Menstrual phase and breast cancer surgery: influence on clinical outcome or pitfall of statistical analysis?

The influence of menstrual status at the time of surgery on the prognosis of women suffering from breast cancer is still discussed controversially. In our patient collective, including 149 patients, we obtained statistically significant results for six different time intervals, indicating that patients who underwent surgery between 11 and 22 days after the last menstrual period (LMP) have a poorer outcome. Focusing on the effect of statistical data evaluation strategy we designed a simulation study to evaluate the amount of type I error (error of a false positive test result) in a multiple testing situation involving a cyclical covariate. Accordingly, we corrected the minimum P-values for the occurring type I error rates. After that correction all six previously significant P-values failed to achieve statistical significance. The impact of different statistical data evaluation strategies in a multiple testing situation is discussed.