Abstract
Background
Valvular heart disease frequently affects older adults and necessitates valve replacement surgery followed by anticoagulation therapy. Although direct oral anticoagulants (DOACs) are increasingly used for atrial fibrillation and venous thromboembolism, they are not approved for use in mechanical heart valves. This study evaluated anticoagulant adherence and clinical outcomes among older patients after valve replacement using Taiwan’s National Health Insurance data.
Methods
This retrospective cohort study used the Taiwan National Health Insurance Research Database and included 4,872 patients aged ≥ 65 years with valvular heart disease who underwent valve replacement between 2010 and 2020. Patients with a history of end-stage renal disease, venous thromboembolism, or ischemic stroke before the index surgery were excluded. Anticoagulant use after discharge was identified, and medication adherence was defined as prescriptions covering ≥ 80% of days during a fixed 180-day post-discharge window. Multivariable logistic regression and Cox proportional hazards models were used to evaluate factors associated with adherence and clinical outcomes, including mortality and readmissions.
Results
Among the included patients, 38.53% did not receive any oral anticoagulant after discharge. Warfarin users demonstrated higher adherence rates than DOAC users. Females and those aged ≥ 75 years showed better adherence. Nonadherence was more common among DOAC users (OR: 1.51; 95% CI: 1.08–2.11) and was associated with higher stroke-related readmissions (DOACs, 7.63%; warfarin, 6.20%), although causal relationships cannot be established. DOAC users had lower observed 1-year mortality rates (4.31% vs. 8.36%) and similar rates of major bleeding events (6.12% vs. 6.34%) compared with warfarin users.
Conclusions
Differences in anticoagulant adherence and clinical outcomes were observed among older patients following valve surgery. Warfarin users showed higher measured adherence, whereas DOAC users had lower observed mortality. Because treatment allocation was not randomized and adherence is closely related to treatment type, the independent effect of adherence on outcomes cannot be completely established. Therefore, these findings should be interpreted cautiously prospective studies should be conducted to elucidate the relationship between adherence and clinical outcomes.
Supplementary Information
The online version contains supplementary material available at 10.1186/s12877-026-07399-6.
Keywords: Valvular heart disease, Anticoagulants, Adherence, Complications, Older adults
Introduction
Valvular heart disease (VHD) involves stenosis or regurgitation in one or more heart valves, with calcification-induced aortic stenosis being the most prevalent, followed by aortic or mitral regurgitation [1]. Artificial heart valves are classified into mechanical and bioprosthetic types. Current guidelines recommend bioprosthetic valves for patients aged ≥ 65 years, making age a key determinant in valve selection [2]. According to guidelines, vitamin K antagonists (VKAs), most commonly warfarin, remain the standard anticoagulant therapy during the early postoperative period after bioprosthetic valve replacement [2, 3]. Nevertheless, studies conducted in Germany, the United States, and Japan have reported inconsistent results regarding the efficacy of direct oral anticoagulants (DOACs) in patients who recently underwent aortic or mitral valve replacement with bioprosthetic valves [4–8]. Furthermore, recent clinical guidelines have cautiously recognized a potential role for DOACs in selected patients with bioprosthetic heart valves and concomitant atrial fibrillation, particularly beyond the early postoperative period and in those with mechanical heart valves [2, 3, 9]. For instance, the RIVER trial reported that DOACs may be considered in stable patients with bioprosthetic mitral valves after surgery [10]. However, these recommendations are nuanced and depend on factors such as valve type and timing after surgery, highlighting continued uncertainty regarding the real-world safety and effectiveness of DOAC use in this population [11].
Warfarin requires regular monitoring of the international normalized ratio (INR) to ensure therapeutic effectiveness. Although this requirement increases treatment complexity, routine INR monitoring also serve as a mechanism to verify appropriate drug use and reinforce medication adherence, which may affect clinical outcomes [2, 4, 12]. In contrast, DOAC therapy does not involve routine laboratory monitoring, which raises concerns regarding unrecognized nonadherence in real-world settings. Despite guideline preferences for warfarin, a small but clinically relevant proportion of patients in routine practice receive DOACs, indicating an evidence gap regarding adherence and outcomes in this population. Poor adherence to anticoagulation therapy may result in serious complications, including major bleeding, thromboembolic events, intracardiac thrombus formation, and postoperative atrial fibrillation (Af), thereby increasing the risk of rehospitalization and adversely affecting the quality of life [13, 14].
Taiwan’s National Health Insurance Research Database (NHIRD) provides a unique opportunity to investigate real-world treatment patterns and outcomes in a large, population-based cohort. Using this nationwide database, the present study aimed to (1) evaluate anticoagulant adherence in older patients after valve surgery, (2) identify factors associated with nonadherence, and (3) examine the associations between adherence patterns and 1-year clinical outcomes, including mortality and hospital readmissions.
Methods
This retrospective cohort study included adults aged ≥ 65 years who were diagnosed with VHD between January 1, 2009, and December 31, 2020.
Data sources
Data were extracted from Taiwan’s National Health Insurance (NHI) program, which covers > 99% of the population [15], via the Health and Welfare Data Science Center (HWDC), which contains deidentified inpatient and outpatient claims. Diagnoses, procedures, and prescriptions were identified from the NHIRD; detailed codes are provided in the supplementary material. Anticoagulant prescription data were retrieved using Anatomical Therapeutic Chemical classification codes. A maximum of three primary diagnoses and procedures, along with the patient’s date of medical visit, pharmacy refill records, prescription dates, laboratory items, and basic sociodemographic information, including insured area, and monthly insurance salary, were obtained. Laboratory parameters, including renal function indices and INR values, were not available in the database.
VHD cohort
The cohort included patients with a primary or secondary diagnosis of VHD, defined as aortic stenosis and aortic regurgitation or mitral stenosis and mitral regurgitation. To ensure accurate case identification, patients were required to have at least one outpatient visit or one hospitalization with a relevant VHD diagnostic code prior to valve replacement surgery. When multiple VHD diagnoses were recorded, the most recent diagnosis before surgery was used to classify VHD type. Patients who underwent surgical valve replacement were identified using procedure codes [16]. The exclusion criteria were as follows: (1) a history of end-stage renal disease, hemodialysis, venous thromboembolism, ischemic stroke, or transcatheter aortic valve implantation (TAVI), (2) missing or unknown sex information, (3) an observation period of < 270 days, (4) insufficient prescription data for exposure assessment.
To ensure accurate exposure classification and to minimize immortal time bias, we predefined a 270-day exposure ascertainment window following hospital discharge. This window comprised the first 90 days after discharge to confirm continued warfarin use, followed by an additional 180-day reference period used to quantify medication adherence and classify patients into adherence groups. All patients were required to have at least 270 days of continuous post-discharge claims data. Individuals with follow-up shorter than 270 days were excluded, as illustrated in the Fig. 1.
Fig. 1.
Flowchart for the enrollment of the study cohort
Clinical outcome follow-up did not begin at the time of hospital discharge. Instead, the index date (time zero) was defined as 270 days after discharge, corresponding to the completion of the exposure ascertainment window. All subsequent follow-up time, incidence rate calculations, and survival analyses were conducted from this index date onward. Events occurring within the 270-day ascertainment period were not included in the outcome analyses.
Outcomes and exposure
The primary outcome was adherence to anticoagulation therapy after valve surgery. The outpatient drug file was used to track the receipt of warfarin and four DOACs (apixaban, rivaroxaban, edoxaban, or dabigatran) post-valve surgery. The treatment period for each drug was calculated, and the interval from prescription to the occurrence of any adverse event was determined. Medication adherence was assessed using the proportion of days covered (PDC). The adherence observation window was defined as the first 180 days following hospital discharge [17]. PDC was calculated as the total number of days with available anticoagulant supply divided by 180 days. Patients with a PDC ≥ 80% were classified as adherent [18]. Because the database did not reliably distinguish valve type (mechanical vs. bioprosthetic) or postoperative indications for long-term anticoagulation (e.g., Af), a fixed 180-day postdischarge window was used to assess adherence. This approach allowed to capture of overall early postoperative anticoagulant use patterns in a heterogeneous elderly population. Although this time frame does not imply that all patients required anticoagulation therapy for 6 months but was selected to standardize exposure assessment across patients with varying clinical indications. Adherence was calculated over the entire 180-day window, irrespective of the timing of adverse events. Clinical outcomes were analyzed separately using time-to-event methods. Patients who switched anticoagulant agents during the 180-day adherence assessment period were excluded from the adherence analysis to ensure consistent exposure measurement. Anticoagulants included warfarin, apixaban, rivaroxaban, edoxaban, or dabigatran, which were recorded if administered during inpatient and outpatient stays. Secondary outcomes included 1-year mortality rates, cardiovascular (CV)-related readmission, and stroke-related readmission. One-year mortality rates were calculated from the date of discharge for VHD. All inpatient, outpatient, and emergency visits were reviewed using electronic health records to determine overall and cause-specific readmission rates. Models were adjusted for age, sex, and major comorbidities. Two distinct analytic time frames were applied in this study for different purposes. Medication adherence was evaluated over a fixed 180-day window following hospital discharge to standardize exposure measurement. Clinical outcomes were evaluated using time-to-event follow-up beginning at discharge. These time frames were analyzed separately and not combined.
Patients were categorized into the following three groups according to prescriptions after discharge: warfarin users, DOAC users, and patients who did not receive any oral anticoagulant (“no anticoagulation” group). The no-anticoagulation group was included in the descriptive analyses to improve the clinical interpretability of treatment patterns. These patients may represent heterogeneous clinical scenarios, including bioprosthetic valve recipients without Af, patients with contraindications to anticoagulation, or individuals with frailty or limited life expectancy. Because medication adherence was defined based on prescription coverage, adherence analyses were restricted to patients who received at least one anticoagulant prescription.
Confounders
Baseline covariates included age at VHD diagnosis, sex, VHD type, and adherence to anticoagulation therapy. Baseline comorbidities were identified using diagnosis codes recorded within the year preceding the index valve surgery. Comorbidities included hypertension, diabetes mellitus, atrial fibrillation (Af), congestive heart failure, transient ischemic attack, vascular disease or peripheral embolism, hyperlipidemia, peptic ulcer disease, coronary artery disease, liver disease, renal disease, dementia, and myocardial infarction. All ICD9 and ICD10 codes are provided in Supplementary Table 1.)
Ethics approval and consent to participate
The study was approved by the Institutional Review Board of National Cheng Kung University Hospital (No. B-ER-111-214) and conducted according to the principles of the Declaration of Helsinki. The study used data from the HWDC Database in Taiwan, maintained by the NHRI. Because the data comprised only deidentified secondary information, the Institutional Review Board of National Cheng Kung University Hospital waived the requirement for informed consent.
Statistical analysis
Frequencies and descriptive statistics were used to summarize the baseline characteristics. Categorical variables were presented as counts with percentages and compared using the chi-square test or Fisher’s exact test, as appropriate.
Anticoagulant adherence was defined as prescription coverage ≥ 80% during a predefined 180-day observation period within the 270-day post-discharge exposure ascertainment window. Covariates included age, sex, Charlson comorbidity index, atrial fibrillation, hypertension, diabetes mellitus, heart failure, renal disease, and other clinically relevant conditions. These variables represent components or proxies of commonly used thromboembolic and bleeding risk scores.
Univariable and multivariable logistic regression analyses were performed to identify factors associated with medication adherence. Mortality and readmission risks were evaluated by estimating cumulative incidence using the Kaplan-Meier method, and differences between groups were assessed using the log-rank test. Incidence rates of clinical outcomes were calculated as events per person-year of follow-up, and Poisson regression models were used to compare differences in incidence rates between groups. All multivariable models were adjusted for age, sex, and relevant comorbidities.
In accordance with the predefined study design, follow-up for all clinical outcomes began 270 days after discharge from the index valve surgery hospitalization, which was defined as the index date (time zero). Only events occurring after this index date were included in the outcome analyses. Patients were followed from time zero until the occurrence of the outcome of interest, death, withdrawal from the National Health Insurance program, or the end of the study period, whichever came first. Person-years were calculated from time zero to censoring. All statistical analysis was performed using SAS 9.4 software (SAS Institute, Cary, NC, USA). P < 0.05 was considered significant.
Results
Overall, 20,361 individuals aged ≥ 65 years who were diagnosed with VHD between 2010 and 2020 were initially identified. After applying the exclusion criteria, 4,872 patients were included in the final analysis (Table 1). The cohort consisted of 2,413 male (49.5%) and 2,459 female (50.5%) patients (mean age, 73.02 years). Of the patients, 2,805 (57.57%) used warfarin, 190 (3.90%) used DOACs, and 1,877 (38.53%) did not receive any anticoagulants. The DOAC group predominantly patients from the northern region, with smaller proportions from the central and southern regions (p < 0.0001). This group also had higher rates of concomitant conditions such as hypertension and Af; however, the proportions of diabetes and vascular diseases did not differ significantly. Patients in the no-anticoagulation group were generally older and had a higher burden of comorbidities than anticoagulant users, suggesting that the clinical risk profiles could have influenced treatment selection.
Table 1.
Baseline characteristics of older patients undergoing valve surgery (all patients) (N = 4872)
| Characteristic | Valve surgery cohort | ||||
|---|---|---|---|---|---|
| Overall(n = 4872) | DOAC | Warfarin | No anticoagulation | ||
| (n = 190, 3.90%) | (n = 2805, 57.57%) | (n = 1877, 38.53%) | p-value | ||
| Age, years | 73.02 (5.57) | 73.27 (6.07) | 73.08 (5.65) | 72.90 (5.39) | 0.4742 |
| Age group | |||||
| 65–74 | 3230 (66.30) | 122 (64.21) | 1862 (66.38) | 1246 (66.38) | 0.0006 |
| 75–84 | 1508 (30.95) | 58 (30.53) | 848 (30.23) | 602 (32.07) | |
| 85 + | 134 (2.75) | 10 (5.26) | 95 (3.39) | 29 (1.55) | |
| Sex | 0.0013 | ||||
| Male | 2413 (49.53) | 93 (48.95) | 1329 (47.38) | 991 (52.80) | |
| Female | 2459 (50.47) | 97 (51.05) | 1476 (52.62) | 886 (47.20) | |
| Geographic region | < .0001 | ||||
| Northern | 2906 (59.65) | 136 (71.58) | 1589 (56.65) | 1181 (62.92) | |
| Central | 871 (17.88) | 18 (9.47) | 541 (19.29) | 312 (16.62) | |
| Southern | 1049 (21.53) | 35 (18.42) | 647 (23.07) | 367 (19.55) | |
| Eastern | 46 (0.94) | < 3* (-) | 28 (1.00) | 17 (0.91) | |
| Monthly income (NTD) | 0.8566 | ||||
| ≤20100 | 1483 (30.44) | 54 (28.42) | 849 (30.27) | 580 (30.90) | |
| 20101–30300 | 2297 (47.15) | 88 (46.32) | 1324 (47.20) | 885 (47.15) | |
| ≥30301 | 1092 (22.41) | 48 (25.26) | 632 (22.53) | 412 (21.95) | |
| Type of VHD | < .0001 | ||||
| Aortic valve stenosis, aortic regurgitation | 1134 (23.28) | 53 (27.89) | 480 (17.11) | 601 (32.02) | |
| Mitral valve stenosis | 3738 (76.72) | 137 (72.11) | 2325 (82.89) | 1276 (67.98) | |
| Comorbidities | |||||
| HTN | 3078 (63.18) | 137 (72.11) | 1648 (58.75) | 1293 (68.89) | < .0001 |
| DM | 1258 (25.82) | 47 (24.74) | 707 (25.20) | 504 (26.85) | 0.4247 |
| Af | 2214 (45.44) | 148 (77.89) | 1448 (51.62) | 618 (32.92) | < .0001 |
| CHF | 3039 (62.38) | 135 (71.05) | 1720 (61.32) | 1184 (63.08) | 0.0200 |
| TIA | 102 (2.09) | 12 (6.32) | 57 (2.03) | 33 (1.76) | 0.0002 |
| Vascular disease/peripheral Embolism | 231 (4.74) | 4 (2.11) | 136 (4.85) | 91 (4.85) | 0.2185 |
| Hyperlipidemia | 1257 (25.80) | 48 (25.26) | 672 (23.96) | 537 (28.61) | 0.0017 |
| Peptic ulcer disease | 656 (13.46) | 30 (15.79) | 382 (13.62) | 244 (13.00) | 0.5255 |
| Coronary artery disease | 2309 (47.39) | 87 (45.79) | 1207 (43.03) | 1015 (54.08) | < .0001 |
| Liver disease | 419 (8.60) | 22 (11.58) | 220 (7.84) | 177 (9.43) | 0.0541 |
| Renal disease | 463 (9.50) | 33 (17.37) | 236 (8.41) | 194 (10.34) | < .0001 |
| Dementia | 120 (2.46) | 7 (3.68) | 66 (2.35) | 47 (2.50) | 0.5132 |
| MI | 224 (4.60) | 5 (2.63) | 114 (4.06) | 105 (5.59) | 0.0208 |
| CCI category | 0.3466 | ||||
| 0 | 574 (11.78) | 16 (8.42) | 344 (12.26) | 214 (11.40) | |
| 1–2 | 2812 (57.72) | 109 (57.37) | 1627 (58.00) | 1076 (57.33) | |
| ≥ 3 | 1486 (30.50) | 65 (34.21) | 834 (29.73) | 587 (31.27) | |
Data are presented as mean (standard deviation). Af atrial fibrillation, CCI Charlson comorbidity index, CHF congestive heart failure, DM diabetes mellitus, HTN hypertension, MI myocardial infarction, NTD New Taiwan Dollar, TIA transient ischemic attack
*In compliance with National Health Insurance database regulations, counts <3 are suppressed to protect patient confidentiality
In patients aged 65–74 years, adherence to anticoagulation therapy was significantly higher among warfarin users (35.45%) than among DOAC users (24.59%, p = 0.0147) (Table 2). Patients with aortic valve stenosis or regurgitation exhibited a warfarin adherence rate of 27.08%, whereas those with mitral valve stenosis had an adherence rate of 36.04%; however, those differences were not significant (p = 0.1972 and p = 0.1465, respectively). Adherence to DOACs was significantly lower among male patients (p = 0.0229).
Table 2.
Univariable analysis comparing the difference in the adherence rate among DOAC and warfarin users
| Characteristic | Valve surgery cohort | ||
|---|---|---|---|
| DOAC (apixaban, edoxaban, dabigatran, and rivaroxaban)(n = 190) | Warfarin(n = 2805) | P-value | |
| Age group, years | |||
| 65–74 | 30 (24.59) | 660 (35.45) | 0.0147 |
| 75–84 | 18 (31.03) | 270 (31.84) | 0.8986 |
| 85 + | 3 (30.00) | 38 (40.00) | 0.7368* |
| Sex | |||
| Male | 21 (22.58) | 453 (34.09) | 0.0229 |
| Female | 30 (30.93) | 515 (34.89) | 0.4268 |
| Geographic region | |||
| Northern | 34 (25.00) | 503 (31.66) | 0.1077 |
| Central | 8 (44.44) | 199 (36.78) | 0.5079 |
| Southern | 9 (25.71) | 248 (38.33) | 0.1336 |
| Eastern | 0 (0.00) | 18 (64.29) | 0.3793* |
| Monthly income (NTD) | |||
| ≤20100 | 19 (35.19) | 297 (34.98) | 0.9758 |
| 20101–30300 | 19 (21.59) | 444 (33.53) | 0.0208 |
| ≥30301–40100 | 13 (27.08) | 227 (35.92) | 0.2169 |
| Type of VHD | |||
| Aortic valve stenosis, aortic regurgitation | 10 (18.87) | 130 (27.08) | 0.1972 |
| Mitral valve stenosis | 41 (29.93) | 838 (36.04) | 0.1465 |
| Comorbidities | |||
| HTN | 40 (29.20) | 546 (33.13) | 0.3461 |
| DM | 9 (19.15) | 256 (36.21) | 0.0177 |
| Af | 45 (30.41) | 535 (36.95) | 0.1150 |
| CHF | 39 (28.89) | 601 (34.94) | 0.1543 |
| TIA | 3 (25.00) | 17 (29.82) | 1.0000* |
| Vascular disease/peripheral Embolism | <3** (-) | 46 (33.82) | 1.0000* |
| Hyperlipidemia | 10 (20.83) | 235 (34.97) | 0.0458 |
| Peptic ulcer disease | 8 (26.67) | 132 (34.55) | 0.3797 |
| Coronary artery disease | 28 (32.18) | 396 (32.81) | 0.9046 |
| Liver disease | 6 (27.27) | 72 (32.73) | 0.6017 |
| Renal disease | 12 (36.36) | 70 (29.66) | 0.4334 |
| Dementia | 0 (0.00) | 24 (36.36) | 0.0877* |
| MI | < 3** (-) | 40 (35.09) | 1.0000* |
| CCI category | |||
| 0 | 5 (31.25) | 121 (35.17) | 0.7477 |
| 1–2 | 23 (21.10) | 547 (33.62) | 0.0071 |
| ≥ 3 | 23 (35.38) | 300 (35.97) | 0.9244 |
Af atrial fibrillation, CCI Charlson comorbidity index, CHF congestive heart failure, DM diabetes mellitus, DOAC direct oral anticoagulant, HTN hypertension, MI myocardial infarction, TIA transient ischemic attack
*Fisher’s exact test
< 3** (-): In compliance with National Health Insurance database regulations, counts < 3 are suppressed to protect patient confidentiality
The risk factors associated with nonadherence to anticoagulant therapy among patients who underwent valve surgery are presented in Table 3. Male patients showed no significant difference in adherence compared with female patients (aOR 1.03, 95% CI 0.88–1.20, p = 0.72). DOAC users were more likely to be nonadherent than warfarin users (aOR 1.51, 95% CI 1.08–2.11, p = 0.0162), indicating significantly higher adherence to warfarin therapy. Af was associated with significantly higher adherence rates (aOR 0.79, 95% CI 0.68–0.92, p = 0.0028) (Table 3).
Table 3.
Multivariable logistic regression analysis of risk factors associated with nonadherence in patients undergoing valve surgery
| Variables | Crude OR | 95% CI | P-value | Adjusted OR | 95% CI | P-value |
|---|---|---|---|---|---|---|
| Sex | ||||||
| Male | 1.06 | (0.91–1.23) | 0.4488 | 1.03 | (0.88–1.20) | 0.7211 |
| Female | Ref. | Ref. | ||||
| Age (years) | ||||||
| 65–74 | Ref. | Ref. | ||||
| 75–84 | 1.14 | (0.97–1.35) | 0.1151 | 1.12 | (0.95–1.33) | 0.1817 |
| 85 + | 0.83 | (0.56–1.25) | 0.3719 | 0.78 | (0.52–1.18) | 0.2392 |
| Oral anticoagulant | ||||||
| DOAC | 1.44 | (1.03–2.00) | 0.0317 | 1.51 | (1.08–2.11) | 0.0162 |
| Warfarin | Ref. | Ref. | ||||
| DM | 0.94 | (0.79–1.11) | 0.4520 | 0.90 | (0.73–1.11) | 0.3293 |
| Af | 0.80 | (0.69–0.93) | 0.0043 | 0.79 | (0.68–0.92) | 0.0028 |
| HTN | 1.14 | (0.98–1.33) | 0.0939 | 1.10 | (0.93–1.28) | 0.2587 |
| Hyperlipidemia | 1.00 | (0.84–1.19) | 1.0000 | 1.01 | (0.82–1.26) | 0.8984 |
| Coronary artery disease | 1.10 | (0.95–1.29) | 0.2055 | 1.07 | (0.91–1.25) | 0.4139 |
| Vascular disease/peripheral embolism | 1.02 | (0.71–1.46) | 0.9083 | 1.00 | (0.69–1.43) | 0.9820 |
| Dementia | 1.05 | (0.64–1.73) | 0.8355 | 1.01 | (0.61–1.66) | 0.9757 |
| Liver disease | 1.09 | (0.82–1.45) | 0.5395 | 1.09 | (0.82–1.44) | 0.5712 |
Adjusted for sex, age, and comorbidities
Af atrial fibrillation, CI confidence interval, DOAC direct oral anticoagulants, DM diabetes mellitus, HTN hypertension, OR odds ratio
The clinical outcomes are presented in Table 4. The rates of major gastrointestinal bleeding were higher in warfarin users (6.34%) than in DOAC users (6.12%). The 1-year incidence rate (IR) for all-cause mortality was 6.77% across the cohort, with lower rates among DOAC users (4.31%) than among warfarin users (8.36%). Cox proportional hazards models identified advanced age (≥ 85 years, adjusted HR, 3.85; 95% CI, 2.96–5.01) and comorbid dementia (adjusted HR, 1.71; 95% CI, 1.18–2.49) as significant predictors of mortality. The readmission IR was 31.23% across the cohort, which was higher among DOAC users (41.57%) than among warfarin users (36.46%). The CV-related readmission IR was 3.82% across the cohort and was lower among DOAC users (3.03%) than among warfarin users (4.53%). The stroke-related readmission IR was 5.10% across the cohort and was higher among DOAC users (7.63%) than among warfarin users (6.20%).
Table 4.
Clinical outcomes during the first year after valve surgery
| Outcomes | All (N = 4872) | DOAC-based therapy | Warfarin-based therapy | |||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Events | Person-years of follow-up | Incidence Rate (per 100 person-year) | Events | Person-years of follow-up | Incidence Rate (per 100 person-year) | Events | Person-years of follow-up | Incidence Rate (per 100 person-year) | p-value | |
| Deep venous thrombosis | 15 | 4496 | 0.33 (0.33–0.34) | 0 | 168 | 0.00 | 11 | 2588 | 0.42 (0.42–0.43) | - |
| Pulmonary embolism | 1 | 4502 | 0.02 (0.02–0.02) | 0 | 168 | 0.00 | 0 | 2594 | 0.00 | - |
| Ischemic stroke | 108 | 4450 | 2.43 (2.42–2.43) | 5 | 166 | 3.01 (2.81–3.22) | 76 | 2557 | 2.97 (2.96–2.99) | 0.7736 |
| Peripheral embolism | 25 | 4491 | 0.56 (0.55–0.56) | 0 | 168 | 0.00 | 21 | 2584 | 0.81 (0.81–0.82) | - |
| Hemorrhagic stroke | 44 | 4483 | 0.98 (0.98–0.99) | 0 | 168 | 0.00 | 35 | 2578 | 1.36 (1.35–1.37) | - |
| Major bleeding, gastrointestinal | 211 | 4410 | 4.79 (4.78–4.79) | 10 | 163 | 6.12 (5.82–6.41) | 160 | 2524 | 6.34 (6.32–6.36) | 0.3129 |
| Intracranial bleeding, Traumatic | 36 | 4487 | 0.80 (0.80–0.81) | 3 | 167 | 1.80 (1.64–1.96) | 27 | 2582 | 1.05 (1.04–1.05) | < 0.0001 |
| Major bleeding, other | 209 | 4407 | 4.74 (4.73–4.75) | 12 | 163 | 7.38 (7.05–7.70) | 148 | 2525 | 5.86 (5.84–5.88) | < 0.0001 |
| Death | 1262 | 18,652 | 6.77 (6.76–6.77) | 18 | 417 | 4.31 (4.22–4.41) | 920 | 11,002 | 8.36 (8.36–8.37) | < 0.0001 |
| Readmission | 3130 | 10,023 | 31.23 (31.22–31.24) | 98 | 236 | 41.57 (41.03–42.11) | 1970 | 5403 | 36.46 (36.44–36.49) | < 0.0001 |
| CV-related readmission | 662 | 17,330 | 3.82 (3.82–3.82) | 12 | 396 | 3.03 (2.94–3.11) | 456 | 10,076 | 4.53 (4.52–4.53) | < 0.0001 |
| Stroke-related readmission | 858 | 16,809 | 5.10 (5.10–5.11) | 28 | 367 | 7.63 (7.48–7.78) | 601 | 9688 | 6.20 (6.20–6.21) | < 0.0001 |
The outcomes (death, readmission, CV-related readmission, and Stroke-related readmission) were assessed beginning 270 days after hospital discharge (defined as the index date). Patients were followed from the index date until the occurrence of the outcome of interest, death, withdrawal from the National Health Insurance program, or December 31, 2020, whichever came first. For 1-year outcome analyses, follow-up was limited to 365 days after the index date. CV cardiovascular event, DOAC direct oral anticoagulant
Kaplan–Meier curves were used to illustrate time-to-event patterns according to anticoagulant adherence status (Fig. 2A and E). Nonadherent patients demonstrated lower event-free survival than adherent patients during follow-up. Consistent with the Kaplan–Meier analyses, adjusted Poisson regression models showed that non-adherence to anticoagulation therapy was significantly associated with increased incidence rates of adverse outcomes (Table 4). Associations were quantified using incidence rate ratios (IRRs) with 95% confidence intervals.
Fig. 2.
Kaplan–Meier curves showing event-free survival according to anticoagulant adherence status during follow-up. Time was measured from the date of hospital discharge. Numbers at risk are shown at prespecified time points. (A) Overall survival. (B) Freedom from major gastrointestinal bleeding. (C) Freedom from other major bleeding events. (D) Freedom from cardiovascular-related readmission. (E) Freedom from stroke-related readmission
Discussion
Adherence to anticoagulation therapy is a critical determinant of clinical outcomes following valve replacement. In our population-based cohort of older adults, warfarin users aged 65–74 years demonstrated higher adherence (35.45%) than DOAC users (24.59%). Poor adherence was consistently associated with adverse clinical outcomes. The relatively lower adherence among DOAC users may reflect differences in prescription practices and patient education, although DOACs do not require routine INR monitoring. Previous studies have demonstrated that adherence to DOACs varies across healthcare systems and is affected by patient education and caregiver involvement [5]. Banerjee et al. reported that women generally exhibit higher medication adherence than men and that older adults may benefit from caregiver support, whereas cognitive decline and polypharmacy may hinder adherence [19]. These findings support our observation that patient education and structured follow-up are important for ensuring adherence, even with DOAC. It should also be noted that the relatively low adherence rate observed at 180 days may partly reflect differences in clinical indications for oral anticoagulation following valve replacement. Contemporary guidelines generally recommend anticoagulation for the first three months after bioprosthetic valve implantation, whereas long-term anticoagulation is primarily indicated for mechanical valves or specific comorbid conditions [2, 9]. Because the database used in this study does not provide information on valve type, it is possible that some patients appropriately discontinued anticoagulation therapy after the early postoperative period according to physician recommendations. In such cases, treatment discontinuation may be misclassified as non-adherence in the claims-based adherence measure.
A substantial proportion of patients did not receive anticoagulation after valve surgery. This subgroup likely represents a clinically distinct population with lower thromboembolic risk, higher bleeding risk, or advanced frailty [9]. Because treatment allocation in observational studies is nonrandomized, baseline differences and unmeasured factors may introduce selection bias or confounding by indication [20]. Therefore, differences in clinical outcomes between treatment groups should be interpreted cautiously, as part of the observed effect may reflect underlying patient risk profiles rather than pharmacologic properties alone. Inclusion of the no-anticoagulation group enhances external validity and reflects real-world geriatric practice [21].
Because anticoagulant exposure was treated as a time-fixed variable after discharge, immortal time bias cannot be completely excluded. Patients must survive until prescription initiation, potentially underestimating event rates in the treated groups. Although anticoagulant therapy is typically initiated shortly after discharge in clinical practice, the potential of residual bias should be acknowledged when interpreting associations between anticoagulant type and outcomes.
Female patients demonstrated higher adherence rates than male patients, which is consistent with previous findings of sex-related differences in health-seeking behavior [12]. The higher proportion of DOAC prescriptions in northern Taiwan suggests regional disparities, including greater access to tertiary medical centers and subspecialty care [22]. Af emerged as a protective factor against nonadherence (aOR, 0.79, p = 0.0028), possibly reflecting heightened awareness among patients of their elevated thromboembolic risk and prompting better compliance [23].
Although warfarin users demonstrated higher adherence, DOAC users exhibited lower 1-year mortality (4.31% vs. 8.36%) and slightly lower rates of major gastrointestinal bleeding (6.12% vs. 6.34%). This apparent discrepancy should be interpreted with causion. In observational studies, treatment selection is influenced by clinical characteristics, and DOAC recipients may differ systematically from warfarin users in frailty, functional status, or other unmeasured prognostic factors [24, 25]. Pharmacological differences may also contribute, as DOACs have more predictable pharmacokinetics and are associated with lower rates of intracranial bleeding, which may affect survival independently of measured adherence [26]. In contrast, although INR monitoring may increase measured adherence among warfarin users, it does not completely mitigate comorbidity-related risks in older populations [27]. Moreover, adherence may have different clinical implications across anticoagulant classes. Warfarin has a narrow therapeutic window and depends on sustained INR control, whereas DOACs exhibit more predictable pharmacokinetics and may be less sensitive to short treatment interruptions [28, 29]. Therefore, identical adherence metrics may not translate into equivalent clinical consequences across treatment groups. Additionally, adherence patterns may vary depending on medication type. The present study focused specifically on oral anticoagulants, and adherence to other chronic medications was not evaluated. Future research may consider comparing adherence across different medication classes to better understand whether the observed patterns are medication-specific or reflect broader medication-taking behaviors in older adults.
Traumatic intracranial bleeding was less frequent among DOAC users than among warfarin users (1.80% vs. 1.05%), which is consistent with previous studies reporting fewer severe bleeding events in bioprosthetic valve populations [6, 7]. Nevertheless, DOAC users exhibited higher stroke-related readmission rates despite lower overall mortality. Several mechanisms may explain this observation, which should be interpreted cautiously. First, lower mortality among DOAC users has been consistently reported in randomized trials and meta-analyses, largely driven by reduced fatal bleeding [24]. Second, stroke-related admissions may include events of varying severity, whereas mortality is strongly affected by fatal bleeding, infection, and comorbidity burden. Third, real-world underdosing or off-label dose reductions common in Asian older adults, may increase thromboembolic risk and while reducing bleeding risk [30–33]. Finally, residual confounding and treatment selection bias may persist, as patients prescribed DOACs may have lower frailty or better functional status, which are factors incompletely captured in claims data [24]. Overall, the coexistence of lower mortality and higher stroke-related readmission is not necessarily contradictory in observational geriatric populations, where different mechanisms affect survival and nonfatal events.
The improved safety profiles of DOACs, particularly lower rates of severe bleeding, may partially account for these observations. Nonetheless, these findings differ from those of Yokoyama et al., who reported that DOACs demonstrated comparable efficacy to warfarin in preventing ischemic stroke in patients with Af [8].
The 2021 European Society of Cardiology (ESC) and European Association for Cardio-Thoracic Surgery (EACTS) guidelines emphasize that anticoagulation management in older patients requires careful balancing of thromboembolic and bleeding risks [9]. In this vulnerable population, medication adherence is a marker of treatment effectiveness and safety. Suboptimal adherence may cause catastrophic ischemic or bleeding events more readily in older patients than in younger patients. Our findings provide real-world evidence supporting individualized long-term anticoagulation strategies for geriatric patients who underwent valve surgery.
Lower CV-related readmission rates among DOAC users (3.03% vs. 4.53%) along with a reduced incidence of peripheral embolism and deep vein thrombosis further support the pharmacokinetic stability of DOACs [34, 35]. These advantages may contribute to improved clinical stability in older adults.
The use of Taiwan’s NHIRD enabled the evaluation of adherence patterns and outcomes in a large, representative geriatric valve surgery cohort, improving the generalizability of our findings to routine clinical practice.
Limitations
This study has several key limitations. First, this was a retrospective, nonrandomized analysis using secondary claims data; therefore, causal inference cannot be established, and residual confounding related to treatment selection cannot be fully excluded despite multivariable adjustment. Differences in baseline characteristics between DOAC and warfarin users may reflect underlying clinical decision-making rather than the effects of the treatments. Second, mortality was common in this older population, and death may act as a competing risk for nonfatal outcomes such as stroke-related readmission, CV-related readmission, and major bleeding. Because competing risk analyses were not performed, the HRs derived from Cox models should be interpreted cautiously as descriptive associations rather than causal estimates. Third, the health insurance database did not provide information to reliably distinguish mechanical from bioprosthetic valves. This limitation is clinically important because DOACs are not recommended for patients with mechanical heart valves, and misclassification of valve type could affect the interpretation of safety and effectiveness outcomes. Although DOAC prescriptions after valve surgery are more commonly observed in patients with bioprosthetic valves in real-world practice, the possibility of inappropriate prescribing or coding limitations cannot be completely excluded. Fourth, although Af was included as a baseline comorbidity, patients with a previous history of venous thromboembolism or ischemic stroke were excluded to reduce heterogeneity related to alternative indications for anticoagulation. This approach may have resulted in a more selected postoperative valve surgery population, potentially limiting the generalizability of the findings to all patients receiving anticoagulation therapy after valve surgery. Fifth, important clinical variables such as frailty status, laboratory parameters, body weight, renal function trends, and validated thromboembolic or bleeding risk scores (e.g., CHA₂DS₂-VASc or HAS-BLED) were not directly available. Although several comorbidities were included as proxy variables, residual confounding due to unmeasured factors such as disease severity and functional status remains possible. Furthermore, we could not fully adjust for concomitant medications such as antiplatelet agents, nonsteroidal anti-inflammatory drugs, or corticosteroids, which may influence bleeding risk and anticoagulation management. Therefore, residual confounding from unmeasured medication exposure cannot be completely excluded.
Sixth, medication adherence was estimated using prescription claims data, which reflects medication dispensing rather than actual intake. Factors such as cognitive function, functional ability, social support, and patient preferences that may affect adherence could not be evaluated. In addition, the claims database does not provide information on valve type (mechanical vs. bioprosthetic) or physician treatment intentions. Current clinical guidelines generally recommend anticoagulation for approximately the first three months after bioprosthetic valve replacement, whereas longer treatment is typically required for mechanical valves. Consequently, some patients who discontinued anticoagulation after the early postoperative period may have done so appropriately according to physician recommendations. In the present analysis, such cases could have been misclassified as reduced adherence when measured over the 180-day observation period. Finally, because anticoagulant exposure was classified after discharge and treated as a time-fixed variable, immortal time bias cannot be completely excluded. Although anticoagulant therapy is typically initiated shortly after discharge in routine practice, which may reduce the magnitude of this bias, it may still result in underestimation of event rates in the treated groups.
In addition, this study used NHIRD data available through 2020; therefore, more recent changes in anticoagulation practice may not be fully captured. Nevertheless, anticoagulation management after valve surgery has not fundamentally changed during this period, and the findings remain relevant to contemporary clinical practice. Future studies using clinical registries or prospective designs with more detailed procedural information, laboratory data, and time-varying exposure modeling are warranted to validate these findings and further clarify the relationship between anticoagulant use and medication adherence.
Conclusions
In this nationwide cohort of older patients undergoing valve surgery, differences in anticoagulant adherence and clinical outcomes were observed between treatment groups. Warfarin users showed higher measured adherence, whereas DOAC users had lower observed mortality but differing readmission patterns. Because treatment allocation was not randomized and adherence is closely linked to treatment type, the independent effect of adherence on clinical outcomes cannot be fully established. Therefore, these observed differences should be interpreted cautiously and may reflect a combination of patient selection, pharmacologic differences, and residual confounding. Future prospective studies incorporating time-varying exposure assessment are required to better elucidate the relationship between adherence and clinical outcomes.
Supplementary Information
Authors’ contributions
Jun-Neng Roan designed the study, interpreted the data, and contributed to article revision. Yu-Sheng Hu contributed to article revision. Tzu-Jung Chuang performed statistical analyses and contributed to article revision. Yi-Lin Wu designed the study and contributed to article revision. Yi-Ching Yang contributed to article revision. Han-Chang Ku designed the study, interpreted the data, wrote the first draft of the article, and contributed to article revision.
Funding
This study was supported in part by grants from the National Health Research Institute (NHRI-13A1-CG-CO-04-2225-1).
Data availability
The datasets generated and/or analyzed in this study are not publicly available due to legal and ethical restrictions imposed by the HWDC, Ministry of Health and Welfare, Taiwan. However, access to the data can be requested from the HWDC (https://dep.mohw.gov.tw/DOS/cp-2516-59203-113.html) subject to approval from the Ministry of Health and Welfare and institutional ethics committees.
Declarations
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Sable C. Update on prevention and management of rheumatic heart disease. Pediatr Clin North Am. 2020;67:843–53. [DOI] [PubMed] [Google Scholar]
- 2.Otto CM, Nishimura RA, Bonow RO, et al. 2020 ACC/AHA Guideline for the Management of Patients With Valvular Heart Disease: Executive Summary: A Report of the American College of Cardiology/American Heart Association Joint Committee on Clinical Practice Guidelines. Circulation. 2021;143:e35–71. [DOI] [PubMed] [Google Scholar]
- 3.Ryu R, Tran R. DOACs in mechanical and bioprosthetic heart valves: a narrative review of emerging data and future directions. Clin Appl Thromb Hemost. 2022;28:10760296221103578. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Gerfer S, Djordjevic I, Eghbalzadeh K, Mader N, Wahlers T, Kuhn E. Direct oral anticoagulation in atrial fibrillation and heart valve surgery-a meta-analysis and systematic review. Ther Adv Cardiovasc Dis. 2022;16:17539447221093963. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Izumi C, Miyake M, Fujita T, et al. Antithrombotic therapy for patients with atrial fibrillation and bioprosthetic valves- real-world data from the multicenter, prospective, observational BPV-AF registry. Circ J. 2022;86:440–8. [DOI] [PubMed] [Google Scholar]
- 6.Kanaan DM, Cook BM, Kelly J, Malloy R. Evaluation of prescribing practices and outcomes using direct-acting oral anticoagulants after cardiac surgery. Clin Ther. 2021;43:e209–16. [DOI] [PubMed] [Google Scholar]
- 7.Pasciolla S, Zizza LF, Le T, Wright K. Comparison of the efficacy and safety of direct oral anticoagulants and warfarin after bioprosthetic valve replacements. Clin Drug Investig. 2020;40:839–45. [DOI] [PubMed] [Google Scholar]
- 8.Yokoyama Y, Briasoulis A, Ueyama H, et al. Direct oral anticoagulants versus vitamin K antagonists in patients with atrial fibrillation and bioprosthetic valves: a meta-analysis. J Thorac Cardiovasc Surg. 2023;165:2052–9. [DOI] [PubMed] [Google Scholar]
- 9.Vahanian A, Beyersdorf F, Praz F, et al. 2021 ESC/EACTS Guidelines for the management of valvular heart disease. Eur Heart J. 2022;43:561–632. [DOI] [PubMed] [Google Scholar]
- 10.Guimarães HP, Lopes RD, de Barros E, Silva PGM, et al. Rivaroxaban in Patients with Atrial Fibrillation and a Bioprosthetic Mitral Valve. N Engl J Med. 2020;383:2117–26. [DOI] [PubMed] [Google Scholar]
- 11.Dangas GD, Tijssen JGP, Wöhrle J, et al. A Controlled Trial of Rivaroxaban after Transcatheter Aortic-Valve Replacement. N Engl J Med. 2020;382:120–9. [DOI] [PubMed] [Google Scholar]
- 12.Ingason AB, Hreinsson JP, Lund SH, et al. Comparison of medication adherence to different oral anticoagulants: population-based cohort study. BMJ Open. 2023;13:e065700. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Chan N, Sobieraj-Teague M, Eikelboom JW. Direct oral anticoagulants: evidence and unresolved issues. Lancet. 2020;396:1767–76. [DOI] [PubMed] [Google Scholar]
- 14.Ng DL, Gan GG, Chai CS, et al. Comparing quality of life and treatment satisfaction between patients on warfarin and direct oral anticoagulants: a cross-sectional study. Patient Prefer Adherence. 2019;13:1363–73. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Wang TH, Tsai YT, Lee PC. Health big data in Taiwan: a national health insurance research database. J Formos Med Assoc. 2023;122:296–8. [DOI] [PubMed] [Google Scholar]
- 16.Rome BN, Gagne JJ, Avorn J, Kesselheim AS. Non-warfarin oral anticoagulant copayments and adherence in atrial fibrillation: a population-based cohort study. Am Heart J. 2021;233:109–21. [DOI] [PubMed] [Google Scholar]
- 17.Choi JM, Lee SH, Jang YJ, Kang M, Choi JH. Medication Adherence and Clinical Outcome of Fixed-Dose Combination vs. Free Combination of Angiotensin Receptor Blocker and Statin. Circ J. 2021;85:595–603. [DOI] [PubMed] [Google Scholar]
- 18.Quirós López R, Formiga Pérez F, Beyer-Westendorf J. Adherence and persistence with direct oral anticoagulants by dose regimen: A systematic review. Br J Clin Pharmacol. 2025;91:1096–113. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Banerjee A, Benedetto V, Gichuru P, et al. Adherence and persistence to direct oral anticoagulants in atrial fibrillation: a population-based study. Heart. 2020;106:119–26. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Shrank WH, Patrick AR, Brookhart MA. Healthy user and related biases in observational studies of preventive interventions: a primer for physicians. J Gen Intern Med. 2011;26:546–50. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Sherman RE, Anderson SA, Dal Pan GJ, et al. Real-World Evidence - What Is It and What Can It Tell Us? N Engl J Med. 2016;375:2293–7. [DOI] [PubMed] [Google Scholar]
- 22.Liu HT, Lee HL, Wang YC, Chou SC, Chou CC, Non-Vitamin K. Antagonist Oral Anticoagulants for Thromboembolic Prevention in Patients with Atrial Fibrillation and Concomitant Mitral Stenosis: A Retrospective Observational Study. Acta Cardiol Sin. 2025;41:622–33. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Guardia Martínez P, Avilés Toscano AL, Martínez Mayoral MA, Moltó Miralles J. DOAC versus VKA in patients with atrial fibrillation and bioprosthetic valves: a systematic review and meta-analysis. Rev Esp Cardiol (Engl Ed). 2023;76:690–9. [DOI] [PubMed] [Google Scholar]
- 24.Zeng S, Zheng Y, Jiang J, Ma J, Zhu W, Cai X. Effectiveness and Safety of DOACs vs. Warfarin in Patients With Atrial Fibrillation and Frailty: A Systematic Review and Meta-Analysis. Front Cardiovasc Med. 2022;9:907197. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Hu T, Chen C, Maduray K, Han W, Chen T, Zhong J. Comparative effectiveness and safety of DOACs vs. VKAs in treatment of left ventricular thrombus- a meta-analysis update. Thromb J. 2024;22:23. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Miyazaki M, Matsuo K, Uchiyama M, et al. Inappropriate direct oral anticoagulant dosing in atrial fibrillation patients is associated with prescriptions for outpatients rather than inpatients: a single-center retrospective cohort study. J Pharm Health Care Sci. 2020;6:2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Martha JW, Pranata R, Raffaelo WM, Wibowo A, Akbar MR. Direct Acting Oral Anticoagulant vs. Warfarin in the Prevention of Thromboembolism in Patients With Non-valvular Atrial Fibrillation With Valvular Heart Disease-A Systematic Review and Meta-Analysis. Front Cardiovasc Med. 2021;8:764356. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Burn J, Pirmohamed M. Direct oral anticoagulants versus warfarin: is new always better than the old? Open Heart. 2018;5:e000712. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Bortman LV, Mitchell F, Naveiro S, et al. Direct Oral Anticoagulants: An Updated Systematic Review of Their Clinical Pharmacology and Clinical Effectiveness and Safety in Patients With Nonvalvular Atrial Fibrillation. J Clin Pharmacol. 2023;63:383–96. [DOI] [PubMed] [Google Scholar]
- 30.Mentias A, Saad M, Michael M, et al. Direct oral anticoagulants versus warfarin in patients with atrial fibrillation and valve replacement or repair. J Am Heart Assoc. 2022;11:e026666. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Liew A, O’Donnell M, Douketis J. Comparing mortality in patients with atrial fibrillation who are receiving a direct-acting oral anticoagulant or warfarin: a meta-analysis of randomized trials. J Thromb Haemost. 2014;12:1419–24. [DOI] [PubMed] [Google Scholar]
- 32.Chao TF, Chan NY, Chan YH, et al. Direct Oral Anticoagulant Dosing in Patients With Atrial Fibrillation: An Asian Perspective. JACC Asia. 2023;3:707–23. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Xu W, Lv M, Wu T, et al. Off-label dose direct oral anticoagulants and clinical outcomes in Asian patients with atrial fibrillation: A new evidence of Asian dose. Int J Cardiol. 2023;371:184–90. [DOI] [PubMed] [Google Scholar]
- 34.Gavrilova A, Meisters J, Latkovskis G, Urtāne I. Stability of direct oral anticoagulants concentrations in blood samples for accessibility expansion of chromogenic assays. Medicina. 2023;59:1339. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Hegstad S, Fuskevåg OM, Amundsen S, Gule M, Spigset O, Helland A. Stability of direct oral anticoagulants and antiarrhythmic drugs in serum collected in standard (Nongel) serum tubes versus tubes containing gel separators. Ther Drug Monit. 2022;44:328–34. [DOI] [PubMed] [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 datasets generated and/or analyzed in this study are not publicly available due to legal and ethical restrictions imposed by the HWDC, Ministry of Health and Welfare, Taiwan. However, access to the data can be requested from the HWDC (https://dep.mohw.gov.tw/DOS/cp-2516-59203-113.html) subject to approval from the Ministry of Health and Welfare and institutional ethics committees.