Association of Anticoagulant Therapy With Risk of Fracture Among Patients With Atrial Fibrillation | Atrial Fibrillation | JAMA Internal Medicine | JAMA Network
[Skip to Navigation]
Figure 1.  Adjusted Hazard Ratios (HRs) for Incident Hip Fracture Among Matched Direct Oral Anticoagulant (DOAC) Users
Adjusted Hazard Ratios (HRs) for Incident Hip Fracture Among Matched Direct Oral Anticoagulant (DOAC) Users

Patients were identified by initial oral anticoagulant therapy for the treatment of atrial fibrillation and stratified by subgroups of interest from the MarketScan Commercial Claims and Encounters and Medicare Supplemental and Coordination of Benefits databases from January 1, 2010, through September 30, 2015. Models were adjusted for age, sex, CHA2DS2-VASc (congestive heart failure, hypertension, age [>65 years = 1 point; >75 years = 2 points], diabetes, and previous stroke/transient ischemic attack [2 points], vascular disease) score, high-dimensional propensity score, and frailty.

Figure 2.  Adjusted Hazard Ratios (HRs) for Incident Hospitalized Fracture Among Matched Direct Oral Anticoagulant (DOAC) Users
Adjusted Hazard Ratios (HRs) for Incident Hospitalized Fracture Among Matched Direct Oral Anticoagulant (DOAC) Users

Patients were identified by initial oral anticoagulant therapy for the treatment of atrial fibrillation and stratified by subgroups of interest from the MarketScan Commercial Claims and Encounters and Medicare Supplemental and Coordination of Benefits databases from January 1, 2010, through September 30, 2015. Models were adjusted for age, sex, CHA2DS2-VASc (congestive heart failure, hypertension, age [>65 years = 1 point; >75 years = 2 points], diabetes, and previous stroke/transient ischemic attack [2 points], vascular disease) score, high-dimensional propensity score, and frailty.

Table 1.  Characteristics of Patients With Atrial Fibrillation by Anticoagulant Prescribeda
Characteristics of Patients With Atrial Fibrillation by Anticoagulant Prescribeda
Table 2.  Adjusted HRs for Incident Fracture Comparing New Users of DOACs vs Warfarin Among Patients With Nonvalvular Atrial Fibrillationa
Adjusted HRs for Incident Fracture Comparing New Users of DOACs vs Warfarin Among Patients With Nonvalvular Atrial Fibrillationa
Table 3.  Adjusted HRs for Incident Fracture Comparing New Users of DOACs Among Patients With Nonvalvular Atrial Fibrillationa
Adjusted HRs for Incident Fracture Comparing New Users of DOACs Among Patients With Nonvalvular Atrial Fibrillationa
1.
Weng  L-C, Preis  SR, Hulme  OL,  et al.  Genetic predisposition, clinical risk factor burden, and lifetime risk of atrial fibrillation.  Circulation. 2018;137(10):1027-1038. doi:10.1161/CIRCULATIONAHA.117.031431PubMedGoogle ScholarCrossref
2.
Mou  L, Norby  FL, Chen  LY,  et al.  Lifetime risk of atrial fibrillation by race and socioeconomic status: ARIC study (Atherosclerosis Risk in Communities).  Circ Arrhythm Electrophysiol. 2018;11(7):e006350. doi:10.1161/CIRCEP.118.006350PubMedGoogle Scholar
3.
Colilla  S, Crow  A, Petkun  W, Singer  DE, Simon  T, Liu  X.  Estimates of current and future incidence and prevalence of atrial fibrillation in the US adult population.  Am J Cardiol. 2013;112(8):1142-1147. doi:10.1016/j.amjcard.2013.05.063PubMedGoogle ScholarCrossref
4.
January  CT, Wann  LS, Alpert  JS,  et al; American College of Cardiology/American Heart Association Task Force on Practice Guidelines.  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.  J Am Coll Cardiol. 2014;64(21):e1-e76. doi:10.1016/j.jacc.2014.03.022PubMedGoogle ScholarCrossref
5.
Connolly  SJ, Ezekowitz  MD, Yusuf  S,  et al; RE-LY Steering Committee and Investigators.  Dabigatran versus warfarin in patients with atrial fibrillation.  N Engl J Med. 2009;361(12):1139-1151. doi:10.1056/NEJMoa0905561PubMedGoogle ScholarCrossref
6.
Patel  MR, Mahaffey  KW, Garg  J,  et al; ROCKET AF Investigators.  Rivaroxaban versus warfarin in nonvalvular atrial fibrillation.  N Engl J Med. 2011;365(10):883-891. doi:10.1056/NEJMoa1009638PubMedGoogle ScholarCrossref
7.
Granger  CB, Alexander  JH, McMurray  JJV,  et al; ARISTOTLE Committees and Investigators.  Apixaban versus warfarin in patients with atrial fibrillation.  N Engl J Med. 2011;365(11):981-992. doi:10.1056/NEJMoa1107039PubMedGoogle ScholarCrossref
8.
Giugliano  RP, Ruff  CT, Braunwald  E,  et al; ENGAGE AF-TIMI 48 Investigators.  Edoxaban versus warfarin in patients with atrial fibrillation.  N Engl J Med. 2013;369(22):2093-2104. doi:10.1056/NEJMoa1310907PubMedGoogle ScholarCrossref
9.
Bengtson  LGS, Lutsey  PL, Chen  LY, MacLehose  RF, Alonso  A.  Comparative effectiveness of dabigatran and rivaroxaban versus warfarin for the treatment of non-valvular atrial fibrillation.  J Cardiol. 2017;69(6):868-876. doi:10.1016/j.jjcc.2016.08.010PubMedGoogle ScholarCrossref
10.
Norby  FL, Bengtson  LGS, Lutsey  PL,  et al.  Comparative effectiveness of rivaroxaban versus warfarin or dabigatran for the treatment of patients with non-valvular atrial fibrillation.  BMC Cardiovasc Disord. 2017;17(1):238. doi:10.1186/s12872-017-0672-5PubMedGoogle ScholarCrossref
11.
Namba  S, Yamaoka-Tojo  M, Kakizaki  R,  et al.  Effects on bone metabolism markers and arterial stiffness by switching to rivaroxaban from warfarin in patients with atrial fibrillation.  Heart Vessels. 2017;32(8):977-982. doi:10.1007/s00380-017-0950-2PubMedGoogle ScholarCrossref
12.
Lau  WC, Chan  EW, Cheung  CL,  et al.  Association between dabigatran vs warfarin and risk of osteoporotic fractures among patients with nonvalvular atrial fibrillation.  JAMA. 2017;317(11):1151-1158. doi:10.1001/jama.2017.1363PubMedGoogle ScholarCrossref
13.
Fiordellisi  W, White  K, Schweizer  M.  A systematic review and meta-analysis of the association between vitamin K antagonist use and fracture.  J Gen Intern Med. 2019;34(2):304-311. doi:10.1007/s11606-018-4758-2PubMedGoogle ScholarCrossref
14.
Veronese  N, Bano  G, Bertozzo  G,  et al.  Vitamin K antagonists’ use and fracture risk: results from a systematic review and meta-analysis.  J Thromb Haemost. 2015;13(9):1665-1675. doi:10.1111/jth.13052PubMedGoogle ScholarCrossref
15.
Sugiyama  T.  Osteoporotic fractures associated with dabigatran vs warfarin.  JAMA. 2017;318(1):90-91. doi:10.1001/jama.2017.6908PubMedGoogle ScholarCrossref
16.
IBM Watson Health. IBM MarketScan research databases for health services researchers (white paper). https://www.ibm.com/downloads/cas/6KNYVVQ2. Published April 2019. Accessed May 2, 2019.
17.
Jensen  PN, Johnson  K, Floyd  J, Heckbert  SR, Carnahan  R, Dublin  S.  A systematic review of validated methods for identifying atrial fibrillation using administrative data.  Pharmacoepidemiol Drug Saf. 2012;21(suppl 1):141-147. doi:10.1002/pds.2317PubMedGoogle ScholarCrossref
18.
Hernán  MA, Alonso  A, Logan  R,  et al.  Observational studies analyzed like randomized experiments: an application to postmenopausal hormone therapy and coronary heart disease.  Epidemiology. 2008;19(6):766-779. doi:10.1097/EDE.0b013e3181875e61PubMedGoogle ScholarCrossref
19.
Garg  RK, Glazer  NL, Wiggins  KL,  et al.  Ascertainment of warfarin and aspirin use by medical record review compared with automated pharmacy data.  Pharmacoepidemiol Drug Saf. 2011;20(3):313-316. doi:10.1002/pds.2041PubMedGoogle ScholarCrossref
20.
Division of Biomedical Statistics and Informatics, Mayo Clinic Research. Biomedical Statistics and Informatics [computer program]. http://bioinformaticstools.mayo.edu/research/gmatch/. Updated July 23, 2018. Accessed August 1, 2019.
21.
Cunningham  A, Stein  CM, Chung  CP, Daugherty  JR, Smalley  WE, Ray  WA.  An automated database case definition for serious bleeding related to oral anticoagulant use.  Pharmacoepidemiol Drug Saf. 2011;20(6):560-566. doi:10.1002/pds.2109PubMedGoogle ScholarCrossref
22.
Quan  H, Sundararajan  V, Halfon  P,  et al.  Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data.  Med Care. 2005;43(11):1130-1139. doi:10.1097/01.mlr.0000182534.19832.83PubMedGoogle ScholarCrossref
23.
Lip  GY, Nieuwlaat  R, Pisters  R, Lane  DA, Crijns  HJ.  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(2):263-272. doi:10.1378/chest.09-1584PubMedGoogle ScholarCrossref
24.
Kim  DH, Schneeweiss  S.  Measuring frailty using claims data for pharmacoepidemiologic studies of mortality in older adults: evidence and recommendations.  Pharmacoepidemiol Drug Saf. 2014;23(9):891-901. doi:10.1002/pds.3674PubMedGoogle ScholarCrossref
25.
Schneeweiss  S, Rassen  JA, Glynn  RJ, Avorn  J, Mogun  H, Brookhart  MA.  High-dimensional propensity score adjustment in studies of treatment effects using health care claims data.  Epidemiology. 2009;20(4):512-522. doi:10.1097/EDE.0b013e3181a663ccPubMedGoogle ScholarCrossref
26.
Bross  ID.  Spurious effects from an extraneous variable.  J Chronic Dis. 1966;19(6):637-647. doi:10.1016/0021-9681(66)90062-2PubMedGoogle ScholarCrossref
27.
VanderWeele  TJ, Ding  P.  Sensitivity analysis in observational research: introducing the E-value.  Ann Intern Med. 2017;167(4):268-274. doi:10.7326/M16-2607PubMedGoogle ScholarCrossref
28.
Mathur  MB, Ding  P, Riddell  CA, VanderWeele  TJ.  Web Site and R package for computing E-values.  Epidemiology. 2018;29(5):e45-e47. doi:10.1097/EDE.0000000000000864PubMedGoogle ScholarCrossref
29.
Watts  NB.  Adverse bone effects of medications used to treat non-skeletal disorders.  Osteoporos Int. 2017;28(10):2741-2746. doi:10.1007/s00198-017-4171-4PubMedGoogle ScholarCrossref
30.
Sugiyama  T, Kugimiya  F, Kono  S, Kim  YT, Oda  H.  Warfarin use and fracture risk: an evidence-based mechanistic insight.  Osteoporos Int. 2015;26(3):1231-1232. doi:10.1007/s00198-014-2912-1PubMedGoogle ScholarCrossref
31.
Sugiyama  T, Takaki  T, Sakanaka  K,  et al.  Warfarin-induced impairment of cortical bone material quality and compensatory adaptation of cortical bone structure to mechanical stimuli.  J Endocrinol. 2007;194(1):213-222. doi:10.1677/JOE-07-0119PubMedGoogle ScholarCrossref
32.
Rey-Sanchez  P, Lavado-Garcia  JM, Canal-Macias  ML, Rodriguez-Dominguez  MT, Bote-Mohedano  JL, Pedrera-Zamorano  JD.  Ultrasound bone mass in patients undergoing chronic therapy with oral anticoagulants.  J Bone Miner Metab. 2011;29(5):546-551. doi:10.1007/s00774-010-0250-8PubMedGoogle ScholarCrossref
33.
Namba  S, Yamaoka-Tojo  M, Hashikata  T,  et al.  Long-term warfarin therapy and biomarkers for osteoporosis and atherosclerosis.  BBA Clin. 2015;4:76-80. doi:10.1016/j.bbacli.2015.08.002PubMedGoogle ScholarCrossref
34.
Fusaro  M, Dalle Carbonare  L, Dusso  A,  et al.  Differential effects of dabigatran and warfarin on bone volume and structure in rats with normal renal function.  PLoS One. 2015;10(8):e0133847. doi:10.1371/journal.pone.0133847PubMedGoogle Scholar
35.
Lucenteforte  E, Bettiol  A, Lombardi  N, Mugelli  A, Vannacci  A.  Risk of bone fractures among users of oral anticoagulants: an administrative database cohort study.  Eur J Intern Med. 2017;44:e30-e31. doi:10.1016/j.ejim.2017.07.022PubMedGoogle ScholarCrossref
36.
Steffel  J, Giugliano  RP, Braunwald  E,  et al.  Edoxaban versus warfarin in atrial fibrillation patients at risk of falling: ENGAGE AF-TIMI 48 analysis.  J Am Coll Cardiol. 2016;68(11):1169-1178. doi:10.1016/j.jacc.2016.06.034PubMedGoogle ScholarCrossref
37.
Gage  BF, Birman-Deych  E, Radford  MJ, Nilasena  DS, Binder  EF.  Risk of osteoporotic fracture in elderly patients taking warfarin: results from the National Registry of Atrial Fibrillation 2.  Arch Intern Med. 2006;166(2):241-246. doi:10.1001/archinte.166.2.241PubMedGoogle ScholarCrossref
38.
Rejnmark  L, Vestergaard  P, Mosekilde  L.  Fracture risk in users of oral anticoagulants: a nationwide case-control study.  Int J Cardiol. 2007;118(3):338-344. doi:10.1016/j.ijcard.2006.07.022PubMedGoogle ScholarCrossref
39.
Mamdani  M, Upshur  REG, Anderson  G, Bartle  BR, Laupacis  A.  Warfarin therapy and risk of hip fracture among elderly patients.  Pharmacotherapy. 2003;23(1):1-4. doi:10.1592/phco.23.1.1.31922PubMedGoogle ScholarCrossref
40.
Misra  D, Zhang  Y, Peloquin  C, Choi  HK, Kiel  DP, Neogi  T.  Incident long-term warfarin use and risk of osteoporotic fractures: propensity-score matched cohort of elders with new onset atrial fibrillation.  Osteoporos Int. 2014;25(6):1677-1684. doi:10.1007/s00198-014-2662-0PubMedGoogle ScholarCrossref
41.
Zalesak  M, Siu  K, Francis  K,  et al.  Higher persistence in newly diagnosed nonvalvular atrial fibrillation patients treated with dabigatran versus warfarin.  Circ Cardiovasc Qual Outcomes. 2013;6(5):567-574. doi:10.1161/CIRCOUTCOMES.113.000192PubMedGoogle ScholarCrossref
42.
Cummings  SR, Melton  LJ.  Epidemiology and outcomes of osteoporotic fractures.  Lancet. 2002;359(9319):1761-1767. doi:10.1016/S0140-6736(02)08657-9PubMedGoogle ScholarCrossref
43.
Rachner  TD, Khosla  S, Hofbauer  LC.  Osteoporosis: now and the future.  Lancet. 2011;377(9773):1276-1287. doi:10.1016/S0140-6736(10)62349-5PubMedGoogle ScholarCrossref
44.
Virnig  B, Durham  SB, Folsom  AR, Cerhan  J.  Linking the Iowa Women’s Health Study cohort to Medicare data: linkage results and application to hip fracture.  Am J Epidemiol. 2010;172(3):327-333. doi:10.1093/aje/kwq111PubMedGoogle ScholarCrossref
Limit 200 characters
Limit 25 characters
Conflicts of Interest Disclosure

Identify all potential conflicts of interest that might be relevant to your comment.

Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.

Err on the side of full disclosure.

If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.

Not all submitted comments are published. Please see our commenting policy for details.

Limit 140 characters
Limit 3600 characters or approximately 600 words
    1 Comment for this article
    Not DOAC has less fractures but warfarin has more fractures
    Rajeev Gupta, MBBS, MD, DM (Cardiology) | Mediclinic Al-Jowhara Hospital, Al Ain, UAE
    Dear Editor,
    The message of the study, "Apixaban compared with warfarin use was associated with lower fracture risk," may easily be misunderstood. As a matter of fact, DOACs don't protect against fractures in any way. Warfarin increases the risk of fractures by inhibiting Vitamin K- dependant bone proteins. (1) Therefore, on the comparison, it must be stated clearly and mentioned: warfarin compared with apixaban was associated with a higher risk of fracture.
    Reference:
    1. Gage BF et al. Risk of osteoporotic fracture in elderly patients taking warfarin. Arch Intern Med. 2006; 166 (2): 241-246.
    CONFLICT OF INTEREST: None Reported
    Original Investigation
    November 25, 2019

    Association of Anticoagulant Therapy With Risk of Fracture Among Patients With Atrial Fibrillation

    Author Affiliations
    • 1Division of Epidemiology and Community Health, University of Minnesota School of Public Health, Minneapolis
    • 2Center for Care Delivery and Outcomes Research, Minneapolis Veterans Affairs Health Care System, Minneapolis, Minnesota
    • 3Department of Medicine, University of Minnesota School of Medicine, Minneapolis
    • 4Rollins School of Public Health, Department of Epidemiology, Emory University, Atlanta, Georgia
    JAMA Intern Med. 2020;180(2):245-253. doi:10.1001/jamainternmed.2019.5679
    Key Points

    Question  Are oral anticoagulants differentially associated with risk of fracture among patients with atrial fibrillation?

    Findings  In this comparative effectiveness cohort study of 167 275 patients with atrial fibrillation, direct oral anticoagulants (ie, dabigatran, rivaroxaban, and apixaban) were associated with modestly lower fracture risk compared with warfarin. This protective association was more pronounced among patients with atrial fibrillation who also had a diagnosis of osteoporosis; among the direct oral anticoagulants, fracture risk was lowest among apixaban users.

    Meaning  These findings add to speculation that warfarin may be harmful to bone health and suggest that direct oral anticoagulants may be preferred among patients with atrial fibrillation and high fracture risk.

    Abstract

    Importance  Warfarin is prescribed to patients with atrial fibrillation (AF) for the prevention of cardioembolic complications. Whether warfarin adversely affects bone health is controversial. The availability of alternate direct oral anticoagulant (DOAC) options now make it possible to evaluate the comparative safety of warfarin in association with fracture risk.

    Objective  To test the hypothesis that, among patients with nonvalvular AF, use of DOACs vs warfarin is associated with lower risk of incident fracture.

    Design, Setting, and Participants  This comparative effectiveness cohort study used the MarketScan administrative claims databases to identify patients with nonvalvular AF and who were prescribed oral anticoagulants from January 1, 2010, through September 30, 2015. To reduce confounding, patients were matched on age, sex, CHA2DS2-VASc (congestive heart failure, hypertension, age [>65 years = 1 point; >75 years = 2 points], diabetes, and previous stroke/transient ischemic attack [2 points], vascular disease) score, and high-dimensional propensity scores. The final analysis included 167 275 patients with AF. Data were analyzed from February 27, 2019 to September 18, 2019.

    Exposures  Warfarin and DOACs (dabigatran etexilate, rivaroxaban, and apixaban).

    Main Outcomes and Measures  Incident hip fracture, fracture requiring hospitalization, and all clinical fractures (identified using inpatient or outpatient claims) defined by International Classification of Diseases, Ninth Revision, Clinical Modification codes.

    Results  In the study population of 167 275 patients with AF (38.0% women and 62.0% men; mean [SD] age, 68.9 [12.5] years), a total of 817 hip fractures, 2013 hospitalized fractures, and 7294 total fractures occurred during a mean (SD) follow-up of 16.9 (13.7) months. In multivariable-adjusted, propensity score–matched Cox proportional hazards regression models, relative to new users of warfarin, new users of DOACs tended to be at lower risk of fractures requiring hospitalization (hazard ratio [HR], 0.87; 95% CI, 0.79-0.96) and all clinical fractures (HR, 0.93; 95% CI, 0.88-0.98), whereas the association with hip fractures (HR, 0.91; 95% CI, 0.78-1.07) was not statistically significant. When comparing individual DOACs with warfarin, the strongest findings were for apixaban (HR for hip fracture, 0.67 [95% CI, 0.45-0.98]; HR for fractures requiring hospitalization, 0.60 [95% CI, 0.47-0.78]; and HR for all clinical fractures, 0.86 [95% CI, 0.75-0.98]). In subgroup analyses, DOACs appeared more beneficial among patients with AF who also had a diagnosis of osteoporosis than among those without a diagnosis of osteoporosis.

    Conclusions and Relevance  In this real-world population of 167 275 patients with AF, use of DOACs—particularly apixaban—compared with warfarin use was associated with lower fracture risk. These associations were more pronounced among patients with a diagnosis of osteoporosis. Given the potential adverse effects of warfarin on bone health, these findings suggest that caution should be used when prescribing warfarin to patients with AF at high risk of fracture.

    Introduction

    An estimated 1 in 3 to 1 in 5 individuals will develop atrial fibrillation (AF) in their lifetime,1,2 and the prevalence of AF in the United States is projected to exceed 12 million by 2030.3 In addition to rhythm and/or rate control, anticoagulant therapy for the prevention of stroke and cardioembolic complications is a mainstay of nonvalvular AF management.4 Oral anticoagulant (OAC) options now include vitamin K antagonists (typically warfarin in the United States) and several direct OACs (DOACs), namely dabigatran etexilate, rivaroxaban, apixaban, and edoxaban tosylate. Direct OACs have been shown to be as effective as warfarin for the prevention of stroke and cardioembolic complications in patients with nonvalvular AF and have been associated with a lower risk of bleeding events in randomized clinical trials5-8 and large, observational comparative effectiveness studies.9,10

    Concerns have been raised about whether warfarin use may be associated with impaired bone quality11 and increased fracture risk, although the existing evidence is inconsistent. This hypothesis was bolstered by a 2017 publication12 that, using a population of patients with AF from Hong Kong, reported that dabigatran was associated with a significantly lower risk of fracture compared with warfarin. However, only 32 events were reported among dabigatran users. Conversely, a 2019 meta-analysis of 4 studies13 found no difference in fracture risk between use of a DOAC vs warfarin. Likewise, findings from comparisons of warfarin users and individuals not prescribed an OAC have been inconsistent.13,14 To our knowledge, no studies have yet conducted comparisons among different DOAC users, although doing so is important15 because prescribers and patients must decide among the various OAC options.

    Given present uncertainty regarding whether fracture risk differs according to type of OAC therapy and the need to optimize AF management to avoid unintended consequences and maximize quality of life, we used data from the MarketScan administrative claims databases to test the hypothesis that, among patients with nonvalvular AF, use of DOACs vs warfarin is associated with lower risk of incident clinical fracture. Head-to-head comparisons of individual DOACs were also conducted.

    Methods
    Study Population

    We used MarketScan Commercial Claims and Encounters and MarketScan Medicare Supplemental and Coordination of Benefits databases (IBM Corporation) from January 1, 2010, through September 30, 2015. The Commercial Claims and Encounters database consists of employer- and health plan–sourced data, whereas the Medicare Supplemental and Coordination of Benefits database includes beneficiaries with Medicare supplemental insurance paid for by employers.16 Missing from these databases are individuals with no insurance, and individuals working at small companies are underrepresented. Because this is a commercially insured population, age at AF is younger than that typically seen in clinical practice. The databases contain individual-level, deidentified health care claims information compliant with the Health Insurance Portability and Accountability Act, inclusive of enrollment records and inpatient, outpatient, ancillary, and drug claims, which are linked via individual-level identifiers. The University of Minnesota institutional review board deemed this analysis exempt from review and informed consent because the data were deidentified. This comparative effectiveness cohort study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.

    To identify AF, we required at least 1 inpatient claim for AF or 2 outpatient claims for AF from 7 to 365 days apart (International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM] codes 427.3, 427.31, and 427.32 in any position). The positive predictive value (PPV) of this definition is approximately 90%, and the sensitivity is approximately 80%.17 We excluded patients with ICD-9-CM codes indicating valvular disease or valve repair or replacement owing to the lack of approval of DOACs for valvular AF.

    The initial sample included 956 884 patients with nonvalvular AF aged 18 to 99 years. The analytic sample consisted of 546 214 patients once we restricted the population to individuals receiving OACs from January 1, 2010, to September 30, 2015; 217 962 patients after requiring at least 3 months of continuous enrollment before the first OAC prescription; and 212 995 patients after excluding those with fracture codes (any fracture in any position) in the 90-day run-in period. These 212 995 patients were eligible for matching (eFigure 1 in the Supplement).

    Anticoagulant Use

    New users of OACs (ie, individuals without known prior exposure to OACs) were categorized according to the first OAC they were prescribed. To mimic the intention-to-treat approach of a randomized clinical trial,18 participants remained in this drug category for the entire analysis. Validity of warfarin claims in administrative databases is excellent (sensitivity, 94%; PPV, 99%).19 Validity for DOACs claims has not yet been established.

    Initial Matching

    We matched each DOAC new user with as many as 3 warfarin-only new users by age (±3 years), sex, time since database enrollment (±90 days), and drug initiation date (±90 days). Matching was performed separately for each DOAC by using an automated greedy matching algorithm.20 The matching process was repeated for each head-to-head DOAC comparison using 1:1 matching.

    Fracture Ascertainment

    Outcomes of interest were (1) hip fractures (inpatient ICD-9-CM code 820), (2) fractures requiring hospitalization (inpatient codes only), and (3) all clinical fractures (inpatient and outpatient codes). Only fractures in the primary position were counted. The all-clinical and hospitalized fracture outcomes were identified by the presence of ICD-9-CM codes 733.1x, 733.93 to 733.98, or 800 to 829. Fractures associated with E codes (E800-E848) were not included, nor were fractures of toes, fingers, and skull or face (ICD-9-CM codes 733.94, 800, 801, 802, 803, 815, 816, 817, 825, or 826).

    Identification of Prespecified Covariates

    Prespecified covariates were derived using information before the date of initiation of OAC therapy (minimum, 90 days) obtained from all MarketScan data sources (ie, demographic data, inpatient, outpatient, and pharmacy claims). Validated published algorithms were used to define numerous prespecified comorbidities, prior procedures, and medications (pharmacy prescription fills).21,22Table 1 lists prespecified covariates; eTable 1 in the Supplement provides details of the codes used to derive these variables. Osteoporosis was defined by ICD-9-CM inpatient or outpatients codes 715 or 733 or by pharmacy claims indicative of osteoporosis treatment (ie, bisphosphonates [alendronate sodium, risedronate sodium, zoledronic acid, and ibandronate sodium], raloxifene hydrochloride, denosumab, and teriparatide). The CHA2DS2-VASc (congestive heart failure, hypertension, age [>65 years = 1 point; >75 years = 2 points], diabetes, and previous stroke/transient ischemic attack [2 points], vascular disease) score23 and a frailty index24 were calculated.

    Creation of High-Dimensional Propensity Scores and Rematching

    Separate high-dimensional propensity scores (HDPS)25 were calculated for each of the 6 main comparisons (individual DOACs vs warfarin or vs other DOACs) using the following steps25:

    1. MarketScan information was categorized into 5 domains: inpatient diagnostic codes, inpatient procedure codes, outpatient diagnostic codes, outpatient procedure codes, and medications. Within each domain, we selected the 200 most prevalent conditions, resulting in 1000 covariates.

    2. Covariates were then empirically ranked based on their potential for controlling confounding (ie, strength of the covariate-outcome association and prevalence of the covariate).26 The top 500 covariates were selected based on this ordering.

    3. The 500 covariates along with the prespecified covariates (also listed in Table 1) were included in a regression model to calculate the probability of exposure (ie, receiving a specific DOAC vs warfarin or the reference DOAC).

    As noted above, for the purpose of defining an index date and collecting covariate information at the time of initiation of drug therapy, we initially matched OAC users by age, sex, enrollment date, and OAC therapy initiation date. After computing the HDPSs, we rematched patients according to each outcome-specific HDPS to make the participant characteristics even more similar at the time of therapy initiation. A greedy matching technique with a caliper of 0.25 SD of each HDPS was used,20 with patients matched with 1 comparison patient (ie, users of warfarin or the reference DOAC).

    Statistical Analysis

    Data were analyzed from February 27, 2019 to September 18, 2019. Cox proportional hazards regression was used to compare the OACs according to risk of incident fracture. Follow-up began at the date of the first OAC prescription, and person-time accrued until incident fracture, health plan disenrollment (including due to death), or the end of study follow-up (September 30, 2015), whichever came first. Models were adjusted for age (continuous), sex, CHA2DS2-VASc score, HDPS, and frailty. Separate models also compared DOACs by dosage. Multiplicative interactions were evaluated by sex, age (<75 vs ≥75 years), and whether the claims indicated a diagnosis of osteoporosis (yes vs no). The Cox proportional hazards regression assumption was checked using Schoenfeld residuals, and there was no evidence of violations.

    Sensitivity analyses were also conducted by (1) adjusting for prevalent fracture (claims indicating fracture before the 90-day run-in period), (2) adjusting for diagnosis of osteoporosis, (3) excluding those with any prevalent fracture, (4) requiring at least 180 days (compared with 90 days in the primary analysis) of continuous enrollment before the first OAC prescription, and (5) censoring participants on evidence of nonadherence (ie, a gap in prescription refills). E values were also calculated to gauge the likelihood that unmeasured confounding explained our findings.27,28 Statistical analyses were performed with SAS, version 9.3 (SAS Institute Inc).

    Results

    Among the 212 995 patients with AF eligible for initial matching, a total of 167 275 patients were matched (38.0% women and 62.0% men; mean [SD] age, 68.9 [12.5] years), initially and by HDPS, and are included in the present analysis (eFigure 1 in the Supplement). Of these, 18.9% were prescribed dabigatran; 21.1%, rivaroxaban; 10.6%, apixaban; and 49.4%, warfarin. Table 1 displays characteristics of these matched participants by OAC initially prescribed. Individuals prescribed warfarin (mean [SD] age, 70.2 [12.3] years) and apixaban (mean [SD] age, 69.1 [12.6] years) tended to be older than patients prescribed dabigatran (mean [SD] age, 67.0 [12.4] years) and rivaroxaban (mean [SD] age, 67.7 [12.3] years). The mean (SD) CHA2DS2-VASC scores followed a similar pattern: 3.6 (2.0) for warfarin, 3.4 (2.0) for apixaban, 3.0 (1.9) for dabigatran, and 3.1 (1.9) for rivaroxaban.

    During a mean (SD) follow-up of 16.9 (13.7) months, 817 hip fractures, 2013 fractures requiring hospitalization, and 7294 clinical fractures occurred. The incidence per 1000 person-years was 3.7 for hip fracture, 9.3 for hospitalized fracture, and 35.7 for any clinical fracture. Adjusted hazard ratios (HRs) and 95% CIs for incident fracture according to DOACs vs warfarin are shown in Table 2. Relative to warfarin, being prescribed a DOAC was generally associated with lower risk of fractures requiring hospitalization (HR, 0.87; 95% CI, 0.79-0.96), all clinical fractures (HR, 0.93; 95% CI, 0.88-0.98), and hip fractures (HR, 0.91; 95% CI, 0.78-1.07). When comparing dabigatran with warfarin, there was some evidence of lower risk of fractures requiring hospitalization with dabigatran (HR, 0.88; 95% CI, 0.78-1.00), although for hip and all clinical fractures, the estimates were near the null. Rivaroxaban use compared with warfarin use was associated with lower risk of fractures requiring hospitalization (HR, 0.80; 95% CI, 0.69-0.93) and all clinical fractures (HR, 0.82; 95% CI, 0.76-0.89), although the estimate for hip fractures (HR, 0.89; 95% CI, 0.71-1.12) was not statistically significant. Apixaban use was associated with lower risk of all fracture outcomes compared with warfarin, including hip fractures (HR, 0.67; 95% CI, 0.45-0.98), fractures requiring hospitalization (HR, 0.60; 95% CI, 0.47-0.78), and all clinical fractures (HR, 0.86; 95% CI, 0.75-0.98). These findings were robust to a number of sensitivity analyses, including adjusting for prevalent fracture, adjusting for the diagnosis of osteoporosis, excluding those with prevalent fracture, and requiring a 6-month run-in (eTable 2 in the Supplement). Likewise, findings were similar when we censored participants at evidence of nonadherence; with a mean (SD) follow-up of 7.1 (8.0) months, there were 220 hip fractures, 525 fractures requiring hospitalization, and 2232 clinical fractures. The HRs for DOAC users vs matched warfarin users were 0.91 (95% CI, 0.70-1.19) for hip fracture, 0.81 (95% CI, 0.69-0.97) for inpatient fracture, and 0.91 (95% CI, 0.84-0.99) for all fractures.

    In analyses stratified by subgroups of interest (Figure 1 and eTable 3 in the Supplement), risk of hip fracture associated with DOAC use was lower among patients with AF who had claims indicating an osteoporosis diagnosis (HR, 0.74; 95% CI, 0.58-0.96) compared with those without osteoporosis (HR, 1.06; 95% CI, 0.86-1.30) (P = .03 for interaction). This finding was similar when evaluating fractures requiring hospitalization (HR for no osteoporosis, 0.97 [95% CI, 0.85-1.11]; HR for osteoporosis, 0.75 [95% CI, 0.64-0.88]; P = .59 for interaction) (Figure 2 and eTable 3 in the Supplement). Although interactions were not always significant, some evidence suggested that the lower risk associated with DOAC use was more pronounced in women than in men. Subgroup analyses for the all clinical fractures outcome are provided in eTable 3 and eFigure 2 in the Supplement. Table 3 provides head-to-head comparisons of the specific DOACs associated with fracture risk. Little evidence suggested that fracture risk varied according to type of DOAC prescribed.

    Precision was poorer for analyses comparing the association of different DOAC doses and warfarin with fracture risk (eTable 4 in the Supplement). Nevertheless, we found some evidence of dose response, whereby risk (vs warfarin) was lower for standard doses than for reduced doses of rivaroxaban. For example, the standard dose of rivaroxaban (20 mg) had an HR of 0.78 (95% CI, 0.58-1.05), while for the 15 mg dose, the HR was 1.02 (95% CI, 0.74-1.40).

    E values were calculated to evaluate how strong unmeasured confounding would have to be to negate the observed results. For the association of DOACs vs warfarin with fractures requiring hospitalization, the E value was 1.56, indicating that residual confounding could explain the observed association if an unmeasured covariate had a relative risk association at least as large as 1.56 with OAC choice (DOACs vs warfarin) and fractures requiring hospitalization. For all clinical fractures and hip fractures, the E values were 1.36 and 1.43, respectively. To put these results in context, in a model adjusted for Table 1 covariates, hypertension was associated with an HR for fracture of 0.79 (95% CI, 0.69-0.91), diabetes with an HR for fracture of 1.16 (95% CI, 1.02-1.31), and ischemic stroke with an HR for fracture of 1.09 (95% CI, 0.92-1.28). Even established risk factors for fracture, such as dementia (HR, 1.43; 95% CI, 1.08-1.91), had an association weaker than the E value, although the upper limit of the confidence interval exceeds the E value.

    Discussion

    Use of the DOACs, and particularly apixaban, was associated with lower incidence of clinical fracture relative to warfarin in this retrospective analysis of patients with AF who had health care insurance coverage. Although the lower risk in DOAC users vs warfarin users did not reach statistical significance for the outcome of hip fracture, precision was poorer for this outcome. In general, associations were more pronounced among patients with AF with a diagnosis of osteoporosis relative to those without. The strongest effect estimates were observed when comparing apixaban and warfarin; this finding was not hypothesized and, as such, warrants further scrutiny. We found no apparent differences in fracture risk in head-to-head DOAC comparisons. Collectively, our findings support the notion that warfarin may be harmful to bone health.

    Speculation that long-term warfarin use might increase fracture risk stems from warfarin’s role as a vitamin K antagonist.29 Vitamin K is important in posttranslational glutamination of osteocalcin, the major noncollagenous bone matrix protein. Warfarin interferes with this process and consequently inhibits the activation of bone matrix proteins.30 Experimental rat models31 and observational human studies32,33 have suggested that warfarin reduces blood osteocalcin levels and may impair bone quality. Few studies have directly compared the effects of DOACs vs warfarin on indicators of bone health. Switching from warfarin to rivaroxaban was associated with an increase in bone formation markers and a decrease in bone resorption markers in a study of 21 patients with AF.11 An animal study also found dabigatran to be advantageous over warfarin on numerous markers of bone health.34

    Our findings that warfarin is associated with greater fracture risk relative to DOACs is consistent with the evaluation of patients with AF from Hong Kong12 that showed greater fracture risk with use of warfarin than dabigatran. Conversely, a smaller study of patients with AF in the Florence, Italy, metropolitan area35 identified no difference in fracture risk by OAC. In post hoc analysis of the Effective Anticoagulation With Factor Xa Next Generation in Atrial Fibrillation–Thrombolysis in Myocardial Infarction 48 trial,36 findings suggested that edoxaban vs warfarin lowered fracture risk (HR, 0.88; 95% CI, 0.75-1.03) among participants at increased risk of falls but not among those at low fall risk. The present findings enhance the current literature in that we had far more fracture events among DOAC users (ie, 356 hip, 863 requiring hospitalization, and 3350 all clinical) than previous analyses and therefore reasonable precision to evaluate subgroups of interest.

    Head-to-head DOAC comparisons yielded no statistically significant differences in fracture risk, although findings suggested that apixaban was the most advantageous. The lack of robust association in comparisons of DOACs is not surprising given the hypothesis that warfarin has a detrimental effect on bone health and the lack of known pathways by which DOACs may influence bone metabolism. To our knowledge, no previous studies have yet evaluated head-to-head comparisons of DOACs and fracture risk.

    The hypothesis that warfarin is deleterious to bone health is not without controversy. Comparisons of warfarin users with individuals not prescribed an OAC have provided inconsistent results,14 with some studies reporting a link between warfarin and fracture37,38 and others showing no association.13,39,40 An alternate hypothesis is that perhaps instead of warfarin being harmful, DOACs may be beneficial. Although the mechanism is unclear, DOACs may influence the likelihood of fracture by reducing fall risk.12,15,36

    Our findings were robust to numerous sensitivity analyses, including when censoring at evidence of nonadherence, which is important given high OAC discontinuation rates among patients with AF.41 Some evidence suggests a dose-response association, although this effect may be due to uncontrolled confounding because patients prescribed reduced doses tend to be older and have more comorbidities.4 In subgroup analyses, a significant interaction occurred whereby DOACs were more beneficial among participants with diagnosed osteoporosis than those without. Differential effects of DOACs by osteoporosis status were not expected and should be evaluated in independent populations.

    Strengths and Limitations

    Strengths of the present analysis are the novel head-to-head DOAC comparisons, the large sample of patients with AF who had insurance coverage and consequent relatively large number of fracture events, and evaluation of the risks associated with these OACs in a real-world population that may be more generalizable across patients with AF and health care professionals than a randomized clinical trial population. Uncontrolled confounding is a chief potential limitation; as such, we used a multipronged approach to minimize confounding (ie, initially matching on key patient characteristics, creating and then rematching on empirically derived HDPS, and finally adjusting for the HDPS and other key patient characteristics). Misclassification is also an important consideration. Unfortunately, we were unable to locate a fracture definition that had been validated in a US population. As such, our definition was based on codes identified in a Taiwanese validation study (PPV, 100% for hip fracture, 86% for vertebral fracture, 100% for wrist and forearm fracture, and 100% for humerus fracture)12 and expert opinion. Last, edoxaban was not included in the present analysis because there were few users in our database.

    Conclusions

    Optimizing care of patients with AF is paramount. In the present study, use of DOACs, and particularly apixaban, was associated with lower fracture risk than use of warfarin. Furthermore, some evidence suggested that this association was more pronounced among patients with AF who had a diagnosis of osteoporosis. Given the detrimental effects of fracture in the elderly,42-44 caution should be used when prescribing warfarin to patients with AF at elevated fracture risk.

    Back to top
    Article Information

    Accepted for Publication: September 26, 2019.

    Corresponding Author: Pamela L. Lutsey, PhD, Division of Epidemiology and Community Health, University of Minnesota School of Public Health, 1300 S Second St, Ste 300, Minneapolis, MN 55454 (lutsey@umn.edu).

    Published Online: November 25, 2019. doi:10.1001/jamainternmed.2019.5679

    Author Contributions: Dr Lutsey and Ms Norby had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

    Concept and design: Lutsey, Norby, MacLehose, Alonso.

    Acquisition, analysis, or interpretation of data: Lutsey, Norby, Ensrud, Diem, Chen, Alonso.

    Drafting of the manuscript: Lutsey.

    Critical revision of the manuscript for important intellectual content: Norby, Ensrud, MacLehose, Diem, Chen, Alonso.

    Statistical analysis: Lutsey, Norby, MacLehose.

    Obtained funding: Alonso.

    Administrative, technical, or material support: Lutsey, Alonso.

    Supervision: Lutsey, MacLehose, Alonso.

    Conflict of Interest Disclosures: Dr Lutsey reported receiving grants from the National Institutes of Health (NIH) during the conduct of the study. Dr MacLehose reported receiving grants from the NIH during the conduct of the study. Dr Diem reported receiving grants from the NIH during the conduct of the study. Dr Chen reported receiving grants from the NIH during the conduct of the study. Dr Alonso reported receiving grants from the NIH and American Heart Association during the conduct of the study. No other disclosures were reported.

    Funding/Support: This study was supported by grants R01HL131579 (principal investigator, Dr Lutsey) and R01-HL122200 (principal investigator, Dr Alonso) from the National Heart, Lung, and Blood Institute, NIH; by grant R21 AG058445 from the National Institute on Aging; and grant 16EIA26410001 from the American Heart Association (Dr Alonso).

    Role of the Funder/Sponsor: The sponsors had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

    References
    1.
    Weng  L-C, Preis  SR, Hulme  OL,  et al.  Genetic predisposition, clinical risk factor burden, and lifetime risk of atrial fibrillation.  Circulation. 2018;137(10):1027-1038. doi:10.1161/CIRCULATIONAHA.117.031431PubMedGoogle ScholarCrossref
    2.
    Mou  L, Norby  FL, Chen  LY,  et al.  Lifetime risk of atrial fibrillation by race and socioeconomic status: ARIC study (Atherosclerosis Risk in Communities).  Circ Arrhythm Electrophysiol. 2018;11(7):e006350. doi:10.1161/CIRCEP.118.006350PubMedGoogle Scholar
    3.
    Colilla  S, Crow  A, Petkun  W, Singer  DE, Simon  T, Liu  X.  Estimates of current and future incidence and prevalence of atrial fibrillation in the US adult population.  Am J Cardiol. 2013;112(8):1142-1147. doi:10.1016/j.amjcard.2013.05.063PubMedGoogle ScholarCrossref
    4.
    January  CT, Wann  LS, Alpert  JS,  et al; American College of Cardiology/American Heart Association Task Force on Practice Guidelines.  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.  J Am Coll Cardiol. 2014;64(21):e1-e76. doi:10.1016/j.jacc.2014.03.022PubMedGoogle ScholarCrossref
    5.
    Connolly  SJ, Ezekowitz  MD, Yusuf  S,  et al; RE-LY Steering Committee and Investigators.  Dabigatran versus warfarin in patients with atrial fibrillation.  N Engl J Med. 2009;361(12):1139-1151. doi:10.1056/NEJMoa0905561PubMedGoogle ScholarCrossref
    6.
    Patel  MR, Mahaffey  KW, Garg  J,  et al; ROCKET AF Investigators.  Rivaroxaban versus warfarin in nonvalvular atrial fibrillation.  N Engl J Med. 2011;365(10):883-891. doi:10.1056/NEJMoa1009638PubMedGoogle ScholarCrossref
    7.
    Granger  CB, Alexander  JH, McMurray  JJV,  et al; ARISTOTLE Committees and Investigators.  Apixaban versus warfarin in patients with atrial fibrillation.  N Engl J Med. 2011;365(11):981-992. doi:10.1056/NEJMoa1107039PubMedGoogle ScholarCrossref
    8.
    Giugliano  RP, Ruff  CT, Braunwald  E,  et al; ENGAGE AF-TIMI 48 Investigators.  Edoxaban versus warfarin in patients with atrial fibrillation.  N Engl J Med. 2013;369(22):2093-2104. doi:10.1056/NEJMoa1310907PubMedGoogle ScholarCrossref
    9.
    Bengtson  LGS, Lutsey  PL, Chen  LY, MacLehose  RF, Alonso  A.  Comparative effectiveness of dabigatran and rivaroxaban versus warfarin for the treatment of non-valvular atrial fibrillation.  J Cardiol. 2017;69(6):868-876. doi:10.1016/j.jjcc.2016.08.010PubMedGoogle ScholarCrossref
    10.
    Norby  FL, Bengtson  LGS, Lutsey  PL,  et al.  Comparative effectiveness of rivaroxaban versus warfarin or dabigatran for the treatment of patients with non-valvular atrial fibrillation.  BMC Cardiovasc Disord. 2017;17(1):238. doi:10.1186/s12872-017-0672-5PubMedGoogle ScholarCrossref
    11.
    Namba  S, Yamaoka-Tojo  M, Kakizaki  R,  et al.  Effects on bone metabolism markers and arterial stiffness by switching to rivaroxaban from warfarin in patients with atrial fibrillation.  Heart Vessels. 2017;32(8):977-982. doi:10.1007/s00380-017-0950-2PubMedGoogle ScholarCrossref
    12.
    Lau  WC, Chan  EW, Cheung  CL,  et al.  Association between dabigatran vs warfarin and risk of osteoporotic fractures among patients with nonvalvular atrial fibrillation.  JAMA. 2017;317(11):1151-1158. doi:10.1001/jama.2017.1363PubMedGoogle ScholarCrossref
    13.
    Fiordellisi  W, White  K, Schweizer  M.  A systematic review and meta-analysis of the association between vitamin K antagonist use and fracture.  J Gen Intern Med. 2019;34(2):304-311. doi:10.1007/s11606-018-4758-2PubMedGoogle ScholarCrossref
    14.
    Veronese  N, Bano  G, Bertozzo  G,  et al.  Vitamin K antagonists’ use and fracture risk: results from a systematic review and meta-analysis.  J Thromb Haemost. 2015;13(9):1665-1675. doi:10.1111/jth.13052PubMedGoogle ScholarCrossref
    15.
    Sugiyama  T.  Osteoporotic fractures associated with dabigatran vs warfarin.  JAMA. 2017;318(1):90-91. doi:10.1001/jama.2017.6908PubMedGoogle ScholarCrossref
    16.
    IBM Watson Health. IBM MarketScan research databases for health services researchers (white paper). https://www.ibm.com/downloads/cas/6KNYVVQ2. Published April 2019. Accessed May 2, 2019.
    17.
    Jensen  PN, Johnson  K, Floyd  J, Heckbert  SR, Carnahan  R, Dublin  S.  A systematic review of validated methods for identifying atrial fibrillation using administrative data.  Pharmacoepidemiol Drug Saf. 2012;21(suppl 1):141-147. doi:10.1002/pds.2317PubMedGoogle ScholarCrossref
    18.
    Hernán  MA, Alonso  A, Logan  R,  et al.  Observational studies analyzed like randomized experiments: an application to postmenopausal hormone therapy and coronary heart disease.  Epidemiology. 2008;19(6):766-779. doi:10.1097/EDE.0b013e3181875e61PubMedGoogle ScholarCrossref
    19.
    Garg  RK, Glazer  NL, Wiggins  KL,  et al.  Ascertainment of warfarin and aspirin use by medical record review compared with automated pharmacy data.  Pharmacoepidemiol Drug Saf. 2011;20(3):313-316. doi:10.1002/pds.2041PubMedGoogle ScholarCrossref
    20.
    Division of Biomedical Statistics and Informatics, Mayo Clinic Research. Biomedical Statistics and Informatics [computer program]. http://bioinformaticstools.mayo.edu/research/gmatch/. Updated July 23, 2018. Accessed August 1, 2019.
    21.
    Cunningham  A, Stein  CM, Chung  CP, Daugherty  JR, Smalley  WE, Ray  WA.  An automated database case definition for serious bleeding related to oral anticoagulant use.  Pharmacoepidemiol Drug Saf. 2011;20(6):560-566. doi:10.1002/pds.2109PubMedGoogle ScholarCrossref
    22.
    Quan  H, Sundararajan  V, Halfon  P,  et al.  Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data.  Med Care. 2005;43(11):1130-1139. doi:10.1097/01.mlr.0000182534.19832.83PubMedGoogle ScholarCrossref
    23.
    Lip  GY, Nieuwlaat  R, Pisters  R, Lane  DA, Crijns  HJ.  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(2):263-272. doi:10.1378/chest.09-1584PubMedGoogle ScholarCrossref
    24.
    Kim  DH, Schneeweiss  S.  Measuring frailty using claims data for pharmacoepidemiologic studies of mortality in older adults: evidence and recommendations.  Pharmacoepidemiol Drug Saf. 2014;23(9):891-901. doi:10.1002/pds.3674PubMedGoogle ScholarCrossref
    25.
    Schneeweiss  S, Rassen  JA, Glynn  RJ, Avorn  J, Mogun  H, Brookhart  MA.  High-dimensional propensity score adjustment in studies of treatment effects using health care claims data.  Epidemiology. 2009;20(4):512-522. doi:10.1097/EDE.0b013e3181a663ccPubMedGoogle ScholarCrossref
    26.
    Bross  ID.  Spurious effects from an extraneous variable.  J Chronic Dis. 1966;19(6):637-647. doi:10.1016/0021-9681(66)90062-2PubMedGoogle ScholarCrossref
    27.
    VanderWeele  TJ, Ding  P.  Sensitivity analysis in observational research: introducing the E-value.  Ann Intern Med. 2017;167(4):268-274. doi:10.7326/M16-2607PubMedGoogle ScholarCrossref
    28.
    Mathur  MB, Ding  P, Riddell  CA, VanderWeele  TJ.  Web Site and R package for computing E-values.  Epidemiology. 2018;29(5):e45-e47. doi:10.1097/EDE.0000000000000864PubMedGoogle ScholarCrossref
    29.
    Watts  NB.  Adverse bone effects of medications used to treat non-skeletal disorders.  Osteoporos Int. 2017;28(10):2741-2746. doi:10.1007/s00198-017-4171-4PubMedGoogle ScholarCrossref
    30.
    Sugiyama  T, Kugimiya  F, Kono  S, Kim  YT, Oda  H.  Warfarin use and fracture risk: an evidence-based mechanistic insight.  Osteoporos Int. 2015;26(3):1231-1232. doi:10.1007/s00198-014-2912-1PubMedGoogle ScholarCrossref
    31.
    Sugiyama  T, Takaki  T, Sakanaka  K,  et al.  Warfarin-induced impairment of cortical bone material quality and compensatory adaptation of cortical bone structure to mechanical stimuli.  J Endocrinol. 2007;194(1):213-222. doi:10.1677/JOE-07-0119PubMedGoogle ScholarCrossref
    32.
    Rey-Sanchez  P, Lavado-Garcia  JM, Canal-Macias  ML, Rodriguez-Dominguez  MT, Bote-Mohedano  JL, Pedrera-Zamorano  JD.  Ultrasound bone mass in patients undergoing chronic therapy with oral anticoagulants.  J Bone Miner Metab. 2011;29(5):546-551. doi:10.1007/s00774-010-0250-8PubMedGoogle ScholarCrossref
    33.
    Namba  S, Yamaoka-Tojo  M, Hashikata  T,  et al.  Long-term warfarin therapy and biomarkers for osteoporosis and atherosclerosis.  BBA Clin. 2015;4:76-80. doi:10.1016/j.bbacli.2015.08.002PubMedGoogle ScholarCrossref
    34.
    Fusaro  M, Dalle Carbonare  L, Dusso  A,  et al.  Differential effects of dabigatran and warfarin on bone volume and structure in rats with normal renal function.  PLoS One. 2015;10(8):e0133847. doi:10.1371/journal.pone.0133847PubMedGoogle Scholar
    35.
    Lucenteforte  E, Bettiol  A, Lombardi  N, Mugelli  A, Vannacci  A.  Risk of bone fractures among users of oral anticoagulants: an administrative database cohort study.  Eur J Intern Med. 2017;44:e30-e31. doi:10.1016/j.ejim.2017.07.022PubMedGoogle ScholarCrossref
    36.
    Steffel  J, Giugliano  RP, Braunwald  E,  et al.  Edoxaban versus warfarin in atrial fibrillation patients at risk of falling: ENGAGE AF-TIMI 48 analysis.  J Am Coll Cardiol. 2016;68(11):1169-1178. doi:10.1016/j.jacc.2016.06.034PubMedGoogle ScholarCrossref
    37.
    Gage  BF, Birman-Deych  E, Radford  MJ, Nilasena  DS, Binder  EF.  Risk of osteoporotic fracture in elderly patients taking warfarin: results from the National Registry of Atrial Fibrillation 2.  Arch Intern Med. 2006;166(2):241-246. doi:10.1001/archinte.166.2.241PubMedGoogle ScholarCrossref
    38.
    Rejnmark  L, Vestergaard  P, Mosekilde  L.  Fracture risk in users of oral anticoagulants: a nationwide case-control study.  Int J Cardiol. 2007;118(3):338-344. doi:10.1016/j.ijcard.2006.07.022PubMedGoogle ScholarCrossref
    39.
    Mamdani  M, Upshur  REG, Anderson  G, Bartle  BR, Laupacis  A.  Warfarin therapy and risk of hip fracture among elderly patients.  Pharmacotherapy. 2003;23(1):1-4. doi:10.1592/phco.23.1.1.31922PubMedGoogle ScholarCrossref
    40.
    Misra  D, Zhang  Y, Peloquin  C, Choi  HK, Kiel  DP, Neogi  T.  Incident long-term warfarin use and risk of osteoporotic fractures: propensity-score matched cohort of elders with new onset atrial fibrillation.  Osteoporos Int. 2014;25(6):1677-1684. doi:10.1007/s00198-014-2662-0PubMedGoogle ScholarCrossref
    41.
    Zalesak  M, Siu  K, Francis  K,  et al.  Higher persistence in newly diagnosed nonvalvular atrial fibrillation patients treated with dabigatran versus warfarin.  Circ Cardiovasc Qual Outcomes. 2013;6(5):567-574. doi:10.1161/CIRCOUTCOMES.113.000192PubMedGoogle ScholarCrossref
    42.
    Cummings  SR, Melton  LJ.  Epidemiology and outcomes of osteoporotic fractures.  Lancet. 2002;359(9319):1761-1767. doi:10.1016/S0140-6736(02)08657-9PubMedGoogle ScholarCrossref
    43.
    Rachner  TD, Khosla  S, Hofbauer  LC.  Osteoporosis: now and the future.  Lancet. 2011;377(9773):1276-1287. doi:10.1016/S0140-6736(10)62349-5PubMedGoogle ScholarCrossref
    44.
    Virnig  B, Durham  SB, Folsom  AR, Cerhan  J.  Linking the Iowa Women’s Health Study cohort to Medicare data: linkage results and application to hip fracture.  Am J Epidemiol. 2010;172(3):327-333. doi:10.1093/aje/kwq111PubMedGoogle ScholarCrossref
    ×