Definition and epidemiology of treatment-resistant depression.

Our results suggest that between 29% and 46% of depressed patients fail to respond fully with antidepressant treatment of adequate dose and duration. In particular, although partial response appears to occur in 12% to 15% of the depressed patients studied, nonresponse is observed in 19% to 34% of this population. The prevalence of treatment-resistant depression derived from studies using ITT analysis is likely to be an overestimate of the actual occurrence of the phenomenon, as these rates also reflect the outcome of patients who were treated inadequately or were intolerant to the treatment. On the other hand, data derived from studies using completer analysis are likely to generate under-estimates of the prevalence of this phenomenon, as patients may have dropped out before completion of the study due to lack of efficacy. One could, therefore, guess that the actual rates of treatment resistance in the clinical population of depressed patients probably lie between these two types of estimates. From an epidemiologic point of view, because the prevalence of depression has been estimated to vary from 2.6% to 5.5% in men and from 6.0% to 11.8% in women, one must conclude that treatment-resistant depression is a very common clinical problem that is likely to affect more than one third of depressed patients. In summary, treatment-resistant depression patients can be defined as those who fail to respond to standard doses (i.e., significantly superior to placebo in double-blind studies) of antidepressants administered continuously for at least 6 weeks. Additional requirements of this operational definition are an accurate diagnosis of depressive disorder, patient compliance with the treatment, the use of valid outcome measures, and therapeutic range of antidepressant blood levels for the tricyclic antidepressants. Finally, symptomatic improvement that is equal or greater than 25% and less than 50% qualifies as partial response, and less than 25% symptomatic improvement is complete nonresponse.