Last week I attended the European Society of Cardiology congress in Amsterdam. ESC is one of the largest cardiology meetings. I attend mostly in the role of a physician journalist. Well, not really a journalist, more like an opinion writer and critical appraiser.
I'm asking a possible naive question, regarding coumadin vs. direct inhibitors, as this is outside of my field, but is really more a general question: is there a possibility of introducing a bias in this type of trial design, by randomizing drug switches. I'm thinking that, by definition, you are taking a population who is at least not doing terribly on a specific drug, (so chopping off one end of the bell curve) and possibly even doing well, and potentially exposing them to a different drug where all outcomes are in play. Is this a valid concern?
Another way to ask this: Imagine the opposite scenario, would we expect to see the same results? Is it possible that if you took the same patient population, except they were patients who happened to be on the direct inhibitors, and then randomized to switch to coumadin, that this trial would show that COUMADIN has the increased risks? Is the initial drug given a leg up, regardless of their relative profiles?
Is it common to test for Factor V Leiden or other genetic propensities before prescribing one or the other? My late husband had this, and took warfarin from 2008 until his death. Does this factor into decision to prescribe one or the other drug?
Wonderful and informative summary and commentary, per usual. This highlights for me the gulf btw biologic plausibility and “extrapolation”, Vs actual evidence. Of course, clinical scenarios require clinical decisions, and the vast majority of those scenarios will not be mappable directly onto any clinical trial evidence. So decisions made using imperfect evidence is the norm. But the issue is one you’ve often spoken of, that of dogma, and the presumption of having “the right answer” absent evidence, and the further therapeutic fashion that promotes uniformity even in the absence of evidence. Individual patients require individualized decisions, not cookbook answers based on population averages from studies not of direct relevance and applicability. Latter day guidelines in many cardiology domains come to mind.
I had to chuckle at the recent ESC cardiomyopathy guideline, wrt SCD prevention in HCM, and calling out the evidence-free (or at least evidence-poor) reliance on apical aneurysms, LGE, and low EF, espoused by the last AHA guidelines.
As for this study, this changes my practice today. And now it makes me wonder about de novo starts in older frail patients with new AF, as to which form of OAC to choose. Yet another of those evidence free zones.
About time some sense returns to medical studies....but one question is important to ask ourselves....were the pharma maker’s studies if any so supportive of using their new drug on the most fragile ? Yet here we have a well designed study that appears to be hazardous to those same people who have been put on these newer drugs....just like the COVID jab. Is anyone going to expose that??
May be an important study but missing critical issues. Warfarin and Coumadin, being Vitamin K antagonists, are ultimately quite unhealthy. Proven to accelerate osteoporosis and probably CAD. You are disturbing healthy calcium flow and metabolism. You need to pay attention to Matrix GLA and ucOC pathways. So, not as simple as presented.
And it was not 69%. It was a 6% different. HR are so misleading. Relative vs Absolute statistics will mislead you every time.
Excellent reply. A long article. But easily summed up as thus. If you want to "sell" anything from products to scientific studies use relative statistics. It's always more impressive. So it's now HR, RR with a dose of CL (confidence limits) superseding p-values. It is a time honored but dishonest statistical tradition. And they all know it.
Thank you for writing on this--it IS an important study. Now we get to see whether it changes practice. I have a subset of anti-coagulated patients who are quite stable on warfarin so I usually suggest they stay on it, and many of these folks do.
A major bleed was 69% more common with DOACs than with warfarin! This is huge. I can’t wait to dig into this study. Thanks for the news John.
One of the worries of using warfarin is the higher risk of intracranial bleeds. What percentage of severe bleeds between these groups was intracranial bleeds? A drug like dabigatran significantly raises the risk of GI bleeds. Does that bear out in this study? Apixaban is the DOAC with the best safety profile and it gets used a great deal, consequently. Even apixaban is riskier than warfarin in this population? Fascinating. Were any patients those with heart valve abnormalities that favor the use of warfarin? I really need to read this study. Cheers John.
Agree Micheal. The subdural hematoma in the “frail”, generally after a fall, has often more morbidity and mortality. The expensive or non existant reversibility of doac’s in GU bleeds is the other shoe.
I now have a latte and a paper review on my Labor Day morning agenda!
So I’m guessing Andexxa was not the game changer it was hoped to be? Idarucizumab (Praxbind) completely reverses dabigatran within 4 hours, according to published reports. More work is needed, of course. More work is always needed.
Glad to read about this study. There are so many cases where the new great drug is pushed onto patients via Big Pharma marketing without adequate studies. In fact that is the norm.
In fact, plain aspirin may be a superior solution for many patients.
For example, this study showed plain aspirin as SUPERIOR but you have to read the whole abstract and study to see how much better aspirin is.
Aspirin Plus Clopidogrel Versus Aspirin Alone After Coronary Artery Bypass Grafting
The Clopidogrel After Surgery for Coronary Artery Disease
Patients were older who were in the aspirin-only arm. Despite that, this arm did as well as the Clopidogrel+ aspirin arm, except in that arm there was one case of severe bleeding (but none in the just-aspirin arm.)
Aspirin is one of the handful of true miracle drugs. That is why there is so much abuse heaped on it. the mistake in the study I’m citing, from the point of view of the drug companies, is that they allowed this study to proceed with an aspirin-only arm.
Anyhow, great work and appreciate your analysis very much.
Unscientifically my comment relates to my many years of talking to patients as a pharmacist. They hated taking Warfarin and found it difficult to get the dose right. They hated having to go to the hospital to get tested and when the new drugs came out they got right on board even though they cost much more and there was no antidote to a bleeding issue. Most of these patients were old but since they were ambulatory they were not frail.
I've been on Warfarin for 3 years now and can certainly see how many people would find it aggravating. My INR is usually just out of range (either high or low) so I'm generally at the clinic every 2 weeks. I'm always careful about cutting myself and little scrapes are a nusiance but that would be the same for any anticoagulant.
On the plus side, I've got to know everyone at the clinic and a visit there is enjoyable. If I don't need a pharmacist consult and dosage change, my cost per visit is $1.89. My Warfarin prescription is $1.57 so I'm usually digging in my pocket for change to pay for it. I'm also convinced that the stuff is saving my life and that's always a plus.
Thanks for highlighting quality science. My question, as someone who takes care of a lot of folks with frailty, is do these interventions actually provide overall benefit and value in the elderly who are somewhere on the continuum with an arrhythmia.
Not being an MD, my simple mind wouldn't have even thought to do such a study for this reason: If DOACs demonstrate less bleeding in younger, healthier people, I would logically think that they are even MORE likely to demonstrate less bleeding in the frail elderly, since they would seem more susceptible. Either I'm not logical, or biology isn't logical! So this suggests one needs to do not just one study per intervention, but many different studies per intervention for different groups of people based on age, comorbidities (and current medications?), and gender. As discussed in previous articles, one shouldn't do them all in one study since it yields multiple comparisons, which increases type 1 errors. Yikes! I hope somebody will step up and pay for all those studies!
Well, again a fwiw comment, but at 65 I took apixaban for 3 months after an episode of atrial flutter. It was an experience I never want to have again. Insomnia, chest pain, difficulty breathing, terrible fatigue and it took 3 months to realize the apixaban was the culprit. And I know other people who had similar experiences. These new anticoagulants can be utterly unbearable. I am not a frail person but if I had been, this might have done me in.
I'm so sorry to hear about your terrible experience with apixaban. Just to be clear, I wasn't saying apixaban was better for older folks, not that 65 is old! :) The FRAIL AF study clearly showed it resulted in more bleeding in that population. But in younger folks, RCT's demonstrated the opposite. It's just so unexpected and paradoxical, and shows the need to study every intervention with every population, if only somebody will pay for all that research! Maybe AI will come to the rescue and be able to figure out who will do better on what medication; i.e., precision medicine. Let's hope so! Stay well!
I do wonder if those awful side effects that I list might not be as lethal as the bleeding, in all age groups, to some extent. I agree that the study as conducted was a good one, and necessary. And pharma is probably not interested in funding such.
Very nice post John. One minor quibble: "Major bleed was 69% more in ..." implies something different than a hazard ratio of 1.69 to most people. I'd recommend reporting "The hazard rate of major bleed was at least 1.23 x larger in DOAC (with 0.95 confidence; point estimate 1.69), with a cumulative incidence at ... months raised from ... in the control arm to ... in DOAC (0.95 CI for difference in cumulative risk ...)." Supplementing that with a Bayesian probability that the risk difference is clinically important would be nice.
How did minor bleeds fare? What would be the result of an analysis that used bleeding more as a continuous variable (e.g., longitudinal analysis)? Or an analysis that distinguished GI bleeding from the worse cerebral bleeding (e.g., ordinal longitudinal analysis)?
I'm asking a possible naive question, regarding coumadin vs. direct inhibitors, as this is outside of my field, but is really more a general question: is there a possibility of introducing a bias in this type of trial design, by randomizing drug switches. I'm thinking that, by definition, you are taking a population who is at least not doing terribly on a specific drug, (so chopping off one end of the bell curve) and possibly even doing well, and potentially exposing them to a different drug where all outcomes are in play. Is this a valid concern?
Another way to ask this: Imagine the opposite scenario, would we expect to see the same results? Is it possible that if you took the same patient population, except they were patients who happened to be on the direct inhibitors, and then randomized to switch to coumadin, that this trial would show that COUMADIN has the increased risks? Is the initial drug given a leg up, regardless of their relative profiles?
Is it common to test for Factor V Leiden or other genetic propensities before prescribing one or the other? My late husband had this, and took warfarin from 2008 until his death. Does this factor into decision to prescribe one or the other drug?
Conversely what would be the likely outcome for switching from Xarelto to the vitamin K's for a non-frail, long term, 80+, user, of that DOAC, ?
🤣
Wonderful and informative summary and commentary, per usual. This highlights for me the gulf btw biologic plausibility and “extrapolation”, Vs actual evidence. Of course, clinical scenarios require clinical decisions, and the vast majority of those scenarios will not be mappable directly onto any clinical trial evidence. So decisions made using imperfect evidence is the norm. But the issue is one you’ve often spoken of, that of dogma, and the presumption of having “the right answer” absent evidence, and the further therapeutic fashion that promotes uniformity even in the absence of evidence. Individual patients require individualized decisions, not cookbook answers based on population averages from studies not of direct relevance and applicability. Latter day guidelines in many cardiology domains come to mind.
I had to chuckle at the recent ESC cardiomyopathy guideline, wrt SCD prevention in HCM, and calling out the evidence-free (or at least evidence-poor) reliance on apical aneurysms, LGE, and low EF, espoused by the last AHA guidelines.
As for this study, this changes my practice today. And now it makes me wonder about de novo starts in older frail patients with new AF, as to which form of OAC to choose. Yet another of those evidence free zones.
About time some sense returns to medical studies....but one question is important to ask ourselves....were the pharma maker’s studies if any so supportive of using their new drug on the most fragile ? Yet here we have a well designed study that appears to be hazardous to those same people who have been put on these newer drugs....just like the COVID jab. Is anyone going to expose that??
May be an important study but missing critical issues. Warfarin and Coumadin, being Vitamin K antagonists, are ultimately quite unhealthy. Proven to accelerate osteoporosis and probably CAD. You are disturbing healthy calcium flow and metabolism. You need to pay attention to Matrix GLA and ucOC pathways. So, not as simple as presented.
And it was not 69%. It was a 6% different. HR are so misleading. Relative vs Absolute statistics will mislead you every time.
https://www.cureus.com/articles/141648-historical-review-of-the-use-of-relative-risk-statistics-in-the-portrayal-of-the-purported-hazards-of-high-ldl-cholesterol-and-the-benefits-of-lipid-lowering-therapy#!/
Philip Miller…so right you are about RR v AR and their usage to mislead.
And in a lot more instances than just this.
Gawd…we’re easily led astray!
Excellent reply. A long article. But easily summed up as thus. If you want to "sell" anything from products to scientific studies use relative statistics. It's always more impressive. So it's now HR, RR with a dose of CL (confidence limits) superseding p-values. It is a time honored but dishonest statistical tradition. And they all know it.
Thank you for writing on this--it IS an important study. Now we get to see whether it changes practice. I have a subset of anti-coagulated patients who are quite stable on warfarin so I usually suggest they stay on it, and many of these folks do.
Frail -AF. That is quite the descriptive name for this study.
Ha! That’s what I came here to say.
Every patient is an N=1!
A major bleed was 69% more common with DOACs than with warfarin! This is huge. I can’t wait to dig into this study. Thanks for the news John.
One of the worries of using warfarin is the higher risk of intracranial bleeds. What percentage of severe bleeds between these groups was intracranial bleeds? A drug like dabigatran significantly raises the risk of GI bleeds. Does that bear out in this study? Apixaban is the DOAC with the best safety profile and it gets used a great deal, consequently. Even apixaban is riskier than warfarin in this population? Fascinating. Were any patients those with heart valve abnormalities that favor the use of warfarin? I really need to read this study. Cheers John.
No it was not 69% more common.
Agree Micheal. The subdural hematoma in the “frail”, generally after a fall, has often more morbidity and mortality. The expensive or non existant reversibility of doac’s in GU bleeds is the other shoe.
I now have a latte and a paper review on my Labor Day morning agenda!
So I’m guessing Andexxa was not the game changer it was hoped to be? Idarucizumab (Praxbind) completely reverses dabigatran within 4 hours, according to published reports. More work is needed, of course. More work is always needed.
Glad to read about this study. There are so many cases where the new great drug is pushed onto patients via Big Pharma marketing without adequate studies. In fact that is the norm.
In fact, plain aspirin may be a superior solution for many patients.
For example, this study showed plain aspirin as SUPERIOR but you have to read the whole abstract and study to see how much better aspirin is.
Aspirin Plus Clopidogrel Versus Aspirin Alone After Coronary Artery Bypass Grafting
The Clopidogrel After Surgery for Coronary Artery Disease
(CASCADE) Trial
https://www.ahajournals.org/doi/pdf/10.1161/CIRCULATIONAHA.110.978007
Patients were older who were in the aspirin-only arm. Despite that, this arm did as well as the Clopidogrel+ aspirin arm, except in that arm there was one case of severe bleeding (but none in the just-aspirin arm.)
Aspirin is one of the handful of true miracle drugs. That is why there is so much abuse heaped on it. the mistake in the study I’m citing, from the point of view of the drug companies, is that they allowed this study to proceed with an aspirin-only arm.
Anyhow, great work and appreciate your analysis very much.
Unscientifically my comment relates to my many years of talking to patients as a pharmacist. They hated taking Warfarin and found it difficult to get the dose right. They hated having to go to the hospital to get tested and when the new drugs came out they got right on board even though they cost much more and there was no antidote to a bleeding issue. Most of these patients were old but since they were ambulatory they were not frail.
I've been on Warfarin for 3 years now and can certainly see how many people would find it aggravating. My INR is usually just out of range (either high or low) so I'm generally at the clinic every 2 weeks. I'm always careful about cutting myself and little scrapes are a nusiance but that would be the same for any anticoagulant.
On the plus side, I've got to know everyone at the clinic and a visit there is enjoyable. If I don't need a pharmacist consult and dosage change, my cost per visit is $1.89. My Warfarin prescription is $1.57 so I'm usually digging in my pocket for change to pay for it. I'm also convinced that the stuff is saving my life and that's always a plus.
Thanks for highlighting quality science. My question, as someone who takes care of a lot of folks with frailty, is do these interventions actually provide overall benefit and value in the elderly who are somewhere on the continuum with an arrhythmia.
Not being an MD, my simple mind wouldn't have even thought to do such a study for this reason: If DOACs demonstrate less bleeding in younger, healthier people, I would logically think that they are even MORE likely to demonstrate less bleeding in the frail elderly, since they would seem more susceptible. Either I'm not logical, or biology isn't logical! So this suggests one needs to do not just one study per intervention, but many different studies per intervention for different groups of people based on age, comorbidities (and current medications?), and gender. As discussed in previous articles, one shouldn't do them all in one study since it yields multiple comparisons, which increases type 1 errors. Yikes! I hope somebody will step up and pay for all those studies!
https://www.sofpromed.com/how-much-does-a-clinical-trial-cost
Well, again a fwiw comment, but at 65 I took apixaban for 3 months after an episode of atrial flutter. It was an experience I never want to have again. Insomnia, chest pain, difficulty breathing, terrible fatigue and it took 3 months to realize the apixaban was the culprit. And I know other people who had similar experiences. These new anticoagulants can be utterly unbearable. I am not a frail person but if I had been, this might have done me in.
I'm so sorry to hear about your terrible experience with apixaban. Just to be clear, I wasn't saying apixaban was better for older folks, not that 65 is old! :) The FRAIL AF study clearly showed it resulted in more bleeding in that population. But in younger folks, RCT's demonstrated the opposite. It's just so unexpected and paradoxical, and shows the need to study every intervention with every population, if only somebody will pay for all that research! Maybe AI will come to the rescue and be able to figure out who will do better on what medication; i.e., precision medicine. Let's hope so! Stay well!
I do wonder if those awful side effects that I list might not be as lethal as the bleeding, in all age groups, to some extent. I agree that the study as conducted was a good one, and necessary. And pharma is probably not interested in funding such.
And yeah, 65 is old alright!
Haven't you heard - 65 is the new 45! 😬
https://www.deseret.com/2015/3/24/20561317/baby-boomers-rejoice-65-is-the-new-45
Inside every old person is a young one who wonders what the hell happened.
Thank you Zade, that made me laugh out loud 😀.
Very nice post John. One minor quibble: "Major bleed was 69% more in ..." implies something different than a hazard ratio of 1.69 to most people. I'd recommend reporting "The hazard rate of major bleed was at least 1.23 x larger in DOAC (with 0.95 confidence; point estimate 1.69), with a cumulative incidence at ... months raised from ... in the control arm to ... in DOAC (0.95 CI for difference in cumulative risk ...)." Supplementing that with a Bayesian probability that the risk difference is clinically important would be nice.
How did minor bleeds fare? What would be the result of an analysis that used bleeding more as a continuous variable (e.g., longitudinal analysis)? Or an analysis that distinguished GI bleeding from the worse cerebral bleeding (e.g., ordinal longitudinal analysis)?
Professor —I appreciate that you take the time to read and comment. I learn each time. And so do our readers. Thank you.