Pinned Tweet
@fluidloading @bentatoo31 @Denis_Faust @Procto_Log The story of pathological consensus and shortcutting the RCT method of Hill/Fischer with the Petty/Bone science. The lumping paradigm.
The Physician’s War: The Story of the Hidden Battle between Physicians and a Science Based on Pathological Consensus a.co/d/8OtJlVk
The Physician’s War: The Story of the Hidden Battle between Physicians and a Science Based on Pathological Consensus a.co/d/8OtJlVk
4
9
48
11,387
Scott @BerryConsultant, given your involvement in REMAP-CAP, we would welcome your thoughts on an important patient safety question.
Critical care physicians need to understand the cause of decades of discordant findings of syndrome RCTs such as CAPE COD and REMAP-CAP. This new framework in the link below describes a third-layer estimand, E3: a treatment effect averaged across the mixture of underlying diseases admitted by a syndrome-based enrollment gate.
When treatment effects differ across diseases, changing their proportions can change, and potentially reverse, the average effect, even when disease-specific effects remain unchanged: the cause mixture paradox.
Could this help explain these discordant findings? More fundamentally, what justifies transporting E3 through guidelines to patients whose disease mixture differs from the trial’s?
The concern is potential patient harm from transporting a mixture dependent average (E3) without establishing its applicability. This does not establish why these particular trials differed, but it raises a question critical care physicians need answered.
pubmed.ncbi.nlm.nih.gov/42464342/
Critical care physicians need to understand the cause of decades of discordant findings of syndrome RCTs such as CAPE COD and REMAP-CAP. This new framework in the link below describes a third-layer estimand, E3: a treatment effect averaged across the mixture of underlying diseases admitted by a syndrome-based enrollment gate.
When treatment effects differ across diseases, changing their proportions can change, and potentially reverse, the average effect, even when disease-specific effects remain unchanged: the cause mixture paradox.
Could this help explain these discordant findings? More fundamentally, what justifies transporting E3 through guidelines to patients whose disease mixture differs from the trial’s?
The concern is potential patient harm from transporting a mixture dependent average (E3) without establishing its applicability. This does not establish why these particular trials differed, but it raises a question critical care physicians need answered.
pubmed.ncbi.nlm.nih.gov/42464342/
Critical care physicians must learn the risks of guidelines derived from cause-agnostic RCTs of synthetic syndromes.
Yet, surprisingly, after much effort, I have discovered that statisticians are not going to timely teach this math despite recognizing its solid foundation.
Perhaps this is due to the fact that the math challenges the broadly applied Grade A branding of all RCT without regard to causal structure.
To whom can I turn to facilitate reform of the present trial design paradigm?
1
3
6
686
Critical care physicians must learn the risks of guidelines derived from cause-agnostic RCTs of synthetic syndromes.
Yet, surprisingly, after much effort, I have discovered that statisticians are not going to timely teach this math despite recognizing its solid foundation.
Perhaps this is due to the fact that the math challenges the broadly applied Grade A branding of all RCT without regard to causal structure.
To whom can I turn to facilitate reform of the present trial design paradigm?
Yet, surprisingly, after much effort, I have discovered that statisticians are not going to timely teach this math despite recognizing its solid foundation.
Perhaps this is due to the fact that the math challenges the broadly applied Grade A branding of all RCT without regard to causal structure.
To whom can I turn to facilitate reform of the present trial design paradigm?
3
2
17
2,407
Public access increasingly includes artificial intelligence access to the full article. If you care about making medical knowledge accessible to clinicians and the public, publish openly or ensure a legal full text version is freely available. Paywalls can block both readers and the AI tools they use to understand the evidence.
Say goodbye to excessive open-access fees just to make your research available to the public.
1
0
3
401
How did Alexander win against the Persian army? He shifted far to the right and then took a wedge of elite Greek cavalry straight in after Darius.
This is the methodology parallel to the “E3 Paradigm” as a tool to induce causal formalism in RCT design.
The Persian Army is the E2 RCT Dogma: A massive, entrenched, heavyweight apparatus relying on established grant pathways, broad pragmatism, decades of consensus, and status-quo inertia.
Fighting a standard battle is a losing strategy: Trying to debate every cause agnostic trial one-by-one across a massive literature is like fighting Darius’s whole line head on. It is absorbed and even ignored by the leaders without breaking the system.
“The E3 Wedge”: The E3 framework and formal mathematical causal structures aren't just another argument; they are the sharp, unyielding wedge.
By targeting this single deep structural flaw in traditional RCT design, the mathematical invalidity of averaging different diseases & mechanisms, it pierces straight to the core command layer of elite trial methodology.
When we strike the structural core with mathematical certainty, the pathological apparatus of critical care syndrome RCT without causal formalism will not hold itself together.
This is the methodology parallel to the “E3 Paradigm” as a tool to induce causal formalism in RCT design.
The Persian Army is the E2 RCT Dogma: A massive, entrenched, heavyweight apparatus relying on established grant pathways, broad pragmatism, decades of consensus, and status-quo inertia.
Fighting a standard battle is a losing strategy: Trying to debate every cause agnostic trial one-by-one across a massive literature is like fighting Darius’s whole line head on. It is absorbed and even ignored by the leaders without breaking the system.
“The E3 Wedge”: The E3 framework and formal mathematical causal structures aren't just another argument; they are the sharp, unyielding wedge.
By targeting this single deep structural flaw in traditional RCT design, the mathematical invalidity of averaging different diseases & mechanisms, it pierces straight to the core command layer of elite trial methodology.
When we strike the structural core with mathematical certainty, the pathological apparatus of critical care syndrome RCT without causal formalism will not hold itself together.
1
0
2
393
Repost of “Cause-agnostic auto mechanics”
After decades of engine failures, trialists define “Engine Overheating Syndrome” (EOS): temperature ≥120°C.
Engine pathophysiologists and causal inference experts call for research into the causes, including low coolant and/or low oil.
Trialists and cause-agnostic statisticians win the grant instead, randomizing 1,000 overheated engines to coolant or oil.
Average outcomes are identical. The New England Journal of Mechanics announces: “coolant and oil are equally effective for EOS!”
The next trial happens to enroll more coolant-deficient engines. A meta-analysis follows.
Guidelines declare: “Evidence-based mechanics (EBM) has established coolant as the standard treatment for EOS.”
After decades of engine failures, trialists define “Engine Overheating Syndrome” (EOS): temperature ≥120°C.
Engine pathophysiologists and causal inference experts call for research into the causes, including low coolant and/or low oil.
Trialists and cause-agnostic statisticians win the grant instead, randomizing 1,000 overheated engines to coolant or oil.
Average outcomes are identical. The New England Journal of Mechanics announces: “coolant and oil are equally effective for EOS!”
The next trial happens to enroll more coolant-deficient engines. A meta-analysis follows.
Guidelines declare: “Evidence-based mechanics (EBM) has established coolant as the standard treatment for EOS.”
0
1
3
246
Lawrence Lynn retweeted
@ThinkingCCn@NEJMs cars overheating and funds experts from the Petty/Bone school of mechanics. The experts name the problem “auto overheating syndrome” (AOS).
They know overheating can result from low water or low oil. Nevertheless, they randomize the next thousand overheating cars to receive water or oil without distinguishing the cause.
The average outcomes are identical. They announce: “Water and oil are equally effective for AOS.”
After 30 years of failure of cause agnostic RCT, the public eventually realizes that equal outcomes across a mixture of problems do not make the remedies interchangeable.
They replace the experts with auto pathophysiologists, who begin by asking why each engine is overheating.
They know overheating can result from low water or low oil. Nevertheless, they randomize the next thousand overheating cars to receive water or oil without distinguishing the cause.
The average outcomes are identical. They announce: “Water and oil are equally effective for AOS.”
After 30 years of failure of cause agnostic RCT, the public eventually realizes that equal outcomes across a mixture of problems do not make the remedies interchangeable.
They replace the experts with auto pathophysiologists, who begin by asking why each engine is overheating.
0
1
3
75
One sees on X toward an evolving intellectual dominance of the young expert pathophysiologists with new enlightened knowledge and interrogation of the literature and of complex relational pathophysiology.
“Pathophysiologists” are the new class of critical care experts. It’s now your field to reform. Salute.
“Pathophysiologists” are the new class of critical care experts. It’s now your field to reform. Salute.
@ThinkingCCf@ross_prager lived through the arrogant decades of consensus sanctioned, cause-agnostic RCTs, it is great to see a paradigm shift finally emerging driven by this new generation of pathophysiologists.
One wonders what the past 34 years might have yielded had sepsis RCT science not been held so tightly by the leadership that shaped it with their guesses in 1992, and successive generations trained to defend its simplistic threshold assumptions and the pathological RCT design it generated.
How much progress was lost in 34 years while consensus authority sustained a profoundly oversimplified framework whose biological and RCT foundations, (coupled with a lack of understand of RCT fundamentals) controlled the science.
When we look at the progress made in cardiology over the past 30 years and see that we are still doing sepsis RCT (testing aspirin) using the 1996 SOFA score as the gate we should hide our heads in shame for our failure to the public.
2
0
6
371
Hello @yudapearl @soboleffspaces and @eliasbareinboim.
I have been looking for a published RCT through which your teams could demonstrate why explicit causal formalism should be a standard requirement for trial design and safe transport.
The 2025 REMAP-CAP hydrocortisone trial offers a revealing case:
1. A broad (cause agnostic) nonspecific threshold gate that does not require selection of a disease or mechanism.
2. Bayesian Time Machine (BTM)
3. Response-adaptive randomization (RAR).
4. Not even post-gate physiological funneling to establish a definable target for transport.
The structural question is simple: whose treatment effects drive the allocation of treatment to incoming (RAR) patients and then to transport.
Note that each interval can generate a different disease-mixture effect, E₃(t), even when disease-specific effects remain unchanged. If allocation responds to benefit in an earlier mixture, what protects a later mixture in which that treatment may cause net harm.
This is the “canary” which warns of an evolving state of unbridled methodological hubris free of the required biology (cause) considerations.
This is how to make the case for reform concrete: dissect this published, high profile, RCT design and expose how ambitious methodology without causal formalism can generate unsafe transport.
I have been looking for a published RCT through which your teams could demonstrate why explicit causal formalism should be a standard requirement for trial design and safe transport.
The 2025 REMAP-CAP hydrocortisone trial offers a revealing case:
1. A broad (cause agnostic) nonspecific threshold gate that does not require selection of a disease or mechanism.
2. Bayesian Time Machine (BTM)
3. Response-adaptive randomization (RAR).
4. Not even post-gate physiological funneling to establish a definable target for transport.
The structural question is simple: whose treatment effects drive the allocation of treatment to incoming (RAR) patients and then to transport.
Note that each interval can generate a different disease-mixture effect, E₃(t), even when disease-specific effects remain unchanged. If allocation responds to benefit in an earlier mixture, what protects a later mixture in which that treatment may cause net harm.
This is the “canary” which warns of an evolving state of unbridled methodological hubris free of the required biology (cause) considerations.
This is how to make the case for reform concrete: dissect this published, high profile, RCT design and expose how ambitious methodology without causal formalism can generate unsafe transport.
The structure of novel platform cause agnostic RCT with “Bayesian Time Machine” (BTM) and “response-adaptive randomization” (RAR) as used in
the 2025 REMAP CAP platform RCT produces a potentially toxic brew.
Each interval can generate a different disease-mixture effect, E₃(t), even when disease-specific effects remain unchanged.
If RAR responds to earlier outcomes, a treatment that benefited a prior disease mixture may receive greater allocation in a later mixture where it causes net harm within the same platform trial.
To induce timely reform, ground your arguments in a structural dissection of a published “toxic brew”. Expose the structure, what the entry gate mixes, what temporal borrowing assumes, and whose treatment effects drive subsequent allocation.
@yudapearl @eliasbareinboim @soboleffspaces
discourse.datamethods.org/t/bayesian-tim…
0
0
4
461
The structure of novel platform cause agnostic RCT with “Bayesian Time Machine” (BTM) and “response-adaptive randomization” (RAR) as used in
the 2025 REMAP CAP platform RCT produces a potentially toxic brew.
Each interval can generate a different disease-mixture effect, E₃(t), even when disease-specific effects remain unchanged.
If RAR responds to earlier outcomes, a treatment that benefited a prior disease mixture may receive greater allocation in a later mixture where it causes net harm within the same platform trial.
To induce timely reform, ground your arguments in a structural dissection of a published “toxic brew”. Expose the structure, what the entry gate mixes, what temporal borrowing assumes, and whose treatment effects drive subsequent allocation.
@yudapearl @eliasbareinboim @soboleffspaces
discourse.datamethods.org/t/bayesian-tim…
the 2025 REMAP CAP platform RCT produces a potentially toxic brew.
Each interval can generate a different disease-mixture effect, E₃(t), even when disease-specific effects remain unchanged.
If RAR responds to earlier outcomes, a treatment that benefited a prior disease mixture may receive greater allocation in a later mixture where it causes net harm within the same platform trial.
To induce timely reform, ground your arguments in a structural dissection of a published “toxic brew”. Expose the structure, what the entry gate mixes, what temporal borrowing assumes, and whose treatment effects drive subsequent allocation.
@yudapearl @eliasbareinboim @soboleffspaces
discourse.datamethods.org/t/bayesian-tim…
0
1
2
529
Lawrence Lynn retweeted
"Today, Oct. 10, is the birthday of my son, Daniel Pearl, who would have been 63 years old today.
Lacking eloquent words to describe his life, his work, and his vision until his brutal murder in 2002, I am posting a poem I wrote in his memory, called 'The Lion's Den': ucla.in/4AT0ddW
followed by David Broza's recording, written for that poem: youtube.com/watch?v=L0yUsB…, ending with Benjamin Lapidus "Kaddish Para Daniel" open.spotify.com/track/2VOZK7Mi…"
Join me in Yitgadal V'Yitkadash Shmei Rabah
Lacking eloquent words to describe his life, his work, and his vision until his brutal murder in 2002, I am posting a poem I wrote in his memory, called 'The Lion's Den': ucla.in/4AT0ddW
followed by David Broza's recording, written for that poem: youtube.com/watch?v=L0yUsB…, ending with Benjamin Lapidus "Kaddish Para Daniel" open.spotify.com/track/2VOZK7Mi…"
Join me in Yitgadal V'Yitkadash Shmei Rabah
517
1,907
11,536
271,782
Question for @stephensenn and @BerryConsultant (or anyone else) re: the Bayesian Time Machine applied to a cause agnostic RCT (CAR) at datamethods. Join the conversation.
discourse.datamethods.org/t/bayesian-tim…
discourse.datamethods.org/t/bayesian-tim…
0
0
0
161
“The RCTs power lies not in form, randomization, blinding, or sample size alone, but in its causal grounding: a true RCT defines its population, interventions, and outcomes within a framework that respects causal structure.”
This….and not: “An RCT is causal by design!” is the definitively functional statement public funding in the 21st century.
It follows that without structural causal formalization there is no means to assure the public is getting the RCT quality it is paying for. Therefore @NIH funding should require formal evidence that a RCT “respects causal structure”.
@yudapearl @f2harrell @eliasbareinboim @DrJBhattacharya
This….and not: “An RCT is causal by design!” is the definitively functional statement public funding in the 21st century.
It follows that without structural causal formalization there is no means to assure the public is getting the RCT quality it is paying for. Therefore @NIH funding should require formal evidence that a RCT “respects causal structure”.
@yudapearl @f2harrell @eliasbareinboim @DrJBhattacharya
1
1
5
307
Responsive to pushback my point is not technical it is (as I said in the post) “functional”. When public funds back a RCT, the public isn't paying for research that is “causal by design” for an internal statistical artifact bound only to the study's specific 500 patient sample.
They are paying for actionable, safe clinical guidelines that transport to a target population. They are paying for gate to guideline “G2G” research.
G2G should be the 21st century funding standard.
They are paying for actionable, safe clinical guidelines that transport to a target population. They are paying for gate to guideline “G2G” research.
G2G should be the 21st century funding standard.
0
0
0
60
Can a Bayesian reject causal formalism of RCT? I argue that she cannot.
Don’t miss the discussion. Clearly datamethods is willing to publish alternative views because the cause agnostic RCT (CAR) is sacrosanct. This aspect of the RCT causal formalism debate is “troll kryptonite”.
@ildiazm @yudapearl @eliasbareinboim
discourse.datamethods.org/t/causal-forma…
Don’t miss the discussion. Clearly datamethods is willing to publish alternative views because the cause agnostic RCT (CAR) is sacrosanct. This aspect of the RCT causal formalism debate is “troll kryptonite”.
@ildiazm @yudapearl @eliasbareinboim
discourse.datamethods.org/t/causal-forma…
2
0
5
598
This is an example of classic clinical trial discourse. It’s a useful article which recognizes that randomization does not guarantee transportability. Yet it exemplifies a recurring ambiguity in statistical discourse: “consider heterogeneity” and “assess applicability” offer correct cautions without actionable criteria. No failure mode analysis methodology is embodied in these words.
The “cause agnostic RCT” framework makes the questions concrete…. Are there diseases/causes/mechanisms which are were mixed and why, mathematically, should that pooled estimate apply to the guideline population?
Without Pearlean causal formalism requiring explicit causal structures and transport assumptions, the discourse is safe as it permits agreement without requiring any change in practice.
The “cause agnostic RCT” framework makes the questions concrete…. Are there diseases/causes/mechanisms which are were mixed and why, mathematically, should that pooled estimate apply to the guideline population?
Without Pearlean causal formalism requiring explicit causal structures and transport assumptions, the discourse is safe as it permits agreement without requiring any change in practice.
Randomized, controlled, and real
jclinepi.com/article/S0895-…
0
1
4
874
I agree Brad. A petition may help. But they are convinced (falsely) that they hold the academic high ground. Reform begins by showing how their preferred pathological cause agnostic RCT science itself can fail patients. This they will struggle to ignore.
Our mathematical failure mode analysis traces syndrome RCTs from entry gate to guideline (G2G) showing how mixing distinct causes in a RCT can produce an E3 estimand that becomes unsafe when converted into a guideline.
The analysis is available to read. It’s time to bring causal formalism to critical care RCTs , take the intellectual hill, and challenge ambiguous pathological cause agnostic RCT based guidelines with the cold steel of causal mathematics.
Our mathematical failure mode analysis traces syndrome RCTs from entry gate to guideline (G2G) showing how mixing distinct causes in a RCT can produce an E3 estimand that becomes unsafe when converted into a guideline.
The analysis is available to read. It’s time to bring causal formalism to critical care RCTs , take the intellectual hill, and challenge ambiguous pathological cause agnostic RCT based guidelines with the cold steel of causal mathematics.
@AnilMakam I hear you. But there is reason for hope. The petition could help. Let’s connect when you have a minute.
1
0
4
1,288
Here @yudapearl @eliasbareinboim is another excellent critique. Here we see an example of critical care guidelines derived without causal formalism of the RCTs and massaged by meta analysis to tease out a treatment guideline based on a blunt, multi-parameter responsive, ratio (P/F).
These are studies of paralysis of mechanically ventilated patients with ARDS, a 1967 synthetic syndrome defined by a guessed cause-agnostic threshold set gate.
These are studies of paralysis of mechanically ventilated patients with ARDS, a 1967 synthetic syndrome defined by a guessed cause-agnostic threshold set gate.
The Case of the Obscured Bias. New post discussing the recent SCCM Guidelines on NMBAs in ARDS. emcrit.org/emnerd/the-cas…
1
2
10
1,077
Great work. Another blow to ~2 decades of critical care guidelines based on the “cause agnostic RCT” (with guessed cause agnostic gates), no intervening failure mode analysis or causal formalism and blind transport to a guideline based on the cause agnostic gate.
@yudapearl @eliasbareinboim
note that what I have taught you about critical care RCT science is true. You can help.
@yudapearl @eliasbareinboim
note that what I have taught you about critical care RCT science is true. You can help.
How strong is the evidence behind the Surviving Sepsis Campaign?
In our new paper, we analyzed more than two decades of SSC guidelines, including 554 statements across seven publications.
The result is striking.
From 2008 to 2026, the proportion of recommendations supported by fell from 15.8% to 2.3%, while strong recommendations decreased from 69.7% to 18.0%.
Perhaps even more importantly, in the 2026 guideline, one-third of strong recommendations are still supported by low- or very-low-certainty of evidence. This means we should be very careful about confusing guideline recommendations, implementation priorities, and certainty of evidence.
Guidelines should support clinical judgment — not replace it.
Happy to see this work published in the Journal of Critical Care.
sciencedirect.com/science/articl…
2
1
11
681