Is (biomedical) research a good cause area to fund?

At first glance, funding biomedical research, or indeed research more generally, may seem like a good choice of cause area. In this post I look at the following:

• Cancer research appears to be substantially less cost-effective than the SoGive recommended charities; this is not based on a very rigorous assessment (see below). It seems that the cost per DALY (i.e. cost per extra year of healthy life) is around a few thousand dollars for cancer research). For comparison, the cost per DALY for a SoGive recommended charity could easily be below $100.
• Other research funding opportunities may do much more good per pound invested: the example of the development of a vaccine may be much better, with a cost per DALY that is roughly in line with a SoGive recommended charity

Why the big disparity between these different types of research? It seems to be because once an area becomes crowded, the effectiveness of the research hits diminishing marginal returns (and note that these calculations are based on averages, and not on the marginal next pound -- if returns are diminishing as I suspect, then these estimates may be generous). The idea that research may hit diminishing marginal returns has some support from, e.g. this Stanford paper (h/t Let's Fund).

If you would like to see more data suggesting that cancer research is indeed a crowded funding area, see /thinkingaboutcharity/2017/08/donating-to-medical-research-heres-why.html.

Overall, this suggests that the range of impactfulness of research is quite substantial. I would tentatively suggest that the research sector is less prone to work which has minimal impact, however this is a complex area which we don't have space to cover in this post.

To put this post within the context of other types of charitable intervention, see this post.


Cancer research

Below I set out three estimates for the cost per DALY (defined below) of cancer research:

• One recent study looking at one additional drug for a cancer site suggests a cost effectiveness of $1,635 to $2,820 per DALY
• One very rough model conducted by SoGive suggests an average cost per DALY of $2,000; this considered cancer research globally
• Another very rough model conducted by SoGive suggests an average cost per DALY of £4,000; this considered cancer research in the UK

I cannot stress enough that the SoGive models are extremely crude -- they are intended to give a rough sense of orders of magnitude, not to be a precise figure.

These findings seem to suggest that the cost effectiveness of cancer research is likely to be around a few thousand dollars per DALY.

A DALY is a somewhat technical term that is essentially equivalent to a year of healthy life. More from the WHO here or from wikipedia here.

Study looking at one additional drug for a cancer site 

One recent study looked at the impact of new cancer drugs launched and found that one additional drug for a cancer site launched during 2006–2010 is estimated to have reduced the number of burden of disease to due cancer by at that site by 5.8% with a cost-effectiveness of $1,635 per disability adjusted life year averted (i.e. healthy year of life gained).

Cancer research -- considering cancer research globally

This calc estimates cancer research cost effectiveness by comparing cancer research spend with overall change in cancer mortality. This assumes that all the change in cancer mortality is purely down to research, and not down to other (e.g. behavioural) factors. This very crude assumption is likely to dwarf the impact of the many other assumptions made in these calcs.

Cancer survival at 10 years has gone from 24% to 50% over the course of 40 years (source: https://www.cancerresearchuk.org/health-professional/cancer-statistics/survival/all-cancers-combined)

The total amount spent on cancer research globally was estimated to be EUR 14 billion in 2004/5 (source: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5527789/) which we can call about $15 billion

There are currently 100 million people in the world with cancer (source: https://ourworldindata.org/cancer) I have based my calculations on this (current) prevalence rate, which is generous (arguably a better choice is 45 million, which was the prevalence in 1990 (ibid) in which case the cost would be doubled.

In this calc I assume that the improvement in cancer mortality is entirely down to research. This is known to be overly generous -- part of it is because of lifestyle changes such as reductions in smoking rates (see, e.g., this US website https://www.cancer.org/latest-news/facts-and-figures-2019.html, which says "The drop in cancer mortality is mostly due to steady reductions in smoking and advances in early detection and treatment.")

Based on these assumptions, the cost per DALY averted comes to around $2000.

Note that although I'm saying cost per DALY averted, it's more like cost per year of life lost averted, however since a cancer diagnosis often led to a swift death in the early 70s (i.e. 50% would be dead within a year, source https://www.thelancet.com/journals/lancet/article/PIIS0140-6736(14)61396-9/fulltext), this seems reasonable

To see the calculations see this spreadsheet: https://docs.google.com/spreadsheets/d/1oNzIwQMiKwlZxj5_dpnsYmyzT-3ieXWVScp47wPSQkA/edit#gid=0

Cancer research -- considering cancer research globally

Just to repeat what I said above: This calc estimates cancer research cost effectiveness by comparing cancer research spend with overall change in cancer mortality. This assumes that all the change in cancer mortality is purely down to research, and not down to other (e.g. behavioural) factors. This very crude assumption is likely to dwarf the impact of the many other assumptions made in these calcs.

Prior to doing the global calc I had done a similar calc based on the UK only, which you can find here https://docs.google.com/spreadsheets/d/1YY46iDaOto6z5yjz3fXTxpES3ndkwASZCbBX6T83ESY/edit#gid=0. This found a cost per DALY of around £4,000. As the calculations here are so crude, I would consider this to be roughly the same number as the $2,000 found for the global calcs, and so this roughly corroborates the figure.


Note that I chose cancer partly because more is written about it, which makes it easier to research.


Development of a vaccine

It seems reasonable to assume that the cost of developing a vaccine is around $1 billion (see https://forum.effectivealtruism.org/posts/BjBmcfwg2awqPJLin/how-much-does-it-cost-to-research-and-develop-a-vaccine)

If that vaccine were to tackle something like TB or malaria, the current DALY burden is around 45 million DALYs (source: http://ihmeuw.org/4spa  )

This suggests a cost per daly of around $22, assuming that it leads to full elimination

If you then assume that the vaccine eliminates half of the disease (and this piece refers to 60% https://forum.effectivealtruism.org/posts/ERptCBkG5Xdmuywd9/how-beneficial-have-vaccines-been), this changes the estimate to $44.

More arguments to suggest that research can be cost effective

Lastly, here's a quote from https://lets-fund.org/better-science/, which also supports the view that research can be highly cost-effective:

For instance, there were almost 1 million AIDS-related deaths in 2017 globally[12]. Thus, a naïve back-of-the-envelope calculation suggests that speeding up the invention of an affordable cure or vaccine by one year might have the potential to save a million lives. Even if the costs of speeding up the arrival of a vaccine was $1 billion, this would lead to a marginal cost-effectiveness of less than $1000 per life saved ($1 billion ÷ 1 million). It is marginal because about $1 billion is already being spent on HIV/AIDS vaccine research annually[13] and so the average cost-effectiveness would be much lower. This model is just a crude oversimplification of a very complex issue intended to give a rough idea of orders of magnitude, but it is roughly in line with estimates of both historical cost-effectiveness of vaccination roll-out (e.g., smallpox eradication was fantastically cost-effective with some estimates suggesting that the cost per death averted was less $100[14]) and also estimates of future vaccine development (see [15],[16] for more on the cost-effectiveness of developing vaccines).