Table 3

Sample sizes to detect a significant difference in overall mortality. Numeric Results with Proportion Lost to Follow Up = 0.0200







Hazard Two-sided


Power
N
N1
N2
S1
S2
Ratio
Alpha
Beta

0.8002
1520
760
760
0.8200
0.8800
0.6442
0.0167
0.1998
0.8004
1089
545
544
0.8200
0.8900
0.5872
0.0167
0.1996
0.8003
2125
1063
1062
0.8300
0.8800
0.6861
0.0167
0.1997
0.8005
1437
719
718
0.8300
0.8900
0.6254
0.0167
0.1995
0.8001
3219
1610
1609
0.8400
0.8800
0.7332
0.0167
0.1999
0.8000
2002
1001
1001
0.8400
0.8900
0.6684
0.0167
0.2000
0.8001
5543
2772
2771
0.8500
0.8800
0.7866
0.0167
0.1999
0.8001
3025
1513
1512
0.8500
0.8900
0.7170
0.0167
0.1999
0.8000
12068
6034
6034
0.8600
0.8800
0.8476
0.0167
0.2000
0.8001
5194
2597
2597
0.8600
0.8900
0.7727
0.0167
0.1999
0.8000
46657
23329
23328
0.8700
0.8800
0.9179
0.0167
0.2000
0.8000
11271
5636
5635
0.8700
0.8900
0.8368
0.0167
0.2000

Summary Statements: A two-sided log rank test with an overall sample size of 2002 subjects (of which 1001 are in group 1 and 1001 are in group 2) achieves 80% power at a 0.0167 significance level to detect a difference of 0.0500 between 0.8400 and 0.8900 – the proportions surviving in groups 1 and 2, respectively. This corresponds to a hazard ratio of 0.6684. The proportion of patients lost during follow up was 0.0200.

Software: PASS 2000, Power Analysis and Sample Size for Windows; NCSS, Kaysville, Utah

Raymond et al. Trials 2008 9:43   doi:10.1186/1745-6215-9-43