Excellent "springboard"(Caution: Buzzword alert) for thinking about the often-ignored issue of having a test with power appropriately built in to detect outcome X in huge Groups of patients as opposed to detecting X in arbitrarily chosen, small Subgroups during the almost traditional post-trial deep massage of data, with or without incense burning in the massage parlor.
No guideline fits all patients, in fact no guideline fits any individual patient perfectly. Flexibility is not just desirable, it's mandatory. This is where physician training, insight, and expert judgment is required. Brent James taught me this many years ago. His work is worth revisiting. Reward and punishment tied to top-down metrics is a bad idea in the practice of medicine. Want more? See Don Berwick's "The Toxicity of Pay for Performance" D.M. Berwick. Qual Manag Health Care. 1995 Fall.
The HF space has been hugely disappointing for me insofar as societies and guidelines venturing far beyond the foundational studies and evidence, and the engagement in groupthink and cheerleading.
There is not 1 RCT investigating the benefits of rapid initiation and/or inpatient initiation of “quadruple therapy” on clinical endpoints, never mind hard endpoints (the best we’ve gotten is surrogates like NT-BNP). Yet opinion leaders are tripping over themselves strongly advocating such a strategy. It makes a mockery of the very concept of EBM itself. It makes me cynical and suspicious of the motivations of the societies and guideline writers when such egregious liberties are being represented in work-product that should instead be a neutral distillation of the evidence.
As an old-fashioned cardiologist now long retired, I recall treating many patients with atrial fibrillation and various degrees of heart failure with digoxin. When ACE inhibitors came out I would add that for patients with orthopnea or other congestive symptoms. The vast majority of patients seemed to do quite well with this regimen. Digoxin was cheap, effective and well tolerated. I tried some on calcium blockers when they were being touted but, in my purely anecdotal experience, they were not as effective or as well tolerated as the digoxin. Of course in those days it was absolutely forbidden to give beta blockers to anyone with heart failure. I wonder if anyone ever ran a study comparing the newer drugs with digoxin. Probably not likely as the newer and sexier (and vastly cheaper) drugs soon took over the market.
John, since atrial fibrillation causes an irregular rhythm, the initial treatment is to decrease the heart rate to diminish myocardial oxygen consumption. Slowing down the heart rate tends to normalize the rhythm in patients with AF. Beta blockers have traditionally been the drugs of choice (atenolol or metoprolol, usually), with non-DHP calcium channel blocks (diltiazem or verapamil) as clear second-choice medications. I wonder, given the tendency of a heart in atrial fibrillation to throw blood clots, if aggressively slowing the heart rate increases the clot risk of patients, leading to their demise. Could this be part of the picture? Whatever benefits beta-blockers confer are offset by the increased clot risk when the heart is slowed too aggressively?
My eyes tend to glaze over when I read the words "meta-analysis". Offhand I can't recall any that were not just a bunch of studies with insignificant differences that were cobbled together in order to engineer a statistically significant result. Having dealt with the heterogeneity of heart failure over a forty year career, I am extremely skeptical about the possibility of matching groups in any study of significant size. Also curious about the definition of "low bias study".
I agree with Ernest Curtis' skeptical opinion regarding what ACTUALLY goes in when a bunch of studies are cobbled together and then the massed data analyzed using various types of statistical Kung Fu.
“I advocate for the use of evidence in the strongest terms. But this does not equate to blindly following overly simplistic algorithms, especially when tied to financial incentives. These are a blight on the evidence-based practice movement.”
This is such an important paragraph. Having worked in pain management AND addiction for well over two decades, I can remember the “evidence” presented that chronic pain + opioids RARELY led to addiction; an issue I argued loudly at the time. I was often reminded of the “evidence...”
My question was WHOSE evidence? We did not know what we know now back then, and the EBM protocol was pushed in my face time and again. Blind overprescribing of opioids certainly has shown evidence that the EBM in practice by many back then wasn’t working. I’m glad to see your excellent essay that can be applied to all areas. Blind Faith should just be the band, not medical practice (yes I just aged myself!)
"...blindly following overly simplistic algorithms, especially when tied to financial incentives." Is exactly how big bureaucratic organizations (alas, the vast majority of healthcare providers worldwide) approach EBM by default.
Do you all think it's an offset of people who might benefit from meds vs those who got worse? You hope that someone does not prescribe a beta blocker to a patient with AFib before assessing the heart rate based on a EMR box . Reminds me of my EP colleague who asks if my consult was a consult to see if a patient needs a pacemaker or a work order to implant . I cared for many elderly people in Texas "L" towns. The "L" is the name of the town and the 80 year olds live with mom and dad ! I would ask for a pacer knowing the current HR was causing symptoms but not syncope etc. The patients were "ok" before but dramatically better after implant. The reason we are paid more than people who just follow the guidelines is because the guidelines are supposed to help us REMEMBER the option to treat a certain way NOT tell us what to do. The problem that everyone forgets is the guideline suggestion is only as good as the data input. The history and physical are not done by the chart but by the physician. An "asymptomatic" 350 pound patient who does not even walk to mailbox is not "asymptomatic".... great article as usual
Since this was a post hoc analysis, how were the placebo and beta blocker groups in Afib matched for baseline characteristics? Perhaps the beta blocker group was sicker.
Love how John Mandrola advocates for careful use of evidenced based medicine when caring for actual patients. The EMR can be a curse - automatic recommendations based on one diagnosis are necessarily mindless.
The problem witn evidence based medicine and guidelines is that they have not be taken as "laws".
HF is an excellent example about treatment and reproducibility: one question. The evidence favoring the use of ARNI is the same than with ACEi? Another one, the up titriation concept developed with ACEi and beta blockers is completely over in order to favor the adition of ARNI an Gliflozins? Do we throw away decades of investigation ?
What about any morbidity benefit (decreased hospitalizations for decompensated HF)? A lot of emphasis is placed on mortality benefit in cardiology trials. Morbidity benefit (if it can be proven) would be very important for the patient (improved quality of life) and for CMS (since HF is the biggest contributor to CMS budget). I'm not sure that the meta analysis 'proves' GDMT is inappropriate for patients with HF and AF. Perhaps a more nuanced discussion would be to clarify endpoints (goals of therapy). Sometimes death is not the worst outcome; prolonged debilitation and / or suffering can be far worse outcome for many. This is not at all to dismiss premature death in young and middle aged as a source of much grief and suffering.
I’m not sure that evidence-based medicine makes any sense, John.
The problem is, that it requires explanatory theories first, and that is often ignored. Studies are done that get financed by Big Pharma largely. There are some that are done without that funding, and there are some older studies that were done before the massive corruption.
But for example, for heart failure, I am convinced that small amounts of T3 can reverse HF. Yet the studies aren’t really done because there is no money in that, and no explanatory theories that are currently in vogue supporting it.
There is certainly evidence for decades that raising T3 is the solution to HF. For instance several studies showed risk of death higher with lower T3.
All-cause mortality was 23% (n = 64) after a mean (±SD) of 12 ± 7 months of follow-up; 47 (73%) of the patients died from cardiac causes. The mean ejection fraction was lower in those patients who died than in those who survived (26% ± 8% vs. 31% ± 8%, P < 0.001), as were levels of total T3 (1.0 ± 0.4 nmol/L vs. 1.3 ± 0.3 nmol/L, P < 0.001) and free T3 (3.2 ± 1.4 pmol/L vs. 3.7 ± 1.0 pmol/L, P < 0.001). In a multivariate model, ejection fraction (odds ratio [OR] = 2.0 per 10% decrease; 95% confidence interval [CI]: 1.4 to 2.8 per 10% decrease; P < 0.001) and total T3 level (OR = 0.3 per 1-nmol/L increase; 95% CI: 0.1 to 0.5 per 1-nmol/L increase; P < 0.001) were the only independent predictors of all-cause mortality. In an alternative model using free T3 levels, ejection fraction (OR = 1.9; 95% CI: 1.4 to 2.7; P < 0.001) and free T3 level (OR = 0.6 per 1 pmol/L; 95% CI: 0.5 to 0.8 per 1 pmol/L; P <0.02) were associated with all-cause mortality. When we considered cardiac mortality alone, male sex (OR = 3.5; 95% CI: 1.7 to 13; P < 0.04), ejection fraction (OR = 1.7; 95% CI: 1.2 to 2.5; P < 0.006), and total T3 level (OR = 0.3; 95% CI: 0.2 to 0.7; P < 0.002) were independent predictors with the multivariate model.
Conclusion
Low T3 levels are an independent predictor of mortality in patients with chronic heart failure, adding prognostic information to conventional clinical and functional cardiac parameters.
And yet, there are virtually no doctors who pay any attention to this. HF is a result of very low cellular energetics and often accomplices type 2 diabetes and other metabolic issues that come from high fat burning, minimal oxidative phosphorylation, high glycolysis producing lactate, high build up of lactate…
The explanatory theory would suggest, let’s give patients small doses of T3 and get their cellular energetics back up. And it works. But nobody is doing it, nobody talks about it, because the “evidence” isn’t there because Big Pharma doesn’t want the evidence to be there.
That’s the problem with the “evidence based” approach.
Excellent "springboard"(Caution: Buzzword alert) for thinking about the often-ignored issue of having a test with power appropriately built in to detect outcome X in huge Groups of patients as opposed to detecting X in arbitrarily chosen, small Subgroups during the almost traditional post-trial deep massage of data, with or without incense burning in the massage parlor.
No guideline fits all patients, in fact no guideline fits any individual patient perfectly. Flexibility is not just desirable, it's mandatory. This is where physician training, insight, and expert judgment is required. Brent James taught me this many years ago. His work is worth revisiting. Reward and punishment tied to top-down metrics is a bad idea in the practice of medicine. Want more? See Don Berwick's "The Toxicity of Pay for Performance" D.M. Berwick. Qual Manag Health Care. 1995 Fall.
The HF space has been hugely disappointing for me insofar as societies and guidelines venturing far beyond the foundational studies and evidence, and the engagement in groupthink and cheerleading.
There is not 1 RCT investigating the benefits of rapid initiation and/or inpatient initiation of “quadruple therapy” on clinical endpoints, never mind hard endpoints (the best we’ve gotten is surrogates like NT-BNP). Yet opinion leaders are tripping over themselves strongly advocating such a strategy. It makes a mockery of the very concept of EBM itself. It makes me cynical and suspicious of the motivations of the societies and guideline writers when such egregious liberties are being represented in work-product that should instead be a neutral distillation of the evidence.
last sentence should read ,,,,(and vastly more expensive)...
As an old-fashioned cardiologist now long retired, I recall treating many patients with atrial fibrillation and various degrees of heart failure with digoxin. When ACE inhibitors came out I would add that for patients with orthopnea or other congestive symptoms. The vast majority of patients seemed to do quite well with this regimen. Digoxin was cheap, effective and well tolerated. I tried some on calcium blockers when they were being touted but, in my purely anecdotal experience, they were not as effective or as well tolerated as the digoxin. Of course in those days it was absolutely forbidden to give beta blockers to anyone with heart failure. I wonder if anyone ever ran a study comparing the newer drugs with digoxin. Probably not likely as the newer and sexier (and vastly cheaper) drugs soon took over the market.
John, since atrial fibrillation causes an irregular rhythm, the initial treatment is to decrease the heart rate to diminish myocardial oxygen consumption. Slowing down the heart rate tends to normalize the rhythm in patients with AF. Beta blockers have traditionally been the drugs of choice (atenolol or metoprolol, usually), with non-DHP calcium channel blocks (diltiazem or verapamil) as clear second-choice medications. I wonder, given the tendency of a heart in atrial fibrillation to throw blood clots, if aggressively slowing the heart rate increases the clot risk of patients, leading to their demise. Could this be part of the picture? Whatever benefits beta-blockers confer are offset by the increased clot risk when the heart is slowed too aggressively?
My eyes tend to glaze over when I read the words "meta-analysis". Offhand I can't recall any that were not just a bunch of studies with insignificant differences that were cobbled together in order to engineer a statistically significant result. Having dealt with the heterogeneity of heart failure over a forty year career, I am extremely skeptical about the possibility of matching groups in any study of significant size. Also curious about the definition of "low bias study".
I agree with Ernest Curtis' skeptical opinion regarding what ACTUALLY goes in when a bunch of studies are cobbled together and then the massed data analyzed using various types of statistical Kung Fu.
“I advocate for the use of evidence in the strongest terms. But this does not equate to blindly following overly simplistic algorithms, especially when tied to financial incentives. These are a blight on the evidence-based practice movement.”
This is such an important paragraph. Having worked in pain management AND addiction for well over two decades, I can remember the “evidence” presented that chronic pain + opioids RARELY led to addiction; an issue I argued loudly at the time. I was often reminded of the “evidence...”
My question was WHOSE evidence? We did not know what we know now back then, and the EBM protocol was pushed in my face time and again. Blind overprescribing of opioids certainly has shown evidence that the EBM in practice by many back then wasn’t working. I’m glad to see your excellent essay that can be applied to all areas. Blind Faith should just be the band, not medical practice (yes I just aged myself!)
"...blindly following overly simplistic algorithms, especially when tied to financial incentives." Is exactly how big bureaucratic organizations (alas, the vast majority of healthcare providers worldwide) approach EBM by default.
And don't get me started on nuances like dosing, timing of doses and genetic differences in response!
Do you all think it's an offset of people who might benefit from meds vs those who got worse? You hope that someone does not prescribe a beta blocker to a patient with AFib before assessing the heart rate based on a EMR box . Reminds me of my EP colleague who asks if my consult was a consult to see if a patient needs a pacemaker or a work order to implant . I cared for many elderly people in Texas "L" towns. The "L" is the name of the town and the 80 year olds live with mom and dad ! I would ask for a pacer knowing the current HR was causing symptoms but not syncope etc. The patients were "ok" before but dramatically better after implant. The reason we are paid more than people who just follow the guidelines is because the guidelines are supposed to help us REMEMBER the option to treat a certain way NOT tell us what to do. The problem that everyone forgets is the guideline suggestion is only as good as the data input. The history and physical are not done by the chart but by the physician. An "asymptomatic" 350 pound patient who does not even walk to mailbox is not "asymptomatic".... great article as usual
Since this was a post hoc analysis, how were the placebo and beta blocker groups in Afib matched for baseline characteristics? Perhaps the beta blocker group was sicker.
It was a meta-analysis of 11 randomized trials. Assuming randomization was proper, the groups should be matched.
Respectfully, patients with heart failure were randomized. Those with Afib would represent a post hoc subgroup analysis, subject to confounding.
In any case, may I say I consider your work to be among the most invaluable in Cardiology. Thank you.
And unless the specialists are critically reviewing such studies, everybody else relies on their expert guidance which can be wrong.
Love how John Mandrola advocates for careful use of evidenced based medicine when caring for actual patients. The EMR can be a curse - automatic recommendations based on one diagnosis are necessarily mindless.
I called our EMR “clickbait medicine”
Interesting results...
The problem witn evidence based medicine and guidelines is that they have not be taken as "laws".
HF is an excellent example about treatment and reproducibility: one question. The evidence favoring the use of ARNI is the same than with ACEi? Another one, the up titriation concept developed with ACEi and beta blockers is completely over in order to favor the adition of ARNI an Gliflozins? Do we throw away decades of investigation ?
The Paradigm HF study which championed Entresto is one of Vinay Prasad's teaching examples of a weak study with inappropriate conclusions.
https://www.youtube.com/watch?v=HkDMn_QP0Uk
Thank you very much for the video! Excellent analysis and criticism! We are not alone!
In my practice I still use ACEi as first choice.
What about any morbidity benefit (decreased hospitalizations for decompensated HF)? A lot of emphasis is placed on mortality benefit in cardiology trials. Morbidity benefit (if it can be proven) would be very important for the patient (improved quality of life) and for CMS (since HF is the biggest contributor to CMS budget). I'm not sure that the meta analysis 'proves' GDMT is inappropriate for patients with HF and AF. Perhaps a more nuanced discussion would be to clarify endpoints (goals of therapy). Sometimes death is not the worst outcome; prolonged debilitation and / or suffering can be far worse outcome for many. This is not at all to dismiss premature death in young and middle aged as a source of much grief and suffering.
I’m not sure that evidence-based medicine makes any sense, John.
The problem is, that it requires explanatory theories first, and that is often ignored. Studies are done that get financed by Big Pharma largely. There are some that are done without that funding, and there are some older studies that were done before the massive corruption.
But for example, for heart failure, I am convinced that small amounts of T3 can reverse HF. Yet the studies aren’t really done because there is no money in that, and no explanatory theories that are currently in vogue supporting it.
There is certainly evidence for decades that raising T3 is the solution to HF. For instance several studies showed risk of death higher with lower T3.
https://www.sciencedirect.com/science/article/abs/pii/S0002934304006576
All-cause mortality was 23% (n = 64) after a mean (±SD) of 12 ± 7 months of follow-up; 47 (73%) of the patients died from cardiac causes. The mean ejection fraction was lower in those patients who died than in those who survived (26% ± 8% vs. 31% ± 8%, P < 0.001), as were levels of total T3 (1.0 ± 0.4 nmol/L vs. 1.3 ± 0.3 nmol/L, P < 0.001) and free T3 (3.2 ± 1.4 pmol/L vs. 3.7 ± 1.0 pmol/L, P < 0.001). In a multivariate model, ejection fraction (odds ratio [OR] = 2.0 per 10% decrease; 95% confidence interval [CI]: 1.4 to 2.8 per 10% decrease; P < 0.001) and total T3 level (OR = 0.3 per 1-nmol/L increase; 95% CI: 0.1 to 0.5 per 1-nmol/L increase; P < 0.001) were the only independent predictors of all-cause mortality. In an alternative model using free T3 levels, ejection fraction (OR = 1.9; 95% CI: 1.4 to 2.7; P < 0.001) and free T3 level (OR = 0.6 per 1 pmol/L; 95% CI: 0.5 to 0.8 per 1 pmol/L; P <0.02) were associated with all-cause mortality. When we considered cardiac mortality alone, male sex (OR = 3.5; 95% CI: 1.7 to 13; P < 0.04), ejection fraction (OR = 1.7; 95% CI: 1.2 to 2.5; P < 0.006), and total T3 level (OR = 0.3; 95% CI: 0.2 to 0.7; P < 0.002) were independent predictors with the multivariate model.
Conclusion
Low T3 levels are an independent predictor of mortality in patients with chronic heart failure, adding prognostic information to conventional clinical and functional cardiac parameters.
——
It’s safe
https://www.sciencedirect.com/science/article/abs/pii/S0002914997009508
And it works
https://www.ingentaconnect.com/content/ben/prc/2008/00000003/00000001/art00003
And yet, there are virtually no doctors who pay any attention to this. HF is a result of very low cellular energetics and often accomplices type 2 diabetes and other metabolic issues that come from high fat burning, minimal oxidative phosphorylation, high glycolysis producing lactate, high build up of lactate…
The explanatory theory would suggest, let’s give patients small doses of T3 and get their cellular energetics back up. And it works. But nobody is doing it, nobody talks about it, because the “evidence” isn’t there because Big Pharma doesn’t want the evidence to be there.
That’s the problem with the “evidence based” approach.