This is the clearest explanation of odds and risk ratio on the internet. The video is very clear on why odds ratio is used in case-control study. I hope I am not being picky, but I do wish the author would use "for" and "against" terms and two whole numbers in discussing odds ratio. Yes, the video is the best. I am being picky.
+Rahul Patwari (Rahul's EM) Using your heart attack numbers, there are 10 patients in the group. 6 patients had a heart attack and 4 did not. I think gamblers would say the odds "for" a heart attack are 6 to 4. Gamblers might state the same relationship as the odds "against" a heart attack are 4 to 6.
I'm not a gambler, but I believe there are number of ways gamblers express odds. I was just going for the way the statisticians do it, but your point is well taken. Thanks!
I just want you to know how much I appreciate your videos on biostats. As an epidemiology instructor I love using your videos for supplemental instructional material. I wish you were my instructor in my day!
Absolutely fantastic explanation. You have a natural talent for teaching, not just delivery but the clear breaking down into relevant sub-concepts and anticipating questions. Bravo.
The concepts couldn't be explained more clearly! Odds and risk ratios have confused me greatly throughout medical school and this is the first time I've ever had an insight and intuition into the concepts. Thanks so much!
Thank you so much. There is a bunch of information on other courses butI couldn't find any playlist for epidemiology than yours.thanks for sharing it free🙏
I am also a MPH student overwhelmed over Epi module, this guy is soo good. His calm way of explaining it without rushing through it definitely helps as do the examples as they are all relevant
I don't know why we cant use the risk in case control study? Viewer may take granted by seeing this video that Odds should be used for case control study while risk/probabilities should use for Cohort studies. Both can be used in either case , its not the study which decide what to use. Both has different way of expressing same measures
This is absolutely amazing. The best explanation on OR and RR with relation to case-control and cohort studies. I guess cross-sectional studies would use OR as well and randomised controlled trials would use RR since they are an upgrade of the cohort study. Additionally, could you please explain how would we interpret the results of RR and OR? Using your example, would RR of 3 be interpreted as "those who take pills have 3 times more risk to develop MI than those who do not take pills" and would OR of 3 be interpreted as "those who take pills are 3 times more like to develop MI than those who do not take the pills"? Thank you for your videos.
Hi Rahul, thanks for your good explanation. It really helps me a lot. Just a quick question. As you mentioned at 15:40, if the number in confidence intervals is 1 which means it is not statistically significant, which means there is no different, is it same if it is more than 1, like the example you gave 11.2, it is more than one and not statistically significant, does it mean it has no different too? Thanks Rahul.
Awesome! Can not having mammograms and getting breast cancer be calculated with these scenarios? I really enjoyed the thoroughness and diagrams, which helped to clarify odds and risks.
I still don't understand why you can't calculate the relative risk from a case-control study. Once you put your data (exposed and non-exposed, diseased and controls) in a contingency table, you should be able to manipulate the data as you want, in the same way as you do in a cohort study, since both study designs use the same instrument (the contingency table) to analyze the data.
I guess you 've worked it out now, as it was so long ago that you posted your comment - but I had the same query, and I'm making this reply is in case anyone has the query too. In a case-control study, the numbers of people who are affected/not affected is more up to you. Maybe you have a group of patients who have had a heart attack (10 in his example), then you go looking for other patients who didnt have a heart attack to compare them with. Maybe you have 10 (as in his example) - or maybe you keep on looking, and find 20, or 30 etc. These will change the numbers in the contingency table and you will get different risk ratios (because you are deliberately searching out more people who didnt have a heart attack). However the caculated odds ratio will stay the same.
You made biostatistics understandable, can you have the whole series for biostat? Im searching for more video from you. Dr. Rahuh you are a good teacher. Thanks you very much for your gift
Wow, you are different. Thank you for such a great video. One question. If you do a retrospective analysis but you look at the effect of exposure on the outcome. (in my case, the risk of cancer recurrence depending on the duration of chemotherapy) Would you still use risk or logistic regression or what?