Music therapy for improving mental health in offenders: protocol for a systematic review and meta-analysis

Abstract

This is a protocol for a systematic review of the effects of music therapy on
offenders. Based on randomised controlled trials, the review aims to assess the
effectiveness of music therapy on adolescent and adult offenders in custodial
institutions including forensic psychiatric hospitals, and offenders or
probationers in the community. The outcomes to be evaluated include alleviated
symptoms of mental illness, psychosocial competencies and reduced recidivism.

Full text

!
1!
Music therapy for improving mental health in offenders: protocol for a
systematic review and meta-analysis
XiJing Chen1,Helen Leith1,Leif Edvard Aarø2,Terje Manger3,Christian Gold4
1. Department of Communication and Psychology, Aalborg University, Aalborg, Denmark.
2. Division of Mental Health, Norwegian Institute of Public Health, Oslo/Bergen, Norway.
3. Department of Psychosocial Science, University of Bergen, Norway.
4. Grieg Academy Music Therapy Research Centre (GAMUT), Uni Health, Uni Research,
Bergen, Norway
Contact address: Christian Gold, GAMUT – The Grieg Academy Music Therapy Research
Centre, Uni Research Health, Uni Research, Lars Hilles gate 3, Bergen, 5015, Norway.
christian.gold@uni.no

!
2!
Abstract
This is a protocol for a systematic review of the effects of music therapy on offenders. Based
on randomised controlled trials, the review aims to assess the effectiveness of music therapy
on adolescent and adult offenders in custodial institutions including forensic psychiatric
hospitals, and offenders or probationers in the community. The outcomes to be evaluated
include alleviated symptoms of mental illness, psychosocial competencies and reduced
recidivism.
Note: This protocol was developed with and for the Cochrane Collaboration in 2012-2013,
but was not published in the Cochrane Database of Systematic Reviews. Co-author H.L. died
in 2014 and the remaining authors would like to acknowledge her contribution. She did
approve the current version of the text in 2013.

!
3!
BACKGROUND
Description of the Condition
An offender is a person who has been convicted of a criminal offence or who has been
adjudged to have committed an act of juvenile delinquency. More than 10.65 million
offenders were held in penal institutions throughout the world in 2008 (Walmsley 2009).
Criminal justice systems usually deal with juvenile offenders differently to adult offenders,
and the age of criminal responsibility varies greatly. The United Nations (UN) considers
criminal responsibility below the age of 12 years unacceptable, but not all countries adhere to
that and the age of criminal responsibility ranges from six to 18 years across countries
(Prison Reform Trust 2011, p34).
The prevalence of mental health problems is high in the offending population (James
2006; WHO 2007). Although a small proportion of offenders with a diagnosed mental illness
are treatedin specialist forensic psychiatric hospitals, most offenders with mental health
problems are serving custodial or community sentences. Many such mental health problems
will be either undiagnosed or will not fully meet the criteria of a mental disorder.
However,high prevalence rates of mental disorders are also likely to signify a high
prevalence of other mental health problems. For example, up to 90% of offenders in prison in
the UK (95% in the case of young offenders) have a diagnosis of mental illness or a drug
and/oralcohol dependency-related mental or behavioural disorder (Prison Reform Trust 2011).
In China, 49% to 71% of male offenders have a personality disorder (Wang 2007). In the US,
56% of offenders in prison have symptoms of mental disorder or a history of treatment for
mental disorder, 43% have symptoms or a history of mania, 23% have a history of major
depression and 15% have experienced psychotic symptoms such as hallucinations or delu-
sions (James 2006). They often show paranoia, suicidal ideation, self-isolating and
self-harming behaviour (Frühwald 2005). Many prisoners also have symptoms that resemble
the negative symptoms of schizophrenia -blunted affect, low motivation and poor social
relationships – which makes it difficult for them to engage in prison rehabilitation
programmes (Ward 2007; Leith 2011). Low self-esteem, poor impulse control and related
behavioural problems, as well as a limited ability to resolve conflicts constructively, are
prominent examples of mental health-related problems common among prisoners. In addition,
low empathy is strongly related to aggressive and offending behaviour (Jolliffe 2004). Mental
health problems are a significant issue for a majority of offenders and therefore something

!
4!
that needs to be addressed alongside (or in some instances instead of) offending behaviour
(Priebe 2008).
Psychological characteristics of offenders vary according to age, gender and crime type.
Young offenders are more likely than adults to have mental health problems and suicidal
tendencies (Prison Reform Trust 2011). Female offenders tend to have more mental health
problems than male offenders and are also more likely to have experienced emotional,
physical, sexual or financial abuse (or a combination) (James 2006). Juvenile offenders tend
to have attention and hyperactivity problems, academic difficulties, oppositional behaviour,
peer relationship and social skills deficits, cognitive and attributional deficiencies, anger
management problems and impulsivity (Wyatt 2002). In terms of crime type, sexual of-
fenders seem to be more likely to have poor social skills and low self-esteem than other
offenders (Worling 2001).
Description of the Intervention
Music therapy is commonly defined as a “systematic process of intervention wherein the
therapist helps the client to improve health, using music experiences and the relationships
that develop through them as dynamic forces of change” (Bruscia 1998). This definition
encompasses all theoretical models of music therapy and includes both treatment and
prevention/health promotion. However, it does not include music interventions at
an ’auxiliary level’ (Bruscia 1998), such as music listening without the presence of a
therapist; such interventions are sometimes referred to as music medicine (Gold 2009; Gold
2011a). Music therapists are specifically trained to intervene within the music (Gold 2007),
and it is therefore seen as important that the intervention is administered by an appropriately
credentialed music therapist to ensure the quality of the intervention.
The music experiences offered in music therapy often include active music making, either
as spontaneous creative expression (for example, free or structured improvisation; Albornoz
2011), as a compositional process (for example, songwriting; Edgerton 1990; Baker 2011) or
as reproduction of existing musical material (for example, singing; Clark 2012) in individual
or group settings. However, listening to music can also be a central modality and often serves
as an anchor point for verbal reflection (Blom 2011). The method of working, as well as the
level of structure and the degree of emphasis on the music itself versus discussion of personal
issues, may vary. These are typically adapted in response to the client’s needs and wishes, but
have also been an issue of debate among music therapists (Mössler 2011a). Music therapy

!
5!
has a long history of application in mental health (Gold 2009; Mössler 2011b) and also in
correctional institutions, where its use dates back to at least the 1930s (Codding 2002).
Various authors have described the use of music therapy methods such as improvisation
(Hoskyns 1988), songs and metaphoric imagery (Chambers 2008), creating and performing
music (O’Grady 2009) and music relaxation (Thaut 1989a) in music therapy with offenders,
either in individual or small group settings (Rio 2002).
Goals of music therapy with offenders are usually broad. According to a survey among
49 American music therapists, the following are typical: to provide “a non-threatening,
motivating reality focus for use of leisure time and release of energy” (100%), to promote
“self-esteem” (94%), “self-control” (91%), “appropriate release of tension, stress and
anxiety” and “coping skills” (91%); 32 further goals were listed by Codding 2002. Some
researchers have suggested that goals of interventions for offenders should be categorised
according to whether or not they are risk or protective factors for future criminal behaviour
(Bonta 2007). Examples of risk factors are antisocial personality patterns, pro-criminal
attitudes, substance abuse, poor family relationships, poor behavioural control; examples of
factors that help people to move away from crime (protective factors) are positive social
orientation, intolerant attitude towards deviance, a flexible personality, good coping skills and
self-calming ability (Bonta 2007). A dynamic risk factor is a risk factor that can change.
Negative emotions and thoughts, low coping skills, lack of anger management ability, lack of
empathy and impulsivity may be examples of dynamic risk factors. A qualitative study that
collected the views of creative arts therapists working with offenders found music therapy to
be focusing on positive developmental goals such as coping skills and emotion regulation
(Smeijsters 2011).
However, the link between mental health issues and criminal risk is not always entirely
clear. There are also numerous reports of music therapy focusing on other goals that are
relevant for the prisoner’s health but may or may not be related to the risk of future criminal
behaviour, such as emotional expression and assertive behaviour (Cohen 1987), mood,
relaxation and insight (Thaut 1989a) or decision-making style (Moss 2004). A certain tension
between goals related to mental health and goals related to criminal risk may be due to the
double nature of therapies for offenders. Many studies of music therapy for offenders
explicitly targeted a population who also had a mental health problem (for example, Thaut
1989a; Codding 2002); even where this was not the case, difficult life histories and traumatic

!
6!
experiences were almost ubiquitous and an important focus of therapeutic work (for example,
O’Grady 2009).
How the Intervention Might Work
Music therapy is a complex intervention. Explaining mechanisms of change in complex
interventions is often not straightforward because of multiple components interacting with
each other, and because the context influences the outcome (Craig 2008, Gold 2012).
Although a model has been suggested for explaining change in externalising behaviour
problems through cognitive-behavioural music therapy (Hakvoort 2013), a more general
model of change for music therapy with offenders has not been proposed previously.
Figure 1 shows an attempt to explain the pathways in which music therapy might work
for offenders. Starting from the definition cited above (Bruscia 1998), the two basic aspects
of music therapy are interpersonal interaction and music experiences (first column in the
figure). Both aspects are connected to each other and cannot easily be separated, although the
extent to which either the music or the interaction aspects is the main agent might vary. Basic
guiding principles (above the first column) include analogy, metaphor, and aesthetics.
Analogy and metaphor are two closely related concepts that basically state that the way
people express themselves in music is related to the way they act in other situations. Thus,
the way people act in general, or the repertoire they have for interacting with others, may be
changed or expanded by changing the way they interact in music (Smeijsters 2012). Analogy
is a more elementary concept than metaphor and is related to very early relational learning
(Stern 2010). In addition, aesthetic experience (where ’aesthetic’ is understood much more
broadly than to be ’pleasing’ in a narrow sense) is seen as a basic part of the human condition
(Aigen 2007). This might explain why music therapy is often found to be motivating for
clients who are not easily motivated for other therapies. Motivation is therefore an
overarching principle towards achieving direct outcomes (second column). Interpersonal
interaction through music has -by means of the basic principles described -a direct and
immediate quality that lends itself well to learning very basic communicative abilities (Stern
2010), to experience and be able to give social support (Procter 2011), and to improve the
ability to understand and share the feelings of others, i.e. empathy. It is proposed that
improvement in mental health outcomes (third column) occurs primarily through those direct
outcomes. Both externalising (e.g., aggressive or impulsive behaviour) and internalising
problems (e.g. anxiety, depression) may be influenced. These are separate but connected

!
7!
(i.e. correlated and interacting with each other). Finally, recidivism and quality of life are
downstream outcomes that might be influenced indirectly by means of reduced externalising
and internalising mental health problems.
Figure 1. Conceptual model of music therapy for offenders
In summary, using music as a medium in therapy might be beneficial because:
Music is a “language of emotions” (Juslin 2010) and may help people develop the ability
to perceive, express and change emotions (e.g. Erkkilä 2011; Mössler 2011b; Gold 2013).
For offenders who have difficulty experiencing and identifying emotions, not necessarily
with a diagnosed disorder, music therapy can offer a safe and contained space in which the
client is supported in identifying and exploring emotional states as well as learning to express
and regulate them in a positive way (Loth 1994; Loth 1996). Strong and potentially
destructive emotions that might otherwise be expressed through self-harming or challenging
behaviour can find positive and appropriate expression in music (Leith 2011).
Music-making is also highly social in nature (Stern 2010, Procter 2011). Making music
together may therefore be an effective way of improving the ability to build and sustain
social relationships, to communicate with and to relate to others. Because music is social,

!
8!
music therapy may enhance offenders’ listening and sharing skills. It may foster flexibility
and sensitivity to others within the musical relationship with the therapist and, in group
therapy, with other group members. These interpersonal relationships, even attachments
developed in music experiences, have significant therapeutic meanings, especially for those
who were lacking these experiences in the past (Rio 2002). It also creates a bridge for the
offenders to link their life inside of prison and outside realities (O’Grady 2011, Tuastad
2013). It may strengthen an awareness of self in relation to others and a client’s confidence in
their own ability to build relationships and to find positive ways of making their needs known
(Leith 2011).
Music is a non-verbal medium. It conveys meanings, but not in the same way as verbal
language. This makes it motivating for clients who are unmotivated for verbal therapies
(Gold 2013). In a setting where verbal self-disclosure can be detrimental for the individual,
purely verbal therapy may be of limited value. Music therapy may be valuable for clients
who respond best to action-oriented forms of expression (Nolan 1983). It may create a
healthy and non-threatening environment in which offenders can overcome anxiety and
resistance, increase motivational ties to reality and gain insight through a learning process
that is perceived as meaningful. Music also provides multimodal experiences, such as
feelings, internal images and body sensations (Skaggs 1997).
The current literature does not suggest any differences in mechanisms of change between
age groups, genders, or types of offences. Therefore it appears justified to combine them in
one review. If future research should suggest substantial differences between those groups,
this will be reconsidered.
Why It Is Important to Do this Review
Both the application of and research on music therapy for offenders seem to be increasing.
Numerous case reports and some outcome studies have suggested beneficial effects, but no
systematic review of effectiveness has been conducted yet. The present review is therefore
necessary to inform policy-makers and practitioners, as well as to guide future research in the
field.
OBJECTIVES
To assess the effectiveness of music therapy on adolescent and adult offenders in custodial
institutions including forensic psychiatric hospitals, and offenders or probationers in the

!
9!
community. The outcomes to be evaluated include alleviated symptoms of mental illness,
psychosocial competencies and reduced recidivism.
METHODS
Criteria for Considering Studies for this Review
Types of studies
Randomised controlled trials (RCTs), including quasi-randomised controlled trials (for
example, allocation by the person’s date of birth, by the day of the week or month of the year,
by a person’s medical record number or alternate allocation). Studies will be considered
regardless of sample size. It is likely that any existing studies in this field will be small and
underpowered pilot studies, but such studies may still contain important information for
building larger studies in the future (Gold 2004b). The limited power of small studies will be
considered in the interpretation.
Types of participants
Adult and juvenile offenders irrespective of offence, gender or nationality, with or without
formally diagnosed mental illness, in any institutional setting (for example, forensic
psychiatry, prison, a correctional institution) or in the community. The starting age for
criminal responsibility varies considerably across countries (Prison Reform Trust 2011, p.
34), but is 14 in many countries (Hazel 2008). We therefore aim to include people from 14
years upwards if in a community setting. Participants in institutional settings will be included
if the institution serves adult or juvenile offenders as defined in the place where the study was
conducted.
Types of interventions
1. Music therapy (either alone or in addition to a standard care that would also be provided in
the comparison condition). Music therapy is defined as “a systematic process of interventions
wherein the therapist helps the client to promote health, using musical experiences and the
relationships that develop through them as dynamic forces of change” (Bruscia 1998). The
methods of music therapy include improvisation, re-creating, composing, listening and
variations of them (Bruscia 1998). Music therapy as offered in the trials must conform to this
definition; in particular it must include music experiences as well as a relationship with an
appropriately credentialed music therapist.
2. Standard care, no treatment or any kind of ’placebo’ therapy. The term placebo therapy

!
10!
does not refer to a medication. Instead, it is a kind of pseudo therapy aimed at controlling for
the therapy’s unspecific effects, for example by constructing an intervention that is similar to
the active interventions but supposedly does not contain its active ingredients. Examples in
music therapy can be found (Gold 2006) but their relevance is controversial (Gold 2011b).
See also Wampold (2001) for a further discussion of ’placebo’ therapies.
Types of outcome measures
General considerations
All primary outcomes are marked with an asterisk (*), indicating that we will include them in
a ’Summary of findings’ table. If there are several time points for the same outcome from the
same study, we will categorise them as short term (up to 12 weeks), medium term (13 to 26
weeks) or long term (more than 26 weeks after randomisation), as in a previous review
(Mössler 2011b). When outcomes are reported on a scale, they should either be based on an
independent observer (that is, not the music therapist) or on self-reports. Scales should
preferably be standardised and published with known reliability and validity.
Primary outcomes
Primary outcomes are those that have the greatest importance. They must be relevant for the
participants and related to the intervention’s goals. The following primary outcomes are
based on the most common goals of music therapy for offenders (Codding 2002; Smeijsters
2011) as well as on the proposed model of change (Figure 1 and How the intervention might
work).
1. Self-concept/self-esteem/self-efficacy *
Almost all music therapists working in correctional settings (94%, Codding 2002) endorse
self-esteem as an important goal. Related constructs are self-concept and self-efficacy. The
three constructs are often used interchangeably and are therefore considered together in this
review. Self-esteem emphasises more the affective components of judging one’s self-worth,
whereas self-concept and self-efficacy emphasize more the cognitive aspects. Sources dis-
agree on whether self-esteem is a part of self-concept or vice versa (Judge 2001). Low
self-esteem may contribute to externalising behaviour problems and delinquency, but also to
internalising problems such as depression (Baumeister 2003). However, self-efficacy might
be a clearer predictor for behavioural outcomes than self-esteem (Bandura 1997). This
outcome is related to participants’ sense of identity, which is often impaired or not fully
developed. We consider it therefore as a relevant outcome in its own right, not only as a

!
11!
predictor of downstream outcomes. However, ceasing to see oneself as an offender and
developing a more positive identity may also be linked to desistance from crime, especially
for people with deeply entrenched identities as offenders (Maruna 2003). This outcome is
subjective in nature and will normally be measured on a self-report scale as a continuous
variable.
2. Behaviour management *
This includes ’self-control’ and appropriate management of aggressive or self-harming
behaviour and impulsivity. This outcome was described as important by 91% of music
therapists (Codding 2002), and some music therapy approaches focus primarily on this
outcome (Hakvoort 2013). It is an objective, observable outcome of immediate relevance to
prison management and very likely relevant for criminal risk, as the offence itself is a deviant
behaviour. It may be measured as continuous or count data, based on independent observer
reports or possibly through self-reports.
3. Anxiety *
Among mental health-related outcomes, anxiety seems to be an important. “Appropriate
release of tension, stress and anxiety” has been endorsed as a relevant goal by 91% of music
therapists (Codding 2002). Along with depression, it is one of the most important
internalising behaviour problems. Both anxiety and depression might be considered as
primary outcomes because they are closely related. However, given that the number of
primary outcomes should be limited, we made a choice to prioritize anxiety as it seems to be
more often endorsed as an outcome and possibly also more prevalent among offenders.
4. Empathy *
Empathy is defined as “the ability to understand and share the feelings of another” (Oxford
2010). It is a direct outcome of music therapy because to work with music experiences is to
work with the ability to understand and share feelings. Many offenders, specifically those
with antisocial personality traits, have insufficiently developed empathy. To “develop trust
and empathy” is a relevant goal for 84% of music therapists (Codding 2002).
5. Any adverse event *
Little is known about the potential adverse effects of music therapy (Edwards 2011). No
specific adverse effects of music therapy have been described in the literature. Therefore, any
adverse events will be reported here.

!
12!
Secondary outcomes
Secondary outcomes concern additional effects of music therapy that may be relevant only
for some offenders or that are relevant but not directly targeted by music therapy.
1. Depression
Many offenders suffer from negative mood states other than anxiety, such as symptoms of
depression. “To reduce the number and length of depressive episodes” was endorsed as a
relevant goal by 56% of music therapists (Codding 2002). Music therapy may be well-suited
to address depressive symptoms (Erkkilä 2011). Together with anxiety, it is an important
internalising problem.
2. Quality of life
Quality of life is a general goal that applies to almost any health intervention for most
conditions. Improved quality of life is also likely associated with reduced criminal risk (Ward
2004). Many aspects of offenders’ quality of life may be beyond the areas that music therapy
can address, and it does not seem to be among the most common goals of music therapists in
the field (Codding 2002). It is however a downstream outcome that may result from
improved mental health. In some contexts it may be controversial whether quality of life of
offenders should be improved at all.
3. Substance abuse
Substance abuse is an internalising mental health problem. It is often considered together
with anxiety and depression (Albornoz 2011). Many offenders suffer from it, and it also
contributes to offending behaviour.
4. Recidivism
To reduce future criminal activity should be an ultimate goal of any intervention for
offenders, including music therapy. However, this outcome will also be influenced by many
external factors. Therefore recidivism is considered as a downstream effect, by means of
direct and mental health outcomes (as shown in Figure 1), rather than as a direct outcome.
Behaviour management (listed under primary outcomes) may be considered an early
indicator of this outcome.
Search Methods for Identification of Studies
Electronic searches
We will search the following databases, with no date or language restrictions: Cochrane

!
13!
Central Register of Controlled Trials (CENTRAL), part of The Cochrane Library; MEDLINE;
EMBASE; PsycINFO; LILACS; CINAHL; ERIC; Sociological Abstracts; International
Bibliography of Social Sciences; National Criminal Justice Reference Service Abstracts;
RILM Abstracts of Music Literature; Social Science Citation Index; SCOPUS; Conference
Proceedings Citation Index -Social Science & Humanitee; WorldCat (theses search); Rutgers
School of Law Grey Literature Database; ClinicalTrials.gov; ICTRP; metaRegister of
Controlled Trials.
We will search MEDLINE using the following search strategy, which will be adapted for
the other databases using appropriate syntax and controlled vocabulary. The strategy will be
simplified for databases and websites that do not support complex search strings. Language
and date limits will not be applied and we will not use a study methods filter. We will seek
translations of papers when necessary. 1. music therapy/ 2. music$.tw. 3. (guided imagery
adj3 music).tw. 4. BMGIM or GIM.tw. 5. (vibro-acoustic$ or vibroacoustic$).tw. 6. music/ 7.
(sing or singing or song$ or choral$ or choir$).tw. 8. (percussion$ or rhythm$ or tempo).tw.
9.melod$.tw. 10. improvis$.tw. 11. (Nordoff-Robbin$ or Bonny$).tw. 12. ((auditory or
acoustic or sound$) adj5 (stimulat$ or cue$)).tw. 13. ((play$ or learn$) adj3 instrument$).tw.
14. or/1-13 15. prisoners/ 16. prison/ 17. juvenile delinquency/ 18. (borstal$ or convict$ or
correctional$ or criminal$ or custody OR custodial OR delinquen$ or detain$ or
detention$ or gaol$ or imprison$ or incarcerat$ or inmate$ or in-mate$ OR jail$ or
offenc$ OR offens$ OR offender$ or penal$ or penitentiar$ or prison$ or probation$ or
recidivid$ OR reformatory or (reform adj school$) or (secure adj accommodation)).tw. 19.
residential treatment/ 20. psychiatric hospitals/ 21. (forensic adj3 (hospital$ or patient$)).tw.
22. or/15-21 23 14 and 22.
Searching other resources
Grey literature
We will search the websites of relevant professional and research organisations to retrieve
grey literature, including the World Federation of Music Therapy, the American Music
Therapy Association, the European Music Therapy Confederation, the Chinese Music
Therapy Association, the Hong Kong Music Therapy Association, the Japanese Music
Therapy Association, the Korean Music Therapy Association, the Canadian Association for
Music Therapy, the Australian Music Therapy Association, the Music Therapy Association
of Taiwan, the British Association for Music Therapy and Music in Prisons

!
14!
(www.musicinprisons.org.uk). We will also search the criminal justice grey literature
database of Rutgers School of Law (law-library.rutgers.edu/cj/gray/).
Handsearching
We regularly monitor the content of following journals and will continue to do so (XJC: all
issues of the Journal of Music Therapy since the journal started in 1964; HL: British Journal
of Music Therapy since 1980; Australian Journal of Music Therapy since 2006; CG: Nordic
Journal of Music Therapy since the journal started in 1992, Musiktherapeutische Umschau
since 1998). Additionally, extensive handsearching of music therapy journals was performed
earlier for related reviews (Gold 2004a; Maratos 2008; Gold 2009; Mössler 2011b) and any
relevant records retrieved there will be re-used.
Correspondence
We will attempt to contact the authors of relevant studies for additional or missed studies and
for further resources and information.
Reference lists
Any potentially relevant studies in the reference lists of studies retrieved by the searches will
be followed up.
Data Collection and Analysis
Selection of studies
Duplicate records of the same report will be removed using reference management software.
Two review authors (XJC and HL) will independently examine titles and abstracts and
exclude those reports that are clearly not randomised studies, not about offenders or not about
music therapy. Two review authors (XJC and HL) will then independently examine the full
text of potentially relevant reports and decide which studies meet the eligibility criteria. If
disagreement occurs, all three review authors will discuss to reach a resolution. Where
information on an eligible study is missing, we will attempt to contact the trial authors to
obtain that information.
Data extraction and management
Two review authors (XJC and HL) will independently extract data for each trial using a data
extraction form to collect information about the population, interventions, randomisation
method, blinding, sample size, outcome measures, follow-up duration, attrition and handling
of missing data, and methods of analysis.

!
15!
Assessment of risk of bias in included studies
We will use The Cochrane Collaboration’s tool for assessing risk of bias (Higgins 2011).
Two review authors (XJC and HL) will independently assess the risk of bias within each
included study based on the following domains, with a review author’s judgments presented
as answers of ’high’, ’low’ or ’unclear’ risk of bias. Any disagreements will be resolved by
discussion and disagreements will be arbitrated by a third review author (CG). The tool will
be used to assess the following domains: randomisation, allocation concealment, blinding of
outcome assessment, incomplete outcome data, and selective outcome reporting and other
sources of bias.
Randomisation
Randomisation will receive the following judgments. ‘Low’ when participants were allocated
through a truly randomised sequence (such as computer-generated random numbers, a ran-
dom numbers table, coin-tossing). ‘Unclear’ when the randomisation method was not clearly
stated or unknown. ‘High’ when randomisation did not use an appropriate method of
sequence generation.
Allocation concealment
‘Low’ when allocation concealment was clearly stated in the study. ‘Unclear’ when
allocation concealment was not clearly stated or unknown. ‘High’ when allocation was not
concealed from either participants or researchers before informed consent, or from
researchers before decisions about inclusion were made.
Blinding
a) Blinding of assessors. For all outcomes that are not self-reports, we will examine whether
blinding of outcome assessors was attempted, and if and how success of blinding was verified.
Quality of blinding will receive the following judgments. ‘Low’ when assessors were blind to
the treatment conditions. ‘Unclear’ when blinding of assessors was not reported. ‘High’ when
assessors were not blind to treatment conditions.
b) Blinding of therapists and clients. We will also consider whether therapists and clients
were blinded to the intervention, even though we do not know of any method to ensure
blinding of therapists and clients in a psychosocial intervention (Gold 2011c).
Addressing incomplete outcome data
Assessment will take into account whether researchers used intention-to-treat analyses by
including measurements from all the participants, including those who did not participate

!
16!
fully in the treatment protocol. Those studies where the researchers did not use
intention-to-treat analyses and it is not possible to conduct them with the available data will
be identified. The adequacy of the way the authors of the trials dealt with missing data will
receive the following judgments. ‘Low’ when intention-to-treat analyses were used or can be
performed using available data. ‘Unclear’ when information about whether intention-to-treat
analyses were performed was not available and cannot be acquired by contacting the
researchers of the study. ‘High’ when intention-to-treat analyses were not performed and
cannot be done using available data. Balance of drop-outs and reasons for dropping out will
be explored. We will use sensitivity analysis to assess the impact of drop-outs.
Selective reporting
The likelihood that the authors of the trial omitted some of the collected data when presenting
the results will be determined and will receive the following judgments. ‘Low’ when all
collected data seem to be reported. ‘Unclear’ when it is not clear whether other data were
collected and not reported. ‘High’ when the data from some measures used in the trial are not
reported.
Other bias
Assessment will determine whether any other bias is present in the trial. In particular,
appropriate administration of the intervention (adequacy of music therapy method and
training) will be assessed as an important issue that may affect the results (Mössler 2011b).
Inappropriate administration of an intervention is one of the examples of potential biases
listed in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011,
Section 8.15.1.5). Baseline imbalance will also be assessed as a potential source of bias.
Measures of Treatment Effect
Dichotomous data
For dichotomous data, we will calculate an odds ratio (OR) with a 95% confidence interval
(CI) for each outcome in each trial (Higgins 2011). From our knowledge of the field, risk
ratios (RR) are rarely reported and if so, they are reported with the original cross table data,
making it possible to calculate any measure of association. Risk hazards are even more
uncommon. If any such measures should occur we will analyse them in line with the
Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011).
Continuous data

!
17!
When means and standard deviations (SD) are provided in the study, we will analyse
continuous data. Effect sizes (that is, standardised mean differences (SMD)) will be used
because they facilitate clinical interpretation (Cohen 1988; Gold 2004b) and because they
allow for combining results from different scales for the same outcome. (The type of effect
size used will be Hedges’ g, which is similar to Cohen’s d but with a small-sample bias
correction.)
Unit of Analysis Issues
Cluster randomisation
If cluster-randomised trials are included, we will examine whether the clustering was taken
into account in the analysis. If clustering has not been taken into account in the published
analysis, we will attempt to adjust the analysis using the design effect. For this adjustment,
we will ideally use an estimate of intra-class correlation from the same study, but where this
is not available, we will attempt to find and use an external estimate from a similar study
(Ukoumunne 1999).
Studies with multiple intervention groups
For studies where there are multiple intervention (or control) groups, we will merge groups
(if deemed similar enough) to avoid multiple comparisons. The data from the same group
will not be analysed twice in the same meta-analysis. A separate meta-analysis will be done
for each comparison.
Multiple time points
If outcomes were measured more than once in the same study, we will not combine different
time points in the same meta-analysis, but will use different meta-analyses to avoid unit of
analysis issues. If studies used an appropriate analysis that takes the dependence of
longitudinal data into account (for example, generalised estimating equations or linear mixed
models), such results will be reported narratively.
Multiple measures of the same outcome
If more than one measure was used to assess the same outcome, and these measures were
interchangeable (for example, both were standardised scales, had the same level of blinding
and were used at the same time point), we will only use the measure that was identified as
primary in the original study. Where this is not possible, we will attempt to make an informed
judgment as to which measure was likely to be intended as primary.

!
18!
Dealing with missing data
We will assess missing data in the included studies. The meta-analysis will be based on the
data of all original participants when it is possible. If a study reports outcomes only for
participants completing the trial, or only for participants who followed the protocol, efforts
will be made to contact study authors to request missing data. We will report the reasons and
types of missing data. If it is not possible to conduct an intention-to-treat analysis in some
trials, we will use sensitivity analysis to assess the potential bias introduced by those trials.
Assessment of heterogeneity
We will assess heterogeneity in the following areas: clinical diversity, methodological
diversity and statistical heterogeneity. We will assess the first two types of heterogeneity
prior to combining trials in a meta-analysis. We will assess statistical heterogeneity in a
meta-analysis using the I2 statistic, a descriptive measure that represents the percentage of
variability that is due to heterogeneity rather than sampling error or chance. The percentages
obtained using the I2 statistic will be interpreted according to the Cochrane Handbook for
Systematic Reviews of Interventions (Higgins 2011, Section 9.5.2). The Chi2 test will be
used in addition to assess the presence of heterogeneity. Because this test is often
underpowered when there are few studies, a P value smaller than 0.10 will be interpreted as
an indication of possible heterogeneity of intervention effects. We will conduct subgroup
analyses to explore the sources of heterogeneity. We have not defined a fixed cut-off value
for I2. As described in the Cochrane Handbook for Systematic Reviews of Interventions
(Higgins 2011, Section 9.5.2, p. 278), there cannot be one percentage at which heterogeneity
is “too high”. Rather, the importance of the observed value depends on the magnitude and
direction of effects as well as the strength of evidence for heterogeneity (P value or CI).
Assessment of reporting biases
For avoiding publication bias, we will make efforts to obtain and include data from
unpublished trials that meet the inclusion criteria. We will use a funnel plot to assess the
likelihood of publication bias in meta-analyses with enough trials. We will include a
triangular 95% confidence region based on a fixed-effect meta-analysis in the funnel plot, and
different plotting symbols will be used to identify different subgroups if applicable. Funnel
plots will be interpreted cautiously as recommended in the Cochrane Handbook for
Systematic Reviews of Interventions (Higgins 2011, section 10.4.5), and if asymmetries are
found we will examine clinical variation in the studies.

!
19!
Data synthesis
We will conduct a meta-analysis to combine data from studies using the same comparisons
and outcomes. Different comparisons and outcomes will be meta-analysed separately. There
are divergent views on the question of random versus fixed effects. Although random effects
are often seen as an appropriate solution when there is unexplained heterogeneity, the
Cochrane Handbook cautions that “a random-effects model does not ‘take account’ of the
heterogeneity in the sense that it is no longer an issue” (Higgins 2011, 9.5.4, p. 280). Some
have even argued that “combining heterogeneous studies using the random-effects model is a
mistake and leads to inconclusive meta-analyses” (Al Khalaf 2011). Fixed-effects models are
also more straightforward to interpret (Higgins 2011). However, it is not possible to resolve
this controversy as part of this review. We will therefore follow the advice given by the
editor and the statistical reviewer: We will, by default, apply both fixed and random effects,
and if the 95% CI of the random-effects analysis includes the 95% CI of the fixed-effect
analysis, report only the random-effect analysis.
Subgroup analysis and investigation of heterogeneity
We aim to limit subgroup analyses to the most important ones. We have selected age and
gender, but others (e.g. age of onset of offending behaviour) may also be relevant. Should we
find heterogeneity we will examine the following subgroups.
1. Young offenders versus adult offenders. Young offenders are an important subgroup
because they are more likely to have mental health problems. Although varying cut-offs have
been used in the literature, it seems to be most common to regard those under 21 years of age
as young offenders (Prison Reform Trust 2011, pp. 37-38; a cut-off of 25 years has also been
used in the literature) and we therefore aim to use this cut-off.
2. Male versus female offenders. Women are more likely than men to have mental health
problems, which often find expression in self-harming behaviour (James 2006). Women
prisoners in the UK commit around 50% of self-harm incidents although they represent only
5% of the total prison population (Smee 2009). The level of importance and nature of
association of criminogenic factors may also differ between genders (Blanchette 2001).
Sensitivity analysis
We will use a sensitivity analysis to investigate if trial characteristics such as unclear
methodology or missing data may have influenced the results of the analyses, by repeating
the analysis with the problematic studies removed. In particular, we will use sensitivity

!
20!
analyses to examine the influence of overall risk of bias (excluding studies with a high risk of
bias) and quality of the intervention (excluding studies that relied on inappropriate music
therapy methods or training level). Randomisation method and other bias (quality of the
intervention) are anticipated to be the most important sources of bias in this area.
ACKNOWLEDGEMENTS
James Tyler Carpenter and Laurien Hakvoort provided valuable feedback on an earlier
version of this protocol.

!
21!
References
Aigen K. (2007). In defense of beauty: A role for the aesthetic in music therapy theory. Part I.
Nordic Journal of Music Therapy, 16(2), 112–128.
Al Khalaf M. M., Thalib L., & Doi SAR. (2011). Combining heterogenous studies using the
random-effects model is a mistake and leads to inconclusive meta-analyses. Journal of
Clinical Epidemiology, 64(2), 119–123.
Albornoz, Y. (2011). The effects of group improvisational music therapy on depression in
adolescents and adults with substance abuse: a randomized controlled trial. Nordic Journal of
Music Therapy, 20(3), 208–224.
Baker F., & Krout R. E. (2011). Collaborative peer lyric writing during music therapy
training: a tool for facilitating students’ reflections about clinical practicum experiences.
Nordic Journal of Music Therapy, 20(1), 62–89.
Bandura, A. (1997). The exercise of control. New York, NY: Freeman.
Baumeister, R. F., Campbell, J. D., Krueger, J. I., & Vohs, K. D. (2003). Does high
self-esteem cause better performance, interpersonal success, happiness, or healthier lifestyles?
Psychological Science in the Public Interest, 4(1), 1–44.
Blanchette, K. (2001). Classifying Female Offenders for Effective Intervention: Application
of the Case-based Principles of Risk and Need [Comprehensive paper submitted in partial
fulfilment of PhD]. 205.193.117.157/text/ pblct/forum/e141/141h˙e.pdf. (accessed 3 October
2012). Ottawa, Canada: Carleton University.
Blom, K. M. (2011). Transpersonal-spiritual BMGIM experiences and the process of
surrender. Nordic Journal of Music Therapy, 20(2), 185–203.
Bonta J, & Andrews D. A. (2007). Risk-need-responsivity Model for Offender Assessment
and Rehabilitation. Public Safety Canada, 1–30.
Bruscia, K. E. (1998). Defining Music Therapy (2nd Edition). Gilsum, NH: Barcelona
Publishers.
Chambers, C. (2008). Song and Metaphoric Imagery in Forensic Music Therapy [PhD thesis].
Nottingham, UK: University of Nottingham.
Clark, I., & Harding, K. (2012). Psychosocial outcomes of active singing interventions for
therapeutic purposes: a systematic review of the literature. Nordic Journal of Music Therapy,
21 (1), 80–98.
Codding, P. (2002). A comprehensive survey of music therapists participating in correctional

!
22!
psychiatry: demographics, conditions of employment, service provision, assessment,
therapeutic objectives, and related values of the therapist. Music Therapy Perspectives, 20(1),
56–68.
Cohen, J. M. (1987). Music therapy with the over controlled offender: theory and practice.
Arts in Psychotherapy, 14, 215–221.
Cohen, J. (1988). Statistical Power Analysis for the Behavioral Sciences (2nd Edition).
Hillsdale, NJ: Lawrence Erlbaum.
Craig, P., Dieppe, P., Macintyre, S., Michie, S., Nazareth, I., & Petticrew, M. (2008).
Developing and evaluating complex interventions: the new Medical Research Council
guidance. BMJ, 337, a1655.
Edgerton, C. D. (1990). Creative group songwriting. Music Therapy Perspectives 8, 15–9.
Edwards, J. (2011). A music and health perspective on music’s perceived “goodness”. Nordic
Journal of Music Therapy, 20(1), 90–101.
Erkkilä, J., Punkanen, M., Fachner, J., Ala-Ruona, E., Pöntiö, I., Tervaniemi, M., et al.
(2011). Individual music therapy for depression: randomised controlled trial. British Journal
of Psychiatry, 199(2), 132–139.
Frühwald, S., & Frottier, P. (2005). Suicide in prison. The Lancet, 366, 1242–1243.
Gold, C., Voracek, M., & Wigram, T. (2004a). Effects of music therapy for children and
adolescents with psychopathology: a meta-analysis. Journal of Child Psychology and
Psychiatry and Allied Disciplines, 45(6), 1054–1063.
Gold, C. (2004b). The use of effect sizes in music therapy research. Music Therapy
Perspectives, 22(2), 91–95.
Gold, C., Wigram, T., & Elefant, C. (2006). Music therapy for autistic spectrum disorder.
Cochrane Database of Systematic Reviews, 2. [DOI: 10.1002/ 14651858.CD004381.pub2]
Gold, C., Wigram, T., & Voracek, M. (2007). Predictors of change in music therapy with
children and adolescents: The role of therapeutic techniques. Psychology and Psychotherapy:
Theory, Research and Practice, 80(4), 577–589.
Gold, C., Solli, H. P., Kruger, V., & Lie, S. A. (2009). Dose-response relationship in music
therapy for people with serious mental disorders: systematic review and meta-analysis.
Clinical Psychology Review, 29(3), 193–207.
Gold, C., Erkkilä, J., Bonde, L. O., Trondalen, G., Maratos, A., Crawford, M. J. (2011a).
Music therapy or music medicine? Psychotherapy and Psychosomatics, 80(5), 304. [DOI:

!
23!
10.1159/000323166]
Gold, C., Wampold, B., & Erkkilä, J. (2011b). Designing pragmatic and meaningful trials of
psychosocial interventions. Reply to BJP eLetter by Sekhri and Jackson. British Journal of
Psychiatry, eLetter. [DOI: 10.1192/ bjp.bp.110.085431]
Gold, C., & Erkkilä, J. (2011c) Authors’ reply to “Refurbishing the masked RCT design for
psychological interventions”. British Journal of Psychiatry, 199(6), 515–516. [DOI:
10.1192/bjp.199.6.515]
Gold, C. (2011d). Special section: music therapy for people with autistic spectrum disorder.
Nordic Journal of Music Therapy, 20(2), 105–107.
Gold, C., Erkkilä, J., & Crawford, M. (2012). Shifting effects in three randomised controlled
trials of a complex intervention. Acta Psychiatrica Scandinavica,126(5), 307–314.
Gold, C., Mössler, K., Grocke, D., Helda,l T. O., Tjemsland, L., Aarre, T., et al. (2013).
Individual music therapy for mental health care clients with low therapy motivation:
Multi-centre randomised controlled trial. Psychotherapy and Psychosomatics, 82(5),
319-331.
Hakvoort, L., & Bogaerts, S. (2013). Theoretical foundations and workable assumptions for
cognitive-behavioral music therapy in forensic psychiatry. Arts in Psychotherapy, 40(2),
192-200.
Hazel, N. (2008). Cross-national comparison of youth justice, 30-2.
www.yjb.gov.uk/Publications/Resources/Downloads/ Cross˙national˙final.pdf. (accessed 3
October 2012).
Higgins, J. P. T., & Green, S. (Eds.). (2011). Cochrane Handbook for Systematic Reviews of
Interventions (Version 5.1.0) [updated March 2011]. The Cochrane Collaboration. Available
from www.cochrane-handbook.org.
Hoskyns, S. (1988). Studying group music therapy with adult offenders: research in progress.
Psychology of Music, 16, 25–41.
James, D. J., & Glaze, L. E. (2006). Mental health problems of prison and jail inmates.
Bureau of Justice Statistics Special Report. U.S. Department of Justice, 1–3.
Jolliffe, D., & Farrington, D. P. (2004). Empathy and offending: a systematic review and
meta-analysis. Aggression and Violent Behavior, 9, 441–476.
Judge, T. A., Bono, J. E. (2001). Relationship of core self-evaluations traits-self-esteem,
generalized self-efficacy, locus of control, and emotional stability-with job satisfaction and

!
24!
job performance: A meta-analysis. Journal of Applied Psychology, 86(1), 80–92.
Juslin, P. N., & Sloboda, J. A. (2010). Handbook of Music and Emotion. Oxford, UK: Oxford
University Press.
Leith, H. (2007). What distinguishes a music therapist from a community musician?
Unpublished manuscript.
Leith, H. (2011). Mixed Methods Exploration of Music Therapy and the Resettlement of
Women Prisoners with Non-Psychotic Mental Health Problems [Elaborate proposal for PhD
study]. Book of Abstracts: PhD Course in Music Therapy Research, 2011 April 28 -May 5.
Aalborg, Denmark: Aalborg University.
Loth, H. (1994). Music therapy and forensic psychiatry - choice, denial and the law. Journal
of British Music Therapy, 8 (10), 10–18.
Loth, H. (1996). Music therapy. In: Cordess, C., & Cox, M. (Eds.), Forensic Psychotherapy:
Crime, Psychodynamics and the Offender Patient. London, UK: Jessica Kingsley Publishers,
561–566.
Maratos, A. S., Gold, C., Wang, X., & Crawford, M. J. (2008). Music therapy for depression.
Cochrane Database of Systematic Reviews, 1. [DOI: 10.1002/ 14651858.CD004517.pub2]
Maruna, S., & Farrall, S. (2003). Desistance from crime: a theoretical reformulation. Kölner
Zeitschrift für Soziologie und Sozialpsychologie, 43, 171–194.
Mcniff, S. (2007). Empathy with the shadow: engaging and transforming difficulties through
art. Journal of Humanistic Psychology. 47(3), 392–299.
Moss, L. S. (2004). The Effects of Lyrics Analysis and Group Discussion on Values
Clarification and Decision Making in Adolescent Delinquents [Master’s thesis].Tallahassee,
FL: Florida State University.
Mössler, K. (2011a). “I am a psychotherapeutically oriented music therapist”: theory
construction and its influence on professional identity formation under the example of the
Viennese School of Music Therapy. Nordic Journal of Music Therapy, 20(2), 155–184.
Mössler, K., Chen, X. J., Heldal, T. O., & Gold, C. (2011b). Music therapy for people with
schizophrenia and schizophrenia-like disorders. Cochrane Database of Systematic Reviews ,
12. [DOI: 10.1002/14651858.CD004025.pub3]
Nolan, P. (1983). Insight therapy: guided imagery and music in a forensic psychiatric setting.
Music Therapy, 3(1), 43–51.
O’Grady, L. (2009). The Therapeutic Potentials of Creating and Performing Music with

!
25!
Women in Prison: A Qualitative Case Study [PhD thesis]. Melbourne, Australia: University
of Melbourne.
O’Grady, L. (2011). The therapeutic potentials of creating and performing music with
women in prison: a qualitative case study. Qualitative Inquiries in Music Therapy, 6,
122–152.
Oxford Dictionary of English (3rd Ed). (2010). Oxford, UK: Oxford University Press.
Priebe, S., Frottier, P., Gaddini, A., Kilian, R., Lauber, C., Martínez-Leal, R., et al. (2008).
Mental health care institutions in nine European countries, 2002 to 2006. Psychiatric Services,
59(5), 570–573.
Prison Reform Trust. (2011). Bromley Briefings Prison Factfile.
www.prisonreformtrust.org.uk/Portals/0/Documents/Fact%20File%20June%202011%20web.
pdf. (accessed 3 October 2012).
Procter, S. (2011). Reparative musicing: thinking on the usefulness of social capital theory
within music therapy. Nordic Journal of Music Therapy, 20(3), 242–262.
Rio, R. E., & Stenny, K. M. (2002). Music therapy for juvenile offenders in residential
treatment. Music Therapy Perspectives, 20, 89–97.
Skaggs, R. (1997). Music-centered creative arts in a sex offender treatment program for male
juveniles. Music Therapy Perspectives, 15, 73–78.
Smee, S. (2009). Engendering Justice - From Policy to Practice: Final Report on the
Commission on Women and the Criminal Justice System. Fawcett Society.
Smeijsters, H., Kil, J., Kurstjens, H., Welten, J., & Willemars, G. (2011). Arts therapies for
young offenders in secure care: a practice-based research. Arts in Psychotherapy, 38, 41–51.
Smeijsters, H. (2012). Analogy and metaphor in music therapy: Theory and practice. Nordic
Journal of Music Therapy, 21(3), 227–249.
Stern, D. (2010). The issue of vitality. Nordic Journal of Music Therapy, 19(2), 88–102.
Thaut, M. H. (1989a). The influence of music therapy interventions on self-rated changes in
relaxation, affect, and thought in psychiatric prisoner-patients. Journal of Music Therapy, 26,
155–166.
Tuastad, L., & O’Grady, L. (2013). Music therapy inside and outside prisons - a freedom
practice? Nordic Journal of Music Therapy, 22(3), 210-232.
Ukoumunne, O. C., Gulliford, M. C., Chinn, S., Sterne, J. A. C., & Burney, P. G. J. (1999).
Methods for evaluating area-wide and organisation-based intervention in health and health

!
26!
care: a systematic review. Health Technology Assessment, 3(5), 1–75.
Walmsley, R. (2009). World Prison Population List. (8th Ed.), London: International Centre
for Prison Studies.
Wampold, B. E. (2001). The Great Psychotherapy Debate: Models, Methods and Findings.
Mahwah, NJ: Lawrence Erlbaum Associates.
Wang, B., Li, B. H., & Hu, J. M. (2007). The research on personality disorders of the
recidivists. Journal of Psychiatry, 20(2), 81–3.
Ward, T., & Brown, M. (2004). The good lives model and conceptual issues in offender
rehabilitation. Psychology, Crime & Law, 10(3), 243–257.
Ward, T., & Maruna, S. (2007). Rehabilitation: Beyond the Risk Paradigm. Abingdon, UK:
Routledge.
World Health Organization. (2007). Mental health in prison. Mental Health, Human Rights
and Legislation Information Sheet, 1-7. www.who.int/mental˙health/policy/
legislation/policy/en/. (accessed 3 October 2012).
Worling, J. R. (2001). Personality-based typology of adolescent male sexual offenders:
differences in recidivism rates, victim-selection characteristics, and personal victimization
histories. Sexual Abuse: A Journal of Research and Treatment, 13(3), 149–166.
Wyatt, J. G. (2002). From the field: clinical resources for music therapy with juvenile
offenders. Music Therapy Perspectives, 20, 80–88.
CONTRIBUTIONS OF AUTHORS
XJC conceived the review, drafted the first version of the protocol and suggested selection
criteria. HL and CG helped with drafting and re-writing the protocol, provided feedback on
selection criteria and other methods details. LEA and TTM helped with improving the
sections on how the intervention might work and description of outcomes.
DECLARATIONS OF INTEREST
XJC, HL and CG are trained music therapists. XJC and CG are involved in a potentially
eligible ongoing study (ClinicalTrials.gov Identifier: NCT01633125). CG is also involved in
another potentially eligible study (ISRCTN22518605) and serves as an associate editor of the
Cochrane Developmental, Psychosocial and Learning Problems Group. HL works as a music
therapist with women in prison and is conducting mixed methods research into music therapy

!
27!
and the resettlement of women prisoners with non-psychotic mental health problems. LEA
and TTM declare that they have no conflict of interest.
SOURCES OF SUPPORT
Internal sources
PhD Programme in Music Therapy, Aalborg University, Denmark.
External sources
GC Rieber Foundation, Bergen, Norway.
The Research Council of Norway, Norway.