The Effects of Improvisational Music Therapy Clinical Trial: Needs for Future Research

The recent study published in the Journal of the American Medical Association on the effectiveness of improvisational music therapy for reducing symptoms in children on the spectrum has triggered a great deal of reaction. Since improvisation is such a cornerstone of our work, we wanted to share our view.

First: This is a Necessary First Step in Investigating when Improvisation may be an Effective Treatment

Congratulations to the research team. It is an important first step for the field that this study-a major music therapy clinical trial- was successfully completed and published in a top tier journal. Though the findings were disappointing, this is the beginning of the path to validating that improvisation is in fact a bona fide treatment for children on the spectrum. Early outcomes studies on recognized treatment modalities often find that the intervention being measured was not effective. For instance, the field of psychotherapy had early studies showing that it was failing to be an agent of change for patients (Eysenck 1952).  For an explanation of how this happened in the field of psychotherapy, click here.

Given that fact, the finding in this study is not unusual.Early outcome studies in all health fields are typically one-size-fits-all treatment thatproceed under the assumption that all patients/clients are created equally. These studies seldom find positive results.

We now have an important opportunity to build on this study. We need to look at it carefully and critique it so we can take the next step in the research process. To paraphrase Dr Perry Wilson, assistant professor of medicine at Yale in his review, this study has the potential to have a positive impact, to “drive the field in new directions.”  For his review of the study click here. He points out problems with the study and how helpful this study can be to propel the development of the field of music therapy

Second generation studies are more able to answer the question: "What treatment-specific benefits occur for which groups of patients under what conditions?" This is the next step that needs to be taken. This essay will end with a   description of the steps we plan on taking, after examining closely the study under discussion.

Second: There are Flaws in this Study to Address when Taking the Next Step

A closer look at the study reveals the following:

Limitations with the measurement tool

The ADOS, though well established as a diagnostic tool, is not considered an optimal tool to measure progress in therapy, a point the authors of this study make in the discussion section. It was designed to assess and diagnose, not measure progress (Payakachat 2012).  For more on this click here and here.

Also, since it has not been tested in any other language but English, it cannot be assumed as valid for use in this study since the treatment took place in several countries where English was not the primary language. There are many differences in social context that were not taken into account.

Limitations with the Assessors

It is not clear if the ADOS assessors from the non-English speaking countries were trained to use it since the trainings for its use are only in English. And the assessors themselves had various degrees of experience, with only 3 countries out of the 9 in total having a level of expertise.

Limitations with the Participant Pool

The children varied greatly in terms of functioning, particularly around expressive language. It is possible that baseline symptom level was related to treatment response. For example,it is possible that children with intact speech or higher levels of cognitive functioning responded differently to the intervention than children with limited speech, or limited cognitive functioning.  Subgroup analyses are essential in investigating response to autism treatments, given the heterogeneity of the population. One-size-fits all approaches are particularly limited in autism spectrum research.

Limitations creating Clinical Intervention Protocol

It is notoriously difficult to create a protocol for improvisational music therapy. It is exponentially more difficult with children on the spectrum. Improvising allows the therapist to vary every element of music--tempo, dynamics, articulation, timbre, melody, harmony, accompaniment pattern, register, etc.-- at any moment. It is not possible to standardize or control for all of these variables. Though the clinical interventions of the therapist were described in this study (synchronizing, mirroring, grounding), there was no discussion of how these interventions were created musically. There may have been a wide disparity in terms of how therapists musically created the interventions they intended. No information about the actual music that was created is included in the study. There was no notation, no discussion of specific musical elements. Qualities inherent in the music created by therapists are a vital component in attuning to clients and establishing therapeutic rapport. It is impossible to tell from this study the quality of the music created by the therapist and how effective the musical experience was in engaging the participants.

Another challenge in creating a protocol is that each person with autism comes to sessions with unique strengths and challenges. When you’ve met one person with autism, you’ve met one person with autism” speaks to the difficulty in generalizing and creating the specific protocols necessary for outcome studies such as the one under discussion. Each participant comes to therapy with their own unique relationship to music, and this needs to be considered when creating protocols to allow for flexibility in the improvisational interventions. Improvising allows the therapist to have choices and respond sensitively and flexibly. The more specific the protocol is, the more likely that the clinical benefits of improvisation dissipate.

Limitations Implementing the Intervention

Clearly it was a tremendous undertaking to include participants, therapists, assessors and researchers from 9 different countries and the authors of this study need to be congratulated for this achievement. But we know little about the musical skill, improvisational abilities or music therapy training of the therapists who implemented the study. For example, only 4 of the 29 music therapists in the study were recognized by the author as Nordoff-Robbins trained. The ability to listen and respond musically with sensitivity and clinical intention is a specialized skill set. It takes time and experience for music therapists to learn to improvise effectively in sessions. The therapists who participated in the study varied greatly in terms of years of experience and training. Treatment is a “black box” in this study. No information was given regarding the quality of the music making or improvisational abilities of the therapists, nor specific training in improvisation, the amount of time they had already been utilizing improvisation in their practice, previous experience working with autistic children, or what kind of supervisory supports they were receiving before becoming part of the treatment team.

There was no discussion regarding cultural differences in how music sounds and is played in each country despite the wide variety of cultures of the clients and therapists who participated in the study.

Another problem related to controlling the intervention was the different numbers of sessions participants received, and that participants missed. It appears as if half of the scheduled music therapy sessions were not given. In fact there was no discussion at all of the impact of missed sessions as a clinical variable

The authors do acknowledge that the lack of consistency and implementation between different music therapists in different locations might have had an effect on the overall measurement of the music therapy intervention. It also acknowledges that the total length of the study (5 months) as probably being too short.

The problem of testing a low-quality behavioral intervention in a large well designed outcome study which is not then followed by second generation studies can have major negative impacts on a field.  This was the case in the ADHD literature, when the Multimodal Study tested stimulant medication against a weak behavioral intervention. (Multimodal Treatment of ADHD (MTA) study. The main findings from this study were published in December 1999).  Because stimulant medication outperformed the limited behavioral intervention, and because second generation large clinical trial studies did not follow, the field was influenced for decades in believing medication was a superior treatment.  It is vital that second generation studies are fielded to provide a more in-depth perspective on high-quality interventions.

Limitations in outcome measures

Symptom reduction may not be a sensitive or appropriate outcome measure. Indeed, the authors of this study acknowledged that functional abilities are more important than reduction of symptom severity and that this study only addresses the latter. In our view it was a serious omission that functional gains and improved quality of life were not measured.  

It is important to note that the goal of reducing symptoms is considered by autism advocates to be an antiquated approach to helping individuals with autism live more satisfying lives. Autistic people (Silberman 2015) do not want to be stopped when flapping or forced to look at someone's eyes rather than look away. Seen through the lens of neurodiversity, these actions are helpful and not to be reduced or extinguished.  ["Autistic people" is the preferred term articulated by those in the Neurodiversity movement. See Silberman, “NeuroTribes-The Legacy of Autism and the Future of Neurodiversity”]

Rather than looking solely to reduce symptoms, music therapists work to identify and build on strengths. By listening closely to the often subtle musical expression that an autistic person brings to the improvisational encounter, and creating a musical path for further development, music therapists are helping to support new abilities and a healthier way for autistic people to be in relation with others. The brief mention of the findings of the qualitative report suggests that this in fact may have occurred.

Limitations in process measures

From our point of view, this may be the most problematic element of the study under discussion. Engagement is a key ingredient for the success of any therapy. Research shows that the more engaged a client is, the more benefits he or she is likely to achieve (Ardito and Rabellino 2011, Luborsky 1985). With a music-centered intervention like improvisation, a participant needs to be engaged in the music making process for the intervention to be successful. We know from clinical experience that increased musical engagement results in increased attention, awareness, responsiveness, organization, and flexibility in music therapy participants.

Yet this study did not measure musical engagement. It did not determine how many of the participants became significantly more engaged with music making in the sessions. We do not know how clients became more engaged. We do not know how much their level of engagement increased. It could be that some participants did become more engaged and some didn’t.  Gains by those who did become more engaged and benefitted from the intervention may not have been detected when combined in the overall analysis with those that did not. Without looking at engagement, the researchers undertaking this study were, in our view, taking a black box approach to the intervention that they were investigating.  

To summarize, the most serious problems in the study are the following. Despite its scope and strong experimental design, we don’t know anything about the quality of the musical and improvisational skill of the therapists, the quality of the music making interventions of the therapists, or the therapeutic process-- the increased engagement in music making of the clients.

Third: How we can Build on this Study

Two primary directions for future research emerge from these findings-one on outcome measures and the other being a need for music therapy process measures.

Outcome measures in future studies should focus on gains in strengths, increased functioning, and increased quality of life. Baseline symptom levels may be a moderator of outcomes, but not a sensitive outcome measure itself. Therapeutic process must be measured so that those who are making gains within treatment can be contrasted on outcomes in their daily lives with those who did not show gains in the therapy process.

We believe increased engagement with music making within improvisational music therapy is the most promising avenue for assessing the process of treatment. Musical engagement is a primary event that takes place during improvisation and one that holds the key to understanding the benefits of improvisation.

To build on this study and answer the questions that will move the research process forward, future research is needed to investigate:

  • What gains in engagement typically occur in high quality improvisational music therapy?

  • What gains in functioning and quality of life are associated with increased engagement with music making?

  • What factors promote engagement?

For a client to be engaged in the improvisational music making process, the structure and quality of the music is extremely important. The quality of the music needs to correlate to the emotional and energetic qualities of the client. The ability of the therapist to listen for cues from the client and create an aesthetically formed improvisation that invites participation is essential. Future studies need to include more detail on the actual music created in session. Understanding more about the music itself will help to improve the intervention protocol for subsequent trials.   

Empirical evidence from years of clinical observation indicates that increased engagement in the music making process leads to increases in flexibility, responsiveness, attention, awareness, and organization. These are essential to improving social interaction and communication. We are currently developing a research tool that grows out of our long standing examination of communication and social interaction in music therapy sessions. The Music Engagement Scale is designed to accurately and simply measure how a participant’s level of engagement increases during a music therapy session that includes improvisation as an intervention.  A sensitive engagement measure will also allow for evaluating therapeutic effectiveness across different schools/approaches to music therapy.  Increased engagement, regardless of approach (e.g., Nordoff-Robbins, Analytic, Benenzon, behavioral) can serve as “common denominator” of therapeutic quality and effectiveness across the field.

From our perspective, we want to know which groups of children on the spectrum increase their social connection with the primary people in their lives -- family members, teachers, peers -- when they increase their engagement with music during treatment in Nordoff-Robbinsmusic therapy. This is the next step that needs to be researched, and we need to construct the best method to measure this.


Paul Nordoff and Clive Robbins are recognized in the field of music therapy as pioneers of the clinical intervention of improvisation (Nordoff and Robbins 2007). Less well known is that they maintained a scientist-practitioner model, focusing on  measuring gains occurring in sessions as well as innovative clinical techniques. Broadly speaking, they focused on the question “How are the therapist and client engaged in music making?” In their holistic case studies they asked and found a variety of answers to the question “How does increased musical engagement lead to overall improvement in the child’s life?” These early questions and findings of Nordoff and Robbins continue to be relevant for our current research at the Center.

To build on the Nordoff-Robbins clinical and research tradition, and to address gaps in the current study under discussion, we need to:

  1. Effectively measure engagement with music-making in Nordoff-Robbins Music Therapy

  2. Answer the question of what gains in engagement with music making typically occur in effective improvisational music therapy

  3. Identify what conditions promote increased engagement

  4. Identify improvements in function and quality of life that are associated with increased engagement with music making in improvisational music therapy

- Dr. Alan Turry, Managing Director


Ardito, R. & Rabellino, D. (2011). Therapeutic alliance and outcome of psychotherapy: Historical excurcus, measurements, and prospects for research. Frontiers of Psychology, 270(2).

Bieleninik, L., Geretsegger, M., Mosler, K., et al., (2017). Effects of improvisational music therapy vs. enhanced standard care on symptom severity among children with autism spectrum disorder: The TIME-A randomized clinical trial. Journal of the American Medical Association, 318(6), 525-535.

Eysenck, H.J. (1952). The effects of psychotherapy: An evaluation. Journal of Consulting Psychology, 16, 319-324.

Luborsky, L., McLennan, A.T., et al (1985). Therapist success and its determinants. Archives of General Psychiatry, 42(6), 602-611.

Nordoff, P. & Robbins, C. (2007). Creative music therapy: A guide to fostering clinical musicianship. Gilsum, NH: Barcelona Publishers.

Payakachat, N., Tilford, J., Kovacs, E, & Kuhlthau,K. (2012). Autism spectrum disorders: A review of measures for clinical, health services and cost-effective applications. Expert Review of Pharmacoeconomics & Outcomes Research, 12(4), 485-503.

Pelham, W.E., (1999).  The NIMH multimodal treatment study for attention-deficit hyperactivity disorder: Just say yes to drugs alone? Canadian Journal of Psychiatry, 44(10), 981-990.

Silberman, S. (2015). NeuroTribes: The legacy of autism and the future of neurodiversity. New York, NY: Penguin Random House LLC.

Wilson, P. (2017). Why didn’t music therapy help autistic kids?  Maybe the researchers failed, not the therapy. MedPage Today.  Retreived from: