Bases for Calculation of Risk Estimates Used in " the Meaning of Radiation at Bikini Atoll "

Page 1

BASES FOR CALCULATION OF RISK ESTIMATES USED IN “THE MEANING OF RADIATION AT BIKINI ATOLL” 409864 Estimates of cancer and birth defect} risks for the Bikini populations were based on a number of assumptions. Some of these assumptions resulted from consultation with other scientists including members of o ~ the BEIR committees. had not been accepted by any U.S. government agency. lie elected to use the values as given in BEIR-I rather than the revised values based on increased age of the population shown in Table V-4 of 7 2. For estimates of cancer risk both the relative risk coefficient and the absolute risk coefficient were used to give a range of estimated risk. The absolute risk coefficient gives a lower value, is less variable with the population and is not dependent upon the spontaneous cancer incidence, which is not known for the Bikini population. The relative risk coefficient gives a high value, but since it is based on the spontaneous cancer incidence, which is unknown for the Bikini population, it is probably less reliable than the estimates calculated from the absolute risk coefficients. 3. For estimating increased cancer incidence, the bone marrow dose was used because it was slightly higher than the whole body dose. This probably introduced a small element of conservation.

Page 2

4. For estimating birth defects neither BEIR-I or BEIR-111 is very clear about what is meant by parental dose, thus it is not clear whether birth defects should be based on the dose to one parent or both parents. In the latter case, the 30-year whole body dose would be doubled, We assumed the BEIR-I risk of 0.2% rem was based on both parents being irradiated. Also because we believed the risk coefficient from BEIR-I was already conservative to use the 30-year whole based on comparisons with BEIR-111, we elected body dose as provided us--not doubled. 5. For the 140 persons who returned to Bikini and were removed in August 1978, it was assumed that no children will be conceived by persons above age 40, that 300 children will be born after August 1978, and that all children born will be offspring of parents, both of whom returned to Bikini . The parental dose was obtained as follows: Average dose to males < 40years old = 1.36 rem Average dose to females < 40years old = 1.08 rem Total parental dose = 2.44 rem Parental dose used in calculations = 1.22 rem 6. The average dose values for persons who lived on Bikini were calculated from individual dose data (whole body and bone marrow) for 50 males and 49 females. These values are tabulated in the,~ppendix. 7. The spontaneous incidence of birth defects was taken to be 10.7% of all live births from BEIR-111. 8. The normal incidence of cancer deaths was assumed to be 15%. A value less than the approximately 20% given for the U.S. population

Page 3

was used because the Bikini people have been and will probably be exposed to much lower limits of environmental carcinogens than people living in the U.S. and because of limited medical services and prevalence of other risks such as drowning, poisoning, etc. Other causes of death are probably higher in the Bikini population than in the U.S. population. We also suspected the average life span was less than in the U.S. population, which might tend to reduce the number of cancers that would occur in the elderly. 9. The largest dose a person might receive in a year was estimated to be three times the average dose. Data in the appendix for individuals show that the highest individual dose is more than twice the average but less than three times. To estimate the number of births, deaths and the magnitude of the Bikini population after 30 years, information was used from the final draft of the Marshall Islands five year health plan prepared by the Trust Territories’ Department of Health Services’ Office of Health Planning and the Resources Department. The document is undated, but the presence of data from 1976 indicates that it must have been prepared in the period of 1977 to 1979 when we received it. It was noted that there are apparent inconsistencies among several of the different tables. For example, Table III-1 gives data for the Marshall Islands for the period 1955-1975 and Table III-5 gives data for the infant mortality rate for 1976. In Table 111-1, the infant death rate per “1000births for 1970 through 1975 is given as 28.3, 33.6, 25.4, 46.4, 21.”1and 37.0. However, Table III-5 indicates the infant mortality rate to be only 17.04. We used the data of Table 111-1 in the

Page 4

following estimates; because it is more complete and it provides a selfconsistent set of data. However, in view of the discrepancies, the results can only be considered as approximations. This probably makes little real difference in view of the uncertainties in the risk coefficients that were used. There is also a bias built into the data because of the inclusion of Ebye and Majaro in the overall Marshall Island rates. This arises from the

Page 5

The number of births in 30 years are given by: where x is the time between O and 30. This gives Similarly, the number of deaths in the 30 year period would be: One other datum needed is the reduction in 30 year dose to those born after the return because of the decrease in radiation levels and the smaller amount of time in the 30 year period that is spent on the island. For this, the total population dose for those born after returning assuming an initial dose rate of 1 rad/year is given by: A is the half-life of decrease of the radiation dose, taken here as Because this integral cannot be solved analytical, an approximate solution was obtained by calculating this function for each of 30 years and summing. This gave 8949 rads for the total population including the original 550. The total dose received by the original 550, assuming that

Page 6

all live for the 30 years, is For those born after the return, the population would be the difference between the total population in 30 years, the number of deaths and the original 550 peopls or 1134. Thus, the per capita dose for this group is 8949/1134 = 7.9 rads. For the original 550, the per capita dose is 11,902/550 = 22 rads. The ratio of these two to give an estimate of the fraction of the full 30 year dose received by the children is 0.36. The assumption of no deaths in the original 550 returning was made for simplicity and the lack of good death rate data. Me also compared the age characteristics of the Marshallese from Table IV-3 and the U.S. population in 1970. This comparison is given in the attached curve. The slopes are similar above age 35 but the magnitudes are distorted by the high birth rate in the Marshall Islands. However, in terms of the relative risk the similar slopes suggest that if the natural cancer rates in the two populations are similar, the relative risk for people above 35 in both populations would be similar because most of the cancer occurs at ages from about 40 and above. However, the magnitude of the relative risk in the U.S. used for the Marshallese will be high by a factor of somewhere around 2-3 because of the distortion caused by the very high proportion of young people who have a relatively low natural cancer incidence. Using the preceding calculations were made for ot,her population s for a population of 550, calculations zes. For a population of 550 (from preceding):

Page 7

Deaths in 30 years = 164-160 Births in 30 years = 1277-1300 For a population of 140 (the number that returned to Bikini): For a population of 235: At the time the 13ikini book was prepared no agency in the U.S. government had accepted the risk coefficients in BEIR-111. Thus we were constrained to use risk coefficients from BEIR-I. Nhile not included in the printed book, risk estimates based on BEIR-111 were calculated for comparison purposes. The following gives the origin of the risk coefficients used.

Page 8

Cancer deaths/106 person rem From the above the minimium estimate of cancer risk would be given by a risk coefficient of 87/106 person rem and the maximum by 458/106 person rem. Thus, these two risk coefficients were used to define a range of ,~-~~~~~~o—year reproduc~l%e generation would cause in the first generation 100-1800 cases of dominant diseases and defects per year (3.6 million births/year) or 5 times this amount at equilibrium. The 1800 cases represent an increase of 0.05% incidence per year first generation and 0.25% at

Page 9

In addit-ion there would be a few chromosomal defects and recessive diseases and a few congenial defects due to a single gene defect$ and chromosome aberrations. e The total incidence at equilibrium is 1100 to 27~000/year. These at equilibrium,the maximum would be 0.75% or 0.15% in the first generation. . These are equivalent to 0.15% per rem at equilibrium and 0.03%/rem in the first generation.

Page 10

Lifetime Risk of Cancer Death Single exposure to Continous Exposure 4 1 rem per generation (1 rem parental exposure) per 106 live offspring 5 to 75 birth defects, this is 0.0005--0.0075%--First l Since the spontaneous rate is given as 10.7%, in the U.S. population, 1 rem will increase the rate from 10.7% to 10.7005--10.7075%. . In terms of the spontaneous rate 1 rem per generation 9ives -~ IV. CALCULATIONSOF RISK Table 1 gives the radiation dose values provided by Dr. Robison for use in developing estimates of increased health risks in the Bikini population. &++-RISKS FOR 14 DIFFERENT LIVING CONDITIONS Table 3 sh ws the calculations for estimates of increased cancer risk for 14 di~rent living conditions.

Page 11

Table 3 gives the calculations for the estimates of birth defects. Table 4 gives risk estimates based on BEIR-111 risk coefficients. These were calculated for compa~~}ion purposes only and was not used in the Bikini book. The highest estimates for cancer risk result from using the linear relative risk model and are about the same as those given in Table 2 for the relative risk model. The lowest estimates result from the linear-quadratic absolute risk model and are slightly less than those for the absolute model in Table 2. Thus, as far as estimates of cancer risk are cone ‘n~~ed,those obtained using risk coefficients from BEIR-I are in the same general range as those obtained using risk coefficients Risk estimates for birth defects obtained using the risk factor from BE~R-I gives values about three times those obtained using the upper value of the range of risk factors given in BEIR-111. If BEIR-111 risk factors for bith defects represent a more enlightened assessment of this potential consequence of radiation exposure than the factor taken from BEIR-I f~r overall health defects, then the estimates ~ivi-~ in the Bikini book may be conservative by a factor of three.