Health Risks of Ionizing Radiation: - Clark University

86 Radiation Workers
Figure 6-14. A facility at Hanford for treating persons
injured by embedded radioactive particles (circa 1967). In
this shielded operating cell, a mock patient is flanked by a
surgeon (right) and a radiation monitor (left) (www.eh.doe.
gov/.../photos/handord/fig1.html).
internal and external radiation exposure.
Pooled international data. Cardis et al. (1995)
pooled data from the Canadian, UK and US cohorts.
This combined cohort of 95,673 workers included
fewer Canadian workers (11,355) than the analyses
of Asmore et al. (1998) and Sont et al. (2001), but
included virtually the entire cohorts analyzed by
Gilbert et al. (1993) and Carpenter et al. (1994,
1998). In the combined group there were significant
trends with dose for leukemia, multiple myeloma and
uterine cancer 34 . There was also a significant doseresponse
trend for noncancer circulatory disease.
The pooled estimates of risk for leukemia and solid
cancer (all cancer except leukemia) were 1.55/Sv
(90% CI –0.2-4.7) and –0.07 (90% CI –0.4-0.3). The
central estimate for solid cancer risk from the threecountry
cohort is negative and the upper confidence
limit is less than the atomic bomb survivors estimate
of 0.47/Sv (Preston et al. 2003). The leukemia risk
by subtype is shown in Table 6-1 and is discussed
further below.
6.5 Discussion
The studies reviewed in this section paint a variety
34 The estimated ERR of multiple myeloma was 4.2/Sv (90% CI 0.3-14.4).
of pictures, and this variety can be partly explained
by differences in the analytical methods that were
employed by the authors. This is particularly true
in the case of Hanford; utilizing basically the same
data pool Gilbert et al. (1993) found no significant
evidence of heath risks aside from multiple
myeloma while Kneale and Stewart (1993) found
risks greater than would be expected based on the
atomic bomb survivor data and also detected an
increased sensitivity in association with exposure at
older ages.
Perhaps more important than methodological
differences are differences among facilities. The
range of radiological and chemical exposures that
workers experience depends on the type of activity
a facility is engaged in; some workers experience
external exposure to gamma radiation and other
workers inhale one or more radionuclides. In
addition, exposure monitoring has varied by
facility and over time. Interpreting the results of
facility-specific studies is complicated by these real
differences and is further complicated by the role
of chance and small numbers of workers at risk. In
general, inconsistencies are expected and individual
results should be judged in a broader context.
The kidney cancer mortality results from a few
of the studies described above provide a useful
illustration. Dupree-Ellis et al. (2000) reported that
there was a significantly positive dose-response trend
for kidney cancer mortality among Mallinckrodt
Chemical Works workers. This was based on ten
deaths and had wide a confidence interval (ERR
10.5/Sv, 0.6-57.4). Omar et al. (1999), on the other
hand, reported the reverse relationship in Sellafield
workers. This study found a significantly negative
dose-response relationship based on thirteen deaths.
Follow-up time was very similar in these two studies
(1942-1993 and 1947-1992) and so was mean dose
(48 and 130 mSv), so something else must explain
the difference in the results. This could be study
design, although this would probably not result in
such dramatic differences. But the difference could
also be chance. Although Mancuso et al. (1977) and
Loomis and Wolf (1996) found evidence suggesting
a kidney cancer risk at Hanford and Oak Ridge,
this result has not been confirmed in other studies.
The international pooled workers study found no