hyun_seoul said:The gap between trial results and real-world results is consistent and it is not fraud.
Agreed, and subgroup analyses deserve particular suspicion. With enough subgroups something is significant by chance, and pre-registered subgroups are a different animal from ones found afterwards.