Changes in anticoagulant prescription patterns over time for patients with atrial fibrillation around the world

Abstract Background Prescribing patterns for stroke prevention in atrial fibrillation (AF) patients evolved with approval of non‐Vitamin K antagonist oral anticoagulants (NOACs) over time. Objectives To assess changes in anticoagulant prescription patterns in various geographical regions upon first approval of a NOAC and to analyze the evolution of oral anticoagulants (OACs) use over time in relation to CHA2DS2‐VASc and HAS‐BLED risk profiles. Methods Global Registry on Long‐Term Oral Antithrombotic Treatment in Patients with Atrial Fibrillation (GLORIA‐AF) Phases II and III reported data on antithrombotic therapy for patients with newly diagnosed AF and ≥1 stroke risk factor. We focused on sites enrolling patients in both phases and reported treatment patterns for the first 4 years after initial NOAC approval. Results From GLORIA‐AF Phases II and III, 27 432 patients were eligible for this analysis. When contrasting the first year with the fourth year of enrolment, the proportion of NOAC prescriptions increased in Asia from 29.2% to 60.8%, in Europe from 53.4% to 75.8%, in North America from 49.0% to 73.9% and in Latin America from 55.7% to 71.1%. The proportion of Vitamin K antagonists (VKAs) use decreased across all regions over time, in Asia from 26.0% to 9.8%, in Europe from 35.5% to 16.8%, in North America from 28.9% to 12.1%, and in Latin America from 32.4% to 17.8%. In the multivariable analysis, factors associated with NOAC prescription were as follows: enrolment year, type of site, region, stroke and bleeding risk scores, and type and categorization of AF. Conclusions During 4 years after the approval of the first NOAC, NOAC use increased, while VKA use decreased, across all regions.

VKA in Phase II showed substantial overlap, as measured by comparison of propensity score distributions. During Phase III, follow-up data were collected for up to 3 years regardless of prescribed antithrombotic therapy. 2 Adults with nonvalvular AF and ≥1 CHA 2 DS 2 -VASc risk factor score for stroke were included. Stroke and bleeding risks were assessed using the CHA 2 DS 2 -VASc and HAS-BLED risk scores. 3,4 Patients were managed according to local practice. This report includes regions and sites enrolling patients during Phases II and

III.
Standard electronic case reports forms (eCRFs) were used to collect patients' baseline characteristics and follow-up observation F I G U R E 1 (A) Enrolment period in Phases II and III of the GLORIA-AF registry per country in calendar dates. (B) Enrolment period in Phases II and III of the GLORIA-AF registry per country in years after NOAC availability data. Baseline therapy was the treatment prescribed for long-term anticoagulation subsequent to the diagnosis of AF and recorded at the baseline visit.
Time zero in a participating country was set to the date of the baseline visit for the first patient in each country. The first year of enrolment for a participating country was the first year after time zero in that country. Most countries continued enrolment for up to 4 years.
In this paper, we classify newly enrolled patients according to which prescribed treatment they received at their baseline visit, by year of enrolment. first year of enrolment. 5 Factors associated with OAC prescription patterns over time were evaluated using log-binomial regression models, providing estimates of relative probability for NOAC prescription (vs. VKA prescription). Both univariate and multivariable log-binomial regression analyses were fit to evaluate the crude as well as adjusted probability ratios together with 95% confidence intervals (CIs).

| Statistical analysis
Missing data were imputed using multiple imputation. The imputation model was constructed with 56 baseline patient characteristic variables including those used in the multivariable analyses. The COPY method was used to obtain approximate maximum likelihood estimates when log-binomial models failed to converge. 6 Statistical analyses were performed using SAS version 9.4 (SAS Institute, Inc, Cary, NC).

| RE SULTS
There were 27 432 eligible patients who enrolled in GLORIA-AF during Phases II and III and who qualified to be included in this F I G U R E 2 Temporal trends of antithrombotic therapy prescription globally. NOAC, non-Vitamin K antagonist oral anticoagulants; OAC, oral anticoagulation; VKA, Vitamin K antagonists analysis. Of 8969 patients who enrolled in the first year, 46.6% were prescribed NOAC, 31.9% were prescribed VKA, and 21.5% were prescribed no OAC. Of 4388 patients enrolled in the fourth year, 71.6% received NOAC, 14.1% received VKA, and 14.3% received no OAC ( Figure 2). A similar trend in treatment pattern over time, ie, increase in NOAC and decrease in VKA, was observed for Europe and North America. From the third to fourth year, an increase in NOAC prescriptions and a decrease in VKA or no OAC prescription was reported in Asia ( Figure 3). The prevalence of non-OAC slightly decreased from Years 1-4, except Latin America ( Figure 2).
Baseline characteristics of patients prescribed NOAC by region are summarized in Table 1. Paroxysmal AF was less prevalent in patients with NOAC during the first year of enrolment than in patients with NOAC during the last year of enrolment in Europe, North America, and Latin America. The standardized differences for stroke and bleeding risk scores (CHA 2 DS 2 -VASc and HAS-BLED) between the last and first year for NOAC patients were small in Europe, North America, and Latin America (less than 0.1).
Baseline characteristics of patients prescribed VKA by region are shown in Table 2. The standardized differences for CHA 2 DS 2 -VASc were small between the patients enrolled during the last and first year in North America, while they were more than 0.1 in Europe, Asia and Latin America. The standardized differences for HAS-BLED were more than 0.1 between last and first year in Asia, North America, and Latin America.
Prescription of oral antithrombotic treatment by region is presented in Table 3 and Figure 3. A decrease in no OAC use including acetylsalicylic acid (ASA) was reported in Asia between the third and fourth year of enrolment. An increase in NOAC use and decrease in VKA and no OAC use including ASA was present in Europe and North America. An increase in NOAC and a decrease in VKA were reported between second and fourth year in Latin America.

| Factors associated with NOAC prescription in phases II and III
Results from univariate analyses are presented in Table 4. In the multivariable log-binomial regression analysis, factors strongly associated with increased prescription of NOAC were as follows: enrolment year, type of site (higher probability outside of a university hospital, such as GP/primary care, specialist office, community hospital, and other), and region (higher prescription probability in North America compared with Europe) ( Table 4).
Factors associated with decreased prescription of NOAC were the following: HAS-BLED score ≥3 (compared with HAS-BLED score <3), categorization of AF (lower probability of symptomatic AF compared with asymptomatic AF), CHA 2 DS 2 -VASc score ≥2 (compared with CHA 2 DS 2 -VASc score <2), and type of AF (lower probability of persistent or permanent AF compared with paroxysmal AF) ( Table 4).          Medical treatment reimbursed by, n (%)

| Prescription of antithrombotics over time by HAS-BLED score class
Regional patterns of prescription of antithrombotic drugs over time by HAS-BLED score class are presented in Table S3. The interval changes in those with HAS-BLED scores ≥3 who received NOACs in Asia were +4.1%. The corresponding interval changes for Europe, North America and Latin America were +20.7, +20.3, and +22.5%. The interval changes in those with HAS-BLED scores ≥3 who received VKAs in Asia were −9.7%. The corresponding interval changes for Europe, North America, and Latin America were −18.5%, −17.2%, and −21.3% (Table S4).

| D ISCUSS I ON
We found that use of NOAC increased and VKA decreased over time The use of NOAC appears to have increased over time in Europe.
This finding is consistent with other reports. [7][8][9] After the release of NOAC, the prevalence of NOAC use rose steadily in Japan. such as CHF, diabetes or vascular disease, than those who use NOAC during their last year of enrolment in Europe. In other studies, patients prescribed VKAs also had more comorbidities than those prescribed NOACs. 13 -14 Similar to our study, patients prescribed VK A were more likely to have permanent AF than those prescribed NOAC in each region. 13 -14 Interestingly, in Korean patients those who used VK As were less likely to have prior stroke/TIA/systemic embolism than those who used NOACs. 15 In our study, the proportion of patients who were prescribed a reduced dose of NOAC is highest in Asia, a finding that could be related to smaller body size in Asian patients. The risk of major bleeding seems to be higher in Asian patients medicated with VKAs than in non-Asian patients. 16 In one study, lower NOAC doses were frequently used in Asian patients in routine daily practice. However, unjustified underdosing of apixaban was associated with a less apparent clinical benefit over warfarin in patients. 17 The data from baseline Phase II of GLORIA-AF showed that considerable numbers of patients were not treated with OAC, especially in Asia and North America. 18  Furthermore, year of enrolment, type of site, region, type and categorization of AF, and stroke and bleeding risks are associated with NOAC prescription in the combined Phase II and III data in our analysis. A similar pattern was found in another report where persistent or permanent AF was inversely associated with NOAC prescription. 13 Also, the year of enrollment was associated with NOAC prescription.

| Limitations
These findings may not generalize to the entire global nonvalvular AF patient population or even to the patient population of the participating countries, because the study is restricted to patients with a CHA 2 DS 2 -VASc score ≥1. Furthermore, this analysis represents only a snapshot of the prescribing practice in the course of treatment and does not take into account treatment continuation, switching or adherence. These issues were addressed in other reports from GLORIA-AF. Data on the reasons for OAC non-prescription were not collected.

| CON CLUS IONS
In this global registry of prospectively enrolled AF patients, NOACs have been more commonly prescribed than VKA. During 4 years after approval of the first NOAC for stroke prevention in AF, NOAC use increased over time, while VKA use decreased across all regions.

ACK N OWLED G EM ENTS
The study was funded by Boehringer Ingelheim. The authors thank the patients who participated in this trial, their families, the investigators, study co-ordinators, and study teams.