I am very new to R and epidemiology and am trying to create a model of how metabolic syndrome is associated with disease 'Y'.

I have been asked to calculate a standardised odds ratio for the effect of BMI on 'Y'.

My model is

METABOLICfit<- glm("Y"~AGE+SEX+BMI+DM+HTN+SMOKER,data=MyData,family=binomial(link="logit"))

exp(coefficients(beta(METABOLICfit)))

       Estimate Std. Error      z value Pr(>|z|)

(Intercept) 0.002283022   1.034872 8.693510e-78 1.000000
SEX.z       0.982666829   1.032443 5.783098e-01 1.793081
AGE_2018.z  1.073146291   1.034939 7.811410e+00 1.040626
DM.z        1.170444708   1.022757 1.090366e+03 1.000000
HTN.z       1.227583387   1.033393 5.140717e+02 1.000000
SMOKER.z    1.014942921   1.032379 1.592752e+00 1.899517
BMI.z       1.227770586   1.028971 1.319658e+03 1.000000

I receive a standardised odds ratio for BMI of 1.23. My question is, is it possible to only standardise specific variables for example only BMI? This is because my other variables e.g. DM, HTN and SMOKER are binary (i.e. either 0 or 1). Is it then reasonable to present the data by showing some odds ratios as standardised and others not?

I am also unsure how to calculate a confidence interval for the standardised odds ratio. I tried using this:

exp(confint(beta(WHRnewfit))) 2.5 % 97.5 %

Another question if, if I am also looking at biochemical variables e.g. Cholesterol, HDL, LDL etc. do these also need to be standardised to allow comparison between variables?

Further, when adjusting a model for age and BMI would it be sensible to use standardised inputs for age and BMI? Would this change the output?

Thanks in advance