Document wDpeJjxEaJaedkpg0E4npyngD

September 29,2006 Page 3 ATTACHMENT 2 Tables and excerpts from the draft protocols that contain raw data C-8 blood levels. Preliminary analyses were conducted on the pregnancies data including 18,276 women yielding 48,007 pregnancies. Of those pregnancies, 17,263 women reported 41,408 pregnancies occurring in 1960 or later. Nearly 40% of the women reported three or more pregnancies, and 23% reported only one pregnancy. Among the reported pregnancies, 11.0% (4,569) ended in miscarriage and 0.69 (287) ended in stillbirth. Among live births, 13.7% (4,824) were reported to be preterm, 5.9% (2,069) were reported to be low birth weight, and 1.7% (598) were term low birth weight. Preeclampsia was reported for 1,528 pregnancies. Birth defects were reported among 1,490 pregnancies, with the most common being congenital heart defects (275), club foot (91), eye defect (81), genitourinary defects (75), and oral clefts (52). Given that these preliminary analyses were based on 52,000 records out of an expected total of 69,000 participants, constituting 75.4% of the total, we generated the following estimates for the total number available for the analysis by multiplying the numbers by 1.33: 55,073 pregnancies after 1960 6,077 miscarriages 382 stillbirths 48,614 live births 6,416 preterm births 2,752 low birth weight births 795 term low birth weight births 2,032 preeclampsia 1,982 birth defects 366 congenital heart defects 121 club foot 108 eye defects 100 genitourinary defects 69 oral clefts 3.8 Power Power is computed for 2 scenarios: comparing exposed to some extent (n=80k) to not exposed (n=420k) comparing top 20% exposed areas (n=l 6k) to less exposed (n=64k) W0794491.1 P-2 September 29, 2006 Page 4 Power is presented in Table 3.4 as the chance of being able to detect a significant result for a modest increase in RR (1.25) in scenario 1 and a more substantial increase in RR 2.0 in scenario 2. [With alpha 0.05, two sided. Calculated with `sampsize' in STATA.] Note that this underestimates power relative to analyses that consider the full range of exposure or ordered categories, but ignores more complex models with multiple covariates which can reduce power somewhat. Nonetheless, these estimates convey a good impression of the magnitudes of association we will be capable of detecting. Table 3.4 Summary of power calculations B lad d er B reast (fem ) K id n ey L iver P a n c re a s T e s te s A ll % P o w e r to d e te c t R R of: 1 .2 5 (80k vs 420k) 88 100 65 25 54 30 8 9 (R R 1 .0 5 ) 2 .0 (16k vs. 64k) 88 100 94 50 87 58 100 (R R 1 .25 ) 3.1 Study population Brookmar are currently nearing completion of the C8 health project. They anticipate completing data collection in May/June 2006, and carrying out some data cleaning and validation over the subsequent months. We can expect a provisional data set from Brookmar shortly after June and a final dataset around December 2006. We have received and have been able to examine provisional data for the first 30,000 participants and (for fewer variables 50,000) which have been very useful for assessing patterns of available data. C8 has been measured in all participants, and included in the questionnaire were questions on whether they had every been diagnosed with cancer, a list of specific named cancers, and in each case the year of diagnosis was requested. Based on reported diagnoses of cancer in the first 30,000 participants, the reported numbers and prevalence (all ages) of each cancer site is given in table 3.1 Of the sites of primary interest liver and pancreas have low survival so self reported cases in survivors will be uninformative. This leaves Bladder, Breast, Kidney, Skin and Testes with 106, 559, 113, 1135 and 59 cases, respectively. W0794491.1 P-3 September 29,2006 Page 5 Table 3.1 Numbers and prevalence of participants reporting diagnoses of cancer in first tranche of 30,629 participants in C8 health project and estimated numbers to be expected in complete dataset. Sites with more interest for follow up are highlighted and the typical 5 year % survival is shown, for these sites. S ite MM B lood Bone B rain B reast (fern ) C ervix C olon Esophagus G all b lad d er K idney Larynx Leukem ia iN i Lungs Lym phom a M elan o m a M outh O varian P a g g rp a s P ro s ta te R ectum S kin Stom ach fiih s T h y ro id U terin e T o ta l #/ 3 0 .6 2 9 47 27 17 13 248 319 110 19 1 50 19 27 3 56 65 142 19 71 2 174 20 504 8 26 34 94 2115 % prevalen ce 0 .1 5 0 .0 9 0 .0 6 0 .0 4 0 .81 1 .0 4 0 .3 6 0 .0 6 0 .0 1 0 .1 6 0 .0 6 0 .0 9 0 .0 1 0 .1 8 0 .2 1 0 .4 6 0 .0 6 0 .2 3 0 .0 1 0 .5 7 0 .0 7 1 .65 0 .0 3 0 .0 8 0 .11 0 .3 1 6 .9 1 #/ 6 9 ,0 0 0 106 61 38 29 559 719 248 43 2 113 43 61 7 126 146 320 43 160 5 392 45 1135 18 59 77 212 4765 % 5 y e a r survival (A C S 1 995-2001) 82 88 65 9 92 4 100 96 Figure 3.1 Distribution of log C8 in blood W0794491.1 s Dgcl w