A significantly number of depressed patients does not have a complete response to an adequate monotherapy with a first line antidepressant. We have too few controlled trials to evaluate when to switch to a different class of antidepressants or adding another treatment modality to the current treatment to which the patient is resistant. The symptom profile of these patients developing only partial response has been insufficiently analysed, because the summed total scores of the outcome depression scales have been considered sufficient.

To make a psychometric analysis at baseline to discriminate at the level of the individual symptoms between complete responders and non-responders in patients who have not responded to a first line monotherapy and who hereafter have to be augmented.

Baseline scores on the Hamilton Depression Scale (HAM-D) and the Montgomery-Åsberg Depression Scale (MADRS) were analysed by Principal Component Analysis (PCA) without rotation procedure. Within both scales the symptoms are more or less positively correlated and it was therefore the non-general principal component we looked for to classify the treatment resistant patients.

Two principal components were identified. The first component was a general factor while the second was a bi-directorial factor separating pure depression symptoms and arousal-related symptoms (including sleep problems, concentrations problems, and somatic anxiety symptoms).

It is very important to monitor both the pure depression symptoms and the arousal-related symptoms to prevent the development of treatment resistant depression. In this respect the total scores on HAM-D17 and MADRS are not in itself sufficient.