In late 2012, I issued a Freedom of Information Act request for records pertaining to the deaths of inmates in the Federal Bureau of Prisons’ privatized, immigrant-only facilities. For years I’d heard stories about neglect inside these prisons, but access to the facilities is limited and thus the accusations have been hard to confirm. For more than a year, the BOP delayed answering my request, until 2014 when I filed a federal lawsuit to compel the BOP to release the records. In 2015, BOP released more than 9,000 pages of healthcare records that it had received from its contractors. These pages are the medical files for 103 men who died inside our nation’s immigrant-only contract prisons.

Medical doctors reviewed the files and independently rendered judgments on the adequacy of medical care, for each case. In 25 cases, at least two reviewers identified inadequacies that likely contributed to premature death of federal inmates held in contract prisons. Below is a FAQ to guide understanding of this largely unknown, complicated prison system, and the troublingly poor care provided inside it.

What are these prisons?

The federal prison system currently includes 11 prisons called Criminal Alien Requirement facilities. They are distinct from the rest of system for two reasons. First, private companies manage day-to-day life inside, including medical care. Nearly every other Bureau of Prisons facility is operated directly by the BOP and its staff. Second, the prisoners inside all have one thing in common: They lack US citizenship.

The prisons are also distinct from immigration detention centers, where federal authorities hold immigrants pending deportation. Rather, the inmates are people who have been convicted of federal crimes and who also happen to be noncitizen immigrants. They are all men; women are mainstreamed into the regular system.

Why do we have them in the first place?

Many of the men held in these immigrant-only prisons are incarcerated for crimes anyone could commit—drugs, fraud, burglary, etc. But 40 percent of them in 2014 were locked up on immigration-related crimes for which only noncitizens can be convicted. Mostly, they are guilty of a crime called “illegal reentry,” or crossing the border after having already been deported.

For decades, returning to the United States after a previous deportation has been a criminal offense. Yet until the 1990s, the crime was rarely prosecuted. Instead, border crossing was treated as a civil matter; the consequence was deportation, not prosecution. In the late 1980s, Congress began a decade-long process to expand criminal penalties for border crossing, and by 1996, returning after deportation was punishable by years of imprisonment.

In the 1990s, Congress also began allocating unprecedented resources to the southern border, eventually expanding the border patrol more than fivefold. Arrests for reentry began to rise. Between the early 1990s and 2004, border prosecutions rose from fewer than 4,000 a year to 31,000. Then in 2005, George W. Bush’s administration exploded the number.

Bush launched Operation Streamline, a Border Patrol partnership with US attorneys and courts, to ramp up prosecutions not only of serial crossers and immigrants with past criminal convictions, but even of first-time crossers. Any foreigner caught entering the US without permission could be filtered through a criminal proceeding and charged with a misdemeanor. Each consecutive crossing led to a longer sentence. By 2013, under President Obama, criminal prosecutions for border-crossing had reached a record high: 91,000.

Today in Del Rio, Texas, the birthplace of Operation Streamline, men and woman detained by Border Patrol are filtered into courtrooms in groups of 70 or 80, wrists and ankles shackled, with headphones for translation. Within a few hours, nearly all have pleaded to misdemeanor “illegal entry” charges. In separate proceedings, repeat crossers are sentenced to longer felony terms for “illegal reentry.” Many of them, land in the BOP’s private prisons.

Why are these facilities segregated and privatized?

The privatization experiment began in the late 1990s. President Bill Clinton had promised to rein in the size of the federal workforce, but then he signed a broad crime bill that promised to grow an already bloated federal prison system. So, buried deep inside its 1996 congressional budget request, the White House proffered a plan to hire contractors to run several facilities. “The Federal Prison System,” it read, “will expand its capacity and cut costs through privatization.”

In 1997, the BOP issued its first contract for an immigrant-only prison. “Criminal aliens were an appropriate group for housing in privately operated institutions where there are somewhat fewer programs offered to prepare offenders to successfully reintegrate into U.S. Communities,” the BOP explained in a 2014 statement. By 2013, the BOP had contracted out 12 of these prisons operated by three companies, costing taxpayers $625 million a year. Today there are 11 facilities in operation.

Has privatization saved taxpayers money?

Probably not. According to the BOP’s own research and records of internal BOP deliberations, any cost savings the privately run facilities have produced may well have been outstripped by the additional costs of monitoring the companies.

Costs to the BOP may be impacted too by a barrage of management problems that have been besieged the prisons since they began. An early study by the BOP found that the first privately operated BOP prison, a California facility called Taft, faced unusually high rates of inmate unrest and drug use. Researchers found that for nearly three years, the worst performing area of operation at Taft was medical care, which at least once failed to meet the federal government’s baseline requirements.

Between 2008 and 2015, prisoners in the private, immigrant-only facilities staged at least five riots, often as a result of unanswered complaints about medical care. In 2012, a riot turned deadly—a guard was killed by prisoners. In February 2015, inmates at the Willacy facility in Texas’ Rio Grande Valley staged a protest over healthcare that turned into a riot. The prison was left uninhabitable and the BOP ended its contract with its operator, Management & Training Corporation.

Do the prosecutions deter border crossing?

Nobody seems to know about the long-term effect. “The theory was perhaps, if we aggressively prosecuted all of those case, that it would deter persons from coming back into the United States,” said John Klassen, who was the a US attorney in Midland, Texas, from 2003 until 2015. “I have no idea whether it had that effect.”

One man I met, Eloy Flores, offers a hint as to why the prosecutions would fail as a deterrent. Flores was prosecuted for illegal entry twice—first for five days and then for four months—as he made an attempt to return to Maryland, where his three US-citizen children were born.

“You’re not being prosecuted for being a bad person or because you do not have good reasons,” Judge Victor Garcia told Flores, in an audiotape of a November 19, 2013, hearing. Judge Garcia nonetheless sentenced him to four months in prison. “I cannot stop you from coming back. But I can tell you what will happen if you do, and that’s you’re going to be in prison.”

When the Department of Homeland Security’s Office of Inspector General tried to measure the impact of Operation Streamline in 2015, it found that the Border Patrol’s methodology for tracking repeat border crossing was flawed. Flores, who is back in Mexico, told me that prison merely slowed down his inevitable return to his family.

How are noncitizens in federal prisons treated differently from other inmates?

Our investigation found that the federal contract prisons operate under a different and less stringent set of rules than the rest of the BOP. The BOP’s contract solicitations only require the operators to follow a fraction of the BOP’s rules for its regular facilities. Many of the BOP’s policies on mental health and suicide prevention, for example, are either listed as guides or not included at all. As a result, one former BOP official told me, “Mental health care is much more difficult to come by in the contract [prisons].”

The contracts also leave decisions about staffing plans, including medical staffing, largely up to the contractors. BOP’s rules for its own prisons say that night and weekend clinic shifts are to be covered by registered nurses or emergency medical technicians, who are trained to handle emergencies. But the records we reviewed and interviews I conducted show that the for-profit prisons sometimes use less-trained, lower-paid licensed vocational nurses to cover work that RN’s would perform in the BOP prisons. In 19 of the 103 deaths we reviewed, at least one doctor flagged the overextension of LVNs as an inadequacy in care.

Corrections officials I talked to told me that less stringent rules are applied to contract facilities in order to keep costs low. But even when the BOP does require the prison companies to follow specific BOP rules, the facilities have been found repeatedly in violation. Federal government monitoring reports show the prisons failing time and again to follow BOP policies on infectious disease treatment, to which the prisons are contractually obligated.

What was our methodology for analyzing death records from these prisons?

To analyze deaths in the Bureau of Prisons’ (BOP) immigrant-only facilities, reporter Seth Freed Wessler requested from the BOP the full medical files for every inmate who died in custody since the Bureau first opened a contract facility. He also requested the psychiatric records in the case of suicides. He received medical records for 103 cases from 1998 to 2014, out of a total of at least 137 total deaths in all BOP contract facilities during this period, a total of more than 9,000 pages. While the private contractors that run these facilities turn over the medical records to the BOP in cases of inmate death, in some cases the files may not have been complete. We sent each of the 103 files to one of a panel of 23 independent reviewers—17 medical doctors and six psychiatrists, who reviewed the suicides. The reviewers found that 26 case files contained too little information to render judgment.

Of the remaining 77 cases, if the first reviewer found evidence that the medical care aligned with recognized medical standards, we put the case aside. We sent the remainder to a second reviewer. If the second reviewer found evidence of adequate medical care, we put the case aside. In all, at least one medical reviewer found evidence of adequate care in 39 cases.

In 38 of the cases, the two reviewers were unanimous in finding evidence of inadequate medical care. We asked these reviewers a second question: Did the inadequacy contribute to the patient’s death? The reviewers agreed, in 25 cases, that there was evidence of inadequate care and that the care likely contributed to an inmate’s premature death.

In some instances, internal mortality reviews that were contained in the records released confirmed weaknesses in medical care.

How did the Bureau of Prisons respond to these findings?

We contacted the Bureau of Prisons by e-mail and phone a total of 10 times over six weeks requesting an interview. We also sent a detailed list of questions. We received acknowledgement that the questions were received, but never received any other response.