WPC 2 ZBn X&m P7&PBibliogrphy66BibliographyN=(Nt{h *NxV{$E Doc Init{66Initialize Document Style{h *NxV{$E    I. 1. A. a.(1)(a) i) a)DocumentҲa1DocumentE+N xVDocument Stylex~N=(Nt{h !*NF *  ׃  2qea2DocumentE+N xVDocument Stylex~N=(Nt{h !*N*    a3DocumentE+N xVDocument Stylex~N=(Nt{h !*N0     a4DocumentE+N xVDocument Stylex~N=(Nt{h !*N   . a5DocumentE+N xVDocument Stylex~N=(Nt{h !*N  2Ke p p ca6DocumentE+N xVDocument Stylex~N=(Nt{h !*N  a7DocumentE+N xVDocument Stylex~N=(Nt{h !*N ` ` ` a8DocumentE+N xVDocument Stylex~N=(Nt{h !*N ` ` ` Tech Init"66Initialize Technical Style{h "*NxV{$E  1 .1 .1 .1 .1 .1 .1 .1 Technical2 } -^ a1Technical+N xVTechnical Document Style~N=(Nt{h #*N 4!     a2Technical+N xVTechnical Document Style~N=(Nt{h #*N *    a3Technical+N xVTechnical Document Style~N=(Nt{h #*N'   a4Technical+N xVTechnical Document Style~N=(Nt{h #*N&   28 $  .  a5Technical+N xVTechnical Document Style~N=(Nt{h #*N&   . a6Technical+N xVTechnical Document Style~N=(Nt{h #*N&!"  . a7Technical+N xVTechnical Document Style~N=(Nt{h #*N&#$  . a8Technical+N xVTechnical Document Style~N=(Nt{h #*N&%&  . 2o4j 0Pleading{$66Header for numbered pleading paper*NxV{$E'(   ,#&m P7&P# X  y*dddyy*dddy H\1 H\2 H\3 H\4 H\5 H\6 H\7 H\8 H\9 H10 H11 H12 H13 H14 H15 H16 H17 H18 H19 H20 H21 H22 H23 H24 H25 H26 H27 H28   Ӽa1Right Par+N xVRight-Aligned Paragraph NumbersNt{h /*N8)*@   a2Right Par+N xVRight-Aligned Paragraph NumbersNt{h /*NA+,@` `  ` ` ` a3Right Par+N xVRight-Aligned Paragraph NumbersNt{h /*NJ-.` ` @  ` `  2Na4Right Par+N xVRight-Aligned Paragraph NumbersNt{h /*NS/0` `  @  a5Right Par+N xVRight-Aligned Paragraph NumbersNt{h /*N\12` `  @hh# hhh a6Right Par+N xVRight-Aligned Paragraph NumbersNt{h /*Ne34` `  hh#@( hh# a7Right Par+N xVRight-Aligned Paragraph NumbersNt{h /*Nn56` `  hh#(@- ( 2G 0a8Right Par+N xVRight-Aligned Paragraph NumbersNt{h /*Nw78` `  hh#(-@pp2 -ppp Њ&m P7&P) `(CG Times (Scalable)&&m P7&P) `(CG Times (Scalable)&`\  PQP)\  ` TmsRmn 10pt (F)`\  PQP)\  ` TmsRmn 10pt (F)2y3|Z#`\  PQP#S!#`\  PQP#j)DS & APPENDIX D ă ! STATISTICAL FORMULAE ă     APPENDIX D. STATISTICAL FORMULAE ă D.1. CELL FREQUENCIES The observed outcome of a casecontrol study or a cohort study may be depicted in a 2x2 table, where a, b, c, and d are cell frequencies. " ETS ExposedЃ ^ Dddxp  Dddxp ^ @@YesNoP P Lung Cancer PresentYesab`  `  Nocd  D.2. CASECONTROL STUDIES The true (but unknown) odds ratio is estimated by the observed odds ratio (OR), ^'OR = ad/bc. A confidence interval on the (true) odds ratio may be calculated from the normal approximation to the distribution of log(OR), the natural logarithm of OR (Woolf, 1955). The variance of log(OR) is estimated by 0 Var(log(OR)) = 1/a + 1/b + 1/c + 1/d and the standard error by its square root, !SE(log(OR)) = (Var(log(OR))). Approximate 90% confidence limits are given by "log(OR)  1.645 SE(log(OR)). The value 1.645 is replaced by 1.96 for 95% confidence limits and, in general, by Z/2 for 100(1- )% confidence limits. The confidence bounds obtained in this way are sometimes called logit limits (Breslow and Day, 1980, p. 134). Significance level (pvalue) of a test for effect, i.e., Ho: (true) odds ratio = 1 against the alternative Ha: (true) odds ratio > 1, is the area under the standard normal curve to the right of the value of the test statistic, given by log(OR)/SE(log(RR)). If the (true) odds ratios are assumed to be equal in k studies, then a pooled estimate is calculated from !log(OR(P)) = wi log(OR)i/wiă where the summations are on i, from 1 to k; OR(P) is the pooled estimate; log(OR)i is the logarithm of OR from the ith study; and wi = (Var(log(OR)i))é1 is the weight of the ith study (Breslow and Day, 1980). D.3. COHORT STUDIES The true (but unknown) relative risk is estimated by the observed relative risk (RR), $RR = (a/a+c)/(b/b+d). A confidence interval on the (true) relative risk may be calculated from the normal approximation to the distribution of log(RR), using the analogue of Woolf's method referred to above (Katz et al., 1978). The variance of log(RR) is estimated by, Var(log(RR)) = c/(a2 + ac) + d/(b2 + bd) and the standard error by its square root, !SE(log(RR)) = (Var(log(RR))). The remaining calculations follow the description for casecontrol studies in Section D.2 with "odds ratio" and "OR" replaced by "relative risk" and "RR," respectively. The pooled estimate of relative risk from both casecontrol and cohort studies is calculated by the same methodology for pooling estimates from casecontrol studies or from cohort studies separately, i.e., the logarithm of each individual estimate is weighted inversely proportional to its estimated variance (Kleinbaum et al., 1982).