Article ID: CJ-17-1172
Background: Asian patients on warfarin therapy usually have lower international normalized ratio (INR) intensities than those recommended by Western clinical practice guidelines. This study evaluated whether a high INR reduces the incidence of thromboembolism (TE) or bleeding events in Asian patients with high CHA2DS2-VASc scores after valve surgery.
Methods and Results: Data of adult patients after valve surgery were retrieved from an integrated healthcare information system of a single hospital between 2014 and 2016. The INR was derived from the closest laboratory data before the index outpatient-clinic visit date. The endpoint of every record was determined as emergency room visit or hospitalization because of TE or bleeding event. A total of 37 TE or bleeding events were retrieved from 8,207 records; the annual incidence rate were 1.2% and 2.8% for low (0–2) and high (3–8) CHA2DS2-VASc score groups, respectively (P=0.007). The incidence rates were lowest for both groups at an INR of 1.5–2.0. High INR intensities did not reduce TE or bleeding incidence. INR >3.0 was associated with increased TE or bleeding incidence in the high-score group (6.8%/year vs. 2.0%/year, P=0.079).
Conclusions: The optimal INR is 1.5–2.5 for low- or high-score Asian patients after valve surgery. INR >3.0 was associated with increased TE or bleeding incidence in the high-score group.
Warfarin remains the anticoagulant of choice for patients with mechanical valve replacement (mVR), despite the availability of multiple novel oral anticoagulants. However, it is also one of the medications causing emergency department visits for adverse drug events.1,2 The updated (2017) American Heart Association/American College of Cardiology (AHA/ACC) guidelines recommend a target international normalized ratio (INR) of 2.5–3.0 for patients after mVR,3 and the 2017 European Society of Cardiology (ESC) guidelines recommend a target mean INR of 2.5–4.0 for low- to high-risk patients with prosthetic valves.4 By contrast, many studies of Asian populations have suggested a lower target INR for patients with prosthetic valves than that recommended by the AHA/ACC or ESC guidelines.5,6 We questioned whether this suggestion was based merely on clinical experience of early patients with low risk of thromboembolism (TE) or if a disparity exists in thrombogenicity between different ethnic groups, thus causing differences in the optimal INR for different ethnic groups. Therefore, in the present study, we analyzed an electronic healthcare database to evaluate the adverse events with different INR intensities in low- or high-risk patients with TE.
The CHA2DS2-VASc score is an established scoring system for estimating the risk of stroke in nonvalvular atrial fibrillation (AF) patients, and is widely used to decide the indication of anticoagulation therapy to prevent ischemic stroke in AF patients.7,8 In recent years, several studies had demonstrated the CHA2DS2-VASc score as a simple and useful tool to predict ischemic stroke and death in other patient groups, including coronary artery disease patients undergoing coronary artery bypass grafting surgery,9,10 patient with peripheral artery disease,11 and uremic patients under hemodialysis.12 The aim of this study was to evaluate whether high INR can reduce the incidence of TE or bleeding events in Asian patients with high CHA2DS2-VASc scores after valve surgery.
Data for the present study were retrieved from the integrated healthcare information system of the National Taiwan University Hospital (NTUH-IHIS) between 2014 and 2016; it includes delinked data of patient registration, pharmacy, billing, laboratory information, and pathology information systems, in addition to clinical information from the outpatient, inpatient, and emergency departments. Data of all patients aged over 20 years with a history of valve surgery were collected, and patients who were regularly followed up at outpatient department (OPD) clinics with a warfarin prescription were included in this study. Valve surgery included valve repair, mVR, or xenograft valve replacement. Only long-term warfarin prescription for a 3-month period was included in the study. Medical disease histories including hypertension, diabetes, congestive heart failure (CHF), peripheral artery occlusive disease (PAOD), and AF were retrieved using the International Classification of Diseases, 9th Revision, Clinical Modification-encoded diagnoses from the NTUH-IHIS database. Details regarding previous heart valve surgery (operation date, operative methods, valve types, and valve brands) were retrieved from our registration operative database.
According to a previous study by Lin et al,13 the ischemic stroke rate is much lower in Taiwanese with low CHA2DS2-VASc score (0.35%, 0.50%, 0.91%, 1.45% per year for score 0, 1, 2, 3, respectively) as compared with that in Caucasians7 (0.20%, 0.60%, 2.5%, 3.7% per year for score 0, 1, 2, 3, respectively). Concerning the high risk of bleeding in Asian populations, we chose CHA2DS2-VASc 0–2 as a low ischemic stroke risk group and CHA2DS2-VASc ≥3 as a high-risk group.
Warfarin dose and prescription date were retrieved for every data item. The INR for every data item was derived from the closest laboratory data before the index OPD visit date.
Warfarin dose fluctuation was defined as the standard deviation (SD) of all the prescribed warfarin doses for the same patient divided by the mean dose. INR fluctuation was defined as the SD of all estimated INR intensities for the same patient divided by the mean INR.
The endpoint of every record was determined using the number of emergency room (ER) visits or hospitalizations in the study hospital with a main diagnosis associated with TE or bleeding within 3 months from the index OPD date. Patients who lived far away from the study hospital (with a ground transportation time >1 h), as identified using the registered residential address, were excluded from the study in order to reduce the probability of including patients who were unable to travel to the study hospital from the first-aid hospitals.
The study project was approved by the Institutional Research Board of NTUH (NTUH-IRB-201609067RINA).
Statistical AnalysisAll continuous variables are presented as the mean±SD or with 95% confidence intervals, and were examined with Student’s t-test. Categorical variables were analyzed with χ2 tests. In multivariate logistic regression for significant variables of adverse events (TE or bleeding), all variables with P<0.05 by univariate analysis were put into the multivariate logistic regression with backward stepwise analysis. Hazard ratio (HR) was calculated by Cox proportional hazard model. Statistical significance was defined at P<0.05. The statistical analyses were performed using MedCalc Statistical Software version 17.8.6 (MedCalc Software bvba, Ostend, Belgium).
The study data set contained 8,207 records from 808 patients, with 4–14 records for each patient, and the mean number of records was 10.1±2.8. Low CHA2DS2-VASc score (defined as 0–2) and high CHA2DS2-VASc score (defined as 3–8) were observed in 66.7% and 33.3%, respectively, of the patients. The demographic data for the low- and high-score groups are listed in Table 1. In summary, the high-score group had a higher prevalence of potential risk factors, including age, CHF, hypertension, diabetes, previous stroke, and PAOD, than did the low-score group. AF was observed in 22.9% and 34.3% of the patients in the low- and high-score groups, respectively (P<0.001). The incidence rate of mVR was higher in the low-score group than in the high-score group (90.3% vs. 79.2%, P<0.001), and the incidence rate of xenograft valve was higher in the high-score than in the low-score group (3.9% vs. 10.6% in the low- and high-score groups, respectively, P<0.001).
| CHA2DS2-VASc 0–2 (5,472) |
CHA2DS2-VASc 3–8 (2,735) |
P value | |
|---|---|---|---|
| Age (years) | 56.7±11.7 | 70.2±10.0 | <0.001 |
| >65 | 1,191 (21.8%) | 2,129 (77.8%) | <0.001 |
| >75 | 224 (4.1%) | 1,052 (38.5%) | <0.001 |
| CHF | 771 (14.1%) | 1,041 (38.1%) | <0.0001 |
| Hypertension | 519 (9.5%) | 1,104 (40.4%) | <0.0001 |
| Diabetes | 438 (8.0%) | 1,005 (36.7%) | <0.0001 |
| Previous stroke | 131 (2.4%) | 937 (34.3%) | <0.0001 |
| PAOD | 299 (5.5%) | 901 (32.9%) | <0.0001 |
| CHA2DS2-VASc | 1.08±0.75 | 3.98±1.16 | <0.001 |
| Female sex | 2,216 (40.5%) | 1,782 (65.2%) | <0.001 |
| With antiplatelet drugs | 208 (3.8%) | 195 (7.1%) | <0.001 |
| With AF | 1,252 (22.9%) | 939 (34.3%) | <0.001 |
| Mechanical valve replacement | 4,939 (90.3%) | 2,260 (79.2%) | <0.001 |
| Aortic | 2,198 (40.2%) | 1,154 (42.2%) | |
| Mitral | 2,065 (37.7%) | 880 (32.2%) | |
| Double | 676 (12.4%) | 226 (8.3%) | |
| Xenograft valve replacement | 214 (3.9%) | 291 (10.6%) | |
| Aortic | 58 (1.1%) | 63 (2.3%) | |
| Mitral | 135 (2.5%) | 138 (5.0%) | |
| Double | 21 (0.4%) | 90 (3.3%) | |
| Tricuspid | 9 (0.2%) | 0 (0.0%) | |
| Mitral valve repair | 309 (5.6%) | 185 (6.8%) |
AF, atrial fibrillation; CHF, congestive heart failure; PAOD, peripheral artery disease.
The distribution of INR is presented in Table 2 and Figure 1. A low INR strategy is followed in our institute. Accordingly, an INR of 1.5–2.0 was observed in more than one-third of the patients in both the low- and high-score groups, followed by an INR of 2.0–2.5 and INR <1.5. In only 6% of cases was the INR data higher than 3.0.
| CHA2DS2-VASc 0–2 (5,472) |
CHA2DS2-VASc 3–8 (2,735) |
P value | |
|---|---|---|---|
| Warfarin dose (day) | 3.35±1.39 | 2.69±1.28 | <0.001 |
| Warfarin dose fluctuation (%) | 8.9 (8.7, 9.2) | 11.4 (10.9, 11.8) | <0.001 |
| INR | 1.95±0.59 | 1.92±0.66 | 0.08 |
| INR fluctuation (%) | 22.2 (21.9, 22.5) | 24.3 (23.8, 24.7) | <0.001 |
| INR level | 0.05 | ||
| <1.5 | 20.2% | 24.5% | |
| 1.5–2.0 | 38.3% | 36.7% | |
| 2.0–2.5 | 24.8% | 21.4% | |
| 2.5–3.0 | 11.1% | 10.7% | |
| >3.0 | 5.5% | 6.7% |
Warfarin fluctuation is presented as standard deviation (SD) of all prescribed warfarin doses divided by its mean for each patient. INR fluctuation is presented as SD of all INR values divided by its mean for each patient. INR, international normalized ratio.
International normalized ratio (INR) distribution in groups with CHA2DS2-VASc scores of 0–2 (A) and 3–8 (B).
The low-score group required higher warfarin doses than did the high-score group to achieve comparable INR intensities (warfarin dose: 3.35±1.39 vs. 2.69±1.28 mg in the low- and high-score groups, respectively, P<0.001; INR: 1.95±0.59 vs. 1.92±0.66 in the low- and high-score groups, respectively, P=0.08) (Table 2). The warfarin dose fluctuation was lower in the low-score group than in the high-score group (8.9% vs. 11.4%, P<0.001). In particular, the INR fluctuation was higher in the high-score group than in the low-score group (24.3% vs. 22.2%, P<0.001) (Table 2, Figure 2).
International normalized ratio (INR) fluctuation, defined as standard deviation (SD) of all INR values divided by the mean value for each patient, for (A) CHA2DS2-VASc score 0–2 group (low-score group) and (B) CHA2DS2-VASc score 3–8 group (high-score group). There is greater fluctuation in the high-score group than in the low-score group.
In total, 37 TE or bleeding events were retrieved from the database, comprising 19 TE events (ischemic cerebrovascular accident [CVA], limb ischemia, and bowel ischemia in 12, 4, and 3 patients, respectively) and 18 bleeding events (hemorrhagic CVA, gastrointestinal bleeding, and traumatic ecchymosis in 3, 9, and 6 patients, respectively), accounting for an annual TE rate of 0.93%, annual bleeding rate of 0.88%, and annual TE or bleeding rate of 1.80% for the study group. The annual TE rate, bleeding rate, and TE or bleeding rate were 0.89%, 0.94%, and 1.83% for those with mVR. The annual TE rate, bleeding rate, and TE or bleeding rate were 0.80%, 0.48%, and 1.29% for those with AF, and the annual TE rate, bleeding rate, and TE or bleeding rate were 0.98%, 1.05%, and 2.03% for those without AF (P=0.71, 0.21, 0.25, respectively). The annual TE rate was 0.00%, 0.85%, 0.90%, 0.99%, 1.43%, 2.26%, 0.00%, 3.81%, 19.1% for CHA2DS2-VASc score 0–8, respectively. The annual bleeding rate was 0.90%, 0.68%, 0.22%, 1.65%, 0.47%, 1.39%, 0.00%, 3.81%, 0.00% for CHA2DS2-VASc score 0–8, respectively. The annual TE or bleeding rate was 0.90%, 1.53%, 1.12%, 2.64%, 1.90%, 5.65%, 0.00%, 7.62%, 19.1% for CHA2DS2-VASc score 0–8, respectively.
Figure 3 illustrates the annual TE or bleeding incidence for the different INR data. The annual incidence rate of TE or bleeding events was lowest at an INR of 1.5–2.0 (1.0% per year), followed by INR of 2.0–2.5 (1.6% per year). When the INR was <1.5, the TE or bleeding event incidence rate was higher than that at INR of 1.5–2.0 (2.3% per year, P=0.09). When the INR was >2.5, the incidence rate was significantly higher than that at INR of 1.5–2.0 (3.1% and 3.3% per year at an INR of 2.5–3.0 and INR >3.0, P=0.02 and 0.04, respectively) (Figure 3A). After stratification by CHA2DS2-VASc score, the annual incidence rates of TE or bleeding events were 1.2% and 2.8% for the low- and high-score groups, respectively (P=0.007). The TE or bleeding incidence rates were lowest at an INR of 1.5–2.0 for the low- and high-score groups (0.6% per year, 2.0% per year, respectively) (Figure 3B). An increase in the INR did not reduce the TE or bleeding incidence rates. By contrast, an INR >3.0 was associated with a higher TE or bleeding incidence rate than an INR of 1.5–2.0 in the high-score group (6.8% per year vs. 2.0% per year, P=0.08) (Figure 3D); this difference was associated with a higher bleeding rate in the high-score group (4.4% per year vs. 0.8% per year, P=0.05) (Figure 3F).
(A–C) Annual thromboembolism (TE) or bleeding rate, annual TE rate, and annual bleeding rate for all data. (D–F) Data stratified by CHA2DS2-VASc 0–2 (solid line) and CHA2DS2-VASc 3–8 (dotted line) scores. Hazard ratio (HR) with 95% confidence interval shown for all data in (A–C) and for high CHA2DS2-VASc group in (D–F). *Statistically significant with a trend, **statistically significantly different. HR with P<0.10 indicated with asterisks.
Notably, 403 (4.9%) of the 8,207 data sets were associated with concomitant antiplatelet use (low-dose aspirin or clopidogrel) (Table 1). None of these patients had TE or bleeding events, which was lower than in the remaining records in the data set (37/7,804, 0.5%), but did not reach statistical significance (P=0.17).
Table 3 and Figure 4 demonstrate the univariate and multivariate analyses of potential variables for TE or bleeding rate. Univariate analysis revealed that a high CHA2DS2-VASc score, female sex, high warfarin dose fluctuation, and high INR fluctuation were associated with high HRs for TE or bleeding incidence, whereas a high warfarin dose and INR of 1.5–2.0 (compared with other INR) were associated with a relatively low HRs (Figure 4). In the multivariate analysis, only CHA2DS2-VASc score, high warfarin dose fluctuation, and high INR fluctuation were significant risk factors for TE or bleeding incidence, whereas female sex, high warfarin dose, and INR of 1.5–2.0 were not (Table 3).
| Univariate analysis | Multivariate analysis | |||
|---|---|---|---|---|
| HR (95% CI) | P value | HR (95% CI) | P value | |
| TE or bleeding | ||||
| CHA2DS2-VASc | 1.29 (1.09–1.53) | <0.01 | 1.24 (1.03–1.48) | 0.02 |
| Female sex | 2.49 (1.23–5.06) | <0.01 | NS | |
| Mechanical valve | 1.16 (0.41–3.27) | 0.785 | ||
| Atrial fibrillation | 0.63 (0.29–1.39) | 0.254 | ||
| Warfarin fluctuation (10%) | 1.83 (1.50–2.29) | <0.01 | 1.68 (1.35–2.09) | <0.01 |
| INR fluctuation (10%) | 1.50 (1.20–1.89) | <0.01 | 1.25 (0.98–1.61) | 0.07 |
| Warfarin dose (1 mg) | 0.69 (0.52–0.93) | <0.01 | NS | |
| INR 1.5–2.0 | 0.45 (0.21–0.99) | 0.030 | NS | |
| TE | ||||
| CHA2DS2-VASc | 1.38 (1.10–1.73) | <0.01 | 1.32 (1.05–1.69) | 0.02 |
| Female sex | 1.81 (0.71–4.59) | 0.214 | NS | |
| Mechanical valve | 0.75 (0.22–2.57) | 0.642 | NS | |
| Atrial fibrillation | 0.82 (0.30–2.28) | 0.705 | NS | |
| Warfarin fluctuation (10%) | 1.74 (1.30–2.30) | <0.01 | 1.64 (1.23–2.18) | <0.01 |
| INR fluctuation (10%) | 1.35 (0.96–1.91) | 0.086 | NS | |
| Warfarin dose (1 mg) | 0.82 (0.57–1.18) | 0.289 | NS | |
| INR 1.5–2.0 | 0.44 (0.14–1.32) | 0.143 | NS | |
| Bleeding | ||||
| CHA2DS2-VASc | 1.20 (0.94–1.54) | 0.15 | NS | |
| Female sex | 3.69 (1.21–11.2) | 0.02 | 2.99 (0.97–9.17) | 0.056* |
| Mechanical valve | 2.38 (0.32–17.9) | 0.40 | NS | |
| Atrial fibrillation | 0.46 (0.13–1.59) | 0.22 | NS | |
| Warfarin fluctuation (10%) | 1.92 (1.45–2.54) | <0.01 | 1.72 (1.27–2.34) | <0.01 |
| INR fluctuation (10%) | 1.65 (1.22–2.22) | <0.01 | 1.35 (0.96–1.90) | 0.081* |
| Warfarin dose (1 mg) | 0.80 (0.55–1.17) | 0.25 | NS | |
| INR 1.5–2.0 | 0.47 (0.15–1.43) | 0.18 | NS | |
*With a trend. CI, confidence interval; HR, hazard ratio; INR, international normalized ratio; TE, thromboembolism.
Hazard ratio for thromboembolism and bleeding events. ★Statistically significant risk factors. CHF, congestive heart failure; INR, international normalized ratio; PAOD, peripheral artery occlusive disease; y/o, years old.
The main finding of the present study was that the high-score group was associated with an increased TE or bleeding rate more than the low-score group. An INR of 1.5–2.0 was associated with the lowest TE or bleeding rate for both groups, but an INR >3.0 was associated with high TE or bleeding rates in the high-score group. Data from this study suggested an optimal INR of 1.5–2.5 for both low- and high-score groups of Asian patients after valve surgery.
The findings of our study are consistent with those from the ATRIA study,14 which indicated that adjustment of INR targets according to the CHA2DS2 score for patients with nonvalvular AF was not required. The difference between the ATRIA study14 and the present study is the ideal INR associated with the lowest TE or bleeding rate; the ATRIA study suggested a target INR of 2.0–3.0, whereas the current study results suggested a target INR of 1.5–2.5. One reason for this difference is that, in the present study, the ethnicity of almost the entire study population was Chinese, whereas that of the study population in the ATRIA study was mainly Caucasian.15
In Western countries, a target INR of 2.0–3.0 is usually recommended to prevent TE events in patients with nonvalvular AF.16 A meta-analysis based on American and European studies concluded that after mVR with new-generation valves, patients should receive warfarin to achieve a target INR of 2.5–3.5.17 The current (2017) AHA/ACC guidelines also recommend an INR of 2.5 for patients who have undergone aortic valve replacement (AVR) and have a low risk of TE events, 3.0 for patients who have undergone AVR and have an additional risk of TE events, and 3.0 for patients who have undergone mitral valve replacement.3 The optimal therapeutic range of INR with the use of warfarin has not been fully established in Asians, although an INR 2.0–3.0 is recommended as the optimal therapeutic range by the Taiwan Heart Rhythm Society and the Taiwan Society of Cardiology for the management of AF.18
A study of Japanese registry data showed that INR intensities ranging from 1.6 to 2.6 were effective in preventing TE events in patients with nonvalvular and valvular AF.6,19 Another study based on a cohort of Chinese patients also demonstrated that INR intensities of 1.8–2.4 appeared to be associated with the lowest incidence rate of major bleeding or TE events.20 Other studies of venous TE (VTE) events have also demonstrated a lower prevalence rate in the Asian population than in the Caucasian population,21,22 thus indicating a disparity in the thrombogenicity and response to warfarin between Asian and Western populations.
The reasons for the disparity in response to warfarin between Asian and Western populations are multifactorial, according to a literature review. In addition to the effect of various foods, alcohol, and differences in liver function, the use of herbal medications and complementary and alternative medicine (CAM) may play a role. Numerous herbal medications or CAM treatments popular in Asian countries have been identified as having antiplatelet or antithrombotic effects; therefore, they potentially interact with warfarin and increases the risk of bleeding events.18,23,24 Another potential explanation is the polymorphism of pharmacogenetics such as CYP2C9, CYP4F2, and VKORC1 genetic variants.25 According to previous studies, the total variability in warfarin dose explained by CYP2C9*3, VKORC1-1639G>A, and CYP4F2*3 is estimated at 40–63% in the Asian population.26 Furthermore, a higher number of genetic variants predisposing individuals to VTE is found in Europe and America than in Asia.22 Overall, Chinese patients required a 40–50% lower maintenance dose of warfarin than did Caucasian patients.22
An extensive review by Saffian et al27 summarized 22 algorithms for warfarin dosing prediction; of these, only 5 algorithms included female sex as a predictive factor, whereas age and weight (or body surface area [BSA]) were universally considered predictive factors. Female sex was shown to be associated with a higher HR of TE or bleeding events than male sex in the present study (Figure 4). Female sex in the present study possibly indicated a lower BSA, so would be associated with a lower warfarin dose (3.09±1.39 mg vs. 3.17±1.39 mg for female and male, respectively, P=0.010), higher warfarin dose fluctuation (0.11±0.11 vs. 0.09±0.10, P<0.001), higher INR (1.96±0.64 vs. 1.92±0.59, P<0.001), and higher INR fluctuation (0.23±0.11 vs. 0.22±0.10, P<0.001).
High INR fluctuation levels were found to be associated with an increased risk of TE or bleeding events in the present study, a finding similar to those in other studies.28–31 An explanation of INR fluctuation and the pathogenesis of TE events29 stated that a transient decrease in anticoagulation to subtherapeutic levels induces thrombus formation on the surface of the prosthetic valve. With subsequent increases in the anticoagulation level to the therapeutic range, the thrombus becomes less adherent to the valve surface, resulting in embolization. A previous study demonstrated that fluctuation in the INR may influence the prothrombotic state, indicated by the D-dimer level.28 Other studies have demonstrated that unstable anticoagulation intensity levels are associated with increased TE or bleeding event incidence.30,31 INR fluctuation tended to predict the risk of TE or bleeding events. Thus, reducing INR fluctuation is a stronger objective than achieving a specific INR after valve replacement surgery. Because high-intensity anticoagulation is associated with high levels of fluctuation,27 a low INR target with or without an antiplatelet agent would reduce INR fluctuation.
A notable finding in the present study was that in 403 records in the data set (4.9%) there was concomitant antiplatelet use (low-dose aspirin or clopidogrel), with 0.0% of the respective patients being associated with TE or bleeding events, compared with 0.5% in the remaining records in the data set (P=0.17). Although not statistically significant, we believe that this finding may shed light on the optimal anticoagulant strategy for patients with prosthetic valves. The 2017 AHA/ACC updated guidelines on valvular heart disease recommend 75–100 mg of aspirin daily in addition to anticoagulation with a vitamin K antagonist in patients with a mechanical valve prosthesis (Class I).3 However, a systemic review32 showed that adding antiplatelet therapy to oral anticoagulation reduces the risk of systemic embolism or death among patients with prosthetic heart valves, although it increases the major bleeding risk. Most of the studies included in the review reported a moderate INR (2.0–3.0). We question whether low-dose antiplatelet agents (e.g., 100 mg of aspirin) combined with low INR (1.5–2.5) might result in a low risk of TE or bleeding events. This hypothesis requires further validation because of the small sample size of patients using antiplatelet agents in the present study.
Regarding the HR of TE for xenograft valve (1,008 data sets), it was not lower as compared with those with a mechanical valve (7,199 data sets) (HR of TE for mechanical, 0.746; 95% CI 0.217–2.566) (Figure 4), we believe one important reason is that the xenograft group in the present study had higher CHA2DS2-VASc scores (2.618±1.871) as compared with the mechanical valve group (1.968±1.589) (P<0.001), indicating the xenograft group, which needed long-term warfarin use, was at high risk of TE according to their demographic characteristics. Whether the new-oral anticoagulants (NOAC) can decrease this group’s TE risk while not increasing the bleeding risk needs further study to answer.
Study LimitationsThe adverse event rates might have been underestimated because only patients sent to the study hospital’s ER were included. To reduce this bias, patients who lived outside the vicinity of the hospital (≈50 km in geographical distance) were excluded from the analysis. In this study, we did not explore concomitantly prescribed medication, such as amiodarone or phenytoin, which may influence the pharmacokinetic effects of warfarin. INR fluctuation >20% suggests the probability of overlap among the INR groups and a difference between the estimated INR and average INR. The INR estimated at the ER for each adverse event can reflect its true level. In the present study, time in the therapeutic range was not calculated because the target INR for each patient or each doctor was not constant.
The results of the present study indicated an optimal INR of 1.5–2.5 for Asian patients after valve surgery for both low- and high-score groups. Even for the high-risk group, INR of 1.5–2.5 and avoidance of warfarin fluctuation may have an acceptable clinical outcome. An increase in the INR was not associated with a decrease in the TE or bleeding event incidence for the high-score group. To avoid fluctuation of both INR and warfarin, low-intensity anticoagulation with an antiplatelet agent might be a treatment option; however, this hypothesis requires additional clinical data for confirmation.
We are grateful to the staff of the Department of Medical Research, NTUH, for providing us with the data from the integrated health care information system (NTUH-IHIS).
This work was partly supported by NTUH Research Grants NTUH-106-05.
All authors report no potential financial and nonfinancial conflicts of interest.
