[bombcar]

It can be more efficient but the question remains if it's more likely to kill than otherwise (not all travel miles are equal).

In one state 34% of traffic fatalities involved alcohol. So Uber may do more for that than grocery delivery.

Grocery delivery drivers may also be better drivers because drive more often. Or, they may be overworked and rushing to complete deliveries.

[wizofaus]

Stats on % of car accidents (esp. fatalities) involving delivery vehicles would be good to see actually, they must be available somewhere?

[bombcar]

The DOT almost certainly tracks "real" delivery drivers (think: semi trucks, etc), but who knows how much gets reported from fly-by-night outfits like Uber and Doordash.

https://cdan.dot.gov/query could probably return something useful if poked long enough.