Absolutely and that has been a direction which some are taking and that may offer one of the greatest early opportunity given the size of the retrospective databases.
The problem is that lumping has been mandated to define the target heterogeneous syndrome to unify or harmonize the worlds research. That consensus is linked to funding. The consensus triage then bakes-in the heterogeneity.
This is why many desire a clean non lumped slate to do the kind of work you are suggesting.
This is a great discussion and I second Erinâs motion that we not pivot to writing about designing better research programs and trials in ARDS and sepsis.
My understanding of Erinâs request was that he was asking for exactly that.
I demurred, as the goal of the talk is general apical error related reform. The specific design issues, as a function of this reform, being best left to the experts.
This discussion needs a reframing or a return to its original frame. It is crucial to apprehend the context when many people engage in asynchronous text exchanges.
RCTs were developed to test interventions within homogenous biological constructs. Petty and Bone proposed RCTs could be used to test interventions within homogenous clinical constructs.
This is no trivial modification from the original method.
I like to use the expression disease model to denote the biological construct that gives a disease its unique taxonomy. Such models must be capable of articulating the clinical presentation, the diagnostic workup, and plausible intervention targets. Clinical constructs are not disease models.
For instance, one may declare that âStrong Headacheâ is a clinical construct and then enroll 'Strong Headache" patients to test the same intervention in all patients, never caring about common biological mechanisms. This is the âPetty and Boneâ approach to headaches.
The Petty and Bone modification is a scientific conjecture per se, on the realm of epistemology.
Refuting a conjecture that failed to provide valid predictions of the empirical world is the unavoidable duty of the scientist.
Beautifully articulated @Rafael_Leite. Yet this is only the âPetty Apical Errorâ (the disease lumping dimension) of the pathological RCT.
The 33year old âBone Apical Errorâ is the second dimension which teaches âhow to lumpâ .
The Bone technique teaches to derive (as by expert guessing) a set of non-disease specific thresholds and then lump different diseases by using that derived set of thresholds as a triage tool
Combining these apical errors together by a consensus task force creates a âheterogeneous syndromeâ which, because it is an artifact we call a âsynthetic syndromeâ.
Combining Petty error 1 with Bone error 2 a consensus task force generates the latest heterogeneous syndrome which is therefore:
A overarching disease agnostic plurality of sets of a second plurality of sets of different diseases which fall within the scope of set of triage thresholds as amended from time to time by pathological consensus.â
It follows therefore that:
(The PettyBone RCT is an âRCT MIMICâ testing an intervention applied to a heterogeneous syndrome which is a clinical construct comprised of a set of different diseases lumped by a triage set of non-disease specific thresholds).
Contrary to those suggesting in this forum that this Langmuir pathological science is known, there were 3 (THREE) task force meetings using both Petty and Boneâs apical errors together to generate triage threshold sets for lumping diseases for harmonizing worldwide RCT in the past three years. They produced three new heterogeneous syndromes for RCT.
The PettyBone pathological science cannot be known if they are still teaching, funding, intellectually colonizing, and doing it.
However I will close my posts on this subject unless anyone in the future wants to have a substantive debate or discussion related to the actual **math and datamethods **.
Before I close I will provide a brief glossary in another post which explains the terms used to expose the pathological science.
âThe mathematics is not there till we put it there.â - Arthur Eddington
This is literally what the global research effort in sepsis immunology is striving for, and has been the stated aim of the field for well over a decade. Itâs the conversation I have most often with the authors of the sepsis-3 paper. I donât know what else you could possibly want? Are you aware of the ongoing research in this area?
This is such a strong and demonstrably untrue statement. With respect, itâs hard to debate this from a rational position as itâs such a ridiculous statement. It has the flavour of âIâm the only one who knows the truthâ. Which isnât a rational position to hold when global research efforts require the input of dozens, if not hundreds, of minds to complete successfully. The mass of back and forth critical scientific reasoning in medicine and intensive care medicine in particular is vast. I appreciate that there is research waste. And the interface between science and academia is complex. Just because some of these positions donât align with your personal view doesnât render them invalid. Doctors, clinical scientists and scientists from a broad range of fields are engaged in massive efforts of directed scientific research from lab to clinical studies, experimental and observational etc.
To suggest that these people canât think scientifically undermines your central thesis to such a high degree. Iâm honestly perplexed by the statement.
Iâm starting to think we are talking cross purposes here. The sepsis research community is well aware of the heterogeneity issue and are directing large scale research efforts to better understand it.
I agree with this sentiment. I havenât actually seen much here that isnât framed as speculative.
If there are concrete proposals of how you propose to move forward in terms of trial design or patient selection, then Iâm all ears. At present, I donât think there is enough here to write an article because itâs all been said before and there isnât any evidence provided to support the arguments beyond opinions (whether or not I agree with those opinions is irrelevant).
We seem to be stuck in a loop.
So, with the greatest of respect. Iâll bow out.
From now on, letâs tone it down and stick to outlining and detailing future research programs and study designs that will result in real progress rather than any repetition of past thoughts. I hope you donât bow out @Doc_Ed .
Hi Doc_Ed
I am used to being called ridiculous etc. I donât care. You are talking to someone whoâs reading bad research in our field for +25 years. We must discuss a generation of scientists (starting with J-L Vincent) who wasted their (ours?) lives in unfruitful models like sepsis and ARDS.
I have studied sepsis immunology and started contributing to the sub-field of purinergic signaling in sepsis, before realizing the sepsis construct is a sinking ship. So, yes, I am aware of ongoing research and I am saying with all the words: it will fail. They wonât reach a disease model using a clinical construct like Sepsis-3 or any other. You canât define a disease as a âdysregulated response to infectionâ and expect someone will find a treatment.
I keep it. I am not suggesting they canât think scientifically. I am affirming it. I am not pretending to be the clever guy in the room. Instead, I had the humility to take a few steps back and try to grasp what was wrong in a field that canât produce innovation. I see a widespread poor understanding of how science works. Our âscientistsâ should try reading epistemology to understand their mistakes. I extend the recommendation to the biostatisticians because they also canât see what is wrong, yet, they think advanced statistics can compensate for bad hypotheses derived from poor clinical reasoning.
I recommend Popper, the Vienna inductivists, Kuhn, etc. It wonât take more than a weekend to get started and it will do more for our field than the silly âprecision medicineâ subgroups they are testing in several trials like Panther, Precise, Pallete, etc.
I thank God I donât need to network, so be free to think or say anything you like about me, but do it elsewhere, not here as asked by Professor Farrel.
However, if you want to discover how I got to such an extreme position, I invite you to my substack.
My proposition is that we shouldnât study ARDS or sepsis patients as belonging to a unique syndrome. Itâs a bold proposition, to abandon the âheterogeneous syndromeâ and the clinical constructs like Sepsis-3. One interesting (maybe unintentional) approach was the STAMINA trial, where they only included pneumonia/ARDS patients. ARDS could be split into several independent nosologies to be studied individually, and patients shouldnât be triaged using PaO2/FiO2 thresholds, at the risk of missing the whole picture of patients with similar pathophysiology.
What a wonderful raw science debate. Raw science disagreement about fundamental methodology in the open is as rare as Bigfoot in critical care.
Yet this is exactly what the public deserves. According to the CDC, 13000 patients died from influenza this season yet there are no RCT to inform physicians the role or lack of thereof of corticosteroids in reducing mortality of severe influenza pneumonia.
This makes those who understand that the research funds were mistakenly spent, for decades on Pettyâs ARDS clinical construct rightfully upset when the beribboned remain recalcitrant and refuse to openly introspect and reform. .
In contrast those âin the dogma boxâ shrug, and say the classic line âthe data are scantâ and âthe trials are comingâ.
These battles between those who want introspection and rapid reform, and those who are comfortable with a non introspection based, slow pace has always existed.
Imagine the polar opposite dispositions of two young engineer groups who had been diligently working on much needed bridge for a decade, only to discover the concrete provided by the trusted leaders was not real concrete.
One group would be nonplussed and say âOh well, letâs keep tryingâ, while another group would be outraged at their lost work and failure for the public and would demand introspection so this would not happen again.
So this is a healthy debate and one expected when social forces do not prevent open debate. I hope it is widely read because it lays bare the present polar state of the fundamental methodology camps of critical care science.
Very well. Here is the template. Follow the guidelines set forth here where treatment is tested for an objectively measurable phenotype of a specific objectively definable disease.
In this case the phenotype is alveolar tuberculosis. This is rare in the US today but if you have been a pulmonologist as long as I have you would have seen it. It was common in the past.
Note he was phenotyping downward from the disease of tuberculosis and even pulmonary tuberculosis. He describes the need for a measurable homogeneous thread.
Erin read this landmark article very closely. Note the emphasis on objective (& non lumped) input measurements and output measurements. Follow the guidelines here for influenza A pneumonia, and each of the other pneumonias.
Donât lump MRSA pneumonia with pneumococcal pneumonia or Influenza pneumonia. Rather study each of them separately. Study aspiration associated respiratory infiltrates in the same way. Modify as needed based on pragmatic of measurement but do not violate the homogeneous thread. ARDS is a severity indication, an artifact.
Work with statisticians to be sure the Bradford Hill method is applied to the best extent possible as early in the trial as objective diagnostic data allows.
If a treatable phenotype of a disease is found look for that phenotype in other diseases and test them each by the Bradford Hill method.
Start with Hill. Then test lumped groups only on objective evidence of probable broader efficacy based on study of individual disease. Bayesian methods may be applied once an objective disease is the target and priors can be defined.
Thatâs exactly what I would do in conjunction with experts given the 33 years of triage threshold based lumping shortcut failure and which is not a valid technique and is, in any regard, antithetical to the RCT function. But first I would widely promulgate the PettyBone mistake to release the indoctrinated. With this effort I would empathize that a valid RCT must have a homogeneous thread.
If a new RCT is chosen, such as the PettyBone RCT this must be validated mathematically before it can reenter the field but it must be so identified separately by branding to avoid confusion with a standard Bradford Hill RCT.
Of course, I would defer to those expert statisticians who have escaped the lumping shortcut box. I hope others will bring ideas now that we are past the turbulence associated with bursting the 33 year dogma bubble of the research based on âdisease lumping by triage threshold set eraâ.
Again read the history. Spend time following the progression of the RCT. Study the methods. Abandon the shortcuts. Fix the hole in the hull.
I have read this entire thread and find it interesting. Iâll hazard to contribute.
The central thesis is that lumping many different pathological processes into synthetic syndromes and then studying these syndromes is so problematic as to be practically useless. For arguments sake let us assume I am utterly convinced. But as with other readers, I am a bit perplexed as to what the alternative proposition is, and, I too, find that the responses to these calls for an alternative weirdly combative and unsubstantial.
We are mainly discussing ARDS and sepsis, and yet the examples those who oppose the âpetty/boneâ RCT offer as a counter point (e.g. tuberculosis) are disease where you can reasonably wait to confirm the pathological cause before treating. When I as a clinician start treating what I presume to be sepsis or ARDS, I do not in most cases know the cause.
Can we have an honest discussion about what inclusion criteria would look like in a critical care trial for severe acute pathology that avoids the pitfalls of âpetty/boneâ? Let us start with a specific example to help my understanding, what are your thoughts about the recently published BALANCE trial .
Thanks for contributing. As you point out this is a complex problem with considerable nuance so we need many more thoughtful participants in this important discussion. Your identification of the Balance Trial is especially illustrative.
First, the use of Bradford Hillâs tuberculosis treatment presentation of the RCT was simply to present the ideal template. Hillâs brilliance is more evident when considered through a contemporary lens.
So if we start with Hillâs template we can identify a group of specific diseases in Pettyâs ARDS which can be diagnosed rapidly. For example, post aspiration, post burn, COVID pneumonia, Influenza A pneumonia, post trauma, pancreatitis. Each of these can be studied individually by multi-center RCT.
Petty did not include bacterial pneumonia and Pettyâs heterogeneous syndrome was greatly expanded by using P/F triage. So over the decades we further progressed in the broadening direction, away from the homogeneous thread which Petty and Bone had already abandoned.
In the Balance Trial there is lumping but not in the âPettyBone senseâ. Petty and Bone taught to lump by non-disease specific chest radiograph and/or non-disease specific thresholds. This was the pitfall which enabled lumping inflation. There is no limit to the broadening in PettyBone based research.
The Balance Trial lumps specifically diagnosed bacteremia with evidence based exclusions (such as Staph. aureus and source control considerations). This is very close to a Bradford Hill RCT but in my view it should have different branding to give the clinician notice that some disease lumping was present. Iâm not sure what that branding should be. I call these collectively, âRCT Modsâ.
The âPettyBone RCTâ brand which I think is absolutely necessary for the clinician (if this RCT mod is not going to be abandoned) refers to a specific type of very broad lumping by non-disease specific triage threshold set.
The thesis here is that RCT should move as closely to Hillâs method as possible.
We can see right away how far a PettyBone mod has moved from a real RCT, where abdominal infection due to perforation after a colonoscopy is lumped with a candida line infection and GAS bacteremia associated with toxic shock.
So the fact that the Balance Trial was able to âthread that heterogeneity needleâ is a wonderful example of movement by the designers, as closely to a real RCT function as possible.
With infection there are dimensions by which this âintelligent lumpingâ can be accomplished. For example the trial could be limited to a source as @Doc_Ed suggested with specified bacterial subsets .
Yet the search for common pathways might be best started with a single disease (eg E. coli bacteremia with a UTI source) thereby setting the rest of the disease related unknown variables as homogeneous across all the trial as possible. Of course this is not the real world but this is rather a means to find the homogeneous biological threads with lumping to follow as defined by evidence.
Thank you again for bringing the Balance Trial to the discussion as it helps illustrate the two polar extremes of Bradford Hill RCT and PettyBone RCT. It also shows the need to brand RCT to alert clinicians so they are immediately aware if a âRCT Modâ is being reported rather than the classic RCT.
I wanted to comment on this. I understand the debate is unusual but the purpose of the original post was to teach the PettyBone RCT Mod not to discuss alternatives.
First the pitfall of non disease specific threshold set lumping with substantially unmitigated dimensionality had to be accepted by the reader before we can discuss alternatives. This is why the discussion kept circling back to the original intent of the post. This is also why fireworks developed relating to the tension that always develops when real fundamental dogma which virtually everyone believed and taught are challenged.
Some commenters seemed to desire to âiceâ the discussion of PettyBone and to divert discussion to alternatives without debating or accepting the need for alternatives as if this was relevant to the Langmuir pathological science nature of the PettyBone RCT.
There is always the deflection to, âWhat is the alternativeâ without the acceptance of the state of pathological science. This deflection prevents introspection.
This type of fireworks seems weird today in the era of consensus driven critical care science but it was once common in science.
My mentor at Wash U Stl. was a proud member of the now defunct âantidogmatic societyâ. Scientists of that time were sceptics and they wore that badge on their chest so everyone was warned that their dogma could be challenged all the way back to its origin. His famous quote was, âWe oppose dogma at every levelâ.
This was once a highly productive and very enjoyable way to âdo scienceâ. The road is wide open because the mentors are not shepherds, they are teachers. In contrast we were all indoctrinated in the opinions of our mentors and shepherds Petty and Bone and their presently extant acolytes. Any sheep that desire to stray are unlikely to be fed.
Today the NIH grant system contributed to the decline of the learned skills of anti-dogmatism which are often replaced with mentor based teaching about networking and consensus task-force class promotion.
This is one of the reasons a deep, historically probing, anti-dogmatic challenge to the fundamental trial methodology in critical care discussed in the open seems weird in 2025 critical care science.
Grok discusses RCT mods.
Lawrence Lynn discusses the issue of misbranding in medical science, specifically regarding modifications (mods) of Randomized Controlled Trials (RCTs). He argues that calling all mods just âRCTsâ inflates their value and misleads clinicians.
He mentions the âPettyBone RCT modâ as an example, suggesting a need for proper branding of RCT mods to avoid confusion and ensure clarity in medical research.
Lynn questions what other RCT mods exist and proposes a naming convention for the mod of the Balance Trial, highlighting a gap in the standardization of RCT terminology in medical literature.
As an endocrinologist, I would like to address the fact that the issue at hand is not limited to the intensive care unit (ICU). E.G. Polycystic Ovary Syndrome (PCOS) is prevalent and is diagnosed according to the Rotterdam criteria, which require the fulfilment of two out of three specific criteria for diagnosis:
Oligo- and/or anovulation,
Clinical and/or biochemical indicators of hyperandrogenism,
Presence of polycystic ovaries as observed via ultrasound.
We also recognise that there are subtypes of the disease. However, this is the clinical practice, and the challenge is that we would need to change the entire paradigm to suit RCT.
Furthermore, what about composite outcomes? Major Adverse Cardiovascular Events (MACE) are frequently used and typically encompass myocardial infarction, stroke, and cardiovascular death, with some definitions also including additional events. I believe that the same issue exists as with the Petty & Bone trials.
Hello Dr. Zarkovic.
Excellent point. In many fields, where separation of specific diseases was not forthcoming, clinical practice evolved around presentation and clinical manifestation which we all called syndromes. Over time specific diseases were separated off from the syndrome. In some fields like cardiology the syndromes (like CHF) have become subordinate to the primary diseases so PettyBone is no longer a problem. (Although you make an interesting point about MACE).
We have learned that the measurements (criteria) defining these syndromes are not mathematically suitable for entry into a RCT function . In essence, concepts or rule-of-thumb heuristic may be quite useful clinically but an RCT cannot be snapped onto a heuristic. This is what happened in critical care.
Yet (correct me if I am mistaken) it seems that there is a homogeneous causative thread of hyperandrogenism with PCOS and with MACE there is a homogeneous causative thread of atherosclerosis (given that other causes of stroke etc excluded).
So at least an RCT of PCOS (with proper disease exclusions) is not at the extreme level of the PettyBone RCT where the subjects are defined by a task force generated triage set of non-disease specific thresholds. There is no homogeneous causative thread in a PB RCT.
Have the RCT been reproducible in the study of PCOS?
You make a beautiful point about the potential for PettyBone RCT mod to creep into disciplines.
The PettyBone RCT mod is the easiest RCT mod. because subjects are simply scooped up by triage. The case finding and disease diagnosing work and bottleneck are eliminated making this RCT mod the easiest path to a grant.
The statistician should always ask if the criteria capture a single disease or a set of different diseases.
For those who have been following this thread might not have ever seen a PettyBone RCT mod.
So here is a classic PettyBone RCT mod. Not even Bone would have performed this RCT because it extends the use of a guessed threshold set to the output not just the input (as in the classic PettyBone RCT mod). Itâs a guessed âthreshold sandwichâ of sorts, whereas the original PettyBone RCT mod only used a guessed set of thresholds as a triage for the subjects.
This present trial uses the 1996 guessed SOFA threshold summation score as the primary outcome. (That by itself has never worked.) see the discussion of the use of guessed threshold sets as measurements broadly in critical science.
Recall that in modern PettyBone science the guessed SOFA score is used as an input (as sepsis 3), an independent variable, and a primary or secondary output. These studies all fail.
The instant Iloprost study (This multicenter PettyBone RCT mod) tests a treatment (Iloprost) for massive SET of different diseases captured by a 1st GUESSED SET of triage thresholds. The primary outcome is the max daily SOFA score comprised of a 2nd GUESSED SET of thresholds.
Results are (secondary outcome of mortality)
âMortality at 90 days in the iloprost group was 57% (81 of 142) vs 51% (70 of 136) in the placebo groupâ
Here is the graph of primary outcome (daily mean max SOFA in the time domain).
Note: As patients worsen this SOFA score rises but as patients die they apparently drop out of the calculation so the âmean max SOFAâ falls as more patients die. It appears, athough I could not find it stated, that this is the mean max SOFA of the daily residual alive.
ââââ-
This is a beautiful illustration of PettyBone science.
Guessed triage set thresholds as the input measurement .
Guessed set of thresholds as the primary output.
The treated set derived from triage is disease and source agnostic.
Of course it failed, as all PettyBone RCT mod fail.
The worst part is that Iloprost might work for one or more diseases hidden in the mix. Who knows?
This PettyBone RCT mod was published in 2024, on the 35 year anniversary of Boneâs original failed methylprednisilone trial for âsepsisâ as defined by his guessed triage threshold set of SIRS.
Note, in the discussion of the trial, there is no discussion of the use of a PettyBone RCT mod as a potential limitation of the trial.
So most respectfully, both @ESMD and @Doc_Ed apparent opinions that this pathological science is known is not correct. Itâs entrenched.
Hence the outrage of @Rafael_Leite who was unknowingly indoctrinated and tricked to waste years of research (as most were) is the expected response.
Complacency and complicit resignation to the dogma out of social fear after 35 years of failure relevant the study of one of the greatest killers of adults and children, should be condemned.
âââ
Now think about a real Bradford Hill RCT testing Iloprost in the treatment Ecoli bacteremia due to UTI with mortality as an endpoint.
Treatment could start immediately pending future results and be withdrawn if E coli is not identified.
Notice how a non-disease agnostic RCT (a Bradford Hill RCT) provides a more homogeneous thread of unknown disease related variables and how a disease agnostic RCT is contaminated with massive unknown disease related heterogeneity which varies from one PB RCT to the next depending on the mix of diseases captured by the triage threshold set.
In any case, itâs time for all with sufficient courage to consider and debate, in the open, why the PettyBone RCT mod always fails and how to go forward without the PettyBone RCT mod.
Here is a classic PettyBone RCT, published this month, testing Aspirin to a set of different diseases captured by the triage threshold set of SOFA (sepsis3). The outcome was also the SOFA score so this is also a âPettyBone SOFA sandwich.â
The PettyBone RCT SOFA sandwich:
Input measurement-
âSepsis was defined according to the Sepsis-3 definition by the presence of a suspected or confirmed infection and an increase in the SOFA score of two points or more.â
Output measurement â
âThe primary outcome was the change in the Sequential Organ Failure Assessment (SOFA) score between day 0 and day 7 or date of discharge/death.â
Of course it was all negative but increased bleeding which caused stoppage. (Some of the subjects were harmed) .
(This harm and waste @ESMD is why I am âcognitively stuckâ. )
Here is how to do a PettyBone RCT mod.
Find a drug and create a hypothesis
Triage the patients using SOFA. (sepsis 3)
Randomize the triaged patients.
Give drug and placebo.
Measure SOFA as an outcome.
This is the easiest method of RCT performance. It was originally made in the USA. These Brazilians were intellectually colonized to use this pathological science.
Very sad. Harm to the subjects with no hope of success. Thatâs the PettyBone RCT!
If they had used SOFA as an outcome it would have been a tiny bit better. They did not; they used change in SOFA, which SOFA was not designed to be used for.