Washing out high leukaemia risks in Dumfries and Galloway

Washing out high leukaemia risks in Dumfries and Galloway

[This page is from the late 1990s but the cover-ups continue]

During the row about the Irish Sea effect, LLRC met WCISU's consultant epidemiologist David Adams-Jones, who used to run the Scottish Cancer Registry.
Dan Kenny's archives have produced a 1987 letter from Mr Adams-Jones to the Dumfries and Galloway Health Board who had, understandably, become a bit worried when Dr Heasman told the Dounreay Inquiry that the DG7.3 post code area contained the highest excess leukaemia risk (age 0 - 24) of any post code in Scotland. Mr Adams-Jones was reassuring:

"These figures should not however be interpreted as a cause for alarm because the figures for the Castle Douglas and Dounreay areas are not comparable.
The figures relating to postcode sector DG7.3 were obtained from an analysis of the distribution of leukaemia cases in all 898 postcode sectors in Scotland. We have established that this distribution is essentially random. An inevitable characteristic of any such a random distribution is that high incidence will occur by chance in some areas while no cases at all will occur in others.
The Dounreay investigation on the other hand centred on a hypothesis about a single geographical area in which a source of potential influence on leukaemia risk was already known to exist. The observed figure for Dounreay could therefore legitimately be compared with that expected by chance.
Perhaps I can illustrate this with an example from a completely different field. Suppose that there had been a suspicion that a member of the staff of ERNIE as corrupt and that it was known that he came from Dounreay. The number of successful bonds in the Dounreay area could then be legitimately compared with the number expected on rational rates and a statistically significant excess would indeed give cause for concern. If, during the course of the study to establish the randomness of the procedure, an excess of successful bonds had been observed in Castle Douglas, balanced by deficits elsewhere, this would have been of no practical significance because, by the nature of a random process, some areas must have had higher rates than others. [We assume, of course, that no-one had suspected another corrupt employee with connections in Castle Douglas before the data were examined.]
To conclude, in the absence of a hypothesis relating to leukaemia risk which is specific to DG7.3, the high incidence of leukaemia registrations in the 0 - 24 age group in that area can only be attributed to chance variations in the overall distribution of the disease. Indeed, it was in this context that Dr Heasman referred to DG7.3 whilst giving evidence to the Dounreay inquiry. The figures we have provided do not therefore, in my opinion, represent a cause for alarm."

The figures Mr Adams-Jones gave were

Period Observed Cases Expected Cases Relative Risk
1968 - 73 2 0.27 7.7
1974 - 78 1 0.22 4.5
1979 - 84 1 0.2 5
1968 - 84 4 0.68 5.88
Leukaemia in young people (0 - 24 years of age) in the DG7.3 postcode area 1968 - 84 showing a persistently high risk.

The idea that there is necessarily nothing to worry about if you find a cluster without first forming a hypothesis is plainly nonsensical.
The truth is that DG7.3 does have a source of potential influence on leukaemia risk - Sellafield lies a few miles down the coast emitting 75% of Europe's entire annual inventory of man-made radioactivity.
It is interesting to note that Heasman and Adams-Jones identify the cluster within a postcode area. Concerns about clusters in Northampton and other places have been dismissed on the grounds that parents and scaremongering reporters have tightened the boundaries (i.e. reduced the geographical area to the area in which the disease is actually found) in order, it is alleged, to make the cluster look more significant than it "really" is. The conventional method is to take a local authority ward or a postcode area as the study area. Thus in Newbury the authorities observe that the leukaemia cases (which make an obvious linear feature on the map) fall in two wards: they are therefore two clusters, each of which fails the statistical significance test. In Northampton the parents and reporters were accused of boundary tightening because they insisted on looking just at Pembroke Road, where all the cases were, whereas the Health Authority insisted on loosening the boundary - diluting the population of Pembroke Road into the far larger population of the postcode. This makes the excess risk look much smaller.
Even if we ignore the question "What is it about postcodes and ward boundaries that make them relevant to the aetiology of any disease?", the DG7.3 cluster cannot be written off by this boundary loosening technique. David Adams-Jones writes it off in another way - it's just part of the random distribution.
The clear purpose of this approach to epidemiology is to avoid having to look for causes. An obvious next move in DG7.3 would be to look for associations with the sea, and at levels of radioactive contamination of any mud flats and sands in the area. Someone should find out whether there is a tidal slack nearby, where heavy particles such as plutonium would settle out (as at Lafan Sands in north Wales; about 200 times as much plutonium as other parts of the Welsh coast)? But since officially there's "no cause for alarm" who will be looking? Not the people who are paid fat salaries to protect public health - that's for sure.


The cover-ups continue (June 2007)
Note: ERNIE = Electronic Random Number Indicating Equipment - the Premium Bond prize winners' picker from pre National Lottery days

If you are seeing this page full screen (i.e. without a navigation bar on the left) you can't see how the rest of the site is organised.
This Home page link takes you to the index page, which has links to all the topics we discuss on the site [only use it if this page is full screen]
Use the Flawed Epidemiology button to see what else we have to say on this topic.