Nearly Four in Ten Deaths Worldwide Are Not Registered. Almost Everything Else Is Built on That Count.

by Karim Haddad · August 28, 2026

World Affairs Correspondent, The Globe Ledger

Somewhere today a person will die, be buried by their family, and leave no trace in any official record. Nobody will have done anything wrong. There will simply have been no registrar within reach, or no requirement to reach one, or no document to bring. The death will not appear in a national statistic, will not adjust a life-expectancy estimate, will not contribute a cause of death to any database, and will not close the person's legal existence — which in practice means their land, their pension and their marriage may stay legally open long after they are gone.

This is not a rare event. According to the World Health Organization's fact sheet on civil registration and vital statistics, published on 27 February 2024, nearly 40 per cent of the world's deaths are not registered. That single sentence sits underneath an enormous amount of what we think we know about humanity, because a civil register is the original source for a great many numbers that are usually presented as simple facts.

Civil registration is the continuous, compulsory recording of vital events — principally births and deaths, and usually marriages and divorces — by a designated civil authority. It is deliberately boring. It is also one of the few pieces of state machinery that does two entirely different jobs at once, and it is the collision between those two jobs that explains why it is so frequently broken.

One ledger, two jobs

The first job is individual and legal. A birth record is the document that makes a person legible to the state: it establishes name, age, parentage and, usually, nationality. It is the thing a school asks for, an employer asks for, a bank asks for, a border asks for. A death record is its mirror — the instrument that lets an estate be settled, a widow inherit, a pension stop, a remarriage take place.

The second job is statistical and collective. Aggregate the individual records and you get vital statistics: birth rates, death rates, infant mortality, life expectancy, cause-of-death distributions. These are the inputs to health planning, to epidemic detection, to the population denominators used in almost every per-capita figure published about a country.

The two jobs pull in different directions. The legal function is driven by individual incentive — people register when they need the paper. The statistical function requires completeness, which means it needs the people who have no immediate reason to register to register anyway. A system optimised for the first will systematically under-count exactly the populations that matter most to the second: the rural, the poor, the displaced, the very old and the very young.

The WHO's regional figures show the pattern clearly. On births, the fact sheet reports registration rates of 98 per cent in the WHO European Region, 96 per cent in the Region of the Americas and 91 per cent in the Western Pacific Region, falling to 78 per cent in South-East Asia, 69 per cent in the Eastern Mediterranean Region and 44 per cent in the African Region. WHO estimates that 36 million babies born each year are not registered, and attributes that directly to capacity: only about half of countries, it says, are able to register at least 90 per cent of births.

On deaths the gradient is steeper. Drawing on the SCORE assessment of health data capacity in 164 countries, WHO reports registered-death coverage of 98 per cent in the European Region, 91 per cent in the Americas and 82 per cent in the Western Pacific, but 61 per cent in South-East Asia, 55 per cent in the Eastern Mediterranean Region and 10 per cent in the African Region. Deaths are harder than births for a structural reason: a birth produces a living person with a lifetime of future need for the document, while a death produces a document that only somebody else will ever need.

Cause of death is harder still. WHO reports that in low-income countries only 8 per cent of reported deaths have a documented cause. A death can be registered as a death without anyone medically certifying what the person died of — and cause is the part that public health actually runs on.

What fills the gap, and what that costs

Where registration fails, estimation takes over. This is why so many global health figures are described as modelled: the model is doing the work the registry did not. That is a legitimate response to missing data, and the alternative — publishing nothing — would be worse. But it changes what the number is. A modelled mortality figure is an inference from surveys, censuses, sample registration schemes and statistical relationships observed elsewhere. It carries uncertainty that a headline rarely carries with it.

The pandemic made the dependency visible. WHO's estimate that 14.9 million excess deaths were associated with the COVID-19 pandemic between January 2020 and December 2021 was necessary precisely because reported deaths were not a reliable count. The same fact sheet notes that 84 of 197 countries — 43 per cent — lacked the mortality data needed to contribute to that estimate. The number that mattered most in a generation had to be reconstructed, because in much of the world the ledger was not there to read.

There is a second, quieter cost, and it falls on individuals rather than on statisticians. If the birth was never registered, the person may have no way to prove who they are. The World Bank's Identification for Development programme puts the number of people worldwide without an official ID at approximately 800 million in its 2025 dataset, which is based on the 2024 round of the Global Findex Database, and reports that at least 2.8 billion people do not have access to a government-recognised digital identity they can use to transact securely online.

The composition of that group tracks the registration gap almost exactly. In a February 2023 post accompanying the programme's earlier estimate of 850 million people without official ID, World Bank researchers Julia Clark, Anna Metz and Claire Casher reported that more than half of those without ID are unregistered children, that the shortfall is concentrated in low- and lower-middle-income economies in Sub-Saharan Africa and South Asia, and that women in low-income countries are eight percentage points less likely than men to hold an ID.

Numbers that move because the measurement moved

Here is where a reader should slow down. The identification estimate has fallen from 850 million, published on the basis of estimates released at the end of 2022, to approximately 800 million in the 2025 dataset. It would be easy to read that as fifty million people acquiring documents. Nothing published alongside either figure supports that reading. The two estimates rest on different data rounds — the earlier one on an assessment ID4D describes as extensively updated in 2021 and 2022 with new data sources and individual-level Global Findex data, the later one on the 2024 Global Findex round — and the programme publishes no reconciliation of the two. Where an estimate and its instrument change at the same time, the difference between them cannot be read as movement in the world.

The birth-registration figures carry a comparable caution. UNICEF, whose global birth-registration data was last updated in June 2026, reports that around 150 million children under the age of five — roughly two in ten — remain unregistered. UNICEF also flags several problems with how that is measured: registration and possession of a birth certificate are not the same thing, and survey respondents sometimes confuse a birth certificate with a health card or another document, or treat notifying a church or village chief as formal registration. For some countries the latest comparable observation dates back as far as 2000, and for others it is as recent as 2012, so the figures describe registration status at the time of collection rather than the situation today.

None of that makes the numbers useless. It makes them the wrong kind of precise. The defensible reading of the global registration statistics is directional and structural: coverage is near-universal in high-income systems, thins sharply across South and South-East Asia and the Eastern Mediterranean, and is weakest by a wide margin in sub-Saharan Africa; deaths are counted far worse than births; and cause of death is counted far worse than either.

Why the gap persists is unglamorous. Registration requires an office a person can physically reach, staff to sit in it, a legal deadline short enough to matter and lenient enough to comply with, and a fee structure that does not price out the households least likely to register. Every one of those is a recurring budget line in a ministry that competes with clinics and schools, and none of them produces a visible result in the year it is funded. The payoff of a functioning register is that in ten years the country's health statistics are its own rather than someone else's estimate of them.

That is the honest case for treating civil registration as infrastructure rather than paperwork. Roads and power grids announce themselves. A registry announces itself only in the negative — in the pension that cannot be claimed, the inheritance that cannot be transferred, the epidemic that is noticed late, and the global death toll that has to be modelled because nobody wrote it down.