The world health report

Chapter 4

Putting it all together -- what is possible?

Estimates of the joint effects of selected risk factors

The multicausal nature of disease often provides a choice among different preventive strategies and offers great potential benefit from simultaneous interventions. For example, modest reductions in blood pressure, obesity, cholesterol and tobacco use would more than halve cardiovascular disease incidence, if these reductions were population-wide and simultaneous. This section includes an assessment of gains in healthy life expectancy attributable to the leading 20 risk factors considered here.

As outlined previously, typically, population attributable fractions add up to less than the sum of components, because many diseases are caused by more than one risk factor. This is shown graphically in Figure 4.11, which shows the individual and joint contributions of three major risk factors to each major burden of disease outcome groups (group I: communicable, maternal, perinatal and nutritional conditions; group II: noncommunicable conditions; and group III: injuries) in three broad combinations of regions -- demographically developed, developing low mortality, and developing high mortality. The size of each circle is proportional to the absolute disease burden.

This figure clearly shows how these selected major risks are responsible for a large fraction of current global disease burden, both across levels of development and type of outcome. It also shows how burden may be caused by more than one risk factor. The grouping by broad disease outcomes conceals some of the substantial population attributable fractions within the component clusters of disease. For example, of all childhood communicable diseases (including acute lower respiratory infection), 50% can be attributed to underweight, 23% to unsafe water, sanitation and hygiene, 13% to indoor smoke from solid fuels, and 63% to the joint effects of all three of these major risk factors. Similarly, 50% of cardiovascular diseases among those above the age of 30 years can be attributed to suboptimal blood pressure, 31% to high cholesterol and 14% to tobacco, yet the estimated joint effects of these three risks amount to about 65% of cardiovascular diseases in this group.

Using the assumptions outlined in Chapter 2, approximately 47% of global mortality can be attributed to the 20 leading risk factors and more than one-third attributed to the leading 10 risk factors. The likely impact of the 20 leading risks from the selected factors was estimated for 2000 in terms of potential gain in healthy life expectancy, as shown in Figure 4.12.

Had these risks not existed, then healthy life expectancy in 2000 might have been, on average, almost a decade greater globally. However, the gain varied considerably across regions, with the countries currently facing the world's largest risks to health having many times more healthy life years to gain than the richest countries. Thus the leading 20 risks were estimated to be responsible for 16 years lost in healthy life expectancy in AFR-E compared with slightly more than four years in WPR-A. Most of this was attributable to the leading few risks -- for example, about 14 years lost in healthy life expectancy in AFR-E and 11 in AFR-D were attributable to the leading five risks in those regions. Notable also were the high mortality European regions of EUR-B and EUR-C, with particularly large attributable burden of healthy life expectancy, principally as a result of their large burden resulting from tobacco, alcohol, cholesterol and other major risks for noncommunicable diseases.

Such joint estimates have considerable uncertainty associated with them. As well as the technical assumptions necessary in making these estimates with limited data, the time-related issues should also be considered, with sequential rather than simultaneous changes occurring in real life. Thus there is the capacity of improved health to beget health. For example, improvements in nutritional status of children in developing countries might well lead to improved ability to avoid and reduce other risks in adulthood as well as the large, immediate threats of communicable diseases. For these reasons, it seems likely that these are conservative estimates of joint effects of major risks on healthy life expectancy.

The distribution of risks across levels of poverty as measured in this report, both within and between regions, suggests they are likely to explain a large proportion of current inequity in healthy life expectancy. The multicausal nature of many diseases means that tackling major risks at a population-wide level offers opportunities to lessen these differentials, whatever their initial cause. The Commission on Macroeconomics and Health recently estimated that a 10% increase in life expectancy might increase GDP by 0.3% in the poorest countries of the world (1). It is clear that many different combinations of reductions in these major risks could increase healthy life expectancy by at least 10% in these countries, especially if they were simultaneous and population-wide. Indeed, at least a quarter of all disease burden can be attributed to the leading three risks in high mortality developing areas and in developed regions, and at least one sixth in low mortality developing regions. Furthermore, these potential gains are averaged over a whole population, even though many people die from other causes. The average gain in healthy life expectancy would be much greater among those with averted events.

Estimates of avoidable burden

Current action to focus on risks to health can change the future but not alter the past. It is possible to avoid future disease burden, but nothing can be done about attributable burden. The main policy use of attributable burden estimates should therefore be to help assess avoidable burden. In addition to the uncertainty involved in estimating attributable burden, making estimates of avoidable burden is particularly challenging because of uncertainty concerning predictions in risk factors and burden, and reversibility of risks. Despite these reservations, the policy relevance of avoidable burden information is considerable and justifies making estimates, given that appropriate caution will be exercised regarding their uncertainty. To maximize policy relevance, estimates can be made particularly for small-to-moderate risk factor reductions; that is, those that are likely to be achievable in the short term. A full range of estimates is essential, however, since, for example, a 5% distributional transition for one risk factor may be cost-effective in one region, whereas a 50% distributional transition may be cost-effective in another. Similarly, in one region, the same resources might be required to achieve a distributional transition of 1% for one risk factor as to achieve a 10% transition for another. Wide ranges of risk reductions have been assessed in the following chapter. As an example, the likely effects of a 25% distributional transition are estimated: that is, a 25% transition from current levels towards the theoretical minimum that occurs in 2000 and is maintained relative to "business as usual" exposure projections.

In this chapter, business as usual, or "drift", was first estimated to calculate what attributable burden would be in future years if there were no change in current trends in risk factor levels and distributions. For example, without further action it is predicted that in 2020 the disease burden attributable to tobacco will be nearly double its current levels. Similarly, there will be a one-third increase in the loss of healthy life as a result of overweight and obesity in 2020 compared with 2000. In contrast, 130 million DALYs per year are currently attributable to underweight, while it is estimated that 90 million will occur from this risk in 2010 even with all the benefits of economic development. Avoidable burden estimates the effects of changes in terms of deviations in risk levels from these predictions. Thus, avoidable burden is defined here as the fraction of total disease burden in a particular year that could be avoided with a specific reduction in current and future exposure compared to predicted current trends. The main estimates here are for a 25% distributional transition -- roughly equated as a reduction of one quarter in current and future risk levels. The initial avoidable burden estimates are summarized in Tables 4.9 and 4.10 and Figure 4.13.

These estimates show, firstly, that underweight will remain one of the leading causes of avoidable burden in 2010 and 2020. This is despite the fact that the estimated global burdens attributable to childhood diseases, diarrhoea and other major causes of childhood mortality are expected to form a considerably lower proportion of the global disease burden in 2010 and 2020. For example, the business as usual trend for burden attributable to underweight suggests that it will be responsible for 90 million DALYs in 2010 and more than 60 million DALYs in 2020, with disease rates continuing to decline, but with increased population sizes. The risk factors of unsafe water, sanitation and hygiene, and indoor smoke from solid fuels assume lesser though still very substantial roles as causes of avoidable burden, as the exposure levels are predicted to decrease with economic development. The associated mortality and morbidity are also proportionally less important as a result of declining levels of related risk factors. Nonetheless, the avoidable burden remains substantial. Because these risks are high in the poor, both within and between countries, efforts to tackle them now are likely to reduce inequality significantly in the future.

The 10 leading risk factors in terms of avoidable burden in 2010 and 2020 are broadly similar to the 10 leading causes of attributable burden in 2000, although the ordering changes somewhat, reflecting expectations of demographic and social development. Most noticeably, the ranking of avoidable burden from reduction in unsafe sex is extremely high, making it the leading cause of avoidable burden and reflecting the benefits of preventing transmission and the continuing predicted epidemic of HIV/AIDS in some places where current effects are small but large increases may occur. If the benefits of reducing undernutrition and unsafe sex are additive, then a 25% reduction in these two risk factors alone would avoid an estimated 5% of global disease burden in 2010. These benefits would be substantially concentrated in sub-Saharan Africa, where the improvement in healthy life expectancy would be even greater.

The potential avoidable burden from decreases in the prevalence of unsafe sex are both substantial and rapid. For example, with a one-quarter reduction, a substantial number of deaths would be averted in 2010. These would mostly occur in young and middle-aged adults, and so the avoidable disease burden in terms of DALYs is even more substantial. Similarly, most of the benefits of reduction in alcohol consumption are rapidly achieved, since most of the attributable burden is to the result of injuries or neuropsychiatric diseases. One quarter reduction in alcohol use from its current trend could result in approximately 15 million fewer DALYs in 2010. Shifting distributions of blood pressure and cholesterol by only a quarter of the distance towards the theoretical minimum from their current trends (on average by 5--10 mmHg systolic pressure or 0.3--0.6 mmol/l total cholesterol) could avert considerable disease burden. Such population-wide reductions could together avert a loss of tens of millions of years of healthy life, with most or all of the full potential reached before 2005 and the effects being approximately additive. Strategies to achieve this are outlined in the following chapter.

Another important feature of these estimates is the importance of reduction in tobacco use now. The benefits, although more delayed than those resulting from reduction of some other risks, are very large and long-lasting. This is seen in the estimated tens of millions of healthy life years to be saved in 2010 and 2020 as a result of preventing and reducing tobacco use. The potential avoidable burden from some other risks closely maps the attributable burden. For the risk factors that predominantly affect cardiovascular diseases (inadequate fruit and vegetable intake, physical inactivity, overweight, blood pressure and cholesterol) and for alcohol, the amount of disease burden avoidable in 2010 from a 25% reduction starting in 2000 is about one-third of the attributable burden in 2000. This "avoidability" is lower for underweight, micronutrient deficiencies, unsafe water, sanitation and hygiene, and indoor smoke from solid fuels -- reflecting changing disease patterns as a result of the assumed demographic and social development -- and for tobacco use, reflecting delayed benefits from cessation. In contrast, it is much higher for unsafe sex, reflecting the benefits of reduced communicable disease transmission and the predicted continuing HIV/AIDS epidemic.

However, these analyses only map out the potential for gain -- what is required next are effective and cost-effective interventions to realize this potential.

The need for cost-effectiveness analyses

Large gains in health are not possible without focusing on efforts to diminish large threats to health. These analyses have shown some major causes of disease and injury burden. While the risk factors were selected from a countless array of possible risks, there are, of course, many other distal factors (for example, lack of education) or proximal factors (for example, fat intake or osteoporosis) that lead to substantial disease burden and were not estimated in this work. However, there may be relatively few others that have population attributable fractions of more than 5% of all disease and injury burden in a particular region.

While many big challenges to health remain, there are also many different ways of meeting them -- involving personal health interventions, non-personal health interventions, and intersectoral action. Not everything can be done in all settings, so some way of setting priorities needs to be found. The next chapter identifies costs and the impact on population health of a variety of interventions, as the basis on which to develop strategies to reduce risk.