Abstract
Background
Off-label underdosing of direct oral anticoagulants (DOACs) is common among Asian patients with atrial fibrillation (AF), partly reflecting concerns about bleeding, yet Southeast Asian data remain limited. This study aimed to characterize DOAC dosing patterns, identify predictors of off-label underdosing, and evaluate associated effectiveness and safety outcomes.
Methods
We conducted a retrospective cohort using electronic health records (EHRs) from two tertiary hospitals in Thailand (2015–2023). Incident AF patients initiating dabigatran, rivaroxaban, apixaban, or edoxaban were included. Doses were classified as on-label, underdosed, or overdosed based on guideline criteria. Propensity score-based inverse probability of treatment weighting (PS-IPTW) Cox proportional hazards regression models was used to evaluate ischemic stroke or systemic embolism (ISSE) and bleeding outcomes.
Results
Among 553 patients, 20.4% were underdosed, 6.3% overdosed, and 73.2% received on-label dosing. Rivaroxaban was most frequently underdosed. Older age and diabetes mellitus independently predicted underdosing. Compared to on-label dosing, underdosing was not associated with ISSE (adjusted hazard ratio [aHR] 0.48, 95% CI 0.11–2.16; p = 0.336) or bleeding (aHR 1.06, 95% CI 0.30–3.73; p = 0.924).
Conclusions
Off-label underdosing occurred in one-fifth of patients but was not associated with significant differences in ISSE or bleeding. Larger, prospective studies in broader Asian and non-Asian populations are warranted.
Keywords: Direct oral anticoagulants, atrial fibrillation, stroke, bleeding, underdosing, off-label
PLAIN LANGUAGE SUMMARY
This study was conducted because doctors sometimes prescribe lower-than-recommended doses of direct oral anticoagulants (DOACs) for people living with atrial fibrillation (AF), especially in Asian populations, due to concerns about bleeding. However, it is unclear how often this happens in real-world practice in Thailand, what factors influence these decisions, and whether using lower doses affects patient outcomes such as stroke or bleeding.
The researchers used electronic health records (EHRs) from two hospitals in Thailand to follow 553 people newly diagnosed with AF who started DOAC treatment between 2015 and 2023. They found that most people (about 73%) received recommended doses, but around 20% received lower-than-recommended doses, most commonly rivaroxaban. Older age and having diabetes were linked to a higher likelihood of receiving a lower dose. Importantly, when comparing people who received lower doses with those who received recommended doses, there were no meaningful differences in the risks of stroke or bleeding.
These results suggest that although prescribing lower doses is relatively common, it may not provide clear benefits or harms in terms of major clinical outcomes. This highlights the need for careful, individualized decision-making rather than routine dose reduction based on concern alone. This study adds important real-world evidence from Southeast Asia, where data are still limited. It helps clinicians, researchers, and policymakers better understand prescribing patterns and supports the need for larger studies to guide safe and effective DOAC use across diverse populations.
ARTICLE HIGHLIGHTS
Older age and diabetes independently predicted off-label underdosing, with rivaroxaban being the most frequently underdosed agent.
While 73% received on-label DOACs, a significant 20.4% were prescribed off-label underdoses, highlighting a persistent clinical trend in Asian patients.
Off-label underdosing showed no statistically significant difference in risks for ischemic stroke, systemic embolism, or bleeding compared to on-label regimens.
These findings highlight the need for larger, multicenter real-world studies to confirm clinical outcomes and inform individualized DOAC dosing strategies.
Graphical Abstract
1. Introduction
Atrial fibrillation (AF) is a common type of cardiac arrhythmia, particularly prevalent in older people [1,2]. It has been reported that over 59 million individuals worldwide in 2019 affected by AF. It is also participated that the number of AF cases could rise to 15.9 million by 2050 in the US, increase to 17.9 million in Europe by 2060, and reach 72 million by 2050 in Asia. AF is also linked to several serious complications, most notably a fivefold increased risk of stroke [2]. Current clinical practice guidelines recommend the use of oral anticoagulants (OACs) for the prevention of ischemic stroke or thromboembolism. There are two main classes of OACs commonly used in clinical practice: vitamin K antagonists, such as warfarin, and direct oral anticoagulants (DOACs), including dabigatran, rivaroxaban, apixaban, and edoxaban [1,2]. Randomized controlled trials demonstrated that all DOACs, when used at standard doses, were non-inferior to warfarin in reducing the risk of stroke or systemic embolism, and were associated with a lower risk or similar risk of bleeding compared to warfarin [3–6].
Nevertheless, there has been a growing trend in the use of off-label underdosed DOACs, the use of a lower-than-recommended dose without meeting guideline-specified dose reduction criteria, in current clinical practice. This may be attributed to concerns about bleeding complications, particularly among older people, those with impaired renal function, or individuals with low body weight. Real-world data suggested that underdosed DOACs were prescribed more frequently in Asian populations compared to Western populations. One possible explanation is that Asians generally have lower body surface area and body than their Western counterparts, which may predispose them to a higher risk of bleeding complications. For instance, a meta-analysis in Asian patients with AF reported a high prevalence of off-label underdosed DOAC prescriptions, reaching up to 63.7% [7], whereas in the United States and Europe, the prevalence was approximately 10% [8–10].
In terms of effectiveness and safety, observational studies conducted in Taiwan reported that the use of off-label underdosed DOACs was associated with an increased risk of ischemic stroke or systemic embolism (ISSE) compared to standard doses [11]. Conversely, a study conducted in the United Kingdom found no significant difference in ISSE risk between patients receiving underdosed versus standard-dose DOACs [10]. Previous meta-analyses also found no significant difference in the risk of ISSE between underdosed and standard-dose DOACs [7,12]. However, contradict the findings of another meta-analysis, which reported an increased ISSE risk associated with underdosed DOACs when compared with standard doses [7]. Across all studies, the risk of major bleeding, which is a key safety outcome, did not differ significantly between underdosed and standard-dose DOAC groups. Given the conflicting evidence regarding the effectiveness and safety of underdosed DOACs, along with the limited number of studies conducted in Thailand, which may serve as one of the representative settings for Asian populations, further investigation is warranted. Therefore, to provide an overall understanding of the use of reduced-dose regimens, this study aimed to examine the prescribing patterns of DOACs, including both on-label and off-label dosing, identify predictors of off-label underdosing, and compare the effectiveness and safety of off-label underdosing of DOACs versus on-label dosing among patients with AF.
2. Patients and methods
2.1. Study design and ethical considerations
This retrospective multicenter cohort study was conducted at two tertiary care hospitals in Thailand, utilizing data extracted from electronic health records (EHRs). The study was conducted in compliance with the ethical principles articulated in the Declaration of Helsinki. Ethical approval was granted by the University of Phayao Human Ethic Committee (HREC-UP-HSST 1.1/004/68); the Ethic Committee of Chiangrai Prachanukroh Hospital, Chiangrai, Thailand (EC CRH 104/67 Ex); and the Ethic Committee of Lampang Hospital, Lampang, Thailand (No. 149/67). This study was reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines (Supplementary Table S1) [13].
2.2. Settings and data sources
This study utilized data from two tertiary care hospitals in Thailand. The first, Lampang Hospital, is a 745-bed facility located in Lampang Province. The second, Chiangrai Prachanukroh Hospital, is situated in Chiangrai Province and has a capacity of 773 beds. Both hospitals operate under the Ministry of Public Health. The EHR databases from these hospitals included comprehensive patient-level data, encompassing demographic information, comorbidities, prescribed medications listed in each hospital’s formulary, as well as laboratory test results from both inpatient and outpatient services.
2.3. Study population
This study used data collected from 1 January 2015 to 31 December 2023. Patients with a first-time diagnosis of AF within the study period were identified using the International Classification of Diseases, Tenth Revision (ICD-10) codes, I48. The inclusion criteria were as follows: (i) patients aged 18 years or older at the time of AF diagnosis; (ii) patients who initiated treatment with any DOACs including dabigatran, rivaroxaban, apixaban, or edoxaban, on or after the date of AF diagnosis, with the first DOAC prescription designated as the index date; and (iii) patients who received at least two DOAC prescriptions, to ensure continuity of care and follow-up at the hospitals. Exclusion criteria were AF patient who: (i) had an observation period less than 12 months prior to AF diagnosis to confirm incident AF cases; (ii) presence of mitral stenosis or a mechanical prosthetic heart valve; (iii) had a history of end-stage renal disease; (iv) received warfarin as the initial OAC; (v) received any OAC prior to AF diagnosis, and (vi) missing data on body weight or serum creatinine.
2.4. Exposure and baseline covariates
DOAC dosing was classified according to international guideline recommendations for stroke prevention in patients with AF [1,2], which are used as the reference dosing framework in the participating tertiary care hospitals. Standard or on-label dosing was defined as the recommended dose for each DOAC: dabigatran 150 mg twice daily, rivaroxaban 20 mg once daily, apixaban 5 mg twice daily, and edoxaban 60 mg once daily. On-label dosing also encompassed dose reductions made in accordance with guideline-recommended criteria. Details of dose reduction recommendations are provided in Supplementary Table S2. The second group consisted of patients who received a lower-than-standard dose of DOACs, referred to as off-label underdosing, which indicated dose reductions not aligned with the recommended clinical guidelines for each specific DOAC. The third group comprised patients receiving off-label overdosed DOACs, defined as prescriptions exceeding the dose recommended by international guidelines [1,2]. The covariates included in this study encompassed patient demographics, comorbidities, and concomitant medications. Demographic variables comprised age, body weight, body mass index (BMI), serum creatinine, and creatinine clearance (CrCl), the latter of which was estimated using the Cockcroft–Gault equation [14]. Comorbid conditions included congestive heart failure, hypertension, diabetes mellitus, dyslipidemia, history of intracranial hemorrhage (ICH), ischemic stroke, chronic kidney disease, and vascular disease. Concomitant medication use included antiplatelet agents, nonsteroidal anti-inflammatory drugs (NSAIDs), angiotensin-converting enzyme inhibitors (ACEIs), angiotensin II receptor blockers (ARBs), beta-blockers, and proton pump inhibitors (PPIs). Information on comorbidities and concurrent medications was collected within one year prior to or on the index date, defined as the initial prescription date of DOACs. Baseline risk scores were also assessed, including the CHA2DS2–VA score [1] and a modified HAS-BLED (excluding international normalized ratio results) score [2].
2.5. Study outcomes and follow-up
The effectiveness outcome was hospitalization due to ISSE, while the safety outcome was the occurrence of any bleeding event. ISSE was defined as a composite of ischemic stroke and systemic arterial embolism. Any bleeding was defined as the occurrence of a hospital diagnosis coded with ICD-10 bleeding codes indicating systemic/hematologic, intracranial, gastrointestinal, genitourinary, ophthalmic, musculoskeletal, or other clinically significant hemorrhage. Full code list is provided in Supplementary Table S3. Patients were followed from the index date, defined as the first date of DOAC initiation, until the first occurrence of any of the following: the outcome of interest, discontinuation of the index DOAC, switching to warfarin, changing to a different DOAC, or the end of the study period (31 December 2023), whichever occurred first. Discontinuation of the index DOAC was defined as the absence of subsequent prescription records beyond the days supplied by the most recent index DOAC prescription.
2.6. Statistical analysis
The proportion of patients receiving on-label dosing, off-label underdosing, and off-label overdosing of DOACs was calculated by dividing the number of AF patients receiving each dosing category by the total number of AF patients prescribed any DOAC. These proportions were further stratified by DOAC type, age group (<60, 60–74, and ≥75 years), and sex.
The full study cohort (N = 553) was used to describe overall DOAC prescribing patterns. For subsequent analyses, including factors associated with off-label underdosing and comparative clinical outcomes, patients receiving off-label overdosing (n = 35, 6.3% of the cohort) were excluded because the primary clinical question focused on off-label underdosing versus on-label dosing and the small number of overdosed patients precluded reliable comparative analysis. Therefore, the analytic cohort included 518 patients receiving either on-label dosing or off-label underdosing. Baseline characteristics of patients receiving on-label dosing and off-label underdosing were summarized using descriptive statistics. Categorical variables were presented as frequencies and percentages, while continuous variables were expressed as mean ± standard deviation (SD) or median with interquartile range (IQR), depending on data distribution. Comparisons between groups were evaluated using the absolute standardized difference (ASD), with an ASD threshold of <0.1 considered negligible [15]. Potential predictors of off-label underdosing were examined using stepwise logistic regression analysis. A probability-of-removal threshold of 0.157 was applied, corresponding to the Akaike Information Criterion (AIC) for model parsimony [16]. The results were presented as adjusted odds ratios (aORs) with 95% confidence intervals (CIs). A subgroup analysis was also performed to explore factors associated with off-label underdosing of rivaroxaban, which was the most frequently prescribed DOAC in this dosing category.
The incidence of ISSE and any bleeding events in the on-label and off-label underdosed groups was reported as incidence rates, calculated by dividing the number of incident cases by the total follow-up time for each group. The primary analysis was conducted using propensity score-based inverse probability of treatment weighting (PS-IPTW) to address potential confounding by indication [17]. The propensity score for receiving an off-label underdosed DOAC was estimated using a logistic regression model that included baseline characteristics and concomitant medications listed in Table 1, except for congestive heart failure, which was excluded from the propensity model due to a zero-cell count (complete separation) in the off-label underdosed group. Covariate balance before and after weighting was assessed using ASD, with values below 0.10 considered indicative of acceptable balance [15]. For any covariates that remained imbalanced after weighting, further adjustment was performed in the final outcome models to ensure a doubly robust estimation. Weighted Cox proportional hazards regression models were used to estimate the associations between off-label underdosed dosing and clinical outcomes, including ISSE and any bleeding events. Notably, chronic kidney disease was excluded from the multivariable adjustment in the bleeding outcome model due to a zero-cell event count among patients with chronic kidney disease. Results were expressed as hazard ratios (HRs) with 95% CIs. As a sensitivity analysis, conventional multivariable Cox proportional hazards regression models were performed to examine the robustness of the primary findings. These models were adjusted for age, sex, body weight, CrCl, hypertension, congestive heart failure, dyslipidemia, diabetes mellitus, prior ICH, prior ischemic stroke, chronic kidney disease, prior bleeding, vascular disease, and concomitant use of antiplatelets, ACEIs/ARBs, beta blockers, calcium channel blockers, NSAIDs, and PPIs. All analyses were performed using STATA version 18.0 (StataCorp LP, College Station, TX). A two-sided p value < 0.05 was considered statistically significant.
Table 1.
Baseline characteristics of patients prescribed on-label DOACs (n = 405) and those prescribed off-label underdosed DOACs (n = 113).
| Characteristics | Total (N = 518) | On-label dosing (n = 405) |
Off-label underdosing (n = 113) |
ASD |
|---|---|---|---|---|
| Demographics | ||||
| Age, mean (±SD), years | 71.9 (10.2) | 72.1 (10.7) | 71.2 (7.9) | 0.093 |
| < 60 years | 61 (11.8) | 54 (13.3) | 7 (6.2) | 0.242 |
| 60–74.9 years | 260 (50.2) | 190 (46.9) | 70 (62.0) | 0.305 |
| 75 years or more | 197 (38.0) | 161 (39.8) | 36 (31.9) | 0.165 |
| Female | 266 (51.4) | 207 (51.1) | 59 (52.2) | 0.022 |
| Body weight, mean (±SD), kg | 61.0 (14.4) | 60.9 (14.5) | 61.4 (14.1) | 0.035 |
| BMI, mean (±SD), kg/m2 | 24.5 (7.8) | 24.5 (8.5) | 24.7 (4.7) | 0.037 |
| <18.5 | 45 (8.7) | 40 (9.9) | 5 (4.4) | 0.213 |
| 18.5–22.9 | 158 (30.5) | 125 (30.9) | 33 (29.2) | 0.036 |
| 23.0–24.9 | 99 (19.1) | 78 (19.3) | 21 (18.6) | 0.017 |
| ≥ 25.0 | 216 (41.7) | 162 (40.0) | 54 (47.8) | 0.157 |
| CHA2DS2-VASc Score, median (IQR)† | 2 (1–3) | 2 (1–3) | 2 (2–3) | 0.040 |
| mHASBLED score, median (IQR)‡ | 2 (1–2) | 2 (1–2) | 2 (1–2) | 0.122 |
| Serum creatinine, mean (±SD) (mg/dL) | 0.97 (0.3) | 0.98 (0.3) | 0.92 (0.3) | 0.199 |
| Creatinine clearance, mean (±SD) (mL/min) | 60.3 (26.4) | 59.8 (27.6) | 62.2 (21.5) | 0.096 |
| < 30 | 43 (8.3) | 35 (8.6) | 8 (7.1) | 0.058 |
| 30–49 | 136 (26.3) | 121 (29.9) | 15 (13.3) | 0.412 |
| > 49 | 339 (65.4) | 249 (61.5) | 90 (79.7) | 0.407 |
| Comorbid conditions, n (%) | ||||
| Hypertension | 246 (47.5) | 194 (47.9) | 52 (46.0) | 0.038 |
| Congestive heart failure | 35 (6.8) | 35 (8.6) | 0 (0.0) | 0.434 |
| Dyslipidemia | 159 (30.7) | 131 (32.4) | 28 (24.8) | 0.168 |
| Diabetes mellitus | 17 (3.3) | 11 (2.7) | 6 (5.3) | 0.132 |
| Intracranial hemorrhage | 7 (1.4) | 6 (1.5) | 1 (0.9) | 0.055 |
| Ischemic stroke | 15 (2.9) | 14 (3.5) | 1 (0.9) | 0.177 |
| Chronic kidney disease | 18 (3.5) | 17 (4.2) | 1 (0.9) | 0.211 |
| History of bleeding | 9 (1.7) | 8 (2.0) | 1 (0.9) | 0.092 |
| Vascular disease | 58 (11.2) | 43 (10.6) | 15 (13.3) | 0.082 |
| Concomitant medications, n (%) | ||||
| Antiplatelets | 105 (20.3) | 82 (20.3) | 23 (20.4) | 0.003 |
| ACEIs/ARBs | 180 (34.8) | 152 (37.5) | 28 (24.8) | 0.277 |
| Beta-blockers | 374 (72.2) | 290 (71.6) | 84 (74.3) | 0.061 |
| CCBs | 148 (28.6) | 115 (28.4) | 33 (29.2) | 0.018 |
| NSAIDs | 37 (7.1) | 34 (8.4) | 3 (2.7) | 0.253 |
| PPIs | 250 (48.3) | 204 (50.4) | 46 (40.7) | 0.194 |
ACEIs/ARBs: angiotensin-converting enzyme inhibitors/angiotensin ii receptor blockers; ASD: absolute standardized differences; BMI: body mass index; CCBs: calcium channel blockers; IQR: interquartile range; NSAIDs: nonsteroidal anti-inflammatory drugs; PPIs: proton pump inhibitors; SD: standard deviation
†CHA2DS2-VASc indicates patients with congestive cardiac failure, hypertension, age ≥75 years (doubled), diabetes mellitus, age 65–74 years, prior stroke or TIA or SE (doubled), vascular disease, and gender category (women). CHA2DS2-VASc score ranges from 0 to 9 (higher score indicates a higher risk for stroke).
‡Modified HAS-BLED indicates patients with hypertension, renal disease, liver disease, prior stroke, prior major bleeding, age > 65 years, medications that predispose to bleeding (NSAIDs or antiplatelet drugs), alcohol use (labile INR not included). Modified HAS-BLED score ranges from 0 to 8 (as labile INR not included in calculation), a higher score indicates a higher risk for bleeding.
3. Results
3.1. Baseline characteristics
A total of 15,230 new cases of AF were identified from the EHRs of two tertiary care hospitals. After applying the inclusion criteria, 553 patients with incident AF who were new users of DOACs were included in the study. Among these, 405 patients received on-label dosing, while 148 patients received off-label dosing. Specifically, 113 patients were classified as receiving off-label underdosing, and 35 patients as off-label overdosing (Figure 1). Among patients included in the analytic cohort comparing on-label dosing and off-label underdosing (n = 518), the mean age was 71.9 years, and 51.4% were female. The mean body weight was 61 kg. The median CHA2DS2-VASc and modified HAS-BLED scores were both 2. The mean CrCl was 60.3 mL/min. Hypertension was the most prevalent comorbidity, observed in 47.5% of patients, whereas a history of stroke was identified in only 2.9%. Concomitant medication use included antiplatelet agents in 20.3% of patients and PPIs in 48.3% (Table 1).
Figure 1.
Study flow diagram of population enrollment. AF: atrial fibrillation; DOACs: direct oral anticoagulants; OAC: oral anticoagulants
3.2. Prescribing patterns of DOACs
Among the 553 new users of DOACs, the most frequently prescribed agent was rivaroxaban (n = 353, 63.8%), followed by dabigatran (n = 106, 19.2%), edoxaban (n = 89, 16.1%), and apixaban (n = 5, 0.9%), respectively. The majority of prescriptions were in the on-label dosing category (73.2%), whereas off-label underdosing and off-label overdosing accounted for 20.4% and 6.3%, respectively (Figure 2(A)). Among patients receiving off-label underdosed DOACs (n = 113), rivaroxaban was the most commonly prescribed agent (n = 57, 50.4%), followed by dabigatran (n = 50, 44.2%). For rivaroxaban, off-label underdosing was mainly due to the use of 10 mg once daily (n = 6) or 15 mg once daily in patients with CrCl > 49 mL/min (n = 51), which fell outside the recommended criteria for dose adjustment. Conversely, off-label overdosing occurred in patients prescribed rivaroxaban 20 mg once daily despite having CrCl ≤ 49 mL/min (n = 27). Among those prescribed dabigatran, off-label underdosing primarily involved the use of dabigatran 110 mg twice daily in patients who did not meet the dose-reduction criteria (n = 49). A smaller number of patients received dabigatran 75 mg twice daily (n = 1). Off-label overdosing with dabigatran occurred in two patients who received 150 mg twice daily despite meeting the criteria for dose reduction due to advanced age (≥80 years) or high bleeding risk (Supplementary Table S4).
Figure 2.
Distribution of DOAC dosing overall (2 A), by DOAC type (2B), age group (2 C), and sex (2D). DOACs: direct oral anticoagulants
In contrast, when dosing patterns were examined with each DOAC group, dabigatran had the highest proportion of off-label underdosing (47.2%), despite rivaroxaban accounting for the largest number of underdosed patients overall. Among rivaroxaban users, 76.2% received on-label dosing, while 16.1% and 7.6% received off-label underdosing and overdosing, respectively. Edoxaban demonstrated equal proportion of off-label underdosing and overdosing (6.7% each), whereas apixaban was prescribed exclusively at on-label doses in this cohort (Figure 2(B) and Supplementary Table S5). When stratified by age group, the majority of patients receiving DOACs were aged 60–74 years (n = 277). Across all age groups, on-label dosing was most frequently used, accounting for 85.7%, 68.6%, and 75.6% in patients aged <60, 60–74, and ≥75 years, respectively. The highest proportion of off-label underdosing was observed in the 60–74 years group (25.3%), followed by those aged ≥75 years (16.9%) and <60 years (11.1%) (Figure 2(C) and Supplementary Table S6). When stratified by sex, both males and females had similar distributions of DOAC dosing. On-label dosing was prescribed to 72.4% of females and 74.2% of males. Likewise, the proportions of off-label underdosing and overdosing were comparable between the two sexes (Figure 2(D) and Supplementary Table S6).
3.3. Predictors for prescribing off-label underdosing DOAC
When comparing the baseline characteristics between patients prescribed on-label DOACs and those prescribed off-label underdosed DOACs, the mean age, body weight, BMI, and CrCl were generally comparable between groups. However, a higher proportion of patients aged 60–74.9 years was observed among those who received off-label underdosed DOACs compared with those on on-label dosing. Regarding comorbid conditions, the overall distribution was largely similar between the two dosing groups, except for congestive heart failure, dyslipidemia, diabetes mellitus, ischemic stroke, and chronic kidney disease. In terms of concomitant medications, patients receiving on-label DOACs had a higher proportion of ACEIs/ARBs, NSAIDs, and PPIs use compared with those receiving off-label underdosed DOACs (Table 1). Following stepwise multivariable logistic regression analysis, diabetes mellitus was found to be significantly associated with an increased likelihood of receiving off-label underdosed DOACs (aOR = 3.27, 95% CI: 1.02–10.54, p = 0.047). Similarly, patients aged 60–74.9 years were more likely to receive off-label underdosed DOACs compared with those aged <60 years (aOR = 2.67, 95% CI: 1.12–6.32, p = 0.026). In contrast, patients with CrCl between 30 and 49 mL/min (aOR = 0.32, 95% CI: 0.17–0.61, p = 0.001) and those receiving ACEIs/ARBs were less likely to be prescribed off-label underdosed DOACs (aOR = 0.53, 95% CI: 0.32–0.88, p = 0.015). In the subgroup analysis restricted to rivaroxaban users, age 60–74.9 years remained significantly associated with off-label underdosing (aOR = 5.49, 95% CI: 1.21–25.03, p = 0.028), with an even stronger association observed among those aged ≥ 75 years. Conversely, lower CrCl and concomitant use of ACEIs/ARBs were associated with a reduced likelihood of receiving off-label underdosed rivaroxaban (Table 2).
Table 2.
Stepwise multivariable logistic regression analysis for predictors of off-label underdosing DOACs and rivaroxaban.
| Predictors | Off-label underdosing DOAC as a group |
Off-label underdosing rivaroxaban |
||
|---|---|---|---|---|
| aOR (95% CI) | p Value | aOR (95% CI) | p Value | |
| Age, year | ||||
| < 60 years | Reference | Reference | ||
| 60–74.9 years | 2.67 (1.12–6.32) | 0.026 | 5.49 (1.21–25.03) | 0.028 |
| 75 years or more | 2.14 (0.85–5.37) | 0.105 | 10.74 (2.15–53.66) | 0.004 |
| Female | 1.70 (0.88–3.29) | 0.115 | ||
| BMI, kg/m2 | ||||
| <18.5 | 0.46 (0.17–1.27) | 0.134 | ||
| 18.5–22.9 | Reference | |||
| 23.0–24.9 | ||||
| ≥ 25.0 | ||||
| Creatinine clearance, mL/min | ||||
| < 30 | 0.15 (0.03–0.78) | 0.025 | ||
| 30–49 | 0.32 (0.17–0.61) | 0.001 | 0.06 (0.01–0.21) | <0.001 |
| ≥ 50 | Reference | Reference | ||
| Dyslipidemia | 0.63 (0.38– 1.04) | 0.073 | ||
| Diabetes mellitus | 3.27 (1.02–10.54) | 0.047 | ||
| Ischemic stroke | 0.14 (0.02– 1.28) | 0.083 | ||
| Chronic kidney disease | 0.14 (0.02–1.15) | 0.067 | ||
| Vascular disease | 1.70 (0.85–3.39) | 0.133 | 3.20 (1.30–7.87) | 0.011 |
| Antiplatelets | ||||
| ACEIs/ARBs | 0.53 (0.32–0.88) | 0.015 | 0.44 (0.21–0.90) | 0.026 |
| Beta-blockers | 1.88 (0.87–4.07) | 0.107 | ||
| NSAIDs | 0.37 (0.11–1.29) | 0.121 | ||
ACEIs: angiotensin-converting enzyme inhibitors; aOR: adjusted odds ratio; ARBs: angiotensin receptor blockers; BMI: body mass index; CI: confidence interval; DOAC: direct oral anticoagulant; NSAIDs: nonsteroidal anti-inflammatory drugs.
Model was derived using stepwise logistic regression (entry/removal criterion p < 0.157) to identify independent predictors of off-label underdosing DOAC use. Only variables retained in the final model are shown.
3.4. Comparative effectiveness and safety outcomes between on-label dosing of DOAC and off-label underdosing of DOAC
For ISSE outcomes, the mean follow-up for ISSE was 1.24 years in the on-label group and 1.35 years in the off-label underdosed group (overall mean 1.26 years). The incidence rate of ISSE was lower in the off-label underdosed group (1.32 events per 100 person-years) than in the on-label dosing group (2.99 events per 100 person-years). However, this difference was not statistically significant after PS-IPTW (adjusted hazard ratio [aHR] = 0.48, 95% CI: 0.11–2.16, p = 0.336). Similarly, for any bleeding outcomes, the mean follow-up was 1.26 years in the on-label group and 1.31 years in the off-label underdosed group (overall mean 1.28 years). The incidence rate of any bleeding was comparable between the off-label underdosed and on-label dosing groups (2.02 vs. 2.34 events per 100 person-years, respectively). After PS-IPTW, no statistically significant association was observed (aHR = 1.06, 95% CI: 0.30–3.73, p = 0.924). In a sensitivity analysis using covariate adjustment based on baseline characteristics, the findings remained consistent with the primary analysis, showing no significant difference between off-label underdosing and on-label dosing of DOACs in either ISSE outcomes (aHR = 0.54, 95% CI: 0.12–2.53, p = 0.436) or any bleeding outcomes (aHR = 0.88, 95% CI: 0.23–3.45, p = 0.857). Covariate balance before and after PS-IPTW was assessed using ASDs and is presented in Supplementary Table S7 and the comparative outcome estimates are presented as a forest plot in Figure 3. In addition, a simplified graphical summary is provided in Figure 4 to integrate the dose classification, selected patient characteristics relevant to off-label underdosing, and clinical outcomes during follow-up.
Figure 3.
Forest plot of ISSE and any bleeding comparing off-label underdosing with on-label dosing of DOACs. The covariate-adjusted Cox models included age, sex, body weight, CrCl, hypertension, congestive heart failure, dyslipidemia, diabetes mellitus, prior intracranial hemorrhage, prior ischemic stroke, chronic kidney disease, prior bleeding, vascular disease, and concomitant use of antiplatelets, ACEIs/ARBs, beta blockers, calcium channel blockers, NSAIDs, and PPIs. aHR: adjusted hazard ratio; CI: confidence interval; DOACs: direct oral anticoagulants; HR: hazard ratio; ISSE: ischemic stroke or systemic embolism; PS-IPTW: propensity score inverse probability of treatment weighting; PY: person-years; Ref: reference
Figure 4.
Simplified graphical summary of patient profiles and clinical outcomes during follow-up. ACEIs/ARBs: angiotensin-converting enzyme inhibitors/angiotensin II receptor blockers; AF: atrial fibrillation; aHR: adjusted hazard ratio; CHF: congestive heart failure; CI: confidence interval; CKD: chronic kidney disease; CrCl: creatinine clearance; DOACs: direct oral anticoagulants; ISSE: ischemic stroke/systemic embolism; NSAIDs: nonsteroidal anti-inflammatory drugs; PS-IPTW: propensity score–based inverse probability of treatment weighting; PY: person-years.
4. Discussion
4.1. Main findings
This electronic health record–based cohort from two tertiary-care hospitals in Thailand examined DOAC prescribing dosing patterns, predictors of off-label underdosing, and comparative outcomes between off-label underdosed and on-label dosing among patients with AF. Among incident AF patients initiating DOAC, rivaroxaban was the most frequently prescribed agent (63.8%), followed by dabigatran, edoxaban, and apixaban. Overall, 73.2% received on-label dosing, whereas 20.4% received off-label underdosing. Approximately half of the underdosed prescriptions involved rivaroxaban most commonly 15 mg once daily despite CrCl > 49 mL/min. In multivariable analyses, older age, diabetes mellitus, and vascular disease were independently associated with higher odds of off-label underdosing. With respect to clinical outcomes, off-label underdosing was not associated with increased risks of ISSE or any bleeding compared with on-label dosing.
4.2. Interpretation of the findings and comparisons with previous studies
Thai data on DOAC utilization vary across settings and time. A nationwide analysis of DOAC use for any indication reported dabigatran as the most frequently prescribed agent (46.3%), followed by rivaroxaban (28.5%) and apixaban (25.2%) [18]. By contrast, a tertiary-care study conducted in 2021–2023 found rivaroxaban predominated (39.2%) [19]. Our findings aligned with this latter pattern, with an even higher proportion of rivaroxaban use (63.8%). Rivaroxaban emerged as the most frequently prescribed DOAC in our cohort largely due to hospital formulary and administrative constraints. In both hospitals, rivaroxaban was initially the only listed DOAC, with other agents added later and requiring written approval for each prescription, which discouraged their routine use. Consequently, prescribers preferentially selected rivaroxaban, which was already fully integrated into the prescribing system. By contrast, a previous nationwide analysis reported dabigatran as the most commonly used DOAC [18]. This may partly reflect that the data were collected from university hospitals, likely reflecting data from university hospitals where most DOACs are available and physicians can more readily tailor drug choice to patients’ clinical profiles. Moreover, compared with earlier Thai cohorts, off-label underdosing was reported at 22.2% [18] and 16.7% [19], and off-label overdosing at 10.7% [18] and 20% [19] In our cohort, off-label underdosing was comparable (20.4%), whereas off-label overdosing was lower (6.3%). When these findings are considered alongside studies from other Asian countries, the prevalence of Asian patients with AF who were prescribed DOACs at suboptimal doses has been shown to range widely, from 3.7% to 63.7%, with an average approximately 20–30% [7]. A recent meta-analysis likewise estimated off-label underdosing across DOACs at about 20%, with off-label overdosing at roughly 5%, which is consistent with our findings [20]. Notably, high rates of off-label underdosing were not confined to Asian populations [21]; they were also observed in non-Asian patients with AF, although the rates tended to be lower outside Asia (31% in Asia, 16% in Europe, and 7% in North America) [20]. Several factors may explain these differences: heterogeneous definitions of off-label underdosing across studies, variation in local prescribing practices and care settings, country-specific approved dose regimens, and the specific DOACs evaluated (single-agent vs. pooled analyses), all of which influence estimated prevalence. The relatively higher underdosing rates observed in Asian cohorts may reflect lower average body weight and a greater perceived risk of bleeding. Prior reports indicated that patients of Asian ancestry had an increased risk of anticoagulant-associated bleeding compared with other racial groups, which may prompt clinicians to select doses below label recommendations out of concern for hemorrhage [22–25]. This prescribing behavior may be partly explained by the so-called Asian or East Asian paradox, which describes the distinct risk–benefit profile of antithrombotic therapy in Asian patients, particularly the heightened concern regarding ICH and other serious bleeding complications [26]. Previous data reported that Asian patients with AF may have a higher susceptibility to spontaneous and anticoagulant-related ICH compared with non-Asian populations [22,27]. Such concern may lead clinicians in Asian practice to adopt a more conservative dosing approach and to prescribe reduced DOAC doses empirically, even when patients do not meet guideline-defined dose-reduction criteria. In our cohort, off-label underdosing was frequently observed among patients with preserved renal function, suggesting that dose reduction may have reflected perceived bleeding risk and clinical caution rather than formal renal dose-adjustment criteria.
According to DOAC type, rivaroxaban showed the highest rate of off-label underdosing (50.4%), driven largely by prescriptions of 15 mg once daily in patients with CrCl > 49 mL/min. This pattern aligned with prior meta-analyses reporting that rivaroxaban was the DOAC most frequently underdosed off-label, with rates similar to apixaban [20]. Several factors may underline this observation. First, rivaroxaban is widely used in routine practice in Thailand [19], including in the institutions contributing to this study. Second, some clinicians may be influenced by the Japanese dosing scheme from J-ROCKET AF [28], in which the approved standard dose is rivaroxaban 15 mg once daily, reduced to 10 mg once daily when CrCl ≤ 49 mL/min; applying that scheme outside Japan means that 15 mg once daily in patients with preserved renal function would be classified as off-label underdosing under international guidelines. Third, our cohort consisted mainly of older adults (mean age 71.9 years) with relatively low body weight (mean 61 kg), which may increase clinicians’ concerns about bleeding. Given reports that rivaroxaban may have a less favorable bleeding profile than some other DOACs [29,30], prescribers may have chosen reduced doses after considering patient-specific factors.
In our cohort, individuals aged 60–74.9 years had higher odds of receiving off-label reduced-dose DOACs than those aged <60 years, consistent with prior studies identifying age as an independent predictor of inappropriate low-dose prescribing [31–33]. We also observed that very old age (≥75 years) was specifically associated with off-label underdosing of rivaroxaban. In clinical practice, advanced age was frequently cited as a primary rationale for prescribing off-label reduced-dose rivaroxaban. This pattern appeared to reflect precautionary dosing in the setting of perceived bleeding susceptibility among older adults, encompassing concerns about hemorrhagic risk, frailty, fall risk, polypharmacy, and borderline renal function even when these factors did not meet label-specified criteria for dose reduction. Older patients also tended to have lower body weight and diminished renal function, both of which signal higher bleeding risk. Together, these considerations likely motivated a protective dose-lowering strategy based on the intuitive premise that greater frailty warrants a lower dose, despite the absence of label-concordant indications for reduction [31].
In addition to age, diabetes mellitus was associated with higher odds of receiving off-label underdosing of DOACs. A plausible explanation is that diabetes is accompanied by complications such as diabetic nephropathy, leading to reduced renal function. Because several DOACs particularly rivaroxaban require dose adjustment based on renal function, clinicians may be inclined to reduce the dose in patients with diabetes. Prior studies identified impaired renal function as a principal predictor of inappropriate dosing, most often underdosing [32]. In practice, diabetes may serve as a clinical cue for fragility (e.g., presumed renal impairment, bleeding propensity, and polypharmacy), prompting dose reduction even when formal criteria for dose adjustment are not met, in an effort to avoid adverse events. We also found that vascular disease was associated with off-label underdosing of rivaroxaban. Patients with vascular disease such as prior myocardial infarction or peripheral artery disease commonly received antiplatelet therapy in addition to anticoagulation, which increases bleeding risk. Given concerns about bleeding with rivaroxaban, particularly gastrointestinal bleeding, clinicians may down-titrate rivaroxaban to mitigate this risk. This observation aligned with previous work from a lower-income setting, in which vascular disease predicted inappropriate DOAC dosing (OR = 2.28; 95% CI, 1.11–4.67; p = 0.024) [32]. Conversely, our data suggested that reduced renal function (CrCl ≤ 49 mL/min) was associated with receiving standard, on-label dosing compared with patients with CrCl ≥ 50 mL/min. This pattern likely reflected both DOAC pharmacokinetics and prescriber behavior: several commonly used DOACs in our cohort, including edoxaban and rivaroxaban, mandate renal dose adjustment. When CrCl was ≤ 49 mL/min, clinicians appeared to appropriately reduce the dose per labeling to limit drug accumulation and lower bleeding risk; accordingly, patients with impaired renal function were more often classified as receiving on-label dosing. Notably, prior studies reported that off-label underdosing tended to occur in patients with preserved renal function, those in whom clinicians had greater latitude to lower the dose despite the absence of a labeling indication [34]. We observed a similar pattern: the most frequent reduction of rivaroxaban from 20 mg to 15 mg occurred among patients with CrCl > 49 mL/min (n = 51). These findings implied that prescribers generally adhered to label-recommended dose reductions in patients with true renal impairment, while off-label dose reductions without a renal indication may be applied to patients with preserved renal function due to concerns such as frailty or perceived bleeding risk. This raised the concern that off-label underdosing could compromise the effectiveness of stroke and systemic embolism prevention.
Our study found that off-label underdosing of DOACs was not associated with a higher risk of ISSE compared with on-label dosing. This agreed with several meta-analyses of observational studies reporting no significant difference [12,35,36] but contrasted with others showing increased ISSE with underdosing [7,37–39]. Evidence restricted to Asian populations was likewise mixed: one analysis observed a higher ISSE risk with underdosing [7], whereas another reported no difference [12]. A recent analysis in Japanese patients suggested no statistically significant difference in the risk of stroke or systemic embolism between off-label underdosing and on-label dosing [40]. These discrepancies likely reflect several factors: (i) heterogeneous definitions of underdose and overdose, including inconsistent renal function assessment (e.g., use of eGFR rather than Cockcroft–Gault CrCl), which can misclassify dosing appropriateness [41–43]; (ii) cross-jurisdictional differences in approved dosing regimens, such that off-label underdosing in one country may be on-label in another; and (iii) population differences in the bleeding–thrombosis tradeoff, East Asian patients tended to have greater bleeding susceptibility and lower thrombotic risk at comparable anticoagulant exposure, encouraging dose reductions and potentially attenuating any signal for increased ISSE [44]. Residual confounding and channeling bias in observational studies may further contribute to divergent estimates across studies. Hence, conclusions about off-label dose reduction should be interpreted in the context of the specific DOAC, local label, and study design quality. For safety outcomes, off-label underdosing was often pursued to reduce bleeding; however, our study did not find a reduction in any bleeding with off-label underdosing compared with on-label dosing. This observation was consistent with prior meta-analyses, which generally showed that off-label underdosing was not associated with lower rates of bleeding [7,12,36,37,39,40].
4.3. Strengths and limitations
This study has several limitations. First, baseline data on body weight and serum creatinine were missing for a subset of patients, limiting complete assessment of label-concordant DOAC dosing. In routine practice, documentation of body weight and serum creatinine at initiation (and periodically thereafter) is essential to estimate renal function and ensure appropriate dosing. Second, although patient-level factors, such as age, renal function, comorbidities, and concomitant medications were available, physician-related determinants of DOAC dosing could not be evaluated. Information on prescriber specialty, specialist consultation, approval processes, and physician-level prescribing preferences was not captured in the structured electronic health record database. Moreover, both study sites were tertiary-care hospitals, limiting variability in hospital-level characteristics. In the Thai healthcare context, DOACs are non-essential medicines, and their prescribing often requires specialist consultation or approval, particularly by internal medicine or cardiology specialists; however, these processes could not be directly verified in the present dataset. Future real-world studies should incorporate prescriber-level and institutional-level data to better clarify how clinical judgment, specialist involvement, and hospital policies influence off-label DOAC underdosing. Third, the number of outcome events was relatively small, including both ISSE and bleeding events, which limited the statistical power to detect between-group differences. Although PS-IPTW improved covariate balance for most baseline characteristics, some covariates, including congestive heart failure and chronic kidney disease, remained imbalanced after weighting. Further adjustment for these variables was limited by zero-cell or sparse-event issues; in particular, congestive heart failure showed complete separation in the off-label underdosing group, and chronic kidney disease could not be included in the bleeding outcome model because no bleeding events occurred among patients with chronic kidney disease. Therefore, residual confounding cannot be fully excluded, and the non-significant findings should not be interpreted as evidence of equivalence, benefit, or absence of harm. These findings should be interpreted cautiously and validated in larger cohorts with sufficient outcome events. Pragmatically, assembling large DOAC cohorts in Thailand can be challenging because DOACs remain largely outside the National List of Essential Medicines and often require out-of-pocket payment or are available only through restricted reimbursement schemes that cover a minority of the population. Where adequate sample sizes are feasible, future studies should evaluate effectiveness and safety by individual DOAC and conduct cost-effectiveness analyses in Thai healthcare settings. Fourth, because this study was conducted in tertiary-care hospitals, generalizability to primary or secondary care settings may be limited; however, these centers account for a substantial proportion of DOAC prescribing in Thailand, which likely enhances internal validity for the local context. Fifth, this study did not identify or analyze the subgroup of patients with coronary artery disease undergoing percutaneous coronary intervention who require oral anticoagulation. In this population, the PIONEER AF-PCI trial [45] supports rivaroxaban 15 mg once daily plus a P2Y12 inhibitor in patients with normal renal function, with a reduction to 10 mg once daily for CrCl 30–50 mL/min. These regimens are considered on-label for such patients but may have been classified as off-label in our analysis. Sixth, findings are most directly applicable to Asian populations; extrapolation to non-Asian settings should be made with caution. Finally, as with all observational studies, the possibility of residual or unmeasured confounding could not be excluded.
The study also has important strengths. To our knowledge, this is the first to evaluate the comparative effectiveness of off-label underdosed versus on-label DOACs specifically among Thai patients with AF. We employed a new-user design in real-world practice, thereby minimizing biases associated with prior anticoagulant exposure (e.g., patients switching from warfarin with labile INR or heightened bleeding susceptibility). The analysis adds evidence from a Southeast-Asian, upper-middle-income setting in which contemporary DOAC data remain limited. Findings were robust across sensitivity analyses, including propensity score-adjusted analysis, which yielded estimates consistent with the primary results.
4.4. Implications for practice
Based on our findings, off-label underdosing of DOACs did not reduce bleeding, contrary to the clinical intent of dose reduction. Although the risk of stroke did not differ from on-label dosing in this cohort, the broader evidence base remained heterogeneous. Multiple meta-analyses particularly outside Japan linked underdosing to higher thromboembolic risk without a bleeding benefit. Accordingly, current data supports a default strategy of guideline-concordant, label-based dosing, with bleeding-risk mitigation pursued through optimization of modifiable factors rather than off-label dose reduction, unless and until robust evidence demonstrates a superior net clinical benefit of underdosing.
The use of off-label DOAC underdosing, often driven by clinicians’ concerns about bleeding risk, highlights an important unmet need for anticoagulant strategies that can better balance thromboembolic protection and bleeding safety. In this context, factor XI or factor XIa inhibitors (FXI/FXIa inhibitors) have emerged as potential future anticoagulant strategies for patients with AF in whom bleeding risk is a major clinical concern. At present, these agents are not yet routinely recommended for stroke prevention in patients with AF. Recent systematic reviews comparing FXI/FXIa inhibitors with DOACs suggest that these agents may offer a more favorable bleeding profile. However, this potential safety advantage should be balanced against remaining uncertainty regarding their efficacy for stroke and systemic embolism prevention, which may be lower than that of DOACs in patients with AF [46,47]. For patients in whom concerns about bleeding currently contribute to off-label DOAC underdosing, appropriately dosed FXI/FXIa inhibition may eventually provide a more rational strategy to address both thromboembolic and bleeding risks. Nevertheless, FXI/FXIa inhibitors should not currently be viewed as substitutes for guideline-concordant DOAC dosing or as justification for off-label DOAC underdosing. Further long-term randomized controlled trials are needed to clarify their efficacy, optimal dosing, safety, and net clinical benefit, particularly in patients at high risk of bleeding. Until such evidence is available, practical efforts should continue to focus on accurate DOAC dose assessment and on-label prescribing.
In addition, dosing should be determined by Cockcroft–Gault CrCl calculated with actual body weight, with baseline weight and serum creatinine documented to ensure appropriate drug selection and dosing. Further, stewardship should target groups prone to underdosing (e.g., adults aged 60–74.9 years and those with diabetes), using pharmacist review and clinical decision support to promote on-label prescribing while individualizing bleeding-risk management.
5. Conclusions
In this real-world cohort of patients with AF, approximately one-fifth initiated DOAC at off-label underdosed regimen, most commonly with rivaroxaban, often despite preserved renal function. Older age and diabetes mellitus were independently associated with underdosing, suggesting a cautious, risk-averse prescribing pattern in routine practice. Off-label underdosing was not associated with statistically significant differences in thromboembolic or bleeding outcomes compared with on-label dosing; however, these findings should be interpreted cautiously given limited event rates. Future large-scale, multicenter studies are warranted to improve generalizability and to evaluate net clinical and economic outcomes across diverse populations.
Declaration of generative AI and AI-assisted technologies in the writing process
During the preparation of this work the author(s) used ChatGPT (GPT-5.5 Thinking, OpenAI) for English language proofreading. After using this tool/service, the author(s) reviewed and edited the content as needed and take(s) full responsibility for the content of the published article.
Supplementary Material
Funding Statement
This manuscript was supported by the School of Pharmaceutical Sciences, University of Phayao. The funder had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Author contributions: CRediT
Conceptualization: PM, TK, PB, Sirayut Phattanasobhon, Supisara Pholprasittito, SW; Methodology: PM, TK, PB, Sirayut Phattanasobhon, Supisara Pholprasittito, SW; Data curation: PM, TK, PB, Sirayut Phattanasobhon; Formal analysis and investigation: PM, TK, PB, Sirayut Phattanasobhon; Writing - original draft preparation: PM with input from all authors; Writing - review and editing: all authors; Project administration: PM; Funding acquisition: PM, TK, PB; Resources: Supisara Pholprasittito, SW, NB. All authors approved the final manuscript and agreed to be accountable for all aspects of the work.
Disclosure statement
The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.
Ethics approval
The study was performed in line with the principles of the Declaration of Helsinki. Ethical approval was granted by the University of Phayao Human Ethic Committee (HREC-UP-HSST 1.1/004/68); the Ethic Committee of Chiangrai Prachanukroh Hospital, Chiangrai, Thailand (EC CRH 104/67 Ex); and the Ethic Committee of Lampang Hospital, Lampang, Thailand (No. 149/67).
Consent to participate
As this retrospective analysis used routinely collected EHRs with no direct patient contact and all data were de-identified prior to analysis, the requirement for written informed consent was waived by the Institutional Review Boards of the participating institutions.
Consent to publish
All authors have read and approved the final version of the manuscript and consent to its publication.
Data availability statement
The data that support the findings of this study are available from the corresponding author, upon reasonable request.
References
Papers of special note have been highlighted as either of interest (•) or of considerable interest (••) to readers.
- 1.Van Gelder IC, Rienstra M, Bunting KV, 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]
- 2.Joglar JA, Chung MK, Armbruster AL, 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]
- 3.Connolly SJ, Ezekowitz MD, 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]
- 4.Giugliano RP, Ruff CT, 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]
- 5.Granger CB, Alexander JH, McMurray JJ, 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]
- 6.Patel MR, Mahaffey KW, Garg J, et al. Rivaroxaban versus warfarin in nonvalvular atrial fibrillation. N Engl J Med. 2011;365(10):883–891. doi: 10.1056/NEJMoa1009638 [DOI] [PubMed] [Google Scholar]
- 7.Mongkhon P, Singkham N, Ponok K, et al. Comparative effectiveness and safety of off-label underdosed direct oral anticoagulants in asian patients with atrial fibrillation: a systematic review and meta-analysis. Drug Saf. 2025;48(1):25–42. doi: 10.1007/s40264-024-01476-8 [DOI] [PubMed] [Google Scholar]; •• This systematic review and meta-analysis provides directly relevant evidence on the effectiveness and safety of off-label DOAC underdosing specifically among Asian patients with atrial fibrillation.
- 8.Sandhu A, Kaltenbach LA, Chiswell K, et al. Off-label dosing of direct oral anticoagulants among inpatients with atrial fibrillation in the united states. Circ Cardiovasc Qual Outcomes. 202316(12):e010062. doi: 10.1161/CIRCOUTCOMES.123.010062 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Vinter N, Linder M, Andersen M, et al. Classification and characteristics of on-label and off-label apixaban use in Denmark and Sweden. Pharmacoepidemiol Drug Saf. 2019;28(6):867–878. doi: 10.1002/pds.4778 [DOI] [PubMed] [Google Scholar]
- 10.van den Dries CJ, Pajouheshnia R, van den Ham HA, et al. Safety of off-label dose reduction of non-vitamin K antagonist oral anticoagulants in patients with atrial fibrillation. Br J Clin Pharmacol. 2023;89(2):751–761. doi: 10.1111/bcp.15534 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Chan YH, Chao TF, Chen SW, et al. Off-label dosing of non-vitamin K antagonist oral anticoagulants and clinical outcomes in Asian patients with atrial fibrillation. Heart Rhythm. 2020;17(12):2102–2110. doi: 10.1016/j.hrthm.2020.07.022 [DOI] [PubMed] [Google Scholar]; •• This large real-world study evaluated off-label DOAC dosing and associated clinical outcomes in Asian patients, making it one of the closest comparative studies to the present research.
- 12.Shen NN, Ferroni E, Amidei CB, et al. An updated pooled analysis of off-label under and over-dosed direct oral anticoagulants in patients with atrial fibrillation. Clin Appl Thromb Hemost. 2023;29:10760296231179439. doi: 10.1177/10760296231179439 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.von Elm E, Altman DG, Egger M, et al. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. J Clin Epidemiol. 2008;61(4):344–349. doi: 10.1016/j.jclinepi.2007.11.008 [DOI] [PubMed] [Google Scholar]
- 14.Cockcroft DW, Gault MH.. Prediction of creatinine clearance from serum creatinine. Nephron. 1976;16(1):31–41. doi: 10.1159/000180580 [DOI] [PubMed] [Google Scholar]
- 15.Austin PC. Using the standardized difference to compare the prevalence of a binary variable between two groups in observational research. Commun Stat Simul Comput. 2009;38(6):1228–1234. doi: 10.1080/03610910902859574 [DOI] [Google Scholar]
- 16.Collins GS, Reitsma JB, Altman DG, et al. Transparent reporting of a multivariable prediction model for individual prognosis or diagnosis (TRIPOD): the TRIPOD Statement. BMC Med. 201513(1):1. doi: 10.1186/s12916-014-0241-z [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Austin PC, Stuart EA.. Moving towards best practice when using inverse probability of treatment weighting (IPTW) using the propensity score to estimate causal treatment effects in observational studies. Stat Med. 2015;34(28):3661–3679. doi: 10.1002/sim.6607 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Wattanaruengchai P, Nathisuwan S, Rattanavipanon W, et al. Prescriber compliance to direct oral anticoagulant labels and impact on outcomes in Thailand. Br J Clin Pharmacol. 2021;87(3):1390–1400. doi: 10.1111/bcp.14535 [DOI] [PubMed] [Google Scholar]; •• This study provides highly relevant Thai evidence on adherence to DOAC dosing recommendations and the clinical consequences of noncompliant prescribing.
- 19.Pongsathabordee C, Saringkarn P, Ratanapornsompong K, et al. Appropriateness of direct oral anticoagulant dosing in patients with atrial fibrillation at a tertiary care hospital in Thailand. Explor Res Clin Soc Pharm. 2024;16:100507. doi: 10.1016/j.rcsop.2024.100507 [DOI] [PMC free article] [PubMed] [Google Scholar]; • This recent Thai study describes the appropriateness and patterns of DOAC dosing in a tertiary-care setting and provides important national context for the present findings.
- 20.Shen NN, Zhang C, Hang Y, et al. Real-world prevalence of direct oral anticoagulant off-label doses in atrial fibrillation: an epidemiological meta-analysis. Front Pharmacol. 2021;12:581293. doi: 10.3389/fphar.2021.581293 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Arbel R, Sergienko R, Hammerman A, et al. Effectiveness and safety of off-label dose-reduced direct oral anticoagulants in atrial fibrillation. Am J Med. 2019;132(7):847–855.e3. doi: 10.1016/j.amjmed.2019.01.025 [DOI] [PubMed] [Google Scholar]
- 22.Hori M, Connolly SJ, Zhu J, et al. Dabigatran versus warfarin: effects on ischemic and hemorrhagic strokes and bleeding in Asians and non-Asians with atrial fibrillation. Stroke. 2013;44(7):1891–1896. doi: 10.1161/STROKEAHA.113.000990 [DOI] [PubMed] [Google Scholar]
- 23.Wong KS, Hu DY, Oomman A, et al. Rivaroxaban for stroke prevention in East Asian patients from the ROCKET AF trial. Stroke. 2014;45(6):1739–1747. doi: 10.1161/STROKEAHA.113.002968 [DOI] [PubMed] [Google Scholar]
- 24.Yamashita T, Koretsune Y, Yang Y, et al. Edoxaban vs. warfarin in East Asian patients with atrial fibrillation- an ENGAGE AF-TIMI 48 subanalysis. Circ J. 2016;80(4):860–869. doi: 10.1253/circj.CJ-15-1082 [DOI] [PubMed] [Google Scholar]
- 25.Goto S, Zhu J, Liu L, et al. Efficacy and safety of apixaban compared with warfarin for stroke prevention in patients with atrial fibrillation from East Asia: a subanalysis of the Apixaban for Reduction in Stroke and Other Thromboembolic Events in Atrial Fibrillation (ARISTOTLE) Trial. Am Heart J. 2014;168(3):303–309. doi: 10.1016/j.ahj.2014.06.005 [DOI] [PubMed] [Google Scholar]
- 26.Kim HK, Tantry US, Smith SC, Jr., et al. The East Asian paradox: an updated position statement on the challenges to the current antithrombotic strategy in patients with cardiovascular disease. Thromb Haemost. 2021;121(4):422–432. doi: 10.1055/s-0040-1718729 [DOI] [PubMed] [Google Scholar]
- 27.Li YG, Lee SR, Choi EK, et al. Stroke prevention in atrial fibrillation: focus on Asian patients. Korean Circ J. 2018;48(8):665–684. doi: 10.4070/kcj.2018.0190 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Hori M, Matsumoto M, Tanahashi N, et al. Rivaroxaban vs. warfarin in Japanese patients with atrial fibrillation – the J-ROCKET AF study. Circ J. 2012;76(9):2104–2111. doi: 10.1253/circj.cj-12-0454 [DOI] [PubMed] [Google Scholar]
- 29.Menichelli D, Del Sole F, Di Rocco A, et al. Real-world safety and efficacy of direct oral anticoagulants in atrial fibrillation: a systematic review and meta-analysis of 605 771 patients. Eur Heart J Cardiovasc Pharmacother. 2021;7(FI1):f11–f19. doi: 10.1093/ehjcvp/pvab002 [DOI] [PubMed] [Google Scholar]
- 30.Ray WA, Chung CP, Stein CM, et al. Association of rivaroxaban vs apixaban with major ischemic or hemorrhagic events in patients with atrial fibrillation. JAMA. 2021;326(23):2395–2404. doi: 10.1001/jama.2021.21222 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Poli D, Antonucci E, Ageno W, et al. Inappropriate underdosing of direct oral anticoagulants in atrial fibrillation patients: results from the START2-AF registry. JCM. 2024;13(7):2009. doi: 10.3390/jcm13072009 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Valdez Baez FJ, Santana Mejia GM, Suárez Fuster L, et al. Determinants of inappropriate dosing of direct oral anticoagulants in non-valvular atrial fibrillation in a low-income country. Cureus. 2024;16(11):e74526. doi: 10.7759/cureus.74526 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Sanghai S, Wong C, Wang Z, et al. Rates of potentially inappropriate dosing of direct-acting oral anticoagulants and associations with geriatric conditions among older patients with atrial fibrillation: the SAGE-AF study. J Am Heart Assoc. 2020;9(6):e014108. doi: 10.1161/JAHA.119.014108 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Kim JY, Kim J, Park SJ, et al. Comparison of high- and low-dose rivaroxaban regimens in elderly east asian patients with atrial fibrillation. J Korean Med Sci. 2024;39(8):e72. doi: 10.3346/jkms.2024.39.e72 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Shen NN, Zhang C, Wang N, et al. Effectiveness and safety of under or over-dosing of direct oral anticoagulants in atrial fibrillation: a systematic review and meta-analysis of 148909 patients from 10 real-world studies. Front Pharmacol. 2021;12:645479. doi: 10.3389/fphar.2021.645479 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Pereira MQ, David C, Almeida AG, et al. Clinical effects of off-label reduced doses of direct oral anticoagulants: a systematic review and meta-analysis. Int J Cardiol. 2022;362:76–82. doi: 10.1016/j.ijcard.2022.04.062 [DOI] [PubMed] [Google Scholar]
- 37.Liu X, Huang M, Ye C, et al. Effect of non-recommended doses versus recommended doses of direct oral anticoagulants in atrial fibrillation patients: a meta-analysis. Clin Cardiol. 2021;44(4):472–480. doi: 10.1002/clc.23586 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Zhang XL, Zhang XW, Wang TY, et al. Off-label under- and overdosing of direct oral anticoagulants in patients with atrial fibrillation: a meta-analysis. Circ Cardiovasc Qual Outcomes. 2021;14(12):e007971. doi: 10.1161/CIRCOUTCOMES.121.007971 [DOI] [PubMed] [Google Scholar]
- 39.Sang C, Chen J, Sun J, et al. Off-label underdosing of four individual NOACs in patients with nonvalvular atrial fibrillation: a systematic review and meta-analysis of observational studies. Eur J Clin Invest. 2022;52(10):e13819. doi: 10.1111/eci.13819 [DOI] [PubMed] [Google Scholar]
- 40.Sairaku A, Kimura Y, Nakano Y.. Clinical outcomes of off-label DOAC underdosing in Japanese patients with atrial fibrillation: a systematic review and meta-analysis. J Thromb Thrombolysis. 2025;58(6):709–720. doi: 10.1007/s11239-025-03107-0 [DOI] [PubMed] [Google Scholar]
- 41.Chao TF, Chan NY, Chan YH, et al. Direct oral anticoagulant dosing in patients with atrial fibrillation: an Asian perspective. JACC Asia. 2023;3(5):707–723. doi: 10.1016/j.jacasi.2023.08.007 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Chan YH, Chao TF, Lee HF, et al. Different renal function equations and dosing of direct oral anticoagulants in atrial fibrillation. JACC Asia. 2022;2(1):46–58. doi: 10.1016/j.jacasi.2021.11.006 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Homan K, Seeley R, Fisher L, et al. Safety of direct-acting oral anticoagulant (DOAC) prescribing: OpenSAFELY-TPP analysis of 20.5 million adults’ electronic health records. BJGP Open. 2024;8(2):BJGPO.2023.0163. doi: 10.3399/BJGPO.2023.0163 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Gong Y, Jeong YH, Wang TD, et al. Position statement on antiplatelet therapy for East Asians with coronary artery disease. JACC Asia. 2025. ;5(7):821–846. doi: 10.1016/j.jacasi.2025.04.010 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Gibson CM, Mehran R, Bode C, et al. Prevention of bleeding in patients with atrial fibrillation undergoing PCI. N Engl J Med. 2016;375(25):2423–2434. doi: 10.1056/NEJMoa1611594 [DOI] [PubMed] [Google Scholar]
- 46.de Alcântara JPTL, Götz G, Amaral PEO, et al. Factor XI inhibitors versus direct oral anticoagulants for the prevention of thromboembolic events and safety in patients with atrial fibrillation: a systematic review and meta-analysis of randomized controlled trials. Heart Lung. 2026;78:102739. doi: 10.1016/j.hrtlng.2026.102739 [DOI] [PubMed] [Google Scholar]
- 47.Parizad R, Hatwal J, Bodagh H, et al. Factor XI inhibitors versus direct oral anticoagulants for stroke prevention in atrial fibrillation: a systematic review of efficacy and safety. Explor Neuroprot Ther. 2026;6:1004148. doi: 10.37349/ent.2026.1004148 [DOI] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The data that support the findings of this study are available from the corresponding author, upon reasonable request.