6
The Participant Cohort
The basic comparison is between groups of military personnel: those who participated in Operation CROSSROADS and a comparison cohort chosen from among those who did not. In this section we describe the process of participant cohort selection in detail, since misclassification in participant selection would decrease the likelihood of observing any exposure-outcome association that might exist.
The Defense Nuclear Agency (DNA) defined "participation" according to a 1989 announcement in the Federal Register,11 to include official military
activities onsite between 1 July 1946 and 31 August 1946 or involving materiel used at the tests. It provided rosters of CROSSROADS participants to the Medical Follow-up Agency (MFUA) in two stages: an initial file in 1986 and the final file in 1994.
When this study was begun, each military service maintained its own Nuclear Test Personnel Review (NTPR) activity, responsible, among other things, for identifying veterans involved in the above-ground nuclear test program. In 1986, Navy and Army NTPR offices provided data to MFUA for approximately 40,500 alleged nonduplicate participants. Air Corps personnel were covered by Army records at that time; the Navy provided Marine Corps records.
In 1992, work was interrupted on this project.12 By the time MFUA resumed the study in 1994, DNA had consolidated the service-based NTPR teams into a single effort. DNA, via the consolidated NTPR, performed extensive data clean-up efforts to bring participant data from different branches into a common format; to update, verify, and correct data; and to create a unique identifier (heretofore nonexistent) for each record. In August 1994, DNA supplied MFUA with that updated participant list. It is the basis of the current study.
For both the 1986 file and the 1994 file, DNA identified the ships that participated at Bikini Atoll during the CROSSROADS operational period. Individuals assigned to those ships at that time plus some individuals serving at Kwajalein and Enewetak islands to support CROSSROADS were identified using the following data sources: "service personnel records, unit diaries, ship deck logs, ship/unit muster rolls, ship/unit officer lists, unit histories, morning reports, and operation participant listings."13 Data items sought were name, rank, and service number, although date and place of birth were also noted in the 1986 file. The quality of the various identifiers varied; other identifiers, such as Social Security Number, were noted when available.
A participant cohort numbering 42,548 was developed across all military branches, using the individuals identified in the 1994 file; data items for those individuals were augmented with 1986 information as appropriate.
|
(C) Service as a member of the garrison or maintenance forces on Eniwetok during the periods June 21, 1951 through July 1, 1952, August 7, 1956 through August 7, 1957 or November 1, 1958 through April 30, 1959. (D) Assignment to official military duties at Naval Shipyards involving the decontamination of ships that participated in Operation Crossroads. (v) The term 'operational period' means: ... (B) For operation CROSSROADS the period July 1, 1946 through August 31, 1946. (Authority 38 U.S.C. 312)" |
12 |
Circumstances—including the illness and death of the principal investigator, Dr. Dennis Robinette and federal funding shortages—caused this study to be put on hold. |
13 |
DNA memorandum dated 11 July 1994 (see Appendix B). |
It is important to realize that some of the individuals in the roster of atomic veterans may have identified themselves to DNA and subsequently become enrolled in the NTPR program as valid participants when their involvement with atomic testing had been verified. We refer to this in our discussion of the findings of this study (Chapter 11).
Independent Verification of Participant Cohort
To examine the validity of the ongoing participant ascertainment process, MFUA staff then worked to characterize the variation between the initial 1986 and final 1994 files. Based on the verification process that we describe in detail in Appendix E, we estimate that approximately 2 to 4 percent of the individuals included in the 1994 participant cohort were not, in fact, participants.
Two non-DNA sources were sought to enable MFUA to estimate an independent assessment of DNA's false negative (missing actual participants) rate. The National Association of Atomic Veterans (NAAV) has run mail-in surveys (Appendix A) to develop a list of radiation-exposed veterans; it provided this list to MFUA. Also, MFUA placed announcements in largecirculation veterans' magazines, requesting "atomic veterans" to write in and identify themselves. Details of the work comparing the DNA, NAAV, and magazine write-in groups are presented in Appendix E. Overall, the matching of flames indicated a high degree of agreement and, therefore, confidence in using the DNA cohort.
Exclusion of "Post-CROSSROADS" Participants
To maintain clarity of cohort definition, this study does not include socalled "post-CROSSROADS" participants, those military personnel who arrived in the designated area after the formal cut-off date of the operation but within the six-month period 1 September 1946 through 28 February 1947.
Crossroads Participants who Participated in other Nuclear Tests
This report on the CROSSROADS test uses "participation" status as a general proxy for exposure (see expanded discussion in Chapter 8). Some CROSSROADS participants also attended nuclear tests other than CROSSROADS (Table 6-1). For perspective, we present the distribution of CROSSROADS participants by their presence at additional atomic tests.
TABLE 6-1. Distribution of Participation in Nuclear Test Series
Number of Nuclear Test Series in which Person Participated |
CROSSROADS Participants |
|
Number |
Percent |
|
CROSSROADS only |
38,203 |
89.8 |
CROSSROADS plus 1 other |
3,913 |
9.2 |
CROSSROADS plus 2 others |
333 |
0.8 |
CROSSROADS plus 3 others |
38 |
0.1 |
CROSSROADS plus 4 others |
14 |
0.03 |
CROSSROADS plus 5 others |
8 |
0.02 |
CROSSROADS plus 6 others |
10 |
0.02 |
CROSSROADS plus 7 others |
1 |
0.002 |
CROSSROADS plus 8 others |
3 |
0.007 |
CROSSROADS plus 9 others |
1 |
0.002 |
CROSSROADS plus 10 others |
1 |
0.002 |
CROSSROADS plus 11 others |
1 |
0.002 |
Total |
42,526* |
~100 |
* This count includes all participants, not only Navy personnel. |
Note that 90 percent (n = 38,203) of CROSSROADS participants did not participate in any other military nuclear test. Of the remaining 10 percent (n = 4,323) who did, most (90.5 percent) were in only one other test (with almost half of those qualifying only by their inclusion in post-CROSSROADS activities, which we consider a separate test series for classification purposes).
Primary Analysis—Navy Personnel Only
The participant cohort was 91 percent Navy personnel. Because Navy records were the most complete (as will be discussed in Chapter 9), primary analyses for this report are limited to the 38,662 participating Navy personnel
Exclusivity of Participant and Comparison Cohorts
The comparison roster was checked against the participant roster. The 205 Navy individuals appearing on both were deleted from the control roster for this report's analyses.