Abstract
Background
Apixaban is a direct oral anticoagulant for stroke prevention in patients with atrial fibrillation (AF) or atrial flutter (AFL), with dose reduction being recommended based on age, body weight, and creatinine.
Objectives
The objective of the study was to describe the use and appropriateness of reduced-dose apixaban.
Methods
Using electronic health record data from PCORnet health systems in the United States, patients with AF/AFL treated with apixaban between 2015 and 2019 were identified. Patients were categorized by dose and appropriateness of dose reduction. Exploratory outcomes included death, ischemic stroke, hemorrhagic stroke, intracranial hemorrhage, and major bleeding.
Results
Among 280,709 patients with AF/AFL, 45,947 (16.4%) received apixaban (84.6% full-dose; 15.4% reduced-dose). Among reduced-dose recipients, 61.0% did not meet the criteria for dose reduction. Over a median follow-up of 2.2 years, reduced-dose patients had higher unadjusted event rates than full-dose patients, including death, ischemic stroke, and major bleeding. Compared with those on appropriate full-dose apixaban as the reference group, inappropriate low-dose apixaban was associated with higher mortality (adjusted HR: 1.77; 95% CI: 1.66-1.88), whereas differences in ischemic stroke (adjusted HR: 1.05; 95% CI: 0.90-1.23), intracranial hemorrhage (adjusted HR: 0.95; 95% CI: 0.71-1.28), and major bleeding (adjusted HR: 1.02; 95% CI: 0.89-1.17) were attenuated.
Conclusions
In clinical practice, reduced-dose apixaban accounts for approximately 15% of prescriptions, with the majority (61%) prescribed in individuals who do not meet the labeling criteria for dose reduction.
Key Words: anticoagulation, apixaban, atrial fibrillation, stroke prevention
Central Illustration
Atrial fibrillation (AF) and atrial flutter (AFL) are the most common sustained cardiac arrhythmias worldwide, affecting an estimated 59 million individuals, with prevalence having tripled over the past 50 years.1,2 Approximately 20% of ischemic strokes are attributable to AF/AFL,3 prompting substantial efforts to optimize the use of oral anticoagulation for stroke prevention in this population.4,5 Despite strong guideline recommendations, a substantial proportion of patients at elevated stroke risk remain untreated with oral anticoagulation.6 Contributing factors include underestimation of ischemic stroke risk and heightened concern regarding bleeding complications.7
To mitigate bleeding risk, pivotal apixaban trials permitted dose reduction in patients meeting at least 2 of the following criteria: age ≥80 years, body weight ≤60 kg, or serum creatinine ≥1.5 mg/dL.8,9 These criteria were intended to identify individuals at risk of increased apixaban exposure due to age-related pharmacokinetic changes, low body weight, or impaired renal clearance. Pharmacokinetic studies have shown modest increases in apixaban exposure among older adults, individuals with low body weight, and those with renal impairment, characterized by higher area under the concentration-time curve and prolonged elimination half-life.10, 11, 12
In the AVERROES (Apixaban vs Acetylsalicylic Acid to Prevent Stroke in Atrial Fibrillation Patients Who Have Failed or Are Unsuitable for Vitamin K Antagonist Treated) trial, fewer than 7% of patients randomized to apixaban met the criteria for dose reduction, limiting inference regarding clinical outcomes with reduced-dose therapy.9 Similarly, in the ARISTOTLE (Apixaban for Reduction in Stroke and Other Thromboembolic Events in Atrial Fibrillation) trial, only 831 individuals (4%) received reduced-dose apixaban, with no evidence of effect modification by dose for stroke prevention or major bleeding, despite lower predicted drug concentrations.8,13 In contrast, in routine clinical practice, reduced-dose apixaban accounts for approximately 20% of apixaban prescriptions for AF, with observational data suggesting that more than half of these prescriptions may not meet the labeling criteria for dose reduction.14,15 Prior studies have suggested that both inappropriate and appropriate dose reductions of direct oral anticoagulants (DOACs) may be associated with higher rates of ischemic stroke.16,17 Randomized experience with alternative DOAC dosing strategies provides important context for interpreting observational comparisons of reduced-dose anticoagulation. In RE-LY, dabigatran 110 mg twice daily provided similar protection against stroke or systemic embolism compared with warfarin and was associated with lower major bleeding, whereas dabigatran 150 mg twice daily provided greater stroke prevention with similar major bleeding.18 In ENGAGE AF-TIMI 48,19 both higher- and lower-dose edoxaban regimens were noninferior to warfarin for stroke or systemic embolism, with dose-related reductions in bleeding and less ischemic stroke protection with the lower-dose regimen.
Accordingly, this study sought to characterize contemporary apixaban dosing patterns and the appropriateness of dose reduction among individuals with AF/AFL treated in large U.S. health systems.
Methods
Patient population and data sources
Using electronic health record (EHR) data from 7 large health systems participating in the Patient-Centered Clinical Research Network (PCORnet), patients with AF or AFL treated with apixaban were identified between January 1, 2015, and December 31, 2019. AF/AFL was defined by the presence of at least 1 inpatient encounter or 2 outpatient encounters within a 1-year period containing an AF/AFL diagnosis code.20 Apixaban use was ascertained from medication records occurring from 24 months before cohort entry through 6 months after cohort entry. Apixaban dosing was categorized as full-dose (5 mg twice daily) or reduced-dose (2.5 mg twice daily). Individuals with records indicating both dosing regimens within the exposure assessment window were categorized as receiving full-dose apixaban. Appropriateness of reduced-dose apixaban was defined according to labeling criteria as the presence of at least 2 of the following: age ≥80 years; body weight ≤60 kg; or serum creatinine ≥1.5 mg/dL.
This study was approved by the Duke University Health System Institutional Review Board (PRO00106775).
Outcomes
Outcomes evaluated included death, ischemic stroke, hemorrhagic stroke, intracranial hemorrhage, and major bleeding. Death was identified using EHR data and supplemented through linkage with Datavant to improve completeness and accuracy. All nonfatal clinical outcomes were identified using administrative diagnosis and procedure codes from final hospital discharge records (Supplemental Appendix).
Major bleeding was defined as 1 or more of the following: hospitalization with a diagnosis of intracranial hemorrhage; hospitalization with a diagnosis of gastrointestinal bleeding accompanied by a blood transfusion within 7 days of admission or death within 30 days of admission; or a procedure to control bleeding accompanied by a blood transfusion within 7 days of the procedure or death within 30 days of the procedure. Recurrent clinical events were considered distinct if they occurred during separate hospital admissions at least 30 days apart.
Baseline characteristics and clinical outcomes were summarized according to apixaban dosing regimen and appropriateness of reduced-dose apixaban use. For nonfatal outcomes, death was treated as a competing risk.
Statistical analysis
Baseline characteristics were summarized according to apixaban dosing regimen and appropriateness of reduced-dose apixaban use. Continuous variables are reported as median (IQR) and were compared using the Wilcoxon rank sum test or Kruskal-Wallis test, as appropriate. Categorical variables are reported as counts (percentages) and were compared using the chi-square test or Fisher exact test. Multivariable regression models were constructed to estimate adjusted HRs across dosing groups. Covariates included demographics, comorbidities, laboratory values, and concomitant medications. In a sensitivity analyses, we undertook an additional multivariable model that included age, body weight, and serum creatinine. As these variables define dose-reduction criteria and are directly related to treatment assignment, this approach allowed assessment of the robustness of findings while explicitly accounting for potential overadjustment. Continuous covariates were modeled using restricted cubic splines to account for potential nonlinearity. Proportional hazards assumptions were assessed using weighted Schoenfeld residuals from cause-specific analyses. No meaningful violations were identified for hemorrhagic stroke, intracranial hemorrhage, or major bleeding. Modest time-varying effects were observed for ischemic stroke and mortality; however, the directionality of HRs remained consistent throughout follow-up. Data on adjustment covariates were complete except for serum creatinine, hemoglobin, and weight, which were complete in >94% of patients. Missing values were filled in with single imputation: missing creatinine was imputed using sex, age group, and renal disease status stratified medians; missing hemoglobin was imputed using sex and age group stratified medians; missing weight was imputed using sex stratified medians.
Follow-up for mortality started on the baseline date and was censored at 5 years. For nonfatal outcomes, cumulative incidence functions were used, treating death as a competing risk. Recurrent nonfatal events were summarized using cumulative mean functions, which estimate the average number of events per 100 individuals as a function of time from cohort entry.
Average event rates were calculated as the number of events per 100 person-years and are reported with 95% CIs. Event rates were estimated using Poisson regression models with generalized estimating equations and robust variance estimates to account for overdispersion and within-individual correlation.
SAS (version 9.4; SAS Institute) was used for data summaries and analysis.
Results
Patient population
Of the 280,709 patients with AF/AFL, 45,947 were treated with apixaban and had documented dosing. Of these, 38,861 (84.6%) received full-dose apixaban and 7,086 (15.4%) received reduced-dose apixaban.
Patients treated with reduced-dose apixaban were older, more frequently females, had lower body mass index, worse renal function, and a higher burden of comorbid conditions than those treated with full-dose apixaban (Table 1). Among patients receiving reduced-dose apixaban, 4,321 (61.0%) did not meet the labeling criteria for dose reduction. Those meeting the criteria for appropriate dose reduction were older, more frequently females, had lower body mass index, worse renal function, and a higher comorbidity burden than those receiving reduced-dose apixaban without meeting dose-reduction criteria (Table 1). Among patients receiving reduced-dose apixaban, 19.0% met no dose-reduction criteria, 41.9% met 1 criterion, 36.6% met 22 criteria, and 2.4% met all 3 criteria (Table 2). Among patients receiving inappropriate reduced-dose apixaban, median values were 9 years below the age threshold, 30.7 kg above the weight threshold, and 0.50 mg/dL below the creatinine threshold (Table 2).
Table 1.
Baseline Characteristics Stratified by Apixaban Dosing Regimen
| Appropriate Full-Dose Apixaban (n = 37,277) | Inappropriate Full-Dose Apixaban (n = 1,584) | P Valuea | Appropriate Reduced-Dose Apixaban (n = 2,765) | Inappropriate Reduced-Dose Apixaban (n = 4,321) | P Valueb | |
|---|---|---|---|---|---|---|
| AF/AFL history | ||||||
| Incident diagnosis | 32,451 (87.1%) | 1,334 (84.2%) | 0.001 | 2,342 (84.7%) | 3,543 (82.0%) | <0.001 |
| Chronic | 5,643/35,035 (16.1%) | 222/1,496 (14.8%) | <0.001 | 531/2,637 (20.1%) | 686/4,137 (16.6%) | <0.001 |
| Paroxysmal | 13,083/35,035 (37.3%) | 494/1,496 (33.0%) | 887/2,637 (33.6%) | 1,377/4,137 (33.3%) | ||
| Other | 16,309/35,035 (46.6%) | 780/1,496 (52.1%) | 1,219/2,637 (46.2%) | 2074/4,137 (50.1%) | ||
| AF codes | 35,035 (94.0%) | 1,496 (94.4%) | 0.45 | 2,637 (95.4%) | 4,137 (95.7%) | <0.001 |
| AFL codes | 8,201 (22.0%) | 333 (21.0%) | 0.36 | 474 (17.1%) | 771 (17.8%) | <0.001 |
| Demographics | ||||||
| Age (years) | 71.0 (64.0, 77.0) | 84.0 (81.0, 87.0) | <0.001 | 86.0 (83.0, 89.0) | 79.0 (71.0, 85.0) | <0.001 |
| Female | 14,834 (39.8%) | 896 (56.6%) | <0.001 | 1,878 (67.9%) | 2,212 (51.2%) | <0.001 |
| Race | 0.56 | 0.147 | ||||
| White | 34,105/36,880 (92.5%) | 1,461/1,568 (93.2%) | 2,542/2,740 (92.8%) | 3,974/4,292 (92.6%) | ||
| Black | 2008/36,880 (5.4%) | 79/1,568 (5.0%) | 127/2,740 (4.6%) | 223/4,292 (5.2%) | ||
| Other | 767/36,880 (2.1%) | 28/1,568 (1.8%) | 71/2,740 (2.6%) | 95/4,292 (2.2%) | ||
| Hispanic | 517/36,639 (1.4%) | 23/1,560 (1.5%) | 0.83 | 39/2,718 (1.4%) | 78/4,249 (1.8%) | 0.093 |
| BMI | 29.9 (26.3, 34.8) [36,960] | 24.2 (21.2, 29.1) [1,527] | <0.001 | 23.1 (20.6, 27.0) [2,712] | 28.0 (24.7, 32.5) [4,279] | <0.001 |
| CHA2DS2-VASc score | 3 (2, 4) | 5 (4, 6) | <0.001 | 5 (4, 6) | 4 (3, 5) | <0.001 |
| HAS-BLED score | 2 (2, 3) | 3 (2, 4) | <0.001 | 3 (2, 4) | 3 (2, 4) | <0.001 |
| Comorbidities | ||||||
| Heart failure | 10,376 (27.8%) | 704 (44.4%) | <0.001 | 1,198 (43.3%) | 1,505 (34.8%) | <0.001 |
| Hypertension | 27,600 (74.0%) | 1,348 (85.1%) | <0.001 | 2,360 (85.4%) | 3,494 (80.9%) | <0.001 |
| Ischemic stroke | 3,068 (8.2%) | 179 (11.3%) | <0.001 | 341 (12.3%) | 381 (8.8%) | <0.001 |
| Hemorrhagic stroke | 281 (0.8%) | 14 (0.9%) | 0.56 | 25 (0.9%) | 31 (0.7%) | 0.653 |
| Myocardial infarction | 4,663 (12.5%) | 278 (17.6%) | <0.001 | 461 (16.7%) | 610 (14.1%) | <0.001 |
| Diabetes | 10,045 (26.9%) | 412 (26.0%) | 0.41 | 700 (25.3%) | 1,297 (30.0%) | <0.001 |
| Prior major bleeding | 517 (1.4%) | 33 (2.1%) | 0.02 | 57 (2.1%) | 103 (2.4%) | <0.001 |
| GI bleeding | 762 (2.0%) | 74 (4.7%) | <0.001 | 108 (3.9%) | 170 (3.9%) | <0.001 |
| Dementia/Alzheimer | 991 (2.7%) | 173 (10.9%) | <0.001 | 300 (10.8%) | 267 (6.2%) | <0.001 |
| Cancer | 5,889 (15.8%) | 298 (18.8%) | 0.001 | 477 (17.3%) | 749 (17.3%) | 0.015 |
| Rheumatic mitral stenosis | 242 (0.6%) | 14 (0.9%) | 0.26 | 33 (1.2%) | 42 (1.0%) | <0.001 |
| Laboratory measurements | ||||||
| Serum creatinine (mg/dL) | 1.0 (0.8, 1.2) [34,858] | 1.5 (1.0, 1.8) [1,549] | <0.001 | 1.5 (0.9, 1.8) [2,664] | 1.1 (0.9, 1.4) [4,036] | <0.001 |
| Hemoglobin, g/dL | 13.7 (12.3, 14.9) [32,679] | 12.5 (11.2, 13.8) [1,491] | <0.001 | 12.3 (11.0, 13.5) [2,460] | 12.8 (11.4, 14.1) [3,719] | <0.001 |
| Medications | ||||||
| Aspirin | 18,664 (50.1%) | 860 (54.3%) | <0.001 | 1,375 (49.7%) | 2,149 (49.7%) | 0.733 |
| P2Y12 inhibitor (clopidogrel, ticagrelor, prasugrel) | 4,179 (11.2%) | 216 (13.6%) | 0.003 | 360 (13.0%) | 543 (12.6%) | 0.002 |
| NSAIDs | 10,032 (26.9%) | 336 (21.2%) | <0.001 | 436 (15.8%) | 925 (21.4%) | <0.001 |
| Antiarrhythmic drugs | 14,065 (37.7%) | 456 (28.8%) | <0.001 | 707 (25.6%) | 1,185 (27.4%) | <0.001 |
| PPI | 15,824 (42.4%) | 706 (44.6%) | 0.09 | 1,124 (40.7%) | 1938 (44.9%) | 0.001 |
AF = atrial fibrillation; AFL = atrial flutter; BMI = body mass index; GI = gastrointestinal; NSAIDs = nonsteroidal anti-inflammatory drugs; PPI = proton pump inhibitor.
Testing differences between appropriate and inappropriate full-dose.
Testing differences between appropriate and inappropriate reduced-dose apixaban.
Table 2.
Proximity to Apixaban Dose-Reduction Criteria Among Patients Receiving Reduced-Dose Therapy
| Inappropriate Reduced-Dose Apixaban 2.5 mga |
Appropriate Reduced-Dose Apixaban 2.5 mga |
|||
|---|---|---|---|---|
| 0 Criteria Metb | 1 Criterion Metb | 2 Criteria Metb | 3 Criteria Metb | |
| Number of patients with recorded apixaban dosing | 1,349 | 2,972 | 2,596 | 169 |
| Age ≥80 years | ||||
| Yes | 0 (0.0%) | 2013 (67.7%) | 2,521 (97.1%) | 169 (100.0%) |
| No | 1,349 (100.0%) | 959 (32.3%) | 75 (2.9%) | 0 (0.0%) |
| Age deviation: years below 80 | 9.0 (5.0, 16.0) [1,349] | 7.0 (3.0, 14.0) [959] | 9.0 (3.0, 16.0) [75] | - |
| Weight ≤60 kg | ||||
| Yes | 0 (0.0%) | 260 (8.7%) | 1,494 (57.6%) | 169 (100.0%) |
| No | 1,349 (100.0%) | 2,712 (91.3%) | 1,102 (42.4%) | - |
| Weight deviation: kg above 60 | 30.7 (17.2, 47.9) [1,349] | 17.6 (7.3, 30.7) [2,712] | 21.6 (12.3, 33.4) [1,102] | - |
| Creatinine ≥1.5 mg/dL | ||||
| Yes | 0 (0.0%) | 699 (23.5%) | 1,177 (45.3%) | 169 (100.0%) |
| No | 1,349 (100.0%) | 2,273 (76.5%) | 1,419 (54.7%) | - |
| Creatinine deviation: mg/dL below 1.5 | 0.50 (0.30, 0.69) [1,263] | 0.44 (0.28, 0.64) [2,074] | 0.60 (0.40, 0.70) [1,318] | - |
Values are n (%) or median (25th percentile, 75th percentile) [N].
Apixaban dose based on having at least one code with apixaban 2.5 or 5 mg within 24 months prior to or 6 months after eligibility for AF/AFL cohort. If both 2.5 and 5.0 mg were recorded, the subject was classified in the 5 mg category.
Criteria include: age ≥80 years, weight ≤60 kg, baseline creatinine ≥1.5 mg/mL.
Among patients receiving full-dose apixaban, 1,584 (4.1%) met ≥2 labeling criteria for dose reduction and would have been eligible for reduced-dose therapy.
Clinical outcomes
Over a median follow-up of 2.2 years (IQR: 1.3-3.7), unadjusted event rates were higher among patients receiving reduced-dose compared with full-dose apixaban across all outcomes, including death (15.4 vs 6.3 per 100 person-years), ischemic stroke (2.3 vs 1.6 per 100 person-years), and major bleeding (3.1 vs 2.2 per 100 person-years) (Table 2).
Among patients receiving reduced-dose apixaban, those meeting the labeling criteria had higher event rates than those not meeting criteria, including death (21.6 vs 12.2 per 100 person-years), ischemic stroke (2.9 vs 2.0 per 100 person-years), and major bleeding (3.8 vs 2.7 per 100 person-years) (Table 3).
Table 3.
Clinical Outcomes Stratified by Apixaban Dosing Regimen
| Outcome | Group | 5-Year CIF (%) | Event Rate/100 PY | Adjusted HR | P Value |
|---|---|---|---|---|---|
| Death | Appropriate full-dose apixaban | 25.1 | 5.9 | Ref | — |
| Inappropriate full-dose apixaban | 62.6 | 19.4 | 2.22 (2.04-2.41) | <0.0001 | |
| Appropriate low-dose apixaban | 65.4 | 21.6 | 2.56 (2.40-2.73) | <0.0001 | |
| Inappropriate low-dose apixaban | 43.7 | 12.2 | 1.77 (1.66-1.88) | <0.0001 | |
| Ischemic stroke | Appropriate full-dose apixaban | 4.7 | 1.6 | Ref | — |
| Inappropriate full-dose apixaban | 7.4 | 2.9 | 1.26 (1.01-1.58) | 0.043 | |
| Appropriate low-dose apixaban | 6.7 | 2.9 | 1.10 (0.92-1.33) | 0.298 | |
| Inappropriate low-dose apixaban | 5.5 | 2.0 | 1.05 (0.90-1.23) | 0.542 | |
| Hemorrhagic stroke | Appropriate full-dose apixaban | 1.2 | 0.3 | Ref | — |
| Inappropriate full-dose apixaban | 1.2 | 0.4 | 0.92 (0.52-1.62) | 0.762 | |
| Appropriate low-dose apixaban | 1.6 | 0.6 | 1.40 (0.97-2.02) | 0.076 | |
| Inappropriate low-dose apixaban | 0.7 | 0.2 | 0.66 (0.44-1.00) | 0.050 | |
| Intracranial hemorrhage | Appropriate full-dose apixaban | 1.7 | 0.5 | Ref | — |
| Inappropriate full-dose apixaban | 1.6 | 0.6 | 0.95 (0.59-1.53) | 0.843 | |
| Appropriate low-dose apixaban | 2.3 | 0.9 | 1.32 (0.96-1.83) | 0.087 | |
| Inappropriate low-dose apixaban | 1.5 | 0.6 | 0.95 (0.71-1.28) | 0.738 | |
| Major bleeding | Appropriate full-dose apixaban | 6.6 | 2.1 | Ref | — |
| Inappropriate full-dose apixaban | 9.5 | 4.1 | 1.20 (0.98-1.46) | 0.072 | |
| Appropriate low-dose apixaban | 8.9 | 3.8 | 1.14 (0.97-1.33) | 0.111 | |
| Inappropriate low-dose apixaban | 7.5 | 2.7 | 1.02 (0.89-1.17) | 0.794 |
Model adjusted for sex, calendar year of study, heart failure, hypertension, diabetes, prior stroke, prior major bleeding or GI bleeding, dementia/Alzheimer’s, cancer, hemoglobin, aspirin use, P2Y12 inhibitor use in addition to apixaban dosing. Death was modeled with cox proportional hazard model. All other outcomes were modeled with Fine-Grey subdistribution hazard model.
CIF% = cumulative incidence function percentage; PY = person-years.
In the adjusted model, differences between dosing groups were attenuated, particularly for nonfatal outcomes, whereas higher mortality persisted among reduced-dose groups. Compared with the appropriate full-dose apixaban population, the adjusted HRs for death were 2.22 (95% CI: 2.04-2.41) for inappropriate full-dose, 2.56 (2.40-2.73) for appropriate reduced-dose, and 1.77 (1.66-1.88) for inappropriate reduced-dose apixaban (Table 3). For ischemic stroke, adjusted HRs were 1.26 (1.01-1.58), 1.10 (0.92-1.33), and 1.05 (0.90-1.23), respectively, whereas for major bleeding, adjusted HRs were 1.20 (0.98-1.46), 1.14 (0.97-1.33), and 1.02 (0.89-1.17). In a sensitivity analysis, incorporating age, body weight, and serum creatinine into the model further attenuated the results toward the null (Supplemental Table 1).
Discussion
In this large, contemporary cohort of patients with AF/AFL treated with apixaban across multiple U.S. health systems, reduced-dose apixaban accounted for approximately 15% of prescriptions, of which 61% did not meet the labeling criteria for dose reduction (Central Illustration). Notably, 19% of patients receiving reduced-dose apixaban did not meet any dose-reduction criteria. Patients receiving reduced-dose apixaban represented clinically distinct, higher-risk populations, with substantially higher unadjusted rates of mortality, ischemic stroke, and bleeding. These findings highlight marked heterogeneity in real-world dosing practices and uncertainty regarding the clinical implications of off-label dose reduction.
Central Illustration.
Clinical Use and Appropriateness of Reduced-Dose Apixaban in AF/AFL
Among 280,709 patients with atrial fibrillation (AF) or atrial flutter (AFL), 45,947 (16.4%) received apixaban. Of these, 38,861 (84.6%) were treated with full-dose (5 mg twice daily) and 7,086 (15.4%) with reduced-dose (2.5 mg twice daily). Among reduced-dose recipients, 2,765 (39.0%) met the dose-reduction criteria, whereas 4,321 (61.0%) did not. Dose-reduction criteria included ≥2 of: age ≥80 years, body weight ≤60 kg, or serum creatinine ≥1.5 mg/dL. Reduced-dose apixaban was associated with higher unadjusted event rates vs full-dose, including death (15.4 vs 6.3 per 100 person-years), ischemic stroke (2.3 vs 1.6), and major bleeding (3.1 vs 2.2). After adjustment, differences in nonfatal outcomes were attenuated, while higher mortality persisted. These findings highlight frequent off-label reduced dosing and substantial heterogeneity in real-world anticoagulation practices. BID = twice a day.
Reduced-dose apixaban was infrequently used in the pivotal randomized trials that established its efficacy and safety for stroke prevention in AF. In AVERROES, fewer than 7% of individuals randomized to apixaban met the criteria for dose reduction, limiting the ability to draw conclusions regarding outcomes with reduced-dose therapy.9 Importantly, despite enrollment of patients considered unsuitable for warfarin, event rates in AVERROES were relatively modest, underscoring the differences between trial populations and the higher-risk patients who often receive reduced-dose apixaban in clinical practice. Similarly, in ARISTOTLE, only 831 participants (4%) received reduced-dose apixaban or warfarin, and no evidence of effect modification by dose was observed for either efficacy or major bleeding, despite lower achieved apixaban concentrations in this subgroup.8,13 These trials therefore provide limited insight into reduced-dose apixaban in broader, higher-risk populations encountered in routine practice. In ARISTOTLE, 22.8% of patients receiving full-dose apixaban met 1 dose-reduction criterion and had numerically higher risks of ischemic stroke and bleeding, yet retained benefit compared with warfarin.8,21 These findings highlight the challenges of extrapolating trial-based dosing strategies to the more complex, higher-risk populations encountered in routine clinical practice.
Observational studies have shown that reduced-dose DOACs are commonly prescribed and often discordant with labeling recommendations, with inappropriate dosing reported in 12% to 18% of patients and up to half of reduced-dose prescriptions in some cohorts.15,16 Claims-based analyses have similarly reported associations between underdosing and higher stroke risk in selected populations.22 Randomized dose-comparison trials provide important context. In RE-LY18 and ENGAGE AF-TIMI 48,19 where lower-dose anticoagulation is associated with modestly reduced thromboembolic protection but lower bleeding. A similar pattern was observed in ELDER-CARE AF,23 where very low-dose edoxaban reduced stroke risk compared with placebo with an acceptable increase in bleeding in patients not suitable for warfarin therapy. In contrast, patients receiving reduced-dose apixaban in our cohort had higher crude rates of both ischemic and bleeding outcomes. This pattern is unlikely to reflect a causal effect of lower-dose apixaban and instead likely reflects confounding by indication, frailty, and clinician selection of therapy, supported by attenuation after multivariable adjustment.
The high prevalence of reduced-dose apixaban use outside of labeling criteria likely reflects the complexity of anticoagulation decision-making in contemporary practice. Clinicians frequently care for patients with competing risks, including frailty and prior bleeding, that are not fully captured in structured data. Consistent with this, most patients receiving reduced-dose apixaban without meeting criteria were not near dose-reduction thresholds, suggesting that off-label dosing reflects broader clinical judgment rather than borderline values alone.
Study Limitations
Several limitations should be considered when interpreting these findings. First, as an observational analysis using EHR data, this study is subject to unambiguous confounding by indication as well as residual confounding from clinical factors not reliably captured in structured data, including frailty, functional status, fall risk, clinician judgment, and patient preferences. Second, outcomes were identified using administrative diagnosis and procedure codes rather than adjudication, which may introduce misclassification. However, these approaches are commonly used in large-scale observational studies and the use of an active comparison group design mitigates this concern. Third, apixaban exposure was based on medication records and did not account for adherence, dose changes over time, or temporary interruptions, and a proportion of individuals had missing dosing information and were excluded from the analysis. Finally, although the study included multiple large U.S. health systems, findings may not be generalizable to settings with different patient populations or prescribing practices.
Reduced-dose apixaban is frequently prescribed outside labeling criteria (61% of cases), with these patients representing a clinically distinct, higher-risk population. Substantial heterogeneity in real-world dosing underscores the need for improved evidence to guide anticoagulant use in populations underrepresented in randomized trials. Ongoing evaluation of factor XI inhibitors may help address this gap by offering anticoagulation strategies with a more favorable balance between thromboembolic protection and bleeding risk.24
Perspectives.
COMPETENCY IN MEDICAL KNOWLEDGE: In patients with AF/AFL, apixaban dose reduction is recommended only when at least 2 labeling criteria are present: age ≥80 years, body weight ≤60 kg, or serum creatinine ≥1.5 mg/dL. In this contemporary multicenter cohort, reduced-dose apixaban was prescribed in approximately 15% of treated patients, and most reduced-dose recipients did not meet labeling criteria for dose reduction.
TRANSLATIONAL OUTLOOK: Future studies should determine whether electronic health record–based decision support, pharmacist-led anticoagulation review, or other implementation strategies can reduce inappropriate apixaban dose reduction and improve patient-centered outcomes.
Funding support and author disclosures
The PCORnet Study reported in this publication was powered by PCORnet, which has been developed with funding from the Patient- Centered Outcomes Research Institute (PCORI). The views presented in this publication are solely the responsibility of the author(s) and do not necessarily represent the views of other organizations participating in, collaborating with, or funding PCORnet or of PCORI. Dr Alexander has received research grants through Duke University from Artivion/CryoLife, Bayer, Bristol-Myers Squibb, CSL Behring, Ferring, the U.S. FDA, and the U.S. NIH; and has received advisory board, honoraria, or consulting payments from AtriCure, Bristol-Myers Squibb, Curis, Eli Lilly, Humacyte, Novostia, Theravance, and Veralox. Dr Coppolecchia is employed by Bayer US LLC. Dr Jones has received research grants to DCRI that support salary: Bayer, Boehringer Ingelheim, Merck, Novartis, National Institutes of Health, Patient-Centered Outcomes Research Institute; and has served consulting/advisory roles with American College of Physicians, Amplitude Vascular Solutions, Merck, U.S. Department of Justice. Dr Steinberg has received salary support from the NIH/NHLBI (#R21HL172288, #1R01HL177105); research support from Abbott, Boston Scientific, Biosense-Webster, Sanofi, and PaceMate; and consulting to Sanofi, Bayer, Boston Scientific, Element Science, Milestone, and AltaThera. Dr Patel reports research grants from Bayer, NHLBI, Idorsia, Regeneron; and consulting with Regeneron. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
Footnotes
The authors attest they are in compliance with human studies committees and animal welfare regulations of the authors’ institutions and Food and Drug Administration guidelines, including patient consent where appropriate. For more information, visit the Author Center.
Appendix
For a supplemental table, please see the online version of this paper.
Appendix
References
- 1.Linz D., Gawalko M., Betz K., et al. Atrial fibrillation: epidemiology, screening and digital health. Lancet Reg Health Eur. 2024;37 doi: 10.1016/j.lanepe.2023.100786. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Schnabel R.B., Yin X., Gona P., et al. 50 year trends in atrial fibrillation prevalence, incidence, risk factors, and mortality in the Framingham Heart Study: a cohort study. Lancet. 2015;386(9989):154–162. doi: 10.1016/S0140-6736(14)61774-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Marini C., De Santis F., Sacco S., et al. Contribution of atrial fibrillation to incidence and outcome of ischemic stroke. Stroke. 2005;36(6):1115–1119. doi: 10.1161/01.STR.0000166053.83476.4a. [DOI] [PubMed] [Google Scholar]
- 4.Joglar J.A., Chung M.K., Armbruster A.L., et al. 2023 ACC/AHA/ACCP/HRS guideline for the diagnosis and management of atrial fibrillation: a report of the American College of Cardiology/American Heart Association Joint Committee on Clinical Practice guidelines. Circulation. 2024;149(1):e1–e156. doi: 10.1161/CIR.0000000000001193. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Van Gelder I.C., Rienstra M., Bunting K.V., et al. 2024 ESC guidelines for the management of atrial fibrillation developed in collaboration with the European Association for Cardio-Thoracic Surgery (EACTS): developed by the task force for the management of atrial fibrillation of the European Society of Cardiology (ESC), with the special contribution of the European Heart Rhythm Association (EHRA) of the ESC. Endorsed by the European Stroke Organisation (ESO) Eur Heart J. 2024;45(36):3314–3414. doi: 10.1093/eurheartj/ehae176. [DOI] [PubMed] [Google Scholar]
- 6.Navar A.M., Kolkailah A.A., Overton R., et al. Trends in oral anticoagulant use among 436 864 patients with Atrial Fibrillation in Community Practice, 2011 to 2020. J Am Heart Assoc. 2022;11(22) doi: 10.1161/JAHA.122.026723. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Cannon C.P., Kim J.M., Lee J.J., et al. Patients and their physician’s perspectives about oral anticoagulation in patients with atrial fibrillation not receiving an anticoagulant. JAMA Network Open. 2023;6(4):e239638. doi: 10.1001/jamanetworkopen.2023.9638. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Granger C.B., Alexander J.H., McMurray J.J.V., et al. Apixaban versus Warfarin in Patients with Atrial Fibrillation. N Engl J Med. 2011;365(11):981–992. doi: 10.1056/NEJMoa1107039. [DOI] [PubMed] [Google Scholar]
- 9.Connolly S.J., Eikelboom J., Joyner C., et al. Apixaban in Patients with Atrial Fibrillation. N Engl J Med. 2011;364(9):806–817. doi: 10.1056/NEJMoa1007432. [DOI] [PubMed] [Google Scholar]
- 10.Frost C.E., Song Y., Shenker A., et al. Effects of age and sex on the single-dose pharmacokinetics and pharmacodynamics of apixaban. Clin Pharmacokinet. 2015;54(6):651–662. doi: 10.1007/s40262-014-0228-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Upreti V.V., Wang J., Barrett Y.C., et al. Effect of extremes of body weight on the pharmacokinetics, pharmacodynamics, safety and tolerability of apixaban in healthy subjects. Br J Clin Pharmacol. 2013;76(6):908–916. doi: 10.1111/bcp.12114. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Chang M., Yu Z., Shenker A., et al. Effect of renal impairment on the pharmacokinetics, pharmacodynamics, and safety of apixaban. J Clin Pharmacol. 2016;56(5):637–645. doi: 10.1002/jcph.633. [DOI] [PubMed] [Google Scholar]
- 13.Zeitouni M., Giczewska A., Lopes Renato D., et al. Clinical and pharmacological effects of apixaban dose adjustment in the ARISTOTLE trial. JACC. 2020;75(10):1145–1155. doi: 10.1016/j.jacc.2019.12.060. [DOI] [PubMed] [Google Scholar]
- 14.Li X., Keshishian A., Hamilton M., et al. Apixaban 5 and 2.5 mg twice-daily versus warfarin for stroke prevention in nonvalvular atrial fibrillation patients: comparative effectiveness and safety evaluated using a propensity-score-matched approach. PLoS One. 2018;13(1) doi: 10.1371/journal.pone.0191722. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Guenoun M., Cohen S., Villaceque M., et al. Characteristics of patients with atrial fibrillation treated with direct oral anticoagulants and new insights into inappropriate dosing: results from the French National Prospective Registry: PAFF. Europace. 2023;25(10) doi: 10.1093/europace/euad302. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Steinberg Benjamin A., Shrader P., Thomas L., et al. Off-label dosing of non-vitamin K antagonist oral anticoagulants and adverse outcomes. JACC. 2016;68(24):2597–2604. doi: 10.1016/j.jacc.2016.09.966. [DOI] [PubMed] [Google Scholar]
- 17.Steffel J., Ruff Christian T., Yin O., et al. Randomized, double-blind comparison of half-dose versus full-dose edoxaban in 14,014 patients with atrial fibrillation. JACC. 2021;77(9):1197–1207. doi: 10.1016/j.jacc.2020.12.053. [DOI] [PubMed] [Google Scholar]
- 18.Connolly S.J., Ezekowitz M.D., Yusuf S., et al. Dabigatran versus warfarin in patients with atrial fibrillation. N Engl J Med. 2009;361(12):1139–1151. doi: 10.1056/NEJMoa0905561. [DOI] [PubMed] [Google Scholar]
- 19.Giugliano R.P., Ruff C.T., Braunwald E., et al. Edoxaban versus warfarin in patients with atrial fibrillation. N Engl J Med. 2013;369(22):2093–2104. doi: 10.1056/NEJMoa1310907. [DOI] [PubMed] [Google Scholar]
- 20.Piccini J.P., Hammill B.G., Sinner M.F., et al. Incidence and prevalence of atrial fibrillation and associated mortality among medicare beneficiaries: 1993–2007. Circ Cardiovasc Qual Outcomes. 2012;5(1):85–93. doi: 10.1161/CIRCOUTCOMES.111.962688. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Alexander J.H., Andersson U., Lopes R.D., et al. Apixaban 5 mg twice daily and clinical outcomes in patients with atrial fibrillation and advanced age, low body weight, or high creatinine: a secondary analysis of a randomized clinical trial. JAMA Cardiol. 2016;1(6):673–681. doi: 10.1001/jamacardio.2016.1829. [DOI] [PubMed] [Google Scholar]
- 22.Yao X., Shah Nilay D., Sangaralingham Lindsey R., Gersh Bernard J., Noseworthy Peter A. Non–vitamin K antagonist oral anticoagulant dosing in patients with atrial fibrillation and renal dysfunction. JACC. 2017;69(23):2779–2790. doi: 10.1016/j.jacc.2017.03.600. [DOI] [PubMed] [Google Scholar]
- 23.Okumura K., Akao M., Yoshida T., et al. Low-dose edoxaban in very elderly patients with atrial fibrillation. N Engl J Med. 2020;383(18):1735–1745. doi: 10.1056/NEJMoa2012883. [DOI] [PubMed] [Google Scholar]
- 24.Harrington J., Piccini J.P., Alexander J.H., Granger C.B., Patel M.R. Clinical evaluation of factor XIa inhibitor drugs: JACC review topic of the week. J Am Coll Cardiol. 2023;81(8):771–779. doi: 10.1016/j.jacc.2022.11.057. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.