That's my point. If inclusion criteria were so good, then why not just repeat the design with a larger n?