Get Informed

Subscribe to our newsletters for regular updates, analysis and context straight to your email.

Close Newsletter Signup

DOJ Admits It Has No Idea How Many People Die in Law Enforcement Custody

Thousands of deaths in jails, prisons, and police custody have gone uncounted in recent years. Now the DOJ is calling for changes to federal law.

Emiliano Bar via Unsplash

DOJ Admits It Has No Idea How Many People Die in Law Enforcement Custody

Thousands of deaths in jails, prisons, and police custody have gone uncounted in recent years. Now the DOJ is calling for changes to federal law.


The federal government has failed to count thousands of deaths in law enforcement custody over the past three years, the Department of Justice conceded in a report published last week.

The report, authored by the Office of the Attorney General, offers an unprecedented accounting of the government’s neglect in tracking deaths in U.S. prisons and jails and at the hands of police, as required under the Death in Custody Reporting Act. Since October 2019, the DOJ has missed at least 18 percent of all deaths in state prisons, 39 percent of deaths in local jails, and between 62 and 71 percent of deaths in police custody, according to the report, which compared the number of deaths reported under DCRA with publicly available sources of information.

In total, more than 5,000 deaths in the criminal legal system have gone uncounted over the past nearly three years. Fifteen states failed to report any arrest-related deaths in that period, and seven states failed to report any deaths in local jails.

“This underreporting is widespread, and not the result of a small number of lagging or uncooperative states,” the report says. The DOJ claims that without changes to DCRA, it will be unable to collect “accurate and complete information” on deaths in custody.

The findings seem likely to bolster criticism from congressional lawmakers who have accused the DOJ of “preventable” failures in implementing the law. On Tuesday, the Senate Permanent Subcommittee on Investigations is set to hold a hearing on Capitol Hill that will include testimony from the DOJ official in charge of collecting in-custody death data, as well as two witnesses whose family members died in custody in Georgia and Louisiana.

Shortly before Tuesday’s hearing, the subcommittee—led by Sen. Jon Ossoff, a Georgia Democrat—released its own findings on DCRA implementation, documenting nearly 1,000 deaths in custody that went uncounted in the 2021 fiscal year alone. The subcommittee report found that 70 percent of deaths reported to the DOJ were missing at least one piece of information required by DCRA. In roughly 40 percent of cases, no cause of death was reported.

The DOJ report attributes its failure to track deaths under DCRA to “unintended consequences” of the legislation, which “have degraded and hindered the Department’s ability to produce complete and accurate information.”

Congress passed DCRA in December 2014, prompted in part by the high-profile police killings of Michael Brown and Eric Garner. The legislation requires states and federal law enforcement agencies to report all deaths that occur in prisons, jails, and immigration detention facilities, as well as all fatalities that occur during police interactions with civilians.

States that fail to report data to the federal government risk losing up to 10 percent of their federal funding under the Edward Byrne Memorial Justice Assistance Grant (JAG) Program. Byrne JAG funds are the primary source of federal money for nearly every part of a state’s criminal legal system, including police, prosecutors, public defenders, jails, prisons, and probation and parole systems.

Federal efforts to collect in-custody death data have yielded mixed results over the past two decades. A voluntary data collection program by the DOJ’s Bureau of Justice Statistics (BJS) received nearly 100-percent response rates from state prisons and local jails, documenting more than 85,000 deaths between 2000 and 2019, when the program ended. In contrast, a BJS effort to track arrest-related deaths by relying on reports from state governments was abandoned in 2014 after an internal study found that states were failing to report roughly half of all arrest-related deaths.

In 2016, the BJS announced plans to improve reporting by collecting data directly from state and local law enforcement agencies. The BJS was also supposed to supplement the data reported by the agencies using open-source methods, such as databases like Mapping Police Violence. However, as The Appeal reported in 2020, the decision to tie in-custody death data to federal grant funding under the 2014 DCRA unintentionally scuttled this plan. Because the BJS can only collect data for research purposes, DOJ lawyers concluded that the agency could no longer be involved in tracking in-custody deaths. Instead, the responsibility would fall on the Bureau of Justice Assistance, which oversees the Byrne JAG Program but has little experience managing large-scale data collections.

Beset by internal debates over how to best implement the legislation, the DOJ failed to collect any data from states under DCRA for nearly five years after the law went into effect. A proposal released in December 2016 relied on a combination of centralized state-level reporting of death data and open-source data collection. But Donald Trump’s DOJ quickly scrapped this plan, replacing it with a stripped-down version that relied exclusively on quarterly reports from state officials, who were expected to collect data on all three categories of reportable deaths. The new plan also eliminated the voluntary data collection program for prisons and jails under BJS, meaning that states—rather than individual local agencies—would be solely responsible for compiling statistics.

According to the DOJ’s report last week, this reliance on state-level reporting meant that the DOJ had no choice but to replace a program that had nearly a 100-percent response rate with one that relied on state officials “using varied strategies [that] collectively have proven to be ineffective in producing complete and accurate information.“

In a DOJ survey published in the report, state officials responsible for collecting mortality data under DCRA reported widespread noncompliance among local jails and law enforcement agencies. Two-thirds of surveyed officials said that they lacked any leverage to force uncooperative local agencies to comply. Earlier this year, a review of state policies on in-custody death reporting by The Appeal found that just 15 states have laws that require correctional facilities to report in-custody deaths, leaving them with little recourse to compel agencies to comply with DCRA standards.

Because of these systemic failures, the DOJ says that it would be “unfair” to penalize states for failing to report deaths as required by DCRA. “The penalty could be applied to a state, even when that state may be fully reporting to BJA what it has received from local agencies and where the lack of reporting is occurring only at the local level,” the report says.

The absence of official data has left journalists and researchers to rely on open-source data to fill critical gaps in our understanding of the extent of deaths in law enforcement custody. A 2020 investigation by Reuters documented more than 7,500 deaths in 500 U.S. jails between 2008 and 2019, finding a 35-percent increase in death rates over the past decade. A report earlier this year by PennLive revealed that officials in Pennsylvania jails were routinely failing to report deaths that occurred after people in their custody were hospitalized.

With no official statistics on deaths in prison and jails since the beginning of the pandemic, nonprofit organizations have had to lead the way in tracking the impact of the COVID-19 pandemic on incarcerated people. Figures collected by the UCLA Law COVID Behind Bars Data Project using publicly available information suggest that at least 3,000 incarcerated people have died from COVID since the start of the pandemic. But this figure may be missing thousands of additional deaths, as there is no consistent standard for reporting COVID-related fatalities in custody.

DCRA’s failures have also left the public with no official data on the number of people killed by police each year. Although data collected by Mapping Police Violence using open-source methods finds that police have killed at least 8,224 people since DCRA became law in December 2014, this figure is almost certainly a substantial undercount.

“[Journalists and nonprofits] have stepped in to attempt the data collection that DOJ is statutorily obligated and best situated to do, as DOJ has the resources, expertise, and tools to facilitate compliance and conduct cross-jurisdictional data analysis,” the Senate report says.

To correct these issues, the DOJ report proposes reforms to DCRA that would allow the DOJ to collect data directly from law enforcement and correctional agencies. Funding would also be provided to help state and local agencies improve their data collection practices, and penalties for noncompliance would be made mandatory. In the meantime, however, the DOJ says it intends to leave the public in the dark.

“DOJ is not required to—and has no specific plans to—publish any state and local custodial death information for FY 2020, FY 2021, or beyond,” the Senate report says.

Support The Appeal

If you valued this article, please help us produce more journalism like this by making a contribution today.