The Risk of Lower Gastrointestinal Bleeding in Low-dose Aspirin Users

W.-C. Chen; K.-H. Lin; Y.-T. Huang; T.-J. Tsai; W.-C. Sun; S.-K. Chuah; D.-C. Wu; P.-I. Hsu


Aliment Pharmacol Ther. 2017;45(12):1542-1550. 

In This Article

Materials and Methods

Study Design

The National Health Insurance Research Database (NHIRD) of Taiwan contains the detailed health care data of more than 23 million enrollees (more than 99% of the Taiwanese population). We compared low-dose ASA users with controls matched by age, gender, and enrollment time in a 1:5 ratio. This cohort was selected from 1 million randomly sampled subjects from the NHIRD and was representative of the whole Taiwanese population. Diagnosis was based on International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM) codes, which were validated in a previous study by Lanas et al. with an accuracy rate of approximately 95%.[10] The inclusion criteria were as followed: (1) new low-dose ASA (75–325 mg daily) users and controls matched by age, sex, and enrollment time between January 1, 2000 and December 31, 2006; (2) aged ≥20 years. The exclusion criteria were: (1) active GIB at enrollment; (2) malignant tumour of the gastrointestinal tract (ICD-9-CM codes: 150.xx, 151.xx, 152.xx, 153.xx, 154.xx); (3) disease associated with alcohol (291.xx, 303.xx, 305.xx, 571.0, 571.1, 571.2, 571.3); (4) inflammatory bowel disease (556.x, 555.x); (5) radiation gastroenteritis or colitis (558.1); (6) intestinal vascular insufficiency (557.xx) and (7) coagulopathy (286.xx) before low-dose ASA use.


Comorbidities were identified according to the receipt of an ICD-9-CM code once for hospitalisation and emergency room claims or three times for out-patient claims before enrollment in the study. The comorbidities included coronary artery disease (ICD-9-CM codes: 411.xx-414.xx), ischaemic stroke (433.xx, 434.xx), diabetes mellitus (250.xx), hypertension (401.xx-405.xx), dyslipidemia (272.0, 272.01, 272.3, 272.4), chronic kidney disease (585, 586, 588.8,588.9, 250.4, 274.1, 403.x1, 404.x2, 404.x3, and 440.1), liver cirrhosis (571.2, 571.5, and 571.6), chronic obstructive pulmonary disease (491.xx, 492.xx, 494.xx, and 496.xx), peptic ulcer disease without complication (531.30, 531.70, 531.90, 532.30, 532.70, 532.90, 533.30, 533.70, and 533.90) and history of peptic ulcer bleeding (531.0, 531.00, 531.01, 531.2x, 531.4x, 531.6x, 532.0, 532.00, 532.01, 532.2x, 532.4x, 532.6x, 533.00, 533.01, 533.2x, 533.4x, 533.6x, 534.0, 534.00, 534.01, 534.2x, 534.4x, and 534.6x). Data on medication prescriptions including clopidogrel, ticlopidine, warfarin, NSAIDs, COX-2 inhibitors, steroids, proton pump inhibitors (PPIs), histamine-2 receptor antagonists (H2RAs), nitrates, alendronate, selective serotonin reuptake inhibitors (SSRIs) and calcium channel blockers use, were retrieved from the database. Medication use was defined as hospitalisation or out-patient prescription of medication for at least 2 weeks prior to the month before the index LGIB in patients with LGIB and at least 2 weeks during the observation period in patients without LGIB.

End-point of the Study

The end-point of this study was LGIB development according to the top three diagnosis codes during hospitalisation or emergency room visits (ICD-9-CM codes: 562.02, 562.03, 562.12, 562.13, 569.86, 569.3, 569.85, 578.1 and 578.9).

Statistical Analysis

Demographic data on low-dose ASA users and controls were summarised as mean and standard deviations for continuous variables and numbers and percentages for categorical variables. Categorical data were compared using chi-squared or Fisher's exact tests. Continuous variables with normal distributions were compared using the independent Student's t test and continuous variables without normal distributions were compared using Mann-Whitney U test. A Kaplan-Meier model was constructed to examine the time from enrollment to the end-point. A log-rank test was performed to compare the end-point events between low-dose ASA users and controls. In addition, univariate and multi-variate Cox proportional hazard regression models were developed to evaluate the independent predictors of LGIB in low-dose ASA users and all enrollees with adjustments for age, gender, comorbidities including coronary artery disease, ischaemic stroke, diabetes, hypertension, chronic kidney disease, liver cirrhosis, chronic obstructive pulmonary disease, dyslipidemia, peptic ulcer disease, history of peptic ulcer bleeding, and concomitant use of clopidogrel, ticlopidine, NSAIDs, warfarin, COX-2 inhibitors, steroids, SSRIs, alendronate, PPIs, H2RAs, nitrates and calcium channel blockers. Significance was defined as a P<.05 for all two-tailed tests. The analyses were conducted using IBM spss Statistics (version 20; IBM Corporation, Armonk, NY, USA).

Ethical Aspects

This study was approved by the institutional review board of Kaohsiung Veterans General Hospital (VGHKS16-CT11–04) and was conducted according to the principles of the Declaration of Helsinki. No informed consent was given because the data were analysed anonymously from the NHIRD. The results were presented in part at the 46th annual meeting of The Gastroenterological Society of Taiwan.