@Terry Shaneyfelt Thank you for this video, which is very helpful for my exam. At 10:35, I am a bit confused why it concluded spurious association when unadjusted RR 0.208 and adjusted RR 0.39, which means the adjustment increases the RR. According to the rule of thumb from your presentation at 9:17, if the adjustment increases the RR or OR or remains stable, we can be more confident in the validity of the association. So, I thought the conclusion should be there are some associations instead of spurious. Please kindly correct me if my understanding is wrong.
I am a PhD student studying Epidemiology preparing for my comprehensive exams and came along your video. Love how you put these concepts into REAL WORLD clinical issues. Thank you so much for the easy to understand and relevant info :)
If I understand relative risk right, then at 0:58 where you say that when RR = 1, there's "no risk of outcome", I would actually say that it means that the exposure has no effect on the risk of the outcome.
@@manarahmed6888 What example is this? Its not in the video. If an RR is 0.58 compared to 1 in the control group, it means that the risk of getting the specific disease is 42% lower with the intervention. If an OR is 0.58 it means that the diseased person was 42% less likely to be exposed to what ever the researchers fear has caused the disease.
At minute 3:28 shouldn't we say "the outcome was 2.3 (instead of 3.3) times more likely to occur in the exposed group than in the unexposed group"? why didn't we apply the "1-RR" formula here?
Could it be that there is some misinterpretation at 10:35? Your unadjusted RR is lower than the adjusted (0.208 versus 0.39) - so basically the treatment reduces the risk of getting the disease by roughly 80% (unadjusted) versus 61% (adjusted). Doesn't that mean that the adjustment gives an "increased risk"? So your conclusion here would be wrong - or did I just get confused?
Im having the same issue. First off it would make no sense that we should be suspicious of an adjusted RR that is smaller than the crude RR. I would be suspicious if the adjusted RR was larger than the crude. If you are trying to eliminate confounders then the adjusted RR should be smaller than the crude.
wish to thank you for this video, was simple to understand, interesting, and helped me heaps for my exam, I now have a basic understanding of two concepts that frustrated and confused me to no end.
Hello Dr. Terry I would like to find an explanation the relationship between relative risk and odds ratio and i still find it hard to understand the difference between them and when each indicate protection or not thanks
Terry, thank you for your explanation. Can a take away be: The odds of someone being a smoker after being diagnosed with lung cancer is X while the risk of someone having lung cancer after smoking is Y?
Thank you for this video, which is very helpful for my exam. At 10:35, I am a bit confused why it concluded spurious association when unadjusted RR 0.208 and adjusted RR 0.39, which means the adjustment increases the RR. According to the rule of thumb from your presentation at 9:17, if the adjustment increases the RR or OR or remains stable, we can be more confident in the validity of the association. So, I thought the conclusion should be there are some associations instead of spurious. Please kindly correct me if my understanding is wrong.
Great vid !!!! I found it really helpful. Your way of explanation is phenomenal and unique. I've many courses but you are quite different, in a good way of course. Many thanks
Very helpful! thank you. You showed example on multivariate adjusted relative risk in this video. just wondering how to do multivariate adjusted relative risk on SPSS. Tips please
2 points: 1) (a/b) / (c/d) = ad/bc 2) think about a case control study which is what I am referring to when defining an odds ratio (ignore the output of logistic regression which is an OR also). Case control studies begin with cases (those with disease) and control (those without). They go back in time to determine exposure status. There is no disease development. It already happened.
I have a question at 6:09 mins, what does it mean that the risk of the women of CAD is 20%? does it mean that she has a risk of 20% over control group then shouldn't the RR be 1.2. By using a RR 0.2, it means that her risk is 80% less and not 20% more ( Because 20% cannot be written as 0.2 ratio because its 20% of something). so then her risk of developing CAD after E/P should be 1.2*0.39=0.468. Am i thinking about it in the right manner?
Yes this is definitely a mistake. Do you perhaps have an answer for my question? "At 9:20 you write we should be suspicious if the an adjusted RR is smaller than the unadjusted (or crude RR), which makes no sense to me. Next slide the crude RR is 0.2 and the adjusted is 0.39 which is bigger, not smaller than the 0.2. I know he says the 0.39 represent a decline compared to the 0.2 but that makes no sense to me.
Correct me if I'm misunderstanding, but if the baseline risk is 30% ( roughly 1:2 ) and the exposure increases the risk threefold, would the new risk be (3:2) or 60%, not 90% as in the video? Perhaps I'm not understanding something here. Thanks.
Baseline risk is the risk in control arm correct? RR= Intervention arm risk/ Control arm risk. Final risk= RR x Baseline risk = Intervention arm risk/ Control arm risk x Baseline risk (Control arm risk) = Intervention arm risk. So the final risk is equal to the calculated risk in intervention arm?
At 9:20 you write we should be suspicious if the an adjusted RR is smaller than the unadjusted (or crude RR), which makes no sense to me. Next slide the crude RR is 0.2 and the adjusted is 0.39 which is bigger, not smaller than the 0.2. I know you say the 0.39 represent a decline compared to the 0.2 but that makes no sense to me. So its a decline if the adjusted RR is smaller than (1.0 - 0.2) 0.8?
Hello professor, According to the example of result shown in this video, could you please come back and try to interpret all numbers that are in brackets?
Hello Professor, I have a question. Can we apply RR in case-control studies as well or could not, because RR= RR in exposed/ RR in non-exposed? Is it so???
Thanks for the helpful video, Terry. Quick question, what would you suggest is the best way to calculate relative risk involving continuous, binomial and ordinal explanatory variables with fall incidence as an outcome which is a count data. It's a retrospective study. I'm using proc genmod in SAS and haven't been able to find a related video. Thanks and appreciate your advice.
Me too, if you have the answer pls tell me lol. He says that the 0.39 represents a decline in risk because you need to say "1 minus the relative risk" - so its 1 - 0.2 = 0.8, and 0.8 is larger than 0.39 thereby representing a decline in risk (atleast I think this is what he means). But then howcome he could directly compare the 0.2 with the 0.22 (the age adjusted)? Shouldnt he then also do the same math here since an adjustment has also been made here? Im quite confused also.
Regarding your discussion of odds ratio, the odds ratio does represent the ratio of odds of disease by exposure when it's a prevalence odds ratio in a cross-sectional study. Just an FYI - otherwise, great video
Terri, I Liked your video and think it is a great review too. I want to know if RR is used in observation studies as well as experimental trial as well. My understanding is that OR is used for reporting case control studies, whereas RR is suitable for both experimental and cohort studies. Thanks
According to the confidence interval, there is no significative difference between 0.22 and 0.39. When comparing 0.208 against adjusted and non-adjusted RR, I would be misleading if I conclude that this association has evidence of a "spurious association". The confidence interval suggests that the variance is pretty big in that study.
I am interested in using this video as part of a maintenance of certification test. This will require that the video be posted for at least three years. Do you have any plans to take this down?
If a confounder is a third variable that is connected to the exposure and creates the same outcome as the disease, thereby skewing the result - how can the removal of a confounder possibly lead to a increase in the relative risk? Removal of the confounder would lead to a smaller risk of a the given disease after adjustment, i.e a RR that is smaller. But your example says we should be worried if the adjustment produces a decline in the RR. Several people in the comment section are wondering the same thing. Does the word "decline in RR" not mean the logical thing here or what?
very helpful but i wish you went more into detail on OR... RR was very well explained but OR was jumped through leaving me wondering about the difference between the two... an example would have been helpful
h+csdrt20 The difference between two are: 1. Odds ratio (OR) is a measure of odds of outcome with exposure while Risk ratio (RR) is the probability of the outcome. 2. OR is calculated in a case control study, while RR, in a cohort study. 3. Since we already have the outcome status known in a case control study and determine the exposure status, OR can not lead to the conclusion that it is the exposure which is resulting in outcome. It only measures that people with outcome have odds of having exposure. So it is a poor measure of causality. 4. Cohort studies, on the other hand have exposure status known and we follow up the study population until outcome occurrence or study completion. Here since we have the exposure and outcome in chronologically prospective order (exposure preceding outcome), RR is a robust measure of causality. Hope it is helpful.
The OR is the odds of being a case among the exposed (A/B) over the odds of being a case among the unexposed (C/D) [1]! It is not the odds of exposure with disease over odds of exposure without disease as the video states. [1] Aschengrau, A. Seage, G. R. (2008) Essentials of Epidemiology in Public Health. Sudbury, MA: Jones and Bartlett Publishers.
Does "incidence of outcome with exposure" mean those who have been vaccinated where exposure means exposure to vaccine? By "incidence of outcome without exposure" do you mean mean those who have not been vaccinated?
By "cohort studies" do you mean clinical trials in a controlled environment. By "case control" do you mean studies in the overall population being vaccinated, say.
Odds ratio used in case control and logistic regression. Rel risk used in cohort studies and RCTs
Sir difference between cohort study and case control?
but can't we used it OR in cohort and RR in Case control, when both are just mathematical terms?
Thanks alot Sir. Where can I find more such direct research paper derived examples & explanations?
@Terry Shaneyfelt Thank you for this video, which is very helpful for my exam. At 10:35, I am a bit confused why it concluded spurious association when unadjusted RR 0.208 and adjusted RR 0.39, which means the adjustment increases the RR. According to the rule of thumb from your presentation at 9:17, if the adjustment increases the RR or OR or remains stable, we can be more confident in the validity of the association. So, I thought the conclusion should be there are some associations instead of spurious. Please kindly correct me if my understanding is wrong.
I am a PhD student studying Epidemiology preparing for my comprehensive exams and came along your video. Love how you put these concepts into REAL WORLD clinical issues. Thank you so much for the easy to understand and relevant info :)
Wow! This should be the top video when you look up this concept. Super concise! Thank you so much!
You are a legend, your videos have helped me in my assignment so much, not sure where i would be without these videos
Thank you so much for your help. I tested out and passed our EBP class with the help of your videos.
this was very concise, intelligible and helpful. Thanks a bunch!
i have an exam in 3 hours and this video helped me clear a few concepts which i had doubt about .. thanks ! :)
If I understand relative risk right, then at 0:58 where you say that when RR = 1, there's "no risk of outcome", I would actually say that it means that the exposure has no effect on the risk of the outcome.
True. You said it better than I did
If the RR is 0.58, this means that there is reduced outcome by 42%?!
@@manarahmed6888 What example is this? Its not in the video. If an RR is 0.58 compared to 1 in the control group, it means that the risk of getting the specific disease is 42% lower with the intervention. If an OR is 0.58 it means that the diseased person was 42% less likely to be exposed to what ever the researchers fear has caused the disease.
Thank you so much!! I'm presenting a research project tomorrow and this video was exactly what I needed.
A little late 🤭 but how did it go?
Wonderful video sir. Extremely insightful, interactive & informative. Thanks a great deal for sharing.
Perfect explanation of the terms and clinical applications!
MD that understands statistics.
Much praise
Loving your analytical explanation…. Thank you
At minute 3:28 shouldn't we say "the outcome was 2.3 (instead of 3.3) times more likely to occur in the exposed group than in the unexposed group"? why didn't we apply the "1-RR" formula here?
+Mohamed Kandil I have the same thoughts as you. I think we should say the outcome was 2.3 times more
1*3.3=3.3, so the outcome was 3.3 TIMES more likely, BUT it's probability is INCREASED BY 2.3 (1+2.3=3.3) when compared to the unexposed number
Excellent video..thanks
Very clear explantion. Thank you.
Amazing presentation Doctor Terry!
*Only 18* 👇👇👇
555646.loveisreal.ru
Could it be that there is some misinterpretation at 10:35? Your unadjusted RR is lower than the adjusted (0.208 versus 0.39) - so basically the treatment reduces the risk of getting the disease by roughly 80% (unadjusted) versus 61% (adjusted). Doesn't that mean that the adjustment gives an "increased risk"? So your conclusion here would be wrong - or did I just get confused?
Im having the same issue. First off it would make no sense that we should be suspicious of an adjusted RR that is smaller than the crude RR. I would be suspicious if the adjusted RR was larger than the crude. If you are trying to eliminate confounders then the adjusted RR should be smaller than the crude.
Hi Terry. Thank you so much for wonderful and simplified explanation. This is exactly the preparation I needed for my exam. May God bless you...
wish to thank you for this video, was simple to understand, interesting, and helped me heaps for my exam, I now have a basic understanding of two concepts that frustrated and confused me to no end.
You just saved my grade! Thank you!
Explanations are so clear and precise:) Thank you.
Thank you for the explanation! It REALLY does help!
Thanks alot. Where can I find more such direct research paper derived examples & explanations?
Thank you for a very clear explanation. Very helpful indeed!
Can you apply the relative risk reduction formula to odds ratio too?
Thank you so much Prof. I learned a lot from your VDO.
Hello Dr. Terry I would like to find an explanation the relationship between relative risk and odds ratio and i still find it hard to understand the difference between them and when each indicate protection or not thanks
Honestly though do you have the time to calculate this risk for individual patients in a normally packed workday?
Terry, thank you for your explanation. Can a take away be: The odds of someone being a smoker after being diagnosed with lung cancer is X while the risk of someone having lung cancer after smoking is Y?
This was great! Super helpful! -Nurse Anesthesia
Thank you for teaching this so well!
You are a great teacher!
Hi. Why did you multiply 0.39 and not 0.61 for the RR in the lady taking E/P? Thank you
great vid! But, can you explain how 'per allele OR' is calculated in case-control study; e.g. association of one SNP with cancer risk?
THIS WAS SO HELPFUL!!!!!! THANK YOU
Thank you for this video, which is very helpful for my exam. At 10:35, I am a bit confused why it concluded spurious association when unadjusted RR 0.208 and adjusted RR 0.39, which means the adjustment increases the RR. According to the rule of thumb from your presentation at 9:17, if the adjustment increases the RR or OR or remains stable, we can be more confident in the validity of the association. So, I thought the conclusion should be there are some associations instead of spurious. Please kindly correct me if my understanding is wrong.
Awesome very well explained thank you
In the question given to compare the adjusted with unadjusted 0.208 multivariate adjusted was bigger 0.3 then y it was spurious ??
What does outcome mean for the lay person? Does it mean getting the disease?
For your example at 3:22, I would interpret that as "3.3 times as likely" but not "3.3 times more likely'
Thank you for the video. Well done and I learned a lot.
how about when the incidence in the unexposed is zero? And if you have a very small sample size, with an incidence of only 13.3% say?
For a non-medical student, What do you mean by exposed group? Do you mean not having had a vaccine?
Great vid !!!! I found it really helpful. Your way of explanation is phenomenal and unique. I've many courses but you are quite different, in a good way of course. Many thanks
Hello Terry, Is ODD ratio the same as Hazard ration?
Very helpful! thank you. You showed example on multivariate adjusted relative risk in this video. just wondering how to do multivariate adjusted relative risk on SPSS. Tips please
What are age-adjusted and multivariate-adjusted? Also, What is 'person-year'? And, what is the use of it?
I like very much the video, your explanation is crystal clear : )
Can you tell me what the difference is between the Hazard Ratio and the Odds Ratio????
Great video! Thank you very much! I will be sending people to it!
2 points:
1) (a/b) / (c/d) = ad/bc
2) think about a case control study which is what I am referring to when defining an odds ratio (ignore the output of logistic regression which is an OR also). Case control studies begin with cases (those with disease) and control (those without). They go back in time to determine exposure status. There is no disease development. It already happened.
I have a question at 6:09 mins, what does it mean that the risk of the women of CAD is 20%? does it mean that she has a risk of 20% over control group then shouldn't the RR be 1.2. By using a RR 0.2, it means that her risk is 80% less and not 20% more ( Because 20% cannot be written as 0.2 ratio because its 20% of something). so then her risk of developing CAD after E/P should be 1.2*0.39=0.468. Am i thinking about it in the right manner?
Excellent point raised Neha.
Neha Singhal i had that exact same thought!
Yes this is definitely a mistake. Do you perhaps have an answer for my question?
"At 9:20 you write we should be suspicious if the an adjusted RR is smaller than the unadjusted (or crude RR), which makes no sense to me. Next slide the crude RR is 0.2 and the adjusted is 0.39 which is bigger, not smaller than the 0.2. I know he says the 0.39 represent a decline compared to the 0.2 but that makes no sense to me.
Your an amazing teacher! Thank you
Correct me if I'm misunderstanding, but if the baseline risk is 30% ( roughly 1:2 ) and the exposure increases the risk threefold, would the new risk be (3:2) or 60%, not 90% as in the video? Perhaps I'm not understanding something here. Thanks.
Baseline risk is the risk in control arm correct? RR= Intervention arm risk/ Control arm risk. Final risk= RR x Baseline risk = Intervention arm risk/ Control arm risk x Baseline risk (Control arm risk) = Intervention arm risk. So the final risk is equal to the calculated risk in intervention arm?
thanks for the thorough explanation
At 9:20 you write we should be suspicious if the an adjusted RR is smaller than the unadjusted (or crude RR), which makes no sense to me. Next slide the crude RR is 0.2 and the adjusted is 0.39 which is bigger, not smaller than the 0.2. I know you say the 0.39 represent a decline compared to the 0.2 but that makes no sense to me. So its a decline if the adjusted RR is smaller than (1.0 - 0.2) 0.8?
Hello professor, According to the example of result shown in this video, could you please come back and try to interpret all numbers that are in brackets?
best explanation out there
Hello Professor, I have a question. Can we apply RR in case-control studies as well or could not, because RR= RR in exposed/ RR in non-exposed? Is it so???
Thanks for the helpful video, Terry. Quick question, what would you suggest is the best way to calculate relative risk involving continuous, binomial and ordinal explanatory variables with fall incidence as an outcome which is a count data. It's a retrospective study. I'm using proc genmod in SAS and haven't been able to find a related video. Thanks and appreciate your advice.
How to find odd ratio in minitab
i dont get why the last adjusted value is greater than the unadjusted but yet this make syou more wary? can someone help
Me too, if you have the answer pls tell me lol.
He says that the 0.39 represents a decline in risk because you need to say "1 minus the relative risk" - so its 1 - 0.2 = 0.8, and 0.8 is larger than 0.39 thereby representing a decline in risk (atleast I think this is what he means). But then howcome he could directly compare the 0.2 with the 0.22 (the age adjusted)? Shouldnt he then also do the same math here since an adjustment has also been made here? Im quite confused also.
Regarding your discussion of odds ratio, the odds ratio does represent the ratio of odds of disease by exposure when it's a prevalence odds ratio in a cross-sectional study. Just an FYI - otherwise, great video
Terri,
I Liked your video and think it is a great review too.
I want to know if RR is used in observation studies as well as experimental trial as well. My understanding is that OR is used for reporting case control studies, whereas RR is suitable for both experimental and cohort studies.
Thanks
Important presentation, Thank you!
According to the confidence interval, there is no significative difference between 0.22 and 0.39. When comparing 0.208 against adjusted and non-adjusted RR, I would be misleading if I conclude that this association has evidence of a "spurious association". The confidence interval suggests that the variance is pretty big in that study.
I am interested in using this video as part of a maintenance of certification test. This will require that the video be posted for at least three years. Do you have any plans to take this down?
it is really nice material
Thanks doctor, so informative
great videos.. thank you
This is very good!! thanks
Thank you so much. Very clear.
If a confounder is a third variable that is connected to the exposure and creates the same outcome as the disease, thereby skewing the result - how can the removal of a confounder possibly lead to a increase in the relative risk? Removal of the confounder would lead to a smaller risk of a the given disease after adjustment, i.e a RR that is smaller. But your example says we should be worried if the adjustment produces a decline in the RR. Several people in the comment section are wondering the same thing. Does the word "decline in RR" not mean the logical thing here or what?
very helpful but i wish you went more into detail on OR... RR was very well explained but OR was jumped through leaving me wondering about the difference between the two... an example would have been helpful
h+csdrt20 The difference between two are:
1. Odds ratio (OR) is a measure of odds of outcome with exposure while Risk ratio (RR) is the probability of the outcome.
2. OR is calculated in a case control study, while RR, in a cohort study.
3. Since we already have the outcome status known in a case control study and determine the exposure status, OR can not lead to the conclusion that it is the exposure which is resulting in outcome. It only measures that people with outcome have odds of having exposure. So it is a poor measure of causality.
4. Cohort studies, on the other hand have exposure status known and we follow up the study population until outcome occurrence or study completion. Here since we have the exposure and outcome in chronologically prospective order (exposure preceding outcome), RR is a robust measure of causality.
Hope it is helpful.
Terrific Terry!
hello there thank you so much for your videos
very informative.
The OR is the odds of being a case among the exposed (A/B) over the odds of being a case among the unexposed (C/D) [1]! It is not the odds of exposure with disease over odds of exposure without disease as the video states.
[1] Aschengrau, A. Seage, G. R. (2008) Essentials of Epidemiology in Public Health. Sudbury, MA: Jones and Bartlett Publishers.
Amazing. thank you very much.
Does "incidence of outcome with exposure" mean those who have been vaccinated where exposure means exposure to vaccine? By "incidence of outcome without exposure" do you mean mean those who have not been vaccinated?
Thank you so much Sir finally I got it
This was great!! Thank you :)
Thank you, it reallt helped me
amazing
Thanks a lot
Thank you!
Thank you
thanks for vid, was helpful. -er md.
Wish you interpreted the OR, you did RR but skimmed past OR.
Whichever video is viewed the most means it is the probably the most confusing lol
comfort
SUR THUN KOO VARY MUCK FUR HALP SUR
lectures
By "cohort studies" do you mean clinical trials in a controlled environment. By "case control" do you mean studies in the overall population being vaccinated, say.
thank you
Thank you