# Table 2 Statistical power for detecting a gene by radiation dose interaction from various study designs

Proportion of gene carriers = 0.5% Proportion of gene carriers = 1%
Study design PP = 100%, PN = 100% PP = 90%, PN = 90% PP = 100%, PN = 100% PP = 90%, PN = 90%
Int RR at 2 Gy 70 100 70 100 16 20 16 20
NCC 3 51 55 48 59 56 63 65 71
CM 1:2 66 67 68 73 75 80 74 76
CM 2:1 32 31 43 45 37 47 48 51
Proportion of gene carriers = 2% Proportion of gene carriers = 3%
Study design PP = 100%, PN = 100% PP = 90%, PN = 90% PP = 100%, PN = 100% PP = 90%, PN = 90%
Int RR at 2 Gy 8 10 8 10 6 8 6 8
NCC 3 67 84 68 84 73 88 74 89
CM 1:2 78 86 83 86 82 90 80 92
CM 2:1 58 69 71 73 63 79 67 81
Proportion of gene carriers = 5% Proportion of gene carriers = 10%
Study design PP = 100%, PN = 100% PP = 90%, PN = 90% PP = 100%, PN = 100% PP = 90%, PN = 90%
Int RR at 2 Gy 5 6 5 6 3 4 3 4
NCC 3 81 88 81 91 71 96 72 86
CM 1:2 88 92 88 94 70 95 74 93
CM 2:1 75 87 81 90 64 89 67 83
1. The power for detecting a gene by radiation dose interaction varies as a function of the proportion of gene carriers, interaction rate ratio at 2 Gy (Int RR at 2 Gy), and predictive positive (PP) and predictive negative (PN) values for registry-noted radiation treatment for true treatment status. These estimates are based on 700 risk sets on a simulation study with 200 trials for each set of parameters. The proportion of registry-noted radiation treated is 40%, and the rate ratio for 2 Gy exposure in non-carriers is 1.5, which corresponds to the 95th centile of a χ2, four degrees of freedom distribution. CM 1:2, counter-matched design in which one member of the triplet is unexposed and two are exposed; CM 2:1, counter-matched design in which two members of the triplet are unexposed and one is exposed; NCC 3, nested case-control study in which each case is randomly matched to two controls.