J Thromb Thrombolysis (2014) 38:226–234 DOI 10.1007/s11239-013-1017-6
Bleeding events and associated factors in a cohort of adult patients taking warfarin in Sarawak, Malaysia Frances Edwards • Paul Arkell • Alan Yean Yip Fong • Lesley M. Roberts • David Gendy • Christina Siew-Hie Wong • Joanna Chee Yien Ngu • Lee Len Tiong • Faridha Mohd Salleh Bibi Lana Yin Hui Lai • Tiong Kiam Ong • Michael Abouyannis
•
Published online: 15 November 2013 Ó Springer Science+Business Media New York 2013
Abstract Evidence is emerging that rates of adverse events in patients taking warfarin may vary with ethnicity. This study investigated the rates of bleeds and thromboembolic events, the international normalised ratio (INR) status and the relationship between INR and bleeding events in Malaysia. Patients attending INR clinic at the Heart Centre, Sarawak General Hospital were enrolled on an ad hoc basis from May 2010 and followed up for 1 year. At each routine visit, INR was recorded and screening for bleeding or thromboembolism occurred. Variables relating to INR control were used as predictors of bleeds in logistic regression models. 125 patients contributed to 140 personyears of follow-up. The rates of major bleed, thromboembolic event and minor bleed per 100 person-years of follow-up were 1.4, 0.75 and 34.3. The median time at target range calculated using the Rosendaal method was 61.6 % (IQR 44.6–74.1 %). Of the out-of-range readings, 30.0 %
F. Edwards D. Gendy M. Abouyannis c/o Dr. L Roberts, Primary Care Clinical Sciences, School of Health and Population Sciences, College of Medical and Dental Sciences, University of Birmingham, Edgbaston, Birmingham B15 2TT, UK P. Arkell (&) c/o Diane Jackson, West Midlands Deanery, Clinical Education Centre, University Hospital North Staffs, Stoke-on-Trent ST4 6QG, UK e-mail: [email protected] A. Y. Y. Fong C. S.-H. Wong J. C. Y. Ngu L. L. Tiong F. M. S. Bibi L. Y. H. Lai T. K. Ong Sarawak Heart Centre, Kuching, Sarawak, Malaysia L. M. Roberts Primary Care Clinical Sciences, School of Health and Population Sciences, College of Medical and Dental Sciences, University of Birmingham, Edgbaston, Birmingham B15 2TT, UK
123
were below range and 15.4 % were above. INR variability, (standard deviation of individuals’ mean INR), was the best predictor of bleeding events, with an odds ratio of 3.21 (95 % CI 1.10–9.38). Low rates of both major bleeds and thromboembolic events were recorded, in addition to a substantial number of INR readings under the recommended target range. This may suggest that the recommended INR ranges may not represent the optimal warfarin intensity for this population and that a lower intensity of therapy, as observed in this cohort, could be beneficial in preventing adverse events. Keywords Anticoagulation Warfarin Haemorrhage Malaysia
Introduction Warfarin is the most commonly prescribed anticoagulant worldwide for individuals at risk of thromboembolic events [1]. Unfortunately the therapeutic index is narrow, with suboptimal treatment risking thromboembolism, and overtreatment threatening haemorrhage [2, 3]. According to US and UK guidelines, an international normalised ratio (INR) range of 2.0–3.0 is recommended, although higher targets are needed in populations at increased risk of thromboembolism, such as those with mechanical heart valve replacement [4, 5]. This target range has been applied worldwide, although data was predominantly collected from the European and North American regions [6–8]. It has been established that ethnicity influences warfarin sensitivity [9–11], and the US Food and Drug Administration (FDA) states that Asian populations need lower doses of warfarin than Caucasian populations to achieve the target INR [12]. A study of low dose Warfarin based in Japan,
Bleeding events and associated factors in a cohort of adult
found that major haemorrhage was more frequent than in Western populations [13]. Cohort data from China suggested a target INR range of 1.8–2.4 to be associated with lowest rates of major bleeding or thromboembolic events [14]. Additionally, one study in Taiwan reported that an INR of \2.0 was not associated with increased thromboembolic events [15], as it has been in Caucasians [16]. Although data regarding use of anticoagulation in Malaysia is scarce, incidence of cardiovascular disease is increasing, and this is likely to be associated with increased use of warfarin therapy [17]. A recent cross sectional study estimated prevalence of asymptomatic atrial fibrillation (AF) at 0.75 % in the Malaysian primary care setting [18]. While modern alternatives to warfarin, such as dabigatran and rivaroxaban,are increasingly being used in Europe and America [19], their relatively higher cost limit their widespread use in developing countries, such as Malaysia. This study is the first to describe rates of adverse events in a Malaysian population taking Warfarin. The Malaysian population is diverse and composed of people of Malay, Chinese, Indian and Indigenous tribal origin, and data here will add to Asia’s evidence base for oral anticoagulation use. Aims 1.
2. 3.
To describe the rates of both haemorrhagic and thromboembolic adverse events in a Malaysian population taking warfarin To describe the intensity of warfarin therapy in the sample population To investigate predictors of adverse events, relating both to INR status and to the risk factors which have been previously established in Western populations
Methods Study setting Participants were drawn from attendants of the cardiology anticoagulation clinic at the Heart Centre of Sarawak General Hospital (SGH) in Kuching, Malaysia. This is the main tertiary referral centre for the state of Sarawak and the only public hospital serving the city of Kuching. The American Antithrombotic Therapy and Prevention of Thrombosis (AATPT) guidelines are followed for target ranges, dose initiation and dose titration [4]. Patients with INR values within their target range were managed by pharmacists, while those outside target range were seen by a doctor. Study subjects, procedures and measurements Recruitment was conducted by a dedicated research pharmacist on an ad hoc basis between May 2010 and
227
December 2011. Data were collected prospectively up to and including the visit after the participant had been in the study for a year. Patients were eligible to be enrolled if they were currently taking warfarin, were over 18 years and spoke one of three main languages (Bahasa Malaysia, Mandarin or English). At the enrolment visit the following data were collected: demographics (age, gender, ethnicity, and occupation), history of warfarin use (date started, dose, and indication), bleeding history, past medical history, alcohol use, weight and laboratory measurements (full blood count, liver function tests, urea and electrolytes, and INR). Participants returned according to the timeframe of their routine care, and on these subsequent visits the INR and current warfarin dose was recorded. At every visit the clinician followed a proforma to identify bleeding events. All participants were asked if they had experienced the following: bloody or black stool, blood in the urine, coughing blood, bleeding gums, skin haematoma, bruising skin, nose bleed, or heavy or irregular vaginal bleeding. Participants were asked of any hospital admissions at every visit, and all such cases underwent a review of the medical records by the research team to identify haemorrhagic or thrombotic events. Participants who stopped attending the clinic were followed up by telephone by the research team to identify adverse events or deaths. Events were then checked with the hospital records. Statistical analysis Only participants completing at least 1 year of follow-up were included, and person-years was the unit of analysis. Participants were censored from the point of time following the occurrence of a major bleed or thrombotic event. Participant demographics and clinical characteristics were described using proportions, counts, means with standard deviations (SD), and medians with interquartile ranges (IQR) as appropriate. The overall percentage time at target range (TTR) was calculated using the method described by Rosendaal et al. [18], which uses linear interpolation to estimate the time spent at each INR value. INRpro provided the software to calculate this [19]. TTR was also calculated using the traditional method as described in the Chest guidelines: number of INR measurements within target range divided by total number of INR measurements [4]. Bleeds were categorised as either major or minor according to the definition of the International Society on Thrombosis and Haemostasis [20]. As such a major bleed was defined as either a fatal bleed, a symptomatic bleed in a critical area (intracranial, intraspinal, intraocular, retroperitoneal, intraarticular, pericardial or intramuscular with compartment syndrome) or a bleed causing a fall in
123
228
F. Edwards et al.
haemoglobin of 2 g/dL or leading to transfusion of two or more units of whole blood or red cells. Rates of major bleeds, minor bleeds and thromboembolic events per person-year on warfarin were calculated with 95 % confidence intervals (CI). Univariate logistic regression was undertaken using the outcome ‘any bleed’ to include major or minor bleeds. The following variables were entered to identify predictors of bleeding: demographics (age, gender and ethnicity), characteristics of warfarin use (number of years on warfarin, warfarin dose and INR target range) and seven variables relating to INR control and variability. These variables were selected because they have been used in other studies, are clinically relevant and show biological plausibility, and they were as follows: [4, 21–24]
Table 1 Comparison of included participants with those lost to follow up Included participants
(5) (6) (7)
TTR (Rosendaal method) TTR (Traditional method) More than or equal to 1 INR reading above target Percentage of INR readings above target range (% above range) Mean of all INR values during the study period for each participant (mean INR) The INR range for each participant (INR range) The SD of the mean INR for each participant (SD of INR)
Variables reaching p \ 0.10 level of significance were entered into multivariate logistic regression using the forward stepwise model analysis to identify independent predictors. Data were analysed using SPSS (v19) for Windows.
58.4
54.8
0.144
Gender (% male)
51.2
50.0
0.899
Target INR range (% range 2.5–3.5)
29.6
27.8
0.832
Enrolment INR (mean)
2.56
2.31
0.153
Table 2 Demographic variables and warfarin indication Mean
SD
58.4
13.3 Number
%
Male
64
51.2
Female
61
48.8
Malay
58
46.4
Chinese
47
37.6
Indigenous tribes
19
15.2
1
0.8
Operator/fabricator/labourer Service operations
19 8
15.2 6.4
Technical/sales/admin support
11
Gender
Ethnicity
Indian Occupation
Managerial/professional
Ethics
p value
Age (mean)
Age
(1) (2) (3) (4)
Participants lost to follow up
None/retired
8.8
5
4
82
65.6
Indication for warfarin
Ethical approval was obtained from the Ministry of Health Medical Research and Ethics Committee, Malaysia and the Population Sciences and Humanities BMedSc Internal Ethics Committee, University of Birmingham, UK. Informed written consent was gained from every participant before enrolment.
Replaced valve Of which C1 mechanical Valve lesion Congestive heart failure
60
48.0
51
40.8
18
14.4
3
2.4
Atrial fibrillation
79
63.2
Cardiac thrombus
9
7.2
Deep vein thrombosis
2
1.6
Pulmonary embolism
3
2.4
Results
Recurrent stroke/transient ischaemic attack
4
3.2
Participant characteristics
Chronic rheumatic heart disease Other
6 1
4.8 0.8
Target INR range
One hundred and fifty-two patients consented to enrol in the study, and 125 of these were included in the final analysis. The INR values had not been recorded for two (1.3 %) participants, five (3.3 %) participants transferred to a different clinic before the end of their first year of followup, four (2.6 %) stopped taking warfarin, three died of unrelated or unknown causes (2.0 %) and thirteen (8.6 %) participants were lost to follow-up.
123
2–3
88
70.4
2.50–3.50
37
29.6
Median
IQR
Years since started warfarin
7
4–14
Warfarin dose
3
2.25–3.5
Measures of warfarin use
Bleeding events and associated factors in a cohort of adult
There were no significant differences in age, gender, enrolment INR, or INR target group between participants who were excluded or included in the study, as shown in Table 1. Participants contributed to a total of 140 personyears of data. The median period of follow-up was 1 year and 26 days (IQR 1 year 13 days to 1 year 56 days) and the median number of follow-up visits was 9 (IQR 6–11). Table 2 describes participants. In summary, the mean age of participants was 58.4 years (SD 13.3), and 51.2 % were male. The majority described themselves as Malay (46.4 %), 37.6 % as Chinese and 15.2 % as from indigenous tribal groups. The most frequent indications for warfarin (non-mutually exclusive) were AF (63.2 %), valve replacement (48.0 %) and heart valve lesions (14.4 %, largely mitral regurgitation and stenosis). Most participants (70.4 %) had a target INR range of 2.0–3.0, and the remainder had a target range of 2.5–3.5 based on
a
the high risk indications for warfarin outlined in the AATPT guidelines [4]. The median time participants had been taking warfarin before enrolment was 7 years (IQR 4–14), and the median dose was 3 mg (IQR 2.25–3.5). Population INR status In total 1,232 INR measurements were recorded, with a median of 9 recordings per patient (IQR 7–12). The median INR for the dataset was 2.39 (IQR 2.01–2.91), while the median for the group with INR target range of 2.0–3.0 was 2.33 (IQR 1.95–2.88) and the median for the group with a target range 2.5–3.5 was 2.52 (IQR 2.12–2.96). The median of the individual’s TTR calculated using the Rosendaal method was 61.6 %, (IQR 44.6–74.1 %). Using the classical method it was 50.0 % (IQR 40.0–66.7 %). Using the Rosendaal method the median TTR for the group with an INR target range of 2.0–3.0 was 66.6 % (IQR
140 120
Frequency
100 80 60 40 20
0.8-1.0 1.0-1.2 1.2-1.4 1.4-1.6 1.6-1.8 1.8-2.0 2.0-2.2 2.2-2.4 2.4-2.6 2.6-2.8 2.8-3.0 3.0-3.2 3.2-3.4 3.4-3.6 3.6-3.8 3.8-4.0 4.0-4.2 4.2-4.4 4.4-4.6 4.6-4.8 4.8-5.0 5.0-5.2 5.2-5.4 5.4-5.6 5.6-5.8 5.8-6.0 6.0-6.2 6.2-6.4 6.4-6.6 6.6-6.8 6.8-7.0 7.0-7.2 7.2-7.4
0
INR value
b
50 45 40 35 30 25 20 15 10 5 0 0.8-1.0 1.0-1.2 1.2-1.4 1.4-1.6 1.6-1.8 1.8-2.0 2.0-2.2 2.2-2.4 2.4-2.6 2.6-2.8 2.8-3.0 3.0-3.2 3.2-3.4 3.4-3.6 3.6-3.8 3.8-4.0 4.0-4.2 4.2-4.4 4.4-4.6 4.6-4.8 4.8-5.0 5.0-5.2 5.2-5.4 5.4-5.6 5.6-5.8 5.8-6.0 6.0-6.2 6.2-6.4 6.4-6.6 6.6-6.8 6.8-7.0 7.0-7.2 7.2-7.4
Frequency
Fig. 1 a Distribution of INR values (target group 2.0–3.0). Dotted lines indicate readings within target range. b Distribution of INR values (target group 2.5–3.5). Dotted lines indicate readings within target range
229
INR value
123
230
F. Edwards et al.
Table 3 Types and frequencies of the major and minor bleeds and the thromboembolic event Number of people who had an event
Rate per 100 personyears of follow-up
Skin bruising
14
10.0
Epistaxis Haemoptysis
9 6
6.43 4.29
Bleeding gums
6
4.29
Skin haematoma
5
3.57
Haematuria
4
2.86
Menorrhagia
2
1.43
Blood in stool
2
1.43
Minor bleeds
Total
48
34.3
Major bleeds Menorrhagia needing transfusion
1
0.714
Intracranial haemorrhage
1
0.714
Total
2
1.43
Pulmonary embolism
1
0.714
Total Total events
1 50
0.714 35.7
Thromboembolic event
47.9–80.4 %) and for the group with a target range of 2.5–3.5 was 46.6 % (IQR 36.2–62.7 %). The difference in TTR between the groups was statistically significant (Mann–Whitney U test p = 0.001). Of the out-of-range readings 30.0 % (95 % CI 27.4–32.5) were below the target range and 15.4 % (95 % CI 13.4–17.4) were above. Histograms showing INR readings for the groups of patients with target ranges of 2.0–3.0 and 2.5–3.5 are shown in Fig. 1a, b respectively. Adverse events Only two major bleeds were recorded during the study period, equating to a rate of 1.43 per 100 person-years of anticoagulation (95 % CI 0.537–3.39). These were an intracranial haemorrhage, confirmed on CT brain and resulting in death, and menorrhagia requiring a transfusion of two units of blood. Additionally, five participants were admitted to Sarawak General Hospital during the study period for haemorrhagic events which resulted in transfusion of fresh frozen plasma (FFP); although these did not meet the criteria for major bleeds. These events were a nosebleed, a gastric bleed, a skin haematoma and two incidents of haematuria. If these had been included as major bleeds, the overall rate would be 5 per 100 person-years (95 % CI 3.15–6.84).
123
There was one thromboembolic event, a pulmonary embolus, giving an incidence of 0.75 per 100 person-years of anticoagulation (95 % CI 0.681–2.11). Forty-eight minor bleeds were reported by 30 participants (23.2 %) representing 34.3 per 100 person-years of anticoagulation (95 % CI 26.9–42.5) (Table 3). Seventeen individuals had one minor bleed, nine experienced two bleeds and three and one suffered three and four bleeds respectively. Risk factors for bleeding outcomes In all variables except mean INR, the INR control was worse in the group who had a bleed, compared to the group who did not (Table 4). However none of these differences, except SD of the mean INR were significant at the 5 % level. The univariate analysis identified six variables to be included in the adjusted analysis, with p \ 0.10 for predicting any bleed. These were: female gender [odds ratio (OR) 3.45, 95 % CI 1.43–8.29], increasing number of years since starting warfarin (OR 1.06, 96 % CI 0.994–1.13), being in the lower INR target range group (OR 2.68, 95 % CI 0.914–7.65), having more than 1 reading out of target range (OR 2.44, 95 % CI 0.852–6.97), having a greater range of INR values (OR 1.26, 95 % CI 0.983–1.62) and an increased SD of mean INR over the year (a measure of within-person variation in INR) (OR 3.049, 95 % CI 1.11–8.38). Table 4 shows the OR and 95 % CI for all the variables that were analysed for an association with bleeding events. In the multivariate logistic regression model, SD of mean INR (OR 3.21, 95 % CI 1.10–9.38) and female gender (OR 3.52, 95 % CI 1.43–8.64) were significant predictors of a bleed at the 5 % level (Nagelkerke R2 = 0.145).
Discussion Relationship of adverse events to anticoagulation intensity This prospective observational cohort study recorded the rates of bleeding and thromboembolic events in a typical, Malaysian anticoagulation clinic serving a diverse and representative population. The rate of major bleeds was 1.43 per 100 person-years (95 % CI 0.537–3.39). Although it is difficult to compare bleeding rates between populations which differ in terms of age or indications for warfarin, in 2012 a systematic review and meta-analysis of randomised controlled trials from Europe and North America, estimated the annual rate of major bleeds to be 4.7 % (95 % CI 3.8–6.0) [25]. Furthermore, as only randomised controlled trials were included, exclusion of high-
Bleeding events and associated factors in a cohort of adult
231
Table 4 Univariate regression analysis showing predictors of haemorrhagic outcome Variable
Univariate logistic regression analysis No bleed
Bleed
OR
95 % CI
Multivariate logistic regression analysis p values
OR
95 % CI
p value
Measures of INR control % TTR Rosendaal method (mean)
62.0
56.9
0.347
0.053–2.26
0.269
% TTR traditional method (mean) C1 Reading above target (%)a
54.1 68.1
48.9 83.9
0.271 2.44
0.034–2.13 0.852–6.97
0.215 0.097
% INR readings above target (mean)
13.1
17.4
9.78
0.533–179
0.125
Mean INR (mean)
2.20
2.14
0.750
0.290–1.94
0.552
INR range (mean)a
2.32
2.93
1.26
0.983–1.62
0.068
SD of mean INR over 1 year (mean)a
0.679
0.858
3.05
1.11–8.38
0.031
3.21
1.10–9.38
0.03
3.51
1.43–8.64
0.006
Demographic variables Gender (% female)a
41.5
71.0
3.45
1.43–8.29
0.006
Age (mean)
57.6
59.7
1.01
0.982–1.04
0.433
Malay
46.8
45.2
1.193
0.340–4.19
0.783
Chinese
36.2
41.9
1.202
0.500–2.89
0.682
1.1
0
0
–
–
16.0
12.9
0.838
0.239–2.94
0.783
10.6 2.94
1.06 0.986
0.994–1.13 0.687–1.42
0.075 0.938
16.1
2.68
0.941–7.65
0.065
Ethnicity (%)
Indian Indigenous tribe
0.949
Other possible predictors Years on warfarin (mean)a Warfarin dose (mean) INR target range (% of patients in the higher target range)a
8.20 2.96 34.0
INR international normalised ratio, TTR time at target range, SD standard deviation a
The variable is a significant predictor of a haemorrhagic outcome at the 10 % level
risk patients and better INR control may have led to underestimation of true rates. A non-systematic review of observational studies undertaken in America calculated a major bleeding rate of 3.6 per 100 person-years [26]. There are various explanations for the low rate of major bleeding reported in this cohort. Haemorrhage resulted in admission to SGH in five cases despite not meeting the study criteria for a major bleed [20]. Some may advocate such cases being defined as a major bleed, in which case the rate would be 5 per 100 person-years (95 % CI 3.15–6.84). Furthermore, although there were no significant differences in age, gender, enrolment INR or INR target group between participants who were lost to followup and those included in the study, 13 participants (8.6 %) were lost to follow-up. Although all attempts were made to identify bleeding events or deaths in these participants, underestimation of outcomes cannot be excluded. In this study, the rate of non-major bleeding was 33.6 per 100 person-years of anticoagulation (95 % CI 25.7–41.4). Few Asian studies have published minor bleeding rates, as definitions of minor bleeding can vary and comparisons are difficult to make. Studies from Europe and America have demonstrated lower rates than reported in this study,
between 6.2 and 19 per 100 person-years [6, 27–29]. High rates of these events are important for the patient and can have a significant impact on quality of life [30]. The rate of thromboembolic events was only 0.75 per 100 person-years of anticoagulation (95 % CI 0.681–2.11). A systematic review of 16 studies from Europe, America and Asia reported the overall rate of thromboembolism at 2.6 per 100 person-years (95 % CI 1.8–3.6) when the INR was between 2.0 and 3.0 and they reported a rise to 9.0 per 100 person-years (95 % CI 6.1–13.4) when the INR was \2.0 [31]. Since 35.5 % of INR readings in this population were below 2.0, the low rate of thromboembolic events we have observed here is unexpected. The reported rates could be at least partly attributable to loss to follow-up, leading to an underestimation of adverse events but there could be another explanation. The degree to which this population was anticoagulated is an important factor which may have influenced the rates of adverse events. INR was less well controlled than reported in other settings [27, 32–35]. Only 48.6 % (95 % CI 46.1–51.1) of results were in range, and the majority of the out of range readings were below the target (Fig. 1a, b). Significantly fewer readings were within range for the
123
232
group with a higher INR target. Adequate anticoagulation was not achieved in either group, but in particular the patients in the group with the higher target were not anticoagulated sufficiently to combat the extra risk of thromboembolism they have. Qualitative investigation into the reasons behind why many patients are being treated subtherapeutically was beyond the scope of this research. However, on discussion with the clinicians in the cardiology department at SGH, it was apparent that INR values below 2.5 were felt to be beneficial in this region. Differences in risk profiles of both thromboembolism and bleeding events between Asians and Caucasians could explain many of the findings outlined above. There is evidence to show that rates of bleeding on warfarin are higher in Asian patients, in particular warfarin-related intracranial haemorrhage [36]. The difference in risk of thromboembolic disease has also been shown to be higher in Asian populations than the global levels in three large trials [37–39]. This study reports rates of both major bleed and thromboembolism that are lower than expected. However the INR values indicate that the population has a lower intensity of anticoagulation than is recommended by Western guidelines and this could be a contributing factor to explain the low rates of adverse events in this population. Other studies have shown the benefit of lower INR target ranges in certain populations. A study in Japan concluded that a target range of 1.6–2.6 resulted in the least adverse events, including both bleeding and thromboembolism, in elderly AF patients [40] and a study in China showed that rates of thromboembolic events were lowest in the INR range 1.5–1.9, showing that even low levels of warfarin can have a beneficial effect in preventing thromboembolism.
F. Edwards et al.
1.03–9.82), suggesting that participants with a highly variable INR are at highest risk of adverse events. Serial INR measurements have been shown to be poor predictors of bleeds, since the rise in INR usually occurs too close to the bleeding event to be identified [41]. Using the variability of the INR could therefore be of clinical relevance in predicting patients who are at higher risk of bleeding, and in whom more effort needs to be made to control their INR. Amongst variables not directly measuring INR control, female gender was the only significant predictor of a bleeding outcome, with an OR of 3.51. The reasons for this are unclear, although it is probably due to social rather than biological differences between the genders. It could be due to differences in reporting of bleeds; perhaps females are more likely to report bruising or heavy vaginal bleeding than their male counterparts. Further research into this subject would be interesting. Limitations These results must be interpreted with care, due to a number of limitations of the study. The sample size was smaller than hoped and the loss to follow up of 8.6 % is noteworthy. Additionally, any conclusions about rates of thromboembolism are difficult to make, due to only one event occurring in the study. The multiple regression model was not particularly convincing, since it explained only 15 % of the variability when predicting bleeds. Although the study tried to reduce reporting bias and eliminate both patient and staff differences in perceptions of what constitutes a minor bleed by asking questions from a proforma, this will never be completely successful, as the mentioning of some bleeds, such as bruising, will always depend on the individual in question as to what is normal for them.
INR control as a risk factor for bleeding outcome Conclusion The importance of tight INR control in preventing haemorrhagic and thromboembolic complications is established. However, before now, measures of INR control which may confer an increased risk of bleeding have not been explored in an Asian population. Unfortunately, rates of major bleeding were too low to use this as a primary outcome, and therefore minor bleeds and major bleeds were combined and analysis was carried using the rates of any bleeding event. Of the seven variables relating to INR control which were investigated, three were significant at the 10 % level in univariate analysis, but the SD of the INR was the only one which was a significant predictor of a bleed outcome at the 5 % level. As the SD of mean INR recorded over the year increased by a unit, the odds of experiencing a bleed increased by just over three times, (OR 3.18, 95 % CI
123
Low rates of major bleeds and thromboembolic events were recorded in this observational cohort study, in addition to a substantial number of INR readings under the recommended target range. Although the study was small, and there was some loss to follow-up, these results suggest that the recommended INR ranges may not represent the optimal warfarin intensity for this population. A lower intensity of therapy, which was in fact observed in this cohort, could be beneficial in terms of preventing adverse events. Further studies, especially multi centre trials in Asia have previously been recommended [42], and will be useful in investigating lower target ranges in Asian populations. The importance of INR variability as an indicator of bleeding risk, rather than one-off measurements, is also advocated.
Bleeding events and associated factors in a cohort of adult Acknowledgments This study is registered under the Malaysian National Medical Research Register (NMRR-10-714-6755). We acknowledge the support from staff of the Department of Cardiology, Sarawak General Hospital, who are involved in this study. We are grateful for permission by the Director General of Health, Malaysia, for the publication of this manuscript.
References 1. Hirsh J (1991) Oral anticoagulant drugs. N Engl J Med 324:1865–1875 2. Levine MN, Raskob G, Landefeld S, Kearon C (2001) Hemorrhagic complications of anticoagulant treatment. Chest 119:108S– 121S 3. Schulman S, Beyth RJ, Kearon C, Levine MN (2008) Hemorrhagic complications of anticoagulant and thrombolytic treatment: American College of Chest Physicians Evidence-Based Clinical Practice Guidelines. Chest 133:257S–298S 4. Ageno W, Gallus AS, Wittkowsky A, Crowther M, Hylek EM, Palareti G (2012) Oral anticoagulant therapy: antithrombotic Therapy and Prevention of Thrombosis, 9th ed: American College of Chest Physicians Evidence-Based Clinical Practice Guidelines. Chest 141:e44S–e88S 5. Keeling D, Baglin T, Tait C, Watson H, Perry D, Baglin C et al (2011) Guidelines on oral anticoagulation with warfarin—fourth edition. Br J Haematol 154:311–324 6. Connolly SJ, Laupacis A, Gent M, Roberts RS, Cairns JA, Joyner C (1991) Canadian atrial fibrillation anticoagulation (CAFA) Study. J Am Coll Cardiol 18:349–355 7. Kearon C, Ginsberg JS, Kovacs MJ, Anderson DR, Wells P, Julian JA et al (2003) Comparison of low-intensity warfarin therapy with conventional-intensity warfarin therapy for longterm prevention of recurrent venous thromboembolism. N Engl J Med 349:631–639 8. Petersen P, Boysen G, Godtfredsen J, Andersen ED, Andersen B (1989) Placebo-controlled, randomised trial of warfarin and aspirin for prevention of thromboembolic complications in chronic atrial fibrillation. The Copenhagen AFASAK study. Lancet 1:175–179 9. Ansell J, Hirsh J, Poller L, Bussey H, Jacobson A, Hylek E (2004) The pharmacology and management of the vitamin K antagonists: the Seventh ACCP Conference on Antithrombotic and Thrombolytic Therapy. Chest 126:204S–233S 10. Dang MT, Hambleton J, Kayser SR (2005) The influence of ethnicity on warfarin dosage requirement. Ann Pharmacother 39:1008–1012 11. Johnson JA (2008) Ethnic differences in cardiovascular drug response: potential contribution of pharmacogenetics. Circulation 118:1383–1393 12. United States Food and Drug Administration. Warfarin Product Labelling. 4-1-2006. Ref Type: Online Source 13. Suzuki S, Yamashita T, Kato T, Fujino T, Sagara K, Sawada H et al (2007) Incidence of major bleeding complication of warfarin therapy in Japanese patients with atrial fibrillation. Circ J 71:761–765 14. You JH, Chan FW, Wong RS, Cheng G (2005) Is INR between 2.0 and 3.0 the optimal level for Chinese patients on warfarin therapy for moderate-intensity anticoagulation? Br J Clin Pharmacol 59:582–587 15. Yu HY, Liu CH, Chen YS, Wang SS, Chu SH, Lin FY (2005) Relationship of international normalized ratio to bleeding and thromboembolism rates in Taiwanese patients receiving vitamin K antagonist after mechanical valve replacement. J Formos Med Assoc 104:236–243
233 16. Scottish Intercollegiate Guidelines Network. Antithrombotic Therapy. 8-5-2012. Ref Type: Online Source 17. Khor GL (2001) Cardiovascular epidemiology in the Asia-Pacific region. Asia Pac J Clin Nutr 10:76–80 18. Rosendaal FR, Cannegieter SC, van der Meer FJ, Brie¨t E (1993) A method to determine the optimal intensity of oral anticoagulant therapy. J Thromb Haemost 69:236–239 19. INR Pro. INR Pro. 2013. 28-7-2013. Ref Type: Online Source 20. Schulman S, Kearon C (2005) Definition of major bleeding in clinical investigations of antihemostatic medicinal products in non-surgical patients. J Thromb Haemost 3:692–694 21. Witt DM, Delate T, Clark NP, Martell C, Tran T, Crowther MA et al (2009) Outcomes and predictors of very stable INR control during chronic anticoagulation therapy. Blood 114:952–956 22. Lip GY, Frison L, Halperin JL, Lane DA (2011) Comparative validation of a novel risk score for predicting bleeding risk in anticoagulated patients with atrial fibrillation: the HAS-BLED (Hypertension, Abnormal Renal/Liver Function, Stroke, Bleeding History or Predisposition, Labile INR, Elderly, Drugs/Alcohol Concomitantly) score. J Am Coll Cardiol 57:173–180 23. Rose AJ, Berlowitz DR, Miller DR, Hylek EM, Ozonoff A, Zhao S et al (2012) INR targets and site-level anticoagulation control: results from the Veterans AffaiRs Study to Improve Anticoagulation (VARIA). J Thromb Haemost 10:590–595 24. Dauphin C, Legault B, Jaffeux P, Motreff P, Azarnoush K, Joly H et al (2008) Comparison of INR stability between self-monitoring and standard laboratory method: preliminary results of a prospective study in 67 mechanical heart valve patients. Arch Cardiovasc Dis 101:753–761 25. Warkentin AE, Donadini MP, Spencer FA, Lim W, Crowther M (2012) Bleeding risk in randomized controlled trials comparing warfarin and aspirin: a systematic review and meta-analysis. J Thromb Haemost 10:512–520 26. Landefeld CS, Beyth RJ (1993) Anticoagulant-related bleeding: clinical epidemiology, prediction, and prevention. Am J Med 95:315–328 27. Palareti G, Leali N, Coccheri S, Poggi M, Manotti C, D’Angelo A et al (1996) Bleeding complications of oral anticoagulant treatment: an inception-cohort, prospective collaborative study (ISCOAT). Italian Study on Complications of Oral Anticoagulant Therapy. Lancet 348:423–428 28. Evans A, Kalra L (2001) Are the results of randomized controlled trials on anticoagulation in patients with atrial fibrillation generalizable to clinical practice? Arch Intern Med 161:1443–1447 29. Gottlieb LK, Salem-Schatz S (1994) Anticoagulation in atrial fibrillation. Does efficacy in clinical trials translate into effectiveness in practice? Arch Intern Med 154:1945–1953 30. Stewart RA (2009) Underestimation of the effect of bleeding in clinical trials. Lancet 374:1145–1146 31. Oake N, Jennings A, Forster AJ, Fergusson D, Doucette S (2008) Anticoagulation intensity and outcomes among patients prescribed oral anticoagulant therapy: a systematic review and metaanalysis. CMAJ 179:235–244 32. Kim JH, Song YB, Shin DH, Kim JS, Choi JO, On YK et al (2009) How well does the target INR level maintain in warfarintreated patients with non-valvular atrial fibrillation? Yonsei Med J 50:83–88 33. Jones M, McEwan P, Morgan CL, Peters JR, Goodfellow J, Currie CJ (2005) Evaluation of the pattern of treatment, level of anticoagulation control, and outcome of treatment with warfarin in patients with non-valvar atrial fibrillation: a record linkage study in a large British population. Heart 91:472–477 34. Reynolds MW, Fahrbach K, Hauch O, Wygant G, Estok R, Cella C et al (2004) Warfarin anticoagulation and outcomes in patients with atrial fibrillation: a systematic review and metaanalysis. Chest 126:1938–1945
123
234 35. Okumura K, Komatsu T, Yamashita T, Okuyama Y, Harada M, Konta Y et al (2011) Time in the therapeutic range during warfarin therapy in Japanese patients with non-valvular atrial fibrillation. A multicenter study of its status and infuential factors. Circ J 75:2087–2094 36. Shen AY, Yao JF, Brar SS, Jorgensen MB, Chen W (2007) Racial/ethnic differences in the risk of intracranial hemorrhage among patients with atrial fibrillation. J Am Coll Cardiol 50:309–315 37. Connolly SJ, Ezekowitz MD, Yusuf S, Eikelboom J, Oldgren J, Parekh A et al (2009) Dabigatran versus warfarin in patients with atrial fibrillation. N Engl J Med 361:1139–1151 38. Patel MR, Mahaffey KW, Garg J, Pan G, Singer DE, Hacke W et al (2011) Rivaroxaban versus warfarin in nonvalvular atrial fibrillation. N Engl J Med 365:883–891
123
F. Edwards et al. 39. Granger CB, Alexander JH, McMurray JJ, Lopes RD, Hylek EM, Hanna M et al (2011) Apixaban versus warfarin in patients with atrial fibrillation. N Engl J Med 365:981–992 40. Yasaka M, Minematsu K, Yamaguchi T (2001) Optimal intensity of international normalized ratio in warfarin therapy for secondary prevention of stroke in patients with non-valvular atrial fibrillation. Intern Med 40:1183–1188 41. Kucher N, Connolly S, Beckman JA, Cheng LH, Tsilimingras KV, Fanikos J et al (2004) International normalized ratio increase before warfarin-associated hemorrhage: brief and subtle. Arch Intern Med 164:2176–2179 42. Kwong JS, Yu CM (2013) The need for multicentre cardiovascular clinical trials in Asia. Nat Rev Cardiol 10:355–362
Bleeding events and associated factors in a cohort of adult patients taking warfarin in Sarawak, Malaysia Frances Edwards • Paul Arkell • Alan Yean Yip Fong • Lesley M. Roberts • David Gendy • Christina Siew-Hie Wong • Joanna Chee Yien Ngu • Lee Len Tiong • Faridha Mohd Salleh Bibi Lana Yin Hui Lai • Tiong Kiam Ong • Michael Abouyannis
•
Published online: 15 November 2013 Ó Springer Science+Business Media New York 2013
Abstract Evidence is emerging that rates of adverse events in patients taking warfarin may vary with ethnicity. This study investigated the rates of bleeds and thromboembolic events, the international normalised ratio (INR) status and the relationship between INR and bleeding events in Malaysia. Patients attending INR clinic at the Heart Centre, Sarawak General Hospital were enrolled on an ad hoc basis from May 2010 and followed up for 1 year. At each routine visit, INR was recorded and screening for bleeding or thromboembolism occurred. Variables relating to INR control were used as predictors of bleeds in logistic regression models. 125 patients contributed to 140 personyears of follow-up. The rates of major bleed, thromboembolic event and minor bleed per 100 person-years of follow-up were 1.4, 0.75 and 34.3. The median time at target range calculated using the Rosendaal method was 61.6 % (IQR 44.6–74.1 %). Of the out-of-range readings, 30.0 %
F. Edwards D. Gendy M. Abouyannis c/o Dr. L Roberts, Primary Care Clinical Sciences, School of Health and Population Sciences, College of Medical and Dental Sciences, University of Birmingham, Edgbaston, Birmingham B15 2TT, UK P. Arkell (&) c/o Diane Jackson, West Midlands Deanery, Clinical Education Centre, University Hospital North Staffs, Stoke-on-Trent ST4 6QG, UK e-mail: [email protected] A. Y. Y. Fong C. S.-H. Wong J. C. Y. Ngu L. L. Tiong F. M. S. Bibi L. Y. H. Lai T. K. Ong Sarawak Heart Centre, Kuching, Sarawak, Malaysia L. M. Roberts Primary Care Clinical Sciences, School of Health and Population Sciences, College of Medical and Dental Sciences, University of Birmingham, Edgbaston, Birmingham B15 2TT, UK
123
were below range and 15.4 % were above. INR variability, (standard deviation of individuals’ mean INR), was the best predictor of bleeding events, with an odds ratio of 3.21 (95 % CI 1.10–9.38). Low rates of both major bleeds and thromboembolic events were recorded, in addition to a substantial number of INR readings under the recommended target range. This may suggest that the recommended INR ranges may not represent the optimal warfarin intensity for this population and that a lower intensity of therapy, as observed in this cohort, could be beneficial in preventing adverse events. Keywords Anticoagulation Warfarin Haemorrhage Malaysia
Introduction Warfarin is the most commonly prescribed anticoagulant worldwide for individuals at risk of thromboembolic events [1]. Unfortunately the therapeutic index is narrow, with suboptimal treatment risking thromboembolism, and overtreatment threatening haemorrhage [2, 3]. According to US and UK guidelines, an international normalised ratio (INR) range of 2.0–3.0 is recommended, although higher targets are needed in populations at increased risk of thromboembolism, such as those with mechanical heart valve replacement [4, 5]. This target range has been applied worldwide, although data was predominantly collected from the European and North American regions [6–8]. It has been established that ethnicity influences warfarin sensitivity [9–11], and the US Food and Drug Administration (FDA) states that Asian populations need lower doses of warfarin than Caucasian populations to achieve the target INR [12]. A study of low dose Warfarin based in Japan,
Bleeding events and associated factors in a cohort of adult
found that major haemorrhage was more frequent than in Western populations [13]. Cohort data from China suggested a target INR range of 1.8–2.4 to be associated with lowest rates of major bleeding or thromboembolic events [14]. Additionally, one study in Taiwan reported that an INR of \2.0 was not associated with increased thromboembolic events [15], as it has been in Caucasians [16]. Although data regarding use of anticoagulation in Malaysia is scarce, incidence of cardiovascular disease is increasing, and this is likely to be associated with increased use of warfarin therapy [17]. A recent cross sectional study estimated prevalence of asymptomatic atrial fibrillation (AF) at 0.75 % in the Malaysian primary care setting [18]. While modern alternatives to warfarin, such as dabigatran and rivaroxaban,are increasingly being used in Europe and America [19], their relatively higher cost limit their widespread use in developing countries, such as Malaysia. This study is the first to describe rates of adverse events in a Malaysian population taking Warfarin. The Malaysian population is diverse and composed of people of Malay, Chinese, Indian and Indigenous tribal origin, and data here will add to Asia’s evidence base for oral anticoagulation use. Aims 1.
2. 3.
To describe the rates of both haemorrhagic and thromboembolic adverse events in a Malaysian population taking warfarin To describe the intensity of warfarin therapy in the sample population To investigate predictors of adverse events, relating both to INR status and to the risk factors which have been previously established in Western populations
Methods Study setting Participants were drawn from attendants of the cardiology anticoagulation clinic at the Heart Centre of Sarawak General Hospital (SGH) in Kuching, Malaysia. This is the main tertiary referral centre for the state of Sarawak and the only public hospital serving the city of Kuching. The American Antithrombotic Therapy and Prevention of Thrombosis (AATPT) guidelines are followed for target ranges, dose initiation and dose titration [4]. Patients with INR values within their target range were managed by pharmacists, while those outside target range were seen by a doctor. Study subjects, procedures and measurements Recruitment was conducted by a dedicated research pharmacist on an ad hoc basis between May 2010 and
227
December 2011. Data were collected prospectively up to and including the visit after the participant had been in the study for a year. Patients were eligible to be enrolled if they were currently taking warfarin, were over 18 years and spoke one of three main languages (Bahasa Malaysia, Mandarin or English). At the enrolment visit the following data were collected: demographics (age, gender, ethnicity, and occupation), history of warfarin use (date started, dose, and indication), bleeding history, past medical history, alcohol use, weight and laboratory measurements (full blood count, liver function tests, urea and electrolytes, and INR). Participants returned according to the timeframe of their routine care, and on these subsequent visits the INR and current warfarin dose was recorded. At every visit the clinician followed a proforma to identify bleeding events. All participants were asked if they had experienced the following: bloody or black stool, blood in the urine, coughing blood, bleeding gums, skin haematoma, bruising skin, nose bleed, or heavy or irregular vaginal bleeding. Participants were asked of any hospital admissions at every visit, and all such cases underwent a review of the medical records by the research team to identify haemorrhagic or thrombotic events. Participants who stopped attending the clinic were followed up by telephone by the research team to identify adverse events or deaths. Events were then checked with the hospital records. Statistical analysis Only participants completing at least 1 year of follow-up were included, and person-years was the unit of analysis. Participants were censored from the point of time following the occurrence of a major bleed or thrombotic event. Participant demographics and clinical characteristics were described using proportions, counts, means with standard deviations (SD), and medians with interquartile ranges (IQR) as appropriate. The overall percentage time at target range (TTR) was calculated using the method described by Rosendaal et al. [18], which uses linear interpolation to estimate the time spent at each INR value. INRpro provided the software to calculate this [19]. TTR was also calculated using the traditional method as described in the Chest guidelines: number of INR measurements within target range divided by total number of INR measurements [4]. Bleeds were categorised as either major or minor according to the definition of the International Society on Thrombosis and Haemostasis [20]. As such a major bleed was defined as either a fatal bleed, a symptomatic bleed in a critical area (intracranial, intraspinal, intraocular, retroperitoneal, intraarticular, pericardial or intramuscular with compartment syndrome) or a bleed causing a fall in
123
228
F. Edwards et al.
haemoglobin of 2 g/dL or leading to transfusion of two or more units of whole blood or red cells. Rates of major bleeds, minor bleeds and thromboembolic events per person-year on warfarin were calculated with 95 % confidence intervals (CI). Univariate logistic regression was undertaken using the outcome ‘any bleed’ to include major or minor bleeds. The following variables were entered to identify predictors of bleeding: demographics (age, gender and ethnicity), characteristics of warfarin use (number of years on warfarin, warfarin dose and INR target range) and seven variables relating to INR control and variability. These variables were selected because they have been used in other studies, are clinically relevant and show biological plausibility, and they were as follows: [4, 21–24]
Table 1 Comparison of included participants with those lost to follow up Included participants
(5) (6) (7)
TTR (Rosendaal method) TTR (Traditional method) More than or equal to 1 INR reading above target Percentage of INR readings above target range (% above range) Mean of all INR values during the study period for each participant (mean INR) The INR range for each participant (INR range) The SD of the mean INR for each participant (SD of INR)
Variables reaching p \ 0.10 level of significance were entered into multivariate logistic regression using the forward stepwise model analysis to identify independent predictors. Data were analysed using SPSS (v19) for Windows.
58.4
54.8
0.144
Gender (% male)
51.2
50.0
0.899
Target INR range (% range 2.5–3.5)
29.6
27.8
0.832
Enrolment INR (mean)
2.56
2.31
0.153
Table 2 Demographic variables and warfarin indication Mean
SD
58.4
13.3 Number
%
Male
64
51.2
Female
61
48.8
Malay
58
46.4
Chinese
47
37.6
Indigenous tribes
19
15.2
1
0.8
Operator/fabricator/labourer Service operations
19 8
15.2 6.4
Technical/sales/admin support
11
Gender
Ethnicity
Indian Occupation
Managerial/professional
Ethics
p value
Age (mean)
Age
(1) (2) (3) (4)
Participants lost to follow up
None/retired
8.8
5
4
82
65.6
Indication for warfarin
Ethical approval was obtained from the Ministry of Health Medical Research and Ethics Committee, Malaysia and the Population Sciences and Humanities BMedSc Internal Ethics Committee, University of Birmingham, UK. Informed written consent was gained from every participant before enrolment.
Replaced valve Of which C1 mechanical Valve lesion Congestive heart failure
60
48.0
51
40.8
18
14.4
3
2.4
Atrial fibrillation
79
63.2
Cardiac thrombus
9
7.2
Deep vein thrombosis
2
1.6
Pulmonary embolism
3
2.4
Results
Recurrent stroke/transient ischaemic attack
4
3.2
Participant characteristics
Chronic rheumatic heart disease Other
6 1
4.8 0.8
Target INR range
One hundred and fifty-two patients consented to enrol in the study, and 125 of these were included in the final analysis. The INR values had not been recorded for two (1.3 %) participants, five (3.3 %) participants transferred to a different clinic before the end of their first year of followup, four (2.6 %) stopped taking warfarin, three died of unrelated or unknown causes (2.0 %) and thirteen (8.6 %) participants were lost to follow-up.
123
2–3
88
70.4
2.50–3.50
37
29.6
Median
IQR
Years since started warfarin
7
4–14
Warfarin dose
3
2.25–3.5
Measures of warfarin use
Bleeding events and associated factors in a cohort of adult
There were no significant differences in age, gender, enrolment INR, or INR target group between participants who were excluded or included in the study, as shown in Table 1. Participants contributed to a total of 140 personyears of data. The median period of follow-up was 1 year and 26 days (IQR 1 year 13 days to 1 year 56 days) and the median number of follow-up visits was 9 (IQR 6–11). Table 2 describes participants. In summary, the mean age of participants was 58.4 years (SD 13.3), and 51.2 % were male. The majority described themselves as Malay (46.4 %), 37.6 % as Chinese and 15.2 % as from indigenous tribal groups. The most frequent indications for warfarin (non-mutually exclusive) were AF (63.2 %), valve replacement (48.0 %) and heart valve lesions (14.4 %, largely mitral regurgitation and stenosis). Most participants (70.4 %) had a target INR range of 2.0–3.0, and the remainder had a target range of 2.5–3.5 based on
a
the high risk indications for warfarin outlined in the AATPT guidelines [4]. The median time participants had been taking warfarin before enrolment was 7 years (IQR 4–14), and the median dose was 3 mg (IQR 2.25–3.5). Population INR status In total 1,232 INR measurements were recorded, with a median of 9 recordings per patient (IQR 7–12). The median INR for the dataset was 2.39 (IQR 2.01–2.91), while the median for the group with INR target range of 2.0–3.0 was 2.33 (IQR 1.95–2.88) and the median for the group with a target range 2.5–3.5 was 2.52 (IQR 2.12–2.96). The median of the individual’s TTR calculated using the Rosendaal method was 61.6 %, (IQR 44.6–74.1 %). Using the classical method it was 50.0 % (IQR 40.0–66.7 %). Using the Rosendaal method the median TTR for the group with an INR target range of 2.0–3.0 was 66.6 % (IQR
140 120
Frequency
100 80 60 40 20
0.8-1.0 1.0-1.2 1.2-1.4 1.4-1.6 1.6-1.8 1.8-2.0 2.0-2.2 2.2-2.4 2.4-2.6 2.6-2.8 2.8-3.0 3.0-3.2 3.2-3.4 3.4-3.6 3.6-3.8 3.8-4.0 4.0-4.2 4.2-4.4 4.4-4.6 4.6-4.8 4.8-5.0 5.0-5.2 5.2-5.4 5.4-5.6 5.6-5.8 5.8-6.0 6.0-6.2 6.2-6.4 6.4-6.6 6.6-6.8 6.8-7.0 7.0-7.2 7.2-7.4
0
INR value
b
50 45 40 35 30 25 20 15 10 5 0 0.8-1.0 1.0-1.2 1.2-1.4 1.4-1.6 1.6-1.8 1.8-2.0 2.0-2.2 2.2-2.4 2.4-2.6 2.6-2.8 2.8-3.0 3.0-3.2 3.2-3.4 3.4-3.6 3.6-3.8 3.8-4.0 4.0-4.2 4.2-4.4 4.4-4.6 4.6-4.8 4.8-5.0 5.0-5.2 5.2-5.4 5.4-5.6 5.6-5.8 5.8-6.0 6.0-6.2 6.2-6.4 6.4-6.6 6.6-6.8 6.8-7.0 7.0-7.2 7.2-7.4
Frequency
Fig. 1 a Distribution of INR values (target group 2.0–3.0). Dotted lines indicate readings within target range. b Distribution of INR values (target group 2.5–3.5). Dotted lines indicate readings within target range
229
INR value
123
230
F. Edwards et al.
Table 3 Types and frequencies of the major and minor bleeds and the thromboembolic event Number of people who had an event
Rate per 100 personyears of follow-up
Skin bruising
14
10.0
Epistaxis Haemoptysis
9 6
6.43 4.29
Bleeding gums
6
4.29
Skin haematoma
5
3.57
Haematuria
4
2.86
Menorrhagia
2
1.43
Blood in stool
2
1.43
Minor bleeds
Total
48
34.3
Major bleeds Menorrhagia needing transfusion
1
0.714
Intracranial haemorrhage
1
0.714
Total
2
1.43
Pulmonary embolism
1
0.714
Total Total events
1 50
0.714 35.7
Thromboembolic event
47.9–80.4 %) and for the group with a target range of 2.5–3.5 was 46.6 % (IQR 36.2–62.7 %). The difference in TTR between the groups was statistically significant (Mann–Whitney U test p = 0.001). Of the out-of-range readings 30.0 % (95 % CI 27.4–32.5) were below the target range and 15.4 % (95 % CI 13.4–17.4) were above. Histograms showing INR readings for the groups of patients with target ranges of 2.0–3.0 and 2.5–3.5 are shown in Fig. 1a, b respectively. Adverse events Only two major bleeds were recorded during the study period, equating to a rate of 1.43 per 100 person-years of anticoagulation (95 % CI 0.537–3.39). These were an intracranial haemorrhage, confirmed on CT brain and resulting in death, and menorrhagia requiring a transfusion of two units of blood. Additionally, five participants were admitted to Sarawak General Hospital during the study period for haemorrhagic events which resulted in transfusion of fresh frozen plasma (FFP); although these did not meet the criteria for major bleeds. These events were a nosebleed, a gastric bleed, a skin haematoma and two incidents of haematuria. If these had been included as major bleeds, the overall rate would be 5 per 100 person-years (95 % CI 3.15–6.84).
123
There was one thromboembolic event, a pulmonary embolus, giving an incidence of 0.75 per 100 person-years of anticoagulation (95 % CI 0.681–2.11). Forty-eight minor bleeds were reported by 30 participants (23.2 %) representing 34.3 per 100 person-years of anticoagulation (95 % CI 26.9–42.5) (Table 3). Seventeen individuals had one minor bleed, nine experienced two bleeds and three and one suffered three and four bleeds respectively. Risk factors for bleeding outcomes In all variables except mean INR, the INR control was worse in the group who had a bleed, compared to the group who did not (Table 4). However none of these differences, except SD of the mean INR were significant at the 5 % level. The univariate analysis identified six variables to be included in the adjusted analysis, with p \ 0.10 for predicting any bleed. These were: female gender [odds ratio (OR) 3.45, 95 % CI 1.43–8.29], increasing number of years since starting warfarin (OR 1.06, 96 % CI 0.994–1.13), being in the lower INR target range group (OR 2.68, 95 % CI 0.914–7.65), having more than 1 reading out of target range (OR 2.44, 95 % CI 0.852–6.97), having a greater range of INR values (OR 1.26, 95 % CI 0.983–1.62) and an increased SD of mean INR over the year (a measure of within-person variation in INR) (OR 3.049, 95 % CI 1.11–8.38). Table 4 shows the OR and 95 % CI for all the variables that were analysed for an association with bleeding events. In the multivariate logistic regression model, SD of mean INR (OR 3.21, 95 % CI 1.10–9.38) and female gender (OR 3.52, 95 % CI 1.43–8.64) were significant predictors of a bleed at the 5 % level (Nagelkerke R2 = 0.145).
Discussion Relationship of adverse events to anticoagulation intensity This prospective observational cohort study recorded the rates of bleeding and thromboembolic events in a typical, Malaysian anticoagulation clinic serving a diverse and representative population. The rate of major bleeds was 1.43 per 100 person-years (95 % CI 0.537–3.39). Although it is difficult to compare bleeding rates between populations which differ in terms of age or indications for warfarin, in 2012 a systematic review and meta-analysis of randomised controlled trials from Europe and North America, estimated the annual rate of major bleeds to be 4.7 % (95 % CI 3.8–6.0) [25]. Furthermore, as only randomised controlled trials were included, exclusion of high-
Bleeding events and associated factors in a cohort of adult
231
Table 4 Univariate regression analysis showing predictors of haemorrhagic outcome Variable
Univariate logistic regression analysis No bleed
Bleed
OR
95 % CI
Multivariate logistic regression analysis p values
OR
95 % CI
p value
Measures of INR control % TTR Rosendaal method (mean)
62.0
56.9
0.347
0.053–2.26
0.269
% TTR traditional method (mean) C1 Reading above target (%)a
54.1 68.1
48.9 83.9
0.271 2.44
0.034–2.13 0.852–6.97
0.215 0.097
% INR readings above target (mean)
13.1
17.4
9.78
0.533–179
0.125
Mean INR (mean)
2.20
2.14
0.750
0.290–1.94
0.552
INR range (mean)a
2.32
2.93
1.26
0.983–1.62
0.068
SD of mean INR over 1 year (mean)a
0.679
0.858
3.05
1.11–8.38
0.031
3.21
1.10–9.38
0.03
3.51
1.43–8.64
0.006
Demographic variables Gender (% female)a
41.5
71.0
3.45
1.43–8.29
0.006
Age (mean)
57.6
59.7
1.01
0.982–1.04
0.433
Malay
46.8
45.2
1.193
0.340–4.19
0.783
Chinese
36.2
41.9
1.202
0.500–2.89
0.682
1.1
0
0
–
–
16.0
12.9
0.838
0.239–2.94
0.783
10.6 2.94
1.06 0.986
0.994–1.13 0.687–1.42
0.075 0.938
16.1
2.68
0.941–7.65
0.065
Ethnicity (%)
Indian Indigenous tribe
0.949
Other possible predictors Years on warfarin (mean)a Warfarin dose (mean) INR target range (% of patients in the higher target range)a
8.20 2.96 34.0
INR international normalised ratio, TTR time at target range, SD standard deviation a
The variable is a significant predictor of a haemorrhagic outcome at the 10 % level
risk patients and better INR control may have led to underestimation of true rates. A non-systematic review of observational studies undertaken in America calculated a major bleeding rate of 3.6 per 100 person-years [26]. There are various explanations for the low rate of major bleeding reported in this cohort. Haemorrhage resulted in admission to SGH in five cases despite not meeting the study criteria for a major bleed [20]. Some may advocate such cases being defined as a major bleed, in which case the rate would be 5 per 100 person-years (95 % CI 3.15–6.84). Furthermore, although there were no significant differences in age, gender, enrolment INR or INR target group between participants who were lost to followup and those included in the study, 13 participants (8.6 %) were lost to follow-up. Although all attempts were made to identify bleeding events or deaths in these participants, underestimation of outcomes cannot be excluded. In this study, the rate of non-major bleeding was 33.6 per 100 person-years of anticoagulation (95 % CI 25.7–41.4). Few Asian studies have published minor bleeding rates, as definitions of minor bleeding can vary and comparisons are difficult to make. Studies from Europe and America have demonstrated lower rates than reported in this study,
between 6.2 and 19 per 100 person-years [6, 27–29]. High rates of these events are important for the patient and can have a significant impact on quality of life [30]. The rate of thromboembolic events was only 0.75 per 100 person-years of anticoagulation (95 % CI 0.681–2.11). A systematic review of 16 studies from Europe, America and Asia reported the overall rate of thromboembolism at 2.6 per 100 person-years (95 % CI 1.8–3.6) when the INR was between 2.0 and 3.0 and they reported a rise to 9.0 per 100 person-years (95 % CI 6.1–13.4) when the INR was \2.0 [31]. Since 35.5 % of INR readings in this population were below 2.0, the low rate of thromboembolic events we have observed here is unexpected. The reported rates could be at least partly attributable to loss to follow-up, leading to an underestimation of adverse events but there could be another explanation. The degree to which this population was anticoagulated is an important factor which may have influenced the rates of adverse events. INR was less well controlled than reported in other settings [27, 32–35]. Only 48.6 % (95 % CI 46.1–51.1) of results were in range, and the majority of the out of range readings were below the target (Fig. 1a, b). Significantly fewer readings were within range for the
123
232
group with a higher INR target. Adequate anticoagulation was not achieved in either group, but in particular the patients in the group with the higher target were not anticoagulated sufficiently to combat the extra risk of thromboembolism they have. Qualitative investigation into the reasons behind why many patients are being treated subtherapeutically was beyond the scope of this research. However, on discussion with the clinicians in the cardiology department at SGH, it was apparent that INR values below 2.5 were felt to be beneficial in this region. Differences in risk profiles of both thromboembolism and bleeding events between Asians and Caucasians could explain many of the findings outlined above. There is evidence to show that rates of bleeding on warfarin are higher in Asian patients, in particular warfarin-related intracranial haemorrhage [36]. The difference in risk of thromboembolic disease has also been shown to be higher in Asian populations than the global levels in three large trials [37–39]. This study reports rates of both major bleed and thromboembolism that are lower than expected. However the INR values indicate that the population has a lower intensity of anticoagulation than is recommended by Western guidelines and this could be a contributing factor to explain the low rates of adverse events in this population. Other studies have shown the benefit of lower INR target ranges in certain populations. A study in Japan concluded that a target range of 1.6–2.6 resulted in the least adverse events, including both bleeding and thromboembolism, in elderly AF patients [40] and a study in China showed that rates of thromboembolic events were lowest in the INR range 1.5–1.9, showing that even low levels of warfarin can have a beneficial effect in preventing thromboembolism.
F. Edwards et al.
1.03–9.82), suggesting that participants with a highly variable INR are at highest risk of adverse events. Serial INR measurements have been shown to be poor predictors of bleeds, since the rise in INR usually occurs too close to the bleeding event to be identified [41]. Using the variability of the INR could therefore be of clinical relevance in predicting patients who are at higher risk of bleeding, and in whom more effort needs to be made to control their INR. Amongst variables not directly measuring INR control, female gender was the only significant predictor of a bleeding outcome, with an OR of 3.51. The reasons for this are unclear, although it is probably due to social rather than biological differences between the genders. It could be due to differences in reporting of bleeds; perhaps females are more likely to report bruising or heavy vaginal bleeding than their male counterparts. Further research into this subject would be interesting. Limitations These results must be interpreted with care, due to a number of limitations of the study. The sample size was smaller than hoped and the loss to follow up of 8.6 % is noteworthy. Additionally, any conclusions about rates of thromboembolism are difficult to make, due to only one event occurring in the study. The multiple regression model was not particularly convincing, since it explained only 15 % of the variability when predicting bleeds. Although the study tried to reduce reporting bias and eliminate both patient and staff differences in perceptions of what constitutes a minor bleed by asking questions from a proforma, this will never be completely successful, as the mentioning of some bleeds, such as bruising, will always depend on the individual in question as to what is normal for them.
INR control as a risk factor for bleeding outcome Conclusion The importance of tight INR control in preventing haemorrhagic and thromboembolic complications is established. However, before now, measures of INR control which may confer an increased risk of bleeding have not been explored in an Asian population. Unfortunately, rates of major bleeding were too low to use this as a primary outcome, and therefore minor bleeds and major bleeds were combined and analysis was carried using the rates of any bleeding event. Of the seven variables relating to INR control which were investigated, three were significant at the 10 % level in univariate analysis, but the SD of the INR was the only one which was a significant predictor of a bleed outcome at the 5 % level. As the SD of mean INR recorded over the year increased by a unit, the odds of experiencing a bleed increased by just over three times, (OR 3.18, 95 % CI
123
Low rates of major bleeds and thromboembolic events were recorded in this observational cohort study, in addition to a substantial number of INR readings under the recommended target range. Although the study was small, and there was some loss to follow-up, these results suggest that the recommended INR ranges may not represent the optimal warfarin intensity for this population. A lower intensity of therapy, which was in fact observed in this cohort, could be beneficial in terms of preventing adverse events. Further studies, especially multi centre trials in Asia have previously been recommended [42], and will be useful in investigating lower target ranges in Asian populations. The importance of INR variability as an indicator of bleeding risk, rather than one-off measurements, is also advocated.
Bleeding events and associated factors in a cohort of adult Acknowledgments This study is registered under the Malaysian National Medical Research Register (NMRR-10-714-6755). We acknowledge the support from staff of the Department of Cardiology, Sarawak General Hospital, who are involved in this study. We are grateful for permission by the Director General of Health, Malaysia, for the publication of this manuscript.
References 1. Hirsh J (1991) Oral anticoagulant drugs. N Engl J Med 324:1865–1875 2. Levine MN, Raskob G, Landefeld S, Kearon C (2001) Hemorrhagic complications of anticoagulant treatment. Chest 119:108S– 121S 3. Schulman S, Beyth RJ, Kearon C, Levine MN (2008) Hemorrhagic complications of anticoagulant and thrombolytic treatment: American College of Chest Physicians Evidence-Based Clinical Practice Guidelines. Chest 133:257S–298S 4. Ageno W, Gallus AS, Wittkowsky A, Crowther M, Hylek EM, Palareti G (2012) Oral anticoagulant therapy: antithrombotic Therapy and Prevention of Thrombosis, 9th ed: American College of Chest Physicians Evidence-Based Clinical Practice Guidelines. Chest 141:e44S–e88S 5. Keeling D, Baglin T, Tait C, Watson H, Perry D, Baglin C et al (2011) Guidelines on oral anticoagulation with warfarin—fourth edition. Br J Haematol 154:311–324 6. Connolly SJ, Laupacis A, Gent M, Roberts RS, Cairns JA, Joyner C (1991) Canadian atrial fibrillation anticoagulation (CAFA) Study. J Am Coll Cardiol 18:349–355 7. Kearon C, Ginsberg JS, Kovacs MJ, Anderson DR, Wells P, Julian JA et al (2003) Comparison of low-intensity warfarin therapy with conventional-intensity warfarin therapy for longterm prevention of recurrent venous thromboembolism. N Engl J Med 349:631–639 8. Petersen P, Boysen G, Godtfredsen J, Andersen ED, Andersen B (1989) Placebo-controlled, randomised trial of warfarin and aspirin for prevention of thromboembolic complications in chronic atrial fibrillation. The Copenhagen AFASAK study. Lancet 1:175–179 9. Ansell J, Hirsh J, Poller L, Bussey H, Jacobson A, Hylek E (2004) The pharmacology and management of the vitamin K antagonists: the Seventh ACCP Conference on Antithrombotic and Thrombolytic Therapy. Chest 126:204S–233S 10. Dang MT, Hambleton J, Kayser SR (2005) The influence of ethnicity on warfarin dosage requirement. Ann Pharmacother 39:1008–1012 11. Johnson JA (2008) Ethnic differences in cardiovascular drug response: potential contribution of pharmacogenetics. Circulation 118:1383–1393 12. United States Food and Drug Administration. Warfarin Product Labelling. 4-1-2006. Ref Type: Online Source 13. Suzuki S, Yamashita T, Kato T, Fujino T, Sagara K, Sawada H et al (2007) Incidence of major bleeding complication of warfarin therapy in Japanese patients with atrial fibrillation. Circ J 71:761–765 14. You JH, Chan FW, Wong RS, Cheng G (2005) Is INR between 2.0 and 3.0 the optimal level for Chinese patients on warfarin therapy for moderate-intensity anticoagulation? Br J Clin Pharmacol 59:582–587 15. Yu HY, Liu CH, Chen YS, Wang SS, Chu SH, Lin FY (2005) Relationship of international normalized ratio to bleeding and thromboembolism rates in Taiwanese patients receiving vitamin K antagonist after mechanical valve replacement. J Formos Med Assoc 104:236–243
233 16. Scottish Intercollegiate Guidelines Network. Antithrombotic Therapy. 8-5-2012. Ref Type: Online Source 17. Khor GL (2001) Cardiovascular epidemiology in the Asia-Pacific region. Asia Pac J Clin Nutr 10:76–80 18. Rosendaal FR, Cannegieter SC, van der Meer FJ, Brie¨t E (1993) A method to determine the optimal intensity of oral anticoagulant therapy. J Thromb Haemost 69:236–239 19. INR Pro. INR Pro. 2013. 28-7-2013. Ref Type: Online Source 20. Schulman S, Kearon C (2005) Definition of major bleeding in clinical investigations of antihemostatic medicinal products in non-surgical patients. J Thromb Haemost 3:692–694 21. Witt DM, Delate T, Clark NP, Martell C, Tran T, Crowther MA et al (2009) Outcomes and predictors of very stable INR control during chronic anticoagulation therapy. Blood 114:952–956 22. Lip GY, Frison L, Halperin JL, Lane DA (2011) Comparative validation of a novel risk score for predicting bleeding risk in anticoagulated patients with atrial fibrillation: the HAS-BLED (Hypertension, Abnormal Renal/Liver Function, Stroke, Bleeding History or Predisposition, Labile INR, Elderly, Drugs/Alcohol Concomitantly) score. J Am Coll Cardiol 57:173–180 23. Rose AJ, Berlowitz DR, Miller DR, Hylek EM, Ozonoff A, Zhao S et al (2012) INR targets and site-level anticoagulation control: results from the Veterans AffaiRs Study to Improve Anticoagulation (VARIA). J Thromb Haemost 10:590–595 24. Dauphin C, Legault B, Jaffeux P, Motreff P, Azarnoush K, Joly H et al (2008) Comparison of INR stability between self-monitoring and standard laboratory method: preliminary results of a prospective study in 67 mechanical heart valve patients. Arch Cardiovasc Dis 101:753–761 25. Warkentin AE, Donadini MP, Spencer FA, Lim W, Crowther M (2012) Bleeding risk in randomized controlled trials comparing warfarin and aspirin: a systematic review and meta-analysis. J Thromb Haemost 10:512–520 26. Landefeld CS, Beyth RJ (1993) Anticoagulant-related bleeding: clinical epidemiology, prediction, and prevention. Am J Med 95:315–328 27. Palareti G, Leali N, Coccheri S, Poggi M, Manotti C, D’Angelo A et al (1996) Bleeding complications of oral anticoagulant treatment: an inception-cohort, prospective collaborative study (ISCOAT). Italian Study on Complications of Oral Anticoagulant Therapy. Lancet 348:423–428 28. Evans A, Kalra L (2001) Are the results of randomized controlled trials on anticoagulation in patients with atrial fibrillation generalizable to clinical practice? Arch Intern Med 161:1443–1447 29. Gottlieb LK, Salem-Schatz S (1994) Anticoagulation in atrial fibrillation. Does efficacy in clinical trials translate into effectiveness in practice? Arch Intern Med 154:1945–1953 30. Stewart RA (2009) Underestimation of the effect of bleeding in clinical trials. Lancet 374:1145–1146 31. Oake N, Jennings A, Forster AJ, Fergusson D, Doucette S (2008) Anticoagulation intensity and outcomes among patients prescribed oral anticoagulant therapy: a systematic review and metaanalysis. CMAJ 179:235–244 32. Kim JH, Song YB, Shin DH, Kim JS, Choi JO, On YK et al (2009) How well does the target INR level maintain in warfarintreated patients with non-valvular atrial fibrillation? Yonsei Med J 50:83–88 33. Jones M, McEwan P, Morgan CL, Peters JR, Goodfellow J, Currie CJ (2005) Evaluation of the pattern of treatment, level of anticoagulation control, and outcome of treatment with warfarin in patients with non-valvar atrial fibrillation: a record linkage study in a large British population. Heart 91:472–477 34. Reynolds MW, Fahrbach K, Hauch O, Wygant G, Estok R, Cella C et al (2004) Warfarin anticoagulation and outcomes in patients with atrial fibrillation: a systematic review and metaanalysis. Chest 126:1938–1945
123
234 35. Okumura K, Komatsu T, Yamashita T, Okuyama Y, Harada M, Konta Y et al (2011) Time in the therapeutic range during warfarin therapy in Japanese patients with non-valvular atrial fibrillation. A multicenter study of its status and infuential factors. Circ J 75:2087–2094 36. Shen AY, Yao JF, Brar SS, Jorgensen MB, Chen W (2007) Racial/ethnic differences in the risk of intracranial hemorrhage among patients with atrial fibrillation. J Am Coll Cardiol 50:309–315 37. Connolly SJ, Ezekowitz MD, Yusuf S, Eikelboom J, Oldgren J, Parekh A et al (2009) Dabigatran versus warfarin in patients with atrial fibrillation. N Engl J Med 361:1139–1151 38. Patel MR, Mahaffey KW, Garg J, Pan G, Singer DE, Hacke W et al (2011) Rivaroxaban versus warfarin in nonvalvular atrial fibrillation. N Engl J Med 365:883–891
123
F. Edwards et al. 39. Granger CB, Alexander JH, McMurray JJ, Lopes RD, Hylek EM, Hanna M et al (2011) Apixaban versus warfarin in patients with atrial fibrillation. N Engl J Med 365:981–992 40. Yasaka M, Minematsu K, Yamaguchi T (2001) Optimal intensity of international normalized ratio in warfarin therapy for secondary prevention of stroke in patients with non-valvular atrial fibrillation. Intern Med 40:1183–1188 41. Kucher N, Connolly S, Beckman JA, Cheng LH, Tsilimingras KV, Fanikos J et al (2004) International normalized ratio increase before warfarin-associated hemorrhage: brief and subtle. Arch Intern Med 164:2176–2179 42. Kwong JS, Yu CM (2013) The need for multicentre cardiovascular clinical trials in Asia. Nat Rev Cardiol 10:355–362
