Stalin had a point

As of the 23rd May 2022 this website is archived and will receive no further updates.

was produced by the Winton programme for the public understanding of risk based in the Statistical Laboratory in the University of Cambridge. The aim was to help improve the way that uncertainty and risk are discussed in society, and show how probability and statistics can be both useful and entertaining.

Many of the animations were produced using Flash and will no longer work.

This appeared in The Times, 3rd January 2011. I have added links to the (fascinating) sources.

What should we listen to? Media stories about the return of "killer swine flu": 39 dead so far, 42 children under five in critical care beds, and five times the number of flu cases compared with this time last year? Or the official line that swine flu is now just the new seasonal flu that will probably only last a few more weeks, and that there is no need to vaccinate healthy children?

Maybe they all could be right. The term "swine flu" still carries a shiver factor, but we now know after the experience of last year that it is generally mild, with one death for every 2,000 people who get ill, but has a tendency to affect some younger people severely. Half the reported fatalities were less than 42 years old.

The focus on emotional stories about individual victims may be missing something important. The Office for National Statistics has just reported that in the week up to December 10 there were 11,193 deaths in England and Wales from all causes: this is an extraordinary jump of around 2,000 in a week and is 10 per cent higher than usual at this time of year. Things could get even worse: consultation rates for flu-like illnesses are tracking those of 1999/2000, which was a very severe flu season with 48,000 excess deaths compared to 25,000 last winter which was around average (swine flu made, statistically, a negligible contribution).

Sadly it's remarkably easy to talk about these deaths - hundreds every day, mostly old people getting pneumonia and dying earlier than they need have - as if they are in some way inevitable. Stalin had a point.

Sometimes, statistics do make news, which makes it even more woeful when they later appear hopelessly wrong. In July 2009 Sir Liam Donaldson, then the Chief Medical Officer, reported that the Government was planning for a "reasonable worst case scenario" of up to 65,000 deaths from swine flu. The actual total was 500.

The official post-mortems of the handling of the pandemic reveal behind-the-scenes disputes between Sir Liam and the mathematical modellers. Each have tried to distance themselves from the 65,000 figure. The modellers say the CMO did a naïve calculation using out-dated risk estimates that assumed an implausible "perfect storm" of a pandemic. Sir Liam claims he didn't like the modellers' estimates so he had his own team do back-of-the-envelope calculations which he did not publicise.

It would be funny if it were not so important. But will they sort out the mess before the next real pandemic? I certainly wouldn't want to predict that.

Free tags: 


Maybe you can help me understand this. The figure used is the "excess winter deaths" - "calculated as winter deaths (deaths occurring in December to March) minus the average of non-winter deaths..." How useful is this without the other figures that go into it? How much of the variation in the "excess" could be due to variations in the "average of non-winter deaths". Maybe the non-winter figures should be shown as well. And is the difference (winter - non-winter) better than, say, the ratio? If the overall population size is varies each year then "excess" deaths could rise/fall even if the ratio of winter/non-winter deaths stays the same.

You write that swine flu "has a tendency to affect some younger people severely. Half the reported fatalities were less than 42 years old." Do you mean 4 years old? I can't see how 50% of reported fatalities being <42 yrs old supports the view that the flu affects younger people more severely (the median age of the UK population is around 40).

To the commenter of 10 January: I suspect he doesn't mean 4 years old, but 42 as stated. The median age of the population is probably not what's being compared with here; assessing swine flu against all flu is what's important. To know whether 50% of fatalities being under 42 means it kills younger people more than other strains of flu we would need to know the median figure for flu deaths as a whole. I presume the author knows it but it was omitted here due to newspaper space constraints, so we just have to infer from his words that it's higher than 42. This is not surprising: we all know anecdotally that flu tends to be a dangerous disease for the old and frail and usually only an inconvenience for the young. I suspect most flu deaths are amongst the over-60s, and I bet that in the under-60s who die people in their 50s are disproportionately represented. A strain of flu that kills evenly across the board can be meaningfully said to affect the young particularly severely, since that's unexpected.