There are numerous pharmacological treatment options for the manic episodes but not the depressive episodes (
2). Lithium, divalproex, carbamazepine, lamotrigine, quetiapine, olanzapine, and fluoxetine exert some beneficial effects (
3–
6), while the use of antidepressants is controversial (
7,
8). Despite some differences in first-line choices, pharmacological treatment algorithms for bipolar depression include the same pharmacological agents (
3,
9–
12). It is difficult to document systematic evidence for the selection of any specific therapeutic choice for treatment-resistant bipolar depression (
13).
Many clinicians regard electroconvulsive therapy (ECT) as the most effective acute treatment in severe treatment-resistant mood and psychotic disorders (
14). The use of ECT in bipolar depression has not been extensively studied, but for severe refractory bipolar depression it is a second-line option in most guidelines (
3,
9–
12). These recommendations are based on clinical experience and the results from nonrandomized studies and a meta-analysis comparing the efficacy of ECT in unipolar versus bipolar depression (
15–
18). To our knowledge, no randomized controlled trials of ECT for the treatment of bipolar depression have been reported.
This study compared efficacy outcomes of ECT and algorithm-based pharmacological treatment in treatment-resistant bipolar depression, using repeated Montgomery-Åsberg Depression Rating Scale (MADRS) (
19) assessments as the primary outcome after a 6-week intervention period and, as secondary outcomes, repeated assessments with the Inventory of Depressive Symptomatology–Clinician-Rated, 30-item version (
20), and the Clinical Global Impressions Scale for Bipolar Disorder (CGI-BP) (
21), the response and remission rates, and the times to response and remission.
Discussion
To our knowledge, this is the first randomized controlled trial comparing the effects of ECT and pharmacological treatment in treatment-resistant bipolar depression. The main finding is that ECT is more effective than pharmacological treatment in the acute phase.
Using a linear mixed-effects modeling approach, we found that the mean MADRS score, the primary outcome measure, differed by 6.6 points between the ECT and pharmacological treatment groups. Similarly significant differences of 9.4 and 0.7 points between the two treatment groups were found for the secondary outcome measures: the 30-item Inventory of Depressive Symptomatology–Clinician Rated and CGI-BP scores, respectively. In a meta-analysis (
35) of a mixed group of patients with unipolar or bipolar depression, ECT was found to be significantly more effective than pharmacological treatment, with a mean difference of 5.2 points (95% CI=1.4–8.9) on the Hamilton Depression Rating Scale (
36).
There was no difference between the (low) remission rates of the ECT group (34.8%) and the pharmacological treatment group (30.0%). The response rate for ECT was significantly higher at 73.9%, which was considered a relatively successful outcome in this cohort of ill patients, in comparison to the response rate of 35.0% for the medication group. It should be noted that measures based on response or remission are a dichotomization of the MADRS score and thus generally produce less powerful results than linear mixed-effects analyses. The response and remission rates for ECT in the present study are consistent with those found by Medda and coauthors (
17) in an open trial of the effects of ECT on medication-resistant depression or mixed states in patients with bipolar disorder subtype I, at about 70% and 30%, respectively. Comparable results have also found been in patients suffering from treatment-resistant unipolar depression (
37). In contrast, patient groups not defined as having medication resistance often have somewhat higher response rates and substantially higher remission rates (
15,
16). This underscores the importance of describing the degree of treatment resistance in patients when comparing the effects of interventions in depression.
In the survival analyses, the times to response and remission did not differ significantly between the ECT and pharmacological treatment groups, but there was a tendency for both times to be shorter in the ECT group. These survival analyses involved time to the first occurrence of response or remission, and patients who dropped out were censored. Measurements made before such individuals drop out may contain valuable information, and we included them in the survival analysis. The results are not directly comparable to those of other studies, because of the analysis method used and the lack of previous randomized controlled trials comparing the effects of ECT and pharmacological treatment in bipolar depression. However, a rapid effect of ECT is often claimed, and a small study of patients with unipolar or bipolar depression found quicker responses among patients randomly assigned to ECT than in those assigned to paroxetine (
38).
Unilateral placement of stimulation electrodes has been found to be slightly less effective than bilateral placement (
35). However, studies using unilateral ECT at a low dose or with a short interelectrode distance, techniques known to be less effective, were included. Although a minority of patients do not respond to right unilateral ECT and need to be crossed over to bilateral ECT, a large scale study (
39) found that high-dose unilateral ECT with d’Elia electrode placement (
26) was as effective as bilateral ECT. Therefore, the modest remission rate found for ECT in our study was probably due not to the use of unilateral electrode placement but, rather, to the selection of patients with a low potential for remission.
The present study was subject to some limitations. Neither the patients nor the researchers were blinded, which may have biased the treatment outcomes. However, this is unlikely since the video-based ratings by the blinded raters were strongly correlated with the results of the initial evaluation. A group receiving sham ECT, to control for a possible placebo response in patients and bias in evaluators, was not included because of ethical considerations. The relatively small study group and high dropout rates limit the statistical power of the analysis and may be a source of type 2 errors (
22). Despite there being few differences in the characteristics of depressive episodes between bipolar disorder subtypes I and II (
40), and particularly in treatment-resistant depression, the inclusion of both subtypes may introduce heterogeneity. However, there are no indications that this should bias the findings, with this instead leading to type 2 errors. The most severely depressed patients were not included because of their inability to give informed consent or their psychiatrists’ opinion that they were in urgent need of ECT. We suspect that their exclusion reduced the observed effect of ECT, since there are some indications that ECT is particularly beneficial in cases of severe depression. Finally, the indications for and attitudes toward ECT in Norway may differ from those in other countries, with implications for the generalizability of the results of this study.
The requirement that ECT be indicated for patients according to the responsible psychiatrist may have caused bias across the centers. The number of patients recruited from each center was too low to correct for any such response differences. However, the severity of depression at inclusion did not differ significantly among the centers. Finally, although we did apply a recognized treatment algorithm, we cannot rule out that the algorithm used was not optimal. Thus, our study clearly requires replication with alternative algorithms and medication dosages.
The main strength of the current study is its randomized controlled design. Furthermore, the study was initiated by researchers and financed by public research funds and the participating hospitals. The psychiatric health care system in Norway is publicly funded and based on catchment area, ensuring a representative sample of patients with severe treatment-resistant bipolar depression. The use of an algorithm-based pharmacological treatment as a comparison condition for ECT made it possible for the researchers to include patients with resistance to several medications. This design ensures that the results may also be generalized to patients exposed to a high lifetime number of pharmacological treatments, which is common in bipolar disorder (
41).
To conclude, the current results show that ECT is more effective than pharmacological treatment in the acute phase of treatment-resistant bipolar depression, which supports ECT as a treatment option. The low remission rates found in this study highlight the need for research focusing on the detection of new and more effective treatment options for treatment-resistant bipolar depression.