Things are relatively quiet on the AI front, so I figured it’s time to check in on some other things that have been going on, including various developments at the FDA.
New 2026 ACC/AHA guidelines came out with instructions on how to identify and treat dyslipidemia (which includes but is not restricted to hyperlipidemia) earlier than is currently being done, which aligns mainstream medical practice more with what the lipid hawks are saying. If these guidelines are properly implemented it could be very impactful towards lowering ASCVD.
> It doesn’t make any sense for bicycling and swimming to have about zero effect
I'd guess we're looking at the combination of a reduction in health-related mortality (saturating with time spent cycling/swimming the same way that of walking does) and an increase in accident-related mortality (traffic collisions/drowning respectively) roughly proportional to time spent cycling/swimming.
Suboptimal social behaviors are being misclassified as 'on the Autism spectrum' as some kind of catch-all when boys disproportionately receive the standard ASD diagnosis. The reality is we are facing a different spectrum affecting both boys and girls; this is how you know it is not ASD, but a more pernicious Social Attention Deficit Disorder with proximate cause being monetized social media designed to disadvantage normal cognitive processes.
"Nobody has a very good explanation for why the industry continues this practice."
As a former federal regulator in a different, but just as contentious field, I can tell you. It is probably tied back to an event that occurred a long time ago (I can't say specifically which one), where someone was hurt by something, and the investigation revealed that some of the data had been mis-read, mis-understood, or just plain faked. And the public outcry went out that "We must make sure that this never happens again!"
So, all of the data now has to be checked, so that an error that happened in the past "never happens again". This is common across many fields where politicians and activists get involved, and is a common reaction by the bureaucrats, who are responsible for "make[ing] sure this never happens again."
Just thought of a good example - thalidomide, where there was no checks to see if it might cause birth defects in children before it was used on pregnant women. The data review was not done properly.
Who, here, is willing to have something bad happen to their spouse or child or other close family members, which turns out might have not happened if the data had been checked? Put your hands up.
I don't think that's what the review means. It sounded to me like they're manually checking every data point that the trial generated and submitted, not checking to make sure the trial didn't miss a category of people like "pregnant women".
I do otherwise agree completely that that's how regulations happen, and that it's bad that they happen that way.
Ideally there'd be cost-benefit analysis and other forms of thoughtful discussion on how this should work, but public panic, safetyism, and politicians being lawyers and not economists seem to obviate the possibility.
You are right - I did not elicidate ALL of the ways that data can be missed, misused, misapplied, misinterpreted, or otherwise tortured. My point was exactly what you said, someone made a mistake in dealing with data, somewhere, and it probably caused a giant scandal, which led to "never again".
Unfortunately, we still have lots of people in fields like economics and other social sciences, and in medicine, being "creative" with statistics. I used to have to deal with these sorts of issues, in regulating nuclear plants, and it is amazing what people do. Even in cases where there is no money or lives at stake, only personal pride and especially professional reputation, people do stuff that is very dodgy.
I will give you another example. In the nuclear business, QA is one of the several "great gods" that is applied in extreme ways, because if something breaks because it wasn't built correctly (to the specification), you can create a royal mess and lose a lot of money. The nuclear QA requirements came from the Nuclear Navy and a guy named Rickover, who INSISTED that everything be checked to make sure it was built to spec. And this practice transferred to the civilian nuclear industry and is the subject of current attack by people who want to be able to buy parts out of the bins at Home Depot or Harbor Freight.
I have a nephew who works for a company that makes parts for equipment in nuclear plants. I had a long talk with him, and he did not understand why the nuclear plants only buy equipment that was designed 50 years ago, and not the latest and greatest. His company makes a fortune selling old designs to nuclear plants, but they also spend a fortune verifying every dimension and every material test result, and the results of every performance test.
I explained to him that there are literally warehouses full of technical analyses of plants that use that equipment. And the performance specs for the equipment are used to determine the plant performance. If ANYTHING in those specs is changed, then someone has to go thru ALL of those analyses, and redo them to understand what has changed, why it has changed, and how that affects the analysis results. And if there is not enough margin in the design to accomodate the change, then they have to ask the government for permission to change it. Which is a real PITA. None of it is done with spreadsheets. Lots of it was done 50 years ago, on paper, with sliderules.
So, NO ONE WANTS to use the latest and greatest, because of the pain it would cause to make the changes. This is what it means to provide Quality Assurance.
A link to a new food poisoning event, with the FDA asking for a recall, which the company has announced it will not do. The comments give you some idea of how a problem that involves children and other innocent victims, develops. Note that there does not seem to be a firm accusation from the FDA - they are still studying what has happened - so it is all about "associations" and "linkages". The FDA has not issued any new regulations, yet.
The comments are the type that activate the "we have to make sure this never happens again" steamroller.
As a lawyer, strongly agree. If you drop this (especially as one of the first companies to do so) and get involved in litigation (almost inevitable) the opposing counsel is going to have the data compared or (seek via discovery) your decision to not compare and by the end of the case they're gonna lay it out before the decider of fact that you could have checked you inputs, but decided to merely sample them and these 35 instances of different data may have led to a different outcome in this case.
I do not intend to defend this, but to point out how it'll go down.
The only fix here is a law that fixes liability or provides for an alternative pathway+safeharbor. The FDA can say "you don't need to do this" all they want, it won't slow down a jury from smacking someone down one second.
Do you think a jury trial demand would survive a motion to dismiss if the procedural record showed FDA review before and after the fact that showed the trial's audit methods in compliance with these promulgated standards?
Those aren't even* Federal regulations. They're non-binding guidance docs. In the discovery phase if I saw a document where the drug company relied on those documents (stamped on each page with "Contains Nonbinding Recommendations") and changed their behavior based on them, it'd be "ok this is hot-harmful" (on a neutral (98% of docs)/hot harmful (1%)/hot helpful scale (1%)).
"Do you think a jury trial demand would survive a motion to dismiss if the procedural record showed FDA review before and after the fact that showed the trial's audit methods in compliance with these promulgated standards?"
I think this would be one of many arguments in a case, and it would not be a key claim. It's just one more thing you'd check off during discovery to see if you can sweeten the pot a bit by throwing it in.
Putting it another way: Someone is going to get injured and sue, right? Regardless of what you do, if you treat enough people eventually you're getting sued. Cutting out this kind of check (without a law or actual federal regulation promulgated via notice and comment rulemaking (which would be hella suspect here since the FDA can't change local court procedures or liability standards)) is just one of many things a good plaintiff's attorney is going to look for to swing a marginal case to a a good case or a good case to a strong case.
The base case would need to be good enough to get into discovery though.
To develop on this a bit more: "don't provide a sufficient alternative pathway?"
What I think you'd need is something like (in a law):
"A drug company that engages in source data verification via random sampling according to X method, Y method or a substantially similar method, upon demonstration of the sufficiency and compliance of their source data verification via random sampling, may not have claims made against them based in whole or in part upon deficiencies in their source verification absent a specific showing that said company either failed to appropriately perform the source verification process via random sampling or knew or should have known why it was inappropriate in this case."
Would need to be tuned of course, but something blocking people from arguing this at all (absent a specific showing of why). Yeah it'll lead to some litigation about what it means, but if this part of the process is really so expensive I think this kind of a shield would protect the companies against weak/meritless arguments in a way that a nonbinding guidance document doesn't.
* This sounds harsher than I intended, I mean it in terms of the hierarchy of "Supreme Law of the Land>Laws>Federal Regs>[Morass of non Federal Regulation rules/binding documents>Nonbinding agency opinions>I heard it from some regulator over coffee" it's pretty down there.
Thalidomide is not a "good example" of the motivation for Source Data Verification; the historical record is completely inconsistent with that claim.
First, thalidomide was never approved in the US because an FDA reviewer, Frances Kelsey, raised justified concerns about missing studies about the rate of metabolizing the drug. Kelsey was later appointed the Director of Investigative Drugs (later Director of Scientific Investigations) and now Phase I trials (*not* the huge expensive ones!) have been largely influenced by Kelsey's academically-inspired process, and this is relatively good.
What is not good is that public outcry -- "Who, here, is willing to have something bad happen to their spouse or child?" -- gave Senator Estes Kefauver political cover to pass the 1962 Kefauver Harris Amendment. Careful, reviewed safety studies had been required since 1938, when a drug *never tested in humans before commercial release* contained poisonous antifreeze as its solvent and killed more then a hundred people. The 30s were a wild west, and the 1938 Federal Food, Drug, and Cosmetics Act put an end to that. But Senator Kefauver had a different issue -- he believed that big pharma was making too much profit at the expense of the American people, for selling new, higher-priced versions of drugs that didn't work any better than the old versions. So in the wake of Frances Kelsey's correctly keeping thalidomide out of the US, he pushed through a new law requiring every drug to be tested for *efficacy* before approval -- those are the hugely expensive, $750mln+ trials that dominate the process. Thalidomide never had an issue with efficacy, and the Kefauver-Harris amendment was more a political response than a logically warranted legislation.
The changes that happened within the FDA to elevate Dr. Frances Kelsey's review methods as the agency-wide standard ensured that any future thalidomides would get caught the same way she caught thalidomide in 1970. The legislative response -- using exactly this kind of "your spouse or child" rhetoric -- added most of a billion dollars to the cost to develop a new drug and has led to dramatically fewer new medicines in the world today.
And by the way? I raise my hand as willing to have something bad happen to my spouse, which turns out might have not happened if the data had been checked. She takes a GLP-1 drug which is currently in Phase III trials by Eli Lilly, based on Phase I and Phase II studies showing it is safe, and has lower side effects than the already-approved drugs. The Phase III trial hasn't been reviewed, and the Source Data Verification hasn't been done, but she's looked at the non-Phase-III data, and I've looked at the data, and I think she is doing something better for her health, notwithstanding that she has a risk of having something happen to her, that might not have happened if the data had been checked.
I also don't buy the "who is willing..." rhetoric when it was the Obama-era FDA that said "yeah, we can check data integrity with organized audits to catch errors and fabrications without checking 100% source data verification at each site visit". I don't think anyone who thinks about the issue for five minutes would honestly believe that the 2013 FDA was too reckless in their 22-page report (https://www.fda.gov/media/116754/download), the 13-page follow-up report in 2019 (https://www.fda.gov/media/121479/download), or the EMA in their 13-page 2013 report (https://www.ema.europa.eu/en/documents/scientific-guideline/reflection-paper-risk-based-quality-management-clinical-trials_en.pdf). The agency has its issues, but these reports are sober and aware of the very real costs of wasteful methods, and cites real-world evidence that wasting time and resources on inefficient verification theater makes it less likely -- not more likely -- that real errors will be caught.
Thanks for explaining thalidomide. It came to mind after I did my original post, and I just did a quick check on it, but not a deep dive.
I don't think we are too far apart on this issue. My question was intended to point out what happens in the public-political sphere as a result of events like thalidomide, which leads to rules that some people might disagree with 60 years later. I think there is a term for it "Chesterton's Fence" - don't take down a fence until you understand why it was erected.
And I hope your wife does well - I take one of those drugs, and besides dropping my weight way down, it seems to have done away with a lot of aches and pains that I thought I would also have as I get old. They seem to be miracle drugs, at least for some people.
Same case for mental health in schools. "It's all because of poor Timmy back in 1997, why didn't anyone ask if he was doing ok? So tragic. Anyways, please consider and respond to the following questions: (1) How often do you have dark, morbid thoughts regarding your own death: (a) constantly, (b) every day, (c) most days, ..."
Agreed on both points. Within reason (and my wife is an extremely reasonable person) my wife's opinion of my attractiveness is by far the most important consideration.
I highly recommend the November 2025 Cheeky Pint podcast interview with Dave Ricks, CEO of Eli Lilly. Ricks went into a lot of detail on clinical trial costs, the R&D process, and drug pricing.
If I'm allowed a bit of self-promotion, I'll also put forward my May 2025 interview of Meri Beckwith, CEO of Chicago research organization of Lindus Health. We also get into the details of how the sausage gets made (and the costs get run up) at the actual level of clinical research practice:
Some inconsistency here - in response to a "medical freedom" X poster claiming that the FDA must suppress vaccine development, Zvi agrees with anither poster who says "and how does the existence of an mRNA vaccine affect your freedom?", calling it obvious nonsense. And yet, when the subject of lab-grown meat comes up, Zvi is sympathetic to (though not strictly in agreement with) those who oppose it because they fear that consumption of superior meat from animals will be outlawed.
Surely the "medical freedom" people are concerned that once such vaccines are available, they will be forced into taking them. From their perspective, it makes sense to fight against these vaccines every step of the way so that they can't ever arrive at a point where they can be mandated.
Personally, I think both situations are very silly and we should be supporting the development of both, but I'd at least like to see a consistent viewpoint from Zvi even if takes the other view.
Anyone know the name of the study in the 'Walk it Off' section? It looks like some doctor just learned about splines and was really enthusiastic to apply them.
Wasn't Jim McKelvey trying to take an engineer's/startup approach to trying to crush the cost of drug trials? Haven't connected with him lately, but forwarded him this.
On observational studies' estimates of dosage effects from exercise:
There are obvious selection effects. Your capacity and inclination to move is caused by health. There's also harm; exercise depletes your resources, which eventually compromises health when it overcomes your recovery capacity. Since the forwards causality is complex, and the selection effects are straightforward, the early minimum in these graphs is mostly information about the point at which direct harm starts to become more important than selection effects on the margin; seems like at the right end there's further selection for elite athlete levels of health in many cases.
An exercise type selection effect: Swimming is frequently chosen as a low-impact exercise tolerated well by sick and elderly people.
The main mechanistic argument I've seen for exercise type and quantity is Chris Masterjohn's argument that maintaining basic competence at many movements is helpful both for local soft tissue health and for injury prevention:
ETA: My opinion is possibly a steelman of Masterjohn's "local soft tissue" mechanism argument, influenced by conversations with Valentin Rozlomii (at visceralcure.com ) and personal experience of his work. The gist is that using or otherwise perturbing a muscle improves circulation and therefore oxygenation in not only that muscle but organs adjacent to it. In addition, if a muscle is chronically tight because weak and underused, it can chronically impinge on a nerve, causing referred pain and chronic tension elsewhere, which then impairs circulation and therefore oxygenation around the referred site.
The main mechanistic argument? Am I missing something, or why do not the metabolic and cardiorespiratory and muscular benefits count as a really solid argument?
The cardiorespiratory arguments are a particularly important *example* of functional fitness, but don't really explain the full data; they'd predict diminishing returns for relatively pure cardio, and not much benefit to anything else. The muscular benefits are a lower level of explanation and aren't sufficient to establish a benefit unless you already have some specific performance goals. The metabolic argument often seems kind of backwards and confused to me; "poor glucose handling" can mean "low glucose requirement," in which case, good news, you can save on food; maybe you mean something much better organized, in which case please point me to the argument.
Re: "45 minutes a day of walking gives you most of the all-cause mortality benefits of exercise", I think that the chart implies more like 10-15 minutes of walking. The X-axis is "MET hours/week", not hours/week. ChatGPT suggests that the MET score of walking used by the paper was probably 2.5-4.5 (it varies with pace). So I think that the walking chart is just indicating that there's a penalty for sedentary lifestyle, more than saying anything about the benefits of non-trivial levels of walking.
For some reason those studying exercise love to use wild units like "MET hours" or "calories expended" instead of just "minutes", this despite the fact that time is overwhelmingly the most common reason that people give for not exercising. When comparing two exercises, they never just give the two groups different exercises for the same amount of time, they have to be "energy matched" or whatever.
It's a free country and people should be able to purchase plastic surgeries if they want...but, man, after helping someone recover from a pretty gnarly one that involved stitches from cutting out bone...did she look better after the (long, painful, frankly kinda disgusting) recovery period? Yeah, she did, went from maybe a 6 to a 7.5. Was it "worth it" in the long run though, versus cultivating a better self-image, improving other markers of health, spending the same amount of money to get a non-trashy wardrobe? (Not having years of aftercare and follow-up procedures?) I dunno, you tell me. Dangerous path to go down, caveat emptor.
On the other hand, this same freedom lets me sign up for experimental trials off a mere Substack post, and I think that's great. Nothing acute or life-threatening, just a human-safety dosage trial, but it was kinda cool to merely...fill out a basic questionnaire, get some pills in the mail, and off to the races. No involvement with government, no hassles going through a big institution (well, not on my end anyway), just direct For Science(tm) to the people. I hope if I ever end up at the end of my rope, then by that time, the legal landscape won't be as hostile as what that one Substacker with terminal cancer faced. The one who died trying to right-to-try experimental Phase 3 trials and wrote about it up to the end. Wish I could recall the name. JD, JB something?
I'm confused. I don't expect anything to be a sole explanation. More specifically, whilst I expect cardio to show diminishing (but slowly!) returns, so does everything? I think this is by far the best established empirically.
And the muscle benefits are hard to extract cleanly, but mechanistically, they are very intuitive: they partly englobe your joints one, but also include less injuries and longer automony which are both heavily associated with longer lives.
Moderna spent $750 million on a study involving 41,000 people after the FDA approved the design. However, the goalposts were moved after the money had been spent. Unpredictable approvals are more costly for innovation than slow ones.
New 2026 ACC/AHA guidelines came out with instructions on how to identify and treat dyslipidemia (which includes but is not restricted to hyperlipidemia) earlier than is currently being done, which aligns mainstream medical practice more with what the lipid hawks are saying. If these guidelines are properly implemented it could be very impactful towards lowering ASCVD.
> It doesn’t make any sense for bicycling and swimming to have about zero effect
I'd guess we're looking at the combination of a reduction in health-related mortality (saturating with time spent cycling/swimming the same way that of walking does) and an increase in accident-related mortality (traffic collisions/drowning respectively) roughly proportional to time spent cycling/swimming.
That was my thought as well.
Suboptimal social behaviors are being misclassified as 'on the Autism spectrum' as some kind of catch-all when boys disproportionately receive the standard ASD diagnosis. The reality is we are facing a different spectrum affecting both boys and girls; this is how you know it is not ASD, but a more pernicious Social Attention Deficit Disorder with proximate cause being monetized social media designed to disadvantage normal cognitive processes.
"Nobody has a very good explanation for why the industry continues this practice."
As a former federal regulator in a different, but just as contentious field, I can tell you. It is probably tied back to an event that occurred a long time ago (I can't say specifically which one), where someone was hurt by something, and the investigation revealed that some of the data had been mis-read, mis-understood, or just plain faked. And the public outcry went out that "We must make sure that this never happens again!"
So, all of the data now has to be checked, so that an error that happened in the past "never happens again". This is common across many fields where politicians and activists get involved, and is a common reaction by the bureaucrats, who are responsible for "make[ing] sure this never happens again."
Just thought of a good example - thalidomide, where there was no checks to see if it might cause birth defects in children before it was used on pregnant women. The data review was not done properly.
Who, here, is willing to have something bad happen to their spouse or child or other close family members, which turns out might have not happened if the data had been checked? Put your hands up.
I don't think that's what the review means. It sounded to me like they're manually checking every data point that the trial generated and submitted, not checking to make sure the trial didn't miss a category of people like "pregnant women".
I do otherwise agree completely that that's how regulations happen, and that it's bad that they happen that way.
Ideally there'd be cost-benefit analysis and other forms of thoughtful discussion on how this should work, but public panic, safetyism, and politicians being lawyers and not economists seem to obviate the possibility.
You are right - I did not elicidate ALL of the ways that data can be missed, misused, misapplied, misinterpreted, or otherwise tortured. My point was exactly what you said, someone made a mistake in dealing with data, somewhere, and it probably caused a giant scandal, which led to "never again".
Unfortunately, we still have lots of people in fields like economics and other social sciences, and in medicine, being "creative" with statistics. I used to have to deal with these sorts of issues, in regulating nuclear plants, and it is amazing what people do. Even in cases where there is no money or lives at stake, only personal pride and especially professional reputation, people do stuff that is very dodgy.
I will give you another example. In the nuclear business, QA is one of the several "great gods" that is applied in extreme ways, because if something breaks because it wasn't built correctly (to the specification), you can create a royal mess and lose a lot of money. The nuclear QA requirements came from the Nuclear Navy and a guy named Rickover, who INSISTED that everything be checked to make sure it was built to spec. And this practice transferred to the civilian nuclear industry and is the subject of current attack by people who want to be able to buy parts out of the bins at Home Depot or Harbor Freight.
I have a nephew who works for a company that makes parts for equipment in nuclear plants. I had a long talk with him, and he did not understand why the nuclear plants only buy equipment that was designed 50 years ago, and not the latest and greatest. His company makes a fortune selling old designs to nuclear plants, but they also spend a fortune verifying every dimension and every material test result, and the results of every performance test.
I explained to him that there are literally warehouses full of technical analyses of plants that use that equipment. And the performance specs for the equipment are used to determine the plant performance. If ANYTHING in those specs is changed, then someone has to go thru ALL of those analyses, and redo them to understand what has changed, why it has changed, and how that affects the analysis results. And if there is not enough margin in the design to accomodate the change, then they have to ask the government for permission to change it. Which is a real PITA. None of it is done with spreadsheets. Lots of it was done 50 years ago, on paper, with sliderules.
So, NO ONE WANTS to use the latest and greatest, because of the pain it would cause to make the changes. This is what it means to provide Quality Assurance.
https://arstechnica.com/health/2026/03/fda-links-raw-cheese-to-outbreak-makers-100-disagree-refuse-recall
A link to a new food poisoning event, with the FDA asking for a recall, which the company has announced it will not do. The comments give you some idea of how a problem that involves children and other innocent victims, develops. Note that there does not seem to be a firm accusation from the FDA - they are still studying what has happened - so it is all about "associations" and "linkages". The FDA has not issued any new regulations, yet.
The comments are the type that activate the "we have to make sure this never happens again" steamroller.
As a lawyer, strongly agree. If you drop this (especially as one of the first companies to do so) and get involved in litigation (almost inevitable) the opposing counsel is going to have the data compared or (seek via discovery) your decision to not compare and by the end of the case they're gonna lay it out before the decider of fact that you could have checked you inputs, but decided to merely sample them and these 35 instances of different data may have led to a different outcome in this case.
I do not intend to defend this, but to point out how it'll go down.
The only fix here is a law that fixes liability or provides for an alternative pathway+safeharbor. The FDA can say "you don't need to do this" all they want, it won't slow down a jury from smacking someone down one second.
Do you think that https://www.fda.gov/media/116754/download (2013 FDA guidance) and https://www.fda.gov/media/121479/download (2023 further guidance) don't provide a sufficient alternative pathway?
Do you think a jury trial demand would survive a motion to dismiss if the procedural record showed FDA review before and after the fact that showed the trial's audit methods in compliance with these promulgated standards?
Those aren't even* Federal regulations. They're non-binding guidance docs. In the discovery phase if I saw a document where the drug company relied on those documents (stamped on each page with "Contains Nonbinding Recommendations") and changed their behavior based on them, it'd be "ok this is hot-harmful" (on a neutral (98% of docs)/hot harmful (1%)/hot helpful scale (1%)).
"Do you think a jury trial demand would survive a motion to dismiss if the procedural record showed FDA review before and after the fact that showed the trial's audit methods in compliance with these promulgated standards?"
I think this would be one of many arguments in a case, and it would not be a key claim. It's just one more thing you'd check off during discovery to see if you can sweeten the pot a bit by throwing it in.
Putting it another way: Someone is going to get injured and sue, right? Regardless of what you do, if you treat enough people eventually you're getting sued. Cutting out this kind of check (without a law or actual federal regulation promulgated via notice and comment rulemaking (which would be hella suspect here since the FDA can't change local court procedures or liability standards)) is just one of many things a good plaintiff's attorney is going to look for to swing a marginal case to a a good case or a good case to a strong case.
The base case would need to be good enough to get into discovery though.
To develop on this a bit more: "don't provide a sufficient alternative pathway?"
What I think you'd need is something like (in a law):
"A drug company that engages in source data verification via random sampling according to X method, Y method or a substantially similar method, upon demonstration of the sufficiency and compliance of their source data verification via random sampling, may not have claims made against them based in whole or in part upon deficiencies in their source verification absent a specific showing that said company either failed to appropriately perform the source verification process via random sampling or knew or should have known why it was inappropriate in this case."
Would need to be tuned of course, but something blocking people from arguing this at all (absent a specific showing of why). Yeah it'll lead to some litigation about what it means, but if this part of the process is really so expensive I think this kind of a shield would protect the companies against weak/meritless arguments in a way that a nonbinding guidance document doesn't.
* This sounds harsher than I intended, I mean it in terms of the hierarchy of "Supreme Law of the Land>Laws>Federal Regs>[Morass of non Federal Regulation rules/binding documents>Nonbinding agency opinions>I heard it from some regulator over coffee" it's pretty down there.
Thalidomide is not a "good example" of the motivation for Source Data Verification; the historical record is completely inconsistent with that claim.
First, thalidomide was never approved in the US because an FDA reviewer, Frances Kelsey, raised justified concerns about missing studies about the rate of metabolizing the drug. Kelsey was later appointed the Director of Investigative Drugs (later Director of Scientific Investigations) and now Phase I trials (*not* the huge expensive ones!) have been largely influenced by Kelsey's academically-inspired process, and this is relatively good.
What is not good is that public outcry -- "Who, here, is willing to have something bad happen to their spouse or child?" -- gave Senator Estes Kefauver political cover to pass the 1962 Kefauver Harris Amendment. Careful, reviewed safety studies had been required since 1938, when a drug *never tested in humans before commercial release* contained poisonous antifreeze as its solvent and killed more then a hundred people. The 30s were a wild west, and the 1938 Federal Food, Drug, and Cosmetics Act put an end to that. But Senator Kefauver had a different issue -- he believed that big pharma was making too much profit at the expense of the American people, for selling new, higher-priced versions of drugs that didn't work any better than the old versions. So in the wake of Frances Kelsey's correctly keeping thalidomide out of the US, he pushed through a new law requiring every drug to be tested for *efficacy* before approval -- those are the hugely expensive, $750mln+ trials that dominate the process. Thalidomide never had an issue with efficacy, and the Kefauver-Harris amendment was more a political response than a logically warranted legislation.
The changes that happened within the FDA to elevate Dr. Frances Kelsey's review methods as the agency-wide standard ensured that any future thalidomides would get caught the same way she caught thalidomide in 1970. The legislative response -- using exactly this kind of "your spouse or child" rhetoric -- added most of a billion dollars to the cost to develop a new drug and has led to dramatically fewer new medicines in the world today.
And by the way? I raise my hand as willing to have something bad happen to my spouse, which turns out might have not happened if the data had been checked. She takes a GLP-1 drug which is currently in Phase III trials by Eli Lilly, based on Phase I and Phase II studies showing it is safe, and has lower side effects than the already-approved drugs. The Phase III trial hasn't been reviewed, and the Source Data Verification hasn't been done, but she's looked at the non-Phase-III data, and I've looked at the data, and I think she is doing something better for her health, notwithstanding that she has a risk of having something happen to her, that might not have happened if the data had been checked.
I also don't buy the "who is willing..." rhetoric when it was the Obama-era FDA that said "yeah, we can check data integrity with organized audits to catch errors and fabrications without checking 100% source data verification at each site visit". I don't think anyone who thinks about the issue for five minutes would honestly believe that the 2013 FDA was too reckless in their 22-page report (https://www.fda.gov/media/116754/download), the 13-page follow-up report in 2019 (https://www.fda.gov/media/121479/download), or the EMA in their 13-page 2013 report (https://www.ema.europa.eu/en/documents/scientific-guideline/reflection-paper-risk-based-quality-management-clinical-trials_en.pdf). The agency has its issues, but these reports are sober and aware of the very real costs of wasteful methods, and cites real-world evidence that wasting time and resources on inefficient verification theater makes it less likely -- not more likely -- that real errors will be caught.
Thanks for explaining thalidomide. It came to mind after I did my original post, and I just did a quick check on it, but not a deep dive.
I don't think we are too far apart on this issue. My question was intended to point out what happens in the public-political sphere as a result of events like thalidomide, which leads to rules that some people might disagree with 60 years later. I think there is a term for it "Chesterton's Fence" - don't take down a fence until you understand why it was erected.
And I hope your wife does well - I take one of those drugs, and besides dropping my weight way down, it seems to have done away with a lot of aches and pains that I thought I would also have as I get old. They seem to be miracle drugs, at least for some people.
Same case for mental health in schools. "It's all because of poor Timmy back in 1997, why didn't anyone ask if he was doing ok? So tragic. Anyways, please consider and respond to the following questions: (1) How often do you have dark, morbid thoughts regarding your own death: (a) constantly, (b) every day, (c) most days, ..."
"Also the person who most has to look at you will be you, so your preferences matter."
Incorrect. I am 100% sure my wife looks at me more than I look in the mirror. So maybe her preferences matter more. Also true on clothing.
Also plausible your wife cares more about how you look than you do (and not just because she sees you more often than you look in the mirror)
Very plausible, even propable.
Agreed on both points. Within reason (and my wife is an extremely reasonable person) my wife's opinion of my attractiveness is by far the most important consideration.
But don't tell your wife you care more about how she looks than she does
Podcast episode for this post: https://dwatvpodcast.substack.com/p/medical-roundup-7
I highly recommend the November 2025 Cheeky Pint podcast interview with Dave Ricks, CEO of Eli Lilly. Ricks went into a lot of detail on clinical trial costs, the R&D process, and drug pricing.
Video link and full transcript here: https://cheekypint.substack.com/p/dave-ricks-ceo-of-eli-lilly-on-glp
If I'm allowed a bit of self-promotion, I'll also put forward my May 2025 interview of Meri Beckwith, CEO of Chicago research organization of Lindus Health. We also get into the details of how the sausage gets made (and the costs get run up) at the actual level of clinical research practice:
https://developmentandresearch.bio/meri-beckwith/ (and especially https://developmentandresearch.bio/meri-beckwith/#decentralized-trials about the source-data-verification issue).
Some inconsistency here - in response to a "medical freedom" X poster claiming that the FDA must suppress vaccine development, Zvi agrees with anither poster who says "and how does the existence of an mRNA vaccine affect your freedom?", calling it obvious nonsense. And yet, when the subject of lab-grown meat comes up, Zvi is sympathetic to (though not strictly in agreement with) those who oppose it because they fear that consumption of superior meat from animals will be outlawed.
Surely the "medical freedom" people are concerned that once such vaccines are available, they will be forced into taking them. From their perspective, it makes sense to fight against these vaccines every step of the way so that they can't ever arrive at a point where they can be mandated.
Personally, I think both situations are very silly and we should be supporting the development of both, but I'd at least like to see a consistent viewpoint from Zvi even if takes the other view.
That is a good argument for not taking anyone dumb enough to not see the difference seriously
Anyone know the name of the study in the 'Walk it Off' section? It looks like some doctor just learned about splines and was really enthusiastic to apply them.
https://bmjmedicine.bmj.com/content/5/1/e001513
Thank you kindly!
Wasn't Jim McKelvey trying to take an engineer's/startup approach to trying to crush the cost of drug trials? Haven't connected with him lately, but forwarded him this.
On observational studies' estimates of dosage effects from exercise:
There are obvious selection effects. Your capacity and inclination to move is caused by health. There's also harm; exercise depletes your resources, which eventually compromises health when it overcomes your recovery capacity. Since the forwards causality is complex, and the selection effects are straightforward, the early minimum in these graphs is mostly information about the point at which direct harm starts to become more important than selection effects on the margin; seems like at the right end there's further selection for elite athlete levels of health in many cases.
An exercise type selection effect: Swimming is frequently chosen as a low-impact exercise tolerated well by sick and elderly people.
The main mechanistic argument I've seen for exercise type and quantity is Chris Masterjohn's argument that maintaining basic competence at many movements is helpful both for local soft tissue health and for injury prevention:
https://chrismasterjohnphd.substack.com/p/what-everyone-should-be-doing-for
See also Masterjohn's appearance on Joe Rogan episode 2420: https://open.spotify.com/episode/0ME2AXkR3e6nQQYMjbewjt
He cites as empirical support the longevity advantage of elite gymnasts vs other athletes: https://pmc.ncbi.nlm.nih.gov/articles/PMC11979035/
I made a similar argument here on Twitter: https://x.com/ben_r_hoffman/status/2015616670450987316
ETA: My opinion is possibly a steelman of Masterjohn's "local soft tissue" mechanism argument, influenced by conversations with Valentin Rozlomii (at visceralcure.com ) and personal experience of his work. The gist is that using or otherwise perturbing a muscle improves circulation and therefore oxygenation in not only that muscle but organs adjacent to it. In addition, if a muscle is chronically tight because weak and underused, it can chronically impinge on a nerve, causing referred pain and chronic tension elsewhere, which then impairs circulation and therefore oxygenation around the referred site.
The main mechanistic argument? Am I missing something, or why do not the metabolic and cardiorespiratory and muscular benefits count as a really solid argument?
The cardiorespiratory arguments are a particularly important *example* of functional fitness, but don't really explain the full data; they'd predict diminishing returns for relatively pure cardio, and not much benefit to anything else. The muscular benefits are a lower level of explanation and aren't sufficient to establish a benefit unless you already have some specific performance goals. The metabolic argument often seems kind of backwards and confused to me; "poor glucose handling" can mean "low glucose requirement," in which case, good news, you can save on food; maybe you mean something much better organized, in which case please point me to the argument.
Re: "45 minutes a day of walking gives you most of the all-cause mortality benefits of exercise", I think that the chart implies more like 10-15 minutes of walking. The X-axis is "MET hours/week", not hours/week. ChatGPT suggests that the MET score of walking used by the paper was probably 2.5-4.5 (it varies with pace). So I think that the walking chart is just indicating that there's a penalty for sedentary lifestyle, more than saying anything about the benefits of non-trivial levels of walking.
https://bmjmedicine.bmj.com/content/bmjmed/5/1/e001513/DC1/embed/inline-supplementary-material-1.pdf
on page 3 it says they used a score of 3
For some reason those studying exercise love to use wild units like "MET hours" or "calories expended" instead of just "minutes", this despite the fact that time is overwhelmingly the most common reason that people give for not exercising. When comparing two exercises, they never just give the two groups different exercises for the same amount of time, they have to be "energy matched" or whatever.
It's a free country and people should be able to purchase plastic surgeries if they want...but, man, after helping someone recover from a pretty gnarly one that involved stitches from cutting out bone...did she look better after the (long, painful, frankly kinda disgusting) recovery period? Yeah, she did, went from maybe a 6 to a 7.5. Was it "worth it" in the long run though, versus cultivating a better self-image, improving other markers of health, spending the same amount of money to get a non-trashy wardrobe? (Not having years of aftercare and follow-up procedures?) I dunno, you tell me. Dangerous path to go down, caveat emptor.
On the other hand, this same freedom lets me sign up for experimental trials off a mere Substack post, and I think that's great. Nothing acute or life-threatening, just a human-safety dosage trial, but it was kinda cool to merely...fill out a basic questionnaire, get some pills in the mail, and off to the races. No involvement with government, no hassles going through a big institution (well, not on my end anyway), just direct For Science(tm) to the people. I hope if I ever end up at the end of my rope, then by that time, the legal landscape won't be as hostile as what that one Substacker with terminal cancer faced. The one who died trying to right-to-try experimental Phase 3 trials and wrote about it up to the end. Wish I could recall the name. JD, JB something?
I'm confused. I don't expect anything to be a sole explanation. More specifically, whilst I expect cardio to show diminishing (but slowly!) returns, so does everything? I think this is by far the best established empirically.
And the muscle benefits are hard to extract cleanly, but mechanistically, they are very intuitive: they partly englobe your joints one, but also include less injuries and longer automony which are both heavily associated with longer lives.
Moderna spent $750 million on a study involving 41,000 people after the FDA approved the design. However, the goalposts were moved after the money had been spent. Unpredictable approvals are more costly for innovation than slow ones.