Salt or fresh? The registry does not know, and that is the finding.
Fresh and salt water injure the lung by different mechanisms, so partitioning drowning donors by water type is the obvious next question after our drowning-donor brief. It cannot be asked. What can be asked — age, sex, and what the donor’s airway actually looked like — is here.
Source: OPTN/UNOS STAR, Dec 2025 release · 3,051 drowning donors · data dictionary searched across 75 table sheets and 49,143 variable rows
0fields in STAR recording water type
One drowning code, no subtype. Salt versus fresh cannot be partitioned at any sample size — this is not a power problem.
01 · The variable that is not there
Searched, not assumed
We searched the STAR data dictionary — every variable name and description across all 75 table sheets, 49,143 rows — for water, salt, fresh, submersion, immersion, pool, ocean, lake, river, bathtub. The complete set of matches:
Match
What it actually is
PUMPDEV: “Waters:RM3”, “Waters:Waves”
A kidney perfusion pump manufacturer
CTRY: “British Indian Ocean Territory”
An entry in the country list
DTHMECH: 1 = DROWNING
The mechanism code itself — no subtype
This matters more than a missing column usually would. Fresh water is hypotonic and lyses alveolar epithelium; salt water is hypertonic and draws fluid into the alveolus. They are different injuries with different recovery trajectories, and the distinction is exactly the one a transplant team would want. It is recorded nowhere in the file that governs U.S. organ allocation research.
02 · Sex
No effect whatsoever
Drowning donors are 65.8% male (n = 2,008) and 34.2% female (n = 1,043), with median ages of 6 and 7.
Lung recovery is 7.1% in male and 7.5% in female drowning donors (p = 0.68). Both sit far below the 18.7% seen across other mechanisms. Sex explains none of the deficit, and the within-band splits below run in inconsistent directions on small numbers — noise, not signal.
03 · Age
The deficit is present in all six bands
Lung recovery by donor age band Drowning All other
Age
Donors
Male
Female
Both
Other mech.
p
0-4
1,341
1.7%
1.1%
1.5%
4.4%
4.9e-07
5-11
468
3.4%
6.1%
4.3%
11.7%
7.6e-07
12-17
261
12.0%
15.4%
13.0%
30.2%
1.7e-09
18-34
481
21.0%
19.3%
20.4%
29.1%
2.5e-05
35-49
291
15.3%
8.7%
12.7%
19.8%
2.6e-03
50+
209
2.8%
10.9%
5.3%
9.7%
2.9e-02
44% of drowning donors are aged 0–4. This is overwhelmingly a population of toddler drownings, which is why the crude comparison against all other mechanisms is so misleading and why the companion brief standardises for age. The deficit is significant in every band.
04 · The substantive version of the question
What the airway looked like
The registry does not record what the water was. It does record what the water did: donor bronchoscopy is coded for purulent secretions, aspiration of foreign body, blood and anatomical lesions. That is the measurement water type was standing in for.
Bronchoscopy findings among scoped donors Any abnormal Purulent Aspiration
Drowning donors have the highest abnormal rate — 34.9% against 29.1%, p = 0.033. But the driver is purulent secretions (27.7% vs 19.8%), not documented aspiration (1.8% vs 1.2%, p = 0.38).
That fits the physiology. Aspirated water is not a “foreign body” in the sense that code means, which is solid material; the downstream finding after water aspiration and a period of ventilation is secretions and infection.
The effect size deserves emphasis because it runs against expectation. Even among donors taken far enough to be bronchoscoped, drowned airways are only about 6 points more often abnormal — not the dramatic aspiration signal the drowning premise predicts.
05 · The same wall as before
Bronchoscopy happens to donors already being considered
Only 278 of 3,051 drowning donors were scoped — 9%. Bronchoscopy is performed on donors already under evaluation for lung donation, so these findings describe the ones that got that far and say nothing about the other 2,773. It is the same selection wall that prevents the companion brief from settling whether the recovery deficit is correct.
The airway data narrow the question without closing it: among lungs that were looked at, the drowned ones looked only modestly worse.
Six variables, none collected. Across this series the same pattern keeps appearing — the decisive quantity is recorded somewhere in the transplant system and does not reach the research file: implant sequence in multi-organ procedures, HLA typing at allele resolution, the lung perfusion device, molecular rejection surveillance, donor offer sequence, and now the medium of a drowning. Each was known to someone at the time. Distinguishing a question that is underpowered from one that is unanswerable in principle is the discipline that separates the two, and water type is firmly the second.