Table 2 |
|||||||||
|
Sensitivity analysis of threshold for margin of error and outcome measure |
|||||||||
|
Recurrence and mortality |
|||||||||
|
|
|||||||||
|
Threshold (±) |
Right before |
Wrong before |
Stayed right |
Stayed wrong |
Got right |
Got wrong |
Right after |
Wrong after |
McNemar p-value |
|
5% |
2 |
18 |
1 |
12 |
6 |
1 |
7 |
13 |
0.125 |
|
10% |
4 |
16 |
3 |
6 |
10 |
1 |
13 |
7 |
0.012 |
|
Recurrence |
|||||||||
|
5% |
6 |
14 |
3 |
5 |
9 |
3 |
12 |
8 |
0.146 |
|
10% |
9 |
11 |
7 |
3 |
8 |
2 |
15 |
5 |
0.109 |
|
Mortality |
|||||||||
|
5% |
5 |
15 |
4 |
9 |
6 |
1 |
10 |
10 |
0.125 |
|
10% |
7 |
13 |
7 |
5 |
8 |
0 |
15 |
5 |
0.008 |
|
Recurrence or Mortality |
|||||||||
|
5% |
9 |
11 |
7 |
3 |
8 |
2 |
15 |
5 |
0.110 |
|
10% |
12 |
8 |
11 |
2 |
6 |
1 |
17 |
3 |
0.125 |
|
|
|||||||||
|
This table shows how results change moving from a ± 5% to a ± 10% accuracy threshold within an outcome measure, and how results change depending on which outcome measure is the focus of the analysis. |
|||||||||
|
Belkora et al. BMC Cancer 2009 9:127 doi:10.1186/1471-2407-9-127 |
|||||||||
