Hi,

You can look for collinearity diagnostics for correlation among IVs before looking at interactions

You can not investigate if B has an effect on O if you don't experimentally control B. In your experimental design you can only investigate the correlation between B and O, and if this correlation is impacted by the presence of D (this is the same as asking if the effect of D depends on B). This is addressed by the interaction term in the model. But this is interpretable only when the relationships between D and O and between B and O are really linear. Otherwise the interaction term will only account for non-linear aspects of these relationships what is not what you want to know.

I would first study the relationship between B and O thoroughly; it might not be linear at all... And then see how D disturbes that relationship. Don't assume a simple linear relationship; it's mostly wrong.

If you want, I can take a look at your data.