Safety and efficacy of non–vitamin K antagonist oral anticoagulants compared with well-controlled warfarin in Thai patients with atrial fibrillation

We used retrospective data from Thai patients with AF aged ≥18 years who received OACs at the Central Chest Institute of Thailand from January 2017 to December 2019. We excluded patients with severe chronic kidney disease (creatinine clearance <30 mL/min/1.73 m²), rheumatic mitral stenosis, mechanical prosthetic heart valve, thrombocytopenia (platelet counts <100,000/mm³), heparin-induced thrombocytopenia, myeloproliferative disorders (essential thrombocythemia, chronic myeloid leukemia, polycythemia vera, agnogenic myeloid metaplasia), hyperviscosity syndrome, discontinuation of OAC due to invasive procedures or surgery more than 3 days, and taking warfarin with TTR <65%. The patients included where chosen using convenience sampling following the hospital number of each patient, which was found from a diagnostic code search and were divided into 2 groups: a group receiving NOACs and a group receiving well-controlled warfarin treatment. Demographic data including sex, age, medical history, CHA₂DS₂-VASc score, HAS-BLED score, SAMe-TT₂R₂ score, previous treatment and medications, thromboembolic events, bleeding events, and all-cause mortality were collected to cover a period of 12 months. TTR was calculated using the method described by Rosendaal et al. [10]. The primary outcome was total bleeding or thromboembolic events or both. Total bleeding events included major or minor bleeding. Major bleeding events were defined following the International Society on Thrombosis and Haemostasis (ISTH) definition for major bleeding [11]. Minor bleeding events were defined as those involving bleeding other than ISTH-defined major bleeding. Thromboembolic events included ischemic stroke or systemic embolization. The secondary outcome was all-cause mortality, total bleeding events including major or minor bleeding, and thromboembolic events including ischemic stroke or systemic embolization. This study was conducted in compliance with the Declaration of Helsinki and its contemporary amendments (2013) and reported following the recommendations of the RECORD and STROBE statements [12, 13]. The Human Research Ethics Committee of the Central Chest Institute of Thailand approved the study protocol (approval No. 018/2563).

We chose equal-sized groups of patients with AF who had received NOACs or well-controlled warfarin treatment and who were eligible for our study by convenience sampling following a sample size calculation. Of the 180 AF patients receiving OAC, 90 received NOACs and 90 received well-controlled warfarin treatment. Nearly 60% of all patients were male. The average age of all patients was 72.8 ± 10.1 years. The average CHA₂DS₂-VASc and HAS-BLED scores were 4 and 2, respectively. The average TTR in the group receiving well-controlled warfarin was 84.9% ± 9.8%. Most patients with AF receiving NOACs were prescribed rivaroxaban and apixaban (62.2%). More patients were taking nonsteroidal anti-inflammatory drugs (NSAIDs) in the NOACs group than in the well-controlled warfarin group. Baseline characteristics are shown in Table 1.

The group that received well-controlled warfarin treatment had more thromboembolic or total bleeding events or both than the group that received NOACs (odds ratio [OR] 3.17; 95% confidence interval [CI] 2.27–4.07; P = 0.01). There were more minor bleeding events in patients who received well-controlled warfarin (OR 3.75; 95% CI 2.79–4.71; P = 0.01). However, there was no difference in thromboembolic events, major bleeding, or all-cause mortality between the 2 groups (Table 2 and Figure 1).

Figure 1

Previous randomized controlled trials have demonstrated that in patients with AF, NOACs were not inferior to treatment with warfarin when considering stroke or systemic embolization or both and major bleeding events [4, 5, 6, 7]. However, a limitation of previous NOAC trials was the poor quality of anticoagulation control in the warfarin arm; a mean TTR <65% was observed in these trials [4, 5, 6, 7]. Moreover, the GARFIELD-AF registry has shown that patients receiving a vitamin K antagonist (66.7% of patients took warfarin) having a TTR <65% had more stroke or systemic embolism, major bleeding, and all-cause mortality than those with a TTR ≥65% [19].

To our knowledge, this is the first study to find that Thai patients with AF who received NOACs had significantly less frequent thromboembolic or total bleeding events or both than those who received well-controlled warfarin treatment, which is consistent with previous reports [14, 15]. In patients with NOACs, the decreased primary outcome was driven by a lower minor bleeding event rate than in patients with well-controlled warfarin, while the frequency of thromboembolic and major bleeding events was comparable between the groups.

The patients with AF under well-controlled warfarin treatment had more antiplatelets than those treated with NOACs. This might lead to increased minor bleeding in those with well-controlled warfarin treatment. However, the patients receiving NOACs also had a history of NSAIDs use, while those under well-controlled warfarin treatment had no history of receiving NSAIDs. This implies that NOACs may have a greater effect in lowering bleeding risk in patients who use NSAIDs than in those who receive warfarin. To date, COX-2 inhibitor use is widespread in many countries and may be used in patients who receive NOACs because of physicians' concerns about gastrointestinal bleeding from conventional NSAIDs, which may manifest as occasional instances of minor bleeding. The possibility of other effects should be clarified.

There are several limitations to the present study. First, we used a relatively small sample size, resulting in a low frequency of thromboembolic and major bleeding events between the NOACs and well-controlled warfarin groups. Second, this study was retrospective, and there may be discrepancies and incompleteness of the information provided, and causality cannot be demonstrated. Third, the patients' baseline characteristics between the 2 groups were different, leading to confounding factors such as age, HAS-BLED score, and use of NSAIDs. A propensity score matching analysis should be conducted to reduce selection bias, and outcomes should be analyzed for competing risk. However, because of the limited sample size, the baseline characteristics of these study patients cannot be reanalyzed. Moreover, simple random sampling could be potentially vulnerable to sampling error because it was not stratified following confounding factors. The patients in this trial were of Thai ancestry, limiting generalizability. Nevertheless, this study was the first, to our knowledge, to show the lower frequency of thromboembolic or total bleeding or both in patients with AF who received NOACs than in those under well-controlled warfarin treatment. Moreover, most patients with AF in the present study had a SAMe-TT₂R₂ score of ≥3, implying that NOACs resulted in fewer thromboembolic or total bleeding events or both than well-controlled warfarin treatment, despite a high SAMe-TT₂R₂ score.

Thai patients with AF who received NOACs had fewer thromboembolic or total bleeding events or both than those who received well-controlled warfarin treatment. However, there was no difference in the frequency of thromboembolic events and major bleeding between the 2 groups.