Table AV

       Comparison of Parameter Estimates For ERR Main Effects
           Model1 All Cancer with Ten Year Lag (N=28,347)
   White Males Ever Employed at X-10 or Y-12 Between 1943 and 1984
   

                           Internal2           External/Internal3	
       TERM                Estimate      SE     Estimate    SE 
 

 
      Age                    5.20      0.22     -0.27     0.38

    1900                     0.538     0.132     -9.2     13.4
    1900--09                 0.731     0.102     -4.3     10.3
B - 1910--19                 0.728     0.093    -10.1      9.3
    1920--29                 0.565     0.103    -25.0     10.3
    1930+                    0.072     0.181    -66.2     17.8

S4- Nonmonthly                   0         .        0        .
    Monthly vs Nonmonthly    -41.9       8.3    -41.3      8.3 

L - Worked at Least 1 Year       0         .        0        .
    Worked less than 1 Year   11.8       9.1     11.3      9.1

IG- Eligible Not Monitored       0         .        0        .
    Eligible and Monitored     5.8       7.5      4.3      7.5
    Not Eligible               2.6       8.7      2.9      8.7

F - X-10 Only                    0         .        0        .
    Y-12 Only vs X-10 Only    15.6       8.6     15.3      8.6
    Multiple vs X-10 Only      6.1       7.8      5.9      7.8

D5- External Dose (Sv)        1.52      0.82     1.45     0.81
    95% CI                (0.18, 3.59 )      (0.15, 3.48) 

  1. ERR main effects model: rate = exp[Age Term +B+S+L+IG+F] (1+D).
  2. For internal analysis baseline rates are estimated using log (age/52.5) for the age term, so that for B=1910-19, exp(.743)=2.11 cancer deaths per 1000. Note that for U.S. white males born in 1915 the cancer mortality rate at age 52.5 is 2.24/1000.
  3. For External/Internal model baseline rates are estimated using log (known rates from vital statistics for U.S. white males) as an "offset" as part of the age term. Consequently, for B=1910-19, exp(-8.5/100)=.907, is the estimated SMR for the 1915 birth cohort at the reference level of the other terms. The age variable, defined as (age-52.5)/100, was included to reflect any systematic age related departures from the external rates.
  4. The coefficients for the factors S, L, IG, and F are relative risk estimates in L% units, i.e. the cancer relative risk estimate for monthly vs non-monthly workers is exp(-41.9/100)=0.66.
  5. ERR per Sv with 95% likelihood based confidence intervals.