Estimate of the number of patients

When we designed the Valkyrie trial several different considerations were made on how to calculate the number of patients to be recruited for the trial. In the end, the decision fell on the way as described in the project [methodology][meto] (translated from portuguese):<br> - #### 5.7.Sample size:<br> - The sample size does not depend on the initial question, which is not comparative. However, in order to ascertain secondary outcomes, it was necessary to calculate the expected number of patients to be recruited. Using data available from patients already treated in our hospital (as discussed above) as a pilot group, assuming a hazard ratio of 0.05 and choosing a statistical power of 80% and 0.5% alpha the minimum number of patients required in the experimental group (in a 1:2 design) is 16 patients. We expect to require 2 years to recruit this sample. The comparison group to be used will be the historic control group of patients treated at our institution between 2000 and 2013, with schemes other than HIT (number of patients 31, suitable for this experimental design). The powerSurvEpi package of R language was used to calculate the sample size (R Development Core Team, 2012).(excerpt)