Deployment Type vs. Outcome (where is it?)

Why bother with all these so-called scientific studies (often twisted to suit a goal, such as using bizarrely-small sample sizes to avoid the risk of finding moderate [~3% [?]] risks) when we already have a huge street-level experiment going on?

The authorities have, or should have, access to all the data needed to answer the following questions:

• How many subjects have mysteriously fallen over dead when the police officer held up a taser and threatened to (but didn't actually) use it? According to Taser, there should be no difference in the death rate.

• How many subjects have died after being tasered in Touch Torture ('Drive Stun' [sic]) mode to an extremity such as an arm or leg? According to Taser, there should be no difference in the death rate "...since the deaths have nothing to do with the taser."

• How many volunteers or trainees have died after being carefully tasered in the back, or down one leg? Different population preclude simple comparisons, but such carefully stage-managed (FAKE) examples shouldn't be allowed to dilute the real-world statistics.
  

• Of all the several hundreds of cases where someone has mysteriously died due to some sort of internal factor (as opposed to falling off a bridge), is there any evidence of a statistical surplus of full-on taser deployments to the chest or thorax?

Such an honest evaluation of the detailed reports, with all the Taser denominator washing [LINK] accounted for, would provide much better evidence of the real-world safety (or not) of the taser.

Based on the many news reports that I've seen, I believe that the answer would be clear-cut and obvious (and frightening, for both sides...).

And perhaps that's why we haven't seen a report covering this sort of very straightforward analysis of Deployment Type versus Outcome.