Thanks for this - I now feel at least 50% more confident in my comment the other day on the monthly links post point out that "a key issue in trial design is that lots of people want to be in a study, and simply lie to the doctors"
In your conclusion, you note this isn't the system you'd like it to be. Is there a description by you or others elsewhere of a possibly superior alternative?
If the core issue is misaligned incentives, could a couple of small tests move the needle without a policy fight?
Example: run an arm where pay is guaranteed for non-serious AEs and there’s a small, fixed “diligence” stipend for timely logs, then compared AE capture, PK/PD outliers, and dropouts to status quo. I'd think reporting would improve when volunteers aren’t worried about losing compensation.
Example: a simple cross-site operational snapshot -- Is dosing order randomized or first-come; what’s the alternate pay policy; how broad are exclusions; how are washouts enforced. Then correlate those knobs with screen-fail rates, AE capture, and protocol deviations. Even a modest, pre-registered audit might show what actually matters.
On reforms, would sponsors/CROs be open to a meaningful day-rate for alternates plus transparent randomized queueing to reduce the “race to screen”? Could contraception language be tuned to mechanism and half-life (ICH/EMA style) with a clear lay rationale? And is there room for a lightweight healthy-volunteer registry via central IRBs or sponsor contracts ---just enough to prevent dual-enrollment and stacked washouts - without waiting on federal law?
Have versions of these have been tried and failed? They seem testable and pleasantly boring -- the best kind of fix.
I'm not sure what you mean by "pay is guaranteed for non-serious AEs." If you mean that participants are guaranteed to be paid for their participation in the study, I don't think that would help much if they're worried about being ruled out of future studies. If you mean that participants would be paid for reporting AEs, that seems like a perverse incentive guaranteed to result in an explosion of AE reports. The people conducting the study don't know the true rate of AEs they're trying to capture (otherwise they wouldn't have to try to study it!) and they have to be careful that they don't set up incentives that result in overshooting the true rate as well as undershooting it.
Oh, I see. I’m not proposing pay-per-AE, just AE-neutral compensation (base pay plus a small fixed compliance stipend regardless of what’s reported) plus a future-eligibility guardrail so routine grade-1/2, expected AEs don’t automatically hurt eligibility unless a clinician flags a real contraindication. Pilot that and track diary timeliness, concordance with labs/PK, and dropouts vs. status quo.
That seems worth testing, but this strikes me as something of a coordination problem. If research participants systematically expect clinics not to be trustworthy to cooperate with (because from their perspective, they're usually not,) then a single clinic guaranteeing that reporting AEs won't hurt eligibility at that clinic may not see much difference, because they're still working with the same participant population, whose expectations are normed on all the other clinics.
If one clinic tests a bunch of variations in procedure, and sees no differences in results, it might be a mistake to conclude that there would be no differences in results if changes were propagated across the whole clinical landscape, and there was enough time for those changes to be reflected in the expectations of participants.
Totally agree on the coordination problem. A single site won’t reset priors. Would it make sense to treat the sponsor/CRO (or a central IRB) as the unit of change and run a staggered multi-site pilot where several clinics adopt AE-neutral pay plus a grade-1/2 eligibility guardrail with a dated, public policy? Start dates randomized by site; evaluate with a simple diff-in-diff on diary timeliness, lab/PK concordance, and dropouts.
To make the commitment credible to volunteers, what about a “portable guarantee” across participating sites, i.e. a benign AE reported at Site A can’t be used to exclude at Sites B/C unless a clinician adjudicates a real contraindication...and implement the same workflow via shared ePRO/payment vendors so it’s visible and consistent?
If that still shows no movement, then the issue likely sits deeper than pay/eligibility fear; if it does, you’ve got a template ready to scale.
I enjoyed this review, it's a pretty niche topic but one I'm familiar with because ~20 years ago I participated in a handful of these types of clinical trials and now I'm helping to plan them.
I've done about a dozen or so Phase 1 trials, and two groups of people I would say you missed in your "who does these trials" sections is:
1) Backpackers, and
2) Foreign students
I think the whole 'no experience required' part attracts less "shady" characters than you seem to have encountered, and more just people who need to make money asap and can't go via the conventional route. I did my trials mostly during COVID in one of the most locked down cities in the world (Melbourne) so that might have contributed.
By contrast I don't think I ever actually encountered personally anyone who seemed all that shady to me. I heard a few stories of people shagging in the bathrooms but beyond that the overall vibe of the people seemed mostly extraverted and friendly.
The whole 'lie at every step of the application process' rings incredibly true though. I have recruited many family members into doing these trials and the one piece of advice I always give is, no matter the question, if it's about having a health problem always say no. They once asked if I had ever smoked, and wanting not to sound too unbelievable I said I had vaped once a few months ago. That was written down as "smoker" for literally years afterwards.
This could just be the fact that you were in Australia? I assume the author was in the US. I would be very surprised if backpackers were a significant portion of this population, as people mostly do not do that kind of urban transit-hopping backpacking in the US in the first place. Students I am sure are a significant component though. If it's at all similar to the population who donates plasma, sites near major universities are likely mostly students and then mostly 'shady characters' at all other sites. But students also likely do not do the cross-country professional research participant who tries to squeeze in as many as possible.
I knew people in college who participated in trials at a company called Pharmaco. My roommate got her wisdom teeth out for free this way.
Pharmaco was located a few thousand feet from the mall and there was a little joke about a path beaten between the two by all the coeds.
Fast forward to being old, and the women I know (only women) who do these things tend to be retired or laid-off and interested in “gigs” of all sorts: the Census, poll work, substitute teaching.
One of my girlfriends, a freelancer, at sixty was also arduously climbing up two flights of stairs to deliver beer and Jell-O cups to kids too lazy to leave their apartments.
Boomer women were hard workers!
Some people, by no means suffering for funds, yet get antsy without any sort of paycheck.
I'm reminded of an interview with Leonard Nimoy's son about his dad's advice that the answer to every question you are asked during an audition interview is "Yes," which went something a little like this:
So, Leonard, we'd love for you to play Crazy Horse in our Custer's Last Stand movie. Can we ask you just a few questions?
Yes.
You know how to ride a horse?
Yes.
And you are of course an expert archer?
Yes.
And you are an expert archer on horseback?
Yes.
And you can shout Sioux curses with a perfect accent while shooting an arrow on horseback at George Armstrong Custer?
Yes.
After all, Dad used to say, they won't start filming for a few weeks, so you can always take lessons.
re "Whenever I’ve been tempted to develop a low opinion of their judgment, I’ve had to temper that with the knowledge that many of these people have apparently accumulated much more disposable income in the process than I have."
Sounds a good system to me! They get to study a much more realistic cross-section of the population than they would if nobody lied. It catches all the major problems (ie the ones that they test for and the ones you can't hide). As Douglas Adams said "Any remaining problems are your own."
And, as a bonus, the company is protected against being sued.
And you're providing gainful to people with criminal convictions who otherwise struggle to get jobs.
Maybe we should just adjust our vocabulary and consistently talk about "phase I subjects" instead of "healthy subjects" when referencing these studies. Some experts already do that (though I'm not sure if it's something a reviewer would flag), but I don't think it's filtered into public discourse yet.
It might be even better if we started attaching a few boilerplate words for those new to the topic:
"No side effects were observed in phase I subjects dosed with 5 mg of exampeline (Examplechuck et al., 2025). However, phase I subjects are known to differ from the healthy general population (Examplovich and Chang, 2022)."
Yes, what I'm wondering here is, how much would these flaws in safety trials cause problems if the US were to switch back to a "safety trials only" system as is often suggested?
A delightful accretion of safetyism, the cynic in me is thoroughly amused. But also, it's impressive that even an arrangement this corrupt and perverse could still be be made decently workable with enough effort.
Will the author stay anonymous? Or will the author reveal themselves and vault into the meta-business of teaching folks how to participate in Phase I trials!?
The two people who distrust white doctors reminded me of a Braver Angels discussion with two vax skeptics in 2021. One was white and MAGA, the other was black and BLM. In case you don't recall, Biden had prioritized getting black and old people vaccinated first because they had worse outcomes in aggregate.
The source of the black guy's skepticism emerged when he said "when has the government EVER prioritized the health of black people?"
The bizarre thing is that a lot of times the reference point in articles (which are now on repeat) about black mistrust of the medical establishment, is the Tuskegee syphilis study - which was itself a prioritizing of the health of black people, among whom syphilis was concentrated, at least in the South. However dumb was the worship of the study later on, to obliviously or callously continue it when treatment was established, the whole thing would not be a millstone around our necks forever and ever if there had been no such original “prioritization”.
re "Do they have any reason to think that ADHD might be associated with an increased risk? I asked, and their answer was, not at all, but better safe than sorry."
Ozempic seems to have somewhat rare but very severe side effects (that include severe depression and suicidal ideation *caused* by the drug) specifically in autistic people, and autism tends to be strongly correlated with ADHD, so there's that.
I kinda feel like you'd rather find that out in a clinical setting than patients at home. But clearly that goes entirely against recruiting only healthy subjects.
The link for “the asshole filter” goes to a copy of another blogger’s post. They disclose that they copied it, but don’t say they had permission to publish a copy. It seems… on brand? Anyway, I think it would be better to link to the original:
I knew sometime who participated in these studies and he said the participants would bite the pills so they knew who was getting placebo or the real thing based on the taste.
Very interesting read, thanks! I wonder if it would help if there was a country-wide database of people who participated and everyone could participate at most once in life
Yeah that’s the one thing I was seeing that could change the incentives - make the pay higher but have a nationwide tracking system that locks you out from any others for a while.
Phase I = adversarial equilibrium: participants exploit for money, clinics shield from liability, pharma tolerates distortions, yet the system limps forward because downstream filters mop up worst failures.
This is an interesting analysis of why “paying people to do medicine-adjacent things” (eg: compensated donations) can get so much pushback: behind knee-jerk “ick”-type reactions, there’s maybe some sense that the interaction between “people optimizing for getting paid,” the typical mild dysfunction you’d see in any large organization, and the sheer stakes of healthcare, can produce quite the perverse and unanticipated outcome.
I'm suddenly imagining a dystopia in which these problems are dealt with by drafting participants the way we assign people to jury duty. Of course that would be entirely against medical ethics as we know them, but this is a dystopia. (Whether drafting people to jury duty should be against legal ethics is another question.)
Looks like this got posted as its own comment rather than under what it's replying to.
thanks, i'll delete.
Thanks for this - I now feel at least 50% more confident in my comment the other day on the monthly links post point out that "a key issue in trial design is that lots of people want to be in a study, and simply lie to the doctors"
https://www.astralcodexten.com/p/links-for-september-2025/comment/152285966
Interesting.
In your conclusion, you note this isn't the system you'd like it to be. Is there a description by you or others elsewhere of a possibly superior alternative?
If the core issue is misaligned incentives, could a couple of small tests move the needle without a policy fight?
Example: run an arm where pay is guaranteed for non-serious AEs and there’s a small, fixed “diligence” stipend for timely logs, then compared AE capture, PK/PD outliers, and dropouts to status quo. I'd think reporting would improve when volunteers aren’t worried about losing compensation.
Example: a simple cross-site operational snapshot -- Is dosing order randomized or first-come; what’s the alternate pay policy; how broad are exclusions; how are washouts enforced. Then correlate those knobs with screen-fail rates, AE capture, and protocol deviations. Even a modest, pre-registered audit might show what actually matters.
On reforms, would sponsors/CROs be open to a meaningful day-rate for alternates plus transparent randomized queueing to reduce the “race to screen”? Could contraception language be tuned to mechanism and half-life (ICH/EMA style) with a clear lay rationale? And is there room for a lightweight healthy-volunteer registry via central IRBs or sponsor contracts ---just enough to prevent dual-enrollment and stacked washouts - without waiting on federal law?
Have versions of these have been tried and failed? They seem testable and pleasantly boring -- the best kind of fix.
I'm not sure what you mean by "pay is guaranteed for non-serious AEs." If you mean that participants are guaranteed to be paid for their participation in the study, I don't think that would help much if they're worried about being ruled out of future studies. If you mean that participants would be paid for reporting AEs, that seems like a perverse incentive guaranteed to result in an explosion of AE reports. The people conducting the study don't know the true rate of AEs they're trying to capture (otherwise they wouldn't have to try to study it!) and they have to be careful that they don't set up incentives that result in overshooting the true rate as well as undershooting it.
Oh, I see. I’m not proposing pay-per-AE, just AE-neutral compensation (base pay plus a small fixed compliance stipend regardless of what’s reported) plus a future-eligibility guardrail so routine grade-1/2, expected AEs don’t automatically hurt eligibility unless a clinician flags a real contraindication. Pilot that and track diary timeliness, concordance with labs/PK, and dropouts vs. status quo.
That seems worth testing, but this strikes me as something of a coordination problem. If research participants systematically expect clinics not to be trustworthy to cooperate with (because from their perspective, they're usually not,) then a single clinic guaranteeing that reporting AEs won't hurt eligibility at that clinic may not see much difference, because they're still working with the same participant population, whose expectations are normed on all the other clinics.
If one clinic tests a bunch of variations in procedure, and sees no differences in results, it might be a mistake to conclude that there would be no differences in results if changes were propagated across the whole clinical landscape, and there was enough time for those changes to be reflected in the expectations of participants.
Totally agree on the coordination problem. A single site won’t reset priors. Would it make sense to treat the sponsor/CRO (or a central IRB) as the unit of change and run a staggered multi-site pilot where several clinics adopt AE-neutral pay plus a grade-1/2 eligibility guardrail with a dated, public policy? Start dates randomized by site; evaluate with a simple diff-in-diff on diary timeliness, lab/PK concordance, and dropouts.
To make the commitment credible to volunteers, what about a “portable guarantee” across participating sites, i.e. a benign AE reported at Site A can’t be used to exclude at Sites B/C unless a clinician adjudicates a real contraindication...and implement the same workflow via shared ePRO/payment vendors so it’s visible and consistent?
If that still shows no movement, then the issue likely sits deeper than pay/eligibility fear; if it does, you’ve got a template ready to scale.
I enjoyed this review, it's a pretty niche topic but one I'm familiar with because ~20 years ago I participated in a handful of these types of clinical trials and now I'm helping to plan them.
I've done about a dozen or so Phase 1 trials, and two groups of people I would say you missed in your "who does these trials" sections is:
1) Backpackers, and
2) Foreign students
I think the whole 'no experience required' part attracts less "shady" characters than you seem to have encountered, and more just people who need to make money asap and can't go via the conventional route. I did my trials mostly during COVID in one of the most locked down cities in the world (Melbourne) so that might have contributed.
By contrast I don't think I ever actually encountered personally anyone who seemed all that shady to me. I heard a few stories of people shagging in the bathrooms but beyond that the overall vibe of the people seemed mostly extraverted and friendly.
The whole 'lie at every step of the application process' rings incredibly true though. I have recruited many family members into doing these trials and the one piece of advice I always give is, no matter the question, if it's about having a health problem always say no. They once asked if I had ever smoked, and wanting not to sound too unbelievable I said I had vaped once a few months ago. That was written down as "smoker" for literally years afterwards.
This could just be the fact that you were in Australia? I assume the author was in the US. I would be very surprised if backpackers were a significant portion of this population, as people mostly do not do that kind of urban transit-hopping backpacking in the US in the first place. Students I am sure are a significant component though. If it's at all similar to the population who donates plasma, sites near major universities are likely mostly students and then mostly 'shady characters' at all other sites. But students also likely do not do the cross-country professional research participant who tries to squeeze in as many as possible.
I knew people in college who participated in trials at a company called Pharmaco. My roommate got her wisdom teeth out for free this way.
Pharmaco was located a few thousand feet from the mall and there was a little joke about a path beaten between the two by all the coeds.
Fast forward to being old, and the women I know (only women) who do these things tend to be retired or laid-off and interested in “gigs” of all sorts: the Census, poll work, substitute teaching.
One of my girlfriends, a freelancer, at sixty was also arduously climbing up two flights of stairs to deliver beer and Jell-O cups to kids too lazy to leave their apartments.
Boomer women were hard workers!
Some people, by no means suffering for funds, yet get antsy without any sort of paycheck.
I'm reminded of an interview with Leonard Nimoy's son about his dad's advice that the answer to every question you are asked during an audition interview is "Yes," which went something a little like this:
So, Leonard, we'd love for you to play Crazy Horse in our Custer's Last Stand movie. Can we ask you just a few questions?
Yes.
You know how to ride a horse?
Yes.
And you are of course an expert archer?
Yes.
And you are an expert archer on horseback?
Yes.
And you can shout Sioux curses with a perfect accent while shooting an arrow on horseback at George Armstrong Custer?
Yes.
After all, Dad used to say, they won't start filming for a few weeks, so you can always take lessons.
re "Whenever I’ve been tempted to develop a low opinion of their judgment, I’ve had to temper that with the knowledge that many of these people have apparently accumulated much more disposable income in the process than I have."
I can't help but feel a little https://xkcd.com/1827/
Sounds a good system to me! They get to study a much more realistic cross-section of the population than they would if nobody lied. It catches all the major problems (ie the ones that they test for and the ones you can't hide). As Douglas Adams said "Any remaining problems are your own."
And, as a bonus, the company is protected against being sued.
And you're providing gainful to people with criminal convictions who otherwise struggle to get jobs.
Maybe we should just adjust our vocabulary and consistently talk about "phase I subjects" instead of "healthy subjects" when referencing these studies. Some experts already do that (though I'm not sure if it's something a reviewer would flag), but I don't think it's filtered into public discourse yet.
It might be even better if we started attaching a few boilerplate words for those new to the topic:
"No side effects were observed in phase I subjects dosed with 5 mg of exampeline (Examplechuck et al., 2025). However, phase I subjects are known to differ from the healthy general population (Examplovich and Chang, 2022)."
How common is it for the drug to get rejected at Phase One?
Approximately 40% failure rate
Yes, what I'm wondering here is, how much would these flaws in safety trials cause problems if the US were to switch back to a "safety trials only" system as is often suggested?
A delightful accretion of safetyism, the cynic in me is thoroughly amused. But also, it's impressive that even an arrangement this corrupt and perverse could still be be made decently workable with enough effort.
Will the author stay anonymous? Or will the author reveal themselves and vault into the meta-business of teaching folks how to participate in Phase I trials!?
The two people who distrust white doctors reminded me of a Braver Angels discussion with two vax skeptics in 2021. One was white and MAGA, the other was black and BLM. In case you don't recall, Biden had prioritized getting black and old people vaccinated first because they had worse outcomes in aggregate.
The source of the black guy's skepticism emerged when he said "when has the government EVER prioritized the health of black people?"
Damned if you do, damned if you don't.
The bizarre thing is that a lot of times the reference point in articles (which are now on repeat) about black mistrust of the medical establishment, is the Tuskegee syphilis study - which was itself a prioritizing of the health of black people, among whom syphilis was concentrated, at least in the South. However dumb was the worship of the study later on, to obliviously or callously continue it when treatment was established, the whole thing would not be a millstone around our necks forever and ever if there had been no such original “prioritization”.
re "Do they have any reason to think that ADHD might be associated with an increased risk? I asked, and their answer was, not at all, but better safe than sorry."
Ozempic seems to have somewhat rare but very severe side effects (that include severe depression and suicidal ideation *caused* by the drug) specifically in autistic people, and autism tends to be strongly correlated with ADHD, so there's that.
I kinda feel like you'd rather find that out in a clinical setting than patients at home. But clearly that goes entirely against recruiting only healthy subjects.
The link for “the asshole filter” goes to a copy of another blogger’s post. They disclose that they copied it, but don’t say they had permission to publish a copy. It seems… on brand? Anyway, I think it would be better to link to the original:
https://siderea.dreamwidth.org/1209794.html
I knew sometime who participated in these studies and he said the participants would bite the pills so they knew who was getting placebo or the real thing based on the taste.
Very interesting read, thanks! I wonder if it would help if there was a country-wide database of people who participated and everyone could participate at most once in life
Yeah that’s the one thing I was seeing that could change the incentives - make the pay higher but have a nationwide tracking system that locks you out from any others for a while.
I think you'd probably very quickly run out of eligible participants, because the vast majority of people wouldn't participate even once.
O1 — Structure
Phase I = safety/tolerability testing, healthy volunteers only, pre-efficacy.
Clinics = liability minimizers; participants = money maximizers.
O2 — Incentive Misalignment
Clinics overload exclusion criteria (“better safe than sorry”).
Honesty → exclusion; lying → access + pay.
Result: honesty systematically punished.
O3 — Participant Profile
Demographic: mostly male, itinerant, anti-authority, fringe investors.
Social: distrustful, contrarian, peer-networked information culture.
Not poor but opportunistic; lump-sum earners with irregular lifestyles.
O4 — Techniques of Evasion
Conceal diagnoses, meds, supplements.
Rotate across sites to bypass washout rules.
Underreport or hide adverse effects.
Share hacks to outsmart staff and maximize inclusion.
O5 — System Outcome
Data inputs skewed: populations cleaner on paper than in reality.
Safety signals muffled by underreporting + dishonest baselines.
Still functional because:
Labs catch acute harms.
Later-phase patient trials filter major risks.
O6 — Meta-State
Equilibrium: neither catastrophic nor pristine.
Science runs on dishonest substrate tolerated as “good enough.”
Cost = eroded validity, hidden risks, gray-zone ethics.
Phase I = adversarial equilibrium: participants exploit for money, clinics shield from liability, pharma tolerates distortions, yet the system limps forward because downstream filters mop up worst failures.
This is an interesting analysis of why “paying people to do medicine-adjacent things” (eg: compensated donations) can get so much pushback: behind knee-jerk “ick”-type reactions, there’s maybe some sense that the interaction between “people optimizing for getting paid,” the typical mild dysfunction you’d see in any large organization, and the sheer stakes of healthcare, can produce quite the perverse and unanticipated outcome.
I'm suddenly imagining a dystopia in which these problems are dealt with by drafting participants the way we assign people to jury duty. Of course that would be entirely against medical ethics as we know them, but this is a dystopia. (Whether drafting people to jury duty should be against legal ethics is another question.)