Jump to
Abstract
Background— Warfarin is effective in the prevention of stroke in atrial fibrillation but is under used in clinical care. Concerns exist that published rates of hemorrhage may not reflect real-world practice. Few patients ≥80 years of age were enrolled in trials, and studies of prevalent use largely reflect a warfarin-tolerant subset. We sought to define the tolerability of warfarin among an elderly inception cohort with atrial fibrillation.
Methods and Results— Consecutive patients who started warfarin were identified from January 2001 to June 2003 and followed for 1 year. Patients had to be ≥65 years of age, have established care at the study institution, and have their warfarin managed on-site. Outcomes included major hemorrhage, time to termination of warfarin, and reason for discontinuation. Of 472 patients, 32% were ≥80 years of age, and 91% had ≥1 stroke risk factor. The cumulative incidence of major hemorrhage for patients ≥80 years of age was 13.1 per 100 person-years and 4.7 for those <80 years of age (P=0.009). The first 90 days of warfarin, age ≥80 years, and international normalized ratio (INR) ≥4.0 were associated with increased risk despite trial-level anticoagulation control. Within the first year, 26% of patients ≥80 years of age stopped taking warfarin. Perceived safety issues accounted for 81% of them. Rates of major hemorrhage and warfarin termination were highest among patients with CHADS2 scores (an acronym for congestive heart failure, hypertension, age ≥75, diabetes mellitus, and prior stroke or transient ischemic attack) of ≥3.
Conclusions— Rates of hemorrhage derived from younger noninception cohorts underestimate the bleeding that occurs in practice. This finding coupled with the short-term tolerability of warfarin likely contributes to its underutilization. Stroke prevention among elderly patients with atrial fibrillation remains a challenging and pressing health concern.
Received July 21, 2006; accepted March 13, 2007.
Warfarin is highly effective in the prevention of stroke in atrial fibrillation (AF).1,2 Despite its proven benefit, studies attest to its underutilization particularly among elderly individuals who face the highest risk.3–6 In a recent study of 21 teaching, 13 community, and 4 Veterans Administration hospitals in 28 states, the use of warfarin at discharge was 54% even among patients considered to be at highest risk.7 Age >80 years and perceived bleeding risk were negative predictors of warfarin use. In a recent prospective study, 51% of patients were discharged on warfarin. Of those not on warfarin, 83% had ≥2 risk factors for stroke and 23% had taken warfarin in the past but were unable to tolerate it in the long term.8 Stroke is the leading cause of disability and the third leading cause of death in the United States with an estimated annual total cost of 57.9 billion dollars.9 Given the aging of our population, it is projected that 7.5 million individuals will have AF in the United States by the year 2020 on the basis of an expected prevalence of 13.5% for individuals ≥75 years of age, and 18.2% for those ≥85 years.10
Editorial p 2684
Clinical Perspective p 2696
Rates of major hemorrhage from randomized trials and observational cohorts have been reassuringly low.11–15 However, published rates may be underestimates, as few patients >80 years of age were enrolled and few cohort studies include the initial phase of therapy, which is reported to convey the highest risk.13,16–19 Observational noninception cohort studies focus on prevalent warfarin use among individuals who have proven they can tolerate anticoagulant therapy. Moreover, recent trials have also largely enrolled patients on vitamin K antagonists at study entry.20–22
Higher rates of hemorrhage with incident warfarin use may in part underlie the reported underutilization of warfarin from cross-sectional studies and noninception cohorts. Long-term tolerability of anticoagulant therapy is largely unknown and may also be a contributing factor. We conducted an inception cohort study to define the rate of major hemorrhage in patients at warfarin initiation and to define the risk of bleeding in the early phase of therapy. We also focused on the tolerability of warfarin during the first year and determined the physician reason for discontinuation with attention to perceived safety issues.
Methods
Design and Patients
To be eligible, patients had to be ≥65 years of age, have AF verified by ECG, be new to warfarin (or had taken none within the previous 12 months), have their care established at the study institution, and have their warfarin managed by the on-site anticoagulation clinic. Potentially eligible patients were prospectively identified through daily searches of electronic admission notes of patients admitted to the medical service of Massachusetts General Hospital from January 2001 to June 2003. We intentionally did not screen patients admitted to the stroke or surgical service given their higher baseline risk of bleeding. Outpatients who started warfarin were identified at the time of the faxed physician referral to the anticoagulation clinic and were tracked to the first appointment. Referred ambulatory patients with AF-related stroke were eligible only if the event had occurred at least 4 weeks before the first appointment.
Outcomes
Patients were enrolled on the first day of warfarin and followed through the first year. Outcomes included major hemorrhage, time to termination of warfarin, and physician reason for discontinuation. Major hemorrhage was defined as fatal, hospitalization with transfusion of ≥2 units of packed red blood cells, or involvement of a critical site (ie, intracranial, retroperitoneal, intraspinal, intraocular, pericardial, or atraumatic intra-articular hemorrhage). The international normalized ratio (INR) at the time of the event was recorded in addition to other potentially contributing factors such as a fall. The reason for warfarin termination was recorded from the medical record. If the reason was not explicitly stated, the physician was queried directly. Reasons for termination included major hemorrhage, other hemorrhage, falls, patient nonadherence with medication or monitoring, coagulopathy, or sustained sinus rhythm.
Patient Characteristics
Demographic and clinical characteristics extracted from the medical record included known risk factors for stroke (hypertension, prior stroke, heart failure, diabetes mellitus) and potential risk factors for hemorrhage (prior hemorrhage, liver disease, falls, active alcohol abuse, active malignancy, renal impairment, dementia). Stroke risk scores were calculated according to the CHADS2 scheme (congestive heart failure=1 point, hypertension=1, age ≥75=1, diabetes mellitus=1, and prior stroke or transient ischemic attack=2).23 Concurrent medications were obtained from the discharge summary, electronic medication list, and physician office record and were verified by the patient or family member at the first anticoagulation appointment. Use of aspirin or nonsteroidal antiinflammatory medications was also recorded.
Anticoagulation Intensity and Warfarin Management
Per clinic routine, all patients attended a mandatory 60-minute educational session on warfarin. Time spent in the INR ranges <2.0, 2.0 to 3.0, 3.1 to <4.0, and ≥4.0 was calculated with linear interpolation following the method described by Rosendaal et al.24 This method does not interpolate an INR if the interval between 2 consecutive measurements exceeds 8 weeks.
Statistical Analysis
Baseline differences between age groups were evaluated with χ2 tests for categorical variables and t tests for continuous variables. Crude event rates were calculated for different age groups, INR categories, time since warfarin initiation, and CHADS2 categories by division of the number of first events by the total person-years of follow-up in the specified category. The INR at the time of event was unobtainable for 1 patient and this patient was excluded from analyses that involved this variable. Poisson regression models with generalized estimating equations were used to calculate incidence rate ratios (IRR) and 95% confidence intervals (CIs). The number of events precluded use of a multivariable regression model.
Cumulative incidence of major bleeding and intolerability of therapy, defined as stopping warfarin therapy for safety reasons, were both estimated with the Kaplan-Meier method. Time to major bleeding was censored at the time of cessation of warfarin therapy, death, or transfer of medical care, whereas time to cessation of warfarin was censored for reasons other than safety, death, or transfer of medical care. Log-rank tests were used to assess differences between the 2 age groups. For all analyses, a 2-sided probability value <0.05 was considered statistically significant. Analyses were performed with Stata statistical software, release 8.0 (Stata Corporation, College Station, Tex.).
Role of the Funding Source
The sponsor had no role in the design and conduct of the study, the collection, analysis, and interpretation of data, or in the preparation, review, and approval of the manuscript. The study was approved by the institutional review board at Massachusetts General Hospital. The observational nature of the study did not require written informed consent.
The authors had full access to and take full responsibility for the integrity of the data. All authors have read and agree to the manuscript as written.
Results
During the study period, 533 patients were identified and 472 were enrolled; 61% of patients had their warfarin managed elsewhere (eg, at a skilled nursing facility) and were not included. A total of 47% of patients were female and 54% were >75 years of age (32% were ≥80 years of age) (Table 1). A total of 59% (n=279) of patients presented with the first documented episode of AF and 35% (n=165) with a recurrent episode. Thirty-five percent of patients had documented coronary artery disease, and 40% overall were on aspirin for primary or secondary prevention.25 Compared with risk factors for stroke, potential risk factors for hemorrhage were less common except for older age. Of the 472 patients, 33% were enrolled at hospital discharge; 42% of these were ≥80 years of age. Symptoms related to an uncontrolled ventricular rate prompted admission for the majority of patients. A total of 90% of patients ≥80 years had a CHADS2 score ≥2. With the use of the Outpatient Bleeding Risk Index, 95.3% of patients would have been classified as intermediate risk and 4.7% as high risk for major hemorrhage at the start of warfarin.26
TABLE 1. Clinical Characteristics of Patients at the Start of Warfarin Therapy
Follow-up was complete for 100% of the cohort. During the first year of therapy, 134 patients were taken off warfarin, 16 died of unrelated causes, 16 transferred their warfarin management, and 306 patients remained on warfarin at the end of the 1-year observation period. The total observation time was 360 person-years.
Anticoagulation Control
The total number of INR measurements was 10 031. Person-time by INR category could not be determined for 21 person-years (5.7%) of the total time on warfarin. During the study, 58% of person-time was spent within the INR range 2.0 to 3.0, 29% below 2.0, 11% within 3.1 to <4.0, and 2% ≥4.0.
Major Hemorrhage
During the first year, 26 patients sustained a major hemorrhage (9 intracranial, 11 gastrointestinal, 1 retroperitoneal, 1 hemothorax after a fall, 1 ocular, 1 hemarthrosis, and 2 epistaxis that required transfusions). The rate of major hemorrhage was 7.2 per 100 person-years (95% CI 4.9 to 10.6), and the rate of intracranial hemorrhage was 2.5% (95% CI 1.1 to 4.7). Patients ≥80 years of age experienced higher rates of major bleeding compared with younger patients (13.08 per 100 person-years versus 4.75 per 100 person-years, P=0.010). Figure 1 shows the cumulative major hemorrhage rate over time between the 2 age groups (P=0.009). Table 2 presents rates of hemorrhage by 3 risk factors, INR range, age group, and time since warfarin initiation. Risk of hemorrhage was increased by INR ≥4.0 (IRR 19.34, 95% CI 8.26 to 45.34), older age (IRR 2.75, 95% CI 1.27 to 5.95), and the first 90 days of therapy (IRR 3.31, 95% CI 1.51 to 7.25). INR ≥4.0 increased the risk, although only 2% of person-time was spent in this range. Fifteen of the 26 major hemorrhages (58%) occurred within 90 days of warfarin initiation, 11 within 30 days, and 7 within the first 2 weeks. Six of these 7 patients were enrolled at the time of hospital discharge; 2 were bridged with heparin. INR elevation was correlated with the other 2 risk factors. Of the 8 patients who experienced a major hemorrhage with an INR ≥4.0, 5 patients sustained the hemorrhage within 90 days, 5 patients were ≥80 years of age, and 3 patients had all 3 risk factors. Of the 15 hemorrhagic events that occurred within the first 90 days, the INR was ≥4.0 in 5 of them. However, age and time since warfarin initiation remained significant predictors when analysis was restricted to INR <4 (Table 2).
Figure 1. Cumulative incidence of major bleeding among patients aged ≥80 years and <80 years (n=472). Numbers below graph are the number of patients without bleeding who continued on warfarin at that time point (P=0.009, log-rank test).
TABLE 2. Incidence Rates and Incidence Rate Ratios for Major Hemorrhage Among Patients Newly Starting Warfarin
In the overall cohort, 40% of patients were on aspirin. Twelve of the 26 events (46%) occurred on aspirin and included 5 gastrointestinal and 4 intracranial bleeds. The dose was 81 mg for 8 of these patients. Nine of the 12 events occurred with an INR <4 and 6 patients were <80 years of age. Of the total 9 intracranial bleeds, 3 were fatal, 6 were intracerebral, and 2 were associated with documented falls. INR values were available for 8 patients and exceeded 4.0 in 3 of them. One intracerebral bleed occurred within the first 30 days, and 8 of 9 bleeds occurred in patients who were ≥75 years of age (3 were ≥80 years of age). Overall, the rate of major hemorrhage was higher in those with CHADS2 scores of ≥3 (IRR 8.19, 95% CI 3.37 to 19.88) (Table 3).
TABLE 3. Distribution of Major Hemorrhagic Events and Warfarin Terminations Due to Perceived Safety Concerns by CHADS2 Score
Tolerability of Warfarin Among Patients ≥80 Years of Age
By the end of the first year, 134 patients had been taken off warfarin. Sustained sinus rhythm was the predominant reason for those <80 years of age, 63% (57 of 91 patients), compared with those ≥80 years of age, 19% (8 of 43 patients). Concerns related to safety accounted for 81% of those patients ≥80 years of age who stopped therapy during the first year (17 bleeding complications, 9 falls, 5 nonadherence with drug or monitoring, 3 coagulopathy, and 1 dermatologic reaction) compared with 37% of patients <80 years of age (P<0.001). Figure 2 shows the smoothed hazard estimates over time for the 2 age groups. The risk of stopping warfarin peaked early and then, beginning at 6 months, approximated that of younger patients. With death and maintenance of sinus rhythm excluded as reasons for warfarin cessation, 26% of patients ≥80 years of age stopped warfarin within the first year. Similar to major hemorrhage, the rate of warfarin termination for perceived safety concerns was higher among patients with CHADS2 scores of ≥3 (IRR 2.27, 95% CI 1.41 to 3.66) (Table 3).
Figure 2. Risk of stopping warfarin in the first year on the basis of perceived safety concerns by age. Numbers below graph are the number of patients on warfarin at that time point (P<0.001, log-rank test).
Discussion
We found the rate of major hemorrhage on warfarin to be higher than previously reported. The aggregate rate was 7.2 per 100 person-years and 13.08% versus 4.75% for patients ≥80 years of age compared with patients <80 years of age. The first 90 days were associated with a 3-fold increased risk. Although only 2% of person-years were spent in an INR range of ≥4.0, INR was a strong risk factor. The long-term tolerability of warfarin among individuals ≥80 years of age has not previously been assessed. We found that 26% of these patients were taken off warfarin; concerns related to safety accounted for 81% of them.
The higher rate of major hemorrhage is likely attributable to the advanced age of our study population and restriction of our cohort to patients who started warfarin (Table 4). A total of 32% (n=153) of patients were ≥80 years of age compared with a total of 20 patients >75 in the pooled analysis of the first 5 randomized trials. Higher rates of hemorrhage were found in the Stroke Prevention in Atrial Fibrillation II Study, which consisted of 2 parallel trials of patients aged >75 years and ≤75 years.27 Of note, the trials used an INR target range of 2.0 to 4.5. Annual rates of major bleeding were 4.2% and 1.7%, respectively. In our study, we reaffirmed the time-dependence of risk that has been shown in other studies. Underestimation of early adverse events and early cessation of warfarin among higher-risk patients likely contribute to the lower rates of bleeding reported from studies with prevalent user designs that largely reflect a warfarin-tolerant subset. This issue is particularly relevant given recent trials that have enrolled up to 84% of patients on warfarin at baseline.20–22 Given the findings of our study, differences in treatment effect would be anticipated between prevalent users and those new to therapy.22 New warfarin status may also help to explain the finding of increased bleeding among patients with a first episode of AF.15 Our study highlights the importance of new-user designs in the evaluation of newer antithrombotic drugs to minimize survivor bias and to better assess drug effects in the early period.28,29 The rate of major hemorrhage found in our study is similar to that of other inception cohorts with several caveats (Table 4). These studies included patients with different indications and higher INR target intensities and, in the Landefeld16 and Beyth26 studies, 100% of patients were identified at hospital discharge. In contrast, our study was restricted to patients with AF, target INR of 2.0 to 3.0, and 33% were identified at discharge. The distinct difference is the age of the patients enrolled: 6% versus 32% were ≥80 years of age. Elevated INR is a firmly established risk factor for hemorrhage. The percent of person-years spent in an INR range >3.0 in our study, 13%, was nearly identical to that reported in 2 recent trials, 12%21 and 15%,22 which emphasizes the quality of anticoagulation that was achieved.
TABLE 4. Rates of Major Hemorrhage by Study Design, Age, and Proportion of Patients New to Warfarin
A potential limitation of our study is that it was conducted at a single academic center that may not reflect other settings. However, the rate of bleeding found in our study may be an underestimate of that experienced in practice. We did not include patients whose warfarin was managed outside of our clinic. Because the clinic does not manage the warfarin of long-term care residents, these potentially frailer patients were not included. Patients had to have their longitudinal care established at the study institution to minimize referral bias, and those with higher baseline risk of bleeding (eg, acute stroke, postoperative AF) were not included. We do not believe that physicians were overly aggressive in the use of warfarin, given the lower prevalence of risk factors for hemorrhage at baseline compared with the high prevalence of risk factors for ischemic stroke. In addition, as we had previously reported, the prevalence of risk factors for bleeding was considerably higher among patients not prescribed warfarin at hospital discharge compared with that of patients enrolled in the current study: prior hemorrhage 32% versus 10%, prior fall 38% versus 4%, cognitive impairment 22% versus 3%, renal dysfunction 29% versus 11%.8 Only 3 of 26 patients who sustained a major hemorrhage would have been classified as high risk by the Outpatient Bleeding Risk Index. However, it is of note that this index was derived from a cohort with a mean age of 61 years, in contrast to the mean age of 77 years in our study.
Despite these limitations, our study has several key strengths. We consecutively identified and followed all eligible patients from the first day of warfarin to ensure complete capture of adverse events. Warfarin management was provided by an experienced staff who worked in a long-established and well-organized clinic. We were able to directly ascertain use of aspirin, an important and often underreported confounder because of its over-the-counter status. Of the 472 patients, 32% were ≥80 years of age, which constitutes the largest proportion of patients in this age group represented in a prospective real-world study of AF.
Implications for Management of Anticoagulation in the Elderly Patient
In our study, patients at highest risk of stroke also experienced most of the bleeding, which illustrates the clinical complexity of these patients. Because cardioembolic strokes are associated with a 30-day mortality of 24% and significant disability among those who survive,30,31 decisions not to prescribe warfarin for patients in this high-risk group should be largely influenced by hemorrhagic outcomes that are equal in magnitude to the irreversible sequelae of an ischemic stroke and not indiscriminate comparison of stroke rates to aggregate rates of bleeding. The rate of intracranial hemorrhage in our study was 2.5% (95% CI 1.1 to 4.7) and 1.7% for intracerebral hemorrhage. Strategies to decrease this risk need to be aggressively sought and implemented. Control of blood pressure has been shown to significantly reduce this risk.32 The benefit of the addition of aspirin to warfarin for cardiovascular disease needs to be rigorously defined and justified.33–35 Recent guidelines advise against this combination for older patients with AF and stable coronary disease.36 Interventions to reduce falls are important as these patients experience higher rates of intracranial hemorrhage, 2.8% (95% CI 1.9 to 4.1).37,38 Although use of lower INR targets has been suggested to offset bleeding risk, this strategy has not been shown to decrease hemorrhage.31,36,39–41 An INR interval of 2.0 to 2.5 in theory seems like a rational compromise, but this degree of precision would be difficult to achieve in practice given warfarin’s variable dose response.31,42 Because elderly patients are slower to normalize an elevated INR, more aggressive management of excessive anticoagulation in this age group seems prudent.43 Vigilant monitoring in the early phase of therapy will help reduce bleeding during this risk-prone period. Further research of age-related vasculopathies that predispose to intracerebral hemorrhage is needed.33,39,44–46 A reduction of the risk of extracranial hemorrhage is critically important as these events are associated with significant morbidity and often precipitate termination of therapy. Thresholds for discontinuation of warfarin may differ among physicians, patients, and families and thus warrant further study.47 Our findings also highlight the need for continued research into the mechanisms of AF and precipitants of thrombus formation, insights that may elucidate novel pathways for stroke prevention without the attendant risk of hemorrhage.48
Conclusions
Published rates of major hemorrhage derived from younger noninception cohorts underestimate the bleeding that occurs in clinical practice. Higher rates of bleeding and short-term tolerability likely contribute to the reported underutilization of warfarin among elderly patients with AF. Stroke prevention among the highest-risk patients remains a challenge. Given the aging of the population, stroke prevention in AF is a pressing health concern.
Acknowledgments
The authors would like to thank Dr Robert Hughes, director of the Massachusetts General Hospital Anticoagulation Management Service, and the clinic nurses for their dedication to patient care.
Source of Funding
The Robert Wood Johnson Foundation Generalist Physician Faculty Scholars Program Grant No. 039174.
Disclosures
Dr Hylek has served on advisory boards for Bristol-Myers Squibb and has received research support from AstraZeneca and Bristol-Myers Squibb. The other authors report no conflicts.
References
CLINICAL PERSPECTIVE
Warfarin is effective in the prevention of stroke in atrial fibrillation but is under used in clinical care. Concerns exist that published rates of hemorrhage may not reflect real-world practice. Few patients ≥80 years of age were enrolled in trials, and studies of prevalent use largely reflect a warfarin-tolerant subset. We sought to define the tolerability of warfarin among an elderly inception cohort with atrial fibrillation. Consecutive patients ≥65 years of age who started warfarin were identified and followed for 1 year. Study outcomes included major hemorrhage, time to termination of warfarin, and reason for discontinuation. Of 472 patients enrolled, 32% were ≥80 years of age, and 91% had ≥1 stroke risk factor. The cumulative incidence of major hemorrhage for patients ≥80 years of age was 13.1 per 100 person-years and 4.7 for those <80 years of age (P=0.009). Despite trial-level anticoagulation control, we found that the first 90 days of warfarin, age ≥80 years, and international normalized ratio (INR) ≥4.0 were each associated with increased risk of hemorrhage. Within the first year, 26% of patients ≥80 years of age stopped taking warfarin; perceived safety issues accounted for 81% of these. Rates of major hemorrhage and warfarin termination were highest among patients with CHADS2 scores (an acronym for congestive heart failure, hypertension, age ≥75, diabetes mellitus, and prior stroke or transient ischemic attack) of ≥3. Rates of hemorrhage derived from younger noninception cohorts underestimate the bleeding that occurs in real-world practice. This finding, coupled with the short-term tolerability of warfarin, likely contributes to its underutilization. Stroke prevention among elderly patients with atrial fibrillation remains a challenging and pressing health concern.
Footnotes
-
Guest Editor for this article was Douglas P. Zipes, MD.
This Issue
Jump to
Article Tools
Share this Article
- Major Hemorrhage and Tolerability of Warfarin in the First Year of Therapy Among Elderly Patients With Atrial Fibrillation
