i spent hours trying to understand the difference between confounding and effect modifier . your video explained the difference very simply and clearrly Thank you so much. Keep up your good work!
Thank you so so much noor for this! Though I was a little confused till the very end but that Reye Syndrome example made everything clear! If you can add like 3-4 more examples they would really help. Best!
Such a great explanation. My particular case is funny because right now i'm doing a question from UWorld Qbank for Step 1 from Biostatistics and is exactly the same as your example. Thanks for the insights!
I'm a student and I'm not sure if I'm right, but in the last example, if age was a confounder, we would see the same effect, and the results wouldn't differ between children and adults. So it's not just if we didn't see association.
Can you tell me if this is right: Confounding: The relation you see is not real, there is something else that is the actual cause of relation. Effect Modification: The relation you see is real, but this relation will only be seen when a modifier is present/absent.
Very clear and thanks. but I still have a quick question. could we say there is no association between determinants and outcomes regardless of confounders? (which is mentioned at 8:52 in this video.) I think the etiologic research is interested in finding the causal relationship between the determinant and outcomes. The researchers have to try to eliminate the effect that the confounders make in the occurrence relation but should we say the opinion above?
In 8:50, there’s a true association only in the presence of the effect modifier (age). There’s no true association in adults. Aspirin is the determinant, liver failure is the outcome and there’s no confounders
Loved it! Made more sense and last example of Reyes sx summarized it all. So to avoid confounding, do u use stratification? U hinted on it somehow. Thanks lots
Thank you Solomon! To avoid confounding from the start we match all variables (some of which are potential confounders) except the variable we are interested in. So we should match all smokers together and then start asking about alcohol use. Now if we didn’t match from the start but suspected there may be a confounder after we saw the results then stratification should eliminate the confounding effect. I hope this makes sense
Hello Noor. In the example of effect modification, there is no increased risk of DVT in patients treated with Estrogen who dont smoke but increased risk in those who smoke. This shows that estrogen dosent lead to DVT alone who dont smoke. Cant this be called as cofounding due to smoking?
the part of the question that says "In non-smokers, no increased risk of DVT is evident with the use of drug RR:0.96" Implies that the drug doesnt actually have an effect. While in effect modification the primary variable [drug] has an effect and the effect modifier plays on the extent of the effect either by increasing it or decreasing it... do you get what I mean?
@@acingmedicine I did watch the entire video. I really love your other videos. I felt that this video was worded a little complicated. I watched it a couple of times and I understood it though. With peace and love 💞
