Table 5

Designs with minimum cost among designs with 80% power of detecting the interaction

Gene-surrogate
Flexible two-phase design options
Expected Phase One counts
Power#
Cost*

Spec
Sens
n0
n1
ρ0
ρ1
N0
N1



70%
80%
700
500
90%
80%
5163
1525
81%
1534
70%
90%
600
500
90%
80%
4412
1472
80%
1394
80%
60%
700
300
90%
90%
7721
1491
80%
1461
80%
70%
600
300
90%
80%
6585
1259
80%
1292
80%
80%
500
300
90%
80%
5461
1200
80%
1133
80%
90%
400
400
90%
80%
4348
1528
81%
1094
90%
60%
400
400
70%
50%
6667
1683
81%
1217
90%
70%
500
300
50%
50%
5896
1169
80%
1153
90%
80%
500
300
40%
60%
4673
1307
80%
1099
90%
90%
500
300
40%
50%
4630
1019
82%
1082

# Analysis approach: Maximum likelihood

* the study cost is computed as the sum of the number of screened subjects divided by 20 plus the number of subjects included in Phase Two.

ρ0 is the proportion of SG+ controls included in Phase Two

ρ1 is the proportion of SG+ cases included in Phase Two

Wild et al. Epidemiologic Perspectives & Innovations 2008 5:4   doi:10.1186/1742-5573-5-4