music is medicine of mind, which never ever been found to be incorrect, hence no need discussion among scientists working in the area of psychiatric disorders.
It depends on the aims of the music therapy and the pathology of the client. If a client is referred following stroke to encourage fine motor skill re-development, reliable physical assessment tools would exist to be able to give a pre- and post-therapy measure (though there's also the question of whether MT is having an effect independent of standard treatment/care...there are a few RCTs, but more would be lovely). Depression symptoms can likewise be quantified through psychological tests pre- and post-therapy; whereas if a child with autism is referred to encourage social skill development, there's not really a standardized measure that can be used. If therapeutic aims are strictly musical, it's usually up to the therapist. Many may have individual scales or infer efficacy from qualitative session descriptions, but there's no one measure for music therapy's efficacy. Did you have a specific client group in mind?
In my work at a mental health hospital, outcomes are measured the same as other disciplines... reduced symptoms, reduced readmission rates, reduced falls, reduced infection etc. As an example, I have attached a research trial I did in 2005 looking at the effects of music therapy in dementia. There, my clinical goal was to reduce or reverse cognitive decline, as assessed by blinded assessors and in comparison to a control group who did not receive the music therapy.