Disproportionality Analysis
What does it mean in plain English?
In pharmacovigilance, disproportionality analysis is a statistical method used during signal detection. It compares how often a specific drug and a specific side effect are reported together against how often you'd expect to see that same pairing purely by chance, based on everything else in the safety database. If a combination shows up far more than chance alone would predict, it gets flagged as a potential "signal" worth a closer look.
Safety scientists use a few different versions of this math: the Reporting Odds Ratio (ROR) and the Empirical Bayes Geometric Mean (EBGM) are two of the most common. None of them proves the drug caused the reaction on their own; they simply point a flashlight at combinations that look statistically unusual, so a human assessor knows exactly where to look next.
Spot the signal
Five drug–event pairs, all looking roughly normal at rest, sitting right around the expected reporting baseline. Hover each bar to reveal what the numbers actually say.
Hover a bar to reveal the real countCommon Misconceptions
"The data shows this drug is linked to Liver Injury way more than expected, so that's proof, right? Pull it from the market now!"
Not quite. A statistical signal just means the pairing shows up more than chance alone would predict; it's a lead, not a verdict. Every flagged signal still goes through a full causality assessment, checking timing, alternative explanations, and biological plausibility, before anyone concludes the drug is actually responsible.
