When national statistics show that deaths by suicide increased by 25% from one year to the next, the instinct is to assume something has gone terribly wrong. Are we in the middle of a social crisis? Has there been a policy failure? But in Czechia, the explanation is simpler and, in a way, more troubling: the original statistics were incomplete to begin with.
Šárka Daňková, Senior Analyst at the Institute of Health Information and Statistics of the Czech Republic (ÚZIS), has spent more than 2 decades working with causes of death statistics. Her meticulous comparison of 2 mismatched datasets – one from her institute and one from the police – produced a clearer picture of suicide mortality in the country.
A gap in the numbers
Šárka first came to ÚZIS in the early 2000s as a student. “I was interested in suicide, and I was writing my thesis about suicide mortality. I was looking for data and more detailed information,” she explains. She has been working there ever since, and, since 2024, leading the data collection working group at WHO/Europe’s Health Information Network. In 2021, WHO designated ÚZIS as a WHO Collaborating Centre for the Family of International Classifications.
Šárka Daňková, Senior Analyst at the Institute of Health Information and Statistics of the Czech Republic (ÚZIS).
At some point, Šárka noticed that ÚZIS and the Czech police had substantially different data on suicide – a gap that had been growing since 2010. This discrepancy has its roots in how deaths are certified in Czechia. When a body is found, a physician attends the scene and registers a preliminary cause of death on the basis of whatever information is immediately available. The police, meanwhile, begin a separate investigation.
“It can be really difficult for a physician to decide if the death was due to accident, murder or suicide,” Šárka explains. Even a forensic pathologist performing an autopsy may not have access to information about a person’s mental health history or family situation. The 2 investigations proceed in parallel, with little exchange between them. As a result, death certificates are sometimes completed without the context that would make the cause clear, and deaths by suicide end up coded as accidents or deaths of undetermined intent.
It was the category of undetermined intent that first alerted Šárka to the scale of the problem. By the time she began her data linkage work, cases recorded as undetermined intent had doubled in Czechia since 2016. “I had a feeling that I could improve the statistics and help to understand what those cases actually were,” she says.
Detective work
The data linkage exercise took about 4 months and involved matching individual death records from the ÚZIS registry with police data. The linkage is conducted under strict confidentiality safeguards, and only aggregated, non-identifiable findings are published. On paper, this sounds straightforward, but in practice, it was anything but.
“One issue is to link the data, but then you have to evaluate the data. That was the most complicated part,” Šárka explains, comparing it to detective work. If the data was contradictory or incomplete, she would personally consult with the certifying physician and the police to see whose account was more likely to be correct. She also looked for other evidence that might indicate suicidal ideation, such as a note or a message left by the person. Records from toxicology were also considered.
When Šárka wasn’t confident that the police evidence was sufficient to warrant a change, she left the record as it was. As the main purpose of the linkage exercise was to have more accurate national data, Šárka updated only the records held in the national database rather than the original death certificates. As a result, most death certificates remained unchanged.
The collaboration with the police, she notes, was easy. “They wanted to improve their statistics as well, and they wanted to lower the discrepancies.” In the end, the statistical cause-of-death records held by ÚZIS were updated, with significant effect on the aggregate statistics: the number of deaths by suicide rose from 1249 to 1561.
As uncertainly fell, recorded suicides rose
In 2024, improved data linkage shifted some deaths previously classified as being of undetermined intent into the suicide statistics. As a result, recorded suicide deaths increased by 25%, while deaths of undetermined intent decreased by 14% - reflecting better classification.
Fig.1: Recorded suicide statistics rose by 25%, while deaths of undetermined intent decreased by 14%. Source: WHO
A better picture
The corrected figures revealed not just a larger national total of deaths by suicide, but also a geographic pattern. Some regions that appeared to have low suicide rates actually had very high rates of deaths coded as undetermined intent, meaning that the true suicide mortality rate in those areas had been hidden.
This matters enormously for resource allocation. Šárka describes a teacher from one such region, whose school had experienced multiple suicide attempts among pupils and applied for funding to hire a psychologist. The application was denied on the grounds that the local suicide statistics did not justify the expenditure. “The improvement in the statistics could really help that school receive the funding now. This is why we need correct and complete data when it comes to suicide prevention. We need to know where and how to target prevention efforts, in which age group and in which region,” Šárka says.
The new data also surfaced a pattern that Šárka had not expected: many young people who had died at railway tracks had been coded as accidents when in fact they had died by suicide. This finding is particularly concerning because in 2022, suicide was the leading cause of death among people aged 15–24 in the WHO European Region – for young women and young men alike. Accurately identifying these deaths is essential for understanding where and how young people are at risk, and for designing interventions that can reach them. Railway operators, Šárka notes, have already started exploring AI-based systems to detect unusual behaviour on platforms. More accurate data on the scale of the problem, she hopes, will strengthen the case for that kind of work.
Deaths from transport accidents also declined
Between 2023 and 2024, deaths classified as transport accidents decreased by 12%. This is consistent with evidence from the linkage exercise that some deaths previously recorded as accidents were subsequently identified as suicides.
Fig.2: Deaths classified as transport accidents decreased by 12%, as some of them were subsequently identified as deaths by suicide. Source: WHO
For policy-makers receiving the revised statistics, Šárka has a clear message: this is not evidence of a sudden crisis, but a measurement correction. “Take it as a new number, a new piece of information about the real size of the issue. Don’t look at it as a trend.” Comparing the new figures to historical data would be misleading. Instead, she urges decision-makers to focus on the demographic and geographic details of the new data, and to use it as the baseline from which future trends will be measured. “We have to wait and see if suicide mortality is increasing, decreasing or stagnating,” she says.
The case for data linkage
Data linkage exercises like this one are not unique to suicide mortality. Šárka does similar work for tuberculosis, cancer and other conditions. This is why completing a death certificate carefully and thoughtfully is so important. “I believe doctors should spend time with patients and not with papers,” she says. But at the same time, she wants physicians to have a better understanding of how death certificates that they fill out are used. “It’s not just bureaucracy. Someone, somewhere, reads it and makes decisions based on what is written there,” she explains. She also hopes for a more systematic approach to data linkage in the future, where uncertain cases could be reviewed by a third independent body that would bring the available data together.
Reducing fragmentation, strengthening links across information systems and translating data and evidence into policy action have been key priorities of the WHO Country Cooperation Strategy for Czechia 2024–2030 and of the long-standing work of WHO with national partners.
Czechia may not be alone when it comes to incomplete data for suicide mortality. Deaths classified as being of undetermined intent are an important signal of cause-of-death data quality, and some may actually be suicides that could not be identified at the time of death registration. Also, countries report mortality data for different years and with varying levels of timeliness and completeness, making it difficult to produce a meaningful current estimate for the entire Region.
The issue can instead be examined country by country. WHO’s Analysing Mortality and Causes of Death (ANACoD3) tool helps countries systematically assess the quality of their mortality data. It can highlight unusually high proportions of deaths assigned to undetermined or poorly specified causes, as well as inconsistencies by age, sex, cause and external-cause categories. These signals do not prove that deaths have been misclassified, but they can identify where closer investigation and, potentially, linkage with police, forensic or other administrative records may be warranted. Czechia’s experience demonstrates what that next stage of investigation can uncover.
Having a more accurate picture is an important step forward. Czechia can now direct resources more precisely towards the regions, age groups and settings where they are most needed. Deaths previously obscured by uncertain classifications are visible in the statistics. And that visibility can help ensure that someone at risk receives support before it is too late.