I generally agree with your post, but this stat is a fair way off. Your 5% is based on fatalities : races started. The odds of being killed would be fatalities : drivers starting races.
1018 races with 20 drivers on average (don’t know how accurate that is) = 20,360 starts. To 52 fatalities, that’s 0.2%. So if there’s been 5 keels lost over the years the risk of dying in an F1 race is something like 5 times greater than losing your keel in a Hobart.
Of course the comparison isn’t apples for apples, so I’m not sure how helpful it is at all. IMO it’s enough to just note a 0.04% failure