Will crowd-sourced online search data become a better 'sentinel' for adverse event reporting?

Article

Academic study finds evidence of an early warning system in anonymized Web searches

Researchers at Columbia University, Stanford University and Microsoft Corp. have reported that analysis of search-engine phrases could become a crowd-sourced method of detecting adverse events from drug interactions, and potentially long before FDA’s MedWatch system picks up the signal. The development could be something of an end-run around FDA’s Sentinel Initiative, a effort started in 2008 to extract adverse event signals from patient records, such as those recorded in electronic health-record (EHR) systems now being installed in many health systems nationally. And that, in turn, is an end-run around one of the main obstacles in generating adverse-event reports (AERs) in the first place: the willingness and dutifulness of medical doctors to file them with FDA’s reporting system. (Various studies have estimated that as few as 10% of adverse events get reported.)

According to the study, published in the Journal of the Medical Informatics Assn. (Am Med Inform Assoc doi:10.1136/amiajnl-2012-001482), and reported in the New York Times, analysis of 82 million searches drawn in 2010 on Google, Yahoo and Microsoft’s Bing, cross-referenced between paroxetine (an antidepressant) and pravastatin (a cholesterol-fighting drug) showed an increased tendency to also search for common symptoms of hyperglycemia (such as elevated blood surgar or blurry vision). About 10% of the searches involved both drugs and the hyperglycemia indications, versus half that for one drug and the symptoms. (Paroxetine [the branded product, Paxil, has been off-patent for about a decade] has a medication guide that includes hyperglycemia as a risk factor, but the guide does not mention a drug-drug interaction with pravastatin.) The researchers call this an “early clue;” however, they knew what they were looking for: Stanford researchers had found the same connection between paroxetine and pravastain in 2011 after studying FDA AERs using similar data-mining techniques. (The later study used data available only before that study was conducted.)

FDA’s Sentinel Initiative has already aggregated over 159 million health records; now the challenge is to figure out how to draw out such signals. The study authors conclude that “Compared to analyses of other sources such as electronic health records (EHR), logs are inexpensive to collect and mine. The results demonstrate that logs of the search activities of populations of computer users can contribute to drug safety surveillance.”

Recent Videos
Related Content
© 2024 MJH Life Sciences

All rights reserved.