What is a confounder? - Epidemiology tutorial to learn the basics in only 5 minutes!

[@bg_1886]

Very clear presentation of what is seemingly a confusing topic. Many thanks.

[@nienkedeglas_mdphd]

thanks for the compliment!

[@arthurwalker636]

Thank you this video makes sense I actually comprehend confounding; keep up the good work.

[@RawaMuhsin]

Thank you for the clear and simple explanation.

[@ICQwhereRU]

Thanks, great and informative video.

[@nienkedeglas_mdphd]

Thank you :)!

[@gwynbird5361]

Thanks for the great video's! I have a question about predictive/causal (aetiological) studies: is it not done to look at both prediction and aetiology in one study? Study setup: prospective cohort study with pre-operative and (4month follow up) postoperative questionnaires. I want to know if I can use a questinonaire as a predictor for symptom score, so I will use a stepwise or backward multiple regression for all possible predictors and identify which model gives the best prediction of symptom score, evaluating if my questionnaire is the best predictor. Yet, some people have a preexisting illness or preexisting allergies, and I suspect these individuals might be at risk for developing symptoms (i.e. getting a higher symptom score). If I want to identify a possible causal relationship between these preexisting illnesses and allergies, I would like to adjust for confounders (in this case age, BMI and intoxications), but I don't know if this is highly unusual to do in a study where my main focus is the predictive value of a questionnaire. Thanks for your time!

[@gwynbird5361]

Furthermore, I would like to evaluate the effect of certain variables on recovery of symptoms (i.e. preoperative symptom score - postoperative symptom score (difference)). However, for one of my variables (duration of self-reported symptoms pre-operative), I am not sure if this can necessarily be defined as 'an exposure'. Therefore, i am not sure if I should be adjusting this variable for possible confounders. Thanks!

[@nienkedeglas_mdphd]

I get your point, indeed in many studies the two are used together. Of course, a causal factor can also be used as a predictor, as it is more likely to be cood in predicting your outcome. The important thing here is that you keep very clear in your manuscript about the distinction between the two. If you intend to develop a predictionmodel to predict your outcome, the choice of covariates will be different compared to if you truly want to prove causal relations, since in the last case, you will have to think in more detail about the possible confounders (and causal pathways). I hope this helps a little bit!

[@nienkedeglas_mdphd]

I think the duration of preoperative symptoms can be considered as an exposure, so yes I do think you should adjust for confounders if you want to look at the causal relation between this exposure and your outcome.

[@gwynbird5361]

@@nienkedeglas_mdphd Thanks for your clear and understandable comments! They are of great help. Keep up the videos!

[@wayneaskin780]

Very informative

[@jameelkifayatullah4819]

Nice video make more

[@marcomahdi435]

Thank you.

[@darixkuijs4075]

Geweldig uitgelegd

[@nienkedeglas_mdphd]

Bedankt!

[@danielkirunda5983]

0:42 very informative