We appreciate the comments of Alegre et al regarding our recent study. They rightly point to the fact that patients taking proton pump inhibitors (PPIs) are often sicker and have more co-morbidities than patients who are not, although in our study, traditional cardiovascular risk factors (e.g., hypertension, diabetes mellitus) were generally similar. An important exception was baseline hematocrit, which was significantly lower in the PPI group and may be a surrogate for patients with histories of peptic ulcer disease (PUD). Although we included this factor in our multivariate Cox regression for major adverse cardiac events, we did not have information regarding history of PUD, gastrointestinal bleeding, or gastroesophageal reflux disease (GERD).

Isolating PUD or GERD as an important cardiovascular risk factor, however, is problematic. Specifically, many patients with cardiovascular disease are receiving dual-antiplatelet therapy, which by default increases their risk for gastrointestinal bleeding. This does not prove, however, that PUD or GERD is an independent risk factor for cardiovascular mortality. In addition, the recent cohort study by Ray et al showing an increased prevalence of esophageal disease in PPI users is hardly surprising; one would expect higher rates of PUD and GERD in PPI users. Indeed, the fact that almost all patients with PUD or GERD are taking PPIs would, at least in our relatively small cohort, lead to a problem of collinearity.

We do agree, however, that the underuse of PPI prophylaxis in the setting of percutaneous coronary intervention may be hazardous. Our unpublished data show 30-day mortality of 20% in patients with gastrointestinal bleeding complicating percutaneous coronary intervention, so any strategy that effectively reduces this risk is sorely needed. The indiscriminate use of PPIs for prophylaxis, however, is concerning in light of our data and other retrospective data in concert with platelet reactivity studies. We therefore hold that the avoidance of strong cytochrome P450 2C19 inhibitors such as omeprazole and esomeprazole in patients taking clopidogrel, as recommended by the United States Food and Drug Administration, is therefore quite reasonable.