Health Risks of Ionizing Radiation: - Clark University

48 Atomic Bomb Survivors
approximating a supralinear curve. For example,
over the dose range 5-20 mSv the ERR estimate is
2.6/Sv and it declines steadily as dose increases 4 .
Solid cancer mortality depends on age at exposure
and gender; specifically, risks are higher for females
and for childhood exposure. The 2003 analysis
generated an average ERR estimate of 0.47/Sv
(averaged for both sexes and assuming exposure at
age 30). As in the 1996 analysis, the dose-response
curve was steeper at lower doses so that the ERR
estimate was 0.74/Sv (0.1-1.5) for doses less than
120 mSv. The effect of age at exposure was described
as a decrease in the ERR estimate by 36% per 10-yr
increase in age at exposure. Exposures in infancy,
for example, would be associated with a solid
cancer mortality ERR of ~1.9/Sv 5 . Delongchamp
et al. (1997) calculated a similar ERR estimate of
1.4/Sv (90% CI 0.4-3.1) for adult cancer mortality
after exposures in early childhood (less than 6 years
old).
The dose-response pattern of leukemia mortality
is best described with a nonlinear model so that
one estimate of ERR/Sv would be misleading.
Pierce et al. (1996) presented summary estimates
of leukemia mortality risk with an ERR of 4.6/Sv
(90% CI 3.3-6.4), apparently based on a linear
model. The effect of age at exposure on leukemia
mortality was complicated. Exposures in childhood
were associated with higher risks in the first 10 or 20
years after exposure. If we consider the risk over the
entire follow-up period to date, however, then risks
are about the same for all ages at exposure. This
pattern can also be described by saying that leukemia
mortality risks declined over time, and this decline
was steeper for people exposed in childhood.
4.3 Noncancer mortality
Noncancer mortality through 1990 was assessed
by Shimizu et al. (1999) and additional followup
through 1997 was discussed by Preston et al.
(2003). The pattern of deaths from noncancer
diseases showed a strong healthy survivor effect.
This was evident in a lower noncancer mortality
rate among proximal survivors for a few years after
the bombings; the people that were strong enough
to survive the bombings were apparently healthier
than average. This effect diminished over time; to
account for this effect noncancer mortality risks
were estimated based on data from 1968-1997
(Preston et al. 2003). Generally the noncancer
mortality rates were associated with dose, although
the effect was smaller than that seen for cancer, and
the dose-response pattern could be described as
linear. Deaths from stroke, heart disease, respiratory
disease, and digestive disease increased by 10-20%
per Sv (ERR 0.1-0.2/Sv). Blood diseases were an
exception, showing a stronger effect with an ERR
of 1.9/Sv (1.2-2.9; Shimizu et al. 1999). There was
some evidence, not significant, of an effect of age at
exposure that was similar to that seen with cancer
mortality.
4.4.1 Solid cancer incidence
Thompson et al. (1994) analyzed the incidence of
solid tumors over the period 1958-1987. Generally,
there was a linear dose-response relationship for
cancer incidence with an ERR of 0.63/Sv (0.52-
0.74). There was a two-fold greater ERR for females
(0.84/Sv) than for males (0.38/Sv), and a strong
effect of age at exposure (Figure 4-5) 6 .
Figure 4-5. ERR estimates for total solid cancer incidence
by gender and age at exposure (based on data from
Thompson et al. 1994).
4 At doses of 20-50, 50-100, 100-200, and 200-500 mSv the ERR estimates are 1.6, 0.60, 0.43, and 0.38/Sv,
respectively.
5 Based on the following model: ERR = 0.50d{exp[-0.045(agex-30)]} (Preston et al. 2003).
6 The ERR estimates were 1.90, 1.21, 0.58, and 0.37/Sv for ages 0-9, 10-19, 20-39 and 40+ at exposure.