Projections of Global Health Outcomes from 2005 to 2060 Using the International

Futures Integrated Forecasting Model

October 18, 2010

Barry B. Hughes, Professor and Director, Frederick S. Pardee Center for International Futures,
Josef Korbel School of International Studies, University of Denver

Randall Kuhn, Assistant Professor and Director, Global Health Affairs Program, Josef Korbel
School of International Studies, University of Denver

Cecilia Mosca Peterson, Frederick S. Pardee Center for International Futures, Josef Korbel
School of International Studies, University of Denver

Dale S. Rothman, Associate Professor, Frederick S. Pardee Center for International Futures,
Josef Korbel School of International Studies, University of Denver

José Roberto Solórzano, Frederick S. Pardee Center for International Futures, Josef Korbel
School of International Studies, University of Denver

Colin D. Mathers, Senior Scientist, Evidence and Information for Policy Cluster, World Health
Organization

Janet R. Dickson, Frederick S. Pardee Center for International Futures, Josef Korbel School of
International Studies, University of Denver

Corresponding Author:

Randall Kuhn
Josef Korbel School of International Studies
2201 S. Gaylord St.
Denver, CO 80208
(303) 871-2061
rkuhn@du.edu

Running Head: Projections of Global Health Outcomes

1
2
Abstract

Background: Long-term forecasts of mortality and disease burdens are essential for setting
current and future health system priorities, yet few forecasts cover a wide range of nations over a
long time-span or situate changes in mortality into an integrated framework. Building on the
work of the Global Burden of Disease project (GBD), we developed an integrated health
forecasting model as part of the larger International Futures (IFs) modeling system.

Methods: IFs health begins with historical relationships between economic/social development
and cause-specific mortality used by GBD, but we build forecasts from endogenous projections
of these drivers, incorporating forward linkages from health outcomes themselves back to inputs
like population and economic growth. The hybrid IFs system adds alternative structural
formulations for causes not well served by regression models (e.g., HIV/AIDS) and accounts for
changes in proximate health risk factors. We forecast to 2100 but report findings to 2060.

Findings: The base model projects that deaths related to communicable disease (CD) will
decline by 50 percent while those related to both non-communicable diseases (NCD) and injuries
more than double. Considerable cross-national convergence in life expectancy is expected.
Climate-induced variations in agricultural yield cause surprisingly little excess childhood CD
mortality, though we do not explore other climate-health pathways. Optimistic and pessimistic
scenarios deviate considerably, with the former producing, in 2060, 39 million fewer deaths and
a 20 percent relative increase in GDP per capita, in spite of a billion additional people. South
Asia would experience the greatest mortality and economic benefit.

Interpretation: While reduction of CD risk factors remains critical over the short-term, long-
term success depends on targeting NCD and injury risk factors and broader social determinants
of health. Economic growth effects are always positive, but are modest and should not constitute
a singular rationale for health investment. Economic impacts could be enhanced by sequenced
investments in family planning, nutrition, and education.

Conclusions: Long-term, integrated forecasting advances our understanding of the connections

between health and other markers of human progress, offering powerful insight into our current
path and key points of leverage for future improvements.

3
Introduction

Long-term forecasts of mortality and disease burdens are essential for setting current and future
health system priorities, yet few forecasts cover a wide range of nations over a long time-span.
Even fewer situate changes in age-specific mortality rates into an integrated framework to
account for the effects of mortality variation on population size, age-structure, and drivers of
mortality such as income. This paper describes an approach to addressing this gap by building an
integrated, long-range health forecasting model into the existing International Futures (IFs)
global modeling system (1). IFs is a software tool whose central purpose is to facilitate
exploration of possible global futures through the creation and analysis of alternative scenarios.
Alongside health, IFs incorporates models of population, economics, education, energy, food and
agriculture, aspects of the environment, and socio-political change and represents dynamic
connections among them. We build on the groundbreaking work of WHO’s Global Burden of
Disease (GBD) project, which has produced the only published global forecasts of regional and
cause-specific health outcomes to date (2,3). GBD was not, however, designed to produce long-
term, integrated forecasts; available analyses extend to 2030, now only 20 years distant.

The need for long-term, integrated forecasting is evident in emerging health risks and population
trends. The epidemiologic transition in leading causes of death from communicable disease (CD)
to non-communicable disease (NCD) and injuries demands that models capture complex, long-
term relationships like the effects of global agriculture production on obesity-related mortality
decades later; the effects of infrastructural investment on road traffic accidents; and the potential
effects of anthropogenic climate change on a constellation of CD and NCD risks (3-5). The shift
from CDs to NCDs also relates to ongoing processes of population growth and aging, which
determine both the number and distribution of deaths in a society. Finally, the burden of disease
and population can have profound forward impacts on subsequent economic growth, thereby
further altering health trajectories in a synergistic fashion (6) .

Many wealthy nations produce long-term forecasts of age-sex-cause-specific death rates and the
economic consequences of aging, yet few are integrated to explore scenarios in which, for
example, a major shift in tobacco consumption could simultaneously affect mortality, morbidity,
population size/structure, and productivity. Few poor countries produce any long-range forecasts,
yet many are in the midst or on the threshold of the above transitions. Finally, no forecast
explores these issues in a global system in which rich and poor countries interact.

In spite of this dearth of forecasting tools, the global community has begun to set global health
targets that are both long-term and based on an integrated understanding of health risk. In 2009,
the WHO Commission on Social Determinants of Health (CSDH) set ambitious targets for the
reduction of levels and gaps in life expectancy, under-5 mortality (U5MR), and adult mortality
(AMR, the probability of dying between age 15 and 49) (7). CSDH stressed the role of the social
environment in driving mortality, yet no model exists to evaluate these pathways.

To address this gap, we extend the work of GBD in a number of ways. IFs forecasts to 2100,
allows end-users to explore country-level outcomes, embeds mortality and morbidity patterns
within larger global systems, forecasts proximate drivers (e.g., obesity, child underweight), and
replaces some regression-based formulations with richer structural formulations. The flexible

4
structure of IFs allows users to vary model assumptions and undertake alternative scenario
analysis. This paper introduces the IFs health model and presents results from our base case,
optimistic, and pessimistic scenarios out to 2060. The model is freely available online at
www.ifs.du.edu, along with complete technical documentation and results (8,9).

Methods

GBD broke new ground with its forecasting approach, measures, and methodology. It was the
first multi-country forecast to disaggregate mortality into multiple causes of death, important
because the driver-outcome relationships vary with cause of death as well as with age and sex.
(10). Rather than relying heavily on extrapolative techniques, GBD employed a structural,
regression-based projection model using exogenous projections of independent distal-driver
variables to forecast health outcomes. This approach led to the development of three alternative
scenarios of the future mortality and morbidity for over 100 diseases based on differing income
and education assumptions for the eight global regions of their analysis. Mathers and Loncar
(2006) incorporated more extensive data, updated driver-variable forecasts, created regression
models specific to low- and lower-middle-income countries, and developed separate projection
models for diseases such as HIV/AIDS that are not easily forecast using distal drivers (3).

We take another step forward by integrating GDB formulations into the existing IFs forecasting
framework and explicitly capturing many of the linkages between health and other dimensions of
human progress. IFs draws upon standard modeling approaches from a wide range of disciplines
including economics, politics, population, agriculture, energy, technology, education, and the
environment. For example , the demographic model incorporates a true “cohort-component”
representation, tracking country-specific populations and events (including birth, death, and
migration) over time by age and sex (11). IFs Global Health begins with the distal driver-
outcome formulations developed for the GBD project. These can be represented as:

ln( M a ,k ,i , R ) = C a , k ,i + β 1 * ln(YR ) + β 2 * ln( HC R ) + β 3 * (ln(YR )) 2 + β 4 * T + β 5 * ln( SI a , k , R ) (1)

where M is Mortality Level in deaths per 100,000 for a given age group a, sex k, cause i and
country or Region R; Y is GDP per capita; HC is Total Years of Adult Education (for adults 25
and older) ; T is time (Year-1900); and SI is Smoking Impact. Deaths are distributed from GBD's
large age categories into five-year age categories (up to age 100, infants separated from 1-4) so
that decedents can be removed from population age-sex structures. While linkage of health
outcomes into our existing cohort-component system is itself a substantial advancement, we
further incorporate improved forecasts of distal and proximate health drivers, structural
representations of key health issues, and forward linkages from health to population and
development, as shown in Figure 1.

In contrast to GBD, IFs includes endogenous forecasts of distal drivers of health. The existing
IFs framework already included the basic linkages required to integrate changes in health to
changes in the drivers of health, for instance the effect of an additional person on the labor-
capital ratio or an additional elder or child on the dependency ratio. We update that system to
address a broader range of pathways through which mortality and disease are known to impact
economic growth (6,12,13), including linkages from reduced mortality/morbidity to reduced

5
fertility (14,15), increased human capital and productivity (16,17), and increased capital stocks
(18,19), as in recent general equilibrium modeling papers by economists such as Weil (18).

Structural formulations capture the broader structural determinants of mortality and ensure stable
results at least to 2060. The distal-driver approach may perform poorly when mortality rates do
not monotonically increase or decrease, often because a complex and perhaps sequential set of
factors drive morbidity and mortality patterns. For HIV/AIDS, a two-stage forecast models rising
and falling prevalence around a peak year drawn from the Spectrum Policy Modeling System
and captures secular trends in case-fatality rates (20). To capture the transitional nature of road-
traffic accident deaths, projections are tied to the growth of the vehicle fleet (strongly related to
income growth, but saturating at higher levels) and to the declining rate of accidents and deaths
per vehicle (related to income and infrastructural investment). Structural models also forecast
smoking impact and the effect of health systems on U5MR.

Finally, distinct proximate driver forecasts capture some of the immediate pathways through
which distal drivers impact mortality (21). Even as traditional risk factors such as undernutrition
and unsafe water/sanitation decline with rising income, other risks such as smoking and obesity
may rise and then possibly fall (22-26). Proximate risk factors are also subject to programmatic
intervention, suggesting a need to model their impacts independent of levels of societal
development (22). We therefore explicitly model eight risk factors identified in the Comparative
Risk Assessment (CRA) project: childhood underweight; body mass index; smoking; unsafe
water, sanitation, hygiene; urban air pollution; indoor air pollution from household use of solid
fuels; global climate change; and vehicle ownership and fatality rate (23). We do not address all
risk factors or all health outcomes related to the selected risk factors due to limitations of data
and knowledge. For instance, we model the effects of anthropogenic climate change only on
undernutrition, but not on vector-borne disease, heat-related illness, or natural disaster risk.

We build our approach using relative risks (RRs) to estimate the population attributable fraction
(PAF) of mortality associated with a risk factor. We adjust base mortality forecasts to account for
deviations in risk factor prevalence between detailed risk factor forecasts emerging from the IFs
system and forecasts based on distal drivers alone. For example, we estimate child undernutrition
using a model of food availability that accounts for demand factors such as GDP per capita and
supply factors including technology, weather, and climate change. Looking deeper, national
changes in temperature and precipitation are determined from global temperature changes, which
are driven by emissions of carbon dioxide from land use change and fossil fuel consumption.

Findings

Base Case Forecast

Figure 2 depicts historical and forecast life expectancy at birth (LEB) by region, from 1960
through 2100. Trends for males and females suggest dramatic improvements in all regions. We
forecast particular improvement in poor regions, especially Sub-Saharan Africa (SSA), the result
of a rapid decline in HIV/AIDS and CDs more generally. Although most sources suggest that the
HIV/AIDS epidemic may have peaked, the recency of this trend limits our confidence in the
forecast, as we explore in our pessimistic scenario below. We forecast a gradual slowing of

6
improvements in life expectancy in high-income countries, particularly for males. Comparing
our results for 2050 to those from UN Population Division, we anticipate comparable but slightly
better life expectancy outcomes, with high-income countries performing about one year better
and poor countries performing up to two years better.

CSDH set the goal of reducing the gap in life expectancy at birth between the 60 longest- and
shortest-lived countries by 10 years, or from 18.8 years in 2000 to 8.8 years in 2040. Figure 3
depicts this gap historically and in the IFs base case forecast. This goal appears unlikely to be
met until after 2060 (in fact, not until near the end of the century). This is due to the aggressive
goals set by CSDH and continued advancement in the life expectancies of high income countries.

Prospects for achieving such ambitious goals will depend on continued reductions in CD
mortality and action against the rising burden of NCD and injuries mortality. Globally, IFs
forecasts a shift away from CD deaths to NCD (already the major cause category in 2005),
shown in Figure 4. We forecast a CD reduction of just over 40 percent by 2030 and almost 70
percent by 2060, in spite of substantial population increase. This is consistent with historical
patterns of progress against most CDs, though considerable uncertainty attends to the pace of
reduction in HIV/AIDS and malaria. Even in SSA, we forecast that the balance of deaths will
shift to NCDs by around 2030; by 2060 NCD deaths would outnumber CD by more than 5 to 1.
These shifts reflect changing age-specific death rates and an older population structure.

Scenario Analysis

To leverage the integrative qualities of the IFs system and address the well-known "optimism
bias" existing in many forecasts (24,25), our scenario analysis moves beyond the symmetrical,
upside-downside variations incorporated in other studies to consider realistic scenarios. We
incorporate two symmetric elements aimed at capturing variations in technology (via a 50%
increase or decrease in the pace of mortality reduction due to technology over time) and in the
proximate drivers of health (via a one standard error increase or decrease in each proximate risk
factor, phased in over time). To better capture potential positive human action above and beyond
our eight proximate drivers, we allow countries currently underperforming their projections, such
as the Russian Federation and South Africa, to gradually converge to expectation (26-28). Two
further adjustments capture a realistic pessimistic scenario, particularly for low income countries.
First, to account for lingering effects of the Great Recession (2008-2011 in the IFs base case), we
model reductions in GDP growth in all countries and greater reductions in low income countries.
Finally, although we introduce no direct change in biological assumptions, our pessimistic
scenario incorporates slowed reductions in CD mortality, particularly for HIV/AIDS (29).

The optimistic and pessimistic scenarios carry significant global implications. Annual global
deaths in the pessimistic scenario grow to be 34 million more by 2060 than in the optimistic
scenario. In terms of death rates, the gap is still larger, because population diverges markedly
between the two scenarios, from just over 9 billion (pessimistic) to just over 10 billion
(optimistic), compared to a base case value of 9.4 billion. Of 1 billion additional people in the
optimistic scenario, the great majority—about 800 million—are 65 and older, with an additional
236 million working-age adults and 39 million fewer people under 15. Due to population aging
and the high probability of some reduction in CD risks, both scenarios suggest an ongoing global

7
shift from CD to NCD burdens. In neither scenario do CDs in 2060 account for more than 12
percent of deaths, though CD burdens remain much higher in the pessimistic scenario.

Figure 5 depicts variation in critical life-course mortality probabilities based on forecast life table
estimates. While U5MR is a well-known target of the Millennium Development Goals,
continued aging and CD mortality reduction push our attention to trends in AMR. CSDH set
explicit targets for reducing U5MR by 95% and AMR by 50% in all country, gender, and social
groups between 2004 and 2040. To address a broader spectrum of life-course mortality trends,
we also track the life table probability of dying between age 60 and 79 as an indicator of older
adult mortality. Relative differences in mortality probabilities are very large across all regions,
with mortality probabilities in the two scenarios varying quite often by a factor of 2 or even
more. Differences are especially great for U5MR, particularly for SSA and South Asia. For low-
income countries, the pessimistic scenario is extremely pessimistic, with an actual increase in
child deaths in coming years and a large gap, exceeding three million child deaths annually for
most of the forecast horizon, between the patterns of that scenario and the optimistic one.

Scenarios variations in AMR are equally striking, with SSA and especially South Asia showing
the greatest deviation. For SSA, the scope of the HIV/AIDS epidemic is so sweeping and the
likelihood of some progress so great that adult mortality falls from 388 per 1000 in 2005 to 219
per 1000 in 2060 even in the pessimistic scenario; in the optimistic scenario it falls much further,
to 105 per 1000. In South Asia and in Europe and Central Asia, where NCDs largely drive adult
mortality, the two scenarios differ dramatically in the rate of progress over the next half-century.
In South Asia for instance, AMR in 2005 was 217 per 1000. In the optimistic scenario, South
Asia’s AMR plunges to 64 per 1000 in 2060, comparable to today’s high-income societies,
whereas in the pessimistic scenario the rate only falls to 165 per 1000.

Regional life expectancy also differs dramatically by scenario, as shown in Figure 6. In 2005,
aggregate life expectancy for SSA was 28 years behind that of high-income countries (52 versus
80) while South Asia experienced a 15-year deficit (65 versus 80). IFs forecasts considerable
convergence in the optimistic scenario, with SSA's disadvantage narrowing to 12 years by 2060
(80 versus 92) and South Asia narrowing to only 5 years (87 versus 92). In the pessimistic
scenario, however, SSA's life expectancy remains 23 years behind today's high-income countries
(63 versus 86). Variation is even more striking for South Asia, which in a pessimistic scenario
would remain 15 years behind high-income countries (71 versus 86), and would lose ground with
respect to SSA. This reflects both the high likelihood of some improvement in HIV/AIDS
mortality in SSA and the dependence of future gains in South Asia on NCD mortality reductions.

Proximate Risk Factor Variation

While variations in proximate risk factor prevalence included in the optimistic and pessimistic
scenarios weigh heavily on the pace of short-term mortality reductions, their effects tend to erode
over time. This is especially true for CD mortality, which is projected to decrease eventually
based on distal drivers alone but would decline much faster under favorable proximate driver
scenarios. We estimate, for instance, that between 2005 and 2060, 131.6 million cumulative CD
deaths, or 23.4% of total CD deaths, could be eliminated by gradually reducing four proximate
drivers (childhood underweight; unsafe water, sanitation, hygiene; indoor air pollution; global

8
climate change). This relatively high population attributable fraction masks substantial variation
over time, from 35.6% of CD deaths averted in 2010 to only 7.9% CD deaths in 2060.

Our forecast of substantial declines in baseline CD mortality and childhood underweight sets the
context for our climate change impact forecast. As noted above, we only estimate the effects of
climate change on childhood underweight through the pathway of diminished food production.
We analyze impacts in the context of a fully integrated social, economic, and environmental
structure. In our base run, the atmospheric concentration of carbon dioxide rises to 550 parts per
million in 2060, translating to a further increase in global temperature of 1.6 degrees Celsius
from 2005, with specific temperature and precipitation changes at country level. Figure 7
presents the projected effect of this additional climate change on mortality from communicable
diseases other than HIV/AIDS among children under 5. This particular result ignores any
potential positive carbon fertilization effects on yield. We employ a 10-year moving average to
smooth the effects of periodic mortality spikes in climate-affected countries with low food
reserves. Such spikes should be anticipated, but not necessarily in the exact years we forecast.

The annual number of additional child CD deaths rises to above 50,000 by 2030, peaking in 2050
at over 70,000, and declining thereafter. The vast majority of additional deaths occurs in South
Asia and SSA, both characterized by greater food insecurity and higher baseline levels of CD
mortality. The overall pattern reflects the increasing relative risks of CD mortality due to
childhood underweight as a result of falling yields due to climate change, but also declining
baseline levels of CD mortality and of childhood underweight due to increased food production.
This should not be taken as conclusive evidence of limited climate change impacts on mortality,
but it does point to the importance of accounting for baseline improvements in childhood
underweight, which is not done in a number of other climate impact forecasts (30).

Economic Impacts

One final contribution of an integrated forecast is the opportunity to evaluate the potential social
and economic impacts of health scenarios. Our analysis suggests that an optimistic health
scenario results in positive economic returns relative to the base case, in spite of the requirements
associated with extra population. Figure 8 presents the potential magnitude of the positive impact
by showing the ratios of GDP per capita of the optimistic scenario to the adjusted base case. In
all regions except East Asia and Pacific, the optimistic scenario increases per capita GDP relative
to the adjusted base case. The different result in East Asia and Pacific flows from the large
number of older adults (and of the elderly who are beyond the 60-79 age range) that China will
experience relative to its working-age population in coming years—most of the reduced
mortality for the region occurs in those age categories and intensifies the fiscal pressures the
elderly will likely place on the society. The same phenomenon appears to a lesser degree in high-
income countries, leading to an absence of economic difference between the two scenarios.

South Asia would benefit most in the optimistic scenario, followed by sub-Saharan Africa and
the Middle East and North Africa. The swing in GDP per capita for South Asia between the two
scenarios reaches 37 percent in 2060 in spite of increased population. South Asia’s relative gains
stem from the imminent arrival of demographic dividend cohorts into prime working ages; their

9
health will matter considerably. Sub-Saharan Africa would experience about a 22 percent swing
in GDP by 2060, while Middle East and North Africa would gain about 15 percent.

Simultaneous attention to other aspects of development, including fertility reduction and

increased food production (and therefore nutrition of larger populations), could further enhance
the gains shown for the optimistic scenario by itself. The incremental gain for sub-Saharan
Africa would be especially great, moving the 22 percent improvement in GDP per capita from
health alone to 36 percent.

The economic boost of the optimistic scenario accumulates over more than a generation, with
effects peaking for most regions and the world as a whole by 2060. Given the more rapid aging
of populations in the optimistic scenario, dependency ratios rise and economic gains diminish
after about 2040. Sub-Saharan Africa experiences peak relative economic gains in 2065.

Discussion and Conclusion

Although forecasting of human population size and characteristics routinely extends to mid-
century, and often to the end of the century or beyond, forecasting of health has for the most part
not looked beyond 2030, and has been relatively rare in general. Yet an increasing number of
global actors and governments take a long-term approach to setting goals for health, as with the
recent WHO Commission on Social Determinants of Health (CSDH), which set goals to 2040.
Societies and global actors not only want to understand the possible future health of citizens,
they want to know how to improve it.

Our analysis reinforces and extends our understanding of changing global and regional mortality
patterns, both the trends toward improvement almost everywhere and toward reduction in the
burden from communicable diseases and (relative) increase in those from non-communicable
diseases and injuries. As dramatic as improvements are in our base case, however, they will not
lead to achievement of the goals of the Commission on Social Determinants of Health except in
the optimistic scenario. We have found that convergence, the closing of the gaps in mortality
patterns between less- and more-developed countries is likely to occur very slowly in our base
case, quite substantially (but by no means fully) across the next half century in our optimistic
scenario, and not at all in our pessimistic scenario.

Because human action is such a large part of the difference in the assumptions that underlie the
optimistic and pessimistic scenarios, action on multiple fronts is critical to convergence,
particularly for regions such as South Asia that sit at the threshold of epidemiologic transition.
Our analysis shows how effective proximate drivers interventions can be over the short-term, but
these impacts diminish in the long-term. While proximate drivers should be added to our model
and new proximate targets will emerge, long-term success nonetheless requires higher-level
action, as spelled out by CSDH. Remaining excess risks are no doubt related to intersectoral
factors such as governance, health systems, and social environment (often called "super distal
drivers") that are currently captured in the convergence term of our optimistic scenario.

The benefits of an intersectoral, interdisciplinary understanding of health futures is illustrated

both in our exploration of climate change impacts on mortality and of economic impacts of

10
health. In the case of climate change, our relatively low estimates of additional child CD deaths
resulting from food insecurity illustrate the need for climate impact studies to account for
broader processes of change like the ongoing trend of improvements in childhood underweight.
While such improvements are no more inevitable than any of our other findings, they do reflect a
plausible trajectory that is not fully addressed by most existing climate impact studies. As
evidence improves, the IFs health forecast will incorporate further connections climate-related
outcomes and conditions such as vector-borne disease, heat-related illness, and natural disaster.

In the case of economic impacts, our model suggests a modest but positive contribution of health
to economic growth, particularly in South Asia and Sub-Saharan Africa. Although improvements
in health, by themselves, should be adequate to justify personal and social investments in them,
debate remains as to whether such investments also generate societal economic returns. Our
extensive analysis of possible pathways for such return, in combination with maintaining full
accounting of the diversion of such investments from other possible uses of funds, suggests that
they should in fact generate positive returns. These impacts could be enhanced if investments in
family planning, nutrition, and education were to dampen the magnitude and burden of short-
term population growth, particularly in areas yet to experience any fertility decline. The modest
level of economic return, however, suggests the need for care in using economic returns as a
simple justification for health investments; the best justification remains better and longer lives.

There is, of course, more that we wish to do or see done. That includes the extension of the set of
proximate drivers to include such drivers as alcohol use and physical activity level. We could
also better capture the social and political context of health, including representation of sub-
national variation (beyond sex differences), accounting for national inequality as a distal driver,
and modeling spatial and social transmission of health risks from nation to nation. Our scenario
analysis would benefit from deeper exploration of a variety of extreme events, including both
major plagues and dramatic breakthroughs in life extension. More immediately, IFs will soon be
able to evaluate the potential health and economic impacts of specific policy and governance
scenarios beginning with the Global Framework Convention on Tobacco Control. We also hope
extend and refine of our forward linkage analysis to provide a more systematic consideration of
the synergistic relationships between health, economics, and institutional change that help
explain recent economic breakthroughs in East Asia and Europe (6,12).

Finally, let us reiterate that the system used for this analysis is available for others to use in
replication or alternative analysis, and also to develop further.

Authors' contributions: ICMJE criteria for authorship read and met: RK BBH CMP DSR JRS
JRD CDM. Agree with the manuscript’s results and conclusions: RK BBH CMP DSR JRS JRD
CDM. Designed the experiments/the study: RK BBH CMP DSR JRS. Analyzed the data: BBH
CMP DSR JRS. Wrote the first draft of the paper: RK CMP. Contributed to the writing of the
paper: RK BBH CMP DSR JRD CDM.

Conflict of interest: none.

11
Role of funding source: This work was supported in part by the Frederick S. Pardee Center for
International Futures at the University of Denver. No funding bodies had any role in study
design, data collection and analysis, decision to publish, or preparation of the manuscript.

Ethics committee approval: not applicable.

12
References

1. Hughes BB, Hillebrand EE. Exploring And Shaping International Futures. Boulder, CO:
Paradigm Publishing; 2006.

2. World Health Organization. The Global Burden of Disease: 2004 Update. Geneva: World
Health Organization; 2008.

3. Mathers CD, Loncar D. Projections of Global Mortality and Burden of Disease from 2002 to
2030. PLoS Med. 2006;3(11):2011-2030.

4. Omran AR. The epidemiologic transition. A theory of the Epidemiology of population

change. 1971. Bull World Health Organ. 2001;79(2):161-170.

5. Prüss-Üstün A, Corvalán CF. Preventing Disease Through Healthy Environments: Towards

an Estimate of the Environmental Burden of Disease. Geneva: World Health Organization;
2006.

6. Sachs JD. Macroeconomics and health: investing in health for economic development:
report. World Health Organization; 2001.

7. Marmot M, Friel S, Bell R, Houweling TA, Taylor S. Closing the gap in a generation: health
equity through action on the social determinants of health. The Lancet. 2008 Nov
8;372(9650):1661-1669.

8. Hughes BB, Kuhn R, Peterson CM, Rothman DS, Solorzano JR, Dickson J. Improving
Global Health: Patterns in Potential Human Progress Volume 3. Boulder, CO: Paradigm
Publishers; 2011.

9. Hughes BB, Kuhn R, Peterson CM, Rothman DS, Solorzano JR, Dickson J. Forecasting
Global Health in IFs: Technical Documentation. Denver, CO: University of Denver; 2010.

10. Murray CJL, Lopez AD, editors. The Global Burden of Disease: A Comprehensive
Assessment of Mortality and Disability from Diseases, Injuries, and Risk Factors in 1990
and Projected to 2020. Cambridge, MA: Harvard University Press; 1996.

11. Preston SH, Heuveline P, Guillot M. Demography: Measuring and Modeling Population
Processes. Malden, MA: Blackwell Pub; 2001.

12. Fogel RW, Costa DL. A Theory of Technophysio Evolution, With Some Implications for
Forecasting Population, Health Care Costs, and Pension Costs. Demography. 1997
Feb;34(1):49-66.

13. Sachs J, Malaney P. The economic and social burden of malaria. Nature.
2002;415(6872):680–685.

13
14. Soares RR. Mortality reductions, educational attainment, and fertility choice. American
Economic Review. 2005;:580–601.

15. Soares RR. The effect of longevity on schooling and fertility: evidence from the Brazilian
Demographic and Health Survey. Journal of Population Economics. 2006;19(1):71–97.

16. Deaton A. Height, health, and development. Proceedings of the National Academy of
Sciences. 2007;104(33):13232.

17. Grantham-McGregor S, Cheung YB, Cueto S, Glewwe P, Richter L, Strupp B, et al. Child
Development in Developing Countries 1: Developmental Potential in the First 5 Years for
Children in Developing Countries. The Lancet. 2007;369(9555):60–70.

18. Weil DN. Accounting for the effect of health on economic growth. Quarterly Journal of
Economics. 2007 Aug;122(3):1265-1306.

19. Bloom DE, Canning D, Graham B. Longevity and life-cycle savings. The Scandinavian
Journal of Economics. 2003;:319–338.

20. Stover J, Johnson P, Zaba B, Zwahlen M, Dabis F, Ekpini RE. The Spectrum Projection
Package: Improvements in Estimating Mortality, ART Needs, PMTCT Impact and
Uncertainty Bounds. Sexually Transmitted Infections. 2008;84(Supplement 1):i24-i30.

21. Kunitz SJ. Explanations and ideologies of mortality patterns. Population and Development
Review. 1987;13(3):379-408.

22. Soper FL, Kerr JA. Building the health bridge: selections from the works of Fred L. Soper.
Indiana University Press; 1970.

23. Ezzati M, Lopez AD, Rodgers A, Murray CJL, editors. Comparative Quantification of
Health Risks: Global and Regional Burden of Disease Attributable to Selected Major Risk
Factors. Geneva: World Health Organization; 2004.

24. Armstrong JS. Principles of Forecasting: A Handbook for Researchers and Practitioners.
Boston: Kluwer Academic Publishers; 2001.

25. Coates J. Coming to Grips with the Future. Research-Technology Management.

2004;47(5):23-32.

26. Shkolnikov Giovanni A, Vladimir M. Causes of the Russian mortality crisis: evidence and
interpretations. World Development. 1998;26(11):1995-2011.

27. Zaridze D, Maximovitch D, Lazarev A, Igitov V, Boroda A, Boreham J, et al. Alcohol

poisoning is a main determinant of recent mortality trends in Russia: evidence from a
detailed analysis of mortality statistics and autopsies. International journal of epidemiology.
2009;38(1):143.

14
28. Kuhn R. Routes to Low Mortality in Poor Countries Revisited: A 25-Year Persepective.
2010;

29. Fauci A. Infectious diseases: considerations for the 21st century. Clinical Infectious
Diseases. 2001;32:675-685.

30. Nelson GC, Rosegrant MW, Koo J, Robertson R, Sulser T, Zhu T, et al. Climate Change
Impact on Agriculture and Costs of Adaptation. Washington DC: International Food Policy
Research Institute (IFPRI); 2009.

15