Data regarding the outcomes of restarting anticoagulation in patients who develop gastrointestinal bleeding (GIB) while anticoagulated are sparse. We hypothesized that restarting anticoagulation in these patients is associated with better outcomes. This is a retrospective cohort study that enrolled subjects who developed GIB while on anticoagulation from 2005 to 2010. Atrial fibrillation was defined by history and electrocardiography on presentation. GIB was defined as a decrease in hemoglobin by 2 g, visible bleeding, or positive endoscopic evaluation. Time-to-event adjusted analyses were performed to find an association of restarting warfarin and recurrent GIB, arterial thromboembolism, and mortality. Stratified analysis by duration of interruption of warfarin was also performed. Overall, 1,329 patients (mean age 76 years, women 45%) developed major GIB. Warfarin was restarted in 653 cases (49.1%). Restarting warfarin was associated with decreased thromboembolism (hazard ratio [HR] 0.71, 95% confidence interval [CI] 0.54 to 0.93, p = 0.01) and reduced mortality (HR 0.67, 95% CI 0.56 to 0.81, p <0.0001) but not recurrent GIB (HR 1.18, 95% CI 0.94 to 1.10, p = 0.47). When the outcomes were stratified by duration of warfarin interruption, restarting warfarin after 7 days was not associated with increased risk of GIB but was associated with decreased risk of mortality and thromboembolism compared with resuming after 30 days of interruption. Decision to restart warfarin after an episode of major GIB is associated with improved survival and decreased thromboembolism without increased risk of GIB after 7 days of interruption.
In patients with nonvalvular atrial fibrillation (AF), warfarin has been the anticoagulant agent of choice for the last 3 decades. Elderly patients appear to have a greater propensity for complications of therapy such as GIB on one end or cerebrovascular events on the other end of the therapeutic spectrum. GIB occurs in 5% to 15% of patients on long-term anticoagulation. Literature is scarce about the various outcomes in this specific population and controversy exists regarding various outcomes based on durations of interruption of warfarin. Short-term interruption has been shown to increase mortality and thromboembolism. It is important to understand the risks of development of recurrent gastrointestinal hemorrhage, mortality, and thromboembolism in the context of resuming warfarin in this particular group. Warfarin has shown to not only improve cardiovascular mortality but also quality of life. We hypothesized that warfarin treatment prevents thromboembolism and has mortality benefit over the subjects who are not restarted on warfarin. Also, we investigated the various durations of interruption and the risk of GIB, thromboembolism, and mortality.
Methods
This is a retrospective cohort study that evaluated the patients enrolled at the anticoagulation clinic of Henry Ford Health System with a large catchment area serving all socioeconomic strata, covering majority of Southeast Michigan, United States. Patients who developed major GIB (defined in the following) while taking warfarin and then had evidence of resolution of major GIB (defined as stability of hemoglobin levels with <1 g decrease of hemoglobin for 48 hours) were included in the study. Patients who died within the first 72 hours of GIB, hospice, postoperative or valvular AF, patients in whom primary indication for anticoagulation was any reason other than nonvalvular AF, and patients in whom warfarin was not interrupted for at least 48 hours were excluded.
Data were initially obtained from insurance claims of warfarin that were evaluated at Henry Ford Health System from January 2005 to December 2010. Patients who had commenced using warfarin for at least 1 year and with at least 2 prescriptions of warfarin within 3 months of contact with a physician were enrolled. In the absence of mortality, at least 2 years of available follow-up after that initial contact in the electronic health record system was a prerequisite. The database query was further narrowed using International Classification of Diseases, Ninth Revision, code 427.3, 427.31, and 427.32 for AF. They were then evaluated for the indications of anticoagulation. Patients other than those with nonvalvular AF as the primary reason for anticoagulation were excluded. The charts of these patients were evaluated to identify major GIB that led to discontinuation of warfarin therapy for at least 2 days. Eventually, 1,329 patients satisfied these criteria. They were reevaluated for accuracy by identifying either a history of AF or atrial flutter in the anticoagulation clinic notes or AF or atrial flutter on the electrocardiogram. A detailed chart review of these patients was carried out by the physician-investigators, and information on demographics, medication use, and other clinical variables was obtained. Index GIB was defined as information collected on the day of GIB and subsequently at the time of resolution of index GIB. Information for CHADS 2 and HAS-BLED was collected at the resolution of index GIB. The consortium diagram is available in the Supplementary material . Index GIB was defined as a decrease in hemoglobin of ≥2 g/dl and/or transfusion of ≥2 units of packed red blood cells with at least one of the following: hematemesis, melena, hematochezia, bright red blood per rectum, blood in nasogastric aspirate, or bleeding documented during an endoscopic procedure. The study protocol was approved by the institutional review board.
Restarting warfarin was defined as prescription of warfarin with objective evidence of increase in international normalized ratio to ≥2.0 with evidence of at least 2 days of discontinuation of warfarin as observed by chart review. Patients who interrupted warfarin after 1 month of restarting warfarin (54; 4.1%) were included in the group that restarted warfarin. Patients who started warfarin after 6 months of interruption (39, 2.9%) were included in the group that did not restart warfarin (warfarin cessation group). Patients who interrupted warfarin within the first month (12; 0.9%) were included in the warfarin cessation group.
Recurrent major GIB was defined as any of the following: (1) >2 g of hemoglobin decrease from the last known hemoglobin level warranting hospitalization, (2) need for blood transfusion of at least 2 units, and (3) visible bleeding by health personnel or endoscopic evidence of stigmata of recent bleeding in the form of visible bleeding or clot. We decided to analyze GIB within 90 days, as propensity for recurrent GIB should be short term. Also, 12% of the patients developed recurrent GIB and we only considered the index GIB.
Thromboembolism was defined as venous thromboembolism (pulmonary embolism and deep venous thrombosis), arterial thromboembolism, or stroke or transient ischemic attack. Stroke or transient ischemic attack was defined as any of the following: (1) recent stroke on brain computed tomogram, (2) on brain magnetic resonance image, and (3) diagnosis of stroke or transient ischemic attack by a neurologist. This outcome was obtained over 1-year duration and only the first episode of thromboembolism was considered as an outcome. Mortality data were updated from the Social Security Death Index, death certificates, or hospital notes. Outcomes were evaluated by 2 blind reviewers (FK and ZA). These were then rechecked and conflicts resolved by consensus.
Descriptive statistics were used to summarize the data. Continuous variables are expressed as mean ± SD and categorical variables are mentioned in percentages. They were analyzed using t test for continuous variables and Fisher’s exact test or chi-square test for categorical variables, as appropriate. Univariate analyses and adjusted Cox proportional analyses were conducted to enunciate the association between restarting warfarin therapy and outcomes. These analyses were adjusted for age, gender, race, Charlson co-morbidity index, number of blood product transfusions, international normalized ratio on admission, and CHADS 2 and HAS-BLED scores. Cumulative incidences were evaluated using Kaplan-Meier method. Time of interruption of warfarin along with restarting warfarin status was added as an interaction term in the Cox proportional hazards model to investigate the association of time in development of thromboembolism, mortality, and recurrent GIB. Because the interaction was significant, we divided the time of interruption of warfarin into 5 groups (<7, 7 to 15, 15 to 21, 21 to 30, and >30 days). Incidences were calculated by the formula: incidence per 100 person-years. Incidence rate ratios were calculated by the formula: incidence rate per 100 person-years of/incidence rate per 100 person-years in patients who interrupted warfarin for >30 days. Kaplan-Meier curves were produced to evaluate the differences in the cumulative incidences of recurrent GIB, mortality, and thromboembolism stratified by time duration of interruption of warfarin and cessation of warfarin use. All analyses were performed using PASW, version 18.0 (SPSS Inc., Chicago, Illinois), and JMP Pro 10.0 (SAS Institute, Cary, North Carolina). The study was analyzed on the basis of the initial group assigned to the patient (similar to intention-to-treat analysis) and p value of <0.05 was considered significant.
Results
Overall, 1,329 patients (mean age 75 ± 11 years, women 47%) developed GIB during January 2005 to December 2010. There were 653 patients (49%) who were restarted on warfarin after a median duration of 50 days (interquartile range 21 to 78). The median CHADS 2 and HAS-BLED scores were 3. Endoscopic evaluation was performed in 883 cases (66.4%). Chronic AF was present in 751 (52.6%), persistent AF in 345 (25.9%), and paroxysmal AF in 233 patients (17.5%). Caucasians, patients with concomitant upper and lower sources for GIB, diabetics, patients with renal disease, history of coronary artery disease, and history of falls were less likely to be restarted on warfarin (p <0.05; Table 1 ). Overall, there were greater co-morbidity burdens in the patients who were not restarted on warfarin. Major reasons for not restarting warfarin were physician preference (18%) and patient’s inability to follow up with the anticoagulation clinic (19%).
| Variable | Warfarin Restarted | p Value | |
|---|---|---|---|
| Yes (n = 653) | No (n = 676) | ||
| Age (yrs) | 74.8 ± 10.7 | 75.3 ± 10.7 | 0.43 |
| Men | 364 (55.7) | 336 (49.7) | 0.03 |
| White | 413 (63.2) | 437 (64.6) | 0.60 |
| Hyperlipidemia | 169 (25.9) | 196 (29.0) | 0.30 |
| Previous coronary artery disease | 54 (8.2) | 89 (13.1) | 0.005 |
| Smoker | 124 (19.0) | 158 (23.4) | 0.06 |
| Framingham risk score | 11.26 ± 8.13 | 12.20 ± 8.6 | 0.04 |
| History of falls | 69 (10.6) | 124 (18.3) | 0.0001 |
| Cancer | 52 (7.9) | 65 (9.6) | 0.33 |
| INR at the time of gastrointestinal bleed | 2.87 | 3.06 | 0.22 |
| CHADS 2 (median score) | 3 (2–4) | 3 (2–4) | 0.08 |
| Previous congestive heart failure | 156 (23.9) | 181 (26.7) | 0.23 |
| Hypertension | 575 (88.0) | 624 (95.6) | 0.06 |
| Diabetes mellitus | 211 (32.3) | 233 (34.5) | 0.42 |
| Thromboembolism | 190 (29.1) | 208 (30.7) | 0.1 |
| Peripheral vascular disease | 89 (13.6) | 96 (14.2) | 0.81 |
| HAS-BLED (median score) | 3 (1–4) | 3 (1–4) | 0.09 |
| Labile INR | 198 (30.3) | 218 (32.2) | 0.47 |
| Alcohol use | 43 (6.6) | 56 (8.2) | 0.25 |
| Abnormal renal function | 109 (16.7) | 146 (21.6) | 0.02 |
| Abnormal liver function | 25 (3.8) | 36 (4.4) | 0.23 |
| Previous major gastrointestinal bleed | 85 (13.0) | 99 (14.6) | 0.42 |
| TTR before gastrointestinal bleed | 57.5 | 58 | 0.65 |
| Systolic blood pressure (mm Hg) ∗ | 144.6 ± 28.9 | 145.8 ± 21.8 | 0.80 |
| Medications on admission | |||
| Aspirin | 154 (23.5) | 169 (25) | 0.56 |
| Nonsteroidal analgesic drug use | 33 (5.0) | 36 (4.9) | 0.90 |
| Antiplatelets | 15 (2.3) | 19 (2.8) | 0.60 |
| Proton pump inhibitor use | 79 (12.3) | 112 (16.5) | 0.02 |
| H 2 blocker use | 24 (3.7) | 33 (4.8) | 0.28 |
| Location of the bleed | |||
| Esophageal | 20 (3) | 23 (3.4) | 0.75 |
| Stomach and duodenum | 213 (32.6) | 216 (32.0) | 0.81 |
| Small intestine | 12 (1.8) | 12 (1.8) | 1.00 |
| Colon and rectum | 342 (52.3) | 359 (53.1) | 0.86 |
| Upper and lower | 15 (2.3) | 28 (4.1) | 0.06 |
| Obscure | 51 (7.8) | 38 (5.6) | 0.12 |
| Clinical presentation | |||
| Occult | 24 (3.7) | 44 (6.5) | 0.02 |
| Hematochezia | 96 (14.7) | 105 (15.5) | 0.70 |
| Melena | 213 (32.7) | 225 (33.3) | 0.82 |
| Hematemesis | 86 (12.4) | 103 (15.2) | 0.30 |
| Blood in stools | 244 (37.4) | 264 (39.0) | 0.53 |
| Diagnosis on endoscopic evaluation | 13 (2.0) | 15 (2.2) | 0.85 |
| Nasogastric aspiration | 31 (4.7) | 49 (7.2) | 0.06 |
| Anemia on presentation | 85 (13.0) | 93 (13.7) | 0.74 |
| Setting of presentation | |||
| Clinic | 47 (7.2) | 62 (9.2) | 0.20 |
| Emergency room | 112 (17.2) | 254 (37.6) | 0.0001 |
| Hospital | 494 (75.6) | 360 (53.2) | 0.0001 |
| Intensive care unit admission | 123 (18.8) | 139 (19.1) | 0.45 |
| Management | |||
| No blood transfusion | 235 (36.0) | 264 (39.1) | 0.27 |
| No fresh frozen plasma | 336 (51.5) | 377 (55.7) | 0.12 |
| No vitamin K | 474 (72.6) | 482 (71.3) | 0.68 |
| Length of stay (days) | 4.7 ± 4.6 | 4.5 ± 4.0 | 0.27 |
∗ Mean systolic blood pressure after resolution of GIB during the hospital stay.
In total, 463 deaths occurred over a period of 2 years. Restarting warfarin was independently associated with decreased mortality (adjusted hazard ratio 0.66, 95% confidence interval 0.55 to 0.80, p <0.0001) compared with warfarin cessation group. The survival analysis is shown in Figure 1 . There were 4 thromboembolism-related deaths all in the patients in whom warfarin was stopped, whereas, there was only 1 death attributable to massive gastrointestinal hemorrhage in the patients who resumed warfarin. There was no significant difference in survival by the duration of interruption of warfarin <30 days; however, as the interruption duration became longer, the survival approached toward the group that did not resume warfarin ( Figure 2 ).
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
