Eighty-five depressed women were treated with amitriptyline for four weeks. Prediction of outcome was examined using three alternative depressive typologies.
The first typology was a classification previously derived by multivariate cluster analysis and comprising four groups: psychotic depressives, anxious depressives, hostile depressives, and young depressives with personality disorder. This typology predicted outcome significantly. Psychotic depressives showed the greatest improvement and anxious depressives the least, with the remaining two groups intermediate. When effects of initial severity were controlled for by analysis of covariance, the poor response of anxious depressives was the main finding.
Clinical diagnoses of psychotic and neurotic depression were also used to classify patients. These assignments overlapped with the cluster analysis typology, psychotic depressives corresponding in both typologies, while anxious depressives, hostile depressives, and young depressives with personality disorder all tended to be diagnosed as neurotic depressives. However, the psychotic-neurotic dichotomy did not significantly predict improvement, apparently because it failed to isolate the anxious depressives from the heterogeneity of neurotic depression.
Patients were also assigned to the three groups of retarded, anxious and hostile depressives previously described by Overall and Hollister. This typology overlapped, although somewhat weakly, in group membership with the four group cluster analysis typology. Prediction could not be tested adequately since there were very few retarded depressives in this sample.
The findings are consistent with other reports that neurotic depressives show worse response than psychotic depressives to tricyclic antidepressants, but indicate that the poor response is particularly associated with one group of anxious depressives within the heterogeneity subsumed in neurotic depression.