SAN DIEGO—Veterans with atrial fibrillation who switch from warfarin to dabigatran may have a higher risk of bleeding, compared with veterans who continue taking warfarin, according to research presented at the 2014 International Stroke Conference. The risk of death, however, may be slightly lower among veterans with atrial fibrillation who switch from warfarin to dabigatran, compared with those who continue taking warfarin.
Although the lower odds of death among patients taking dabigatran are an important finding, “the decision to recommend dabigatran requires careful evaluation of patient characteristics that may increase the likelihood of gastrointestinal hemorrhage,” said Mary Vaughan Sarrazin, PhD, Core Investigator at the Iowa City Veterans Affairs Center for Access and Delivery Research and Evaluation and Associate Research Professor in Internal Medicine at the University of Iowa in Iowa City.
Mary Vaughan Sarrazin, PhD
A Retrospective Analysis of Veterans’ Data
Dr. Sarrazin and colleagues conducted a longitudinal analysis of a retrospective cohort of patients at Veterans Affairs (VA) medical centers to evaluate the relative risks of bleeding (ie, gastrointestinal, intracranial, and other hemorrhage) and death for patients with atrial fibrillation who switch to dabigatran after at least six months on warfarin, compared with patients who continue taking warfarin. Using national, VA, administrative, laboratory, and pharmacy data from fiscal years 2010 to 2012, the investigators identified approximately 85,000 patients who met the study criteria. Patients who had severe renal disease or a mechanical valve were excluded.
During the 15-month follow-up, 1,394 patients (approximately 1.7% of the study population) switched from warfarin to dabigatran. Dabigatran and warfarin exposure were defined at the start of each follow-up week using national VA pharmacy data. Patients who initiated dabigatran were assumed to stay on dabigatran, and patients who never initiated dabigatran were assumed to stay on warfarin. In sensitivity analysis, the researchers censored patients who stopped taking their medication.
The occurrence of each outcome was defined separately for each type of bleeding event and for death, and the dates of the outcomes were identified using VA data. Patient characteristics were calculated at baseline and over time. The characteristics included age, race, distance to the nearest VA medical center, International Normalized Ratio (INR) time in therapeutic range, mean creatinine glomerular filtration rate, CHADS2 stroke risk score, HAS-BLED bleeding risk score, and comorbid conditions.
Odds of Bleeding Were Higher With Dabigatran
In some ways, patients who switched to dabigatran were healthier than patients who did not switch, said Dr. Sarrazin. Patients who switched were slightly younger, had lower mean CHADS2 risk scores, and were less likely to have kidney disease. On the other hand, patients who switched to dabigatran had more markers of heart disease, such as greater likelihood of dysrhythmia, heart failure, or prior myocardial infarction. Patients who switched to dabigatran also had slightly lower time in therapeutic INR range and were more likely to live 60 miles or farther from the nearest VA medical center.
Of the 1,394 patients who switched to dabigatran, 131 had a bleeding event, and 49 patients died during weeks of dabigatran use. More than 10,000 bleeding events and nearly 7,000 deaths were associated with warfarin. The unadjusted relative odds of any bleeding were 1.47 times higher, and the unadjusted odds of gastrointestinal hemorrhage were 1.74 times higher, during weeks of dabigatran use. The investigators found no significant difference between the drugs in the odds of intracranial or other hemorrhage, although the occurrence of these events was rare.
After the researchers controlled for baseline and time-varying covariates using marginal structural model weights and logistic regression, they found that the relative odds of bleeding were still higher with dabigatran than with warfarin. The relative odds of death, however, were lower during weeks of dabigatran use in unadjusted analysis and in risk-adjusted analysis.
Sensitivity analysis that censored patients who ceased their medication produced similar results to the primary analysis, although the lower odds of death for participants who switched to dabigatran were no longer statistically significant. An analysis using a composite outcome defined as bleeding or death demonstrated that a higher likelihood of bleeding was associated with dabigatran, but the difference was small. “In our primary analysis, we found a 1.61-times higher odds of gastrointestinal hemorrhage with dabigatran, compared with a 1.26-higher odds using the composite outcome,” said Dr. Sarrazin. She and her colleagues found no significant difference in the odds of bleeding when they defined bleeding events based on inpatient claims alone.
Investigators Did Not Examine Bleeding Severity
Among the study’s limitations is its low power to detect significant differences for rare outcomes, such as intracranial bleeding or inpatient bleeding, said Dr. Sarrazin. In addition, the researchers’ assessment of bleeding history and outcomes depended on the accuracy of the ICD-9 diagnosis codes for bleeding, and the investigators did not examine bleeding severity. Finally, despite the use of multivariable analyses to control for differences in pre-existing risk factors between patients who did and did not switch to dabigatran, differences in unmeasured risk factors could account for the higher bleeding risk or lower death in patients who initiated dabigatran.