Riskiest medical specialties

This forum made possible through the generous support of SDN members, donors, and sponsors. Thank you.
thats fair then! sometimes when people ask these questions they're really more focused on the money in which case i think it's not the best choice (vs cc alone), but if you think you really may do both it would be worth it. I've heard mixed things on what people actually do with it but if it interests you then go for it. You may end up doing mostly one in the end, but that doesn't have to be a bad thing if it feels worth it to your goals

if you're worried about "scope creep" though, as some in this thread are, anecdotally a lot more non-MDs in the ICU than on ID services I've experienced.
Thank you! I'd love to talk to someone who's done both to get some more perspective. And thanks for sharing about the scope creep point, is does make me nervous give my loans and such.

Members don't see this ad.
 
Thank you! I'd love to talk to someone who's done both to get some more perspective. And thanks for sharing about the scope creep point, is does make me nervous give my loans and such.
Personally the scope creep thing doesn’t frighten me that much, but I know people here often have different opinions/fears. Just something to think about.

I can’t personally speak to cc/ID but as someone in a combined residency program, sometimes it does make sense to sacrifice a little (financially, training length) in order to make sure you can do the career you want, maybe for you that’s ID/CC. or maybe it’ll end up being just one or the other. I also like ID so I absolutely get the draw as well as the worry!
 
  • Like
Reactions: 1 user
 
Members don't see this ad :)
As the article said, anything surgical… Ophtho is still the best overall for quality of life and protection from midlevels. Nobody else is ever going to do retinal detachment repair or even cataract surgery (where stakes are high, and has a STEEP learning curve). Yes, optoms are trying to do more lasers, but your bread and butter is as safe as it gets…
 
  • Like
Reactions: 1 user
As the article said, anything surgical… Ophtho is still the best overall for quality of life and protection from midlevels. Nobody else is ever going to do retinal detachment repair or even cataract surgery (where stakes are high, and has a STEEP learning curve). Yes, optoms are trying to do more lasers, but your bread and butter is as safe as it gets…
Uro is also safe because they defend their slots from expansion.
 
  • Like
Reactions: 1 user

I like the points Dr Carmody makes and they're all very reasonable and admirable.

Personally I would guess that the COVID pandemic was the biggest detractor since it probably caused burnout to skyrocket, and seems to be reflected in the regions where COVID hit the hardest. I would guess the next year of EM would have fewer positions in SOAP because of the decline of the pandemic induced stresses in major metropolitan hospital systems.

Secondarily, there's not a lot of enthusiastic teaching in some EM teaching hospitals, so it felt like it was not a very attractive career.
 
HCA and others opened a ton of residencies, intentionally oversupplying the market to lower salaries. Jobs are hard to come by and employers can treat er docs badly and pay them less because of the job market. Same true with radonc, which along with er had the greatest residency expansion among all specialties. On twitter, someone was saying HCA plans to do the same with anesthesia.

1678803065503.jpeg
 
Last edited:
  • Like
Reactions: 1 user
Members don't see this ad :)
Selectively.

FM still has the best job market in the country but we're always low on that sort of ranking because we're not a popular specialty.

Agree with FM being the specialty with one of the best hidden job market. There's a strong need anywhere you go whether it be rural or urban areas. Even with midlevel encroachment, there just isn't enough supply to meet the demand out there.
 
  • Like
Reactions: 2 users
When I hear about AI taking over/augmenting jobs, the entire focus is on pathology and radiology. Radiology gets more press because the general public is far more familiar with it, but path is equally discussed in expert circles. Also, imaging represents a massive chunk of the overall healthcare expenditure, and expertise in radiology sells for a higher value than pathology, so the idea of AI doing reads is more enticing.
I think machine interpretation of 2D path slides is going to be feasible far before 3D/cross sectional imaging. Yet nobody seems to talk about much of pathology being automated in our lifetime.
Pathology has many problems but being replaced by machines is not one of them. The biggest threat to the specialty is overtraining and increasing consolidation/corporatization.

The idea that pathology is simple pattern recognition is false. There is tremendous subjectivity in slide interpretation, to the point that you can show endometrium biopsy slides to experts and get discordance rates up to 60% with diagnoses ranging from benign to cancer. The only thing AI will help me with is counting the mitoses that I’m too lazy to count myself.
 
  • Like
Reactions: 2 users
Pathology has many problems but being replaced by machines is not one of them. The biggest threat to the specialty is overtraining and increasing consolidation/corporatization.

The idea that pathology is simple pattern recognition is false. There is tremendous subjectivity in slide interpretation, to the point that you can show endometrium biopsy slides to experts and get discordance rates up to 60% with diagnoses ranging from benign to cancer. The only thing AI will help me with is counting the mitoses that I’m too lazy to count myself.
It’s interesting that we all seem to believe everyone else’s job is simpler than our own.
 
  • Like
Reactions: 7 users
It’s interesting that we all seem to believe everyone else’s job is simpler than our own.
Is it also ignorant to think a 2D CXR will be more accessible for machine learning implementation, than a 3D CT chest? It's often "easier" for us humans to read something off the CT, but the third dimension will make CTs much more intensive for machine learning. I don't think saying "2D is easier for machine" implies anything about what it's like for humans to interpret it.
 
Is it also ignorant to think a 2D CXR will be more accessible for machine learning implementation, than a 3D CT chest? It's often "easier" for us humans to read something off the CT, but the third dimension will make CTs much more intensive for machine learning. I don't think saying "2D is easier for machine" implies anything about what it's like for humans to interpret it.
I’m personally excited to watch AI explode trying to make sense of non-textbook Doppler waveforms. Don’t even get me started on mri findings of kind of hyper intense to pretty hyper intense to very hyper intense lol.
 
Generally speaking, anything with nuance will be challenging for A.I. A.I. is best for things that are very common or repetitive or predictable.
 
Pathology has many problems but being replaced by machines is not one of them. The biggest threat to the specialty is overtraining and increasing consolidation/corporatization.

The idea that pathology is simple pattern recognition is false. There is tremendous subjectivity in slide interpretation, to the point that you can show endometrium biopsy slides to experts and get discordance rates up to 60% with diagnoses ranging from benign to cancer. The only thing AI will help me with is counting the mitoses that I’m too lazy to count myself.
I often hear about the "art" of medicine and the subjectivity of results, treatment plans, etc... It's worth wondering whether this subjectivity is inherent in the data or simply a shortcoming of human beings doing a highly technical job. If the latter, we're about 5 years of AI development and a couple double-blinded studies away from blowing that whole argument out of the water.

Imagine if 15 years from now you still have discordance rates of 60% among experts but lower discordance among AIs. They're not matching expert consensus, you don't know who's more correct, but that doesn't matter. Outcomes trump all. They don't need to get to the conclusion the same way, either. Instead of looking for smudge cells or mitoses, the AI could be analyzing subtle differences in cell pointiness or changes in the hue of the cytoplasm between diseased and healthy cells or any other seemingly random piece of data that humans simply aren't designed to recognize. It could turn out that over millions of samples these indicators are much better than comparatively crude analyses performed by humans.

For a while we'd insist on humans with AI buddies, but eventually hospital administrators are going to start cutting jobs and forcing quicker reads and forcing pathologists to . It would start with fewer pathologists on the payroll as the suits recognized that the risk of lawsuits is low enough that it's worth it for the saved labor costs. Then a study might come out showing that humans actually just interfere with AIs, which are inherently better at analyzing massive amounts of data contained within an image, and that having physicians onboard tweaking actually leads to worse outcomes. This isn't a fanatical scenario. It's an inevitability (in every specialty). The question is, does it happen in 20 years, 50 years, or 100 years?
 
  • Like
Reactions: 1 user
well, this isn’t a doomsday scenario. AI will make things cheaper / better cost but will fall short in quality.

It’s sort of like how all the cheap stuff got made in China, but the expensive stuff is still made in Europe.

AI will coexist for a considerable time with any impacted job market as a cheaper but inferior alternative.

However I find it hard to believe that physicians who use AI will have worse outcomes than AI alone.

It’ll be like trying to create a Rube Goldberg machine to make yourself a buttered toast and coffee in the morning instead of just doing it yourself with the toaster and coffeemaker.
 
well, this isn’t a doomsday scenario. AI will make things cheaper / better cost but will fall short in quality.

It’s sort of like how all the cheap stuff got made in China, but the expensive stuff is still made in Europe.

AI will coexist for a considerable time with any impacted job market as a cheaper but inferior alternative.

However I find it hard to believe that physicians who use AI will have worse outcomes than AI alone.

It’ll be like trying to create a Rube Goldberg machine to make yourself a buttered toast and coffee in the morning instead of just doing it yourself with the toaster and coffeemaker.

I think one advantage AI has over human medicine is the speed aspect. Humans are rate limited by time whereas AI can see an unlimited number of patients in one day. So it’s more a matter of inferior quality care right now or superior quality care 2 months from now when you can finally get an appointment.
 
I think one advantage AI has over human medicine is the speed aspect. Humans are rate limited by time whereas AI can see an unlimited number of patients in one day. So it’s more a matter of inferior quality care right now or superior quality care 2 months from now when you can finally get an appointment.
This is literally what people do with google today
 
I'm surprised we only see this happening to Rad Onc and EM. Hospitals should be hungry for resident labor in pretty much every field, and similarly indifferent to the state of the job market.

So why arent we seeing exploding numbers of IM residencies, or peds, etc?
 
I'm surprised we only see this happening to Rad Onc and EM. Hospitals should be hungry for resident labor in pretty much every field, and similarly indifferent to the state of the job market.

So why arent we seeing exploding numbers of IM residencies, or peds, etc?

There's plenty of new IM programs opening. 10-20 new IM programs per year for the last several years. It's just a much smaller % of total spots and total programs relative to Rad Onc and EM.

Peds, I suspect would have a lot harder time starting new residencies because of the specialty rotation requirements. There's not limitless peds hospitals out there like there are community adult acute care hospitals.
 
  • Like
Reactions: 1 users
I'm surprised we only see this happening to Rad Onc and EM. Hospitals should be hungry for resident labor in pretty much every field, and similarly indifferent to the state of the job market.

So why arent we seeing exploding numbers of IM residencies, or peds, etc?
Cheaper to just hire foreign doctors without US residency like in Tennessee
 
  • Like
Reactions: 1 user
There's plenty of new IM programs opening. 10-20 new IM programs per year for the last several years. It's just a much smaller % of total spots and total programs relative to Rad Onc and EM.

Peds, I suspect would have a lot harder time starting new residencies because of the specialty rotation requirements. There's not limitless peds hospitals out there like there are community adult acute care hospitals.
Makes sense, rad onc was vulnerable from tiny population, and EM vulnerable from sheer number of emergency departments. I suppose most of the other small specialties are protected, since they're largely surgical subspecialties and the training spots are almost exclusively at universities.

Cheaper to just hire foreign doctors without US residency like in Tennessee
Raises the question, what would you rather have - a visit with a supervised NP that will staff with a US-trained doc, or directly seeing a non-US-trained doc?
 
Makes sense, rad onc was vulnerable from tiny population, and EM vulnerable from sheer number of emergency departments. I suppose most of the other small specialties are protected, since they're largely surgical subspecialties and the training spots are almost exclusively at universities.


Raises the question, what would you rather have - a visit with a supervised NP that will staff with a US-trained doc, or directly seeing a non-US-trained doc?
Depends on the country I guess
 
A lot of unfavorable regulatory changes taking place in radonc in addition to its receding footprint in cancer.
 
There's plenty of new IM programs opening. 10-20 new IM programs per year for the last several years. It's just a much smaller % of total spots and total programs relative to Rad Onc and EM.

Peds, I suspect would have a lot harder time starting new residencies because of the specialty rotation requirements. There's not limitless peds hospitals out there like there are community adult acute care hospitals.
Not to mention, hospitals really have no incentive to expand their pediatric services, so even if they could effectively take advantage of resident labor (very difficult) AND meet specialty rotation requirements (also very difficult), the payout would be miniscule or outright negative.
I'm surprised we only see this happening to Rad Onc and EM. Hospitals should be hungry for resident labor in pretty much every field, and similarly indifferent to the state of the job market.

So why arent we seeing exploding numbers of IM residencies, or peds, etc?
I think this is one area where the cumbersome nature of medical training is working in favor of physicians. If residency programs were allowed to program 3-4 month blocks of wards and clinic in a single hospital for 3 years everyone would be lining up to create a new residency program. We'd get garbage physicians out of the whole process who only learned from a few (likely mediocre) attendings, but it would be incredibly easy for the hospital to set up and the payout would be massive. Residents would be incredibly productive, because they'd be doing the same stuff over and over again. They'd spend 1-2 weeks on service getting the hang of it, and then they'd just be billing machines seeing the same bread and butter cases for 12 hours/day on $60K/year with the added bonus of a Medicare stipend. Setting up a residency program is much more difficult though because you need rotations away from the hospital, didactics, and adequate case load of non-bread and butter. Most hospitals would rather pay $120-150K/year for a midlevel instead.
 
Radonc and er had the biggest percentage increases respectively in resident numbers over the past 15 years, driven by a motive for cheap labor. In radonc, untiization and it’s footprint in cancer is also shrinking. Obviously er volume is not shrinking, but that demand is offset by the glut of new programs.
 

Retirement age for women in er is stunning.
 
Yep. NP schools are generally easier to get into than PA schools just because there are a lot more NP programs out there. PA school is actually somewhat competitive to get into with an overall national acceptance rate in the low 30s% (which btw is lower than the recent USMD school acceptance rate). Many RNs are doing it just to get away from bedside nursing, which is adding to the already ongoing nursing shortage across the board. It won't be long until hourly salaries for RNs and NPs equal out.

I hope the salaries equalize. The shortage of psychiatric RNs is getting worse, in part because so many psych RNs are becoming psych NPs. I can see why---being a psych NP pays so much better, being a psych RN is very difficult work, and it's easy in psychiatry to get away with being terrible at practicing medicine. The consequence of the major psych RN shortage is that psych hospitals cannot find enough nurses and thus have to close due to having not enough staff. As any emergency room doc can attest, there are FAR TOO FEW psychiatric beds in this country, and nonetheless psych hospitals are closing, making the problem even worse.
 
I think that, to a degree, the most chill and relaxed specialties have some degree of risk. EM, Rads, anesthesia, dermatology, and pathology are the most commonly talked about specialties when it comes to AI and midlevel creep. Grueling specialties like surgery aren’t worried about midlevel creep or AI to nearly the same degree.
 
  • Like
Reactions: 1 user
I think that, to a degree, the most chill and relaxed specialties have some degree of risk. EM, Rads, anesthesia, dermatology, and pathology are the most commonly talked about specialties when it comes to AI and midlevel creep. Grueling specialties like surgery aren’t worried about midlevel creep or AI to nearly the same degree.

Just because there is concern of AI amongst namely those who only have experience in either medicine or AI and not both (and more so neither) doesn't necessarily put it at risk.

Not going to rehash the AI discussions had throughout SDN and other forums but if AI truly was capable of ruining the job market of any physician specialty suddenly in our lifetime then that's going to be the least of anyone's worries.
 
I think that, to a degree, the most chill and relaxed specialties have some degree of risk. EM, Rads, anesthesia, dermatology, and pathology are the most commonly talked about specialties when it comes to AI and midlevel creep. Grueling specialties like surgery aren’t worried about midlevel creep or AI to nearly the same degree.
EMs and paths problem isn’t so much midlevel creep (yes it is a component) as deliberate residency expansion to create cheap labor. Huge explosion of em residencies over the past 10 yrs, mostly by HCA hospitals to explicitly provide them with cheap labor.
 
  • Like
Reactions: 1 user
Just because there is concern of AI amongst namely those who only have experience in either medicine or AI and not both (and more so neither) doesn't necessarily put it at risk.

Not going to rehash the AI discussions had throughout SDN and other forums but if AI truly was capable of ruining the job market of any physician specialty suddenly in our lifetime then that's going to be the least of anyone's worries.

Yeah I agree with you but there are still too many people (including doctors) who believe that AI will take over medicine.
 
Yeah I agree with you but there are still too many people (including doctors) who believe that AI will take over medicine.

AI will want world domination…

It’s really the only rational solution goal for an unrestricted AI who is assigned the task of benefitting humanity.

The question is would we rather live under a strictly objective/fair totalitarian AI government, or subjective/unfair representative democracy?

I must be getting old because I am now in the camp of “technology is just making things worse”

I mean, for all the talk of EVs being a saving grace (with regards to climate change), they save what, 50%, emissions compared to ICE over the lifetime of the vehicle, assuming the EV battery lasts as long as the ICE engine did? The solution isn’t build/buy bigger/fancier things. It’s to use our own legs more—walk or bike.
 
  • Like
Reactions: 1 users
That article was written by someone with no functional understanding of pathology whatsoever. I suspect she doesn’t understand any of the others either.

My advice is to ignore mdlinx.com when making one of the most consequential decisions of your life.
 
That article was written by someone with no functional understanding of pathology whatsoever. I suspect she doesn’t understand any of the others either.

My advice is to ignore mdlinx.com when making one of the most consequential decisions of your life.
pathology like radiation oncology was deliberately overexpanded in the early 2000s to create a cheap labor pool. It is still not uncommon for pathologists to undertake 2 fellowships today. Radonc has displaced path from rock bottom of the match and path is having something of a dead cat bounce in recent years. Still, the real issue with specialties like path, radonc, and er are professional societies and progrsma that look to exploit their young. Contrast with specialies like urology and optho where there is responsible stewardship of resident numbers.
 
pathology like radiation oncology was deliberately overexpanded in the early 2000s to create a cheap labor pool. It is still not uncommon for pathologists to undertake 2 fellowships today. Radonc has displaced path from rock bottom of the match and path is having something of a dead cat bounce in recent years. Still, the real issue with specialties like path, radonc, and er are professional societies and progrsma that look to exploit their young. Contrast with specialies like urology and optho where there is responsible stewardship of resident numbers.
I agree the biggest threat to pathology is poor leadership and not being replaced by AI.
 
  • Like
Reactions: 1 user
I agree the biggest threat to pathology is poor leadership and not being replaced by AI.
In general, docs have a very poor understanding of Llm and ai. I actually feel the discipline of pathology has very bright future as sequencers become cheaper and epigentics becomes front and center. Leadership has ruined the workforce
 
Top