Where the deaths fall

Demographics of the Grave is the block of five bars in the middle of the Ledger. Infancy, school age, the long middle of life, the years from 50 to 69, and 70 and over. Each bar carries a percentage, and a count that ticks up while you watch. The tall bar is easy to read as the dangerous age, and the short one as the safe age. The percentages are answering a different question. They say what share of deaths1 falls in each age band2.

A share of deaths is not the chance of dying at that age. Ages 15 to 49 hold a large part of the people alive and only about a seventh of the deaths. Ages 0 to 4 hold a much smaller part of the living and still about 8 per cent of the deaths. The chance of dying is a death rate3. The bars are not death rates. The same mix-up, applied to countries rather than ages, is the subject of Why Germany looks deadlier than Kenya.

What the bars are using

The methodology page is already plain about this. The five percentages are fixed, and they are rounded approximations of the global pattern: 8.3 per cent at ages 0 to 4, 1.7 per cent at 5 to 14, 10 per cent at 15 to 49, 30 per cent at 50 to 69, and 50 per cent at 70 and over. They do not change by cause. A road death and a dementia death are painted with the same age brush. The count ticking beside each bar is that share of the deaths since you arrived. It is not a count of deaths at that age.

The bars are also drawn against the tallest one, so the 70-and-over bar fills its track and the others are scaled to it. The percentage printed beside the label is the share. The length of the bar is that share compared with the largest share.

Set beside a published series, the sketch is close at the ends and looser in the middle. The figures below are modelled4 estimates from the Global Burden of Disease study, as compiled by Our World in Data, for the same five bands. In 2023 that series puts about 60 million deaths in the world. The Ledger's all-cause baseline is the round 61.5 million discussed in Why 1.95 deaths per second?. These are neighbouring totals from different reconstructions. The shares below use the age series' own total, so they are not the dashboard percentages multiplied by 61.5 million.

The parts are rounded, and together they are the series' total of about 60 million. The school-age bar and the two ends are near the printed figures. The two middle bars are the rough ones. Ages 15 to 49 are printed as 10 per cent and come out at 13.5. Ages 50 to 69 are printed as 30 per cent and come out at 26. The same gap is there in 2019, before the pandemic year, so it is not an artefact of 2021. It is the rounding.

At the 2023 figures, that is about 12,800 deaths a day under five, about 2,400 a day at ages 5 to 14, about 22,200 a day at 15 to 49, about 43,000 a day at 50 to 69, and about 84,000 a day at 70 and over.

The bands are not the same length

The groups are the ones the global cause tables use, which is why they are uneven: five years, then ten, then thirty-five, then twenty, then every age after 70. Comparing 1.5 per cent with 13.5 per cent as if the bands were the same stretch of life makes the middle look more dangerous than the pile per year of age suggests, and makes school age look even quieter than it is.

Spread across the years inside each band, 2023 is about 930,000 deaths for each year of age under five, about 87,000 for each year from 5 to 14, about 230,000 for each year from 15 to 49, and about 780,000 for each year from 50 to 69. The last band cannot be divided this way, because it has no end. This is still a pile, not a risk. It does not divide by how many people are that age. A year of age with more people in it contributes more deaths at the same chance of dying. What it does show is the shape: a high count in infancy, a trough in the school years, and a climb from there.

Infancy and childhood, ages 0 to 4

About 4.7 million deaths, and about 8 per cent of all of them, from the five years before a child turns five. The infant-mortality article is about a different fraction: the share of children who die, counted per 1,000 live births. That article uses the UN figure of 4.9 million under-five deaths in 2024, about 13,400 a day, and a global rate of 3.7 per cent of children born. This series, from a different model, puts 2023 at about 4.7 million. Close, and not the same number. One is a risk of dying in childhood. The other is a share of every death in the world.

The causes are the ones that article sets out: being born too early, lack of oxygen during birth, infection in the first weeks, and, among children who survive that first month, pneumonia, diarrhoea and malaria. On the UN figures used there, just under half of under-five deaths are now in the first 28 days, 2.3 million of the 4.9 million. The bar on the Ledger is the whole of those five years, taken together.

Youth and adolescence, ages 5 to 14

This is the shortest bar, and the quietest stretch in the table. Ten years of life, about 0.9 million deaths, about 1.5 per cent of the total, about 2,400 a day. That is roughly a fifth of the daily count under five, from twice as many years of age.

The WHO fact sheet on road traffic injuries calls them the leading cause of death for children and young adults aged 5 to 29. That span is wider than this bar. It includes the school years and the decade after them, and the road deaths of older adults sit on the road counter with everyone else. Injury, including drowning, is a large part of what little death there is here. The bar is short because even the leading causes at these ages are small beside infancy on one side and old age on the other.

Prime and midlife, ages 15 to 49

Thirty-five years, and about 13.5 per cent of deaths. The dashboard prints 10. This is the band where the causes stop looking like old age, and have not yet become it. Road injury runs through it. So do maternal deaths, suicide, interpersonal violence and HIV, alongside the start of heart disease and cancer. The Ledger's cause counters do not split any of those by age. A line in the feed about a rider in Bangkok and a line about a heart in Hamburg are given the same age mix as each other.

In some countries this band is the largest share of deaths, because old age has not yet taken over the pile and childhood has loosened its hold. Kenya, in this series for 2023, puts about a third of its deaths at ages 15 to 49, about 15 per cent under five, and about 26 per cent after 70. Japan puts about 2 per cent of deaths in the whole of ages 15 to 49.

The years from 50 to 69

Twenty years, about 15.7 million deaths, about 26 per cent of the total. The dashboard prints 30. This is where the leading causes of the world table are already doing most of the work: heart disease, stroke, cancer, chronic lung disease, diabetes and kidney disease. The global targets often talk about death before 70 from those diseases. This band is not that indicator. The target usually starts at 30, and this bar starts at 50. What they share is the idea of a death that arrived before old age had become the usual time.

Falls belong here and in the next band. The WHO fact sheet says the highest fall death rates are in people over 60. The Ledger gives every fall the same age split as every heart attack.

Seventy and over

About 30.7 million deaths, 51 per cent of the series, about 84,000 a day. The dashboard prints 50, and for once the rounding is tight. This is the bar that fills the track.

The causes are the same chronic diseases, carried further, with dementia among them. Alzheimer's disease and other dementias are a disease of advanced age that was named for a woman of 51. The deaths gathered on that counter are concentrated long after 51. "Old age" is not itself a cause on a certificate. A death at 80 is still written as a heart attack, a stroke, a cancer, a fall, or dementia, which is why the cause counters and this bar can both be true.

The bar is tall for two reasons that are easy to fold into one. The chance of dying in a given year is highest at these ages. And enough people now reach them that the deaths accumulate here. A population that survives childhood, and then survives to 70, will bury most of its dead after 70 even while the chance of dying at each older age slowly falls. That is the same mechanism as the rise in the annual total in the piece on 1.95 a second: more people, and older people, produce more deaths.

The world average is not a country

The five bars are one world, held still. A high-income country and a country where children still die in large numbers do not draw this picture.

India sits near the world, and weighs heavily in it. Japan would fill almost the whole track with the last bar. In Nigeria and Chad the first bar is still the largest. Germany's crude death rate looks high next to Kenya's for the reason in the other article: Germans are older. This table is that fact as a pile of deaths, rather than as a rate.

How the pile moved

The chart on the dashboard cannot show change. The shares are fixed. In the series behind this article, the change since 1980 is the whole story of the picture.

In 1980 the under-five deaths and the deaths after 70 were about the same size: about 14 million and about 13 million. By 2023 the second had become about 31 million and the first about 4.7 million. The middle bands moved much less. People did not stop dying. The deaths moved up the age range, which is what a long fall in child mortality looks like when it is drawn as shares. The longer history of that fall, back before 1980, is in the infant-mortality article. What a life expectancy built from these age-by-age chances actually measures is in the piece on life expectancy.

The total in this series was about 55 million deaths in 2019, about 66 million in 2021, and about 60 million in 2023. Every adult band was higher in 2021 than in 2019. The bars on the Ledger stay where they were printed. A year like 2021 does not appear on them.

More from the Ledger

Last reviewed: 5 October 2026