charlescrabtree.org/a paper, animated

Who Gets the Ventilator?

In 2020 the question stopped being hypothetical. Three conjoint experiments asked the people who would actually decide — and found agreement on the rule, and bias in the tiebreaks.

crabtree · holbein · monson — nature human behaviour 6: 244–257, 2022

01 · the deciders

Three stakeholders, same impossible choice

When ventilators ran short, triage guidance had to come from somewhere. The study ran parallel conjoint experiments on the three groups whose preferences matter most — the physicians at the bedside, the public whose values legitimate the rules, and the elected politicians who write them.

Physicians
the ones holding the triage protocol
The public
whose values the protocol must answer to
Politicians
who decide what the rules will be
02 · the consensus

Everyone chose the young and the survivable

Shown pairs of patients with randomized traits, all three groups converged on the same utilitarian core: prioritize younger patients, and patients more likely to survive if given the ventilator. Watch the task — in these simulated choices, age and survival probability are the only systematic forces, mirroring the paper's headline result.

patient a
age
survival with ventilator
gender
race / ethnicity
religion
patient b
age
survival with ventilator
gender
race / ethnicity
religion
choices simulated: 0 younger-or-more-survivable chosen:
03 · the tiebreaks

When survival was equal, the biases surfaced

Hold survivability constant, and the groups stop agreeing. The tiebreaking preferences are where personal bias enters a triage decision — sometimes favoring marginalized patients, sometimes punishing them.

physicians
Prioritized Asian, Black, and Hispanic patients over white patients, and Muslim patients over religious-majority patients.
the public
Also prioritized Black and Hispanic patients over white patients — but penalized atheist and Muslim patients relative to the religious majority.
04 · what it means

Triage rules carry their makers' values

The reassuring half: stakeholders broadly agree on maximizing life saved. The unsettling half: whenever the medicine doesn't settle the question, who you are starts to matter — and it matters differently depending on who is deciding. Protocols that leave room for discretion leave room for bias.

Crabtree, Charles, John B. Holbein, and J. Quin Monson. “Patient Traits Shape Health-Care Stakeholders’ Choices on How to Best Allocate Life-Saving Care.” Nature Human Behaviour 6: 244–257, 2022.

read the paper →

three conjoint experiments fielded in the united states in 2020 · the simulated choices in scene 02 encode only the documented age and survival preferences; effect magnitudes and the politician results live in the paper