In a first-of-its-kind investigation built on 1,500 public records requests, Reuters documented more than 7,500 inmate deaths from 2008-2019 at more than 500 local jails – every major jail in the U.S. and the 10 largest in each state. Two-thirds of those who died were awaiting trial and presumed innocent. Mortality rates were higher in jails that privatized the management of their medical services using big correctional healthcare contractors. Deaths of women inmates, trapped in male-oriented detention systems, soared to record highs. Thousands perished from preventable causes. Black inmates died in disproportionate numbers.
Congressional leaders and state legislators proposed several reforms to address the revelations of Dying Inside. Our database also was used by local journalists, watchdog groups and civil rights advocates to identify local jails with high inmate death rates and highlight those problems in news reports and independent studies.
In West Virginia and Georgia, Reuters’ data spurred state legislators to draft bills setting new standards for jail oversight. In Mississippi, after Reuters published previously unreported surveillance video and details on the 2018 death of Harvey Hill, a state legislator and veteran Congressman pushed for criminal charges against the jail guards who beat him. “It’s just disheartening to see this happen to another African American male,” said Mississippi lawmaker Chris Bell. The Justice Department is now investigating the death and a decision on whether to issue indictments is expected this year.
In Washington, congressional leaders pushed the U.S. Justice Department to strengthen its tracking of jail deaths. Reuters’ reporting “underscores the need to overhaul our pretrial detention practices,” said House Judiciary Committee Chairman Jerrold Nadler. Senator Richard Blumenthal said Congress should consider restricting DOJ funding if it fails to address systemic reporting problems.
NaphCare Inc, a correctional healthcare company cited in the Reuters investigation, vowed to take “corrective initiatives” to reduce death rates at jails under its care after reading our report.
“The increase in deaths in custody reflected in (Reuters’) statistics – in particular of people of African descent, of women and of people in pretrial detention – is especially alarming, not least because many of these deaths are preventable,” the United Nations’ Human Rights Office said in a statement, urging jails to “reduce over-reliance on pre-trial detention.”
Follow-up stories listed at the bottom of the main project page captured some of this impact: https://www.reuters.com/investigates/section/usa-jails/
Our records’ requests yielded thousands of pages of documents and more than 150,000 data points that had to be keyed in and analyzed. We broke the data into two sets:
* A “jails” database capturing annual data on inmate populations, mortality and healthcare for each of 500-plus jails in our sample. The data, spanning 2008-2019, included yearly headcounts, deaths by type (suicide; homicide; medical/natural; drugs/alcohol; accident; other), and details on any healthcare contractors employed in each year.
* A “deaths” database with details on all of the more than 7,500 deaths we documented. For each death, we captured, when available, the inmate’s name, age, race, gender, date of death, cause of death, length of incarceration and custody status (convicted vs. awaiting trial).
We built both databases in Google Sheets, and we used folders in Google Docs to organize text records for each jail, such as investigation documents, autopsy reports and correspondence. The videos we collected also were archived in Google Docs.
We did our own data analysis, primarily using PostgreSQL and Microsoft Excel. We consulted with independent experts to validate some of our statistical models, which we designed in R-Studio. For example, in comparing death rates at jails with privatized healthcare to those with publicly managed medical services, we developed a statistical model that controlled for the size of the jail population, the size of the county, the county mortality rates and the year. We published sidebars explaining our methodology and naming any experts who reviewed our work.
We also open-sourced our data for journalists, policymakers and other stakeholders, providing downloadable ZIP files containing the “jails” dataset, the “deaths” dataset and record layouts, both nationally and by state. Additionally, we offered easily shareable PDFs with summary data on deaths by jail and state.
What was the hardest part of this project?
Our series is the most in-depth exploration ever published of U.S. jail deaths and the 1,500 records requests behind the reporting posed extraordinary challenges. We focused on 523 jails and sent three rounds of requests over 18 months, each gathering new details. Achieving the project’s nearly 100% response rate required persistence and, at times, pressure of litigation.
To determine our universe of jails, we used the U.S. Justice Department’s Census of Local Jails to identify all facilities with average populations of at least 750 inmates, as well as the 10 largest jails in each state. Then we reviewed every jail’s records policies and wrote tailored requests for each.
Many jails were reluctant to provide the information we requested – records on all inmate fatalities over a 12-year span, including the circumstances of each death and other details. We often had to contact jails repeatedly, cajoling officials for months to gain cooperation.
We ultimately collected more than 150,000 data points; each had to be manually entered, confirmed and fact checked.
As our data came together, we identified specific jails as case studies to illustrate various problems and trends. This required interviews at multiple jails and reviews of lawsuits, official audits and local news coverage.
We reported at more than a dozen locations. To uncover the story of neglect and death in a Savannah jail, three reporters spent six weeks in the city developing sources to corroborate assertions from legal filings and other documents. In Mississippi, we obtained previously unreported jail video and investigation reports to reveal how guards fatally beat Harvey Hill, held for trespassing. We made multiple trips to Georgia’s Cobb County to report the story of Chinedu Efoagui, who died after 512 days behind bars, unconvicted, in a traffic case.
What can others learn from this project?
Until now, reporters have never been able to identify the nation’s deadliest jails or compare mortality rates across facilities. The U.S. Bureau of Justice Statistics has collected data on every death in U.S. jails for more than 20 years, but under an obscure statutory restriction, jail-by-jail mortality data has never been released – not to the public, not to policy makers, not even to Justice Department lawyers who are supposed to take legal action against jails with unconstitutional conditions.
We have replicated the government’s restricted data for more than 500 of the nation’s largest jails and made it available to other journalists. This marks the first time reporters have had access to this sort of data so they can readily examine and compare mortality and health care at their local jails. Those jails, typically run by local sheriffs or police departments, often are among the least transparent components of the U.S. criminal justice system, facing a fraction of the accountability required of state and federal prisons.
In addition to making our data available to other news organizations, we have provided a roadmap for how to conduct these investigations. Our stories, including three pieces detailing our methodology, illustrate how mortality and health care data can be used to assess a jail’s performance. And, just as important, we explain the limits of that data and the strategies jails use to keep deaths off the books or mask their true cause.
Connor Sheets, an investigative reporter for the Alabama Media Group and member of ProPublica’s Local Reporting Network, is among the reporters who have used our data to launch their own stories on local jail mortality.
“Thanks to (Reuters) for their new data on jail deaths,” Sheets wrote on Twitter. “What a great resource and important dataset.”