The use of 95% CI or 90% CI for drug product development - a controversial issue?

For review and approval of drug products, a 95% confidence interval approach for evaluation of new drugs is commonly used, while a 90% confidence interval approach is considered for assessment of generic drugs and biosimilar products. In the past decade, FDA has been challenged for adopting different standards (i.e., 5% type-I error rate for new drugs and 10% type-I error rate for generics/biosimilars) for regulatory submissions of drugs and biologics. This note intends to clarify the confusion by pointing out the fundamental differences between (i) the concepts of point hypotheses and interval hypotheses, and (ii) the concepts of interval hypotheses testing and confidence interval approach. In general, the method of interval hypotheses testing is not equivalent to the confidence interval approach although they may be operationally equivalent under certain conditions.