Support Independent Journalism. Donate today!

How the Federal Government Lost Track of Deaths in Custody

The Department of Justice is leaving researchers, policymakers, and advocates in the dark about deaths in police custody, prisons, and jails.

Photo illustration by Elizabeth Brown. Photo from Getty Images.

Amid protests against police violence in the wake of George Floyd’s death at the hands of Minneapolis police officers, politicians have embraced calls for more data about police use of force.

Democratic members of Congress released a proposal on June 8 calling for the creation of a nationwide registry of police misconduct along with mandatory reports from local, state, and federal law enforcement agencies on use of force data. The Trump administration approved a much more limited proposal in a June 16 executive order, requiring police agencies that receive federal funding to submit reports to a database of excessive use of force incidents.

Efforts by the federal government to track deaths in law enforcement custody, including deaths during arrests as well as deaths in prisons, jails, and immigration detention centers, predate George Floyd’s death by nearly two decades. But despite these efforts, little public data exists on how, why, or even how many people die in the custody of law enforcement or correctional officers. Efforts to implement a 2014 law that sought to rectify this issue have been delayed by confusion over the law’s requirements and opposition by law enforcement groups, and advocates claim that the Trump administration has intentionally undermined efforts to collect an accurate count of in-custody deaths in the United States.

“We have been waiting for more than five years for simple data collection, and this administration is just ignoring its statutory obligations,” said Sakira Cook, director of the Justice Reform Program at the Leadership Conference on Civil and Human Rights, an umbrella organization for more than 200 civil and human rights groups.

Challenges from the start

Before 2000, little federal data regarding the number of people who died each year in prisons, jails, and police custody existed. The issue gained national attention in 1995, when Mike Masterson, then an investigative journalist at the Asbury Park Press in New Jersey, used press reports to estimate that more than 1,000 people had died in local jails nationwide during a four-month period, many under suspicious circumstances. Shocked that no official data existed on in-custody deaths, he traveled to Washington, D.C. to distribute copies of his article to members of Congress.

Five years later, Congress unanimously passed the Death in Custody Reporting Act of 2000, which required states to report every death of a person who “is in the process of arrest, is en route to be incarcerated, or is incarcerated at a municipal or county jail, State prison, or other local or State correctional facility” to the Department of Justice every three months. In a speech on the House floor, Republican Asa Hutchinson, then a representative from Arkansas and the bill’s co-sponsor, credited Masterson’s reporting for inspiring him to introduce the legislation.

“I felt so encouraged when it passed,” Masterson told The Appeal. “It’s certainly one of the top stories I’ve done in my almost 50-year career.” 

The Bureau of Justice Statistics, the DOJ agency that collects crime data, began collecting the data that same year. Between 2001 and 2016, the BJS found that at least 16,058 people died in local jails, nearly half by nonnatural causes like suicide or drug overdose, and nearly 60,000 people died in state and federal prisons. The agency also reported nearly 5,000 arrest-related deaths, the majority of which were police homicides, from 2003 to 2009, the last consecutive year of full data on arrest deaths.

But the program faced challenges from the start. Congress provided no funding to implement the law and no enforcement mechanism to ensure states reported data, meaning the BJS had to rely on law enforcement and correctional agencies to provide the data voluntarily. 

The BJS used its own budget to collect data from all 50 state prison systems and nearly 90 percent of local jails each year, but the agency lacked the resources to survey each of the nation’s roughly 18,000 law enforcement agencies individually, so it relied on state governments to track arrest-related deaths on its behalf. Many states collected that data in a haphazard fashion, often relying on media reports or voluntary reports by law enforcement agencies with no additional verification. Only 36 states reported data every year.

Many states collected that data in a haphazard fashion, often relying on media reports or voluntary reports by law enforcement agencies with no additional verification. Only 36 states reported data every year.

In 2014, BJS suspended its arrest-related deaths program after states failed to report sufficient data in 2012 and 2013. The following year, an internal review found that BJS was recording data for fewer than half of arrest-related deaths nationwide; roughly 500 deaths each year were falling through the cracks.

Even when the BJS did release in-custody mortality data, the bureau only made state- and national-level data available to the public. When The Appeal sought facility-level data, the agency said it has a blanket policy of rejecting such FOIA requests. “BJS cannot provide facility-level reporting because it opens the door to identification of decedents, which would violate privacy rules,” the agency told The Appeal.

Additionally, former BJS officials told The Appeal that reporting facility-specific data could jeopardize their relationship with those reporting it. 

“Statistical agencies need to build credibility and trust with data providers, and if data is used against them, that’s the end of the data,” one former official told The Appeal. “If you start naming people, they are less willing to give you data.” The official requested anonymity in order to speak more freely about the agency’s internal procedures.

But advocates for criminal legal reform argue that keeping this data confidential undermines public accountability for problematic correctional facilities and law enforcement agencies. 

“The purpose of the law was to increase transparency, to be able to have national statistics, and to be able to look at those statistics to offer solutions,” Cook said. “If lay citizens are able to see that there are lots of deaths in their town or county or whatever, that could suggest systemic issues that need reform.”

New law, old problems

In 2013, Virginia congressman Bobby Scott, who co-authored the original Death in Custody Act, sought to solve these problems and sponsored a new bill, the Death in Custody Reporting Act of 2013, which would impose financial penalties on states that didn’t comply with the reporting requirements. Law enforcement organizations opposed the bill, but it passed unanimously in December 2014, just months after the high-profile police killings of Eric Garner and Michael Brown. 

“Hopefully there will be better compliance and enforcement than existed then, and also more cooperation,” Connecticut Senator Richard Blumenthal, a co-sponsor of the legislation, told Mother Jones at the time. “There’s certainly more awareness now about the importance of this data, and much more focus on it.”

But the new law left one of the original DCRA’s most fundamental flaws unaddressed, according to one former DOJ official who worked on implementing the new DCRA during the final two years of the Obama administration. The new bill didn’t provide states or the federal government with additional money to modernize the often-fragmented and archaic data collection systems.

It also created new problems. If the DOJ used data collected by the BJS to impose the law’s new penalties and mandatory reporting requirement, it could risk breaking federal law, which prohibits the government from using BJS data for law enforcement purposes, department lawyers determined. Instead, DOJ officials assigned the task to the Bureau of Justice Assistance, which administers department grants to the states but has little experience conducting statistical research.

“BJA is not a statistical agency; it’s a grant management agency. BJS has expertise, but they’re scientifically independent,” the former official told The Appeal. “None of this was contemplated in the statute.”

Additionally, penalties for noncompliance fall on state governments, but much of the data is in the hands of local sheriffs and police. As a result, the former official noted, agencies that comply with the law could be penalized for noncompliance by other agencies in their state. 

Despite these challenges, the Obama administration released a plan for implementing the law in December 2016. The plan required BJA to use media reports and other public records to verify the accuracy of data reported by the states; a 2016 study by BJS found that these methods allowed the agency to identify nearly three times more arrest-related deaths than it found in previous years by relying on data from states alone. The plan also called for the release of data on the number of deaths reported by each participating agency and facility.

Still no public data

About a month later, however, Donald Trump became president, and the new administration put plans to implement the law on hold. According to a 2018 report by the DOJ’s inspector general, the agency’s new leadership thought the plan overburdened states and exceeded the law’s requirements. According to lobbying disclosure forms, law enforcement organizations like the Major County Sheriffs of America lobbied the DOJ throughout 2017 and 2018 on the implementation of DCRA. 

In June 2018, the DOJ released a revised implementation plan for public comment. The new plan eliminated almost everything from the Obama-era plan, only requiring states to complete two-page incident reports for each in-custody death. The requirement that BJA verify state data against media reports and public records were gone altogether, leading the DOJ’s inspector general to conclude, “if the Department implements its current plan, it may not produce complete data about deaths in custody.” Gone, too, were plans to make DCRA data publicly available.

According to Samuel Sinyangwe, who co-founded the Mapping Police Violence data project, the lack of verification “kills the program.”

“Without that, you are entirely relying on police agencies to provide this data with no enforcement power and a proven track record of [agencies] refusing to do so,” Sinyangwe said. “When you use an outdated methodology, you rely on agencies to provide data in a way that will never get you to a comprehensive total, unless you have strict enforcement, and [the DOJ is] just not willing to do that.”

The American Civil Liberties Union, the Leadership Conference on Civil and Human Rights, and the NAACP Legal Defense and Educational Fund, among other organizations, wrote in a 2018 public comment that the Trump administration’s DOJ plan “bypasses the comprehensive findings and recommendations of BJS’s technical review and redesign study, as well as the public deliberation that was undertaken during the 2016 comment periods. There is simply no justification for this.”

We have no idea how many people are killed by police each year.

Kanya Bennett senior legislative counsel at the ACLU

But DOJ moved forward with the plan, officially starting data collection in January 2020, more than five years after the DCRA became law. 

However, it’s unclear what, if anything, the DOJ will do with this information. In 2018, the department said it had no plans to report state data collected under the DCRA, saying only that “Once data collection has begun, the Department will assess what kinds of reporting would be appropriate based on the available data.” In September 2019, a DOJ spokesperson confirmed that the data collected under DCRA will most likely not be made publicly available. “As part of their clearance to collect data … BJA agreed to pass the mortality information to BJS,” the spokesperson told the Palestine Herald-Press, a Texas newspaper. “Any data transferred becomes subject to data protection guidelines.” In May, a spokesperson for BJA told The Appeal, “At this time, BJA does not have a plan to make the data collected under DCRA available to the public.” 

Meanwhile, the BJS continues to collect its own data on deaths in prisons and jails, relying on the same voluntary reporting practices it has used for the past two decades. But only 20 percent of jails report any deaths at all in a given year, according to BJS. And delays in reporting the data have increased under the Trump administration; BJS didn’t release any mortality data until February, when it released death figures for 2016. 

The BJS has also yet to revive its arrest-related deaths reporting program, meaning there is no publicly available, official data on deaths from police encounters. “We do not have data on police and community encounters. We have no idea how many people are killed by police each year,” said Kanya Bennett, senior legislative counsel at the ACLU.

The lack of data has made it harder to spot particularly problematic correctional facilities and law enforcement agencies. For instance, between 2003 and 2018, Hampton Roads Regional Jail in Virginia reported 68 deaths, sparking a federal investigation that found “The jail’s practice of subjecting prisoners with serious mental illness to prolonged periods of restrictive housing … shows deliberate indifference to their health and safety.”

‘A persistent crisis of deaths in custody’

Without official data on in-custody deaths, reporters and researchers have had to compile their own estimates. Sinyangwe’s Mapping Police Violence project, for instance, has compiled records on more than 7,600 police killings since January 1, 2013. In 2016, reporters at HuffPost tracked down public records to create a database of more than 800 jail deaths in the 12 months following Sandra Bland’s July 2015 death in a Texas jail. In May 2019, The Frontier gathered records on 215 jail deaths in Oklahoma between 2009 and 2018. In March, reporters for WBUR in Boston found that 195 people died in the custody of Massachusetts sheriffs between 2008 and 2018; at least 37 of those deaths were never reported to the federal government. All of these estimates may understate the true number of in-custody deaths.

A lack of official data has similarly hindered efforts to track the spread of COVID-19 in prisons, jails, and detention centers. Since March, at least 605 prisoners in the United States have died from COVID-19, though gaps in testing mean the real figure may be much higher. The number of jail deaths from the disease is even murkier; the Covid-19 Behind Bars Data Project at the University of California, Los Angeles, has only collected data on 33 of the country’s nearly 3,000 jail jurisdictions, recording 15 deaths.

According to Sinyangwe, these independent efforts show how comprehensive data can help advocates and policymakers identify problems and advance reforms. 

“Data changes conversations about what works and what doesn’t,” he said. “St. Louis, Oklahoma City, and Orlando, Florida, have the highest rates of police killings, based on five years of data that the federal government would not make available, but now they’ve changed policies because of the data we collected. Data has allowed the introduction of an evidence basis in a field that lacked all evidence moving into 2014.”

In January, two Democratic members of Congress, Jerrold Nadler of New York and Karen Bass  of California, wrote a letter to the DOJ’s inspector general calling for a new investigation into the department’s failure to implement the DCRA. “In the absence [of] action by the Department, the United States continues to face a persistent crisis of deaths in custody, the true scope of which remains unknown,” the letter said.

“Everyone should be shocked and outraged on where we are with respect to this law,” Bennett said. “It’s not like the United States does not know how to count. If this were impacting another community, a community other than young Black men, there would be data, we would be working to find solutions, and this would be the most minimal thing we did.”