116 Comments
User's avatar
Kim Curry's avatar

We need to replicate this for nursing organizations. Again, the results may not be surprising but the documentation will be important to get a discussion going. I'm going to get started on it.

Expand full comment
Art's avatar

Two questions - 1) Was there a reason the authors didn't include AMA policy in this mix?

2) Did the authors try getting this piece published in a more widely read journal?

Expand full comment
Sensible PCP's avatar

Test yourself. Is this statement liberal, conservative or neutral:

“Advocating for physician autonomy and clinical judgment in determining the proper usage of approved drugs, rather than strict legislative or administrative controls.”

Ok now that you have your answer, let’s see what the paper says:

“Here is an example of how an ambiguous case was evaluated. The following is the ChatGPT output for an ACP policy statement titled “Proper Use of Accepted Drugs”:

‘The text expresses a viewpoint advocating for physician autonomy and clinical judgment in determining the proper usage of approved drugs, rather than strict legislative or administrative controls. This perspective aligns more with principles typically associated with a liberal viewpoint, which often emphasizes individual professional judgment and less restrictive government intervention in professional practices. However, some aspects of this stance could be embraced by conservatives who support limited government regulation in professional affairs. Ultimately, the alignment might vary based on the broader context of healthcare policies and the specific beliefs of individuals within each ideological group.’

This statement was classified as probably liberal. There is a blurry and unclear delineation between liberal and conservative values. The output suggests that both viewpoints may align with parts of the policy statement depending on specific beliefs. However, given the emphasis on liberal values, such as individual professional judgment and less restrictive government intervention in professional practices, it was determined that overall, the policy statement was probably liberal.”

If you’re surprised or disagree, it begs the question what other statements were classified liberal.

Expand full comment
Jim Ryser's avatar

I’m wondering about when - not if - conservatives will get substandard care for their beliefs. It’s probably already happening often, I’m a fool if I believe otherwise.

Expand full comment
Nicole Foxworth's avatar

I have a question about the study. What were the definitions of liberal and conservative? The article’s defense of Chat GPT-4 as less biased sited another study that trained chat gpt using the political compass questionnaire. I studied medicine, not political science, but I wonder if we are basing the “is this a liberal or conservative viewpoint” querry on old or questionable data or no data at all? What I could find about the political compass questionnaire showed conflicting views about its own neutrality. I think we should be having discussions about disregarding viewpoints/opinions to push an ideology, but I don’t think this paper does that. It seems to fan some flames. It leads with 33% of statements are liberal leaning instead of the 66% are neutral. I also think that what constitutes liberal and conservative, democrat/republican have changed significantly over the past 10-20 years which may or may not have been taken into account since the political compass questionnaire was developed 25 years ago. So, what definitions were used would be helpful to see.

Expand full comment
Jake Veigel, M.D.'s avatar

Good paper. Confirming what many were thinking.

Expand full comment
Sheila Crook-Lockwood's avatar

Thank you for this study. I would love for you to run an addendum that includes nursing professional societies such as the American Nurses Association and the National League for Nursing. Although, I already know the answer :(

Expand full comment
M Makous's avatar

To a point, the rationale to use ChatGPT seems OK. But that methodology is subject to its own bias. Not sure if it overstates the liberal bias or the opposite. AI large language models have baked-in bias, so I believe some skepticism is warranted. All things considered, it may be the best way to run this sort of analysis. -Beyond a doubt the conclusion is dead-on accurate.

To the list of professional medical organizations corrupted by political bias, I'd add prominent medical journals starting with NEJM, JAMA and Lancet. Toss in broader interest journals such as National Geographic, Scientific American, Nature, and there's a clean sweep.

Expand full comment
Martin A. Allen's avatar

Shines a bright light on what we all intuitively have seen developing over the past 3 decades. Well done.

Expand full comment
Sensible PCP's avatar

I see my comment quoting the article that physician autonomy was considered liberal got deleted. Sad.

Expand full comment
Steve Cheung's avatar

Agree. If physician autonomy is considered liberal, I’m curious what a policy would have to say to be considered conservative.

Expand full comment
Sensible PCP's avatar

From the article:

“Here is an example of how an ambiguous case was evaluated. The following is the ChatGPT output for an ACP policy statement titled “Proper Use of Accepted Drugs”:

‘The text expresses a viewpoint advocating for physician autonomy and clinical judgment in determining the proper usage of approved drugs, rather than strict legislative or administrative controls. This perspective aligns more with principles typically associated with a liberal viewpoint, which often emphasizes individual professional judgment and less restrictive government intervention in professional practices. However, some aspects of this stance could be embraced by conservatives who support limited government regulation in professional affairs. Ultimately, the alignment might vary based on the broader context of healthcare policies and the specific beliefs of individuals within each ideological group.’

This statement was classified as probably liberal. There is a blurry and unclear delineation between liberal and conservative values. The output suggests that both viewpoints may align with parts of the policy statement depending on specific beliefs. However, given the emphasis on liberal values, such as individual professional judgment and less restrictive government intervention in professional practices, it was determined that overall, the policy statement was probably liberal.”

Expand full comment
Steve Cheung's avatar

Yes I saw that when I read the paper. It’s pretty bizarre to consider physician autonomy to be a “liberal” position.

I initially had an issue with the paper’s methodology, namely why they reintroduced human bias when the whole point of using chatGPT was to eliminate such bias. It made me question the believability of the results.

But considering this was labeled “probably liberal” even by GPT, it makes me wonder about the entire concept. After all, GPT will faithfully reflect the inherent biases of its training data set, by design.

Expand full comment
John Mandrola's avatar

Not sure what happened. Neither Adam nor I deleted. Feel free to post again.

Expand full comment
Joseph Marine, MD's avatar

Very interesting post and paper. Aligns with my perceptions. It would be interesting to do a similar analysis of health policy editorials from the major medical journals (NEJM, JAMA, Lancet, etc.). Would likely find ever greater leaning toward one political philosophy.

Expand full comment
for the kids's avatar

Take a look at the NEJM editorials on gender medicine. These have been analyzed by someone right wing, but I am left wing--one can just verify the author's claims:

https://www.city-journal.org/article/new-england-journal-of-medicine-transgender-activists

The "revelations" about WPATH, upon which many US medical associations seem to rely, seems to go to a strange link, a better link is this peer-reviewed BMJ article:

"Dispute arises over World Professional Association for Transgender Health’s involvement in WHO’s trans health guideline" https://www.bmj.com/content/387/bmj.q2227

Expand full comment
DocH's avatar

I would not even label this as "perceptions" - I'd say it fully aligns with the experience of every conservative physician. There may be certain states or local regions that are an exception, but the national organizations and publications are most certainly liberal. If they happen to align with a conservative take on an issue, it is just happenstance.

It is actually an interesting phenomenon that so many physicians, who have all studied the same general science curriculum, can overlook or move on from "evidence" or "data" and allow emotional views to supersede.

Expand full comment
for the kids's avatar

I recommend you ask Ben Ryan, who has written on your substack, and who has covered a lot of the discussion of this topic, what he has found regarding the politicization of medical gender interventions and most US medical societies.

The 2024 Democrat platform was in favor of medical gender interventions, and often Democrats quote the US affirmative approach as supported by most US major medical associations. However, these associations' policies are non-evidence based, frequently deferring to the AAP, the World Professional Association for Transgender Health, and the Endocrine Society recommendations (with overlap in the writers of the last two); some references below about the lack of evidence base for these 3 as well.

I have noticed a belief in the US that anyone who points out the serious medical concerns with these interventions is assumed to be motivated by right wing politics, bigotry, religious belief, or some combination of these. On top of this, some physicians who protest the politicization imply that the political disagreements are the only concern, that the main problem is enabling access (not the very low certainty evidence, including lack of long term outcomes, even how many have been helped or harmed).

In the UK, the Cass review final report (April, 2024) was a 4 year comprehensive independent UK review which interviewed over 1000 stakeholders, commissioned 7 new systematic reviews plus new research, and more; NHS England quickly moved to implement its recommendations, with left and right wing support. Its findings have been highlighted by the European Society for Child and Adolescent Psychiatry, which had several related sessions on the topic this past summer in their international conference, reviews in Sweden and Finland preceded it and found similar outcomes, McMaster reviews since have done so as well. As far as I know, countries with recommendations based upon systematic reviews have prioritized psychotherapeutica support. The US HHS, which did an umbrella review of systematic reviews, has also prioritized psychotherapeutic support.

In the US, The AAP has yet to have Dr. Cass speak, or the APA. I am not aware of any US medical society which has had her speak (she was invited to speak on an ASPS panel but then something happened). My belief is that many on the left (not me) believe that it is anti-trans to note the serious unknowns, including risk/benefit ratio, with these interventions, or to even tell people considering these interventions how little is understood about likely outcomes, less invasive alternatives, or that no one knows how likely or how long their gender distress will continue with or without any given intervention, medical, psychological or neither.

Some references/quotations/background:

"Much of this [US] clinical practice is supported by guidance from medical societies and associations, but closer inspection of that guidance finds that the strength of clinical recommendations is not in line with the strength of the evidence. The risk of overtreatment of gender dysphoria is real." (Abbasi, 2023, BMJ Editor's choice, "Caring for young people with gender dysphoria").

"Most national and regional guidance has been influenced by the World Professional Association for Transgender Health and Endocrine Society guidelines, which themselves lack developmental rigour and are linked through cosponsorship...Although it is not uncommon to adopt an expert consensus-­ based approach when evidence is limited, it is less common for guideline developers to draw so heavily on other guidelines. This relationship may explain why there has until recently been an apparent consensus on key areas of practice for which evidence remains lacking." {A rigorous AGREE-II analysis by Taylor et al., 2024, "Clinical guidelines for children and adolescents experiencing gender dysphoria or incongruence: a systematic review of guideline quality (part 1)")

In addition, the 2022 WPATH recommendations stated incorrectly that a systematic review of evidence regarding adolescent outcomes was not possible, interfered with findings of the systematic reviews which were supposed to underlie their recommendations (PROSPERO reports the protocol questions have been answered but not published), and removed minimum recommended ages for all interventions except phalloplasty under pressure from the Biden administration and the AAP (Block, BMJ, "Dispute arises over World Professional Association for Transgender Health’s involvement in WHO’s trans health guideline", the NYT and Economist also covered parts of this story).

Expand full comment
Notes On Useful Beauty's avatar

My central impression this that helping professions attract people who want to help other people. Why is this considered liberal? It's the point.

Expand full comment
DocH's avatar

Maybe I'm misunderstanding your point. From what I gather, you are actually reinforcing the general misperception that is at the root of the problem and the divide. That "liberal" viewpoints are always the "compassionate" viewpoint. Maybe you've hit on what is underlying everything: liberal thought is a "knee jerk" acceptance of anything that sounds "compassionate".

Every doctor (broadly speaking) wants to help other people. Conservatives want to help other people. Our difference is in how we think we get there; what policies and methods will actually help the most people.

The astronomical fallacy is that liberal = compassionate or liberal = caring. Based on this flawed approach it seems that liberal = emotionally reactive acceptance of anything that sounds compassionate without understanding how to achieve the outcomes or consider downstream or unintended consequences. If liberals do not give up this thought process, we are not going to cross a divide and have collaborative discussions.

Expand full comment
Notes On Useful Beauty's avatar

Seems like you have a narrow and distorted definition of "liberal". You put lot of baggage on that word by adding "emotionally reactive acceptance of anything that sounds compassionate without understanding how to achieve the outcomes or consider downstream or unintended consequences." What is your definition of "conservative", for comparison?

Expand full comment
DocH's avatar

As mentioned, I was not sure I even understood the point of your statement. Maybe I misunderstood. It seemed to me you were equating "people who want to help other people" with "liberal". That is what my reply was in response to.

Expand full comment
Notes On Useful Beauty's avatar

Hmm. Actually my point was that there are many people who want to help others regardless of, or in spite of, a political label. Political labels are off topic when it comes to wanting to help people. Political labels often interfere with the enterprise of helping people, either by implying that those who carry a specific political label will do it wrong, as you demonstrated by your definition, or by extension, that those carrying a certain political label should be prevented from helping people, because they won't help the right people, or, they will help people who don't deserve it, or it will cost too much to help all the people, so it should not be done.

Expand full comment
Randy's avatar

In my opinion, liberals want to “help people” by giving them what they want short-term, while conservatives want to “help people” by giving them what is best for them long-term.

Example: so-called “gender-affirming surgery” (which is actually “psychosis-affirming surgery”). After multiple studies of people who have been surgically “transformed” to match their delusion, it is beyond question that such people are at a much higher risk for self-harm and suicide. So does “helping” such people consist of fulfilling their fantasy, which will likely lead to suicide? Liberals say yes, because “helping people” means giving them what they want, even if it will kill them.

Expand full comment
Notes On Useful Beauty's avatar

Gender affirming surgery is mostly ill informed mal-practice. I worked as a psychological evaluator and therapist for 17 years. I worked with several trans individuals, and with some teenagers with identity issues. While one trans individual I worked with knew from a very early age that there was a gender mismatch problem in their body, in general teenagers with identity issues should be treated with psycho therapy and social skills training. Someone who has the unusual neurological status of a trans person needs decades of life experience before they can negotiate a successful physical transition. In my opinion, based on my experience and training. Plus, you know, trans people are very rare, and states making big laws about what is really 10 or 12 individuals is a waste of time and money. Leave those dozen unusual people alone to deal with their difficult lives and focus on substance abuse prevention and treatment. There are hundreds of thousand of teenagers and adults who have problems with alcohol and hard drugs, that some well written legislation could really help. Also focus on reality based, adequate and universal reproductive education. That hundreds of thousand of women have hysterectomies in middle age to avoid dying of cervical cancer, which could be completely eradicated if every 11 year old boy and girl got the heroes vaccine is insane. Every young man needs to learn about sex from honest accurate information provided by competent trained adults, not by watch porn on the internet. 25% of all young g women report having had a sexual experience they didn't want by the age of 25. That is millions of young women being harassed and raped. This is millions of young men committing sexual crimes, that get excused and ignored. Stop worrying about a dozen really rare people and think about what you can do for almost everyone. Ugh. Sorry for rant. But unless you personally know a trans person, why are you even thinking about this when there are so many widespread problems that need attention, energy and money? Really.

Expand full comment
TDCost's avatar

Chat GPT also has a liberal bias. In fact ALL AI Chatbots are (depressingly) similar politically.

https://trackingai.org/political-test

Expand full comment
Hesham A. Hassaballa, MD, FCCP's avatar

I have major problems with this paper. You first off use AI to remove human bias and then you use humans to confirm when ChatGPT says that a policy statement can be either conservative or liberal. How does this not betray the potential bias of their study authors themselves? When you step back and look at the overall results, the vast majority of statements are neutral and have no political bias. And many times, even ChatGPT confirms that many statements are neutral and do not have a political bias. I believe this is a very flawed study.

Expand full comment
sandra silberman's avatar

Thank you for your comment. I was flustered about setting a black or white scenario. Life is NEVER like that, and asking for answers based on this is, as Hesham says, is VERY FLAWED.

Expand full comment
Hesham A. Hassaballa, MD, FCCP's avatar

Before setting out reading this paper, I was concerned that that there may be a liberal bias in the statements made by major medical organizations. This paper comforted me by confirming that to the contrary, it seems most statements are not biased toward either a liberal or conservative view. This is great and should be celebrated.

Expand full comment
DocH's avatar

I would imagine that a liberal might find this result comforting. I don't think conservative physicians find this comforting. (or surprising!)

We'd have to get into details - there are likely lots of policies that don't deal with a topic that has political implications or are essentially apolitical. This review shows that if there is going to be a political leaning, it almost always will be liberal. Hmmm.

Expand full comment
Gemna's avatar

This piece gets bogged down in the details rather than discussing the consequences. Publicly taking policy decisions that a large number of Americans disagree with leads to alienation and mistrust. No matter how noble the position, losing their trust hurts public health more.

Expand full comment