Member? Log in...

Join or renew today

Resources: Articles

Investigating mental illness in America’s jails Date: 10/01/18

Gary Harki

After reporting on a young man with schizophrenia and bipolar disorder who died in jail, Gary Harki, an investigative reporter for The Virginian-Pilot, used his O’Brien Fellowship in Public Service Journalism at Marquette University to lead an investigation into how people with mental illness are treated in jails around the country.

Through numerous FOIAs and phone calls, Harki, and the students working with him, learned that just eight states that kept data on people with mental illness who died in jail. There were many states that tracked jail deaths in general but did not specify if the person was mentally ill.

The result is a tour de force of data visualization, interactives, narrative and a painstakingly built database of information about the 404 people with known mental illness who have died in U.S. jails since 2010. Harki and the former students answered questions about the project from AHCJ's Emily Willingham.

What brought you to this story to begin with?

Gary Harki (investigative reporter for The Virginian-Pilot, fellow in the O’Brien Fellowship in Public Service Journalism at Marquette University):

In August 2015 a young man named Jamycheal Mitchell died in the Hampton Roads Regional Jail in Portsmouth, Va. I was switching over to The Pilot's newly formed investigative team at the time and started digging into what happened to him.

Mitchell had stolen about $5 worth of snacks – a zebra cake and a Mountain Dew – back in April from a 7-Eleven that he thought was owned by his father. He had schizophrenia and bipolar disorder and, once in jail, he was placed alone in a cell, where his condition deteriorated.

He died there, and to me it was just appalling. The water in his cell had been turned off, and a court order sending him to a mental hospital was lost in the mail. A second court order was tucked into a drawer and forgotten until after he died.

So it really started there – just asking why and how did this happen.

I started writing a lot about the intersection of Virginia’s mental health care and criminal justice systems.

Then I applied for the O'Brien Fellowship in Public Service Journalism at Marquette University and was accepted for the 2017–18 school year. That gave me the time and resources to look at the issue from a broader perspective. It also gave me the help of three Marquette students, seniors Alexandria Bursiek and Rebecca Carballo and grad student Diana Dombrowski. Without their contributions, I couldn’t have built the database. It took an enormous amount of time.

How did you navigate the trauma of the people themselves and their consent (or in the case of their death, the next-of-kin) consent to be included in the story. How did you deal with the multiple sensitivities around that?

Gary Harki:

Where we quote relatives in the stories, we usually approached families through their lawyers. In many cases they had spoken to the media before. In some cases we called victims’ relatives directly. We talked a lot about how to do this over the year.

Reporting on victims of trauma or family members who have lost their loved ones is always difficult. You want to let them know you’d like to talk to them and why, but you don’t want to do it in a way that pressures them or causes more grief. I always approach people in these in the same way: I tell them who I am and why I want to talk to them, and I try to be genuine – I am here to do a job. I am sympathetic, but I don’t try to be their best friend. I tell them I can't make any promises, that all I can do is write about what happened so that the public knows. I usually say something like, “Obviously I think it’s important for the public to know your story. I hope you think so too.” I’ve found people appreciate the directness and honesty in those situations, even if they decide not to talk to me.

And I think a lot of people come to see themselves how I see them - as having a chance to push the public, politicians and law enforcement to address or at least acknowledge a terrible situation that they'd rather ignore.

One thing I have found with people suffering from the loss of someone in this type of situation – they usually want to talk about what happened.

First, they want the public to know that they cared about the victim. In many of these cases, jail officials will, through statements, put some of the blame back on the families – implying, if not outright stating, that if they had taken better care of their loved one, he or she would not have gone to jail or would have been in better shape and would not have died. The families want people to know that’s not true.

Second, they want people to know that what happened to their loved one could happen to others.

Third, they want some form of justice, and they hope that telling their story could prevent this type of thing from happening again and hold people in the mental health and criminal justice systems accountable for these deaths. Sometimes that happens, sometimes not. Again, I don’t make promises.

It is for similar reasons that Jan Green, who I wrote about in the second story in the series, wanted to talk to us. Rebecca Carballo and I traveled to Minnesota to interview her. I know it was extremely painful for – she later told me she had nightmares for three days after the interview – but she really wanted to do it.

Jan lived through conditions that killed so many inmates listed in the database.

She wanted people to know that her family loved her and was trying to help her during her years-long ordeal. She wanted people to know this could happen to them or someone they loved. And she wanted to help other people by telling her story.

For the database itself we relied on a lot of court records and news stories. Most of the 404 deaths we tracked were first spotted in a news report. Gathering the data consisted of looking at what had already been written in news accounts and court filings. There was also a database in Texas that had names and mental health diagnoses that was very helpful.

We did not try to contact each of the families of the people counted. I would have liked to do that, but even with the time and resources available, it didn’t seem practical given everything we had to do to make the database viable.

The collection of video and images is striking, horrifying, tragic and heartbreaking. What processes did you go through to obtain them, especially the video sequences you included?

Gary Harki:


I had help tracking down photos from one of my students, Alexandria Bursiek, through part of the summer. She mainly called jails looking for mug shots. We also got photos from attorneys who were suing jails on behalf of families who lost loved one, and in one or two cases from families. Some shots were also on file in photo libraries we had access to.

As for the videos – there are a lot more out there on YouTube than what we were able to use. We took videos found there and credited the news organization that originally posted them. There were really so many deaths in the database where videos had ended up online, thanks to news organizations or lawsuits, that we had no trouble finding them.

Going through those and figuring out what we could use and wanted to use was probably the worst part of working on this project.

The database you built is remarkable. Can you walk us through how you began to build it--maybe some definitions you used to categorize, parameters you set, how you accessed the data?

Gary Harki:

Thank you!

It would not have been possible to build the database without the help of the students that worked with me. It took that much time and person power. I could easily have still been sitting at my desk entering in data by hand had they not been so dedicated.

I decided to build the database because I wanted to capture the scope of the problem in a way that would grab the public's attention. That a high percentage of people in jails had a mental illness wasn't news.

I needed to show outcomes – what jailing people who need mental health care actually looks like.

Anecdotally, I knew the system was killing people in terrible ways because I kept seeing similar stories crop up elsewhere after Mitchell died. That’s why I thought we could count the deaths.

I would loved to have included something like use-of-force incidents involving people with mental illness, but those incidents don't even make the news. And ultimately, we had to create our database using newspaper accounts and court filings.

We started searching for data on deaths of people with mental illness in jails at the state and federal level and didn't find much.

It was Rebecca Carballo's job to make sure we had contacted each state and found out whether it included mental illness indicators in its jail death data. For some states that meant sending out Freedom of Information Act requests. In others, it meant calling several agencies to confirm that they too did not keep the data.


Rebecca Carballo (Recent Marquette University graduate and a reporter for the Charleston Gazette-Mail in West Virginia):

After numerous FOIAs and phone calls, we found there were eight states that kept data on people with mental illness who died in jail. There were many states that tracked jail deaths in general but did not specify if the person was mentally ill.

My search would usually start with the department of corrections. They would then refer me to the medical examiner, state police, attorney general, etc. Eventually, I would find someone who would tell me they had it or no such record existed. If the agency thought they had it, I would usually have to FOIA for it.  Many state agencies said they did not track jail deaths and that we would have to go county by county to find that number. If that was the case, we did not count them because we were looking to see if records were being kept at the state level.

Gary Harki:

Going county-by-county was just not feasible for the time and resources we had. We tried to be as ambitious as possible while still making the project manageable.

After learning that Texas miraculously had a database that included deaths in local jails and indicators of mental illness, I went back and made additional calls to some states just to double-check. I must have spent a day calling various agencies in California because I just couldn’t believe they didn’t have anything.

I made the final determination if a case was included in the database or not. There were three basic criteria:

  1. An indication of mental illness prior to their time in jail custody. This information generally came from a lawsuit or news report. We always wanted a court document if we could get one. We'd also use statements by family members or jail officials indicating prior mental illness.

  2. Death while in jail custody. There were a couple of cases where a severe trauma happened in jail and the person was transferred to a hospital. We included those as well.

  3. These deaths had to have occurred between Jan. 1, 2010, and July 1, 2018. I felt that that was as far back as we could go and be reasonably assured we were finding most of the cases that were written about.

Beyond basic demographic information, I wanted to see if there were patterns surrounding the deaths, as well as outcomes: Was anyone charged with a crime or sued because of what happened?

There are currently over 70 fields in the database, including notes fields on cause of death, yes/no fields for various types of mental illness and fields indicating whether the person was shocked or Tazered, and whether there was an indication that the death was caused by some form of drug withdrawal. We also tracked outcomes – was a lawsuit filed or was anyone charged with a crime?

We had most of the fields set up before or at the very beginning of data collection. But there are a couple of fields we created while reading through stories and lawsuits.


Diana Dombrowski (Recent grad student at Marquette University, now a reporter for The Sheboygan Press in Wisconsin):

The crazy part about the database was that as we kept adding deaths and talking about the situations we were finding, we found more patterns. Some of these warranted new categories all on their own, but that always meant going back through the deaths we'd already documented to satisfy that category.

I noticed one trend when we were at least halfway through the database entries. In some cases family or friends came to the jails in person and sometimes with medication for their loved one. These cases really stuck out to me because they were instances where people who knew these inmates best were coming with medications and warnings that their loved one needed specific medication or treatment.

Whether there was some kind of warning and whether medication was brought by someone to the jail became two separate fields in the database. I had to go back through the two hundred-some deaths we'd already entered to check this, but I think it was worth it, and it resulted in a separate story that ran with the database.

And then, of course, the data visualization – it makes the story into a moving picture, a visualization on a national scale – can you talk a little about the tools in terms of education, personal skill, design, and software – that went into it? That's a big question, I realize, so maybe take one example or two.

Gary Harki:

The data visualization started off with a discussion between Will Houp, who at the time was in charge of creating digital graphics and interactive projects at The Pilot, and me. We'd worked together on this type of thing before – taking data and walking the reader through it in a (hopefully) compelling visualization.

I tried to think of the graphic as if it were a scene in a movie. At first I wanted the camera focused on one or two people struggling and ultimately dying with mental illness in jail. Then the camera pulls out to show you a whole sea of people who died in similar ways. First you get the reader to relate to one person; then you show them just how big the problem is.

Will Houp

Will Houp (former interactive graphics editor at The Virginian-Pilot, now an interactive producer at CNN):

To add onto what Gary said, we were hoping to create a visualization that explains the data in the most effective way. We started with a scrolling, story-style project that showed statistics but didn't communicate the individual experiences. But Gary and the students had collected such compelling video from the different cases in the database that he and I decided those needed to lead the visualization.

On the technical side, the project was built by me and coded from scratch in JavaScript using CSS and HTML as well (the core web technologies). The code runs a few calculations when the page loads and then watches while you scroll to determine which part of the visualization to show next. That took several iterations to get the timing right.

 I hadn't worked on something that moved as you scrolled like this one does, but I had seen it done by other publications and wanted to give it a shot. Almost everything I know about coding and building data visualizations has been self-taught. I took a couple of classes in graduate school but mostly I would see something I liked and try to replicate it or create my own version.

What drove your choices of the protagonists for your focus?

Gary Harki:

We really tried to find people who represented patterns we saw in the data, and also cases that felt like they deserved more attention than they'd gotten so far. I knew I wanted to do a section on Jamycheal Mitchell because his death really started me down the path to building the database and reporting this out on a larger scale. Also, it was important to tell part of the story from inside a jail and get the perspective of people who work there. We visited the jail and interviewed people there in January for the purpose of doing just that. For the other people used in the first story, it was a bit overwhelming. There are so many heinous cases.

Rebecca Carballo:

This might not be the most inspiring answer, but I used the filtering feature in Excel (or the graphic Will made) to find trends. For instance, when I wrote the story about those who died in isolation, I filtered it to see what the most common cause of death was. It was suicide. I then filtered it further to find which cases had lawsuits filed. It was ideal to look at cases with lawsuits because the court documents provide background information, and the lawyers can sometimes get you in touch with a family member.

Diana Dombrowski:

A lot of who I ended up interviewing had to do with who I could get a hold of. I identified several people in the database who satisfied certain situations whose families I would be interested in contacting and after running them through LexisNexis we hit a lot of dead ends. I left a lot of messages for either the wrong person or people who just weren't interested in talking to me. The people who ended up in the stories were sometimes the only ones who would return calls.

What was the timeline for building the database and reporting this out, from idea to publication? How did you map out those logistics?

Rebecca Carballo:

We began working on this in September 2017. Gary gave an outline of the kind of cases we were looking for – someone who was mentally ill who died in jail no earlier than 2010. We had discussions about what we were classifying as a mental illness. There were meetings where we grappled with if addiction should count as a mental illness. We mulled over whether we could assume someone was mentally ill if they died by suicide. We discussed various ideas for fields. Once we had a clear picture what we were looking for, we began dividing up the states between the four of us. It started off with us each finding and recording ten deaths a week. As the year went on we adjusted that weekly quota.

Diana Dombrowski:

Like Becca said, we started out with 10 deaths a week, but things changed as we went on. After a couple months, Gary was the one finding the deaths and then letting us fill in the details in the database. Entering the deaths was by far the most time consuming part of the database. We were only a couple weeks away from graduating in May when we finally finished the database - and even after that some were still added.

 Gary Harki:

I kept my editor back at The Virginian-Pilot, Bill Henry, in the loop, but most of the planning I did myself in Milwaukee. My original intention, back in October-November, was to publish in late June. As the database grew, both in the number of deaths included and in the number of fields we were tracking, that date was moved back.

I’ve reported out a number of large projects and basically I manage all of them through a rather detailed task list, and a list of stories and elements I want to include. By the end each story and graphic had a page or two of notes that kept inventory on what elements we needed and what tasks had and had not been completed. I worked off of it and kept track of what Will and the students were working on from there. Towards the end that task list expanded to others in the newsroom that were involved.

Lexy – Alexandria Bursiek, the other student who worked on this project and who is currently on her honeymoon – also helped out with project management by organizing and reading through a ton of reports, lawsuits and other material and organizing and prioritizing it. It was a tough job and not something that I think appeals to everyone. Luckily I knew Lexy well enough by the time I asked her to take on that task that I knew she’d do it well.

How did you take care of yourselves and your own mental health while serving witness to so such a huge scale of human misery? Or did you?

Gary Harki:

I don't know if I am the best at this, but I try very hard not to dwell on these cases when I'm not working. That said, when you're doing a project like this you're almost always working near the publication. I was very, very happy to take a week off after and travel to see family.

For me, the most important thing is to remember why we are doing this. It helps to feel like you can help the situation by writing about it. If I felt this were all hopeless, that what we were writing wouldn't have some impact, I don't think I could do it.

Also, I think it's safe to say we all enjoyed each other's company and had fun working together, despite the subject matter. As awful as this is, if journalism isn't fun on some level it's not worth doing.

Rebecca Carballo:

The team dynamic really helped me. In a strange way, it was a bonding experience. If I read a case that was particularly terrible, there were three people I could talk to about it, who could relate.  Also, when I was collecting state data, every FOIA rejection was motivation for me to keep pushing. There were so many states that weren't monitoring this issue. The hope that what we were doing would start a conversation, was enough to keep me going.

Diana Dombrowski:

It didn't take long for me to realize that I was reading about some pretty intense situations. When I'd talk about what I was doing with people outside of the project I would get a lot of horrified reactions, but talking to our team made everything a lot easier because we were all dealing with it at the same time. I don't think I've ever worked so well with a group before. I felt like we really meshed and became friends and having us all researching something so intense and just plain sad probably accelerated that. There were definitely some weeks we all felt burnt out, but Gary especially always reminded us of the end goal and why this was so important.