What are the WHO-UMC causality categories and the Kramer algorithm?

WHO-UMC and Kramer are two established methods for judging whether a drug caused an adverse event. The WHO-UMC system, from the Uppsala Monitoring Centre, sorts a case into six categories: certain, probable/likely, possible, unlikely, conditional/unclassified and unassessable/unclassifiable. It weighs the time relationship to drug intake, whether disease or other drugs could explain the event, the response to withdrawal (dechallenge) and, for the 'certain' category, rechallenge where necessary. The Kramer algorithm (Kramer et al., JAMA 1979) scores six axes: previous experience with the drug, alternative causes, timing, drug levels or evidence of overdose, dechallenge and rechallenge; the total places the case in definite, probable, possible or unlikely. Neither yields a precise probability; both make the reasoning explicit and repeatable. Burna AI applies both to every drug in a regimen. Naranjo is compared with both at burna.ai/blog/naranjo-who-umc-kramer-causality-scales.