Picture a line chart with two lines on it. The first shows the number of children who die before their fifth birthday each year, worldwide. It starts high on the left, in 1990, and slopes down steeply towards the present day. Anyone glancing at it would call it one of the great success stories of modern health.
Now look at the slope in the last stretch of that line. It is flatter. Not flat, but noticeably shallower than the section before it. That change in angle is this week’s chart, and it is the part most people miss.
The data
The numbers come from the United Nations Inter-agency Group for Child Mortality Estimation (UN IGME), the joint effort of UNICEF, the WHO, the World Bank Group and the UN Population Division that produces the official global estimates. Its 2024 report, Levels and Trends in Child Mortality, covers data up to 2023. These are the figures we would plot:
- Under-five deaths in 1990: 13.0 million
- Under-five deaths in 2000: 10.1 million
- Under-five deaths in 2023: 4.8 million, of which 2.3 million were newborns in their first month
- Global under-five mortality rate in 2000: 77 deaths per 1,000 live births
- Global under-five mortality rate in 2023: 37 deaths per 1,000 live births, a 52% decline
- Change in the annual rate of reduction between the Millennium Development Goal era (2000 to 2015) and the Sustainable Development Goal era (2015 to 2023): down 42%
- Countries projected to miss the SDG target of 25 under-five deaths per 1,000 live births by 2030 if current trends continue: 60
The single takeaway
The world has not stopped making progress on child survival. It has slowed down, sharply, and the slowdown matters more than the headline.
A 42% fall in the annual rate of reduction means that each year since 2015 has delivered much less improvement than each year between 2000 and 2015 did. The report projects that on the SDG-era trend, an estimated 30 million children will die before their fifth birthday by 2030.
Put simply: the easy part of the curve is behind us.
Why the line bends
The report itself points to a set of pressures: conflict, climate change, the cost of living, economic instability, institutional fragility and worsening humanitarian crises. It also highlights that progress has slowed in some regions, including sub-Saharan Africa, where a large share of child deaths now occur.
There is also a structural reason visible inside the data. The report notes that newborn mortality has fallen more slowly than mortality among older children. The neonatal mortality rate fell 44% between 2000 and 2023, from 31 to 17 deaths per 1,000 live births, while mortality between one month and five years fell 58%.
That gap tells you something about what kind of health system the next phase requires. Many of the big gains of the 2000s came from interventions that could be delivered at scale through campaigns and community programmes: vaccines, bed nets, oral rehydration, treatment for pneumonia and malaria. Reducing newborn deaths depends more heavily on what happens around birth: skilled attendants, functioning facilities, resuscitation, care for small and sick babies. That is harder, slower and more expensive to build. It depends on the everyday quality of a health system rather than a single well-run campaign.
What the chart does not show
A single global line hides enormous variation. Some countries have cut child mortality dramatically since 2015. Others have stalled or gone backwards. The global slowdown is partly a story about where the remaining deaths are concentrated: increasingly in fragile and conflict-affected settings, where delivering any service is hard.
It also does not show uncertainty. The UN IGME publishes ranges around its estimates, and data in many of the highest-burden countries still relies on household surveys conducted years apart. The trend is robust; the exact figures for any single year in any single country are less so.
And it cannot show the counterfactual. The report estimates that if child mortality had stayed at 1990 levels, millions more children would have died. The slowdown is real, but it is a slowdown from a period of extraordinary progress, not a reversal of it.
Reading the slope properly
The “annual rate of reduction” is the figure doing the real work in this chart, and it is easy to misread. It is a percentage decline per year, not a count of deaths avoided. That distinction matters because a common defence of the slowdown goes like this: when mortality is high, it is easy to cut it by large numbers; once it is lower, each further reduction is naturally smaller.
That argument holds for absolute numbers but not for percentage rates. Falling from 77 to 70 deaths per 1,000 and falling from 37 to 33.6 are both roughly a 9% reduction. A constant annual rate of reduction produces a curve that flattens in absolute terms but keeps shrinking by the same proportion each year. The UN IGME finding is that the proportion itself has fallen, by 42%. The slowdown is not an artefact of arithmetic. Something in the underlying drivers has changed.
A rough illustration of what this means. At the 2023 global rate of 37 per 1,000, a country cutting mortality by 4% a year would take around a decade to reach 25. A country cutting it by 2% a year would take about twice as long. For the countries already well above the global average, the difference between those two paths is the difference between meeting the 2030 target and missing it by a generation. The figures here are illustrative, but the logic is why small changes in the rate matter so much.
The chart we would put next to it
If we had room for a second panel, it would split the global line into two: deaths in the first month of life, and deaths between one month and five years.
The second line would fall much more steeply than the first. By 2023, newborns accounted for 2.3 million of the 4.8 million under-five deaths, close to half. As older-child mortality keeps falling faster than neonatal mortality, the share of deaths that happen in the first four weeks keeps rising. Child survival is increasingly a newborn survival problem.
The reason is visible in what kills newborns. The leading causes, according to the WHO, are complications of preterm birth, complications during labour and birth, and infections. None of these responds to a single commodity delivered in a campaign. They respond to a trained attendant at birth, a facility with electricity and oxygen, someone who can resuscitate a baby who does not breathe, and a ward that can keep a small baby warm and fed. That is the health system itself, not a programme within it.
The strongest counterargument
It would be wrong to read this chart as proof that the old approach has run out of road. There is a respectable argument that the slowdown is partly about where children are born rather than what systems are doing.
Births are increasingly concentrated in countries where mortality is highest, particularly in sub-Saharan Africa, whose share of the world’s births has been rising. A global rate is a weighted average. If more children are born in places where survival is harder, the global line flattens even if every individual country maintains its pace. Some of the slowdown is a composition effect.
That argument has limits. The UN IGME report also shows that many individual countries have slowed, and that progress is weakest in fragile and conflict-affected settings. Composition explains part of the flattening, not all of it. And it does not make the policy problem easier. It means the remaining deaths are concentrated in exactly the places where health systems are weakest and delivery is hardest.
Second-order effects of a flatter line
A slower decline has consequences that go beyond the number of deaths.
It changes the politics of funding. Global health spending has historically been easiest to raise when progress is fast and visible. A flattening curve is harder to sell, precisely at the moment when the remaining work is more expensive per life saved. There is a risk of a vicious circle: slower progress weakens the case for funding, which slows progress further.
It changes what success looks like to funders. Campaign-based interventions produce outputs that are easy to count: doses delivered, nets distributed. Strengthening care around birth produces outputs that are harder to count and slower to show up in mortality data. Funders who reward quick, measurable results may unintentionally steer money away from where the remaining deaths are.
It changes the data problem. The countries with the highest mortality often rely on household surveys conducted several years apart. As attention shifts to newborns, the gaps in birth and death registration become more costly, because it is hard to improve care around birth when you cannot count reliably who was born and who died.
What each actor can take from it
For a ministry of health in a high-burden country, the chart argues for treating labour wards and newborn units as core infrastructure, with the same political priority once given to immunisation campaigns.
For donors, it argues for patience and for funding metrics that credit system-building, such as the quality of care around birth, rather than only counting commodities delivered.
For researchers and statisticians, it argues for investment in civil registration and vital statistics, so that the next version of this chart rests on counted births and deaths rather than extrapolation between surveys.
What to watch next
The next UN IGME estimates will show whether the SDG-era pace has held, slowed further or begun to recover. The indicator to watch is the neonatal mortality rate and its annual rate of reduction. If that line starts to bend down faster, the global curve will follow. If it does not, the 2030 targets will slip further out of reach.
Why it matters for health leaders
Three implications seem hard to avoid.
First, the strategies that produced the steep part of the curve will not, on their own, produce the next part. The remaining gains depend more on facility-based care around birth and on reaching fragile settings.
Second, funding decisions made now shape the slope of the line for years. The report explicitly frames the next stage as needing greater effort and investment than before, at a time when stable funding is uncertain.
Third, the SDG target is still a useful discipline. Sixty countries off track by 2030 is not an abstraction. It is a list of places where a concerted push could still bend the line back down.
One practical test for any plan that claims to address this chart: ask what share of its budget goes to care around birth, and how it will measure whether that care is getting better. Plans that describe newborn survival as a priority but spend mainly on commodities and campaigns are still fighting the last phase of the curve.
Over to you
If you work in maternal, newborn or child health: what do you think is flattening the curve in your setting? Is it financing, workforce, facility quality, conflict or something the global data does not capture? Tell us in the comments, and tell us which country you are writing from.
