• How do you use cost-effectiveness in practice?

    In my recent post on preventing suicide, I reported that the cost of the Garrett Lee Smith (GLS) program is $450,000 per life saved. I compared this to the value of saving a life used by the Department of Transportation — $9,000,000 — and concluded that GLS is cost-effective and it should be reauthorized by the US Congress.

    But in email, Austin questioned whether this is all that cost-effectiveness requires.*

    What if GLS is cost-effective but it’s the 9,473,347th most cost-effective thing we could do that we don’t. What if we can only afford to do the first 134,283 of those things. Shouldn’t we do them in cost-effectiveness order? (Probably not, but this is what a pure CEA [Cost-Effectiveness Analysis] argument would lead to, right?)

    Then, despite GLS’s cost-effectiveness, we should argue that we ought NOT do it because doing so would entail a larger CEA-based opportunity cost. We might fail to save five people to save three.

    I agree with Austin that CEA says do things in CEA order. But even though I invoked CEA in my post, I haven’t justified suicide prevention in that way. Why not?

    1. I don’t know how to order the set of possible policies in CE rank. I don’t even have the list of possible policies. So I can’t carry out the “pursue policies in rank CEA order” procedure.
    2. I’m confident that among the policies the US Congress  funds, there are many that are less cost-effective than GLS, and some that fall below the CE criterion for the value of a life saved. So, if Congress shifted spending from (say) ethanol subsidies to GLS, that would save lives. So my argument that the cost of GLS is less than the value of a life saved at least points to the opportunity of a CE shift in spending.

    Funding GLS is may not be the optimal thing to do next. But given the limits on our knowledge and the current mix of our policies, GLS is a good thing to do. An old proverb: “Don’t make the best the enemy of the good.”

    But there are interesting rejoinders to both of my numbered points.

    Rejoinder to argument 1. “This is a straw man argument. You don’t have to know all possible policies and their CE rank. All you need to do is make your best effort to find the most CE thing you can do, and then do that. For example, the Give Well Foundation estimates that the cost per child-life-saved of an anti-malarial bed net is about $2800. If so, you could save more than 150 children in malarial regions for the cost of preventing one youth suicide in America.”

    This is the Effective Altruism argument and it is a formidable challenge. I hope to write about it soon.

    Rejoinder to argument 2. “This argument assumes that the substitution of a better policy (GLS) for a worse (ethanol subsidies) will occur. You don’t know this and there is little evidence that government works that way.”

    I agree with this rejoinder. It is not sufficient for the health policy community to think only about what the best policy is. We also need to think about the process of policy making and how to make it more evidence-informed. This blog is a small step toward that goal. The mission of TIE, as I understand it, is to provide a more effective way to translate and diffuse scientific knowledge so that it can be used by clinicians and policy makers.

    *Nota Bene: Austin is not arguing that we should make all of our decisions using CEA (see here). He’s asking this: if you are going to do CEA, how should it be done?


    Comments closed