Cadmore media player playing audio How to Use a Noniferiority Trial: Interview With Dr Gordon Guyatt
Audio Player
Accessibility
Using forms mode, you can utilize keyboard shortcuts to make usability more efficient.
Space bar will play and pause the audio. Pressing escape will close all displays and pressing it again will jump to the end of the player.
You can press A to open the Keyboard Shortcut Dialog which also provides an audio description of the player and all of its features.
Transcript
Result Count
of
Search
Segments
Result Count
of
Search
Name:
How to Use a Noniferiority Trial: Interview With Dr Gordon Guyatt
Description:
How to Use a Noniferiority Trial: Interview With Dr Gordon Guyatt
Thumbnail URL:
https://cadmoremediastorage.blob.core.windows.net/905ef4fe-774e-4dce-a323-3c41d7f7a353/thumbnails/905ef4fe-774e-4dce-a323-3c41d7f7a353.jpg?sv=2019-02-02&sr=c&sig=f%2Bwdo4Lg3MoRGqOyt7zFtLKw%2B1nfM6OCUurijtZEdfE%3D&st=2024-12-21T14%3A04%3A20Z&se=2024-12-21T18%3A09%3A20Z&sp=r
Duration:
T00H21M26S
Embed URL:
https://stream.cadmore.media/player/905ef4fe-774e-4dce-a323-3c41d7f7a353
Content URL:
https://cadmoreoriginalmedia.blob.core.windows.net/905ef4fe-774e-4dce-a323-3c41d7f7a353/12413001.mp3?sv=2019-02-02&sr=c&sig=pi7d%2FCbAsau5NlMYuTDy78vQw4auXStM7Udh9vfUEZE%3D&st=2024-12-21T14%3A04%3A20Z&se=2024-12-21T16%3A09%3A20Z&sp=r
Upload Date:
2022-02-28T00:00:00.0000000
Transcript:
Language: EN.
Segment:0 .
>> This is Ed Livingston, Deputy Editor of Clinical Reviews in Education for JAMA with today's JAMAevidence Podcast. [ Music ] So, I'm very honored to speak today with Gordon Guyatt, the Principal Editor of the User's Guide to the Medical Literature to discuss Chapter 8, How to Use a Noninferiority Trial. I think this is a critically important chapter because I think many clinicians get confused by these concepts. And I know I certainly did.
So, let me start, Gord, by asking you about what you're testing when you test an equivalence trial. >> Well, equivalence trial as opposed to noninferiority tries to establish that there is really no difference between an intervention and a control. And that is very challenging because defining no difference is very difficult. And basically equivalence trials have more or less gone out of fashion and essentially been replaced by noninferiority.
Noninferiority does not try to establish equivalence, but rather says if the new treatment is not as good as the old treatment with respect to the primary outcome, is the difference small enough that it is still acceptable? >> The concept of noninferiority is very complicated because of the terminology. There's a lot of double negatives, and it's very, very hard to follow. As a clinician, what I really want to know is if new drug B is the same as the old drug A. So, why is it so difficult to not just compare things along those lines?
>> Well, because in equivalence trials, you are trying to establish both boundaries. So, let's say you have two treatments, A and B. And if it favors A, you're saying it's plus 1%. And if it favors B, it's minus 1%. And you say, okay, I'm ready to interpret equivalence if it's between plus 1% and minus 1%. Well, that forces you to show that it's really in an extremely narrow range on both sides.
Noninferiority is easier because it just says are we on one side of the minus 1%. It would be plus 2% or plus 3%. But are we on just one side of the minus 1%? So, it's the difference between, in a conventional superiority trial, we just say, okay, can we say that we're above 0 effect. Simply the difference with noninferiority, are we above and effect of minus 1 or minus 2 or whatever.
We don't have both boundaries. The equivalence trial makes it more difficult because you're setting two different boundaries and saying that the effect is between those two boundaries. >> So, that implies a directionality for noninferiority. So, you're looking at one side. So, could you explain to me, when comparing a new treatment to a standard treatment, one I guess is on the left side of the minus sign, the other one so the right, so the boundary is a negative 1, could you explain some of that? >> Sure. So, typically, traditionally, a new treatment was supposed to be better than existing treatment with what one might call the primary endpoint or endpoints.
So, we're trying to prevent stroke. We're trying to prevent heart attacks. And we want to make sure that we do, in fact, prevent them. The new treatment often is more costly, it often has adverse effects. But we're ready to tolerate that if the effect of the new treatment is sufficiently better than the old one. With noninferiority, what the new treatment has to offer is not that it is better with respect to primary endpoints, like reducing stroke or myocardial infarction, or even mortality, but rather it has other advantages.
Those other advantages may be decreased burden, so you don't have to be as careful. Anticoagulants is a great example. You don't have the INR monitoring of Warfarin. You don't have to be worried about drug interactions quite as much. You don't have to be worried about your alcohol quite as much. So, that's one advantage, a burden advantage. And the other advantage can be a toxicity advantage. So, we're saying, okay, the advantages of this new drug are not that it's better in reducing stroke or myocardial infarction, but that it has burden or toxicity.
And then we say if those burden or toxicity advantages are sufficiently great, we are ready to accept that the treatment may not be quite as good in reducing what it's supposed to do primarily, which is reducing stroke or myocardial infarction. So, with the burden and toxicity advantages, we're ready to accept a small or perhaps not so small if the burden in toxicity advantages are very large increase in strokes and myocardial infarctions.
In other words, that the new treatment is less effective with respect to the primary endpoint than what exists currently. >> So, that difference is the noninferiority margin. And that margin is basically saying that one treatment is just as good as another treatment. But how do you go about determining what that margin should be? >> Well, I can tell you when you say how it is determined, I can tell you how we think it should be determined.
We think that clinicians should consider the benefits in terms of toxicity and burden and say what do we think that our patients are willing to accept. So, it's really a value judgment, the novel anticoagulants are a great example to illustrate this I think. So, traditional treatments are Warfarin. Major disadvantages, you have to have your INR checked all the time, you have to be very worried about drug interactions, and the novel anticoagulants do not have those problems.
So, what the clinician should ask is, in thinking in a noninferiority frame of mind, how much, say it's an atrial fibrillation, you're mainly worried about stroke, how much of an increase in stroke relative to Warfarin would my patients be willing to accept for the advantages of getting away from all the burdens associated with Warfarin? And like any time when there's benefits and downsides of treatments, which there almost always is, it's a value and preference judgment.
And when you think about it then, how would we make that? Well, the greater the burden and toxicity advantages or less burden, less toxicity advantages of a new drug, the more we're willing to accept a decrement in benefit. So, big burden in toxicity advantages we're ready to accept a little more loss of benefit in an outcome like stroke or myocardial infarction. It also then depends on what we are-- what we are losing.
So, we might not want to lose very much if it were death that they were losing benefits, we're going to have more death. We might not be willing to lose very much if it was stroke or particularly severe stroke, maybe more with myocardial infarction, or maybe if it was just hospital admission we might be willing to lose more. So, they're, so, it's a tradeoff that essentially involves values and preferences. But for whatever, for many, there may be a number of reasons.
That's not how the Food and Drug Administration wants to do it. Food and Drug Administration, in a way, a much more complicated statistical approach, which says, okay, what was the effect of the, from say a meta-analysis, what was the benefit of the previous accepted treatment? And we are, and we're willing to give up half of the accepted, half of the benefit of the previously accepted treatment.
That's how we set our noninferiority margin. That and in doing so they ignore the issues of how much toxicity and burden benefit you're getting with the new treatment, and they ignore what the outcome is. So, we don't think that's terribly satisfactory. And what the clinician might want to know how the FDA does it. But what the clinician should think about is what are the values and preferences of my patient given the burden and toxicity benefits of the new treatment, how much would my patients be willing to give up in terms of the benefits in reducing death, stroke, myocardial infarction, or whatever it is.
>> So, how do you recommend that a clinician interpret a negative noninferiority trial? And the reason I ask is we've published a couple of papers in the last year at the Journal wherein the investigators took a guess at what they thought the noninferiority margin should be for a particular intervention. And then when they did the trials, they were, they missed the margin by just a little bit. So, because they had predefined that margin, they had to conclude that they were not able to demonstrate noninferiority, yet they were very substantial statistically significant effects of the interventions.
Given that the noninferiority margin is established somewhat subjectively, how do you think the practicing doctor should interpret this sort of thing when they see those kinds of conclusions in a journal article? >> Well, the issue is whether noninferiority has been established or not established. Do you accept the noninferiority margin? So, one example we use in the paper, in the User's Guide, is a trial that looked at a new way of treating endometrial carcinoma.
And the new treatment had fewer gastrointestinal effects. And the investigator said for this reduction in gastrointestinal effects, we think patients would be, or we said our noninferiority, they don't frame it in what patients think, although we believe they should, where we set our noninferiority margin at a 6% increase in the recurrence of the carcinoma. And the clinician would have to say, are my patients ready to accept that?
Well, they met their criteria. The boundary of the confidence interval said, well, it could be as much as a 5% increase. When we presented that to clinicians, they think patients would not be ready to accept a 5% increase in recurrence for the decreased GI toxicity with the new treatment. So, that is the question that clinicians should ask themselves. If they miss the noninferiority margin, in other words, they say, okay, we think, we set our noninferiority margin at 3%, and it turns out that the treatment may be as much as 4% worse, clinician has to say, well, with my patients, if there were 4% more strokes or myocardial infarctions or recurrences of carcinoma or whatever, would my patients be willing to accept that for the toxicity and burden advantages of the treatment?
>> That is a really great explanation. Thank you for that, Gordon. That really puts it into perspective because we struggled with this on a couple of papers this year. >> If you think about what your patients are trading off, I'm delighted that you like the explanation, because we think that that's the way to, the way to frame it is to think of what our patients would be willing to accept. >> We published this major randomized control trial of antibiotic treatment for appendicitis. Investigators picked a noninferiority margin of 24%.
They really didn't have any basis for that. They just picked it. Because that was the best they could do because there's no trial evidence to guide them. So, they picked it. And it turned out that the margin was like 26%. So, it was technically a negative trial and had to be presented as such. But roughly three-quarters of all patients who were enrolled in the trial never needed surgery. And so unfortunately the trial was perceived by the surgical community as negative, because we said it was negative, but three-quarters of patients didn't need surgery.
And I think that's a pretty big thing. So, my recommendation to people was you need to look at the actual numbers and forget about [inaudible]. >> Absolutely. As a matter of fact, so there was a situation where you would have said their noninferiority margin was perhaps too conservative. Unfortunately, in most situations, it's in the other direction. People are ready to accept, to conclude noninferiority when patients might not be inclined to accept it.
But your point is absolutely right. Either way, look at it, think of what it means in terms of the benefits and downsides and how your patients would feel about it. >> Yeah, so the bottom line is it requires some subjectivity and common sense in interpreting these journal articles, irrespective of what we say. >> Yes, absolutely, as in fact every journal article requires that when it involves tradeoffs between benefits and downsides, which with virtually every treatment we have, it does involve tradeoffs between benefits and downsides.
>> Great. Well, let me see if you can help me with another concept I've struggled with. And this relates to intention to treat and how it factors into noninferiority. And I'm going to start with intention to treat by itself. So, again, as a clinician, maybe not a sophisticated trialist or epidemiologist, I would look at intention to treat studies when I was practicing surgery and think who really cares about that? I don't really care that they were randomized to an operation or medicine and they may have crossed over, and maybe they were randomized to medicine and they got surgery and they had a good outcome.
I really only want to know the effect of the surgery. And so I have troubles as a clinician buying into the concept of intent to treat. I really was interested in a protocol analysis. I wanted to know how the intervention actually worked. How do you explain that to clinicians? >> Well, the disadvantage of a podcast is I don't get to teach with a whiteboard, which makes it a bit more challenging. But the concept is that the purpose of randomization is to create groups that are prognostically balanced.
And that's why we bother with randomized trials. That's why we don't just do observational studies and see people who happen to have gotten a treatment who didn't because we're worried about prognostic imbalance. As soon as you start taking people out who were randomized, you risk losing the prognostic balance that randomization created in the first place. So, if you take out the people who didn't take their treatment, you are removing people that unless they are prognostically identical to those who remain, who did take their treatment, you're going to bias the results.
If the people who didn't take treatment were, in fact, prognostically worse, they were the people destined to do badly, which as it turns out they usually are, then you're going to get rid of the prognostically bad people from the treatment group and make the treatment look better, even if there's no effect at all, you're going to make treatment look good simply by getting rid of the prognostically poor people, people who are destined to do badly, and you get a misleading result.
In other words, you throw away-- you went to all the trouble to do a randomized trial. Why? To create prognostic balance. And now you throw away what you, what the intent was when you randomized. You basically take it back to an observational study by throwing out people who are likely to have a different prognosis. So, the reason for the intention to treat is to maintain the prognostic balance that randomization created in the first place.
>> So, that really helps me understand it. It makes a tremendous amount of sense. So, you convinced me that intention to treat is the right way to look at a clinical trial. Why is it not necessarily the right way to look at a noninferiority trial? >> Well, it is and it isn't. So, the downside of intention to treat is that it will underestimate the effects if everybody took the treatment. So, you're right, we're interested in the physicians and patients we think are interested.
What happens if I take the treatment? But unfortunately if we just look at those people, as I just described, it throws away the balance that randomization created in the first place. So, intention to treat is not a perfect solution in that it really doesn't tell us what we're interested in. It does give us an unbiased result. But part of the reason I think we're relatively comfortable with that is it's a conservative conclusion. So, if a trial, some of the people don't take the treatment and the treatment still looks good, we're pretty comfortable, because we think if everybody would take the treatment, if anything, the treatment would look even better.
So, it gives us a conservative conclusion. And I think that's part of why in conventional superiority trials we're comfortable with the intention to treat. In the noninferiority, picture what would happen if a large proportion of the people did not take the conventional treatment, the standard treatment. Well, now you're losing all the benefits of the standard treatment, and it will be very easy to show noninferiority.
Now we have an anticonservative situation with the intention to treat. We risk say okay we can adopt this new treatment, it's no worse than the existing treatment, they're only a small amount worse, when, in fact, it would be a lot worse if people only took the standard treatment. So, people, I think, are less comfortable with intention to treat, even though it maintains prognostic balance, gives us an unbiased assessment, but it is anticonservative in contrast to the standard superiority trial.
And that's why people like to say not that the protocol may be biased, but because of the anticonservative nature of intention to treat in the context of noninferiority, if we conclude noninferiority, we'd like also to look at a protocol to feel safer that the protocol supports the intention to treat. [ Music ] >> This is Ed Livingston, Deputy Editor for Clinical Reviews and Education of JAMA. You've just heard the most recent JAMAevidence Podcast on noninferiority trials.
This is a very complicated topic and not necessarily easy to follow. I encourage our listeners to look at the June 16th, 2015, issue of JAMA, where there is a JAMA guide to statistics and methods summary of noninferiority trials written by Drs. Amy Kaji and Roger Lewis. This article is written for practicing clinicians who don't have a statistical background and provides an easier way for clinicians to understand these concepts, which are not at all intuitive.
Thank you for listening. Make sure to go to the iTunes store or Stitcher and subscribe to the JAMA Clinical Reviews Podcasts. These are informative podcasts that cover clinical topics that provide in-depth coverage of clinical topics. We encourage you to write reviews of these podcasts so that we get feedback and can make them even better. [ Music ]
Segment:0 .
>> This is Ed Livingston, Deputy Editor of Clinical Reviews in Education for JAMA with today's JAMAevidence Podcast. [ Music ] So, I'm very honored to speak today with Gordon Guyatt, the Principal Editor of the User's Guide to the Medical Literature to discuss Chapter 8, How to Use a Noninferiority Trial. I think this is a critically important chapter because I think many clinicians get confused by these concepts. And I know I certainly did.
So, let me start, Gord, by asking you about what you're testing when you test an equivalence trial. >> Well, equivalence trial as opposed to noninferiority tries to establish that there is really no difference between an intervention and a control. And that is very challenging because defining no difference is very difficult. And basically equivalence trials have more or less gone out of fashion and essentially been replaced by noninferiority.
Noninferiority does not try to establish equivalence, but rather says if the new treatment is not as good as the old treatment with respect to the primary outcome, is the difference small enough that it is still acceptable? >> The concept of noninferiority is very complicated because of the terminology. There's a lot of double negatives, and it's very, very hard to follow. As a clinician, what I really want to know is if new drug B is the same as the old drug A. So, why is it so difficult to not just compare things along those lines?
>> Well, because in equivalence trials, you are trying to establish both boundaries. So, let's say you have two treatments, A and B. And if it favors A, you're saying it's plus 1%. And if it favors B, it's minus 1%. And you say, okay, I'm ready to interpret equivalence if it's between plus 1% and minus 1%. Well, that forces you to show that it's really in an extremely narrow range on both sides.
Noninferiority is easier because it just says are we on one side of the minus 1%. It would be plus 2% or plus 3%. But are we on just one side of the minus 1%? So, it's the difference between, in a conventional superiority trial, we just say, okay, can we say that we're above 0 effect. Simply the difference with noninferiority, are we above and effect of minus 1 or minus 2 or whatever.
We don't have both boundaries. The equivalence trial makes it more difficult because you're setting two different boundaries and saying that the effect is between those two boundaries. >> So, that implies a directionality for noninferiority. So, you're looking at one side. So, could you explain to me, when comparing a new treatment to a standard treatment, one I guess is on the left side of the minus sign, the other one so the right, so the boundary is a negative 1, could you explain some of that? >> Sure. So, typically, traditionally, a new treatment was supposed to be better than existing treatment with what one might call the primary endpoint or endpoints.
So, we're trying to prevent stroke. We're trying to prevent heart attacks. And we want to make sure that we do, in fact, prevent them. The new treatment often is more costly, it often has adverse effects. But we're ready to tolerate that if the effect of the new treatment is sufficiently better than the old one. With noninferiority, what the new treatment has to offer is not that it is better with respect to primary endpoints, like reducing stroke or myocardial infarction, or even mortality, but rather it has other advantages.
Those other advantages may be decreased burden, so you don't have to be as careful. Anticoagulants is a great example. You don't have the INR monitoring of Warfarin. You don't have to be worried about drug interactions quite as much. You don't have to be worried about your alcohol quite as much. So, that's one advantage, a burden advantage. And the other advantage can be a toxicity advantage. So, we're saying, okay, the advantages of this new drug are not that it's better in reducing stroke or myocardial infarction, but that it has burden or toxicity.
And then we say if those burden or toxicity advantages are sufficiently great, we are ready to accept that the treatment may not be quite as good in reducing what it's supposed to do primarily, which is reducing stroke or myocardial infarction. So, with the burden and toxicity advantages, we're ready to accept a small or perhaps not so small if the burden in toxicity advantages are very large increase in strokes and myocardial infarctions.
In other words, that the new treatment is less effective with respect to the primary endpoint than what exists currently. >> So, that difference is the noninferiority margin. And that margin is basically saying that one treatment is just as good as another treatment. But how do you go about determining what that margin should be? >> Well, I can tell you when you say how it is determined, I can tell you how we think it should be determined.
We think that clinicians should consider the benefits in terms of toxicity and burden and say what do we think that our patients are willing to accept. So, it's really a value judgment, the novel anticoagulants are a great example to illustrate this I think. So, traditional treatments are Warfarin. Major disadvantages, you have to have your INR checked all the time, you have to be very worried about drug interactions, and the novel anticoagulants do not have those problems.
So, what the clinician should ask is, in thinking in a noninferiority frame of mind, how much, say it's an atrial fibrillation, you're mainly worried about stroke, how much of an increase in stroke relative to Warfarin would my patients be willing to accept for the advantages of getting away from all the burdens associated with Warfarin? And like any time when there's benefits and downsides of treatments, which there almost always is, it's a value and preference judgment.
And when you think about it then, how would we make that? Well, the greater the burden and toxicity advantages or less burden, less toxicity advantages of a new drug, the more we're willing to accept a decrement in benefit. So, big burden in toxicity advantages we're ready to accept a little more loss of benefit in an outcome like stroke or myocardial infarction. It also then depends on what we are-- what we are losing.
So, we might not want to lose very much if it were death that they were losing benefits, we're going to have more death. We might not be willing to lose very much if it was stroke or particularly severe stroke, maybe more with myocardial infarction, or maybe if it was just hospital admission we might be willing to lose more. So, they're, so, it's a tradeoff that essentially involves values and preferences. But for whatever, for many, there may be a number of reasons.
That's not how the Food and Drug Administration wants to do it. Food and Drug Administration, in a way, a much more complicated statistical approach, which says, okay, what was the effect of the, from say a meta-analysis, what was the benefit of the previous accepted treatment? And we are, and we're willing to give up half of the accepted, half of the benefit of the previously accepted treatment.
That's how we set our noninferiority margin. That and in doing so they ignore the issues of how much toxicity and burden benefit you're getting with the new treatment, and they ignore what the outcome is. So, we don't think that's terribly satisfactory. And what the clinician might want to know how the FDA does it. But what the clinician should think about is what are the values and preferences of my patient given the burden and toxicity benefits of the new treatment, how much would my patients be willing to give up in terms of the benefits in reducing death, stroke, myocardial infarction, or whatever it is.
>> So, how do you recommend that a clinician interpret a negative noninferiority trial? And the reason I ask is we've published a couple of papers in the last year at the Journal wherein the investigators took a guess at what they thought the noninferiority margin should be for a particular intervention. And then when they did the trials, they were, they missed the margin by just a little bit. So, because they had predefined that margin, they had to conclude that they were not able to demonstrate noninferiority, yet they were very substantial statistically significant effects of the interventions.
Given that the noninferiority margin is established somewhat subjectively, how do you think the practicing doctor should interpret this sort of thing when they see those kinds of conclusions in a journal article? >> Well, the issue is whether noninferiority has been established or not established. Do you accept the noninferiority margin? So, one example we use in the paper, in the User's Guide, is a trial that looked at a new way of treating endometrial carcinoma.
And the new treatment had fewer gastrointestinal effects. And the investigator said for this reduction in gastrointestinal effects, we think patients would be, or we said our noninferiority, they don't frame it in what patients think, although we believe they should, where we set our noninferiority margin at a 6% increase in the recurrence of the carcinoma. And the clinician would have to say, are my patients ready to accept that?
Well, they met their criteria. The boundary of the confidence interval said, well, it could be as much as a 5% increase. When we presented that to clinicians, they think patients would not be ready to accept a 5% increase in recurrence for the decreased GI toxicity with the new treatment. So, that is the question that clinicians should ask themselves. If they miss the noninferiority margin, in other words, they say, okay, we think, we set our noninferiority margin at 3%, and it turns out that the treatment may be as much as 4% worse, clinician has to say, well, with my patients, if there were 4% more strokes or myocardial infarctions or recurrences of carcinoma or whatever, would my patients be willing to accept that for the toxicity and burden advantages of the treatment?
>> That is a really great explanation. Thank you for that, Gordon. That really puts it into perspective because we struggled with this on a couple of papers this year. >> If you think about what your patients are trading off, I'm delighted that you like the explanation, because we think that that's the way to, the way to frame it is to think of what our patients would be willing to accept. >> We published this major randomized control trial of antibiotic treatment for appendicitis. Investigators picked a noninferiority margin of 24%.
They really didn't have any basis for that. They just picked it. Because that was the best they could do because there's no trial evidence to guide them. So, they picked it. And it turned out that the margin was like 26%. So, it was technically a negative trial and had to be presented as such. But roughly three-quarters of all patients who were enrolled in the trial never needed surgery. And so unfortunately the trial was perceived by the surgical community as negative, because we said it was negative, but three-quarters of patients didn't need surgery.
And I think that's a pretty big thing. So, my recommendation to people was you need to look at the actual numbers and forget about [inaudible]. >> Absolutely. As a matter of fact, so there was a situation where you would have said their noninferiority margin was perhaps too conservative. Unfortunately, in most situations, it's in the other direction. People are ready to accept, to conclude noninferiority when patients might not be inclined to accept it.
But your point is absolutely right. Either way, look at it, think of what it means in terms of the benefits and downsides and how your patients would feel about it. >> Yeah, so the bottom line is it requires some subjectivity and common sense in interpreting these journal articles, irrespective of what we say. >> Yes, absolutely, as in fact every journal article requires that when it involves tradeoffs between benefits and downsides, which with virtually every treatment we have, it does involve tradeoffs between benefits and downsides.
>> Great. Well, let me see if you can help me with another concept I've struggled with. And this relates to intention to treat and how it factors into noninferiority. And I'm going to start with intention to treat by itself. So, again, as a clinician, maybe not a sophisticated trialist or epidemiologist, I would look at intention to treat studies when I was practicing surgery and think who really cares about that? I don't really care that they were randomized to an operation or medicine and they may have crossed over, and maybe they were randomized to medicine and they got surgery and they had a good outcome.
I really only want to know the effect of the surgery. And so I have troubles as a clinician buying into the concept of intent to treat. I really was interested in a protocol analysis. I wanted to know how the intervention actually worked. How do you explain that to clinicians? >> Well, the disadvantage of a podcast is I don't get to teach with a whiteboard, which makes it a bit more challenging. But the concept is that the purpose of randomization is to create groups that are prognostically balanced.
And that's why we bother with randomized trials. That's why we don't just do observational studies and see people who happen to have gotten a treatment who didn't because we're worried about prognostic imbalance. As soon as you start taking people out who were randomized, you risk losing the prognostic balance that randomization created in the first place. So, if you take out the people who didn't take their treatment, you are removing people that unless they are prognostically identical to those who remain, who did take their treatment, you're going to bias the results.
If the people who didn't take treatment were, in fact, prognostically worse, they were the people destined to do badly, which as it turns out they usually are, then you're going to get rid of the prognostically bad people from the treatment group and make the treatment look better, even if there's no effect at all, you're going to make treatment look good simply by getting rid of the prognostically poor people, people who are destined to do badly, and you get a misleading result.
In other words, you throw away-- you went to all the trouble to do a randomized trial. Why? To create prognostic balance. And now you throw away what you, what the intent was when you randomized. You basically take it back to an observational study by throwing out people who are likely to have a different prognosis. So, the reason for the intention to treat is to maintain the prognostic balance that randomization created in the first place.
>> So, that really helps me understand it. It makes a tremendous amount of sense. So, you convinced me that intention to treat is the right way to look at a clinical trial. Why is it not necessarily the right way to look at a noninferiority trial? >> Well, it is and it isn't. So, the downside of intention to treat is that it will underestimate the effects if everybody took the treatment. So, you're right, we're interested in the physicians and patients we think are interested.
What happens if I take the treatment? But unfortunately if we just look at those people, as I just described, it throws away the balance that randomization created in the first place. So, intention to treat is not a perfect solution in that it really doesn't tell us what we're interested in. It does give us an unbiased result. But part of the reason I think we're relatively comfortable with that is it's a conservative conclusion. So, if a trial, some of the people don't take the treatment and the treatment still looks good, we're pretty comfortable, because we think if everybody would take the treatment, if anything, the treatment would look even better.
So, it gives us a conservative conclusion. And I think that's part of why in conventional superiority trials we're comfortable with the intention to treat. In the noninferiority, picture what would happen if a large proportion of the people did not take the conventional treatment, the standard treatment. Well, now you're losing all the benefits of the standard treatment, and it will be very easy to show noninferiority.
Now we have an anticonservative situation with the intention to treat. We risk say okay we can adopt this new treatment, it's no worse than the existing treatment, they're only a small amount worse, when, in fact, it would be a lot worse if people only took the standard treatment. So, people, I think, are less comfortable with intention to treat, even though it maintains prognostic balance, gives us an unbiased assessment, but it is anticonservative in contrast to the standard superiority trial.
And that's why people like to say not that the protocol may be biased, but because of the anticonservative nature of intention to treat in the context of noninferiority, if we conclude noninferiority, we'd like also to look at a protocol to feel safer that the protocol supports the intention to treat. [ Music ] >> This is Ed Livingston, Deputy Editor for Clinical Reviews and Education of JAMA. You've just heard the most recent JAMAevidence Podcast on noninferiority trials.
This is a very complicated topic and not necessarily easy to follow. I encourage our listeners to look at the June 16th, 2015, issue of JAMA, where there is a JAMA guide to statistics and methods summary of noninferiority trials written by Drs. Amy Kaji and Roger Lewis. This article is written for practicing clinicians who don't have a statistical background and provides an easier way for clinicians to understand these concepts, which are not at all intuitive.
Thank you for listening. Make sure to go to the iTunes store or Stitcher and subscribe to the JAMA Clinical Reviews Podcasts. These are informative podcasts that cover clinical topics that provide in-depth coverage of clinical topics. We encourage you to write reviews of these podcasts so that we get feedback and can make them even better. [ Music ]
Form Title Download
Media
Transcript
Form Title Bookmark
Entire Media
Segment(s)
Custom Clip
Start At:
End At:
Form Title Share
Entire Media
Segment(s)
Custom Clip
Start At:
End At:
Form Title Accessibility
The cadmore media player is a professional video and audio player experience designed to deliver media as well as relevant transcript, segmentation, and metadata. On the screen is a traditional html audio player with buttons such as play, pause, forward, captioning, language and play speed selection and more. Adjacent to the audio player are generally tabs, one of which shows a scrolling transcript of the audio and the other shows any segmentation of the media, sometimes referred to as chapters. At the top of the audio player is a menu bar containing an Explore button and Tools button. The Explore button allows you to dig deeper into the audio by opening up an explore dialog which contains an about tab which contains rich metadata about the media, a segments tab which contains further information about the audio, and a related tab which contains links to other media related to this media. The tools button opens a menu with options for sharing the media on social media, creating bookmarking links, creating embed codes for your website, generating citations, and more.
Keyboard Shortcuts
Keyboard shortcuts only work while you are in forms mode. By entering forms mode, you will be able to quickly access portions of the media player with single key commands.
Spacebar: Play/Pause Toggle
E: Explore Menu,
O: Tools Menu
T: Transcript Tab,
S: Segments Tab
F: Forward 15,
R: Rewind 15
L: Languages,
P: Playspeed
M: Mute Toggle,
V: Volume Bar
N: Video Only Toggle,
B: Full Screen Toggle
C: Captions Toggle
Escape: Closes dialogs and menus
Z: Next Transcript Chapter
X: Previous Transcript Chapter
Q: Next Transcript Paragraph
W: Previous Transcript Paragraph
Form Title Embed Media
Entire Media
Segment(s)
Custom Clip
Start At:
End At:
Embed Size
-
You Size - Responsive
-
We Size - Responsive
-
Picture Overlay Popup
-
400 x 225
-
512 x 288
-
400 x 700
-
768 x 1024
-
1024 x 512
-
1280 x 608
Form Title Cite
No citation provided.
Info
Segments
Related
Thumbnails
Accessibility
Using forms mode, you can utilize keyboard shortcuts to make usability more efficient.
Space bar will play and pause the media.
You can press A to open the Accessibility Dialog.
This describes many other keyboard shortcuts for media controls.
Exit Clip Mode
The video is currently set-up to show just a clip from 0:00 to video end. You have requested to see a portion of the media outside of the clip.
Click OK to switch to the full media or click if you would like to stay inside the clip. You can always go back to the clip with Tools and Clip Mode