Speed of electroconvulsive therapy for depression: Effects of electrode placement

Electroconvulsive therapy (ECT) is a rapidly effective treatment for severe depression. Treatment with right unilateral (RUL) or bitemporal (BT) ECT may explain individual differences in speed of ECT effectiveness. There is limited evidence for demographic and clinical factors that predict speed of response and remission with ECT. We aimed to investigate differences in speed of improvement as well as achieving response and remission between twice‐weekly brief‐pulse high‐dose (6 × seizure threshold) RUL ECT and moderate‐dose (1.5 × seizure threshold) BT ECT. We also explored demographic and clinical characteristics that predict speed of response and remission.


| 445
FOX and MCLOUGHLIn all depressive symptom dimensions. 5 The rapid effectiveness of ECT has direct clinical implications, particularly among patients with severe depression who require urgent treatment.
When measuring the speed of ECT effectiveness, numerous definitions and methodologies can be applied. 6 While speed of improvement with ECT is commonly described as the rate of change in depression severity over time, speed of ECT response or remission refers to the time taken to meet the defined criteria for response or remission. A reduction of at least 50% from baseline score on a mood rating scale such as the Hamilton Depression Rating Scale (HDRS) is typically used to define ECT response. 7 Remission with ECT is usually indicated by a HDRS score equal to or below 10. 7 Notably, there is variability in the trajectories of how quickly individuals improve with and respond to ECT. 8 Individual differences may be partially explained by variations in ECT parameters. Although brief-pulse ECT is more efficacious for treating depression than ultrabrief pulse ECT, 9,10 there may not be a significant difference in speed of response between brief-pulse and ultrabrief pulse unilateral ECT. 11 Beyond pulse-width, variations in electrode placement may influence the speed of ECT effectiveness. 12 Bitemporal (BT) ECT is the most widely used form of brief-pulse ECT electrode placement worldwide. 13 However, cognitive outcomes are more favourable with right unilateral (RUL) ECT than BT ECT. 2 Meta-analyses of relevant trials show that there are no significant differences between brief-pulse RUL ECT and BT ECT in terms of changes in depression rating scores, rates of response and remission, or the numbers of ECT sessions per course. 2 Yet, the differences between brief-pulse RUL ECT and BT ECT in the speed of symptom improvement, response and remission have rarely been studied and remain uncertain.
To our knowledge, a study from the Consortium for Research in Electroconvulsive Therapy (CORE) in the USA is the only study to date that directly investigated whether there was superior speed of improvement in depression symptoms with thrice-weekly BT or RUL ECT. 12 Compared with RUL ECT, BT ECT was associated with significantly lower HDRS scores at earlier time points in the ECT course. 12 From this, it was inferred that depression improved faster with BT ECT. 12 These findings have the potential to guide clinical decisionmaking. Faster speed of improvement might indicate consideration of BT, rather than RUL, electrode placement among individuals where urgent improvement is required, even though BT ECT may cause more cognitive side-effects. 2 However, the overall evidence supporting a more rapid speed of improvement with BT ECT is limited.
In addition to ECT parameters, demographic and clinical factors may influence how quickly patients meet response criteria with ECT. ECT is typically indicated among patients with medication-resistant depression. 14 However, no studies to date have investigated if treatment-resistant depression predicts a more rapid ECT response. Faster speed of response to ECT was predicted by higher baseline depression severity 15 but not by gender, current depressive episode duration or presence of psychotic symptoms. 11 There is conflicting evidence for age 11,16 and depression polarity 8,11,17,18 as predictors of speed of ECT response. It has been proposed that a superior overall response to ECT is potentially associated with levels of suicidality, 19 a positive family history of mood disorder 20 and presence of melancholic features, such as psychomotor disturbances. 7 However, the influence of suicidality, family history of mood disorder and melancholia on speed of ECT response have yet to be examined. Overall, clinical studies that investigated the predictors of speed of ECT response are scarce.
Relative to speed of response, variables that predict speed of remission to ECT have been less investigated. Remission status is indicative of a transition to sustained recovery 21 and predicts a lower risk of future depressive relapse. 22 Thus, the initial treatment goal with ECT is to achieve remission of depression symptoms. 23 Among depressed patients, the presence of psychotic symptoms predicts faster speed of remission to ECT. 24 To our knowledge, no studies to date have investigated whether speed of remission to ECT is predicted by differences in ECT electrode placement.

| Aims of the study
This study aimed to investigate the differences in speed of improvement as well as achieving response and remission between twice-weekly brief-pulse high-dose right

Significant outcomes
• There was no significance difference between BT and RUL ECT in speed of depression improvement, response or remission. • Speed of ECT effectiveness may not be a pertinent factor when making the clinical decision to prescribe BT or RUL ECT for depression. • Exploratory analyses revealed multiple demographic and clinical characteristics did not predict speed of response or remission with ECT.

Limitations
• Findings were based on secondary analyses of previous trial data. • For exploratory analyses, the sample size was reduced for group comparisons and the significance levels of multiple comparisons were not corrected for.
unilateral electroconvulsive therapy and moderate-dose bitemporal electroconvulsive therapy. We hypothesised that faster speeds of improvement, response and remission would be associated with bitemporal electroconvulsive therapy when compared to right unilateral electroconvulsive therapy. Additionally, we conducted exploratory analyses to identify demographic and clinical characteristics that may predict speed of response and remission with electroconvulsive therapy.

| Participants
Patients with depression were recruited over a 5-year period as part of the EFFECT-DEP Trial (Enhancing the Effectiveness of ECT in Severe Depression; ISRCTN23577151), which has been previously described. 25 Briefly, EFFECT-Dep was a pragmatic, patient-and rater-blinded, non-inferiority trial comparing twice-weekly, brief-pulse, moderate-dose BT (1.5 × seizure threshold) and high-dose RUL (6 × seizure threshold) ECT in real-world practice. Ethical approval for this study was provided by St Patrick's University Hospital Research Ethics Committee. Inclusion criteria for the EFFECT-Dep Trial included being over the age of 18 years, referral for ECT, meeting the diagnostic criteria for major depressive episode (unipolar or bipolar) using the Structured Clinical Interview for DSM-IV Axis I Disorders (SCID-CV), 26 and having a baseline score ≥21 on the HDRS, 24-item version (HDRS-24). 27 Exclusion criteria included being medically unfit for general anaesthesia or ECT, substance or alcohol misuse in the previous six months, ECT in the previous six months, a history of schizophrenia, schizoaffective disorder, or neurodegenerative or other neurological disorder, involuntary patient status and an inability or refusal to consent. Written informed consent was obtained from all patients prior to study participation.

| ECT protocol
Twice weekly, brief-pulse (1.0-msec pulse width, current amplitude 800 mA) ECT was delivered with handheld electrodes using a Mecta 5000 M device (Mecta Corp., Portland, Ore.; maximum 1200 mC). Methohexital (0.75-1.0 mg/kg) anaesthesia and succinylcholine (0.5-1.0 mg/kg) for muscle relaxation was used to administer ECT. 28 Patients underwent up to 12 ECT sessions, as determined by consultation between the patient and referring clinician. During the trial, usual antidepressant treatments were maintained for all patients.

| Clinical assessments
The following pre-ECT (baseline) demographic and clinical variables were obtained for each patient: age; gender; duration of depressive episode (in weeks); presence of a family history of mood disorder (yes/no; defined as patients' selfreport of any first-degree relative with a previous diagnosis of depression and/or bipolar disorder or death by suicide); treatment-resistant depression (yes/no; Antidepressant Treatment History Form) 29 ; HDRS-24 score; and levels of suicidality (HDRS-24 item 11), psychomotor retardation (HDRS-24 item 16) and agitation (HDRS-24 item 17). As previously suggested, a composite score using the HDRS-24 items 16 and 17 (ie psychomotor retardation and agitation) was calculated to capture psychomotor disturbances, as a core feature of melancholia. 7 Presence of bipolar depression, psychotic symptoms and melancholia were assessed as 'present or not present' using the SCID-CV. 26 To classify remission and response status, HDRS-24 scores were obtained weekly during the allocated ECT course (ie after every second ECT session). Response was defined as ≥60% decrease from the baseline HDRS-24 score and a score ≤16. Remission was defined as ≥60% decrease from the baseline HDRS-24 score and a score ≤10. A more conservative definition of sustained remission, used in the original report of the EFFECT-Dep Trial and that required scores to be maintained for two consecutive weeks, 25 was also used for analyses.

| Statistical analysis
IBM SPSS Statistics software version 25 was used to conduct statistical analyses (IBM Corporation). Normality distributions were first investigated for all variables using skewness coefficients, histograms and Q-Q plots. Non-parametric equivalent tests were conducted when distributions were potentially non-normal in sample sizes below 30. 30 Where appropriate, means with standard deviations (SD) or number (%) per ECT electrode placement group (ie RUL or BT ECT) were reported for baseline demographic and clinical characteristics. When comparing RUL and BT ECT group means, the sample size at baseline was large enough to detect a clinically relevant effect size of at least 0.50 with 0.80 power. 31 Independent sample t tests were conducted to compare the mean times to achieve response and remission between RUL and BT ECT groups. To examine speed of improvement of HDRS-24 scores, mean differences between RUL and BT ECT groups in weekly HDRS-24 scores were determined using independent sample t-tests and a Mann-Whitney U test. Cohen's d was used to measure the effect size of pair-wise comparisons. 31 The Benjamini-Hochberg (B-H) method was used where appropriate to control for the false discovery rate with multiple comparisons. 32 Pearson's chi-square and Fisher's exact tests were used to compare the proportion of patients who met criteria for response and remission at each treatment week.
To compare the speed of response and remission trajectories of RUL and BT ECT groups, Cox proportional hazard model analysis was conducted. 6 The separate outcome measures were as follows: (i) time to response and (ii) time to remission. Patients were classified as censored observations if they did not achieve the outcome of interest (ie response or remission) or if information on the outcome of interest was not available. Univariate Cox regression analyses were firstly conducted to determine whether the following variables were potential covariates to be included in multivariate analyses: gender, age, baseline HDRS-24 score, duration of current depressive episode, presence of a family history of mood disorder, treatmentresistant depression, bipolar depression, psychotic symptoms, melancholia and levels of suicidality, psychomotor retardation, agitation and psychomotor disturbances.
After investigating the presence of relevant covariates, univariate Cox regression analysis was conducted to investigate whether ECT electrode placement group predicted speed of response. To investigate whether ECT electrode placement group predicted speed of remission, multivariate Cox regression analysis was used. For all univariate and multivariate Cox regression analyses, hazard ratios (HR) and their 95% confidence intervals (CI) were reported. Given the exploratory nature of Cox regression analyses, adjustments for multiple comparisons were not conducted. 33 For all tests, statistical significance was defined as p < 0.05, with two-tailed p-values used.

| Baseline demographic and clinical characteristics
Sixty-nine patients were assigned to RUL ECT and 69 to BT ECT. The rate of drop out from randomisation to end of treatment was 2.9%. Table 1 summarises the baseline characteristics of patients within RUL and BT ECT groups. As reported previously, 25 the RUL and BT ECT groups were well balanced for demographic and clinical characteristics.

| Time to response and remission
Despite deviations from normality, sample sizes above 30 were sufficiently large to conduct independent sample t tests. 30

HDRS-24 scores
Based on the number of patients who continued to have ECT at each treatment week, Figure 1 illustrates the HDRS-24 scores for RUL and BT ECT groups at baseline and at each treatment week. Independent sample t tests, using B-H adjusted P-values, and a Mann-Whitney U test revealed no significant differences between BT and RUL ECT groups in HDRS-24 scores at any time point for patients who continued to have ECT at each treatment week (Table 2).
Using the more conservative definition of remission, the time taken for participants to achieve sustained remission range from two to seven weeks. This included a proportion of participants who achieved sustained remission based on HDRS-24 scores recorded one week after final ECT (ie one week after ECT course completion). Table S1 shows the weekly cumulative frequency and percentage of patients within RUL and BT ECT groups that achieved sustained remission with ECT at each weekly assessment. With the conservative definition of sustained remission, the proportion of remitters did not differ between patients who received RUL ECT and BT ECT after two (p = 0.50, Fisher's exact test), three (χ 2 = 0.47, p = 0.49), four (χ 2 = 0.04, p = 0.85), five (χ 2 = 0.03, p = 0.86), six (χ 2 = 0.27, p = 0.61) or seven weeks (χ 2 = 0.26, p = 0.61).

| Speed of response to ECT
Univariate Cox regression analyses revealed that the following continuous variables did not predict speed of response to ECT: age, baseline HDRS-24 score, duration of current depressive episode and levels of suicidality, psychomotor retardation, agitation or psychomotor disturbances (Table S2). When investigating categorical variables, speed of response to ECT was not predicted by gender, family history of mood disorder, treatment-resistant depression, bipolar depression, psychotic symptoms or melancholic symptoms (Table S2). When using a median split of 58 years, there were no differences in speed of response between younger (20-57 years) and older (58-88 years) patients (Table S2).
As the hazard ratios of variables shown in Table S2 indicated an absence of potential covariates, multivariate Cox regression analysis was not suitable to determine if RUL or BT ECT group predicted speed of response. Thus, univariate Cox regression was conducted to investigate if speed of F I G U R E 1 Mean HDRS-24 scores with standard deviations for RUL and BT ECT groups at baseline and treatment weeks. See Table 2

Week 4 n (%)
Week 5/Total n (%)  response was predicted by electrode placement group. As shown in Figure 2A, ECT electrode placement group did not predict speed of response to ECT among patients who met response criteria (HR = 0.79, 95% CI = 0.53-1.19, p = 0.25).

| Speed of remission with ECT
Using univariate Cox regression analyses, the following variables did not predict speed of remission with ECT: gender, age, age over 58, baseline HDRS-24 score, duration of current depressive episode, family history of mood disorder, treatment-resistant depression, bipolar depression, psychosis, melancholia and levels of psychomotor retardation, agitation or psychomotor disturbances (Table S2). Greater levels of suicidality predicted a slower speed of remission with ECT (HR = 0.81, 95% CI = 0.67-0.99, p = 0.04). Multivariate Cox regression analysis was conducted to determine if speed of remission, among those who met remission criteria, was predicted by BT or RUL ECT group, while controlling for the confounding effect of suicidality ( Figure 2B). When controlling for levels of suicidality, ECT group did not predict speed of remission (HR = 0.80, 95% CI = 0.51-1.23, p = 0.30).

| Effects of electrode placement on speed of ECT
When comparing speed of ECT effectiveness between twiceweekly brief-pulse high-dose RUL ECT and moderate-dose BT ECT, the proportion of responders and remitters did not significantly differ between groups at any time point. There were also no differences between RUL and BT ECT in the speeds of improvement, response or remission with ECT. This contrasted with our study hypothesis and was inconsistent with earlier findings that depression improved faster with BT ECT. 12 Conflicting findings may be explained by methodological differences between trials. While the EFFECT-Dep trial recorded weekly HDRS-24 scores for twice-weekly ECT, the CORE study calculated time to response based on HDRS-24 assessments at every ECT visit (ie thrice-weekly). 12 Thus, measuring changes in depressive symptoms at a narrower time interval may be more sensitive to detecting speed of improvement with ECT. Nonetheless, weekly depression assessments indicated that speed of improvement was faster with BT ECT in the CORE trial. 12 It is also possible that thrice-weekly, but not twice-weekly, BL ECT is associated with some initial earlier improvements in depression when compared to RUL ECT. 12 The advantages of the current study included that we conducted a pragmatic trial examining ECT in real-world practice, with continuation of patients' psychotropic medications. Additionally, we had a relatively high retention rate that reduced attrition bias (97%, compared with 74% in the CORE study). 12 There was no difference between our study and the CORE study in ECT dosage strategies. 12 The mean number of ECT session was also consistent with the number of ECT session reported in previous trials that compared BT and RUL ECT. 2 Another strength of the present study is the statistical methods employed. We utilised survival analysis, which is sensitive to detecting group differences in the speed of ECT antidepressant activity. 6 Survival analysis in our study further indicated that there was no effect of electrode placement on the likelihood of achieving earlier response or remission. Given the limited research available, subsequent studies should employ survival analysis to investigate if BT or RUL ECT predicts faster ECT effectiveness.

| Exploratory analyses
In addition to ECT electrode placement, exploratory analyses revealed for the first time that speed of response and remission to ECT were not predicted by family history of mood disorder in a first-degree relative, treatment-resistant depression, presence of melancholia or levels of psychomotor retardation, agitation and psychomotor disturbances. Patients' gender, duration of current depressive episode and presence of psychotic symptoms also did not predict speed of ECT response, which was in line with prior research. 11 We found that age was not associated with speed of ECT response. This contrasted previous findings, in which one study reported faster speed of ECT response among older patients 11 and another study reported a faster ECT response among younger patients. 16 There are a number of possible explanations for the inconsistencies reported on the association between age and speed of ECT response, including differences in study design. For example, the previously reported effect of older age on faster speed of ECT response was based on pooled data from multiple studies, in which patients were not all randomised or blinded to ECT trial arms. 11 It was reported that younger age predicted faster response to ECT among a smaller sample (N = 48), by comparing the age of early and late responders based on an arbitrarily determined number of ECT sessions. 16 From our study, which directly investigated the probability of faster ECT response with age, it cannot be concluded that age predicts speed of response or remission to ECT.
Although bipolar depression has been previously associated with speed of ECT response, 8,17 we did not find any effect of depression polarity. The ability to compare studies that investigated the effect of depression polarity on speed of ECT response is limited by differences in patient characteristics and study design. 18 For example, previous studies included a relatively smaller sample of patients with bipolar disorder 17 and data that were pooled from multiple studies. 8 The lack of association between speed of response to ECT and baseline depression severity was also inconsistent with earlier findings. 15 We did not replicate the findings of an association between psychotic symptoms and speed of remission to ECT. 24 While suicidality severity can be rapidly reduced with ECT, 19 levels of suicidality did not predict speed of ECT response. Our exploratory analyses revealed that higher levels of suicidality may predict slower speed of remission and sustained remission with ECT. Although intriguing, these preliminary findings would not survive correction for multiple comparisons. Furthermore, the present study did not include patients who were actively suicidal or who would have had high levels of suicidality. Overall, there is a dearth of research on speed of remission to ECT. The exploratory analyses in the present study are the first to indicate that numerous demographic and clinical characteristics do not predict speed of remission to ECT. Future research is warranted, given that remission status is considered the gold standard for determining depression treatment effectiveness. [21][22][23] Compared with previous studies, 3,8,11,12,[15][16][17] we used a more conservative definition of ECT response. Specifically, this study used a definition of response that required a relatively greater decrease in baseline HDRS-24 scores and a lower threshold score. Across the literature, ECT response and remission have been heterogeneously defined. 7 These discrepancies in definitions may explain inconsistent findings on the predictors of speed of response and remission to ECT. There is a lack of consensus on how to optimally define response and remission to ECT. To enable cross-study comparisons, universal definitions for response and remission to ECT should be established.

| Limitations
Our study has a number of limitations. Firstly, findings were based on secondary analyses of previous trial data. 25 Thus, data were not collected with the primary objective of investigating speed of improvement, response and remission with ECT. Although the EFFECT-Dep trial included a relatively large sample size at baseline (N = 138), there was a reduction in sample size with weekly HDRS-24 assessments. However, the trial dropout rate was very low. The reduction in sample size over time reflects real-world practice in which patients receive a variable number of ECT sessions that is determined by clinicians in consultation with patients. Furthermore, this study was sufficiently powered at all time points to detect the previously reported effect of ECT electrode placement on speed of improvement with RUL or BT ECT. 12 Additional limitations of exploratory analyses were that the sample size was reduced for group comparisons and the significance levels of multiple comparisons were not corrected for. Failure to correct for multiple comparisons was justified due to the exploratory nature of analyses. 33 Future research should conduct hypothesis-driven analyses that are adequately powered and adjusted a priori to account for multiple testing.
To conclude, this study adds to the currently very limited body of research on speed of improvement, response and remission with ECT. Speed of response and remission to ECT were not predicted by numerous clinical and demographic characteristics, apart possibly from greater levels of suicidality. However, the limitations of exploratory analyses should be recognised when interpreting results. Replication of findings with a larger sample size is required.
This research has the potential to inform clinical decisionmaking. Given the findings, it appeared that speed of improvement, response and remission may not be relevant factors when deciding between BT and RUL electrode placement for ECT. As cognitive outcomes are more favourable with RUL ECT, 2 our findings further support consideration of RUL ECT when determining whether to prescribe briefpulse BT or RUL ECT for treating depression.