The Effects of Sellafield on Cancer Incidence in Ireland from 1994-1996

Analysis of National Cancer Registry Ireland

Small Areas Data

Chris Busby PhD

Rachel Kaleta BSc

Helen Rowe BA

Report 2000/12

Green Audit

Aberystwyth: November 2000

1. Background

From 1998, Green Audit has undertaken research in connection with the legal case Short and Others vs BNFL and Others. This research has examined the evidence that radioactive discharges from the BNFL reprocessing plant at Sellafield in Cumbria have been and will continue to affect the health of those people living near the shores of the Irish Sea. The work initially focussed on the distribution of cancer incidence in persons living in Wales between 1974 and 1989 and examined official small area cancer incidence data obtained from the Wales Cancer Registry, a division of the Welsh Office.

Results of this small area study showed that cancer incidence in Wales, over the period, was significantly affected by proximity to the sea. These results were summarised in Busby 2000. In particular, the results showed that it was proximity to large areas of intertidal sediment such as estuaries, inlets and mud banks that was associated with excess cancer risk. The trend with distance from the coast was quite curious and specific. Within the spatial resolution of the data it was possible to infer that it was the very narrow 800metre wide strip that carried the highest risk. Transects of cancer risk from the seaboard inland showed highest risk in the proximal narrow band falling sharply and flattening for inland populations. Attention was drawn in Busby 2000 to the similarity between this trend in cancer risk and various published measurements which showed the penetration inland of radioactive particles derived from the sea. Thus the phenomenon of sea-to-land transfer of radioactive material was tied to the measured cancer excess in coastal populations living near radioactive mud banks in the intertidal zone.

Attempts were also made to obtain and analyse small area data for Ireland: however there were two major problems here. First, there has been no national cancer registry in Ireland until 1994, and so comparable data to the Welsh i.e. for the years of peak discharges from Sellafield, were not available. Second, following the establishment of a national cancer registry in Cork, it was by no means clear that small area data for the few years after 1994 would be made available in a form that would enable the two countries to be compared. This is because of the unfortunate development in protocols for releasing such data, which occurred following the release to Green Audit of the Welsh data. This required the Cancer Registries to refuse to release any data which might allow any case to be identified. This ‘confidentiality’ requirement was originally interpreted by the cancer registries in the UK as meaning that no data could be released to areas smaller than entire Counties. Later a more precise formulation required that the denominator in any analysis of cancer rate in any area might not be smaller than 1000 persons in the category of risk. This UK Cancer Registry protocol was adopted by the National Cancer Registry Ireland in discussions with Green Audit in 1999 about the release of their small area data for Ireland for 1994-1996. In order to make the best use of the data within this requirement, Green Audit used population census data to make up their own small areas which would approximate the 1000 denominator size requirement to enable an analysis of the 0-4 year old group living near the east coast of Ireland and NCRI agreed to supply these data. Green Audit aggregated the District Electoral Divisions (DED) for which census population data was available into units with base population in the 0-4 age groups greater than 1000 individuals. The aggregation was done in a way that minimised the East West distance of the final unit at the expense on the North South distance, in order to obtain the greatest East West resolution.

Prior to this, some work on an unconnected area in England had thrown some light on the hypothesis. Early in 2000, Green Audit obtained cancer mortality data for census ward areas in England and Wales from the UK Office for National Statistics. This enabled an analysis of an area in Somerset near the Hinkley Point nuclear power station which discharge to a large coastal mud bank. These data contained the numbers of deaths from all malignancy, breast , prostate and lung cancer and these were analysed to examine Standardised Mortality Ratios by distance from the mud bank. The results showed the same sea coast effect which had appeared in the Welsh data.

In order to examine the effect of proximity to a contaminated mud bank in Ireland and to look at small area cancer risk with a spatial resolution that was more precise than that available from cancer registry data it was decided to undertake a questionnaire survey of such an area. The area that was chosen was Carlingford and Greenore in County Louth. This area was broadly that which had been analysed earlier (Busby, 1999, 2000) by calculating cancer risk in the period 1965-1986 using the data provided by the local GP, Dr Andrew MacDonald (MacDonald, 1997). This showed that the cancer incidence rate for certain types of cancer and particularly leukemia was well above the national average. In addition, many people living in the area had drawn attention at public meetings and in letters to newspapers to what they perceive as high levels of cancer. Indeed, the widely-held belief that Sellafield has caused cancer and other illnesses in the area eventually resulted in the formation of the Dundalk-based STAD group who are supporting the plaintiffs in the court case.

An advantage of a questionnaire approach is that the results are not dependent on cancer registration and ultimately each case can be named should there be any argument about accuracy. Results were reported to the litigants this year (Busby and Rowe 2000). The sea coast effect was graphic: people living within 100metres of the contaminated mud banks in Carlingford Lough reported more than 2.5 times the relative risk from cancer than the inland group (p = 0.004) whilst the trend with distance was similar to that found in Wales and at Hinkley Point and was highly significant (p = .0009).

Shortly after this analysis was carried out, the National Cancer Registry supplied the incidence data for the small areas which had been defined by Green Audit to conform to the confidentiality requirements. Data were supplied for all cancers, brain cancers and leukemia by sex and 10- year age group but also included the numbers of cases in the 0-4 age group. Results from analysis of this data is reported here.

2. Strategy and Method

General

The hypothesis to be tested, based on earlier findings (Busby 2000, Busby and Rowe 2000 and Busby et al, 2000a, 2000b and 2000c) was that there was increased cancer incidence risk with distance from the Irish Sea, particularly that part of the Irish Sea in the north east, where the highest levels of Sellafield radioactivity are measured in coastal intertidal sediment. Because of the National Cancer Registry’s requirements for aggregation of wards, the spatial resolution of the data supplied was generally not sufficiently high to examine distances from the coast less than 5km. This is a distance which is significantly greater than that in which the sharp decrease in air concentration of resuspended radioisotopes has been shown to occur (Eakins and Lally, 1984) and which can be seen in the Carlingford study (Busby and Rowe 2000) and in some of the earlier work. There is a 100-fold drop in airborne plutonium in the 2km distance from the Irish Sea in Cumbria (Eakins and Lally 1984) and although their measurements were made in Cumbria, they showed that it was the seaspray resuspended particles, scavenged from the sea bed in the intertidal zone that was responsible for the effect. The drop out of particulate aerosols originating from the sea from the point of resuspension is such that heavy particles > 10 microns precipitate within the 2km region (depending on wind speed) but that smaller particles are capable of travelling very large distances. Particles origination from the sea contribute about 20-30% of all PM10 particles in England and Wales. Measurements by Eakins et al.,1984 and others (Priest et al 1997, Popplewell et al., 1988) support this in that that the trend in airborne resuspended radioactive material continues to fall more gradually with distance from the coast up to at least 100km. Therefore , if Sellafield material were causing cancer in Irish populations, we should expect to see a significant negative trend with distance from the east coast, little trend with distance from the south coast and no trend with distance from the west coast. For this reason we included west coast county areas as a control for the effect we are investigating.

Small Area Choice

Counties in Ireland which were divided into small areas for this study are listed in Table 1 and the small areas themselves in Appendix A together with a map. By inspection of the DED map of Ireland, supplied by the Ordnance Survey, 23 counties were divided into 270 small area groups of DEDs labelled GADEDS (Green Audit DEDs). Table 2 shows the DED composition of two examples, the GADEDs labelled A1 and A2 in County Carlow. These groups and their DED composition, were supplied to the Irish National Cancer Registry (NCRI) in Cork in 1998 who agreed to process their postcoded data so as to supply the numbers of cases registered in the period 1994-1996 in these GADEDs. The work was apparently carried out by a separate organisation and was checked by NCRI. The final files were given to Green Audit in June 2000 and these contained data for all malignancies, all leukemias and brain cancers by sex and ten year age group but also included the age group 0-4 which had been specifically requested. Following preliminary analysis of this data it was discovered that results for 18 out of the total 270 GADEDs were missing. Data for these areas was requested and most of this data has now (Nov 2000) been supplied but has not been included in this study. The numbers of cases involved represents less than 2% of the number of cases studied in the total area and therefore cannot alter the results significantly. The main blocks of missing data may be seen in Table 1.

In addition, the NCRI stated that 536 cancers including 21 brain cancers and 14 leukemias could not be assigned to a DED: most of these cases were in the elderly. In October 2000 INCI informed us that they had discovered some limitations in the accuracy of the geocoding that had been carried out by the contracting organization. They assigned cases which could not be accurately geocoded to a DED in the geographical centre of the area and stated that in practice this meant that cancer numbers might be artificially high in the GADEDs Dublin County Borough North A1, Cork County Borough B5, Galway County Borough E1 and Waterford County Borough B3. This was a fair way of distributing the cases since these DEDs have high populations and the extra cases would thus have little effect on their relative risk values. On the other hand, it is possible that moving these cases could have artificially reduced the relative risk in coastal wards. We feel that, in summary, the number of cases either missing or moved to the centre of the county is so small in proportion to the total number of cases that it is unlikely that they would significantly alter the conclusions of the study.

There were 20655 cancer cases in the study area between 1994 and 1996 excluding the 9504 cases in the Dublin/ Dun Laoghaire/ Rathdown/ Fingal area. 536 cases represents only 2.6% of these. The latter area was excluded for the study for two reasons. First, the coastal position of the area would have biased any attempt to examine for coastal trends due to the large population. Second, it was thought that the largely metropolitan and semi industrial nature of the area would have resulted in cancer stresses that had little to do with the causes we are examining. This was the reason we excluded the industrial areas of South Wales from the Welsh study reported in Busby 2000.

Relative Risk and Seacoast Trend in Risk

The Relative Risk was calculated in the standard way. First, the mean rate for the cancer in question for the whole of Ireland and for each sex and 5- or 10-year age group over the period 1994-1996 was obtained from published reports (NCRI 1998, 1999). These rates were then used to multiply each sex and age group GADED population obtained from the 1996 census population of Irish DEDs by NCRI and supplied to Green Audit along with the cancer data. Multiplying this result by three gave the expected number of cases in each sex and age group in each GADED in the study for the period 1994-1996. The observed totals, divided by this expected total, gave the Relative Risk for each GADED for the sex and age groups and cancer sites.

For all cancers, because of the large numbers, we were able to examine risk trends in age groups and by sex. For the two small incidence cancers, because the rates were not sinificantly different for the sexes we looked at persons but were also able to examine young persons <44 and old persons > 44 as two groups. In all cases we also examined the risks in the 0-4 age group.

All GADEDS were coded as East West or South on the basis of establishing risk by distance from these three coasts. The East coast was defined from County Louth to County Wexford; South Coast was from Waterford to West Cork and West Coast included Clare and Galway. Distance from the nearest respective sea coast was measured from the area centre of the GADED and division of GADEDs in the centre of the country was decided upon by ensuring that at areas were included which permitted a trend to be established for at least 60km distance from the respective sea coast. In the case of the East coast, since this was the hypothesised exposed coast, this range was extended to 120km.

In addition, each GADED was coded according to its County, enabling us to examine the effects in Louth or Louth and Meath and also with distance along the coast south from Louth which has the highest measured concentrations of Sellafield radioisotopes in intertidal sediment.

The method used to establish trend with distance from the sea is similar to that employed in the Welsh study, Busby 2000. Because the GADEDS vary in size, it is impossible to establish risk by distance from the sea on the basis of regression methods using relative risk in each GADED. Thus groups of GADEDS are aggregated together according to their inclusion in distance bands and the total number expected and observed cases compared to produce a relative risk for the distance band population as a whole. Unlike the regression method employed in the Welsh study, trend by distance is examined statistically here in a way that was developed for the Hinkley Point study. This is to use the statistic Chi Squared for Linear Trend in Proportions, which is a version of the statistical procedure developed by Mantel and Haenszl (Schlesselman, 1982). This method was much in vogue as a method for examining trends in dose response data and also for epidemiology but has fallen out of use in the last fifteen years with the development of regression methods. For cancer studies, the latter suffer from major shortcomings because of the problem of standardisation for age and the insensitivity of standard regression methods to the numbers representing each data point. We addressed this problem in the Welsh study by weighting the points by the expectation in the small area which it represented: nevertheless, there was loss of information and statistical power in this procedure. An account of the procedure we have used in this study is given in Appendix B.

In addition, we have presented the results graphically and through power series or LOESS local regression fits of the data points for relative risk for each distance band aggregate. We have also tested coastal strips against inland blocks for East West and South aggregates of GADEDS.

3. Results

3.1 All malignancy excluding non-melanoma skin cancer.

3.1.1 Study area totals

The study area population totals were made up of 1,392,430 females and 1,420,352 males. The number of registrations in the study area by sex and age group are given in Table 3. The total numbers over the three year period were 15810 females and 14349 males. This gave a cancer sex ratio Female/Male of 1.10. The totals for the whole country for this period, recorded in the NCRI reference volumes were 7721 and 6756 per year giving a total of 23163 and 20268 cases in the three years, a sex ratio of 1.14. Note that there are more female cancer registrations than male, despite the male population being smaller.

3.1.2 Age and sex trend anomalies in Ireland

For most cancer sites, the risk of cancer increases with age. This is because, assuming constant carcinogenic stress, the mutation probability is merely a function of the number of cells at risk which is the number of cells in a clone which carries the requisite pre-cancerous mutation. The relationship and its basis in biology are discussed in Busby 1995. For all malignancies, all cancers aggregated together, the rate with age increases as a power relationship: this is seen in the two curves in Fig 1, which compare data for England and Wales 1994 with the average 1994-96 NCRI published data for males in all Ireland. Note the similarity, which is what would be expected for two similar genetic population types exposed to similar carcinogenic stresses or pollutants. However, there is a marked anomaly apparent when the females are compared in Fig 2. For female cancer in Ireland, the rates diverge markedly from the baseline rate expected on the basis of the England and Wales populations in two groups: those aged between 25 and 40 in 1995 and also those aged over 80. In order to investigate this further, we look to see where in Ireland the effect is most pronounced. The women aged 25-35 in the period 1994-1996 were born in the period 1959-1971, and since cancer increases sharply with age over the age range 25-35, we assume that the effect is driven by the older women in this group. This points to 1959-63 as the period of birth. This period was that of the global weapons fallout, but since we are looking for a specific Ireland effect, the most likely source of risk is the 1957 Windscale fire, which caused considerable releases of radioactivity to precipitate on the eastern side of the country, particularly in County Louth. Anomalous increases in Down’s syndrome births to Windscale exposed mothers have already been reported for Dundalk (Sheehan, 1983). In Fig 3 we show a comparison of Co. Louth with a west coast control, Co. Galway and it is clear that the Irish female cancer anomaly is much higher in County Louth than Galway, in support of the hypothesis. Thus the source of this effect is Windscale, now called Sellafield.

Why does the effect only show in women? The answer may be that apart from leukemias in the very young, female cancers are the first cancers to show in any exposed population. The main cancer showing in this age group will be cervical cancer and although we do not have data on this cancer in the INCR file we can look at the 1998 publication on the INCR to examine the risk maps and compare county Louth and County Galway. We see immediately that the highest rates for cervical cancer are in County Louth, and they remain high all down the east coast of Ireland. This is in comparison with low rates in Galway and on the west coast. Tables in INCR 1998 give Standardised Incidence Ratios of 155 for Louth compared with 99 for Galway and 81 for Clare. However, we cannot be certain that the cause is not lifestyle, since sexual promiscuity is a known factor in this disease and there may be a variation in this factor between the east and west coasts.

3.1.3 All malignancy by area and distance from the sea in adults.

Following division of the GADEDs into groups by distance from the seacoast in the east, west and south coasts, it is clear that the seacoast effect on cancer is present on the east coast but not on the south or west coasts. Tables 4, 5 and 6 show the numbers of cases of all ages in males and females observed and expected on the East, West and South coast GADED groups by shortest distance to the relevant seaboard. Fig. 4 shows a plot of this data for the three areas for females with a power fit to the data points, and Fig. 5 shows the same for males.

In order to examine the age group 25-34, who were seen to exhibit excess risk in the examination of the overall data and in County Louth, we examine the cases and risks in the three areas in Tables 7, 8 and 9. The plots of the risks in the young women aged 25-34 are shown in Fig 6. The numbers of cases in the men of this age group were too small to plot a significant trend.

The interesting variation with risk in men and women by distance from the sea on the East coast is stratified by age group in Table 10 and plotted for women in Fig 7 and men in Fig 8. The existence of the differential seacoast effect is clear for women in both the 25-44 age groups and in those over 75: it is present in men in the 25-44 age group also but it is clearly absent in men over the age of 45. For men and women combined, the effect is shown in the panel in Fig 9.

The results of statistical tests on the trends in adult cancer by distance from the sea in the three groups are given in Table 11. These show that for women, the seacoast effect on all malignancies was present on the east coast for women but not on the south or west coasts. For men, curiously, the effect was largely absent, except for those aged 25-44 on the east coast.

3.1.4 Childhood cancer.

For the age group 0-4 and 0-14, separate analyses were performed on the aggregated male and female populations. Although the numbers were small in these groups, there were no clear effects of living near the sea in either age group, even in the area of hypothesised highest risk, County Louth. For the 0-14s in the east coast area of analysis there were 26 cases of both sexes in the 0-20km strip with 32.8 expected and 71 cases in the whole of the interior above 20km with 78.7 expected, the trend being fairly flat along the transect.

3.2 Brain tumours

The total number of registrations in the three areas we studied are given in Table 12.

Because of the small numbers of cases, numbers in the distance bands used for all malignancy were too small to provide meaningful trends. Accordingly, we have defined wider distance bands for this site and have aggregated risk into three bands, namely <20km, 20-60km and >60km. Results for the three areas are given in Table 13. There is slight evidence for a trend away from the sea on the East coast but closer examination of the data suggests that there is no sharp fall in risk close to the sea on the whole east coast, as there is with the all malignancy result. The hypothesised high-risk area of County Louth does show some evidence of high coastal risk since 19 out of the 22 registrations are in the 5km strip. However the numbers are too small to draw any real conclusions. Separate analysis for the children and young persons showed no significant effects.

3.3 All Leukemias

Table 14 gives the numbers of cases and the expectations in the three study areas for all ages.

As with the brain tumours, because of the small numbers of cases, numbers in the distance bands used for all malignancy were too small to provide meaningful trends. Accordingly, we have defined wider distance bands for this site and have aggregated risk into three bands, namely <20km, 20-60km and >60km. Table 15 gives the results. As with the brain tumours, there were no significant trends in the age groups or by distance from the sea in County Louth.

3.4 Cooley Peninsula

The area Louth A1 represents the Cooley peninsula, which contains the wards of Carlingford and Greenore, among others. The recent Green Audit/STAD cancer survey of these two wards established that the Relative Risk of cancers reported in the survey for the period 1989-1999 was 0.86. The figure shown for the same cancers here for the three years 1994-1996 is 0.65.

4. Discussion

It is rare in epidemiology that a prior hypothesis is supported so spectacularly as the seacoast hypothesis has been by the results reported here. There is high risk on the east coast, where the Irish Sea is contaminated by Sellafield. Risk is highest in the north east, where the contamination is highest. There is little excess risk on the south coast and none at all on the west coast, where Sellafield isotopes are largely absent. However, it is also rare for the results of a new investigation to not provide new questions and this also is the case here.

The results of this small area study show that there is a highly significant sea-coast effect on the east coast for men and women combined. This effect is absent on the south and west coasts and therefore supports the prior hypothesis that exposure to radioactive particles transferred ashore by sea-to-land transfer increases the risk of cancer. The risk is highest in the north, County Louth, where the contamination is greatest. The effect is, however, largely driven by cancers in women. For men, curiously, the effect is only present in the age groups 25-34 and 35-44. These age groups also shows an alarmingly large cohort effect in the women, and together, these findings suggest that this group, those who were aged 25-34 and to a lesser extent 35-44 in the period of the study data (1994-1996), had been exposed to some carcinogen. This effect was also apparent in the Carlingford survey data where 6 cases of cancer aged between 27 and 43 were born between 1956 and 1960. This is the period when they would have been exposed to fallout material from the 1957 Windscale fire either as young children, or through genetic damage to their parents. In the case of the peak effect found in the present study, the years of birth of the cohort are between 1960 and 1971 with a range back to 1950.

The data does not show a seacoast effect for the Brain tumours or the Leukemias, cancer sites which are traditionally associated with radiation. High levels of risk for these cancers was not found in the Carlingford survey either, where the highest risk was for colon cancer and lymphoma. However, the exposure onset lag is quite small for leukemia and unknown for brain tumours.

The absence of a sea coast effect on the east coast for older men requires explanation, and without a breakdown of the data into cancer types, and the examination of trends prior to 1994, this is difficult. There is no doubt that there is a very significant effect in women. There have been reports of anomalous health indicators found in women on the north east coast (Grehan 1998, 2000) before, and it does seem as the women of County Louth have been exposed to some harmful agent at the time of the Windscale fire. We noticed that there is a sharp change in the sex ratio of the populations of the coastal region on the east coast of Ireland. There are more men than women in this 5km strip with the Male/Female ratio being 1.05 in the <3km band, 1.04 in the <5km band but changing to the normal human population ratio of 0.97 for the whole of the inland bands. This variation in the population requires some investigation as it is a well known indicator for genetic damage. The preponderance of cancer in women, suggests that it is women’s cancers that are involved. In the 25-44 age group this is cervical cancer and breast cancer, and indeed, County Louth has the highest cervical cancer rate in Ireland by a very large margin. Therefore one explanation for the surprising difference between the women and the men may just be that the female cancers are the first to become clinically evident, as their age onset is earlier. We might therefore, on this basis, expect the male seacoast effect to develop in the next ten years as the lung cancers and prostate cancers in the males begin to be diagnosed.

We are still left having to explain the fact that the effect is high in the old women but not in the old men. There are two possible explanations. The first is that the men and women have been differentially exposed. This could have occurred if the pattern of work in the area involved the men working away from home in areas inland in the period 1955-75. In order to examine this we are investigating the age and sex breakdown of the east coast County Louth populations from 1951-1991: these data have not yet been received. In this situation we assume that the women in the period stayed at home. The second possibility is that the data are a snapshot of the situation, and that those men who represent the old age group are survivors, and that the more cancer prone weaker proportion have already been diagnosed or have died. The peak levels of radiation from Sellafield arrived in Ireland with the wind and the rain in 1957, or washed up on the coast from the late 1970s. Since then levels have fallen in the intertidal sediment surface, and exposure will have been less. We are now looking at data on cancer incidence some twenty years after the peaks in licensed emissions and 40 years after the Windscale fire. This is rather late for the detective to arrive at the scene of the crime and still find anything. However, we have found a great deal. The seacoast effect, discovered in the Welsh data for 1974-89 and in the Somerset data for Hinkley Point, and then, graphically in the Carlingford Survey, has now been found in all of the women and some of the men on the east coast, but not the south and west coasts. This is very strong evidence that the hypothesis that Sellafield has caused cancer in Ireland is correct. More support would come from an analysis at DED level to investigate the seacoast effect with more resolution. For further background and supporting evidence ,with explanations of the mechanisms involved please see the earlier reports listed in the reference section.

5. Acknowledgements

We are grateful to Dr Harry Comber and the staff of the National Cancer Registry in Cork for arranging to have the data aggregated to the small areas we defined. We would like also to thank the Ordnance Survey, Dublin for kindly providing maps, Mr Ollan Herr of Dundalk and Mr John Gorman of Dublin for assistance in obtaining ward boundaries, Mary Heanue, Evelyn Cronin and others at the Central Statistical Office, Dublin, for help with population and mortality data. We thank the litigants and their solicitors for their bravery and hard work in pursuing this case. We are grateful also to the Irish State (Department of Public Enterprise) for financial support.

6. References

Busby C.C, (1999) Radiation from Sellafield and Cancer near the Irish Sea. The First Annual progress report from the Irish Sea Group in support of the litigation Short and Others vs BNFL and Others. Unpublished report.

Busby C.C, (2000) Radiation from Sellafield and Cancer near the Irish Sea. The Second Annual progress report from the Irish Sea Group in support of the litigation Short and Others vs BNFL and Others. Unpublished report.

Busby C, Dorfman P, Rowe H (2000a) Cancer Mortality and Proximity to Hinkley point Nuclear Power Station in Somerset: Part I Breast Cancer. Occasional Paper 2000/2 Aberystwyth: Green Audit

Busby C, Dorfman P, Rowe H (2000b) Cancer Mortality and Proximity to Hinkley point Nuclear Power Station in Somerset: Part II Prostate Cancer. Occasional Paper 2000/3 Aberystwyth: Green Audit

Busby C, Dorfman P, Rowe H (2000c) Cancer Mortality and Proximity to Hinkley point Nuclear Power Station in Somerset: Part III All malignancies, lung and stomach cancer. Summary Occasional Paper 2000/4 Aberystwyth: Green Audit

Busby C, Rowe H (2000) Cancer Incidence in Carlingford and Greenore, County Louth: Results of the STAD/ Green Audit Questionnaire Report 2000/06 Aberystwyth: Green Audit

Eakins, J. D., Lally, A. E., Cambray, R. S., Kilworth, D., Morrison, R. T., and Pratley, F. (1984), ‘Plutonium in sheep faeces as an indicator of deposition on vegetation’, Journal of Environmental Radioactivity, 87-105.

Eakins, J.D and Lally, A.E., (1984), 'The transfer to land of actinide bearing sediments from the Irish Sea by spray.' Science of the Total Environment 35 23-32

Grehan Mary (1999,2000) Pers comm.

MacDonald, A. (1997), A Twenty-Year Survey of a Rural General Practice in Ireland, unpublished.National Cancer Registry Ireland (NCRI) (1998) Cancer in Ireland 1995 Cork: National Cancer Registry Board

National Cancer Registry Ireland (NCRI) (1999) Cancer in Ireland 1996 Cork: National Cancer Registry Board

Priest, N. D., O’Donnell, R.G., Mitchell, P. I., Strange, L., Fox, A., Henshaw, D. L., and Long, S. C. (1997), ‘Variations in the concentration of plutonium, strontium-90 and total alpha emitters in human teeth collected within the British Isles’, Science of the Total Environment, 201, 235-243.

Popplewell, DS, Ham GJ, Dodd NJ, Shuttler SD (1988) ‘Plutonium and Cs-137 in autopsy tissues in Great Britain’ Sci. Tot. Environment 70 321-34

Sheehan, P. M. E. and Hilary, I. B. (1983), `An Unusual Cluster of Down's Syndrome, Born to Past Students of an Irish Boarding School', British Medical Journal, 287 (12 Nov.).

Schlesselman J. (1982) Case Control Studies p200 Oxford: University Press

Executive Summary

The study examined cancer incidence in 270 small areas in 23 counties of the Republic of Ireland for the years 1994-1996 inclusive. Data was made available by the National Cancer Registry Ireland by sex and ten year age groups for all malignancies, brain tumours and all leukemias. In addition, data was made available for children aged 0-4.

The analysis sought to examine the hypothesis that living on the East Coast, near the Irish Sea, involved significant excess risk of cancer, due to internal exposure to man-made radioactivity from sea-to-land transfer of radioisotopes from Sellafield in Cumbria, UK. To this end, the total area was divided into three groups made up of small areas whose area centroids were at different distances from the East, South and West coasts of Ireland. By calculating Age Standardised Relative Risks for men and women of different ages in these areas it was possible to determine the trend in cancer risk for the three cancer data groups by distance from the sea coast.

The results clearly demonstrated a significant cancer risk sea-coast effect on the East Coast but not on the West or South coasts for women of all ages combined. The effect was highly significant. For bands of increasing distance from the East Coast centered at 2.4, 3.2, 6.4, 8.1, 13.1, 31, 51, 71 92, 128km the relative risks for women of all ages were 1.4, 1.27, 0.86, 0.86, 0.9, 1.0, 1.03, 1.15 and 0.97. Chi-square for trend was 10.7; p = 0.001. Testing the coastal <5km group against the >5 km group gave Chi-square = 44.3; p = 0.00000. A curious excess risk existed on the eastern sea coast in younger women in the age groups 25-34 and 35-44. The trend with distance in the 25-34 group for the same distance bands as those listed above gave 1.66, 1.53, 0.53, 0.49, 0.73, 0.9, 1.02, 0.72, 1.14, 0.56 . Chi-square for trend was 11.13; p = 0.0009. Test of coastal <5km against non-coastal in this age group gave RR = 1.62; p = 0.000025. This trend with distance is similar in shape to that which we previously found in Welsh incidence data for 1974-89, cancer mortality data for Somerset from 1994-99 and very-small-area mapped questionnaire data from Carlingford and Greenore, Co. Louth. The trend correlates with the function describing inland penetration of plutonium particles in seaspray

For the South and West coast groups there was no discernible trend with distance from the sea in either women or men of any age group, sugesting that the cancer excess was due to an environmental agent present on the east coast. The effect was clearest in the comparison of County Louth with a control county of Galway.

For men of all ages combined, the effect is absent, although it is there for the age group 25-34 and 35-44 on the East Coast. This age group represents the cohort most affected by the releases from the Windscale fire in 1957, and these groups were identified as a high risk group in the Carlingford and Greenore survey results also.

For children aged 0-4 and 0-14, there is no significant sea-coast effect, nor is there any effect for brain tumours and leukemias. It is argued that this may be a consequence of the exposure/expression lad for these diseases.

It is concluded that the results for all malignancies support the hypothesis that living near the East coast of Ireland carries excess risk of cancer for women of all ages and men of the age group 25-44, and that this excess risk is driven by some factor which is specific to the narrow coastal strip bordering the Irish Sea. The finding supports the hypothesis that ecposure to radiation from Sellafield may be responsible.