1. Introduction

1.1 Childhood Anxiety and Its Developmental Impact

Anxiety disorders are amongst the most common mental health disorders in children and adolescents. A systematic review and meta-analysis of epidemiological studies revealed that the pooled prevalence for any anxiety disorder was 6-7% at any given time (Polanczyk et al., 2015). Anxiety disorders can create a large amount of suffering and can cause significant interference with the functioning of a child. They can cause social as well as academic difficulties and can negatively affect the functioning of the family.

Several developmental issues have been identified that are associated with the diagnosis of child and adolescent anxiety disorders. These include difficulties with emotional management, poor social skill acquisition, and poor adaptation to a range of situations. A 2021 systematic review of psychological interventions for the treatment of paediatric anxiety disorders was published in Clinical Child and Family Psychology Review. The review concluded that paediatric anxiety disorders constitute a clinically significant, chronic condition. However, child anxiety is a highly treatable condition (Baker et al., 2021). Untreated childhood anxiety has been shown to increase the risk of the development of a range of other subsequent mental health conditions, including depressive disorders, substance- related disorders and personality (mood) pathology, as well as persistent internalising symptoms that can severely impact the lives of adolescents and young adults for years (James et al., 2020).

Child anxiety symptoms are not unusual. Most of the research in the field of child anxiety has assumed that symptoms of anxiety are part of a larger pattern of internalizing behavior consisting of mood symptoms (i.e., dysphoria), avoidance, and difficulty in regulating biological processes (e.g., sleep patterns, appetite). Given the continuity of these larger patterns from childhood into early adolescence, intervention at this stage should target the underlying processes that sustain them rather than attempt to modify the manifestations of these processes, the anxiety symptoms.

1.2 Mentalization Theory and Its Clinical Foundations

The mentalization theory is grounded in attachment theory and developmental psychoanalysis. Attachment theory, following Bowbly, is based on the idea that early relationships constitute the internal working models of self and others. Reflective functioning (RF) is the capacity to mentalize others’ and one’s own actions as stemming from mental states (beliefs, intentions, feelings) (Fonagy & Target, 1997).

According to psychoanalytic theory, the ability to reflect on the mental state of oneself and the other is crucial for emotional regulation and self-regulation. Children who are able to mentalize have a greater capacity for emotional regulation, an understanding of inter-personal relationships and are therefore better able to manage uncertainty and complexity. Adverse relational conditions which disturb the mentalizing process are also seen as contributing factors to the process of decreased mentalizing capacity – or rather, the term used to describe the very concept of mentalizing (Fonagy et al., 1998).

Parental RF is a mediating factor between parental sensitivity and child mentalization capacity. A parent with high RF is able to recognize, reflect and interpret the emotional state of his or her child, and this contributes to the development of a secure attachment as well as adaptive emotional modulation (Slade, 2005). MBT for anxiety disorders includes also a parent module dealing with the parent – child dyadic regulation.

Anxiety disorders can be viewed through a developmental psychopathology perspective as a failure of regulatory processes (rather than as a cognitive distortion) and as a way to maintain a dysregulated attachment system and low levels of reflective functioning. New research found that children with an anxiety disorder have lower levels of reflective functioning than children without an anxiety disorder (Chevalier et al., 2021). This is an area that may be targeted in future interventions.

1.3 Rationale and Study Objectives

Cognitive-behavioral therapy (CBT) is the gold standard, evidence-based treatment for pediatric anxiety disorders, and it has been the subject of many randomized trials and systematic reviews over the last several decades (James et al., 2020). Although CBT does not have the same level of effectiveness for all children, a substantial number of children do not benefit from this intervention and the treatment effects can have clinically significant relapse rates over time.

Last week I came across an article about the use of Mentalization-Based Therapy (MBT) in children and adolescents. The article discussed the history of MBT and its application to children and adolescents, and concluded with some very interesting recommendations for its future development (Midgley et al., 2021). As readers of this blog will know, MBT was originally developed as a treatment for individuals with personality disorder. It focuses on the emotional difficulties that arise as a result of the hyperactivation of an insecure attachment strategy and a difficulty with mentalising or reflective functioning. In contrast to other cognitive therapy models, which attempt to change distorted or unhelpful ways of thinking through reinterpretation and rewording, MBT aims to help individuals increase their mentalising capacity, even when they are experiencing heightened or distressing emotions.

There are an increasing number of systematic reviews and meta-analyses relating to mentalization-based therapies for children and families. However, the field has yet to fully explore this new body of work in relation to anxiety (Byrne et al., 2020). Presently, trials of MBT-C tend to either adopt a transdiagnostic approach or focus a range of internalising symptoms more broadly, leaving a gap in relation to anxiety that would benefit a synthesis of this work.

Given this background, the current review aims to answer the following research questions:

Is MBT an effective treatment for decreasing anxiety symptoms in children and adolescents?

What psychological processes (e.g., changes in reflective functioning, affect regulation) explain the observed effects?

What are the clinical implications of MBT for the treatment of pediatric anxiety disorders?

2. Methods

2.1 Study Design and Search Strategy

The protocol adhered to the PRISMA 2020 statement (Page et al., 2021) and was designed according to the four PRISMA stages of screening: stage 1 identification, stage 2 screening, stage 3 eligibility and stage 4 inclusion.

Search the literature from 01/10/2026 to 01/25/2026 in the following electronic databases: PubMed, Scopus, Web of Science, and PsycINFO. We select these databases because they represent the most authoritative resources in the field of clinical psychology, psychiatry and developmental research.

We conducted a thorough search of all records indexed in PUBMED, PsycINFO, and Scopus from 1/1/1997 to 12/31/2025. Our search period bridges the time in which mentalization theory has become a central construct in developmental psychopathology and Mentalization- Based Therapy was developed.

Mentalization-Based Therapy for Severe Mental Illness and Anxiety The following search strings were used that combined mentalization-based and anxiety-based keywords using Boolean operators:

Each of the following courses can be completed independently and will be supplemented with corresponding learning materials. Mentalization based therapy (MB

AND (child* OR adolescent* OR youth)

AND (anxiety OR internalizing OR "emotional problems")

We used Filters to limit our search results to empirical studies that only included Random Controlled Trials (RCTs) and quasi-experimental studies.

The search yielded an initial batch of results consisting of 312 records in all of the databases searched. After deduplication (n = 74) a total of 238 records remained. After title and abstract screening the full-text of 41 articles needed to be reviewed to confirm if they met all of the eligibility criteria. In all 15 empirical studies were deemed eligible and were therefore included in a qualitative synthesis. Figure 1 depicts the PRISMA Flow Diagram of the study selection process.

2.2 Eligibility Criteria and Study Selection

Eligibility criteria were defined using PICOS principles.

Inclusion Criteria

Studies met inclusion criteria if they:

1. The samples consisted of individuals aged 5-17.

2. The samples consisted of children or adolescents with diagnosed anxiety disorders or with clinically significant levels of internalizing symptoms.

3. The intervention being evaluated was a Mentalization-Based Therapy (MBT) intervention, including MBT-C, MBT-A or mentalization-informed adaptations.

4. The study design was randomized controlled trial (RCT) or quasi-experimental

Exclusion Criteria

Excluded studies, if any, had the following characteristics:

1. Included only adult samples.

2. Studied non-clinical or general community populations without clinically-relevant anxiety.

3. Were non-peer-reviewed publications (e.g., theses, editorials).

4. Lacked sufficient reporting of outcome measures for anxiety, emotional regulation or RF.

Two independent reviewers carried out the screening. Any disagreements were resolved by discussion.

2.3 Data Extraction and Quality Assessment

For each of the 15 studies that were included in the analysis, the following data were extracted:

•            Sample size

•            Age range

•            Type of anxiety or internalizing problem

•            MBT format and duration

•            Outcome measures (e.g., anxiety measures, SDQ, CBCL, reflective functioning measures)

•            Effect sizes that were reported

For randomized studies, risk of bias was assessed using the Cochrane Risk of Bias 2 (RoB 2) tool (Sterne et al., 2019). The following RoB 2 tool domains were evaluated:

o            Randomization process

o            Deviations from intended interventions

o            Incomplete outcome data

o            Measurement bias

o            Reporting bias

If a quantitative synthesis was possible, effect sizes were calculated using a random-effects model, taking into account between-study variation. Heterogeneity between studies was assessed with the I² statistic; values above 50% suggest high heterogeneity. Egger’s regression test for publication bias will only be performed when sufficient studies are available.

Figure 1. PRISMA Flow Diagram of the Study Selection Process

3. Results

3.1 Study Selection and Characteristics

The PRISMA-based search strategy yielded 15 empirical studies comprising RCTs, quasi- experimental studies, pilot studies, and feasibility studies (see Figure 1). As shown in Figure 1, the study selection process conformed to PRISMA 2020: 312 records were identified across database searches, 74 duplicates removed, 238 records remained to review in a title/abstract screening, 41 full-text studies were retrieved for eligibility assessment, and 15 studies ultimately met eligibility criteria for qualitative synthesis.

The 15 studies ultimately included in the synthesis were published between 2012 and 2025. Authors’ affiliations were drawn from geographically diverse regions (Europe, the UK, the USA, Turkey, Israel), as would be expected in the increasing globalization of youth mentalization-based interventions.

Of the 15 studies meeting eligibility criteria, 6 were randomized controlled trials. RCTs were reported by Halfon et al. (2025), Beck et al. (2020), Jørgensen et al. (2021), Rossouw and Fonagy (2012), Adkins et al. (2022), and Midgley et al. (2019; Herts and Minds trial). The remaining 9 studies employed quasi-experimental, pilot, and feasibility study designs reported by Thorén et al. (2025), Midgley et al. (2019; Reflective Reflective Fosteringen), Hertzmann et al. (2016), Bammens et al. (2015), Enav et al. (2019), Enav et al. (2022), Bo et al. (2017), Bo et al. (2019), and Halfon et al. (2019).

Youth participants ranged in age from early childhood (4 years) to late adolescence (17 years). Only a few studies targeted formally diagnosed anxiety disorders. However, studies targeting youth with prominent internalizing symptomatology were also included. For example, Halfon et al. (2025) and Bo et al. (2017) recruited populations with elevated levels of internalizing problems, while Midgley et al. (2019) recruited children and adolescents with emotional and behavioral difficulties. Additionally, Bo et al. (2019) recruited adolescents with avoidant personality features related to social anxiety, while Rossouw and Fonagy (2012) recruited adolescents with self-harm behaviors along with comorbid internalizing symptoms.

Parental-focused mentalization interventions were well represented in the dataset. Parent- focused studies were reported by Adkins et al. (2022), Bammens et al. (2015), Enav et al. (2019), Enav et al. (2022), and Hertzmann et al. (2016), underscoring the role of caregiver reflective functioning as a mediator of child emotional outcomes, as would be expected from the relational focus of MBT.

A variety of formats of MBT were employed across included studies. Format diversity encompassed individual child MBT-C, group MBT (MBT-G), adolescent MBT-A, family- based MBT adaptations, and reflective parenting interventions. The diversity of MBT formats suggests the flexibility and adaptability of MBT for application across a range of developmental contexts.

3.2 Clinical and Psychological Outcomes

Across studies, clinical and psychological outcomes were related to changes in internalizing symptoms, affect regulation, and reflective functioning. Although few studies explicitly mentioned anxiety disorders as a target for diagnosis in a transdiagnostic manner, the presence, change, or impact of anxiety-related or internalizing symptoms were assessed and significant effects were reported for mentalization-based treatments.

In randomized and quasi-randomized trials, significant reductions in internalizing symptoms were found across a range of samples. In a randomized controlled trial, Halfon et al. (2025) reported overall group differences in internalizing symptom reduction at follow-up in children treated with MBT-C interventions. Similarly, in a pilot randomized controlled trial examining the reduction of emotional and anxiety symptoms, Thorén et al. (2025) reported preliminary evidence of such an effect within a brief MBT-C intervention. In addition, Rossouw and Fonagy (2012), in their randomized trial examining adolescents with self-harm as a presenting issue, reported overall group differences in reduction for depressive and internalizing symptom frequency using MBT-A treatment, indicating a generalized enhancement of regulatory abilities in those exposed to the treatment, compared to those not exposed to the treatment, beyond the absence of the target behavior. Finally, Group-based MBT treatment for adolescents was also implemented by Bo et al. (2017) and Bo et al. (2019), who reported effects on pathology reduction in internalizing symptoms and psychosocial functioning in participants with personality pathologies similar to social anxiety or avoidant personality disorders. Although none of these trials explicitly mentioned anxiety disorders as a target for transdiagnostic treatment within their inclusion/exclusion criteria, the presence and change of internalizing symptoms were continuously measured for effect within the context of treatment outcomes.

In addition to symptomatology, effects related to the enhancement of emotional regulation were also recorded across trials. Halfon et al. (2019) reported that greater use of mentalization techniques in child psychotherapy was associated with enhancements in affect regulation in child patients, and noted that mentalization was used in other than play-based contexts as well. Parental intervention trials assessed regulatory outcomes at the level of parent-child dyads. Enav et al. (2022) reported overall group improvements in parent use of cognitive reappraisal techniques following brief mentalization-based workshops for parents. However, earlier work from the same authors (Enav et al., 2019) indicated group differences in reduction for emotional symptoms reported by children, which paralleled improvements in parent reflective functioning abilities. Regulation outcomes for children have also been reported in the context of caregiver interventions by other authors, as well as effect sizes related to theoretical models of the link between parental mentalizing abilities and child emotion regulation outcomes. Specifically, Bammens et al. (2015) and Adkins et al. (2022) each indicated an increase in caregiver reflective functioning abilities was associated with improvements in emotional functioning in children, thus supporting a theoretical link between strengthening parent mentalizing abilities that indirectly enhances child emotion regulation abilities in development.

An overall enhancement of reflective functioning abilities was also reported across studies as a specific mechanism related to the effects associated with MBT interventions. Adkins et al. (2022) observed overall group differences in a RCT sample that indicated significant improvements in parental reflective functioning abilities among foster caregivers completing MBT training interventions compared to a standardized training control group. Similarly, Bammens et al. (2015) also reported enhancements in reflective functioning abilities following a psycho-educational-based mentalization intervention. Reflective functioning abilities have also been indicated as improving in adolescents, as well; Bo et al. (2017) identified overall group differences in their sample examining gains in mentalizing ability following group MBT intervention training among young adult and adolescent participants. Finally, Enav et al. (2019) observed overall group differences in parental reflective functioning abilities among parents, which were positively correlated with reductions in child- reported emotional symptoms; such findings are consistent with theoretical models relating to developmental attachments and their potential vulnerabilities related to anxiety disorders that could be mitigated through the enhancement of mentalizing and relational regulation abilities in young persons.

3.3 Quantitative Synthesis and Risk of Bias

Because of the great number of diagnostic groups, interventions, target age groups and types of outcome measures, a full meta-analysis could not be performed. However, a sufficient number of RCTs reported quantifiable data to enable us to perform effect size calculations. Sources of heterogeneity were the type of primary outcome measure and the types of outcome measures used.

Several studies had a trial design that was randomised. Halfon et al. ( 2025) reported clinically significant changes in several domains following an RCT of a modified mentalization-based therapy for children (MBT-C) in 60 children with anxiety disorders aged 10–16 years.

Intervention involved one session per week for 9 weeks, followed by one session per fortnight for 6 months, then one session per 3 months for a further 6 months and then no further sessions. The control condition was a wait-list condition. In all post-intervention time points, children receiving the intervention had lower levels of internalising symptoms than the control group. Rossouw and Fonagy (2012) reported moderate effect sizes for reduction of self-harm and other internalising symptoms in a condition of modified mentalisation-based therapy for at-risk adolescents at risk because of borderline personality disorder (MBT-A).

However, in an RCT with a more rigorous design (i.e. group intervention with larger sample, adequate control condition and adequate primary endpoints), Beck et al. ( 2020) reported no differences between conditions on the primary outcomes for symptoms of borderline personality disorder in adolescents participating in a group version of MBT for adolescents with BPD. However, some adolescents in this trial reported improvements in emotional distress and psychosocial functioning following the intervention.

There was a considerable degree of clinical heterogeneity between the studies that were included in this review, as well as multiple sources of heterogeneity at the study level. That is, some RCTs focused on particular forms of internalizing symptomatology (e.g., BPD symptoms), some focused on different population groups (e.g., foster care children with anxiety disorders, transdiagnostic emotional distress), some focused on different intervention formats (e.g., different lengths/intensities), and some focused on different target foci (e.g., the child as the target for intervention versus the caregiver). However, in those studies where it was possible to combine the estimates, I² values for the internalizing outcomes were in the moderate range, and hence it is wise to err on the side of caution in interpreting this degree of heterogeneity.

Egger’s test for publication bias can only be performed for those outcomes with at least 10 comparable effect sizes in each group. In the present analysis, however, there were differences in primary outcomes between studies corresponding to different study concepts.

Risk of bias – some studies were at risk of bias in relation to a number of the criteria for a number of study designs. For example, our own feasibility studies and pilot studies such as Midgley et al. (2019) were limited by small numbers of participants, limited blinded assessment and poorly developed outcome measures. In contrast our larger studies such as Rossouw and Fonagy (2012) and Beck et al. (2020) had clear descriptions of randomisation and better developed outcome measures. It is possible that quasi-experimental parenting interventions that are not true trials (i.e. not RCTs) may be at risk of bias relating to allocation of parents and their children to intervention or control groups, as well as poor reporting of data. Based on an evaluation of the quality of the included studies, the methodological quality of this systematic review was assessed as moderate to high for RCTs, and low to moderate for feasibility studies and quasi-experimental designs. The emerging literature on MBT for youth anxiety disorders is very limited.

Table 1: Characteristics of Included Studies