An observational study has found that in patients using low-dose aspirin after a first myocardial infarction (MI), use of proton-pump inhibitors (PPIs) was associated with increased risk of recurrent cardiovascular events.

Patients discharged after MI are normally prescribed low-dose aspirin, as this has been shown to reduce the risk of recurrent events. Due to the increased risk of gastro-intestinal bleeding associated with anti-platelet drugs, treatment with acid-suppressive drugs such as PPI is recommended for those with other factors that increase the risk of GI bleeding. Some evidence suggests that PPI may reduce the effect of low-dose aspirin, however it is conflicting, so the authors carried out a retrospective population-based study comparing outcomes in post-MI patients treated with aspirin according to their use of PPI. They obtained data from Danish healthcare and population registries to assemble a cohort of patients aged over 30 who were discharged from hospital after a first MI between 1997 and 2006 taking low-dose aspirin. They were classified according to PPI use and non-use: those taking clopidogrel were excluded in view of the separate concerns over the combination of clopidogrel and PPI. Users of H2-receptor antagonists (H2RA) were also identified to provide an active control group, and propensity scoring was used to match PPI users with similar controls. Primary outcome was the combined end point of death from cardiovascular causes and readmission to hospital for myocardial infarction or stroke.

There were 97,499 patients admitted with a first MI during the study period, of whom 19,925 filled a prescription for aspirin within 30 days of discharge and were eligible for the analysis. Just over a fifth (21.6%) filled a prescription for a PPI during the subsequent year, and 3.3% filled a prescription for a H2RA. PPI users tended to be older, were more likely to be female, had more medical treatment and had more co-morbidity than non-users. Over the follow-up period, 3,366 of the study population (16.9%) had a recurrent cardiovascular event. Compared to non-users, those who used a PPI had a higher risk of an event, with a hazard ratio (HR) of 1.46 (95% CI 1.33 to 1.61; P<0.001).  In the propensity score matched model patients were matched 1:1, and in this analysis the HR was 1.61 (95% CI 1.45 to 1.79; P<0.001). A sensitivity analysis showed no increase in risk related to use of H2RA (HR 1.04; 95% CI, 0.79 to 1.38; P=0.78), and calculation suggested that an unmeasured confounder existing in 20% of the population would have to increase the risk 4-fold to have the same effect.

The authors conclude that in patients treated with low dose aspirin after a first MI, use of PPI is associated with an increased risk of a recurrent cardiovascular event. They discuss potential mechanisms for the effect, suggesting alterations in aspirin bioavailability as a major possibility. They also discuss potential unmeasured confounders, including smoking status, body mass index, and lipid levels, but suggest that the comparison with H2RA users reduces the risk of unmeasured confounding. Finally, they suggest that further studies are needed to examine this possible interaction further.

[Editor's comment: suggestive, but definitely not conclusive. In any observational study, unmeasured confounding cannot be dismissed, and the reliability of propensity matching and estimates from sensitivity analyses are highly dependent on assumptions used. It is also arguable whether patients treated with H2RA are likely to be directly comparable to those treated with PPI. Further confirmation is needed - ideally from randomised controlled trials, although these are unlikely; meanwhile these results should not stop patients who need PPI from receiving them. Bandolier has argued that because of study limitations, relative risks of less than two should be viewed with caution, especially if from observational studies (Bandolier 2007; 14 (7): 1-3). ]