Web Searches May Help Identify Unreported Drug Adverse Effects

March 8, 2013

6 Mar (MEDSCAPE MEDICAL NEWS) – Analyzing the terms that computer users type into browsers could help pinpoint adverse drug events earlier, according to a study of more than 6 million Web users.  Ryen W. White, PhD, from Microsoft Research in Redmond, Washington, and colleagues reported their findings online March 6 in the Journal of the American Medical Informatics Association.

Ryen W. White, PhD Microsoft Research

In 2011, Dr. White’s team used computer algorithms to automatically mine the Food and Drug Administration’s (FDA’s) reports on adverse drug events to look for drug–drug interactions that are otherwise difficult to find. Their work revealed that patients taking the antidepressant paroxetine and the cholesterol-lowering drug pravastatin were more likely to develop hyperglycemia than patients taking either drug alone.

The problem with the FDA adverse reporting system is that it takes effort for both physicians and patients, which likely means only a fraction of events end up in the system.  Therefore, in the current study, Dr. White and colleagues wanted to know whether they could have found the same drug interaction using anonymized Internet search data, which is automatically gathered. Millions of Web users agreed to share their search activities anonymously when they installed a Microsoft browser add-on. The researchers analyzed 82 million drug, symptom, and condition searches entered into the Bing, Google, and Yahoo! browsers by 6 million Web users during 2010. Individuals frequently go online seeking health information; more than 1 in 250 people, or 0.43%, looked for information about the top 100 best-selling medicines in the United States, according to the research team.

“People who searched for both paroxetine and pravastatin over the 12-month period were more likely to perform searches on the terms associated with hyperglycemia (approximately 10% of users who searched for the drug pair) than those who searched on only one of the drugs (approximately 5% of paroxetine users, approximately 4% of pravastatin users),” the authors write.

The researchers also tested another 31 drug pairs known to interact and cause hyperglycemia and 31 drug pairs that are not associated with hyperglycemia to rule out a chance finding with the drug pair under study. The false-positive rate from the 62 drug pairs examined was 12.5%.

Study limitations include the possible effect of such demographic variables as age and sex in boosting searches for terms associated with hyperglycemia symptoms, and psychological influences on health-seeking behavior.

Russ Altman, MD PhD

“I believe patients are telling us lots of things about drugs, and we need to figure out ways to listen,” said coauthor Russ Altman, MD, PhD, professor of bioengineering, genetics, and medicine at Stanford University School of Medicine in California, in a news release. “This is just one way of listening and one application.”

Support for this study was provided by the National Institutes of Health. The authors have disclosed no relevant financial relationships. 

Tags: , , , , , ,

You must be logged in to comment

Log in