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 61
11 Findings In this chapter, we present the findings from our standardized mortality ra- tio and proportional hazards analyses.! We begin with tests of the primary study endpoints and the presentation of some descriptive data. The findings from these analyses suggested additional avenues of investigation, the results of which are then reported. TESTS OF PREDETERMINED PRINCIPAL ENDPOINTS We had determined in advance of data collection that participant versus ref- erent mortality rates would be formally compared for three endpoints all causes, all malignancies, and all leukemia minus chronic lymphoid leukemia (CLL). We had further decided to examine these outcomes using both standard- ized mortality ratios (SMRs) and proportional hazards analyses. SMRs were used to compare mortality rates for participant and referent subjects with U.S. white male population rates.2 However, these SMRs take into account only age and calendar year of death, and the mortality rates of the U.S. general population are generally higher than those of military veterans. When a comparable referent group is available, proportional hazards analyses allow for simultaneous control of design variables (via stratification) as well as tighter control for age differ- ences (via covariate adjustment) and thus provide a better basis for estimating the difference in mortality experience of the two groups. SMR data in Table 11-1 show that the SMRs for both all-cause mortality and all-malignancy mortality were almost equal for participants and referents, whereas participants had a higher SMR for leukemia death (0.75) than did refer iSee Chapter 9 and Appendix C for explanation of these methods. 2See Chapter 9 for discussion of the use of white male rates. 61
OCR for page 62
62 THE FIVE SERIES STUDY ents (0.651. As anticipated, all SMRs are less than 1.00, indicating that both the participant and the referent cohorts had lower mortality than the general popula- tion the "healthy veteran effect" (Seltzer and Jablon, 1974, 1977~. The propor- tional hazards analyses also show with estimated hazard ratios (HRs~that participants and referents were at similar risk of all-cause mortality (HR = 1.00) and all-malignancy mortality (HR = 1.02), and participants had an estimated 14 percent higher risk of leukemia death (HR = 1.14~. However, none of these haz- ard ratios is significantly different from 1.00, indicating that there were no sta- tistically significant differences between the participant and referent cohorts on these outcome measures. DESCRIPTIVE ANALYSES OF PREDETER1VIINED ADDITIONAL MORTALITY ENDPOINTS We also looked at a number of mortality endpoints that had been examined in other studies, particularly cancer endpoints. Table 11-2 shows SMRs for broad categories of noncancer causes of death. Only one SMR was greater than 1.00 symptoms, signs, and ill-defined conditions. Proportional hazards analy- sis shows only two significant differences. Participants had a significantly higher risk of death due to external causes (1.08; 95% CI 1.02-1.16) and a sig- nificantly lower risk of death due to unknown cause (0.62; 95% CI 0.57-0.67~. The latter is not surprising because it reflects the fact that the causes were miss- ing for 4.5 percent of participant deaths versus 7.3 percent of referent deaths, a statistically significant difference (see Table 8-4 and related discussion). Partici- pants also had higher risks of death due to diseases of the musculoskeletal sys- tem (1.43; 95% CI 0.86-2.38) and congenital anomalies (1.59; 95% CI 0.72- 3.51~; however, these are based on relatively small numbers of deaths. Table 11-3 shows data for various cancer mortality endpoints in some de- tail. Again, almost all SMRs are less than 1.00. The six cancer sites with esti- mated hazard ratios greater than 1.2 are nasal cancer (2.64; 95% CI 1.02-6.82), thyroid cancer (2.33; 95% CI 0.83~.55), cancer of the testes (1.62; 95% CI 0.59-4.46), male breast cancer (1.39; 95% CI 0.53-3.66), bone cancer (1.21; 95% CI 0.57-2.60), and prostate cancer (1.20; 95% CI 1.03-1.40~. Only the risks of death due to nasal cancer and prostate cancer were significantly higher among participants than referents. Among the hematologic cancers, the risks for all leukemia (1.15; 95% CI 0.93-1.43) and leukemia minus chronic lymphoid leukemia (CLL) (1.14; 95% CI 0.90-1.44) were both elevated, albeit not signifi- cantly, among participants. Table 11-4 shows a more detailed breakdown of leukemia deaths by sub- Wpe, as available from death certificates. The highest hazard ratios are associ- ated with two types of acute leukemia: lymphoid leukemia excluding CLL (2.05; 95% CI 0.71-5.92) and myeloid leukemia excluding chronic myeloid leukemia (1.44; 95% CI 1.00-2.09~.
OCR for page 63
63 o Ct sir Ct Cd o o V o Cal By X no o G) r9 ~ o ~ I o _ o C`O ~ o N~ o =6 C=~: ~ ~ no ¢ V) U' ~ 5-, Ct So, 1 g sly 50 -9 ~ O ~ in o o ~ ~ Cal Z ~ · ~ 1 Ct ~L . C) ·~ ~ o ~ ~ rid o o .... ~4 1 1 1 1 Go of ~ o ~ ~ AS ~ .... o o o o _ _> ~ o o o .... ~_4_~ U) rid ~ ~ ~ . . . . o o o o ~ ~ ~ ~o ~ o ~ ~ ~ ~ - ~ - ) ~ o o o o v) 4~ 'e a, o v cr v u, . - o au e v · 4 o oo ~ v) ~ oo oo ~ ~ o ~ ~ oo ~ ~ - o oo c~ oo ~ o l i - o o ~ o - ~ v, · - c) u) ~ 'e c<, · cq · ~'e g ~ ~ - - - ¢ ¢ ¢ oo o o ~ l o^ o o o e c~ ~ - s~ c~ e e e c) ~ - o~ . . - .s~ c~c~o' ) ~o .~ =O c~- o so ~ ~ c~ ~ o ~o ~ C4 =0 ~ 11 ~t4 c~8 °- ~`:: ~ V ~¢ -^ C~ . ~ Cd ~ ~ V .= C2 ~ >, _ V =,9 Ce .. .o ~ ~ ~Q C~ .= o s~ Ct .= _ C 's ~ _' ~ ~ o ._ ~ G) ~ U, a' Cd ._ 3
OCR for page 64
64 sit To, o C=t N Ct Cal US o V .~ U. ~2> ·O ._ ~ ~ o Ct O C<' ~ V .N Ct ~ O ~ Cal V: ~ V Cal ~ CO Cal sol ~ m ~ O g I ED ,- 0 ~ O ~ O . ~ Z ~ 4~ lo, ._ c) ED ~ o .~ Ct me V cr ~ _' =0 So: U.= ~ -9 ~ ~ 0 ~ =0 V .= ·C=d V) CQ (D Coo 0 ~ , ~v vo o e v 'e o cO v 'e o m C~ ~ ~O O _ O C ~ ~ _ O _ _ ____ _______ I II IL II II I I oo0 ~r oo ~ o0~ ~cr~ ... .. .. .. . . o o o o o o o o o o o ~x_ _ ~_ _~ ~oo ~ ~ oo cr 0 oo 0 ~ 0 ~ 0 0 0 ~ . . . . . . . . . . . o-o ~o-o---o ao ~ ~ ~ ~ _ . . . . . . . .. . . o o o o o o o oo o o ~oo~ ~ ~_oo oo o ~ ~oo _ ~_ ~ o oo- _ oo~ _ ~_~ - , . . .. .... . .. o o o o o o o o o o o _ ~_ ~o - . - ao _ o ~_ ~OX _ <5N - _ ~oo crx ro O ~ oo ~oo ~ - t- _ I I ~ I I I t I I I ~o o o o o o o o o o O ~- d oo O ~c-] o _ ~C-] ~s o o . ~a, ·= ~ E E o .E ~E e ~ _ ~° _ -~ ~ E s ` Ce o ~ ~ , = o ~ ~ ~ o ~ ~ ~ ~ ~ C,q ~ ~ ~ ~ ~ o . ~ o o o o o o ~ =, . ~ . ~ ~ ~ ~ ~ Cq Cq C~ . ~ . ~ ~ o C~ CO U) C~ _ _ ~._ _ ._ ._ ._ ._ ._ ~ 2 m ~ ~ 2 ~ ~ ~ ~ ~
OCR for page 65
65 ~oo- ~ r ~O- ~O ....... - ~--o- ILIIIII oo ~oo ~oo ~o ~ oooo-oo _~_~_~_~ oo ~oooo ~o ~ooo ~o ....... o--o_o- ~oooo_ ax ~_ ..... . ooo-o o ~oo- - ~-C~t- __ - oo ~_ o ..... . ooo-o o ~CS ~oo _ ~_ooo C~' o ~ ~ ~oo - cr~ ~cr.ONON o ~ ~ ~1 1 1 o 0 0 oO ~- d" oO °° ~1 U. ~ . . o ._ ~. - C ~s~ o Cq o ~ . ~ Ce Ce ~o ,y ~Ct o _ ~ c-> u] o o c.) 'e ~ g s: u, c~ ~ ·~ ~ ~ ~ ~ o ce 5 ~q U~ ~c) ~o ~ ~ co~ ~ ~ ~ ~ ~ 5 Ct o ._ C) U) Ct C) ._ _ ~ 4_ s~ '0°. ~so~ o So ~ ~o .° V) ~o~ ~ h_4 to Ct 11 ~ b4 a ~c~ ~ _^ ~ ~ ~ .~ >~=o ~ -~ ::s~ Ct ° o Ct Ct ~ 11 ~ ct ~ =O V =8 :O ^ =~;
OCR for page 66
66 o ._ Cal N Sea, ~ Z Ct ~ ~ Cal O '-1 '- V ~ ~ O V) ~ V O ,: am .~ ~ Ct O ·_ ~ .~ ~ a= - ~ ) Ct ~ O V) U) Cal Em v ~ S ~ in to 1 a' ~ 0 ~ O O 50 -9 ~ Cal .~ E-. To c o ._ Cal it ~ V ~ _ o v PA o v :O Ct ~ 1 au U) Cd is at 0 ~ c~ cq ~ ) o ~ cn o v 1 v e o c~ 'e v ~n e - o au z 'e ~ - cd ~ u~ ~ ~ o o ~ ~ ~ ~ o ~ ~ ~ ~ ~ o oo ~ ~ ~ ~ o - ~ ~ ~ ~ ~ oo ~ o oo ~ o ~ ~ oo ~ ~ ~ ~ ~ - ~ ~ ~ - ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ t cr~ 0 ~ c~ . . . . . . . . . . . . . . . . . . . . O O O O O O O O O O O O ~ O O O O O O O _~ _~ _~ _~ N~ ~ _~ ~ ~ ~ ~ _~ ~_ ~ ~ ~ ~ ~ ~ ~ ~ oO oo ~ cr ~ ~ ~ _ ~ ~ ~ ~ c~ 0 cr ~ ~ ~ 0 ~ ~ 0 ~ ~ ~ ~ ~ 0 ~ 0 0 _ . . . . . . . . . . . . . . . . . . . . o ~ o o o o ~ o o ~ o o ~ o o - ~ - ~ ~ oC ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ o ~ ~ ~ o oo ~ ~ c~ ~ ~ ~ ~ oo ~ ~ ~ ~ oo ~ ~ ~ ~ o o o o o o o o o o o o o o o o o o o ~ oo ~ ~ ~ o c~ ~ o ~ ~ ~ o ~ ~ ~ - ~ ~ ~ ~ o oo oo ~ oo ~ o o o o o o ~ ~ ~ o oo ~ ~ o ~ ~ ~ ~ ~ ~ ~ oo ~oo ~ oo oo o o o o o o ~ o o o o o o oo o ~ ~ ~ o ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ o ~ ~ ~ ~ ~ ~ oo ~ ~ c~ ~ ~ ~ ~ ~ oo ~ ~ o ~ O ~4 ~ ~ - ~ ~ o ~ o o ~ ~ ~ ~ ~ ~ ~ o o ~ ~ o o ~ ~ ~ ~ ~ ~ ~ ~ ~ u~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ t~ t~ _4 _~ _~ _~ _~ ~ _4 _I ~ ~ _~ ~ _ _4 _ ~4 _ ~ _ C~ Ct au ~O ~ ~ ~,_ ~ '= == ~ _ ~ ~ ~ D ~ ~ ~ ~ =, ~ ~ ~ _ . ~m
OCR for page 67
67 ~ _ C~ ~ o o _~ _~ ~ o _ _ o ~ o~ ~ ~ ~ ~ ~ o ~ ~ o . . . . . . _ ~' _ _ _ _ l l l l l l o . . . . . . _ o o o o o _~ _~ _~ _~ ~ o ~ o o o ~ ~ ~ oo o _ oo . . . . . . _ _ o _ _ o l ~o _~ rn _ - ~ ~ . . . _ _ _ I I ~ oo . . . o o o _/ ~ o o oo . . . _ _ o ~o ~ o .° I I oI 3 . ~ o o oo C~ o ~ ~U. o . . . . ~ ~ ~ ~ - , ~ oo ~ ~ oo C~ ~ ~ ~ ~ ~ oo . . . . . . . . . . . . . . .. o o o o o o o o o o o o o o oo - oo cr~ C~ ~ ~ ~ ~o ~_ Ir, ~t _ _ _ o . C~ Cq ~s . s .o ~q ~ oo ~ oo ~ ~o ~ - s 11 · - .^ c)~ .= .= oG et ~ c) ~ d- ~ 'e ~ o .s ·~e ~ ~ v ~ ~ 'e <=D == ~ ~ Z o ~ ~ ~ - ~ ~ ~ oo oo - o ~ o ~ ~ ~ ~ oo ~ ~ ~ ~ ~ o ~ ~ ~ oo ~ ~ . . . . . . . . . . . . . . .. - o o o o o o o - o o o o o oo ~ ~ o o ~ oo ~ ~ c~ ~ oo ~ ~ ~ t ~ ~- ~ - ~ - ~- - oo - ~ c~ ~ - ct u, 'e s~ - e~ 4 - s~ oo o - ~ ~ oo ~ o oo oo oo oo ~ - - - - - e u, o c) G E- - _ ~ C~ _ _ oo ~ o o 1 o o - ~ o o o o C) q) . ~ ~.O 'e W ~ o ~ ~ -= m Y ~ m ~ E Z X ~ l o ~o ~ o o o Ct Ct Ct ._ U) ~ ~_ ~C~ ._ 4 - ~Ct o C ~. Cq ~ _ Ct s~ O G, U) ~ ~ ~< - o, ~ _ ~ . ~ .° ~ .= ~o o ~ ~ o Ct Cq ¢ a' V ~ U~ ,,= C~ Ce s~ ~ Cq ~ Ce ~ ~; o > ~ o ~ ~ ,~ o C) ~ c ) -
OCR for page 68
68 o o .~ N o ,S _ D ~o Ct ~ 4~ 1 ' V Ct ~ ~ 1 ~ Us -9 O O ~ ~0 ~ ·~ Z ~ ~L, :O En ~ o Cal He V \ ~ o Ct ~ _' A U. o V Cal V o Ct V of ~ ~ ~or O ~ O . . . . . . . . . . ~ I ~ I ~I t I I I ~ o oo ~Cat o .... ... . . . ooo-~ ~o~ ~n ~c~ o . . . .. . . . . . - ~- -- o ~ ~ ~- ~ ~o oo o o v) oo ~ ~ ~ v) ~ oo oo . . . . . o o o o o o - ~- ~ ~t l (~N o o o o o o l co~ o o · c~] o v .= c) x ce .s ~ - - ~ - 4~> ~ a' v >` ~ o ~ oo ~ oo oo ~ ~ ~ o o o o o ~ ~ ~ oo c~ - . . ... o o o o o ~oo oo o ~ ~ ~ oo ~ o o o o o - - c) e a' ~ ~ - a ~ ~ _t ~ ~ ~0 C) ~ Ct ~ O ~ Ct . - Ct o E~ _` O .~ 11 V Cd . ~ .= _ ·0 .= So~ 11 V . ~ .~ ·0 _ C) .= 11 ~ .o V ·~ Ct ,= 0 c' S ~ 11 .° _4 ~ V ~ .. ~ . E~ 0 zO Ct o .~ g . C~ S~ U: C~ ~4 C~ Ct C) . - C~ C~ S~ a' U, 5 - =0, C~ - ~0 C) ~ Ct O . - .s: ~3 s~ N t - O CO _ ~ e~ 0 O O O C) ~V O . O ,s: au ~ '= ~ C~ ,=0 ~ ~ O Ct Ct C) . _ _ 4 - C~ C) ~ ;> _ ~ C) g :O >` Ct ~ ~ t4 V) ~Ct~
OCR for page 69
FINDINGS 69 INVESTIGATING LEUKEMIA RISK BY LAND AND SEA SERIES PARTICIPATION We stated earlier that although we had not identified any unambiguous proxy measures for radiation dose, we had considered the possibility that the different test series or the land and sea series represent differences in exposure experience. Exposures were not uniformly distributed within series. Operation PLUMBBOB, for example, consisted of 30 tests, including safety tests that pro- duced negligible yields as well as detonations of up to 44 kt (Harris et al., 1981~. A formal test of heterogeneity of leukemia minus CLL risks among the individ- ual test series was not statistically significant (x2 = 7.191, 4 df,p = .13~; thus, we removed the series-specific analyses from the core presentation of this report. To be complete, however, risk estimates of individual test series, service branch, and paygrade groupings are presented in Appendix E. Our interest in the land-sea dichotomy was based on the possibility that the exposure experience was qualitatively and perhaps quantitatively-different at the two test sites. In Table 11-5, land series participation is defined as attendance at UPSHOT-KNOTHOLE, PLUMBBOB, or any other test series conducted at the Nevada Test Site, regardless of a participant's selection series. Similarly, atten- dance at GREENHOUSE, CASTLE, REDWING, or any other series conducted at the Pacific Proving Ground constituted sea series participation, regardless of se- lection series. Members of the referent cohort were classified only by their coun- terpart's selection series. Thus, because a participant may have attended both land and sea tests, the number of land series participants plus sea series participants is greater than the total number of participants. Approximately 5 percent of the par- ticipant cohort members were associated with both land and sea senes. TABLE 11-5. Number of Participants Who Participated in Any Land Series and in Any Sea Series, by Selection Series Selection Series Selection Any Land Any Sea Series Series Series GREENHOUSE Sea 9,5287819,528 UPSHOT-KNOTHOLE Land 18,47318,473681 CASTLE Sea 15,68549015,685 REDWING Sea 12,9231,16012,923 PLUMBBOB Land 11,55911,559481 Total 68,16832,46339,298 Participants in both land and +2,431+1,162 sea senes NOTE: Boldface signifies personnel counted in both "any land" and "any sea" cate- gories. No member of the Referent Cohort participated in any test series. To maintain com- parability between cohorts, referents are assigned the selection series of the matched participant unit.
OCR for page 70
70 THE FIVE SERIES STUDY Table 11-6 shows that for all-cause mortality, land series participants have a slightly, but significantly, lower risk relative to referents (HR = 0.96; 95% CI 0.93-0.99), whereas sea series participants have a slightly higher, also significant, risk relative to sea series referents (HR = 1.03; 95% CI 1.00-1.06~. Neither land series nor sea series participants had a significantly higher risk of all-malignancy death (HR= 1.00 and 1.04, respectively). However, land series participants show a statistically significant increase in the hazard ratio for death due to leukemia, 1.49 (1.0~2.13~; for sea series participants, the hazard ratio was 0.92 (0.67-1.27), not significantly different from 1.00. Thus, participation in a land series is associated with a significantly higher risk of leukemia death, while participation in a sea se- ries is associated with a significantly increased all-cause death rate. INVESTIGATING LEUKEMIA RISK BY TIME SINCE FIRST PARTICIPATION AND AGE AT FIRST PARTICIPATION To explore the increased, but not statistically significant, risk of leukemia observed among participants relative to referents, we looked for patterns conso- nant with other research findings on the association of radiation and leukemia. We therefore fit a model using time-dependent covariates to estimate the risk of leukemia death in three periods relative to first series participation: less than 5 years, 5 years to less than 15 years, and 15 years or more after participation. We created a model to estimate the risk of leukemia mortality for three ranges of age at first participation: less than 20 years of age, 20 to 25 years, and 25 years and older. Even with this study's very large cohort, there were insufficient early leu- kemia deaths to yield a definitive picture. Tables 11-7 and 11-8 (see page 72) show that the relationship between participant status and leukemia mortality does not seem to follow anticipated patterns of latency and age at exposure (Boice, 1996~.
OCR for page 71
71 Go so . 'e He ~ - · o o · - 'e cd ~ To .~ 'e · of · ~ 'v . - ~ v) CQ ~ -9 v: o ~ · ~ 'e - ~ ~ · - ~ - ; 'e En ~ - ^ 11 s Ace I v] o^ 11 tact au au v) 'e do .~ in v At\ ~ Go Do of o ~ it o ·c~ ma v 'e ~ o - ~ of ce au o ~ ~ to to ~ - ~ - o l l 0 cr ~ . . . to to to to . . . to ~ o ~ . . . ~ ~ o . . . o o ~ ~o o ~ ~ o ~ . . . o ~ - oo ~ ~ .O E ~ ~ s.= C~ ~ C<~ ~SO ¢ ¢ ~ ° - Ct ~ o . ~ ~ o . ° U~ . .= U: ~ s Ct Cq Ct =._ ~ ~ c~ U~ ca ce · ~ o =, o 3 :~ ~ O~ ~ $: ] ~ ~ =o co' ~ ~ ~ C ~ ~ ~ ~o ID 3 C E c~s ~ ~ .O ~ C O E =, c4 = ¢ ~o U, m_ ~, ~o 8 2 5 ~. ~ ~ a' ~ O ) C) · s~ o o ~ a' o o C) _' ._ ~ _ o == C~ U, ~ ° a., O-= ~ ~ au ~ c; ct .o ~ U) _ Ct _ Ce C~ ~ cn O 'c) ~ =-,=.= Ce ~ o ° Ct ° . - . _ ,= =. Cd o ~ ~ .O C~ o Ce Cd ~ 4- Ct.~ g C) ~ C~ _ cL, s~ ~ ~ ·- C~ . ~ C := o ;> C~ ~ C) ~m~= V,= =~ a' Cd CO Ct ·~ S~ ~ Ct
OCR for page 72
72 THE FIVE SERIES STUDY TABLE 11-7. Hazard Ratios for Leukemia, Excluding Chronic Lymphoid Leukemia, by Time After First Exposure Hazard Ratio* (95% CI) Time After First Exposure <5 years 5 to <15 years 215 years 0.80 (0.2~2.42) 1.09 (0.59-2.02) 1.16 (0.90-1.49) NOTE: CI = confidence interval. *Participant cohort relative to referent cohort. Proportional hazards model controls for series, service branch, and paygrade by stratification and age, age squared, and age cubed by covariate adjustment. TABLE 11-8. Hazard Ratios for Leukemia, Excluding Chronic Lymphoid Leukemia, by Age at First Participation First Participation Hazard Ratio* Age (years) (95% CI) <20 20to<25 225 1.18 (0.52-2.70) 1.22 (0.81-1.85) 1.08 (0.79-1.47) NOTE: CI = confidence interval. *Participant cohort relative to referent cohort. Proportional hazards model controls for series, service branch, and paygrade by stratification and age, age squared, and age cubed by covariate adjustment.
Representative terms from entire chapter: