I am conducting a study in which I test two different treatments A and B. The outcome is binary (success, fail). We want to demonstrate if treatment B is non-inferior to A with a margin of the proportions of 5%. In our study, we obtained the following table:
| Treatment | Success | Fail | Total |
|---|---|---|---|
| Treat A | 266 | 472 | 738 |
| Treat B | 232 | 435 | 667 |
| Total | 498 | 907 | 1405 |
Given these proportions, I obtained a p-value, given the article "Methods for Equivalence and Noninferiority Testing by Gisela Tunes da Silva, Brent R. Logan, John P. Klein" off 0.007 for the non-inferiority margin.
However, I needed to control for several variables and afterwards, I performed a propensity score matching, in which I obtained an OR of 0.9 [0.7;1.1]. The OR for the non-inferiority margin would be 0.59, and therefore I can reject the null hypothesis. I want to obtain a p-value, given this non-inferiority margin. Do you have any ideas on how to perform it? If it can be implemented in R that would be great.
