Sample Size Estimation
To estimate the trial sample size, EF75 values were measured pre- and
post- treatment in seven patients clinically prescribed albuterol who
otherwise met study inclusion criteria. We observed a mean absolute
change in EF75 of 1.3L/min ± 0.8 and powered the trial using a
conservative target of approximately half that value, 0.6L/min. Sample
size estimations were performed in a two-step process. We first
calculated the necessary number of participants using a paired t-test
for two correlated means. Based on an alpha of 0.05 and correlation of
0.5, a sample size of at least 21 infants provided ≥90% power to detect
a mean difference of 0.6ml/min ±0.8 using a baseline EF75 mean of
2.6L/min and a post-treatment mean of 3.2L/min. We then used simulation
with 200 replications to model the necessary number of subjects for a
generalized linear model based on an estimated correlation of 0.5
between each of the mean EF75 values for the 3 treatment doses. Based on
this calculation, a sample size of between 24 and 25 subjects provided
approximately 80% power to detect a mean difference of 0.6L/min between
the 3 treatment doses. A sample size of 24 patients was selected to
ensure equal representation of the 6 possible treatment ordering
schemes.