SPECIFIC COMMENTS ON VOLUME II: THE TECHNICAL APPENDIXES

As previously stated, the appendixes to the final Fernald report contain the technical details associated with the dose reconstruction. Because these appendixes are likely to be of interest to others involved in dose-reconstruction projects, the National Research Council committee offers several suggestions.

Appendix H: Particle Size Distributions for Dust Collectors

RAC has chosen to accomplish several things in this appendix, including fitting the cascade-impactor data by exact polynomial equations, simulating mathematically the action of an Andersen impactor by describing the jet configurations and air flows to match the air-sampling process, and comparing this approach to conventional particle-size analysis by using lognormal distribution assumptions and probit analyses to obtain mass median aerodynamic diameters (MMADs). This is an acceptable approach to obtain a more-accurate description of the particles collected on each impactor run, but it is unconventional, and operationally it might be cumbersome or time-consuming.

Several basic statements made in this appendix should be emphasized, as follows.

Page H-2, paragraph 4: “Rather than making the prior assumption that all particle size distributions were lognormal, we fitted cubic polynomials to the log-probability-transformed cumulative sampler data in order to represent the distributions by continuous functions for further calculations. . . . We did not consider possible distortion of the distribution of particles that entered the sampler.”

Page H-1, paragraph 6, and page H-2, paragraph 1: “We have examined the properties of the Andersen Mark II cascade impactor, and we have developed a simulation of the instrument to study its potential for distorting the sampled distribution. We have found no reason to alter our approach to representing particle size distributions. ” This statement indicates that the authors were aware of the “in-field” distortion of particle size distributions that actually occur on sampling, but the theoretical description of the instrument does not indicate such losses, so this theoretical model of the impactor is a proper approach.



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 30
A REVIEW OF THE RADIOLOGICAL ASSESSMENTS CORPORATION'S FERNALD DOSE RECONSTRUCTION REPORT SPECIFIC COMMENTS ON VOLUME II: THE TECHNICAL APPENDIXES As previously stated, the appendixes to the final Fernald report contain the technical details associated with the dose reconstruction. Because these appendixes are likely to be of interest to others involved in dose-reconstruction projects, the National Research Council committee offers several suggestions. Appendix H: Particle Size Distributions for Dust Collectors RAC has chosen to accomplish several things in this appendix, including fitting the cascade-impactor data by exact polynomial equations, simulating mathematically the action of an Andersen impactor by describing the jet configurations and air flows to match the air-sampling process, and comparing this approach to conventional particle-size analysis by using lognormal distribution assumptions and probit analyses to obtain mass median aerodynamic diameters (MMADs). This is an acceptable approach to obtain a more-accurate description of the particles collected on each impactor run, but it is unconventional, and operationally it might be cumbersome or time-consuming. Several basic statements made in this appendix should be emphasized, as follows. Page H-2, paragraph 4: “Rather than making the prior assumption that all particle size distributions were lognormal, we fitted cubic polynomials to the log-probability-transformed cumulative sampler data in order to represent the distributions by continuous functions for further calculations. . . . We did not consider possible distortion of the distribution of particles that entered the sampler.” Page H-1, paragraph 6, and page H-2, paragraph 1: “We have examined the properties of the Andersen Mark II cascade impactor, and we have developed a simulation of the instrument to study its potential for distorting the sampled distribution. We have found no reason to alter our approach to representing particle size distributions. ” This statement indicates that the authors were aware of the “in-field” distortion of particle size distributions that actually occur on sampling, but the theoretical description of the instrument does not indicate such losses, so this theoretical model of the impactor is a proper approach.

OCR for page 30
A REVIEW OF THE RADIOLOGICAL ASSESSMENTS CORPORATION'S FERNALD DOSE RECONSTRUCTION REPORT Page H-4, paragraph 2: “When we first examined the sampler data from the 1985 NKES study we fitted lognormal distributions to the data as a matter of course, but a critical examination of the plots revealed some cases in which the lognormal distribution was obviously not an appropriate representation of the data (whether the data were properly representative of the distribution taken in by the sampler is a separate question).” Page H-4, paragraph 3: “Knutson and Lioy (1989) discourage the routine assumption of lognormality for particle size data. The authors are of the opinion that the lognormal distribution has been overused. . . . There are significant departures from lognormality. Such departures are quite obvious in the FMPC cascade-impactor data.” All the above quotations from the text are favorable to the proposed new mathematical approach to analyzing particle-size data as collected by Andersen's cascade impactor. The new approach is considered a scientific breakthrough by the RAC authors, and there is nothing inherently wrong with it. For decades, in fact, the industrial-hygiene field has produced many schemes for describing particle-size data, and this is another acceptable one. The traditional lognormal approach is simple for the field measurements and for a rapid assessment of exposure conditions, and it is likely to continue to be widely used. A further explanation of this text concerning the new approach is best presented on page H-10, paragraph 3. First, it is known that particles entering an impactor tend to bounce on collection stages and that when this happens, the particles can shatter and be carried on to later stages. They then show up as particles of smaller mass than those which were in the distribution that entered the impactor. The lognormal approach generally eliminates the larger-particle fraction because of this. The authors recognize this when they state that “when particle bounce occurs, distorted efficiency curves are effectively being substituted for the theoretical curves on which the analysis is based.” However, little contribution is attributed to this phenomenon, as seen on page H-10, paragraph 4: “In the FMPC sampling, however, we believe the comparisons of the sampling data with the simulated data points, which were computed with the theoretical curves, cast doubt on the presence of such levels of degradation of collection efficiency.” This theoretical simulation of the impactor does not account for particle bounce, and the polynomial equations fit the data without accounting for any.

OCR for page 30
A REVIEW OF THE RADIOLOGICAL ASSESSMENTS CORPORATION'S FERNALD DOSE RECONSTRUCTION REPORT A few statements are perhaps not necessary in the report. For instance, page H-11, paragraph 2 states that “in every case, the simulated data points closely approximate the distribution curve from which they were derived.” It would be surprising if they did not. In accepting any problems with the simulation, however, the authors state that “this result reassures us that accepting at face value the sample points from the 1985 NKES study is a reasonable procedure for drawing conclusions about the form of the sampled distribution. It says nothing, of course, about the relationship of the distribution at the time of sampling with comparable distributions over time.” All in all, the computer simulation of the Andersen cascade impactor with the mathematics presented is acceptable. The fact that it is not a conventional approach based on an assumption of lognormality in particle size distributions does not detract from the result. Part of the problem, if there is any, in using such a simulation technique might be related to the fitting of the analytic physical results with the biologic model for deposition and retention after inhalation. Appendix I: Dosimetric Methods The following must be borne in mind when combining the above analysis of particle size with the modeling for deposition of particles in various regions of the respiratory tract. Many statements that might be interpreted as matters of opinion by someone applying them to hazard analysis will follow. Some examples are given here. There is confusion between the authors and the information documented by the Task Group on Lung Dynamics, the basis for the deposition model in ICRP-30 presented on page I-9 in this appendix. For instance, figure I-5 shows the fraction of respiratory tract deposition as a function of MMAD; the results for geometric standard deviations range from 1 to 4.5. We are interested primarily in pulmonary deposition of uranium particles and tracheobronchial deposition of radon progeny. From the figure, using a particle distribution of 2 µm for uranium, deposition in the pulmonary region would vary from about 20% to 30% —a factor of 1.5—in going from a standard deviation of 1 to 4.5. In the smaller particle-size region, such as with radon progeny, the same particle size for absorbed progeny yields deposition of about

OCR for page 30
A REVIEW OF THE RADIOLOGICAL ASSESSMENTS CORPORATION'S FERNALD DOSE RECONSTRUCTION REPORT 5% to 10% in the tracheobronchial region. If the unattached progeny are of concern, a particle diameter of 0.02 µm would yield deposition in the tracheobronchial region of 15% to 25% of that inhaled. Some statements in appendix I might not be adequately explained: Page I-10, paragraph 1: “At the expense of accepting this range of variability, it is possible to eliminate the GSD (geometric standard deviation) of the distribution as a parameter.” The ranges were discussed immediately above. Page I-9, paragraph 2: “From these calculations, the Task Group concluded that when the mass median aerodynamic diameter (MMAD) of a distribution is specified, the deposition fraction is relatively insensitive to the GSD of the distribution, within the range of 1.2 and 4.5.” It is not clear where this originated, but presumably it was in the original task group report; if so, it is not reinforced by figure I-5 discussed above. It is odd that if a polynomial equation is used to fit the particle-size data, no geometric standard deviations may be derived therefrom, and that it is then concluded that the distribution of the particle size distribution is not important for deposition in the respiratory tract. Perhaps a factor of 1.5 is not important, but it appears to be so in fitting the particle distribution. Appendix S: Lifetime Risks of Fatal Cancer for Individual Scenarios at the Feed Materials Production Center Page S-4, table S-2: These risk estimates, which are based on data from the Japanese atomic bomb survivors, are dominated by the digestive tract, with the colon having 17% and the stomach 27% of the total risk. In western populations that have been studied, the percentage of total radiation risk associated with those sites appears to be much smaller. It would have been more appropriate to use risk estimates from western populations. However, given that lung exposure proves to be the main concern in this assessment, this point is minor. Page S-7: The discussion of risk estimates from uranium-worker studies is inappropriately selective in the data that it emphasizes. In each study, RAC focuses on some

OCR for page 30
A REVIEW OF THE RADIOLOGICAL ASSESSMENTS CORPORATION'S FERNALD DOSE RECONSTRUCTION REPORT subgroup finding with a suggestion of a risk and barely mentions the overall negative findings. For example, regarding the case-control analysis by Cookfair and others (1984), the text focuses on a small subgroup that appeared to show an increased risk (paragraph 2), rather than using the risk estimate for the whole series. Similarly, for the Archer and others (1973) and Waxweiler and others (1983) studies, the text essentially ignores any comparison of the observed deficit in lung cancer with the BEIR IV risk estimate (NRC 1988), but instead focuses on a small apparent excess of lymphohematopoietic cancers in the studies. It should also be noted that the Waxweiler paper includes and updates most, if not all, of the Archer cohort; so there are not 2 independent findings of excess lymphohematopoietic cancers, as the manner of presentation suggests. Page S-11 and figure S-1: This material is vague in describing risk versus age without specifying whether the age is age at irradiation or attained age at observation. Not until a page later is it made clear that it is age at irradiation. Pages S-11 and S-12: RAC should provide a table of the adjustment factors for age and sex, so that its calculations could be verified. RAC provides voluminous tables for many other items (such as tables I-1S to I-43S, M-21 and M-22, N-27 to N-31, and R2) but not for these key factors. In addition, figure S-1 for these factors does not show the relative-risk coefficients for ages 0-9 years, although they are obviously important for later calculations. Table S-14: An explanation of a “truncated normal” distribution would be helpful.