Limitations of controlled augmentation trials in schizophrenia