07 May 2020 3 9K Report

Hi,

I am not sure whats the best way to "control" for pubertal stage and gender in my analysis.

Actually, puberty is classified in 5 stages (1=prepubertal to 5=fully grown). To make it easier and also get more or less two groups with the same size, I also classified them into 1=prepubertal and 2=pubertal(stages2-5), which is valid from a clinical perspective.

So, what I want to do:

I want to see how continous marker X correlates with Insulin resistance (continous) AND if continous marker X is able to predict insulin resistance.

However, I am a bit stuck how to best implement pubertal status and gender in this analysis. it is likely that pubertal stage is somehow a confounding factor (since it is associated with insulin resistance), and maybe also gender.

--> How do I solve this? do a partial correlation with gender and puberty as cofactors? add them as first independent variables in a logistic regression model? how do I see/can say that puberty really influences the relation between marker X and insulin resistance?

If puberty is a confounder, it would then be possible to do two subgroup-analysis seperately evaluating the marker in prepubertal and pubertal children. (?)

thanks in advance!

More Ju Li's questions See All
Similar questions and discussions