Risk of major adverse limb events in patients with type 2 diabetes mellitus receiving sodium glucose cotransporter 2 inhibitors and glucagon-like peptide-1 receptor agonists: A population-based retrospective cohort study.

Background: Both sodium glucose cotransporter 2 inhibitors (SGLT-2i) and glucagon-like peptide-1 receptor agonists (GLP-1RA) have cardiovascular protective effects in patients with type 2 diabetes mellitus. However, the comparative risk of GLP-1RA versus SGLT-2i for major adverse limb events remains unknown. Materials and methods: We studied a nationwide cohort involving 123,048 diabetes patients 20-100 years of age who initiated a SGLT-2i or GLP-1RA during 2012 and 2017. The patients in the two groups were matched by propensity score (PS), and incidence rates for hospitalization for major adverse limb events, critical limb ischemia (CLI) and lower extremity amputation (LEA), were assessed. Cox proportional hazards regression was applied to estimate hazard ratios (HRs) between patients receiving SGLT-2i as compared with GLP-1RA. The modification effects of age, a history of established cardiovascular disease, and chronic kidney disease were examined. In addition, use of dipeptidyl peptidase-4 inhibitor (DPP-4i) was chosen as a second active comparator. Results: After PS-matching, a total of 13,378 SGLT-2i and 13,378 GLP-1RA initiators were identified. Use of SGLT-2i was not associated with an increased risk for hospitalization for CLI and LEA, either compared with GLP-1RA (HR, 1.13; 95% CI, 0.77-1.65 and 1.27; 95% CI, 0.63-2.55, respectively) or compared with DPP-4i use (HR, 1.06; 95% CI, 0.75-1.50 and HR, 0.80; 95% CI, 0.42-1.53, respectively). Although the study was underpowered to explore potential effect modification, a trend of higher risks for LEA was noted among SGLT-2i users with cardiovascular disease as compared with either GLP-1RA or DPP-4i. Conclusion: Use of SGLT-2i was not associated with higher risks for hospitalization for CLI and LEA as compared with reference drugs. Further large-scale studies are needed for a precise risk estimation.

Incidence of hospitalization for infection among patients with hepatitis B or C virus infection without cirrhosis in Taiwan: A cohort study.

BACKGROUND: Infection is a major complication in liver cirrhosis and causes major morbidity and mortality. However, the incidence and mortality related to these conditions in patients infected with hepatitis C virus (HCV) are unclear, as is whether antiviral therapy could change their infection risk. METHODS AND FINDINGS: In this community-based cohort study, a total of 115,336 adults (mean age 52.2 years; 35.6% men) without cirrhosis participating in the New Taipei City Health Screening in 2005-2008 were classified as having noncirrhotic HCV (NC-HCV) (n = 2,839), noncirrhotic hepatitis B virus (NC-HBV) (n = 8,316), or no HBV or HCV infection (NBNC) (n = 104,181). Participants were followed to their first hospitalization for infection or death after data linkage with the Taiwan National Health Insurance Research Database (NHIRD) and Death Registry. A Cox proportional hazard regression model, adjusted for age, sex, body mass index (BMI), smoking, alcohol consumption, education level, diabetes, renal function, systemic steroids, and history of hospitalization, was used to calculate hazard ratios (HRs) and 95% confidence intervals (CIs) for overall and individual sites of infection and infection-related mortality. The reference group was NBNC participants with normal to mildly elevated alanine aminotransferase (ALT) (<1.5 times upper normal limit [UNL]) levels. To further address the impact of antiviral treatment on infection risk, we conducted analyses of data from the nationwide NHIRD and compared the risks for hospitalization because of infections and infection-related deaths between patients with HCV who received antiviral therapy (n = 20,264) and those who remained untreated (n = 104,360). During a median 8.2-year follow-up, the incidence of hospitalization for infection was substantially higher in NC-HCV patients. Compared to the reference group, NC-HCV was associated with a significantly higher risk for hospitalization because of overall infections (adjusted HR: 1.22; 95% CI: 1.12-1.33), but we observed no increased risk for patients in the NC-HBV (adjusted HR: 0.94; 95% CI: 0.88-1.01) or NBNC group with moderate to markedly elevated ALT levels (adjusted HR: 1.03; 95% CI: 0.93-1.14). For specific sites of infection, the NC-HCV group had increased risks for septicemia and lower respiratory tract, reproductive, and urinary tract infections. We noted no increased risk for infection-related death among patients with NC-HCV. Patients with HCV who received antiviral therapy had significantly reduced infection-related hospitalization and death risks (adjusted HR: 0.79; 95% CI: 0.73-0.84 for infection-related hospitalization and adjusted HR: 0.08; 95% CI: 0.04-0.16 for infection-related deaths). Study limitations include the exclusion of patients with cirrhosis from the cohort, the possibility of unmeasured confounding, and the lack of information on direct-acting antiviral agents (DAAs). CONCLUSIONS: In this study, patients with NC-HCV were at increased risk for hospitalization for infection, while no increased risk was observed for NC-HBV-infected patients.

Comparing the risks of hospitalized heart failure associated with glinide, sulfonylurea, and acarbose use in type 2 diabetes: A nationwide study.

BACKGROUND: Increasing evidence suggests that certain newer anti-diabetic drugs are associated with an increased risk of hospitalized heart failure (HHF). However, the potential risks associated with the use of sulfonylurea and glinide have not been carefully evaluated. METHODS: A retrospective cohort study using the Taiwan National Health Insurance claims database was conducted to examine the risks of HHF among newly diagnosed type 2 diabetic patients who initiated glinide, sulfonylurea, or acarbose therapy during 2006-2012. The outcome of interest was hospitalization due to heart failure after treatment initiation, defined by ICD-9-CM code. A Cox proportional hazard regression model was used to calculate the hazard ratio (HR) and 95% confidence interval (CI) using acarbose as the reference group. RESULTS: A total of 25,638 glinide, 272,140 sulfonylurea, and 29,376 acarbose initiators were included in the analysis. Patients who initiated glinide had the highest crude HHF incidence. In the analysis adjusted for baseline differences, a significantly higher risk of HHF was found for glinide (adjusted HR, 1.53; 95% CI, 1.24-1.88), but not for sulfonylurea (adjusted HR, 0.94; 95% CI, 0.80-1.11), as compared with acarbose. The elevated risk remained consistent across different subgroups of patients as well as several sensitivity analyses including exploring the impact of potential unmeasured confounding. CONCLUSIONS: These findings indicated that, as compared with acarbose, glinide may be associated with a higher risk of HHF for type 2 diabetic patients. Further researchis needed to fully evaluate the risks and benefits of glinide therapy relative to other oral anti-diabetic agents.

Comparison of the effects of serotonin-norepinephrine reuptake inhibitors versus selective serotonin reuptake inhibitors on cerebrovascular events.

BACKGROUND: Use of selective serotonin reuptake inhibitors (SSRIs) has been associated with an increased risk of intracranial hemorrhage. However, little is known about cerebrovascular risk in users of serotonin-norepinephrine reuptake inhibitors (SNRIs). Our aim was to determine the differential risk of cerebrovascular events between SSRIs and SNRIs. METHOD: A nationwide population-based cohort study was conducted in adult patients who started taking SSRIs or SNRIs during the time period 2005 through 2009. The outcome of interest was defined by the first hospitalization diagnosis for ischemic stroke (ICD-9-CM codes 433, 434, 436) or intracranial hemorrhage (ICD-9-CM codes 430, 431, 432). We used a Cox regression model with time-varying medication use and adjusted for stroke risk factors to estimate the hazard ratios (HRs) of ischemic stroke and intracranial hemorrhage associated with SNRI use, using SSRI use as a reference. RESULTS: Among 582,650 SSRI and 76,920 SNRI initiators with an average follow-up period of 3.2 years, there was a nonsignificantly increased trend toward intracranial hemorrhage (adjusted HR = 1.24 [95% CI, 0.97-1.58]) in SNRI users compared to SSRI users. The risk of ischemic stroke was comparable between the 2 treatment groups (adjusted HR = 1.01 [0.90-1.12]). Similar results were obtained in sensitivity analyses, considering a dose-response relation, allowance of a 7-day grace period between study drug discontinuation and outcome occurrence, and restriction to exclusive users, who remained on the initial treatment. In the subgroup analysis, there was an increased incidence of intracranial hemorrhages in SNRI users compared to SSRI users in patients without prior depression (adjusted HR = 1.63 [1.14-2.32]). CONCLUSIONS: Use of SNRIs is not associated with an increased risk of either ischemic stroke or intracranial hemorrhage as compared to use of SSRIs in adult patients with depression or anxiety. However, SNRIs should be used cautiously in patients without depression.

Risk of Intracranial Hemorrhage From Statin Use in Asians: A Nationwide Cohort Study.

BACKGROUND: Reports of statin usage and increased risk of intracranial hemorrhage (ICH) have been inconsistent. This study examined potential associations between statin usage and the risk of ICH in subjects without a previous history of stroke. METHODS AND RESULTS: Patients initiating statin therapy between 2005 and 2009 without a previous history of ischemic or hemorrhagic stroke were identified from Taiwan's National Health Insurance database. Participants were stratified by advanced age (≥70 years), sex, and diagnosed hypertension. The outcome of interest was hospital admission for ICH (International Classification of Diseases, Ninth Revision, Clinical Modification codes 430, 431, 432). Cox regression models were applied to estimate the hazard ratio of ICH. The cumulative statin dosage stratified by quartile and adjusted for baseline disease risk score served as the primary variable using the lowest quartile of cumulative dosage as a reference. There were 1 096 547 statin initiators with an average follow-up of 3.3 years. The adjusted hazard ratio for ICH between the highest and the lowest quartile was nonsignificant at 1.06 with a 95% confidence interval spanning 1.00 (0.94-1.19). Similar nonsignificant results were found in sensitivity analyses using different outcome definitions or model adjustments, reinforcing the robustness of the study findings. Subgroup analysis identified an excess of ICH frequency in patients without diagnosed hypertension (adjusted hazard ratio 1.36 [1.11-1.67]). CONCLUSIONS: In general, no association was observed between cumulative statin use and the risk of ICH among subjects without a previous history of stroke. An increased risk was identified among the nonhypertensive cohort, but this finding should be interpreted with caution.

Severe hepatic injury associated with different statins in patients with chronic liver disease: a nationwide population-based cohort study.

BACKGROUND AND AIM: The hepatotoxicity of statins in patients with chronic liver diseases remains unclear. In this study, we aimed to estimate the risk of severe hepatic injury associated with different statins in patients with chronic liver disease. METHODS: A nationwide population-based cohort study was conducted by analyzing the Taiwan National Health Insurance database. A total of 37,929 subjects with chronic liver disease who started statin therapy were identified during the period of January 1, 2005 to December 31, 2009. Outcome was defined as hospitalization due to liver injury. RESULTS: During a total of 118,772 person-years of follow-up, 912 incident cases of hospitalization due to hepatic injury are identified. The incidence rate was 2.95, 2.49, 2.92, 1.94, 2.65, and 2.52 per 100,000 person-days for atorvastatin, lovastatin, fluvastatin, pravastatin, simvastatin, and rosuvastatin initiators, respectively. Overall, there was no difference in the incidence associated with different statins. However, when each statin was further categorized to high (≧ 0.5 defined daily dose) or low (< 0.5 defined daily dose) mean daily dose, only high-dose atorvastatin was significantly associated with increased risk of hospitalization due to hepatic injury (hazard ratio, 1.62; 95% confidence interval, 1.29, 2.03) as compared with low-dose atorvastatin. CONCLUSION: The overall incidence of hospitalization due to severe hepatic injury was low among statin initiators with chronic liver disease. Only high-dose atorvastatin was associated with increased risk.

Comparing individual angiotensin-converting enzyme inhibitors with losartan in the risk of hospitalization for pneumonia and related mortality: a nationwide cohort study.

BACKGROUND: Angiotensin-converting enzyme (ACE) inhibitors might decrease the risk of pneumonia, but head-to-head comparisons with angiotensin receptor blockers (ARBs) were seldom made. The objective of this study was to evaluate incidence of pneumonia and mortality for different ACE inhibitors as compared to losartan, an ARB that has similar indications. METHODS: Adult patients aged more than 20 years who initiated ACE inhibitors and losartan between 1 January 2004 and 31 December 2009 were identified from Taiwan's National Health Insurance Database. The outcomes of interest were hospitalization for pneumonia and pneumonia-related mortality. Participants were followed from treatment initiation to the earliest of outcome occurrence, death or disenrollment, treatment discontinuation, switching to other ACE inhibitors or ARBs, or study termination (31 December 31 2010). Proportional-hazards regression model was used to calculate the hazard ratios and their 95% confidence intervals (CIs), adjusted on baseline characteristics. RESULTS: A total of 1,192,082 ACE inhibitors and 175,668 losartan initiators were included. The risk of hospitalization for pneumonia was significantly higher for captopril (hazard ratio 1.94, 95% CI 1.82-2.07), enalapril (hazard ratio 1.14, 95% CI 1.07-1.22), fosinopril (hazard ratio 1.11, 95% CI 1.02-1.21), perindopril (hazard ratio 1.14, 95% CI 1.04-1.25), and ramipril (hazard ratio 1.11, 95% CI 1.02-1.22), as compared with losartan. Captopril was associated with a significantly increased risk of pneumonia mortality (hazard ratio 2.43, 95% CI 1.79-3.31). CONCLUSIONS: Treatment with ACE inhibitors is not associated with a lower risk of pneumonia incidence and mortality as compared with losartan.

Antihypertensive agents and risk of Parkinson's disease: a nationwide cohort study.

BACKGROUND AND PURPOSE: Hypertension has been associated with Parkinson's disease (PD), but data on antihypertensive drugs and PD are inconclusive. We aim to evaluate antihypertensive drugs for an association with PD in hypertensive patients. METHODS: Hypertensive patients who were free of PD, dementia and stroke were recruited from 2005-2006 using Taiwan National Health Insurance Database. We examined the association between the use of calcium channel blockers (CCBs), angiotensin converting enzyme inhibitors (ACEIs), angiotensin receptor blockers (ARBs) and the incidence of PD using beta-blockers as the reference. Cox regression model with time-varying medication use was applied. RESULTS: Among 65,001 hypertensive patients with a mean follow-up period of 4.6 years, use of dihydropyridine CCBs, but not non-dihydropyridine CCBs, was associated with a reduced risk of PD (adjusted hazard ratio [aHR]  = 0.71; 95% CI, 0.57-0.90). Additionally, use of central-acting CCBs, rather than peripheral-acting ones, was associated with a decreased risk of PD (aHR = .69 [55-0.87]. Further decreased association was observed for higher cumulative doses of felodipine (aHR = 0.54 [0.36-0.80]) and amlodipine (aHR = 0.60 [0.45-0.79]). There was no association between the use of ACEIs (aHR = 0.80 [0.64-1.00]) or ARBs (aHR = 0.86 [0.69-1.08]) with PD. A potentially decreased association was only found for higher cumulative use of ACEIs (HR = 0.52 [0.34-0.80]) and ARBs (HR = 0.52 [0.33-0.80]). CONCLUSIONS: Our study suggests centrally-acting dihydropyridine CCB use and high cumulative doses of ACEIs and ARBs may associate with a decreased incidence of PD in hypertensive patients. Further long-term follow-up studies are needed to confirm the potential beneficial effects of antihypertensive agents in PD.

Continuation of statin therapy and a decreased risk of atrial fibrillation/flutter in patients with and without chronic kidney disease.

BACKGROUND: To contain cost, Taiwan's previous National Health Insurance Reimbursement Policy requested that physicians discontinue their patients' statin therapy once the serum cholesterol had reached appropriate levels. This allowed us to evaluate the association between statin continuation and the occurrence of atrial fibrillation/flutter and whether it was modified by chronic kidney disease (CKD) status. METHODS: Patients who initiated statin therapy between January 1, 2001 and December 31, 2009 were identified from a random sample of one million subjects in the Taiwan National Health Insurance Research Database. The outcome was atrial fibrillation/flutter. A proportional hazard regression model with time-varying statin use was applied to estimate the hazard ratios (HR) and 95% confidence intervals (CIs) for atrial fibrillation/flutter according to current statin use versus treatment discontinuation, adjusted for baseline disease risk scores and time-varying covariates. RESULTS: A total of 6767 CKD and 63,678 non-CKD patients initiating statin therapy were included and followed for an average of 4.0 years. A total of 1118 participants experienced new-onset atrial fibrillation/flutter. The incidence of atrial fibrillation/flutter was approximately 2 fold higher in the CKD patients. Continuation of statin therapy was associated with a 22% (adjusted hazard ratio 0.78; 95% CI: 0.65-0.93) and 57% (adjusted HR 0.43; 95% CI: 0.27-0.68) decrease in atrial fibrillation/flutter hazard as compared with discontinuation in non-CKD and CKD patients, respectively. CONCLUSIONS: Continuation of statin therapy was associated with a decreased risk of atrial fibrillation/flutter among CKD and non-CKD patients. However, further randomized studies are still needed to assess the association.

Effects of selective serotonin reuptake inhibitors versus tricyclic antidepressants on cerebrovascular events: a nationwide population-based cohort study.

Depression is a common disorder worldwide and is strongly associated with stroke. Use of antidepressants could potentially decrease the risk of stroke in patients with depression. However, the role of selective serotonin reuptake inhibitors (SSRIs), the most frequently prescribed antidepressant in this era, in the risk of stroke showed inconsistent results. We aimed to assess the association between the use of different types of antidepressants, SSRIs and tricyclic antidepressants (TCAs), and the risk of cerebrovascular events in patients with depression or anxiety. A nationwide population-based cohort study was retrospectively conducted in patients with depression or anxiety who started to take SSRIs and TCAs identified from the Taiwan National Health Insurance claims database (2001-2009). We examined the association between the 2 types of antidepressants and incidence of stroke using a proportional hazard model adjusted for stroke risk factors. Among the 24,662 SSRI and 14,736 TCA initiators, the crude incidence rate for stroke was 10.03 and 13.77 per 100 person-years, respectively. Selective serotonin reuptake inhibitor use was not associated with risk of stroke as compared with TCAs in the time-fixed analysis. After adjusting for baseline propensity scores in the time-varying analysis, SSRI use significantly reduced risk of stroke as compared with TCAs with the adjusted hazard ratio of 0.67 (95% confidence interval, 0.47-0.96). The effect persisted even after considering the antidepressant dosage (hazard ratio, 0.65 [0.42 to 0.99]). In summary, use of SSRIs was associated with a reduced risk for stroke, as compared with TCAs, in this specific disease population.

Discontinuation of statin therapy associates with Parkinson disease: a population-based study.

OBJECTIVE: To evaluate the effect of discontinuing statin therapy on incidence of Parkinson disease (PD) in statin users. METHODS: Participants who were free of PD and initiated statin therapy were recruited between 2001 and 2008. We examined the association between discontinuing use of statins with different lipophilicity and the incidence of PD using the Cox regression model with time-varying statin use. RESULTS: Among the 43,810 statin initiators, the incidence rate for PD was 1.68 and 3.52 per 1,000,000 person-days for lipophilic and hydrophilic statins, respectively. Continuation of lipophilic statins was associated with a decreased risk of PD (hazard ratio [HR] 0.42 [95% confidence interval 0.27-0.64]) as compared with statin discontinuation, which was not modified by comorbidities or medications. There was no association between hydrophilic statins and occurrence of PD. Among lipophilic statins, a significant association was observed for simvastatin (HR 0.23 [0.07-0.73]) and atorvastatin (HR 0.33 [0.17-0.65]), especially in female users (HR 0.11 [0.02-0.80] for simvastatin; HR 0.24 [0.09-0.64] for atorvastatin). As for atorvastatin users, the beneficial effect was seen in the elderly subgroup (HR 0.42 [0.21-0.87]). However, long-term use of statins, either lipophilic or hydrophilic, was not significantly associated with PD in a dose/duration-response relation. CONCLUSIONS: Continuation of lipophilic statin therapy was associated with a decreased incidence of PD as compared to discontinuation in statin users, especially in subgroups of women and elderly. Long-term follow-up study is needed to clarify the potential beneficial role of lipophilic statins in PD.

Statins of high versus low cholesterol-lowering efficacy and the development of severe renal failure.

BACKGROUND: The objective of this nationwide retrospective cohort study was to examine the renal outcomes of HMG-CoA reductase inhibitor (statin) initiators. METHODS: The patients who started to take statins with high cholesterol-lowering efficacy (atorvastatin and rosuvastatin) and low efficacy (lovastatin, simvastatin, pravastatin, and fluvastatin) between 1 January 2001 and 31 December 2008 were identified from the Taiwan National Health Insurance claims database. The outcome of interest was severe renal failure, defined as the composite endpoint of hemodialysis, peritoneal dialysis, and kidney transplantation. A proportional hazard regression model was applied to estimate the incidence ratio between the two groups, adjusted for the propensity scores based upon baseline characteristics. RESULTS: Among of the 26,007 and 42,249 statin initiators, the crude incidence rate for developing severe renal failure was 0.65 and 0.46 per 100 person-years for the high-efficacy and low-efficacy groups, respectively. Despite that these two groups had comparable risk for myocardial infarction (hazard ratio: 1.06, 95%CI: 0.92­1.21), there was a 13% increased hazard for developing severe renal failure in the rosuvastatin and atorvastatin initiators (hazard ratio: 1.13, 95%CI: 1.02­1.26). The increased risk associated with these two statins was consistent across different risk groups (diabetes, chronic kidney disease, and ischemic heart disease). CONCLUSIONS: Statins with high cholesterol-lowering efficacy might increase the risk for developing severe renal failure. An alternative explanation is that the renal risk cannot be ameliorated as much as cardiovascular risk. Further follow-up studies or meta-analyses are needed to solve the controversy.

Antidepressant use and the risk of upper gastrointestinal bleeding in psychiatric patients: a nationwide cohort study in Taiwan.

The magnitude of risk between selective serotonin reuptake inhibitors and upper gastrointestinal bleeding (UGIB) is still unknown in patients with psychiatric diseases. The aim of this study was to quantify the risk of UGIB induced by use of antidepressants with different affinities for serotonin transporters in psychiatric patients using Taiwan's nationwide health insurance claims database. We conducted a propensity score- matched retrospective cohort study and identified 304,606 psychiatric patients who initiated antidepressant treatment during the 2005-2006 period. Antidepressants were classified as high- (HA group), intermediate- (IA group), or low-affinity (LA group) serotonin reuptake inhibitors. Patients in the LA group were matched 1:1 to those in the HA and IA groups according to their propensity scores. Subjects who were successfully matched were followed up from the date of antidepressant initiation to first hospitalization for UGIB, drug discontinuation, transition to or addition of antidepressants in another group, or the study's end (whichever occurred first). A total of 153,486 psychiatric patients were successfully matched, and 498 first UGIB events were identified. Compared with the LA group, patients in the HA group had a higher risk for UGIB (hazard ratio [HR], 1.38; 95% confidence interval [CI], 1.11-1.71). The HR (95% CI) of the IA group was 1.11 (95% CI, 0.88-1.41). The trend for elevated UGIB risk with increasing affinity of serotonin transporters was statistically significant (P < 0.01). Elderly patients and those with prior UGIB history were more susceptible to the harmful effects. Our findings suggest that the use of high-affinity serotonin reuptake inhibitors may increase the risk for UGIB in psychiatric patients.

Utilization of antidepressants in Taiwan: a nationwide population-based survey from 2000 to 2009.

OBJECTIVE: This study examined trends in antidepressant utilization in Taiwan between 2000 and 2009. METHODS: We collected data from the National Health Insurance Research Database of all incident antidepressant use. We described the incidence, prevalence, and therapeutic indications of antidepressant use. We tested the trends by using logistic regression analyses with adjustment for age and sex. RESULTS: Overall, the prevalence of antidepressant use per 1000 persons increased from 32.1 in 2000 to 46.3 in 2009; however, the incidence per 1000 persons declined from 20.8 in 2000 to 16.5 in 2009. For antidepressant classes, the prescriptions of selective serotonin reuptake inhibitors and other new agents increased during the study periods; however, the use of tricyclic antidepressants, trazodone, and monoamine oxidase inhibitors declined. For therapeutic indications, we found increasing trends of antidepressant use for sleep and adjustment disorders; however, the rates of antidepressant use for mood disorder, anxiety disorders, and non-psychiatric conditions decreased. CONCLUSIONS: Regarding the new use of antidepressants in Taiwan between 2000 and 2009, we found that the decreased incidence of antidepressant use might be attributed mainly to the decreased use of tricyclic antidepressants and monoamine oxidase inhibitors. Decreased use among middle-aged and older persons and fewer antidepressant prescriptions for non-psychiatric conditions were also noted.

Non-steroidal anti-inflammatory drugs use and risk of upper gastrointestinal adverse events in cirrhotic patients.

BACKGROUND/AIMS: The upper gastrointestinal (GI) toxicity associated with non-steroidal anti-inflammatory drugs (NSAID) use among cirrhotic patients remains unclear. The objective of this study was to evaluate the risk of upper GI adverse events associated with celecoxib and oral and parenteral non-selective NSAIDs in cirrhotic patients. METHODS: All the patients aged ≥ 20 years with a diagnosis of cirrhosis hospitalized for variceal bleeding and non-variceal upper GI adverse events (oesophageal, gastric, duodenal ulcer, bleeding; gastritis and duodenitis) in 2006 were identified using ICD-9-CM diagnosis codes from inpatient claims from the Taiwan National Health Insurance Database. In the case-crossover study design, the case period was defined as 1-30 days and the control period as 31-60 days before the date of hospitalization. The information for individual NSAID use was obtained from the outpatient pharmacy prescription database. Adjusted self-matched odds ratios (OR) and their 95% confidence intervals (CI) were estimated with a conditional logistic regression model. RESULTS: A total of 4876 cirrhotic patients were included. The adjusted OR (95% CI) was 1.44 (0.89-2.31) for celecoxib, 1.87 (1.66-2.11) for oral non-selective NSAIDs and 1.90 (1.55-2.32) for parenteral NSAIDs overall. Risks were similar for variceal and non-variceal events. Concomitant use of proton pump inhibitors and histamine-2 receptor antagonists tended to decrease the upper GI toxicity associated with non-selective NSAIDs and celecoxib. CONCLUSION: The use of nsNSAIDs but not celecoxib was associated with a two-fold increased risk of variceal and non-variceal upper GI events among cirrhotic patients.