U.S. flag

An official website of the United States government

NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Cover of Psychotherapy: Three approaches evaluated

Psychotherapy: Three approaches evaluated


This document presents a review of the work of the expert group convened by INSERM through the collective expert evaluation procedure to answer the questions raised by the Direction générale de la santé (DGS; general directorate of health) on the evaluation of psychotherapies.

It is based on the scientific information available the last 6 months of 2003. The document base for this expert evaluation consisted of approximately 1,000 articles and documents.

The INSERM collective expert evaluation centre co-ordinated this collective work with the Département animation et partenariat scientifique (DAPS; department for facilitation and scientific partnership) to instruct the dossier and with the documentation service of the Département de l'information scientifique et de la communication (DISC; department for scientific information and communication) for the literature search.


Psychotherapies are widely used treatments in healthcare practice for mental disorders in adults, adolescents, and children. They are used alongside drug therapy for some severe disorders (schizophrenia, bipolar disorder, etc.) and as an alternative to drug therapy for other, less severe disorders, or for those in which drug therapies are not used (for example, personality disorders).

In France, psychotherapies are generally recommended to patients by medical psychiatrists, psychologists, general practitioners, or other healthcare professionals, although spontaneous requests also occur. The percentage of these latter cannot be quantified because there are no data available on this subject. Psychotherapies are usually practised on an outpatient basis in the setting of care from psychiatrists and psychologists and on an institutional basis by different parties (nurses, psychologists, etc.), often under the responsibility of a psychiatrist. Psychotherapies are not included in the listing of technical procedures in French healthcare regulations, with the exception of group therapies. A category "psychiatric consultation", which does not specify the type of care administered by the psychiatrist in the consultation, does, however, exist.

On an international scale and according to published scientific works, psychotherapies are performed by psychiatrists and psychologists, and to a lesser extent in the United Kingdom and United States by specialist nurses (nurse therapists), social workers, or specialised counsellors and by students as part of psychotherapy research projects, under close supervision. Finally, in some research work, reference is made to general practitioners who have received brief training in applying methods that have already been tested and are suitable for general medical practice in healthcare or prevention.

Like other treatments, much scientific work has been conducted on the different psychotherapy methods. Some of this work has sought to evaluate the effectiveness of the practices under different conditions.

In the mental health plan implemented by the Minister of Health in 2001, the Direction générale de la santé (DGS; general health directorate) approached INSERM to produce a current overview of the international literature on aspects of evaluating the effectiveness of different psychotherapeutic approaches. Two French associations, Unafam1 and Fnap-psy,2 worked with the DGS in this approach. With the agreement of these partners, the scope of the expert assessment covered three major psychotherapeutic approaches—the psychodynamic (psychoanalytical) approach, the cognitive-behavioural approach, and family and couple therapy—often used to care for defined disorders of adults, adolescents, or children.

To respond to this request, INSERM convened an expert group in a collective expert evaluation procedure. The group consists of psychiatrists, psychologists, epidemiologists, and bio-statisticians. This expert group structured the analysis of the international literature around the following questions.

How can we envisage evaluation of psychotherapies in terms of efficacy?

What are the different types of studies that enable assessment of the efficacy of the psychotherapies?

What are the methodological difficulties encountered in such an evaluation?

What are the historical stages of the assessment of efficacy of the psychotherapies?

What are the theoretical references for the psychodynamic (psychoanalytical), cognitive-behavioural, and family approaches?

What information is present in the literature about assessment of the efficacy of the psychodynamic (psychoanalytical), cognitive-behavioural, and family approaches?

What information is present in the literature about the comparative assessment of the efficacy of these different psychotherapeutic approaches?

What information is present in the literature about evaluation of the efficacy of these three psychotherapeutic approaches for different diseases?

What information is present in the literature about evaluation of the efficacy of these psychotherapies in children and adolescents?

We collated more than 1,000 articles from an independent interrogation of the international databases conducted by the collective expert evaluation centre. The experts were asked to supplement this bibliography within their own field of competence and within the scope of the objectives of the expert assessment. The experts presented a critical analysis and review of the published work on international and national scales on the different features of the scope of the assessment during 11 working meetings that were organised between the months of May 2002 and December 2003.

Expert advisory group and authors

Expert group and authors

Olivier CANCEIL, University Hospital Department of Mental Health and Therapeutics, sector 75G14, St. Anne hospital centre, Paris

Jean COTTRAUX, Anxiety Treatment Unit, Pierre Wertheimer Neurology Hospital, Lyon university hospital centre

Bruno FALISSARD, PSIGIM, Faculty of Medicine, Paris XI, Department of Public Health, Paul Brousse hospital, AP-HP, Villejuif

Martine FLAMENT, Mental Health Research Institute of the Ottawa University, Ottawa, Canada

Jacques MIERMONT, Federation of Family Therapy Services, Paul Guiraud Specialist Hospital Centre, Villejuif

Joel SWENDSEN, Clinical Psychology and Psychopathology Laboratory, Victor Segalen Bordeaux 2 University, Institut universitaire de France

Mardjane TEHERANI, Department of Psychiatry, Bichat-Claude-Bernard University Hospital Centre, Paris

Jean-Michel THURIN, Psychiatrist, Paris

The following presented communications

Daniel WIDLöCHER, Pierre and Marie-Curie University, Paris

David SERVAN-SCHREIBER, Clinical Psychiatrist, Faculty of Medicine, Pittsburgh University, United States

Ivy BLACKBURN, Cognitive and Behavioural Therapies Centre, Newcastle upon Tyne, United Kingdom

Scientific and editorial co-ordination

Fabienne BONNIN, Scientific associate at the INSERM Collective Expert Evaluation Centre, Xavier-Bichat faculty of medicine

Catherine CHENU, Expert evaluation Lead at the INSERM Collective Expert Evaluation Centre, Xavier-Bichat Faculty of Medicine

Jeanne ETIEMBLE, Director of the INSERM Collective Expert Evaluation Centre, Xavier-Bichat Faculty of Medicine

Catherine POUZAT, Scientific associate at the INSERM Collective Expert Evaluation Centre, Xavier-Bichat Faculty of Medicine

Literature and technical assistance

Chantal RONDET-GRELLIER, documentalist at the INSERM Collective Expert Evaluation Centre, Xavier-Bichat Faculty of Medicine


An evaluation of the effects of the psychotherapies appears to be required to guide public health decisions and fulfil the wishes of patients who want to know how effective the treatments offered are.

The expert group has analysed three psychotherapy approaches from the work available in the literature providing the basis for a scientific evaluation of their efficacy: the psychodynamic (psychoanalytical) approach, the cognitive-behavioural approach, and the family and couple approach. These psychotherapies have in common their length of use and the solid nature of their theoretical conceptualisation, the existence of specific training in their practice by clinicians, and their widespread use within the field of healthcare.

A second a priori option of the expert group was to direct its attention toward application of these psychotherapeutic methods to the treatment of mental disease in adults, a field in which the literature is most advanced in terms of efficacy studies. The following disorders were considered in this expert evaluation: anxiety disorders, mood disorders, schizophrenia, eating disorders, personality disorders, and alcohol dependency. In addition, the group was also careful to consider and report on specific work conducted in the age bands for each type of disease that is also seen in children or adolescents. Similarly, some work relates to autism and other invasive developmental disorders, hyperactivity, and conduct disorders.

A scientific assessment of the efficacy of psychotherapies assumes: first, that the characteristics of the patients are included (what diseases are the studies based on, level of severity of the disease and possibly its co-morbidities); and second, that the levels of improvement of the patients at the end of treatment are known. The description of the target diseases and definition of the objectives of treatment may differ depending on the studies and according to the underlying theoretical frameworks of the psychotherapeutic approaches. This may make comparisons between treatments difficult. Nevertheless, insofar as a therapy is proposed for a given syndrome, improvement in the syndrome represents a common standard to assess the different therapies.

A number of factors may influence the course of a psychotherapy and therefore its assessment: the nature and severity of the disorder, life events, family and social environment, the placebo effect, the treatment method or technique used, the therapeutic relationship with positive or negative combination effects of treatments, and biological changes. These features are also described in the analysis of the different approaches.

How do we see the methodological problems in evaluating psychotherapies?

The scientific assessment of a therapy raises at least three methodological questions: What is the definition of the population of patients to be treated? How do we measure the efficacy of the therapy? How do we prove this efficacy?

The definition of the population of patients treated (equivalent to conventional inclusion and exclusion criteria) partly determines the clinical use of the results from the studies. First, diagnoses must be used that are as close as possible to the widest consensual terminological definitions so that the conclusions drawn from the study can be easily generalised. Second, it is essential that the diagnostic process be conducted with a minimum of ambiguity to guarantee the reproducibility of the experiment. In practice, these two constraints are often difficult to reconcile: the diagnosis categories constructed from an optimal reproducibility basis such as the DSM (American association of psychiatry) or CIM (World Health Organisation) are not necessarily those that are most widely used in everyday clinical practice, particularly in France. Although the disorders examined in this evaluation from the available literature are among the most common, we must add that a number of varied psychological disorders and symptoms seen in psychotherapy cannot be categorised in terms of the syndromes or diseases that have already been defined.

The choice of efficacy measurement is undoubtedly the most important methodological issue. This raises several questions, the first being to determine whether it is legitimate to use quantitative measurements to describe the improvement in a patient during psychotherapeutic management. Compared to the very great complexity of the subject, we must be very cautious about the value we attach to these measurements in the field of psychotherapy. The measurements in reality are only the numerical representation of a characteristic. Commonsense suggests that as an initial approximation, we should observe whether a patient is "more" or "less" improved in a given feature of their functioning. A numerical system can then be used to grade the clinical improvement. Furthermore, we must of course be sure that this measurement reliably reflects the improvement, in other words the measurement must be valid. Although this measurement deals with subjective and not objective characteristics, such a subjective measurement of efficacy can be validated. The measurement is, however, always subject to the defining theory that its designers have explicitly or implicitly used. This point is essential, because if we are considering an assessment of psychotherapy, bias may exist because of either antagonism or congruence between the defining theory of the measurement instrument and the theoretical support of the psychotherapy being studied.

The question of proof of efficacy is linked to the partly random nature of any patient's response to a treatment. If a difference in efficacy is observed between two groups of treated patients, the question that arises is whether this difference is or is not compatible with spontaneous variations in efficacy that are seen between patients for the same treatment. This problem is often resolved in practice using random allocation of treatments and a statistical test to determine the significance of the difference in efficacy. This unavoidable use of statistics assumes that the effect being studied is reproducible. In the case of psychotherapies, because the patient (or the patient–therapist couple) is individual in his/her path through life and normal mental functioning: in that situation how can we envisage reproducible experiments? In reality this question exists for any demonstration of treatment efficacy. If, for example, we assess the efficacy of an antibiotic in the treatment of pulmonary tuberculosis, the study will have to last for one or more years. If an investigator wishes to reproduce the study once the results have been published, it is possible that the ecology of the organism has changed and that the experiment would no longer still be entirely the same. The concept of reproducibility in clinical research is weakened compared to the classical experimental sciences such as physics, chemistry, or biology. This relative weakening, however, is not enough to make a scientific process unusable, as was clearly shown by advances in our knowledge of treatments during the 19th century.

Overall, there is no clear conceptual restriction to the use of scientific evaluation of efficacy for a psychotherapy. It is possible to envisage testing the hypothesis of the efficacy of these treatments in the context of refutable, reproducible experiments. Subtleties must, however, be introduced into the confirmation: first, the reproducibility of the study is not 100%, although this problem is not specific to the field of psychotherapies; second, the definitions of the patients to be treated are not always consensual; this may occasionally hinder the clinical use of the results. Finally, the studies are only interpretable if the measurements taken are valid. In the context of psychotherapies, these measurements are sometimes based on a subjective "phenotype", and their validity is easier to demonstrate if they are incorporated into a theoretical field that is compatible with the psychotherapy being studied. The criteria for assessment of efficacy most frequently used in the literature involve symptomatic factors that cannot claim to be universally relevant. This is undoubtedly a limitation, although we must recognise that the symptom remains an irrefutable factor in the state of patients' mental health.

What are the different types of studies taken from the literature to evaluate the psychotherapies?

Evaluation of the efficacy of a treatment relies on comparative controlled trials. It is practically impossible, however, to compare an active psychotherapy with a psychotherapeutically inert "placebo" in a double-blind, randomised trial based on the models of pharmacological trials, because relationship and situation effects and the expectations of the therapists and patients are active components in any psychotherapeutic system. Different types of "control group" are therefore found in trials, such as the "placebo attention" group, in which the patients have minimal contact with the therapist, who does not use the factors that are assumed to be active in the therapy in question. This helps to remove simple patient management effects. We also see, as a comparison group, the group of patients still on the waiting list, which only receives simple telephone contact for several months (this may raise ethical problems and often results in drop-outs to another therapy). It is common in trials to compare a therapy to the "treatment as usual (TAU)" of the trouble in question. In addition, we must recognise that comparison of psychotherapy with a chemotherapy may produce bias in favour of the psychotherapy if the patients in the study have almost all received chemotherapies without effect (and may, for this reason, come looking for another treatment).

Double blinding is only possible in evaluation of psychotherapies in the situation where a psychotherapy is compared with medical drug treatment versus the same therapy against a placebo. In this situation, the evaluation concerns the interaction between the psychotherapy and medical drugs.

To answer the difficulties of independent and blind evaluation of the test hypotheses, some studies measure the patients' beliefs and those of their therapists in the treatment to which the patient was randomised at the start of treatment and then study the correlation between these measurements and the results. The psychotherapeutic placebo must have characteristics that make it as similar as possible to a genuine therapy: the placebo must be credible.

A certain number of factors relating to the attitude and behaviour of the therapist toward the patient have long been considered therapeutic. Force of persuasion, the ability to create a family atmosphere, warmth, empathy, genuineness of sentiment, and a positive view of the patient have all been reported. To this must be added social-occupational status, credibility, setting, and renown. These factors have not been greatly studied empirically. The most recent studies refer to the "therapeutic alliance", which describes the nature and quality of the interaction between the patient and the therapist. The therapeutic alliance in analytical therapy forms the context in which the "transference" may be expressed (the bringing out of unconscious desires and of the problem that is central to the cure). This relies upon mutual involvement of the psychoanalyst and patient in seeking out the causes of the disorder and planned future changes using the cure process. In cognitive-behavioural therapy, the therapeutic alliance describes an empirical relationship of empirical collaboration between the patient and therapist (similar to that of two scientists working on a common problem), which is used for the basis of learning, leading to cognitive changes in the person. The "therapeutic alliance" relationship defined in this way is a necessary but not sufficient condition for change. In family therapy, the therapeutic alliance is based on respecting styles of interaction, value and belief systems, forms of family knowledge and know-how, and the construction of hypotheses that can be changed, depending on the experiences shared.

In terms of the measurement of the effects of psychotherapies, the studies use many evaluation scales of symptoms, behaviour, and methods of peoples' psychological and interpersonal functioning. These scales have been validated for various psychopathological problems. They may be completed either by the clinician or by the patient him/herself. Personality questionnaires or ad hoc measurements are also found in some studies, depending on the hypotheses tested. In vivo behavioural tests provide a direct measurement of a person's performances and may be very different from the evaluation scales.

Correctly conducted evaluation studies report many criteria and measurements, allowing the range of conclusions to be extended. Some of these studies also analyse features of the therapeutic process in detail. Alongside changes in scores from continuous scales, the studies refer occasionally to general discontinuous criteria, criteria for good results, or "end point criteria". A single positive/negative end point criterion (success/failure) may be used, or alternatively a principle end point and secondary end points.

Statistically significant changes measured in a group using a scale may occasionally only reflect mediocre clinical results, the mean value of which is sufficient to make statistical tests significant if the statistical power is high because of the large number of patients included. Conversely, lack of change in the mean value of a scale score may more rarely be accompanied by clinically beneficial changes in some patients or in a subgroup of patients. Expressing the magnitude of the effect obtained in the "average" subject in the study for people receiving the treatment or its comparator (placebo or other treatment) is close to the effect size and is necessary information in addition to the classical statistical tests.

Evaluation criteria for controlled therapeutic trials (from Foa and Meadows, 1997 – revised by Maxfield et al. 2002)

Criteria Score/10
Clearly defined symptoms0 0.5 1
Validated measurements0 0.5 1
Independent blind evaluator0 0.5 1
Trained reliable evaluator0 0.5 1
Treatment present in a manual0 0.5 1
Randomisation0 0.5 1
Compliance with treatment0 0.5 1
No other concomitant treatment0 0.5 1
Multi-modal evaluation measurements and interviews0 0.5 1
Optimal length of treatment0 0.5 1

Meta-analysis is a quantitative approach to a literature review that estimates the magnitude of the effect obtained in the "treated subject" compared to the "control subject" from the effect size. This analysis is based on the concept that all of the studies represent a quantum of information connected to the aim of the research subject and that each study provides its own contribution. The assumption made is that all of the studies included represent a sample of all possible studies on the subject in question.

Meta-analysis therefore involves combining the studies, coding the results, and calculating the amplitude or treatment effect size. For a given criterion that is being studied at the end of treatment, this represents the difference in mean values 3 between the study group and its comparator group (control or other treatment group). The criterion is generally the score on an evaluation scale. The effect size therefore reflects the possible gain by the treatment group compared to the control group. An effect size is considered to be small between 0.20 and 0.50, average between 0.50 and 0.80, and large above 0.80. In some studies, the effect size is calculated by comparing the pre- and post-treatment scores. 4 This effect size tends to produce a higher value than the comparison of active treatment versus control situation because the placebo effect is not subtracted.

Distribution of scores and effect size.


Distribution of scores and effect size.

The aim of meta-analysis is essentially to solve the problem of discordant results by providing more detailed information about the magnitude of the effects. It should also be useful for identifying responders. Starting from these points, the quality criteria for a meta-analysis can then be defined.

Proposed golden rules to evaluate the quality of meta-analyses

Criteria Score/7
Inclusion of all quality studies on the subject0 0.5 1
Clearly defined assessment criteria (end points)0 0.5 1
Use of appropriate statistical methods0 0.5 1
Taking statistical power into account0 0.5 1
Comparison of effect sizes0 0.5 1
Test of study comparability0 0.5 1
Estimation of unpublished studies0 0.5 1

What are the different forms of the psychodynamic and psychoanalytical approach?

The psychodynamic approach brings together practices ranging from traditional psychoanalysis to psychodynamic (psychoanalytical) psychotherapies, both long and brief. The general underlying psychoanalytical theory to these psychotherapeutic practices is common to them all.

Psychoanalysis was born at the end of the 19th century with Freud's work on hysteria and interpretation of dreams, etc. From its very origin and through the precise description of psychological effects and the formulation of hypotheses about the subconscious mechanisms underlying them, Freud was intending to integrate psychoanalysis into the scientific approach. Psychoanalysis now has a century of history of contribution to psychiatric care. It developed more intensely after the Second World War. The stages in development of psychotherapy have been characterised by clinical variants linked to the evolution of models or terminological classification factors.

The psychodynamic (psychoanalytical) psychotherapies stress a patient's awareness of the psychological conflicts he/she is suffering from, together with acquisition of new psychological and developmental abilities (relating to construction of self and symbolisation). All psychotherapies of this type are based on psychoanalytical theories, including transference, although they may differ depending on the submodels to which they refer, the specific objectives they are aiming for, and the specific techniques used in reaching these objectives. These psychotherapies adapt to the characteristics of the patient, which are identified through the expression of the transference itself. They are generally broken down into several stages according to the level of psychological organisation of the patient and the relationship interactions established.

Long-term psychotherapies can be distinguished from the brief or short-term psychotherapies (40 or fewer sessions). These latter types of psychotherapy have developed more recently in the United States. They may be centred on an event, or alternatively may be interpretative and centred on personality. The aim of treatment is to acquire insight or to obtain a change in personality, and the techniques used stress interpretation work and analysis of the transference. Focal psychotherapy identifies a central conflict present since childhood, reactivated during adult life and forming the origin of the problem. In this case, the aim is to resolve this problem through a relationship with the therapist, providing new opportunities for emotional assimilation and insight.

Psychoanalytical psychotherapy is a long-term process conducted with a trained psychoanalyst, involving several sessions per week over a period lasting at least one year. These sessions allow expression of subconscious conflicts and addressing fixations (libido-related and ontogenetic) that are brought to the surface in the transference relationship with the therapist. Through the construction of the analytical space and interpretation they lead on to work in a (re)constructive process designed to change psychological structure and organisation.

The different techniques in the psychodynamic (psychoanalytical) approach

Technique Definition
PsychoanalysisInvestigation method involving identification of subconscious significance of a person's words, actions, imaginary constraints (dreams, fantasies, delusions). The method is based on the person's free associations, which guarantees the validity of the interpretation.
Psychoanalytical therapyPsychotherapeutic method based on investigation involving identification of subconscious significance of a person's words, actions, or imaginary productions. This method is characterised by controlled interpretation of defence resistance, transference and desire mechanisms, and dynamics in the identification processes.
Brief psychodynamic therapy (on average 12 sessions, at a frequency of one session per week)Specific therapeutic intervention involving a specific "state" or "problem" to obtain a change in the state or resolution of the problem.
Interpersonal psychodynamic therapy (10 to 12 sessions)The emphasis is placed on the patient's psychosocial and interpersonal experiences.

Therapists' training takes account of the diversity of practices, which range from the psychoanalytical cure to more directive techniques. The term "training" in the psychodynamic (psychoanalytical) approach refers more to transmission of a practice than to communication of knowledge. The practitioner learning to perform psychoanalyses or psychoanalytical psychotherapies must acquire: associative listening allowing him/her to adapt to different techniques; the ability to develop a specific framework for the psychodynamic work; and the ability to define the best suited therapeutic indication to evaluate the person's psychological structure.

Training of psychoanalysts is based on three arms: personal analysis, supervision of cases treated by the candidate, and theoretical training in educational institutes. These institutes are more or less closely linked to the psychoanalysis societies. At the origin of this, the Association psychanalytique internationale (international psychoanalytical association) created in 1910 had the aim of underpinning the principles of training recognised by all of its members. Many divisions and a more decentralised view of the training rules within the association led to a degree of variability in how these principles were applied. Currently in France, psychoanalysts refer to several theoretical frameworks (Freudian, Alderian, Jungian, Kleinian, Lacanian, etc.). These practitioners are grouped into two associations which form part of the Association Psychanalytique Internationale: The Société Psychanalytique de Paris (Paris Psychoanalytical Association), and the Association Psychanalytique de France (French Psychoanalytical Association). In addition, the Lacanians belong to different associations, including one international association. This diversity has led to several methods of training and practice.

Personal analysis is a strict prerequisite to become a psychoanalyst, although its final outcome and practice differ considerably from one "institution" to another. These differences relate in particular to the nature: therapeutic or strictly didactic, or designed to promote the psychoanalytical experience. Over time, these differences have become so great that the different "institutions" no longer necessarily see themselves within a common training.

Supervisions are designed to familiarise the candidate to the practice of psychoanalysis. The objective is neither pure technical training nor a form of psychotherapy but to enable the candidates to transpose the experiences that they have acquired during their analysis into their practice as an analyst.

In terms of theoretical training, the general rule in France is for training outside of any academic framework: free choice of teaching, research training courses and book work, no validation of knowledge. In other countries, the training is often closely derived from university methods or even integrated into university education.

Assessment of the candidate before the start of the personal analysis is scarcely performed anymore, particularly in France. Assessments before performing supervised cures, at the end of each of the cures, and at the end of the course generally lead to the candidate being admitted to a psychoanalytical association.

The entire training, including the personal analysis, lasts for between 5 and 8 years and provides a qualification to people who have already followed university and clinical education.

Long-term psychotherapies run over several years and are applied to complex diseases such as, for example, the severe personality disorders (notably borderline). In borderline personality disorder, the psychotherapy addresses defects that have characterised the early phases of development of the child. These defects result in identity and relationship problems that recur in everyday life situations and are expressed during the psychotherapy. The psychotherapist uses different techniques: expressive, modified analytical, and exploratory. This involves containing, confronting, interpreting, and supporting, depending on the degree of severity with which the disease is expressed. It is first, however, essential to establish a stable therapeutic framework for the treatment to begin. The technique used for the treatment may evolve during the psychotherapy. "Interpretation" may initially be contraindicated and subsequently be effective.

Brief psychodynamic psychotherapies are relatively little used in France, although a large proportion of the evaluation studies refer precisely to practice of these therapies. They range from the most directive types, centred on the event, to those that are more interpretative and centred on personality. Their indications and contraindications (severe disorders, poor motivation for therapy) are very precise.

In focal psychotherapy (David Malan, a pupil of Balint), the start of treatment is preceded by a very major evaluation phase. Identification of precipitating factors, early traumatic experiences, or repeated patterns leads to the definition of an internal conflict that has been present since childhood and that becomes the focal point of treatment. The greater the likelihood that the area of conflict emerges during the transference, the more likely is the result to be positive. The "transference triangle" (transference, current relationship, and past relationship) leads to improvement in the patient's health. 20 to 30 sessions are generally involved, and in a few published cases the therapy has been continued for a year.

Brief psychotherapy with anxiety provocation (Peter Sifneos) concentrates exclusively on the Oedipus complex. During the initial phase of treatment, the therapist has to establish a good relationship with the patient to create a strong therapeutic alliance. The therapist uses confrontations that provoke anxiety to clarify the questions troubling the patient in his/her early life and the current conflict. To undergo this type of therapy, the patient must have a major specific complaint and recognise the psychological nature of his/her symptoms. The patient must be particularly well motivated to change and be seen to be able to interact with the therapist, who evaluates the patient by expressing his/her feelings. Acceptance to make reasonable sacrifices and a realistic expectation of the results of the psychotherapy are also required. The vast majority of treatments of this type involve 12 to 16 sessions and never go beyond 20 sessions. Each session lasts 45 minutes.

In Mann's time-limited psychotherapy, 2 to 4 evaluation sessions are normally undertaken before beginning the psychotherapy. The psychotherapist explains the therapeutic contract and aim of the therapy to the patient and uses classical psychoanalytical psychotherapeutic techniques: defence analysis, interpretation of transference, and reconstruction. The psychotherapy is limited to a total of 12 hours of treatment that are divided up depending on the patient's needs. This may take place in the form of weekly sessions lasting half an hour for 24 weeks, or hourly sessions twice per week for 6 weeks.

Davanloo psychotherapy involves 5 to 40 sessions, depending on the patient's area of conflict. The treatments generally last for between 15 and 25 sessions. It is not recommended that a specific end date be set, but rather that the patient be told clearly that the treatment will be short. Short periods (5 to 15 sessions) are reserved for patients who mostly have an Oedipus conflict.

Brief adaptive psychotherapy is a more cognitive therapy that concentrates on identification of the most inappropriate pattern and on elucidating it in past and current relationships, and very specifically in the relationship between the patient and the therapist. The aim of the therapy is to make the patient able to develop insight into the origins and determining factors of this pattern, to develop more appropriate interpersonal relationships.

Strupp and Binder psychotherapy is based on interpersonal transactions and focuses on a linguistic analysis of the description of relationships. It is therefore based on analysis of the patient's current interpersonal relationships, including the relationship with the practitioner, and internal object representations. It brings attention to patients' withdrawal and detachment characteristics that are considered to be defence mechanisms. It is therefore more centred on interpersonal deficiencies than on intrapsychological conflict.

The Gillieron brief psychodynamic investigation technique is designed to identify the nature of the psychological change desired and the best ways of reaching this with the patient. Its initial results are to enable the patient to construct a request for care that is tailored to the origins of the conflict, to strengthen the therapeutic alliance, and sometimes to resolve the crisis that has brought the patient to consult.

The psychodynamic (psychoanalytical) psychotherapies described above in adults are the ones that are most frequently found in the studies assessing efficacy. The aim of these therapies is to achieve profound, sustainable changes in the patient. They seek to obtain these changes through the use of language (they form part of the so-called verbal therapies) as the instrument to reconstruct the persona.

Psychoanalysis has developed differently in its applications to children. The rule of free associations cannot be applied to children, and the symbolic value of the game has been taken into account. The game therefore becomes the preferred instrument in psychotherapy, including drawings, toys representing human beings, animals, cars and houses, etc. The game is similarly considered to be a means for children to defend themselves against the affects that they are afraid of in the treatment situation.

In principle, frequent sessions are involved. In British and Latin American countries, it is usual to see a child 5 to 6 times per week. Many in France consider that 3 or 4 sessions per week are essential, although it is often impossible to sustain such a frequency for a long period of time. Commonly, children are analysed for 2 sessions per week. Psychoanalysts are therefore able to use more straightforward psychotherapeutic methods in children, although all are derived from psychoanalysis.

They are based on the following principles: expressional psychotherapy uses mostly children's games, although the game is played in the presence of an adult and has a "regressive" function that cannot be ignored. Relational psychotherapy plays a key role in children. The psychoanalytical interpretations in the different psychotherapeutic interventions are always aimed at verbalising affects.

What are the different forms of cognitive-behavioural psychotherapies?

Behavioural and then cognitive therapies were first developed in Britain and Northern European countries at the start of the 1960s. They then spread through all developed countries and have been present in France since the start of the 1970s through private associations, the major two of which are the Association Française de thérapie comportementale et cognitive (AFTCC; French association for behavioural and cognitive therapy), which was founded in 1972, and Association Francophone de formation et de recherche en thérapie comportemental et cognitive (AFFORTHECC; French speaking association for training and research in behavioural and cognitive therapy), founded in 1994. These two associations offer both base training and continuing education in the form of workshops and meetings.

Training in cognitive-behavioural therapies (CBT) in France is aimed at psychiatrists, general practitioners, psychologists, and specialist psychiatric nurses. Orthophonists, psychomotor practitioners, and specialist educators also have access to some of this training. The training is provided privately by AFTCC and AFFORTHECC and in the public sector by university diplomas (UD). In principle, the training lasts for 3 years, following the criteria recommended by the European Association for Behavioural and Cognitive Therapy (EABCT).

Training in cognitive-behavioural therapy according to the criteria of the European Association of Behaviour and Cognitive Therapy (EABCT)

The training lasts for a minimum of 5 years including the base professional training.
Training: 450 hours, 200 of which from a competent therapist.
Development of competencies: 200 hours.
Supervision: 200 hours by a competent therapist.
At least 8 cases supervised, covering 3 types of problems.
Memory: at least 4 cases ( 2,000–4,000 words)
Accreditation by an association (the above training followed by continuing education)
Therapy and personal development.
Is didactic therapy required? The choice is left free to each country.
It is stressed, however, that each therapist must know when to ask for assistance.

Behavioural and cognitive therapies represent the application of the principles derived from experimental psychology to clinical practice. These therapies were initially based on learning theories: classical conditioning, operant conditioning, and social learning theory. They then referred to the cognitive theories of psychological functioning, particularly the information processing model.

The principles of classical conditioning (respondent or Pavlovian) are based on the concept that a certain number of behaviour patterns result from conditioning through the association of stimuli.

According to operant conditioning described at the end of the 1930s by Skinner as an extension of Darwinian theory of natural selection, the living body acts on the environment and the consequences of its action lead it to change its behaviour. Analysis of the maintenance of a behavioural sequence involves studying its consequences, which allow the final result of a behaviour to be understood. An action that has positive consequences tends to be repeated (positive reinforcement), and conversely, negative consequences of an action on the body tend to produce avoidance or escape behaviour from the situation that is liable to produce the unpleasant effects (negative reinforcement). Absence of negative or positive consequences of an action will lead gradually to the disappearance of the action because of the absence of any reinforcement: this is extinction.

The two major principles most frequently used in practice are the principle of difficulty segmentation (this consists, for example, of classifying stages of confronting a situation depending on the anxiety experienced at each stage) and progressive shaping, with positive reinforcement of behaviours by approval from the therapist. This allows the desired aim to be approached gradually and avoids discouragement both of patients and of therapists.

At the end of the 1970s, the importance of learning by imitation of models was demonstrated experimentally by Bandura. The principles of this were extended to clinical problems, and so-called "modelling" techniques are used above all to develop social skills through role-playing. Bandura also developed a general theory of psychotherapeutic change, proposing that there was a specific dimension of mental functioning: perceived self-efficacy. The change takes place depending on whether a person considers him/herself to be capable or incapable of performing a behaviour and whether the person considers that the behaviour will lead to a result.

Cognitive therapies are based on the concept of cognitive schemas. A cognitive schema can be described as a cognitive structure printed on the body through experience. Cognitive maps are stored in the long-term memory and select and process information subconsciously (in the sense of automatically). These schemas represent personal interpretations of reality and influence individual strategies of adaptation. They represent an interaction between behaviours, emotions, attention, and memory. Each psychopathological disorder occurs as a result of inappropriate interpretation of the person's own self, the current environment, and the future. There are, therefore, specific schemas: negative interpretation of events (depression), dangers (phobias, panic attacks), and overresponsibility (obsessive–compulsive disorder). These schemas are characterised by selective attention toward the events that they confirm: they therefore represent a prediction that comes to course.

Description of the different types of cognitive and behavioural therapies

Technique Definition
Cognitive therapyTherapy based on modification of cognitive schemas and processing of information.
Behavioural therapyTherapy based on the principles of conditioning and social learning.
Cognitive-behavioural therapyTherapy based both on learning theories and modification of cognitive maps.
EMDR therapy (eye movement desensitisation and reprocessing therapy)Desensitisation therapy through ocular movements and reprocessing of information.
Group or couple cognitive-behavioural therapyMost CBT can be performed on an individual, group, or couple basis, depending on the indications and specific cases.
Cognitive-behavioural family therapyThis is used above all in the psychoeducational family approach for psychotic patients in a context of psychosocial rehabilitation and in the treatment of certain childhood and adolescent disorders, particularly autism and externalised disorders (hyperactivity, conduct disorders).

Like any therapy, cognitive-behavioural therapy is based on a therapeutic relationship, the components of which are nonspecific: warmth, empathy, genuineness, professionalism, mutual confidence, patient acceptance. These components work concurrently to establish a positive therapeutic alliance and are necessary, but not sufficient. In cognitive-behavioural therapy, the therapeutic relationship is based on the here and now, selection of concrete problems to solve with the patient, and a constant attitude toward testing therapeutic hypotheses established in collaboration with the patient.

A key stage is the functional analysis, which studies the relationships between "behaviours-problems", thoughts, emotions, and the social and physical environment to tailor the application of the general principles based on learning theories and cognitive theories to each patient. Functional analysis grids are used that allow the patient's functioning vis-à-vis his/her behaviours—existing problems (synchrony), their establishment and maintenance in the past (diachronic) to be understood, and the therapeutic process to be guided from hypotheses that are common to the patient and therapist relating to factors involved in triggering and maintaining the disorder.

Cognitive-behavioural therapy can be used in the form of individual, couple, or family therapy. The sessions are limited in number to 10 to 25 (one per week) for anxiety and depression disorders to around 100 sessions (one or two per week) for personality disorders or rehabilitation of psychotic patients. The sessions last from 30 to 60 minutes. For personality disorders and depression, however, sessions lasting 1 hour are recommended. Sessions lasting at least 1 hour and up to 3 hours have been recommended for serious obsessive–compulsive disorders and chronic post-traumatic stress, depending on the individual difficulty of the case.

Many of the cognitive-behavioural techniques that have been developed and used in adults have also been applied to children and/or adolescents. In the best of cases, these techniques have been adapted to take account of the specific developmental features associated with age, and in some cases, specific manuals designed for children or adolescents have been published. In other cases, the treatments are used in their original form or only slightly changed, depending on the discretion of the therapist. Other cognitive-behavioural techniques have been developed directly for children or adolescents and for indications that are specific to this age group.

What techniques are used in family or couple therapy?

The following definition can be offered for the "lowest common denominator" of the family therapies: "Family therapy involves any beneficial form of consultation, either single or repeated, which brings together at least two members who form part of the lives of one or more person(s) suffering distress, one of the persons who is consulting normally being the person in greatest distress".

The beneficial effect of this consultation(s) is assessed from symptoms, the distress, and the problems and relationships of the people who consult. Perception of the benefit may be gained by the people themselves, the people in distress, the therapist involved, and also the broader therapeutic entourage and entourage in the lives of the people concerned.

Family therapies were developed in the United States in the 1950s in psychiatric units and Social Service departments as forms of assistance and care for serious mental disorders considered not to be readily accessible or inaccessible to classical forms of psychotherapy. Although family therapies more specifically target interpersonal problems or behavioural problems, they are also useful for the treatment of different mental disorders. They derive from psychodynamic, cybernetic, and systemic ethnological and anthropological principles. Behavioural and cognitive, humanist and narrative currents then developed, centred on the solution.

Training of therapists is usually delivered by private institutions. This normally takes place in groups of 10 to 15 people with an average of 200 hours per year for 4 years. The universities (psychiatric and clinical psychology) have gradually and relatively sparsely integrated training modules for family therapies into the end of courses. Intensive training lasting 1 to 2 years is also offered in some universities and is not reserved for psychiatrists or clinical psychologists.

Around 20 countries in Europe take part in the organisation of the European Family Therapy Association (EFTA). The dominant current is the ecosystem current. Approximately 200 French professionals belong directly to the EFTA. Most of these are psychologists, social workers, or nurses. The Société Française de thérapie familiale (French family therapy association), which belongs to the EFTA and to the Fédération Française de psychiatrie(FFP; French Federation of Psychiatry), has 300 members, 180 of whom are full members. These latter members must provide evidence of 4 years of training involving 200 hours per year and 4 years of family therapy practice. It consists of more than 50% psychiatrists and also has doctors, nurses, psychologists, social workers, and specialist educators. Some French therapists belong to both associations. There are also several psychoanalytical family therapy currents.

France developed the family approach primarily at the end of the 1970s. Since then the types and techniques of therapeutic intervention with families have diversified and become more complex and interlinked. They are based on eco-etho-anthropological principles.

The most recent currents seek to avoid focusing on the family either by directing the interventions through objectivation of symptoms, conduct and emotional and representational mapping (cognitive-behavioural therapies), or by sharing life experiences, avoiding causalist constructs (humanist, narrative currents centred on the solution), or by broadening the contextual meetings to far wider systems (multi-family sessions, networks). Many changes have taken place between these currents: cognitive-behavioural therapies construct types of meeting adjusted for each objective to be treated (individual, couple, family, multi-family, psychosocial rehabilitation groups); the ecosystem therapies refer more to complexity theory and see the sessions as co-creation and co-evolution devices between family members and therapists, which cannot be reduced to predetermined programmes.

Description of the different types of therapies used in the family approach

Principle currents Definition
Couple therapy and psychodynamic family therapyCentred on insight and/or affective experiences, analysis of resistances, intertransference movements, interfantasy processes, and access to subconscious processes. They are frequently based on intergenerational or multigenerational approaches.
Ecosystem couple and family therapiesCentred on improvement in communications in the here and now, on paradoxical prescriptions, resistances, symptoms, and tasks, all necessary to take account of the ecosystem. They may take on structural, strategic, or narrative forms, centred on the solution.
Cognitive-behavioural couple and family therapiesCentred on improvement of conduct and cognition, evaluation and suppression of symptoms, reduction in the expression of critical and hostile emotions, stress management, learning skills in social relations.
Family psychoeducationCentred on information relating to problems, diseases, treatments, and adaptation attitudes to take faced with disturbances due to the disease.
Humanist family therapiesCentred on the expectations and personality of the patients, their ability for independence and to choose to maintain the symptoms or separate from them.
Eclectic and integrative family therapiesCentred on adjustment of methods, techniques and theories depending on the requirements of the families and treatment projects.
Family therapies for unwilling familiesCentred on learning social contexts within which the request for care or demand for treatment emerges: the family as a meta-therapist, assisting the therapists.
Behavioural and cognitive multi-family therapiesCentred on information exchange, mutual aid, problem-sharing, and methods of facing up to problems; development of inter-family solidarity.
Psychodynamic psychosocial therapiesCentred on psychodrama inspired from psychoanalysis, role playing, and interpretation of transference.
Behavioural and cognitive psychosocial therapiesCentred on learning social skills, social-occupational rehabilitation, and stress management.

Removal of the family's role in the origin of the problems has become radical in the cognitive-behavioural approach. The family is considered to be a group of people with behavioural, emotional, and cognitive maps that may have been disturbed in the presence of mental disorders. In the psychoeducational management of schizophrenic patients and their families, the family is considered to be a "normal" family confronted with a disease or a set of diseases that are cerebral in origin and probably neurodevelopmental in nature. If disturbed intra-family relationships or even disturbed relationships between the family and the social environment exist, these disturbances are considered to be secondary to the disease. The therapists therefore start from the observation of the disease, inform the family about the features of the disease (particularly the importance of genetic and biological factors), its course, and its treatment. They offer psychoeducational advice, demonstrating how reducing emotional overload and criticism is likely to lead to improved problem management.

The starting point for the systemic multi-family therapies proposed by Laqueur at the end of the 1970s is to recreate a community and social space for families and patients faced with isolation, suffering, or distress, which appears to be incommunicable to others. Families are asked to take part in meetings in which information is shared in a questions-and-answers mode. These therapies are offered to families facing the same problem (schizophrenia, eating disorders, mood disorders, etc.). Although the pathological profile of the patients must be consistent to form these multi-family groups, experience has shown that inclusion in the groups should be as random as possible in terms of other characteristics (ethnic, religious, political, socio-economic, intellectual, belief systems, opinions, politics, etc.). The group should preferably be 4 to 7 families in size. Interaction between several families appears to produce faster changes than single-family therapies, which are also performed in some cases. Learning processes are initiated from communication through analogy, indirect interpretation, and cross-identification between members of different families. It appears that this facilitates communication spontaneously, that speaking up is easier, and that the atmosphere is more permissive than if the attention is focused on the single family. The copresence of families facing a recent disease and families with greater experience and maturity with the disease result in these latter families acting as "cotherapists".

McFarlane introduced a distinctly more psychoeducational process than the classic version of systemic approaches to avoid spillover due to unbridled emotional expression. This belongs to the cognitive-behavioural current. The process is designed to reduce interpersonal and social isolation, avoid stigmatisation of patients or other members of the family, support each family by reducing the considerable burden which the disease places on its functioning, remove families from tendencies to overprotect and/or disengage from hostility or criticism, and improve intra-family communication, which is facilitated from the outset by the setting of the multi-family exchange itself.

Network therapies, started by Speck at the end of the 1960s, broadened the therapeutic intervention to all people in the environment of the patient's family. The number of people involved may be as high as 50 or 60. This type of intervention may be considered when other therapeutic methods have failed (individual, family, institutional hospital therapies) or to avoid hospitalisations in highly critical situations: high risk suicide, serious mental disorders with risk of "committing the act". The treatment team consists of 4 to 5 people and seeks to restore the "tribal" relationship of the person, which had been broken by modern society. Giving value to the network allows it to operate as a support against the destructuring distresses, restoring confidence in the ability to establish relationships outside of the family. Three forms of network therapy can be distinguished, depending on whether the primary sector (close contacts of the patient and of the patient's family), secondary sector (professionals meeting the social demand), or a combination of both sectors is involved. Mobilising the entire network has (as in the initial forms of family therapy mobilising the whole family) helped to enrich knowledge on the subject, contexts of life, and the follow-up of people who suffer.

Although very different in terms of origin and orientation, the humanist, narrative, and solution-centred concepts, in common with those above, abandoned focussing on the symptom or searching for its causes. The humanist therapy involves developing each person's potentials, taking account of his/her own strengths and weaknesses, development rates, and life projects. Whether the therapeutic contact is short or long, this avoids the creation of dependencies that would reactivate the relationship maps arising from the past and in contrast would promote updating new processes in keeping with the current problems and difficulties.

In narrative therapy, the personal construction of knowledge is achieved by comparing family and social constructs, which in the final analysis only come from a relativist point of view of the state of the world. As each vision of the world is relative to the relationship contexts in which it is produced, none could claim to be pre-eminent.

In solution-centred therapy, the therapist only takes positive experiences from the past and directs his/her interventions from the present toward the future. The therapist considers that the patient has used the correct solutions and suggests new adjustments that confirm these correct solutions. In this context, a slight change in a person may have consequences on the entire marital or family system, without necessarily meeting the spouse or other members of the family.

In contrast to cognitive-behavioural therapies that seek to objectivise reproducible procedures possibly using evaluation scales and information and learning guides for patients, the systemic therapies see themselves as contextual intervention projects that enable ways of thinking and doing to be readjusted or even invented, depending on the features of each specific clinical situation. The therapeutic project involves delineating the areas of competence and performance of the families and persons involved. This leads to learning of learning (deutero-learning), which patients and their close contacts do not do spontaneously because of self-contradictory instructions in everyday life. Initiating a therapeutic context involves creating a mechanism in which the options can swing between expression of ordinary conversations and construction of viable projects, through comparative exploration of alternative solutions.

Marital behavioural therapy is based on learning of communication within the couple and problem resolution. It offers planning for changes in behaviour to increase satisfactory interactions and reduce destructive and negative interactions. It is not only a strategy for intervention but also a treatment based on social learning.

Cognitive marital therapy is centred on irrational relationship mapping and irrational beliefs. It is often associated with behavioural techniques or even training in expressing emotions.

Psychodynamic marital therapies are centred on emotions or directed toward insight. When the therapy is directed toward insight, it stresses conflicting emotional processes affecting each of the partners considered separately and the interactions between the partners and the broad family unit. This therapy includes individual marital and family functioning in terms of development and maturation processes, collusions and divergent contractual expectations, irrational role assignments, and maladapted relationship rules. The therapists use probing, clarification, and interpretation to discover and explain feelings, beliefs, and expectations that the partners have of themselves, their partners, and their marriage and that may be partially or totally subconscious and be restructured by conscious renegotiation.

Marital therapy focused on emotions is based on the Bowlby attachment theory and sees relationship distress as weak links in which the needs for attachment are unachievable because of rigid ways of interaction that block emotional engagement. The method involves helping each partner to explore and communicate his or her emotional experiences on subjects such as affiliation dependency (proximity and control) in the context of the normal relationship. Valuable attachment needs are clarified, and each person comes to better understand and see his or her partner with greater sympathy. This leads to new and less defensive interactions. This approach appears particularly useful for couples who are not displaying extreme disturbances.

What is the net result of the evaluation studies of psychodynamic (psychoanalytical) psychotherapies?

Started since the end of the 1910s, evaluation of the effects obtained from psychoanalytical psychotherapies in patients has come up against difficulties in analysing the multidimensional features of the changes. Quite specific to this approach, psychotherapy is constructed from the patient and the patient's specific problems and methods of functioning. In the most recent studies, the evaluation involves not only the effect of therapy on symptoms, without differentiating between terminological categories, but also an evaluation of the changes in the psychodynamic structure that underpins the disorder itself. Evaluation instruments for psychodynamic changes have been developed recently (MSI, ECPD, KAPP, etc.), 5 as have instruments that evaluate use of techniques, adherence with the therapeutic method (PACS-SE, TIRS, PTS, GIS, STT), 6 or the therapeutic alliance (CALPAS)5.7 However, the specific instruments to evaluate the psychodynamic aspects are still little used. Some, still preliminary studies have sought to address the role of specific and nonspecific factors in the effects of psychodynamic (psychoanalytical) psychotherapies: do men and women respond in a similar way? What is the impact of a quality of object relationships? What are the interactions between the patient's accounts of a therapy experience and the therapeutic alliance? What is the influence of interpersonal or personal problem typologies and style of principal attachment?

These studies have thrown particular light on the interaction between these different variables. It is likely that early symptomatic improvement plays a role in establishing the therapeutic alliance (these two factors reinforce each other mutually) and thereafter in the effect of the therapy. Some of these parameters, however, appear to have greater prognostic value than others on the results of treatment: the initial quality of object relationships and training of therapists for difficult cases.

Although many research studies have been conducted on long psychodynamic (psychoanalytical) treatments (mostly case and process studies), these have only very recently been extended to studies on clinical populations. Conversely, more evaluation studies have been performed on brief psychodynamic (psychoanalytical) therapies. Three meta-analyses reported the efficacy of brief psychotherapies (on target symptoms, general symptoms, or social adaptation) compared to a placebo (waiting list or no treatment) for a set of disorders. It must be noted, however, that two of these meta-analyses did not examine the effects of psychodynamic psychotherapies independently of those of the non-psychodynamic interpersonal therapy. One of these meta-analyses demonstrated that efficacy was greater for well-trained therapists.

Two meta-analyses that combine studies conducted on stabilised schizophrenic patients followed up on an outpatient basis demonstrated that psychodynamic psychotherapy or psychoanalysis had little or no effect. Psychodynamic (psychoanalytical) psychotherapies produced a very small effect size (0.27). Only one study examined hospitalised patients (in the acute phases) and did not find psychodynamic therapy to have additional effect over drug treatment.

For mild or moderate depression in the adult, one meta-analysis combined psychodynamic therapies and interpersonal studies in the term "verbal therapy" and showed these therapies to have global benefit, although it was not possible to conclude what the efficacy of each of the two types of therapy examined separately was. Three studies on depression in the elderly treated on an outpatient basis (in a meta-analysis) compared psychodynamic (psychoanalytical) therapy or its brief form to placebo or to a waiting list and found no significant, positive benefit.

The association of psychodynamic psychotherapy with antidepressant treatment in patients managed on an outpatient basis (after hospitalisation) has been assessed in major depression in one randomised controlled trial that demonstrated that the combination of both treatments had significant benefit, with an improvement in global functioning and a fall in the hospitalisation rate at the end of treatment. The psychotherapy was administered by well-trained nurses under close supervision. Finally, one randomised controlled trial studied the results of brief psychodynamic interpersonal therapy at 6 months in adults after a self-poisoning suicide attempt. The results showed a reduction in depressive symptoms, suicidal ideation and relapse, and higher satisfaction.

Two controlled trials on the treatment of anxiety disorders were found in the literature. Horowitz short psychodynamic psychotherapy (centred on resolution of intra-psyche conflicts as a result of a traumatic experience) was shown to be effective in patients with post-traumatic stress state compared to a control group. The effects were particularly large on traumatic, avoidance, and somatisation symptoms. The other controlled trial, conducted in patients suffering from panic disorder, demonstrated that addition of brief psychodynamic psychotherapy (centred on psychosocial vulnerability) to drug treatment significantly reduced relapse rates (at 18 months) compared to medical drug treatment alone. However, one uncontrolled trial in people with panic disorder suggested that psychotherapy centred on the panic used as monotherapy (with a manual) to be very effective, and that the gains were maintained at 6 months of follow-up. The absence of a control group in this case, however, makes it impossible to confirm the result or that treatment was effective.

No controlled trials were found in the literature for eating disorders. One cohort study followed more than 1,000 patients with anorexia or bulimia who received psychodynamic (psychoanalytical) therapy during inpatient hospitalisation, for 2 to 3 months. 33% of the anorexic patients and 25% of the bulimic patients had no further symptoms at follow up at two-and-a-half years. The results correlated with specific patient characteristics: older age in anorexics was a predictive indicator for poorer response to treatment; for bulimic patients, impulsivity, the presence of associated symptoms of anorexia, and a large number of previous treatments were associated with less good results whereas good social adaptation was a predictive indicator for improvement. We must remember that this study was limited by the fact that it did not contain a control group and that the observed improvement cannot be attributed with certainty to the therapy.

Personality disorders define several types of very different patients grouped into three categories (A, B, and C) in the DSM. Category A contains the paranoid, schizoid, and schizotypic personalities; category B contains antisocial, borderline, hystrionic, and narcissistic personalities; and category C contains avoiding, dependent, obsessive–compulsive, and nonspecified personalities. This is, therefore, a group of disparate disorders that have the common feature of being primary with respect to the development of other problems such as depression, of occurring during development in childhood or adolescence, and of continuing in adulthood. Patients with a personality disorder have many problems that are liable to vary over time. Evaluation is based on different aspects of their functioning (reduction in number of suicide attempts and self-harm behaviour, quality of object relationships, etc.), the interpretation of which requires particular attention. For example, the increase in attendance at healthcare services may be a sign of improvement at the start of treatment, whereas a reduction in attendance is expected at the end of treatment. These are also chronic disorders, and the effects of therapies may be difficult to interpret because of events, other treatments, etc. or simply age. One study clearly shows that improvement in symptoms and functioning in patients who are treated is associated with better interpersonal relationships, whereas that untreated patients progress toward social withdrawal.

For the personality disorders (all disorders combined), the literature contains one meta-analysis conducted in 2003 that grouped 15 trials, two of which compared psychodynamic therapy to a control condition (waiting list or standard care). The overall effect size (calculated from pre- and post-treatment data for all of the studies in the meta-analysis) was 1.46 for self-assessment measurements and 1.79 for measurements assessed by other people. The effect size compared to control conditions (calculated for both studies) was 1.32 (from self-assessment measurements). Psychodynamic psychotherapy appears to be effective in personality disorders (with two controlled trials).

One controlled trial examined the efficacy of psychoanalytically orientated psychotherapy in a day hospital compared to the standard psychiatric care (monthly consultation with medical drug compliance control) in patients with borderline personality disorders (type B). The treatment was administered by nurses trained in psychiatry (under bi-weekly supervision) although were not formally qualified in psychotherapy. The scores for all evaluation measurements fell significantly in the psychotherapy patients at 6 and 18 months: improvement in depressive symptoms, reduction in suicidal and self-harm acts, reduction in days of inpatient hospitalisation, and improved social and interpersonal functioning.

Interpersonal psychodynamic psychotherapy (psychotherapy derived from the Hobson conversational model) was assessed (in a non-randomised controlled trial) in a group of patients suffering from borderline personality disorders, compared to a "usual treatment" group (support therapy, crisis intervention, cognitive therapy, pharmacotherapy, drug therapy). The psychotherapy was based on the concept that borderline personality disorder occurs as a result of interrupted development of the "Ego" and is designed to promote maturation (discovering, constructing, and object relationships expression of personal reality).

Effects of psychodynamic (psychoanalytical) interventions

Diseases Trials considered Main results
Acute phase1 trialResults not significant.
Stabilised2 meta-analysesLittle or no effect.
Mood disorders
Moderate depression in the adult1 meta-analysisCombined BPT and IT (non-psychodynamic) produced positive results, although no study on psychodynamic therapy alone.
Moderate depression in the elderly1 meta-analysisNo significant result.
Major depression on antidepressants1 controlled trialPositive, significant effect of psychodynamic therapy after hospitalisation on global functioning, reduction in relapses.
Depression associated with attempt suicide1 controlled trialPositive effect of interpersonal psychodynamic therapy on suicidal ideation and relapse rate at 6 months.
Anxiety disorders
Panic disorder on antidepressants1 controlled trialBPT effective in reducing relapses after stopping antidepressant therapy for 9 months.
Post-traumatic stress1 controlled trialBPT effective on symptoms.
Personality disorders
All disorders combined1 meta-analysis (2 controlled trials)Significant effects on overall improvement.
Borderline personality3 controlled trials (1 non-randomised trial)Orientation psychoanalytical psychotherapies effective on all measurements at 6 and 18 months. IT (psychodynamic) effective on diagnostic criteria, maintained from 1 to 5 years; individual and group therapies effective.
Antisocial personality1 controlled trialBrief psychodynamic therapy beneficial to patients presenting with depression.
Avoidant or other type C personality1 controlled trialBPT effective (Davanloo and adaptive psychotherapy), maintained one-and-a-half years after end of treatment.

BPT, brief Psychodynamic therapy; IT, interpersonal therapy.

30% of the patients treated with interpersonal psychodynamic psychotherapy no longer had the DSM diagnostic criteria for personality disorder after 1 year, whereas the patients from the control group had not changed. This improvement was maintained at follow-up at 1 and 5 years.

Interpersonal group psychodynamic psychotherapy has been compared to individual psychotherapy (randomised controlled trial) in the treatment of borderline personality disorders. Treatment lasted for 35 weeks. The psychotherapists were trained and experienced in individual psychodynamic therapy and trained, managed, and supervised for the dynamic group psychotherapy. The results demonstrated a considerable improvement in behavioural indicators, social adaptation, overall symptoms, and depression. No significant difference was found between group and individual psychodynamic therapies at the end of treatment and at 24 months of follow-up.

Patients with personality disorders (mostly type C, a few type B disorders) were treated on an outpatient basis with two forms of brief psychodynamic psychotherapy (Davanloo psychotherapy and adaptive psychotherapy). The results demonstrated a significant improvement in those treated with both forms of therapy compared to a waiting-list control group. No difference was found between the two therapies, and the improvement was maintained one-and-a-half years after the end of treatment. Brief psychodynamic therapy combined with advice was found to be more effective than advice alone in patients with antisocial personality, dependent on opiates and who had associated depression (one controlled trial). Several trials have demonstrated that personality disorders are frequently associated with other disorders (for example, depression), and that this co-morbidity influences the results of therapy and often requires longer treatment.

In children, only retrospective uncontrolled trials (from the Anna Freud Centre in London) have examined the short- and long-term outcome of patients treated with psychoanalysis or psychodynamic (psychoanalytical) psychotherapy. Results showed a 62% improvement in patients treated for 1 year (4 to 6 sessions per week), although the study methodology (no control group) made it impossible to distinguish the effect of treatment from the natural course of the disorder. The results were consistent over several points, a major one of which was patient age: the younger the patient was, the better were the improvement and results obtained.

Effects of psychoanalytical interventions in psychological disorders of children and adolescents

Diseases Trials considered Main results
Disturbing disorders (hyperactivity with attention deficit, conduct disorders)Retrospective non-controlled trial (736 cases)*Improvement observed in 62% of children: not possible to draw conclusion on efficacy in the absence of a control group.
Emotional disorders (anxious and depressive disorders)Retrospective non-controlled trial (763 cases)*The probability of improvement falls with age.
Disturbing disorders, emotional disorders, personality disordersRetrospective non-controlled trial (763 cases)*Better improvement rate in emotional disorders.

* This is the same population.

Similarly, there were fewer treatment drop-outs in the younger patients (<12 years of age). Assistance provided to the parents during the child's treatment was a factor that helped to promote improvement in the child's psychological state.

Because psychoanalysis is a treatment lasting several years and requires major investment for the young patient and his/her family, it is important to define the clinical and environmental conditions that allow the correct indications for use of this treatment to be defined and the expected benefits of this type of treatment in a young child to be identified.

Levels of evidence of psychodynamic (psychoanalytical) therapy in adults

Proven efficacy: established by a meta-analysis and randomised controlled trials.
Personality disorders, particularly borderline personality disorder.
Presumed efficacy: established by randomised controlled trials.
Panic disorder on antidepressants, post-traumatic stress state

What results are obtained with cognitive-behavioural approach techniques?

Many meta-analyses have evaluated cognitive-behavioural therapies (CBT) (21 meta-analyses were identified for the disorders being studied) alongside the randomised control trials. Only the results of the meta-analyses are analysed and, where results are not available we have analysed those of the randomised controlled trials.

CBT has been used very widely in different anxiety disorders, and there are many results available from efficacy assessments of these therapies. For panic disorder and agoraphobia, 3 meta-analyses have demonstrated a significant decline in symptoms in response to CBT compared to control conditions. The most effective therapeutic combination appears to be a combination of in vivo exposure and antidepressants.

The effect of CBT has been compared to that of drug therapy in patients with generalised anxiety (1 meta-analysis). The effect sizes were relatively similar (0.70 for CBT and 0.60 for drug therapy), although the effect was maintained after treatment with CBT whereas the effect of drug therapy disappeared after patients were weaned off therapy. The CBT and drug combination was not evaluated.

In the post-traumatic stress state, 1 meta-analysis combined trials on different types of behavioural and cognitive therapies (behavioural therapies, EMDR, etc.) and drug treatments.

CBT (including EMDR) appeared to be more effective than drug treatment on symptoms of post-traumatic stress. The effects of the psychotherapies were maintained after follow-up for an average of 15 weeks. Another meta-analysis specifically examining EMDR demonstrated that this technique (which is considered to be a variant of behavioural exposure therapy) was effective compared to the control group.

Three meta-analyses have examined obsessive–compulsive disorders. One of these (combining 86 trials from 1970 to 1993) demonstrated no difference between antidepressants prescribed alone, CBT, and a combination of both therapies. The effect size for CBT ranged from 0.70 to 1.46, depending on the criterion evaluated. Another meta-analysis (combining 77 trials between 1973 and 1997) found that the CBT was equivalent or better than treatment with serotonin re-uptake inhibitors. A third meta-analysis demonstrated that serotoninergic drugs, exposure CBT, and cognitive and behavioural therapy all to be similarly effective. To calculate the long-term efficacy of CBT in obsessive–compulsive disorders, the results of 9 cohort studies (controlled) were combined. A 70% improvement rate was found over a follow-up period of 1 to 6 years (average, 3 years), with a 60% mean fall in ritual behaviour. However, as a rule, residual symptoms persisted, and the risk of suffering depression remained unchanged.

Three recent meta-analyses provide an overall view of the short- and long-term effects of CBT in social phobias. One meta-analysis (including 42 trials) demonstrated that cognitive therapy associated with exposure had a greater effect size than placebo (1.06 versus 0.48). Additional improvement was also found at follow-up. Another meta-analysis (combining 24 studies) showed CBT to have an effect size of 0.74 compared to placebo. The CBT did not demonstrate significant differences in efficacy compared to drug therapy. In the third meta-analysis, several CBT methods were compared to control conditions and to drug therapy. The effect size ranged from 0.6 to 1.0 for all of the forms of CBT and from 1.0 to 2.0 for drug therapy. The improvement in symptoms with CBT was maintained over time.

A few, low statistical power, controlled trials exist for the specific phobias (flying, dentist, spiders, heights, claustrophobia). These all show the different types of therapy (cognitive, behavioural, exposure in virtual reality, in vivo exposure therapies, etc.) to be effective.

For moderate or severe depression, the oldest meta-analysis (which included 28 trials) demonstrated that cognitive therapy was superior to waiting list, to drug therapy, and to behavioural therapy. The results of cognitive therapy at the end of treatment were better than those of antidepressants and of waiting-list patients. Behavioural therapy was shown in the most recent meta-analyses to be equivalent in efficacy to cognitive therapy (because the technique performed does in fact often associate behavioural and cognitive methods) and antidepressants, and equal to interpersonal therapy in one meta-analysis.

The efficacy evaluation has also addressed prevention of long-term relapses in depressed patients. The term relapse refers to redevelopment of a complete depressive state between 6 and 9 months after a remission lasting 2 months: a recurrence occurs beyond that period. The effects of cognitive therapy on prevention of relapses were greater than those of the antidepressants (between 1 and 2 years) in 6 controlled trials of 8. On average, 60% of patients treated with drug therapy alone relapsed compared to only 30% of patients treated with cognitive therapy alone or combined with antidepressants.

According to the trials that have examined the effects of cognitive therapy on residual symptoms and recurrences in patients receiving antidepressants, the number of recurrences was significantly lower in the group that received cognitive therapy. The authors concluded that CBT was an alternative to continuing antidepressants.

In patients who suffer a major hospitalised episode of depression, one meta-analysis reported an evaluation conducted on discharge from hospital, which demonstrated that cognitive therapy associated with drug therapy was effective. The results of another meta-analysis showed an effect size of 0.96 for behavioural therapy and 0.85 for cognitive therapy compared to a control group in depression in the elderly. Psychoeducational treatments (information, awareness, improvement in interpersonal functioning, etc.) have recently been developed for patients suffering from bipolar disorder. These have produced beneficial results in terms of the time to the first relapse of mania (65 weeks compared to 17 weeks in the control group).

In two meta-analyses and in controlled trials, the CBT has been shown to have beneficial effects on personality disorders (avoidant, borderline, antisocial). In one of these meta-analyses, the overall effect size for CBT was 1.0 (1.20 for self-evaluation methods and 0.87 for evaluation measurements by other people). Several controlled randomised trials have been conducted with dialectic behavioural therapy (DBT) for borderline personality disorders in women from disadvantaged areas. This therapy involves an eclectic set of techniques based on the behavioural and cognitive principles. A reduction in suicidal and parasuicidal behaviour (35%) was found after 1 year in patients who received DBT compared to 65% in patients who received usual treatment (psychoanalytical treatment or support). These studies also showed a decrease in pathological anger and days of hospital dmission and improved social adjustment in the group treated with DBT.

Effects of cognitive-behavioural interventions

Diseases Trials considered Main results
Anxiety disorders
Agoraphobia2 meta-analysesEfficacy of CBT proven
Panic disorders1 meta-analysisEfficacy of CBT proven; significant reduction in panic attacks
Generalised anxiety disorder1 meta-analysisEfficacy of CBT proven; maintenance of effect after end of treatment
Social phobia3 meta-analysesEfficacy of CBT proven; maintenance of effect during follow-up period
Post-traumatic stress2 meta-analysesEfficacy of CBT proven; maintained at follow-up
Efficacy proven for EMDR (simple variant of CBT)
Obsessive–compulsive disorder3 meta-analysesEfficacy of CBT proven
Specific phobia6 controlled trialsEfficacy of CBT presumed
Mood disorder
Moderate or mild depression, outpatient basis3 meta-analysesEfficacy of CBT proven
Depression, hospitalised1 meta-analysisEfficacy of CBT proven
Depression in the elderly1 meta-analysisEfficacy of CBT proven
Bipolar disorder on psychotropic drugs1 meta-analysisEfficacy of CBT presumed
Chronic schizophrenia on neuroleptics3 meta-analysesEfficacy of CBT proven
Schizophrenia in the acute period, on neuroleptics1 meta-analysisEfficacy of CBT presumed
Personality disorders
Borderline personality2 meta-analyses
5 controlled trials
Efficacy of CBT proven
Avoiding personality1 controlled trialEfficacy of CBT presumed
Alcohol dependency2 meta-analyses
1 review
Efficacy of CBT proven
Eating disorders
Bulimia4 meta-analysesEfficacy of CBT proven in the short term
Binge eating disorder6 controlled trialsEfficacy of CBT presumed
Anorexia1 post-hospitalisation controlled trialPresumed efficacy in the prevention of relapses after weight gain
Anxiety and depressive disorders in children and adolescents
Moderate depressive disorders2 meta-analysesEfficacy of CBT presumed
Anxiety disorders6 controlled trialsEfficacy of CBT presumed, although no trials specific for this type of disorder

Cognitive-behavioural therapies have been evaluated in alcohol-dependent patients. These therapies use desensitisation, positive reinforcement, and motivational and relapse prevention strategies. They are occasionally used in family and couple therapies. Self-efficacy and social anxiety reduction techniques have also been developed. These derive directly from the Dandura theories on social learning and self-efficacy. The Prochaska and DiClemente five-stage model (precontemplation, contemplation, preparation, action, maintenance) also applies to the CBT. Cognitive or behavioural therapies may be put in place at each stage. Several studies (meta-analyses, controlled trials) have demonstrated brief interventions to be more effective than long-term interventions in patients who are motivated for treatment. However, the effects appear to be greater in patients who suffer less severely. There is no evidence that programmes designed to control alcohol intake produce better results than those designed to produce total abstinence.

Behavioural therapies and also, although to a lesser extent, the cognitive therapies used in development and social skills programmes have produced beneficial results (in 3 meta-analyses) on short- and medium-term relapse rates in reducing symptoms and in social re-adaptation of schizophrenic patients. The problem that remains is generalisation of the acquired skills that, although it appears to be real, is still too limited.

For acute schizophrenia, one meta-analysis conducted mostly on recent studies evaluating the effects of cognitive therapies demonstrated that the hospital admission relapse rate was not systematically reduced when cognitive therapy was compared to standard therapy in schizophrenic patients. A significant difference was, however, found in favour of cognitive therapy for earlier discharge from hospital compared to standard treatment. In terms of overall improvement in mental state, a significant difference was found in favour of cognitive therapy compared to standard therapy at 13 and 26 weeks, although this difference was no longer significant at 1 year. The conclusion drawn was that cognitive therapy is a promising treatment that requires additional evaluations.

The use of behavioural and cognitive-behavioural techniques is an integral part of most of the multi-modal treatment programmes for anorexia nervosa, whether on an outpatient or inpatient basis. Five randomised controlled trials have evaluated the efficacy of CBT in anorexic patients and highlight the improved compliance of anorexic patients to CBT compared to other treatments. They were not, however, able to establish the benefit of this type of treatment. The most recent controlled trial was the first empirical evaluation of the efficacy of CBT as a post-hospitalisation treatment for anorexia nervosa in adults. Following weight regain, the relapse rate and treatment discontinuation rate were lower and the overall clinical results were better in the group that received CBT compared to the comparator group.

Four meta-analyses have been published in the treatment of bulimia, combining between 7 and 35 randomised controlled treatment trials (usually women) with a diagnosis of bulimia (occasionally also binge eating disorder or nonspecific bulimic eating disorder). All four meta-analyses concluded that CBT was effective in the short term in reducing symptoms of bulimia (often assessed from the frequency of attacks and of vomiting) and in the associated dysfunctional attitudes and distortions (in the fewer studies which included these measurements). Comparisons were performed pre- and post-treatment or by comparing CBT to the control conditions. The reported effect sizes ranged from 0.55 to 0.74 in the intragroup comparisons (pre- versus post-treatment) and from 0.23 to 0.67 in the intergroup comparisons (CBT versus control). The long-term data are either insufficient or less encouraging than in the short term: in addition, the wide range of follow-up periods and measurements used to calculate the long-term effect size limits the interpretation of present results. One of the meta-analyses compared randomised controlled trials of drug treatment (9 trials) to controlled trials of CBT or behavioural therapy (26 trials). In the short term, CBT produced effect sizes that were greater than for drug therapy for all of the variables examined. Combination of these two treatments was significantly more effective than drug treatment alone for the frequency of attacks and vomiting and more effective than CBT alone for the frequency of attacks, but not for the frequency of vomiting. The general conclusions drawn were that existing research in favour of CBT being effective in bulimia is convincing despite considerable interindividual variability in the magnitude and stability of the response to treatment.

CBT has also been studied in the treatment of binge eating disorder or BED. This syndrome, which has recently been identified as a specific eating disorder, is in fact closer to bulimia than to obesity without binge eating disorder. Because of this, the first research into the treatment of BED concentrated on investigating the efficacy of methods that have already been proven for the treatment of bulimia (CBT and interpersonal psychotherapy). Six controlled trials demonstrated CBT in various forms (individual or group, self-administered or under the supervision of a therapist) to be effective, demonstrated by good compliance with treatment, which is unusual in eating disorders. Addition of physical exercise to CBT and extending the treatment period (one trial) improved the results and helped to produce a greater reduction in the frequency of bulimic attacks and rise in BMI (Body Mass Index). The positive effects of CBT in obese subjects with BED included a reduction in weight (although this was less that in obese patients without BED) and also a significant improvement in subjective perception of state of health and associated quality of life (one trial). Only one trial included a follow-up period (12 months). The frequency of attacks increased slightly during the follow-up period although remained less than the pre-treatment.

Levels of proof of CBT

Proven efficacy: established by one or more meta-analyses or consistent, high statistical power, randomised trials.
  • agoraphobia; panic attacks; social phobias; generalised anxiety; post-traumatic stress; obsessive–compulsive disorder
  • mild or moderate depressive states; acute depressive states; prevention of relapses and recurrences of outpatient depression; hospitalised depression
  • schizophrenia for psychosocial rehabilitation
  • borderline personality in women; alcohol-dependent people
  • bulimia
Presumed efficacy: established by meta-analyses, randomised controlled trials, cohort studies, reviews: some of these studies may be contradictory and require confirmation.
  • specific phobias
  • avoidant personality; antisocial personality
  • schizophrenia in the acute phase (combined with neuroleptics)
  • bipolar disorder (treated with mood-regulating drugs)
  • moderate depressive disorders in children and adolescents
  • anxiety disorders in children and adolescents

The efficacy of CBT in the treatment of depression has been evaluated in children and adolescents in two meta-analyses, the results of which were consistent and indicate that CBT provides significant symptomatic improvement. However, the trials included in these meta-analyses were limited in number, mediocre in quality, and they used samples of patients recruited from the general population rather than those requesting care, and therefore people who were less severely affected. More recent but isolated controlled trials suggest that the response to CBT is better in younger people and in those in whom the consequences of the disorders are less severe, that the parents' involvement in the treatment does not improve results, and that a rapid response to treatment predicts better long-term outcome. We can refer to the CBT as being of presumed efficacy in moderate depressive disorders in children and adolescents, although the CBT cannot currently be recommended as monotherapy in cases of severe depression in young people.

Proof of the efficacy of the CBT in anxiety disorders in children and adolescents is still limited. Although many studies that examine specific phobias exist, most are old and are of more experimental than therapeutic interest, because they were performed on subjects who were not recruited in a clinical context. The more recent studies have used different CBT techniques (systematic desensitisation in an imaginary situation or in vivo, filmed model technique or in vivo model technique, participation modelling, contingency reinforcement management, training in self-control, restructuring of distorted cognition), which have been shown to be effective compared to other treatments or to a waiting list.

In the treatment of fears and phobias, proof of efficacy exists from several consistent randomised trials for two CBT techniques: participation modelling and contingency reinforcement management; presumed efficacy exists for desensitisation in the imaginary or in vivo situation, the in vivo model technique, and the filmed model technique.

Two randomised controlled trials have been published in school phobia, one demonstrating CBT to be superior to a waiting list and the other concluding that CBT but also psychoeducational support, introduced initially as a control condition, were both effective. Only one recent controlled trial in social phobia has shown that treatment called "social effectiveness therapy for children", which combines group training with social skills, individual exposure, and household tasks, was more effective than nonspecific psychotherapy centred on performance anxiety, and that the benefits were maintained at 6 months.

Presumed efficacy exists for a set of anxiety disorders grouped together in the same trials: hyperanxiety, separation anxiety, and childhood avoidant disorder, from 4 controlled trials that demonstrated that an individual treatment programme with CBT was superior to no treatment and from 2 controlled trials demonstrating group CBT to be superior to a waiting list. In addition, an open trial of group CBT combining parents and children in the treatment of separation anxiety (in some cases associated with another anxiety disorder) in pre-adolescents, demonstrated a higher recovery rate after 3 years than at the end of treatment.

In obsessive–compulsive disorder in children and adolescents, presumed efficacy only exists for a CBT programme based on exposure with response prevention (manual-based programme). A single, low power, randomised trial and 9 open trials are consistent, demonstrating that this treatment is effective on obsessional and compulsive symptoms both in the short term (between 25% and 79% of subjects were improved) and in the long term (6 trials, included follow-up periods ranging from 3 months to 14 years). Huge multi-centre randomised controlled trials on the efficacy of CBT, either alone or in association with pharmacological therapy, are currently under way. These should provide more clearly defined proof of the efficacy of CBT in this indication.

Many cognitive-behavioural techniques are used in the psychosocial intervention programmes designed for the treatment of invasive developmental disorders, particularly autism and externalised disorders, i.e., hyperactivity with attention deficit and conduct disorders, in children and adolescents. Because these programmes usually include active participation of the parents, proof supporting the efficacy of CBT in these indications is reported in the context of family therapies.

What are the results obtained with family and couple therapy techniques?

Reviews that analysed the initial trials, published between 1972 and 1983, highlighted the absence or inadequate nature of the control groups, the unreliability of the evaluation methods, the absence of a sufficient period before follow-up assessment, and the poorly defined nature of the theoretical bases used in the comparisons. Since the 1980s, authors have proposed more detailed indications based on comparisons with individual therapies, comparisons between theoretical orientations, differences between these orientations depending on the problems treated, and the effects of essential moderators and methodological choices on the evaluation of results. Work is now being done to examine the actual scientific proof of the effectiveness of couple and family therapies in major disorders and problems in adults (schizophrenia, mood disorders, drug addiction, etc.) and in children and adolescents (autism, anorexia, conduct disorders, etc.).

The largest number of evaluation studies has been performed in schizophrenia. A single robust criterion for efficacy of treatment is often used in the studies: this is the percentage relapse rate observed in a group during a given time period (relapses being defined, for example, as psychiatric hospitalisation). The results are consistent, and it can be stated with a high level of proof that family interventions reduce the percentage of relapse rate in these types of disorders.

Long-term family interventions have significantly better efficacy compared to brief family interventions. Beyond this, the large diversity of the types of intervention (hospital, outpatient, home) trials examined describe interventions that were usually based on the expressed emotion theory. This theory considers schizophrenia to be a brain disease for which the families are not responsible and refers to behavioural and cognitive principles: understanding and management of stressful situations, recognition of disorders and information about therapeutic methods, and family adjustment to the consequences of the disorders and impacts of treatments. Four meta-analyses and three systematic reviews describe the efficacy of behavioural and cognitive-inspired family therapies.

One controlled trial demonstrated that multi-family behavioural and cognitive interventions appear to produce better results than single-family interventions on prevention of relapse with hospitalisation. This does not mean that all families will benefit from multi-family therapy, particularly because some are not ready for it.

Information for schizophrenic patients and their families, about the current state of knowledge available to professionals about the disorders and the types of treatment and therapeutic approaches available, appears to have a clear therapeutic component in a large number of cases. This lies within a global psychosocial treatment, which may involve learning social skills and management of critical situations and an accompanying approach to social-occupational rehabilitation processes, when these can be envisaged.

In addition, increasingly diverse forms of family therapies exist that are still difficult to evaluate in the current state of methods used to assess therapeutic efficacy, particularly because they are continually evolving. These various practices do not involve standardised family follow-up. It is not clear which family follow-up therapeutic technique is most effective, with which type of patients, and with which types of family. There is, therefore, a considerable gap between research into families of schizophrenic patients over the last 30 years and the application of this knowledge to everyday clinical practice.

Two controlled trials compared different forms of family intervention in a wide range of mood disorders. One of these demonstrated the benefit of multi-family therapies. Conversely, the use of intensive couple therapy for women with unipolar depression appears to be less relevant than the use of drug therapy combined with supportive therapy, either individually or with the partner.

In the case of bipolar disorders, 2 controlled trials have shown that family-focused therapy with information about the nature of the problems, and involving the partners, reduces the number of relapses, increases the periods before relapses, and markedly improves depressive symptoms (according to one of these trials, the question of manic symptoms remains controversial).

In forms of major depression with a "critical partner", one controlled trial has shown that systemic interactive marital therapy produces an improvement in symptoms, both at the end of treatment and 2 years later.

In anorexia nervosa, family therapy appears to produce better results at the start of the disorder in young adolescents. It does not produce savings in hospitalisation in some severe, life-threatening forms of the disease. Published studies consider family interventions, either during a hospitalisation or during outpatient treatments. Five controlled trials have demonstrated various family therapies (psychodynamic, cognitive, systemic, psychoeducation) to have a positive effect on weight recovery. In addition, taking the family dynamic into consideration in the treatment plan appears to have a specific effect on overall improvement, not only on weight gain. One systematic review, however, states that in types of the illness in which serious discord exists within the family, it may be preferable to consider psychotherapy in parallel for the anorexic patient and accompanying support, information, and assistance work for the parents. Depending on the case, direct face-to-face meeting between the patient and his/her parents may have a maturing value, or may be destructive. The therapist's clinical judgement on an individual case basis appears to be essential. One systematic review found no clear advantages associated with any particular theoretical orientation or type of therapist for the family therapy of bulimia. The best results appear to be obtained with more sustainable intensely programmed groups, with the addition of external components (for example, individual work).

In alcohol dependency, the trials (grouped in 2 meta-analyses and 2 systematic reviews) have demonstrated proof of the utility of including family members in the three phases of treatment: initiation; primary treatment; post-cure rehabilitation. Proof has accumulated over time on the efficacy of behavioural couple therapies in terms of abstinence, resolution of problems associated with alcohol, quality of relationships, and reductions in separations and divorces when compared to individual treatments. Some projects, such as the CALM (counselling for alcoholics marriages) had significant effects on domestic violence and on reducing hospitalisations and imprisonment. In all cases, it seems advisable to consider multiple treatment methods tailored to each specific situation. The individual approach to the alcohol-dependent patient and the partner, the marital and family approach, multi-marital and multi-family approaches, or a community reinforcement approach may be added, depending on the factors making up the personality of the patients and their close relatives, the family situation, the type of alcohol misuse, possible co-morbidities, and levels of motivation and commitment to the treatment project. Although at present, results for the efficacy of the adjustment process are uncertain and new hypotheses need to be constructed to establish how the variables relating to severity of psychiatric disorders, level of independence, level of support with abstinence, and degree of investment in social relationships can guide toward the most relevant type of therapy. It would be interesting, for example, to test the effects of interaction between the theory of the treatment and the features of a patient.

Effects of family therapies depending on disorder

Disease Trials considered Main results
Schizophrenias4 meta-analyses
6 controlled trials
Significant benefit of behavioural and cognitive family therapies and family psychoeducation in reducing relapses and rehospitalisations.
Anorexia5 controlled trials
1 systematic review
Significant benefit of behavioural and cognitive family therapies and family psychoeducation and ecosystem therapies for patients with anorexia nervosa that had developed less than 3 years previously.
Mood disorders5 controlled trialsSignificant benefit of behavioural and cognitive orientated ecosystem couple and family therapies for bipolar disorder and major depression.
Alcohol dependency2 meta-analyses
2 systematic reviews
Significant benefit of including family members in the treatment of the alcohol-dependent person; significant benefit of behavioural couple therapies.
Childhood autism7 controlled trialsSignificant benefit of behavioural and psychoeducation programmes with parental training in improvement in IQ, educational performance, and social conduct of children with early autism.
Hyperactivity3 controlled trials
2 prospective controlled trials
Significant benefit of behavioural training of parents and combined treatment including medical drugs and intensive management of the child with parents and school.
Conduct disorders8 controlled trialsSignificant benefit of parental learning and training in problem resolution skills.
Anxiety disorders in children1 controlled trialSignificant benefit of family "management" associated with cognitive and behavioural therapy in disappearance of symptoms.

Other trials examined childhood diseases, such as anxiety disorders, hyperactivity, conduct disorders, and autism. One controlled trial demonstrated behavioural and cognitive family therapy to be effective in childhood anxiety disorders. The informed participation of parents as co-therapists appears to represent a considerable advance in psychotherapies for children with separation anxiety, hyperanxiety, or social phobia.

Attention deficit disorder with hyperactivity has received considerable attention in both pharmacological and psychotherapeutic research. Behavioural training of parents, involving teaching them to adopt a contingency reinforcement management system with their hyperactive child, has been shown to be effective on many occasions compared to waiting-list conditions. Training in social skills and problem resolution has been shown to be effective when it forms an integral part of intensive multi-modal treatment programmes (5 controlled trials). Combined treatment associating drug therapy and intensive management including the parents, children, and school has been shown to be significantly superior to behavioural therapy alone in three fields: opposal and aggressive behaviour scored by the parents, internalised symptoms, and reading performance. The presence of a co-morbid anxiety disorder (34% of subjects) also tends to lend an advantage to combined treatment. In addition, combined treatment could help to reduce doses of drugs.

Conduct disorders in children and adolescents cause serious problems for the family, school, and overall society. These include a wide range of behaviours that range from simple opposal or aggression provocation conduct to those as serious as murder. Minor problems in young children often represent the premonitoring developmental signs of serious aggression in adolescents or in adulthood, making their treatment both desirable and justified. The proposed treatments may be orientated toward the parents, the subject him/herself, or his/her environment (for example, school); the approaches are often different depending on whether the children involved are prepubertal or adolescents, and several treatments may be combined. Therapy with parental learning (explaining the principles of social learning and behaviour modification, rewarding desired conduct, and removing parental attention or privileges in situations of undesirable conduct) is more effective than no treatment (waiting list) or even to alternative therapies. In addition, this type of treatment, designed to reduce antisocial behaviour in the person initially referred for care, can also reduce the risk of analogous behaviour developing in the siblings. Another method proposed to treat disruptive conduct in children and adolescents is training in problem resolution skills. This is based on the idea that antisocial conduct is at least in part linked to cognitive processes, such as a tendency to attribute hostility inappropriately to other people and a poor capacity to understand social situations and resolve interpersonal problems. Combined therapy with parental learning and problem resolution training appears to provide additional advantage (8 controlled trials).

Seven controlled trials had been dedicated to intensive educational and behavioural programmes in autistic children, conducted to a large extent by the parents. The first systematic evaluation of an early intervention programme for autism dates from 1987 (UCLA, University of California). The treatment lasted for 2 years, began at home, and involved an intensive intervention of at least 40 hours per week with a therapist face to face with the child. The effectiveness of this programme had a key influence on subsequent works.

Therapies for autism are based to a large extent on the operant behaviour and conditioning principle (Loovas method) and on psychoeducational and behavioural approaches centred on cognitive and developmental acquisition of competencies (TEACCH method). Both methods are used with a view to educational or social normalisation.

Studies that have applied these two approaches have shown substantial gains in cognitive development (IQ) and in the language of children suffering from autism or other invasive developmental problems (IDP). Starting treatment at an early age appears to be a necessary condition for these interventions to be effective.

Regardless of the place where these programmes are conducted (at home or in specialist centres), close co-operation between the parents and professionals over a long period is a prerequisite for success. The gains obtained are generally maintained after treatment is stopped. According to one of the trials, 42% of children who received intensive behavioural treatment (Loovas method) could no longer be distinguished from other children 6 years after the end of treatment.

Nevertheless, despite intensive interventions, the authors found a lack of progress in some children, whereas those who progressed most had the best cognitive competencies at the start. Studies that were dedicated to childhood autism, therefore, appear to confirm that autism does not represent a homogeneous group of diseases and/or handicaps and that there is a very wide range of responses between children and families to intensive therapeutic approaches.

In conclusion, we found that the current forms of family therapy share a number of common points:

  • modest expectation in the objectives for change in demanding and severe pathological situations
  • not accusing the family and respecting individuals, their lifestyles, beliefs, and knowledge systems
  • sharing relevant information between the patients, families, and professionals
  • calming anxiety-generating or stressful situations
  • accompanying the patients and their families over a sufficiently long time if needed
  • acceptance of differentiation and diversification of intervention methods

Levels of proof for family therapy

Proven efficacy: established by one or more meta-analyses and consistent, high statistical power, randomised controlled trials.
SchizophreniaFamily therapy by psychoeducation for the prevention of relapses and rehospitalisations
Alcohol dependencyCouple behavioural therapy
Autism (child)Intensive educational and behavioural programme administered early
Hyperactivity (child)Intensive multi-modal treatment including parental behavioural training
Conduct disorders (Child)Treatment with parental learning
Presumed efficacy: established by meta-analyses, randomised controlled trials, cohort studies, reviews: some of these may be contradictory and require confirmation.
Bipolar disorderCouple and family therapy with psychoeducation
Anorexia nervosaFamily therapy, any orientation (cognitive-behavioural, psychoeducation ecosystem)

Schizophrenias and alcohol dependency in the adult, conduct disorders and drug addictions in the adolescent, and hyperactivity and autism in children clearly benefit from a combination of therapies, among which the family therapies appear to be important. It would be desirable to see research undertaken to evaluate the many forms of family therapy that are based on personal and interactive individuality and that are tailored to relieve distress and give back reasons for living to patients with major life-threatening problems. It is also important to stress that after being in existence for 50 years, the family therapies movement continues to experience rapid change in practice, continuously adapting to the constraints of individual clinical situations, depending on the wide diversity of methods and techniques that have been used and formalised to date.

What data are available on the comparative assessment of the different therapies?

In recent decades, the field of research into comparison of different forms of psychotherapy has advanced considerably and has led to many therapeutic recommendations in various countries. Nevertheless, making comparisons between the different approaches is a difficult process when evaluating the psychotherapies.

In this section, only those trials aiming directly at modifying the diagnostic criteria for a disorder, the symptoms, or various aspects of patient functioning are analysed. These trials include meta-analyses and randomised controlled trials that have made direct comparisons between two types of therapy and those that have compared the efficacy of different psychotherapies compared to a control group within the same trial. The control group can be made up of people who have not received any active treatment (for example, a waiting-list group) or patients who have received "standard" supportive psychotherapy. 8 Comparison with a supportive therapy allows the specific effects of a given treatment to be identified beyond the expected benefits of positive regular contacts with other people.

First, considering the overall effects (all diseases combined), 5 meta-analyses were found based on almost 700 trials conducted over 60 years. These trials were very varied in methodology and mostly involved people suffering from anxiety and depressive disorders. Their findings indicate that psychotherapy (all forms analysed together) is more effective than no treatment. The average result in treated patients was 70% to 80% better than in untreated patients. This difference was statistically significant in all five meta-analyses. However, the overall positive effects of psychotherapies do not mean that all forms of psychotherapy are effective in the same way, or that the psychotherapies are effective when considered on an individual level. The meta-analyses that demonstrated positive effects of psychotherapy also showed that the positive results were usually associated with cognitive-behavioural therapy divided into two categories: those in which the CBT emerged in the leading position in an effect size classification, and those (the more frequent) that were obtained from direct comparisons between different forms of psychotherapy. Beyond these results, the 5 meta-analyses were not able to classify the other psychotherapies in terms of general efficacy.

Although it may be useful to be aware that psychotherapy is effective, good patient management in general is based on an evaluation of efficacy in the specific types of mental disorder.

One analysis demonstrated that the CBT approach was more effective than "verbal" therapies (psychodynamic approaches and gestalt 9 combined) in various anxiety syndromes analysed together, and more effective than supportive therapy. One randomised control trial compared the psychodynamic approach to supportive therapy in a sample of patients suffering from various anxiety disorders (and some depressive disorders) and found no significant difference in efficacy.

In generalised anxiety disorder (GAD), one meta-analysis found that CBT was more effective than psychodynamic therapy at the end of treatment and after 6 months of follow-up (no family therapy studies were included in this meta-analysis). Of the seven randomised controlled trials found in GAD, three compared CBT to psychodynamic therapy and concluded that CBT was more effective. In the four trials that compared CBT with supportive therapy, three found CBT to be superior and one found no difference.

In other trials that compared supportive therapy with CBT, CBT was found to be superior in the treatment of social phobia (3 trials), panic disorder (3 trials), and post-traumatic stress state (2 trials). The literature does not contain trials that compared pharmacodynamic or family approaches to supportive therapy in these disorders, and no randomised controlled trials were found for the specific phobias. For patients with post-traumatic stress state, one randomised controlled trial, which compared psychodynamic therapy with CBT, did not find any significant difference between the two therapies, both of which produced superior results to those in untreated subjects.

All of the trials that have included a patient follow-up period have shown that the therapeutic gains, regardless of therapy, are generally stable over time in most patients treated for anxiety disorder. The same applies to all of the disorders examined in this review. In addition, the trials do not report symptom substitution with any of the therapies studied (including CBT) in the months or years of the follow-up after treatment.

Depression is a heterogeneous syndrome involving various aetiological factors and may be managed with different types of therapy. Nine meta-analyses or systematic reviews have compared different forms of psychotherapy in adults or in the elderly. In four of these trials, CBT was compared to several other forms of therapy, analysed under the single category of "other therapies": interpersonal therapies, psychodynamic therapy, and supportive therapy.

All of these concluded that CBT was more effective at the end of treatment. In addition, one meta-analysis reported CBT to be superior after a follow-up period of 1 to 12 months, with no fall in the effect size that was obtained initially. Two meta-analyses that compared psychodynamic psychotherapy with CBT produced different results. One, which directly compared brief psychodynamic psychotherapy with CBT, found no significant difference, although CBT but not psychodynamic therapy was statistically different to no treatment. The other meta-analysis also found no significant difference between psychodynamic therapy and CBT, although both treatments were more effective than supportive therapy.

Two systematic reviews have compared different forms of psychotherapy in depressive disorders. The first examined 9 trials comparing the psychodynamic approach with CBT: 5 found both approaches to produce equivalent results, and 4 found that CBT was superior. In the other comparisons, CBT was equivalent to interpersonal therapy in one trial and equivalent to supportive therapy in four others. The only trial that compared psychodynamic therapy with supportive therapy found no difference. In the second systematic review, of depression in the elderly, 3 trials compared psychodynamic therapy with CBT and found no difference between the two therapies at the end of treatment, although two of the three trials found greater improvement and improved maintenance of gains on follow-up in patients treated with CBT.

Fifteen recent randomised controlled trials have directly compared the psychodynamic approach with CBT in the treatment of depression. Most of these have included a patient follow-up period ranging between 3 and 18 months after the end of treatment. Almost all involved brief forms of psychotherapy (between 8 and 20 sessions). Most of the trials found that psychodynamic therapy and CBT were associated with improvement in the depression. Eleven trials found CBT to be more effective, either at the end of treatment or in the follow-up period, and no trials found psychodynamic therapy to be more effective. The trials that concluded that CBT was superior only on follow-up or in which the magnitude of result was greater on follow-up than at the end of treatment suggest an "incubation" effect, indicating that the benefits of CBT are not restricted to the active treatment period.

Two of four trials in depressed patients demonstrated CBT to be more effective than supportive therapy, whereas two found no difference. The only trial that compared psychodynamic therapy to supportive therapy found no difference. In the United States National Institute of Mental Health (NIMH) collaborative programme for the treatment of depression, two trials demonstrated that interpersonal therapy and CBT were effective in the management of depression, although it found no significant difference between these two approaches.

Results of comparative evaluations of different approaches (PT, CBT, ET)

Diseases Trials considered Main results
Anxiety disorders
Generalised anxiety disorderPT/CBT: 1 meta-analysis, 3 controlled trialsCBT more effective
CBT/ST: 1 meta-analysis, 4 controlled trialsCBT more effective
Panic disorderCBT/ST: 3 controlled trialsCBT more effective
Social phobiaCBT/ST: 3 controlled trialsCBT more effective
Post-traumatic stressPT/CBT: 1 controlled trialNo difference in efficacy between the two therapies
CBT/ST: 2 controlled trialsCBT more effective
Mood disorders
Depression in adults or the elderlyPT/CBT: 2 meta-analyses, 2 reviews, 15 controlled trialsResults of meta-analyses inconsistent: no difference found (50% of trials) or CBT superior (50% of trials). Results of trials analysed in the reviews: CBT superior in the majority (73%) of the controlled trials.
CBT/"others": 4 meta-analysesCBT superior to "other" therapies
CBT/ST: 1 meta-analysis, 4 controlled trialsCBT more effective in the meta-analysis and in half of the controlled trials.
CBT/IT: 2 controlled trialsNo difference in efficacy between the two therapies.
PT/ST: 1 controlled trialNo difference in efficacy between the two therapies.
Major depression inCBT/FT: 2 controlled trialsInconsistent results
adolescentsCBT/ST: 3 controlled trialsCBT more effective
FT/ST: 2 controlled trialsInconsistent results
CBT/IP: 1 controlled trialNo difference in efficacy between the two therapies
Schizophrenia (non-acute period)PT/CBT/FT: 2 meta-analysesNo difference in efficacy between FT (psychoeducation) and CBT. CBT and FT superior to PT
PT/CBT: 1 controlled trialCBT more effective
FT/ST: 1 controlled trialPsychodynamic therapy more beneficial on functioning of the ego and cognition, less beneficial for relapses.
CBT/SR: 1 controlled trialFamily therapy more effective on improvement of residual symptoms.
Schizophrenia (acute period)CBT/SRT: 1 controlled studyCBT more effective
CBT/ST: 1 controlled studyCBT more effective
CBT/ST psychoeducation: 1 controlled trialDifference not significant: trend toward fewer relapses with CBT.
Eating disorders
BulimiaPT/CBT/FT: 1 meta-analysis, 1 controlled trialNo significant difference between the therapies
PT/CBT: 3 controlled trialsLittle difference between the therapies or inconsistent results
PT/FT/ST: 2 controlled trialsLittle difference between the therapies
CBT/MT: 2 controlled trialsNo difference between the therapies or inconsistent results
CBT/IT: 2 controlled trialsLittle difference between the therapies or inconsistent results
AnorexiaPT/CBT: 2 controlled trialsLittle difference between the therapies or inconsistent results
PT/FT/ST: 1 controlled trialLittle difference between the therapies or inconsistent results, although improved efficacy of family therapy for recent anorexia.

CBT, cognitive-behavioural therapy; PT, psychodynamic therapy; ST, supportive therapy; IT, interpersonal therapy; FT, family therapy; SRT, standard recreation therapy; "others", PT, ST, IT analysed in a single category; MT, motivational therapy.

Schizophrenia is characterised by psychotic symptoms (such as hallucinations or delusions) present during the active phase and frequently by persistent deficits in the area of social or occupational functioning in the residual phase. The different psychotherapies have been evaluated in this disorder as a supplement to standard medical treatments. The two comparative meta-analyses on the subject examined more than 100 trials, including various forms of psychotherapy in the treatment of schizophrenia (symptoms in the acute or residual phase, relapse rates). Both found that family therapies (psychoeducation) had the greatest effects at the end of treatment: the effects of CBT were equivalent or slightly inferior to those of the family approach. Psychoanalytical or psychodynamic treatments were the least effective. However, few if any specific comparisons were performed between therapies (two by two). Six randomised controlled trials of different psychotherapies in schizophrenia were examined: four compared CBT to other psychotherapeutic approaches, with follow up for 9 to 24 months. CBT was found to be more effective than psychodynamic therapy (1 trial), recreation therapy (1 trial), and supportive therapy (2 trials). In one trial that examined the efficacy of psychodynamic therapy compared to so-called "reality adaptation/supportive" therapy with a follow-up period of 2 years, the authors found little differences in efficacy. Finally, a comparison between "personal" therapy (a coping approach or adaptation strategy centred on stress reactivity), family therapy, and supportive therapy conducted over a period of 3 years found none of the therapies to be superior overall to the others, although different gains were found depending on the patient's home conditions (whether the patient lived alone or in a family setting).

The trials have shown several forms of psychotherapy to be effective in eating disorders. The only meta-analysis that examined bulimia did not find any significant difference between the various therapies tested (psychodynamic therapy, CBT, family therapy, standard support). Various forms of psychotherapy (psychodynamic, CBT, family therapy, interpersonal therapy, supportive therapy) were compared in 9 randomised controlled trials on bulimia conducted after 1992 (the date of the meta-analysis described above). Although significant differences were found, the results of the trials were inconsistent, and no general conclusion can be drawn as to the superiority of one approach over another. The literature is far more limited for the treatment of anorexia (3 controlled trials) and does not demonstrate any significant difference in efficacy between the therapies: psychodynamic, CBT, family therapy, or supportive therapy.

Similarly, there is insufficient literature on the other disorders in adults, notably alcohol dependency and personality disorders, to allow the different psychotherapies considered in this review to be compared.

Finally for disorders in children and adolescents, the meta-analyses show that at the end of psychotherapy a significantly larger number of children treated are improved compared to untreated control subjects, and that the mean effect sizes of psychotherapy are similar to those reported for adults. In addition, the effects of psychotherapies are generally superior for CBT compared to non-CBT treatments (psychodynamic therapy and supportive therapy) and slightly superior for family or supportive therapies compared to psychodynamic treatments. Considering the findings by specific disorder, there are relatively few comparisons between approaches except for those on depression and anxiety disorders. For these syndromes, the comparative trials in children and adolescents have tended to demonstrate similar results to those described for adults. For example, the four recent trials of good methodological quality on depression, with follow-up to 2 years, mostly found CBT to be generally equivalent to interpersonal therapy and to be more effective compared to the other approaches. Family therapy was more effective than other therapies in one of these trials, particularly for family problems and parent-children relationships. No trials have included the psychodynamic (psychoanalytical) therapies. Overall, and despite some similarities with the results obtained in adults, the smaller number of trials in children and adolescents and their methodological weaknesses make it impossible to identify the differences in efficacy between the different forms of psychotherapy as clearly.

Levels of proof in adults demonstrating superior efficacy between the three approaches studied, established by comparative trials

Superior efficacy proven
Generalised anxiety disorderCBT versus psychodynamic therapy
Major depressionCBT versus psychodynamic therapy
Superior efficacy presumed
SchizophreniaPsychoeducational family therapy versus psychodynamic therapy
CBT versus psychodynamic therapy

Levels of proof in adults demonstrating superior efficacy of one therapy compared to supportive therapy

Superior efficacy proven
Generalised anxiety disorderCBT
Panic disorderCBT
Social phobiaCBT
Post-traumatic stressCBT
Major depressionCBT
Superior efficacy presumed
SchizophreniaPsychoeducational family therapy

What factors can be assessed to determine which therapy is suitable for which disorder?

For each of the three types of approaches examined in this review—psychodynamic (psychoanalytical therapy), CBT, and family therapy—we have reviewed the trials which have assessed their efficacy compared to a control group. The results of trials comparing the efficacy of these therapies are also described. Finally, these analyses allow us to describe those therapies that are likely to bring patients benefit in their care for each of the disorders.

Published findings on patients suffering from acute phase or hospitalised schizophrenia have established the efficacy of family therapies combined with antipsychotic agents on the 2-year relapse rate (1 meta-analysis), the presumed short-term efficacy of cognitive therapies combined with antipsychotic agents (1 meta-analysis), and the lack of effect of psychodynamic therapies alone or combined with drug treatment (1 good quality meta-analysis, which only, however, used data from 3 old trials, in the absence of other usable published data).

Diseases Trials considered Main results
Schizophrenia3 meta-analyses, CBTCBT moderately effective on the 2-year relapse rate in schizophrenic patients in the acute or hospitalised phase, combined with antipsychotic agents.
CBT effective in stabilised patients followed on an outpatient basis, combined with antipsychotic agents.
1 meta-analysis, CBT and FT (PE
1 meta-analysis, PE
Family therapies effective on the 2-year relapse rate in acute phase or hospitalized schizophrenics combined with antipsychotic agents.
3 meta-analyses, FT (of which 2 with PE)Psychoeducational approach effective (family or patient centred) in stabilised patients followed up on an outpatient basis, combined with antipsychotic agents.
1 meta-analysis, PT
1 systematic review, PT
No efficacy demonstrated for PT on clinical course of schizophrenic patients.

CBT, cognitive-behavioural therapy; FT, family therapy; PE, psychoeducational approach; PT, psychodynamic therapy.

For stabilised schizophrenic patients followed on an outpatient basis, 3 meta-analyses have established the efficacy of the cognitive-behavioural approach in combination with drug therapy on social skills acquisition or improvement in emotion management, with a mean follow-up period of 5 months. The psychoeducational approach combined with drug therapy has also been proven to be effective on the 1- and 2-year relapse rates when the family approach is used (2 meta-analyses) and on the 18-month relapse rate when it is patient centred (1 meta-analysis). The psychodynamic approach has not been established to be effective on the patients' clinical course (from 1 meta-analysis and 1 review), even when combined with antipsychotic agents. Direct comparisons between the various psychotherapeutic approaches themselves have established that cognitive-behavioural therapies and the psychoeducational approach are more effective.

For the mood disorders, the information available on bipolar disorder only relates to the psychoeducational approach, which was shown to be effective in combination with drug therapy on global functioning and compliance with treatment when the psychotherapy was family based (marital) (1 controlled trial) and on the time to development of relapses of mania (but not depression) at 18 months when it was only patient based (1 controlled trial). Notably, no comparisons have been performed between the psychotherapies designed for the treatment of bipolar disorder.

It has been established for depressive disorders in hospitalised patients that cognitive-behavioural therapies combined with antidepressants have effects on depressive symptoms (1 meta-analysis). Family psychoeducation has short-term effects on global patient functioning (1 controlled trial), and psychodynamic therapies have effects on social adaptation and the length of the patient's hospitalisation (1 controlled trial). The level of proof of efficacy in this indication is greater for the CBT, and the controlled trials that compared psychodynamic and cognitive-behavioural approaches have concluded that the cognitive-behavioural approaches are superior.

For mild or moderate depressive disorders treated on an outpatient basis, cognitive therapies have been proven to be more effective than antidepressant treatments (2 meta-analyses). Interpersonal therapies are similar in efficacy to cognitive therapies (1 meta-analysis). There are few data on the psychodynamic therapies. In contrast to the interpersonal therapies, these have not been shown to be similar in efficacy to the CBT (1 meta-analysis). Couple therapies also appear to be effective for people living with a critical partner (1 controlled trial).

The cognitive-behavioural therapies have been studied by far the most in anxiety disorders. Their efficacy is best established in panic disorder, whether or not associated with antidepressants (2 meta-analyses), in general anxiety disorder, whether or not associated with medical drug treatments (1 meta-analysis), in the post-traumatic stress disorder (2 meta-analyses, 1 examining EMDR), in obsessive–compulsive disorders (3 meta-analyses), in social phobias (3 meta-analyses), and in various specific phobias (6 controlled trials).

Diseases Trials considered Main results
Mood disorders
Bipolar disorder2 controlled trials, PE (one of which involving FT)Efficacy of the family psychoeducational or individual approach, in combination with medical drug treatment.
Depressive disorders in hospitalised patients1 meta-analysis, CBTEfficacy of CBT on depressive symptoms
1 controlled trial, PE (FT)Efficacy of short term family psychoeducation on global functioning in combination with medical drug treatment.
1 controlled trial, PTEfficacy of PT on social adaptation and length of hospitalisation, in combination with medical drug treatment
Mild or moderate depressive disorders2 meta-analyses, CBT
1 meta-analysis, CBT and IT
Efficacy of CBT and IT
1 meta-analysis, CBT and VTEfficacy of verbal therapies, including possibly PT
1 meta-analysis, CBT and PT in the elderlyEfficacy of CBT in the depressed elderly
Major depressive disorders1 controlled trial, FT (couple)Efficacy of FT (couple) in subjects living with a critical partner

PE, psychoeducational approach; FT, family therapy; CBT, cognitive-behavioural therapy; PT, psychodynamic therapy; IT, interpersonal therapy; VT, verbal therapy.

The brief psychodynamic therapies are effective when used in combination with antidepressant treatment in preventing relapses of panic disorder, 9 months after the antidepressant treatment has been stopped (1 controlled trial), although with a lower level of proof than the CBT. They also are of presumed efficacy in the post-traumatic stress state (1 controlled trial) but have not to date been studied in other anxiety disorders. Comparisons between the different approaches (including comparative meta-analyses) have also shown the CBT to be the most effective therapies for all of the anxiety disorders.

Diseases Trials considered Main results
Anxiety disorders
Panic disorder1 meta-analysis, CBTEfficacy of CBT in prevention of relapses
1 controlled trial, BPTEfficacy of CBT in prevention of relapses
Generalised anxiety disorder1 meta-analysis, CBTEfficacy of CBT
Post-traumatic stress2 meta-analyses, CBT
1 controlled trial, PT
Efficacy of CBT and PT
Obsessive compulsive disorder3 meta-analyses, CBTEfficacy of CBT
Social phobia3 meta-analyses, CBTEfficacy of CBT
Specific phobia6 controlled trials, CBTEfficacy of CBT

CBT, cognitive-behavioural therapy; BPT, brief psychodynamic therapy; PT, psychodynamic therapy.

Cognitive-behavioural therapies, whether or not combined with drug therapy, have been shown to be effective in bulimia nervosa (6 meta-analyses), as have the interpersonal therapies (1 meta-analysis, 1 controlled trial).

In anorexia nervosa, family therapies have been proven to be effective, but only in young patients in whom the disorder has been present for less than 3 years for up to 5 years of follow-up (3 controlled trials, 1 non-meta-analysis systematic review). The cognitive-behavioural approach has not been shown to be effective on symptoms (disparate results in one non-meta-analytical systematic review and 2 controlled trials), although it has presumed efficacy in preventing relapses (1 controlled trial).

Diseases Trials considered Main results
Eating disorders
Bulimia nervosa6 meta-analyses, CBTEfficacy of CBT clearly established, efficacy of IT established in prevention of relapses
1 meta-analysis, IT
1 controlled trial, IT
1 systematic review, EP
Possible efficacy of the PE approach
Anorexia nervosa5 controlled trials, FTEfficacy of FT established in young patients
1 systematic review, FT (CBT)
1 systematic review, CBT
Efficacy of CBT not demonstrated (except in the form of FT)

CBT, cognitive-behavioural therapy; IT, interpersonal therapy; PT, psychodynamic therapy; PE, psychoeducational approach; FT, family therapy.

Borderline personality has been the most studied of the personality disorders; psychodynamic therapies (1 meta-analysis and 1 controlled trial) have been shown to be effective from 18 months to 4 years of follow-up. The cognitive-behavioural therapies have also been shown to be effective at 1 year of follow-up (1 meta-analysis and 5 controlled trials).

The psychodynamic (psychoanalytical) therapies and cognitive-behavioural therapies appear to be effective at 7 months of follow-up for antisocial personality, when the people concerned are also depressed (1 controlled trial), and at 6 months of follow-up in certain personality disorders (1 meta-analysis and 1 controlled trial).

As for eating disorders, no controlled trials have yet established any one therapy to be more effective than another in personality disorders.

In problems relating to alcohol dependency or abuse, family therapies (2 meta-analyses, 1 systematic review, and 1 systematic trial) and cognitive-behavioural therapies (1 meta-analysis and 1 controlled CBT trial) have shown the family therapies still to have presumed efficacy. Psychoanalysis-derived therapies have not been studied in this situation.

The comparisons between psychotherapies performed to date have concluded that the motivational therapies are as effective as the cognitive-behavioural therapies for problems associated with alcohol abuse or dependency (1 controlled trial).

Diseases Trials considered Main results
Personality disorders 1 meta-analysis, PT and CBT
1 controlled trial, PT, CBT
5 controlled trials, CBT
2 controlled trials, PT
Efficacy of PT and CBT for personality disorders (particularly borderline and antisocial types if associated with depression)
Alcohol dependency 2 meta-analyses and 1 systematic review, FTEfficacy of FT in maintaining abstinence
1 meta-analysis and 1 controlled trial, CBTEfficacy of CBT in maintaining abstinence
1 controlled trial (MT, CBT)Comparable efficacy of MT and CBT

CBT, cognitive-behavioural therapy; PT, psychodynamic therapy; FT, family therapy; MT, motivational therapy.

Finally, no evidence of short- or long-term symptom substitution or displacement was found in the trials evaluated for any therapy or disorder considered by this expert review.

Levels of proof of efficacy of three psychotherapeutic approaches examined in the adult*

Efficacy Proven (1) or Presumed (2)
Schizophrenia (acute phase) with medical drugsFamily psychoeducational therapy on 2-year relapse rates (1)
CBT approach (2)
Schizophrenia (stabilised, followed upon outpatient basis) with medical drugsFamily psychoeducational approach (1)
CBT approach (social skills acquisition, emotion management) (1)
Depression, hospitalised on antidepressantsCBT approach (1)
Bipolar disorder with medical drugsFamily psychoeducation approach (marital) and CBT approach (2)
Moderate depressionCBT approach (1)
Panic disorderCBT approach (1)
Brief psychodynamic approach with antidepressants (2)
Post-traumatic stressCBT approach (of which EMDR) (1)
Brief psychodynamic approach (2)
Anxiety disorders (GAD, OCD, phobias)CBT approach (1)
BulimiaCBT approach (1)
AnorexiaFamily therapies in young patients (2)
CBT approach for prevention of relapses (2)
Personality disordersPsychodynamic approach (1)
CBT approach (1)
Alcohol dependencyFamily therapy and CBT approach in maintenance of abstinence (1)

CBT, cognitive-behavioural therapy; EMDR, eye movement desensitisation and reprocessing.

* See previous tables for levels of proof of the therapies applied to children.


In conclusion, the literature review conducted in this expert evaluation has enabled us to produce a summary of the trials that have examined evaluations of psychodynamic (psychoanalytical) therapy, cognitive-behavioural therapy, and family and couple therapy consistent with current recognised scientific criteria.

The objective of this report is to assist decision-making in public health. It is based on the results of controlled trials conducted in the clinical population that are appropriate for this purpose, and without ignoring the methodological limits of such an exercise that are discussed at the beginning of the review. The major criterion used to evaluate the efficacy of therapies is improvement in clinical symptoms. Other criteria, such as improvement in functioning of the person, the person's quality of life, and social adjustment, have also been taken into account in some of the analyses. The review conducted in this expert evaluation allows the efficacy of each of the three approaches to be assessed, when used alone compared to no treatment (placebo or waiting list) and depending on the disorders in question. Depending on the disorder, some approaches appear to be more effective than others (see the table in the preceding section, Levels of proof of efficacy of three psychotherapeutic approaches examined in the adult).

The conclusions that follow from this analysis and from the review of the evaluation studies contained in the literature represent an information source of use to professionals and users. Although the individual relationship between the person who is suffering and the therapist remains a key factor in the choice and execution of a therapy, informing users and training of therapists must be consistent with the available scientific proof. These are two major factors in improving the healthcare offered and in the networking of different healthcare workers.



Unafam, Union nationale des amis et families de malades psychiques (national union of friends and families of patients with mental diseases).


Fnap-psy, Fédération nationale des associations de patients et ex-patients en psychiatrie (national federation of associations of psychiatry patients and ex-patients).



The effect size is the mean value of the treatment group less the mean value of the control group, divided by the standard deviation of the control group.


This is equal to the difference in score after treatment, less the score before treatment, divided by the standard deviation.


MSI, McGlasham semi-structured interview; ECPD, change in dynamic psychotherapies scale Kapp Karolinsk psychodynamic profile.


PACS-SE, Penn adherence-competence scale for supportive-expressive therapy; TIRS, Therapist intervention rating system; PTS, Perception of technique scale; GIS, General interpersonal skill; STT, Specific therapeutic technique.


CALPAS: California psychotherapy alliance scales


These therapies are known by various names (general psychotherapy, supportive psychotherapy, non-directive therapy, etc.). Some humanist approaches, including the Rogerian therapy, are also considered to be supportive therapies because they consider empathy and the support from therapists as being fundamental mechanisms for change.


An existential phenomenon therapy of F. Perls (1940), based on gaining awareness of acts and emotions and on beliefs and self-acceptance and self-esteem.

Created: 2004.

Copyright © 2004, Institut national de la santé et de la recherche médicale (INSERM)
Bookshelf ID: NBK7123PMID: 21348158


Other titles in this collection

Related information

Similar articles in PubMed

See reviews...See all...

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...