# Cheap ways to save lives

Our research for saving lives in Africa comes out soon. While we’ve mostly stuck to finding the best organization – rather than generalizing about “how to save lives” – we’ve formed a couple informal opinions along the way, and this seems like a good time to share.

First off, I think bed nets are a little overrated as a cheap way to save lives. You may have heard sales pitches like this: “With just $10 you can send a bed net to stop mosquitoes in their tracks. Send a net. Save a life.” (From the Nothing but Nets Campaign.) Or, from Nicholas Kristof: “For$5 you can buy a family a large mosquito net and save several people from malaria.”

The thing is, while it’s true that $5-10 buys a net, that’s a very long way from saving a life. We’re looking at PSI’s net-selling program, which costs between$5-10/net including distribution, marketing, etc., and we’re finding the following things need to be considered:

• Relatively few children die from malaria. That means you have to give out a lot of nets to make a difference in a few lives. By our estimates, you need to distribute about 25 nets to reach a child who would have died. And, some nets are likely distributed to areas where the mortality rate is even lower.
• Not all nets that are distributed are eventually used properly. Distributing something for free is great because everyone gets it, but it’s also likely that many people choose not to use it, and if they do, may use it for something far different than what it’s intended for. Since PSI sells rather than gives nets, we’d guess the concern is smaller than usual with them, but it’s still a concern.
• Nets only save lives if at-risk people sleep under them. Malaria largely kills children under the age of five (as well as pregnant women). If the wrong family members are under the nets – or they’re just not using the nets, period – the nets won’t do any good. Proper use, awareness of who’s at risk, etc. can’t exactly be taken as given, although unlike other distribution campaigns we’ve seen, at least PSI has some data on how often the nets actually get used (about 70% of those who own them use them on a given night, in the region PSI studied).
• Sleeping under a net only reduces your risk of contracting malaria by 50%. You can still get bitten during the day, when you go to the bathroom, etc.
• A net doesn’t last forever. Nets tear, and some need to be retreated with insecticide to remain effective.

When we do all the math (not available yet but will be this coming Monday), we estimate that you don’t end up saving a life per net – you end up saving a life for every 70 or so nets at best, and maybe even more like 300 (so around $500 to$2000). Even PSI’s own estimate of lives saved comes out close to around 70 nets per life saved. And that’s looking at an organization whose customers purchase nets, defraying the cost somewhat and also probably reducing the number of nets that go to waste. Giving nets, though it may ultimately be effective, may involve even higher expenses.

Too expensive? Of course not, $2000 for a life is still a ridiculous deal. We just think you can do better. For example, I think condoms are pretty underrated as a way of saving lives from HIV/AIDS. A lot of the focus in HIV/AIDS is on antiretroviral therapy, an extremely expensive form of continual treatment for existing AIDS patients, but promoting safe sexual behavior of any kind can stop AIDS before it starts for many people, and has many additional benefits as well. Looking at the same organization (PSI), we think their condom marketing is saving lives for more like around$250-1000 a pop – and that isn’t including other benefits, such as:

• Reducing unwanted pregnancies (which also means reducing deaths in childbirth, actually one of the leading killers of African adults).
• Reducing sexually transmitted diseases aside from HIV/AIDS (also a killer and generally a pain in the neck).
• Slowing the spread of HIV/AIDS.

Of course, these numbers and claims rely on a lot of assumptions; our full research will be available within the week, so you’ll get the chance to check it all out then. These are just informal observations. But even though our estimates are rough, it seems to me that all things considered, a successful condom marketing program (or any program that increases safe sexual behavior) is a pretty good life saver, and probably better than the good old nets. I’ll take either one in a heartbeat over antiretriviral therapy, which was all the rage for a long time despite the fact that it’s one of the most expensive and complex ways to help people in a region littered with cheap and simple opportunities.

Then there are things I think might be even more cost-effective, but don’t know much about yet:

• Immunizations. They’re great because they fully protect a child from a disease for the rest of his life (unlike nets which only last for a few years), and as far as we know, the actual vaccine costs little. Measles, for which an effective vaccine already exists is still one of the leading causes of death among children in the developing world. We don’t know why vaccines don’t reach those children – it’s certainly plausible that the areas they live in are so hard to reach that the cost of vaccinating them is extremely high – but we’d like to know more about this, because it’s hard to find a simpler way to save a life.
• Vitamin A supplementation. Research suggests that providing vitamin A supplements (a product that costs less than a quarter) to children under five can reduce child mortaility by pretty huge amounts. We have some questions about the research, as well as the ease of expanding coverage (as in vaccinations), but it’s possible that (unsexy though it may be) reducing Vitamin A deficiency is one of the most promising ways to lower infant mortality and improve general health.

Thoughts?

• Richard Potter on November 28, 2007 at 7:22 am said:

I am contributing to http://www.Lubabalo.com, and greatly appreciate the research and thought provoking entries on the GiveWell blog.

Is there anything in your AIDS research with regard to education? Folks who have visited Africa on behalf of Lubabalo tell me that the Africans they came in contact with have a serious misunderstanding about the transmission of AIDS. They have been told that they were given AIDS by someone who had it when they had sex with that person. Then they are told that if they have sex with someone else, they will give AIDS to that person. True enough, but there seems to be a serious communication gap.

Apparently this has left some Africans with the impression that if they give their AIDS to someone else, then they will no longer have AIDS because they have given it away. Does any of your research confirm this?

• We think that education about HIV/AIDS (because of lack of knowledge and mistaken beliefs about transmission) and promotion of ways to prevent transmission (because even if people understand the way transmission works, they still may choose not to use condoms, get tested for HIV, etc.) are important methods for increasing the likely impact of any program. Organizations we’ve looked at that run programs to fight HIV/AIDS all use some form of education/promotion to supplement their activities (whether its distributing condoms or offering testing services). However, we don’t think that education or promotion alone is enough.

Good organizations monitor not only how many people they reach through their programming but also try to assess the degree to which they successfully change beliefs. The best organizations monitor that as well as changes in behavior (the goal of the education campaign): condom use, frequency of sex with high-risk partners (e.g., prostitutes), and number of partners.

Without seeing evidence of the effect of any education campaign (through changes in beliefs and behavior), I wouldn’t have confidence in its impact. It’s too easy – especially when dealing with cultures very different from out own – to teach and encourage and promote but have no impact (something we’ve seen from some of our applicants).

• Having recently run into this site, I am happy to see that you doing comparisons to see which of the many methods is more cost-effective in saving lives. The hard part is tracking the multitude of background facts necessary. For example, what fraction of the African people believe that they can give AIDS away by having sex.

I am waiting with bated breath to see how many of these facts you were able to assemble in preparation for your forthcoming African research paper.

Thanks for heading in the right direction.

• Carl Shulman on November 28, 2007 at 11:50 am said:

http://www.guttmacher.org/pubs/journals/3013404.html
Here’s an article on the tendency of family planning organizations to channel their resources to price subsidies at the expense of monitoring and evaluated education/promotion.

On nutrient supplementation, it’s worth noting a distinction between products that have to be provided to each individual and the fortification of products with more centralized production, e.g. putting iodine in salt, or iron in flour.

• Carl Shulman on November 28, 2007 at 11:55 am said:

Also, ‘cost per life saved’ is an inferior measure compared to ‘cost per life-year.’ Malaria strikes more children than does AIDS, and those children have more life to lose, etc.

• On ‘cost per life saved’ vs ‘cost per life-year’ it’s important to keep the following in mind:

1. Given the range of diseases that affect young children in sub-Saharan Africa, a child has roughly an 80% chance of reaching age 5. That means that saving a 1 yr old from malaria really only means giving him an ~85% chance of reaching age 5. On the other hand, once you make it past age 5, your chances of living to age ~50 are far greater. So, saving a 20-yr old from HIV may, in fact, save more expected years of life.
2. I don’t think the ‘years’ vs ‘lives’ distinction is practically that important when deciding which interventions to fund because of the inherent uncertainty range of any cost-effectiveness estimate. Except in extreme cases (like saving an 85-yr old), the difference in years saved for two people maxes out at 2-3x, well within the range that would significantly affect my decisions.

On other cost-effective interventions: the interventions I listed were a couple that some of our applicants implement and have high potential cost-effectiveness. I’m sure there are many more interventions with the same (or far better) cost-effectiveness, and I’d like to learn more about them. If you have info, please send it on.

• I really don’t like using ‘years’ more than ‘lives’. The following hypothetical is another reason. Imagine you can save one of two people: a 25-year-old mother with 5 children or a 5-year-old child. Knowing nothing else, I’d expect the 5-year-old to live 20 more years than the 25-year-old. But, that information seems unimportant to me relative to the fact that the mother has children, who depend on her for care. ‘Number of years left to live’ may be easy to measure but it doesn’t help me decide what to do in this case.

• I will defend “cost per life saved” over “cost per life-year” all day long. Here are some reasons.

1. Emotional relevance of metrics is important. That’s how we make sure that our conclusions accord with common sense and emotion, not just logic. Producing a “life-years” figure would leave us with an emotionally empty ranking that we couldn’t perform a “But does this feel right and good?” test on. I know some people think charity can be reduced to logic free of emotion, but we don’t think that.

2. Related: I do NOT think that saving a 5-year-old is more valuable than saving a 20-year-old. I have thoroughly gone over all the arguments for it and I prefer to save a 20-year-old. I am not a utilitarian, I don’t believe that the value of your life is equal to the sum of the values of the moments of your life, and I’d rather save an adult – someone who has an identity and a “story” that can avoid a tragic ending – than a 5-year-old. In fact saving a 5-year-old, who has very little in the way of memories and experiences and unique perspective on the world, seems philosophically closer to creating another person than it does to saving a person from death, and the former seems clearly less valuable to me (I think it’s a tragedy when someone is murdered, but I don’t think it’s a tragedy when someone passes up a chance to get pregnant).

3. Of course, lots of people disagree with my philosophical stance on #2, but that’s really the most important argument for our approach to metrics: our metrics can generally be converted into other metrics. Others can’t. For example, if you know how many people were saved from malaria, along with the context (which we generally provide) of how old these people are likely to be and what other risks they face, you can compute life-years saved on your own. On the other hand, if you provide me with a life-year calculation, I can’t turn that into anything that is emotionally relevant to me. We strive always to describe a charity’s impact by saying how many lives were changed and how they were changed, to make our metrics as convertible as possible into whatever other formal or informal metrics our readers prefer.

• Carl Shulman on November 28, 2007 at 12:54 pm said:

http://big.berkeley.edu/HIVFINAL2.PDF
A working paper comparing HIV prevention and treatment options under budget constraints.

The Global Fund progress report plays up the number of lives saved, but look at page 76: the lives saved have primarily been from tuberculosis. Further it hypes the number of people on antiretroviral therapy, but doesn’t factor in the long term costs of continuing to provide drugs to the same individuals in the future. It claims that it has prevented several times as many HIV infections as it has saved lives (temporarily, and without accounting for future costs) through antiretrovirals, and yet plans to massively expand its funding for the less effective option.

This is from an initiative that emphasizes ‘monitoring’ and ‘rigorous evaluation’ of the individual projects it funds, but ignores huge disproportions in the effectiveness of its different program areas.

http://www.joinred.com/manifesto/

• This comment is attributed to Carl Shulman, not me. For some reason he can’t post a comment (we are looking into this).

This is further reason to prefer a carefully targeted donation to paying a premium for a RED product.

http://www.copenhagenconsensus.com/Default.aspx?ID=223
The bibliography of the Copenhagen Consensus analysis of nutrient supplementation might be a good place to start on nutrients.

Elie,

On your point #1, that’s the reason for using life-years: future mortality, whether in childhood or senescence. On #2, there are actually numerous programs targeting the dying: Mother Theresa is famous for her ‘noble’ work providing hospice-type services. On the positive and negative externalities of lives, you want to factor them in, as you would the fact that people infected with HIV can spread the virus. If you use the same metric to talk about medical care in the U.S. you will often be comparing interventions that add 3-6 months of life versus others that add 40-60 years, and sometimes the costs will be similar.

Holden,

I agree that more is lost with the death of a 20 year old than of an embryo or infant. But to say that a typical 5 year old (with a sense of personal identity, the ability to recognize other minds, to use language, etc) is morally closer to a rock than a typical 20 year old would be very difficult for me. The lives of most people are filled with very standard experiences (unique in the proper names of the players, and particular combinations, but not radically so), unreflective compliance with biological drives, and habit. A similar argument could be made to disregard their welfare in favor of a few with uniquely excellent qualities.

Indeed, philosopher Thomas Hurka reads such a view in Nietzsche, ‘maximax perfectionism,’ that the moral imperative is to increase the height of the peaks of human excellence (perhaps in various fields, or perhaps on a single scale of excellence for lives lived), and the lives of the herd are irrelevant by comparison.

• gaverick on December 2, 2007 at 11:26 am said:

For your project, you might review the findings of DCP2: http://www.dcp2.org/pubs/DCP/2/FullText
I would not focus strictly on mortality, as there are a number of health conditions that cause misery but may not be fatal, and that can be cost-effectively addressed — eg blindness, anemia, TB. I suggest presenting your results in terms of both lives saved and DALYs averted. This is now standard in global health. For a priority-setting exercise of the sort you are performing, I would also try to represent policy interventions that may be effective but are often missing from CEA — for instance, removing barriers to agricultural trade and to immigration.

• First, some housekeeping. Some of the organizations we’ve reviwed for our “saving lives in Africa” cause have asked for a couple more days to review our reports before they go public. We’ve agreed to give them time to respond and correct any mistakes we’ve made. We plan on making our research public later this week.

Gaverick –

We’ve drawn heavily from the Disease Control Priorities Project in our research. The other links you’ve sent look interesting, thanks.

We don’t focus soley on mortality. The “saving lives in Africa” cause focuses on mortaility and debilitating conditions (e.g., blindness). We have tried to quantifiy morbidity, where possible, but haven’t had great luck understanding the distribution of outcomes for someone who, for example, contracts malaria.

On DALYs: we’re not crazy about them because we’re not convinced that “years” (as opposed to “lives”) is the thing that we’re trying to save (see comments above). And, most of the DALY output we’ve seen is relatively black-box (i.e., conclusions about cost-effectiveness and DALYs saved without laying out all the pieces that build up to the DALY estimate). This is particularly true of the DCPP (see this table on the cost-effectiveness of different HIV/AIDS interventions.

That said, given that DALYs are the standard in public health, we present some of our research in terms of DALYs.

We would also like to know more about the policy interventions you bring up, though our priority is reviewing interventions that individual donors might fund.

• Peter Burgess on December 29, 2007 at 12:53 am said:

Dear Colleagues

This dialog about performance metrics is interesting, and thank you for it. Please keep pressure on so that there are more organizations doing relief and development work that have such metrics … albeit not perfect.

You have correctly observed that too much of the available performance metrics is merely about how much activity has been undertaken, rather than how much result has been achieved.

I am associated with the Transparency and Accountability Network (Tr-Ac-Net) that is, inter alia, building a database about socio-economic results in a community … because in the end, a community has to survive on its own without external assistance in perpetuity. These metrics will eventually prove far more useful than those that are derived from implementing organizations that must, above all, justify their performance to their donors.

We are applying these concepts in communities where malaria programs are a priority, in cooperation with the Integrated Malaria Management Consortium (IMMC). Some preliminary cost effectiveness models suggest that an integrated approach to mosquito (vector) and malaria control can improve cost effectiveness significantly … and that most of the single intervention strategies are doomed to failure in the long term unless external funding is going to flow for ever.

It’s really good news that your GiveWell initiative has been launched, and is attracting attention.

Best wishes

Peter Burgess

• Crust on August 1, 2008 at 9:41 am said:

Elie, I think your argument in part (1) of #6 is incorrect. I don’t see how there can be reasonable assumptions under which “saving a 20-yr old from HIV may, in fact, save more expected years of life [than saving a one year old from malaria]”.

Let’s assume that conditional on reaching age 20, the life expectancy is 75 and that of the 15% of one year olds who don’t make age 5 none in fact make it to age 2 (both assumptions are favorable to your viewpoint). I’ll also assume all 5 year olds make it to age 20 (a close approximation to reality). Then the expected value of years of life saved for the 20 year old is 75-20 = 55. For the one year old, it’s 0.85*74 = 62.9 (0.85*(74)), appreciably more.

• Yes, you’re right.

• I’m glad I read this, I was thinking of buying a net after watching the Keifer Sutherland pitch during the debut of this year’s 24. I think my donation could be a little better directed.

• Remorque on February 25, 2009 at 10:43 am said:

Totaly agree with you. Those ads about helping people in Africa are just another marketing way to get your money.Help a homeless during a cold winter night a pay him a night a your local community center. This will have more impact!

• designerfliesen on March 2, 2009 at 12:59 pm said:

Yes I also believe, helping the people in you nearest envirnonment is the best what you can do. But if everybody thinks like that the poor people in africa really get much more problems.

• designerfliesen on March 2, 2009 at 1:00 pm said:

Yes I also believe, helping the people in you nearest envirnonment is the best what you can do. But if everybody thinks like that the poor people in africa really get much mor

• Fußmatte on April 22, 2009 at 10:26 am said:

A mosquito net costs just a few cents and not $5-10! • We’ve generally seen charities quoting$5-10. Do you have a source for the price you quote?

• Ian Turner on April 26, 2009 at 4:47 pm said:

In your report on population services international, you name a price ($650-$1000) on how much it costs to prevent one case of HIV/AIDS via condom marketing and distribution. But condoms don’t only affect HIV/AIDS: They also affect family size and population growth. Do you have any figures or thoughts on how much it costs to prevent one birth?

• Ian, we mention the additional benefit of condoms in our PSI review, but we haven’t researched cost-effectiveness figures for it.

PSI estimates \$58 per unintended pregnancy averted. See Pg ii of their 2006 Cost Effectiveness Report (http://www.psi.org/research/documents/health_impact/psi_cost_effectiveness_2006.pdf).