Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
, 392 (10159), 2052-2090

Forecasting Life Expectancy, Years of Life Lost, and All-Cause and Cause-Specific Mortality for 250 Causes of Death: Reference and Alternative Scenarios for 2016-40 for 195 Countries and Territories

Affiliations

Forecasting Life Expectancy, Years of Life Lost, and All-Cause and Cause-Specific Mortality for 250 Causes of Death: Reference and Alternative Scenarios for 2016-40 for 195 Countries and Territories

Kyle J Foreman et al. Lancet.

Abstract

Background: Understanding potential trajectories in health and drivers of health is crucial to guiding long-term investments and policy implementation. Past work on forecasting has provided an incomplete landscape of future health scenarios, highlighting a need for a more robust modelling platform from which policy options and potential health trajectories can be assessed. This study provides a novel approach to modelling life expectancy, all-cause mortality and cause of death forecasts -and alternative future scenarios-for 250 causes of death from 2016 to 2040 in 195 countries and territories.

Methods: We modelled 250 causes and cause groups organised by the Global Burden of Diseases, Injuries, and Risk Factors Study (GBD) hierarchical cause structure, using GBD 2016 estimates from 1990-2016, to generate predictions for 2017-40. Our modelling framework used data from the GBD 2016 study to systematically account for the relationships between risk factors and health outcomes for 79 independent drivers of health. We developed a three-component model of cause-specific mortality: a component due to changes in risk factors and select interventions; the underlying mortality rate for each cause that is a function of income per capita, educational attainment, and total fertility rate under 25 years and time; and an autoregressive integrated moving average model for unexplained changes correlated with time. We assessed the performance by fitting models with data from 1990-2006 and using these to forecast for 2007-16. Our final model used for generating forecasts and alternative scenarios was fitted to data from 1990-2016. We used this model for 195 countries and territories to generate a reference scenario or forecast through 2040 for each measure by location. Additionally, we generated better health and worse health scenarios based on the 85th and 15th percentiles, respectively, of annualised rates of change across location-years for all the GBD risk factors, income per person, educational attainment, select intervention coverage, and total fertility rate under 25 years in the past. We used the model to generate all-cause age-sex specific mortality, life expectancy, and years of life lost (YLLs) for 250 causes. Scenarios for fertility were also generated and used in a cohort component model to generate population scenarios. For each reference forecast, better health, and worse health scenarios, we generated estimates of mortality and YLLs attributable to each risk factor in the future.

Findings: Globally, most independent drivers of health were forecast to improve by 2040, but 36 were forecast to worsen. As shown by the better health scenarios, greater progress might be possible, yet for some drivers such as high body-mass index (BMI), their toll will rise in the absence of intervention. We forecasted global life expectancy to increase by 4·4 years (95% UI 2·2 to 6·4) for men and 4·4 years (2·1 to 6·4) for women by 2040, but based on better and worse health scenarios, trajectories could range from a gain of 7·8 years (5·9 to 9·8) to a non-significant loss of 0·4 years (-2·8 to 2·2) for men, and an increase of 7·2 years (5·3 to 9·1) to essentially no change (0·1 years [-2·7 to 2·5]) for women. In 2040, Japan, Singapore, Spain, and Switzerland had a forecasted life expectancy exceeding 85 years for both sexes, and 59 countries including China were projected to surpass a life expectancy of 80 years by 2040. At the same time, Central African Republic, Lesotho, Somalia, and Zimbabwe had projected life expectancies below 65 years in 2040, indicating global disparities in survival are likely to persist if current trends hold. Forecasted YLLs showed a rising toll from several non-communicable diseases (NCDs), partly driven by population growth and ageing. Differences between the reference forecast and alternative scenarios were most striking for HIV/AIDS, for which a potential increase of 120·2% (95% UI 67·2-190·3) in YLLs (nearly 118 million) was projected globally from 2016-40 under the worse health scenario. Compared with 2016, NCDs were forecast to account for a greater proportion of YLLs in all GBD regions by 2040 (67·3% of YLLs [95% UI 61·9-72·3] globally); nonetheless, in many lower-income countries, communicable, maternal, neonatal, and nutritional (CMNN) diseases still accounted for a large share of YLLs in 2040 (eg, 53·5% of YLLs [95% UI 48·3-58·5] in Sub-Saharan Africa). There were large gaps for many health risks between the reference forecast and better health scenario for attributable YLLs. In most countries, metabolic risks amenable to health care (eg, high blood pressure and high plasma fasting glucose) and risks best targeted by population-level or intersectoral interventions (eg, tobacco, high BMI, and ambient particulate matter pollution) had some of the largest differences between reference and better health scenarios. The main exception was sub-Saharan Africa, where many risks associated with poverty and lower levels of development (eg, unsafe water and sanitation, household air pollution, and child malnutrition) were projected to still account for substantive disparities between reference and better health scenarios in 2040.

Interpretation: With the present study, we provide a robust, flexible forecasting platform from which reference forecasts and alternative health scenarios can be explored in relation to a wide range of independent drivers of health. Our reference forecast points to overall improvements through 2040 in most countries, yet the range found across better and worse health scenarios renders a precarious vision of the future-a world with accelerating progress from technical innovation but with the potential for worsening health outcomes in the absence of deliberate policy action. For some causes of YLLs, large differences between the reference forecast and alternative scenarios reflect the opportunity to accelerate gains if countries move their trajectories toward better health scenarios-or alarming challenges if countries fall behind their reference forecasts. Generally, decision makers should plan for the likely continued shift toward NCDs and target resources toward the modifiable risks that drive substantial premature mortality. If such modifiable risks are prioritised today, there is opportunity to reduce avoidable mortality in the future. However, CMNN causes and related risks will remain the predominant health priority among lower-income countries. Based on our 2040 worse health scenario, there is a real risk of HIV mortality rebounding if countries lose momentum against the HIV epidemic, jeopardising decades of progress against the disease. Continued technical innovation and increased health spending, including development assistance for health targeted to the world's poorest people, are likely to remain vital components to charting a future where all populations can live full, healthy lives.

Funding: Bill & Melinda Gates Foundation.

Figures

Figure 1
Figure 1
Global distribution across countries of the most influential drivers of health Figure shows: (A) lag-distributed income, (B) educational attainment, (C) total fertility rate under 25 years, (D) Socio-demographic Index, (E) met need for contraception, (F) diphtheria, tetanus, pertussis dose 3 vaccination, (G) Haemophilus influenzae type B vaccination, (H) measles vaccination, (I) pneumococcal conjugate vaccination, and (J) rotavirus vaccination. Colours show distinction between year estimates in the past and scenarios in the future. Data are the 1990 Global Burden of Disease Study (GBD) estimate, the 2016 GBD estimate, the 2040 forecast, the 2040 better health scenario, and the 2040 worse health scenario. Shown are the top 20 risks ranked from 1–20 by the number of risk-attributable years of life lost in 2016, and ordered horizontally, across rows. The estimate for each country is the age-standardised mean value across both sexes, and a Gaussian kernel density estimator produced a distribution from the estimates for all countries. Vertical lines represent the global population-weighted average, whereas the density distribution gives each of the 195 countries equal weight. FPG=fasting plasma glucose. SIR=smoking impact ratio.
Figure 1
Figure 1
Global distribution across countries of the most influential drivers of health Figure shows: (A) lag-distributed income, (B) educational attainment, (C) total fertility rate under 25 years, (D) Socio-demographic Index, (E) met need for contraception, (F) diphtheria, tetanus, pertussis dose 3 vaccination, (G) Haemophilus influenzae type B vaccination, (H) measles vaccination, (I) pneumococcal conjugate vaccination, and (J) rotavirus vaccination. Colours show distinction between year estimates in the past and scenarios in the future. Data are the 1990 Global Burden of Disease Study (GBD) estimate, the 2016 GBD estimate, the 2040 forecast, the 2040 better health scenario, and the 2040 worse health scenario. Shown are the top 20 risks ranked from 1–20 by the number of risk-attributable years of life lost in 2016, and ordered horizontally, across rows. The estimate for each country is the age-standardised mean value across both sexes, and a Gaussian kernel density estimator produced a distribution from the estimates for all countries. Vertical lines represent the global population-weighted average, whereas the density distribution gives each of the 195 countries equal weight. FPG=fasting plasma glucose. SIR=smoking impact ratio.
Figure 1
Figure 1
Global distribution across countries of the most influential drivers of health Figure shows: (A) lag-distributed income, (B) educational attainment, (C) total fertility rate under 25 years, (D) Socio-demographic Index, (E) met need for contraception, (F) diphtheria, tetanus, pertussis dose 3 vaccination, (G) Haemophilus influenzae type B vaccination, (H) measles vaccination, (I) pneumococcal conjugate vaccination, and (J) rotavirus vaccination. Colours show distinction between year estimates in the past and scenarios in the future. Data are the 1990 Global Burden of Disease Study (GBD) estimate, the 2016 GBD estimate, the 2040 forecast, the 2040 better health scenario, and the 2040 worse health scenario. Shown are the top 20 risks ranked from 1–20 by the number of risk-attributable years of life lost in 2016, and ordered horizontally, across rows. The estimate for each country is the age-standardised mean value across both sexes, and a Gaussian kernel density estimator produced a distribution from the estimates for all countries. Vertical lines represent the global population-weighted average, whereas the density distribution gives each of the 195 countries equal weight. FPG=fasting plasma glucose. SIR=smoking impact ratio.
Figure 2
Figure 2
The three model components for China and Australia by age and sex: (A) China, male, all causes; (B) China, female, all causes; (C) Australia, male, all causes; (D) Australia, female, all causes. Figure shows the annualised rates of change (ARCs; errors bars represent 95% UI) for deaths from 1990 to 2016 broken down into the ARC for the underlying death rate and total death rate, and the ARC for the reference scenario for 2016 to 2040 by underlying rate, the underlying rate plus risk attributable mortality, and the underlying rate plus risk attributable mortality plus the autoregressive integrated moving average (ARIMA) component (reference scenario). The ARC is measured in terms of all-cause mortality by location, and rates of change are shown by age group on the x-axis. Rates of change are calculated from 2016 to 2040 in the forecasts and from 1990 to 2016 on past GBD estimates. GBD=Global Burden of Disease Study. EN=early neonatal. LN=late neonatal. PN=post-neonatal. Std=age standardised.
Figure 3
Figure 3
Global distribution of population in 2016 and 2040 reference forecasts, 2040 better health scenario, and 2040 worse health scenario Data are shown (A) by age and sex and by (B) total fertility rate, (C) life expectancy, and (D) population. Triangles within the population pyramid represent the mean age globally for males and females for corresponding years and forecasts. Inlays show total population forecasts, and associated inputs into the population forecast: fertility and life expectancy.
Figure 4
Figure 4
Leading 20 Level 3 causes of YLLs globally in 2016 and 2040 by rank order Figure shows percentage changes in the number of years of life lost (YLLs), all-age, and age-standardised rates. Rectangles are colour-coded based on Global Burden of Disease (GBD) Level 1 cause hierarchy: red=communicable, maternal, neonatal, and nutritional diseases; blue=non-communicable causes; green=injuries. Causes are connected by lines between time periods, with solid lines representing increasing relative rank and dashed lines representing decreasing rank. From 2016 to 2040, three measures of change are shown: percentage change in total number of YLLs, percentage change in the all-age YLL rate, and percentage change in the age-standardised YLL rate. Statistically significant changes are in bold. COPD=chronic obstructive pulmonary disease. Neonatal preterm birth=neonatal disorders due to preterm birth complications.
Figure 5
Figure 5
Evolution of leading causes of global under-5 deaths from 1990 to 2016 and in the 2040 reference forecast, 2040 better health scenario, and 2040 worse health scenario Estimates are reported in millions, with 1990 and 2016 estimates based on Global Burden of Disease Study (GBD) 2016 results. Neonatal preterm birth=neonatal disorders due to preterm birth complications. Congenital=congenital defects. Other neonatal=other neonatal disorders.
Figure 6
Figure 6
Leading 20 risk factors contributing to the global difference in risk-attributable YLLs between the 2040 reference forecast, 2040 better health scenario, and 2040 worse health scenario The differences between reference and better and worse health scenarios are grouped by Global Burden of Disease Study (GBD) Level 2 causes attributable to risks, which are colour coded to correspond with the causes contributing to the change in years of life lost (YLLs) between scenarios for each risk factor. Black solid vertical lines represent all-cause attributable YLLs in the 2040 reference forecast, red dashed vertical lines represent all-cause attributable YLLs in the 2040 worse health scenario, and green dashed vertical lines all-cause attributable YLLs in the 2040 better health scenario.
Figure 7
Figure 7
Global and GBD super-region life expectancy and relative contribution of Level 1 GBD cause groups to total YLLs, 1980–2040, for the reference forecast scenario Each ternary plot represents the relative contribution of years of life lost (YLLs) by Level 1 Global Burden of Disease Study (GBD) cause group in a given year and changes in life expectancy over time as depicted by colour-coded circles sized relative to life expectancy. The closer each circle is to a given corner of the ternary plot—communicable, maternal, neonatal, and nutritional (CMNN), non-communicable diseases (NCDs), and injuries—the greater is the proportion of YLLs due to that Level 1 GBD cause. If CMNN, NCDs, and injuries contributed equally to YLLs (ie, each a third), the circle would be positioned in the middle of the ternary plot. LE=life expectancy.
Figure 8
Figure 8
Life expectancy changes from 2016 to 2040, by the 21 GBD Level 2 causes of death and for each GBD region, for the (A) reference forecast, (B) better health scenario, and (C) worse health scenario Regions are based on Global Burden of Disease Study (GBD) 2016 location hierarchy. Blue vertical lines represent the estimated life expectancy in 2040 for both sexes, and orange vertical lines represent GBD 2016 estimated life expectancy in 2016 for both sexes. Horizontal rectangles are colour-coded by GBD Level 2 causes contributing to the difference in life expectancy between 2016 and 2040, with causes to the left of the orange line contributing to a reduction in life expectancy and causes to the right of the orange line contributing to an increase in life expectancy.
Figure 8
Figure 8
Life expectancy changes from 2016 to 2040, by the 21 GBD Level 2 causes of death and for each GBD region, for the (A) reference forecast, (B) better health scenario, and (C) worse health scenario Regions are based on Global Burden of Disease Study (GBD) 2016 location hierarchy. Blue vertical lines represent the estimated life expectancy in 2040 for both sexes, and orange vertical lines represent GBD 2016 estimated life expectancy in 2016 for both sexes. Horizontal rectangles are colour-coded by GBD Level 2 causes contributing to the difference in life expectancy between 2016 and 2040, with causes to the left of the orange line contributing to a reduction in life expectancy and causes to the right of the orange line contributing to an increase in life expectancy.
Figure 9
Figure 9
Map of life expectancy for both sexes in 2040 based on the reference forecast Key shown in years. ATG=Antigua and Barbuda. FSM=Federated States of Micronesia. LCA=Saint Lucia. TLS=Timor-Leste. TTO=Trinidad and Tobago. VCT=Saint Vincent and the Grenadines.
Figure 10
Figure 10
Map of the differences in life expectancy for both sexes from 2016 to 2040 based on the reference forecast Key shown in years. Legend shown in years. ATG=Antigua and Barbuda. FSM=Federated States of Micronesia. LCA=Saint Lucia. TLS=Timor-Leste. TTO=Trinidad and Tobago. VCT=Saint Vincent and the Grenadines.
Figure 11
Figure 11
Ten leading causes of YLLs in 2040, globally and by GBD region, based on the reference forecast Values are reported in thousands. Causes are listed at the Global Burden of Disease Study (GBD) Level 3 cause hierarchy from GBD 2016, and are colour coded in accordance with their Level 2 categorisation (shading represents GBD Level 2 cause hierarchy). YLLs=years of life lost.
Figure 12
Figure 12
Ten leading risk factors contributing to the largest differences in risk-attributable YLLs between the 2040 reference forecast and 2040 better health scenario, globally and by GBD region Values are reported in thousands. Risks are listed at the Global Burden of Disease Study (GBD) Level 3 risk hierarchy from GBD 2016, and are colour coded in accordance with their Level 1 categorisation (shading represents GBD Level 2 cause hierarchy). YLLs=years of life lost.

Comment in

Similar articles

See all similar articles

Cited by 63 articles

See all "Cited by" articles

References

    1. Lopez AD, Shibuya K, Rao C. Chronic obstructive pulmonary disease: current burden and future projections. Eur Respir J. 2006;27:397–412. - PubMed
    1. Guzman-Castillo M, Ahmadi-Abhari S, Bandosz P. Forecasted trends in disability and life expectancy in England and Wales up to 2025: a modelling study. Lancet Public Health. 2017;2:e307–e313. - PMC - PubMed
    1. Kontis V, Bennett JE, Mathers CD, Li G, Foreman K, Ezzati M. Future life expectancy in 35 industrialised countries: projections with a Bayesian model ensemble. Lancet. 2017;389:1323–1335. - PMC - PubMed
    1. Etkind SN, Bone AE, Gomes B. How many people will need palliative care in 2040? Past trends, future projections and implications for services. BMC Med. 2017;15:102. - PMC - PubMed
    1. Oestergaard MZ, Inoue M, Yoshida S. Neonatal mortality levels for 193 countries in 2009 with trends since 1990: a systematic analysis of progress, projections, and priorities. PLoS Med. 2011;8:e1001080. - PMC - PubMed

MeSH terms

Feedback