Can we talk more about research papers that try to link doctor characteristics-- particularly those that have been the target of discrimination-- such as race, and gender, and sexual orientation-- to patient's outcomes?

All this research is problematic

Let me explain
[thread]

First, to the lay person, I understand why it makes intuitive sense why your health, your life is tied to the your doctor.

Of course, we all believe that, and in some situations it may be acutely true.

A surgeon with a shaky hand may be responsible for my demise! i hope not

However, too often it-- from a medical standpoint-- it is a shaky proposition.

Patients, particularly hospitalized patients are cared for by dozens of health care professionals of all different races, genders, and rank

The sicker they are, the more hands on deck

Many many people are driving health care decisions. Perhaps >20 >40

If you replace some of the planks of the ship, I suspect it is possible you would get different decisions. However, there is scant data for this. It is also possible decisions are quite comparable.

We all learn medicine in the same '2020 cannon of medicine' and differences between Harvard and Oregon are far fewer than similarities

The variation in decisions (if it exists) made may have an impact on outcomes, but it is also possible the change in outcomes are tiny, minuscule, the dreaded -- NS

Off the bat I think that substantive evidence has to be generated that there is any person on the team who is even capable of steering the supertanker on GPS nav.

Even a doctor who is presumably all by themself-- an ed doc-- may be influenced by colleagues working at the same time, as they say 'Hey, let me bounce this off you'

Wedding one doctor to one patient's fate is extremely murky

Its hard enough in a criminal case of malpractice

In an administrative dataset -- omg, it is near impossible!

None of the administrative data techniques actually track key decisions. They use weak proxies like who bills for what

Next, these physician characteristics--- particularly those that have been the target of centuries of discrimination-- are tied to many other things.

The choice of specialty, places of practice, populations cared for.

Our society is divided across these fault lines

Separating the doctor and patient from the circumstances surrounding their place/ timing/ details and communities of employment is a tall task.

Docs who work in underserved communities may have limited resources and worse health outcomes through no fault of their own

Third, the analytic choices here are vast, and malleable.

In a legendary experiment by @BrianNosek and colleagues

When the EXACT SAME DATA were given to several teams investigating a claim of racial bias. These were all the estimates

Analytic flexibility allows for a range of possible outcomes in retrospective data sets, particularly for huge data sets, which will reach significance no matter what-- even trivial differences will be nominally significant.

Fourth, now returning to our question

Studying this topic is fraught. If you ask whether a historically marginalized group has better, worse or similar outcomes, you have to struggle with the aforementioned & also the results...

If you find better or similar outcomes, that would be welcome in our society. Worse outcomes would be a catastrophic finding.

Of course, these worse outcomes could be (and mostly likely are) artifacts, alternatively, it could represent discrimination from the patient's side

However, whatever the reason-- the ability to report worse outcomes is socially difficult in 2020. The potential for bad actors to misuse the findings are immense.

Fifth - when you put this all together, the recipe is problematic.

You have a proposition that is ill defined (linking one doc to one person)

You have massive analytic flexibility that can reach a range of outcomes

You have a strong disincentive to reach one result...

The net result is that these papers are unlikely to offer high credibility.

Sixth, and finally, I believe that this is all well-meaning but misguided.

It is well meaning-- folks want to advance the case for historically marginalized groups

I do too!

But it is misguided because it reduces a discussion that ought to be about justice and equality to a competition to justify justice and equality.

Equality is only worth it if we are better off.
I reject that.

Equality, in and of itself, means we are better off.

In science, we have issued prohibitions on dangerous research like gene editing embryos & decades ago recombinant tech.

I think we should issue a moratorium on research that tests whether immutable characteristics of doctors affect patient outcomes.

The societal risks from misuse are too great.

The ability to reach true causal conclusions is too illusory

And the rationale is accepting a broken foundation for
justice and equality that we ought to reject.

Let me end with an anonymous poll

Do you agree or disagree with me that this entire body of literature is problematic