Non-steroidal Anti-inflammatory Drugs Use and Risk of Upper Gastrointestinal Adverse Events in Cirrhotic Patients

Yen-Chieh Lee; Chia-Hsuin Chang; Jou-Wei Lin; Hsi-Chieh Chen; Min-Shung Lin; Mei-Shu Lai


Liver International. 2012;32(5):859-866. 

In This Article


A total of 1 17 295 adult patients hospitalized for upper GI adverse events in 2006 were identified. Among them, 10 907 subjects fulfilled the criteria for cirrhosis. After further excluding patients with any diagnosis of cancer or coagulopathy before index hospitalization, concomitant diagnosis of trauma or injury, and those who were admitted for any reason 120 days before index hospitalization, a total of 4876 patients were included in the final analysis (Fig. 1). Most of the patients were male (74%), with mean age of 54.8 years. Approximately 38% had chronic viral hepatitis as the potential cause of cirrhosis. Characteristics, co-morbidities and co-medications use among these patients at admission is summarized in Table 1.

Figure 1.

Flow of the study subjects' enrollment.

Crude and adjusted ORs regarding NSAIDs use and upper GI adverse events are shown in Table 2 using conditional logistic regression analysis. We found a modest decrease in risk estimates after controlling for other individual NSAIDs and potential time-varying confounding variables in the multivariable analysis. Use of oral celecoxib was associated with an odds ratio (95% CI) of 1.44 (0.89–2.31) for upper GI adverse event, as compared with 1.87 (1.66–2.11) for oral nsNSAIDs and 1.90 (1.55–2.32) for parenteral NSAIDs overall. The adjusted ORs were 1.40 (1.25–1.56) for acetaminophen use and 1.25 (0.63–2.47) for anti-glaucoma medication use respectively. Additional control of acetaminophen use in the model did not change risk estimates for NSAIDs substantially; the ORs were 1.33 (0.83–2.11) for celecoxib, 1.50 (1.25–1.81) for oral nsNSAIDs and 1.90 (1.55–2.32) for parenteral NSAIDs. With regards to the risk of individual nsNSAIDs, adjusted OR ranged from 2.22 (1.48–3.32) for oral acemetacin, 1.92 (1.52–2.42) for parenteral ketorolac, to around 1.5 for oral ibuprofen and mefenamic acid. No significant difference in risk was found for frequent (≥0.5 DDD/day) and intermittent (<0.5 DDD/day) users of acematacin, diclofenac, ibuprofen and mefenamic acid. There were insufficient numbers of users of other NSAIDs to allow for frequency-response analysis.

Among included patients, 3307 patients were hospitalized for gastroesophageal variceal bleeding and 1569 patients for non-variceal events (oesophageal, gastric, duodenal and peptic ulcer and bleeding, gastritis and duodenitis). Table S2 presents the results after further dividing upper GI adverse outcomes into variceal and non-variceal events. The risks associated with celecoxib, nsNSAIDs overall, and individual NSAIDs were similar for variceal and non-variceal adverse outcomes. Due to small sample sizes after stratification, some risk estimates did not reach statistical significance.

We did not find the effects of celecoxib or nsNSAIDs on upper GI adverse events were modified by age, gender, prior GI events and presence of osteoarthritis among cirrhotic patients (Table 3). The adjusted OR of oral nsNSAIDs in aspirin users was 3.09 (1.27–7.56), whereas it was 2.05 (1.83–2.29) in those who did not take aspirin concomitantly. The risks of oral nsNSAIDs in users of acid suppressive agents (histamine-2 receptor antagonists and proton pump inhibitors) were slightly lower than the risks in those who did not take these agents with overlapping confidence intervals. For celecoxib, risk of upper GI adverse events seemed to be substantially lower in those who take concomitant proton-pump inhibitors (adjusted OR: 094, 95% CI: 0.47–1.90). The potential protective effects of acid suppressive agents on the upper GI risks associated with celecoxib and nsNSAIDs were similar for both variceal and non-variceal events (Table 4).

In sensitivity analysis (Table S3), the positive association between NSAIDs use and upper GI adverse event was evident despite different definition of control period was used. Effects of parenteral NSAIDs were slightly decreased when using 91–120 days prior to index date as the control period.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.