Skip to content

The flu paradox: is the WHO focusing too little on flu?

The WHO is in the news these days thanks to the H1N1 epidemic (alias the swine flu, or the Colbert flu), and it is doing an admirable job coordinating various national agencies in fighting a pandemic. Historically it has been at the forefront of fighting epidemic disease, whether tuberculosis or AIDS. However, since Gro Harlem Brundtland's director-generalship 1998-2003 there has been an increased emphasis on public health, in particular fighting alcohol and tobacco use but also traffic accidents. Has the WHO aimed at the right or wrong problems?

The goal of public health is to prevent disease, prolong life and promote health in a collective way: it aims at population-level health issues rather than individual-level issues, and is usually implemented by societies and organisations. From this perspective it might not matter what kind of public health problem is attacked, as long as the number of deaths and amount of suffering that can be reduced is the biggest given the limited resources of society.

Overall, in 2002, noncommunicable conditions such as cancer and
injuries were responsible for 33 million deaths (cardiovascular disease kills 17 million per year), while communicable
conditions killed 18.4 million. Tobacco kills 4 million, alcohol is probably responsible for about 1.5 million, while AIDS causes 2 million deaths and malaria one million.  The annual flu epidemics cause between 250,000 and 500,000 deaths per year.

By this scale flu does not seem to merit much effort. It is certainly a major killer, worse than all tropical diseases put together, but efforts aimed at reducing cardiovascular disease or smoking would seem to save many more people. Just a 2% reduction of deaths due to cardiovascular disease or a 13% reduction of smoking deaths would correspond to the whole burden of flu. Bunting lifestyle diseases and reducing accidents seems much more efficient than going after infectious diseases.

But there is a problem: the 1918 flu mayhave killed up to 50-100 million over a span of two years. While subsequent pandemics have been smaller (the Asian flu killed 1-1.5 million, the Hong Kong flu 0.75-1 million), there is no fundamental reason why another pandemic could not be as deadly. The problem with pandemic sizes is that they appear to have a so-called power-law distribution: very large pandemics are rare, but not so rare that we can ignore them. In some cases (technically: the exponent is between -2 and -1) the total number of dead in power-law distributed disasters (e.g. wars) is dominated not by the average yearly toll, but the number of victims in the biggest disaster (e.g. about 30% of all war dead over the last 2 centuries are due to WW II).

This has unsettling implications for pandemic flu. We do not know for certain that flu is distributed according to such a power-law, but it is possible. We also know (using statistical theory) that if it is, then past experience is not a reliable guide at all: the annual averages do not give any information about how big pandemics could be, and the fact that the 1918 flu had a certain extent does not tell us this is an upper limit. Worse, in this case it is possible to show that using past statistics systematically underestimates the average. Since pandemics are rare there is also a tendency to under-invest in protection – most years there is no pandemic, and maintaining worldwide vigilance is very expensive. We were lucky this time that the avian flu worry had improved responses: given a few years of no news and the response to the new flu would have been far weaker.

This suggests that if WHO aims at reducing the number of deaths globally, over long spans of time, then spending significant effort on preventing flu (and other) pandemics may be more important than many of the "stable" killers like malaria or alcohol – even when these killers most years are worse problems. It is the exceptional years that matter. WHO has a position and remit that makes it ideal for doing the kind of long-term vigilance local organisations may not be able to maintain. Seatbelts or tobacco may be low-hanging public health fruits, but it is the pandemics that WHO might really matter for.

Share on

2 Comment on this post

  1. I am puzzled by this post. Is the question whether it is ethical to ignore the most efficient approach to public health? I read the above post as criticizing Bruntland for inefficienty and choosing the wrong time horizon, not for an ethical mistake.

  2. I’m not exactly criticizing Bruntland in this post, although I am a bit concerned that widening WHO’s public health concerns may reduce its ability to act on its core competences. The problem with rare, random but severe and global pandemics is that few organisations are able to plan ahead for them, so there will be a tendency to underinvest in them. This makes the few organisations able to deal with this perspective extra valuable: most governments can promote safer traffic, but few organisations are good at pandemics.

    There is also a more fundamental problem with balancing efforts against the ongoing, large number of deaths due to various predictable factors with the occasionally very large but usually small number of deaths due to pandemics and major disasters. Averages may not be relevant if some of the data is power-law distributed, but what measure to use for cost-benefit analysis in these situations remains unclear.

Comments are closed.