Fig 1. Difference between survival analysis in a clinical setting with balanced populations and genomics setting, with unbalanced populations.

(A) In a typical clinical study, two pre-selected groups of similar size are compared. Because the groups are balanced and each has a suitable number of patients, the asymptotic approximation (normal distribution) used in common implementations of the log-rank test gives an accurate approximation of the exact distribution, resulting in accurate p-values. (B) In a genomics study, the two groups are defined by a genetic variant. In many cases, the sizes of the groups are unbalanced, with one group being much larger than the other. In this situation, the asymptotic distribution does not accurately approximate the exact distribution of the log-rank statistic, and the resulting p-values computed from the tail of the distribution (see inset) are inaccurate.