In logistic regression, if the dependent variable is the binary outcome at 10 years, then the model can produce the predicted probability for each observation at 10 years. The average of the individual predicted probabilities is the expected event rate for the study group. You can develop the model on one group and apply it to another group.
It's a bit messier to obtain (and explain) for Cox regression, but it's documented in the PHREG details.
Doc Muhlbaier
Duke