.S.
        • Deitelzweig S.
        • Keshishian A.
        • et al.
        Effectiveness and safety of apixaban versus warfarin in non-valvular atrial fibrillation patients in “real-world” clinical practice.
        Thromb Haemost. 2017; 117: 1072-1082
        • Lip G.Y.H.
        • Keshishian A.
        • Li X.
        • et al.
        Effectiveness and safety of oral anticoagulants among nonvalvular atrial fibrillation patients: the ARISTOPHANES study.
        Stroke. 2018; 49: 2933-2944
        • Thigpen J.L.
        • Dillon C.
        • Forster K.B.
        • et al.
        Validity of International Classification of Disease codes to identify ischemic stroke and intracranial hemorrhage among individuals with associated diagnosis of atrial fibrillation.
        Circ Cardiovasc Qual Outcomes. 2015; 8: 8-14
        • Cunningham A.
        • Stein C.M.
        • Chung C.P.
        • Daugherty J.R.
        • Smalley W.E.
        • Ray W.A.
        An automated database case definition for serious bleeding related to oral anticoagulant use.
        Pharmacoepidemiol Drug Saf. 2011; 20: 560-566
        • Charlson M.E.
        • Pompei P.
        • Ales K.L.
        • MacKenzie C.R.
        A new method of classifying prognostic comorbidity in longitudinal studies: development and validation.
        J Chronic Dis. 1987; 40: 373-383
        • Austin P.C.
        Balance diagnostics for comparing the distribution of baseline covariates between treatment groups in propensity-score matched samples.
        Stat Med. 2009; 28: 3083-3107
        • Austin P.C.
        The use of propensity score methods with survival or time-to-event outcomes: reporting measures of effect similar to those used in randomized experiments.
        Stat Med. 2014; 33: 1242-1258
        • Coleman C.I.
        • Bunz T.J.
        • Eriksson D.
        • Meinecke A.K.
        • Sood N.A.
        Effectiveness and safety of rivaroxaban vs warfarin in people with non-valvular atrial fibrillation and diabetes: an administrative claims database analysis.
        Diabet Med. 2018; 35: 1105-1110
        • Peacock W.F.
        • Tamayo S.
        • Sicignano N.
        • Hopf K.P.
        • Yuan Z.
        • Patel M.
        Comparison of the incidence of major bleeding with rivaroxaban use among nonvalvular atrial fibrillation patients with versus without diabetes mellitus.
        Am J Cardiol. 2017; 119: 753-759
        • Hsu C.C.
        • Hsu P.F.
        • Sung S.H.
        • et al.
        Is there a preferred stroke prevention strategy for diabetic patients with non-valvular atrial fibrillation? Comparing warfarin, dabigatran and rivaroxaban.
        Thromb Haemost. 2018; 118: 72-81
        • Adeboyeje G.
        • Sylwestrzak G.
        • Barron J.J.
        • et al.
        Major bleeding risk during anticoagulation with warfarin, dabigatran, apixaban, or rivaroxaban in patients with nonvalvular atrial fibrillation.
        J Manag Care Spec Pharm. 2017; 23: 968-978
        • Amin A.
        • Keshishian A.
        • Trocio J.
        • et al.
        Risk of stroke/systemic embolism, major bleeding and associated costs in non-valvular atrial fibrillation patients who initiated apixaban, dabigatran or rivaroxaban compared with warfarin in the United States Medicare population.
        Curr Med Res Opin. 2017; 33: 1595-1604
        • Huybrechts K.F.
        • Gopalakrishnan C.
        • Franklin J.M.
        • et al.
        Claims data studies of direct oral anticoagulants can achieve balance in important clinical parameters only observable in electronic health records.
        Clin Pharmacol Ther. 2019; 105: 979-993
        • Fuller J.H.
        • Stevens L.K.
        • Wang S.L.
        Risk factors for cardiovascular mortality and morbidity: the WHO Mutinational Study of Vascular Disease in Diabetes.
        Diabetologia. 2001; 44: S54-S64
        • Janghorbani M.
        • Hu F.B.
        • Willett W.C.
        • et al.
        Prospective study of type 1 and type 2 diabetes and risk of stroke subtypes: the Nurses’ Health study.
        Diabetes Care. 2007; 30: 1730-1735
        • Fangel M.V.
        • Nielsen P.B.
        • Larsen T.B.
        • et al.
        Type 1 versus type 2 diabetes and thromboembolic risk in patients with atrial fibrillation: a Danish nationwide cohort study.
        Int J Cardiol. 2018; 268: 137-142
        • Broder M.S.
        • Neary M.P.
        • Chang E.
        • Cherepanov D.
        • Katznelson L.
        Treatments, complications, and healthcare utilization associated with acromegaly: a study in two large United States databases.
        Pituitary. 2014; 17: 333-341