Why suicide data can lead us astray

On 29 August, the Office for National Statistics (ONS) released new data on suicides in England and Wales for 2023. It showed what the ONS called “the highest rate seen since 1999”.

This stark news was sadly true, but it’s also easily misunderstood. The Guardian reported that the ONS data meant “the rate at which people killed themselves in England and Wales reached the highest level in more than two decades”. The i said there were “6,069 suicides in England and Wales in 2023”. The Telegraph said: “There were 5.7 deaths by suicide per 100,000 females in England and Wales in 2023.”

In fact, none of these claims are really correct, because when the ONS talks about the “suicide rate”, it means the rate at which suicides are being registered, not the rate at which they’re happening. All three articles mentioned this elsewhere, but it’s an important distinction that is often lost.

Complicating matters further, a change in the rules in July 2018 also makes it difficult to rely on comparisons that reach back into the years before that.

The ONS has told us that the higher suicide rate since 2018 is “likely” to be “partly” because of this change, but it is rarely mentioned in the media. And it was often absent from social media posts marking World Suicide Prevention Day on 10 September.  

Altogether, it is no surprise that we have often seen trends in suicide oversimplified by politicians and in newspapers.

Is suicide becoming more common?

There were certainly more suicides per person in the 1980s than there are now. The decline in the rate, whether by registrations or occurrences, has been particularly pronounced in women, whose suicide rate was about half the male rate in the early 1980s, but is now more like a third.

After years of decline, the rate seemed to flatten for both sexes in the late 2000s, however, before rising slightly—at least on paper—from the late 2010s. But the data in this recent period, from roughly 2017 onwards, is particularly hard to read.

Firstly, there is the question of which suicide rate we ought to look at: figures are available both by date of registration and by date of occurrence. Most media reports refer to the rate at which deaths are registered. This figure is always more up-to-date, insofar as it refers to the most recent complete year, but it only describes the rate of registrations—not the rate of deaths.

This is unavoidable, because suicides are both tragic and complex, meaning there is often a delay of months or even years between a death occurring and its subsequent registration as a suicide by a coroner. Of the 6,069 suicide registrations in 2023, only 39% concerned deaths that happened in that year.

The ONS does also publish suicide data by date of occurrence, but it can take years for the trend to emerge. In a note on the most recently published data, the ONS said the 2022 figure was “a slight underestimate of the eventual number of suicides occurring in this year”, while the number of suicides that occurred in 2021 was revised upwards by 9%. 

This may seem like just a technical distinction, but relying on registrations can significantly change the story that the data tells. Looking just at the crude numbers, you could equally say that the rate (of suicide registrations) rose sharply in 2023 to a long-term high, or that the rate (of suicide occurrences) has fallen for the last three years in a row, to 2022. 

We know that the number of suicides in 2022 will rise, but how much it will rise is very hard to say. The rise in registrations in 2023 might mean there’s been a rise in the number of suicides. Or it might mean that coroners’ courts have been catching up with a large number of older cases. Or both might be true at the same time.

A similar issue sometimes caused problems during the pandemic, when for instance an unusually large number of Covid-19 deaths were reported in one day because of a backlog in the system, not because of a sudden spike in deaths occurring. By the same token, the sharp fall in suicide registrations in 2020 did not mean that suicides themselves had fallen by the same amount.

A new method designed to monitor suspected suicide occurrences more quickly, known as near to real-time suspected suicide surveillance, was launched in England in 2023, and may become more useful for understanding recent suicide trends in future. 

The standard of proof

The other issue to be aware of is a change in the standard of proof, which was introduced in coroners’ courts on 26 July 2018.

Before that date, coroners were expected to register a death as a suicide if the evidence showed that’s what it was “beyond all reasonable doubt”. Afterwards, the evidence only had to show that a death was a suicide “on the balance of probabilities”.

As the ONS said shortly afterwards: “It is likely that lowering the standard of proof will result in an increased number of deaths recorded as suicide, possibly creating a discontinuity in our time series.”

Unfortunately, it’s not been easy to establish what effect the change in the rules had. Looking at quarterly data for England, which the ONS also publishes, it is clear that the rate of suicides—and soon afterwards the rate of suicide registrations—rose substantially around 2018. Since then it has remained roughly flat, or perhaps fallen slightly, until the rise in registrations last year.

On the face of it, this looks like the kind of discontinuity that the ONS said was possible, as a consistently higher proportion of deaths began to be recorded as suicides from a specific moment onwards.

Of course, there could have been a rise in the number of suicides occurring at the same time that the rules changed. Indeed that might even explain most of the rise that we can see. And when you look closely at the quarterly data, it doesn’t quite fit the idea that coroners changed their behaviour suddenly in July 2018, because the rise began a few months earlier.

It’s possible to speculate that coroners may have begun to anticipate the new rules before they were officially applied, but we don’t really have good evidence for that.

When we spoke to the ONS about this issue in detail, it told us: “The ONS suicide statistics were impacted in July 2018 by the standard of proof used to determine whether a death was caused by suicide being lowered to the ‘civil standard’; balance of probabilities. Previously a ‘criminal standard’ was applied; beyond all reasonable doubt.

“The increase in the suicide rate in 2018 is likely to have been partly due to this change in the standard of proof, but other factors will have played a part too.”

In short, it’s likely that the suicide rate in the past would have been higher if coroners had been working with the same rules that apply now. How much higher, we can’t say, but it does mean that comparisons between rates before and after 2018 are less reliable.

How to talk about suicide rates

When using suicide data, we recommend always saying whether a rate or a number describes death registrations or the deaths themselves.

It’s important to be clear that the short-term trend of suicide occurrences may be different from the trend in registrations. It’s also important to say that the most recent figures for occurrences may well be revised upwards when new data arrives.

When talking about long-term trends, always bear in mind that the numbers after 2018 may be higher because of the way suicides have been counted since then. 

The chair of the National Suicide Prevention Strategy Advisory Group, Professor Louis Appleby told us: “When suicide rates change, we should start by asking: did anything change about the data? As Full Fact show, how and when we record suicide can influence national rates. Yet one thing never changes—any suicide rate is too high.”

Contacting Samaritans

If you are struggling to cope, please call Samaritans for free on 116 123 (UK and the Republic of Ireland) or contact other sources of support, such as those listed on the NHS help for suicidal thoughts webpage. Support is available around the clock, every day of the year, providing a safe place for you, whoever you are and however you are feeling.

Thanks to Professor Louis Appleby and the ONS Health Team for their help with this article.