Childhood leukaemia worldwide, ages 0-14
also called global childhood leukaemia incidence · IICC-3 leukaemia rate
Leukaemia is the commonest cancer of childhood: about 46 children in every million aged 0-14 are diagnosed each year, which works out to roughly 90,000 children worldwide. Infants under one are a small slice of that, which is why an infant-only drug programme looks unfundable while a childhood-leukaemia platform does not.
The same thing, for a scientist
The IARC International Incidence of Childhood Cancer study (IICC-3) reports a world-standardised leukaemia incidence of 46.4 per million person-years in children aged 0-14, the single commonest childhood cancer group, against an all-cancer WSR of 140.6. Applied to a global 0-14 population of roughly 2 billion, that implies on the order of 90,000 childhood leukaemia diagnoses per year worldwide. Registered all-cancer incidence in this age group rose from 124.0 to 140.6 per million between the 1980s and 2001-10, a change that is at least partly ascertainment.
The denominator, stated first2 deeper
Denominator: children aged 0-14, world standard population, person-years 1. IICC-3 rates are world-standardised incidence rates (WSR) per million person-years in the 0-14 band. They are not per million total population and they are not infant rates.
| Measure | Value (ages 0-14) |
|---|---|
| All childhood cancer, WSR | 140.6 per million person-years 1 |
| Leukaemia, WSR | 46.4 per million person-years 1 |
| CNS tumours, WSR | 28.2 per million 1 |
| Lymphomas, WSR | 15.2 per million 1 |
| Cases contributing to the 0-14 estimate | 284,649 over 2001-10 1 |
To convert to a global count you need a global 0-14 population. Taking ~132 million births a year sustained over fifteen years gives ~2.0 billion children aged 0-14 2, so 46.4 per million implies roughly 90,000 childhood leukaemia diagnoses per year worldwide. Treat this as one significant figure.
Where the infant slice sits inside it
Infant acute leukemia contributes on the order of 5,400 of those ~90,000 cases, about 6% (see epi-infant-leukemia-global-burden). The proportion is small because the 0-14 band is fifteen times as wide as the infant band while the infant rate (~41 per million infants) is comparable to, not vastly above, the 0-14 average.
This is the single most important framing point in the whole impact beat: infant leukemia is not rarer than childhood leukaemia because the rate is lower; it is rarer because the age window is one year wide. Anyone who quotes '3 per million' and '41 per million' as if they contradict each other has confused a rate with a window.
Regional variation, and what it is worth
IICC-3 found incidence varying considerably between and within the 19 geographical areas and ethnic groups it covers, and by cancer type, sex, age and ethnicity 1. Two cautions before anyone builds an aetiological argument on that variation.
- Only 153 of 532 invited registries met the quality bar, so low recorded incidence in a region can mean low ascertainment rather than low risk 1.
- The 1980s-to-2000s rise from 124.0 to 140.6 per million is exactly the pattern that improving diagnosis and registration produces, so it cannot be read as a rise in true risk without further work 1.
The honest use of this dataset is as a denominator and a sanity check, not as evidence about causes.
| metric | incidence |
|---|---|
| value | 46.4 |
| unit | per million person-years, world-standardised |
| population | children aged 0-14 |
| region | worldwide (153 high-quality registries, 62 countries) |
| year | 2017 |
| source | IICC-3, Steliarova-Foucher et al., Lancet Oncol 2017 |
| denominator | children aged 0-14, world standard population |
| all cancer wsr 0 14 | 140.6 |
| cns wsr | 28.2 |
| lymphoma wsr | 15.2 |
| derived global childhood leukaemia cases per year | 90000 |
| infant share of childhood leukaemia PCT | 6 |
- Quoting the 0-14 leukaemia rate when arguing for infant-specific funding inflates the apparent population fifteenfold and will be caught. The rate and the age window must always travel together.
- What does the successor to IICC-3, covering 2011-2020, show for leukaemia in the 0-4 band, and has the regional gradient narrowed as registration improved?
- Can infant-specific rates be extracted from the IICC contributing registries on request, which would replace the derived figure in `epi-infant-leukemia-global-burden` with a measured one?
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
In children aged 0-14, the world-standardised incidence of leukaemia is 46.4 per million person-years, the commonest childhood cancer group, within an all-cancer WSR of 140.6 based on 284,649 cases; the all-cancer WSR rose from 124.0 in the 1980s, and 153 of 532 invited registries met quality standards.
International incidence of childhood cancer, 2001-10: a population-based registry study · Steliarova-Foucher E, Colombet M, Ries LAG, et al. · Lancet Oncology 2017What this evidence cannot show
Covers diagnoses in 2001-10, so it is now a historical baseline. Registry coverage is geographically uneven, with least coverage in the highest-birth regions. Age-specific rates are published for 0-4, 5-9 and 10-14, so the infant year cannot be extracted from the published tables.
Evidence that argues against this (1)
- Registry incidence is diagnosed incidence, and worldwide the two differ by nearly half. A calibrated microsimulation of childhood cancer incidence across 200 countries and territories estimated 397,000 (95% UI 377,000-426,000) incident childhood cancers worldwide in 2015 of which only 224,000 (95% UI 216,000-237,000) were diagnosed, implying that 43% (172,000 of 397,000) were never diagnosed -- ranging from 3% in western Europe and North America to 49% in south Asia and 57% in western Africa. This entry's ~90,000 global childhood leukaemia diagnoses derived from IICC-3 registry rates is therefore a count of children who reach a registry, not of children who get leukaemia, and the shortfall is concentrated in exactly the highest-birth regions where this entry already notes registry coverage is weakest. Abstract read via PubMed; full text not read. Ward ZJ et al. 2019
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset
There are approximately 132 million live births worldwide per year, which sustained across fifteen single-year cohorts gives a global 0-14 population of roughly 2 billion.
World Population Prospects 2024: Summary of Results · United Nations Department of Economic and Social Affairs, Population Division · United Nations 2024What this evidence cannot show
Multiplying current births by fifteen ignores both past higher birth cohorts and child mortality; the two errors partly cancel but the product is an approximation, not the UN's own published 0-14 population.
Evidence that argues against this (1)
- The same simulation gives a better-founded denominator-free forward estimate than births-times-fifteen: 6.7 million (95% UI 6.3-7.2) cases of childhood cancer worldwide between 2015 and 2030, of which 2.9 million (95% UI 2.7-3.3) are projected to be missed at current levels of health-system performance. Using that projection rather than a rate-times-population product avoids the acknowledged cancellation-of-errors problem in this evidence item, and it also changes the entry's framing argument: if 43% of cases are undiagnosed, the 'childhood leukaemia platform looks fundable' calculation is being made on a market that is under-counted by a similar margin outside high-income countries. Abstract read via PubMed. Ward ZJ et al. 2019
Counter-evidence: Lee 2022 (PMID 35084434) shows DUX4-rearrangement frequency varies with genetic ancestry (East Asian OR 1.30, African OR 0.70) across 2428 children, so the 4% figure is population-specific
Counter-evidence (Ward 2019, PMID 30824204): registries recorded 224,000 of an estimated 397,000 global childhood cancers in 2015, so 43% went undiagnosed (3% in western Europe and North America, 57% in western Africa). Registry-rate-times-population gives diagnosed cases, not incident cases, and the shortfall sits in the highest-birth regions.
- supports →How many children this affects
resolves the 3-per-million versus 41-per-million denominator confusion