I've been looking for a good video to understand intent to treat from the past one month. This really helped me a lot. Thank you for sharing your understanding with your audience, Mr. Terry.
You are BRILLIANT. LOVE your videos. I come to you to understand what my professor fails to explain. Thanks for teaching so well and sharing your gift with us.
Thank you so much for posting this video. I'm trying to struggle with research methodology and statistics and your videos help me a lot in understanding my reading.
Thank you very much for your videos. They've been really helpfull. You possess really didactic ways of explaining some concepts that sometimes can be difficult to understand. Greetings from Argentina!
Damn, exactly straight from USMLE Step 3 - UW Biostatistics & Epidemiology section - ITT analysis tested, used to preserve randomization! Thank you doc!
thanks for the clear explanation! I don't like it when I need to watch 10+ minutes of video when I just started trying to understand a new concept, yours is clear & concise, this really helped a lot!!
This raises many questions, the most important of which is whether academics can discuss a logical question and one can admit a mistake. ITT is upheld by some (including Wikipedia) and considered completely wrong by others (like me: Wikipedia would not accept my edits). You can discuss experimental situations and describe the appropriate statistical approach but as stated ITT is fundamentally idiotic. I had suggested that we can be explicit by recognizing that ITT ask the question (as stated in the video) What is the effect of ASSIGNING a drug or intervention? Most readers do not want to know this but if they do, then you can do ITT. However, it must be stated explicitly. And consistently --many articles describe an intervention as assigning patients to adding coconut oil to their food but the article quickly morphs to a study of the effect of coconut oil even paper and let the press say coconut oil is bad for you, that's wrong. Incorrect. If you think there was a placebo effect you have to show that. "May" is not data. It is not science. In the real world, you don't know who took the drug so you must effectively due ITT but we always did that and we don't need a special name. We attribute the effect to the coconut oil because that's the best we can do. When we find out about compliance, we have to do something different. The real world is separate from the intervention. Surgery will have a different effect if it is carried out at Mass General or on a battlefield (God willin').. The real point: 1. ITT requires that if nobody takes the pill, then you must say that the pill has the effect that you measure in these subjects. 2. If the subject told you before the experiment that they cannot, for religious reasons or whatever, take a pill, you would exclude them from the study. Finding out after you start, doesn't change anything. 3. Randomization refers to relevant variables that you are not testing. You intend to break the randomization by measuring response to a new variable. ITT is foolish and should never be done (if you know the details of adherence) unless you emphasize that it is about the intention. (The road to statistical hell is paved...) So, one of us is wrong. One us has to admit a mistake. I am willing if you can answer the objections above. Are you up for resolving this issue? Admittedly, I have the advantage in that I do experimental biochemistry and make two or three mistakes a week which I have to face up to. So, what say you?
Thank you for posting this Terry. I wonder what your thoughts are on the Sussman JA article that was published in BMJ-- Sussman talks about doing (what they call) containment adjustment ITT analyses:doi: 10.1136/bmj.c2073 Their point being that ITT analsyes do not analyze the effect of receiving the treatment but rather the effect of being assigned to the treatment group... whereas containment adjustment ITT anlayses provide an estimate of the effect of receving the treatment.
You preserve randomization but assign treatment-unrelated outcomes to the treatment group. Not sure if the benefit outweighs the loss here. Amazing explanation though.
So when would per protocol be more applicable to a clinical decision than intention to treat? When the per protocol justifies their removal of various subjects and then based on your judgment you make that call? Thanks in advance.
Thanks for the explanation. So ITT is for perserving randomization. But still, how is it fair to include people who died before taking the medicine? I don't get it just yet.
I am still confused how if we don't know the outcome of the dropped out subjects? For example, 100 subjects in control group vs 100 subjects in experimental group are followed for survival after 5 years. If there are 10 dropped out subjects in experimental group and (for the sake of simplicity) the rest of them is alive, how should we calculate the survival rate of experimental group? PP analysis: 90/90 = 100% ITT analysis: 90/100? And how about if I study their mortality rate? PP analysis: 0/90 = 0% ITT analysis: 10/100 = 10% (we count the dropped out subjects as "failure") or 0/100 (because we only include the dropped out as they were randomized, without any outcome)? Thank you
in this example.. why dont we reduce the sample size to 90 in each group during data analysis as 10-10 participants from each group as lost.. can you pls explain that too?
Someone gave you a thumbs down just because. In reality you don't deserve one. You are helping me with my Journal Club Presentation. Are you by chance a Preventive Medicine Physician?
This example actually doesn't work because the design is fundamentally flawed . Each person in the control group should have been matched with an intervention subject and they should have both started treatment (i.e., surgery or ASA) at the same time. The RRR at 2:40 is higher because the time period over which stroke incidence was analysed is different between the groups. This is a tutorial about why it's important to design your study properly not a tutorial about ITTA
Amazing how he can describe something completely idiotic as if it not only made sense but was demanded of reason. There are in fact fewer events in the surgery. Intention to treat is certainly conservative because it's wrong. I doesn't preserve randomization. Not imbalanced prognostic factor. In the experiment, people sensibly want to know what is the effect of surgery compared to aspirin. Instead, ITT answers instead what is the effect of TELLING PEOPLE to have surgery vs. taking aspirin. Is that really what you want to know? Surgery is better. Being assigned to surgery does not depend on the surgery.
Above comments a little bit over the top but it really is true that intention to treat doesn't make sense. Since writing above I see the very useful other stuff this guy wrote so I will try to talk to him directly.
Each person in the control group should have been matched with an intervention subject and they should have both started treatment (i.e., surgery or ASA) at the same time. The RRR at 2:40 is higher because the time period over which stroke incidence was analysed is different between the groups.
The mark of a good teacher is that after listening to him you think the topic was not that hard to explain after all.
Totally agree!!!
And you subscribed straight away after watching a single clip
This guy's awesome, why can't he be my professor?!!
I've been looking for a good video to understand intent to treat from the past one month. This really helped me a lot. Thank you for sharing your understanding with your audience, Mr. Terry.
Years of pharmacy school and my professors couldn't make these scientific journal topics as easy to understand as you do. Great videos, thanks so much
Your videos have been unbelievably helpful. Thanks so much!
Simple, very illustrative and great !
Thank you Prof.!
You are BRILLIANT. LOVE your videos. I come to you to understand what my professor fails to explain. Thanks for teaching so well and sharing your gift with us.
Never really knew what Intention to treat really meant not until I saw this video. This was really helpful
Awesome video, thank you! So clear, love the example you use.
Thanks Dr.Shaneyfelt.Everytime I view your videos I enjoy statistics which I hate before.You are a great teacher.Good luck for you.
Thank for this. In just four minutes you have summarised my 1 day struggle. Thanks
Thanks a lot. Simple and clear.I'm not in the medical sector and this helped me understand.
Very Helpful! I am glad this was the first video I watched.
Absolutely excellent explanation! Thank you so much!!
Thanks Terry, well done!
Great explanation, thanks for posting!
Thank you very much sir for such nice presentation
👏🏽👏🏽👏🏽!!!
Awesome explanation :p
Thank you for spreading the knowledge !!
Thank you so much for posting this video. I'm trying to struggle with research methodology and statistics and your videos help me a lot in understanding my reading.
Very well explained.. thanks a lot
Thank you very much for your videos. They've been really helpfull. You possess really didactic ways of explaining some concepts that sometimes can be difficult to understand. Greetings from Argentina!
Clear & informative. Thanks a lot!
This was a great video, best yet
Damn, exactly straight from USMLE Step 3 - UW Biostatistics & Epidemiology section - ITT analysis tested, used to preserve
randomization! Thank you doc!
thanks for the clear explanation!
I don't like it when I need to watch 10+ minutes of video when I just started trying to understand a new concept, yours is clear & concise, this really helped a lot!!
THANKS!
very helpful. Thanks. I liked the example.
This is a great explanation of ITT. Complex made easy!
Great video and explanation on point!
That was a very nice and smooth explanation..Thank you so much.
Thank you! First time I understood ITT
Excellent explanation, thanks for sharing.
Very good ! Thank you very much!
great explanation! thanks!
THank you so much, that was so helpful. I didn't understand it based how it was taught to me at university
Thank u ,amazing job 🙏🏻
thank you so much for posting this :)
Very good presentation. Thank you
Thank you great video's about Biostatiistics, a very challenging subject.
Really helpful, thanks a ton! :)
thats lovely. thanks for the video
Fantastic explanation!!!!
Great video. Helped me a lot with my research. Wish you were my prof.
Just awesome.
Nice and clear, thank you)
thank you that was brief and clear
Brilliant explanation
Nice video, thank you very much
very good explanation, thank you
thank you this is great!
its very easy now. thank you.
thank you for this video it help a lot
Thank you so much!!!
Thank you!
This raises many questions, the most important of which is whether academics can discuss a logical question and one can admit a mistake. ITT is upheld by some (including Wikipedia) and considered completely wrong by others (like me: Wikipedia would not accept my edits). You can discuss experimental situations and describe the appropriate statistical approach but as stated ITT is fundamentally idiotic. I had suggested that we can be explicit by recognizing that ITT ask the question (as stated in the video)
What is the effect of ASSIGNING a drug or intervention? Most readers do not want to know this but if they do, then you can do ITT. However, it must be stated explicitly. And consistently --many articles describe an intervention as assigning patients to adding coconut oil to their food but the article quickly morphs to a study of the effect of coconut oil even paper and let the press say coconut oil is bad for you, that's wrong. Incorrect. If you think there was a placebo effect you have to show that. "May" is not data. It is not science.
In the real world, you don't know who took the drug so you must effectively due ITT but we always did that and we don't need a special name. We attribute the effect to the coconut oil because that's the best we can do. When we find out about compliance, we have to do something different. The real world is separate from the intervention. Surgery will have a different effect if it is carried out at Mass General or on a battlefield (God willin').. The real point:
1. ITT requires that if nobody takes the pill, then you must say that the pill has the effect that you measure in these subjects.
2. If the subject told you before the experiment that they cannot, for religious reasons or whatever, take a pill, you would exclude them from the study. Finding out after you start, doesn't change anything.
3. Randomization refers to relevant variables that you are not testing. You intend to break the randomization by measuring response to a new variable.
ITT is foolish and should never be done (if you know the details of adherence) unless you emphasize that it is about the intention. (The road to statistical hell is paved...)
So, one of us is wrong. One us has to admit a mistake. I am willing if you can answer the objections above. Are you up for resolving this issue? Admittedly, I have the advantage in that I do experimental biochemistry and make two or three mistakes a week which I have to face up to. So, what say you?
Very well explained thank you.
clear explanation! awesome ! thanks :)
thanks it was very helpful
great explanation
Thank you for posting this Terry. I wonder what your thoughts are on the Sussman JA article that was published in BMJ-- Sussman talks about doing (what they call) containment adjustment ITT analyses:doi: 10.1136/bmj.c2073
Their point being that ITT analsyes do not analyze the effect of receiving the treatment but rather the effect of being assigned to the treatment group... whereas containment adjustment ITT anlayses provide an estimate of the effect of receving the treatment.
this is great
Thank you so much :)
thank you!!
Thanks !
Finaaaaaaaaaaly I understand so thank you 👩🏼🏫
Thank you
Clear 🎉🎉🎉
thank you sir
I came here to learn about this from reading this concept in an RCT study I needed to read, and he explained it easy. Why can't text books do this?
You preserve randomization but assign treatment-unrelated outcomes to the treatment group. Not sure if the benefit outweighs the loss here. Amazing explanation though.
excellent
So when would per protocol be more applicable to a clinical decision than intention to treat? When the per protocol justifies their removal of various subjects and then based on your judgment you make that call? Thanks in advance.
Thanks for the explanation. So ITT is for perserving randomization. But still, how is it fair to include people who died before taking the medicine? I don't get it just yet.
I am still confused how if we don't know the outcome of the dropped out subjects? For example, 100 subjects in control group vs 100 subjects in experimental group are followed for survival after 5 years. If there are 10 dropped out subjects in experimental group and (for the sake of simplicity) the rest of them is alive, how should we calculate the survival rate of experimental group?
PP analysis: 90/90 = 100%
ITT analysis: 90/100?
And how about if I study their mortality rate?
PP analysis: 0/90 = 0%
ITT analysis: 10/100 = 10% (we count the dropped out subjects as "failure") or 0/100 (because we only include the dropped out as they were randomized, without any outcome)?
Thank you
So does intention to treat analysis limit contamination biases?
Thanks for the video, it is still hard to understand because it is so counter intuitive.
in this example.. why dont we reduce the sample size to 90 in each group during data analysis as 10-10 participants from each group as lost.. can you pls explain that too?
Clear cut
Nice explanation. Do you have a video on mITT analysis?
Someone gave you a thumbs down just because. In reality you don't deserve one. You are helping me with my Journal Club Presentation. Are you by chance a Preventive Medicine Physician?
wow ,it is the first time 4 me to get smth clear in biostatistics
that audio at the beginning is always so bad 😆 but great videos! much appreciated
hi! can you tell me specifically how to perform intention to treat in spss?
There's nothing special to do. You analyze it however is appropriate but include all patients in the group,to which they were randomized.
Succinct explanation
Pathoma of Biostatistics
I posted my thoughts in my blog at ebmteacher
This example actually doesn't work because the design is fundamentally flawed . Each person in the control group should have been matched with an intervention subject and they should have both started treatment (i.e., surgery or ASA) at the same time. The RRR at 2:40 is higher because the time period over which stroke incidence was analysed is different between the groups. This is a tutorial about why it's important to design your study properly not a tutorial about ITTA
Amazing how he can describe something completely idiotic as if it not only made sense but was demanded of reason. There are in fact fewer events in the surgery. Intention to treat is certainly conservative because it's wrong. I doesn't preserve randomization. Not imbalanced prognostic factor. In the experiment, people sensibly want to know what is the effect of surgery compared to aspirin. Instead, ITT answers instead what is the effect of TELLING PEOPLE to have surgery vs. taking aspirin. Is that really what you want to know? Surgery is better. Being assigned to surgery does not depend on the surgery.
Above comments a little bit over the top but it really is true that intention to treat doesn't make sense. Since writing above I see the very useful other stuff this guy wrote so I will try to talk to him directly.
Each person in the control group should have been matched with an intervention subject and they should have both started treatment (i.e., surgery or ASA) at the same time. The RRR at 2:40 is higher because the time period over which stroke incidence was analysed is different between the groups.
CovidImages need to be invested more than half19
this is great