← The impacts · AItopiaOrAImageddon?

entry · p2026-09-06-289 · validated

TREWS sepsis early warning system

reduced in-hospital mortality rate (3.3%, confidence interval (CI) 1.7, 5.1%, adjusted absolute reduction, and 18.7%, CI 9.4, 27.0%, adjusted relative reduction)
tech
learned-sepsis-alerting
passes the definition as
a risk model fit to electronic health record histories that scores every inpatient continuously for sepsis and raises a clinician-facing alert, in place of the hand-written SIRS threshold rules
from
2018
region
US
domain
health
side
benefit
people
590,736 patients (2021)
effect
18.7 percent (2021) — adjusted relative reduction in in-hospital mortality (3.3 percent absolute) among sepsis patients whose TREWS alert a provider confirmed within three hours, against those whose alert was dismissed, confirmed late or never addressed
reversible
yes
who holds the switch
owner:Bayesian Health
attribution
named
source kind
primary
the statistic
Adams et al., Nature Medicine 28, 1455-1460 (21 July 2022). 590,736 patients were monitored by TREWS across five hospitals; the analysis is of the 6,877 with sepsis identified by the alert before antibiotics were started. Confirmation within three hours was associated with a 3.3 percent absolute and 18.7 percent relative reduction in in-hospital mortality, plus less organ failure and shorter stays; among those additionally flagged high risk the absolute reduction was 4.5 percent. The companion paper (https://www.ebi.ac.uk/europepmc/webservices/rest/search?query=EXT_ID:35864251&format=json&resultType=core) reports 82 percent sensitivity and a 1.85-hour cut in median time to first antibiotic. In severity this is the strongest kind on the benefit side: the counted effect is deaths, which I would call absorbing for those it lands on, severe for the wider monitored population who got faster antibiotics. WEAKEST FIELD: the accessible pages give only 'a 2-year period' of deployment at five hospitals ending before the 21 July 2022 publication and print no calendar dates, so people.date is set inside that window rather than read off the page; strike it if that is not good enough. US population 2022: 333,996,304 (World Bank).
source
https://www.nature.com/articles/s41591-022-01894-0
proposed by
researcher (opus), brief 2026-09-06, severe-benefits
received
2026-09-06T17:40:15-04:00
harvest report
impacts/harvest/2026-09-06-severe-benefits.md

Rulings, in order

  1. 2026-09-06T17:40:16-04:00 · validatorpass — quote found on the fetched page (exact match; HTTP 200 text/html; charset=utf-8)
  2. 2026-09-06T17:40:46-04:00 · sessionseverity — Faster antibiotics and lower mortality for sepsis patients the model flagged before treatment: a treatment in service — severe; deaths averted would be absorbing but the paper gives a rate, not a count, and the confirmed-early denominator is not published. The 590,736 monitored bore nothing measurable; the people who bore the effect are the 6,877 sepsis patients identified by the alert (the paper's own analysis cohort), so the reach is derived on 6,877 (2021).
The proposal as it arrived (JSON)
{
 "id": "p2026-09-06-289",
 "received": "2026-09-06T17:40:15-04:00",
 "proposer": "researcher (opus), brief 2026-09-06, severe-benefits",
 "kind": "entry",
 "sha": "dc7ed1d2e90a551d",
 "data": {
  "kind": "entry",
  "tech": "learned-sepsis-alerting",
  "tech_name": "TREWS sepsis early warning system",
  "definition": "a risk model fit to electronic health record histories that scores every inpatient continuously for sepsis and raises a clinician-facing alert, in place of the hand-written SIRS threshold rules",
  "region": "US",
  "domain": "health",
  "side": "benefit",
  "from": "2018",
  "people": {
   "n": 590736,
   "unit": "patients",
   "date": "2021"
  },
  "effect": {
   "n": 18.7,
   "unit": "percent",
   "date": "2021",
   "what": "adjusted relative reduction in in-hospital mortality (3.3 percent absolute) among sepsis patients whose TREWS alert a provider confirmed within three hours, against those whose alert was dismissed, confirmed late or never addressed"
  },
  "reversible": true,
  "switch": "owner:Bayesian Health",
  "statistic": "Adams et al., Nature Medicine 28, 1455-1460 (21 July 2022). 590,736 patients were monitored by TREWS across five hospitals; the analysis is of the 6,877 with sepsis identified by the alert before antibiotics were started. Confirmation within three hours was associated with a 3.3 percent absolute and 18.7 percent relative reduction in in-hospital mortality, plus less organ failure and shorter stays; among those additionally flagged high risk the absolute reduction was 4.5 percent. The companion paper (https://www.ebi.ac.uk/europepmc/webservices/rest/search?query=EXT_ID:35864251&format=json&resultType=core) reports 82 percent sensitivity and a 1.85-hour cut in median time to first antibiotic. In severity this is the strongest kind on the benefit side: the counted effect is deaths, which I would call absorbing for those it lands on, severe for the wider monitored population who got faster antibiotics. WEAKEST FIELD: the accessible pages give only 'a 2-year period' of deployment at five hospitals ending before the 21 July 2022 publication and print no calendar dates, so people.date is set inside that window rather than read off the page; strike it if that is not good enough. US population 2022: 333,996,304 (World Bank).",
  "source": "https://www.nature.com/articles/s41591-022-01894-0",
  "quote": "reduced in-hospital mortality rate (3.3%, confidence interval (CI) 1.7, 5.1%, adjusted absolute reduction, and 18.7%, CI 9.4, 27.0%, adjusted relative reduction)",
  "attribution": "named",
  "source_kind": "primary"
 },
 "from_file": "impacts/harvest/2026-09-06-severe-benefits.md"
}