Durée de traitement et évolution clinique en psychothérapie: enjeux de la relation dose-effet

In the last years, reflecting political concerns about allocation of financial resources, clinical research has addressed the question of the relation between number of sessions of psychotherapy (dose) and patient clinical evolution (effect). Howard et al. (1986) conceptualised the path of patient improvement as a negatively accelerating function of treatment length. Barkham et al. (1996) reported a similar dose-effect relationship. More recently, change has been evaluated on a session-by-session basis to overcome the pitfalls of the previous pre-post study designs (Kadera et al. 1996, Anderson and Lambert 2001, Hansen and Lambert 2003). The rate of improvement might vary depending on the characteristics of the symptoms (Kadera et al. 1996, Anderson and Lambert 2001). Kopta et al. (1994) showed that the effective dose varied across different types of symptoms. Fewer sessions were required for improvement in distress (acute) symptoms, and more sessions for characterological symptoms. Patient characteristics might also be important. Psychiatric outpatients were shown to recover or improve more slowly than patients consulting for adjustment problems (Hansen and Lambert 2003). Finally, treatment characteristics might influence patient evolution (Barkham et al. 1996). Methodological issues are also at stake. Defining the primary outcome measure as improvement (Howard et al. 1986, Anderson and Lambert 2001, Hansen and Lambert 2003) or as recovery (Anderson and Lambert 2001, Hansen and Lambert 2003) changed the dose-effect curve. Barkham et al. (1996) observed that session-by-session plots of improvement looked more or less linear. On an individual basis, treatment takes a given number of sessions to reach a good enough level (GEL), at which point the patient and therapist either redeploy therapeutic efforts to other problems or discontinue therapy (Barkham et al. 2006). This endpoint depends on process variables that cannot be apprehended by the data aggregated on the dose-effect curves. They define general profiles for particular populations of patients benefiting from various treatments. They do not predict the evolution of a single patient during his or her treatment. Clinical thinking, type of disorder, its gravity and its clinical characteristics, as well as patient expectations and motivations can still be regarded as actual indications of treatment length. In Switzerland, studies on the dose-effect relationship are needed. Future research should address this issue at a regional or national level.