Collecting All the Required Data from RERF
In order to evaluate cancer-risk and dose-response from the A-Bomb Study (1950-1982), we needed to assemble the raw data from RERF. Since RERF had not published all of them, this task was not routine. It involved four sets of source-material:
The Written Technical Report (TR) :
The RERF Technical Report for the 1950-1982 cancer follow-up is TR-1-86 by Preston and Pierce (Pr86). TR-1-86 is also called Life Span Study (LSS) Report 10. An abbreviated version of TR-1-86 was published in the journal Radiation Research in 1987 (Pr87a).
Analysts using these publications, however, need to avoid three erroneous entries in their Table l:
Error No.1: Hiroshima subjects in the 1-9 kerma-rad dose-category are given as 15,391. This is an error. The correct number is 15,931.
Error No. 2: The mean dose for Hiroshima in the 300-399 kerma-rad dose-category is given as 364.4. This is an error. The correct value is 346.4.
Error No. 3: The mean dose for Hiroshima in the 400+ kerma-rad dose-category is given as 524.2. This is a small error. The correct number is 524.4.
This book uses the correct numbers, of course; they will be seen as entries in our special Table 26-H, which also comes from RERF.
Welcoming Input from Readers :
In mentioning these slips in the the RERF report, we intend no criticism. We certainly hope that anyone who notices such slips in this book will inform us, so that we may correct them before another printing and can thus prevent trouble for others who may use our numbers "downstream." Other types of comment also are welcome.
The Corresponding Floppy Diskette :
Most commendably, RERF has started to make available floppy diskettes which contain far more data than do the written reports. For the 1950-1982 follow-up, we used the RERF diskette entitled "Cancer and Non-Cancer Mortality, R10ALL.DAT" (R-Ten-All). The diskette reports the observations for the 91,231 A-bomb survivors in the Life Span Study, but omits the "Not in City" or NIC group (see Chapter 5).
On the diskette, the 1950-1982 data are arranged in 1,280 rows of entries. The number 1,280 is the product of 2 cities, times 2 sexes, times 5 age-bands, times 8 dose-classes, times 8 four-year follow-ups.
Except for Chapter 17 in this book, we always deal with the aggregate or cumulative observations from 1950-1982. Therefore, we have summed the entries for the eight separate follow-ups, and this reduces the file from 1,280 to 160 rows, or subsets.
We have reproduced these 160 rows of raw data in this book's Chapter 26 as Master Table 26-A,B,C,D as follows:
Table 26-A : Hiroshima Males (40 lines).
Table 26-B : Nagasaki Males (40 lines).
Table 26-C : Hiroshima Females (40 lines).
Table 26-D : Nagasaki Females (40 lines).
* (3) Previously Missing Data --
Persons, True Doses, True Ages ATB
We (and other analysts) also need the actual numbers of persons initially in each of the 160 subsets, their true mean doses, and their true mean ages ATB. These three crucial types of data are not provided on the RERF diskette.
We requested the missing data from RERF, and we received it as nine printout sheets, which are reproduced in Chapter 26 as Tables 26-E through 26-M. This information is not available (as we go to press) anywhere else in the literature. Except for reduction in size and addition of some identifying titles, the sheets are exactly what RERF provided (final version, August 29, 1988).
We would like to express our appreciation for the excellent cooperation of Dr. Donald A. Pierce, who was Chief of RERF's Department of Statistics at that time.
Initial Persons :
The RERF diskette provides person-years (PYR), but not the number of persons initially in each subset in 1950. Therefore the persons in our Master Table 26-A,B,C,D come from the special Tables 26-E through 26-M (where they are called "cases").
True Mean Doses :
The diskette provides fictitious or nominal doses. It uses the mean dose of an entire dose-group whether or not it really describes a particular subset of that dose-group. Therefore, the true mean kerma-doses in Master Table 26-A,B,C,D (Column G) were added by us from the special Tables 26-E through 26-M.
True Mean Ages At-Time-Of-Bombing :
The diskette provides fictitious or nominal ages ATB. It uses the midpoint of an age-band whether or not the midpoint really describes a particular subset of that age-band. Therefore the true mean ages ATB in our Master Table 26-A,B,C,D (Column D) were put there by us from the special Tables 26-E through 26-M.
The data which we received from RERF, on the true mean ages ATB, apply to RERF's customary eight dose-groups. It is easy to see in Table 26-E (and even easier in Table 4-A) that true mean age ATB is not the same in all eight dose-groups. In view of RERF's new practice of subdividing Dose-Groups 2 and 3, the following point deserves emphasis.
A Warning about Subdivision :
If a dose-group is divided into halves (its lower dose-range and its upper dose-range), its two cohorts will probably have different mean ages. When Dose-Group 2 is divided, it is impossible to predict whether the people falling into its lower dose-range are younger, or older, than the people falling into its upper dose-range. The same uncertainty exists about any division of Dose-Group 3.
For the reasons explained in Chapter 11, and demonstrated in Chapter 13, no analysis of risk or dose-response could be properly done with subdivided groups in the absence of the necessary adjustments for age-differences.
The problem does not arise in this book, because we use the customary undivided dose-groups for which true age-data now have been provided by RERF. In our judgment, the number of cancer cases and the dosimetry in this database do not permit subdivision (see Chapter 8, Part 4).
* (4) Missing Data for DS86 Subsets --
Neutron and Gamma Components of Dose
Our approach to application of the new dosimetry was described in Chapter 6. In order to use our approach, we needed to obtain data which have not been published by RERF: The DS86 neutron and gamma components for each subset of persons shown in RERF Technical Report TR-12-87, Appendix Table 2 (Shi87).
We requested this information from J.W. Thiessen, M.D., the Vice-Chairman of RERF and representative of the (U.S.) Department of Energy, Office of Health and Environmental Research (Doe87, p.xvii). He graciously provided it in December 1988.
The printouts from RERF are reproduced in Chapter 26 as our Tables 26-N and 26-"O". Chapter 10 shows the steps which lead from the special Tables 26-N and 26-"O" to the entries in Columns I (Eye) and J of the Master Table 26-A,B,C,D.
A Word of Appreciation :
RERF holds in trust, for humans everywhere, a unique database of great importance for human health. With regard to the courtesy and attention which RERF gives to meeting requests from outside analysts, RERF sets a fine standard indeed for other research institutions.