From Certificate to Statistic: A Reader’s Guide to UK Mortality Data
Photo: U.S. Air Force photo by Senior Airman Ashley Talley, Public domain, via Wikimedia Commons
A death certificate is often treated as the end point of a clinical story. In public health terms, it is closer to a starting point. The document records a medical opinion about why someone died; the statistic derived from it eventually informs estimates of disease burden, service planning and research. Between the two lies a process that is legal, administrative and interpretive. Understanding that process is essential for anyone reading UK mortality data in 2026.
The document before the data
In England and Wales, a medical certificate of cause of death is normally completed by a doctor who attended the deceased. Since 2024, the medical examiner system has been on a statutory footing, adding a layer of scrutiny before registration. Cases referred to a coroner follow a different route, and the cause of death may remain provisional until an inquest concludes. Registrars then record the death, and the Office for National Statistics codes the causes for statistical use. Scotland and Northern Ireland have separate arrangements through National Records of Scotland and the Northern Ireland Statistics and Research Agency.
This matters because mortality data is not a single, uniform record. A figure published for the UK may combine systems with different timings, coding practices and revision cycles. Even within one system, a death registered in one week may not appear in the statistics for several weeks, and coroner cases can take much longer.
Coding turns prose into categories
The certificate records causes in words. Statistical agencies convert those words into codes from the International Classification of Diseases, most commonly ICD-10 in current UK practice, with planning for future transitions. Coding is not a mechanical translation. It requires rules for selecting the underlying cause—the disease or injury that initiated the train of events leading to death—and for recording contributing conditions.
A useful distinction is between the underlying cause and the conditions mentioned on the certificate. Someone may die with pneumonia, frailty and dementia all recorded. The underlying cause might be dementia, with pneumonia as a consequence of immobility; or pneumonia might be selected as underlying if it began the fatal sequence. The headline mortality table usually reports the underlying cause, while multiple-cause data preserves more of the clinical picture.
That single choice can change what a figure appears to say. A rise in deaths “from” pneumonia may reflect changes in how certifying doctors describe the sequence, not necessarily an increase in pneumonia incidence.
Why the headline figure is a choice
Mortality statistics are shaped by several decisions before publication. Crude counts rise as the population ages and grows; age-standardised rates allow fairer comparison over time. Provisional weekly figures are useful for surveillance but are subject to revision. Annual final data are more complete but arrive later. Some outputs report deaths by underlying cause, others by all mentions; some cover place of death, others registration district.
Readers should also note the denominator. A death rate per 100,000 people is not the same as the number of deaths, and a local spike based on small numbers can be statistically unstable. In a district with a few hundred deaths a year, a handful of additional respiratory deaths may look dramatic in percentage terms without indicating a sustained change.
Reading long-term patterns
Mortality data is often used to track dementia, respiratory illness and cardiovascular disease. Each requires caution. Dementia and Alzheimer’s disease are now more frequently recorded than in earlier decades, reflecting greater clinical awareness, changes in certification practice and an ageing population. That does not mean the underlying incidence has risen in a simple, linear way. Similarly, respiratory deaths are sensitive to winter infections, coding fashions and the distinction between pneumonia as an immediate cause and chronic lung disease as the underlying condition.
Cardiovascular disease illustrates a different pattern. Age-standardised mortality has fallen substantially over the long term, but heart disease and stroke remain leading causes of death. A falling rate and a large absolute burden can both be true. Mortality data describes deaths, not the full prevalence of disease among the living.
Delays, revisions and geography
Provisional figures can be revised as late registrations and inquest outcomes are processed. A sudden fall in deaths in one week may reflect reporting delays rather than a genuine improvement. Local authority comparisons can be affected by age structure, deprivation, ethnicity and the availability of care. These factors do not make the data unreliable; they make interpretation conditional.
Practical checks before quoting a figure
Ask whether the number refers to the underlying cause or all mentions. Check whether it is a count or an age-standardised rate. Establish whether the data are provisional or final. Confirm the period, geography and ICD version. Above all, avoid treating one week or one local report as a national trend.
Mortality statistics are a public health record of real value. They are also a mediated record, built from clinical judgements, legal processes and coding rules. Read carefully, they reveal long-term patterns in disease and care. Read carelessly, they can turn a final entry into a misleading headline.
