It is widely accepted that ECT is the most effective short-term treatment for severe depression. "Almost everybody responds" has been the clinical wisdom. While that is an exaggeration, for correctly selected patients given the best care, it is not too far off. Still, refining the accuracy of predicting who will, or more important, who will not, respond is a worthwhile endeavor.
Prognostication is a major part of what physicians do in many fields of medicine, and it is particularly relevant when a treatment or procedure is controversial or anxiety-provoking. Being able to accurately tell a prospective ECT patient how likely he or she is to respond would be helpful. Let's look at what is known about this area, what is controversial, and what remains to be done. I will limit this discussion to the diagnosis of a major depressive episode, either in the context of unipolar or bipolar disorder, and schizoaffective disorder.
Current literature
The literature on ECT response prediction is a field of broken dreams. From the 1950s onward, dozens of investigators have offered up indexes and rating scales, symptom clusters, multiple regression analyses of variables, and even meta-analyses of studies, searching for the holy grail of sensitive and specific ECT response prediction. This literature has been well reviewed both in the American Psychiatric Association's (APA) Task Force Report on ECT1 and by Abrams in the fourth edition of his textbook, Electroconvulsive Therapy.2 The conclusions are similar: systematic attempts to develop clinically useful rating indexes have largely failed, but the prudent, experienced clinician can usually tell who will respond to ECT. The corollary "who will not respond" seems harder to predict with certainty, and definitely not easily enough to deny ECT to a patient who is severely depressed merely because he does not fit the "profile."
Most of the ECT response prediction literature centers on the theme of being able to identify a characteristic "core" type of depression that is highly "biological," very severe, heritable, and typically episodic. Various names have been applied to this textbook type of severe depression, including "melancholic," "endogenous," and "endogenomorphic." Melancholia is typically defined as a type of depressive illness with disturbances of mood, psychomotor activity, and vegetative functions, and with psychosis present in up to 30% of patients.3 Thus, many researchers have offered what were termed "favorable" features for ECT response. They include sudden onset, shorter duration of episode, psychomotor retardation, early awakening, weight loss, older age, paranoia, and delusions (psychosis). Not all have stood the test of replication.
Response predictors
Psychosis is perhaps the best-established predictor of ECT response. More than a dozen studies, from the 1950s to the recent Consortium for Research in ECT (CORE) multicenter study, show better response rates for this severe form of depression, in which delusions, paranoia, or both are present.4-6 As discussed in the APA Task Force Report, this is noteworthy because the pharmacotherapeutic treatment of psychotic depression is particularly difficult.1 Here is a situation in which ECT is preferentially helpful in the patients with the most severe form of the illness. Whether psychosis is a marker of a distinct depressive biological subtype or simply an indicator of disease severity is unclear. The APA Task Force Report reached the oft-repeated conclusion, "The rapid onset and high probability of improvement with ECT makes this treatment of particular value for these patients [with psychosis]."1 Shorter time spent in the current episode has likewise been repeatedly (although not universally) associated with increased ECT response rates, a situation that also pertains to antidepressant medications.7 Some studies have not replicated this finding.8
Older age is another factor shown to be associated with favorable ECT response. This is curious, given the largely opposite finding in many other treatment situations in medicine. It is particularly opportune, since elderly patients often have the most difficulty in tolerating the adverse effects of psychotropic medications. The CORE study replicated this finding.9
Catatonia, a syndrome that has seen a renewed surge of interest in recent years, has been associated with good ECT prognosis.10 This is irrespective of the underlying diagnosis (catatonia may occur in affective disorders, schizophrenia, and some medical illnesses). ECT has been shown to be effective in an uncommon and life-threatening variant of the syndrome called "lethal catatonia."11 While the presence of catatonic features may be a positive prognostic indicator for ECT, these signs and symptoms occur infrequently enough in the depressed population that their absence cannot be used to suggest inferior ECT prognosis.
Negative predictors
On the negative side, a cluster of clinical features has been associated with a poorer response to ECT. These have included hypochondriasis, "neurotic" traits, hysterical personality, anxiety, dependency, lifelong depression, and somatization. In DSM terms, these might be considered characteristic of an Axis II disorder rather than essential features of a primary, melancholic Axis I disorder. In fact, the literature specifically investigating the relationship of outcome with ECT in patients with comorbid borderline personality disorder and depression suggests a poorer outcome acutely but is by no means conclusive.12,13 In the study by Feske and associates,12 patients with personality disorders other than borderline personality disorder responded as well to ECT as those with no personality disorder. Other types of secondary depressions, as in persons with preexisting medical or psychiatric disorders, are believed to have poorer responses to ECT.
The construct of "medication resistance" has received great attention in the literature in recent years, related to both pharmacotherapy and ECT. Data indicating that the degree of medication resistance (history of failed trials of antidepressants) is associated with diminished ECT response have come out of the large ECT trials conducted by Prudic and associates.14 Recently, the CORE group did not replicate this finding,15 and it remains intriguing but cannot yet be used as a factor in deciding whether ECT should be considered. "Decreased response" does not mean poor response in absolute terms, and the vast majority of patients for whom ECT is recommended have failed to remit with antidepressant medications.
After decades of successful ECT use but largely unsuccessful attempts to accurately predict response rates, investigators remain undaunted. Recently, de Vreede and associates16 offered a new index composed of 4 factors: age, psychosis, medication resistance, and personality disorder. They showed "reasonably" good predictive ability, but the overall 58% "poor response" to ECT in that study is anomalous, as is the finding that psychosis was associated with poorer response. The CORE investigators also remain interested in refining ECT response prediction and will continue to study clusters of clinical features from that large data set.
Even without a reliable instrument to accurately predict a patient's likelihood of response to ECT, this work has given clinicians a useful framework in which to make educated clinical judgments about the recommendation for ECT. Does it matter that we cannot tell a patient whether he will have a 60% or a 95% chance of remitting with a course of ECT? Yes, it does. That is why studies of this subject should continue. Of course, a biological marker or laboratory test, which is likely to be developed in the future (recall the dexamethasone(Drug information on dexamethasone) suppression test that almost succeeded), is what we really need.
