Le rapport de l’Inserm, évaluant les psychothérapies, date de février 2004. Notre article propose, à un lectorat francophone, une revue systématique des travaux publiés, depuis 2003, et concernant l’efficacité des psychothérapies psychanalytiques sur les troubles mentaux.

Une recherche systématique sur le site PubMed a identifié 71 articles de méta-analyses, publiés entre 2004 et 2019, inclus. Parmi eux, 25 articles ont été jugés pertinents et sont présentés en détail. Nous avons aussi recherché des études de long terme, publiées depuis 2013 et non inclues dans les méta-analyses. Trois études contrôlées et randomisées, comparant une thérapie psychanalytique à des traitements actifs, ont été identifiées.

Onze articles de méta-analyses ont évalué l’amélioration des symptômes à l’issue du traitement. Ils concluent tous que l’effet des psychothérapies psychanalytiques est robuste. Dix articles de méta-analyses ont comparé une psychothérapie psychanalytique à un traitement actif (e.g. autre psychothérapie, médication). Un seul rapporte une infériorité cliniquement significative des psychothérapies psychanalytiques par rapport aux psychothérapies cognitivo-comportementales. Cinq articles de méta-analyse et trois études randomisées ont comparé des psychothérapies psychanalytiques de long terme (plus d’un an) à divers traitements actifs. Tous ont conclu à une efficacité des psychothérapies psychanalytiques égale ou supérieure à celle des traitements actifs de comparaison.

Les études concernant les psychothérapies psychanalytiques se sont multipliées depuis 2003 dans les revues biomédicales. Elles montrent que, pour la plupart des troubles mentaux fréquents, les psychothérapies psychanalytiques sont aussi efficaces que les psychothérapies cognitivo-comportementales. Cette absence de différence est discutée.

A French governmental institute published, in February 2004, a report assessing the efficacy of psychotherapies in the light of the biomedical literature. It concluded that cognitive psychotherapies effectively cure common mental disorders, while the efficacy of psychodynamic therapies is not proven by scientific studies. Because many French mental health professionals are practicing with reference to psychoanalysis, this conclusion stirred up heated controversy. Since February 2004, numerous studies assessing psychodynamic therapies have been published in peer-reviewed biomedical journals. Moreover, these primary studies have been meta-analyzed in dozens of review articles. Here, we systematically review these meta-analysis articles.

A systematic search for meta-analyses assessing psychodynamic therapies was performed using PubMed and identified 71 articles published from January 2004 to December 2019. Among them, 25 articles were judged to be relevant because they reported meta-analyses assessing the symptoms of common mental disorders in at least three distinct cohorts of adult patients. Although the primary studies included in these 25 meta-analysis articles often overlap, the selection criteria, calculation methods and results always differ between them. Therefore, we reviewed all of them without further selection. From all the meta-analyses reported in these 25 articles, we systematically present here the most compelling ones, i.e. those calculated from the largest number of primary studies. Results were quantified in terms of effect size (i.e. standardized mean difference). Effect sizes below 0.25 were considered as without clinical significance, whereas those superior to 0.8 were regarded as robust. Because short-term psychodynamic therapies had been assessed in 20 meta-analysis articles published until 2017, we did not search for more recent primary studies. However, because the most recent meta-analysis article about long-term psychodynamic therapies was published in 2013, we also searched, using PubMed, for primary studies assessing psychodynamic therapies lasting for at least one year and published from January 2013 to December 2019. Among the 57 publications retrieved by PubMed, three were identified as randomized controlled trials not included in meta-analyses and were extensively described here.

Eight meta-analysis articles have assessed symptom improvement at treatment termination by comparing with baseline symptoms. According to all of them, psychodynamic therapies alleviate symptoms and their effect sizes are always robust. Three meta-analysis articles compared psychodynamic therapies with inactive treatments (e.g. placebo medication, waiting list) and reported clinically significant differences in favor of psychodynamic therapies. Ten meta-analysis articles compared, at treatment termination, psychodynamic therapies to active treatments, including medication and cognitive psychotherapies. Nine of them reported no difference. Only one article concluded that psychodynamic therapies are clinically inferior to cognitive psychotherapies (d=−0.28). Seven meta-analysis articles compared psychodynamic therapies to active treatment at follow-up (i.e. months or years after treatment termination). Five of them reported no significant difference, one reported a medium effect size in favor of psychodynamic therapies over various active treatments (d=0.38), while the other reported a clinically significant difference in favor of cognitive psychotherapies (d=−0.55). Because short-term treatments are often insufficient to prevent relapse, investigations about long-term treatments (i.e. more than one year) are needed, but such published studies are still scarce. Five meta-analysis articles and three primary studies published since 2013 compared long-term psychodynamic therapies to various active treatments of similar duration. According to them, psychodynamic therapies were at least as effective as other active treatments.

A systematic review about psychodynamic therapies, published in 2015 in Lancet Psychiatry, included 64 randomized controlled trials of which 37 were published after 2003. Therefore, most quality studies assessing psychodynamic therapies have been published since 2003 and have been reviewed in recent meta-analysis articles. All together, this recent literature leads to the conclusion that psychodynamic therapies are as effective as active treatments, including cognitive psychotherapies, to help patients suffering from common mental disorders (unipolar depression, anxiety disorders, eating disorders and personality disorders). Beside this overall conclusion, it appears that randomized controlled trials are not well suited for answering why psychotherapies work in some patients but not in others, and how they work in general. Other approaches are needed, including case studies.