Tracking the safety record of new drugs has proven to be a slow, tough task for the US Food and Drug Administration; often it takes years after a drug has been on the market to determine its full extent of side effects, which are reported to the agency mainly by doctors who prescribe it.
Thousands of patients, however, use Twitter to tweet out symptoms they’ve developed from their medications, and Boston Children’s Hospital researchers found in a new study that such tweets could be a potential new way to identify drug problems in real time.
In the FDA-funded study published this week in the journal Drug Safety, the researchers from Boston Children’s, the FDA, and elsewhere searched Twitter posts mentioning 23 commonly used medications — including antidepressants, sleeping pills, and over-the-counter acetaminophen and ibuprofen — as well as vaccines over six months between 2012 and 2013. Out of 60,000 tweets mentioning these drugs or vaccines, 4,401 described side effects that were blamed on the medications or shots.
“The slice of adverse events we saw were not on the severe side,” said study co-author Dr. John Brownstein, an associate professor of pediatrics at Boston Children’s Hospital. “But these may be symptoms that don’t come to FDA’s attention because people don’t think of them as severe events and may not see a doctor for them.”
For example, one person tweeted, “So much for me going to sleep at 12. I am wide awake thanks prednisone and albuterol.”
Sometimes, though, tweets had no validity whatsoever like the person who complained about getting the stomach flu twice after getting the seasonal flu shot — which protects against respiratory, not stomach, viruses.
“We’re developing an automated filtering system with the FDA to help eliminate the 98 percent of tweets that turn out to be noise, or not relevant for adverse event detection,” said study leader Clark Freifeld, a research software developer at Boston Children’s Hospital. Once the system is up and running, he said, it could be used to detect side effect trends without having to search for each medication by name.