In studies such as these with huge intersubject variability, and probably also quite large intrasubject variability from day to day on many scales, it makes sense to look at change from baseline for individuals in the various groups rather than just differences in raw scores between treatment groups.
It seems by a cursory glance that most analyses done were of the latter kind, and I think then a study such as this, with low N, is expected to not show very much effect of any kind. (Since the intersubject variability and potentially also the natural intrasubject variability for most measured scales seem higher than any expected treatment effect).
I am almost certain that e.g. none of the approved and quite convincingly working SSRI:s would have shown any efficacy in a study with this design and similar N.
It seems by a cursory glance that most analyses done were of the latter kind, and I think then a study such as this, with low N, is expected to not show very much effect of any kind. (Since the intersubject variability and potentially also the natural intrasubject variability for most measured scales seem higher than any expected treatment effect).
I am almost certain that e.g. none of the approved and quite convincingly working SSRI:s would have shown any efficacy in a study with this design and similar N.