COMARE Statement on
Green Audit Occasional Paper 2002/5
Cancer in Burnham on Sea North: Results of the PCAH (Parents Concerned
About Hinkley) Questionnaire
This statement was re-revised in October 2004
Summary
COMARE was asked by the Department of Health to consider this Green Audit report in detail and advise on the implications it may have for the health of people living in Burnham on Sea. The Green Audit study reported the results of a household survey of the Burnham North ward, which alleged that there were cancer excesses (breast, kidney, leukaemia and cervix) associated with exposure to man-made radioactivity via the local estuarine sands. However, a study by the South West Cancer Intelligence Service (SWCIS), carried out at the request of the Somerset Coast Primary Care Trust, showed that the Green Audit study only covered a small sample of the cases arising in the ward. Using the complete cancer registration data set for the ward the data showed no cancer excess other than for leukaemia. When this excess of leukaemia cases was studied, the majority of the extra cases proved to be chronic lymphocytic leukaemia (CLL), a cancer not considered by any previous investigators and reviewers to be associated with exposure to radiation.
Whilst the Green Audit report was intended to address the concerns of the local community, and did indeed involve them, it is so deeply flawed that it cannot provide any reliable information or conclusions about rates of cancer in Burnham. We believe it is essential that Green Audit withdraws its report so as not to cause further unjustified local concern. The community's interest in establishing the facts are not well served by studies with such deficiencies and it is essential that future studies should have the highest possible standards of design. We note that the authors appear neither to have consulted the local NHS Research Ethics Committee nor to have had the study protocol peer reviewed. If these steps had been taken, the study deficiencies would almost certainly have been pointed out to them. Also Green Audit should follow normal scientific practice and submit their reports for peer review rather than releasing their reports themselves without such review. All other scientific investigators have to follow these rules.
Furthermore, we wish to comment on Green Audit's frequently expressed, and totally unfounded, allegations concerning cancer registries. The remarks about cancer registries and the release of data are both intemperate and factually incorrect. COMARE wishes to place on record the fact that it has worked for over 17 years with many cancer registries and other organisations holding medical data and rejects in the strongest possible terms the implication that they cannot be trusted.
COMARE has already recommended a study of childhood cancer around nuclear installations in Great Britain (Recommendations 4 & 5, COMARE Third Report, 1989) and this study is currently nearing completion. COMARE will consider the need for further studies once it has examined these results.
1. Background
In July 2002 Green Audit published a report entitled 'Cancer in Burnham on Sea North – Results of the PCAH [ie, Parents Concerned About Hinkley] Questionnaire'. This questionnaire survey was carried out following earlier reports by Green Audit that had alleged there were increased cancer mortality risks attributable to contamination by radioactive material, particularly from the Hinkley Point nuclear power station. The objective of the survey was to estimate the incidence of various forms of cancer in the area and to determine whether the rates were higher than would be expected on the basis of national rates. The questionnaire was based on one developed by Green Audit and the analysis was done by the same group. Volunteers from PCAH carried out the survey.
2. The Green Audit survey results
The Green Audit (GA) survey results were based on a study conducted in spring 2002 by volunteers of PCAH assisting householders in the Burnham North ward to complete a questionnaire. Amongst other information, residents were asked about cancer cases occurring at their own address over the last 10 years. Following publication of our original statement in November 2003, PCAH informed us that 749 householders were sampled, of which 100 per cent responded; the method used to ensure random sampling was not described. The questionnaires representing a total of 1,487 persons surveyed. If this is compared with the 1991 census population of Burnham North, this would be about 30% of all persons living in that area. In fact, cancer case information was provided by the householders dating back to 1971, but only cases reported as being diagnosed in 1996-2001 were used in the analysis. It is not clear if these are cases diagnosed whilst a resident of Burnham. The authors say that these case numbers were compared with expected case numbers using the age-distribution of the surveyed group and the 1997 England and Wales cancer rates. A relative risk was then calculated with its associated probability using a Poisson method. The two data periods analysed, 1996-2001 and 1998-2001, overlap considerably. The results for the two periods are not independent of each other and therefore, the quoted p values do not properly represent the evidence.
Full details are not given for lung cancer, but the results have been taken to suggest significantly higher than expected case numbers of breast, kidney, leukaemia and cervix cancer. No further details of the cases are given, eg, histology, types of leukaemia or whether cervix uteri refers to malignant or pre-malignant disease. The survey goes on to use other data to suggest that cigarette smoking was low in respondents but that over 60% of respondents spent time on the beach and that this was a source of radioactivity and linked to the cancer excesses.
3. South West Cancer Intelligence Service (SWCIS) Study
More recently a study has been conducted by the South West Cancer Intelligence Service (SWCIS) and the South West Public Health Observatory to address broader issues of the cancer levels in Burnham North and surrounding wards in Somerset with respect to proximity to Hinkley Point and the local estuary. This study did not rely on survey work but took the cancer registration data for the whole area. These data are collected routinely for all newly diagnosed cases of cancer and are based on multiple overlapping data sources.
The study is able to provide a broad picture of how Somerset wards vary in their specific cancer rates. They used rates for 1990-1999, adjusted for differences in the age distributions compared with England and Wales rates.
The results for Burnham North were:
Site |
Cancer
Registration SIR* 1990-1999 Burnham North |
|---|---|
| All Sites | 94 (NS) |
| Breast (F) | 111 (NS) |
| Kidney | 128 (NS) |
| Leukaemia | 222 (Sig) |
| Cervix | 99 (NS) |
| Colon | 111 (NS) |
| Prostate | 87 (NS) |
| Lung | 81 (NS) |
*Standardised incidence
ratio, England and Wales = 100
NS = not significant
Sig = Statistically significant at the 1% level (one sided test)
The risk for only one type of cancer, leukaemia, was found to be significantly raised. However, this study examined further wards in the locality of Hinkley (Highbridge, Burnham South and Barrow) and found total numbers of cancers to be high in two wards (Highlands, SIR 118 and Burnham South 114), largely due to high breast cancer rates (133 in both wards). This led the investigators to examine the ward results for all of Somerset where complex patterns of high and low wards were found. However, none of these patterns suggested any link to Hinkley or the nearby estuary. Further examination of leukaemias in Burnham North showed that more than half were cases of chronic lymphocytic leukaemia (CLL), a cancer known not to be associated with radiation. Once these cases were removed from the analysis there was no significant excess. [CLL is an indolent disease, predominantly of the elderly, which may cause few symptoms over long periods. Diagnosis is often the result of investigation on an unrelated complaint and incidence in a geographical area depends on the intensity of such investigations. Hence the baseline incidence is uncertain, particularly for relatively small areas and it is unwise to calculate rates, raised or otherwise, in subgroups in such areas. The overwhelming weight of evidence from international studies leads to the conclusion that CLL is a leukaemia sub-type which has not been shown to be associated with, or induced by, ionising radiation (NRPB, 2003). This fact alone should be sufficient to allay concern that environmental radiation might have played a part in causing such cases observed in this study.]
The excesses of breast cancer, by contrast, which were carefully mapped throughout the county, were shown to occur as a result of the breast cancer screening programme which calls up women for screening every three years according to their place of residence. The excesses were the additional undiagnosed cases unearthed at particular time points by the screening procedure.
4. The two studies
A house-to-house survey that relies on volunteers is bound to be flawed in several ways. The 30% sampled may not be representative of the whole population. Recall of disease by the lay public is not likely to be accurate. The study design does not allow for possible migration in and out of the area. (This is touched upon in the Green Audit report.)
Various independent studies suggest that the ascertainment of cases by cancer registries is generally quite high (up to 95% for some adult cancers and higher for childhood cancers) and thus represents a very different dataset to that gathered by Green Audit (Huggett, 1995). It was possible for the cancer registry to count the number of cancers in the whole population of Burnham North in the time span of the Green Audit study. It is not known if the cancer registry case count includes the cases found by Green Audit. Curiously the Green Audit counts for cervix and kidney were almost the same as the cancer registry figures, indicating that Green Audit's figures are almost certainly erroneously too high, given that they are looking at only 30% of the population. This is not surprising given the methods used. Equally, therefore, it is not surprising the results of the two studies are different.
The Green Audit study is probably biased in several ways and the results show the flaws inherent in the approach they adopted. In commenting on this work we do not wish to criticise either the idea of carrying out such a survey or the involvement of the local community. There are, however, well-recognised methodological problems in such work, and the failure by Green Audit to take account of these will have led to unreliable conclusions. Two principal problems are, first, the possibility of bias in the results if the participants were not truly randomly selected and second, more seriously, the possibility that the self-reported information on medical diagnoses is incompatible with that in medical records. In order to avoid these possible biases, it is essential first to obtain a high response rate either from the whole of the population being surveyed or from a sample that adequately represents it. If this is not done, it has to be assumed that those who do respond are very likely to be in some way different from the general population. This could result in reported cancer rates being either too high or too low, and the reported data on cancer rates should be validated by a check on the survey responses against medical records.
The SWCIS cancer registry study, which was carried out by highly qualified and respected epidemiologists and public health professionals, is reassuring in that it demonstrates conclusively a lack of association between cancer incidence and Hinkley Point and its local estuary. The CLL excess and the post screening breast cancer excess represent no link whatsoever with ionising radiation.
5. The Green Audit survey questionnaire
The questionnaire is badly designed
and the questions insufficiently precise. Two particularly important
points are:
(a) As noted above, no attempt was made to validate the reported
diagnoses by reference to medical records, or even to determine
whether the respondents were referring to the primary diagnosis
or to secondary sites to which cancer may spread, or to malignant
or in situ or benign cancers.
(b) Many of the questions were directed only to persons reporting cancer (or birth defects or infant death) whereas they should have been asked of all respondents. For example, all respondents should have been asked about their length of residence in Burnham North and activities related to the beach.
We strongly criticise Green Audit for not seeking appropriate advice on the design of their study. The remit of NHS Research Ethics Committees (REC) says that if requested to do so, an NHS REC may provide an opinion on the ethics of research studies carried out by, for example, private sector companies, the Medical Research Council (or other public sector organisations), charities or universities. This would cover organisations such as Green Audit. The proposed design and questionnaire for a study such as this should also be externally reviewed.
6. Analysis and conclusions of the Green Audit survey
In addition to the problems relating
to the questionnaire, the Green Audit report contains various
problems, some of which are listed below.
(a) The report confuses "prevalence", ie, the number
of people diagnosed with cancer who are still alive, with "incidence",
ie, the number of newly diagnosed cases of primary cancer in a
given time period (usually a calendar year).
(b) The calculation of the expected
numbers of incidence cases does not take into account migration
in and out of the district over the study period. In part, this
seems to be effectively the problem that the authors describe
as "population leakage", and they are correct in saying
that the failure to detect all cases will lead to an underestimate
of risk.
(c) There is in this study no independent medical and, in particular,
histological verification of cancer diagnosis.
(d) It is not clear whether the responses on which numbers of
cancer cases are based refer only to malignant neoplasms, as apparently
do the expected numbers quoted by Green Audit. In particular,
it is necessary to know whether 'cervix cancer' actually includes
"carcinoma in situ of the cervix" in addition to invasive
(malignant) cancer. To compare numbers based on the inclusion
of in situ cases with expected numbers for true malignant cervical
cancer would be a major error; it could indeed account for the
whole of the excess of this condition quoted in the study.
(e) Table 6 in the Green Audit report on the lifestyle factors of cancer sufferers is meaningless without reference to similar data for those who do not suffer from cancer.
7. Cancer registries and policies concerning confidentiality
Finally, we wish to comment on Green Audit's frequently expressed, and totally unfounded, allegations concerning cancer registries. The remarks about cancer registries and the release of data are both intemperate and factually incorrect.
The reference on the first page of the Green Audit report to "the
perceived secrecy of the cancer registries" conveys a wholly
false impression. It would be more accurate to speak of 'the confidentiality
constraints under which cancer registries necessarily work'. The
references to the Welsh Cancer Registry are equally misleading.
These and other issues have been discussed in an earlier COMARE
Statement (see Incidence
of Childhood Cancer in Wales).
At the beginning of the section headed "Comments in the replies" the authors say "The Cancer Registries claim that people do not want to be identified as cancer sufferers or as the family of a cancer sufferer. This is the rationale behind the Cancer Registries' refusal to release small area data". Both parts of this statement are incorrect. First, cancer registries do not make claims about what people want. Second, Green Audit continues to suggest that it is being treated unreasonably by the cancer registries. The simple fact is that all requests, by outside individuals or bodies, to cancer registries for access to data that do, or might, identify individual patients have to be reviewed by an ethical committee. Cancer registries must comply with the provisions of the Data Protection Act and Section 60 of the Health and Social Care Act, 2001.
Further, the "difference in the application of confidentiality considerations" between data on incidence (from cancer registries) and mortality (from death certificates) is not "hard to understand". Information on the incidence of cancer collected while the patient is alive, is "medical in confidence", and is subject to the provisions of the Data Protection Act 1998. Cancer registries are permitted to process such data without the consent of the patient under the Health and Social Care Act. The release of potentially identifiable data is subject to approval by the Secretary of State for Health who is advised by the Patient Information Advisory Group. While the information in death registers is not held in confidence, the ways in which it may be provided are limited in statute. The Births and Deaths Registration Act 1953 provides for information to be made available only in the form of certificates of individual entries in registers on receipt of an application from a person who has identified which entry they want to buy and on payment of the prescribed fee. Other legislation enables information from the registers to be supplied direct to various organisations - for example, the NHS Act 1977 enables the Secretary of State to receive information for the purposes of managing the health service. Information such as electronic tables containing numbers of deaths at all ages combined for certain causes of death at ward level has been made available under Section 5 of the Census Act 1920.
The Green Audit report refers to data that "has not been filtered through official channels and can therefore be believed". COMARE believes that the doctors and scientists in cancer registries and elsewhere who record and analyse cancer incidence and mortality data have no agenda other than to document and present such data as truthfully and accurately as possible. COMARE wishes to place on record the fact that it has worked with many cancer registries and other organisations holding medical data and rejects in the strongest possible terms the implication that they cannot be trusted.
8. COMARE's conclusions
The Green Audit study and report provide no reliable information whatsoever about cancer in Burnham on Sea.
Some or all of the problems with the study could have been avoided if Green Audit had sent the study design and questionnaire for ethical committee and peer review.
The report should have been submitted to a scientific journal for peer review in the standard way. COMARE believes that the report is so poor scientifically that it would not be acceptable for publication in any reputable professional journal.
The SWCIS study conclusively demonstrates that there is no association between cancer incidence in Burnham on Sea and its local estuary and Hinkley Point Nuclear Power Station.
References
Huggett C (1995). Review
of the Quality and Comparability of Data held by Regional Cancer
Registries. Bristol Cancer Epidemiology Unit, University of Bristol.
NRPB (2003). Risk of leukaemia and related malignancies following radiation exposures: Estimates for the UK population. Report of an Advisory Group on Ionising Radiation. Doc NRPB, 14 (1).
Re-revised 1 October 2004
Top 
