Dropouts sample size

Does basing the sample size calculation on the per-protocol set and then increasing the sample size to allow for dropouts ensure that the per-protocol set will not be subject to bias ... 

No It often makes sense to power for the per-protocol set and then factor upwards to allow for dropouts as this will also ensure that there is enough power for the full analysis set providing any extra patient-to-patient variation in the full analysis set does not counterbalance the increase in sample size, but the analysis based on the per-protocol set is still subject to bias. See Section 8.5.2 for further discussion on this point. 

The chosen sample size reflects (i) the formal sample size calculation (using a residual variance value of 17.5%) based on the pharmacokinetics of ethinylestradiol in lean subjects, (ii) published sample sizes in other studies of this type of study ranged between 12 and 34, (iii) that this study will include overweight and obese subjects, a population who have been suggested to show a higher variability in their pharmacokinetics and in their menstrual cycles, and finally (iv) the plan not to replace dropouts. 

Subjects will drop out of trials for either random (ignorable) reasons or perhaps for a reason attributable to their disease, trial conditions, or other nonignorable factor. Both conditions are important to consider for efficacy trial simulation. In the former case, subjects who drop out (are missing) at random will result in a decrease in total sample size and may affect the study power. In the latter case, nonrandom dropout is considered to be nonignorable in that the reason for dropout is informative to the trial outcome and may bias the results. In the seminal paper by Sheiner (25), an example of nonrandom dropout is presented for an analgesic trial, where those subjects not achieving adequate pain relief were more likely to drop out (i.e., to take rescue medication). 

---