# Sample Size calculation for Non-Inferiority trials

#### AgraRD

##### New Member
Hi! If we are calculating sample size for a pediatric non-inferiority trial, (and there is evidence in adults that the test drug has 10% more efficacy than the comparator drug, and is cheaper as well), can we assume different event rates for the wo drugs in the calculations for the sample size, or do they have to be the same?

#### hlsmith

##### Not a robit
Are you saying in the adult population the two drugs have been revealed different at the 10% mark, but you want to do a non-inferiority test in peds population. Are you setting yourself up for disappointment or do you have a biologically plausible rationale why those results won't generalize to peds (e.g., kids aren't little adults)? Non-inferiority you set a margin (one-sided) that the drugs cannot differ by. If the drug proved out in the adult population that margin better be pretty big, but then that may put into question the conundrum of statistically significant vs. clinically, since the NNT would be 10.

#### AgraRD

##### New Member
Well, the test drug B has an efficacy in adults of around 65%, and the control drug A about 50%, roughly. A is about 50 times more expensive than B. I only want to prove non-inferiority of B to A, to justify replacing A with B. A superiority trial would require a larger sample size. Both drugs are not well established in children.

#### hlsmith

##### Not a robit
OK, your original post was not specific enough for me to follow that part. Well all you have to do is plug in the above rates, balance between groups, power, alpha, and margin for non-inferiority and you should be good to go.

#### AgraRD

##### New Member
That's my question, though. Is it ok for me to assume a higher efficacy for the new drug when I'm calculating sample size? Or do I have to assume equal efficacy for A and B?

#### hlsmith

##### Not a robit
To my knowledge you use the presumed efficacy, so they dont have to be congruent.