(CNN) Debate over police using excessive force in black and Latino communities has been an ongoing theme in America's narrative long before this month's high-profile police shootings of two black men, Alton Sterling and Philando Castile.
However, when it comes to a scientific look at such police actions, the data seem to be lacking. There is no universally accepted and scientifically backed data source for when injuries or deaths occur during police interactions -- and many scientists say there's an important need for one.
"This is a very old problem that we have been trying to solve since the Civil War, through the civil rights movement, the riots of the 1960s, the Rodney King riot," said Ted Miller, principal research scientist at the Pacific Institute for Research.
Miller served as lead author of a provocative new paper published in the journal Injury Prevention on Monday that combined statistics from various hospital, law enforcement and media sources to expand data on police-involved violence. The paper found that while racial minorities were more likely to be stopped by police, the probability of being killed or injured during that stop may not vary by race.
Experts not involved in the paper have praised the approach of creating such a data cocktail but also questioned its findings.
"Using medical data is a creative way to get more information than we otherwise have, but there are some limits to that methodology," said Seth Stoughton, assistant professor at the University of South Carolina School of Law and a former police officer, who was not involved in the new paper.
"The big takeaway is, this tells us how desperately we need accurate, national standardized information about police use of force," he added. "We shouldn't be having to dig through medical coding to get this information."
Miller and his colleagues analyzed data on injuries resulting from police interactions from the 2012 Healthcare Cost and Utilization Project, the U.S. Vital Statistics Mortality Census, the 2011 Police Public Contact Survey, FBI reports on stop-and-search procedures, traffic stop arrest reports in states that did not report their arrests to the FBI, and newspaper reports.
The analysis showed that black people, Native Americans and Latinos were more likely to be stopped and arrested than whites and Asians. The analysis also showed that blacks were more likely than whites to die at the hands of police.
But, according to the new findings, once an actual stop or arrest occurs, that racial disparity disappears, and blacks face no greater risk of injury or death -- a finding that Miller said surprised him. The new paper included data only on traffic stops that resulted in a search, he added.
"The benchmark of using stops is an unreasonably conservative test of bias," said Phillip Atiba Goff, the Franklin A. Thomas Professor in Policing Equity at John Jay College of Criminal Justice, and president of the Center for Policing Equity. "That's because if that bias does exist, you're more likely to stop and arrest someone who is not white. If you want to measure bias, you can't use a benchmark that in itself is biased."
As a Los Angeles Times headline put it, "Blacks are more likely to be killed by police, but that's because they're more likely to be stopped, study says."
But Goff referenced a separate think tank study published last week by the Center for Policing Equity, which he co-authored, that found that the average rate of using force among blacks tends to be 3.6 times as high as among whites and 2.5 times as high as the overall rate.
That previous research showed that police officers employed force in less than 2% of all interactions, and use of force was not associated with crime rates in the area where the force occurred.
Stoughton also said he thought it was interesting that the new paper found no real racial disparities in the overall rates at which officers used force resulting in injury.
"That's a valuable contribution when we're still struggling to understand the racial dynamics of policing," Stoughton said, but he added that the finding shouldn't necessarily be seen as conclusive. "The fact is, there's nothing simple about policing and nothing simple about race relations."
Since the paper involves medical coding, it may be true that officers are more likely to seek medical treatment for whites compared with blacks who are injured, Stoughton said. For instance, a study published in the journal Proceedings of the National Academy of Sciences in April found that doctors were less likely to prescribe pain medications to black patients than white patients, possibly due to an implicit racial bias.
"If there's something similar in policing, there might be a disparity," Stoughton said. "It may be that officers are using more force against blacks, but they aren't getting the same medical treatment. ... That's an open question. I'm not saying that's the case. I'm saying we don't know yet."
The new analysis also revealed that about 55,400 people were killed or treated in a hospital for injuries suffered during legal stop-and-search procedures in 2012, with about 1,063 suffering fatal injuries.
During that time, about 67,000 officers were assaulted while on the job, with an estimated 18,600 receiving medical treatment for injury, and 48 were killed, according to the paper.
But, Stoughton pointed out, by having medical data available only for people who suffered injuries during interactions with officers, the paper offers a view of only medically treated injuries and then compares those with all assaults against officers.
"That's not a great comparison, because of the 67,000 assaults, only about 27% of those involved injury that require any form of medical treatment," Stoughton said.
Miller, lead author of the new paper, said he was surprised to find that, on a typical day, about three people die and 150 are injured during incidents involving police nationwide.
However, the paper shows that injuries in general are more likely to be severe if they result from another person assaulting you than an officer.
"This study adds a clearer picture of the broader range of injuries associated with police action," Miller said. "In showing that injuries in assaults are much more likely to result in hospital admission than injuries during legal intervention, it suggests that police are not typically out of control when they injure someone, although undoubtedly they sometimes are, which is inappropriate."
David Klinger, a professor of criminology and criminal justice at the University of Missouri-St. Louis, agreed that the paper offers an interesting assessment but said there are limitations.
For instance, the paper measures injuries but doesn't take into account whether these injuries may have resulted from resisting arrest, he said. "I don't know about the validity of that data."
Klinger also added that the data didn't include injuries from police dog bites.
"That would be interesting to track that, because there is some research from a number of years ago that reported black suspects were much more likely to be bitten by police dogs," Klinger said. "There may be some racial effect that is not being uncovered because they don't have the dog bites."
All in all, the authors have done the best they can on the limitations of the data that exist, said Alex Piquero, the Ashbel Smith professor of criminology at the University of Texas at Dallas.
"Miller and his colleagues are dealing with a world of very imperfect data, and they are trying to put together that data," he added. "What we should do at the federal level is to get people to put into and populate a database where we can get this information. ... That, to me, is the moral of the story."
For instance, on the city level, the Dallas Police Department was one of the first to publicly release data on officer-involved shootings, Piquero said. "In my mind, when we're having these discussions, we have to have the latest data at the table."