Adam -- I share the exact same feelings regarding guidelines! I wrote a whole thesis about it. Yet, I must say, EBM has originally been conceptualised as the very integration of the best research available (a lot of focus here, I know) and patients' clinical state, circumstances and preferences ;)
Excellent post. You emphasize the minimal benefit at 10 years for many patients to whom we recommend statins - implying guidelines may be too, let’s say liberal, in their approach. However the inverse could also be argued - that we are not accounting for 20,30,even 40 year risk, and that we should be treating earlier and more patients. Of course we don’t have much “evidence” to support that argument but “intuitively” it may be correct.
IMHO, the AHA's numbers are nonsensical. For starters, their 'ASCVD' definition is made up of roughly 30% fatal events. But their data source, a 2012 CTT meta-analysis, found no mortality benefit at ASCVD risk <20% over 10 years. That leaves 'nonfatal MI' as the majority of their remaining benefit they claim—but that is proven not to augur either death or disability (https://researchtranslation.substack.com/p/cardiologys-nonfatal-fatal-mistake). It typically means a troponin leak of, at BEST, uncertain significance. Humans taking statins aim to prevent death and disability, to live longer and better. By this measure roughly 10% of the benefits the AHA is counting from statin trials would be relevant. I'd suggest therefore increasing the NNTs on their graphs by 10x, to about 1,000 and 330. That would be closer, and still likely inflated. At these far more accurate numbers cholesterol reduction is overwhelmingly more harmful than helpful.
I find the lipid guidelines similar to the recently released blood pressure guidelines: designed to make more "patients" rather than improve well-being.
Do you think the guidelines are setting the groundwork to help expand the use of newer lipid lowering therapies for primary prevention by setting the therapeutic threshold lower. Even talking about statin and diabetes risk is reminiscent of the pharmaceutical company playbook to start damaging the reputation of the established out of patent treatment so it can be replaced. A “narrative” shift?
Great piece. I was never an EBM guru but always interested in it; and I am old enough to have seen Feinstein in action. He could be ferocious; as you point out, he was incredibly smart. I share your scepticism about guidelines but it seems to me that a common problem with them is often missed; they are issued for clinical domains for which issuing a guideline is simply wrong; and even when there is a case for a guideline, they are often far too specifically directive given the state of the evidence and far too reliant for their direction on expert opinion rather than science, which, as you say, is likely informed by many considerations other than evidence. The result is damage to practice, damage magnified when guidelines are turned into performance measures. As my colleagues pointed out long ago, guidelines are appropriate “for diseases having significant practice variability and for which a valid evidence base can guide recommendations” as per here: https://jamanetwork.com/journals/jama/article-abstract/183430. As others have pointed out, the most politically inflected guidelines (e.g. transgender care) completely miss the mark of this standard, telling clinicians what to do when there is little or no good science to back up the guideline.
Probably from the original 1992 JAMA paper Adam. Eg. In the Abstract:
“Evidence-based medicine de-emphasizes intuition, unsystematic clinical experience, and pathophysiologic rationale as sufficient grounds for clinical decision making and stresses the examination of evidence from clinical research.”
The new guidelines will not change my practice. The emphasis on CAC scores/ lipoproteins/ use of newer($$) non statin drugs have minimal end point data to support them.
I do agree that we should consider other risk factors (FH/ exercise etc) in discussions with patients. If we consider ARR and NNT for most the Cor A1/2 recommendations, the value is often minimal. These thoughts primarily apply to primary prevention but are not irrelevant to newer non statin drugs for secondary prevention.
I strenuously agree with the whole enchilada. This statement in particular summarizes my sentiments about guidelines succinctly: “The longer I spend with them, the less of a fan I am.”
Guidelines are now no better than recipes. And most people who now use them function like cooks in a kitchen rather than physicians. I see this blind uncritical adherence daily. I have even less charitable things to say about those who write them.
I would love to see the day that guidelines are written by critical appraisal/EBM experts (a platoon of Dr. JMM’s neutral Martians would do nicely)….rather than the conflicted/bought-and-paid-for subject matter experts that pass as guideline writers these days. Of course, the latest iteration of lipid guidelines is simply the most recent example of the archetype.
You had me at hello. Rather with CW, a male age 60 and dyslipidemia. My age is higher than CW, my LDL, while high, is significantly lower than CW's. Let me give you a patient's perspective on confronting dyslipidemia, the guidelines and the PCP interface.
My PCP is reading through my blood test results. He breezes through the lipid portion until I say, "That sounds kind of high". He stops and reflects back on the numbers and agrees. He gets out his handy calculator and tells me my CVD risk. He tells me he wants to do a calcium test, which in his new offices with a 5-storey atrium, they now have onsite capability. He states that, "If you have a score of zero we can talk". Anything other than zero means it is statins for me. Other than a year on 5mg of amlodipine when he said, "I am probably overtreating you", in response to the new push for lower BP scores, I have never taken prescriptions. No health issues. Healthy BMI. Nothing. [BP was then 115/82 and got as low as 85/63 while on the drug.]
In the guidelines, it says I am at moderate risk based on my lipids and age. PCP prescribes a high intensity regimen per the guidelines. In addition, a baby aspirin. Aspirin is nowhere in the guidelines and against EBM to what I can tell. At my age, there is a higher incidence for serious adverse effects from statins. Lower dosage levels are recommended from what I read. That is not the prescription given.
In my reading, maybe as many as half the people who start a statin, discontinue by the end of the first year. The guideline mentions no such thing. Wouldn't that be something important if you are interested in treating dyslipidemia? The only mention of adverse effects is in conjunction with other drugs. Of the people I know, I know 4 who started a statin regimen. One, a woman in her early 60's started but discontinued due to musculoskeletal pain. A leading adverse effect of statins. She lived at home alone for 20-some years and then in independent living for 5. Died 2-months short of 90, with COPD/HF. The guidelines mention 150 minutes of exercise per week. Muscle pain would seem to inhibit the exercise goal for patients in the guidelines.
The guidelines do specifically mention an adverse event. "...incident diabetes as a result of statin initiation is the most common serious adverse event attributed to statins". Of the other 3 people I know on statins, one died at 74 from cancer (UESL). Toward the end, she had no wish for any more days of life. She was pre-diabetic. Another is mid-70's, is pre-diabetic and loves Rueben sandwiches, which he still eats as his lipids are in range. The fourth is a 50-year plus friend. He referred me to my PCP. He went from statins to pre-diabetic and then diabetic. No, these four are not statistics, but it is hard to not be influenced by them. Covid highlighted how harmful diabetes is to whole person health.
As for my dyslipidemia, I have modified my diet and increased my weekly exercise to over 200 minutes. In the first 2-months, I have lost a few pounds and lowered my LDL by 23%. Total cholesterol is now in range, but work remains for LDL. My new regimen is easily sustainable for me, without the potential for adverse effects that might defeat my health goals.
Yes, the guidelines are too long. But from my perspective, its obvious limitations mean they will never be as successful as we should want them to be.
Adam your comments are bang on - that is why our PEER group (peerevidence.ca) has decided over the last 10 years to write "Simplified" guidelines for primary care as we were unable to have any impact/input on national guidelines. Our 2023 simplified lipid guideline https://www.cfp.ca/content/69/10/675 is 11 pages (in contrast to the 134 pages of the 2026 American guideline) and we provide simple web-based tools https://decisionaid.ca/cvd/ that provide clinicians with the numbers they need to do shared-decision making. This tool has been accessed by 100,000 + people over the last year. In other words it is possible to write guidelines that people like Adam would hopefully not say are "divorced" .. "from any consideration of value"
It has been obvious for years that guidelines absolutely cannot be trusted in any medical category linked to politics - any guidelines regarding transgender care, or dealing with vaccines, for instance. Unfortunately, you also can’t trust guidelines linked deeply to new sources of major pharmaceutical profits - ADHD ones, or weight loss ones. But I always figured that the legitimate, respectable, apolitical old fields, like say cardiology and oncology, were trustworthy - then I started reading y’all and learned from Vinay that oncology is full of garbage studies and from you that cardiology is full of garbage studies. Moral of the story: I’m sure at least all the asthma guidelines must be faultless! ; )
Much, much more on the profound flaws of guidelines and the inevitable death of human medicine here:
I think this is a great perspective and illustrates the tension we face when trying to practice EBM.
I share your feelings about guidelines in general—I wrote fairly critically of cancer screening guidelines last September. However, I think that guidelines can and should have a place in medical practice.
Most practicing clinicians will never be able to keep up with the primary literature on their scope of practice, especially generalists like us. Having a panel of experts parse out the evidence and distill it down to something actionable makes a lot of sense. But the authoritative epistemic posture of guideline panels today is a major problem.
“We’re experts so you should do what we say” isn’t (and shouldn’t be) acceptable with the level of distrust in the medical field today. As I argued in my screening guideline essay, guidelines would serve clinicians better by equipping rather than directing. Teach the uncertainty and nuance in the evidence and use clear statistics such as NNT and NNH so that practicing clinicians can more easily apply them to individuals. The blanket approach to recommendations is a disservice, especially when the evidence is weak.
I would also prefer to see simpler guidelines that focus only on the interventions that improve meaningful outcomes based on high-quality evidence. Making what should be a ten page document into a few hundred puts it out of reach for most. Standardize practice around the best evidence. We don’t really need to make recommendations about every obscure item in the topic. Find another document in which to publish that.
Anyway, sorry about the mini-essay in the comments—you got me thinking too much this morning!
Great point. In my own field, I have no problem spotting bs in guidelines as I’m familiar with the source literature. But as soon as I venture outside my clinical area (such as even when I look at PE guidelines), I am completely at the mercy of guideline writers as I don’t have the foundation to separate wheat from chaff. And then I wonder whether their guideline writers are as conflicted as those in my field tend to be. And then I see a study like HI-PEITHO and it becomes more difficult to resist nihilistic urges.
Try arguing over whether or not a patient with pneumococcal pneumonia on ceftriaxone and azithromycin can have the azithromycin stopped. Never mind the issue of whether or not the bactericidal action of the ceftriaxone is inhibited by the azithromycin - I'm just pointing at the pure fact of unnecessary therapy.
"Wait, wait! He has CAP, doesn't he? It says right here...."
In a lecture during my first week in medical school, an older practitioner said, "You will all want to practice according to the literature (that's what they called EBM back in 1972). There are two problems with that: First, half of what's in the literature is wrong; Second, we don't know which half". The next lecture was by a young man who exhorted us all to go into radiology or surgery, because "There's a lab in Pittsburgh with a computer program called "internist". Within five years, this program will make all the diagnoses, no need for medicine".
Later in my career, I worked alongside a very productive group which studied, and published, on Clinical Decision Making (CDM), the premise of which was that iterative hypothesis testing was a) how diagnoses were made, and b) that machines could do it faster.
Now, I hear that AI makes the diagnosis correctly >95% of the time, and when it takes an MD as a collaborator, the percent correct diagnoses fall into the high 20s.
Based on my (extensive) experience with the problem of diagnosis, it seems to me that fact-stacking (the "literature", EBM, CDM, LLM) isn't really the way we do this. It's how medical students and junior residents do it, but that's because they have no experience and no library of personal wrong turns and mistaken assumptions to draw on, and no community of colleagues to assist with their own experiences.
I have, many times, used the quip "the plural of anecdote is not data" - true, as far as it goes. But "data" is only part of diagnosis, and I venture to say, not the largest part.
However, EBM can help one find and try to reduce personal biases which affect how one perceives the evidence, which were in part the reason for EBM in the first place, weren't they?
"Rather than the absence of any harm, it is the expectation of an overall benefit of a medical treatment that is the foundation of the implicit doctor-patient contract. In the context of an expectation of efficacy, powerful cognitive biases can blind clinicians to obvious signs that a treatment is not helping, or may even be harming their patients. With examples from medical history and current clinical controversies, this paper examines how systematic psychological biases can distort not just individual decision making, but perceptions of the evidence base upon which clinical decisions are built. These distortions can perpetuate harmful practices in medicine long after the objective evidence points in a different direction. By becoming aware of these biases and the way they shape perceptions of the evidence base, doctors can reduce the negative impact they may have on the patients in their care."
Yes, I am aware of the problem of cognitive bias. As an English major, I am also aware of the problems of isolation and dehumanization.
In classical Internal Medicine training (now, sadly, dead), one of the common catchphrases was, “YOU are the treatment”, meaning, approaching the patient as a fellow human rather than as an object was part of therapy. You cannot be fully human shorn of your experiences, which, I think, is a better term than “cognitive biases”.
" But the more I read, the more I feel that guidelines are overly influenced by people with either financial or intellectual conflicts of interest. The power on the guideline writing committees is seldom held by experts in research analysis, but rather by content experts. Too many recommendations are “strong but based on weak evidence.”
There are criteria for guideline committees that those with conflict of interests be a minority, right? ("Clinical Practice Guidelines We Can Trust", Graham, 2011, https://www.ncbi.nlm.nih.gov/books/NBK209539/ ). There is also "How to recognize a trustworthy clinical practice guideline" (Lima et al., 2023). You do say "power" rather than "majority" but one could hope that the process would respect the goal? And of course, conflicts of interests should not only be listed but also how they are managed should be provided.
I think also that strong recommendations based upon weak evidence is not considered appropriate except in very limited circumstances (there are something like 5 situations where it is "ok", discussed in this paper: Strong recommendations from low certainty evidence: a cross-sectional analysis of a suite of national guidelines (Chong et al., 2021), https://link.springer.com/article/10.1186/s12874-023-01895-8).
If one is interested in the formal requirements, there is a lecture by Ivan Florez, What is a trustworthy guideline? (https://www.youtube.com/watch?v=smRzaGjQU0Y ). He's a pediatrician, part time teaching at McMaster and professor in Colombia, Director of Cochrane Colombia, etc. In particular, he is the lead on the AGREE collaboration, which assesses guidelines. The talk is general, although it is part of a conference on evidence-based medicine for gender dysphoria in young people (https://segm.org/cme25 ). This field's guidelines are almost all poor quality (See the AGREE-II review of them by Taylor et al., 2024, Clinical guidelines for children and adolescents experiencing gender dysphoria or incongruence: a systematic review of guideline quality (part 1) and (part 2), both available at this site at Archives of Diseases in Childhood by BMJ: https://adc.bmj.com/pages/gender-identity-service-series ).
There is also a really good 2021 article, "How to Interpret and Use a Clinical Practice Guideline or Recommendation: Users' Guides to the Medical Literature" by Brignardello-Petersen, Carrasco,-Labra and Guyatt.
I know this is all theoretical, but if the guideline is not even trustworthy, it's even worse than what you are describing.....
Thank you as always for a nuanced perspectives. My first thought when I heard about these guidelines was the same - is the goal really to put everyone on medication turning them into patients?? Americans already take more medications than other nations for more money and poorer outcomes. If we really wanted to move the needle, we’d spend more of our money on public health initiatives that change how we live.
Adam -- I share the exact same feelings regarding guidelines! I wrote a whole thesis about it. Yet, I must say, EBM has originally been conceptualised as the very integration of the best research available (a lot of focus here, I know) and patients' clinical state, circumstances and preferences ;)
Excellent post. You emphasize the minimal benefit at 10 years for many patients to whom we recommend statins - implying guidelines may be too, let’s say liberal, in their approach. However the inverse could also be argued - that we are not accounting for 20,30,even 40 year risk, and that we should be treating earlier and more patients. Of course we don’t have much “evidence” to support that argument but “intuitively” it may be correct.
IMHO, the AHA's numbers are nonsensical. For starters, their 'ASCVD' definition is made up of roughly 30% fatal events. But their data source, a 2012 CTT meta-analysis, found no mortality benefit at ASCVD risk <20% over 10 years. That leaves 'nonfatal MI' as the majority of their remaining benefit they claim—but that is proven not to augur either death or disability (https://researchtranslation.substack.com/p/cardiologys-nonfatal-fatal-mistake). It typically means a troponin leak of, at BEST, uncertain significance. Humans taking statins aim to prevent death and disability, to live longer and better. By this measure roughly 10% of the benefits the AHA is counting from statin trials would be relevant. I'd suggest therefore increasing the NNTs on their graphs by 10x, to about 1,000 and 330. That would be closer, and still likely inflated. At these far more accurate numbers cholesterol reduction is overwhelmingly more harmful than helpful.
I find the lipid guidelines similar to the recently released blood pressure guidelines: designed to make more "patients" rather than improve well-being.
Do you think the guidelines are setting the groundwork to help expand the use of newer lipid lowering therapies for primary prevention by setting the therapeutic threshold lower. Even talking about statin and diabetes risk is reminiscent of the pharmaceutical company playbook to start damaging the reputation of the established out of patent treatment so it can be replaced. A “narrative” shift?
Great piece. I was never an EBM guru but always interested in it; and I am old enough to have seen Feinstein in action. He could be ferocious; as you point out, he was incredibly smart. I share your scepticism about guidelines but it seems to me that a common problem with them is often missed; they are issued for clinical domains for which issuing a guideline is simply wrong; and even when there is a case for a guideline, they are often far too specifically directive given the state of the evidence and far too reliant for their direction on expert opinion rather than science, which, as you say, is likely informed by many considerations other than evidence. The result is damage to practice, damage magnified when guidelines are turned into performance measures. As my colleagues pointed out long ago, guidelines are appropriate “for diseases having significant practice variability and for which a valid evidence base can guide recommendations” as per here: https://jamanetwork.com/journals/jama/article-abstract/183430. As others have pointed out, the most politically inflected guidelines (e.g. transgender care) completely miss the mark of this standard, telling clinicians what to do when there is little or no good science to back up the guideline.
On “and don’t know where they originated”…
Probably from the original 1992 JAMA paper Adam. Eg. In the Abstract:
“Evidence-based medicine de-emphasizes intuition, unsystematic clinical experience, and pathophysiologic rationale as sufficient grounds for clinical decision making and stresses the examination of evidence from clinical research.”
https://jamanetwork.com/journals/jama/article-abstract/400956
The new guidelines will not change my practice. The emphasis on CAC scores/ lipoproteins/ use of newer($$) non statin drugs have minimal end point data to support them.
I do agree that we should consider other risk factors (FH/ exercise etc) in discussions with patients. If we consider ARR and NNT for most the Cor A1/2 recommendations, the value is often minimal. These thoughts primarily apply to primary prevention but are not irrelevant to newer non statin drugs for secondary prevention.
Fantastic piece. One “like” seems insufficient.
I strenuously agree with the whole enchilada. This statement in particular summarizes my sentiments about guidelines succinctly: “The longer I spend with them, the less of a fan I am.”
Guidelines are now no better than recipes. And most people who now use them function like cooks in a kitchen rather than physicians. I see this blind uncritical adherence daily. I have even less charitable things to say about those who write them.
I would love to see the day that guidelines are written by critical appraisal/EBM experts (a platoon of Dr. JMM’s neutral Martians would do nicely)….rather than the conflicted/bought-and-paid-for subject matter experts that pass as guideline writers these days. Of course, the latest iteration of lipid guidelines is simply the most recent example of the archetype.
You had me at hello. Rather with CW, a male age 60 and dyslipidemia. My age is higher than CW, my LDL, while high, is significantly lower than CW's. Let me give you a patient's perspective on confronting dyslipidemia, the guidelines and the PCP interface.
My PCP is reading through my blood test results. He breezes through the lipid portion until I say, "That sounds kind of high". He stops and reflects back on the numbers and agrees. He gets out his handy calculator and tells me my CVD risk. He tells me he wants to do a calcium test, which in his new offices with a 5-storey atrium, they now have onsite capability. He states that, "If you have a score of zero we can talk". Anything other than zero means it is statins for me. Other than a year on 5mg of amlodipine when he said, "I am probably overtreating you", in response to the new push for lower BP scores, I have never taken prescriptions. No health issues. Healthy BMI. Nothing. [BP was then 115/82 and got as low as 85/63 while on the drug.]
In the guidelines, it says I am at moderate risk based on my lipids and age. PCP prescribes a high intensity regimen per the guidelines. In addition, a baby aspirin. Aspirin is nowhere in the guidelines and against EBM to what I can tell. At my age, there is a higher incidence for serious adverse effects from statins. Lower dosage levels are recommended from what I read. That is not the prescription given.
In my reading, maybe as many as half the people who start a statin, discontinue by the end of the first year. The guideline mentions no such thing. Wouldn't that be something important if you are interested in treating dyslipidemia? The only mention of adverse effects is in conjunction with other drugs. Of the people I know, I know 4 who started a statin regimen. One, a woman in her early 60's started but discontinued due to musculoskeletal pain. A leading adverse effect of statins. She lived at home alone for 20-some years and then in independent living for 5. Died 2-months short of 90, with COPD/HF. The guidelines mention 150 minutes of exercise per week. Muscle pain would seem to inhibit the exercise goal for patients in the guidelines.
The guidelines do specifically mention an adverse event. "...incident diabetes as a result of statin initiation is the most common serious adverse event attributed to statins". Of the other 3 people I know on statins, one died at 74 from cancer (UESL). Toward the end, she had no wish for any more days of life. She was pre-diabetic. Another is mid-70's, is pre-diabetic and loves Rueben sandwiches, which he still eats as his lipids are in range. The fourth is a 50-year plus friend. He referred me to my PCP. He went from statins to pre-diabetic and then diabetic. No, these four are not statistics, but it is hard to not be influenced by them. Covid highlighted how harmful diabetes is to whole person health.
As for my dyslipidemia, I have modified my diet and increased my weekly exercise to over 200 minutes. In the first 2-months, I have lost a few pounds and lowered my LDL by 23%. Total cholesterol is now in range, but work remains for LDL. My new regimen is easily sustainable for me, without the potential for adverse effects that might defeat my health goals.
Yes, the guidelines are too long. But from my perspective, its obvious limitations mean they will never be as successful as we should want them to be.
Adam your comments are bang on - that is why our PEER group (peerevidence.ca) has decided over the last 10 years to write "Simplified" guidelines for primary care as we were unable to have any impact/input on national guidelines. Our 2023 simplified lipid guideline https://www.cfp.ca/content/69/10/675 is 11 pages (in contrast to the 134 pages of the 2026 American guideline) and we provide simple web-based tools https://decisionaid.ca/cvd/ that provide clinicians with the numbers they need to do shared-decision making. This tool has been accessed by 100,000 + people over the last year. In other words it is possible to write guidelines that people like Adam would hopefully not say are "divorced" .. "from any consideration of value"
It has been obvious for years that guidelines absolutely cannot be trusted in any medical category linked to politics - any guidelines regarding transgender care, or dealing with vaccines, for instance. Unfortunately, you also can’t trust guidelines linked deeply to new sources of major pharmaceutical profits - ADHD ones, or weight loss ones. But I always figured that the legitimate, respectable, apolitical old fields, like say cardiology and oncology, were trustworthy - then I started reading y’all and learned from Vinay that oncology is full of garbage studies and from you that cardiology is full of garbage studies. Moral of the story: I’m sure at least all the asthma guidelines must be faultless! ; )
Much, much more on the profound flaws of guidelines and the inevitable death of human medicine here:
https://gaty.substack.com/p/the-three-wise-men-walk-into-a-doctors
I think this is a great perspective and illustrates the tension we face when trying to practice EBM.
I share your feelings about guidelines in general—I wrote fairly critically of cancer screening guidelines last September. However, I think that guidelines can and should have a place in medical practice.
Most practicing clinicians will never be able to keep up with the primary literature on their scope of practice, especially generalists like us. Having a panel of experts parse out the evidence and distill it down to something actionable makes a lot of sense. But the authoritative epistemic posture of guideline panels today is a major problem.
“We’re experts so you should do what we say” isn’t (and shouldn’t be) acceptable with the level of distrust in the medical field today. As I argued in my screening guideline essay, guidelines would serve clinicians better by equipping rather than directing. Teach the uncertainty and nuance in the evidence and use clear statistics such as NNT and NNH so that practicing clinicians can more easily apply them to individuals. The blanket approach to recommendations is a disservice, especially when the evidence is weak.
I would also prefer to see simpler guidelines that focus only on the interventions that improve meaningful outcomes based on high-quality evidence. Making what should be a ten page document into a few hundred puts it out of reach for most. Standardize practice around the best evidence. We don’t really need to make recommendations about every obscure item in the topic. Find another document in which to publish that.
Anyway, sorry about the mini-essay in the comments—you got me thinking too much this morning!
Great point. In my own field, I have no problem spotting bs in guidelines as I’m familiar with the source literature. But as soon as I venture outside my clinical area (such as even when I look at PE guidelines), I am completely at the mercy of guideline writers as I don’t have the foundation to separate wheat from chaff. And then I wonder whether their guideline writers are as conflicted as those in my field tend to be. And then I see a study like HI-PEITHO and it becomes more difficult to resist nihilistic urges.
It’s not nihilistic to point out nonsense being marketed as how doctors should practice.
Re: "guidelines".
Try arguing over whether or not a patient with pneumococcal pneumonia on ceftriaxone and azithromycin can have the azithromycin stopped. Never mind the issue of whether or not the bactericidal action of the ceftriaxone is inhibited by the azithromycin - I'm just pointing at the pure fact of unnecessary therapy.
"Wait, wait! He has CAP, doesn't he? It says right here...."
Great points. Thanks.
In a lecture during my first week in medical school, an older practitioner said, "You will all want to practice according to the literature (that's what they called EBM back in 1972). There are two problems with that: First, half of what's in the literature is wrong; Second, we don't know which half". The next lecture was by a young man who exhorted us all to go into radiology or surgery, because "There's a lab in Pittsburgh with a computer program called "internist". Within five years, this program will make all the diagnoses, no need for medicine".
Later in my career, I worked alongside a very productive group which studied, and published, on Clinical Decision Making (CDM), the premise of which was that iterative hypothesis testing was a) how diagnoses were made, and b) that machines could do it faster.
Now, I hear that AI makes the diagnosis correctly >95% of the time, and when it takes an MD as a collaborator, the percent correct diagnoses fall into the high 20s.
Based on my (extensive) experience with the problem of diagnosis, it seems to me that fact-stacking (the "literature", EBM, CDM, LLM) isn't really the way we do this. It's how medical students and junior residents do it, but that's because they have no experience and no library of personal wrong turns and mistaken assumptions to draw on, and no community of colleagues to assist with their own experiences.
I have, many times, used the quip "the plural of anecdote is not data" - true, as far as it goes. But "data" is only part of diagnosis, and I venture to say, not the largest part.
However, EBM can help one find and try to reduce personal biases which affect how one perceives the evidence, which were in part the reason for EBM in the first place, weren't they?
Regarding these biases, see, for instance, Baxendale's "How to be a Better Doctor: Recognizing How Cognitive Biases Shape-and Distort-Clinical Evidence" (from 2025, https://www.imrpress.com/journal/BJHM/86/2/10.12968/hmed.2024.0743 ).
"Rather than the absence of any harm, it is the expectation of an overall benefit of a medical treatment that is the foundation of the implicit doctor-patient contract. In the context of an expectation of efficacy, powerful cognitive biases can blind clinicians to obvious signs that a treatment is not helping, or may even be harming their patients. With examples from medical history and current clinical controversies, this paper examines how systematic psychological biases can distort not just individual decision making, but perceptions of the evidence base upon which clinical decisions are built. These distortions can perpetuate harmful practices in medicine long after the objective evidence points in a different direction. By becoming aware of these biases and the way they shape perceptions of the evidence base, doctors can reduce the negative impact they may have on the patients in their care."
Yes, I am aware of the problem of cognitive bias. As an English major, I am also aware of the problems of isolation and dehumanization.
In classical Internal Medicine training (now, sadly, dead), one of the common catchphrases was, “YOU are the treatment”, meaning, approaching the patient as a fellow human rather than as an object was part of therapy. You cannot be fully human shorn of your experiences, which, I think, is a better term than “cognitive biases”.
It's a really nice paper..maybe you'd like it.
Thank you!
" But the more I read, the more I feel that guidelines are overly influenced by people with either financial or intellectual conflicts of interest. The power on the guideline writing committees is seldom held by experts in research analysis, but rather by content experts. Too many recommendations are “strong but based on weak evidence.”
There are criteria for guideline committees that those with conflict of interests be a minority, right? ("Clinical Practice Guidelines We Can Trust", Graham, 2011, https://www.ncbi.nlm.nih.gov/books/NBK209539/ ). There is also "How to recognize a trustworthy clinical practice guideline" (Lima et al., 2023). You do say "power" rather than "majority" but one could hope that the process would respect the goal? And of course, conflicts of interests should not only be listed but also how they are managed should be provided.
I think also that strong recommendations based upon weak evidence is not considered appropriate except in very limited circumstances (there are something like 5 situations where it is "ok", discussed in this paper: Strong recommendations from low certainty evidence: a cross-sectional analysis of a suite of national guidelines (Chong et al., 2021), https://link.springer.com/article/10.1186/s12874-023-01895-8).
If one is interested in the formal requirements, there is a lecture by Ivan Florez, What is a trustworthy guideline? (https://www.youtube.com/watch?v=smRzaGjQU0Y ). He's a pediatrician, part time teaching at McMaster and professor in Colombia, Director of Cochrane Colombia, etc. In particular, he is the lead on the AGREE collaboration, which assesses guidelines. The talk is general, although it is part of a conference on evidence-based medicine for gender dysphoria in young people (https://segm.org/cme25 ). This field's guidelines are almost all poor quality (See the AGREE-II review of them by Taylor et al., 2024, Clinical guidelines for children and adolescents experiencing gender dysphoria or incongruence: a systematic review of guideline quality (part 1) and (part 2), both available at this site at Archives of Diseases in Childhood by BMJ: https://adc.bmj.com/pages/gender-identity-service-series ).
There is also a really good 2021 article, "How to Interpret and Use a Clinical Practice Guideline or Recommendation: Users' Guides to the Medical Literature" by Brignardello-Petersen, Carrasco,-Labra and Guyatt.
I know this is all theoretical, but if the guideline is not even trustworthy, it's even worse than what you are describing.....
Thank you as always for a nuanced perspectives. My first thought when I heard about these guidelines was the same - is the goal really to put everyone on medication turning them into patients?? Americans already take more medications than other nations for more money and poorer outcomes. If we really wanted to move the needle, we’d spend more of our money on public health initiatives that change how we live.