RSH Data & Documents

"Low Level
Radiation Health
Effects: Compiling
the Data"

Revision 2
March 30, 1999
by Radiation, Science, and
Health, Inc.,
Edited by J. Muckerheide

1.2.3.2
Medical Patients

"In this paper we report on lung cancer mortality between 1950 and 1987 in a cohort of 64,172 Canadian tuberculosis patients, of whom 39% were exposed to highly fractionated multiple chest fluoroscopies leading to a mean lung radiation dose of 1.02 Sv received at moderate dose rates. These data have been used to estimate the excess relative risk per sievert of lung cancer mortality, and this is compared directly to estimates derived from 75,991 atomic bomb survivors. Based on 1,178 lung cancer deaths in the fluoroscopy study, there was no evidence of any positive association between risk and dose, with the relative risk at 1 Sv being 1.00 (95% confidence interval 0.94, 1.07), which contrasts with that based on the atomic bomb survivors, 1.60 (1.27, 1.99). The difference in effect between the two studies almost certainly did not arise by chance (P = 0.0001). This study provides strong support from data for humans for a substantial fractionation/dose-rate effect for low-linear energy transfer radiation and lung cancer risk. This implies that lung cancer risk from exposures to such radiation at present-day dose rates is likely to be lower than would be predicted by current radiation risk models based on studies of high-dose-rate exposures."

"…A substantial proportion of these patients (39%) received lung absorbed doses from repeated fluoroscopic examination used in conjunction with pneumothorax treatment, with a number of individuals receiving lung doses in excess of 3 Gy. The dose rate involved, of the order of 0.6 mGy per second, is far lower than that of either the atomic bomb survivors or the ankylosing spondylitis study subjects, and is comparable to the lung dose rate involved in current diagnostic X-ray films. However, dose rates to workers in the nuclear industry are substantially lower than this, and hence the term 'moderate' is used to describe the dose rates involved in the fluoroscopy of the tuberculosis patients in the present study. The large size of the fluoroscopy cohort and the substantial lung doses received allow direct evaluation of lung cancer risk from fractionated moderate-dose-rate low-LET ionizing radiation."

"A total of 1,332 lung cancer deaths were observed in the entire Canadian fluoroscopy cohort between 1950 and 1987. Of these, 1,178 occurred 10 or more years after first exposure, compared to an expected value of 1,181, based on Canadian population rates, giving a standardized mortality ratio of 1.00 (95% confidence interval 0.94, 1.06). This lack of association relative to the population was seen for both men (912 observed, 934 expected, standardized mortality ratio = 0.98, 95% confidence interval 0.91, 1.04) and women (266 observed, 247 expected, standardized mortality ratio = 1.08, 95 % confidence interval 0.95, 1.22). The standardized mortality ratios for exposed men and women were 0.95 (347 observed) and 1.00 (108 observed), respectively.

"Table III shows the observed and expected numbers of lung cancer deaths classified by cumulative lung dose. There is no evidence from the results shown in Table III of any substantive increase in risk compared to the Canadian population amongst exposed study subjects, including those who received lung doses in excess of 1 Sv. Further, there is no obvious pattern of increasing risk with increasing dose beyond that which could well have occurred by chance.

Table III

"Table IV shows the results of the fluoroscopy study when internal comparisons are used; i.e., relative risks are estimated for each category of lung dose compared to the unexposed category. The results mirror the standardized mortality ratio analyses. None of the risks for any of the dose categories are elevated meaningfully or significantly compared to the unexposed category. The number of lung cancer deaths is adequate even in the higher-dose categories to lead to upper confidence bounds which exclude relative risks of 1.6 or greater for both sexes combined. The upper confidence bounds for men and women analyzed separately exclude relative risks of 1.9 or greater.

Table IV

"The results of an analysis of the data for the atomic bomb survivors which are directly comparable to the results for the Canadian fluoroscopy study are also presented in Table IV. In the atomic bomb survivors study, there were very few lung cancer deaths at doses above 2 Sv, so this category could not be subdivided as was done for the Canadian fluoroscopy data.

"It is apparent from Table IV that the results of the atomic bomb survivors study are substantially different from those of the Canadian fluoroscopy study. All of the higher-dose categories have elevated risks relative to the lowest category, and risks increase with increasing dose for both men and women…."

"In addition to the categorical analyses described above, analyses were also conducted using the linear excess relative risk model."

"Estimated excess relative risks per sievert for the Canadian fluoroscopy study based on the linear excess relative risk model are shown in Table V. For these excess relative risks the confidence intervals around the point estimates are narrow and exclude values of 0.11 per sievert or greater. There is no evidence of any meaningful positive association with dose for either men or women."

Table V

"From Table V, based on the linear excess relative risk model, there is clearly a substantial difference between the results of the Canadian fluoroscopy study and the atomic bomb survivors study. This difference in effect was tested formally by including an interaction term between dose and study (i.e. Canadian fluoroscopy or atomic bomb survivors). … The difference approached conventional levels of statistical significance for men (P = 0.07) and was highly statistically significant for women (P < 0.0001). It thus appears that the difference between the results of the Canadian fluoroscopy study and the atomic bomb survivors study almost certainly did not arise by chance."

"Finally, it should be noted that a very substantial bias would need to be present to mask a positive association of the magnitude seen in the atomic bomb survivors study. For example, a lung dose of 4.3 Sv (the person-years weighted mean dose in the fluoroscopy study for those with a dose in excess of 3 Sv) would lead to a predicted relative risk of 3.6 (95% confidence interval 2.2, 5.3) based on the data for the atomic bomb survivors, and this contrasts with the observed relative risk amongst the fluoroscopy cohort of 1.0 (95% confidence interval 0.7, 1.5) (Table IV)."

"The existence of a fractionation/dose-rate effect for low-LET radiation is supported from both principles of radiobiology and the data for animal experiments (BEIR V 1990). The present results provide strong epidemiological support for the existence of a substantial fractionation/dose-rate effect in the induction of lung cancer by such radiation. A smaller cohort study of another series of tuberculosis patients in the United States examined by fluoroscopy has also failed to find any radiation-related excess of lung cancer (Davis et al 1989), though the increased sampling variability in that study could have led to a failure to observe a modest positive association by chance….