The study was designed to have 81% power to detect an absolute difference of 10 percentage points in cumulative live-birth proportions between treatment groups (20% in the clomiphene group on the basis of the results for the clomiphene-only group after five cycles in our prior study2 (link) vs. 30% in the letrozole group), with the use of Pearson’s chi-square test at a two-sided significance level of 0.05. We calculated that the analysis would require a sample of 300 patients per treatment group, which we increased to 375 to allow for a dropout rate of 20%. A chi-square test (or Fisher’s exact test if any frequency count was <5) was used for testing differences between the two treatment groups. For all continuous variables, the mean and standard deviation in each group are reported.
We used Kaplan–Meier curves to report the time from randomization to live birth according to treatment group and according to treatment group and tertile of maternal body-mass index (BMI). A log-rank test was used to test the interaction between BMI tertile and study treatment with regard to the time from randomization to live birth. Although these BMI groups were not pre-specified in the study protocol, our previous studies have suggested that BMI affects infertility treatment in women with the polycystic ovary syndrome.2 (link) To avoid type I errors from multiple comparisons, we did not explore various BMI stratifications but instead used simple tertiles. All analyses were performed with the use of SAS software, version 9.2 (SAS Institute). Data were analyzed according to the intention-to-treat principle.
We used Kaplan–Meier curves to report the time from randomization to live birth according to treatment group and according to treatment group and tertile of maternal body-mass index (BMI). A log-rank test was used to test the interaction between BMI tertile and study treatment with regard to the time from randomization to live birth. Although these BMI groups were not pre-specified in the study protocol, our previous studies have suggested that BMI affects infertility treatment in women with the polycystic ovary syndrome.2 (link) To avoid type I errors from multiple comparisons, we did not explore various BMI stratifications but instead used simple tertiles. All analyses were performed with the use of SAS software, version 9.2 (SAS Institute). Data were analyzed according to the intention-to-treat principle.
