Lifelong oral anticoagulation therapy for prevention of thromboembolic events is recommended in all patients who undergo mechanical heart valve replacement [1‐3]. However, bleeding complications that result from excess dose of anticoagulants adversely impacts patients’ quality of life and can cause unnecessary morbidity and mortality [4]. Risk of thromboembolism and bleeding is commensurate with the level of anticoagulation. Current American and European clinical guidelines recommend a higher international normalized ratio (INR) for anticoagulant therapy after mechanical mitral valve replacement, because higher rates of thromboembolic complications were reported when the mechanical valve was in the mitral position, as compared to when the mechanical valve was in the aortic position [5‐8]. A target INR range of 2.5–3.5 is the current recommendation in patients who have undergone mechanical mitral valve replacement [5‐7]. These data were based on results from studies conducted in Western countries; however and according to our review of the literature, data regarding the safety of warfarin in Asian populations remain insufficient. Previous data have shown that Asian population with atrial fibrillation who received warfarin has an increased risk of intracranial hemorrhage up to 4 times compared to Caucasians [9]. Asian population had a greater proportion of intracerebral bleeding as a stroke subtype when compared to Caucasians [10]. There is no clear explanation for the increased risk of intracerebral bleeding in Asian population [10]. The few studies that were conducted in Asian populations who received mechanical heart valve replacement generally recommended a lower INR level as being appropriate for anticoagulant therapy [11‐13]. However, the currently available data from Thai population are insufficient for defining the optimal INR level for anticoagulant therapy in Thai mechanical valve recipients. Accordingly, the aim of this study was to identify the optimal INR level for warfarin therapy after mechanical mitral valve replacement in Thai patients.

Two hundred patients were included and followed till 707 patient-years. Mean duration of follow-up was 3.53 ± 1.27 years. Mean age of patients was 50.30 ± 10.9 years, with a gender proportion breakdown of 116 (58%) females and 84 (42%) males. Two-thirds (70.5%) of patients had rheumatic valve disease. All patients received bileaflet mechanical valves. Left atrial appendage closure was performed during surgery in 19 patients (9.5%). Twenty-seven (13.5%) patients had concomitant aortic valve replacement (AVR), and 147 (73.50%) of patients had atrial fibrillation. Among patients with atrial fibrillation, only 4 (2.7%) were paroxysmal atrial fibrillation in nature. The number was too small to make any comparison for the incidence of thromboembolic events. Hypertension was the most common comorbidity (26.5%). Seven (3.5%) patients received concomitant aspirin. Baseline demographic characteristics, clinical characteristics, and laboratory parameters of Thai mechanical mitral valve replacement patients are shown in Table 1.

A total of 31 thromboembolism and bleeding events (4.38 per 100 patient-years) were documented during the follow-up period (Table 2). Eleven patients experienced 13 thromboembolic events (3.42 per 100 patient-years), and 12 patients experienced 18 total bleeding events (5.50 per 100 patient-years). Intracranial bleeding occurred in 3 patients (2.62 per 100 patient-years). There was one patient who died from intracerebral hemorrhage and another patient who died from a large cerebral infarction. The percentage of patient time spent within INR 2.5–3.4, INR < 2.5, and INR > 3.4 was 41.96, 54.04, and 4%, respectively. We analyzed and compared the incidence density of thromboembolic (Fig. 1a), total bleeding, clinically significant bleeding, and minor bleeding events (Fig. 1b, c, d) between INR groups using chi-square test. The average dose of warfarin in the study sample was 25.6 ± 9.5 mg per week. The frequency of controlled INR of 2.5–3.5 based on standard recommendation was 40.4 ± 16.3%. Average frequency of INR monitoring was once every 3.1 ± 1.0 month. One-hundred ninety-one patients (91.5%) had warfarin dose optimization. The frequency of based on the INR test was 3.31.8 times during the study period. Analysis by linear mixed models (fixed effect) to study the difference in INR levels in patients with and without events at time indicates that the average INR levels of bleeding, minor bleeding and major bleeding group was more than non-event group (p < 0.001) but the average INR levels of thromboembolism group was lower than non-event group (p < 0.001)(Table 3).

The optimal INR range in Thai patients with mechanical mitral valve replacement to be 2.0–3.4. An INR level greater than 3.4 significantly increased the incidence density of total bleeding events (p = 0.001; rate ratio 21.18, 95% confidence interval 6.88–49.46), whereas an INR level less than 2 significantly increased the incidence density of thromboembolic events (p < 0.001; rate ratio 29.00, 95% confidence interval 6.33–269.26). Accordingly, we concluded that the optimal INR level was 2.0–3.4, as show in Fig. 2. The overall event rate was lowest in the 2.0 to 3.4 INR range, with statistically significant differences being observed between INR 2.3 to 4 and < 2 (p < 0.001) and between INR 2.3 to 4 and > 3.4 (p < 0.001) (Fig. 3).

This is the first study in INR levels for warfarin therapy in Thai patients with mechanical mitral valve replacement. This study found that Thai patients with mechanical mitral valve replacement required lower INR levels for warfarin therapy than Western patients according to the clinical guidelines [5‐7]. However, the results of this study are consistent with the results of studies in mechanical mitral valve replacement patients from China and Japan [11‐13]. These observed disparities between Asian and Western populations may be attributable race-related differences.

In a report from China, Xin-Min Zhou, et al., 2005 [11] described a rate of 5.83 per 100 patient-years for bleeding events, and 0.26 per 100 patient-years for thromboembolic events in patients who received a Carbomedics valve replacement. Compared to our data, Zhou’s study included both mechanical mitral and aortic valve, had a shorter time of follow-up (377 vs 706 patient-years), and had fewer thromboembolic event (1 event vs 13 events). They concluded that low-intensity oral anticoagulant (1.4–2.0) in patients with mechanical valves could effectively prevent thromboembolism and markedly decrease bleeding events. The optimal INR range in our study was 2.0–3.4 which is higher than Zhou’s and may be related to the study population who was post mitral valve replacement in all cases in our study but mitral or aortic valve replacement for their study.

Heinrich Koertke, et al., 2010 [19] followed 1137 German patients in a prospective randomized multicenter trial and demonstrated the efficacy and safety of very low, self-managed INR doses in patients with aortic valve replacement (INR target value: 2.0, range: 1.6–2.1), and in patients with mitral valve or double valve replacement (INR target value: 2.3, range: 2.0–2.5). The results of this study is in a similar direction to our study and indicate that even in Caucasians, a carefully managed strategy can reduce the INR target in patients after mechanical valve replacement. In France, the AREVA study reported an incidence of thromboembolic complications that was similar between patients with a target INR range of 2.0–3.0 and patients with a target INR range of 3.0–4.5; however, there were fewer bleeding complications in the low-dose group [20]. The study population of our study were different from the aforementioned studies [11, 19, 20]. They enrolled both mitral and aortic valve replacement but we enrolled only patients with mechanical mitral valve. Majority of patients enrolled in AREVA study were those with aortic valve replacement. Patients with aortic valve replacement usually require a lower INR target than those with mitral valve replacement [21, 22].

Previous Asian studies reported a higher incidence of bleeding complications among Asian warfarin users than among Western users [9]. The overall event rate for ischemic stroke or TIA in the present study was 5.5%, while the overall event rate for bleeding complications was 9% (4.5% clinically significant and 4.5% minor). Therefore, the overall rate of bleeding events found in this study was similar to the rates reported from previously published Asian and Western studies but the rate of thromboembolic event in our study were higher [11‐13, 19, 20]. The higher rate of ischemic stroke may be related to a suboptimal TTR control in our population compared to other studies [23, 24]. Doctors may be fear of bleeding and keep a relatively low INR levels. It has been reported from the GARFIELD AF registry that TTR of Asian population is lower than Caucasians [25]. But despite the lower TTR, Asian population had a higher rate of major bleeding and intracerebral hemorrhage compared to Caucasians [10]. The higher ischemic stroke rate in our study as compared to study from China [11] may be due to the study population. We enrolled patients with mechanical mitral valve prosthesis whereas study from China enrolled both mitral and aortic valve prosthesis. Patients with aortic valve prosthesis had a lower thrombotic complication and required a lower INR target compared to those with mitral valve prosthesis [21, 22].

From the results of the present study, we concluded that 2.0–3.4 is the INR range at which the lowest incidence of thromboembolism and bleeding complications should be anticipated. When we combined the 6 INR groups into 3 INR groups (INR less than 2.0, INR 2.0–3.4, and INR greater than 3.4), INR 2.0–3.4 had significantly lower combined thromboembolic and bleeding complications than both INR greater than 3.4 and INR less than 2.0. The explanation for the lower INR target may be related to a higher bleeding event and intracerebral hemorrhage in Asian population as compared to Caucasians [26]. When we consider the INR level that is most suitable for the balance of risk and benefit ratio towards a lower target than the general recommendation.

This study has several mentionable limitations. First and consistent with the retrospective nature of this study, some patient data may have been missing or incomplete. For example, some physicians or patients may have overlooked or ignored some episodes of minor bleeding, so those episodes would not have been documented. Second, the size of the study population was relatively small. Although we ran power analysis to calculate sample size needed for the primary objective of this study, we felt that the number of patients of 200 is relatively small. Further study with a larger number of study population may be needed to confirm the results of this study. Third, the patients enrolled in this study were from a single center. The nature of hospital and medical practice may be different. Also our center is Thailand’s largest tertiary referral hospital, which means that we are often referred patients with complicated and intransigent conditions. As such, it is possible that our findings may not be generalizable to patients with the same condition in other settings. Lastly, since we excluded patients with history of stroke or bleeding, we did not have data on optimal INR for this group which are considered a high risk group. We excluded patients with history of thromboembolic event and bleeding due to some reasons. For patients with history of stroke, most of them had some degree of residual neurological deficit. This could lead to problem of the documentation of recurrent stroke. For patients with history of bleeding, they could have some underlying disease that prone to recurrent bleeding even without warfarin. Moreover, many patients with history of stroke of bleeding may be a referred case from other hospital and may be difficult for the investigators to collect the data on the time relation of anticoagulant use and the onset of stroke or bleeding. Further prospective study in a larger study population is needed to confirm the findings of this study, and to narrow the optimal INR range in patients with certain demographic and clinical profiles.

The optimal INR level was within the range of 2.0 to 3.4 in our cohort of Thai mechanical mitral valve replacement patients. This INR target may also be applied in Asian population. This suggestion should allow more room for INR adjustment which could avoid bleeding event in patients with mechanical mitral valve without losing the benefit of warfarin.