A Model for Estimating the Risk of Gastrointestinal Bleeding for Patients on Oral Anticoagulants and Interacting Medications

Q: What are the key findings of this study?

The DIOAC-GIB risk model quantified drug-specific gastrointestinal bleeding risk (OR 1.16 to 20.67 for NSAIDs) and correlated significantly with HAS-BLED and DOAC scores (r = 0.5, p < 0.01).

Q: What is the clinical evidence from this study?

Study design: Cohort. Population: Patients receiving oral anticoagulants (warfarin or direct-acting oral anticoagulants) at risk for gastrointestinal bleeding (n=11168). Intervention: DIOAC-GIB (Drug Interactions in Oral AntiCoagulants for Risk of Gastrointestinal Bleeding) risk model and pharmacodynamic interacting medications (NSAIDs, SSRI/SNRIs, aspirin, antiplatelets, corticosteroids, PPIs) vs. HAS-BLED and DOAC bleeding risk scores (and oral anticoagulant monotherapy without interacting medications in meta-analytic model inputs). Primary outcome: Predicted risk of gastrointestinal bleeding (GIB) and criterion-related correlation of DIOAC-GIB score with HAS-BLED and DOAC scores (r = 0.50 (with HAS-BLED) and r = 0.49 (with DOAC score), p=<0.01).

Q: What does this research mean for the field?

The DIOAC-GIB risk model quantifies drug-specific gastrointestinal bleeding risk across drug interactions and correlates significantly with established HAS-BLED and DOAC scores (r = 0.5, p < 0.01). Novelty: ClaimNovelty.METHODOLOGICAL. Consensus alignment: ConsensusAlignment.NEUTRAL.

Q: What question did this study set out to answer?

To develop and validate the DIOAC-GIB risk model to predict gastrointestinal bleeding in patients receiving oral anticoagulants based on interacting medications and patient-specific risk factors.

Ainhoa Gómez-Lumbreras; Abdelrahaman G. Tawfik; Thomas S. Reese; Daniel M. Witt; Guilherme Del Fiol; Daniel C. Malone

doi:10.1002/phar.70205

October 25, 2026Pharmacotherapy The Journal of Human Pharmacology and Drug TherapyOpen Access

A Model for Estimating the Risk of Gastrointestinal Bleeding for Patients on Oral Anticoagulants and Interacting Medications

View Full Paper

Ask AI

Bookmark

Share

Key result

DIOAC-GIB risk model shows ~0.50 correlation with established scores for GI bleeding.

r = 0.50 (with HAS-BLED) and r = 0.49 (with DOAC score)
P<0.01
n=11,168

Why the study?

Pharmacodynamic drug-drug interactions in patients receiving oral anticoagulants may increase the risk of gastrointestinal bleeding, motivating the development and validation of a risk prediction model.

Does the medication-specific DIOAC-GIB risk model accurately estimate gastrointestinal bleeding risk and correlate with established bleeding risk scores (HAS-BLED and DOAC score) in adults receiving oral anticoagulants and interacting medications?

Key Points

To develop and validate the DIOAC-GIB risk model to predict gastrointestinal bleeding in patients receiving oral anticoagulants based on interacting medications and patient-specific risk factors.
Extracted baseline gastrointestinal bleeding rates for apixaban, rivaroxaban, dabigatran, edoxaban, and warfarin from pivotal clinical trials and published meta-analyses.
Obtained risk estimates using random-effects meta-analyses or existing literature for interacting co-medications (NSAIDs, SSRI/SNRIs, aspirin, antiplatelets, corticosteroids, and PPIs), age over 65 years, and bleeding history.
Assessed model performance by calculating correlation coefficients between DIOAC-GIB scores and existing bleeding risk scores (HAS-BLED and DOAC score).
Baseline bleeding risk was 1.16% for apixaban, 2.00% for warfarin, 3.00% for dabigatran, 3.15% for rivaroxaban, and 3.31% for edoxaban.
Gastrointestinal bleeding risk increased with age over 65 (relative risk = 2.50, 95% CI: 1.2–5.5), prior bleeding history (OR = 5.14, 95% CI: 3.92–6.73), antiplatelets (OR = 1.97, 95% CI: 1.73–2.25), corticosteroids (OR = 1.69, 95% CI: 1.53–1.87), SSRI/SNRIs (OR range: 1.31 to 1.50), and NSAIDs (OR range: 1.16 for celecoxib to 20.67 for ketorolac), while proton pump inhibitors reduced risk (OR = 0.67, 95% CI: 0.62–0.74).
DIOAC-GIB scores significantly correlated with both HAS-BLED (r = 0.50) and DOAC scores (r = 0.49; p < 0.01 for both).

Study Design

Type

Cohort (n=11,168)

Multicenter

No

Structured PICO

Does the medication-specific DIOAC-GIB risk model accurately estimate gastrointestinal bleeding risk and correlate with established bleeding risk scores (HAS-BLED and DOAC score) in adults receiving oral anticoagulants and interacting medications?

P

Population

11,168 adults (≥18 years, >62% aged ≥65 years) with an active prescription for warfarin (n=1,558) or a DOAC (n=9,610) evaluated over a 12-month period to validate the DIOAC-GIB gastrointestinal bleeding risk model.

E

Exposure

DIOAC-GIB (Drug Interactions in Oral AntiCoagulants for Risk of Gastrointestinal Bleeding) risk prediction model incorporating specific oral anticoagulants (warfarin, apixaban, rivaroxaban, dabigatran, edoxaban), individual NSAIDs, individual SSRI/SNRIs, aspirin, antiplatelet medications, corticosteroids, proton pump inhibitors (PPIs), age >65 years, and prior GIB history.

C

Comparator

Existing bleeding risk models: HAS-BLED score (evaluated in warfarin users) and DOAC score (evaluated in DOAC users); for individual risk factor meta-analyses, OAC monotherapy or placebo/unexposed control groups.

O

Outcome

Predicted probability/score of gastrointestinal bleeding (GIB) per 100 treated patients and Spearman's rank correlation (criterion-related validity) of the DIOAC-GIB model with HAS-BLED and DOAC scores.safety

Main Result

Effect estimate: r = 0.50 (with HAS-BLED) and r = 0.49 (with DOAC score)

Absolute Event Rate: 5.48% vs 4.95%

p-value: p=<0.01

Limitations

The model was calibrated to predict GIB based only on DOAC pivotal studies in atrial fibrillation, which may not reflect bleeding risk in all types of patients receiving DOACs.
Criterion-related validity was evaluated using only a single electronic health record data source (University of Utah), and uncontrolled hypertension had to be imputed.
The model focuses on pharmacodynamic interactions and does not include medications with pharmacokinetic interactions with oral anticoagulants.
There was considerable variation in the definition of gastrointestinal bleeding across the published literature and high heterogeneity (I² > 85%) in meta-analyses for history of GIB, antiplatelets, and corticosteroids.
The model does not specify individual risks for P2Y12 inhibitors, does not account for duration/time of exposure, and omits certain risk factors such as sex, alcohol use, and uncontrolled hypertension.
Calibrated using DOAC pivotal trials in atrial fibrillation, which may limit generalizability to other OAC indications
Validated in only a single EHR system (University of Utah) with random imputation required for uncontrolled hypertension (10% of hypertensive patients)
Omits pharmacokinetic drug-drug interactions, individual P2Y12 agent differentiation, and exposure duration/time-to-event effects
Considerable variation in GIB definitions across included studies and high heterogeneity (I² > 85%) in meta-analyses for history of GIB, antiplatelet medications, and corticosteroids
Does not include non-modifiable or hard-to-capture risk factors such as sex, alcohol use, or uncontrolled hypertension

Implication

The DIOAC-GIB model provides agent-specific gastrointestinal bleeding risk estimates for patients taking oral anticoagulants and interacting medications, correlating with HAS-BLED and DOAC scores while enabling modifiable, shared prescribing decisions.

Abstract

Introduction Pharmacodynamic drug–drug interactions (DDI) in patients receiving oral anticoagulants (OAC) may increase the risk of gastrointestinal bleeding (GIB). The purpose of this study was to develop and validate a GIB model for patients receiving OACs. Methods A risk model, D rug I nteractions in O ral A nti C oagulants for Risk of G astro i ntestinal B leeding (DIOAC‐GIB), was developed to predict GIB based on interacting medications and other patient‐specific risk factors for patients on OAC. Inputs for the model were obtained using random‐effect meta‐analysis or from the literature for the following attributes: exposure to nonsteroidal anti‐inflammatory drugs (NSAIDs); selective serotonin reuptake inhibitors/serotonin‐norepinephrine reuptake inhibitors (SSRI/SNRIs); aspirin, antiplatelet medications, and proton pump inhibitors (PPIs); age; and GIB history. The model was correlated to existing OAC models (HAS‐BLED and DOAC score). Results GIB risk for OACs was based on pivotal trials (1.16% for apixaban; 3.15% for rivaroxaban; 3.00% for dabigatran; 3.31% for edoxaban; and 2.00% for warfarin). The odds ratio (OR) of GIB for NSAIDs ranged from 1.16 (95% confidence interval [CI]: 0.84–1.61) for celecoxib to 20.67 (95% CI: 14.56–29.34) for ketorolac. ORs for GIB among SSRI/SNRIs ranged from 1.31 (95% CI: 1.07–1.62) for paroxetine to 1.50 (95% CI: 1.32–1.70) for venlafaxine. Risk of GIB also included persons > 65 years of age (relative risk = 2.50, 95% CI: 1.2–5.5); history of GIB (OR = 5.14, 95% CI: 3.92–6.73); exposure to aspirin (OR = 1.31, 95% CI: 0.93–1.85), antiplatelet medications (OR = 1.97, 95% CI: 1.73–2.25), and corticosteroids (OR = 1.69, 95% CI: 1.53–1.87); and PPIs reduced GIB risk (OR = 0.67, 95% CI: 0.62–0.74). Correlations between the DIOAC‐GIB and HAS‐BLED and DOAC scores were 0.5 for both ( p < 0.01). Conclusion DIOAC‐GIB is a novel GIB‐risk model that incorporates pharmacodynamic OAC DDIs to assist clinicians and patients to determine a therapeutic plan while accounting for GIB risk. DIOAC‐GIB correlates with other bleeding risk scores and may help identify high‐risk patients who may benefit from tailored prescribing.

Discussion

No takes yet. Share an insight, caveat, or question.

Cite This Study

Gómez-Lumbreras et al. (2026) conducted a cohort in Patients receiving oral anticoagulants (warfarin or direct-acting oral anticoagulants) at risk for gastrointestinal bleeding (n=11,168). DIOAC-GIB (Drug Interactions in Oral AntiCoagulants for Risk of Gastrointestinal Bleeding) risk model and pharmacodynamic interacting medications (NSAIDs, SSRI/SNRIs, aspirin, antiplatelets, corticosteroids, PPIs) vs. HAS-BLED and DOAC bleeding risk scores (and oral anticoagulant monotherapy without interacting medications in meta-analytic model inputs) was evaluated on Predicted risk of gastrointestinal bleeding (GIB) and criterion-related correlation of DIOAC-GIB score with HAS-BLED and DOAC scores (r = 0.50 (with HAS-BLED) and r = 0.49 (with DOAC score), p=<0.01). The DIOAC-GIB risk model quantified drug-specific gastrointestinal bleeding risk (OR 1.16 to 20.67 for NSAIDs) and correlated significantly with HAS-BLED and DOAC scores (r = 0.5, p < 0.01).

synapsesocial.com/papers/6adf55e1953ea0d7a3d1da2c https://doi.org/10.1002/phar.70205

View Full Paper

Ask AI

Bookmark

Share

A Model for Estimating the Risk of Gastrointestinal Bleeding for Patients on Oral Anticoagulants and Interacting Medications

Key Points

Study Design

Structured PICO

Main Result

Limitations

Implication

Abstract

Discussion

Cite This Study

A Model for Estimating the Risk of Gastrointestinal Bleeding for Patients on Oral Anticoagulants and Interacting Medications

Key Points

Study Design

Structured PICO

Main Result

Limitations

Implication

Abstract

Discussion

Cite This Study

Keep reading

More related papers

More from Pharmacotherapy The Journal of Human Pharmacology and Drug Therapy

Keep reading

More related papers

More from Pharmacotherapy The Journal of Human Pharmacology and Drug Therapy