WILLIAM BRANGHAM: The death of George Floyd, which sparked a national reckoning over policing, resulted from a police technique that is considered non-lethal.

A new investigation led by the Associated Press has now found that is much more common than had been thought.

John Yang has our report.

JOHN YANG: William, the investigation found that, between 2012 and 2021, more than 1,000 people died after police used physical force that was not intended to be lethal.

That includes batons, stun guns, physical restraints and chemical agents.

The oldest victim they found was 95, the youngest 15.

Only 28 of the officers involved were charged, and the police role was cited in only about half of the cases, meaning that many more Americans have died at the hands of police than previously known.

The Associated Press investigation was done with the Howard Center for Investigative Journalism at the University of Maryland and Arizona State University and PBS' "Frontline," whose report on this will air April 30.

AP investigative reporter Reese Dunklin was part of that team.

Reese, so how did the use of force that is not intended to be lethal end up in deaths?

REESE DUNKLIN, Associated Press: Well, we found in hundreds of cases, what happened was officers went too far, too fast, too long, and they were making errors in the way they applied this force that went beyond best safety practices.

So they would hold people down on the ground on their chest in a way where these people couldn't breathe, their hearts couldn't function as properly.

And that's exactly what happened to George Floyd.

Or, in other cases, they would use their stun guns too many times and for too long.

JOHN YANG: When you look at the victims in these cases, is there a common thread about where these instances happen, who the victims are, and what they -- why they were being restrained?

REESE DUNKLIN: Yes, we found that, in a number of cases, the location was actually the homes of those who died, and that was because sometimes they were having medical emergencies, like potential drug overdoses, and their families were calling 911 to get medical assistance.

And the cops arrived.

And in order to try to control the person to get medical help, the cops went too far.

And another thing we found as far as a commonality is that the toll fell hardest on Black Americans.

Even though they represent only 12 percent of the nation's population, we found that they were about a third of those who died.

And probably the last commonality we found was why they ran were in these situations.

And, again, it was mental health crises.

It was a lot of drugs.

And that's what brought them to the attention of law enforcement.

JOHN YANG: In your report, you say that about half these cases, only about half, is the role of the police cited in the -- as either causing or contributing to the death.

In the other cases, what do they say?

How do they attribute these deaths?

REESE DUNKLIN: Well, they attribute it in a couple of ways.

First, some of these were attributed to a controversial label for a condition that some critics say is a way to justify excessive force.

And that label is excited delirium.

And that's meant to be a way to say someone is highly agitated, their heart's racing, they're sweating profusely.

And police and other people who subscribe to this believe that, in and of itself, is life-threatening.

So that's how it is used to sort of shift the blame away from the force.

In other cases, we found someone's preexisting medical conditions were blamed or even their mental health problems.

And then, in other cases, legitimately, they had very toxic levels of drugs in their system, and that made them vulnerable when police used even just minimal force.

JOHN YANG: There's been so much focus on police shootings, of course, shooting people who they're trying to stop or trying to arrest.

Is there anyone who keeps track of these cases, cases where their nonlethal use of force ends up in a death?

REESE DUNKLIN: Well, the federal government is supposed to be tracking it through the DOJ.

But DOJ has struggled for years to track them in an effective, complete way.

DOJ says that's partly because local police departments won't cooperate and send in the information to their states, which can then send it in to the feds.

And then CDC tracks it nominally through its mortality death data.

But that's not really designed to capture everything, so it doesn't capture everything, based on what we found.

JOHN YANG: It sounds like you would say that a lot of these are instances where the officer goes too far or just does it in the wrong way.

Is this basically a training problem?

REESE DUNKLIN: That's what some of the experts tell us.

Many departments in many states get training right, and they have instances like the kind we saw where everyone gets out safe and everyone goes home at the end of the night.

But there are departments that aren't following best practices, and the experts told us that, if they would just modernize, update, and reassess what policing associations like the International Association of Chiefs of Police prescribe, then maybe we would have fewer deaths.

JOHN YANG: We should talk about officers going too far or putting a hold on too long.

Are there examples you can cite?

REESE DUNKLIN: Yes.

We had one example that we mentioned in the article yesterday where a Texas man was shocked up to 11 times with a Taser stun gun.

And he was held in what is known as prone restraint.

That's pinned down on the ground on your chest, where it compresses your lungs, so it makes it tough to breathe.

He was held that way for 22 minutes.

And to give you a frame of reference, George Floyd was held down for roughly nine-and-a-half minutes.

JOHN YANG: Reese Dunklin of the Associated Press, thank you very much.

REESE DUNKLIN: Thank you.