Two numbers sit in the same column of the same Ministry of Justice table, and they shouldn't really be able to coexist. For 2024, the classification "awaiting further information" holds 45 deaths in prison custody. For 2023, the same classification holds 10. Same category, same file, one year apart, a rise of 350% (Source: Ministry of Justice, Safety in Custody -- Deaths_in_prison_custody_1978_to_2024_accessible -- Table_1_1).

Read that quickly and it looks like something went badly wrong last year. It didn't, or at least this number can't tell you that it did. What you are looking at is not a cause of death. It is the absence of one.

Here is the honest reading, and it is the boring one. When someone dies in prison, the death is recorded straight away. The classification, self-inflicted, natural causes, homicide, other, usually waits on an inquest. Inquests take time. So the most recent year in any release will always be stuffed with cases the statisticians cannot yet label, and older years empty out as coroners finish their work. The gap between 45 and 10 is mostly the gap between twelve months of waiting and twenty-four months of waiting. That is the base-effect caveat, and it is real. Anyone who tells you 2024 was uniquely deadly on the strength of this row is reading the table wrong.

But do not stop there, because the interesting figure is the small one.

Ten deaths from 2023 still had no recorded cause when this file was published. Those are deaths that have now had well over a year, in many cases approaching two, and the state still cannot write down in its own statistics how the person died. Each of the 35 extra cases in 2024 will, in all likelihood, resolve into one of the standard categories over the next couple of releases. The ten from 2023 are the ones that haven't. They are the residue.

That matters because of what the classification is for. Self-inflicted deaths trigger different scrutiny to natural causes. Homicides trigger different scrutiny again. Prevention work, the whole point of collecting this data, runs off knowing which is which. Every case parked in "awaiting further information" is a case that cannot yet feed into any of it. It is a person, a bereaved family, and a lesson nobody has been able to learn yet.

These are the most recent available figures, covering deaths up to the end of 2024, and the queue they describe will already have shortened since. That's how the series works. The published number is a snapshot of an administrative process as much as a snapshot of what happened inside prisons.

Which brings me to the thing I'd actually change, and this is my view rather than the data's. A count of unclassified deaths is a useful public statistic in its own right. Publish it as a waiting time, not just a headcount. How many deaths are still unclassified after one year, after two, after three. The 45 would stop looking like a spike and start looking like what it is: a measure of how long it takes this country to say out loud how someone died in its care.