2026 is on track to be the deadliest year on record for people held in ICE’s growing archipelago of civil detention facilities across the country. I began writing about these deaths at the start of this year because I wanted to make sure that the public understood these tragedies as part of a collective trend rather than a series of disconnected individual events.
In doing so, I created a working dataset for myself so that I could simply keep track of the details of the burgeoning cases. But after realising that the public lacked a vetted repository of these cases, I asked our team at Relevant Research to help me expand the data and turn it into a public resource. We did, and that’s how the Detention Mortality dashboard was created. See the previous announcement post here: New Tool Now Available: The ICE Detention Mortality Dashboard.
Adam Sawyer and I held a public briefing on Monday about this resource. I had the opportunity to walk through the decisions that went into the design, why we included some things and not others, and answer questions from an engaged audience. I’m grateful to Adam for hosting and moderating, and thankful for those who could attend.
Although we did walk through the features of the dashboard, most of the discussion focused on the background of ICE detention death reporting and how we collect and process those announcements into data. I was particularly adamant about a few points.
First, I wanted the entirety of the dataset to make it clear that this is not a problem that is limited to the current administration. Detention conditions have been far from adequate during previous administrations, and they will continue to be deadly in the future without systemic improvements and an overall reduction in the numbers of people who are held in the first place. Even so, as I said in the briefing, “This administration in particular has sought to criminalize people in death as they were criminalized in life.”
Second, it’s critical that we understand the limitations of a traditional cause of death analysis and think socially and politically about the relationship between social systems and fatalities. As I tried to illustrate in my conversation with Adam, “Is it fair to say that a detention center is capable of killing someone, or a detention system is capable of killing someone, as opposed to someone had cardiac arrest and their heart killed them? Is it not true that class and racism and patriarchy are causes of death for many people?” This is the kind of question we have to ask continually.
Third, as a practical matter, projects like this exist because the official source information for these deaths can come and go. As I said then, “I never trust that those web pages are going to be there when I need them again.” So gathering, validating, archiving, and visualizing the data is actually one step—not a huge step but a necessary step—toward ensuring that we maintain a public record of these tragedies.
You can listen to the entire discussion above, and I’ve provided several more robust sections below, including:
A summary of FAQs
Links to related resources
An example of how reporters can use this resource
Just a quick note before we get to the Q&A. The best way to support this work is to upgrade to a paid subscription here on Substack. This ensures that I can spend as much time as possible doing the research, writing, and creative work needed to respond to our fast-changing immigration policy landscape.
Upgrade to paid
1. Frequently Asked Questions
Adam Sawyer: What is ICE detention death reporting, and what is the Detention Mortality Dashboard?
Austin Kocher: ICE has two obligations when someone dies in its custody. It publishes a short detainee death notification, which the agency treats as a press release due within roughly 48 hours, and it later files a fuller detainee death report with Congress at around 90 days. I started writing about these deaths case by case, and once the count passed twenty or twenty-five I needed a system to keep track of them. The dashboard grew out of the spreadsheet I built for my own writing. I also archive the source pages, because DHS pages get deleted or quietly edited and I do not trust them to be there when I need them.
AS: Has ICE’s reporting changed since the current administration took office?
AK: I have not tracked this as long as other people have, so I rely on those who have, like Andrew Free, who report that the documents have grown shorter and less detailed, particularly the 90-day reports. We are also seeing ICE run past the 48-hour notification deadline, and some 90-day reports arrive late or thin.
AS: Where can someone find these records themselves?
AK: ICE keeps the notifications on its news releases page, filterable by category, and the fuller reports on a separate detainee death reporting page as PDFs. Structuring them into data is harder than it sounds. We capture the title, the body, the link, the date the person died and the date ICE announced it, so anyone can check the agency against its own deadline. Names are the hardest field, because ICE and news outlets both misspell them, and this is someone’s life. Early in my career I turned sheriff checkpoint reports into maps, which posed the same problem, building a dataset out of documents that were never meant to be one.
Audience: Jails and prisons often move someone to an emergency room before a death is declared, which keeps it off their count. Is ICE doing that?
AK: Yes. Custody is broader than the inside of a detention facility, so we record both the last facility a person was held in and the place they actually died. Medical care is also expensive. Sheriffs across the U.S. South told me repeatedly that one heart attack or one diabetic seizure can blow through a jail’s medical budget for the year, so there is a financial reason to release a medically fragile person rather than treat them. This administration also rolled back the requirement that ICE track deaths within 30 days of release. The morning of this conversation an attorney emailed about a client deported mid-crisis who died at home a week later and asked whether the name could go on the dashboard. I had to say no, because that death produces no notification and no report.
AS: Can you walk us through the site?
AK: The site opens on a map of every facility where a death has been attributed, color-coded for whether the person died inside the facility or elsewhere after a transfer. Filters for state and date range work in real time, so a map can be set up for a presentation and screenshotted. Below that is a chart of custodial deaths over time by fiscal or calendar year, and fiscal year 2026 is the deadliest on record as far as our records go. Below that is the case registry, one page per person, linking to the ICE notification and, where more than 90 days have passed, the fuller report. The whole file downloads as a CSV with latitude and longitude included. The data originally went back to 2018 and now reaches 2003.
AS: Does the dashboard include people who die crossing the border?
AK: No. Border deaths are tracked elsewhere. Pima County is the best single source, and No More Deaths has newer data for the El Paso region. That work has its own methodological difficulties and it is a different dataset, though anyone wanting a fuller picture might combine them.
AS: Does it include deaths from ICE officer-involved shootings, such as Renee Goode?
AK: No, unless the person was already in custody, which would be unusual. This is focused on people who had entered the detention system in some way. Those cases are not in the dataset and are not carried as edge cases either, though they are important in their own right.
AS: Why focus only on immigration detention deaths rather than jails and prisons as well?
AK: Good research is incremental, very much like a Lego block. This one came out of realizing I had built something for my own writing that other people could use, and once you have gone to the trouble of validating it, keeping it private is hard to justify. It is also unfunded and pro bono. Relevant Research pays the bills through client work on immigration data, and public tools are how we give some of that back. Bishal Timsina built the site, and Adam’s work on facility data shaped the counting rules. Other people will take the useful part, or the missing part, and build their own.
Audience: What is a cause of death, and can a detention center be one?
AK: If someone dies of respiratory failure you can write down emphysema and stop. But class, race and gender shape who develops that condition, who gets treated for it and who dies of it, and social scientists have argued for a long time that those are causes too. So it is at least a coherent question whether a detention system is capable of killing someone, as distinct from saying a heart stopped. As a political geographer I take a wide view of this. Many people arrive in detention already carrying elevated vulnerability because of where they live and what work they have to do. That question is much bigger than this project can answer.
Audience: How is maternal mortality handled in ICE detention, and would an infant’s death count as a detention death?
AK: Our dataset holds nothing on it, and ICE’s reporting carries no evidence either way. Zain Lakhani at the Women’s Refugee Commission is who I look to here. There is substantial evidence that pregnant, postpartum and nursing women face serious complications, as do their infants. She has described women so malnourished in detention that by the time they are deported they cannot nurse, which raises questions about a child’s development even where there is no death in custody. It is the right question to ask, and there are people working on it.
AS: Any last word?
AK: Share the resource, and do it in a way that is not sensationalizing or tokenistic, and not overly partisan. Neither political party has clean hands on this issue. I would like it to contribute to reducing this kind of harm. And if you see a mistake, or anything that should be updated, get in touch. We aim for accuracy and would rather fix something.
Migrant memorial on Mexico side of Texas border, bridge to U.S. in the background. Author’s photo.
2. Additional Resources from the Briefing
Sadly, the topic of migrant death is not a new one. I have written about this many times before.
Beyond the Official ICE Detention Death Count, on the three groups of people detention harms, only one of which anyone counts.
Hospitals Should Never Become Places Where Fear Overrides Care, my conversation with the National Immigration Law Center about what happens when ICE brings someone it is already holding into an emergency department.
15 Years of Data on Migrant Deaths, on the border-death databases, written when No More Deaths released fifteen years of El Paso Sector records.
Migrants Are Left to Die at the Borders, on the No More Deaths report about 911 calls placed from the desert, which documents disappearance rather than death.
The Precarious Life of the Migrant, written days after 51 people were found dead in a truck in Texas and 23 at the Melilla border, on what makes migrant life precarious in the first place.
People filled the chat with resources, and I tried to capture them here, along with the ones that came up in discussion.
IncarcerationTransparency.org, shared by Jay Aronson of Carnegie Mellon University, who tracks deaths in Pennsylvania jails because nobody else does.
Death in Custody: How America Ignores the Truth and What We Can Do About It, by Roger A. Mitchell Jr. and Jay Aronson, on the same failures in the criminal legal system, where the documentation is thinner still.
Unbreaking, the volunteer project tracking federal policy changes, where Maggie Vail is one of the immigration page leads.
The KonTerra Group, which Lynne Cripe noted offers counseling and support groups for immigration attorneys and advocates.
Detention Kills, where Andrew Free has documented these deaths far longer than I have.
LA Taco, which has broken several of these cases before ICE posted anything.
Detention Reports, for the facility population data that pairs with a case record, run by Adam Sawyer at Relevant Research.
The El Paso Sector Migrant Death Database, which No More Deaths built out of medical examiner records, CBP files and direct observation.
The Arizona OpenGIS Initiative for Deceased Migrants, a similar interactive database for Arizona.
The IOM’s Missing Migrants Project, which tracks similar data along routes to the E.U. and elsewhere.
3. How Reporters Can Use This Resource
Reporters are an important audience for well-vetted, nonpartisan immigration data. I always invite reporters to join the live briefings and discussion I host. This is intended to contribute to public information. Reach out for direct quotes if needed, but just for my own time management, I encourage reporters to quote from public discussions like this one if they can.
Mathew Miranda of the Sacramento Bee was on the call and published an article that evening, which illustrates exactly how we hope the dashboard can contribute to research and reporting. Mathew reports that all six people who have died in ICE custody in California under this administration died at hospitals, shortly after being held at a federal detention center, four of them after Adelanto and two after the Imperial Regional Detention Facility.
He also reports that California accounts for 40 of the 289 deaths since 2003, second only to Texas, and sets that against the California Department of Justice investigation published in May, which found detention conditions had worsened under the current deportation campaign, and against the detention bills now waiting on the governor’s desk. Thanks, Mathew, for covering it the same day.
Take a minute and share this post so that others have access to this information.