Depression-Focused Psychotherapies in the Treatment of Severe Depression: Are They Contraindicated?
The finding of an apparent association between increased symptom severity and poorer response to CT in the TDCRP is somewhat controversial, for both sociopolitical and empirical reasons. For example, not all studies have observed such an association, and among those that have, only the reports from the TDCRP suggested that this association was specific to CT; that is, it is not apparent for other active treatments.
The issue of illness severity and psychotherapy response is broader than one pertaining to CT. For example, the American Psychiatric Association’s (1993) “Practice Guideline for Major Depressive Disorder in Adults” used severity and/or melancholia as the principal clinical indicator for selection of pharmacotherapy instead of or in addition to psychotherapy. Indeed, some data can be marshaled in support of these recommendations (Thase and Friedman 1999). In their meta-analysis of psychotherapeutic studies of depression, Robinson and colleagues (1990) found that pretreatment severity predicted poor outcomes across various modes of psychotherapy. Similarly, in a meta-analysis pooling the results of more than 600 patients with major depressive disorder treated in standardized protocols at the University of Pittsburgh (Thase et al. 1997b), both IPT and CT were found to be significantly less effective than the combination of IPT and pharmacotherapy among patients scoring 20 or more on the HAM-D. By contrast, psychotherapy alone was nearly as effective as combined treatment in the subset of less severely depressed patients (
Shared Features of Depression-Focused Psychotherapies
This controversy centers partly around the competing formulations of major depression as either a predominantly psychosocial or a predominantly biomedical phenomenon, with adversaries generally taking sides according to psychologic or psychiatric (e.g., American Psychiatric Association 1993) orientations. Such debates are never fully resolved by the results of empirical studies, and, as discussed below, the subject becomes even more controversial when the evidence is mixed.
The proposition that clinical severity can be used to determine the need for pharmacotherapy should have two alternate sources of validation. First, clinical assessments of the signs and symptoms of melancholia or endogenous depression should also be expected to be correlated with poorer psychotherapy outcomes, because the neurovegetative signs and symptoms of melancholia are presumed to be more reflective of disturbances of underlying brain function (Free and Oei 1989; Rush and Weissenburger 1995; Thase and Friedman 1999). Second, an even stronger association should be observed between objective neurobiological correlates of severe or endogenous depression, such as hypercortisolemia (Thase et al. 1996a) or abnormal electroencephalographic sleep patterns.
Among the investigations specifically assessing endogenous subtypes and/or symptoms in patients treated with depression-focused psychotherapies, several have reported the predicted relationship between the more “biologically disturbed” subtype and poorer therapy outcomes (Gallagher and Thompson 1983; McKnight et al. 1992; Prusoff et al. 1980). In addition, endogenously depressed patients were found to be more likely to drop out of psychotherapy than were nonendogenously depressed patients in two other studies. Others’ findings do not support a relationship between endogenous symptoms or subtype diagnoses and poor response to IPT, behavior therapy, or CT. Such a “split ballot” could result from either a small to modest effect size or problems with the reliability or validity of the construct. Both explanations are likely to be operative.
This issue undoubtedly depends on the breadth of syndromes diagnosed as endogenous depression. For example, a majority of the studies examining response to psychotherapies have employed either the more inclusive Research Diagnostic Criteria definition of endogenous depression or symptom scores extracted from rating scales (Kovacs et al. 1981). Blackburn and Moore (1997) were unable to test the predictive value of the Newcastle Diagnostic Index (Carney et al. 1965) classification because only about 10% of their sample met this restrictive subtype. Furthermore, all of these studies have excluded patients with two of the best-validated endogenous subtypes, psychotic depression and bipolar depression.
It is also true that the relationships between various measures of clinical severity or endogenicity and response to antidepressant pharmacotherapy are not that well established. In fact, several groups have reported poorer responses to tricyclic antidepressants in more severely depressed patients than in less severely depressed patients (Abou-Saleh and Coppen 1982; Hollon et al. 1992a; Kocsis et al. 1990; McKnight et al. 1992). In McKnight et al.‘s (1992) comparative study, both CT and amitriptyline were significantly less effective among the patients with melancholia compared with the remainder of the sample.
Studies using electroencephalographic sleep profiles as an indicator of neurobiological dysfunction have generally failed to detect differences between psychotherapy responders and nonresponders, particularly when reduced rapid eye movement (REM) latency is used as the “marker”. In fact, our group found that patients with reduced REM latency had more favorable response to CT on some analyses (Simons et al. 1995). This finding has been tempered by recognition that reduced REM latency is more of a state-independent (trait-like) vulnerability marker than a severity (state-dependent) marker (Kupfer and Ehlers 1989; Thase et al. 1996b).
More promising results have been reported in studies utilizing more sophisticated methods of classifying sleep. The Pittsburgh group has completed a series of studies of depressed outpatients using a composite measure of electroencephalographic sleep dysfunction based on three variables: reduced REM latency, decreased sleep efficiency, and increased REM density (Thase et al. 1997c). In the first two studies (Thase et al. 1996b, 1997a), depressed patients with abnormal sleep profiles were about 20% less likely to respond to CT or IPT than were patients with normal profiles (
Figure 40-8). In the IPT study (Thase et al. 1997a), 75% of the psychotherapy nonresponders with abnormal sleep studies responded to antidepressants, indicating that these patients’ disorders were not refractory to all treatments. In the third study, Buysse et al. (1999) were unable to confirm a relationship between the specific sleep profile developed by Thase et al. (1997c) and response to IPT. However, this study had a relatively homogeneous sample (premenopausal women with recurrent depression) and the prevalence of sleep abnormality in this study (26%) was significantly lower than that observed in the earlier work of Thase et al. (1996a, 1997c). Further, Buysse et al. (1999) did find significant associations between other related sleep abnormalities and poorer IPT response.
Studies using various measures of hypercortisolism as an alternative index of state-dependent neurobiological disturbance have yielded more consistent results. In both inpatient (Robbins et al. 1989; Thase et al. 1996a) and outpatient (Corbishley et al. 1990; McKnight et al. 1992) studies, hypercortisolism has been associated with poorer responses to psychotherapy (e.g.,
Figure 40-9). Because hypercortisolism also has been associated with poorer responses to placebo treatment but not active pharmacotherapy (Ribeiro et al. 1993), these studies provide the strongest source of evidence collected to date in support of more traditional views of the relative merits of psychotherapy and pharmacotherapy. However, it should be noted that measures of hypercortisolism, such as an abnormal dexamethasone suppression test or elevated urinary free cortisol excretion, are abnormal in as few as 15%-30% of depressed outpatients (Thase and Howland 1995). Thus, if there is a subgroup of patients with nonpsychotic major depressive disorder that is too neurobiologically impaired to benefit from psychotherapy, this less responsive grouping is not likely to account for more than one-quarter of psychiatric outpatients.
Revision date: June 22, 2011
Last revised: by Jorge P. Ribeiro, MD