Project description

This year long, collaborative, data-driven investigation revealed for the first time, the scale of homeless deaths in the UK. Over the past year the work has sparked debate, influenced policy change at both a national and local level, and has shone a light on a previously ignored tragedy.

The project started in February 2018, in the midst of a freezing winter. I had what I thought was a simple question: how many people were dying while homeless? For weeks I called round coroners’ offices, hospitals, police, council and local government officials. I soon realised that no official body held the figure. No-one knew how and why these people were dying.

Thus, the Dying Homeless project began. After months of initial work we had created a groundbreaking database where we logged the details of each death. This database was used to build a powerful visualisation, which allowed readers to see more about each person, as well as understanding the scale of the tragedy.

We publicly launched in April, at which point we knew two people a week had died over the winter. We partnered with Channel 4 News and the Guardian, as well as with local outlets.

Using scraping, deep-web searches and FOI we were then able to reveal that, of 83 of the recent deaths in our database, not a single official review had been launched to explore the circumstances of the deaths, despite experts arguing that a review would be best practice. In fact, we found only eight official reviews into homeless deaths by any Safeguarding Adult Board since 2010. The findings prompted several charities and MPs to call for immediate change.

In October 2018 we produced the first-of-its-kind annual figure, revealing that at least 449 people had died homeless across the UK. The story led the Channel 4 News, with a 20 minute segment, and was picked up across the print press, appearing in the Guardian, the Times, the Independent, Huffpost, the Big Issue, the Metro, the Mirror and the BBC, amongst others.

Crucially, we had also opened up our findings to our Bureau Local network who reported out the specific data for their own areas. Collaborators included the Bristol Post, Ferret, the Detail, Yorkshire Post, Leeds Live, the Argus, Belfast Telegraph and others.

Since then we have continued to log deaths. In December we revealed the 2018 figure and a failure to provide support to councils to undertake reviews. Later, in March 2019 we revealed that a third of homeless deaths were from treatable medical conditions.

We now know that at least 800 people have died homeless in the UK in 18 months, around 11 people a week. Importantly, we were also able to tell the stories of those that died. They include an avid gardener, a physicist who had ambitions to work with Stephen Hawking, a 19 year-old living in a tent and a grieving 31 year-old who had lost both his mother and brother.

What makes this project innovative?

This project was hugely innovative in focus, scope and implementation. No-one else had noticed that homeless deaths were not being logged or recorded in an centralised way. On realising this concerning lack of evidence, we decided there was huge public interest in addressing this evidential void. We called on our network of more than 900 Bureau Local members all across the country to join on us on this nationwide collaboration, and over the next year they alerted us to deaths as and when they heard of them. We also created an online form which we circulated to shelters, soup kitchens and homeless outreach teams across the country. This was a huge crowdsourcing exercise and each submission needed to be checked and verified. We also used scraping, deep-web searches and FOIs to ascertain how rarely official reviews into homeless deaths occured. This was a true collaboration and to support our network of local journalists and hyperlocal bloggers, we opened up our database and produced clear “Reporting Recipes” which included context for the data, explainers on how to use the database to extract relevant information for their region, details of our topline findings and quotes from key experts. Later in the project we also recorded audio clips of some of our network members, sharing their own tips and advice on how to approach and report on such a sensitive issue. We were also keen that this project went beyond simply figures and statistics, we wanted to tell the stories of those that have died. My colleague Charles Boutaud used JavaScript's D3.js library to write code to present the stories of these people in a powerful online visualisation. As this was to be an ongoing visualisation growing over time with the project, we needed to make it easy and efficient to update. With this in mind, Charles wrote a Python script that used Googlesheets and GitHub APIs to check if new data had been entered in the Googlesheets fact-checked dataset and if so automatically commit the new data to the Github page repository hosting the visualisation.

What was the impact of your project? How did you measure it?

This year-long, collaborative investigation has had major impact at both a local and national level. We started our series using #makethemcount: a call for an official body to log these deaths. In December, prompted by our work, the Office for National Statistics produced the first ever official data on homeless deaths. This came after we opened our database to the ONS to help them develop their methodology. Our work was cited heavily in their release. The Scottish and Northern Irish statistical bodies have said they are considering producing their own data too. To date, our network has published 100 local stories and the findings were covered in 78 national stories. The day the ONS published its figures, we were cited in more than 800 articles. Over the year the project has repeatedly set the news agenda, provoked debates in parliament, (both houses), and prompted more than 20 MPs to speak out on the issue. We were invited to present the findings to the All Party Parliamentary Group on Homelessness and our work was shown to the UN Special Rapporteur on Poverty. We also presented it at various expert-led conferences. James Brokenshire, the Secretary of State for Housing called our findings “utterly shocking” and in summer the government’s Rough Sleeper Strategy recommended, for the first time, that homeless deaths be reviewed. Now, several councils, including Brighton & Hove, Oxford, Malvern and Leeds have said they will undertake their own reviews into deaths, while others, such as Haringey council, have put in place new protocols to log how and when people die homeless, a move we were told was directly prompted by our work. The work was also received positively by those working with and for homeless people. Several organisations called it a “wake-up call”. The Big Issue called the work “pioneering” and said: “To its huge credit, the Bureau’s research extensively looked into the stories of every one of those deaths”. It named the Bureau Local as one of its Top 100 Changemakers for 2019. Crisis UK said: “it is difficult to overstate the importance of the Dying Homeless Project, which has shed new light on a subject that was ignored for too long”. Our findings were also picked up and used by Comic Relief for Red Nose Day 2019, to help raise vital funds for UK homeless services. Across the country, members of the public have come together to remember those that passed away. In the last year there have been protests about homeless deaths in Belfast, Birmingham, London and Manchester, memorial services in Brighton, Oxford, Luton and London, and physical markers erected in Long Eaton and Northampton. As we draw our year long investigation to a close, we have worked hard to ensure the important work continues. The ONS will now produce official statistics, and we plan to hand over responsibility for the collation of personal stories to the Museum of Homelessness.

Source and methodology

We resolved to log these deaths after discovering that no one else did. But with no official data collection it was not easy. We spent a year talking to scores of people: doctors, charity workers, family members, coroners, soup kitchens, anyone we thought could help us record details of the homeless who had died on the streets, in hostels and in temporary accommodation. First, I worked out a careful methodology, after consulting various expert groups. We decided to use the same definition as the homeless charity Crisis; it defines someone as homeless if they are sleeping rough, or in emergency or temporary accommodation such as hostels and B&Bs, or sofa-surfing. In Northern Ireland, we were only able to count the deaths of people registered as officially homeless by the Housing Executive, most of whom were in temporary accommodation while they waited to be housed. For a year, we attended funerals and inquests, interviewed family members, collected coroner’s reports, shadowed homeless outreach teams and compiled Freedom of Information requests. We travelled up and down the country to Manchester, Glasgow, Brighton, Leeds, Bradford, Stafford, Milton Keynes and beyond, interviewing scores of people. We also coordinated dozens of journalists across the UK to make sure we covered as much of the country as possible. When we were passed a name by the public, we only published if it has been verified by local homelessness charities, the coroner’s office or other officials. We recognised that there was often no clear-cut cause for many of these deaths, and that this was both a highly sensitive and complex issue. We committed to recording these deaths in a respectful and nuanced manner and redacted sensitive information where necessary or when requested to by the family. We asked those who used our database to do the same.

Technologies Used

Javascript (D3.js), Python, Tabula, GoogleForms, Googlesheets and Github APIs.

Project members

Charles Boutaud The Bureau Local Network


Additional links


Click Follow to keep up with the evolution of this project:
you will receive a notification anytime the project leader updates the project page.