A total of 259 patients met the inclusion criteria and was included in the study. Among these 48.6% (n=126) were male. Mean age of all subjects was 67.9 ± 10.4 years. The demographic characteristics of study patients were shown in table 1.
The overall mean INR value was 2.3 ± 0.3. Using Caucasian therapeutic range, 34.5% of all measured INR values were in the therapeutic range, whereas 53.8% were sub-therapeutic, and 11.8% were supra-therapeutic. Overall TTR was 40.2 ± 17.1%. The mean proportion of duration individual’s INR being sub-therapeutic and supra-therapeutic were 47.8 ± 19.0% and 12.3 ± 9.3%, respectively. Among all, 7.7% of subjects were with ideal TTR during the inclusion period.
For Japanese therapeutic range, 44.1% of all measured INR values were in the therapeutic range, whereas 36.2% were sub-therapeutic, and 19.6% were supra-therapeutic. Overall TTR was 49.1 ± 16.1%. The TTR was significantly higher than that of using Caucasian therapeutic range (P<0.001). The mean proportion of duration individual’s INR being sub-therapeutic and supra-therapeutic were 30.6 ± 18.2% and 20.3 ± 11.4%, respectively. There were 12.4% of all subjects with ideal TTR.
Proportion of patients with ideal TTR were found to be statistically different among different subgroups of indications using Caucasian therapeutic range, as shown in table 1. Mean TTR of each indication of warfarin was calculated and compared within each therapeutic range group and among two therapeutic range groups (table 2). Within Caucasian therapeutic range, mean TTR with indication for AF was significantly higher than that of PHV (p<0.001), and that of both AF and PHV (p<0.001). Mean TTR with indication for neither AF nor PHV was also significantly higher than that of PHV (p=0.038). As with Japanese therapeutic range, mean TTR with indication for AF was significantly higher than that of both AF and PHV (p<0.001). Mean TTR using Japanese therapeutic range was significantly higher than that using Caucasian therapeutic range within each indication category.
Predictors of suboptimal TTR
Subjects were divided into four quartiles upon their TTR according to Caucasian therapeutic range (table 3). Predictors were determined upon regression across the four quartiles. Adjusted OR (aOR) for poor TTR was calculated. Results showed that younger age, concurrent use of aspirin, frusemide, famotidine, pantoprazole or simvastatin were associated with poorer TTR.
Impact of TTR on clinical outcome
Clinical outcomes were compared between two therapeutic ranges (table 4). Out of 259 patients, 35.9% of subjects experienced complications including thrombotic events and bleeding complications. Out of the 39 patients with thrombotic events, 41.0% of them had recurrent non-ST elevation myocardial infarction and 33.3% had stroke. For bleeding complications, 68.8% had minor bleeding while 31.3% of them experienced major bleeding episodes. People with ideal TTR had significantly fewer overall complications and bleeding complications, compared to those with non-ideal TTR, in both Caucasian and Japanese therapeutic range. Additionally, there were fewer thrombotic events in patients with ideal TTR using Japanese therapeutic range. All patients who had complications were those with non-ideal TTR, using Caucasian therapeutic range.
Patients were further stratified into quartiles based on TTR using Caucasian therapeutic range (table 5). Percentage of patients having overall complications and bleeding complications decreased from Quartile 1 to Quartile 4. Statistical test showed that there is a trend between each tested clinical outcome and TTR.
Impact of TTR on economic outcome
Healthcare costs were all expressed in terms of United States dollar (USD) per year (USD $1 = Hong Kong Dollar $7.8), as shown in table 4. Including all services related to warfarin, patients on average needed to spend USD $809.9/year. For economic outcomes, cost of INR checkups, cost of clinical visits and total healthcare cost were significantly lower in patients with ideal TTR in both Caucasian and Japanese therapeutic range. Patients with ideal Japanese TTR also had lower cost of hospitalization.
When focusing on Caucasian therapeutic range, healthcare provider needed to pay an addition of USD $530.1/year considering direct healthcare cost related to warfarin for each patient with non-ideal TTR, compared with a patient with ideal TTR. The average additional cost composed of USD $374.3/year for INR checkups and USD $9.0/year for clinic visits. Cost for hospitalization was USD $25.1/year for a patient with non-ideal TTR, compared to USD $0.3/year for a patient with ideal TTR. In conversion, patients with non-ideal TTR were hospitalized for approximately 2 days per year while those with ideal TTR were unlikely to be hospitalized.
A total of 174 completed OAK test, with mean score of 54.1% based on percentage correct out of 19 questions. The mean duration of warfarin therapy of this subgroup of patients within the study period was 4.8 ± 1.4 year. Only 24 patients (13.8%) achieved the satisfactory overall score of ≥75% in the test. Out of 19 questions, only four questions were answered correctly by ≥70% of respondents (table 6). Patients generally knew how to differentiate different strengths of warfarin (question 2, 81.6%), the reason for taking warfarin (question 8, 82.8%), what INR is (question 7, 83.3%) and when they should monitor sign of bleeding (question 15, 70.7%).
However, only half of the respondents knew the importance of dietary modifications (question 4, 50.6%), which vitamin warfarin interacts with (question 5, 44.8%), interpretation of INR test results (questions 9, 52.3%; question 20, 47.1%), when they should seek an immediate medical attention (question 11; 50.6%) and management for missing a dose (question 16, 48.9%). Approximately one-fourth of respondents gave the exactly opposite answer regarding interpretation of INR. Respondents perceived that there would be increased bleeding risk for INR below therapeutic range and increased thrombotic risk for INR above therapeutic range.
Patient knowledge regarding medication interactions with warfarin (questions 6, 35.6%, question 10, 21.8%) and the consequence of missing or skipping a dose (question 1, 28.2%; question 12, 25.3%) was insufficient. Nearly half of respondents believed no effect on missing one dose of warfarin.
Results from multiple linear regression showed that respondents with increasing age (adjusted β = -0.17, 95% CI (-0.23, -0.11), p = 0.001) or comorbid with diabetes (adjusted β = -1.21, 95% CI (-2.29, -0.12), p = 0.03) were more likely to score low in OAK test. On the contrary, respondents with concurrent hypertension (adjusted β = 1.68, 95% CI (0.56, 2.80), p = 0.004) or thyroid dysfunction (adjusted β = 2.38, 95% CI (0.80, 3.97), p = 0.003) were more likely to score high in OAK test. Respondents with better TTR were more likely to score high in OAK, yet marginally insignificant (adjusted β = 2.73, 95% CI (-0.21, 5.68), p = 0.069).