1. The Green Audit Irish Sea Study

This study began in December 1997 as part of the research effort associated with the legal case, Short and Others vs BNFL. It was a contention of the litigants that radioactive pollution of the Irish Sea by the BNFL nuclear reprocessing plant at Sellafield in West Cumbria was a danger to the health of persons living near the east coast of Ireland.

There has been anecdotal evidence of increases in cancer, leukemia and other genetic-based illness near the Irish and Welsh coasts. Because the Irish have had no national cancer registry over the period of operation of Sellafield, it was of interest to look for any sea proximity effect that might support the plaintiffs' claims.

Cancer registry data covering small areas is seldom made available to independent researchers. Between 1992 and 1996, the publication of three books on the effects of low-level radiation from man-made fission-product pollution and cancer in Wales (Busby 1992, 1994, 1995) advanced the thesis that a comparison of cancer incidence across two countries, Wales and England, that were differentially polluted by radioisotopes from the early 1960s global weapons-testing fallout suggested that it was internal exposure to such man-made radiation that was the main cause of the sudden increases in cancer in Wales relative to England, twenty years after the exposure. The resulting controversy put pressure on the Wales Cancer Registry to allow Green Audit access small area data which they had hitherto refused to release. Immediately following this the registry was closed down by the Welsh Office and responsibility for cancer intelligence was handed to a new Wales Cancer Intelligence Unit.

The data obtained by Green Audit was for 230 areas of residence in Wales, based on pre-1974 local authority boundary administrative areas. The period covered was 1974 to 1989 and incidence data was by sex, site and five-year age-group. This level of resolution permits the examination of the health effects in Wales of proximity to the Irish Sea. This part of the Irish Sea Study examines the appropriate age- standardised incidence of twelve different types of cancer as a function of distance from the Irish Sea.

We report here preliminary results for childhood leukemia age 0-4.

2. Method

For the purposes of a preliminary examination of the effect of the Irish Sea on leukemia in children aged 0-4, the south-eastern industrial counties of Gwent, South Glamorgan, Mid-Glamorgan and West Glamorgan were filtered from the database (but see below). This was for two reasons. First because the areas are classified as industrial areas by the Office of National Statistics and therefore comparisons of cancer rates with the rural areas we have chosen for the study might not be epidemiologically appropriate. Second, because the effect we are investigating is an Irish Sea coast effect and these counties border the Bristol Channel which itself takes radioactive emissions from Hinkley Point and Berkeley and Oldbury nuclear power stations. The remaining counties were Clwyd, Gwynedd, Dyfed, Pembrokeshire and Powys. All except Powys border the Irish Sea. Powys was retained as a control for distance since the rural population is comparable with that of the rest of the study area.

Since the size of an administrative area is set on the basis of population, the geographical size of the "Areas of Residence" units varies considerably. Population data for these areas was obtained by aggregating the appropriate wards given by the UK Office of National Statistics and using ward-level populations obtained from the 1981 census.[See explanatory note on the jargon]

Using the population of England and Wales 1979 and the cancer rates for this population as a base, the expected numbers of leukemia cases in the age group 0-4 were then calculated for each small area and totalled over the 16 years 1974-89.

Distance of the approximate centroid of population from the sea coast was then measured and tabulated. The small areas were then split into aggregate groups according to their inclusion within a set of boundaries at varying distances from the coast, each set having approximately the same number of small areas within it. The observed numbers of cases of leukemia in the age group 0-4 over the period 1974-89 were then compared with the total expectation based on the aggregated population at risk and the quotient expressed as 'Relative Risk'. Significance was established by calculating a P value based on the Poisson distribution-based probability that the observed number of cases or more could have occurred given the mean expectation.

3. Results

Results are given in Table 1 and Figure 1. The first column of Table 1 gives the proximal and distal boundaries in kilometres of the strips running parallel with the coast into which an aggregated number of areas are grouped. The observed and expected numbers of cases are given in Columns 3 and 4. In Column 2 is given the mean distance from the sea together with its standard deviation. In the fifth column is given the 'Relative Risk' (Observed/ Expected) and the last two columns show the cumulative Poisson P value and the numbers of areas involved in each aggregation. Because of the existence in Wales of 14 administrative areas consisting of more densely populated seaside communities it was possible to use these to establish relative risk within 0.8 km of the sea.

Table 1. Relative Risk of leukemia (ICD 204-8) in children aged 0-4 over the period 1974-89 at various distances from the Irish Sea.
(Source: Wales Cancer Registry- Welsh Areas of Residence Datafiles).

Range of distance from sea <sd>(km)	Mean distance (Std. dev)	Observed cases	Expected Cases	Relative Risk	P value significance (Poisson)	Number of wards aggregated
<0.8	0.5 (0.1)	23	4.99	4.6	0.0000	14
0.9<sd<2	1.0 (0.0)	19	5.75	3.3	0.0001	10
2.1<sd<5	4 (0.0)	18	6.23	2.9	0.0001	12
5.1<sd<11	8.24 (0.66)	31	13.0	2.4	0.0000	17
11.1<sd<21	16.4 (2.8)	17	5.31	3.2	0.0000	14
21.1<sd<41	34.7 (5.8)	19	7.85	2.4	0.0005	17
41.1<sd<61	51.8 (5.8)	8	5.6	1.4	0.203	19
51<sd<71	61.4 (4.8)	3	3.41	0.9	0.660	14

Figure 1
Leukaemia age 0 - 4 in Wales
Relative risk 1974 - 1989 by distance from the Irish Sea

Mean distance from Irish Sea (Km)

---

4. Discussion

These crude results show a considerable and significant sea proximity effect. Living close to the Irish Sea in Wales appears to give an astonishing 4.6- fold increased risk of childhood leukemia in the 0-4 age group relative to England and Wales 1979. The numbers were sufficiently large in each group to establish a high degree of statistical significance for this result. Moreover, apart from the group within the 11 to 21 km strip, the relative risk fell monotonically towards the English border, some 70 kilometres from the coast. The border with England showed incidence rates over the period very similar to those expected for England and Wales. The trend, despite the 11 - 21km group, was statistically significant ( Chi² for linear trend = 7.75 ; P = 0.005 ). The increased risk in this 11 to 21 km strip was caused by some high levels of childhood leukemia in mountainous areas, particularly in North Wales, summarised in Table 2.
These areas of Snowdonia were particularly badly contaminated by both weapons testing fallout in the period 1959-63 and more recently by the Chernobyl accident where high levels of contamination occurred.(Cawse et al.1988) Three of the cases in this strip occurred immediately after the Chernobyl contamination and are very probably ascribable to it (Busby and Cato, 1998).

Table 2. Leukemia 1974-89 in mountain areas of Snowdonia contributing to high relative risk in 11-21 km sea proximity group.

Area	Observed cases	Expected Cases	Relative Risk
Betws y Coed	3a	0.53	5.7
Llanrwst	4b	0.14	28.5
Nant Conwy	2	0.64	3.1
Ffestiniog	1	0.29	3.45

a One of these observed shortly after Chernobyl
b Two of these registered shortly after Chernobyl
Removal of only these three cases from the 17 observed cases in the 11 to 21 km group reduces the RR of this group to 2.6, roughly the same level as the groups either side.

Clearly these results demonstrate the existence of some factor which increases the risk of childhood leukemia at the edge of the Irish Sea. In view of the known association between leukemia and exposure to ionizing radiation (BEIR V, 1990), exposure to radioisotopes released from BNFL Sellafield and subsequently brought ashore along the coast of Wales seems to be the most likely explanation. Excess risk of childhood leukemia with relative risks from 10 to 15 times control incidence has now been shown near the reprocessing plants at Sellafield and Dounreay (Beral et al, 1993 ) and La Hague. (Viel et al, 1996, 1997).

In the UK, increases in childhood leukemia have also been noted near Aldermaston and Harwell (Busby and Cato, 1997). What all these sites have in common is that, in the form of liquids dusts and gases, they release large quantities of man-made radioactive fission-product isotopes to the environment. Following the discovery of the Sellafield leukemia excess in 1983, a number of government-funded committees have attempted to establish the causes of these leukemias. They have, to date, continued to assert that radiation cannot be the cause and a number of alternative hypotheses have been advanced.(COMARE, 1996) The most widely discussed of these is the 'population mixing hypothesis' of Kinlen (Kinlen,1988) whereby leukemia is assumed to be of viral origin and population mixing in the remote areas where nuclear plant is sited is assumed to reduce herd immunity to such a viral challenge. Applied to results given here it cannot be considered a reasonable explanation since the population mixing in Wales does not fall off monotonically with distance from the sea , nor is it great enough to explain the magnitude of effect.

The basis of the belief that levels of exposure to ionizing radiation cannot be the cause of the nuclear site leukemia clusters is the risk model for radiogenic leukemia presently used by the International Commission on Radiological Protection (ICRP 60, 1990) This model is based on the leukemia yield and other cancer and genetic effects in the survivors of the Hiroshima bomb in 1945. The model has been under considerable pressure in the last ten years, particularly following the observation of health effects after Chernobyl. (STOA, 1998) In particular, Busby argued as early as 1992 that considerable error in the risk factors for radiogenic effects might result from the ICRP assumption that external acute and internal chronic irradiation were equivalent. Other serious criticisms of the risk model of the ICRP have been made by Stewart (STOA, 1998) who has shown that the control group and study group of the Hiroshima survivors' analysis were not drawn from a comparable population, nor from a population that was comparable for risk purposes with those whose exposure is presently being assessed. For a fuller analysis see Busby, 1995.

The radioisotopes that are released in the fission process did not exist on earth prior to 1945, yet many of them mimic elements that are utilised by physiological processes at the molecular level. Moreover, there is a novel class of these isotopes which have sequential double decay schemes and in principle can damage the cellular repair mechanisms by first causing sub-lethal damage and then attacking a second time during the repair replication cycle. This 'Second Event' hazard has been argued to introduce an error in the risk model applied to exposure to such isotopes of between 500 and 1000 times (Busby, 1995, 1997).

The most recent observation which supports the claim of an error in the risk model for leukemia is that of a highly significant increase in infant leukemia in Wales and Scotland following the Chernobyl fallout. This study of a cohort whose exposure was well defined enabled a minimum error of 100 times to be established for the risk for radiogenic leukemia in infants who were exposed in utero to fission product irradiation. (Busby and Cato, 1998).

The contamination of the coastal areas of Wales by radioisotopes from BNFL Sellafield is well established and measurements are made regularly by MAFF and other authorities. Plutonium comes ashore from the Irish Sea and is measured in sheep droppings up to 20km inland (Cawse et al, 1988). Measurements of plutonium made in the 1970s showed a continuous fall-off with distance from the Irish Sea across the whole of Wales and into England.(Cawse et al, 1986) More recently, plutonium measurements in children's teeth show a similar distance effect from Sellafield. (O'Donnell, et al, 1997)

Radioisotopes come ashore in a number of ways and exposure is through inhalation of dust or sea spray and ingestion of contaminated food, water or milk. Recently Viel (Viel et al, 1997) showed that the two main excess-risk factors associated with childhood leukemia in his case-control study of the population near the La Hague reprocessing plant were playing on the beach and eating shellfish. Proximity of a population to an estuary area where there is a large expanse of sand has also been shown to carry excess risk for childhood leukemia by Alexander (Alexander et al, 1990)). In a recent analysis of the leukemia clusters near the weapons facilities in Newbury, Busby has drawn attention to high concentrations of radioactivity on dust particles trapped in passive filters and has suggested that the movement and distribution of small radioactive particles may be influenced by the electrostatic charge acquired by them as they decay and the way in which these micron sized charged dust particles are affected by the earth's electrostatic field. (Busby, 1998a) Positively charged particles resulting from beta-particle emission would be attracted to the ground whilst negatively charged particles, e.g.plutonium oxide, would be repelled and driven into the air , to be blown ashore and collect and concentrate at electrostatic singularities. Radioisotopic pollution from Sellafield may not be the whole cause of the exposure near the sea since weapons fallout in the period 1959-63 will have been washed to the Irish Sea and may contribute to the total pollution burden following resuspension of sand and dust in estuary areas.

The calculation of health detriment by the UK National Radiological Protection Board depends upon the use of the same risk factors derived from Hiroshima. It is their model of exposure that is used to set limits for the licensing of emissions from the plant. The results reported here represent confirmation of major errors in the risk models used and argue for an immediate cessation of releases of these substances to the environment and urgent research on their micro distribution, dispersion and associated health risks.

5. Possible Errors in the Preliminary Analysis

We excluded the south eastern industrial counties of Wales for reasons given earlier. Their inclusion would not have affected the overall results, however, since the Relative Risk for the aggregate of 66 administrative areas of south east Wales was a modest 1.5, with 118 cases observed and , 77 cases expected.

There will be a slight error in using the 1981 census populations throughout. In previous analyses we have found that, because of the trend in birth rate over the whole period, this error in England is not more that 10 percent. We intend to refine this study using a linear interpolation between the 1971, 81 and 91 census data to establish the degree of error involved by using this approximation. Such a level of error is unlikely to affect our conclusions.

The analysis does not make any allowance for the variation of childhood leukemia with socioeconomic group. Again, this can be addressed but, such a refinement is unlikely to affect the overall conclusions. It may be that the kind of person choosing to live by the sea is particularly susceptible to leukemia. However, this cannot explain the effect seen in the continuously declining trend with distance from the sea.

One remaining possibility is error in the cancer registry data. Although the Wales Cancer Registry admitted in 1996 to errors in the registration of bone cancer in Wales, they maintained that their childhood leukemia data had been carefully checked and validated (Cotter, 1996).

Conclusions

Between 1974 and 1989, relative risk of leukemia in children aged 0-4 in coastal areas of Wales adjacent to the Irish Sea was over four times that expected on the basis of national figures for England and Wales (RR = 4.6;
P = 0.0000). Discounting outliers, this effect persisted up to 20 km from the sea and fell off continuously with distance through the whole of Wales to its border with England. The effect is most probably caused by exposure to radioisotopic pollution from the BNFL plant at Sellafield.

---

References

Alexander, F.E, Cartwright, R.A, MacKinnon, P and Ricketts, T.J, Leukemia incidence, social class and estuaries: an ecological analysis. Journal of Public health medicine, 12, 109-117, (1990)

BEIR, 1990, Health Effects of Exposure to Low Levels of Ionizing Radiation, BEIR V Washington: National Academy Press

Beral, V, Roman, E and Bobrow, M (1993), Childhood Cancer and Nuclear Installations. London: British Medical Journal

Busby, C.C (1992), Low level radiation from the nuclear industry: the biological consequences. Aberystwyth: Green Audit

Busby, C.C. (1994), Radiation and Cancer in Wales. Aberystwyth: Green Audit

Busby, C.C. (1995), Wings of Death: Nuclear Pollution and Human Health Aberystwyth: Green Audit

Busby C.C.(1997), in The Health Effects of Low Level Radiation: Proceedings of the Symposium held at the House of Commons, April 24th 1996 ed. R.Bramhall, Aberystwyth: Green Audit

Busby Chris, (1998a), Childhood leukemia and radioactive pollution from the Atomic Weapons facilities at Aldermaston and Burghfield in West Berkshire: Causation and Mechanisms. Occasional Paper 98/1, Aberystwyth: Green Audit

Busby, Chris and Scott Cato, Molly (1997),'Death rates from leukemia are higher than expected around nuclear sites in Berkshire and Oxfordshire.' British Medical Journal, 315: 309

Busby, Chris and Scott Cato, Molly (1998), Increases in Leukemia in infants in Wales and Scotland following Chernobyl: evidence for errors in statutory risk estimates. Occasional Papers No 98/2, Aberystwyth: Green Audit ( on this site)

Cawse, P.A and Horrill, A.D, (1986), A Survey of Caesium-137 and Plutonium in British Soils in 1977 Report /1030 C(10) Harwell: Atomic Energy Research Establishment

Cawse,P.A, Cambray,R.S, Baker, S.J, and Burton, P.J (1988), A Survey of the Background Levels of Radioactivity in Wales, Cardiff, Welsh Office)

COMARE, (1996), Committee on medical Aspects of Radiation in the Environment: Fourth Report. The incidence of cancer and leukemia in young people living in the vicinity of the Sellafield site, West Cumbria. Wetherby: Department of Health

Cotter, Mary, Wales Cancer Registry (1996), personal communication

O'Donnell,R.G, Mitchell, P.I, Priest, N.D, Strange, L, Fox, A , Henshaw, D.L, and Long, S.C (1997), 'Variations in the concentrations of Plutonium, Strontium-90 and total alpha emitters in human teeth colllected within the British Isles.' Science of the Total Environment 201: 235-243

ICRP, (1990), International Commission on Radiological Protection: ICRP 60. 1990 recommendations of the ICRP. Oxford: Pergamon Press

Kinlen, L.J, (1988), 'Evidence for an infective cause of childhood leukemia: comparison of a Scottish New Town with Nuclear reprocessing Sites in Britain', Lancet,ii,1123-7

STOA: European Parliament Science and Technology Options Assessment,(1998), Proceedings of a workshop on survey and evaluation of criticisms of basic safety standards for the protection of workers and members of the public against ionizing radiation. Editor P.A.Assimakopoulos, Brussels: European Parliament

Viel, J-F, Poubel, D. and Carre, A (1996) ,'Incidence of leukemia in young people and the La Hague nuclear reprocessing plant- a sensitivity analysis,' Statistics in Medicine, 14:2459-2472

Viel, J-F and Poubel, Dominique, (1997), 'Case Control Study of Leukemia among young people near the La Hague reprocessing plant: the environmental hypothesis revisited.' British Medical Journal 14: 101-6

---

Correspondence to C.C. Busby, Glyndale, Trinity Road, Aberystwyth, Cardiganshire SY23 1LU UK email christo@grenaudit.org