Play and playing have always been at the core of psychodynamic approaches to working with children (Klein, 1932; A. Freud, 1965; Winnicott, 1971). The reasons for this are simple: the content, structure, and function of play are viewed as providing a window to understanding the nature of the child's anxieties and conflicts, and to assessing the internal and relational capacities he has available to organize and regulate his thoughts, feelings, and intentions. Psychodynamic child psychotherapy had its earliest beginnings nearly a hundred years ago, when Sigmund Freud used the principles of psychoanalysis to understand and ‘treat’ (via the boy's father) the symptoms of Little Hans, a 5-year-old Viennese boy with a dread of horses (S. Freud, 1909). It was Hans's play, drawings, and fantasies that helped Freud uncover the conflicts and anxieties thought to lie beneath the child's fears, and that guided the interpretations of these fears that he passed along to the boy's father.

Freud's treatment of Little Hans was—in essence—the first psychodynamic child therapy, although his reliance upon verbal interpretation would differentiate his approach, derived directly from adult psychoanalysis, from that of psychoanalytically oriented therapy. Pioneered by his daughter, Anna, and another Viennese psychoanalyst, Melanie Klein, psychodynamic child therapy was oriented around discovering the meaning and function of the child's play. Despite enormous differences in their view of early experience and psychic organization, Freud and Klein were together to create the field of child psychoanalysis, and establish it for a time as the primary means of treating children suffering from a wide array of psychological disturbances (see Klein, 1932; A. Freud, 1966–1980). For both, play, like dreams, provided a window to the deepest parts of the child's soul, a ‘royal road’ to the unconscious. They and their followers were the first to fully recognize that children can express in play what they cannot express in words; indeed, until they are nearly adolescent, due to the constraints of development, and the nature of childhood defenses, play is their dominant mode of self-expression. Whereas words and insight were viewed as the primary agents of change in adult psychotherapy, the dynamic and therapeutic aspects of play were thought to be the dominant medium of change in child psychotherapy.

The primary aim of psychodynamic child therapy has been, from the beginning, to allow development to keep moving the child forward (A. Freud, 1965; Winnicott, 1965). Children come to therapy because—whether or not they have specific symptoms, or are more globally delayed or derailed—they are not progressing developmentally, be this manifest in their behavior, their relationships, or their capacity to learn. Most psychodynamic child therapy is aimed at freeing the child from the constraints of his conflicts, deficits, or inhibitions so that he is able to function autonomously and productively in all domains of his functioning.

In the early days of psychodynamic child therapy, verbal interpretation of the unconscious meaning of the child's play was thought crucial to symptom remission and developmental advance. The extreme of this position is best represented by Melanie Klein, who suggested that ‘the child's fantasies, set forward in his play, become more and more free in response to continual interpretation’ (Klein, 1932, p. 18). In this early view, resolution is only achieved via interpretation. This belief was rooted in classical psychoanalytic notions of insight and structural change, an emphasis that has diminished considerably over the course of the past 80 years, although therapists still routinely use language to make sense of children's play. Children also often talk while they play, for playing provides a safe background for talking about difficult topics. But interpretation, per se, is no longer emphasized as the primary agent of change in child work; rather, what is thought to be curative is enhancing the child's symbolic, imaginative, and mentalizing capacities by increasing the range, depth, and emotional richness of his play (see Rogers, 1995). This expansion of the child's capacity to acknowledge various aspects of his self-experience in the safety of play and fantasy is, many believe, what allows developmental progress. Mentalization in play leads to the development of structures for containing feelings and understanding oneself and others (Slade, 1994; Fonagy and Target, 1996b, 1998; Fonagy et al., 2002a).

The capacity to play is rooted in early relationship experience (Slade, 1986, 1987, 1994). Beginning with the earliest playful exchanges with the mother, the child slowly develops the capacity to recognize that he and she have separate and unique minds, and that ideas and feelings are not concrete realities, but rather states that—in play—can be reworked and transformed (Fonagy and Target, 1996b; Target and Fonagy, 1996). The development of these capacities depends upon the establishment of intimate, secure relationships, which permit the discovery of the self and the other, and their separation. In relationships that are disturbed, however, these capacities are also disturbed; putting things into words and into play can be terrifying and disorganizing. And, lacking the presence of a comforting and organizing internalized other, symbolization becomes terrible evidence of one's separateness rather than a means to maintain contact and closeness (Winnicott, 1971; Slade, 1986).

It is for these reasons that the child's capacity to establish a relationship with the therapist (and, conversely, the therapist's capacity to establish a relationship with the child) is central to the treatment (Slade, 1994). Many children arrive knowing that their primary relationships depend upon their either not expressing, or disguising or distorting what they are truly thinking and feeling. The development of the capacity to play in a rich symbolic manner depends upon their experience of the therapist's willingness to both accept and contain the complexity and rawness of their actual internal world. This experience of the other as at once tolerant and regulating is what makes it possible for the child to establish the relationship that fosters the emergence of mentalization and symbolic functioning.

Play therapy is at the core two people, the child and the therapist, playing together. Children enter treatment with varying capacities to play, to talk, and to establish a relationship with the therapist. Most often these variations are linked to the nature and severity of developmental disruptions, emotional disturbance, and trauma. Sometimes the first job of the therapist is to help the child play, even a little. This may mean helping the child with the rudiments of telling a coherent story, it may mean helping him to imagine the inner life of the characters he has created, it may mean helping him find solutions in play that help to contain the intense feelings generated (Slade, 1994). But even when a child is able to play, all playing is not equal: play that is repetitive, devoid of emotion, or designed to inhibit communication (whether explicitly symbolic or not) precludes intimacy with the self or therapist, as does play that is dysregulated, fragmented, and too close to the affect it is meant to transform. In either instance, the child is unable to embrace the ‘pretend mode’, the space between reality and fantasy that allows for transformation, individuation, and true connection with another. Creating this ‘playspace’ with the therapist is the work of psychodynamic child therapy (Winnicott, 1971; Fonagy and Target, 1996b). It is important to note, in this context, that as children enter middle childhood, they may well not favor explicitly symbolic play; rather, they will choose board games or more physical forms of play, such as basketball, sewing, etc. This is not to say that such play cannot become symbolic, or at least invested with dynamic meaning and complexity that can be critical to therapeutic change (see Altman, 1997). The playspace can have all sorts of shapes, but it is the capacity to engage with the therapist in play, and in the creation of shared experience that defines real therapeutic engagement. Of course, as children age, they will begin to prefer talk over play; in fact, it is often in the context of apparently ‘neutral’ activities that they will begin to talk about the things that are bothering them.

Because the relationship is so central to moving development forward, regularity is thought to be an especially crucial aspect of the process of play therapy. Children are typically seen at least once a week, and many clinicians prefer to work with them twice or three times a week, because the processes inherent to the development of the capacity to pretend fully and imaginatively are complex, and require sustained periods of connection with the therapist. In many clinical settings this is simply not feasible, but there is evidence that increased frequency is critical to developmental change in seriously disturbed children (Target and Fonagy, 1994a). Equally critical to the child's progress is consistency. Children find change and disruption difficult, as their defenses are typically relatively tenuous or overly rigid (A. Freud, 1965); in either case, their capacity to engage in treatment is greatly helped by the therapist's sensitivity to the impact of changes in schedule separations and other.

Until relatively recently, there was little consideration in the psychodynamic child therapy literature of how to involve the parents in a child's individual treatment (this despite the fact that parents are almost always involved in children's therapy in some way). Historically, the parent and his or her actual behavior with the child were viewed as extraneous to the treatment process. This had much to do with the history of child psychoanalysis, and in particular with the emphasis within this literature upon both the privacy and exclusiveness of the of the child–therapist relationship, and upon the view that treatment was meant to affect internal processes rather than real relationships. While parents were typically seen occasionally for guidance and general ‘catching up’ on the child's home and school life, there was little conceptualization of how to engage dynamically the parent in the child's treatment so as to change ongoing patterns of interaction and relatedness. In the early days, this was actually frowned upon. However, as clinicians began to recognize the impact of relationships (in the extreme, trauma or abuse) upon child functioning, and as relational, attachment, and family approaches gained ascendancy, such predispositions began to change.

The first clinicians to radically confront the exclusion of parents from the child treatment process were Selma Fraiberg and her colleagues in their work on infant–parent psychotherapy (Fraiberg, 1980; Lieberman and Pawl, 1993). Called in by state welfare authorities to decide on troubled young mothers’ capacities to care for their children, many of whom were showing signs of trauma and abuse at a very young age, Fraiberg and her colleagues were able to affect the parent–child relationship in direct and dramatic ways by working with parents and infants together. They believed that the baby's presence in the room galvanized maternal affects and representations in ways that were transforming and healing, and allowed mothers to separate their own projections from the babies’ affiliative and attachment needs. While this approach was virtually unheard of in the late 1970s, it has now become an accepted mode of working with parents and their infants and toddlers.

Today, therapists working with pre-school and school age children continue to differ in the extent to which they involve parents in the individual psychotherapy of their children, although most believe that establishing and maintaining an alliance with parents is vital (Siskind, 1997; Slade, 1999, in press; Novick and Novick, 2002). However, therapists with differing trainings and orientations, do this differently. For some therapists, typically those who are more psychoanalytically oriented, the domain of the child's individual psychotherapy is still secluded, and parent meetings are less central to the therapy. For others, working from a more object relational and attachment framework, separate but regular (at least monthly, if not more frequently) meetings are more typical (Slade, 1999). Some therapists—following the infant–parent psychotherapy model—involve the parents in the child's actual sessions, not to talk, necessarily, but using the child's play as the means to enhancing relatedness and communication between parent and child (Slade, 1999; Oram, 2000; Chazan, 2002). (As is described in the section below, this approach has much in common with current family therapy approaches.)

The aim of most parent work is to effect change in the dynamics and functioning of the actual parent–child relationship, as such changes are believed intrinsic to development in the child. Clearly, one aspect of this work is to help parents understand critical aspects of their children's development; for example that a 4 year old's lie does not have the same significance or meaning as a 12 year old's. More important, however, successful parent work involves engaging the parent's capacity for reflective functioning (Slade, in press). Parent work helps a parent separate their own subjective experience of the child from the child's own thoughts, intentions and feelings. A parent's subjective experience of the child can be profoundly influenced by their own conflicts, or by the distorting effects of malevolent projections and representations. The work of the therapist is to help the parent hold the child and his or her subjective experience in mind, as distinct from the parent being aware only of their own perspetive. This kind of work can powerfully help the parent to become better at managing the child's feelings and behaviour.

There is rather less research available on the outcome of psychodynamic treatment than of some other approaches with children (Weisz et al., 1992). The most extensive study of intensive psychodynamic treatment was a chart review of more than 700 case records at a psychoanalytic clinic in the United Kingdom (Fonagy and Target, 1994, 1996c; Target and Fonagy, 1994a,b). The observed effects of psychodynamic treatment were impressive, particularly with younger children and those with emotional disorder or those with disruptive disorder, comorbid with anxiety. In addition, intensive treatment appeared more effective for children with emotional disorders which caused significant impairment across contexts. However, children with pervasive developmental disorders or mental retardation appeared to respond poorly to psychodynamic treatment.

Some smaller-scale studies have demonstrated that psychodynamic therapy can bring about improvement in aspects of psychological functioning beyond psychiatric symptomatology. Heinicke (1965; Heinicke and Ramsey-Klee, 1986) demonstrated that general academic performance was superior at 1-year follow-up in children who were treated more frequently in psychodynamic psychotherapy. Moran and Fonagy (Fonagy and Moran, 1990; Moran et al., 1991) demonstrated that children with poorly controlled diabetes could be significantly helped with their metabolic problems by relatively brief, intensive psychodynamic psychotherapy. In a naturalistic study, Lush et al. (1991) offered preliminary evidence that psychodynamic therapy was helpful for children with a history of severe deprivation who were fostered or adopted. Improvements were only noted in the treated group.

An important study from the University of Pisa (Muratori et al., 2003) looked at the effectiveness of an 11-session treatment program for 58 children with anxiety disorder or dysthymic disorder. The treatment was structured, focal psychodynamic psychotherapy, including both family and individual sessions. The control group were referred for community treatment. Measures were taken at baseline, 6 months (end of treatment for the experimental group), and 2 years follow-up. The two key measures were the Children's Global Assessment Scale (CGAS; completed by a blind, independent interviewer who interviewed both child and parent), and Child Behavior Check List (CBCL) completed by the parents. The results revealed a significant difference between the groups, only at follow-up, on both the CGAS and CBCL scales. In addition, the authors report a significantly lower level of service use in the experimental group during the follow-up period. This study is unique in providing a well-matched control group, to assess the effectiveness of psychodynamic psychotherapy.

Negative findings concerning the effectiveness of child psychodynamic therapy were reported by Smyrnios and Kirkby (1993). In this study no significant differences were found at follow-up between a time-limited and a time-unlimited psychodynamic therapy group and a minimal contact control group. The control group families may have had good outcomes because the minimal contact consisted of discussion of an agreed formulation and of how the family could effectively help themselves. Negative outcomes were also reported by Szapocznik et al. (1989), who compared the effectiveness of individual psychodynamic therapy or structural family therapy in treating disruptive adolescents. Both forms of treatment led to significant gains. But at 1-year follow-up, while the child functioning remained improved for both groups, family functioning had deteriorated in the individual therapy group.

Good evidence is available for the success of therapeutic approaches that can be considered indirect implementations of psychoanalytic ideas. For example, Kolvin et al. (1981) demonstrated that psychodynamic group therapy had relatively favorable effects when compared with behavior therapy and parent counseling, particularly on long-term follow-up. In a smaller-scale study of group social relations interventions, Lochman et al. (1993) have reported similarly encouraging results. Interpersonal psychotherapy (IPT), although not a psychodynamic treatment (Klerman et al., 1984) incorporates interpersonal psychodynamic principles. Mufson et al. (1993) have manualized this therapy for depressed adolescents (IPT-A), and a clinical randomized controlled trial has been reported (Mufson et al., 1999). This included 48 referred adolescents with major depression, of whom 32 completed the protocol. The majority of drop-outs came from the control condition, which was ‘clinical monitoring’, effectively a waiting list. An intent-to-treat analysis showed that 75% of patients treated with IPT-A recovered, as judged by Hamilton Rating Scale scores, in comparison with 46% of those in the control group. Other studies have also found IPT to be effective for adolescents, more so on some dimensions than was CBT (Rosselló and Bernal, 1999), and sertraline (Santor and Kusumakar, 2001).

Thus, there is limited evidence on the efficacy of child psychodynamic psychotherapy. However, given the fact that each study reported has methodological shortcomings—such as small sample size, nonstandardized process and outcome assessments, nonrandom assignment, lack of adherence measures—what emerges most powerfully is the need for new outcome studies in this area, applying strict methodological criteria and samples which reflect clinical realities.

Psychodynamic child psychotherapy was the first psychosocial treatment specifically developed for mental disorders for children. Its ambitious aim is the developmental advancement of children whose symptoms are seen as an indication of a failure to progress socially, cognitively, or emotionally. While interpretation and insight represent an important feature of therapeutic process, more central are becoming able to play, and to establish a relationship with a therapist that is richly imbued with symbolic meaning, and aims to extend the child's capacity coherently to represent mental states. These representations allow the child to understand himself and others better, and to gain more control over what happens in his or her relationships as a result. For most child therapists, work with parents is important for both preschool and school-age children, its primary aim being to help parents understand their child's thoughts and feelings. Evidence for psychodynamic child therapy is currently limited but available studies suggest that this approach can be helpful in improving the child's development across domains of functioning, especially interpersonal understanding.

As with other therapeutic approaches with children, CBT with children is shaped by theory, ideology, and traditions of practice. It has its theoretical underpinnings in a number of related research traditions particularly behavioral science (Herbert, 1994), social learning theory (Bandura, 1977), cognitive developmental theory (Bruner, 1990), and the cognitive theory of emotional disorders. Ideologically, CBT practice is described as following a scientific practitioner approach, which emphasizes the importance of empirical methodologies, research evidence, and formal hypothesis testing. This ideological framework has shaped specific traditions of practice such as the promotion of open collaborative practice with clients. As an example, CBT therapists encourage the development of a mutual formulation of the client's problems, professional knowledge sharing with the client, explicit explanations of the treatment model and open testing of individual focused hypotheses about what may produce change. However, compared with social constructionist approaches, CBT would be seen to adopt an expert position with its clients.

In current practice, CBT with children (and their parents) has evolved from a loosely related set of theories, research findings, beliefs, and practice traditions, resulting in a diverse set of therapeutic techniques and practice. Some interventions emphasize the central role of children's cognitions in the etiology and maintenance of childhood disorders and thus aim to change cognitions, whereas others focus more on the behavioral mechanisms thought to be central to achieving change. Thus, Kendall (2000) has defined current practice as ‘the purposeful attempt to preserve the demonstrated positive effects of behavioral therapy within a less doctrinaire context and to incorporate the cognitive activities of the client into the efforts to produce therapeutic change.’

Historically, techniques of change based on behavioral theory, such as behavior modification, preceded more cognitive approaches. Behavior modification (Herbert, 1998) applies the theory of classical and operant reinforcement to a wide range of childhood clinical problems such as anxiety disorders (phobias, obsessive-compulsive disorder) conduct problems and early developmental problems (sleep disturbance, enuresis). This approach is based on the notion that problem behaviors are likely to recur if the consequences of such behaviors are rewarding to the child. Formal treatments of this kind begin with a functional analysis, in which the antecedents and consequences of problem behaviors are systematically recorded so as to determine environmental and transactional patterns and responses that support these behaviors. Interventions are planned to alter these behavioral patterns by focusing on reducing rewarding consequences, and increasing the positive consequences of pro-social behaviors. This approach is most commonly applied by working with the parent, using reported behavior of the child in the school or home environment. Improvements with respect to reduced frequency or severity of problem behaviors are explicitly celebrated or rewarded. For example, parents are encouraged not to respond to angry outbursts or tantrums in young children with ‘rewarding’ responses (attention, raised excitement) and to encourage more pro-social behaviors in achieving wishes or negotiating conflict.

Alternatively, treatment focuses more on the behavior and interactions taking place within the treatment session and explicitly structures sessions as opportunities to change the child's behavior. Most notable of these is the ‘Parent–child Game’ (Jenner, 1999) in which a therapist directly prompts parents (through a one-way screen using an earpiece) to follow behavior modification principles in changing a child's behavior.

Parent training has become one of the most widely used of the behavioral approaches. This method has been most comprehensively developed and evaluated by Webster-Stratton (Webster-Stratton et al., 1989; Webster-Stratton and Herbert, 1993). The training can be delivered to parents either individually or in a group, and is typically brief (eight to 12 sessions) with a carefully prepared curriculum for each session. Video clips are used to illustrate common parent–child conflicts, and the emphasis is on structured ‘homework’ exercises that facilitate the generalization of skills learned in therapy to the family environment. Initial sessions focus on positive interactions between the parent and child, particularly those that occur within the context of play. Behavioral principles of selective attention and reinforcement are illustrated and practised through homework tasks, along with more cognitive components such as problem solving, negotiating turn taking and emotional recognition.

The apparent theoretical simplicity of the original behavioral model was partly due to its nearly exclusive focus on childhood behaviors, rather than upon the relationships in which problematic behaviors occurred. This despite the fact that the intervention was almost always implemented through social interaction between the parent and child. Compared with early developments of behavior modification, current behavioral work tends to include relationship factors much more. Thus, there may be increasingly little difference between systemic interventions that encourage interactional experiments, and behavioral approaches that take account of the parent–child relationship, than may appear from theoretical descriptions of these treatment models.

Although current evidence would suggest a place for parent training in addressing the needs of children with emotional and behavioral difficulties, the exclusive focus on the parent is clearly limiting. Greater effectiveness has been indicated for parent training programs that offer child-focused CBT alongside the parent training (Kazdin et al., 1992; Webster-Stratton and Hammond, 1997). Improved generalization and increased stability of treatment effects produced by parent work may be facilitated by greater emphasis on direct work with the child, including the child's thoughts about current difficulties, and the development of social and problem-solving skills. As will be described in the next section, these limitations have, in part, led to the development of child-based cognitive-behavioral treatments of children.

The CBT model is based on the proposition that childhood emotional disorders are maintained by implicit cognitive biases manifest through fixed core beliefs, dysfunctional assumptions, and automatic thoughts about the world, self, or others resulting in dysfunctional mood states, emotion or social interaction (Friedberg and McClure, 2002).

CBT with children typically has four key components, namely engagement, formulation, learning new skills, and applying change strategies (Kendall, 2000; Friedberg and McClure, 2002). The construction of a shared, comprehensible formulation is central. Problems are defined in terms of a child's thoughts, feelings, and/or behavior, usually linked to specific situations rated by frequency and severity. This enables problems to be addressed sequentially and organized in a hierarchical way that allows the child (and parent) to determine what they are able to cope with. The person (child) in a more global sense is not the problem. This definition of the problem allows for explicit understandings about the solution that is being sought and allows the possibility of the child and the parent achieving ‘success’ by reaching explicit targets of change. Behavioral techniques for noticing and rewarding positive change are usually integrated into this broader CBT approach.

In general, CBT sessions tend to have a more structured ‘curriculum’ than nondirective therapies. The therapist is active, self-disclosing where appropriate, and adopts a psychoeducational, collaborative approach in which a range of activities within the session may be suggested. Kendall (2000) uses the metaphor of the therapist as being like a sports coach in which concepts of practice, preparation, and training are often referred to. The focus is on creating change both within the session but also more crucially in generalizing change to the child's daily life. Practicing anger or anxiety management skills with the therapist in real life situations may be part of the treatment plan, as the intervention is not necessarily confined to the clinic room. In order to support the generalization of new skills to the home environment, the ‘curriculum’ often includes homework and record keeping tasks.

Activities initially may focus upon developing core skills such as: emotional recognition; separating thoughts, feelings, and actions; and activity monitoring and diary keeping. For example, poor discrimination between anxiety and anger feeling states may be more common in children with emotional behavioral difficulties. Similarly, improving a child's ability to regulate emotional states is likely to be dependent on their ability to monitor and notice internal states. Activities supporting strategies for change will be adopted depending on the formulation but may include a combination of behavioral and cognitive techniques such as relaxation training, problem solving, role playing, exposure, behavioral experiments, and testing the evidence for beliefs. Perhaps the most widely applied change technique is problem solving, in which children are guided to consider alternative options, to adopt a position of choice rather than powerlessness and to improve social perspective taking.

Compared with work with adults, the application of this approach to children raises a number of particular challenges. First, in contrast to adults, children are brought to therapy (Kendall, 2000). They do not make independent decisions to seek help for self-identified problems. The description of the ‘problem’ is constructed within a context of their families and caregivers. This is self-evident but has major implications in establishing collaborative practice with the child based on a shared formulation of a child's difficulties. Clearly, children may not ‘collaborate’ if they perceive the reason for therapy as being critical of them, i.e., having a ‘behavior problem’. Second, compared with adults, children's ability to make changes in their lives is restricted by their dependency on parents/caregivers. Third, children's interests and styles of interaction require that therapeutic methods not rely solely only on verbal interaction (Friedberg and McClure, 2002). Some cognitive techniques for adults may be developmentally inappropriate and ineffective with children. There is a need to incorporate both the form and content of children's thinking for the cognitive components of CBT to become applicable. Thus, for younger children, their thinking and expectations of the world and others may be most readily revealed through symbolic play. Similarly, children may need narratives as a way of developing explanations about the world, rather than abstract ideas (Bruner, 1990). Thus, for example, story telling may have a greater role in cognitive restructuring than methods of Socratic questioning appropriate for adult CBT work. Finally, CBT interventions partly rely on the patient being able to report cognitive states in order that distortions can be effectively challenged. In general, children may have less practice (and less interest) in the recall of experience and monitoring internal states than adults. Clearly such therapeutic tasks need to be carefully constructed to be within their cognitive developmental abilities, although the degree to which this restricts the application of cognitive approaches even in young children is far less clear (Meadows, 1993).

As already indicated, there has been a tendency in the child CBT literature to describe CBT independent of the role and relationship of parents and other family members. For example, Lochman et al. (1991) concluded that the ‘most striking deficiency in CBT programs… has been the neglect of children's caregivers, especially parents. Intervening with these caregivers can be critical in strengthening treatment effects and in maintaining the generalization of treatment effects over time.’ In addition, there is some suggestion that involvement of parents may increase treatment effectiveness (Mendlowitz et al., 1999; Barrett et al., 2001). Different CBT approaches with children have proposed different roles for parents that can be broadly identified as facilitator, co-therapist, or patient. As facilitator, the parent is predominantly involved in supporting the child's individual therapy and may meet with the therapist occasionally (Kendall, 2000). As a co-therapist the parent may be actively involved in supporting the child in learning new skills and may be central to providing behavioral feedback and rewards. In such instances, the parent is seen as closely collaborating with the therapist using agreed upon CBT techniques (March and Mulle, 1998; Mendlowitz et al., 1999). Alternatively, parents may be clients receiving treatment to cope with their own difficulties, which may be associated with the child's problem, either as part of a family approach (Barrett, 1998) or individually alongside the child's sessions (Cobham et al., 1998). Typically, parents may be offered CBT to manage their own emotional and behavioral difficulties. In practice, parents may sometimes wish to move between these different roles during a child's treatment and, although some flexibility of relationship with the family is often essential, sudden changes in parental role can be disruptive for the child. In general, much work still needs to be done in developing coherent models of CBT practice that are coherent with family roles, relationships, and individual differences.

Overall, there is considerable evidence for the effectiveness of behavior modification, particularly with respect to conduct problems in younger children (Kazdin, 1985) and for developmental difficulties such as sleep disturbance and enuresis (Christophersen and Mortweet, 2001). The utility of parent training has also been well supported, although the exclusive focus upon the parent clearly limits its impact. Indeed, greater effectiveness has been indicated for parent training programs that offer child-focused CBT alongside the parent training (Kazdin et al., 1992; Webster-Stratton and Hammond, 1997). Improved generalization and increased stability of treatment effects produced by parent work may be facilitated by greater emphasis on direct work with the child, directed toward the child's cognitions about current difficulties and the development of social and problem-solving skills.

There is variable empirical support for the effectiveness of CBT depending on the disorder and the developmental level of the child (Fonagy et al., 2002b). In general, as reviewed in detail in the Fonagy et al. book, the evidence for effectiveness is stronger for moderate, single problem presentations rather than complex chronic problems with high levels of comorbidity. Within these constraints, there is strong and accumulating evidence for the effectiveness of CBT (parent training) as an effective treatment for conduct problems in children under 8 years. For older children (8–12 years) the addition of problem-solving skills training for the child appears to enhance parent training approaches. Similarly, CBT is proving to be an effective treatment for general and specific anxiety disorders both delivered in individual and group settings. In addition, CBT has been shown to improve physical outcomes for children with paediatric conditions and with developmental difficulties such as sleep and toileting difficulties. Evidence for the effectiveness of CBT for depression, for conduct problems in adolescence and attentional problems is less strong. For depression, there is little evidence for pre-adolescent children. For adolescents, CBT may provide benefit for moderate levels of depression but short-term treatments need to be extended to anticipate the risk of relapse. Similarly, for conduct problems in adolescence, CBT packages such as problem-solving and social skills training are unlikely to be sufficient to address moderate to severe levels of difficulty but may contribute to the effectiveness of multimodal treatment approaches that also address family relationships and broader social environment variables such as positive leisure activities. For attentional problems, there is some evidence that CBT may enhance on-task activity and reduce disruptiveness but this is less effective than stimulant medication. However, it may contribute to an overall treatment approach and enable lower doses of medication to be effective.

Much work remains to be done to establish empirical support for many aspects of this theoretical model, as opposed to therapeutic effectiveness (Stallard, 2002). There is some evidence that children with anxiety disorders are more likely to perceive the world as threatening (Kendall and Panichelli-Mindel, 1995) and that children with conduct problems may anticipate ambiguous social situations as indicating hostility (Dodge, 1985). However, there is as of yet sufficient formal evidence either that cognitive therapy both produces cognitive change in children or that this is critical to functional improvement (Stallard, 2002). This is a crucial area for future research.

CBT with children currently encompasses a wide range of interventions to address childhood disorders and distress. In general, there is some evidence of the usefulness of CBT for a number of childhood disorders. More established behavioral approaches such as behavior modification and parent training increasingly include cognitive factors for both parents and children, and the child is placed in a more central position in the therapeutic endeavor. This is ideologically welcome as it conveys respect for the child's perspective and experience. However, it remains unclear whether CBT is yet addressing critical cognitive factors that lead to childhood disorders. A broader theoretical model, which includes processes of attachment, family relationships, and social developmental factors, may better capture the multiple processes that contribute to childhood distress.

There are many excellent accounts of the history and development of systemic practice, from early more positivist roots in cybernetics, communication and systems theories, through to the so-called ‘second order’ therapies that incorporate constructivist and social constructionist models (Hoffman, 1981; Dallos and Draper, 2000). Although family and systemic therapies have become one of the predominant forms of working with children's emotional and behavioral problems, surprisingly little has been written about children's perceptions of family work or about ways in which children might be more fully engaged in the therapeutic process. Most therapeutic models rely heavily on verbal communication and so might be seen to exclude younger children. In the past family therapy has been criticized for ignoring children and, in effect, conducting therapy in their presence without involving them. Children's worlds are often full of play, creativity, and activity, and therapy must presumably incorporate these concepts if it is to be meaningful to children. A number of authors have written about how children might be more actively engaged in the therapeutic process: most arguing strongly for the inclusion of more play, creativity, story telling, and active involvement (O'Brien and Loudon, 1985; Zilbach, 1986; Combrinck-Graham, 1991; Gil, 1994; Wilson, 1998; Context, 2002).

Different schools of family therapy have had to address these concerns in different ways, ways that are congruent with their underlying theoretical principles. Structural family therapy (Minuchin et al., 1967; Minuchin, 1974) assumes that problems in the child arise from underlying problems in the structure and organization of the family. The therapist is interested in how the family makes decisions, and how the boundaries between individuals and subsystems within the family lead to relative engagement or distancing. The therapist is often directive, attending to sequences and patterns of behavior, and seeking to bring about change using techniques such as enactment, escalation, and unbalancing.

Minuchin (1974) developed structural family therapy while working with disorganized and chaotic families in a deprived inner city (New York). Therefore, it is not surprising that he looked to provide clearer structures for families, and that therapists have found these techniques particularly helpful in working with families where children have behavioral problems. Children may not, on the surface, welcome attempts to provide clearer rules and boundaries but the active, directed approach of the therapist in structural family therapy does make it easy to engage children. Techniques such as enactment and the encouragement of family members to practice new ways of behaving and communication in the session ensure that all family members, including even quite small children are actively involved in therapy.

Brief solution-focused therapy (Berg and de Shazer, 1993) assumes that problems are maintained by the way difficulties are viewed and by the repetitive, behavioral sequences surrounding attempts to solve them. Families are seen as constantly changing and it is assumed that families will already have solutions to their own difficulties. The therapist sets clear goals with the family and focuses on solutions not problems, looking for exceptions to the ‘problem-saturated’ story that the child is ‘always’ a problem. Underlying this emphasis on competence and solutions is a focus on challenging unhelpful beliefs about the child and the problem as part of the process of generating new solutions. This focus on solutions can be helpful when working with children who are often worried that being brought for therapy is just another context in which they will be blamed for family difficulties. Solution-focused work is often active and, like structural therapies, can involve tasks and between session homework—these practical activities provide a further opportunity for children to be actively engaged.

Postmodern therapies (Andersen, 1987; Anderson and Goolishan, 1988, 1992) are informed by social constructionism and see language rather than interactional patterns as the system to focus on in therapy. Language does not just describe the family situation but can create and maintain that situation. The therapeutic style is conversational with the intention of creating change through the development of new language. The therapist takes a nonexpert role and asks questions that seek to create new possibilities or alternative understandings. ‘Reflecting team’ conversations are used as a means of sharing the therapy team's alternative stories and explanations with the family without imposing those ideas on the family. Such therapies are linguistically based and it can be difficult, though not impossible, to engage younger children. Reflecting teams may be confusing for younger participants, although a brief report by Marshall and Reimers (2002) suggests that teenagers, at least, find them potentially helpful, understanding, and caring.

Narrative therapy (White and Epston, 1990) draws on the way that we all make sense of our experience by creating personal accounts or narratives. Therapy is a form of conversation that encourages reflection and can transform problem-saturated narratives into more positive accounts. The emphasis on language can be off-putting for children as with other postmodern therapies but techniques such as externalization, which assist in separating the person from the problem, can help the child to feel less blamed and join the child with the family in fighting the problem. Narrative therapists also see those with problems as having expertise in solving them that may help children to feel engaged and less blamed, and the emphasis on narrative suggests the possibility of links with stories and story telling—ideas familiar to children. Narrative therapists also look for ‘unique outcomes and positive exceptions’ concepts similar to the search for solutions and exceptions by solution-focused therapists, and this too may help children to feel less blamed. Larner (1996) draws on child psychotherapy and narrative theory and technique to suggest ways of joining the ‘child's symbolic play as narrative… ‘to the ‘… family story as social text in therapeutic conversation’. He quotes Anderson as describing the therapist's expertise ‘being in conversation with the expertise of the client’ and notes that the expertise of the child is in the ability to play.

There are a few recent studies looking at children's perspectives on therapy. Stith et al. (1996), for example, explored the experience of 16 children from 12 families in a qualitative study. Children, interviewed alone, wanted to be included in therapy and were keen to know more about their families, be involved in generating solutions and not feel blamed for problems. They did not want to be the sole focus of discussion. Even primary school children understood the purpose of therapy and found talking about problems helpful but their willingness to be involved increased with time and with the amount they knew about why their families were coming to therapy. Lobatto (2002) describes a thoughtful qualitative analysis of interviews of six children, aged 8–12 years, in the presence of their parents. She describes the difficulties children had in deciding how and when to participate in the therapeutic conversation, their uncertainty about the rules of therapy, and the importance of toys and play materials in maintaining a safe space for children within therapy. She echoes Wilson (1998) in suggesting the need for clearly stated ground rules about participation in therapy.

In general, family and systemic therapies have not been well evaluated despite their widespread use in clinical settings, and there is a need for more randomized, controlled evaluations. However, the quality of published research is similar to that concerning other psychological treatments and there are sufficient good quality studies to draw some conclusions. There is good evidence for the effectiveness of systemic therapies in the treatment of conduct disorders (particularly in older children, and in relation to offending) and substance misuse. Functional family therapy has been shown to be effective in reducing adolescent offending behavior (Alexander and Parsons, 1973; Parsons and Alexander, 1973; Barton et al., 1985; Gordon et al., 1988) in multiply offending adolescents. Follow-up into early adulthood showed improvements were maintained (Gordon et al., 1995). Multisystemic treatment (MST) comprising detailed individual assessment followed by a combination of therapeutic interventions has been demonstrated to reduce significantly recidivism when compared with treatment as usual. Improvements were maintained at 30-month follow-up and costs of MST were lower than in control groups (Borduin, 1999). MST is intensive and time consuming with sessions held in the family's home and in community locations. It is more than just family therapy, although classical family therapy interventions play a key part and is concerned, more than many interventions, with the family situated in its social context. Stanton and Shadish (1997) systematically reviewed studies of treatments for drug abuse and conclude that family–couples therapy is superior to individual counseling/therapy and peer group therapy for both adults and adolescents. Family therapy was also superior to family psychoeducation and tended to have lower drop-out rates than other treatments.

There is also good evidence for the effectiveness of systemic therapies in the treatment of anorexia nervosa in younger people. Russell et al. (1987) randomly allocated individuals with anorexia nervosa and bulimia nervosa to either family therapy or a ‘nonspecific form of individual therapy’ after discharge from inpatient care. At 1 year, family therapy was found to be more effective than individual therapy in patients whose illness was not chronic and had begun before the age of 19 years. Improvements were maintained at 5-year follow-up (Eisler et al., 1997). Robin et al. (1994), 1999), compared behavioral family systems therapy with a form of individual therapy for anorexia nervosa. In a random allocation study, behavioral family systems therapy produced greater weight gains and higher rates of resumption of menstruation at posttreatment and at 1-year follow-up than the comparison intervention.

In addition there is some support for the effectiveness of systemic treatments in depression, self-harm (where they may have significant cost benefits), and in chronic illness (Cottrell and Boston, 2002; Fonagy et al., 2002b). The existing research also offers some suggestions as to how systemic ideas may contribute to other therapeutic models or the development of integrated approaches. There have been reports that in controlled studies systemic therapies may reduce drop-out and increase engagement and consumer satisfaction (Szapocznik et al., 1988; Henggeler et al., 1996; Harrington et al., 1998). There is also support for the notion that parental involvement is beneficial even if parents are not in the same room as the young person as long as systemic ideas are informing therapy (Robin et al., 1994, 1999; Eisler et al., 2000). Systemic interventions may also have positive effects that are maintained and even increase with time (Szapocznik et al., 1989). This would fit with systemic theory that addressing underlying family interactional patterns will produce lasting change and is in contrast to cognitive-behavioral treatments that require ‘booster sessions’ to maintain change (Fonagy et al., 2002b). These findings suggest that systemic ideas have something useful to offer other theoretical perspectives.

There is evidence that family and systemic therapy is an effective treatment for some young people and systemic ideas can contribute to the delivery of other treatment modalities. However, the best evaluated systemic interventions are the older ‘first order’ structural/strategic models, not the more recent developments using social constructionist and narrative frameworks. It is possible for systemic therapies to ignore children and young people and become marital/adult work in the presence of the child. However, the theoretical models and practical techniques of the current schools of systemic practice all acknowledge the importance of involving children and have all found creative ways of doing this. There is emerging evidence from qualitative research that even quite young children can understand, make sense of, and participate in systemic work. Careful explanation of the purpose and process of therapy, recognition of the expertise of the child and the provision of environments that are child friendly and promote play and creativity should maximize the involvement of children.