Two different methods were compared in the assessment of depressive symptomatology improvement: live naturalistic (N) performed by the patient's therapist, and from videotape record of structured clinical interview (VSI) assessed by an independent rater out of five psychiatrists.

Sixty-one newly admitted depressed inpatients, with a Montgomery and Asberg Depression Rating Scale (MADRS) total score above 20, were assessed before antidepressant treatment (DO), after 10 days (D10) and 4 weeks of treatment (D28). Assessments were based on the MADRS and the Depression Retardation Rating Scale (DRRS) for both N and VSI methods, and on the SCL-90-R for self-rating.

With the MADRS, the N method was shown to be more sensitive to symptomatology change than the VSI method, but the VSI method was more correlated to self-assessment than the N method was. However, these results were not replicated on the DRRS, for which an underscoring with the VSI method was evidenced as compared to the N method.

As shown in other studies, the poorest agreement between the two methods was evidenced at DO, suggesting a “novelty effect” particularly with the VSI method. This “novelty effect” may be all the more pronounced if a personality disorder is associated to depression. Consequently, information concerning each patient before rating videotapes is needed, as well as investigations in the field of depression and personality disorders.

The greater change observed in MADRS with the N method as compared to the VSI method, may be due non-specific factors related mostly to therapist expectations by comparison to neutral raters. This hypothesis should be tested in the placebo group of a double-blind study, and, if confirmed, the use of VSI methods, by minimising non-specific factors of improvement due to therapist expectations, may decrease the placebo response in antidepressant drug trials.