Regular, comparable assessments of causes of death help set health priorities. Updated estimates now cover 292 causes across 204 countries and territories, plus 660 subnational locations, for each year from 1990 to 2023. The analysis combines death counts, rates and years of life lost (YLLs) and adds two intuitive measures: the probability of dying before age 70 years (70q0) and the mean age at death by cause and sex. Together, these metrics show where mortality has improved, where progress slowed during the pandemic and where outcomes differ from what would be expected given population age structures. The results support decisions on preventing premature mortality while highlighting persistent differences between regions, sexes and disease categories.
Updated Measurement Framework
Cause-specific mortality was estimated mainly using the Cause of Death Ensemble model (CODEm), which selects and combines models based on out-of-sample performance. YLLs were calculated by multiplying deaths in each cause–age–sex–location–year cell by standard life expectancy at each age. The 70q0 approach provides comparable estimates of premature mortality by cause and sex. To separate risk from population age structure, the analysis reports both observed mean age at death and an expected mean age obtained by applying global age-specific mortality rates to each population.
Methodological updates addressed pandemic-era data issues. A systematic correction reassigned deaths that were likely misclassified during COVID-19 surges in 2020–22. The COVID-19 framework incorporated more vital registration and preliminary data. All CODEm models were fitted to mortality rates rather than cause fractions. The evidence base rose to 55 761 sources, adding country-years of vital registration, surveillance and verbal autopsy, alongside other data types. The cause hierarchy expanded to include first-time estimates for ulcerative colitis, Crohn’s disease, thyroid disease, other endocrine, metabolic, blood and immune disease and electrocution.
Must Read: Global Shifts in Mortality and Health from 1970 to 2050
Estimates were produced for 21 regions and seven super-regions, for 25 age groups from birth to 95 years and older, by sex and all sexes combined, for every year since 1990. Uncertainty intervals were derived from 250 draws per metric. This full reanalysis supersedes prior rounds, giving comparable methods across the time series and enabling consistent trend assessment through the pandemic and into 2023.
Leading Causes and Premature Mortality
The pandemic temporarily reshaped the global ranking of causes of death. COVID-19 became the leading age-standardised cause at Level 3 in 2021 then fell to 20th by 2023. As COVID-19 receded, ischaemic heart disease and stroke returned to the top positions continuing their long-term decline in global age-standardised mortality rates. Age-standardised mortality also fell markedly for diarrhoeal diseases, tuberculosis, stomach cancer and measles signalling broad gains in prevention and treatment for communicable and vaccine-preventable conditions.
Progress was uneven across causes, sexes and locations. Conflict and terrorism contributed to notable sex differences in selected settings. Neonatal disorders saw large reductions in age-standardised YLL rates yet remained the leading global source of YLLs in most years except 2021 when COVID-19 temporarily dominated. Compared with 1990, total YLLs fell substantially for vaccine-preventable diseases such as diphtheria, pertussis, tetanus and measles.
Premature mortality trends provided further context. Across every super-region and region, all-cause 70q0 declined from 2000 to 2023 with wide variation in pace and level. For females, increases in 70q0 were linked to drug use disorders and to conflict and terrorism. For males, increases in 70q0 were associated with drug use disorders and diabetes. In sub-Saharan Africa, 70q0 increased for many non-communicable diseases (NCDs) and the mean age at death from NCDs was lower than expected for the super-region. In the high-income super-region, 70q0 rose for drug use disorders and the observed mean age at death was lower than expected indicating outcomes worse than would be anticipated after accounting for population age structure. These patterns point to causes and locations where reductions in premature mortality are lagging and where targeted interventions may deliver the greatest benefit.
Mean Age at Death and Regional Gaps
Mean age at death summarises how risk, timing and care interact. Globally, the all-cause mean age at death increased from 46·8 years in 1990 to 63·4 years in 2023. For males it rose from 45·4 to 61·2 years and for females from 48·5 to 65·9 years. These shifts indicate that deaths are occurring at older ages on average though gains are uneven.
In 2023, the high-income super-region recorded the highest all-cause mean age at death at 80·9 years for females and 74·8 years for males. Sub-Saharan Africa had the lowest mean ages at death at 38·0 years for females and 35·6 years for males. Comparing observed and expected mean ages at death helps identify outcomes that are better or worse than predicted by age structure and global risk patterns. Where observed mean age is lower than expected, modifiable factors may be at play including disease characteristics, access to care, socioeconomic conditions and exposures such as high blood pressure or the use of alcohol, tobacco or drugs.
Combining mean age at death with 70q0 and age-standardised rates provides complementary views of timing, risk and burden. Locations with lower-than-expected mean ages for prominent causes may need focused strategies even when headline death rates improve. Where observed mean ages meet or exceed expectations, it may suggest progress consistent with sociodemographic development.
Cause-specific estimates to 2023 show a return of ischaemic heart disease and stroke to the top of global rankings as COVID-19 declined from its 2021 peak position alongside sustained reductions in several communicable and vaccine-preventable diseases. Neonatal disorders remain the largest source of YLLs in most years while drug use disorders, diabetes and deaths from conflict and terrorism are linked to increases in premature mortality for particular populations. Rising mean ages at death mask stark regional gaps with the lowest values in sub-Saharan Africa and the highest in high-income settings. Using 70q0, mean age at death, YLLs and age-standardised rates together clarifies where outcomes diverge from expectations and where interventions could have the greatest impact supporting targeted data-driven policies to reduce preventable mortality.
Source: The Lancet
Image Credit: iStock
