The study found that patients taking drugs known as proton pump inhibitors, such as Nexium, Prilosec and Prevacid, had a 16 percent to 20 percent higher risk of heart attack than those who took H2 blockers, a class that includes Zantac and Tagamet.

Share story

An analysis of millions of electronic health records raises new questions about the safety of certain acid-reflux medications taken by millions of Americans, suggesting they increase the risk of heart attacks.

Researchers from Houston Methodist and Stanford University used computer algorithms to scour some 16 million records from nearly 3 million patients to find which of them had taken medications for gastroesophageal reflux disease, or GERD, and how many of them had heart attacks.

The study found that patients taking drugs known as proton pump inhibitors, such as Nexium, Prilosec and Prevacid, had a 16 percent to 20 percent higher risk of heart attack than those who took H2 blockers, a class that includes Zantac and Tagamet.

The approach used by the researchers can show associations, but it does not prove the drugs caused the illness. The researchers said more conclusive studies are needed to confirm the findings.

They used a technique known as data mining, hailed as the next big advance in drug surveillance. It enables doctors to use electronic health records of millions of patients to investigate connections between drugs and side effects without having to enroll a single patient or deal with the financial and regulatory challenges of a clinical trial.

The study, published in the online journal PLOS ONE, adds to previous concerns about the drugs and supports the earlier findings of Dr. John Cooke, chair of the Department of Cardiovascular Sciences at the Houston Methodist Research Institute and senior author of the study.

Two years ago, Cooke published research using mouse models and human cultures showing that the acid-reflux drugs inhibit the ability of blood vessels to relax. Over time, this could lead to cardiovascular complications. Other studies have shown that proton pump inhibitors increase the risk of a second heart attack.

“These drugs are dangerous and have an increased risk for a number of adverse events, including cardiovascular disease,” Cooke said. “The data that we have suggests caution in the use of these drugs in the general public.”

While the drugs were approved for up to four weeks of use at a time, Cooke said doctors routinely prescribe them for much longer. Consumers buying them over the counter often take them for even minor symptoms and commonly ignore restrictions on duration, he said.

“Our report raises concerns that these drugs — which are available over the counter and are among the mostly commonly prescribed drugs in the world — may not be as safe as we previously thought,” said Dr. Nicholas Leeper, a Stanford vascular-medicine specialist who was the study’s principal investigator.

Doctors not involved in the study said it may be too early to jump to any conclusions.

Although the overall risk of heart attacks in people with no diagnosed heart disease remains low, the number of people taking the drugs makes the findings potentially significant. According to the Food and Drug Administration (FDA), one in 14 Americans takes a proton pump inhibitor.

“If you truly can prove that, then it raises the red flag for a large group of people who are exposed to this medicine,” said Dr. Shahid Rahman, an interventional cardiologist affiliated with Memorial Hermann Northwest Hospital.

The concern with data mining is that while the massive numbers of patients involved in these studies can provide a clearer signal of a problem, they can also magnify flaws in study design that lead to misleading results.

For example, proton pump inhibitors are routinely prescribed in the hospital, so patients taking the drugs may be sicker than those taking H2 blockers; their reflux may have been more severe. Many of the risk factors for reflux disease, such as obesity and smoking, are also risk factors for heart attacks.

“It’s very hard to assure ourselves that the PPI use is not a marker for something else,” said Dr. Elmer Bernstam, associate dean for research at the University of Texas Health Science Center at Houston School of Biomedical Informatics.