Object: Antidepressant efficacy is different in controlled trials (RCTs) and naturalistic studies. This paper reviewed exclusion criteria that are usually applied to RCTs focusing on their limitations in extrapolating data to implement guidelines for the routine treatment of depression. In addition it ascertained the representativeness of a "naturalistic" sample of inpatients treated at our mood disorder center comparing their sociodemographic and clinical characteristics to the features of mood disorder individuals reported by epidemiological studies.Method: 508 patients who met DSM IV criteria for major depressive episode and either bipolar or major depressive disorder were compared to 316 nonpsvchiatric controls on their social profiles. Structured interviews and rating scales were used for the clinical evaluation of the patient sample.Results: Common exclusion criteria of RCTs are: 1) unavailable informed consent 2) age over 65 years 3) diagnosis of bipolar disorder 4) presence of psychotic symptoms 5) mild depression (RDRS<20) 6) comorbid psychiatric disorders (axis I) 7) concurrent medical conditions (axis III). Details of our research sample: 52% of patients had bipolar depression, 49% psychotic features, 86% positive family history of mood disorders, 48% comorbid personality disorders and 66% were classified as severe mood disorders by all these featttres and high recurrence of affective episodes. Depressed patients were not markedly different from controls in their social profiles. General population: about 30% of mood disorder cases in the general population are bipolar and 20%-25% have psychotic depression. Epidemiological studies show that mood disorder patients are more frequently single and unemployed and have lower education level than the normal population.Conclusions: Our naturalistic sample, recruited in a specialized research center, included many patients who are not usually enrolled in RCTs but were still somewhat different from the general population of mood disorder individuals as for sociodemographic and clinical features. Thus, if selection criteria limit the representativeness of RCTs, research setting is the main bias in naturalistic studies.

The need for additional strategies for medication adherence in unipolars, is striking. In fact, as well as bipolar forms, despite the planned long-term use of antidepressants, a variety of evidence suggests that adherence often fails within the first months of treatment after remission from a depressive episode. This requires appropriate patient education and support to develop a greater adherence to treatment. The aim of the present study was to verify the risk of relapse during a 6-months continuation period in two groups of unipolar patients participating (N=58) or not (N=41) in psychoeducational group sessions during the acute treatment of a depressive episode treated with fluvoxamine. At the end of follow-up period we observed a significantly (p<0.005) higher number of relapsed in the group of patients who did not participate in psychoeducational intervention (31.7% vs 13.8%). Among patients who participated in psychoeducational intervention the risk of relapse was affected by the number of psychoeducational group sessions performed (p<0.0001).