ABSTRACT
Objectives
To describe temporal trends in the clinical landscape of vitamin K antagonist (VKA) treatment in the Netherlands, after the introduction of a lower international normalized ratio (INR) target range in 2016 and direct oral anticoagulants.
Methods
Data from 17 Dutch anticoagulation clinics between 2014 and 2022 were individually linked to nationwide healthcare data. Prevalent VKA users were identified on January 1 of each year, forming nine annual cohorts. We assessed baseline patient characteristics, anticoagulation treatment quality (using time in/above/below therapeutic range [TTR/ATR/BTR]), and annual odds ratios of major bleeding and thromboembolism for each cohort.
Results
The number of patients treated with VKAs increased from 101 090 in 2014 to 121 916 in 2017, followed by a decline to 73 649 in 2022. Over time, cohorts were characterized by increasing age and a higher prevalence of comorbidities, with a doubling of kidney disease. After the introduction of the lower INR target range, mean BTR decreased (~30% in 2014 to ~15% in 2022), while mean TTR increased (~54% in 2014 to ~67% in 2022), with stable mean ATR. Across calendar years, the 1‐year odds ratio of major bleeding decreased from 1.71% (95% CI 1.63–1.79) in 2014 to 1.52% (95% CI 1.44–1.61) in 2022, whereas the odds ratio of ischemic stroke/transient ischemic attack/systemic thromboembolism increased from 1.14% (95% CI 1.08–1.21) in 2014 to 1.32% (95% CI 1.24–1.40) in 2022. The 1‐year odds ratio of mortality increased from 5.09 (95% CI 4.95–5.22) to 8.55 (95% CI 8.35–8.75), with an OR of 1.74 (95% CI 1.68–1.81).
Conclusions
VKA users declined between 2014 and 2022. The remaining patients using VKA became older and more comorbid. During this period, INR target ranges were lowered, while temporal changes were observed with decreasing odds ratio of major bleeding and increasing odds ratio of thromboembolism and mortality.
Keywords: acenocoumarol, epidemiology, oral anticoagulants, phenprocoumon, treatment adherence and compliance, treatment outcome
Key Points
The number of VKA users in The Netherlands declined between 2014 and 2022.
The number of VKA users that switched to a DOAC increased over time.
The remaining VKA users were older and have more comorbidities, of which those with kidney disease has almost doubled.
After the INR ranges were lowered in the Netherlands in 2016, the proportion of INR measurements below the target range declined and within increased significantly.
VKA patients had an increased odds ratio of arterial thromboembolism and a decreased major bleeding odds ratio after from 2016 compared with 2014.
Plain Language Summary
Vitamin K antagonists (VKAs) are blood thinners that require monitoring. Monitoring is performed by measuring an internationalized normalized ratio (INR) in blood. Based on the measured INR and the INR target range, the dose of the VKA can be adjusted. Patients using VKAs in the Netherlands are monitored by anticoagulant clinics with a frequency of at least every 6 weeks. New blood thinners (DOACs) were introduced with an advantage that these do not require regular monitoring. DOACs are first choice treatment for indications such as atrial fibrillation and venous thrombosis. However, some patients cannot use DOACs, for reasons such as a limited kidney or liver function or indications requiring VKAs. We aimed to describe if and how the VKA population changed over time. We found that the number of VKA patients declined. The patients that were treated with VKAs became older with more coexisting diseases. One of the diseases that increased most was kidney disease, which doubled in percentage (from approximately 5% to 10%). Over time the risk of bleeding decreased and the risk of thromboembolism increased.
1. Introduction
Vitamin K antagonists (VKAs) dominated oral anticoagulant treatment for several decades. However, since the introduction of direct oral anticoagulants (DOACs), they have been gradually replaced as the first‐line treatment since 2010 [1, 2, 3]. For most indications, DOACs are now preferred over VKAs because of their superior safety and effectiveness [4, 5, 6, 7, 8]. Nevertheless, VKAs are preferred for patients with mechanical heart valves, thrombotic antiphospholipid syndrome, impaired renal and/or liver function, and women at risk of thromboembolic events who desire to have children [9] and a potential safer profile for frail patients [10].
In the Netherlands, acenocoumarol and phenprocoumon are registered, while warfarin is not. Specialized anticoagulation clinics support the monitoring and management of VKA therapy [11]. The decrease in the number and type of patients that remain in the anticoagulant clinics due to an increased DOAC prescription may have consequences for existing quality measurements of patient care in anticoagulation clinics.
Next to the shift in the number and type of VKA users, since 2016 INR target ranges guiding VKA therapy in the Netherlands have been lowered to align with international practice from 2–3.5 (target: 2.5–3.5) and 2.5–4 (target: 3–4) to 2–3 and 2.5–3.5 for the low and high intensity treatment groups, respectively [12]. At present, no evaluation took place of the consequences (e.g., time in therapeutic range, rates of clinical outcomes) of this alteration of the INR target range. Furthermore, emerging evidence has also suggested that the COVID‐19 pandemic adversely affected the quality of VKA treatment [13], consistent with broader disruptions observed across chronic disease management during this period [14]. Given these changes, the present study aimed to describe temporal trends of patient characteristics, anticoagulation treatment quality, and risks of adverse clinical events among the currently treated (prevalent) VKA users in the Netherlands, covering the period of changes in INR target range, start of DOAC use and the COVID‐19 pandemic.
2. Methods
2.1. Data Sources
Routinely collected data covering the period from 2014 to 2022 were obtained from 17 Dutch anticoagulation clinics (Table S1). The data contained information on all INR measurements for all VKAs users at these clinics, including date of INR measurement, INR value, lower and upper INR target ranges, location of measurement, VKA dose, and VKA subtype. In addition, the data included the start and end date (when available) of VKA treatment, as well as the indication(s) for VKA therapy.
The clinic data were individually linked to data from Statistics Netherlands (“Centraal Bureau voor de Statistiek,” CBS) on birthdate, sex, and postal code with a matching rate exceeding 95%, after which all data were pseudonymized. We used the following microdata from CBS: (1) demographic characteristics (2014–2022); (2) standardized household income (2014–2022); (3) in‐hospital diagnoses (2010–2022); (4) outpatient medication dispensing records (2010–2022); and (5) date and primary cause of death (2014–2022). These data sources have been described in detail previously [3]. Comorbidities (or medical history) and clinical outcomes were identified using the International Classification of Diseases (ICD), 10th revision (or 9th revision for some diagnosis records registered before 2013); dispensed medications were identified according to the Anatomical Therapeutic Chemical (ATC) Classification System (Table S2).
The study was conducted in accordance with the Declaration of Helsinki and was approved by the Scientific Committee of the Department of Clinical Epidemiology, Leiden University Medical Center (approval no. A226), with a waiver of informed consent due to the exclusive use of preexisting, deidentified data.
2.2. Study Design and Study Cohorts
To investigate temporal trends among treated VKA users between 2014 and 2022, nine annual cohorts of prevalent (at that moment treated) VKA users were identified using the clinic data. Specifically, for each calendar year (e.g., 2014), January 1st was defined as the index date (baseline). We identified individuals with at least one recorded INR measurement within the 3 months preceding the index date (e.g., between October 1, 2013, and January 1, 2014). These individuals were considered prevalent VKA users for that cohort (e.g., referred to as Cohort 2014), and the most recent INR measurement prior to the index date was defined as the index INR measurement.
Individuals were excluded if (based on the clinic data associated with the index INR measurement, unless otherwise specified, Figure 1): (1) the recorded type of anticoagulant was not a VKA or was missing; (2) the VKA treatment period had ended before the index INR measurement or the index date; (3) the individual did not survive until the index date (determined using CBS data on date of death); (4) when the INR target range was unavailable across the whole treatment period; or (5) a DOAC was dispensed prior to the index date.
FIGURE 1.
Flow chart of the study population. †From the October 1 of the previous calendar year to the January 1 of the current calendar year. ‡The January 1 of 2014–2022. *Based on the latest INR measurement before the index date (i.e., the index INR measurement). The proportion is calculated based on the remaining sample size after applying the previous exclusion criterion.
The same individual could be included in multiple annual cohorts, provided that inclusion and exclusion criteria were satisfied. If an individual had more than one eligible index INR measurement from different clinics (for the same annual cohort identification), one of the clinics was randomly selected to ensure that each individual was included only once per annual cohort. If more than one VKA treatment period covered the index INR measurement, the treatment period with a start date closest to the index date was selected.
2.3. Patient Characteristics at Baseline
For each individual in each annual cohort, patient characteristics at baseline were determined as follows: (1) age, sex, and immigration background, derived from CBS demographic data on the index date; (2) standardized household income, based on CBS household income data in the calendar year of the index date; (3) the CHA2DS2‐VASc score, calculated using CBS data on in‐hospital diagnoses recorded within 4 years prior to the index date; (4) information on VKA treatment, determined from clinic data at the time of the index INR measurement; (5) prior use of antithrombotic agents within 6 months to 4 years before the index date, identified from CBS outpatient medication dispensings; and (6) comorbidities and medical history, determined from CBS in‐hospital diagnosis data within 4 years prior to the index date (Table S2).
For individuals included in multiple annual cohorts, baseline characteristics were consistently (re‐)defined according to the corresponding index dates.
2.4. Anticoagulation Treatment During 1‐Year Follow‐Up
For each individual in each annual cohort, follow‐up began on the corresponding index date (i.e., first of January of that corresponding year) and continued until the end of the calendar year or death, whichever occurred first. Using medication dispensing data from CBS, we identified the number of filled prescriptions for antithrombotic agents and switching to DOACs during the 1‐year follow‐up. Switching to DOACs was defined as the presence of at least one dispensed prescription for DOACs. In addition, if the clinic data indicated that the VKA subtype associated with any INR measurement during follow‐up was actually a DOAC, the individual was also classified as having switched to DOACs. Patients switching to a DOAC are discharged from the registry of their anticoagulation clinic, which stops their data from being collected for our study.
To calculate INR specific variables, we restricted the follow‐up to the duration of VKA treatment only. The end of VKA treatment was defined as the earliest of the following: (1) the recorded end date of VKA treatment in the clinic data; (2) the first occurrence of a DOAC prescription; or (3) 42 days after the last available INR. Among individuals with at least one INR during this observation period, we calculated the proportion of INRs below, within, and above the target range. Among individuals with at least two INRs, we additionally estimated the proportion of observation time spent below (BTR), within (TTR), and above the target range (ATR) using the Rosendaal method [15].
2.5. Clinical Outcomes During 1‐Year Follow‐Up
For each individual in each annual cohort, we used CBS data on in‐hospital diagnoses (restricted to primary/main diagnoses) and the primary cause of death to assess the occurrence of the outcomes. Outcomes were only considered if these occurred after the inclusion date of that specific individual in that particular cohort during the 1‐year follow‐up period. We considered the following outcomes (1) major bleeding; (2) ischemic stroke/transient ischemic attack/systemic thromboembolism; (3) myocardial infarction; (4) VTE; and (5) all‐cause mortality (without considering the cause). Within each annual cohort, each type of clinical event was evaluated separately, and for each event type, only the first occurrence during follow‐up was counted. Given the use of nationwide data, we assumed no loss to follow‐up after ignoring emigration due to lack of data (which should be negligible).
2.6. Statistical Analysis
Summary statistics for baseline patient characteristics and measures of anticoagulant treatment were summarized using means and standard deviations (SDs) or medians and interquartile ranges (IQRs) for continuous variables, and counts (percentages) for categorical variables, stratified by cohort (calendar year). If records lacked INR target ranges, missing target ranges were imputed using available target range information from the same patient measured at nearby time points.
Clinical outcomes were described by calculating the 1‐year cumulative incidence for each outcome within, defined as the number of events divided by the number of patients at baseline, with 95% confidence intervals (CIs) estimated using the Clopper–Pearson exact method. Logistic regression with clustered sandwich standard errors were used to estimate odds ratios (ORs) for clinical outcomes comparing cohorts with Cohort 2014 serving as the reference year. Although the study was descriptive in nature, both crude ORs and ORs from prespecified adjusted models were reported to evaluate possible changes in outcomes over time: (1) Model 1, adjusted for age, sex, immigration background, and standardized household income; (2) Model 2, adjusted for Model 1 covariables plus receiving ≥ 1 prescription of antiplatelet agent (within 6 months before index date, and the interval between 6 months and 4 years before index date); and (3) Model 3, adjusted for Model 2 covariables plus comorbidities including AF, heart failure, aortic plaque, myocardial infarction (history), hypertension, rheumatic mitral stenosis/mechanical heart valves, other valvular heart diseases, ischemic stroke (history), transient ischemic attack, systemic thromboembolism, peripheral artery disease, VTE, diabetes, major bleeding, malignant tumor, overweight/obesity, alcohol abuse, liver diseases, kidney diseases, coagulopathy, autoimmune diseases, and system connective tissue disorders. All statistical analyses were performed using R (version 4.2.3; R Project for Statistical Computing).
3. Results
3.1. Temporal Trends in Baseline Patient Characteristics of Prevalent VKA Users
The number of VKA patients increased from 101 090 in 2014 to 121 916 in 2017, followed by a decline to 73 649 in 2022. Over time, the population of prevalent VKA users aged, with the mean age varying from 74.1 ± 11.9 years in 2014 to 76.1 ± 11.98 in 2019 to 75.0 ± 12.7 years in 2022, while the distributions of sex and immigration background remained largely unchanged (Table 1). Acenocoumarol was the most frequently prescribed VKA (74.9%), followed by phenprocoumon (25.1%), with these proportions remaining stable across the study period. The most common INR target range prior to 2016 was 2.5–3.5 (~75%), followed by 3.0–4.0 (~14%) and 2.0–3.0 (~11%). After the of the INR target range in 2016 in the Netherlands, the 2.0–3.0 range became the most prevalent (~80%), while the proportions of the 2.5–3.5 and 3.0–4.0 ranges declined to 17% and 0.5%, respectively. AF remained the most prevalent indication for VKAs (> 70%). However, its proportion appeared to decrease from 2020. The mean CHA2DS2‐VASc score remained stable throughout the study period. The prevalence of most comorbidities increased gradually over time, particularly kidney disease which almost doubled from 5.5% in 2014 to 10.5% in 2022 (Table S3).
TABLE 1.
Baseline characteristics of individuals receiving vitamin K antagonists by calendar year.
| Variables | 2014 | 2015 | 2016 | 2017 | 2018 | 2019 | 2020 | 2021 | 2022 |
|---|---|---|---|---|---|---|---|---|---|
| n | 101 090 | 109 218 | 117 133 | 121 916 | 111 613 | 111 769 | 107 286 | 86 440 | 73 649 |
| Age (years) | 74.1 ± 11.9 | 74.4 ± 12.0 | 74.8 ± 11.9 | 75.4 ± 11.9 | 75.8 ± 11.8 | 76.1 ± 11.8 | 75.6 ± 12.4 | 75.3 ± 12.5 | 75.0 ± 12.7 |
| Female | 46 535 (46.0) | 50 148 (45.9) | 53 691 (45.8) | 56 010 (45.9) | 51 088 (45.8) | 50 949 (45.6) | 47 947 (44.7) | 38 049 (44.0) | 32 030 (43.5) |
| Immigration background | |||||||||
| Native Dutch | 88 608 (87.7) | 95 258 (87.2) | 102 187 (87.2) | 106 646 (87.5) | 97 369 (87.2) | 96 295 (86.2) | 92 110 (85.9) | 73 462 (85.0) | 62 318 (84.6) |
| First generation of immigrants | 8031 (7.9) | 9017 (8.3) | 9634 (8.2) | 9874 (8.1) | 9239 (8.3) | 9228 (8.3) | 9141 (8.5) | 7886 (9.1) | 6903 (9.4) |
| Second generation of immigrants | 4451 (4.4) | 4943 (4.5) | 5312 (4.5) | 5396 (4.4) | 5005 (4.5) | 6246 (5.6) | 6035 (5.6) | 5092 (5.9) | 4428 (6.0) |
| CHA2DS2‐VASc score | 2.4 ± 1.4 | 2.5 ± 1.5 | 2.5 ± 1.5 | 2.6 ± 1.5 | 2.6 ± 1.5 | 2.6 ± 1.5 | 2.6 ± 1.5 | 2.5 ± 1.5 | 2.5 ± 1.5 |
| Type of VKA | |||||||||
| Acenocoumarol | 73 633 (72.8) | 80 330 (73.6) | 87 525 (74.7) | 93 264 (76.5) | 84 849 (76.0) | 84 556 (75.7) | 80 928 (75.4) | 64 299 (74.4) | 54 316 (73.7) |
| Phenprocoumon | 27 438 (27.1) | 28 864 (26.4) | 29 578 (25.3) | 28 620 (23.5) | 26 739 (24.0) | 27 193 (24.3) | 26 323 (24.5) | 22 109 (25.6) | 19 301 (26.2) |
| Warfarin/Fluindione | 19 (0.0) | 24 (0.0) | 30 (0.0) | 32 (0.0) | 25 (0.0) | 20 (0.0) | 35 (0.0) | 32 (0.0) | 32 (0.0) |
| INR target range | |||||||||
| 2.0–3.0 | 10 109 (10.0) | 11 623 (10.6) | 13 029 (11.1) | 101 787 (83.5) | 92 819 (83.2) | 92 554 (82.8) | 87 520 (81.6) | 69 315 (80.2) | 58 082 (78.9) |
| 2.5–3.5 | 75 457 (74.6) | 81 914 (75.0) | 87 821 (75.0) | 18 576 (15.2) | 17 386 (15.6) | 17 816 (15.9) | 18 228 (17.0) | 15 744 (18.2) | 14 292 (19.4) |
| 3.0–4.0 | 14 557 (14.4) | 14 704 (13.5) | 15 273 (13.0) | 505 (0.4) | 462 (0.4) | 458 (0.4) | 544 (0.5) | 499 (0.6) | 448 (0.6) |
| Planned average VKA dose | 1.7 (0.9–2.6) | 1.7 (0.9–2.6) | 1.8 (1.0–2.6) | 1.7 (0.9–2.5) | 1.7 (0.9–2.5) | 1.6 (0.9–2.4) | 1.7 (0.9–2.4) | 1.6 (0.9–2.4) | 1.6 (0.9–2.5) |
| Acenocoumarol | 2.2 (1.6–2.9) | 2.1 (1.6–2.9) | 2.1 (1.6–2.9) | 2.0 (1.5–2.7) | 2.0 (1.5–2.7) | 2.0 (1.4–2.7) | 2.0 (1.5–2.7) | 2.0 (1.5–2.7) | 2.0 (1.5–2.7) |
| Phenprocoumon | 0.6 (0.4–0.9) | 0.7 (0.4–0.9) | 0.6 (0.4–0.9) | 0.6 (0.4–0.8) | 0.6 (0.4–0.8) | 0.6 (0.4–0.8) | 0.6 (0.4–0.8) | 0.6 (0.4–0.8) | 0.6 (0.4–0.8) |
| Warfarin/Fluindione | 1.4 (0.8–2.3) | 1.3 (1.0–2.3) | 1.3 (0.9–2.1) | 1.3 (0.8–2.3) | 1.4 (0.9–1.9) | 1.4 (0.9–1.8) | 1.2 (0.9–1.9) | 1.6 (0.9–2.9) | 1.0 (0.6–1.6) |
| Recorded indication of VKA treatment a | |||||||||
| Atrial fibrillation | 73 313 (72.5) | 79 359 (72.7) | 85 515 (73.0) | 88 828 (72.9) | 80 469 (72.1) | 79 016 (70.7) | 71 851 (67.0) | 55 315 (64.0) | 44 928 (61.0) |
| Venous thromboembolism | 13 235 (13.1) | 14 844 (13.6) | 16 213 (13.8) | 16 849 (13.8) | 15 497 (13.9) | 16 169 (14.5) | 16 876 (15.7) | 14 602 (16.9) | 13 014 (17.7) |
| Mechanical heart valve | 6488 (6.4) | 6905 (6.3) | 7254 (6.2) | 7598 (6.2) | 7499 (6.7) | 8221 (7.4) | 9493 (8.8) | 8678 (10.0) | 8207 (11.1) |
| Other valve disease | 9095 (9.0) | 9802 (9.0) | 10 394 (8.9) | 10 435 (8.6) | 10 157 (9.1) | 9966 (8.9) | 11 087 (10.3) | 10 305 (11.9) | 9488 (12.9) |
| Other mitral valve disease | 556 (0.6) | 524 (0.5) | 539 (0.5) | 700 (0.6) | 708 (0.6) | 990 (0.9) | 969 (0.9) | 740 (0.9) | 712 (1.0) |
| Comorbidities b | |||||||||
| Atrial fibrillation | 31 611 (31.3) | 35 418 (32.4) | 39 279 (33.5) | 40 620 (33.3) | 35 575 (31.9) | 33 087 (29.6) | 28 413 (26.5) | 20 637 (23.9) | 16 346 (22.2) |
| Hypertension | 15 201 (15.0) | 19 752 (18.1) | 23 945 (20.4) | 26 675 (21.9) | 24 753 (22.2) | 24 792 (22.2) | 23 155 (21.6) | 18 210 (21.1) | 15 673 (21.3) |
| Myocardial infarction (history)/Ischemic stroke (history)/Transient ischemic attack/Systemic thromboembolism/Peripheral artery disease/Aortic plaque | 17 847 (17.7) | 20 319 (18.6) | 22 740 (19.4) | 24 203 (19.9) | 22 170 (19.9) | 21 800 (19.5) | 19 702 (18.4) | 15 378 (17.8) | 12 927 (17.6) |
| Injury/poisoning/Alcohol abuse | 13 593 (13.4) | 15 721 (14.4) | 17 669 (15.1) | 19 378 (15.9) | 18 452 (16.5) | 18 044 (16.1) | 16 832 (15.7) | 12 762 (14.8) | 10 365 (14.1) |
| Diabetes/Thyroid disease/Cushing syndrome | 10 042 (9.9) | 12 466 (11.4) | 14 719 (12.6) | 16 149 (13.2) | 14 973 (13.4) | 15 164 (13.6) | 13 974 (13.0) | 10 748 (12.4) | 8949 (12.2) |
| Heart failure/Chronic pulmonary heart disease | 10 604 (10.5) | 12 270 (11.2) | 13 941 (11.9) | 15 292 (12.5) | 14 303 (12.8) | 14 303 (12.8) | 13 161 (12.3) | 10 120 (11.7) | 8327 (11.3) |
| Rheumatic mitral stenosis/Mechanical heart valves/Other valvular heart diseases | 8164 (8.1) | 9741 (8.9) | 11 505 (9.8) | 12 575 (10.3) | 12 153 (10.9) | 12 296 (11.0) | 12 078 (11.3) | 9637 (11.1) | 8365 (11.4) |
| Kidney diseases | 5605 (5.5) | 7749 (7.1) | 10 150 (8.7) | 12 614 (10.3) | 12 491 (11.2) | 13 056 (11.7) | 12 207 (11.4) | 9329 (10.8) | 7716 (10.5) |
| Anemia/Coagulopathy | 6317 (6.2) | 7977 (7.3) | 10 037 (8.6) | 11 540 (9.5) | 11 350 (10.2) | 11 753 (10.5) | 11 164 (10.4) | 8934 (10.3) | 7711 (10.5) |
| Lung disease | 6865 (6.8) | 8634 (7.9) | 10 317 (8.8) | 11 170 (9.2) | 10 482 (9.4) | 10 698 (9.6) | 10 172 (9.5) | 8029 (9.3) | 6628 (9.0) |
| Malignant tumor | 6063 (6.0) | 6883 (6.3) | 7908 (6.8) | 8329 (6.8) | 7554 (6.8) | 7255 (6.5) | 6631 (6.2) | 5130 (5.9) | 4213 (5.7) |
| Major bleeding | 4865 (4.8) | 5885 (5.4) | 7096 (6.1) | 7739 (6.3) | 7273 (6.5) | 6979 (6.2) | 6378 (5.9) | 4964 (5.7) | 4127 (5.6) |
| Venous thromboembolism | 3407 (3.4) | 4034 (3.7) | 4440 (3.8) | 4364 (3.6) | 3375 (3.0) | 2927 (2.6) | 2423 (2.3) | 1733 (2.0) | 1464 (2.0) |
Note: For readability, selected categories for categorical variables are presented in this table. Full category details are provided in Table S3.
Abbreviations: INR, international normalized ratio; VKA, vitamin K antagonist.
Based on free‐text data from participating anticoagulation clinics. Patients may have multiple indications; timing was unavailable, so some indications may have developed after the index date but were considered as present at index.
Identified from hospital diagnoses (CBS microdata) within 4 years before the index date.
3.2. Temporal Trends in Anticoagulation Treatment Among Prevalent VKA Users
Over the study period, VKA users filled VKA prescriptions at a stable frequency (median 3 fills/year follow‐up; IQR 2–4). The number of INR measurements also remained largely unchanged (median 18; IQR 13–24), although a lower frequency was observed in 2020 (median 16; IQR 11–22). In contrast, the proportion of patients who switched to DOACs within 1 year increased from 0.7% in 2014 to 7.1% in 2022 (Table S4).
The mean of INRs within target range was 59.4% (SD 19.9), while 21.1% (SD 17.1) and 19.5% (SD 14.3) were below and above the target range, respectively. Stratified by calendar year, the proportion of INR measurements below the target range declined significantly, from above 30% in 2014–2015 to below 20% from 2016 onward (Table S4).
Similar trends were observed when VKA exposure was assessed using therapeutic time below, within, and above the target range among patients with at least two INR measurements after the index date (Figure 2). Mean BTR during VKA treatment decreased (from ~30% before 2016 to ~15% thereafter), whereas mean TTR increased over the same period (from ~54% to ~67%), while mean ATR remained stable at approximately 18%.
FIGURE 2.
Distribution of time in INR target range among individuals receiving vitamin K antagonists by calendar year. ATR, time above therapeutic range; BTR, time below therapeutic range; INR, international normalized ratio; TTR, time in therapeutic range.
3.3. Temporal Trends in Annual ORs of Adverse Clinical Events Among Prevalent VKA Users
Across calendar years, irrespective of anticoagulant treatment, the 1‐year OR of major bleeding decreased from 1.71% (95% CI 1.63–1.79) in 2014 to 1.52% (95% CI 1.44–1.61) in 2022, corresponding to a crude OR of 0.89 (95% CI 0.82–0.96, see Table S5). In contrast, the 1‐year OR of ischemic stroke/transient ischemic attack/systemic thromboembolism increased from 1.14% (95% CI 1.08–1.21) to 1.32% (95% CI 1.24–1.40), with a crude OR of 1.16 (95% CI 1.06–1.26). The OR of myocardial infarction and VTE remained largely stable over time (Figure 3). However, the 1‐year OR of all‐cause mortality increased from 5.09% (95% CI 4.95–5.22) in 2014 to 8.55% (95% CI 8.35–8.75) in 2022, yielding a crude OR of 1.74 (95% CI 1.68–1.81), see Table S5.
FIGURE 3.
Temporal trends in risk of clinical events among individuals receiving vitamin K antagonists (at baseline) by calendar year. Note: The yellow shaded area marks the first calendar year when the target range of international normalized ratio during vitamin K antagonist treatment was fully switched to the international standard in the Netherlands. CI, confidence interval; TE, systemic thromboembolism; TIA, transient ischemic attack.
4. Discussion
4.1. Summary of Study Findings and Implications
In this study, we comprehensively characterized temporal trends from 2014 to 2022 in patient characteristics, VKA treatment quality, and OR of adverse clinical events among prevalent VKA users across 17 anticoagulation clinics in the Netherlands. We report four key findings: (1) the number of VKA patients peaked in 2017 and subsequently declined; (2) the VKA population aged, exhibited a higher comorbidity burden and increased annual mortality OR; (3) following the lowering of the INR target range in 2016, INR measurements were more likely to fall within the target range and less likely to fall below it; and (4) in the years after lowering the INR target range compared with before, prevalent VKA users experienced a lower annual OR of major bleeding and a slightly higher annual OR of arterial thromboembolism.
Together, these findings provide a timely update on the descriptive epidemiology of real‐world VKA use in the Netherlands. Our study offers a comprehensive population‐level overview of who receives VKAs, how treatment is delivered, and the prognosis of treated patients. This information is essential for identifying opportunities to improve VKA treatment management and to inform healthcare resource planning, including the organization and capacity of anticoagulation clinics. In addition, these data serve as a valuable benchmark for future studies to evaluate the changes in risks in a changing population and current guidelines.
4.2. Interpretation and Comparison With Existing Literature
Although a decline in VKA use in the DOAC era is expected, temporal trends in VKA treatment have not previously been reported with the level of detail and recency provided in our study, particularly at a (nearly) nationwide scale. The lowering of the INR target range in the Netherlands in 2016 [12], together with the potential impact of the COVID‐19 pandemic on VKA treatment [13], further underscore the relevance of examining contemporary epidemiological patterns of VKA use in the Netherlands. We observed that, following the lowering of the target range in 2016, a greater proportion of time among VKA users was spent within the INR target range and a reduction of time below target. Concurrently, the population appeared to experience a lower OR of major bleeding and a higher OR of arterial thromboembolism. These observations may suggest that the lowering of the INR target in 2016 improved VKA treatment quality (evaluated by TTR) and a slightly reduced the OR of major bleeding, at the cost of an increase in arterial thromboembolism (crude OR 1.16, 95% CI 1.06–1.26) and all‐cause mortality (crude OR 1.74, 95% CI 1.68–1.81). We emphasize however, that the magnitude of these results found in our descriptive study should be interpreted with caution as various events occurred almost simultaneously in our study (changes in target range in 2016, introduction of DOACs in the Netherlands in 2010 and also included the COVID‐19 pandemic).
Our findings are consistent with those of a prior study examining the effects of lowering the INR target range, which reported significantly higher TTR and significantly lower BTR after the adjustment [16]. However, this study did not detect differences in clinical outcomes, likely due to limited sample size. Other studies have reported a decline in major bleeding risk during periods overlapping with our study timeframe [3, 17]. Because our study period partially overlapped with the COVID‐19 pandemic, this could be an alternative explanation. For instance, we observed fewer INR measurements in 2020 and an increased proportion of patients switching to DOACs without a major increase in mortality, consistent with prior literature [18, 19, 20]. However, it is important to emphasize that our study design does not support causal inference regarding the effects of lowering the INR target or the COVID‐19 pandemic on VKA treatment quality or clinical outcomes. Addressing such questions would require a fundamentally different study design.
Another important issue warranting discussion is the shifting patient profiles. For example, the prevalence of kidney disease almost doubled, which may be attributable to the preference of VKAs over DOACs in patients with end‐stage renal disease [9]. Furthermore, this renal disease is important to consider when prescribing low‐molecular‐weight heparins. Therefore, as the population remaining on VKA therapy is aging and exhibiting a higher burden of comorbidities, it may be appropriate to extend the focus of VKA monitoring. Furthermore, prior studies have demonstrated that frail patients are at increased risk of stroke compared with nonfrail individuals [21], and that chronic kidney disease is associated with platelet hyperaggregability [22] and an increased bleeding risk [23]. It should therefore be considered whether the currently widely adopted INR target ranges remain optimal in the context of these evolving patient characteristics.
4.3. Strengths and Limitations
We used registries from multiple sources and most variables were identified using administrative codes making our study subject to potential missing data. For example, identification of clinical outcomes primarily relied on in‐hospital diagnosis data, whereas not all VTEs require hospitalization. Consequently, the absolute VTE OR is likely underestimated. This limitation also applies to several comorbidities identified as baseline patient characteristics (e.g., overweight, diabetes mellitus, and hypertension). However, the extent of missingness is likely consistent over the study period, which makes comparisons across calendar years unbiased. Second, not all anticoagulation clinics in the Netherlands were included in the study, and switching between clinics within the same calendar year was not captured. We assume that the absolute number of affected patients is small and that clinic switching occurs largely at random, thereby minimizing its impact on the observed temporal trends.
5. Conclusion
Between 2014 and 2022, the number of patients in the Netherlands treated with VKAs decreased, while this VKA treated population became older and more comorbid, among which the prevalence of kidney disease almost doubled. During this period, INR target ranges were lowered, and temporal changes were observed with decreasing OR of major bleeding and increasing OR of thromboembolism and mortality. Future studies should be conducted to investigate whether current guidelines should be adapted to this changing patient population.
Author Contributions
G.D.Z., Q.C., S.C.C., and N.v.R. contributed to conceptualization and design of the study. Q.C. was primarily involved in data curation and data analysis. G.D.Z. and Q.C. drafted the initial version of the manuscript. M.J.H.A.K., F.A.K., S.C.C., and N.v.R. contributed to interpretation of data and provided critical revisions for important intellectual content. All authors read and approved the final version of the manuscript.
Funding
Qingui Chen was supported by the Chinese Government Scholarship (201906380148) for his PhD study at the Leiden University Medical Center. The other authors have nothing to report.
Ethics Statement
The study was conducted in accordance with the Declaration of Helsinki and was approved by the Scientific Committee of the Department of Clinical Epidemiology, Leiden University Medical Center (approval No. A226), with a waiver of informed consent due to the exclusive use of preexisting, deidentified data.
Conflicts of Interest
The authors declare no conflicts of interest.
Supporting information
Table S1: The 17 Dutch anticoagulation clinics participating in this study.
Table S2: Codes used for variable identification.
Table S3: Baseline characteristics of individuals receiving vitamin K antagonists by calendar year.
Table S4: Summary statistics of vitamin K antagonist treatment among individuals receiving vitamin K antagonists by calendar year.
Table S5: Clinical outcomes within 1 year after index date among individuals receiving vitamin K antagonists by calendar year.
Acknowledgments
The authors thank the Federation of Dutch Thrombosis Services (“Federatie van Nederlandse Trombosediensten”) and Statistics Netherlands (“Centraal Bureau voor de Statistiek”) for preparing data (including those from the Dutch Hospital Data registry) available for the current study. Qingui Chen was supported by the Chinese Government Scholarship (No. 201906380148) for his PhD study at the Leiden University Medical Center.
Data Availability Statement
The study used data from the Federation of Dutch Thrombosis Services and nonpublic microdata from Statistics Netherlands, but these data cannot be shared directly by the authors. Under certain conditions, the data are accessible for statistical and scientific research. For further information: fnt@fnt.nl and microdata@cbs.nl.
References
- 1. van den Heuvel J. M., Hovels A. M., Buller H. R., Mantel‐Teeuwisse A. K., de Boer A., and Maitland‐van der Zee A. H., “NOACs Replace VKA as Preferred Oral Anticoagulant Among New Patients: A Drug Utilization Study in 560 Pharmacies in The Netherlands,” Thrombosis Journal 16 (2018): 7, 10.1186/s12959-017-0156-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Gadsboll K., Staerk L., Fosbol E. L., et al., “Increased Use of Oral Anticoagulants in Patients With Atrial Fibrillation: Temporal Trends From 2005 to 2015 in Denmark,” European Heart Journal 38, no. 12 (2017): 899–906, 10.1093/eurheartj/ehw658. [DOI] [PubMed] [Google Scholar]
- 3. Chen Q., Toorop M. M. A., Tops L. F., Lijfering W. M., and Cannegieter S. C., “Time Trends in Patient Characteristics, Anticoagulation Treatment, and Prognosis of Incident Nonvalvular Atrial Fibrillation in The Netherlands,” JAMA Network Open 6, no. 4 (2023): e239973, 10.1001/jamanetworkopen.2023.9973. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Ruff C. T., Giugliano R. P., Braunwald E., et al., “Comparison of the Efficacy and Safety of New Oral Anticoagulants With Warfarin in Patients With Atrial Fibrillation: A Meta‐Analysis of Randomised Trials,” Lancet 383, no. 9921 (2014): 955–962, 10.1016/S0140-6736(13)62343-0. [DOI] [PubMed] [Google Scholar]
- 5. van Es N., Coppens M., Schulman S., Middeldorp S., and Buller H. R., “Direct Oral Anticoagulants Compared With Vitamin K Antagonists for Acute Venous Thromboembolism: Evidence From Phase 3 Trials,” Blood 124, no. 12 (2014): 1968–1975, 10.1182/blood-2014-04-571232. [DOI] [PubMed] [Google Scholar]
- 6. Hindley B., Lip G. Y. H., McCloskey A. P., and Penson P. E., “Pharmacokinetics and Pharmacodynamics of Direct Oral Anticoagulants,” Expert Opinion on Drug Metabolism & Toxicology 19, no. 12 (2023): 911–923, 10.1080/17425255.2023.2287472. [DOI] [PubMed] [Google Scholar]
- 7. Van Gelder I. C., Rienstra M., Bunting K. V., et al., “2024 ESC Guidelines for the Management of Atrial Fibrillation Developed in Collaboration With the European Association for Cardio‐Thoracic Surgery (EACTS),” European Heart Journal 45, no. 36 (2024): 3314–3414, 10.1093/eurheartj/ehae176. [DOI] [PubMed] [Google Scholar]
- 8. Ortel T. L., Neumann I., Ageno W., et al., “American Society of Hematology 2020 Guidelines for Management of Venous Thromboembolism: Treatment of Deep Vein Thrombosis and Pulmonary Embolism,” Blood Advances 4, no. 19 (2020): 4693–4738, 10.1182/bloodadvances.2020001830. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Bejjani A., Khairani C. D., Assi A., et al., “When Direct Oral Anticoagulants Should Not Be Standard Treatment: JACC State‐of‐the‐Art Review,” Journal of the American College of Cardiology 83, no. 3 (2024): 444–465, 10.1016/j.jacc.2023.10.038. [DOI] [PubMed] [Google Scholar]
- 10. Joosten L. P. T., van Doorn S., van de Ven P. M., et al., “Safety of Switching From a Vitamin K Antagonist to a Non‐Vitamin K Antagonist Oral Anticoagulant in Frail Older Patients With Atrial Fibrillation: Results of the FRAIL‐AF Randomized Controlled Trial,” Circulation 149, no. 4 (2024): 279–289, 10.1161/CIRCULATIONAHA.123.066485. [DOI] [PubMed] [Google Scholar]
- 11. Rosendaal F. R. v M F. and Cannegieter S. C., “Management of Anticoagulant Therapy: The Dutch Experience,” Journal of Thrombosis and Thrombolysis 2, no. 4 (1996): 265–269, 10.1007/bf01061909. [DOI] [Google Scholar]
- 12. Seelig J., Hemels M. E. W., Xhaet O., et al., “Impact of Different Anticoagulation Management Strategies on Outcomes in Atrial Fibrillation: Dutch and Belgian Results From the GARFIELD‐AF Registry,” Journal of Thrombosis and Haemostasis 18, no. 12 (2020): 3280–3288, 10.1111/jth.15081. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Camilleri E., van Rein N., van der Meer F. J. M., et al., “Stability of Vitamin K Antagonist Anticoagulation After COVID‐19 Diagnosis,” Research and Practice in Thrombosis and Haemostasis 5, no. 7 (2021): e12597, 10.1002/rth2.12597. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Olmastroni E., Galimberti F., Tragni E., Catapano A. L., and Casula M., “Impact of COVID‐19 Pandemic on Adherence to Chronic Therapies: A Systematic Review,” International Journal of Environmental Research and Public Health 20, no. 5 (2023): 3825, 10.3390/ijerph20053825. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Rosendaal F. R., Cannegieter S. C., van der Meer F. J., and Briet E., “A Method to Determine the Optimal Intensity of Oral Anticoagulant Therapy,” Thrombosis and Haemostasis 69, no. 3 (1993): 236–239. [PubMed] [Google Scholar]
- 16. Prins A., Ojik A., From J., et al., “Het Effect Van Aanpassing Van de Therapeutische INR‐Range Op Het Optreden Van Bloedingen en Trombo‐Embolische Complicaties Bij Atriumfibrilleren,” Nederlands Platform Voor Farmaceutisch Onderzoek 5 (2020): a1724. [Google Scholar]
- 17. Peeters M. T., Vroman F., Schreuder T. A., van Oostenbrugge R. J., and Staals J., “Decrease in Incidence of Oral Anticoagulant‐Related Intracerebral Hemorrhage Over the Past Decade in The Netherlands,” European Stroke Journal 7, no. 1 (2022): 20–27, 10.1177/23969873211062011. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Gilboa I., Rodrig K., Hornik‐Lurie T., and Pereg D., “Evaluation of Vitamin K Antagonists Management and Control in Patients With Mechanical Prosthetic Heart Valves During the COVID‐19 Pandemic,” Israel Medical Association Journal 26, no. 9 (2024): 546–550. [PubMed] [Google Scholar]
- 19. Comin Colet J., Sicras Mainar A., Salazar‐Mendiguchia J., et al., “Influence of the COVID‐19 Pandemic on Patients Receiving Oral Anticoagulants for the Treatment of Non‐Valvular Atrial Fibrillation,” International Journal of Cardiology‐Heart & Vessels 51 (2024): 101358, 10.1016/j.ijcha.2024.101358. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Rein N.v. and Jaarverslag F. N. T., “Federatie Van Nederlandse Trombosediensten” (2024).
- 21. Evans N. R., Todd O. M., Minhas J. S., et al., “Frailty and Cerebrovascular Disease: Concepts and Clinical Implications for Stroke Medicine,” International Journal of Stroke 17, no. 3 (2022): 251–259, 10.1177/17474930211034331. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Yagmur E., Frank R. D., Neulen J., Floege J., and Muhlfeld A. S., “Platelet Hyperaggregability Is Highly Prevalent in Patients With Chronic Kidney Disease: An Underestimated Risk Indicator of Thromboembolic Events,” Clinical and Applied Thrombosis/Hemostasis 21, no. 2 (2015): 132–138, 10.1177/1076029613490828. [DOI] [PubMed] [Google Scholar]
- 23. Potpara T. and Camm A. J., “A Double Paradox: End‐Stage Renal Disease‐High Thromboembolic and Bleeding Risk Management Using a Device With Risks of Thromboembolism and Bleeding,” Europace 27 (2025): 27(10), 10.1093/europace/euaf238. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Table S1: The 17 Dutch anticoagulation clinics participating in this study.
Table S2: Codes used for variable identification.
Table S3: Baseline characteristics of individuals receiving vitamin K antagonists by calendar year.
Table S4: Summary statistics of vitamin K antagonist treatment among individuals receiving vitamin K antagonists by calendar year.
Table S5: Clinical outcomes within 1 year after index date among individuals receiving vitamin K antagonists by calendar year.
Data Availability Statement
The study used data from the Federation of Dutch Thrombosis Services and nonpublic microdata from Statistics Netherlands, but these data cannot be shared directly by the authors. Under certain conditions, the data are accessible for statistical and scientific research. For further information: fnt@fnt.nl and microdata@cbs.nl.