Theory 10/Pharmacovigilance
Signals Are a Reporting Ratio
The claim
Disproportionality signal thresholds should scale with the ratio of reports to exposure rather than with report counts, so a drug with high exposure needs proportionally more reports to generate the same signal.
01The mechanism
A signal is a departure from expected frequency. The expectation scales with exposure, so the same count carries different weight depending on how many patients took the drug.
That makes the ratio the threshold, and explains why rare events for common drugs are detected later than common events for rare ones.
02What this is not
Count-threshold signalling
Those use absolute counts. This uses exposure-adjusted counts, which explains asymmetric detection.
03Predictions
- P1Detection lag should track the exposure ratio.
- P2Exposure-adjusted thresholds should detect rare-event signals sooner.
04Refutation
Detection tracking counts at controlled exposure would remove the adjustment.