When should statisticians become involved in a clinical trial? Why?

Paper detalis:

1. When should statisticians become involved in a clinical trial? Why? (i.e., hint, watch Clinical Research Statistics for Non-Statisticians).

2. In your area of clinical practice what type of statistics have you seen used the most, predictive, causal or both (i.e., provide an example as part of your answer)? Why do you think this is?