Harvey Hill wouldn't leave John Finnegan's front yard. He stood in the pouring rain, laughing at the sky, alarming his former boss' wife. Finnegan dialed 911.
"He needs a mental evaluation," the landscaper recalls telling the arriving officer. Instead, Hill was charged with trespassing and jailed on suspicion of a misdemeanor offense that could bring a $500 fine.
It was a death sentence.
The next day, May 6, 2018, Hill's condition worsened. He flew into a rage at the Madison County Detention Center in Canton, Mississippi, throwing a checkerboard and striking a guard with a lunch tray.
Three guards tackled the 36-year-old, pepper sprayed him and kicked him repeatedly in the head. After handcuffing him, two guards slammed Hill into a concrete wall, previously unpublished jail surveillance video shows. They led him to a shower, away from the cameras, and beat him again, still handcuffed, a state investigation found. The guards said Hill was combative, exhibiting surprising strength that required force.
Video showed Hill writhing in pain in the infirmary, where he was assessed by a licensed practical nurse but not given medication. Mississippi law dictates that a doctor or higher credentialed nurse make decisions on medical interventions. But Hill was sent straight to an isolation cell, where a guard pinned him to the floor, removed his handcuffs, and left him lying on the cement. Hill crawled to the toilet. Then he stopped moving.
No one checked him for 46 minutes. When they did, he didn't have a pulse. Within hours, he was dead. And he had a lot of company.
Hill's is one of 7,571 inmate deaths Reuters documented in an unprecedented examination of mortality in more than 500 US jails from 2008 to 2019. Death rates have soared in those lockups, rising 35% over the decade ending last year. Casualties like Hill are typical: held on minor charges and dying without ever getting their day in court. At least two-thirds of the dead inmates identified by Reuters, 4,998 people, were never convicted of the charges on which they were being held.
Unlike state and federal prisons, which hold people convicted of serious crimes, jails are locally run lockups meant to detain people awaiting arraignment or trial, or those serving short sentences. The toll of jail inmates who die without a case resolution subverts a fundamental tenet of the US criminal justice system: innocent until proven guilty.
"A lot of people are dying and they've never been sentenced, and that's obviously a huge problem," said Nils Melzer, the United Nations' special rapporteur on torture and other inhuman punishment, after reviewing the Reuters findings. "You have to provide due process in all of these cases, you have to provide humane detention conditions in all of these cases and you have to provide medical care in all of these cases."
The US Constitution grants inmates core rights, but those provisions are hard to enforce. The Fourteenth Amendment guarantees fair treatment to pre-trial detainees, but "fair" is open to interpretation by judges and juries. The US Supreme Court has ruled that the Eighth Amendment's ban on cruel punishment forbids "deliberate indifference to serious medical needs of prisoners," but proving deliberate negligence is difficult. The Sixth Amendment assures speedy trials, but does not define speedy.
The Reuters analysis revealed a confluence of factors that can turn short jail stays into death sentences. Many jails are not subject to any enforceable standards for their operation or the healthcare they provide. They typically get little if any oversight. And bail requirements trap poorer inmates in pre-trial detention for long periods. Meanwhile, inmate populations have grown sicker, more damaged by mental illness and plagued by addictions.
The 7,571 deaths identified by Reuters reflect those stresses. Most succumbed to illness, sometimes wanting for quality healthcare. More than 2,000 took their own lives amid mental breakdowns, including some 1,500 awaiting trial or indictment. A growing number – more than 1 in 10 last year – died from the acute effects of drugs and alcohol. Nearly 300 died after languishing behind bars, unconvicted, for a year or more.
As with much of the US criminal justice system, the toll behind bars falls disproportionately on Black Americans, such as Hill. White inmates accounted for roughly half the fatalities. African Americans accounted for at least 28%, more than twice their share of the US population, a disparity on par with the high incarceration rate of Blacks. Reuters was not able to identify the race of 9% of inmates who died.
Jail deaths typically draw attention locally but escape scrutiny from outside authorities, a gap in oversight that points to a national problem: America's system for counting and monitoring jail deaths is broken.
A BROKEN SYSTEM OF FEDERAL OVERSIGHT
America's 3,000-plus jails are typically run by county sheriffs or local police. They often are under-equipped and understaffed, starved for funds by local officials who see them as budgetary burdens. A rising share have contracted their healthcare to private companies.
Yet there are no enforceable national standards to ensure jails meet constitutional requirements for inmate health and safety. Only 28 states have adopted their own standards to fill the gap. And much of the oversight that does exist is limited by a curtain of secrecy.
The Justice Department's Bureau of Justice Statistics has collected inmate mortality data for two decades – but statistics for individual jails are withheld from the public, government officials and oversight agencies under a 1984 law limiting the release of BJS data. Agency officials say that discretion is critical because it encourages sheriffs and police to report their deaths data each year.
The secrecy has a cost: Local policy makers can't learn if their jails' death rates are higher than those in similar communities. Groups that advocate for inmates' rights can't get jail-by-jail mortality data to support court cases. The Justice Department's own lawyers, charged with taking legal action when corrections facilities violate constitutional standards, can't readily identify jails where high death counts warrant federal investigation.
"If there's a high death rate, that means there's a problem," said Julie Abbate, former deputy chief of the Justice Department's Special Litigation Section, which enforces civil rights in jails. Publicizing those rates "would make it a lot harder to hide a bad jail."
The Justice Department does issue broad statistical reports on statewide or national trends. But even those fatality numbers don't always tell the full story.
Some jails fail to inform BJS of deaths. Some report them inaccurately, listing homicides or suicides as accidents or illnesses, Reuters found. Justice Department consultant Steve Martin, who has inspected more than 500 US prisons and jails, said that in all the cases he's investigated, he recalls only one homicide being reported accurately. The others were categorized as "medical, respiratory failure, or whatever," he said.
Other jails find other ways to keep deaths off the books, such as "releasing" inmates who have been hospitalized in grave condition, perhaps from a suicide attempt or a medical crisis, so they're not on the jail's roster when they die. Sheriffs sometimes characterize these as "compassionate releases" that allow inmates' families a chance to spend their final hours together without law enforcement supervision.
In all, Reuters identified at least 59 cases across 39 jails in which inmate deaths were not reported to government agencies or included in tallies provided to the news organization.
The Justice Department has grown more secretive about the fatality data under the Trump administration. While BJS never has released jail-by-jail mortality figures, it traditionally has published aggregated statistics every two years or so. The 2016 report wasn't issued until this year.
And, a Justice spokesman said, there are "no plans" to issue any future reports containing even aggregated data on inmate deaths in jails or prisons.
The report delays are "an outrage," said Representative Bobby Scott, a Virginia Democrat who co-authored the original reporting law in 2000 with a Republican colleague. Scott said secrecy was never the goal. He co-authored a 2014 update, which restricts federal grant money when jails don't report deaths and shifts data collection to a different Justice Department agency that would not be restricted from releasing jail-by-jail data. The updated law has yet to be implemented.
"The whole point," Scott said, "is we suspect a lot of the deaths are preventable with certain protocols – better suicide protocols, better healthcare, better guard-to-prisoner ratios. You've got to have information at the jail level. You have no way of really targeting corrective action if you don't."
Because the government won't release jail-by-jail death data, Reuters compiled its own. The news organization tracked jail deaths over the dozen years from 2008 to 2019 to create the largest such database outside of the Justice Department. Reporters filed more than 1,500 records requests to obtain information about deaths in 523 US jails – every jail with an average population of 750 or more inmates, and the 10 largest jails or jail systems in nearly every state. Together, those jails hold an average of some 450,000 inmates a day, or about three out of every five nationwide.
"You've got to have information at the jail level. You have no way of really targeting corrective action if you don't."
One finding: Since the last Justice Department report, for 2016, the death rate in big jails has continued to climb, leaving it up 8% in 2019, the highest point in the 12-year period of 2008-2019 examined by Reuters. In that time, the suicide rate declined as many facilities launched suicide awareness and response initiatives. But the death rate from drug and alcohol overdoses rose about 72% amid the opioid epidemic.
The data also reveals scores of big jails with high death tolls, including two dozen with death rates double the national average.
Such data "would have actually been very helpful for enforcement purposes," said Jonathan Smith, who ran the Justice Department's Special Litigation Section from 2010 to 2015.
RARE SCRUTINY, REFORM
Detailed insight into jail deaths can save lives.
In 2016, the Justice Department began investigating the Hampton Roads Regional Jail in Portsmouth, Virginia, after state Attorney General Mark Herring and local civil rights groups called for a probe following several inmate deaths. Reuters found the jail, which serves five jurisdictions, averaged 3.5 deaths per thousand inmates over the years 2009 to 2019, more than double the national average of 1.5 deaths.
In December 2018, the Justice Department said the 900-bed jail violated inmates' rights by failing to provide adequate medical and mental healthcare. The regional authority that manages the jail agreed to a "consent decree," enforced by a federal judge, to ensure improved treatment of prisoners.
Inmate deaths dropped after the agreement, which required increased staffing, better training and enhanced medical services. The jail reported two fatalities in 2019 and one through this May, down from an average of five a year in the prior four years.
That was one of the Justice Department's last jail investigations. From 2008 to 2018, the department opened 19 investigations into jails, three during President Trump's tenure.
Yet since 2018, it hasn't opened any. A memo circulated in November 2018 by then-Attorney General Jeff Sessions put hurdles in the way of entering consent decrees for overhauling jails. In a telephone interview, Sessions told Reuters the policy he set forth adhered to Supreme Court standards on when consent decrees could be entered, allowing them when "appropriate" and "justified."
In the absence of federal oversight, states have a patchwork of guidelines.
Seventeen states have no rules or oversight mechanisms for local jails, according to Reuters research and a pending study by Michele Deitch, a corrections specialist at the Lyndon B. Johnson School of Public Affairs at the University of Texas. In five other low-population states, all detention facilities are run by state corrections agencies. The other 28 have some form of standards, such as assessing inmates' health on arrival or checking on suicidal inmates at prescribed intervals. Yet those standards often are minimal, and in at least six of the states, the agencies that write them lack enforcement power or the authority to refer substandard jails for investigation.
Deitch said these gaps make comprehensive nationwide statistics all the more important. "You can't have good policy without good data," she said. "Data tells us what is going right and what's going wrong."
A DEATH IN MISSISSIPPI
Without jail-by-jail mortality data, even jails with extraordinary death rates can escape official intervention for years, and local officials can remain blind to the seriousness of problems their facilities face. One example is the Marion County Jail in Indiana, a decrepit 65-year-old facility nicknamed "The Fossil" within the sheriff's department.
Overfilled and understaffed, the Marion County jail had at least 45 deaths from 2009-2019. Yet local officials rejected pleas from two consecutive sheriffs for additional funding to bolster staffing and build a new facility.
Reuters found that the jail is among the two dozen with an average death rate, 3.5 deaths per 1,000 inmates, at least double the national average from 2009 to 2019. And its record was troubling on one of the most challenging problems plaguing jails: suicide, which accounted for more than a quarter of all US jail deaths.
Thomas Shane Miles, a married father of two, struggled for years with mental illness and opioid addiction when he was arrested in 2016 on a misdemeanor drug possession warrant. On his second day in jail, he flung himself down a stairway and swallowed the contents of a chemical ice pack.
Put on suicide watch, Miles was given a "suicide smock" – a heavy hospital-style gown closed with Velcro – and placed in a monitored cell. The jail's policies, as well as American Bar Association guidelines, dictate that suicidal inmates be monitored continuously.
On Day 6, Miles was given a jail uniform for a hearing and escorted down an underground hallway to a holding cell below the adjacent court building – a cell with no video monitor or clear sightlines for deputies. Left alone, he tore a strip of cloth from the collar, looped it over a door hinge and hung himself. He was found unconscious 30 minutes after entering the cell. An internal inquiry said the supervising officer logged his rounds after the fact, leaving it unclear when Miles was checked.
In a wrongful death suit that settled this September, Miles' family argued that despite being identified as a suicide risk, he was given the means and opportunity to kill himself. The sheriff's office denied misconduct and said it admitted no wrongdoing in the settlement; details were not disclosed.
Miles' suicide was the jail's seventh in just under 15 months. The Fossil's suicide rate ranked it among the top 20 jails in the Reuters study.
In 2016, the sheriff called the suicide problem an "epidemic," but county officials denied requests for more funding. While the county knew it had a suicide problem, there was no way to know how it compared. Like all other officials, Marion County's leaders had no access to the Justice Department figures.
The sheriff's jail-management mission often "came in second" in a budget system that pits it against the Indianapolis police department's law enforcement duties, said Frank Mascari, who sits on the City-County Council. "We knew there were some deaths" at the jail, he said, "but we didn't have the statistics" to know the rates were extraordinary.
From 2015 to 2017, the sheriff's budget grew just over 1% a year, audit figures show. The inmate population rose 12% in that time, due to a rise in arrests and to state legislation dictating that some low-level felons serve their sentences in county jails, not state prisons.
"A lot of people are dying and they've never been sentenced, and that's obviously a huge problem."
The sheriff launched suicide-prevention efforts, hired social workers and trained deputies in spotting suicide warnings. From 2017 to 2019, the number of suicides dropped to two a year, but staffing remained critically low as deputies routinely left for better paying jail jobs in nearby suburbs.
Jail deaths remained stubbornly high despite the decline in suicides, reaching six last year, the heaviest toll in more than a decade, driven in part by drug and alcohol overdoses. Still, there has been no state or federal intervention.
In July 2018, Kyra Warner, 30, went quiet about 90 minutes after arriving at the jail. As her limbs twitched, cellmates called for help, telling nurses and deputies that Warner said she had been using methamphetamine and anti-anxiety drug Xanax.
Jail video shows Warner unable to walk on her own as deputies moved her to a monitored isolation cell, where they left her on the floor, still twitching. She lay unresponsive as they checked her periodically over two hours – until medical staff found no pulse. She died of an accidental overdose.
"The officers that are watching aren't medically trained," said Rich Waples, a lawyer handling the family's ongoing wrongful death lawsuit against the sheriff and Wellpath, the company providing the jail's healthcare. "If she'd gotten prompt care, they could have reversed the effects of those drugs."
Jail officials denied wrongdoing and noted in their response to the suit that deputies checked on Warner numerous times, but added they are not medical professionals. Wellpath, also contesting the ongoing suit, denied any misconduct.
"We're not built to be the largest mental health hospital in the state," said Colonel James Martin, who oversees the jail. "We're not built to be the largest detox facility in the state." Yet the jail has "more detox beds than any single hospital in the state."
The jail's shortcomings have been documented, including a county-commissioned review in 2016 that found the Fossil "antiquated," with inadequate staffing and design flaws that severely hamper inmate monitoring. In 2018, after another independent study highlighted the jail's challenges, the county approved a new $580 million criminal justice complex, with dedicated facilities to treat mental illness and substance abuse. In 2022, the Fossil will be history.
Another flaw in the US system for monitoring jail fatalities is misleading disclosure. The John E. Polk Correctional Facility in Florida's Seminole County reported one death to the Justice Department in 2019. But at least one other death at the jail was not reported in its official filings.
On June 2, 2019, Thomas Harry Brill, 56, was found hanging by a bed sheet in his cell. Staff tried but failed to resuscitate him, the jail said. He was pronounced dead at a nearby hospital. Sheriff's spokeswoman Kim Cannaday said he "was released out of our custody" before he died. "Therefore, it would not technically be considered an in-custody death."
Brill's sister, Tracy, was shocked to learn his death was excluded from the jail's official count. "They're trying to avoid responsibility," she told Reuters. "They're playing with the numbers. That's just wrong."
Brill graduated from Eastern Michigan University with a mathematics degree and lived on a sailboat for years, she said. He had been wrestling with mental illness when he flew from his home in San Diego to look at a boat in Florida. Out of money, he was found in a stolen car and arrested, but couldn't afford bail. He died unconvicted of the charge. "He needed $500 to get out," she said. "It was an awful, ridiculous waste that he died."
A DEATH IN MISSISSIPPI
The Reuters death database also points to another benefit of collecting and publishing jail mortality rates: It can identify an unusual number of fatalities at jails that typically have few. One is Mississippi's Madison County Detention Center, where Harvey Hill died after being beaten by guards.
The jail had occasional deaths, and in several years reported none. Yet in 2018, it had two deaths, including an inmate who died of complications from an ectopic pregnancy. Few other jails its size had multiple deaths that year.
Hill grew up in the poorest county in the poorest state in America. West, his town of 185 people, is intersected by a four-lane highway in Mississippi's rural Holmes County. He did landscaping work an hour's drive south in Canton, a city of 13,000 in the state's wealthiest county, where 19th Century antebellum shophouses packed with antiques line a postcard-perfect downtown square.
When he was 18, Hill was arrested on charges of sexual battery and robbery. He pleaded guilty and served 14 years in prison. Friends and family say he began piecing together his life after his 2015 release, taking landscaping jobs with business owner Finnegan. "He was an incredible worker," said Finnegan.
Through the winter of 2017 into the spring, Hill showed signs of mental illness, displayed flashes of paranoia and complained of insomnia, said Finnegan. After he let him go in 2018, Hill started showing up at his home, claiming his old boss owed him millions of dollars. "Harvey, if I had taken your millions, I wouldn't be landscaping. I would be on an island," Finnegan recounts telling him.
Hill kept returning. In May 2018, Finnegan called the Madison Police Department. If he wanted Hill removed, he had to press charges, Finnegan said he was told, so he did. "That's not something I really wanted to do," he said. "Harvey needed to be in a mental hospital."
At the station, Finnegan told the officer he'd drop the charges and take Hill to a mental health facility if they could find a room. Instead Hill was booked into Madison County's jail that Friday morning. "I'll pick you up on Monday," said Finnegan. "And we'll get you some help."
The Madison Police Department said there were "no remarkable or extraordinary events related to his arrest." Mississippi has no standards or oversight for jails.
In their response to a family lawsuit, the guards said their actions were proper under jail policy. Michael Wolf, an attorney for one of the guards, James Ingram, told Reuters that Hill bit and then tried to head butt an officer, "and continued to resist and exhibited unusual strength. The control techniques were consistent with the County use of force continuum." The other guard named in litigation, James Buford, declined to comment.
The family believes the force was unjustified. "Harvey Hill was in handcuffs and beaten to death," said Derek Sells, a lawyer representing the family. "Someone needs to be held responsible."
Hill's death was one of four Reuters identified at the jail over the 12-year period. After he died, the jail filled out a form for the BJS with Hill's name and details including his race, age and charges. The box for "homicide" was left unchecked. Two years later, no "cause of death" has been sent to the BJS, the jail said, citing an ongoing investigation by the Mississippi Bureau of Investigation. No one has been charged.
The family said the jail lied about his death. "They just told us that Harvey had passed and he had had a heart attack," said Katrina Nettles, his younger sister. The jail did not respond to requests for comment. Its medical contractor, Quality Correctional Health Care, and the nurse who treated Hill denied wrongdoing in litigation.
An autopsy ruled Hill's death a homicide, however. The report showed that abrasions speckled his head and chest. Severe internal bleeding swelled his neck. His liver had been lacerated.
The state medical examiner, citing a backlog, didn't release the findings to the family until this June, 25 months after he died and 13 months after the statute of limitations had expired for litigation involving assault. The family filed its ongoing lawsuit last February, before receiving the autopsy.
Told by Reuters of the autopsy's grim findings, Finnegan bent forward and choked back tears. "God Almighty," he said, dragging a hand over his face. "Harvey was a friend."
How Reuters tracked and analysed deaths in America's largest jails
By GRANT SMITH and PETER EISLER
The Reuters examination of deaths in US local jails represents the largest collection and publication of inmate mortality data undertaken outside the federal government.
The news organization filed more than 1,500 public records requests to collect data on inmate populations and deaths from more than 500 local jails. That universe includes the 10 largest jails in each state, as well as any jail in the country with an average daily population of 750 or more inmates.
In all, the Reuters data captures about 60% of the total inmate population in the nation's 3,000-plus jails. Similarly, Reuters data accounts for about 60% of all inmate deaths nationwide, based on the latest national data collected by the US Bureau of Justice Statistics. BJS issues national-level and state-level data on jail deaths, but no statistics for individual jails. The Reuters investigation is the first to provide individual jail death data on a national scale.
Reuters calculated annual death rates at more than 500 jails by dividing the total number of deaths in a given year by the average daily population in the same year – the same formula used by BJS and other experts in criminal justice statistics.
States and local law enforcement agencies have varying definitions for what constitutes a jail death. Reuters counted all deaths that occurred in a jail, as well as deaths of inmates who were hospitalized for injuries or conditions incurred at the jail. When inmates are in life-threatening condition, some jails release them and do not count their subsequent death as an inmate fatality. Reuters, like many jurisdictions across the country, included those cases in its tally of jail deaths.
Reuters received responses from more than 95% of the jurisdictions from which it sought public records. Not all jails were able to provide accurate data on inmate populations for every year covered by the analysis, particularly the earlier years. Data was not available on race for about 9% of inmates who died and for conviction status for about 17% of fatalities. In cases where data was available for adjacent years, Reuters used that information to estimate inmate populations for the years in which no data was provided – a statistical method also used by BJS.
Reuters also used court records and news accounts to identify deaths that were not documented in jails' responses and, in many other cases, to augment information jails did provide. Several dozen unreported deaths were identified in this manner and added to the Reuters tally. Court records and other official records, such as autopsy reports, also were used when available to fill in data that some jails declined to provide, such as cause of death or age.
Reuters also collected information on how healthcare services are provided in each jail, identifying those that relied on private companies to manage and deliver that care. Reuters only considered jails to have privatized or contracted care if they relied on a company to manage and staff the facility's entire healthcare operation. If a jail contracted with individual practitioners for discreet medical services or hired staffing agencies to provide clinicians, Reuters still considered that care to be publicly managed, just as it would if the jail was running its own healthcare operation or relying on a public health agency.
The data captures jails in 44 states plus the District of Columbia. It does not include six other states – five where all detention facilities are managed by unified state corrections agencies (Connecticut, Delaware, Hawaii, Rhode Island and Vermont), and Alaska, which uses a hybrid model that also relies largely on a network of state-run facilities.