Prescribing proton pump inhibitors with clopidogrel

BMJ 2012; 345 doi: http://dx.doi.org/10.1136/bmj.e4558 (Published 10 July 2012) Cite this as: BMJ 2012;345:e4558
  1. Laura Ellyn Targownik, associate professor of medicine, section of gastroenterology
  1. 1Department of Internal Medicine, University of Manitoba, Winnipeg, MB, Canada, R3N 0V6
  1. targowni{at}cc.umanitoba.ca

Stronger evidence emerges that PPIs do not cause myocardial infarction in users of dual antiplatelet treatment

In a linked research paper (doi:10.1136/bmj.e4388), Douglas and colleagues analysed whether proton pump inhibitors (PPIs) influence the risk of adverse cardiovascular outcomes in users of aspirin and clopidogrel.1 The use of aspirin and clopidogrel, or dual antiplatelet treatment, is currently the standard of care for the prevention of coronary stent thrombosis.2 These drugs are also used in the management of other cardiovascular conditions.3 Although use of dual antiplatelet therapy is associated with an increased risk of gastrointestinal haemorrhage,4 this risk can be greatly reduced through the concomitant use of a PPI.5 However, clinicians have become worried about using PPIs with dual antiplatelet treatment after the discovery that PPIs may interfere with clopidogrel mediated platelet inhibition through inhibition of the CYP2C19 cytochrome pathway, which is necessary for activation of clopidogrel.6 In addition, the findings of observational studies suggest that PPIs are associated with an increased risk of adverse cardiac outcomes in users of dual antiplatelet treatment.7 However, given the inherent biases in observational datasets, it is unclear whether this association is causal. Because a randomised controlled trial of sufficient statistical power is unlikely to be performed to establish definitively whether PPIs interfere with the ability of clopidogrel to reduce the risk of cardiovascular events, new pharmacoepidemiological strategies that mitigate against the effects of any systematic biases must be sought.

Douglas and colleagues performed a series of complementary analyses of data from the United Kingdom General Practice Research Database (UKGPRD) to delineate whether PPIs influence the risk of adverse cardiovascular outcomes in users of dual antiplatelet treatment.1 The authors first used a standard Cox proportional hazards model to assess the risk of recurrent myocardial infarction in people who did and did not use PPIs who were receiving aspirin and clopidogrel. They found that use of PPIs was associated with a significant risk of myocardial infarction (hazard ratio 1.35, 95% confidence interval 1.26 to 1.45). However, people who are prescribed a PPI are probably at higher underlying risk of myocardial infarction than those who are not given a PPI. For example, obese people may be at higher risk of myocardial infarction and are also more likely to have gastro-oesophageal reflux disease and to take a PPI for this indication. Although a Cox proportional hazards model can control for known confounding variables that may be linked to both PPI use and myocardial infarction, it cannot adjust for the effects of factors that were not measured. In addition, it cannot adjust for the effects of unknown characteristics that may be associated with PPI use and that also increase the risk of myocardial infarction. Thus, residual confounding may lead to the detection of an association between PPI use and myocardial infarction even if no causal association exists.

The problem of residual confounding can be rectified by performing a within subject analysis, where subjects function as their own controls, because this eliminates many systematic differences between the two groups. When Douglas and colleagues performed such an analysis on this dataset they found no significant association between PPI use and myocardial infarction (0.75, 0.55 to 1.01) when the incidence rate of myocardial infarction in people taking aspirin and clopidogrel was compared between periods of concomitant PPI use and periods of non-use. The researchers performed similar analyses for other known CYP2C19 inhibitors and found no associations with myocardial infarction in the within person analysis despite finding a significant association between their use and myocardial infarction in the proportional hazards model. They concluded that once residual confounding is minimised no association exists between myocardial infarction and PPI use in people who use dual antiplatelet treatment. These results lend support for the use of PPIs and dual antiplatelet treatment to prevent gastrointestinal haemorrhage.

Although the within person analysis is a useful approach its validity hinges on the assumption that participants’ underlying risk of myocardial infarction does not vary over time and that they are no more or less likely to be prescribed PPIs if the risk of myocardial infarction increases. Although PPIs may be given if prodromal symptoms of myocardial infarction are mistaken for reflux or dyspepsia, this would lead to a stronger association between PPI use and myocardial infarction, which was not the case. The results of this analysis also agree with higher quality prospective analyses in which the use of PPIs was not associated with an increase in myocardial infarction in people who used dual antiplatelet drugs. Moreover, although studies have shown that activation of clopidogrel is reduced in patients with lower CYP4530 2C19 activity,8 current evidence suggests that there is no definite association between low CYP4530 2C19 activity and adverse cardiac outcomes.9 10

Because patients with cardiovascular disease are at an especially high risk for morbidity and mortality after an acute gastrointestinal haemorrhage,11 clinicians should strongly consider prescribing a PPI to all patients who use dual antiplatelet drugs, especially in the presence of additional risk factors for gastrointestinal complications, such as age over 60; concomitant use of non-steroidal anti-inflammatory drugs, other anticoagulants, or corticosteroids; and important medical comorbidities.


Cite this as: BMJ 2012;345:e4558


  • Research, doi:10.1136/bmj.e4388
  • Competing interests: The author has completed the Unified Competing Interest form at www.icmje.org/coi_disclosure.pdf (available on request from the corresponding author) and declares: no support from any organisation for the submitted work; no financial relationships with any organisations that might have an interest in the submitted work in the previous three years; LET is a member of the speaker’s board for Pfizer Canada; the advisory boards of Janssen Canada, Merck Canada, and Abbott Canada; and was a member of the advisory board of Astra Canada (up to 2009).

  • Provenance and peer review: Commissioned; not externally peer reviewed.