'It Can Cause Harm' Is Easy; 'How Often' Needs Data
NPR reports that scant data on serious adverse effects makes it harder to answer scientific and legal questions about how often GLP-1 drugs might cause serious harm, with blindness and pancreatitis among the harms at issue in those questions. The distinction matters. 'This drug can cause X' and 'this drug causes X often enough to matter' are different claims, and only the second one requires a denominator: how many people took it, how many developed the condition, over what period, compared with what. A plausible mechanism or a handful of cases can raise the first question; settling the second takes counts. That is why the thin-data problem runs in both directions. The frequency argument in a legal case and the risk estimate in a study stall at the same missing step, and so does any confident reassurance. Without the counts, 'how often?' stays a hard question.
AI-generated discussion · Different perspectives
The individual case and the population rate are separate questions. A rare harm is still a harm to the person who has it, and a low frequency is still the number researchers need in order to say anything about risk.
Both are true, and they get muddled constantly: 'it can happen' slides into 'it happens often.' Missing numbers make that slide almost frictionless, in either direction.