Dual Antiplatelet Therapy and Upper Gastrointestinal Bleeding Risk

Do PPIs Make Any Difference?

Inamul Haq; Fazal-ur-rehman Ali; Shakeel Ahmed; Steven Lindsay; Sudantha Bulugahapitiya


Br J Cardiol. 2013;20(4) 

In This Article


The mean age of patients was 66 years (range 24–96) with a median value of 68 years; 67% were males and 33% females, 74% Caucasians and 26% Asians. In the PPI group, 27 patients (29.7%) were admitted with STEMI and 64 patients (70.3%) with NSTEMI, while the corresponding values for the control group were 39 (45.3%) and 47 (54.7%), respectively. In the PPI group, 39 patients (42.9%) were treated with percutaneous coronary intervention (PCI), four (4.4%) with a coronary artery bypass graft (CABG) and 48 (52.7%) were managed conservatively, while the corresponding values for the control group were 11 (12.8%), five (5.8%) and 70 (81.4%), respectively. Baseline demographics are illustrated in Table 1.

Out of the 177 patients, evidence of upper gastrointestinal bleeding was found in 10 patients, six in the PPI group and four in the control group (odds ratio 1.45, 95% confidence interval [CI] 0.39–5.32, p=0.58). In the PPI group (six patients, 6.6%), the endoscopy findings were as follows: gastritis in four, bleeding angiodysplasia in one and bleeding oesophagitis in one. The results for the control group (four patients, 4.6%) showed gastritis in two, gastric ulcer in one and Mallory Weiss tear in one. None of these patients had a previous history of gastrointestinal bleeding. Table 2 explains these results in more detail.

Out of the six patients admitted with gastrointestinal bleeding, 4/50 (8%) were on lansoprazole, 1/31 (3.2%) on pantoprazole and 1/9 (11.1%) on omeprazole. Gastrointestinal bleeding was sufficiently severe to require a blood transfusion in two patients in the PPI group and one in the control group. In the PPI group, two patients required endoscopic intervention in the form of an adrenaline injection and diathermy, and only one patient was treated with heater probe coagulation and clips. In contrast, only one patient was treated with an adrenaline injection and haemostatic clips in the control group.

There was no statistically significant difference in gastrointestinal bleeding between the two groups, both unadjusted and adjusted for potentially confounding variables. Although there were no significant differences, the differences in outcome between groups were reduced after adjusting for differences in demographics between groups. Table 3 outlines the upper gastrointestinal bleeding episodes, both unadjusted and adjusted for variables.

There was no gastrointestinal bleeding related mortality in any of the study patients. Overall, 12 deaths were observed, seven in the PPI group (four ACS, one pulmonary embolism, one pneumonia, one heart failure) and five in the control group (three ACS, one metastatic disease with unknown primary, one pneumonia).