Thread by Manasvini Singh
Thread
Sizzling new WP with the incredible @sdschwab!
TLDR: The powerful really *are* living their best lives. Using data from military hospitals where both docs and patients have rank (a measure of power), we find that powerful patients get more care and have better outcomes.
π§΅1/9
TLDR: The powerful really *are* living their best lives. Using data from military hospitals where both docs and patients have rank (a measure of power), we find that powerful patients get more care and have better outcomes.
π§΅1/9
Though power is tricky to study in the real-world, we can do so with Military Hospital ED data.
How?
-Hierarchy is an imp way power is enforced in society
-Docs & patients have ranks, which we use to measure their power differential
-Patients as-randomly assigned to docs in EDs
How?
-Hierarchy is an imp way power is enforced in society
-Docs & patients have ranks, which we use to measure their power differential
-Patients as-randomly assigned to docs in EDs
5 big results (methods in paper!):
1) βHigh-powerβ patients, i.e., those who outrank their physician, get
more physician effort (RVUs)
more resources (tests, procedures, opioids)
lower chance of 30-day hosp admission
than βlow-powerβ patients π¨π ππ‘π π¬ππ¦π π«ππ§π€
1) βHigh-powerβ patients, i.e., those who outrank their physician, get
more physician effort (RVUs)
more resources (tests, procedures, opioids)
lower chance of 30-day hosp admission
than βlow-powerβ patients π¨π ππ‘π π¬ππ¦π π«ππ§π€
2) A doc provides more effort to patients ππ’π π‘ promoted to rank R than they do to patients ππππ’π‘ to be promoted to rank R. Patientβs promotion date is unknown to doc, so promotion only changes the power diff b/w them.
(Robust to placebos with 250 βfakeβ promotion dates)
(Robust to placebos with 250 βfakeβ promotion dates)
3) Docs may be reallocating effort away from low-power patients towards high-power patients:
On days that a physician is assigned to a high-power patient, their concurrently seen low-power patients suffer (receive lower effort, are more likely to return to ED within 30 days).
On days that a physician is assigned to a high-power patient, their concurrently seen low-power patients suffer (receive lower effort, are more likely to return to ED within 30 days).
4) Doc-patient concordance on race and sex interacts strongly with patient rank.
Eg: White docs give Black low-power patients less effort than White low-power patients π¨π ππ‘π π¬ππ¦π π«ππ§π€. Outranking their doc allows Black patients to get more effort from White docs.
Eg: White docs give Black low-power patients less effort than White low-power patients π¨π ππ‘π π¬ππ¦π π«ππ§π€. Outranking their doc allows Black patients to get more effort from White docs.
5) Why does this happen? Do the docs respect their high-power pats (status), or are they scared (authority)?
Using date of patient retirement β when status is kept but authority lost β we find that being high-power is beneficial even after retirement! So status plays a role.
Using date of patient retirement β when status is kept but authority lost β we find that being high-power is beneficial even after retirement! So status plays a role.
We do a bunch of robustness checks -
physician FEs (account for physician ability/ practice style)
spec curve using 1200 covariate combinations,
re-running analyses limited to when ED is busiest (to minimize non-random matching),
propensity-matched sample etc β¦
All good!
physician FEs (account for physician ability/ practice style)
spec curve using 1200 covariate combinations,
re-running analyses limited to when ED is busiest (to minimize non-random matching),
propensity-matched sample etc β¦
All good!
Conclusion: the powerful enjoy better outcomes in society by simply possessing power. Given equity concerns, results may be especially relevant to healthcare β¦
but can be applied to any non-healthcare setting with a power imbalance (employee-employer, landlord-renter etc)
/FIN
but can be applied to any non-healthcare setting with a power imbalance (employee-employer, landlord-renter etc)
/FIN
Mentions
See All
Ethan Mollick @emollick
Β·
Mar 20, 2023
Impressive & important paper is able to get at the role of power in health care by looking at the military, where the ranks of doctors & patients are known. The powerful get more attention, which also results in better outcomes, at the expense of the less powerful. Good thread.