Anticoagulants, essential in the treatment of different cardiovascular conditions, increase the risk of gastrointestinal bleeding. This risk varies, according to the type and administration route of the anticoagulant. National studies report bleeding rates between 0.5% and 4.7%, whereas worldwide rates vary between 4% and 17%. The present study characterizes the anticoagulated population with gastrointestinal bleeding, followed at a referral anticoagulation clinic, contributing to the regional epidemiology and clinical decision-making.
Materials and methodsA cross-sectional study was conducted on non-pregnant patients anticoagulated with warfarin, direct-acting oral anticoagulants, or parenteral anticoagulants, and who presented with gastrointestinal bleeding, within the time frame of January 2018 and June 2022.
ResultsThe study included 121 patients. The main indication for anticoagulation was atrial fibrillation (35%) and at least 30% had a previous bleeding episode. The upper gastrointestinal tract was the primary origin of bleeding (51%). There were no statistically significant differences in the therapeutic interventions between the different anticoagulant agents evaluated. A total of 9.1% of patients presented with rebleeding during hospitalization and in-hospital mortality was 2.5%.
ConclusionsOur findings showed that patients with gastrointestinal bleeding associated with anticoagulant use had no relevant differences regarding clinical characteristics, treatment regimen, and outcomes, according to the type of anticoagulant employed. The behavior of gastrointestinal bleeding in anticoagulated patients may be similar to that of non-anticoagulated patients. Further studies comparing these two patient groups are warranted.
Los anticoagulantes, fundamentales en el tratamiento de diversas condiciones cardiovasculares, aumentan el riesgo de sangrado gastrointestinal (GI). Este riesgo varía según el tipo y vía de administración del anticoagulante. Estudios nacionales reportan tasas de sangrado entre 0,5 % y 4,7 %, mientras que globalmente varían entre 4 % y 17 %. Este estudio caracteriza la población anticoagulada con sangrado GI en seguimiento en una clínica de anticoagulación de referencia, aportando a la epidemiología regional y la toma de decisiones clínicas.
Materiales y métodosEstudio de corte transversal en pacientes adultos no gestantes, anticoagulados con Warfarina, anticoagulantes orales directos o parenterales que presentaron sangrado GI entre enero del 2018 y junio del 2022.
ResultadosSe incluyeron 121 pacientes. La principal indicación de anticoagulación fue la fibrilación auricular (35 %) y al menos el 30 % tenían un episodio de sangrado previo. El principal origen de sangrado fue el tracto GI alto (51 %). No hubo diferencias estadísticamente significativas en las intervenciones terapéuticas entre los diferentes agentes anticoagulantes evaluados. El 9,1 % de los pacientes presentaron resangrado durante la hospitalización y la mortalidad hospitalaria fue del 2,5 %.
ConclusionesLos hallazgos de este estudio muestran que el sangrado gastrointestinal asociado al uso de agentes anticoagulantes no muestra diferencias relevantes en las características clínicas, régimen de tratamiento y desenlaces según el tipo de anticoagulante empleado. El comportamiento del sangrado digestivo en pacientes anticoagulados podría ser similar al de pacientes no anticoagulados; se requieren más estudios que comparen estos dos grupos de pacientes.
Vitamin K antagonists (VKAs), such as warfarin; direct-acting oral anticoagulants (DOACs), such as apixaban, dabigatran, edoxaban, and rivaroxaban; and heparins that include unfractionated heparin (UFH) and low-molecular-weight heparins (LMWHs), such as enoxaparin and dalteparin, are used in the comprehensive management of patients with atrial fibrillation (AF), deep vein thrombosis, and prosthetic heart valves.1,2 According to their administration route, they are divided into 2 groups: parenteral and oral.3 The parenteral anticoagulants include the heparins, whereas VKAs and DOACs are oral anticoagulants, and they are subdivided into Xa factor inhibitors and direct thrombin inhibitors.4
Anticoagulants, regardless of their administration route, may increase the risk of gastrointestinal bleeding of endoluminal origin, whether due to ulcers or diverticula, or after therapeutic endoscopic procedures.5
Bleeding risk has been widely studied in the literature, but there is no evidence indicating a significant difference in the risk of gastrointestinal bleeding between the different types of anticoagulants, including DOACs and VKAs. However, differences have been observed in other outcomes, such as general mortality, the risk of cerebrovascular events, and other clinical factors, which may influence the choice of the most adequate anticoagulant treatment for each patient.6,7
Concerning the parenteral anticoagulants, such as LMWHs, the risk of all-cause bleeding is 1–2%, and is greater with therapeutic than with prophylactic doses.8 On the other hand, there is a higher rate of incidence of major bleeding with UFH than with LMWHs.8 Evidence evaluating parenteral options in the context of gastrointestinal bleeding is scarce.
The primary goal of anticoagulation clinics is to optimize the safety and efficacy of anticoagulant treatment. The professionals at these clinics are in charge of determining the most adequate therapeutic regimen for each patient, which includes selecting an individualized dose, taking into account the factors of bleeding and thrombosis risk, possible pharmacologic risks, and the comorbidities present. In addition, they focus on patient follow-up and education, to ensure adequate anticoagulant management.9
At the international level, the reported incidence of gastrointestinal bleeding varies between 4 and 17%, depending on the type of anticoagulant therapy, patient age, and other factors to consider.10 At the national level, there are no studies that specifically evaluate the risk for or presentation of gastrointestinal bleeding in the anticoagulated population. Available data place the risk of major bleeding between 1.16 and 2.46% and non-major bleeding between 8.17 and 37.68%,9,11 keeping in mind that those values are taken from studies with different populations of interest, indications for anticoagulation, and therapies employed.
The present study aims to characterize the ambulatory population treated with anticoagulants who developed gastrointestinal bleeding, contributing to the regional epidemiology and clinical decision-making in this context.
Materials and methodsA cross-sectional study was designed that evaluated the clinical histories of adult patients who received at least one dose of oral or parenteral anticoagulant, were followed at the anticoagulation clinic of the Fundación Valle del Lili in Cali, Colombia, between January 2018 and June 2022, and who within that time frame had been seen at the emergency service for a bleeding episode (any ICD-10 code for gastrointestinal bleeding) or had presented with such an episode during hospitalization. No pregnant or breastfeeding patients were included in the study.
A total of 274 patients were identified, who were followed at the anticoagulation clinic and had an ICD-10 gastrointestinal bleeding diagnosis upon admission to the emergency room or during their hospitalization. After the selection criteria, 121 patients were included in the study (Fig. 1).
Three groups of anticoagulated patients were analyzed: the VKA group, whose indicated medication was warfarin; the DOAC group; and the parenteral anticoagulant group. Data on the diagnostic tests corresponding to bleeding episode onset were collected from the clinical histories, whether as the reason for emergency room consultation or as an event occurring during hospitalization of the patient for other causes. The information was saved on an institutional platform for capturing research data (BD Clinic), guaranteeing confidentiality of the data obtained from the clinical histories and the collection format, through patient codification. Only the research team had access to the information.
VariablesThe independent variables were demographic (sex, age), clinical (comorbidities, past medical histories, indication for anticoagulation, bleeding origin, Glasgow-Blatchford Bleeding Score [GBS]), treatment (medical management, endoscopic management, surgical or angiographic management), and clinical outcome (in-hospital mortality and rebleeding events) characteristics. Notably, the GBS is a tool in clinical decision-making regarding the timing of endoscopy, whose required clinical and laboratory variables were available in the clinical histories of the patients. The dependent variable/outcome was the type of anticoagulant.
For the present study, major bleeding was considered when at least one of the following criteria was met: fatal bleeding, bleeding from a critical organ (intracranial, intraspinal, intraocular, retroperitoneal, intra-articular, pericardial, intramuscular with compartment syndrome), and/or bleeding resulting in a drop in hemoglobin level ≥2 g/dL within a 24 h period or the need for transfusion of ≥2 units of packed red blood cells or their equivalent. Clinically relevant non-major bleeding was that which did not meet the criteria of major bleeding but required medical intervention, caused hospitalization or a non-programmed visit to a healthcare center, or interrupted anticoagulant treatment. Minor bleeding included any bleeding that did not meet the major bleeding or clinically relevant non-major bleeding criteria.
Regarding management, medical management was considered the systemic implementation of medications or blood products (e.g., proton pump inhibitors [PPIs], red blood cell transfusion, tranexamic acid, activated carbon, among others). Endoscopic management included the procedures of sclerotherapy, ligation, hemoclips, argon plasma, and electrocoagulation at the bleeding site. Angiographic management was performed for diagnostic and therapeutic purposes through vascular embolization, when possible, whereas surgical management was performed primarily to control the bleeding at the affected anatomic site when its location was uncertain, endoscopic access was not possible, or the patient’s clinical condition worsened or did not improve with another type of management. The treatments utilized were not mutually exclusive, given that a patient could have received more than one type of treatment.
Concerning outcomes, rebleeding referred to the recurrence of the bleeding episode in the gastrointestinal tract, after initially having achieved hemostasis or the bleeding had been stopped during the hospital stay, and mortality was the in-hospital patient death due to any cause.
Statistical analysisDescriptive statistics were performed, calculating percentages for the qualitative variables, and mean or median with their respective dispersion measurements (standard deviation or interquartile range) according to data distribution, for the quantitative variables. To determine the relation between the dependent variable (type of anticoagulant) and the independent variables, the qualitative variables were compared utilizing the chi-square test or Fisher’s exact test, whereas the quantitative variables were analyzed through the Mann-Whitney U test or Student’s t test. Statistical significance was set at a p < 0.05 and the data were processed using the STATA® v.14 statistics program.
ResultsThe clinical histories of 121 patients who met the selection criteria were evaluated. There was a higher percentage of women (54%) and median patient age was 69 years (IQR: 58−79). Median age in the warfarin group was 62 years (IQR: 56−72), 75 years (IQR: 67−82) in the DOAC group, and 66 years (IQR: 58−78) in the parenteral anticoagulant group, with a statistically significant difference (p = 0.005) (Table 1). The most frequent gastrointestinal comorbidities were previous gastrointestinal bleeding (32%), diverticular disease (13%), and vascular ectasia (13%). No statistically significant differences were found among those findings (Table 1).
Clinical and demographic characteristics and past medical histories of the patients.
| Characteristics | n | General | Warfarin | DOAC | Parenteral anticoagulant | p value* |
|---|---|---|---|---|---|---|
| n (%) | n (%) | n (%) | n (%) | |||
| Total | 121 | 28 (23.1) | 46 (38) | 47 (38.8) | ||
| Sex | 121 | 0.300 | ||||
| Female | 65 (54) | 12 (43) | 24 (52) | 29 (62) | ||
| Male | 56 (46) | 16 (57) | 22 (48) | 18 (38) | ||
| Agea | 121 | 69 (58−79) | 62 (56−72) | 75 (67−82) | 66 (58−78) | 0.005 |
| Gastrointestinal comorbidities | ||||||
| Previous gastrointestinal bleeding | 39 (32) | 8 (29) | 15 (33) | 16 (34) | 0.890 | |
| Diverticular disease | 16 (13) | 1 (3.7) | 9 (20) | 6 (13) | 0.200 | |
| Vascular ectasia | 16 (13) | 3 (11) | 4 (8.7) | 9 (19) | 0.400 | |
| Erosive gastritis | 11 (9.1) | 2 (7.1) | 4 (8.7) | 5 (11) | >0.900 | |
| Previous ulcer | 10 (8.3) | 4 (15) | 4 (8.7) | 2 (4.3) | 0.200 | |
| Gastroesophageal reflux disease | 7 (5.9) | 2 (7.1) | 3 (6.8) | 2 (4.3) | 0.800 | |
| Other comorbidities | ||||||
| High blood pressure | 66 (55) | 13 (48) | 33 (72) | 20 (43) | 0.013 | |
| Solid malignancy | 37 (31) | 3 (11) | 13 (29) | 21 (45) | 0.008 | |
| Diabetes | 27 (22) | 10 (36) | 10 (22) | 7 (15) | 0.110 | |
| Dyslipidemia | 27 (22) | 10 (36) | 12 (26) | 5 (11) | 0.031 | |
| Chronic kidney disease | 14 (12) | 6 (21) | 2 (4.0) | 6 (13) | 0.078 | |
| Liver disease | 14(12) | 2 (7.1) | 3 (6.5) | 9 (19) | 0.200 | |
| Autoimmune disease | 9 (7.5) | 5 (19) | 1 (2.2) | 3 (6.4) | 0.042 | |
| Hematologic malignancy | 2 (1.7) | 0 (0) | 1 (2.2) | 1 (2.1) | >0.900 | |
| Laboratory data | ||||||
| Hemoglobin (g/dL)a | 119 | 10 (13−7.8) | 10.8 (14.2−7.5) | 10.7 (13−8.1) | 9.2 (11−7.7) | 0.081 |
| Platelets (µL)a | 119 | 242,000 (328,000−187,000) | 212,500 (282,000−176,500) | 249,000 (319,500−185,500) | 275,000 (337,000−208,000) | 0.153 |
| Prothrombin time (seconds)a | 59 | 15 (22−13) | 28 (51−20) | 14 (15−13) | 14 (15−12) | <0.001 |
| Partial thromboplastin time (seconds)a | 55 | 36 (44−32) | 46 (52−38) | 35 (38−31) | 33 (36−29) | <0.001 |
DOAC: direct-acting oral anticoagulant.
Regarding other comorbidities, 55% of the patients had high blood pressure and frequency was higher in the patients in the DOAC group (72%) (p = 0.013). Thirty-one percent of the patients presented with solid malignancy, with a significantly higher prevalence in the parenteral anticoagulant group (p = 0.008). Dyslipidemia was present in 22% of the patients and gastrointestinal bleeding was more frequent in the DOAC group (p = 0.031). Lastly, 7.5% of the patients had an autoimmune disease, with a significantly higher frequency in the warfarin group (p = 0.042). There were no statistically significant differences in hemoglobin levels or platelet counts, whereas prothrombin time and partial thromboplastin time values were longer in the warfarin group (p < 0.001) (Table 1).
The primary indications for anticoagulation were AF (36.4%), followed by pulmonary embolism and hypercoagulability conditions (21.5% for each indication), and prosthetic heart valves (6.6%). DOACs were the most widely used medications in patients with AF, with a frequency of 58.7%. There was a greater prevalence of AF in the warfarin group, reaching 35.7%. Parenteral anticoagulants were the most widely used medications in the cases of pulmonary embolism, with a frequency of 36.2%. Those differences were statistically significant (p < 0.001). Antiplatelet agents were given to 17.7% of the patients and 24% of all study patients took a PPI at the onset of the gastrointestinal bleeding episode. In that last patient group, 60% were managed with parenteral anticoagulants, resulting in a statistically significant difference (p < 0.050) (Table 2).
Pharmacologic history and anticoagulation indication.
| Characteristics | n | General | Warfarin | DOAC | Parenteral anticoagulant | p value* |
|---|---|---|---|---|---|---|
| n (%) | n (%) | n (%) | n (%) | |||
| Anticoagulation indication | ||||||
| Total | 121 | 28 (23.1) | 46 (38) | 47 (38.8) | ||
| Anticoagulation indication | 121 | <0.001 | ||||
| Atrial fibrillation | 44 (36.4) | 10 (35.7) | 27 (58.7) | 7 (14.9) | ||
| Prosthetic heart valve | 8 (6.6) | 8 (28.6) | 0 (0) | 0 (0) | ||
| Pulmonary thromboembolism | 26 (21.5) | 0 (0) | 9 (19.6) | 17 (36.2) | ||
| Deep vein thrombosis | 17 (14) | 4 (14.3) | 3 (6.5) | 10 (21.3) | ||
| Hypercoagulability conditiona | 26 (21.5) | 6 (21.4) | 7 (15.2) | 13 (27.6) | ||
| Pharmacologic history | ||||||
| Ambulatory proton pump inhibitors | 121 | 29 (24) | 2 (7.1) | 9 (20) | 18 (38) | 0.005 |
| Acetylsalicylic acid (ASA) | 120 | 19 (16) | 3 (11) | 9 (20) | 7 (15) | 0.700 |
| Ambulatory steroids | 120 | 11 (9.2) | 3 (11) | 1 (2.2) | 7 (15) | 0.082 |
| Ambulatory NSAIDs | 119 | 7 (5.9) | 2 (7.1) | 2 (4.3) | 3 (6.7) | 0.800 |
| Other antiplatelet agents | 121 | 2 (1.7) | 0 (0) | 0 (0) | 2 (4.3) | 0.110 |
DOAC: direct-acting anticoagulant; NSAIDs: nonsteroidal anti-inflammatory drugs.
Fifty-one percent of the cases identified corresponded to upper gastrointestinal bleeding, with the stomach as the main focal point (35%), whereas bleeding in 40% of the patients originated in the colon. The warfarin group and parenteral anticoagulant group had a higher percentage of upper gastrointestinal bleeding, at 62 and 51%, respectively. In the DOAC group, 55% of the patients presented with lower gastrointestinal bleeding. No statistically significant differences were found regarding bleeding location and the anticoagulant received (p = 0.200).
The GBS was utilized for stratifying the bleeding risk and was calculated for 50 patients. The score was higher in the parenteral anticoagulant group, compared with the other agents, but was not statistically significant (p = 0.080) (Table 3).
Bleeding diagnosis and origin.
| Characteristics | n | General | Warfarin | DOAC | Parenteral anticoagulant | p value* |
|---|---|---|---|---|---|---|
| n (%) | n (%) | n (%) | n (%) | |||
| Total | 121 | 28 (23.1) | 46 (38) | 47 (38.8) | ||
| Present type of bleeding | 115 | 0.400 | ||||
| Upper gastrointestinal bleeding | 59 (51) | 16 (62) | 20 (45) | 23 (51) | ||
| Lower gastrointestinal bleeding | 54 (47) | 9 (35) | 24 (55) | 21 (47) | ||
| Upper and lower gastrointestinal bleeding | 2 (1.7) | 1 (3.8) | 0 (0) | 1 (2.2) | ||
| Bleeding site | 101 | 0.259 | ||||
| Esophagus | 4 (4.0) | 2 (9.5) | 1 (2.5) | 1 (2.5) | ||
| Stomach | 36 (35) | 10 (48) | 13 (32.5) | 13 (32.5) | ||
| Duodenum | 5 (5.0) | 1 (5.0) | 4 (10.0) | 0 (0) | ||
| Middle intestine | 4 (4.0) | 1 (4.8) | 1 (2.5) | 2 (5.0) | ||
| Colon | 40 (40) | 4 (19) | 16 (40) | 20 (50) | ||
| Not identifiable | 12 (12) | 3 (14) | 5 (12.5) | 4 (10) | ||
| Glasgow-Blatchford Bleeding Scorea | 50 | 9 (12−6) | 9 (15−8) | 7.5 (9.8−5.2) | 11 (12.5−7) | 0.083 |
DOAC: direct-acting oral anticoagulant.
Of the 81 patients (66.9%) who underwent upper gastrointestinal endoscopy, the main findings were antral erosive gastropathy (32%) and gastric ulcer (30.5%), whereas no positive findings were found in 9 patients. Of the 52 patients (43%) who underwent colonoscopy, the main findings were diverticula (23%) and hemorrhoids (19.2%), and no findings were identified in 7 patients. There were no statistically significant differences between the different anticoagulant agents and the upper gastrointestinal endoscopy and colonoscopy findings (p > 0.500 for the two groups) (Table 4). The patients in whom no lesions were identified in those procedures underwent additional evaluations, detecting vascular ectasia in 2 cases: one through capsule endoscopy in a patient from the DOAC group and one through computed tomography (CT) angiography, in a patient from the parenteral anticoagulant group. The origin of gastrointestinal bleeding could not be identified in 14 patients (11.6%).
Endoscopic findings and bleeding classification.
| Characteristics | n | General | Warfarin | DOAC | Parenteral anticoagulant | p value* |
|---|---|---|---|---|---|---|
| n (%) | n (%) | n (%) | n (%) | |||
| Total | 121 | 28 (23.1) | 46 (38) | 47 (38.8) | ||
| Endoscopic findings | 81 | 0.580 | ||||
| Antral erosive gastropathy | 23 (31.9) | 5 (27.8) | 7 (29.2) | 11 (39.3) | ||
| Gastric ulcer | 22 (30.5) | 8 (44.4) | 7 (29.2) | 7 (25) | ||
| Esophageal varices | 7 (9.7) | 2 (11) | 3 (12.5) | 2 (7.2) | ||
| Gastric carcinoma | 5 (6.9) | 0 (0) | 2 (8.3) | 3 (10.7) | ||
| Duodenal ulcer | 6 (8.4) | 2 (11) | 3 (12.5) | 1 (3.5) | ||
| Vascular ectasia | 4 (5.5) | 1 (5.5) | 0 (0) | 3 (10.7) | ||
| Gastric polyps | 3 (4.2) | 1 (5.5) | 2 (8.3) | 0 (0) | ||
| Erosive esophagitis | 2 (2.8) | 1 (5.5) | 0 (0) | 1 (3.5) | ||
| Colonoscopy findings | 52 | 0.650 | ||||
| Diverticula | 12 (23) | 4 (28.6) | 3 (15.8) | 5 (26.3) | ||
| Hemorrhoids | 10 (19.2) | 4 (28.6) | 2 (10.5) | 4 (21) | ||
| Ulcer | 6 (11.5) | 1 (7.1) | 3 (15.8) | 2 (10.5) | ||
| Vascular ectasia | 5 (9.6) | 0 (0) | 3 (15.8) | 2 (10.5) | ||
| Polyps | 5 (9.6) | 3 (21.4) | 1 (5.3) | 1 (5.3) | ||
| Masses or tumors | 5 (9.6) | 1 (7.1) | 2 (10.5) | 2 (10.5) | ||
| Colitis | 2 (3.9) | 1 (7.1) | 1 (5.3) | 0 (0) | ||
| Forrest classification | 25 | 0.400 | ||||
| I A | 1 (4) | 1 (11.1) | 0 (0) | 0 (0) | ||
| I B | 2 (8) | 0 (0) | 1 (11.1) | 1 (14.3) | ||
| II B | 4 (16) | 1 (11.1) | 2 (22.2) | 1 (14.3) | ||
| II C | 2 (8) | 2 (22.2) | 0 (0) | 0 (0) | ||
| III | 16 (64) | 5 (55.5) | 6 (66.6) | 5 (71.4) |
DOAC: direct-acting oral anticoagulant.
The Forrest classification was applied solely to bleeding due to gastric and duodenal ulcers. Of the 28 patients with said lesions, 3 were not classified. Sixty-four percent of the patients were classified as Forrest III, with no statistically significant differences (p = 0.400) (Table 4).
Of the study population, 19.8% required endoscopic management, and sclerotherapy (33%) and argon plasma (41%) were the main therapeutic methods employed. A total of 1.6% of the patients had a surgical intervention and 1.7% underwent angiography. The rest of the patients received only medical treatment, with a statistically significant difference in favor of the warfarin group, compared with the DOAC and parenteral anticoagulant groups (82, 54, and 69%, respectively; p = 0.044). Importantly, the significant difference in favor of the VKAs was due exclusively to the greater use of vitamin K in the study population (39%; p < 0.001). Thirty-four percent of the patients required packed red blood cell transfusion and 2.5% required platelets, with no statistically significant difference between the types of anticoagulants (p = 0.600 and p = 0.800, respectively) (Table 5).
Patient management and outcomes.
| Characteristics | General | Warfarin | DOAC | Parenteral anticoagulant | p value* |
|---|---|---|---|---|---|
| n (%) | n (%) | n (%) | n (%) | ||
| Total | 121 | 28 (23.1) | 46(38) | 47(38.8) | |
| Managementa | |||||
| Medical management | 79 (66) | 23 (82) | 25 (54) | 31 (69) | 0.044 |
| Proton pump inhibitors | 62 (51) | 16 (57) | 21 (46) | 25 (53) | 0.700 |
| Red blood cell transfusion | 41 (34) | 8 (29) | 15 (33) | 18 (39) | 0.600 |
| Vitamin K | 13 (11) | 11 (39) | 1 (2.2) | 1 (2.1) | <0.001 |
| Fresh frozen plasma transfusion | 7 (5.8) | 1 (3.6) | 2 (4.3) | 4 (8.7) | 0.700 |
| Platelets | 3 (2.5) | 0 (0) | 1 (2.2) | 2 (4.3) | 0.800 |
| Somatostatin | 3 (2.5) | 0 (0) | 2 (4.3) | 1 (2.1) | 0.600 |
| Tranexamic acid | 2 (1.7) | 1 (3.6) | 1 (2.2) | 0 (0) | 0.500 |
| Prothrombin complex concentrates | 1 (0.8) | 0 (0) | 0 (0) | 1 (2.1) | >0.900 |
| Protamine sulfate | 1 (0.8) | 0 (0) | 0 (0) | 1 (2.1) | 0.600 |
| Activated carbon | 1 (0.8) | 0 (0) | 1 (2.2) | 0 (0) | 0.600 |
| Endoscopic management | 24 (19.8) | 2 (7.1) | 12 (26) | 10 (21.2) | 0.300 |
| Argon plasma | 10 (41.6) | 0 (0) | 6 (15.8) | 4 (12.1) | |
| Sclerotherapy | 8 (33) | 1 (4.5) | 5 (13.1) | (6) | |
| Band ligation | 3 (12.5) | 0 (0) | 1 (2.6) | (6) | |
| Hemoclips | 2 (8.3) | 1 (4.5) | 0 (0) | 1 (3.1) | |
| Electrocoagulation | 1 (4.1) | 0 (0) | 0 (0) | 1 (3.1) | |
| Surgical management | 2 (1.6) | 0 (0) | 1 (2.1) | 1 (2.1) | 0.800 |
| Angiographic management | 2 (1.7) | 0 (0) | 1 (2.2) | 1 (2.1) | 0.400 |
| Rebleeding during hospitalization | 10 (9.1) | 2 (7.1) | 7 (15) | 1 (2.1) | 0.400 |
| In-hospital mortality | 3 (2.5) | 0 (0) | 1 (2.2) | 2 (4.3) | 0.800 |
DOAC: direct-acting oral anticoagulant.
Rebleeding occurred in 9.1% of the patients during their hospitalization. In-hospital mortality was 2.5%, corresponding to 2 patients in the parenteral anticoagulant group and one patient in the DOAC group (Table 5).
Discussion and conclusionThe use of anticoagulant medications has been considered a risk factor for presenting with gastrointestinal bleeding.10,12 The type of anticoagulant related to bleeding, its intensity, location, and outcome may be different in each patient.12 Our study evaluated the clinical, diagnostic, and therapeutic characteristics of a group of anticoagulated patients, who were followed at a Latin American referral hospital’s anticoagulation clinic and who presented with gastrointestinal bleeding.
The median patient age was similar to that described in other studies on anticoagulation.9,11 The age of the patients who received DOACs was significantly higher than that of the other therapeutic groups, which could be related to the safety profile of those agents in the geriatric population.13
Regarding comorbidities, the highest percentage of patients with high blood pressure was found in the DOAC group, and their main indication for anticoagulation was AF. This may be related to the risk factors for AF described in the literature, which primarily are advanced age and the prevalence of arterial hypertension.14,15 The parenteral anticoagulants were the most widely employed medications in patients with solid malignancy, possibly explained by the greater experience with and safety profile evidence on those agents in the oncologic population.15,16
In patients with VTE, there was a higher frequency of gastrointestinal bleeding associated with the use of parenteral anticoagulants. This could be explained by the fact that said agents are the recommended prophylactic strategy for thrombosis in patients hospitalized due to acute disease, such as VTE.17
Of the patients who presented with synchronous VTE and solid malignancy, LMWH was the anticoagulation regimen used, which is in line with international treatment recommendations.18
Concerning dyslipidemia, there was a higher percentage of gastrointestinal bleeding in the DOAC group, which could be due to the fact that treatment with DOACs in high-risk cardiovascular patients has been shown to offer a better prognosis.19 That treatment is associated with a reduced risk for major adverse cardiovascular events (MACE) in patients with VTE and also surpasses the efficacy of VKAs in such a setting.20
In patients with autoimmune diseases, there was a larger percentage of gastrointestinal bleeding in the VKA group. That finding may be related to the preference for those agents in diseases that favor prothrombotic states. Antiphospholipid syndrome (APS) stands out, in which VKA use has shown greater efficacy in preventing thrombotic events, especially in patients with triple antibody positivity.21 A clear indication for VKA use has not been demonstrated in the literature in other autoimmune diseases.
The most frequent gastrointestinal comorbidities were previous gastrointestinal bleeding, diverticular disease, vascular ectasia, erosive gastritis, and ulcers, with no significant differences between anticoagulant groups. Those finding concur with reports in the literature on risk factors for gastrointestinal bleeding in anticoagulated populations. A meta-analysis published in 2023 that analyzed 34 studies on anticoagulated populations identified the history of previous bleeding as a moderate risk factor for presenting with gastrointestinal bleeding (OR 3.26; 95% CI 1.86–5.73).22 Gastritis and ulcer disease are described as the most common causes of upper gastrointestinal bleeding in anticoagulated patients, whereas diverticular disease and malignancy are typical findings in patients with lower gastrointestinal bleeding.23
In patients with a pharmacologic history of PPI use, the parenteral anticoagulant group had the higher percentage of gastrointestinal bleeding. This contrasts with reports in the literature, which have evaluated the combination of PPIs with VKAs or DOACs, but with no clear evidence of interactions with parenteral anticoagulants.24,25 Said result could be related to particularities in the indications for PPI use and their distribution among the anticoagulant groups of our study, or to unidentified factors. Additional research is needed to clarify the relation between PPIs and parenteral anticoagulants, in the context of gastrointestinal bleeding. On the other hand, medications, such as acetylsalicylic acid (ASA), nonsteroidal anti-inflammatory drugs (NSAIDs), and steroids, showed no significant differences between the anticoagulant groups, although they have been reported in the literature as risk factors for gastrointestinal bleeding.26,27 This may be explained by the fact that there was a low frequency of use of those medications in our study population.
A larger percentage of upper gastrointestinal bleeding was found in the warfarin group, whereas lower gastrointestinal bleeding was more frequent in the DOAC group. That anatomic bleeding distribution according to anticoagulant type is consistent with reports in the literature. Cohort studies have shown that warfarin is associated with a higher rate of upper gastrointestinal bleeding,28 whereas lower gastrointestinal bleeding tends to be more common in patients with DOAC therapy. In particular, a higher frequency of bleeding of colorectal origin, associated with lesions such as angiodysplasia, has been described with dabigatran use.29
In relation to the GBS, all anticoagulant groups had median scores above 7. Upon analyzing the data, the patients with those scores underwent more invasive interventions, indicating the accuracy of said score for predicting the need for advanced measures.30,31
In our study, more than 30% of the patients with lower gastrointestinal bleeding had colonoscopy findings; the most common were diverticulosis and hemorrhoids. Hashash et al. consistently identified diverticulosis as the main cause of bleeding in anticoagulated patients (64.8%), followed by polyps (30.7%), and hemorrhoids (27.3%).32
Medical management was predominant in our study, which is consistent with that reported in the literature. Studies report that medical management is the most widely implemented strategy in the treatment of gastrointestinal bleeding in anticoagulated patients.33,34
Rebleeding during hospitalization was present in 9.1% of our patients, which is a lower percentage than that reported in previous studies with varying frequency (17.8%, 15%). The difference may be primarily due to follow-up times, given that in the literature, follow-up varied from 30 days to 2 years. Other factors could also influence said result, such as gastrointestinal lesion characteristics or the anticoagulation regimen.35–37
In our study, the mortality rate was below the 8.1% reported by Turcano et al.38 That difference could be explained by the fact that in our study, mortality was registered only during hospitalization due to a bleeding episode, whereas the abovementioned study conducted a follow-up of 30 days. There were no significant differences in mortality, according to the type of anticoagulant utilized, in either of the studies.
Study strengths and limitationsThe present study is one of the first descriptive analyses profiling a southwestern Colombian population of patients receiving oral or parenteral anticoagulants, with follow-up at a clinic specializing in anticoagulation, and who experienced gastrointestinal bleeding. At the national level, the information on this complication has been based primarily on case series. Regarding our results, both mortality and prevalence of rebleeding were evaluated during hospitalization corresponding to the episode of gastrointestinal bleeding analyzed. Even though all patients who met the selection criteria were included, upon categorizing the data according to the different anticoagulant types, the sample was divided into unbalanced groups, limiting the possibility of performing additional analyses. Given that the information was collected retrospectively, it was not possible to stratify the bleeding (major bleeding, clinically relevant non-major bleeding, and minor bleeding) because the medical records of the study subjects lacked certain indispensable data for making that classification.
In conclusion, we found that gastrointestinal bleeding associated with the use of anticoagulant agents showed no relevant differences in clinical characteristics, treatment regimen, and outcome, according to the type of anticoagulant employed.
Moreover, the behavior of gastrointestinal bleeding in anticoagulated patients may be similar to that of non-anticoagulated patients. Further studies are needed to compare those 2 groups of patients.
Ethical considerationsIn compliance with the 1993 resolution Nº 008430 of the Ministry of Health of Colombia, the development of this project was classified as low-risk research. It was approved by the Ethics Committee in Biomedical Research, in act No. 20, October 5, 2022.
Financial disclosureNo specific grants were received from public sector agencies, the business sector, or non-profit organizations in relation to this study.
The authors declare that there is no conflict of interest.
The authors wish to thank the Clínica de Anticoagulación of the Fundación Valle del Lili.
