Miller Breast Cancer Study

An Analysis of Breast Cancer Mortality

Following Fluoroscopy in TB patients

The publication: 

Miller AB, Howe GR, Sherman GJ, et al. Mortality from breast cancer after irradiation during fluoroscopic examination in patients being treated for tuberculosis. N Engl J Med 321:1285-1289 (1989).

 is a study of breast cancer mortality in Canadian TB patients who had been examined with fluoroscopy.  Table 1 gives the breast cancer mortality rate among these patients as a function of radiation dose.  Figure below on the right shows the same data in a graphical form.  Miller et al did not plot the data in a graph.  The graph is from (Cuttler et al. 2003, Can Cancer Be Treated with Low Doses of Radiation? Journal of American Physicians and Surgeons)

Table 1 from:  Miller et al. 1989                                                Graph from Cuttler et al. 2003

The conclusion of Miller et al. paper - as shown in the figure below - is linear model provides best fit to the data.  Text from last page of the Miller et al. paper:

Viewing the data above, do you agree with the conclusion, that linear model provides the best fit?

The same group updated the study with 7 more years of data.  (Howe et al. 1996)

Instead of separating low dose groups like Miller et al did, Howe et al lumped 0.01-0.49 y into a single bin, effectively making the reduction in cancers at 0.15 Gy in Miller study disappear, even if it were there after adding more data.  (See Table III below)

If we are interested in determining the effect of low dose radiation, we need to have data with finer bins, as presented in Miller et al.

BEIR VII ignored the Miller et al. data and used Howe et al. data to claim increased risk of cancer from low dose radiation.   Do you think this is reasonable?