EHR-enabled data dive links stomach blockbusters to heart attacks

Stanford University researchers have dug into 16 million clinical records on 2.9 million patients in search of links between proton pump inhibitors (PPIs) and heart attacks. And the electronic health record (EHR)-enabled data dive has turned up reasons to probe the topic further, with the team linking products such as AstraZeneca's ($AZN) Nexium to a 16% to 20% increase in the risk of heart attack.

EHRs were central to the project. Stanford has kept electronic versions of inpatient and outpatient clinical notes since 1994, giving it a repository of data on 1.8 million patients, 19 million encounters and 35 million diagnoses to mine. The researchers combined this resource with data collected by Practice Fusion, a provider of a free EHR system used primarily by small outpatient clinics. Digging into the data using a pharmacovigilance approach validated by other researchers revealed a correlation between use of PPIs and an elevated risk of heart attack. 

The researchers behind the project have stressed that their approach cannot act as judge, jury and executioner against the class of products, but it does provide a way to pick out safety signals that warrant further investigation. "I will say the association looks to be fairly compelling. What we have seen in this study is that PPI usage is very clearly associated with increased risk of heart attack," Dr. Nick Leeper, a Stanford cardiologist and senior author of the paper, told KQED. A randomized controlled trial comparing PPIs to a placebo would be needed to confirm causation.

Such studies are costly, time-consuming and in some cases unfeasible or unethical, making alternative data-enabled pharmacovigilance methods more practical, particularly for preliminary investigations. These characteristics have led the FDA to invest in its Sentinel Initiative and made EHRs an attractive choice when the Stanford team wanted to see if evidence linking PPIs to heart attacks in preclinical tests was also found in the real world. The approach does carry some significant caveats, though, which make it unwise to draw concrete conclusions from the analysis.

Notably, the team was unable to control for some potentially significant confounding variables, such as obesity and insulin resistance. It is also possible that patients taking PPIs were misdiagnosed, with doctors mistaking angina for acid reflux.

- read the PLOS paper
- here's KQED's article
- and FiercePharma's take