Children in psychodynamic psychotherapy: changes in global functioning

This study was part of the Erica Process and Outcome Study. The aim was to investigate if children's global functioning improves after psychodynamic psychotherapy. Variables that may predict changes in global functioning were examined both statistically and qualitatively, for example, the child's age and gender; diagnosis and comorbidity; treatment variables. The sample consisted of 33 children (five to 10 years of age) who participated in psychodynamic psychotherapy with parallel work with parents. Twenty-nine children had at least one DSM-IV diagnosis, and 15 children had comorbid conditions. At the start of psychotherapy, the most frequent diagnoses were attention disorder and disruptive behaviour. Child psychotherapists rated the children's global functioning using CGAS and HCAM. Large effect sizes were obtained (d = 1.80 and d = 1.98). However, no statistical correlations were found between global functioning and the selected variables. In order to illuminate the complex connection between process and outcome the study was complemented with in-depth case studies where data were taken from questionnaires, completed by the child therapists every third month. Two child therapies were selected: one in which the therapist rated a large change and one in which a small change was rated (CGAS). The analysis showed that important individual change, for example, attainment of formulated goals, was not always reflected in the change rated using the CGAS. Findings suggest that psychodynamic child psychotherapy can be beneficial. However, further research is needed to identify factors that contribute to change in children's global functioning.

Keywords: CGAS; psychotherapy process; case study; outcome; treatment goals; intersubjectivity; EPOS

Psychological problems among children and adolescents are continuously increasing (Kessler et al., [27]; Atladóttir et al., [2]), and by the year 2020 the numbers are predicted to increase by 50% (European Commission, WHO European region and the Ministry of Health of Luxembourg, September 2004). In response to these mental health problems, psychotherapy for children and young people has advanced in recent years, and a wide range of treatment modalities and interventions have been developed (Kazdin, [24]). One such treatment method is psychodynamic child psychotherapy with parallel work with the child's parents.

The rising levels of mental ill-health amongst children and adolescents have also put increased pressure on child and adolescent mental health services to provide information about factors that may contribute to successful or unsuccessful treatment outcomes (Fonagy et al., [18]).

Over the past decade, research on psychotherapy for children and young people has grown rapidly (Kazdin, [24]; Midgley et al., [33]; Tsiantis and Trowell, [42]). In parallel, researchers have examined a number of factors that could affect the outcome of child psychotherapy. However, the results of these investigations are often contradictory and therefore difficult to interpret. For example, in their meta-analyses, Casey and Berman ([10]) and Deakin and Nunes ([12]) examined the impact of gender on psychotherapy outcome. Their findings indicated that girls generally had a better outcome after psychotherapy than boys. Other studies have reported insignificant gender differences (e.g., Weisz et al., [43]). Researchers have also examined the role played by the child's age, and findings have pointed in different directions. Thus, Fonagy and Target ([17]) found that children under nine years of age had a better outcome compared to older children and that younger children derived greatest benefit from intensive treatment. On the other hand, Casey and Berman ([10]) reported that the child's age was unrelated to psychotherapy outcome.

In recent times, thorough reviews of psychotherapy outcome studies have been undertaken (e.g., Fonagy et al., [18]; Kennedy, [25]; Windaus, [44]). Although the review by Fonagy et al. ([18]) showed that proponents of cognitive behavioural therapy more often embrace the need for systematic treatment evaluations compared to proponents of other treatment modalities, a number of recent, well-controlled studies point in a different direction. For example, the studies by Horn et al. ([21]), Kronmüller et al. ([28]), and Winkelmann et al. ([45]) reported a good outcome of psychodynamic time-limited psychotherapy (30 sessions) for children with depression, anxiety and behavioural disorders, respectively.

Other studies of psychodynamic child psychotherapy have examined the effects of frame conditions such as time limitation. Muratori and co-workers (Muratori et al., [34]) showed that a time-limited intervention (11 sessions) had better outcome at a two-year follow-up compared to treatment as usual. A study of children with relatively mild difficulties showed similar results; both time-limited psychotherapy (12 sessions) and minimal contact intervention showed better outcomes compared with time-unlimited psychotherapy (Smyrnios and Kirkby, [41]). In contrast, a number of studies have shown that the length and intensity of the psychotherapeutic treatment is positively related to treatment outcome (Heinicke and Ramsay-Klee, 1986; Boston and Lush, [5]; Fonagy and Target, [17]).

Despite more recent developments, child and adolescent psychotherapy research has been criticised for its limited advances (Kazdin, [22]; Fonagy, [15]; Midgley, [32]). For example, treatment methods tend to be superficially described, and conditions necessary for successful treatment outcomes are seldom addressed (Kazdin, [24]). In order to understand how to judge whether a therapeutic method is suitable for a particular patient it is essential to have a deeper knowledge of the therapy process (Fonagy, [16]). It is also necessary to understand what course of events in therapy is related to change factors and outcome. Certain change factors are related to technique, others stem from what may be termed 'common factors'. These can be the matching between the type of therapy, problem, patient and psychotherapist as well as the creation of a therapeutic alliance (Green, [19]; Meissner, [31]).

The therapeutic alliance is a complex concept that embraces the agreement about goals and method in the therapy, the emotional bond that develops between the patient and therapist, the patient's involvement in the therapy as well as the therapist's ability to empathise with the patient's difficulties (Bordin, [4]). One aspect of the therapeutic alliance is the patient's expectations and the extent to which the patient is ready for change (Llewelyn and Hardy, [29]; Nock and Kazdin, [35]) The creation of a good therapeutic alliance in child therapy can be made more difficult due to the child not being the one initiating the contact or not expressing or experiencing any problems (Carlberg et al., [9]). The child, the school and the parents may all hold different perceptions of the child's problems and the goals of the therapy (Shirk and Karver, [39]). The involvement of the child's family is an important component in the treatment that can also contribute to a good result (Fonagy, [18]; Carlberg et al., [9]).

Moreover, clinically important outcome variables such as quality of life, belief in the future, and self-esteem have proved difficult to quantify (Fonagy et al., [18]). Another problem is that many well-controlled research studies (efficacy-studies) do not mirror complex clinical reality. Studies of child psychotherapies carried out in naturalistic settings (effectiveness-studies) are therefore greatly needed (Kazdin, [24]; Kennedy and Midgley, [26]; Midgley et al., [33]; Carlberg, [8]).

A related problem was discussed by Kazdin ([23]), who highlighted that psychotherapies that are evaluated may not be the ones that are most frequently used in clinical practice (Shirk and Russell, [40]). Up until now, relatively few high-quality studies have demonstrated that psychodynamic child psychotherapy approaches have positive treatment outcomes. One reason likely is that few such studies have been performed (Kazdin, [23]; Kennedy, [25]).

The primary aim of this study is to evaluate changes in children's global functioning after psychodynamic psychotherapy. A second aim is to identify factors that can account for possible changes of children's global functioning after psychodynamic psychotherapy. A third aim is to discuss the discrepancy between effect measures at group level versus analyses of development in individual therapies. The aim for the qualitative part of the study is to contribute to the understanding of the therapeutic process in the outcomes of child psychotherapies.

The questions that were evaluated are:

• 1. Does children's global functioning improve with psychodynamic psychotherapy?
• 2. What are the relationships between the psychotherapist's ratings of global functioning and the child's age and gender, diagnosis and comorbidity, and length and frequency of therapy?
• 3. How can an analysis of processes of therapy based on systematically collected qualitative data of the child-therapist interactions shed light on variations in outcome?

The present paper reports on a naturalistic, multi-centre pre-post study, evaluating children's global functioning after psychotherapy. Quantitative and qualitative methods are used in order to describe factors underlying change.

The target cohort of this study consisted of 33 children. The average child was 7.6 years of age with an age range from five to 10 years. Twenty-two participants were boys (66.7%) and 11 were girls (33.3%). A number of the children in this study had a high level of pathology. Twenty-nine children had at least one DSM-IV diagnosis, and 15 children had comorbid conditions (two ≥ DSM-IV diagnoses). Four of these had three diagnoses. The most frequent Axis I diagnoses at the start of psychotherapy were attention-deficit and disruptive behaviour disorders (n = 15), separation anxiety or reactive attachment disorder (n = 10), and anxiety disorder (n = 6). Four children did not receive an initial diagnosis. Since these children were judged to have poor global functioning according to the CGAS (<70), the lack of diagnoses for this group is probably explained by insufficient diagnostic routines at the mental health service during the time of the study.

Thirty-two psychotherapists worked with the children in this study. Of these, six were men and 26 were women. One psychotherapist had two children in treatment. The psychotherapists' experience of working as child psychotherapists ranged between one and 30 years. All psychotherapists were also health care professionals such as psychologists or social workers. The majority of psychotherapists had specialist training in child psychotherapy. Psychotherapists who had received basic training but not specialist training in child psychotherapy received regular supervision from an experienced child psychotherapist. All psychotherapists had a psychodynamic orientation.

The data that were analysed here were collected in the Erica Process and Outcome Study (EPOS). The EPOS evaluated goal-formulated, time-limited psychotherapy with parallel work with parents (Carlberg, [9]). The families that participated in the EPOS had themselves sought help for their children at child and adolescent mental health services in a number of Swedish cities and in Denmark (n = 4). The EPOS aimed to explore the complex clinical reality in child and adolescent mental health services through combining the study of process and outcome variables.

Child guidance clinics assessed the children following a standardised procedure and decided whether this form of psychotherapy was suitable for the specific child and his or her family. The following inclusion criteria were used. The child was between five and 10 years of age at the start of therapy. The child was given one or two psychotherapy sessions per week from between six months to two and a half years. Parents were expected to participate in their own psychotherapy once a week or at least once fortnightly. Goals and frames for the therapies were to be formulated and documented according to a detailed plan. Every third month, or in 10 cases every sixth month, data on process and outcome were collected using questionnaires completed by both the children's and parents' psychotherapists and also the parents.

About half of the children (45.5%) had one psychotherapy session per week, and the remainder had two sessions per week. The total number of psychotherapy sessions ranged from 20 to 152 sessions. Twelve children received fewer than 50 sessions, and 21 children received between 54 and 152 sessions (median = 68). The parents received between eight and 91 psychotherapy sessions (median = 39).

The psychotherapists and clinical teams carried out assessments according to a specified schedule. The following measures were used in this study:

Children's Global Assessment Scale (CGAS). The CGAS is a widely used measure of global adjustment including three different contexts of functioning: at home, in school, and with peers (Shaffer et al., [38]; Shaffer et al., [37]; Lundh et al., [30]). A 1–100 point measurement scale is used. All children were assessed by the child psychotherapists before and after treatment using information from both the therapy and external sources to rate global functioning. In a previous study of 20 psychotherapies, which were later included as a part of this study, good inter-rater reliability of the CGAS was obtained (ICC 2, 1 = .71) (Nolkrantz Frydman, [36]).

Hampstead Child Adaption Measure (HCAM). The HCAM is a 100-point rating scale. Its psychometric structure owes much to the CGAS instrument developed by Shaffer et al. ([37]). A manual was compiled to guide the rating procedure on one global rating of psychological functioning and 14 subscales.[1] HCAM is loosely based on Anna Freud's notion of developmental lines. The rater makes judgements on the child's adaptional achievements relative to their age, and physical and social circumstances. Fonagy and Target ([17]) describe the CGAS and HCAM scales as highly correlated.

DSM-IV (American Psychiatric Association, [1]) was used and primary diagnosis and co-morbid conditions as well as psychosocial and environmental factors affecting the child were documented.

Paired t-test for within-group comparison of CGAS and HCAM global scores pre and post treatment was used. The effect size of treatment was calculated by dividing the mean change score of CGAS and HCAM by the SD of the change score (Wright and Young, [46]). According to Cohen's ([11]) rule of thumb, effect sizes of 0.20 are considered as small, effect sizes of 0.50 as medium, and effect sizes ≥ 0.80 as large effect estimates.

In order to explore what variables may be related to change in the CGAS total score Pearson's correlation coefficient analysis was computed between the CGAS total score, demographic variables (age, sex), diagnosis and co-morbidity, and treatment variables (i.e. frequency and number of sessions, length in weeks of child therapy and number of sessions for parents).

A thematic analysis (Braun and Clarke, [6]) was carried out on the data from two therapies that were conducted by two child psychotherapists.

The first therapy was with a boy, David, who was eight years old at the start of therapy. The therapy's length was 18 months (94 sessions) at a frequency of twice a week. The parents attended parallel sessions at a frequency of once a week alternating between the mother and father. The therapist rated the CGAS change as 4 points on the scale (61–65).

The second therapy was with a girl, Frida. She was eight years old at the start of therapy. The therapy's length was 15 months (38 sessions) at a frequency of once a week. The parents attended parallel sessions at a frequency of once a week. The therapist rated the CGAS change as 30 points on the scale (45–75).

Qualitative descriptions were extracted from questionnaires that the child therapists completed every third month. The questionnaires were intended to give information about: the therapy's goals and content; important themes; attainment of the goals identified at the start of therapy; descriptions of the alliance between the child and the psychotherapist; the therapist's approach; significant episodes and changes in the therapy as well as circumstances outside the therapy that may have affected the development. Assessment of the child before and after the therapy was analysed using their DSM diagnoses, their CGAS and HCAM ratings as well as psychosocial and environmental factors affecting the child.

The first author undertook the thematic analysis and discussed the analysis in the research group to reach consensus. The thematic analysis aimed to identify, analyse and report themes within data, and used the steps recommended by Braun and Clarke ([6]) to do this. The procedure undertaken was (a) to read the therapists' report several times to become familiar with the data, (b) to formulate initial, tentative codes, (c) to look for themes that linked several codes together, or represented one exceptionally strong code, (d) to review the themes and (e) to define and name themes.

Because data were available from several different sources (therapist questionnaires; diagnosis; CGAS and HCAM ratings; psychosocial and environmental data) the data were processed through triangulation. The purpose was to illuminate topic issues from different angles to obtain a reliable and valid account of each of the two therapies.

The study was approved by the local ethics committee. Parents gave their informed consent. The two detailed accounts of psychotherapy have been anonymised and certain background facts changed in order to maintain anonymity, without altering the content.

A significant difference was found between CGAS pre and post treatment (p < 0.001; Table 1). Twenty-seven of the children had a post treatment change in CGAS of 10 points or more which is considered to be a significant clinical change. Before the treatment commenced, two children showed a CGAS score > 70. It should be noted that a child with a CGAS score < 70 is probably in need of child mental health services (Bird et al., [3]). In comparison, there were 20 children whose after treatment CGAS score > 70.

Table 1. Means and standard deviations of the CGAS and HCAM-global, pre and post treatment, p-values (t-test) and effect size (ES).

The overall effect size of CGAS change was ES = 1.80, indicating a large change (Table 1).

To identify variables that could predict the improvement of global functioning measured with the CGAS, bivariate correlations were calculated. The variables were age, sex, diagnosis, co-morbidity, frequency and number of sessions, length in weeks of child therapy, and number of sessions for parents. No significant correlations were found between any of these variables and CGAS change scores.

The effect size in HCAM global functioning was found to be ES = 1.98 which also indicates a large change in functioning (Table 1). The four subscales that showed the largest change were general mood, and variability of mood (ES = 1.83), ability to tolerate frustration and control impulses (ES = 1.54), development of confidence and self-esteem (ES = 1.42) and ability to cope with very stressful events (ES = 1.30).

The analysis of each of the two therapies is presented below, following a summary of the initial assessment of the child. For both therapies, the thematic analysis has identified a number of process development and change factors.

At the time of starting therapy, David was an eight-year-old boy who was adopted and had lived in Sweden since he was 12 months old. His parents had gone through a very trying separation with conflicts about custody of David. David's mother reported that her son's hyperactivity was a major problem.

In the psychological pre-therapy assessment it became clear that David had certain cognitive difficulties and more specifically he had dyslexia. According to CGAS, David's global functioning level was 61, which indicates a low level of general mental health functioning. The HCAM assessment showed that the areas in which David had the greatest problems were his ability to learn, to develop confidence and self-esteem, and to engage in positive relationships with his parents. He also had severe difficulties with tolerating frustration and controlling impulses. On the other hand, his relationships with siblings and peers, as well as adults outside of the family, were deemed to be good. Play, hobbies and interests were also areas without difficulties. The global HCAM score was 54. At the start of therapy David met the criteria for the DSM-diagnosis of attention/hyperactivity disorder. The assessment indicated that David had a strong maternal attachment and he overcompensated for feelings of abandonment with grandiose fantasies. The assessment suggested that David had attachment problems and difficulties understanding the subtleties and social conventions of communication with other people.

Before therapy the therapist was asked to identify three therapeutic goals. The child therapist formulated the following goals:

• • To establish a strong working alliance through providing a secure and predictable therapeutic frame.
• • To support David's ego development so that he would have the means to achieve comprehensible and meaningful experiences of his socio-cultural surroundings.
• • To provide means and opportunities to work through traumatic experiences that had affected David's inner world, which would enable David to use and develop his resources in the best way possible.

The psychotherapist reported that in the initial phase, David found it difficult to leave his mother in the waiting room, he had a rejecting attitude towards the therapist and he avoided eye contact. At the same time David seemed reluctant to end sessions and often said that he wanted to live in the therapy room. He made use of the materials in the therapy room and played war games with toy soldiers and made paper airplanes. In the early treatment phase his desire to be 'big' was a prominent theme and he expressed disgust at everything that he associated with being an infant like being dependent on others. David found it difficult to ask for help or use the therapeutic space in a cooperative way. The therapist described that she took a stance of patiently waiting and remaining one step behind David. He often refused to verbalise what he wanted and the therapist repeatedly emphasised that she was a separate person and clearly showed that she was unable to understand what David was communicating, or what he wanted, unless David himself talked to her about his thoughts and wishes.

After a few months of therapy David more frequently sought contact with his therapist and he also sought affirmation through eye contact. He latched on to the therapist's interpretations; he began to speak more spontaneously about things that were on his mind, and he also initiated playing together. David expressed and worked on issues to do with his negative self-esteem. A fear of failing and his concerns about being stupid became apparent. In contrast to this, he continued having compensatory fantasies where he engaged with superheroes in an omnipotent way. An ambivalence towards women surfaced and David moved between idealisation and denigration of them. David expressed his longing for closeness to a mother figure, but at the same time closeness to a mother figure also entailed feelings of fear and disgust. In play he portrayed battles between good and evil.

Gradually, the therapist became more direct in taking up David's potential to express his own will. She became more active in naming, offering interpretations and clarifying all his ways of expressing himself whilst maintaining a neutral position. It became apparent how David tried to integrate his conflicting feelings of fear and bravado. He showed an increasing interest in the therapist's observations and interpretations. In the encounter with David the therapist continued to be respectful and constantly receptive to his communications. She initiated his sharing of experience without being intrusive. When the play became destructive the therapist was tolerant to a degree but also set clear boundaries.

After a year of therapy, the theme of David's thoughts about his origins within his birth family and his feelings about adoption surfaced. He explored a recurring theme of feeling different from others and standing out which hindered him in his relationships. David also still struggled with conflicts to do with being a young child and at the same time feeling disgust and intolerance for childish things. At times he wanted to continue to show himself in the guise of impressive superheroes which contrasted with games on a more primitive level such as peek-a-boo. David gave eye contact increasingly more often and he also initiated the sharing of his experiences. A contact characterised by co-operation and trust developed between David and the therapist in which David expressed a wish to be physically close.

In her session notes, the therapist wrote about how she tried to give him her full attention and encouraged his initiative in talking about 'forbidden' and difficult things, for example, fears, violent thoughts and anger towards both his adoptive and biological parents. She described how she opened out this discussion and helped him to see other perspectives which deepened his understanding of his history. David showed more enthusiasm about attending therapy, though some ambivalence remained. The therapist also reported a change that occurred in the countertransference, namely that she began to feel more relaxed and more able to use humour, she was able to be playful and on the same level as David in games and play. The therapy was now characterised by more active contact from David who initiated contests and challenges; a mutual understanding and affinity were emphasised more and more.

In the termination phase David showed more aggressiveness in play and wanted to challenge, which he often expressed in an omnipotent, grandiose way. Existential themes such as life and death, issues to do with the future, as well as recalling his experiences in therapy and his feelings about ending were explored in the final therapy sessions. In this phase, there was a deep and trusting contact where the therapist was able to access many deep reflections but also needed to endure symbolic attacks. Towards termination the therapist's approach became increasingly affirming in character.

At the termination of therapy David's CGAS score was 65 as opposed to a score of 61 at the commencement of treatment. The therapist estimated that all the three goals that were set at the start of the therapy had been attained: David's difficulties in tolerating frustration and controlling impulses were no longer a major problem. His confidence and self-esteem, as well as his relations with his parents, had improved considerably. David had achieved a major change in his ability to engage in relationships with other people. He initiated contact which he was able to sustain and he expressed himself more clearly. The therapist perceived David to be more independent, relaxed and having a more optimistic view of the future; he was able to express a wish for dialogue and to share confidences; he was calmer and focused; he also had a better ability to express himself symbolically. However, the therapist reported that David still met the diagnostic criteria for attention disorder/hyperactivity. According to the HCAM global rating (65) David's major problem was now the ability to learn, which was to be expected given his dyslexia and cognitive difficulties. The psychosocial and environmental factors affecting the child as well as relationship problems between adults in the family, custody and access problems still had a negative impact on David's psychological well-being. David's maternal attachment remained strong and he overcompensated for feelings of abandonment with grandiose fantasies.

During the initial phase the therapist adopted an approach characterised by curiosity that was not intrusive. The therapist showed a genuine interest but tried not to hurry an understanding of David before she knew him well. The therapist maintained the therapeutic boundaries but let David make decisions about the content of sessions and his play was self-directed. The stance taken by the therapist showed David that in order to be able to speak with someone in a genuine way he needed to recognise that the therapist was a separate person in her own right. She introduced many measures that promoted mentalising and by doing this she laid the foundation for a growing inter-subjectivity. In addition, the therapist's affect attunement deepened the therapeutic relationship. Consequently, David became increasingly more interested in their contact and initiated playing together, which in its turn improved the inter-subjective exchange.

The capacity for agency was promoted by David discovering that he had to be an active participant in conveying his needs. The therapist invited him to express in words that he wanted something, to make his needs known. In conjunction with the building up of the alliance the therapist was gradually able to make more demands and to become increasingly more candid in her observations. For David, the therapist became a model of a non-judgmental adult who was able to set boundaries in a relationship while maintaining contact. She conveyed to David that they could be different and still respect one another, which reinforced David's capacity for individuation.

David was allowed to show his destructive side as the therapist conveyed that she was able to endure these aspects of him within safe parameters so that there was no harm done. The therapist kept herself available for close contact during these periods and conveyed that things could be seen from different perspectives.

Work was carried out in the therapy to support a healthy narcissism. David retreated to an early grandiose experience of self in order to then take a step towards a more age-appropriate way of behaving and perceiving himself. His longing for close contact with the therapist became constructive through a process of the therapist not permitting him to be physically close but instead recognising his needs and putting these into words. The therapist maintained her boundaries and in so doing she provided him with a sense of safety in the therapy. This in turn facilitated David in his exploration of his painful early history, the more problematic aspects of his experience and the ways he had developed to protect himself. In the therapeutic relationship, it became possible for him to share his inner world. By gently challenging him, for instance when he was relying on pseudo-grandiosity, the therapist was able to help the strengthening of David's self. The growing humour in the therapy can be regarded as a sign of a deep alliance.

When her therapy started, Frida was eight years old. Frida is a girl with a congenital physical disability. Her male therapist described how Frida presented herself as different from others and she was furious about her disability. She found it hard to make any social contact with her classmates and she often felt lonely. She self-harmed by frequently hitting herself. Frida reported that she did this to attract attention and thought it was better to hurt herself than to talk with others about what she found difficult.

At the start of the therapy Frida was living with relationship problems and somatic ill-health in the family. Her CGAS score was rated at 45 and she had major problems in her relations with her parents, siblings, peers and adults outside the family. Frida had developed poorly in terms of confidence and self-esteem and she had little ability to cope with stressful events. She also had severe difficulty with playing. Her low mood and frequent mood shifts, as well as her poor ability to tolerate frustration and control impulses, caused her significant problems. Her HCAM global score was rated at 39, which is generally indicative of major problems. Her parents reported that Frida's greatest difficulties were in her relationships with peers. They also suggested that she had major emotional problems. At the start of therapy Frida was found to meet the DSM diagnostic criteria for oppositional defiant disorder, anxiety disorder and identity problems.

At the start of therapy, her therapist formulated the following goals:

• • Frida needs to find other forms of expressing herself other than hitting and harming herself. She needs to be able to endure and moderate her feelings and to develop ways to express feelings in relation to her parents and other important people.
• • Frida needs to develop a better ability to play and to fantasise. With an improved ability to play Frida will have the opportunity to integrate thoughts and feelings, to work through anxiety and gain a better understanding of the context in which she lives.
• • Frida needs to increase her potential to think about and understand others in order to, for example, facilitate contact with friends. She needs to develop her inner representations and working models and attain a more developed mentalising ability.

At the start of therapy it was difficult for Frida to engage in any form of cooperative play. She wanted to be in control and tried to dominate the therapist. During the initial three months of treatment Frida drew; she brought photographs from home and played alone with the dolls' house. She sat in the sand box, and delightedly touched the sand, often in high spirits. Important themes were her communication of a strong wish to be allowed to be a baby and to give vent to her fury. Frida expressed her feelings of rage with regard to her disability and her anger towards her parents. She also expressed strong jealousy towards all younger children and infants and she was unable to tolerate sharing her mother's attention. Slowly as Frida's attempts to control the therapist diminished, the contact between them deepened which led to an increased strengthening of positive feeling. The therapist reported a calmer atmosphere in sessions and the sense of scope for development of thought and feeling. However, he described Frida as highly observant and she carefully read the therapist's facial and emotional expressions as if acutely concerned about his reactions to her.

As the therapy continued, the therapist reported that an important focus was to make distinctions between positive feelings and more destructive communications from Frida that began to emerge more in sessions. He set clear limits for Frida's destructiveness and gradually Frida's preoccupation with longing to be a baby again emerged and was closely explored with her. A parental figure that was able to take care of her became important in the therapy and Frida seemed happy to have the exclusive attention of the therapist to explore all her infantile feelings. The rage about her disability continued to be a strong theme. She oscillated between the thought of harming herself or harming her parents. The therapist represented strong boundaries that were containing for Frida in order for her to explore these issues.

Over time there was a decreasing need to establish boundaries as Frida became calmer and also more content. She became more able to express some of her strong feelings about a number of concerns and the therapist received Frida's painful material, which opened up the possibilities for new development and play. For instance, she became able to share with the therapist her experiences of contact with other children and she explored her feelings about being someone who was often teased, questioned and bullied. A wish emerged at this point to be the age she actually was in contrast with the wish to be baby.

Important themes that later surfaced in the therapy were to do with the body, its ability and skill as well as its disability. Frida presented increasingly as a nine-year old girl with age-appropriate reflections. She related to herself in a more nuanced way and was more trusting in her relation to the therapist. Frida often seemed more satisfied; she expressed joy and had a lighter mood. She ceased hitting and harming herself, although she continued to have substantial difficulties in modifying her feelings of hate and envy. However, the therapist experienced that their relationship and the way of being together gradually became more stable and the sessions contained longer periods of calm. Frida became able to express warmer feelings in her relation to her parents. The therapist described how Frida often expressed genuine joy with the playing that began to emerge in both the therapy sessions and in everyday life. Frida established more co-operative relationships in her external world and she was able to reflect and think about other people with empathy. Her disability no longer dominated all her interactions with her environment.

A gravity and sadness emerged in Frida after about a year of therapy. Frida's thoughts and feelings about her disability were expressed in a more integrated and calm way. Dreams and longing emerged about being able to walk and run better. She was mourning the body she did not have. In the psychotherapeutic work, the therapist and Frida met on a deep and serious level to explore these issues. The need for immediate affirmation, strong containing and clear setting of boundaries abated.

In the final months of the therapy the focus was on the fact that the therapy would soon be terminated and Frida brought up thoughts about contact with friends and talked of her dreams for the future. The contact between Frida and the therapist had become increasingly easy, sometimes happy and even a little mischievous. Frida was no longer so scrutinising and attentive in the contact with the therapist. She felt accepted by him which freed her to be more occupied with herself and her life outside of the therapy.

At termination of the therapy Frida's CGAS score was 75 in comparison with a score of 45 at the commencement of treatment. She was no longer weighed down to the same high degree by the many psychosocial and environmental factors that were present at the start of therapy. According to her HCAM assessment after treatment, her problems before therapy had diminished. For instance, her relations with peers and adults outside the family were no longer a substantial problem for Frida. In addition, she had discovered a new ability to play. Her severe difficulty in relating to her parents had greatly diminished. Frida's ability to cope with stressful events and her ability to tolerate frustration had improved to a level described as manageable.

The problems concerning her development of confidence and self-esteem, mood and mood shifts had also diminished considerably, although they remained to some extent. Frida also had a better learning ability. After the therapy the HCAM global score was 81. Frida was no longer deemed to have any of the three DSM-IV diagnoses she had at the start of therapy. The therapist reported that all three goals set at the start of therapy had been attained and he described Frida as essentially functioning well. Frida exhibited more joy, and greater hope and longing for the future. She had better self-esteem and stability and was also able to see herself in relationships and compare herself with friends in a favourable way. She had established significant reciprocal friendships and was able to play better. Her disability was no longer as traumatic for her and it became possible for her to talk about it. Frida and her parents were able to relate to each other on an emotional level and Frida no longer required her parents to affirm her all the time. She was able to express and share strong feelings and cope with both internal and external reality.

In the therapy the therapist aimed for Frida to gain access to and better understand her feelings. The therapist showed that conflicting feelings, both positive and negative, could exist at the same time. He helped Frida to broaden and deepen her emotional register. In the therapy, Frida received help to express her strong negative feelings without having to act them out. The therapist showed with empathy and understanding that he was able to withstand her rage, but set a limit for the physical, concrete enactment of it, thus creating security in the therapeutic setting.

Through a process of sharing Frida's feelings about her key preoccupations, the therapist provided a basis that supported the therapeutic alliance. A benevolent spiral was created gradually in which the therapist felt that he did not need to set limits, which in turn led to Frida being able to rest psychologically.

Frida's need to carefully read the therapist's expressions can be compared with the need of a small child. Here, the therapist behaved like a parent would towards a much younger child. He was containing, receptive and at hand to receive Frida's painful material which opened the way for new development and play. The therapist became both a model and a repairing parent gestalt who gave psychological closeness. Thereafter in the therapy it was possible to discern a qualitative leap where Frida was perceived as age appropriate and increasingly nuanced whilst at the same time, contact between her and the therapist became increasingly relaxed. As her trust in the therapist increased Frida gained a new platform for expressing her deep sadness and this became more processed. Her feelings about her disability were not as central in her emotional experience as it had been previously. It became possible for relationships with other people to come more into focus and Frida could start making use of and develop her ability for empathy.

In the therapy much work was centred on Frida creating new inner working models. This work made it possible for her to cope with her difficulties in a new way, which meant that her self-image changed and her self-esteem was not as fragile as previously. This resulted in Frida not being as dependent on external objects for immediate affirmation; she had developed some inner resources and she became more occupied with herself and her life outside of the therapy in an ordinary way.

A major finding of this study was that the children's global functioning, measured with the CGAS and HCAM, improved substantially after psychodynamic psychotherapy with parallel work with the child's parents. This finding, obtained with two independent measures, yield some support for the benefits of this treatment approach.

The four subscales of HCAM that showed the largest change were general mood, and variability of mood; ability to tolerate frustration and control impulses; development of confidence and self-esteem; and ability to cope with very stressful events.

To identify variables that could predict the improvement of global functioning (CGAS) correlations were calculated. No significant correlations were found between CGAS change scores and age, sex, diagnosis, co-morbidity, frequency and number of sessions, length in weeks of child therapy, or number of sessions for parents. The associations between the child's global functioning and demographic, diagnostic and treatment variables could be examined in future research, using bigger samples.

Qualitative analyses of two of the therapies were carried out in order to obtain an in-depth understanding of the findings. The thematic analysis of one of the two therapies that was selected (David) shows how major improvements that are achieved in therapeutic treatment may not be captured in psychometric measures, for example, the CGAS. The positive change that took place in David in several areas of functioning and experience, such as increased independence, greater happiness and sense of relaxation and an optimistic view of the future, could be described as more psychological than psychiatric. Therefore there is a risk of these changes not being picked up as critical outcome variables in a diagnosis-centred child psychiatry framework. The therapy with David is an example of how a child with an attention/hyperactivity disorder can greatly benefit from child psychotherapy despite the diagnosis remaining after the termination of therapy.

In the other psychotherapy (Frida) there was an even wider scope for improvement as the child had a low score on the global ratings prior to therapy. This therapy also focused upon affects including promoting impulse control, frustration tolerance and self-confidence as well as creating an inner world, which in turn increased her inter-subjectivity. These improvements can contribute to a continued positive effect of the psychotherapy long after its termination. Access to affect and inter-subjective relationships as well as the ability to play should promote the child's natural psychological and emotional development.

The two case descriptions, where the CGAS change rated small or large respectively indicate that:

• 1. The complexity in the therapy process emerged in the analysis. It was difficult to capture a comprehensive picture of development after psychotherapy with simple outcome measures or with psychiatric diagnoses.
• 2. Even in the therapy where the therapist rated a small CGAS change it was apparent that the child benefited from the therapy and developed in important areas.
• 3. The inter-subjectivity between the child and psychotherapist was essential. The contact embraced aspects containing both transference and relating to a real object that together laid the foundation for the therapist to be able to come close to the child on a deeper level.
• 4. The therapist's approach and interventions could be characterised as creating a firm therapeutic frame that involved a human companion where inter-subjectivity could be tested.

An important strength of this study was that it was performed in a naturalistic setting. The child and adolescent mental health services used clinical criteria to select the patients. The children and parents that were selected were judged to be motivated for the significant commitment that this treatment involved. Thus, we do not know how representative these families are of the population of families that usually seek help at child and adolescent mental health services, or of the families to whom the services usually offer child psychotherapy. A clinical evaluation would suggest that the families that participated in the EPOS study represent the broad variation of psychological problems that are found at child and adolescent mental health services. Thus, in contrast to many efficacy-studies, the present study examined children with complicated problems who were selected for psychotherapy. An important task for future child psychotherapy research is to create a better understanding of external factors, such as family and social environment, and their importance for process and treatment outcome. What constitutes a good match between psychotherapist and child and his or her family? Future studies could examine this question and identify the individual psychotherapist's contribution to the development and changes in psychotherapy.

A clear limitation in this study was the lack of a comparison group. To gain a better understanding of the magnitude of the good outcome obtained, it would have been interesting to compare treatment outcome of goal-formulated, time-limited psychotherapy with another treatment. Another weakness was that the psychotherapists themselves rated the children's global functioning with the CGAS. The therapies are thus observed from the psychotherapists' perspective. However, a study of CGAS inter-rater reliability that was based on part of the data used in this study supports the reliability of the ratings (Nolkrantz Frydman, [36]). A third weakness was that the study group was small and therefore a replication study based on a larger data set needs to be undertaken.

It is important to study the mechanisms behind therapeutic change, as there is not a simple connection between process and outcome in psychotherapy. Each psychotherapeutic encounter creates a unique process. Child psychotherapists with the same training can differ significantly in their views of the therapeutic process, the use of therapeutic technique and their approach in the encounter with the child. By giving increasing attention to the therapeutic relationship as a significant change factor, research on variations and differences in therapists' ways of working become important in understanding what happens in child therapy. We would like to highlight the importance of emphasising generic knowledge in psychotherapy training programmes, knowledge that cannot be connected to a particular therapeutic school of thought. It is important that every prospective therapist is given the opportunity during training to examine and develop awareness of her/his own personal psychotherapeutic approach and thereby integrate this with a stable theoretical basis.

This study will hopefully promote research interest amongst clinically active therapists. There is a gap between academic research and clinical work that needs to be bridged. A more clinically integrated research and developmental work setting would fertilise the whole child therapy field.

To conclude, this study suggests that psychodynamic child psychotherapy can be beneficial. These findings should be followed up in studies combining qualitative and quantitative methods to increase our understanding of processes and mechanisms behind these changes.

This work was supported by Marcus and Amalia Wallenberg's Foundation and the Gålö Foundation. We wish to express our gratitude to Katarina Nolkrantz Frydman, Jenny Sima and Agneta Thorén for their contributions to this study.