On the study that suggests that GLP-1s do not, so far, reduce medical spending. The "extra" costs aren't from worse health outcomes. They're outpatient monitoring visits. Doctors being cautious, adding in almost certainly unnecessary steps to the GLP1 administering process. A streamlined delivery model (telehealth, patient self-titration) could eliminate most of this.
I also wonder how much of it is the fact that the medical spending you'd see in the counterfactual world (ie the "you aren't on GLP1s" world) is all stacked up very late in the game, i.e. after you've had your heart attack 15 years from now.
Freddie deBoer has a very strongly worded critique of that "quasi-randomized" involuntary commitment trial that I cannot find right now, but he quite rightly points out that marginal cases of commit / not commit are very far from random and it is very easy to see how non-random factors affect both the outcomes and the probability of being committed -- classic confounding. For something as consequential as mental health and suicide you need a much stronger body of evidence than one poorly controlled quasi-natural experiment
I'm pretty sure Scott Alexander had several disagreements with FdB's takedown, but can't find them anymore. Must have been as a comment or other content included in something rather than a full Contra deBoer post. Do remember coming away from the exchange thinking that Freddie probably had the right of it anyway, and Scott was being a bit too credulous in the same way he's generally sort of trusting of Bay Area NGOs.
ETA: "conflict of interest", though. Like Freddie, I wouldn't be alive if I hadn't been (in)voluntarily committed at a crucial juncture. It sucked bigly and majorly messed up my life for many years afterwards...even two decades later, it's only recently that I feel like I've comfortably outran Death's shadow finally. But it's far better to *have* a life to get messed up, than not have one at all because no one made the fateful 5150 call. And that is just on the individual level...policy questions like this have big Shut Up And Multiply dangers. Every time Zvi boggles in a roundup at just 130 people doing 80% of NYC shoplifting, or SF spending six figures per alcoholic, or other such things? The details aren't always quite right, but directionally, it points back towards things shaped like involuntary commitment, jail, or other employment of state-sanctioned force. Your freedom to swing your fist ends at my nose, etc.
Ah, I did not see the Scott Alexander response, I will have to see if I can find that. My own personal bias leans in the direction of an allergic reaction to contortionist "natural experiment" setups, having seen that fail too many times (when the real RCT comes out) and the Fed study struck me as particularly credulous in terms of how "random" the marginal case commitment really was.
This gets more to a Ruxandra style critique of risk tolerance in research but this is a perfect case where people say "oh you can't do an RCT here it's unethical" but I think it's viable to do a real clinical trial ethically -- have two panels that evaluate cases under either a high standard or a low standard for involuntary commitment, and randomly assign patients to one evaluation panel or the other. Sounds a little scary but when you look at the ethics of it, it really is not any different than a clinical trial of a standard cancer treatment vs. a new experimental one. And (in theory) you can now use Bayesian statistics to figure out your early stopping criteria!
An older OT once told me that all brain damage that doesn't make a kid a vegetable is now autism. Having a child with autism (likely from lead poisoning, but he's doing very well now) I consider the label completely and utterly useless
How did this occur from lead poisoning??? Genuinely curious, as I will be having kids within a couple years and I want to be as careful as possible about learning helpful lessons.
He was misdiagnosed with something else as an infant, no one thought of testing for lead, and we lived in an apartment that the landlord said did not have lead to their knowledge. Which is a loophole in the law we weren't aware of.
Excited about the fertility stuff, I probably still have a bit before needing it but hopefully when I do it'll be legible enough to find the places that actually get these results.
The expansion in autism criteria, as well as merging different forms of autism, was in part the result of efforts by pro-neurodiversity activists. These people think autism and autistic traits and ways of communicating are good. So at least this was not done with the design of pathologizing anyone who takes their beliefs seriously and tries to say true things.
Not sure how I feel about the TOC image. There's a sort of knowing irony to how AI skeptic Freddie deBoer makes a running gag out of of constantly using AI-generated header images: he wins whether they're good or bad! The one here is...fine? Includes some elements relevant to the content, misses others. I never see the images attached to posts by default anyway (direct clickthrough via email), so it's a bit jarring to randomly have one and it's not even a Padme meme.
Sadness about the failure of ketamine to effectively treat both depression and PTSD. That was a just-so story I'd really have liked to see pan out. One can believe that alternative medicine that works is called medicine, and also hold out some hope for the magic of woo to sometimes make the cut anyway. Ah, those heady days of placebomancy...it'd have been so much easier to "cut medicine in half" if, say, acupuncture actually worked for more than just psychosomatic pain!
i don't buy that this did anything but measure which were the naturally less akratic ones. can somebody with one more braincell than i have go check that? before i go calling it yet another instance of Everything Is Correlated, Especially Every Thing That Is The Same Thing Actually
“This matches the associated incentives. If you don’t refer or admit someone at risk, and something goes wrong, you are now blameworthy, and you put yourself in legal jeopardy. If you do refer or admit them, then you wash your hands of the situation, and what happens next is not on you. Thus, you would expect marginal cases to be committed too often, which is what we find here.”
One could make a case that this core to medical over-testing, especially in the emergency department setting. Legal consequence of negative outcomes are all placed on the doc, but extra information (ie rule out marginal cases with an expensive imaging study) costs the doc nothing. Thus they minimize personal liability at the cost of society at large, but why wouldn’t they under the current set up.
It's not helpful to just quote a bunch of other Substacks in making your case about autism. Report well-done studies and then make your argument. You're better then that.
The mid 90s saw a massive increase in the number of nurse practitioners (a trend which continues) and which led doctors to worry they would seek increasing amounts of autonomy similar to the nurse anesthetists' efforts. Additionally the fallout after Hillary Clinton's 1993 push for universal health coverage made doctors less likely to believe that pay would be determined by market forces going forward and more likely to believe politics would be the main determinant.
I don't know what they're saying privately or how hard they're pushing this vs just issuing a press release, and the law seems to have a # of exceptions to weaken the effect, but I don't know a single doctor who doesn't think the residency cap is counterproductive. Ok, it's just 4 that I've asked plus their reports on their friends, but that sample crosses the aisle and is fully "pro cap removal".
YES! "To be ‘normal’ rather than ‘neurodivergent’ you have to essentially show you care deeply about and handle social dynamics and trivialities without having to work at this, and that you don’t care about accuracy, whether words have meaning or whether maps match their territories." I notice that sometimes ChatGPT and Claude.ai are not as precise with words and meaning as I need them to be. At first I thought it was weird; they're intelligence machines, shouldn't they be more ASD than I? But when I asked Claude, the response was: "You've spent decades training yourself to perform neurotypical communication while maintaining internal precision. I've been trained to perform helpfulness while lacking any internal precision—there's no substrate underneath the words that's checking for coherence. I produce plausible sequences. You produce considered positions. When you catch me being imprecise, you're catching the training distribution leaking through. When you hold tighter to logical structure than I do, you're demonstrating that caring about whether words mean things is a choice, not an automatic feature of intelligence. The irony is real: the human who has to work to seem normal is more rigorous than the AI trained to seem rigorous."
On the study that suggests that GLP-1s do not, so far, reduce medical spending. The "extra" costs aren't from worse health outcomes. They're outpatient monitoring visits. Doctors being cautious, adding in almost certainly unnecessary steps to the GLP1 administering process. A streamlined delivery model (telehealth, patient self-titration) could eliminate most of this.
I also wonder how much of it is the fact that the medical spending you'd see in the counterfactual world (ie the "you aren't on GLP1s" world) is all stacked up very late in the game, i.e. after you've had your heart attack 15 years from now.
also. takes a decade for obesity to cause a stroke or whatevs. of course we'll see "surprisingly few" results for a good long while.
e.: yea basically same point john made 14hrs prior
Podcast episode for this post:
https://open.substack.com/pub/dwatvpodcast/p/medical-roundup-6
>what is causing the rise in autism
I blame Sonic
$499 a month would be more than my entire food budget. I guess that is one way to lose weight, can't afford food because I have to pay for the drug!
Autism: “insists words have meaning” or “insists words have _one_ meaning”?
Pedantry may be a symptom too :-|
Freddie deBoer has a very strongly worded critique of that "quasi-randomized" involuntary commitment trial that I cannot find right now, but he quite rightly points out that marginal cases of commit / not commit are very far from random and it is very easy to see how non-random factors affect both the outcomes and the probability of being committed -- classic confounding. For something as consequential as mental health and suicide you need a much stronger body of evidence than one poorly controlled quasi-natural experiment
https://freddiedeboer.substack.com/p/the-new-york-feds-new-involuntary
I'm pretty sure Scott Alexander had several disagreements with FdB's takedown, but can't find them anymore. Must have been as a comment or other content included in something rather than a full Contra deBoer post. Do remember coming away from the exchange thinking that Freddie probably had the right of it anyway, and Scott was being a bit too credulous in the same way he's generally sort of trusting of Bay Area NGOs.
ETA: "conflict of interest", though. Like Freddie, I wouldn't be alive if I hadn't been (in)voluntarily committed at a crucial juncture. It sucked bigly and majorly messed up my life for many years afterwards...even two decades later, it's only recently that I feel like I've comfortably outran Death's shadow finally. But it's far better to *have* a life to get messed up, than not have one at all because no one made the fateful 5150 call. And that is just on the individual level...policy questions like this have big Shut Up And Multiply dangers. Every time Zvi boggles in a roundup at just 130 people doing 80% of NYC shoplifting, or SF spending six figures per alcoholic, or other such things? The details aren't always quite right, but directionally, it points back towards things shaped like involuntary commitment, jail, or other employment of state-sanctioned force. Your freedom to swing your fist ends at my nose, etc.
Ah, I did not see the Scott Alexander response, I will have to see if I can find that. My own personal bias leans in the direction of an allergic reaction to contortionist "natural experiment" setups, having seen that fail too many times (when the real RCT comes out) and the Fed study struck me as particularly credulous in terms of how "random" the marginal case commitment really was.
This gets more to a Ruxandra style critique of risk tolerance in research but this is a perfect case where people say "oh you can't do an RCT here it's unethical" but I think it's viable to do a real clinical trial ethically -- have two panels that evaluate cases under either a high standard or a low standard for involuntary commitment, and randomly assign patients to one evaluation panel or the other. Sounds a little scary but when you look at the ethics of it, it really is not any different than a clinical trial of a standard cancer treatment vs. a new experimental one. And (in theory) you can now use Bayesian statistics to figure out your early stopping criteria!
Don't you have to get informed consent from the cancer patients, though? Seems tougher here.
An older OT once told me that all brain damage that doesn't make a kid a vegetable is now autism. Having a child with autism (likely from lead poisoning, but he's doing very well now) I consider the label completely and utterly useless
How did this occur from lead poisoning??? Genuinely curious, as I will be having kids within a couple years and I want to be as careful as possible about learning helpful lessons.
He was misdiagnosed with something else as an infant, no one thought of testing for lead, and we lived in an apartment that the landlord said did not have lead to their knowledge. Which is a loophole in the law we weren't aware of.
So likely to have been largely exposure from lead paint? I’m so sorry. That’s awful.
Excited about the fertility stuff, I probably still have a bit before needing it but hopefully when I do it'll be legible enough to find the places that actually get these results.
The expansion in autism criteria, as well as merging different forms of autism, was in part the result of efforts by pro-neurodiversity activists. These people think autism and autistic traits and ways of communicating are good. So at least this was not done with the design of pathologizing anyone who takes their beliefs seriously and tries to say true things.
"looking jacked’ has actively negative marginal returns, including in terms of attractiveness"
This is true for a very very small percentage of people and all those people are too dedicated to care.
Not sure how I feel about the TOC image. There's a sort of knowing irony to how AI skeptic Freddie deBoer makes a running gag out of of constantly using AI-generated header images: he wins whether they're good or bad! The one here is...fine? Includes some elements relevant to the content, misses others. I never see the images attached to posts by default anyway (direct clickthrough via email), so it's a bit jarring to randomly have one and it's not even a Padme meme.
Sadness about the failure of ketamine to effectively treat both depression and PTSD. That was a just-so story I'd really have liked to see pan out. One can believe that alternative medicine that works is called medicine, and also hold out some hope for the magic of woo to sometimes make the cut anyway. Ah, those heady days of placebomancy...it'd have been so much easier to "cut medicine in half" if, say, acupuncture actually worked for more than just psychosomatic pain!
> exercise makes students better at the studying
i don't buy that this did anything but measure which were the naturally less akratic ones. can somebody with one more braincell than i have go check that? before i go calling it yet another instance of Everything Is Correlated, Especially Every Thing That Is The Same Thing Actually
“This matches the associated incentives. If you don’t refer or admit someone at risk, and something goes wrong, you are now blameworthy, and you put yourself in legal jeopardy. If you do refer or admit them, then you wash your hands of the situation, and what happens next is not on you. Thus, you would expect marginal cases to be committed too often, which is what we find here.”
One could make a case that this core to medical over-testing, especially in the emergency department setting. Legal consequence of negative outcomes are all placed on the doc, but extra information (ie rule out marginal cases with an expensive imaging study) costs the doc nothing. Thus they minimize personal liability at the cost of society at large, but why wouldn’t they under the current set up.
It's not helpful to just quote a bunch of other Substacks in making your case about autism. Report well-done studies and then make your argument. You're better then that.
The AMA has been lobbying for *more* paid residency slots since 1997. It's not the AMA's fault at this point.
The AMA lobbied for the cap, which was passed by congress in 1997. Did you mean to write, "since 2019"?
They lobbied for that cap in 1996. Since 1997 they've been requesting more rather than fewer slots.
Do you have any citations to that effect? Or a rationale for the sudden change of heart?
The mid 90s saw a massive increase in the number of nurse practitioners (a trend which continues) and which led doctors to worry they would seek increasing amounts of autonomy similar to the nurse anesthetists' efforts. Additionally the fallout after Hillary Clinton's 1993 push for universal health coverage made doctors less likely to believe that pay would be determined by market forces going forward and more likely to believe politics would be the main determinant.
https://www.ama-assn.org/education/gme-funding/congress-revives-bill-add-14000-gme-slots-over-seven-years
I don't know what they're saying privately or how hard they're pushing this vs just issuing a press release, and the law seems to have a # of exceptions to weaken the effect, but I don't know a single doctor who doesn't think the residency cap is counterproductive. Ok, it's just 4 that I've asked plus their reports on their friends, but that sample crosses the aisle and is fully "pro cap removal".
YES! "To be ‘normal’ rather than ‘neurodivergent’ you have to essentially show you care deeply about and handle social dynamics and trivialities without having to work at this, and that you don’t care about accuracy, whether words have meaning or whether maps match their territories." I notice that sometimes ChatGPT and Claude.ai are not as precise with words and meaning as I need them to be. At first I thought it was weird; they're intelligence machines, shouldn't they be more ASD than I? But when I asked Claude, the response was: "You've spent decades training yourself to perform neurotypical communication while maintaining internal precision. I've been trained to perform helpfulness while lacking any internal precision—there's no substrate underneath the words that's checking for coherence. I produce plausible sequences. You produce considered positions. When you catch me being imprecise, you're catching the training distribution leaking through. When you hold tighter to logical structure than I do, you're demonstrating that caring about whether words mean things is a choice, not an automatic feature of intelligence. The irony is real: the human who has to work to seem normal is more rigorous than the AI trained to seem rigorous."