British Nuclear Test Veterans Association
|
|
|
|
|
||||
|
|
|
|
|
|
||||
|
|
|
|
|
|
Mortality and Morbidity of Members of the British Nuclear Tests Veterans Association and the New Zealand Nuclear Tests Veterans Association and their Families
SUE RABBITT ROFF
University of Dundee Centre for Medical Education
Published in Medicine, Conflict and Survival (Volume 15, Supplement 1, July-September 1999)
Part One
Mortality Profile from a Pre-defined Sample of Death Certificates of Veterans of United Kingdom Nuclear Weapons Tests
Background
More than 20,000 members of the United Kingdom Armed Forces - all of them men, most of them in their twenties - were required to be present at UK nuclear weapons tests which took place in Australia (Monte Bello, Maiden Island, Emu Field, Maralinga and Woomera) and at Christmas Island (now Kiribati) between 1952 and 1963, including the 'cleanup operations'.
Methods. An hypothesis-generating retrospective cohort study was carried out on 608 notifications of death received by the British Nuclear Tests Veterans Association (BNTVA) since its formation in 1983 and until the initiation of this study on 1 July 1997. The membership of the BNTVA is approximately ten per cent of the total 22,000 UK men who participated in the UK atmospheric and other nuclear weapons tests and trials.
Findings. Nearly two-thirds of the men in this sample died before the age of 60 years; a similar proportion of men died of cancers which are of possible radiogenic origin. Nearly two-thirds of the men died within 30 years of their presence at the weapons tests; age at the time of the tests (ATT) had no influence on ultimate longevity. A third of the cancer deaths occurred in the 1990s, after the cut-off date of earlier morbidity and mortality studies done by the UK National Radiological Protection Board (NRPB). More than half the deaths from multiple myeloma (N = 21; 55%) and from brain cancers (N = 17; 52.9%) occurred after 1989.
Interpretation. The present study suggests that the effect of exposure to ionising radiation extends into the third and fourth decade, a hypothesis compatible with data on Japanese atomic bomb survivors. Evidence of premature death from radiogenic cancers and other causes within this predefined sample emphasizes the need for continuing follow-up of all nuclear test participants, both epidemiologically and clinically.
Introduction
The initial aim is to develop a working profile of the causes of mortality of a particular membership group by conducting a retrospective cohort study and comparing it to profiles of similar groups of European men.
Methods
The methods used in a retrospective cohort study1 of the causes of death of the 759 vegetarian society members whose deaths were recorded in society records, and whose death certificates could be traced, were emulated in this study.
The NRPB/Ministry of Defence (MOD) cohort of 21,358 men who had participated in the UK nuclear weapons tests2 was constructed from various sources, including the membership rolls of the British Atomic Veterans Association, the British Nuclear Tests Veterans Association (BNTVA) and the Royal British Legion. The present studies are of the 2,087 current members of the BNTVA, 608 of whom are the survivors of deceased veterans. The birth years of the members of the BNTVA are very similar to those of the full cohort: 24% were born before 1930 (28% of full cohort); 72% were born in the 193 Os (66% of the full cohort) and 4% in the 1940s (6% of full cohort). Any age bias in the present sample is therefore slightly towards a younger membership of the organisation than that of the full cohort as reported by the NRPB/MOD.
The 608 death notifications to the British Nuclear Tests Veterans Association (BNTVA) were databased and coded, together with information obtained from a questionnaire which members complete on joining the BNTVA (which specifically requests their consent that the information supplied may be used for research). Death certificates had been submitted by veterans' survivors in 455 instances, and causes of death were ascertained from these and, in five cases, from information supplied from the deceased's family where no death certificate was available.
The pattern of deaths of the 455 death-certificated nuclear veterans was compared with the patterns reported for other male cohorts from the post war period - studies of: nitrate fertiliser workers at a factory in Billingham, UK3; butchers in Geneva, Switzerland4; men who had manufactured mustard gas in the county of Cheshire, UK, during the Second World War5; cancer mortality among members of a vegetarian society1; and mortality among UK veterinarians6.
Results
Longevity or the lack of it
In the BNTVA sample, less than one-fifth of those dying lived beyond 65 years of age. There is a particular concentration of deaths between 50 and 65 years of age. Clustering of death appeared 25 to 30 years after the presence of the veterans at the tests, for those both under and over age 30 at the time of the tests. Longevity was therefore reduced in those under 30 when they attended the tests. The extra longevity in those over the age of 30 at their presence at the tests was, however, statistically experienced before their attendance.
Age at death
 |
Figure 1 |
 |
Figure 2 |
 |
Figure 3 |
 |
Figure 4 |
 |
Figure 5 |
 |
Figure 6 |
Of the 608 deaths notified to the BNTVA, years of birth and death were available for 531 men (87.3%). The numbers of this sub-group dying at various ages are shown in Figure 1. The average age of death was 56.0 years, less than the figure of 57.6 years for males in the former Soviet Union, which has been the subject of some comment7. Of the 531 men, 22% died before the age of 50 years, another 42% died before the age of 60 years, and 29% before the age of 70 years (that is, 64% under the age of 60). The same data are presented as percentages in Figure 2. Almost 60% of the sample died between the ages of 50 and 65 years, in what may be considered as middle age for individuals living in a northern hemisphere industrialized society (Figure 2).
Information on the year of their deaths was available for 543 men (Figure 3), and the age at which the veterans were present at a test or series of tests is known for 535 men. Forty per cent of deaths had occurred by 25 years after attendance at the tests, 63% by 30 years, and 90% by 35 years (Figure 4).
Of the men present at the tests, 173 were more than 30 years old at the time of the tests (ATT). The average life span of this sub-group was 62 years. They averaged 36 years of age ATT, and survived on average 26 years after the tests. About three-fifths of their life span was therefore lived before the tests and two-fifths afterwards.
The remaining 362 men, who were aged less than 30 years ATT, died at an average of 53 years. Their average age ATT was in the mid-twenties. They too therefore survived on average about 26 years after attendance at the tests, but for this latter sub-group 26 years represented only about half of their life span. The longevity overall thus lay disproportionately with the third of the sample who were aged more than 30 years ATT, but was, in a statistical sense, experienced before these men attended the tests.
The trend towards shorter life spans for those who were younger ATT is consistent throughout the cohort. The youngest men, those born in the 1940s, were in their late teens at the time they attended the tests; there were only ten individuals in this sub-sample. Their average life span was only 44.5 years, with a range of 26 to 55 years. They represented 2.0% of the sample, but experienced only 1.6% of the total life years. Men born in the 1930s numbered 287, and were in their 20s ATT. Their average life span was 51.6 years; they experienced only 49.6% of the life years, although constituting 54% of the sample. Another 168 men were horn in the 1920s and were in their thirties ATT. Their average life span was 59.6 years; they experienced 33.9% of the life years though forming only 31.7% of the sample. A further 68 men were born before 1920 and were in their forties or older ATT. The average life span of this last sub-group was 65.6 years, they experienced 14.8% of the life years, but comprised only 12.7% of the sample.
Causes of death
Death certificates were available for 450 veterans and information on causes of death from the families of a further five men. Cancer was certified as the cause of death of 311 men, 68.4 % of the sample of 455 veterans for whom death certificates were available (Figure 5). For over a quarter of the cancers, the primary site could not be specified from the certificate. Most of the cancers are those which are eligible for compensation under Title 38, United States Code Annotated §1112.
The first death from oesophageal cancer did not occur until 1981; the 12 men who died of this cause were aged between 51 and 69 years.
The first death from multiple myeloma (MM) occurred in 1965. The ages of the ten men who died from MM before 1989 ranged from 41 to 74 (mean 58.7). The ages of the 11 men who died from MM after 1990 ranged from 51 to 80 (mean 63.4). Excluding the two who died at 41 years and 80 years, the average age of death from MM would be 55.6 years.
Eighteen men died from leukaemias. Those born earliest (in 1908, 1916, 1918 and 1925) died of chronic lymphatic leukaemia or chronic granulocytic leukaemia (in 1959, 1992, 1993 and 1992 respectively). The remaining 14 died of acute myeloid leukaemia. When the 13 men who died of leukaemias before 1990 are compared with the five who died after 1990, there is a striking difference in average age at death 97 47.4 years for the former and 67.6 years for the latter group. Ages at death ranged from 30 to 61 in the former group and 59 to 76 in the latter group.
While the ages at death of the eight men who died from Hodgkin's disease ranged from 22 to 59 years, two were aged 39, one 47 and one 49 years at death. The first three deaths occurred before 1972. Five men died of non-Hodgkin's lymphoma (NHL) before 1990 and four after; in both subsets the average age of death was 57 years although the ages at death ranged from 46 to 70 years. More detailed information on multiple myeloma, NHL and leukaemia is being prepared for publication.
Seventeen men died of brain cancers, eight before 1990 and nine after. The average age of death of those in the first group was 48.3 years, but it was 58.3 years in the second group. One man died of brain cancer at age 27 (in 1965); one at 36 (1966), four in their forties, eight in their fifties, two in their sixties and one at 70 years. A further 21 men died of cerebrovascular diseases, 14 before 1990 and seven after.
The ages at death of the 12 men who died from cancer of the prostate ranged from 54 to 73 years; the first deaths from this cause occurred in 1978 among men in their late fifties; five of them before age 60.
Three men died from cancer of the testis (aged 35, 38 and 62 years); one man died from cancer of the urethra, aged 26 years; three men died of cancer of the bladder, in 1988, 1989 and 1990 aged 51, 59 and 65 years respectively. Six men died of cancer of the kidney, rwo in their late thirties and the other four in their fifties. Three men died of colorectal cancer in their thirties, four in their forties, and six in their fifties.
The first of the six deaths from melanoma occurred in 1983, the ages of the men at death ranging from 51 to 63 years. Two men, born in 1927 and 1928, died of cancer of the thyroid in 1982 and 1996 respectively. The five who died of cancers of the larynx, parotid, maxilla, mouth, and of nasopharyngeal cancer, all died in 1990 or later and their ages ranged from 57 to 61 years. Two men died from cancer of the breast, one aged 64 and one aged 75, in 1990 and 1993 respectively. The three men who died of bone cancers were aged 59, 60 and 63 years; one (who was born in 1916) died in 1975 and the other two in 1991 and 1992.
Comparisons with other cohorts
Mortality from various categories of disease in this cohort of nuclear test veterans was compared with mortality in the same categories among the other male occupational cohorts cited under Methods (see above). The findings are shown in Figure 6. While 68% of the present sample of test veterans died of neoplasms, only 30% of the sample of nitrate workers3 and butchers44 did so, and only 18% of mustard gas workers5 and veterinarians6, and 16% of vegetarians1. Unlike the nuclear veterans, the other groups are more likely to die of circulatory or respiratory diseases than of cancer.
Discussion
Death certificate studies probably underestimate the incidence of specific causes of mortality. Based on the US-Japan Atomic Bomb Casualty Commission/Radiation Effects Research Foundation series of more than 5,000 necropsies, the accuracy of death certification for 12 disease categories has been examined8 9 10, with assessment of the effect of potential modifying factors on agreement and accuracy. The overall agreement between death certificate and necropsy diagnoses was only 52.5%. The detection rate was highest for neoplasms, but almost 25% of cancers diagnosed at autopsy were nevertheless not cited on death certificates. Confirmation and detection rates were above 70% for neoplasms and external causes of death only. Overall agreement decreased with increasing age at death, and was worse for deaths occurring outside of hospital. There was some suggestion that agreement improved over time, but no indication that radiation dose, sex, city of residence, or inclusion in a biennial clinical examination programme influenced agreement. Some of these factors may well apply in the UK, especially in view of the frequent use of the term 'carcinomatosis' in the death certificates studied in this sample. In general, the present researcher (who carried out all the coding on the data), observed that older death certificates, that is those dated before 1970,tended to be more cursory and less informative than more recent ones. However, there was sufficient information on all the death certificates available to classify the causes of death in the broad categories used by Kinlen and colleagues1 in their study of cancer mortality in a similarly predefined sample of UK vegetarians.
There is a strong suggestion of increased overall mortality in the group studied here (Figure 1), though this needs to be confirmed by a full rereview of the full cohort of UK nuclear test veterans. At the same time, the suggestion in Figure 4 of a clustering of deaths about 30 years after attendance at the tests, irrespective of age at that time, should be carefully examined.
Clearly the number of deaths from neoplasms within the current sample differed from other cohorts of men studied since the Second World War, showing at least a 'doubling effect' even after allowance for any bias inherent in working with a predefined sample (Figure 6). In contemplating the bias issue, it should be noted that although 608 deaths were notified to the membership secretary of the BNTVA, only 455 families submitted death certificates. This suggests that claims for compensation and/or pensions for men and their widows was not an overriding consideration in joining the organization. With regard to the nature of the cancers, the primary site could not be identified from the information available in 80 out of the 311 cancers reported in the present survey (Figure 5). The primary might be recognizable from hospital and autopsy records in some cases. This would be an important part of a re-examination of the cancer mortality and morbidity of the full cohort. Factors in the lifestyle of the veterans predisposing to cancer, in particular their smoking habits, would need to be taken into account, but also the possibility of synergy between these factors and radiation exposure.
Seventeen deaths were attributed to brain cancer in this sample. This is equivalent to the incidence of brain cancers found in a population-based cohort of 959 patients diagnosed with seizure disorders while residents of Rochester, Minnesota, between 1935 and 1979.11 In that sample, for all cancer sites combined, there were 65 cases with a standard morbidity ratio of 1.4. Most of the excess was attributable to a 22-fold increase in the incidence of primary brain tumours in these patients. The incidence of meningioma among Nagasaki atomic bomb survivors in five-year periods since 1975 has been reported as 5.3, 7.3, 10.1 and 14.9 cases per million of population.12 The incidence of meningioma among the Hiroshima atomic bomb survivors also increased between 1975 and 1994. The authors conclude that meningioma is the eighth radiation-related tumour occurring in atomic bomb survivors, along with acute leukaemia, thyroid, breast, lung, stomach, colon and skin cancer.
While increased age ATT does not seem to have served as a protective factor in the latency of the conditions induced by possible radiation exposures in the present sample, young ATT may have rendered those aged less than 30 years more vulnerable. This is compatible with the findings of a study of Scottish women13 given radiation therapy for metropathia haemorrhagica between 1940 and 1960 (average follow-up of 28 years). For some cancers, the standardised mortality ratio (SMR) was higher 30 and more years after irradiation than at five to 29 years, indicating that the effects of exposure persist for more than that time.14
More than a third of the deaths reported in the current survey occurred after 1989 and more than half the deaths from multiple myeloma and brain cancers occurred since then. This finding should be considered in conjunction with the pattern of deaths from non-Hodgkins lymphoma. Revised understanding of the latency of hazards from ionising radiation exposure is being drawn from research on survivors of the Japanese atomic bombings15 16. The present study reports more than double the normal occurrence of cancer deaths in a male 91occupational92 cohort of the post-war years. While this study is intended to be hypothesis-generating, the consistency of the findings on the timings of the deaths in its cohort with those on latency in the Japanese survivor cohort may be considered important. It will be important to compare the data with the studies of Estonian workers who went to Chernobyl.17 18 19
The findings of this study thus strongly suggest that the termination of the NRPB survey2 in December 1990 was premature. In particular, it reports double the mortality from multiple myeloma compared with that stared in the NRPB studies up to December 1990; half of this excess mortality occurred after the NRPB cut-off date.
Read Part Two