{
  "$type": "site.standard.document",
  "bskyPostRef": {
    "cid": "bafyreihnbhesoxxeig2ah2ixmstx7cwswjpe2stthtv7fzoc5exlfa764i",
    "uri": "at://did:plc:wwyqal4cnqhuwyacdj7rqq3n/app.bsky.feed.post/3mfdlgbbfmi72"
  },
  "path": "/t/statin-related-side-effects-the-recent-lancet-publication-is-biased-toward-false-negatives/28635#post_20",
  "publishedAt": "2026-02-19T01:28:17.000Z",
  "site": "https://discourse.datamethods.org",
  "tags": [
    "@ESMD"
  ],
  "textContent": "I think this is **the** crucial question. Trials are not powered for adverse outcomes by political and economic choice. This is unethical; by the “first, do no harm” principle, a medication’s safety profile should be well characterized before clinical use.\n\nThalidomide, opiates, DES..\n\nIt’s complicated, because we don’t always know what the adverse outcomes will be. There should be some hints in the earlier phase studies, but, beyond this, trials should be ongoing, with moving estimates of the probabilities of side effects as they emerge. The trial should not be stopped until essentially the distributions of the major ones (or the most harmful ones) are well estimated. Afterwards, if a new, really harmful adverse outcome jumps out in post-trial surveillance (with some reason to suspect causality, as discussed by @ESMD), it would actually cause the original trial to regain equipoise, and that trial would begin again enrolling new participants.",
  "title": "Statin-related side effects: the recent Lancet publication is biased toward false-negatives"
}