Advertisement

Bleeding Risk Following Systemic Fluconazole or Topical Azoles in Patients with Atrial Fibrillation on Apixaban, Rivaroxaban, or Dabigatran

Open AccessPublished:November 30, 2021DOI:https://doi.org/10.1016/j.amjmed.2021.11.008

      Abstract

      Background

      Bleeding safety in relation to use of systemic fluconazole and topical azoles among patients with atrial fibrillation treated with apixaban, rivaroxaban, or dabigatran is insufficiently explored, despite clinical relevance and several reports suggesting associations.

      Methods

      Using nationwide Danish registers, we identified patients with atrial fibrillation initiated on apixaban, rivaroxaban, or dabigatran from 2012-2018. We investigated associations between bleeding incidents and systemic fluconazole or topical azole treatment using a case-crossover design with 30-day exposure windows and reported odds ratios (OR) with 95% confidence intervals (CI).

      Results

      We included 32,340 (36%), 32,409 (36%), and 24,940 (28%) patients initiated on apixaban, rivaroxaban, and dabigatran, respectively. Patients on apixaban were older (median age: 77 years; interquartile range [IQR] 70-84) compared with rivaroxaban users (median age: 75 years; IQR 68-82) and patients on dabigatran (median age: 73 years; IQR 66-80). Apixaban users had a significantly increased risk of bleeding following exposure to systemic fluconazole: odds ratio (OR) 3.5; 95% confidence interval (CI), 1.4-10.6. No increased risk was found among rivaroxaban and dabigatran users: ORs of 0.9 (95% CI, 0.2-3.0) and 1.7 (95% CI, 0.5-5.6), respectively. As to bleeding risk pertaining to topical azole exposure among apixaban, rivaroxaban, and dabigatran users, no association was found, with corresponding ORs of 0.8 (95% CI, 0.5-1.3); 1.3 (95% CI, 0.9-2.1); and 1.2 (95% CI 0.8-1.8), respectively.

      Conclusion

      In patients with atrial fibrillation on either apixaban, rivaroxaban, or dabigatran, an association between an elevated bleeding risk and use of systemic fluconazole was found among patients on apixaban. We found no increased risk of bleeding following co-exposure to topical azoles.

      Keywords

      Clinical Significance
      • A significant association between bleeding risk and use of systemic fluconazole was found among patients on apixaban. Added to available knowledge, this should encourage physicians to apply close monitoring or other risk mitigation strategies if co-administration is necessary.
      • No increased risk of bleeding following co-exposure to topical azoles was found. Thus, it seems safe—from our data—to administrate topical azoles concomitantly with apixaban, rivaroxaban, and dabigatran.

      Introduction

      Systemic antifungal treatment with most azoles has been associated with an increased risk of bleeding in patients on non-vitamin K antagonist oral anticoagulants (NOAC),
      • Steffel J
      • Verhamme P
      • Potpara TS
      • et al.
      The 2018 European Heart Rhythm Association Practical Guide on the use of non-vitamin K antagonist oral anticoagulants in patients with atrial fibrillation.
      thus, concomitant treatment is contraindicated.

      European Medicines Agency. European public assessment report, Eliquis. Available at: https://www.ema.europa.eu/en/medicines/human/EPAR/eliquis. Accessed June 22, 2021.

      European Medicines Agency. European public assessment report, Pradaxa. Available at: https://www.ema.europa.eu/en/medicines/human/EPAR/pradaxa. Accessed June 22, 2021.

      European Medicines Agency. European public assessment report, Xarelto. Available at: https://www.ema.europa.eu/en/medicines/human/EPAR/xarelto. Accessed June 22, 2021.

      European Medicines Agency. European public assessment report, Lixiana. Available at: https://www.ema.europa.eu/en/medicines/human/EPAR/lixiana. Accessed June 23, 2021.

      The interaction occurs due to azoles’ inhibition of cytochromes P450 3A4 (CYP3A4) and P-glycoprotein (P-gp), which are involved in the metabolism of apixaban, rivaroxaban, and edoxaban (CYP3A4 and P-gp); and dabigatran (P-gp).
      • Gelosa P
      • Castiglioni L
      • Tenconi M
      • et al.
      Pharmacokinetic drug interactions of the non-vitamin K antagonist oral anticoagulants (NOACs).
      ,
      • Fawzy AM
      • Lip GYH
      Pharmacokinetics and pharmacodynamics of oral anticoagulants used in atrial fibrillation.
      Paradoxically, fluconazole, the most commonly used systemic antifungal treatment,
      • Adriaenssens N
      • Coenen S
      • Versporten A
      • Goossens H
      Outpatient systemic antimycotic and antifungal use in Europe: new outcome measure provides new insight.
      is not officially contraindicated in patients on NOAC due to insufficient data.
      • Steffel J
      • Verhamme P
      • Potpara TS
      • et al.
      The 2018 European Heart Rhythm Association Practical Guide on the use of non-vitamin K antagonist oral anticoagulants in patients with atrial fibrillation.

      European Medicines Agency. European public assessment report, Eliquis. Available at: https://www.ema.europa.eu/en/medicines/human/EPAR/eliquis. Accessed June 22, 2021.

      European Medicines Agency. European public assessment report, Pradaxa. Available at: https://www.ema.europa.eu/en/medicines/human/EPAR/pradaxa. Accessed June 22, 2021.

      European Medicines Agency. European public assessment report, Xarelto. Available at: https://www.ema.europa.eu/en/medicines/human/EPAR/xarelto. Accessed June 22, 2021.

      European Medicines Agency. European public assessment report, Lixiana. Available at: https://www.ema.europa.eu/en/medicines/human/EPAR/lixiana. Accessed June 23, 2021.

      ,
      • Fawzy AM
      • Lip GYH
      Pharmacokinetics and pharmacodynamics of oral anticoagulants used in atrial fibrillation.
      However, recent observational studies have associated systemic fluconazole with an increased risk of bleeding in patients on apixaban or rivaroxaban.
      • Chang SH
      • Chou IJ
      • Yeh YH
      • et al.
      Association between use of non–vitamin K oral anticoagulants with and without concurrent medications and risk of major bleeding in nonvalvular atrial fibrillation.
      • Mueck W
      • Kubitza D
      • Becka M
      Co-administration of rivaroxaban with drugs that share its elimination pathways: pharmacokinetic effects in healthy subjects.
      • Fernandez S
      • Lenoir C
      • Samer C
      • Rollason V
      Drug interactions with apixaban: a systematic review of the literature and an analysis of VigiBase, the World Health Organization database of spontaneous safety reports.
      Further, a case report drawing the same line between topically applied miconazole and rivaroxaban
      • Lee K-G
      • Jennifer DL
      Increased bleeding tendency from interaction between rivaroxaban and topical miconazole: case report.
      —along with studies showing non-systemic azoles to interfere with warfarin treatment
      • Iversen DB
      • Hellfritzsch M
      • Stage TB
      • Aabenhus RM
      • Lind BS
      • Pottegård A
      Antimycotic treatment of oral candidiasis in warfarin users.
      ,
      • Miki A
      • Ohtani H
      • Sawada Y
      Warfarin and miconazole oral gel interactions: analysis and therapy recommendations based on clinical data and a pharmacokinetic model.
      —begs the question whether topical azoles could also be associated with elevated bleeding risk in patients on NOACs. Clinical data are scarce on this topic, and with azoles for topical use being available over the counter, it is warranted to investigate any potential association.
      Using Danish nationwide registers, we investigated the risk of bleeding following either systemic antifungal treatment with fluconazole or topical antifungal treatment with any azole agent in atrial fibrillation patients on either apixaban, rivaroxaban, or dabigatran.

      Methods and Materials

      Data Sources

      This study was based on nationwide, Danish administrative health registers previously used by this group.
      • Schmidt M
      • Pedersen L
      • Sørensen HT
      The Danish Civil Registration System as a tool in epidemiology.
      • Schmidt M
      • Schmidt SSAJ
      • Sandegaard JLJ
      • et al.
      The Danish National Patient Registry: a review of content, data quality, and research potential.
      • Kildemoes HW
      • Sorensen HT
      • Hallas J
      The Danish National Prescription Registry.
      • Holt A
      • Blanche P
      • Zareini B
      • et al.
      Effect of long-term beta-blocker treatment following myocardial infarction among stable, optimally treated patients without heart failure in the reperfusion era: a Danish, nationwide cohort study.
      Please see Appendix (available online) for details.

      Population

      We performed a nationwide, case-crossover study during 2012-2018. All patients aged 18-100 years old with a history of atrial fibrillation and no history of NOAC treatment were identified on the day of their first claimed prescription of any of the following NOACs: apixaban, rivaroxaban, or dabigatran. Edoxaban users were not included due to very low use in Denmark.

      Holt A, Blanche P, Zareini B, et al. Gastrointestinal bleeding risk following concomitant treatment with oral glucocorticoids in patients on non-vitamin K oral anticoagulants [online ahead of print]. Heart. 2021 Aug 13;heartjnl-2021-319503. doi:10.1136/heartjnl-2021-319503

      Only patients alive and without a change of NOAC agent 90 days after their first claimed prescription were included.
      To be certain that atrial fibrillation was the most likely indication for NOAC treatment, patients with venous thromboembolisms 6 months prior to—or total hip or knee arthroplasty 5 weeks prior to—date of redeemed prescription were excluded. Patients redeeming 2 different NOAC prescriptions on the first day were also excluded, as well as patients with history of valvular atrial fibrillation or coagulopathy (Figure 1).
      Figure 1
      Figure 1Flowchart of the study cohort.
      NOAC = Non-vitamin K antagonist oral anticoagulants.

      Study Design, Outcome, and Exposure

      The outcome was any bleeding incident requiring hospitalization (Supplementary Table 1, available online). Patients were followed from 90 days after first redeemed prescription of apixaban, rivaroxaban, or dabigatran until study end: December 31, 2018, death, emigration, or change of NOAC agent, whichever came first. Patients changing NOAC agent prior to follow-up start were excluded. Patients with a bleeding incident during follow-up were defined as cases (Figure 1).
      Supplementary Table 1ICD and ATC codes
      DetailsICD-10 and ATC codes used
      Study population
       Atrial fibrillationDefined from diagnosis codesICD-10: I48
       NOACsDefined from ATC-codesDabigatranATC: B01AE07

      Standard dosage: 150mg x 2
      RivaroxabanATC: B01AF01

      Standard dosage: 20mg x 1
      ApixabanATC: B01AF02

      Standard dosage: 5mg x 2
      EdoxabanATC: B01AF03

      Standard dosage: 60mg x 1
      Exclusion criteria
       CoagulopathiesDefined from diagnosis codesICD-10: D65-69
       Hip or knee replacementDefined from procedural codesICD-10: KNFB, KNFC, KNGB, KNGC
       Venous thrombo-embolismsDefined from diagnosis codesICD-10: I26, I801-803, I808-9, I821-23, I828, I829
       Valvular atrial fibrillationDefined from diagnosis codes or procedural codesICD-10: I05-6, Z952, Z954, I080A, I081A, I082A, I083A, KFKD, KFKH, KFMD, KFMH, KFGE, KFJF
      Exposure
       Systemic fluconazoleDefined from ATC codesATC: J02AC01
       Topical azoleDefined from ATC codesATC: A01AB09, D01AC, G01AF02, G01AF02, G01AF04, G01AF05, G01AG02
       Topical azole applied to mucosal tissue, onlyDefined from ATC codesATC: A01AB09, G01AF02, G01AF02, G01AF04, G01AF05, G01AG02
      Outcomes
       BleedingUpper gastrointestinal bleedingICD-10: K250, K252, K254, K256, K260, K262, K264, K266, K270, K272, K274, K276, K280, K282, K284, K286, K290, K920, I850, K228F, K298A, K638F-G, I864A
      Lower gastrointestinal bleedingICD-10: K625, K661, K638B
      Unspecified gastrointestinal bleedingICD-10: D500, K921-22, K638C
      Intracranial bleedingICD-10: I60-62, S064, S066
      Urinary tract bleedingICD-10: N02, R31
      Airway bleedingICD-10: R04
      Acute bleeding anemiaICD-10: D62
      OtherICD-10: I312, H313, H356, H431, H450, J942, H053A, G951A, S368D
      Comorbidities
       StrokeDefined from diagnosis codesICD-10: I63, I64, G45
       Heart failureDefined from diagnosis codesICD-10: I42, I50, J81, I110
       Ischemic heart diseaseDefined from diagnosis codesICD-10: I20-I25
       Prior bleedingDefined from diagnosis codesICD-10: I60-62, N02, R31, R04, D62, I312, H313, H356, H431, H450, S064, S066, J942, H053A, G951A, S368D, K250, K252, K254, K256, K260, K262, K264, K266, K625, K270, K272, K274, K276, K280, K828, K284, K286, K290, K920-22, K661, I850, D500, K228F, K298A, K638B-C, K638F-G, I864A
       Acute coronary syndromeDefined from diagnosis codesICD-10: I21-22, I200
       Liver diseaseDefined from diagnosis codesICD-10: B15-B19, C22, D684C, I982B, K70-K77, DQ618A, Z944
       Diabetes mellitusDefined from diagnosis codes or treatment with glucose lowering drugsICD-10: E10-14

      ATC: A10
       HypertensionDefined from diagnosis codes or the combination treatment with a least two classes of antihypertensive drugs (Adrenergic α-antagonist, non-loop-diuretics, vasodilators, beta- blockers, calcium channel blockers, and renin-angiotensin system inhibitors)ICD-10: I10-13, I15

      ATC: C02A, C02B, C02C, C02DA, C02DB, C02DD, C02DG, C02L, C03A, C03B, C03D, C03E, C03X, C07A, C07B, C07C, C07D, C07F, C08, C09AA, C09BA, C09BB, C09CA, C09DA, C09DB, C09XA02, C09XA52
       Chronic kidney diseaseDefined from diagnosis codesICD-10: E102, E112, E132, E142, I120, M200, M313, M319, M321B, N02-N08, N11-N12, N14, N18-N19, N26, N158-N160, N162-N164, N168, Q612-Q613, Q615, Q619

      ICD-8: 403, 404, 580-4, 590, 223, 25002, 40039, 59009, 59320, 75310-1, 75319
       CancerDefined from all cancer diagnosis codes, excluding non-melanoma skin cancerICD-10: C00-C43, C45-C97
       Chronic obstructive pulmonary diseaseDefined from diagnosis codesICD-10: J42-J44
       Alcohol useDefined from diagnosis codesICD-10: F10, K70, E52, T51, K860, E244, G312, I426, O354, Z714, G621, G721, K292
      Medication
       Acetylsalicylic acidATC: B01AC06
       Non-steroidal anti-inflammatory drugsATC: M01A, except M01AX05
       StatinsATC: C10A
       Beta-blockersATC: C07
       Renin angiotensin system inhibitorsIncluding: angiotensin-converting-enzyme inhibitors, angiotensin-II receptor blockersATC: C09
       Loop-diureticsATC: C03C
       ThiazidesATC: C03A, C07B, C07D, C09XA52, C03EA01
       ADPiATC: B01AC04, B01AC22, B01AC24
      NOAC = Non-vitamin K antagonist oral anticoagulants
      ATC:          Anatomical Therapeutic Chemical system
      ICD-10:       10th revision of the International Classification of Diseases system
      Diagnoses (primary or secondary), surgical procedures, and pharmacotherapy used for defining the study population, comorbidity, concomitant treatment, and outcomes.
      We investigated associations of 2 different exposures with bleeding: systemic fluconazole and topical azoles, including miconazole (oral, skin, and genital formulations), clotrimazole (skin and genital formulations), ketoconazole (skin formulation), and imidazole/triazole combined with corticosteroids (skin formulation) (Supplementary Table 2, available online). Patients were categorized as exposed if they redeemed a prescription within 30 days prior to outcome.
      Supplementary Table 2Distribution of Topical Azole Agents Use in Patients on Apixaban, Rivaroxaban, and Dabigatran
      Topical AzolesApixabanRivaroxabanDabigatran
      Oral formulations (%)
       Miconazole12.111.410.5
      Skin formulations (%)
       Clotrimazole1.72.01.8
       Miconazole19.218.717.1
       Ketoconazole15.816.117.8
       Imidazole/triazole combined with corticosteroids48.849.750.4
      Genital formulations (%)
       Clotrimazole1.61.21.3
       Miconazole0.80.91.1
      Total (n)10,09787218234
      Total number of claimed prescriptions within the first 3 years after initiation on non-vitamin K oral anticoagulants, including agent proportions.

      Characteristics of the Cohort

      Comorbidities were defined binarily as present if any of the following diagnoses were given up to 5 years prior to baseline (90 days after first claimed NOAC prescription): heart failure, acute coronary syndrome, ischemic heart disease, stroke, chronic obstructive pulmonary disease, chronic renal disease, liver disease, alcohol abuse, cancer, or bleeding. Concomitant treatment was defined as a redeemed prescription up to 6 months prior to baseline of the following medicaments: diuretics, antihypertensive drugs, antiplatelet agents, statins, and non-steroidal anti-inflammatory drugs. To capture patients being treated for hypertension and diabetes mellitus outside of hospitals, we defined both conditions as either a diagnosis or as a combination treatment with at least 2 antihypertensive drugs or treatment with a glucose-lowering drug, respectively (Supplementary Table 1).
      The patients’ baseline bleeding risk was illustrated by a modified HAS-BLED (hypertension, abnormal renal or liver function, stroke, bleeding, labile INR [international normalized ratio], elderly, drugs or alcohol) score based on history of hypertension, renal or liver disease, stroke, bleeding, age >65 years, antiplatelet or non-steroidal anti-inflammatory drug use, and alcohol abuse.
      • Lip GYH
      • Skjøth F
      • Nielsen PB
      • Kjældgaard JN
      • Larsen TB
      The HAS-BLED, ATRIA, and ORBIT bleeding scores in atrial fibrillation patients using non-vitamin K antagonist oral anticoagulants.
      CHA2DS2‐VASc (congestive heart failure, hypertension, age ≥75 years, diabetes mellitus, stroke or transient ischemic attack, vascular disease, age 65-74 years, sex category) scores at baseline were presented as well, using information obtained from the registers as described above.
      • Olesen JB
      • Lip GYH
      • Hansen ML
      • et al.
      Validation of risk stratification schemes for predicting stroke and thromboembolism in patients with atrial fibrillation: nationwide cohort study.

      Statistical Methods

      Baseline characteristics were presented with counts and percentages and continuous variables as median with interquartile range (IQR) stratified by NOAC agent.
      We reported proportions of patients redeeming at least 1, 2, 3, and 4 prescriptions of systemic fluconazole and topical azoles according to NOAC agent. These were estimated using appropriate methods to analyze recurrent events in the presence of censored data and competing risk of death.
      • Scheike TH
      • Eriksson F
      • Tribler S
      The mean, variance and correlation for bivariate recurrent event data with a terminal event.
      For our main analyses, we associated concomitant treatment with an azole agent and an NOAC agent with an outcome of bleeding requiring hospitalization using a case-crossover design. The case-crossover design is a “case-only” design that compares an individual's exposure in an index period just prior to the event occurred with a reference period prior to the index period.
      • Maclure M
      The case-crossover design: a method for studying transient effects on the risk of acute events.
      We used 30-day periods separated by a 10-day washout period to avoid lagged effects. Additionally, we also investigated 20- and 40-day periods. With the 30-day analysis as an example, the index period would be 0-30 days prior to the event, and the reference period would be 40-70 days prior to the event. By using each patient as his or her own control, this method adjusts for any unmeasured confounding that does not vary from the start of the reference period to the end of the index period. This makes it a suitable approach for studying the risk of abrupt outcomes following transient exposures.
      • Cadarette SM
      • Maclure M
      • Delaney JAC
      • et al.
      Control yourself: ISPE-endorsed guidance in the application of self-controlled study designs in pharmacoepidemiology.
      The case-crossover design only includes cases with a discordant exposure history, meaning that cases exposed only in the index period support an odds ratio (OR) of >1 and vice versa for cases exposed only in the reference period. Cases with a concordant exposure history, either exposed or not exposed in both periods, did not contribute to the analyses.
      We used R (version 3.5.0 for Windows, R Foundation for Statistical Computing)

      R Core Team. R: a language and environment for statistical computing. Vienna, Austria; R Foundation for Statistical Computing. Available at: https://www.r-project.org/. Accessed May 20, 2020.

      for data management, statistics, and illustrations.

      Ethical Considerations

      Retrospective studies using administrative health databases do not need ethical approval in Denmark. The study was approved by the Data Protection organization of the Capital Region of Denmark (Approval no. P-2019-348).

      Supplementary Analyses

      We performed several supplementary analyses. 1) To account for a possible impact of temporal trends in the prescribing patterns of azole agents, which could bias the case-crossover analysis, we performed additional analyses using a case-time-control design as well.
      • Suissa S
      The case-time-control design.
      We extended the case-crossover analysis with a set of controls matched 1:4 on index day since inclusion (day of event for the matched case), age, sex, and calendar year of inclusion. Temporal evolutions in prescribing patterns, if any, should be accounted for via the control group.
      • Suissa S
      The case-time-control design.
      2) To identify whether topical azoles applied to mucosal tissue (ie, potentially larger and clinically relevant systemic uptake) were associated with an increased risk of bleeding, separate analyses were conducted. Specifically, vaginally or orally applied azoles were sought associated to the risk of bleeding using the case-crossover design in 20-, 30-, and 40-day time windows.

      Results

      Cohort Characteristics

      We included 32,340 (36%), 32,409 (36%), and 24,940 (28%) patients initiated on apixaban, rivaroxaban, and dabigatran, respectively. Patients on apixaban were older (median age: 77 years, interquartile range [IQR] 70-84) compared with rivaroxaban users (median age: 75 years, IQR 68-82), and most markedly when compared with patients on dabigatran (median age: 73 years, IQR 66-80). Comorbidities were more prevalent among apixaban users compared with rivaroxaban and dabigatran, with the 2 latter being comparable (Table).
      TableBaseline Characteristics at Time of Inclusion According to NOAC Agent
      CharacteristicsApixaban(n = 32,340)Rivaroxaban(n = 32,409)Dabigatran(n = 24,940)
      Sex, n (%)17,193 (53.2)18,343 (56.6)14,672 (58.8)
      Age, median [IQR]77 [70–84]75 [68–82]73 [66–80]
      HAS-BLED score, median [IQR]2 [2–3]2 [2–3]2 [2–3]
      CHA₂DS₂-VASc score, median [IQR]4 [3-4]3 [2-4]3 [2-4]
      Standard NOAC

      dosage, n (%)
      21,172 (65.5)25,138 (77.6)14,888 (59.7)
      Comorbidities
       Hypertension29,951 (92.6)29,694 (91.6)23,050 (92.4)
       Diabetes mellitus5769 (17.8)5431 (16.8)3735 (15.0)
       Heart failure6071 (18.8)5370 (16.6)3957 (15.9)
       Acute coronary syndrome2075 (6.4)1863 (5.7)1483 (5.9)
       Ischemic heart disease5970 (18.5)5529 (17.1)4379 (17.6)
       Stroke5253 (16.2)3791 (11.7)3263 (13.1)
       Chronic obstructive

      pulmonary disease
      3718 (11.5)3312 (10.2)2212 (8.9)
       Chronic kidney disease1918 (5.9)1329 (4.1)537 (2.2)
       Cancer4320 (13.4)4178 (12.9)2693 (10.8)
       Prior bleeding4109 (12.7)3574 (11.0)2527 (10.1)
       Alcohol abuse794 (2.5)730 (2.3)609 (2.4)
       Chronic liver disease375 (1.2)362 (1.1)216 (0.9)
      Medication
       Beta-blockers22,496 (69.6)22,627 (69.8)18,063 (72.4)
       RAS inhibitors16,447 (50.9)16,302 (50.3)12,521 (50.2)
       Thiazides4967 (15.4)4882 (15.1)4321 (17.3)
       Loop diuretics11,212 (34.7)9869 (30.5)6780 (27.2)
       Statins13,961 (43.2)13,584 (41.9)10,464 (42.0)
       Acetyl salicylic acid7649 (23.7)7255 (22.4)7337 (29.4)
       Antiplatelets10,106 (31.2)9281 (28.6)8456 (33.9)
       NSAID3488 (10.8)3632 (11.2)3115 (12.5)
      CHA₂DS₂-VASc = congestive heart failure, hypertension, age ≥75 years, diabetes mellitus, stroke or transient ischemic attack, vascular disease, age 65-74 years, sex category; HAS-BLED = hypertension, abnormal renal or liver function, stroke, bleeding, labile INR (international normalized ratio), elderly, drugs or alcohol; NOAC = non-vitamin K antagonist oral anticoagulants; NSAID = non-steroid antiinflammatory drug; RAS = renin angiotensin system.

      Use of Azoles

      Systemic fluconazole use did not differ according to NOAC agent, with 4.9%-5.3% of patients claiming at least one prescription, and 1.5%-1.8% claiming at least 2, within the first 3 years from NOAC initiation (Figure 2A).
      Figure 2
      Figure 2Use of systemic fluconazole and topical azole agents among patients on apixaban, rivaroxaban, and dabigatran. Proportion of patients who claim at least 1, 2, 3, or 4 prescriptions of either (A) systemic fluconazole or (B) any topical azole agent within the first 3 years from inclusion, 90 days after initiation on either apixaban, rivaroxaban, or dabigatran.
      Treatment with topical azoles was more frequent, with 16.1%-17.6% claiming at least one prescription within 3 years of NOAC initiation, and 2.5%-2.9% claiming at least 4 prescriptions. Uptake among apixaban users was slightly more prevalent compared with dabigatran and rivaroxaban (Figure 2B).
      Different agents of topical azoles were equally used in the 3 NOAC groups. Skin-applied imidazole/triazole combined with corticosteroids (50%) and ketoconazole (16%) along with different formulations of miconazole (31%) were the most used topical azole agents (Supplementary Table 2).

      Bleeding

      During the study period, 1990, 2403, and 2277 patients on apixaban, rivaroxaban, and dabigatran, respectively, were hospitalized due to bleeding. Overall, the gastrointestinal tract was the most frequent bleeding site, with proportions of 38.7%, 39.8%, and 49.7%, respectively, among patients on apixaban, rivaroxaban, and dabigatran. Lower gastrointestinal bleeding was more prevalent among patients on dabigatran (21.9%) compared with patients on either apixaban (13.4%) or rivaroxaban (13.7%). Airway bleeding was most common among patients on rivaroxaban (25.6%) compared with apixaban (20.2%) and dabigatran (14.8%) (Supplementary Table 3, available online).
      Supplementary Table 3Distribution of Bleeding Incidents During Follow-Up According to Non-Vitamin K Oral Anticoagulant Agent
      Type of BleedingApixabanRivaroxabanDabigatran
      Gastrointestinal bleeding (%)
       Upper10.910.29.6
       Lower13.413.721.9
       Unspecified14.415.918.2
      Intracranial bleeding (%)13.412.710.8
      Urinary tract bleeding (%)19.717.417.9
      Airway bleeding (%)20.225.614.8
      Acute bleeding anemia (%)6.03.44.8
      Other (%)2.11.21.9
      Total (n)199024032277
      Bleeding incidents according to site of bleeding and non-vitamin K oral anticoagulant agent treatment.
      Among apixaban users, a significantly increased risk of bleeding following exposure to systemic fluconazole was found with a case-crossover odds ratio (OR) of 3.5 (95% confidence interval [CI], 1.4-10.6) for 30-day exposure windows. Amid apixaban users with a bleeding incident exposed to systemic fluconazole in the index period and with a discordant exposure history, gastrointestinal bleedings accounted for 65%, hematuria 10%, airway bleeding 10%, and other bleedings 15% vs 50% (gastrointestinal), 16.7% (hematuria), 16.7% (airway bleeding), and intracranial bleeding (16.7%) for patients exposed in the control period.
      Among rivaroxaban and dabigatran users, the risk was not significantly elevated with case-crossover ORs of 0.9 (95% CI, 0.2-3.0) and 1.7 (95% CI, 0.5-5.6), respectively, for 30-day exposure windows (Figure 3).
      Figure 3
      Figure 3Exposure to systemic fluconazole and the risk of bleeding among patients on apixaban, rivaroxaban, and dabigatran. Data were analyzed and presented using a case-crossover design with 20-, 30-, and 40-day exposure windows.
      *Cases contributing to the analyses were those having a discordant drug exposure history in the index and reference periods.
      CI = confidence intervals.
      When assessing bleeding risk pertaining to topical azole exposure among apixaban, rivaroxaban, and dabigatran users, no association was found with corresponding case-crossover ORs of 0.8 (95% CI, 0.5-1.3), 1.3 (95% CI, 0.9-2.1), and 1.2 (95% CI, 0.8-1.8) for 30-day exposure windows (Figure 4).
      Figure 4
      Figure 4Exposure to topical azole agents and the risk of bleeding among patients on apixaban, rivaroxaban, and dabigatran. Data were analyzed and presented using a case-crossover design with 20-, 30-, and 40-day exposure windows.
      *Cases contributing to the analyses were those having a discordant drug exposure history in the index and reference periods.
      CI = confidence intervals.
      Assessing the same associations using 20- and 40-day exposure windows yielded similar results, yet the significant association between systemic fluconazole and bleeding risk among apixaban users was not found using 40-day exposure windows (OR 1.9; 95% CI, 0.9-4.2). A significant association between topical azoles and bleeding among patients on rivaroxaban was found using 20-day exposure windows (OR 1.8; 95% CI, 1.1-3.2), but not when using 30- or 40-day exposure windows (Figures 3 and 4).

      Supplementary Analyses

      1) Using a case-time-control design to assess associations between azole exposure and the risk of bleeding yielded similar results compared with the main case-crossover analyses. ORs showing a significant association between systemic fluconazole exposure and bleeding among patients on apixaban remained consistent using 20- and 30-day exposure windows. The association between topical azoles and bleeding among patients on rivaroxaban using 20-day exposure windows was not significant in the case-time-control analysis (Supplementary Tables 4 and 5, available online).
      Supplementary Table 4Systemic Fluconazole Exposure and the Risk of Bleeding Among Patients on Apixaban, Rivaroxaban, and Dabigatran, According to the Case-Time-Control Design
      Odds Ratios (95% Confidence Interval)Events Contributing to

      the Analysis
      Cases contributing to the analyses were those having a discordant drug exposure history in the index and reference periods.
      Apixaban
       20-day exposure windows4.5 (1.4-14.6)24
       30-day exposure windows2.8 (1.0-8.3)27
       40-day exposure windows1.4 (0.6-3.4)35
      Rivaroxaban
       20-day exposure windows1.0 (0.3-3.7)13
       30-day exposure windows2.0 (0.8-5.1)30
       40-day exposure windows1.5 (0.6-3.4)37
      Dabigatran
       20-day exposure windows2.6 (0.8-8.8)16
       30-day exposure windows3.0 (1.1-8.4)24
       40-day exposure windows1.5 (0.7-3.3)40
      Data were analyzed and presented using a case-time-control design including a control group matched to cases on age, sex, calendar year, non-vitamin K antagonist oral anticoagulant agent, and follow-up time.
      low asterisk Cases contributing to the analyses were those having a discordant drug exposure history in the index and reference periods.
      Supplementary Table 5Topical Azole Exposure and the Risk of Bleeding Among Patients on Apixaban, Rivaroxaban, and Dabigatran, According to a Case-Time-Control Design
      Odds Ratios (95% Confidence Interval)Events Contributing to

      the Analysis
      Cases contributing to the analyses were those having a discordant drug exposure history in the index and reference periods.
      Apixaban
       20-day exposure windows0.9 (0.5-1.6)53
       30-day exposure windows0.6 (0.4-1.0)85
       40-day exposure windows0.7 (0.4-1.0)104
      Rivaroxaban
       20-day exposure windows1.7 (0.9-3.1)65
       30-day exposure windows1.3 (0.8-2.2)93
       40-day exposure windows1.1 (0.7-1.6)119
      Dabigatran
       20-day exposure windows1.1 (0.6-2.0)66
       30-day exposure windows1.4 (0.9-2.4)91
       40-day exposure windows1.6 (1.0-2.5)106
      Data were analyzed and presented using a case-time-control design including a control group matched to cases on age, sex, calendar year, NOAC agent and follow-up time.
      low asterisk Cases contributing to the analyses were those having a discordant drug exposure history in the index and reference periods.
      2) When assessing topical azoles applied to mucosal tissue in separate case-crossover analyses, no clear association between exposure and the risk of bleeding was found either. However, some of these results were unfit for interpretation due to lack of power (Supplementary Table 6, available online).
      Supplementary Table 6Azoles Applied to Mucosal Tissue and the Risk of Bleeding Among Patients on Apixaban, Rivaroxaban, and Dabigatran, According to the Case-Crossover Design
      Odds Ratios (95% Confidence Interval)Events Contributing to the Analysis
      Cases contributing to the analyses were those having a discordant drug exposure history in the index and reference periods.
      Apixaban
      20-day exposure windows3.0 (0.5-30.4)8
      30-day exposure windows1.8 (0.5-6.8)14
      40-day exposure windows2.3 (0.6-10.0)13
      Rivaroxaban
      20-day exposure windows
      Zero cases exposed in the reference period, making the analysis impossible to run.
      11
      30-day exposure windows1.9 (0.7-5.5)20
      40-day exposure windows1.3 (0.5-3.6)21
      Dabigatran
      20-day exposure windows2.5 (0.4-26.3)7
      30-day exposure windows1.8 (0.4-8.2)11
      40-day exposure windows1.5 (0.5-5.1)15
      Data were analyzed and presented using a case-time-control design including a control group matched to cases on age, sex, calendar year, non-vitamin K antagonist oral anticoagulant agent, and follow-up time.
      low asterisk Cases contributing to the analyses were those having a discordant drug exposure history in the index and reference periods.
      Zero cases exposed in the reference period, making the analysis impossible to run.

      Discussion

      In a nationwide population of patients with atrial fibrillation on NOACs, systemic fluconazole exposure was associated with a significantly elevated risk of bleeding only among patients treated with apixaban in a 30-day exposure window and not rivaroxaban or dabigatran. Further, no consistent association with bleeding risk was found following exposure to topical azoles on either NOAC agent—despite widespread use.

      Systemic Azoles

      Previous studies have linked systemic fluconazole with an increased risk of bleeding both in pooled cohorts of NOAC-treated patients
      • Chang SH
      • Chou IJ
      • Yeh YH
      • et al.
      Association between use of non–vitamin K oral anticoagulants with and without concurrent medications and risk of major bleeding in nonvalvular atrial fibrillation.
      ,
      • Mueck W
      • Kubitza D
      • Becka M
      Co-administration of rivaroxaban with drugs that share its elimination pathways: pharmacokinetic effects in healthy subjects.
      and specifically, among patients on apixaban.
      • Fernandez S
      • Lenoir C
      • Samer C
      • Rollason V
      Drug interactions with apixaban: a systematic review of the literature and an analysis of VigiBase, the World Health Organization database of spontaneous safety reports.
      However, the lack of self-controlled and active comparator designs makes interpretation difficult because unmeasured confounding by indication could be impactful when comparing patients with and without an indication for systemic antifungal treatment.
      • Nørgaard M
      • Ehrenstein V
      • Vandenbroucke JP
      Confounding in observational studies based on large health care databases: problems and potential solutions - a primer for the clinician.
      In our cohort, systemic fluconazole was seldomly used among patients on NOACs. Less than 2% claimed more than one prescription within 3 years from initiation on either NOAC. This could imply that physicians are aware of the contraindication regarding all other systemic azole agents.
      • Steffel J
      • Verhamme P
      • Potpara TS
      • et al.
      The 2018 European Heart Rhythm Association Practical Guide on the use of non-vitamin K antagonist oral anticoagulants in patients with atrial fibrillation.
      Nevertheless, a significantly increased risk of bleeding was associated with systemic fluconazole use among patients on apixaban with gastrointestinal bleeding as the predominant site, suggesting a direction for any warranted risk mitigation and monitoring. A significant association was not found among patients on rivaroxaban or dabigatran. The lack of association could be expected among patients on dabigatran due to the absence of CYP3A4 involvement, but—due to the shared elimination pathway with apixaban—it was surprising that no association was found among patients on rivaroxaban either. An explanation could be that patients on apixaban were older and more comorbid—thus more fragile—compared with patients on rivaroxaban, resulting in a bigger risk of hospitalization. Alternatively, the interaction between systemic fluconazole and rivaroxaban might not be clinically relevant, as proposed previously.
      • Steffel J
      • Verhamme P
      • Potpara TS
      • et al.
      The 2018 European Heart Rhythm Association Practical Guide on the use of non-vitamin K antagonist oral anticoagulants in patients with atrial fibrillation.
      ,
      • Fernandez S
      • Lenoir C
      • Samer C
      • Rollason V
      Drug interactions with apixaban: a systematic review of the literature and an analysis of VigiBase, the World Health Organization database of spontaneous safety reports.
      On the contrary, apixaban has previously been identified as the safest choice in the presence of CYP3A4 and P-gp inhibitors among patients with cancer.
      • Agnelli G
      • Becattini C
      • Meyer G
      • et al.
      Apixaban for the treatment of venous thromboembolism associated with cancer.
      ,
      • Verso M
      • Munoz A
      • Bauersachs R
      • et al.
      Effects of concomitant administration of anticancer agents and apixaban or dalteparin on recurrence and bleeding in patients with cancer-associated venous thromboembolism.
      Everything considered, it seems difficult to derive a clear-cut interpretation from our data regarding systemic fluconazole. However, because these questions cannot be answered through clinical trials, it is vital to provide and interpret all observational data available.
      • Hernán MA
      Causal analyses of existing databases: no power calculations required.

      Topical Azoles

      In general, topical azoles applied to intact skin have not been assumed to be absorbed systemically, but topical azoles applied to oral or vaginal mucosa could be absorbed systemically to a low degree.

      Prescriber's digital reference. Available at: https://www.pdr.net/. Accessed June 7, 2021.

      Disturbingly, a case report along with studies showing an interaction with warfarin have suggested otherwise.
      • Lee K-G
      • Jennifer DL
      Increased bleeding tendency from interaction between rivaroxaban and topical miconazole: case report.
      • Iversen DB
      • Hellfritzsch M
      • Stage TB
      • Aabenhus RM
      • Lind BS
      • Pottegård A
      Antimycotic treatment of oral candidiasis in warfarin users.
      • Miki A
      • Ohtani H
      • Sawada Y
      Warfarin and miconazole oral gel interactions: analysis and therapy recommendations based on clinical data and a pharmacokinetic model.
      In our cohort, the use of topical azoles was much more prevalent than the use of systemic fluconazole. An association between topical azoles and bleeding among patients on rivaroxaban was found using the case-crossover design and the 20-day exposure window, although this could not be reproduced using the case-time-control design or any other exposure windows. Thus, our interpretation is that no consistent association between topical azoles and the risk of bleeding among patients on either apixaban, rivaroxaban, or dabigatran can be concluded. Assessing topical azoles applied to mucosal tissue in a separate analysis yielded no significant findings either. This suggests that topical azoles are not absorbed sufficiently for the interaction to be clinically relevant. This is reassuring due to the widespread use, as well as the over-the-counter availability, of topical azoles.

      Strengths and Limitations

      The main strength of this study is the inclusion of a large cohort of NOAC-treated patients with atrial fibrillation from nationwide registers, which should minimize inclusion and selection bias. Further, we find the interpretation of our results to be much easier and generalizable due to the self-controlled designs used; however, unmeasured confounding might still bias our results. We examined different lengths of exposure windows to elevate the chance of capturing an association, if any was present. Because the timespan from co-exposure to a possible increase in bleeding risk remains an unanswered question, we believe it improves the robustness of our conclusions.
      Not many patients were exposed to systemic fluconazole, which resulted in large CIs, making interpretation difficult for these analyses. We would need very narrow CIs to conclude that no association exists, but on the contrary, statistically significant findings can probably be trusted—despite large CIs. Insufficient power was also a limitation in relation to the supplementary analysis concerning topical azoles applied to mucosal tissue. Reassuringly, this could mean—at least in a Danish context—that the prevalence of co-exposure in relation to a bleeding incident was extremely low.
      We assumed that patients initiated on NOACs continued the treatment throughout the study period. This is a limitation, although non-adherence would draw our results toward neutral associations.

      Clinical Perspectives

      The interaction between most systemic azoles and NOACs is well known, although fluconazole is not contraindicated in guidelines nor product summaries.
      • Steffel J
      • Verhamme P
      • Potpara TS
      • et al.
      The 2018 European Heart Rhythm Association Practical Guide on the use of non-vitamin K antagonist oral anticoagulants in patients with atrial fibrillation.

      European Medicines Agency. European public assessment report, Eliquis. Available at: https://www.ema.europa.eu/en/medicines/human/EPAR/eliquis. Accessed June 22, 2021.

      European Medicines Agency. European public assessment report, Pradaxa. Available at: https://www.ema.europa.eu/en/medicines/human/EPAR/pradaxa. Accessed June 22, 2021.

      European Medicines Agency. European public assessment report, Xarelto. Available at: https://www.ema.europa.eu/en/medicines/human/EPAR/xarelto. Accessed June 22, 2021.

      European Medicines Agency. European public assessment report, Lixiana. Available at: https://www.ema.europa.eu/en/medicines/human/EPAR/lixiana. Accessed June 23, 2021.

      Observational studies are often the best available measure when assessing drug safety following post-market approval.
      • Hernán MA
      Causal analyses of existing databases: no power calculations required.
      ,
      • Zipursky J
      • Juurlink DN
      Studying drug safety in the real world.
      When adding our data of a possible association between bleeding risk and systemic fluconazole use among patients on apixaban to available knowledge,
      • Chang SH
      • Chou IJ
      • Yeh YH
      • et al.
      Association between use of non–vitamin K oral anticoagulants with and without concurrent medications and risk of major bleeding in nonvalvular atrial fibrillation.
      ,
      • Fernandez S
      • Lenoir C
      • Samer C
      • Rollason V
      Drug interactions with apixaban: a systematic review of the literature and an analysis of VigiBase, the World Health Organization database of spontaneous safety reports.
      it could be time to review official recommendations.
      Recent reports have described interactions between NOACs and warfarin and topical azoles;
      • Lee K-G
      • Jennifer DL
      Increased bleeding tendency from interaction between rivaroxaban and topical miconazole: case report.
      • Iversen DB
      • Hellfritzsch M
      • Stage TB
      • Aabenhus RM
      • Lind BS
      • Pottegård A
      Antimycotic treatment of oral candidiasis in warfarin users.
      • Miki A
      • Ohtani H
      • Sawada Y
      Warfarin and miconazole oral gel interactions: analysis and therapy recommendations based on clinical data and a pharmacokinetic model.
      however, our data could not confirm that this interaction is clinically pertinent. This lack of association may, however, need further exploration, particularly regarding topical azoles applied to mucosal tissue among patients on NOACs.

      Conclusion

      In a nationwide cohort of patients with atrial fibrillation on NOACs, a significant association between an elevated bleeding risk and use of systemic fluconazole was found among patients on apixaban only. We found no significantly increased risk of bleeding following co-exposure to topical azoles.

      Appendix

      Contents

      Page 2………………………Description of data sources
      Page 3… …Supplementary Table 1: ICD and ATC codes
      Page 6… Supplementary Table 2: Distribution of topical azole agents use in patients on apixaban, rivaroxaban, and dabigatran.
      Page 7…Supplementary Table 3: Bleeding incidents during follow-up according to non-vitamin K oral anticoagulant agent
      Page 8…Supplementary Table 4: Systemic fluconazole exposure and the risk of bleeding among patients on apixaban, rivaroxaban, and dabigatran, according to the case-time-control design.
      Page 9…Supplementary Table 5: Topical azole exposure and the risk of bleeding among patients on apixaban, rivaroxaban and dabigatran, according to the case-time-control design.
      Page 10…Supplementary Table 6: Azoles applied to mucosal tissue and the risk of bleeding among patients on apixaban, rivaroxaban and dabigatran, according to the case-crossover design

      Description of Data Sources

      All Danish citizens have a personal identification number that enables cross-linkage between registers. The Danish National Patient Register holds information on hospitalizations, including diagnoses and procedural codes and is coded according to the International Classification of Diseases.1 The National Prescription Register holds information on date, amount, and strength of the drug of all claimed prescriptions coded per the Anatomical Therapeutic Chemical classification system.2 The Danish Civil Registration system provides data on date of birth and sex, as well as emigration and immigration history.3 Please see online-only appendix for International Classification of Diseases, Tenth Revision (ICD-10) and Anatomical Therapeutic Chemical (ATC) codes used throughout the study (Supplementary Table 1).

      References

      • 1.
        Schmidt M, Schmidt SSAJ, Sandegaard JLJ, et al. The Danish National Patient Registry: a review of content, data quality, and research potential. Clin Epidemiol. 2015;7:449-490.
      • 2.
        Kildemoes HW, Sorensen HT, Hallas J. The Danish National Prescription Registry. Scand J Public Health. 2011;39(7 suppl):38-41.
      • 3.
        Schmidt M, Pedersen L, Sørensen HT. The Danish Civil Registration System as a tool in epidemiology. Eur J Epidemiol. 2014;29(8):541-549.

      References

        • Steffel J
        • Verhamme P
        • Potpara TS
        • et al.
        The 2018 European Heart Rhythm Association Practical Guide on the use of non-vitamin K antagonist oral anticoagulants in patients with atrial fibrillation.
        Eur Heart J. 2018; 39: 1330-1393
      1. European Medicines Agency. European public assessment report, Eliquis. Available at: https://www.ema.europa.eu/en/medicines/human/EPAR/eliquis. Accessed June 22, 2021.

      2. European Medicines Agency. European public assessment report, Pradaxa. Available at: https://www.ema.europa.eu/en/medicines/human/EPAR/pradaxa. Accessed June 22, 2021.

      3. European Medicines Agency. European public assessment report, Xarelto. Available at: https://www.ema.europa.eu/en/medicines/human/EPAR/xarelto. Accessed June 22, 2021.

      4. European Medicines Agency. European public assessment report, Lixiana. Available at: https://www.ema.europa.eu/en/medicines/human/EPAR/lixiana. Accessed June 23, 2021.

        • Gelosa P
        • Castiglioni L
        • Tenconi M
        • et al.
        Pharmacokinetic drug interactions of the non-vitamin K antagonist oral anticoagulants (NOACs).
        Pharmacol Res. 2018; 135: 60-79
        • Fawzy AM
        • Lip GYH
        Pharmacokinetics and pharmacodynamics of oral anticoagulants used in atrial fibrillation.
        Expert Opin Drug Metab Toxicol. 2019; 15: 381-398
        • Adriaenssens N
        • Coenen S
        • Versporten A
        • Goossens H
        Outpatient systemic antimycotic and antifungal use in Europe: new outcome measure provides new insight.
        Int J Antimicrob Agents. 2013; 42: 466-470
        • Chang SH
        • Chou IJ
        • Yeh YH
        • et al.
        Association between use of non–vitamin K oral anticoagulants with and without concurrent medications and risk of major bleeding in nonvalvular atrial fibrillation.
        JAMA. 2017; 318: 1250-1259
        • Mueck W
        • Kubitza D
        • Becka M
        Co-administration of rivaroxaban with drugs that share its elimination pathways: pharmacokinetic effects in healthy subjects.
        Br J Clin Pharmacol. 2013; 76: 455-466
        • Fernandez S
        • Lenoir C
        • Samer C
        • Rollason V
        Drug interactions with apixaban: a systematic review of the literature and an analysis of VigiBase, the World Health Organization database of spontaneous safety reports.
        Pharmacol Res Perspect. 2020; 8: e00647
        • Lee K-G
        • Jennifer DL
        Increased bleeding tendency from interaction between rivaroxaban and topical miconazole: case report.
        Proc Singapore Healthc. 2015; 24: 121-122
        • Iversen DB
        • Hellfritzsch M
        • Stage TB
        • Aabenhus RM
        • Lind BS
        • Pottegård A
        Antimycotic treatment of oral candidiasis in warfarin users.
        Am J Med. 2021; 134: e308-e312
        • Miki A
        • Ohtani H
        • Sawada Y
        Warfarin and miconazole oral gel interactions: analysis and therapy recommendations based on clinical data and a pharmacokinetic model.
        J Clin Pharm Ther. 2011; 36: 642-650
        • Schmidt M
        • Pedersen L
        • Sørensen HT
        The Danish Civil Registration System as a tool in epidemiology.
        Eur J Epidemiol. 2014; 29: 541-549
        • Schmidt M
        • Schmidt SSAJ
        • Sandegaard JLJ
        • et al.
        The Danish National Patient Registry: a review of content, data quality, and research potential.
        Clin Epidemiol. 2015; 7: 449-490
        • Kildemoes HW
        • Sorensen HT
        • Hallas J
        The Danish National Prescription Registry.
        Scand J Public Health. 2011; 39: 38-41
        • Holt A
        • Blanche P
        • Zareini B
        • et al.
        Effect of long-term beta-blocker treatment following myocardial infarction among stable, optimally treated patients without heart failure in the reperfusion era: a Danish, nationwide cohort study.
        Eur Heart J. 2021; 42: 907-914
      5. Holt A, Blanche P, Zareini B, et al. Gastrointestinal bleeding risk following concomitant treatment with oral glucocorticoids in patients on non-vitamin K oral anticoagulants [online ahead of print]. Heart. 2021 Aug 13;heartjnl-2021-319503. doi:10.1136/heartjnl-2021-319503

        • Lip GYH
        • Skjøth F
        • Nielsen PB
        • Kjældgaard JN
        • Larsen TB
        The HAS-BLED, ATRIA, and ORBIT bleeding scores in atrial fibrillation patients using non-vitamin K antagonist oral anticoagulants.
        Am J Med. 2018; 131: 574.e13-574.e27
        • Olesen JB
        • Lip GYH
        • Hansen ML
        • et al.
        Validation of risk stratification schemes for predicting stroke and thromboembolism in patients with atrial fibrillation: nationwide cohort study.
        Br Med J. 2011; 342: d124
        • Scheike TH
        • Eriksson F
        • Tribler S
        The mean, variance and correlation for bivariate recurrent event data with a terminal event.
        J R Stat Soc Ser C Appl Stat. 2019; 68: 1029-1049
        • Maclure M
        The case-crossover design: a method for studying transient effects on the risk of acute events.
        Am J Epidemiol. 1991; 133: 144-153
        • Cadarette SM
        • Maclure M
        • Delaney JAC
        • et al.
        Control yourself: ISPE-endorsed guidance in the application of self-controlled study designs in pharmacoepidemiology.
        Pharmacoepidemiol Drug Saf. 2021; 30: 671-684
      6. R Core Team. R: a language and environment for statistical computing. Vienna, Austria; R Foundation for Statistical Computing. Available at: https://www.r-project.org/. Accessed May 20, 2020.

        • Suissa S
        The case-time-control design.
        Epidemiology. 1995; 6: 248-253
        • Nørgaard M
        • Ehrenstein V
        • Vandenbroucke JP
        Confounding in observational studies based on large health care databases: problems and potential solutions - a primer for the clinician.
        Clin Epidemiol. 2017; 9: 185-193
        • Agnelli G
        • Becattini C
        • Meyer G
        • et al.
        Apixaban for the treatment of venous thromboembolism associated with cancer.
        N Engl J Med. 2020; 382: 1599-1607
        • Verso M
        • Munoz A
        • Bauersachs R
        • et al.
        Effects of concomitant administration of anticancer agents and apixaban or dalteparin on recurrence and bleeding in patients with cancer-associated venous thromboembolism.
        Eur J Cancer. 2021; 148: 371-381
        • Hernán MA
        Causal analyses of existing databases: no power calculations required.
        J Clin Epidemiol. 2021 Aug 27; (S0895-4356(21)00273-0)https://doi.org/10.1016/j.jclinepi.2021.08.028
      7. Prescriber's digital reference. Available at: https://www.pdr.net/. Accessed June 7, 2021.

        • Zipursky J
        • Juurlink DN
        Studying drug safety in the real world.
        JAMA Intern Med. 2018; 178: 1533-1534