The risks of swine flu? I haven't a clue

DJS, Times May 7th 2009

It could be designed to make me feel inadequate. I am a Professor of Risk, and when my daughter Rosie wanted to spend part of her gap-year working on a newspaper she chose, with that true nose for a story, to go to Mexico. So it’s assumed I know the chances of her and everyone else getting, or even dying of, swine flu. But I just don’t know: risk is such an odd thing - no instrument can measure it but it constantly changes as we find out more information, just as the betting odds on Obama being President oscillated wildly over the year before the election. So what do we really mean by chance and risk anyway?

In some circumstances we can comfortably put a number on risk: if I spend £1 on a lottery ticket, I can calculate from the number of ways the balls can be drawn that there is a 1 in 14 million chance that I’ll win the jackpot. Doing the sums for swine-flu is a different matter: a heavenly Dale Winton doesn’t pull balls with our names on out of a rather large bag, so epidemiologists resort to computer models of how epidemics work.

But now instead of just having pure unavoidable chance, ignorance of the mechanics of the epidemic starts to dominate the calculations. It’s a bit like trying to work out the odds of winning the lottery when you don’t know how many balls there are. The shape of the epidemic would be a lot more predictable if we knew all about this virus, and in particular something called the ‘reproductive number’, which is how many people a case is expected to infect in an unaffected and unprotected population. For example each case of measles would be expected to infect 20 people, which is why the drop the drop in MMR coverage is viewed anxiously, while for small-pox it’s around 5, and SARS around 3. Epidemiologists and insurers have been rushing to estimate this quantity from the limited available data: for this particular virus it seems to be less than 2 so a bit of effort might push it below the magic threshold of 1, when the epidemic should disappear.

Meanwhile my girl in Guadalajara reports that nobody there seems to care much about the reproductive number, and the lack of any clear information has brought a mixture of blind terror and indifference. For every few people not wearing masks someone is wearing four at once, just in case. And it’s never too long before the wearer’s intrinsic Mexican-ness overrides instructions and face masks are yanked down to kiss a cheek or smoke a cigarette. The masks themselves sold out completely on the second day of the scare, leading many people to fashion their own from dish-clothes and bits of string.

The health minister in Guadalajara has only just been forced to admit there may possibly be some local cases, whereas in the UK the papers are providing full histories of every contact – invaluable information for the epidemic model. But our ignorance goes beyond not knowing how infectious the virus is and the proportion of cases that die - the virus could mutate or, the feared outcome, join with avian flu to create a new strain: in spite of the opportunities for flying-pig jokes, this would be no laughing matter.

At least we can think of these possibilities and weigh them up, inevitably using a lot of judgement stirred in with the science. But our journey through ignorance can lead into the pitch-black of deep uncertainty - Rumsfeld’s unknown unknowns. It can be disastrous to believe you have thought of everything - it seems clear that a major reason for the financial crisis was a belief that the risk models were somehow ‘true’ and that the world really worked according to the rules, and there was no preparation for when events did not fit the model.

So we need some humility and to admit we may be wrong. Pundits may mock the level of uncertainty that says the eventual UK body-count could be none or could be a million, but it’s simply an expression of honest ignorance. The need is to have emergency systems that are precautionary at first, and then rapidly adapt to new knowledge obtained from good data. Deciding which vaccine to prepare for the winter flu season will require a delicate balance of risks and benefits – a real gamble in the face of uncertainty.

And even if a judgement is inevitable, the reasoning should at least have some science behind it, unlike Egypt’s slaughter of the innocent pigs. Perhaps even that’s better than the conspiracy theories circulating in Mexico, which invite us to believe that the virus was introduced by either the Americans, the pharmaceutical industry or to distract attention from the drug cartels.

Anyway, my gut feeling is that the chances we will see the girl again are looking quite good. But we’ve sent out Tamiflu just in case.

David Spiegelhalter is Winton Professor of the Public Understanding of Risk at the University of Cambridge. Rosie Spiegelhalter is sticking it out in Mexico.

Free tags: