Clinical trial registry alone is not adequate: on the perception of possible endpoint switching and P-hacking.