Radiation Risk Studies in Military Populations
S. Jablon *
As an introduction, it will be useful to consider why it is desirable to do studies of radiation risk, for which of the possible purposes military populations are appropriate, and why. It has been well known for many years that ionizing radiation is a cause of subsequent cancer, which may not occur until several decades after exposure. What, then, is the need for studies?
There are two principal reasons for doing more studies of radiation risk in humans. First, although we know about the late effects of exposure to radiation in a general way, our quantitative information is somewhat uncertain; we can specify the cancer risk per unit of radiation dose only with an accompanying uncertainty of at least a factor of two. Second, some persons who have been exposed to radiation and who subsequently develop cancer, in the belief that the radiation was the cause of the cancer, claim compensation from the author of the exposure, whether the federal government, a national laboratory, a commercial nuclear power generating utility, or a physician.
We need better and less uncertain quantitative information about the risks of radiation carcinogenesis--how many extra cases of cancer will develop per unit of radiation exposure, and when. The information is necessary in order to set “safe” limits on exposure for persons who are exposed occupationally; so that populations involved in an event like the accident at the Three Mile Island generating plant can be informed, warned, or reassured, depending on the exact circumstances; and to enable physicians to balance risk versus benefit when considering the advisability of a radiological examination such as mammography as a screening procedure for early breast cancer. We need to know what are the kinds
* |
Seymour Jablon is with the National Cancer Institute in Bethesda, Maryland. |
of cancer that are induced by radiation; leukemia is certainly a late effect, as are cancers of the lung and of the stomach, but we are still uncertain about the lymphomas or liver cancer or many others. What are the risks imposed by various levels of radon concentrations in homes, schools, or work places? To obtain more and better data concerning these points we need large-scale studies in which there is good quantitative data concerning the magnitude of individual exposures. Studies of the kind just described can be characterized as “scientific studies.”
The second class of studies, for lack of a better term, can be called “population studies.” Such studies as those of persons who lived near Three Mile Island; of residents of southwest Utah in the 1950's, at the time of above-ground testing of nuclear weapons at the Nevada Test Site; or of veterans who, while in service, were present at nuclear weapons tests belong to this class. Studies of the population around Chernobyl will also be of this kind, since the possibility of deriving reasonably accurate estimates of individual radiation doses seems remote. In all of these instances it seems quite possible that estimates of the probable range of individual radiation doses, and their average, can be obtained, but the results of such studies, considered for their scientific value, will probably be no more than confirmatory, in a general way, of the risks obtained from other studies. If such studies have, at best, marginal value for science, why should they be done? For two reasons:
-
The people involved are concerned, or at least many of them are, since it is general knowledge that radiation is a cause of cancer.
-
Many of those who have been exposed, no matter how low the radiation doses may be in individual cases, will attribute any cancer that may develop to the previous exposure.
Since about 20 percent of all deaths in the United States are caused by cancer, and at least 30 percent of all people have some form of cancer diagnosed during their lifetimes, the potential for claims is enormous. This has been demonstrated with respect to the off-site residents of Utah, veterans who were present at nuclear tests, and the veterans who participated in the occupation of Hiroshima and Nagasaki just a few weeks after the atomic bombings of those cities. The statement that, in some instances at least, the exposure levels were too low to result in detectable effects is met with skepticism. Population studies, if they show that the populations in question have cancer rates that are no different from those of the general population, may demonstrate that claimed excesses of cancer are not well founded; alternatively, they may show that the estimated exposure levels are erroneous or that the generally accepted estimates of risk per unit exposure have been drastically underestimated.
Studies of veterans generally fall into the second class; that is, they are not primarily designed to produce scientific information about cancer risks per rem, if for no other reason than that reasonably accurate information about radiation exposures is hardly ever available. While many of the participants in tests at the Nevada Test Site and at the Pacific test area had film badges, the primary purpose of the badging procedure was to ensure that the participants were not exposed to radiation fields that exceeded certain maximum “allowable ” limits, not to lay the foundation for future estimates of radiation risk per rad. Accordingly, the badges were not handled with the care that would otherwise have been mandatory, records of readings were sometimes lost or otherwise mishandled, and there was little effort at quality control. Great care in handling and reading the film badges, in fact, would not have been appropriate--the procedures that were followed accomplished the task that they were designed for: to ensure that the exposures were less than some stated amount, not to lay the foundation for future studies.
It is clear why little thought was given to obtaining scientific information. The recent BEIR V report (1) estimates that a radiation dose of 0.1 Sv (10 rem) to a population of 100,000 young men would result, roughly, in about an extra thousand deaths from cancer, over the remaining lifetimes of that population. That is, the population 's cancer mortality would be increased by about 5 percent from the 20,000 cancer deaths that can be expected in an unselected population, and to observe the excess, the population would have to be followed for 60 years or more. Given that few veterans who were present at nuclear weapons tests had exposures approaching 10 rem, it seems clear that science would have little to gain from a study of those veterans.
If the risks to the veterans are so small as to be unobservable in a scientific study, what is the problem? Why are we, or they, concerned? Because although we, as scientists, believe, and think we know, that the risks are so small as to be unobservable, members of the general public, including veterans, do not know that, and attempts to persuade them are met with blank disbelief; it is claimed that government officials are liars and that there is a conspiracy to “cover up” the damage that has been done to innocent people. Since, in fact, some officials did, for whatever reasons, conceal from the affected populations facts concerning the magnitude of radioactive fallout, the distrust is understandable.
The purpose of studies of veterans, then, is to determine whether large classes of veterans who have been exposed to ionizing radiation are, on that account, subject to larger risks of subsequent cancer than they otherwise would have been.
So much for the objectives, and the problems, of studies of veterans. What are the advantages? There are two:
-
Individuals are well identified in military files. In addition to the name, a military identifying number, which is now the Social Security number (SSN), is available, which facilitates tracing to learn outcomes.
-
The files of the Veterans Administration, chiefly the Beneficiary Identification and Records Locater Subsystem (BIRLS), are resources that enable a quick and relatively easy determination of whether a given veteran is alive or dead. A statement of the cause of death, transcribed from the death certificate, is often available, and when it is not, there is usually sufficient information to enable a request to the state in which death occurred to obtain a copy of the death certificate itself. Information available from the BIRLS file can be supplemented by resort to the files of the Social Security Administration (SSA), which similarly, provides the information required to enable resort to the state of death. This is an enormous advantage, compared with the difficulty of tracing ordinary civilian rosters, which usually involve long, arduous, and expensive procedures that are not completely successful.
Studies of veterans that have been done usually have been of weapons test participants, but one study of veterans who had occupational exposures has been reported.
X-Ray Technologists
Thousands of soldiers served as x-ray technologists during World War II. Many of these technologists were trained in Army technical training schools so that it was possible to identify them from school records. A roster that included 6,560 former technologists was created along with rosters of pharmacy and medical technologist controls, for whom the educational qualifications were similar to those for the x-ray technologists. Information regarding exposures of the x-ray technicians was, as might be expected, not available. However, by means of a mailed questionnaire for which the response rate was about 65 percent, some information related to exposure was obtained, such as number of months service in the occupation, occupation after military service (19 percent continued to work as x-ray technologists after an average of three years of occupational exposures while in military service), whether or not there was a history of x-ray therapy, the number of children, and their sex.
Two reports were published. The first (2) concerned follow-up data through 1963, that is, eighteen years after 1945, and the second report (3) extended the follow-up by eleven years, to 1974, making a total 29-year experience after separation from service.
Table 1 shows the results concerning mortality derived from the first survey. Attention focuses primarily on leukemia, since that is the
bellwether for late health effects following ionizing radiation exposure. There were 8 deaths among the technicians, while 6.4 would have been expected at concurrent U.S. rates. The technicians did have excessive mortality from lung cancer, whether compared with the United States or with the controls. However, in the absence of information on cigarette smoking, it is difficult to interpret this. The controls had a deficit of lung cancer deaths, which no doubt, reflected the fact that the different occupational groups had different backgrounds.
Table 2 displays the data concerning sex ratio of children. Six to 8 percent of the men on each roster reported having x-ray therapy, usually for such conditions as acne or bursitis, and strangely, among the technologists who reported such therapy, but not among the corresponding controls, the sex ratio of the offspring was significantly reduced. Among those who did not report x-ray therapy there was no evident effect on the sex ratio.
The later report (3) concerned only mortality; there was not a second round of questionnaires. The number of leukemia deaths increased between 1963 and 1974 from 8 to 12 among the technologists, and from 5 to 7 among the controls ( Table 3 ). Although there was a small excess of leukemia deaths among the technologists, as might be expected to result from radiation exposure, the differences were not statistically significant on either occasion. In fact, mortality was very similar in the two rosters; the excess of respiratory cancer mortality that had been seen in the earlier follow-up had disappeared by 1974.
Nuclear Test Participants
In 1980 Caldwell and colleagues at the Centers for Disease Control (CDC) published a report (4) that showed highly significant excesses of both the incidence of and mortality from leukemia in a group of 3,224 veterans who had participated at a test code-named Smoky at the Nevada Test Site in 1957 ( Table 4 ). The report stirred considerable interest, but seemed unexplainable. Although 4 of the 9 men diagnosed with leukemia had radiation doses estimated to be more that 0.5 rem, none were as large as 3 rem. Conventionally accepted risk estimates for the induction by radiation of leukemia could not explain the excess of about 5.5 incident cases above expectation, given the average radiation dose of perhaps 0.5 rem to the entire group. A second report, in 1983 (5), was concerned with mortality from all causes among the former Smoky participants, both from malignant neoplasms and from other causes of death. The veterans had quite low mortality from diseases other than cancer, as has been generally true for groups that have been selected initially for good health ( Table 5 ). The cancer mortality experience, however, was remarkably close to
expectation at national death rates. Examining the experience with respect to particular kinds of malignant disease ( Table 6 ), there were no remarkable findings except for leukemia where, as had been previously observed, the number of deaths was significantly high.
In view of the reports from CDC it seemed important to review the experience of participants at other weapons tests, both of fission devices at the Nevada Test Site and of thermonuclear weapons at the Pacific Proving Ground. It will be recalled that the Bravo test shot of the Castle series in 1956 exceeded the expected yield, and large amounts of radioactive debris were released into the atmosphere. Several Japanese crew members aboard the fishing boat Lucky Dragon were rather heavily dusted; symptoms of acute radiation injury were reported, and one man died. Similarly, some Marshallese on Rongelap and other atolls in the group were also exposed to radioactive fallout severe enough to cause symptoms, and in fact, during follow-up examinations, an increased incidence of thyroid nodules and, later, of thyroid cancer was found.
At the request of the Defense Nuclear Agency a study was undertaken with the objective of learning whether the findings following the Smoky shot were unique to that test or whether similar excesses of leukemia or other cancers might be found among those who had been present at other tests (6).
Five test series were selected, three from the Pacific and two from the Nevada Test Site (NTS); nearly 50,000 participants were identified ( Table 7 ). Radiation doses could be determined only for about two-thirds of the participants ( Table 8 ), but from the available data it was clear that the doses varied greatly among the series, with the Pacific series attended by radiation doses much larger than those at the NTS. Substantial numbers of men had doses estimated at more than 3,000 mrem (3 rem) and several hundred had more than 5 rem.
The excess of leukemia among Smoky participants that had been reported by Caldwell et al. was confirmed ( Table 9 ), but mortality from no other form of cancer was increased, and in fact, there was actually a deficit in total cancer mortality. When the data for the men in all test series were combined, despite the increase among the men who had been at Smoky, there was no excess of deaths due to leukemia (56 deaths, 56.4 expected) and a very sizable deficit of deaths from all forms of cancer, as compared with the number of deaths expected from contemporary U.S. mortality rates (1,046 deaths observed, 1,243.5 expected).
In summary, studies of radiation carcinogenesis in veterans, rooted in their experiences while in military or naval service, are of strategic importance:
-
Very large population groups are needed, and these can be supplied by the military.
-
Very long intervals of follow-up are necessary, and with the aid of such indexes as the VA BIRLS system, it is feasible to study rosters of tens of thousands of veterans over a span of more than 30 years.
-
The federal government undertakes to be responsible for injury to veterans that has its cause in experiences while in military service. Since there is no general consensus regarding the cancer risks that may follow relatively low doses (under 5 rads), only actual study of the experience of veterans can define those risks in relation to background “normal” cancer mortality rates.
REFERENCES
1. National Academy of Sciences National Research Council Health effects of exposure to low levels of ionizing radiation Washington DC : National Academy Press ; BEIR V Report ; 1990
2. Miller, R.W. ; Jablon, S. A search for late radiation effects among men who served as x-ray technologists in the U.S. army during World War II Radiology 96 : 269-274 ; 1970
3. Jablon, S. ; Miller, R.W. Army technologists : 29-year follow-up for cause of death Radiology 126 : 677-679 1978
4. Caldwell, G.G. ; Kelley, D.B. ; Heath, C.W. Jr. Leukemia among participants in military maneuvers at a nuclear bomb test JAMA 244 : 1575-1578 ; 1980
5. Caldwell, G.G. ; Kelley, D. ; Zack, M. ; Falk, H. ; Heath, C.W.Jr. Mortality and cancer frequency among military nuclear test participants, 1957 through 1979 JAMA 250 : 620-624 ; 1983
6. Robinette, C.D. ; Jablon, S. ; Preston, T.L. Studies of participants in nuclear tests National Research Council Medical Follow-up Agency 1985
TABLE 1
NUMBERS OF DEATHS BY CAUSE, OBSERVED vs. EXPECTED, STANDARDIZED FOR AGE, AMONG FORMER X-RAY, PHARMACY, AND MEDICAL TECHNOLOGISTS, 1946-1963
X-Ray |
Pharmacy |
Medical |
||||
Category |
Obs. |
Exp. |
Obs. |
Exp. |
Obs |
Exp. |
Deaths from trauma |
66 |
63.9 |
10 |
15.3 |
53 |
49.8 |
Suicide |
18 |
16.2 |
2 |
4.1 |
12 |
11.7 |
Homicide |
3 |
3.8 |
1 |
0.8 |
4 |
3.4 |
Other |
45 |
... |
7 |
... |
37 |
... |
Deaths from all diseases |
223 |
208.5 |
52 |
52.8 |
141 |
154.7 |
All cancer |
55 |
52.3 |
13 |
13.0 |
37 |
39.7 |
Leukemia |
8 |
6.4 |
0 |
1.5 |
5 |
5.0 |
Other lymphatic |
4 |
7.5 |
5 |
1.8 |
6 |
5.7 |
Respiratory |
17 |
10.5 |
1 |
2.8 |
3 |
7.8 |
Digestive |
13 |
12.1 |
3 |
3.1 |
8 |
8.9 |
Other |
13 |
15.8 |
4 |
3.8 |
15 |
12.3 |
Vascular, CNS |
14 |
11.7 |
3 |
3.0 |
6 |
8.3 |
Arteriosclerotic & degenerative heart disease |
83 |
76.7 |
11 |
19.8 |
59 |
56.6 |
Peptic ulcer |
1 |
1.0 |
1 |
0.3 |
0 |
0.7 |
Cirrhosis, liver |
14 |
11.1 |
4 |
3.0 |
4 |
7.9 |
Nephritis and nephrosis |
4 |
4.5 |
1 |
1.2 |
4 |
3.3 |
Other |
52 |
51.3 |
19 |
12.6 |
31 |
38.2 |
Total in sample |
6,560 |
1,522 |
5,304 |
|||
Source: Miller and Jablon (2). |
TABLE 2
SEX RATIO IN OFFSPRING BORN IN 1946 OR LATER
Category |
Number of Children |
Proportion of Male Births |
Men who reported radiotherapy |
||
X-ray technologists |
539 |
0.4712 * |
Pharmacy technologists |
78 |
0.5641 |
Medical lab technologists |
423 |
0.5130 |
Men who did not report radio therapy |
||
X-ray technologists |
6,539 |
0.5178 |
Pharmacy technologists |
1,510 |
0.5211 |
Medical lab technologists |
6,033 |
0.5173 |
TOTAL |
15,168 |
|
* Difference between this ratio and the total ratio significant at the 5% level, two-tail test. Source: Miller and Jablon (2). |
TABLE 3
NUMBER AND PERCENTAGE OF DEATHS BY CAUSE, 1946-1974
X-ray Technicians |
Controls |
|||
Cause of Death (ICD 7th Revision) |
No. |
% |
No. |
% |
Total |
6,560 |
100.0 |
6,826 |
100.0 |
Deaths, all causes |
792 |
12.10 |
792 |
11.60 |
Deaths by cause |
||||
Accidents, poisonings, and violence (E800-E999) |
107 |
1.63 |
109 |
1.59 |
Disease (001-796) |
657 |
10.02 |
656 |
9.61 |
Malignant neoplasms(all 140-205) |
145 |
2.21 |
158 |
2.31 |
Leukemia (204) |
12 |
0.18 |
7 |
0.10 |
Other lymphatic and hematopoietic (200-203,205) |
11 |
0.17 |
16 |
0.23 |
Respiratory system (160-164) |
41 |
0.62 |
42 |
0.62 |
Digestive system (150-159) |
38 |
0.58 |
47 |
0.69 |
Other malignant neoplasms (remainder 140-205) |
43 |
0.66 |
46 |
0.67 |
Vascular lesions of CNS (330-334) |
37 |
0.56 |
42 |
0.62 |
Arteriosclerotic and degenerative heart disease (420-422) |
283 |
4.31 |
266 |
3.90 |
Cirrhosis of liver (581) |
35 |
0.53 |
33 |
0.48 |
Other disease (remainder 001-795) |
157 |
2.39 |
157 |
2.30 |
Unknown cause |
28 |
0.43 |
27 |
0.40 |
Source: Jablon and Miller (3). |
TABLE 4
LEUKEMIA MORTALITY AND INCIDENCE IN SMOKY PARTICIPANTS
No. of Leukemia Cases |
Observed/Expected Ratio |
|||
Leukemic Cell Type |
Observed |
Expected |
Probability (Poisson) |
|
Mortality |
||||
(all types) |
8 |
2.9 |
2.4 |
.01 |
Incidence |
||||
All types |
9 |
3.5 |
2.3 |
.01 |
AML only |
4 |
1.1 |
3.6 |
.03 |
CML only |
3 |
0.7 |
4.3 |
.03 |
AML and CML |
7 |
1.8 |
3.8 |
.003 |
Source: Caldwell et al. (4). |
TABLE 6
DEATHS FROM NEOPLASMS AMONG SMOKY PARTICIPANTS, 1957-1979
Deaths |
|||
Observed |
Expected |
Ratio |
|
All neoplasms |
64 |
64.3 |
1.00 |
Leukemia |
8 |
3.1 |
2.58 |
Lymphoma |
3 |
4.2 |
0.71 |
Multiple myeloma |
0 |
0.7 |
0.00 |
Digestive cancer |
15 |
15.6 |
0.96 |
Respiratory |
21 |
22.2 |
0.94 |
Brain & nervous system |
5 |
2.9 |
1.72 |
Other |
12 |
12.7 |
0.94 |
Source: Caldwell et al. (5). |
TABLE 7
TEST SERIES INCLUDED IN THIS STUDY, DATES, AND NUMBERS OF PARTICIPANTS
Series |
Date |
Location |
Number of shots |
Number of participants |
Greenhouse |
1951 |
Pacific |
4 |
3,093 |
Upshot-Knothole |
1953 |
Nevada |
11 |
10,365 |
Castle |
1954 |
Pacific |
6 |
11,674 |
Redwing |
1956 |
Pacific |
17 |
10,564 |
Plumbob |
1957 |
Nevada |
30 |
15,165 |
Total |
49,148 * |
|||
*The total is less than the sum because there were 1,713 multiple test participants. Source: Robinette et al. (6). |
TABLE 8
GAMMA RADIATION DOSES (MREM) BY SERIES, PERCENTAGE OF PARTICIPANTS, AND MEAN VALUES
Number with Known |
Mean |
Percent more than |
|||||
Series |
Dose |
Dose |
100 |
300 |
1,000 |
3,000 |
5,000 |
Greenhouse |
2,099 |
1,291 |
87.6 |
81.7 |
55.9 |
8.2 |
3.0 |
Upshot-Knothole |
5,741 |
291 |
10.3 |
9.1 |
7.1 |
3.9 |
1.2 |
Castle |
7,595 |
1,493 |
91.3 |
78.7 |
45.2 |
13.6 |
3.7 |
Redwing |
9,205 |
1,534 |
88.3 |
79.2 |
47.0 |
17.6 |
2.8 |
Plumbob |
9,477 |
538 |
57.1 |
35.9 |
13.4 |
3.8 |
1.4 |
Smoky * |
(3,440) |
571 |
62.5 |
38.8 |
14.1 |
2.8 |
1.1 |
Total ** |
32,577 |
934 |
66.1 |
54.1 |
29.3 |
8.8 |
1.7 |
*Included in total for Plumbob. **Includes 1,540 men with known doses present at more than one series who are counted in each. Source: Robinette et al. (6). |
TABLE 9
OBSERVED AND EXPECTED MORTALITY FROM CANCER BY CALENDAR YEAR: SMOKY