Skip to main content

Table 4 Comparison of the non-stratified and stratified log-rank type test for different covariate types: power

From: On comparison of net survival curves

 

62% events due to ex. haz.

93% events due to ex. haz.

 

LRt

LRt-str

LRt

LRt-str

X∈D, bal (sex); S∉D, bal, bin

0.498

0.501

0.879

0.882

X∈D, bal (sex); S∉D, imbal, bin

0.508

0.510

0.872

0.874

X∈D, bal (sex); S∉D, bal, 10 str

0.501

0.498

0.875

0.878

X∈D, bal (sex); S∉D, bal, bin, NPH

0.514

0.520

0.887

0.893

X∉D, bal, bin; S∈D, bal, bin (sex)

0.483

0.489

0.872

0.875

X∈D, bal (sex); S∈D, (age; 3 str)

0.512

0.513

0.871

0.873

  1. Methods included: non-stratified (LRt) and stratified log-rank type test (LRt-str). ((im)bal = (im)balanced variable, i.e., the groups occur with (un)equal probabilities; bin= binary variable; 10 str = 10 strata; NPH = nonproportional effect). X is the categorical covariate of interest, S the stratification covariate, D denotes the demographic variables
Back to article page

BMC Medical Research Methodology

ISSN: 1471-2288

Contact us