Study: Adverse Drug Event Data Plentiful on Twitter, but Less Useful
May 1, 2014 in News
For the study, researchers scanned Twitter using keywords related to the drugs, collecting about 6.9 million pertinent tweets over a six-month period. The researchers then randomly selected 61,000 of those tweets to manually analyze.
Of the analyzed tweets, the researchers identified 4,400 as adverse events. In comparison, the researchers found that just 1,400 adverse events related to the same drugs were reported to FDA over the same time period.
According to the researchers, the patterns of Twitter data suggested that some users intended to “passively” report the problems. For example, users would sometimes link their complaints to “hashtags and mentions, such as #accutaneprobz and @EssureProblems.”
In addition, some of the comments suggested that Twitter users understood the idea of causation in drug safety, with some remarking that the adverse event was alleviated after they stopped ingesting the drug.
However, the researchers identified several limitations to the Twitter data, including that adverse events reported through the social media site:
- Were not always as specific or as useful as those reported to FDA; and
- Were too varied and complex to be successfully analyzed by a computer, while the overall number of tweets — especially for widely used drugs — eventually could “overwhelm the capacity for human review.”
The researchers said they “do not recommend the wholesale import of individual social media posts into post-marketing safety databases.” However, they said the data could be used with “other post-marketing sources … for idea generation and reasonable hypotheses followed up by formal epidemiologic studies” (Comstock, MobiHealthNews, 4/30).