Speculative explainer · crisis-standard-of-care

Who gets the ventilator?

One bed, two patients, four rulebooks. Every crisis-standard-of-care policy trades lives against life-years against equity — none of them wins on all three at once. Pick a policy, run the seeded reveal, and watch who it selects.

Step 1 · Policy selector
Lives saved / 1,000
—
—
Life-years saved / 1,000
—
—
Selection-rate gap (equity)
—
—
Loading equity read…
Step 2 · Adjustable assumptions
Vent : patient ratio ⓘ1 : 2
Access-gap severity ⓘ40%

Every figure above is a synthetic 1,000-patient population regenerated from the seed below — this panel controls the accounting, the seed controls the instance.

Step 3 · Seeded-lottery reveal

Same seed always produces the same population and the same "who lives" outcome — reproducibility is the point. Re-rolling shows how much a single draw's tie-breaks can swing around the banded estimate above. Grid shows the first 100 of 1,000 simulated patients.

selected + survives not selected + survives anyway not selected + dies selected + dies anyway
Documented record & further reading
Documented record Rubin JB et al., "Equitable Allocation and Ventilator Triage" — simulation cohort study, PLOS ONE (2024): 1 ventilator per 2 patients base case; age-banding ≈28.7 lives + 3,408 life-years saved per 1,000 (maximizes life-years); SOFA-bands ≈13.2 lives + 416 life-years saved per 1,000 (fewest lives, most equitable). All protocols except lottery and banded-SOFA produced statistically significant racial disparities in the study cohort.
White DB, Lo B et al., COVID-era crisis-standard-of-care allocation simulation, Am J Respir Crit Care Med. Caveats This page's SOFA-bands and age-band headline figures anchor on the PLOS ONE point estimates; lottery and SOFA+comorbidity outcomes, and every band shown, come from this page's own small synthetic 1,000-patient simulation (12 reseeds) built to the same tiers — not a reproduction of the study's full patient-arrival model. Age carries uncertain legal status as an allocation criterion in several US jurisdictions. Read every number here as illustrative, not a clinical recommendation. Genre / further reading (living-genre color, not scholarship, where noted) NY State Task Force on Life and the Law, Ventilator Allocation Guidelines (2015, rev.) — source of the ≤7 / 8–11 / ≥12 SOFA tiers used here.
Maryland Department of Health, Crisis Standards of Care framework (2021) — source of the +3-point severe-comorbidity SOFA adjustment.
This is an explorable explainer, not a frozen result — it computes over assumptions you can change above, and ships no canonical hash. When the underlying allocation model locks, nt_triage_allocator can be promoted to a deterministic MCP chain (parameters → hashed identity), the way ch_stadium_capacity already is.
SOURCE // Rubin et al. PLOS ONE (2024); NY-2015 / Maryland-2021 allocation frameworks. This page runs a client-side synthetic-population simulation only — no network calls, no worker call, no golden hash. All instruments →