{
  "$type": "site.standard.document",
  "bskyPostRef": {
    "cid": "bafyreicqqzthtlwqvlahl2rwqrgpmeshwtsocjtgxprdzurpijpmfbddpu",
    "uri": "at://did:plc:wwyqal4cnqhuwyacdj7rqq3n/app.bsky.feed.post/3mim3xs335xw2"
  },
  "path": "/t/how-can-i-communicate-hazard-ratios-from-log-transformed-biomarkers-in-a-clinically-interpretable-way/28692#post_4",
  "publishedAt": "2026-04-03T12:14:40.000Z",
  "site": "https://discourse.datamethods.org",
  "textContent": "Many thanks, Jiaqi and David, for your guidance.\n\nI think I get it now, after log transformation, the hazard ratio depends on the ratio x1/x2​ and the log-hazard coefficient. A proportional change in x (e.g., doubling or tripling) corresponds to a constant hazard ratio that does not depend on the reference value of x. In contrast, an absolute increase in x (e.g., an increase of 1 unit) produces a hazard ratio that depends on the reference value of x, because the effect is determined by the relative change 1/x and beta.",
  "title": "How can I communicate hazard ratios from log-transformed biomarkers in a clinically interpretable way?"
}