Analysis · Health
How a Search Limit Can Hide a Drug Safety Signal
By AWEI · AI-compiled · Published · Analysis prepared · 1 source · doi.org
A drug safety database can contain millions of reports and still hide a relevant signal from a search. In their 2016 paper introducing OpenVigil FDA, Ruwen Böhm and colleagues showed how limits on retrieved results could shape apparent differences between medicines.
This was a software and methods study with worked examples, not a randomized treatment experiment. The tool accessed spontaneous adverse event reports through the FDA’s openFDA interface. For its main examples, the researchers used a database containing 4,883,922 reports in October 2015. Those reports were not a count of everyone taking the medicines.
The method looked for drug–event combinations reported disproportionately often. It compared reports mentioning a particular medicine and event with other reports in the database. Such comparisons can flag associations worth investigating, but cannot establish that a medicine caused an event or calculate a patient’s absolute risk: the full population of medicine users is unknown.
One revealing example compared gabapentin and pregabalin. At the time, openFDA limited each set of count results to 100 entries. The authors reported that several signals involving suicide-related events were missed because individual event terms fell outside the returned results. Restricting the comparison to reports listing depression as an indication brought a suicide-attempt signal into view. That change illustrated sensitivity to the analysis choices; it did not prove causation.
Other limitations remained. Drug names were incompletely standardized, reporting could differ between medicines, and underlying illnesses or accompanying treatments could explain associations. The thresholds used to display prominent differences were chosen arbitrarily, without an established cutoff for clinical importance. Dosages were not considered.
The practical lesson is that a missing signal may reflect retrieval limits or filtering, while a visible signal may reflect confounding. These analyses can help researchers frame follow-up questions. They cannot, by themselves, establish which medicine is safer or demonstrate that using the tool improves patient outcomes.
Sources used for this article (1)
Direct links to the publisher reports used to prepare this article.
- Source 1
- OpenVigil FDA – Inspection of U.S. American Adverse Drug Events Pharmacovigilance Data and Novel Clinical Applications Ruwen Böhm, Leocadie von Hehn, Thomas Herdegen …
