Reporting risk and radiation

This is an archived post that will soon move to a new blog.

The media, I'm sure you know, has an unhealthy pre-occupation with health. Hazards are conjured from thin air, and negligible risks are blown out of proportion. Radiation is one of the best subjects available for a health scare: few people have any understanding of the science behind it, it's a synonym for "danger!", and it could get into your house while you're sleeping, and you'd never know! The Guardian last week reported the "news" that a tiny quantity of radium was found on the Olympic building site in London. This was described as a "contamination event", and was carried despite the fact that the "news" was that nothing had happened and there was no risk to health. The reason that radiation of any kind is considered to be "news" is that a lot of people mistakenly believe that the dose of a hazard and risk to health always have a linear relationship: halve the dose and you halve the risk; and with tiny doses, there is still a risk present. Suppose we simply the situation, and pretend that there is only one type of "radiation" (when actually there are several, and they not all equally hazardous), and in a quantity of one million units our radiation will kill one million people instantly (in reality many/most "deaths from radiation" take the form of cancer in old age). We can not extrapolate from that to predicting that a quantity of 1,000 units it will kill 1,000 people, or that one unit will kill one person. The risk associated with a million units may be much larger than a million times the risk associated with a single unit, or it might be much smaller.

I know, it's an article in a general newspaper: it has to be simplified to keep it readable. I can hardly criticise the journalist for simplifying the science in order to keep an article short and readable, when weblogs like this are also built on that principle. The problem is, there's a fine line between simplifying for the purpose of communication, and being unequivocally wrong. Take this story from earlier this month. The headline is "Windscale radiation 'doubly dangerous'." But that is not what the article is reporting. What is doubled in the article is the estimate of the quantity of radioactive material released during the Winscale fire in 1957. There is no reason why that should translate to a doubling of risk to health, and indeed, despite the quote marks, nowhere in the article does anybody describe the risk as having doubled. The risk will depend on many variables: what the material released was, and thus its half-life, type of emitter, chance of entering the food chain, etc; the quantity of it; how dispersed or concentrated it became; whether it went out to sea, onto farmland, or into cities; whether it was released in bulk, or trickled out; and so on. When the quantity of radiation is doubled, therefore, the danger might be far more than doubled, or it might be altered only slightly. If we plotted a graph of radiation exposure and chance of developing a particular cancer, we might find something like this:

Because there are so many variables, it's difficult to determine just what sort of graph we should get without doing plenty of empirical work first. In the case of Windscale, somebody has measured the variables, and done some calculations: BBC news reports a figure of 200 additional cases of cancer caused by the incident, which has since been reassessed to 240 in light of the "double danger". Of those, just over half will be fatal, and many will not appear until late in life. The news is therefore that will be around twenty more deaths from a class of diseases that already accounts for 13% of all deaths. The reduction in life expectancy in the areas affected will be negligible. That twenty people will die younger than they should have is not a fact that should be ignored: lessons should, and indeed have been learned about nuclear safety. But since everybody is going to die, we should not be worrying about this one hazard out of millions, which will reduce the lifespan of twenty people out of billions. According to Google, more people are killed each year from each of: dog bites, carbon monoxide poisoning from camping stoves, falling down abandoned mine shafts, being crushed by vending machines, drinking pesticides, bee stings, and hot baths.