Asphyxial death does not mark a worse donor lung — and the one result that said it did was confounding.
Hypoxia and aspiration make asphyxial donors the obvious candidate for a death mechanism that damages lungs. Tested four independent ways across 316,248 deceased donors, the premise fails every time. Three of the four point estimates favour the asphyxial group.
Source: OPTN/UNOS STAR, Dec 2025 release · 12,509 asphyxial-mechanism donors (4.0% of all deceased donors, the 7th most common mechanism)
33.6% vs 32.7%severe PGD at 72 h, asphyxial vs all other donors
No difference (p = 0.61). One-year graft survival is nominally higher in the asphyxial group, and none of fourteen donor-CT features separates them.
01 · The question the registry cannot answer
There is no strangulation code
This started as a count — how many donors died from strangulation. The honest answer is that the registry does not record it. ASPHYXIATION is one category pooling hanging, ligature and manual strangulation, positional asphyxia, smoke inhalation and airway obstruction — different physiology, one code.
Circumstance of death among asphyxial-mechanism donors
Circumstance records intent, not method. Suicidal asphyxiation (6,249 donors) is the closest available proxy for hanging; homicidal asphyxiation (154) for strangulation. If strangulation is the question, the answer is bounded above by roughly 154 donors — and even that conflates strangulation with smothering and other homicidal airway obstruction. Reporting a single number would imply a precision the data does not have.
02 · The premise
If any death mechanism marks an injured lung, this is the candidate
Asphyxial death involves profound hypoxia, frequently aspiration of gastric contents, and negative-pressure pulmonary edema is a recognised consequence of forceful inspiration against an obstructed airway. The premise is clinically reasonable. It is also testable four independent ways: utilisation, early graft function, one-year survival, and the donor CT itself.
Characteristic
Asphyxiation
All other mechanisms
Donors with lungs recovered
2,418
56,391
Median age
28
33
Donation after circulatory death
12.8%
6.2%
Median P/F ratio
442
442
P/F below 300
7.8%
8.5%
Two imbalances, pulling opposite ways. Asphyxial donors are five years younger, which predicts better lungs. They are also roughly twice as likely to donate after circulatory death (12.8% against 6.2%), which predicts worse utilisation. The second explains the only significant result below. Donor oxygenation does not separate them at all — and slightly favours the asphyxial donors.
03 · The one significant result
A 1.3-point utilisation gap that is really a DCD gap
At least one lung is recovered from 19.3% of asphyxial donors against 18.6% of the rest — marginally higher. Of lungs recovered, 90.9% are transplanted against 92.1%, at p = 0.0016.
Share of recovered lungs transplanted, pooled and stratified Asphyxiation All other mechanisms
The gap vanishes on stratification, and reverses within brain-death donors. Within DBD it is -0.3 points (p = 0.40) — asphyxial lungs used slightly more. Within DCD it is +0.2 points (p = 0.91). This is Simpson's paradox: asphyxial donors are twice as likely to be DCD, and DCD lungs are transplanted at 71% against 94% for brain-death lungs. The mechanism label was standing in for the donation pathway.
With 111,036 lungs in the denominator, statistical significance is cheap. Effect size and confounding are what matter.
04 · The outcomes
Recipients do not do worse — if anything, slightly better
Outcome
Asphyxiation
All other mechanisms
p
Severe PGD at 72 h
33.6% (n=782)
32.7% (n=13,110)
0.61
Median P/F at 72 h
300
306
—
1-year graft survival
86.2% (n=1,529)
85.1% (n=34,011)
0.25
Severe PGD is the ISHLT grade-3 proxy — P/F below 200 or ECMO at 72 hours — across 13,893 of 39,793 linked transplants with a derivable grade. Both point estimates run opposite to what the clinical premise predicts.
05 · The imaging read
The donor CT does not distinguish them either
Imaging is where a hypoxic or aspiration injury should be most directly visible. Across 5,119 donors with a linked chest CT (261 asphyxial, 4,858 other), all fourteen features of the interpretable HU panel were compared by permutation test on the difference in medians.
None of the fourteen separates the groups after Bonferroni correction — including the aeration and density features that would carry an infiltrate.
One caveat specific to this read. The first version of this comparison ran against a consolidation feature that was structurally incapable of taking a non-zero value, so the feature most directly expressing the hypothesis was not being tested at all. That defect is fixed and the comparison re-run; the null here is from the repaired panel. The earlier version was uninformative rather than negative.
06 · What it means
Permissive, not directive
Nothing here argues for accepting a lung that looks poor. It argues that the death mechanism alone is not a reason to look harder — and that asphyxial donors, five years younger with equal oxygenation, should not be discounted on mechanism.
One thing this cannot settle. That transplanted asphyxial lungs perform normally is consistent both with correct selection having removed the injured ones and with there being little to select against. The lungs that were declined have no outcome, so the two cannot be told apart here. Read the null as "mechanism is not itself informative", not as "no asphyxial lung is ever injured".
One instance of a pattern. Across this series, donor and recipient attributes recorded in the registry repeatedly fail to predict early graft dysfunction — HLA antigen mismatch, preservation solution, every component of the allocation score, donor offer sequence, and now death mechanism. The consistent implication is that the risk of early graft dysfunction lives in donor and preservation variables the registry does not collect, which is where a direct assessment of the organ itself would have to earn its value.