(p.215) Appendix 2 The Cost of Saving a Life
(p.215) Appendix 2 The Cost of Saving a Life
APPENDIX 1 has given an indication of the numbers of people whose lives are threatened by severe material deprivation. What is the cost of saving one of those lives, or preventing a threat to the life of one of those people? The estimates I have seen range from one dollar to $9 million. But those are not answers to the same question. In order to arrive at a meaningful answer, we need to give some thought to just what is being asked.
In Chapter 5 I mentioned some answers that are well towards the lower end of the range. These report the costs of actually administering some effective health interventions, which are extremely small. According to the Rehydration Project, $1 is the cost of administering oral rehydration therapy to treat a child for the dehydration caused by diarrhoea. This is estimated to be effective in 90 per cent of cases, for a condition that currently kills 1.6 million children under the age of 5 each year. The World Health Organization estimates that the cost of the drugs required for an effective course of treatment against tuberculosis can be as low as $10. And UNICEF's figure for vaccinating a child to provide protection ‘from Vitamin A deficiency and a host of deadly diseases, including diphtheria, pertussis, tetanus, polio, measles, childhood tuberculosis, hepatitis B, and Hib (Haemophilus influenzae type b), which is a major cause of pneumonia and meningitis’ is $20–30. In each case, these costs do not merely represent the price of the medicines, but include the staff costs and overheads required to administer them as well.
I also cited figures for emergency disaster relief. The cost of supplying people with food, shelter, sanitation, and basic medical care in emergency relief camps has been estimated by Oxfam at £50–150 for one person for six months (at today's prices), and at the time of writing, the Disasters Emergency Committee estimates that this costs £100 for one family for one month.
A major cause of mortality among the chronically poor is simple malnutrition: this currently kills 6.6 million children under the age of 5 each year. The UN Food and Agriculture Organization (FAO) estimates the cost of feeding the world's 214 million hungriest people at $5.2 billion per year, which is $24 each.
It is not uncommon to see figures such as these represented as the cost of saving a person's life: ‘Relief groups calculate the cost of saving a life in Afghanistan over the next six to eight months at $50.’ However, this is misleading, for several (p.216) reasons. (They are reasons why the agencies I referred to above do not make this kind of claim.) Let us start by looking at two of the more obvious ones.
The first reason is clearest in relation to vaccination and emergency disaster relief. To say that vaccinating children, or giving families emergency supplies, is saving their lives implies that they would have died without these actions. But of course that is not going to be true of everyone who receives these forms of assistance. This point is clearly acknowledged in many emergency relief operation evaluation reports, in explaining why it is not possible to say how much difference the activities of a given agency have made to overall mortality. The second point is the converse of the first. I cannot claim to have saved your life if I save you from one threat only for you to die from another. So it will be misleading to cite the low cost of treating one cause of child mortality as the cost of saving a life, if it is only the cost of treating one of many such causes.
In his book Living High and Letting Die, Peter Unger adopts an approach that addresses these two points. Using figures from the World Bank's Population Health and Nutrition Department and UNICEF, he calculates the annual health care costs of treating a ‘typically sick’ child between the ages of 2 and 6, adds figures to represent the costs of donor- and recipient-country administration, and arrives at a result (in 1996) of $200 per child. Unger's conclusion is that $200 is the cost of taking a ‘typically sick’ 2-year-old and giving him a greater than 90 per cent chance, at age 6, of living to 21.
TWO MORE PROBLEMS
Unger rightly points out that no amount of spending will guarantee that a 2-year-old will reach 21. He might have added that for any ‘typically sick’ 2-year-old, there is no way to guarantee that that child would not have survived to 21 without the intervention. However, similar points hold in any situation in which a person needs life-saving assistance. So if I know that, by spending $200 on a child with a high chance of dying unaided, I can ensure that he reaches an age at which his chances of reaching 21 are over 90 per cent, then it does seem that I can think of this as the cost of saving his life.
However, there remain two reasons why Unger's calculation will not show this. First, his method is to calculate the cost (based on a figure of $16 per year for local health care expenses) of treating a ‘typically sick’ 2-year-old each year until the age of 6, at which point his chances of reaching 21 are over 90 per cent. However, this is not the cost of ensuring that a 2-year-old reaches the age of 6. It is the cost of successfully treating those 2- to 6-year-olds who are successfully treated. The success rates for those treatments may well be high. But unless they are rates of 100 per cent, Unger's $200 figure will not be the cost of giving a 2-year-old a 90 per cent chance of reaching 21.
(p.217) Is that a serious problem? Not obviously. If $200 will significantly raise a 2-year-old's chance of reaching adulthood, that still provides a very strong reason for giving $200.
Another problem, though, is more significant. As I read him, Unger is suggesting that if I give $200 to an aid agency then (on average) one child's life will be saved. However, that is false for a separate reason, which affects all of the low-end estimates mentioned so far. To see this clearly, go back to the ORT figure of $1. This is the cost of administering ORT to a dehydrated child in a health clinic. But it is not the case that if I give an extra $1 to UNICEF, one extra child's life will be saved. There is no queue of dehydrated children at health clinics waiting for $1 donations to fund their treatment. The cost of ensuring that children do not die of dehydration is the cost of ensuring that they receive appropriate treatment when they need it. In part, of course, meeting this cost means ensuring that there are staffed health clinics containing rehydration salts. But the much more expensive part of this cost is the cost of ensuring that the most vulnerable children are at those clinics when they need to be. And the cost of ensuring this includes the cost of creating extra health facilities, securing the opportunities for the poorest people to use them, and supplying the basic forms of health education that ensure that they are used.
The more general point is this. Given the appropriate statistics, we can work out the average cost of treating those children who do get treated, and this cost may be very low. But what we want to know is the cost of extending treatment to those children who are not currently getting it: that is, the marginal cost of treating an extra child. And none of the calculations mentioned so far will help us to do that.
Our question is, What is the marginal cost of saving an extra life, rather than the average cost of saving lives so far? For answers to this question, we can turn to health economics for help, in its attempts to quantify the correlation between health spending and mortality. However, if we expect neat answers, we will be disappointed. Resolving the substantial questions debated by health economists, and doing so for all of the poverty-related diseases that are major causes of global mortality, is a mammoth task.
Two issues make attempts to quantify the correlation between health spending and mortality highly controversial, and explain the wide variation in the figures produced by such attempts. One is the difficulty of answering the question, in relation to any given health intervention, what would have happened without it. That is, what are the other variables we need to control for in order to establish the causal efficacy of a given intervention, and how do we devise a study that controls for them? And the other is the range of further, related assumptions that need to be made, and are made differently by different researchers, about the quality of the data from which they are working, about how to define the costs and assess the coverage of a given intervention, and about how to extrapolate from one set of local conditions to cover others.
(p.218) The result of this is a striking variation in the figures produced by these studies. To take a representative example, a 1991 survey of studies of the cost-effectiveness of anti-malaria interventions found that estimates of the cost of these interventions per year of life saved varied between $2.10 and $259. When this range of uncertainty is compounded for all the other major poverty-related causes of mortality, the resulting figure for protecting poor people from threats from all of those causes is even less clear. However, one thing that does seem plain is that the cost of saving extra lives will need to be thought of as being considerably higher than the low-end figures mentioned above.
One way to try to estimate a lower boundary for this cost is to take the World Bank-derived figure of $25–30 billion per year which the UNDP cites as the amount of increased aid spending it estimates as the cost of reaching the Millennium Development Goals for health. There are three of these goals: to reduce the under-5 mortality rate by two-thirds, to reduce the maternal mortality rate by three-quarters, and to halt and reverse the spread of HIV–AIDS, malaria, and tuberculosis, between 1990 and 2015. Using their current figures for these causes of mortality, and assuming current projections for the rate of population increase (since these goals concern the mortality rate, rather than the absolute incidence), I arrive at an implied figure of $2,500–3,000 per life saved.
However, the method used by the World Bank to arrive at their figure of $25–30 billion is, as before, to estimate the average cost of supplying resources to those already within the system, rather than the incremental cost of bringing those from outside the system into it. And the latter cost is going to be higher. The neediest people—the people who are not being reached by current health spending—are the hardest and most expensive to reach: they are the most physically isolated, the most educationally disadvantaged, and the likeliest to be pre-empted by other people competing for access to the same goods. (This is what poverty is: a powerlessness to get access to basic goods.)
What about an upper boundary? One way to try to establish that is to look at some of the sophisticated work that has been done in wealthy countries to estimate the cost of reducing mortality there, using reliable sources of statistical data. But here the figures run into the millions. For example, a recent authoritative study estimates the cost, through increased welfare spending, of reducing child mortality during the depression of the 1930s in the United States at $2.5 million per life saved—although it acknowledges that others' estimates have ranged as high as $9 million. This gives us an upper boundary for thinking about the costs of saving lives. Presumably the cost of lowering the mortality rate in today's poorest countries, where the per capita health spending is less than $20, is lower than the cost of doing that in a country with less poverty.
This does not get us very far. There are many questions to be asked about the figures that supply these upper and lower boundaries, and they still differ from each other by a factor of 1,000. Moreover, they do not give us guidance on how to think about the cost of saving extra lives through humanitarian relief, where relief operation evaluators are reluctant even to attempt to produce such figures. I am not going to attempt to go further here into the questions that would be needed to resolve these issues. To see why, we need to ask ourselves one further question.
(p.219) DOES THIS MATTER?
Answering these questions matters if you think that the case for requiring me to contribute to aid agencies comes from the fact that my giving money to an aid agency will itself make the difference between life and death for someone. However, in Chapter 4 I argued against that view. There is reason to doubt whether this is the right way to think about giving money to aid agencies; but nonetheless, there remains a strong case for thinking that making such contributions is morally required. If I do so, I am contributing to a collective action of saving lives. We ought collectively to be doing that, and I ought to be joining in.
This leads to two simple observations about the figures I have just been discussing. First, the case for requiring me to contribute comes from the importance of contributing to helping the people who are already being helped, and not from the importance of my performing an action that will lead to an extra person's being helped (desirable though that would obviously be). And even the smallest figures I have been discussing do serve to indicate the average cost of our collectively supplying important forms of help to those who need it. But secondly (as I pointed out in Section 5.3), even if the average cost of our collectively helping one needy person runs into the millions, my own share of that collective cost will still be very small, given the number of people who are contributing towards helping: a fraction of a dollar, rather than a fraction of a cent. And that is enough to show that the Extreme Demand would indeed be extremely demanding, even if the figure that properly represents the cost of saving one poverty-afflicted life is right at the top of the spectrum just discussed.
What does of course matter is whether there are effective things that we can collectively do to help the poor. I argued in Chapter 3 that the answer to this is Yes. And in thinking about the help we can give, we should not concentrate exclusively on threats to people's lives. I acknowledged this at the start of the book; and a major concern in Part II was to emphasize the range of interests apart from interests in life itself that ground requirements of beneficence.
For the figures on low-cost health interventions:
Rehydration Project, <http://www.rehydrate.org/>.
For the figures on disaster relief:
Disasters Emergency Committee, <http://www.dec.org.uk/index.cfm/asset_id, 999/index.html>.
For the FAO figure on feeding the hungry:
UNDP, Human Development Report 2003,90.(p.220)
For the quoted claim about $50 saving a life in Afghanistan:
Bob Geary, ‘Afghanistan Seen through the Eyes of an Aid Worker’.
Peter Unger, Living High and Letting Die: Our Illusion of Innocence,146–8.
For the figures on malaria:
José A. Najera, Bernhard Liese, and Jeffrey S. Hammer, Health Sector Priorities Review: Malaria.
On United States child mortality in 1930s:
Price V. Fishback, Michael R. Haines, and Shawn Kantor, The Welfare of Children during the Great Depression.
The source of the UNDP's costing of the Millennium Development Goals for health is:
Shantayanan Devarajan, Margaret J. Miller, and Eric V. Swanson, Goals for Development: History, Prospects and Costs,16.
(However, note that these authors point out at p. 21: ‘Estimates of the costs of meeting specific human development goals, such as those for education and health, are highly problematic. The relationship between public expenditures and outcomes is complex and empirical evidence from developing countries suggests only a weak link between public spending on education and school enrollments, or between health expenditures and mortality or disease.’)
For relief operation evaluation reports, the British Disasters Emergency Committee publishes the independent evaluation reports it has commissioned to evaluate the relief operations of its twelve member agencies, since 1998, against the principles of the Red Cross code:
Some further sources on the cost-effectiveness of aid:
Active Learning Network for Accountability and Performance in Humanitarian Action, <www.alnap.org/>.
Centre for Research on the Epidemiology of Disasters, >http://www.cred.be/>.
DAC Network on Development Evaluation, <http://www.oecd.org/department/0,2688,en_2649_34435_1_1_1_1_1,00.html>.
FAO Food Security and Agricultural Projects Analysis Service, <http://www.fao.org/es/ESA/fsecur_FSinf_en.htm>.
Overseas Development Institute, <http://www.odi.org.uk/>.