Storm Force 10 Forecast for Sea of Troubles
Report from Radioactive Times Volume 3, Number 2, October 1999 (content not updated)
Radioactive Times Vol 3 No 1 reported the Irish sea coast effect in Wales. In the period 1974- 1989 people of all ages living within 800 metres of the sea in North Wales suffered far higher cancer rates than the national average. The effect was most pronounced in children but also occurred in adults. It was concluded that over 5000 cancer cases had been registered in the coastal strip over the period in excess of the number expected on the basis of national averages for 1979.
The BBC Wales documentary, Sea of Troubles in February 1999 made much of the issue.The new cancer registry, WCISU denied the data. Both Michael Meacher and the Welsh Office referred the matter to COMARE.
Which Data Set is Valid?
In their analysis for COMARE, WCISU concentrated on childhood leukemia age 0-4. But high levels of childhood cancer and brain cancer had also been reported in this age group. WCISU chose to extend their study to 0-14year olds for leukemia, brain tumours and all cancers. They reported, in a paper to COMARE, that there was no sea coast effect and that Green Audit had used incorrect data.
COMARE, clearly wishing to know the source of the disparity between the Wales Cancer Registry (WCR) data which showed the effect and the new WCISU data which seemed not to asked for a copy of the WCR dataset but WCISU could not provide it as it had been wiped from the computer.
For their part, Green Audit declined to provide a complete set of their WCR databases until WCISU also put their cards on the table and release their version. Despite a request to WCISU for this from COMARE this has been refused. Even the Welsh office has been refused. LLRC has been leaked a letter from the Medical Officer for Wales, Dr Ruth Hall, asking Dr John Steward to release the data.
In the absence of any hard evidence, COMARE chose to believe Dr Steward. He was invited to present WCISU’s analysis to the committee in March. On the 21st July, the Welsh Office (now the National Assembly for Wales) published the full COMARE response in a Press Briefing. Five days before the data of this briefing, reporters from the national papers were calling Green Audit to ask what on earth was going on. They had been circulated a confidential press release announcing the forthcoming Welsh Office briefing but asking them to keep quiet about it . Naturally they all believed that something big was in the air and wanted to know what it was. Green Audit thus came to discover the project in time to brief all the newspapers. The Welsh Assembly Press Office refused to allow Radioactive Times to send a reporter since only national media representatives were allowed into the circus tent.
Assembly Health Secretary, Jane Hutt said that COMARE had compared the child leukemia data held by Dr Steward with that maintained by Dr Gerald Draper of the Oxford Childhood Cancer Research Group (CCRG) and had concluded, on the basis of this test that the WCR data being used by Green Audit was incorrect. She said:
CCRG maintains a registry which contains data from a variety of sources and is considered by COMARE to be as complete as practically possible. As the data held by WCISU was validated by that held by CCRG, COMARE concluded that they found no evidence to support the contention of an enhanced incidence of childhood leukemia among Welsh populations living near the coast.
Dr Ruth Hall, Chief Medical Officer for Wales also welcomed the COMARE statement, saying:
COMARE, the specialist independent scientific committee in this field has examined the research and concluded that there is no adverse health effect on the incidence of childhood cancers.
Both Jane Hutt and Ruth Hall expressed the hope that Green Audit would withdraw their conclusions.
Cover-up and Whitewash
Chris Busby, in interviews on Irish Radio, BBCTV Wales and in newspapers called the Welsh Office exercise
a cold blooded and cynical attempt to cover up the existence of a serious environmental problem. He added It is a shame that the new Welsh Assembly cannot do better than this. COMARE has not done its job or it would have noticed that published reports of Dr Steward’s group, WCISU, show that on taking over from WCR they retrospectively removed over 3000 cancer cases from the files without explanation. COMARE have accepted Dr Steward’s analysis without question: his paper was not seen by us nor was it reviewed. They have bankrupted their credibility as an independent watchdog
Following the media reports of the argument, Dr Dafydd Wigley MP, leader of Plaid Cymru in the Assembly and at Westminster called for a public enquiry. He was reported as saying that he had been aware of high levels of cancer and leukaemia in his constituency, which contained Caernarfon and Bangor -- two of the towns flagged up by the Green Audit report as having high levels of radioactivity from Sellafield and also high cancer levels.
COMARE dyke springs leak
Shortly after the National Assembly Press Briefing, a plain brown envelope arrived at Green Audit’s office. It contained the confidential minutes of the COMARE meeting on 18th March where the issue was discussed. For students of psychology, group dynamics, the full minutes are reproduced on the LLRC website. At this meeting, John Steward and David Adams Jones presented the Results of a preliminary study to test the Irish Sea proximity hypothesis of Busby et al. They told the committee thatthey had repeated Green Audit’s analysis and found that the general level of relative risks for leukemia were in line with those expected for England and Wales and that there was no evidence of a decreasing gradient of leukemia risk with distance from the Irish Sea.Risks for childhood cancer and brain tumour incidence were not found to be significantly different from those expected in England and Wales.
Green Audit only saw this paper after the press briefing in July and were not invited to the COMARE meeting to comment. How could two studies of the same phenomenon show two wildly different results?
Statistics, Damned Lies, Lies
There were only certain ways methods available to lose the effect and, as it transpired, WCISU had used them all (see A Tale of Two Studies below).
LLRC has seen Dr Steward’s paper and apart from the completely revised data his team used his claim to have reproduced Green Audit’s methodology is untrue.
· WCISU’s coastal strip is 5km wide compared with the 800metre strip used by Green Audit.
· There are large differences in the numbers of cases recorded, particularly in the north. Together with the loss of cases this dilution of the effect resulted in a much smaller rate for childhood leukemia in the coastal 5km strip than even the rate for England.
In addition, Steward et al used the wrong population figure for a study of cancer from 1974-89, one based on the 1991 census. The Green Audit study had used the 1981 census populations. But the main question is still about the discrepancy in the figures. (see Raising the Dead)
A Tale of Two Studies
Dr Steward's WCISU childhood cancer study was presented to COMARE as a repeat of the Green Audit , Wales Cancer Registry data study. The main cancer excess effect found by Green Audit was in the age group 0-4. This maybe why Dr Steward chose to dilute this group by looking at the 0-14 year olds. However, more effort was needed to dilute or remove enough cases for the result of the 0-14s to show no significant effect. In the two columns below we analyse the differences between the two to examine how this was accomplished. Information bases for the two studies relate the same phenomenon -- All Malignancies age 0-14 for All Wales 1974-89.
|WCISU adjusted database
||Green Audit/ WCR Database
|· Period 1974-89||· Period 1974-89|
|· All malignancies 0-14||· All malignancies 0-14|
|· 1991 census wards aggregated||· 1981 census ward s aggregated|
|· 5-km closest strip to sea||· 800 metre closest strip to sea.|
|· Non significant excess found. Relative Risk in this strip = 1.11||· Significant excess found. Relative Risk in this strip = 1.4 P = 0.01 (RR = 2.2 in the 0-4s)|
|· 1991 Population used totals 569900 children aged 0-14 and 204,800 children aged 0-4. (21% higher than 1981 population)||· 1981 Population used totals 564870 children aged 0-14 and 169,200 children aged 0-4.|
|· Total cancers 0-14 recorded in WCISU datafiles = 1006||· Total cancers 0-14 recorded in WCR datafiles used = 1188|
1188 minus 1006 is equal to a shortfall of 182 cases, or about 15% of the total number recorded by Wales Cancer Registry in their 1994 and 1992 publications. Astonishingly, Steward has even taken 50 cases or more from his own dataset published by WCISU in April 1998, where the number of Childhood cancers for 1974-89 was recorded as 1054.
The exclusion of the non-malignant skin cancers has very little effect. The totals for NMS are about 10 in Wales over the whole period. How did COMARE validate such outrageous differences in Steward's own data? How did Dr Draper confirm these figures? LLRC intends to find out.
This page was last updated May 2001