[Dennis Ton]

For reasons having nothing to do with the issue of what can be done with the original data and what can be done with imputations, this sentence is very wrong. An ROC analysis says nothing whatsoever about the correspondence between the value of a predictor and the associated risk. An ROC analysis measures the ability of the model to discriminate positive and negative outcomes, that's all.

Excuse me for my mistake.
I was meant to ask: should you interpret the ROC analysis as an aggregation of all imputed datasets and regression analyses?
Since you first calculate an aggregated predicted probability using mi predict, and then generate the ROC (assuming that the data on the outcome variable was not imputed and thus is the same across each imputed dataset.
Referring to your response at #2.

[Clyde Schechter]

should you interpret the ROC analysis as an aggregation of all imputed datasets and regression analyses?

Well, being careful about the wording, I would say it as: the ROC area is calculated from a model that is an aggregation of all imputed data sets and regression analyses.