If a Statistical analysis is going to have any reliability at all, it must include data current right up till yesterday evening. Cutting off a few Years at this end fails the "Smell Test". Including Data from Day One of North Sea Operations is valid so long as comparisons use similar Data derived from similar periods of time.
I can see where omitting the Loss of a Rotor Head and a CFIT crash that killed four people might skew the Safety Stats in a more favorable direction than if they were included (and I am just a Dumb Ass Helicopter Pilot).