In his response to my column  on the Liberal Democrats' policy on nuclear power, the party's environment spokesman, Simon Hughes, takes me to task for my "polemical" approach. I can only apologise if my writing style offends. I had supposed that putting things bluntly might be the best way to communicate with politicians who put populism ahead of good sense.
Hughes reiterates his assertion  that nuclear power has detrimental effects on people's health. Unfortunately, the particular study he cites to justify this assertion supports my argument, not his. I will try to explain why below, using numbers where necessary and giving scientific references for other cited studies in the text.
Let's begin by looking in detail at the single German study Hughes bases his anti-nuclear argument on. Published in 2008 and generally referred to as the KiKK study, it appeared to show a relationship between a child's risk of contracting leukaemia and distance between the child's residence and the nearest nuclear power plant.
At first pass, this looks worrying, and campaigning groups quickly picked up on these results to trumpet a 120 per cent increase in childhood cancer rates (Hughes mentions "a doubling of the incidence of childhood leukaemia").
But let's look at the figures, rather than the propaganda. First, the numbers when expressed in percentage terms sound scary, but in absolute terms are rather small. (Yes, I realise that any child getting leukaemia is a terrible thing, but please let's refrain from emotive point-scoring.)
The study looked at children living within a five-kilometre radius of Germany's 16 nuclear installations over a 23-year period. It found 37 cases of leukaemia when statistically 17 would have been expected. That is about one extra case of leukaemia in the entire country each year -- hardly a major indictment of an entire industry, even if nuclear power were the cause of the increased incidence, which it almost certainly isn't.
Second, and more importantly, the study does not suggest any causal relationship between radiation emissions from the power plants and the increased cancers, a point that the authors explicitly acknowledge. Again, at first pass this seems odd. We know that exposure to environmental ionising radiation can cause leukaemia. We know that there seems to be an (admittedly tiny) increase in leukaemia around some nuclear power stations. It is safe to assume that nuclear power stations release some radiation. Case proven, right? Wrong.
Influence of a world-view
Here we go right to the heart of the matter. The assumption that living near a nuclear power station delivers a significant dose of radiation -- implicit in Hughes's response and most anti-nuclear discourse on the subject -- does not stand up to scrutiny.
As the German study itself points out, the annual natural radiation exposure is about 1.4 milliSieverts, and average exposure from medical procedures (X-rays and the like) is about 1.8 milliSieverts. A companion study gives precise figures for the additional doses of radiation for those living near German nuclear power plants: "cumulative exposure to atmospheric discharges" for a 50-year-old person in 1991 living five kilometres from one of the nuclear plants would range from 0.0000019 mSv at Obrigheim station to 0.00032 mSv at Gundremmingen station.
For anyone who is unhappy with numbers (including, I suspect, Simon Hughes), the original study makes very clear that additional "radiation exposure near German nuclear power plants is a factor of 1,000-100,000 less" than natural background levels. Given that these levels vary by more than a factor of ten between different regions for entirely natural reasons (such as differing geology), it is vanishingly unlikely that tiny amounts of additional radiation from nuclear power stations causes leukaemia among children.
So why, in that case, is there a statistical correlation, albeit a very minor one? There is no easy answer to this, another problem with basing one's world-view on science rather than prejudice. Looking at other studies may help, however.
As long ago as 1989 a paper published in the Lancet found that "excess mortality due to leukaemia and Hodgkin's disease in young people who lived near potential [nuclear] sites was similar to that in young people who lived near existing sites". In other words, cancer clusters appeared around sites at the planning stage, well before any radiation exposure might have occurred -- and persisted even if no nuclear facilities were ever built.
Studies have also found that cancer rates around nuclear facilities remained unchanged before and after start-up: in other words, where cancer "clusters" existed, they were already in evidence before any radiation was produced. Once again, this raises more questions than answers, but suggests that radiation exposure has nothing to do with it.
A further piece of evidence comes from the site of the world's worst nuclear accident, at Chernobyl. That children were exposed to major doses of radiation here is not in doubt: one study looking at childhood leukaemia in Belarus, Russia and Ukraine (published in the International Journal of Epidemiology) found that the median exposure to participants in the study was about 10 milliGray (the change in units is to do with biological doses received, but, for our purposes, Grays and Sieverts are roughly comparable) -- about ten times natural background levels. Even so, the authors concluded that their study "provides no convincing evidence of an increased risk of childhood leukaemia as a result of exposure to Chernobyl radiation".
Cause and effect
Why does this matter? Because if children near Chernobyl who were irradiated with doses 100,000 or more times that received by German children living near nuclear plants did not show statistically significant increases in leukaemia, then the conclusion is obvious: once again, it is incredibly unlikely that German nuclear power stations are causing leukaemia in kids. (I say "incredibly unlikely" advisedly: a relationship can never be ruled out with absolute certainty. Science is like that, Simon.)
So what explains the German leukaemia clusters? No one knows. One theory is that population mixing in rural areas, which already show higher-than-average rates of leukaemia, is somehow to blame. A possible infectious agent has been suggested as a transmission pathway for leukaemia; yet none has been identified.
Statistical accidents may also explain much of the mystery. Most of the leukaemia clusters involve tiny numbers of people (another cluster around the Krummel nuclear plant in Germany involved five individuals, while the much-discussed Sellafield cluster involved 12 cases over 40 years), making any statistically significant conclusions incredibly difficult.
Leukaemia clusters have been found in villages in Scotland and Germany which are miles from the nearest nuclear plant, and where no causative agent is suggested. Studies of nuclear sites in other countries, including Canada, France, Israel, Japan, Spain, Sweden and the US, have found no cancer clusters of any sort.
In the interests of brevity, I will skip the rest of Hughes's accusations -- even though I note that they sound a little, er, polemical to me. I only wish to note that, in identifying me as a Conservative supporter, Hughes must also be a convert to telepathy as well as mind control: I have never written anything suggesting that I will vote Conservative; indeed, I have not yet decided which way to turn on 6 May.
I'll leave this with one observation. Science is about testing assumptions; politics is generally about confirming them. I wish it were different, but Simon Hughes will need to demonstrate more open-mindedness in future if he is to provide evidence to the contrary.