Safety analyses included all the patients who had received andexanet. The efficacy analysis population included only patients who retrospectively met both of two criteria: baseline anti-factor Xa activity of at least 75 ng per milliliter (or ≥0.25 IU per milliliter for patients receiving enoxaparin) and confirmed major bleeding at presentation, as determined by the adjudication committee. Initially, a sample of 250 patients was planned, which would provide 80% power to show that the percentage of patients with excellent or good hemostatic efficacy was more than 50%. The sample was adjusted to 350 patients in protocol amendment 4 (January 2017) to meet new regulatory requirements for sufficient numbers of patients for each factor Xa inhibitor and to have at least 120 patients with intracranial hemorrhage in the efficacy analysis population.
Continuous variables are summarized as mean and standard deviation or median and interquartile range; categorical variables are presented as frequencies. Percent change from baseline in anti-factor Xa activity was computed with a two-sided nonparametric confidence interval for the median.10 Percentages of patients with effective hemostasis are presented with a 95% confidence interval calculated with the binomial test. The association between hemostatic efficacy and change in anti-factor Xa activity was examined with the use of receiver-operating-characteristic (ROC) curves.11 (link) Analyses were performed with the use of SAS software, version 9.4 (SAS Institute).