If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Non-Invasive Neurostimulation Therapies Laboratory, University of British Columbia, Vancouver, BC, CanadaDepartment of Psychiatry, University of British Columbia, Vancouver, BC, Canada
Non-Invasive Neurostimulation Therapies Laboratory, University of British Columbia, Vancouver, BC, CanadaCentre for Health Evaluation and Outcomes Sciences, St Paul's Hospital, Vancouver, BC, Canada
Non-Invasive Neurostimulation Therapies Laboratory, University of British Columbia, Vancouver, BC, CanadaDepartment of Psychiatry, University of British Columbia, Vancouver, BC, Canada
Department of Psychiatry, University of British Columbia, Vancouver, BC, CanadaBritish Columbia Psychosis Program, Vancouver, BC, CanadaBritish Columbia Mental Health and Substance Use Services Research Institute, Vancouver, BC, Canada
Non-Invasive Neurostimulation Therapies Laboratory, University of British Columbia, Vancouver, BC, CanadaDepartment of Psychiatry, University of British Columbia, Vancouver, BC, Canada
For treatment-refractory psychosis (TRP), ECT is particularly effective in reducing positive symptoms in patients who are also taking clozapine.
•
ECT can improve negative symptoms in TRP patients with a comorbid depressive episode.
•
Bifrontal and bitemporal ECT are both effective in TRP, and bifrontal may be more effective for positive symptoms in patients taking clozapine.
•
ECT seizure duration is increased by concomitant clozapine and decreased by concomitant anticonvulsant medication.
•
Anticonvulsants did not adversely affect ECT clinical outcomes in our TRP sample.
Abstract
Background
Treatment-refractory psychosis (TRP) is a significant clinical challenge. While clozapine is frequently effective, alternate or augmentation strategies are often necessary. Evidence supports effectiveness of electroconvulsive therapy (ECT), but questions remain about optimal treatment parameters and impacts of concomitant pharmacotherapy.
Objective
/Hypothesis: To analyze the impact of clozapine, anticonvulsant medication, mood state, and ECT electrode placement on outcomes in TRP. We hypothesized that ECT would lead to greater reduction in positive symptoms, particularly in patients receiving clozapine.
Methods
Retrospective study in a tertiary TRP program. The Positive and Negative Syndrome Scale (PANSS) was used for clinical outcomes, with positive subscore as primary outcome. Clinical and ECT data were analyzed using a linear modelling approach, controlling for relevant covariates.
Results
A total of 309 patients were included. ECT plus clozapine associated with greater improvement in positive, general, and total symptoms than ECT alone. ECT associated with greater improvement in negative symptoms in depressed patients. Bifrontal placement was mostly equivalent to bitemporal, with greater reduction of positive symptoms in patients receiving clozapine, and associated with lower electrical dose in patients on anticonvulsants. Clozapine increased seizure duration, while anticonvulsants decreased it. Anticonvulsant use in ECT patients associated with equivalent to slightly improved symptom reduction.
Conclusions
ECT's benefit in TRP may be greatest in patients receiving clozapine. ECT can improve negative symptoms in depressed TRP patients. Bifrontal placement is effective in TRP. Clozapine and anticonvulsants have opposite effects on seizure duration, but anticonvulsants may not adversely affect clinical outcomes of ECT for TRP.
Abbreviations: TRP – Treatment-refractory psychosis; TRS – Treatment-refractory schizophrenia; NNT – Number Needed to Treat; PANSS – Positive and Negative Syndrome Scale.
remains a significant clinical challenge. Clozapine is the gold-standard treatment for patients who do not improve with two or more adequate antipsychotic trials [
Treatment-resistant schizophrenia: treatment response and resistance in psychosis (TRRIP) working group consensus guidelines on diagnosis and terminology.
], while various augmentation strategies have also been studied, including anticonvulsants/mood stabilizers, amisulpride, sulpiride, and aripiprazole for positive symptoms [
Double-blind, placebo-controlled, multicenter trial of selegiline augmentation of antipsychotic medication to treat negative symptoms in outpatients with schizophrenia.
], so it is important to investigate augmentation strategies, both alone and in combination with clozapine, that address residual positive symptoms in TRP.
Electroconvulsive therapy (ECT) has evidence of benefit in TRP, particularly for positive symptoms [
Combined use of electroconvulsive therapy and antipsychotics (both clozapine and non-clozapine) in treatment resistant schizophrenia: a comparative meta-analysis.
]. Some evidence also suggests a possible effect on negative and general symptoms, as defined by the Positive and Negative Syndrome Scale (PANSS), though this finding is not consistent across studies [
Comparative study of effectiveness of augmentation with ECT in clozapine resistant schizophrenia (CRS) and non-clozapine resistant schizophrenia (Non-CRS).
Nonetheless, significant gaps remain in the evidence base for clinical outcomes of ECT in TRP. Accordingly, guideline-based recommendations for this treatment are either cautious or absent [
]. There is limited understanding of interactions between clozapine and other medications with ECT in TRP, an area specifically cited as needing further research by the recent APA Schizophrenia Practice Guidelines [
]. A more recent Cochrane review also cited need for more study of different electrode placements in TRP(16). To our knowledge, no studies have examined bifrontal placement in TRP, despite early RCT evidence demonstrating superiority over bitemporal for both symptom improvement and cognitive outcomes in psychosis more generally [
Double-blind randomized controlled study showing symptomatic and cognitive superiority of bifrontal over bitemporal electrode placement during electroconvulsive therapy for schizophrenia.
]. There is also limited evidence concerning the effects of clozapine, anticonvulsants, and electrode placement on seizure response. Seizure threshold is hypothesized to decrease with clozapine [
We investigated the effects of and interactions between ECT, medications, and mood state on clinical outcomes in a naturalistic sample of patients with highly refractory psychotic disorders. We hypothesized that 1) Patients treated with ECT would demonstrate greater reductions in positive symptoms; and 2) Patients treated with the combination of ECT and clozapine would demonstrate greater symptom reductions than either used alone. We then conducted exploratory analyses of the impact of electrode placement and concomitant medications (including clozapine) on metrics of seizure expression, and of electrode placement and anticonvulsant use on clinical outcomes.
2. Materials and methods
2.1 Study setting and population
This was a retrospective study using clinical data from the British Columbia Psychosis Program (BCPP) in Vancouver BC, Canada. Ethics approval was obtained from the University of British Columbia Clinical Research Ethics Board. The BCPP is a provincial tertiary inpatient program treating the most severe cases of TRP in the province. The study population were inpatients with severe and treatment-refractory psychosis who were referred to the Program by psychiatrists working at secondary inpatient centers or community mental health teams and admitted to the program between 2012 and 2020. Most patients admitted to the Program have a diagnosis of schizophrenia or schizoaffective disorder. A majority of patients have a history of a previous or ongoing clozapine trial, either in the community or during a prior admission.
The BCPP utilizes a comprehensive treatment approach for individuals with refractory psychotic disorders. This includes a formal diagnostic review according to DSM criteria; clozapine challenges and rechallenges as appropriate; trials of alternate antipsychotic agents and augmentation agents, including anticonvulsant medications; evidence-based nonpharmacological strategies including CBT for psychosis and social skills training; and trials of electroconvulsive therapy. Patients being treated with clozapine at the time of admission assessment are continued on this treatment unless significant tolerability or safety issues arise during admission (e.g., myocarditis, agranulocytosis). Electroconvulsive therapy is used as part of the BCPP clinical protocol on the basis of existing primary data supporting use in TRP(16). It is offered to patients who either are experiencing significant residual positive symptoms in spite of a clozapine trial of adequate dose and duration as per the Canadian Schizophrenia Guidelines [
] or who have failed to tolerate an adequate clozapine trial. Patients undergoing ECT continue to receive psychosocial interventions.
2.2 Eligibility criteria
Patients were eligible if they 1) had a psychiatrist-confirmed DSM diagnosis of a primary psychotic disorder (i.e., schizophrenia or schizoaffective disorder) at admission; 2) had complete PANSS data at admission and discharge; and 3) completed their admission during the study period (2014–2020). Patients with a non-primary psychotic disorder (e.g. bipolar disorder) were excluded from analysis.
2.3 ECT procedure
Patients who were offered and consented to a course of ECT were enrolled with the UBC Hospital inpatient ECT service. Where necessary, substitute consent was sought from a patient's substitute decision-maker. Index treatments were delivered biweekly under the supervision of a psychiatrist, anesthesiologist, and anesthetic nurses. Based on evidence supporting bifrontal over bitemporal electrode placement for psychosis [
Double-blind randomized controlled study showing symptomatic and cognitive superiority of bifrontal over bitemporal electrode placement during electroconvulsive therapy for schizophrenia.
] and on improved cognitive outcomes in general, initial bifrontal placement was used in the clinical protocol for these patients. We used the placement method outlined by Abrams & Taylor [
]. Patients who failed to improve were switched to bitemporal. Length of treatment course was clinically determined based on tolerability and evidence of improvement. Anticonvulsants and lithium were held the night before treatment. General anesthetic was achieved with intravenous methohexital (1.0–1.5 mg/kg) and muscle relaxation with succinylcholine (0.5 mg/kg). ECT stimulus was delivered with a Thymatron IV device using brief pulse width (0.5 ms). A titration approach was used at the first ECT session to determine seizure threshold, with the first stimulus delivered at 5–10% of maximum with restimulation at increases of 5–10% as required; stimulus intensity for the second session was defined as 1.5–2.5 times seizure threshold for bifrontal and bilateral placements, and 4–6 times threshold for right unilateral. Further dose increases were at the discretion of the treating psychiatrist based on seizure adequacy (duration of at least 15 s and subjective seizure quality) and clinical progress. Antihypertensive, antiemetic, and analgesic medications were administered at the discretion of the attending anesthesiologist.
2.4 Data collection
Clinical and demographic data were recorded for all BCPP patients as part of routine clinical care. Admission data were captured following an initial assessment period, during which any medications causing significant tolerability or safety issues were stopped. Psychosis symptoms were measured at admission and discharge by the PANSS (see Supplementary Fig. 1). Medication dosages, including total antipsychotic medication load as measured by the defined daily dose method [
], were recorded. ECT data were gathered by treatment staff as part of routine ECT care and patients provided consent to the use of these data in future analysis. Data included concomitant psychiatric medications, anesthetic medications, electrode placement, stimulus parameters, EEG seizure duration, and electrical dose used.
2.5 Outcomes
The primary outcome of interest was change in the positive symptom subscale of the PANSS from admission to discharge. Change in the total PANSS score and in negative- and general-symptom subscale scores from admission to discharge, were secondary outcomes. For the group of patients who received ECT, exploratory outcomes included within- and between-patient differences in seizure duration and electrical dose required to elicit a seizure. For patients who experienced a switch of electrode placement during their ECT course, they were grouped by the placement used for the majority of their treatments.
], an R statistical package. For analyses of PANSS score effects, ANCOVA models were constructed to estimate the relative contribution of hypothesized variables, including an adequate (≥6 sessions) course of ECT, treatment with clozapine, concomitant treatment with an anticonvulsant, and presence or absence of a manic or depressive state. Subgroup analysis of ECT patients was conducted to evaluate the contribution of electrode placement to PANSS outcomes. During model selection, potential covariates were identified based on plausible clinical significance; for each outcome model, candidate covariates were screened for collinearity and tendency towards overfitting and excluded if these effects were observed. Age and duration of illness were screened for all models based on prior plausibility in the ECT literature [
]. Effect size estimates for pairwise comparisons were generated using the marginal means approach. To account for multiple comparisons, the Tukey correction was applied to all model outputs.
For analyses of ECT seizure response, a mixed-effects model for repeated measures was used, with individual patients acting as clusters. Presence of anticonvulsant was treated as a categorical variable, while clozapine dose, benzodiazepine dose (lorazepam equivalents), ECT stimulus electrical dose, and treatment number were treated as continuous variables. Covariate selection was carried out as above. Covariates with significant (p < 0.05) variance were treated as random effects; all others were treated as fixed effects.
3. Results
3.1 Clinical and demographic characteristics
Out of 475 patients who were admitted to the program during the study period, 309 met inclusion criteria for analysis. Patients were excluded for: lack of complete PANSS data at admission and discharge (n = 116) or no primary psychotic disorder diagnosis (n = 50). Most patients had a diagnosis of schizophrenia. Approximately half were taking clozapine at admission, and 27.2% were taking an anticonvulsant. The rate of anticonvulsant use was similar regardless of diagnosis (schizophrenia 26.2%, schizoaffective disorder 30%; χ2 = 0.43, p = 0.51). Patients who received ECT tended to be older, have a longer duration of illness, more symptom severity, and greater treatment resistance. Of patients receiving ECT, 54% were treated concurrently with clozapine (see Table 1). Clozapine discontinuations following admission assessment were rare (overall incidence 2.8%), with 3 occurring in the ECT group and 5 in the non-ECT group between admission and discharge.
Table 1Clinical and demographic characteristics of the study population.
All patients (N = 309)
Received ECT course (n = 50)
No ECT course (n = 259)
Age, years, mean (SD)
38.3 (13.4)
43.9 (13.8)
37.2 (13.1)
Sex, female, n (%)
76 (24.4)
12 (24.0)
64 (24.7)
Diagnosis, schizophrenia, n (%)
229 (74.1)
34 (68.0)
195 (75.3)
Diagnosis, schizoaffective disorder, n (%)
80 (25.9)
16 (32.0)
64 (24.7)
Duration of illness, mean (SD)
15.5 (12.0)
20.4 (13.0)
14.5 (11.7)
History of clozapine use, n (%)
268 (86.7)
48 (96.0)
220 (85.0)
Clozapine use at admission, n (%)
156 (50.5)
27 (54.0)
129 (49.8)
Clozapine daily dosage, mg, mean (SD)
369.8 (163.3)
358.2 (141.1)
372.0 (168.4)
Antipsychotics at admission, defined daily dosesa, mean (SD)
2.3 (1.6)
2.2 (1.2)
2.3 (1.7)
Anticonvulsant at admission, n (%)
84 (27.2)
16 (32.0)
68 (26.3)
Antidepressant at admission, n (%)
98 (31.7)
17 (34.0)
81 (31.3)
Lithium at admission, n (%)
40 (13.9)
7 (14.0)
33 (12.7)
Mood state, euthymic, n (%)
213 (68.9)
25 (50.0)
188 (72.6)
Mood state, depressed, n (%)
42 (13.6)
14 (28.0)
28 (10.8)
Mood state, manic, n (%)
54 (17.5)
11 (22.0)
43 (16.6)
PANSS total score at admission, mean (SD)
95.2 (19.2)
101.4 (18.5)
94.0 (19.2)
PANSS-positive subscore at admission, mean (SD)
24.8 (7.1)
26.4 (6.2)
24.5 (7.2)
CGI-severity at admission, mean (SD)
5.8 (0.8)
6.2 (0.7)
5.8 (0.9)
Degree of resistance at admissionb, mean (SD)
5.3 (0.8)
5.6 (0.6)
5.2 (0.8)
a – Method of standardizing antipsychotic medication dosage, described by Lin and colleagues [
Sixty-one patients (19.7%) received at least one ECT treatment during admission, and 50 (16.2%) received at least six treatments. A further 88 patients (29.5%) were offered ECT during admission but declined. Bifrontal placement was most frequently used, and a majority of ECT patients received exclusively bifrontal treatments (see Table 2).
A two-way ANCOVA model was selected to evaluate the effects of an adequate course of ECT and clozapine treatment on PANSS-positive subscale change during admission. Anticonvulsant treatment, best Global Assessment of Functioning score in the past year, total antipsychotic equivalent dose at discharge, and degree of antipsychotic resistance at admission as measured by the Degree of Resistance Scale proposed by May et al. [
The overall model is shown in Table 3. Two model terms were statistically significant after adjustment for multiple comparisons: ECT x clozapine (F = 6.37, p = 0.01) and anticonvulsant (F = 4.61, p = 0.03). A post-hoc test for unadjusted main effect of ECT x clozapine on positive symptoms using a 2-way ANOVA was significant (p = 0.04), consistent with the full ANCOVA model, while a test for unadjusted effect of anticonvulsant was not significant (p = 0.30), suggesting an impact of covariate adjustment.
Table 3Summary of ANCOVA model for PANSS-positive subscore change.
Term
Sum of Squares
df
Mean Square
F
p
Overall model∗
1056.7
10
105.7
1.97
0.036∗
ECT
106.7
1
106.7
2.40
0.122
Clozapine
13.8
1
13.8
0.31
0.578
Total AED at discharge
145.9
1
145.9
3.28
0.071
DOR at admission
55.5
1
55.5
1.25
0.265
Best GAF in past year
151.3
1
151.3
3.40
0.066
Anticonvulsant∗
204.6
1
204.6
4.61
0.033∗
ECT x Clozapine∗
283.1
1
283.1
6.37
0.012∗
ECT x Anticonvulsant
79.4
1
79.4
1.79
0.182
Clozapine x Anticonvulsant
4.0
1
4.0
0.09
0.764
ECT x Clozapine x Anticonvulsant
12.44
1
12.44
0.28
0.597
Residuals
12708.59
286
44.44
This table summarizes the output of the analysis of covariance (ANCOVA) model for predicting reduction in positive symptoms of schizophrenia, as measured by the Positive Symptoms section of the PANSS (Positive and Negative Syndrome Scale).
ECT – electroconvulsive therapy.
AED – antipsychotic equivalent dose.
DOR – degree of resistance, as measured by the Degree of Resistance scale.
GAF – Global Assessment of Functioning.
df – degrees of freedom.
∗ – significant with a p-value below 0.05, after correction for multiple comparisons.
Effect size estimates were calculated. ECT plus clozapine was associated with significantly greater improvement versus clozapine without ECT (Mean difference = 4.75, Cohen's d = 0.71, 95% CI = 0.24–1.18) (see Fig. 1). The combination of ECT plus anticonvulsant was associated with greater improvement versus patients who received neither (Mean difference = 4.32, Cohen's d = 0.65, 95% CI = 0.11–1.19).
Fig. 1Comparison of PANSS-positive change by ECT and clozapine status.
This figure illustrates the reduction in positive symptoms of schizophrenia, as measured by the Positive Symptoms section of the Positive and Negative Syndrome Scale (PANSS), for patients who did or did not receive electroconvulsive therapy (ECT) or clozapine during inpatient treatment. All data are represented as marginal means calculated from the ANCOVA model.Error bars represent standard error values calculated using the marginal means approach.∗ – significant difference, with a p-value below 0.05 after correction for multiple comparisons.
We also conducted an exploratory analysis of the three-way interaction between ECT, clozapine, anticonvulsants. ECT in patients receiving both clozapine and an anticonvulsant was associated with greater improvement versus ECT without either medication (Mean difference = 7.64, Cohen's d = 1.15, 95% CI = 0.25–2.05), the medication combination without ECT (Mean difference = 6.92, Cohen's d = 1.04, 95% CI = 0.25–1.83), clozapine without ECT or anticonvulsant (Mean difference = 6.90, Cohen's d = 1.03, 95% CI = 0.29–1.78) or treatment with none of the three (Mean difference = 5.58, Cohen's d = 0.84, 95% CI = 0.09–1.58).
Subgroup analysis of ECT patients by electrode placement demonstrated that patients receiving ECT via exclusively or primarily bifrontal experienced similar PANSS-positive score improvement to those receiving bitemporal (Cohen's d = −0.55, 95% CI = −1.36 – 0.25). The addition of clozapine to ECT was associated with improved outcomes in patients receiving primarily bifrontal (Mean difference = 7.73, Cohen's d = 1.03, 95% CI = 0.02–2.04) but not bitemporal (Mean difference = −0.05, Cohen's d = −0.06, 95% CI = −1.39 – 1.36) placement.
A two-way ANCOVA model was selected to evaluate the effects of ECT on PANSS-negative subscale score change. During model evaluation and selection, clozapine and ECT x clozapine did not contribute significantly to the overall model. However, mood state (depressive versus manic versus euthymic) did contribute significantly and was thus selected as the second factor variable (with mood state treated as a nominal, non-hierarchical factor) together with ECT. The only model term to remain significant after covariate adjustment was ECT x mood state (F = 3.62, p = 0.03) (see Supplementary Table 1). A post-hoc test for unadjusted main effect of ECT x mood state on negative symptoms using a 2-way ANOVA was non-significant (p = 0.06), suggesting a modest impact of covariate adjustment.
As above, effect size estimates were calculated. ECT in depressed patients was associated with larger score reduction versus ECT in euthymic patients (Mean difference = 4.74, Cohen's d = 0.78, 95% CI = 0.11–1.45) or ECT in manic patients (Mean difference = 6.13, Cohen's d = 1.01, 95% CI = 0.17–1.86) (see Fig. 2). Furthermore, among depressed patients, ECT was associated with a larger score reduction versus no ECT (Mean difference = 5.44, Cohen's d = 0.90, 95% CI = 0.24–1.56). ECT in depressed patients was also associated with a larger score reduction compared with both manic patients (Mean difference = 4.62, Cohen's d = 0.76, 95% CI = 0.15–1.38) and euthymic patients (Mean difference = 4.30, Cohen's d = 0.71, 95% CI = 0.24–1.56) who did not receive ECT. Among all patients who did not receive ECT, depressed patients did not show a greater reduction in negative symptoms versus euthymic or manic patients.
This figure illustrates the reduction in negative symptoms of schizophrenia, as measured by the Negative Symptoms section of the Positive and Negative Syndrome Scale (PANSS), for patients who did or did not receive electroconvulsive therapy (ECT) during inpatient treatment, grouped by mood state at the time of treatment.All data are represented as marginal means calculated from the ANCOVA model.Error bars represent standard error values calculated using the marginal means approach.∗ – significant difference, with a p-value below 0.05 after correction for multiple comparisons.
Subgroup analysis of electrode placement within ECT patients revealed no difference between patients treated with primarily bifrontal versus bitemporal overall (Cohen's d = −0.58, 95% CI = −1.32 – 0.17) or within euthymic or manic patients. However, in depressed patients, bitemporal placement was associated with greater improvement in negative symptoms (Mean difference = 15.98, Cohen's d = 2.16, 95% CI = 0.81–3.51).
In order to investigate whether diagnosis (schizoaffective disorder versus schizophrenia) had a moderating effect on our finding that depressed patients experienced greater improvement with ECT, we conducted further sensitivity analyses, treating diagnosis first as an individual factor covariate in the overall model and next as a combined term (diagnosis x ECT) replacing mood state. In both analyses, diagnosis did not contribute significantly to the overall model.
A two-way ANCOVA model was selected to evaluate the effects of ECT on PANSS-general subscale score change, with clozapine as the second factor variable. ECT and clozapine did not reach significance as individual contributors to the model, but the 2-way term ECT x clozapine was significant (see Supplementary Table 2). A post-hoc test for unadjusted main effect of ECT x clozapine on general symptoms using a 2-way ANOVA was significant (p = 0.049), consistent with the full ANCOVA model.
ECT plus clozapine was associated with a significant effect on PANSS-general reduction versus ECT without clozapine (Mean difference = 5.71, Cohen's d = 0.63, 95% CI = 0.05–1.22), clozapine without ECT (Mean difference = 5.74, Cohen's d = 0.64, 95% CI = 0.21–1.06), or neither treatment (Mean difference = 4.85, Cohen's d = 0.54, 95% CI = 0.10–0.97).
Subgroup analysis of electrode placement within ECT patients demonstrated that primary or exclusive use of bifrontal was similar to bitemporal for PANSS-general score improvement (Cohen's d = −0.55, 95% CI = −1.36 – 0.25).
3.5 Secondary outcomes: PANSS-total score change
A two-way ANCOVA model was selected to evaluate the effects of ECT on PANSS-total subscale score change, with ECT and clozapine as the factor variables. As with PANSS-general, only ECT x clozapine reached significance as an individual contributor to this model (see Supplementary Table 3). A post-hoc test for unadjusted main effect of ECT x clozapine on positive symptoms using a 2-way ANOVA was significant for the ECT term (p = 0.03), but not ECT x clozapine, suggesting a partial impact of covariate adjustment.
Estimates of effect size on PANSS-total for this term showed a moderate effect of ECT plus clozapine versus clozapine without ECT on the total PANSS score (Mean difference = 10.77, Cohen's d = 0.59, 95% CI = 0.12–1.06), with other comparisons not reaching significance.
Subgroup analysis of electrode placement within ECT patients revealed no significant difference between patients treated with primarily bifrontal versus bitemporal either overall or stratified by clozapine use.
3.6 Secondary outcomes: ECT seizure duration
In the linear mixed model for EEG seizure duration, ECT electrode placement and anticonvulsant use were included as factor variables, and clozapine dose, lorazepam equivalent dose, electrical dose, and treatment number were included as continuous covariates. Of these, only lorazepam equivalent dose showed significant variance and was thus treated as a random-effect variable.
Clozapine dose (F = 5.51, p = 0.02), electrical dose (F = 10.67, p < 0.01), and anticonvulsant use (F = 6.09, p = 0.01) were significant contributors to the model, while lorazepam equivalents, treatment number, and electrode placement were non-significant (see Supplementary Table 4).
Parameter effect estimates for the terms reaching significance were as follows: for clozapine, an increase in seizure duration of 0.03 s/mg (95% CI = 0.004–0.05); for electrical dose, a decrease of 0.034 s/millicoulomb (mC) (95% CI = −0.014 to −0.05); and for anticonvulsant use, a decrease in seizure duration of 10.55 s versus no anticonvulsant (95% CI = 2.17–18.93).
3.7 Secondary outcomes: ECT electrical dose
In the model for electrical dose, ECT electrode placement and anticonvulsant use were included as factor variables, and clozapine dose, lorazepam equivalents, and treatment number were included as continuous covariates. Of these, only lorazepam equivalent dose showed significant variance and was thus treated as a random-effect variable.
Clozapine dose (F = 25.83, p < 0.01), treatment number (F = 181.28, p < 0.01), anticonvulsant use (F = 23.84, p < 0.01), and electrode placement (F = 12.06, p < 0.01) were significant contributors to the model, while lorazepam equivalent dose was non-significant. The 2-way term anticonvulsant x electrode placement was also significant (F = 6.52, p < 0.02) (see Supplementary Table 5).
Parameter effect estimates for the terms reaching significance were as follows: for clozapine, an increase in electrical dose of 0.2 mC/mg (95% CI = 0.12–0.27); for treatment number, an increase of 5.46 mC/treatment (95% CI = 4.66–6.25); for anticonvulsant use, an increase of 111.25 mC versus no anticonvulsant (95% CI = 66.59–155.91); and for electrode placement, an increase of 42.74 mC for bitemporal versus bifrontal (95% CI = 22.22–63.26) and an increase of 79.44 mC for right unilateral versus bifrontal (95% CI = 27.55–131.33).
Finally, post-hoc comparisons were conducted to estimate the effect size of electrode placement on electrical dose with and without concomitant anticonvulsant use present. Bifrontal placement was associated with a decrease in mean dose by 78.21 mC versus bitemporal placement when an anticonvulsant was present (SE = 17.4, p < 0.01); however, no significant difference was observed between these placements when an anticonvulsant was not present. The remaining post-hoc comparisons between placements were non-significant for effect on electrical dose.
4. Discussion
4.1 Significance
Our pragmatic retrospective study supports the clinical effectiveness of ECT for TRP. One of the main strengths of our study was its relatively large sample size compared to other retrospective studies [
Comparative study of effectiveness of augmentation with ECT in clozapine resistant schizophrenia (CRS) and non-clozapine resistant schizophrenia (Non-CRS).
], which allowed sufficient power to examine subgroups and conduct post-hoc comparisons to estimate the impact of different medications and ECT electrode placements on clinical outcomes. We have identified patients who may experience a benefit from ECT on negative and general symptoms, for which evidence is currently limited. This is also the first study to our knowledge to examine the individual and interactional effects of clozapine, other medications, and electrode placement on metrics of seizure expression and seizure threshold in TRP.
There is limited evidence regarding optimal ECT electrode placement for TRP. Bifrontal may be preferable in non-treatment-resistant psychosis with a larger degree of symptom improvement and fewer cognitive adverse effects [
Double-blind randomized controlled study showing symptomatic and cognitive superiority of bifrontal over bitemporal electrode placement during electroconvulsive therapy for schizophrenia.
], but this effect has not been investigated in TRP patients to our knowledge. Our study provides the first evidence of clinical outcomes for bifrontal versus bitemporal ECT in a large clinical sample of patients with refractory psychotic illness.
The combination of these analyses allows for a balanced consideration of the risks and benefits of ECT treatment for refractory psychosis, considering both clinical and procedural factors. This is important to contextualize the use of ECT in a population that is characterized by high rates of treatment resistance, comorbidity, and polypharmacy.
4.2 Effects of ECT on positive symptoms
For our primary outcome, we observed a significant impact of the combination of ECT and concomitant clozapine on the PANSS-positive subscore. Effect size estimates suggest that the addition of ECT to clozapine therapy was superior to the use of clozapine alone. Interestingly, in our population ECT in the absence of clozapine was not associated with additional benefit after adjustment for confounders and multiple comparisons, which may suggest an additive effect of the ECT-clozapine combination in highly refractory patients. Our data thus provide support for hypothesis #2 above, that this treatment combination would be superior to either used alone. This has been posited elsewhere [
]. Our study provides support for this hypothesis in a pragmatic clinical setting consisting of patients with highly refractory psychosis, and provides a rationale for further testing of this hypothesis in a prospective research design.
4.3 Effects of ECT on negative, general, and overall symptoms
With respect to negative symptoms (PANSS-negative subscore), we saw no overall effect of ECT on negative symptoms. However, patients with a depressive mood state at admission did show greater improvement in negative symptoms with ECT irrespective of whether they had a diagnosis of schizophrenia or schizoaffective disorder. This mixed finding provides insight into an area of inconsistency in prior reports of ECT in psychotic disorders, with some studies finding no effect of ECT on negative symptoms [
Comparative study of effectiveness of augmentation with ECT in clozapine resistant schizophrenia (CRS) and non-clozapine resistant schizophrenia (Non-CRS).
]. These studies generally did not separate analyses by mood state or disorder. There are several possible interpretations for our findings. First, there was the possibility that diagnosis might be a confounding factor influencing this finding; for instance, that ECT patients with schizoaffective disorder experienced greater negative symptom reduction than those with schizophrenia because depression was more common in this group. However, sensitivity analysis suggested that diagnosis did not exert a significant impact on the degree of negative symptom change following ECT in depressed patients; an intriguing corollary hypothesis is whether the negative symptom scale is capturing a transdiagnostic mood dimension that responds to ECT. This hypothesis converges with a recent meta-analysis on the phenomenological overlap between negative and mood symptoms in schizophrenia as measured by the PANSS [
]. It is also possible that a depressive mood state can amplify negative symptoms in a reversible manner which is responsive to ECT.
Our analyses of general symptoms (PANSS-general subscore) and total symptoms (PANSS-total score) demonstrated similar findings to that for positive symptoms, namely that the effect of ECT and clozapine in combination was larger than either alone. Effect sizes were significant for similar comparisons as for positive symptoms, with ECT plus clozapine versus clozapine alone showing the largest effect. Furthermore, for total score, estimated mean differences were roughly double those seen in the positive subscore analysis, increasing our confidence in the model's consistency across different symptom domains.
4.4 Effects of electrode placement on clinical outcomes and seizure response
Our analysis revealed that patients treated primarily or exclusively with bifrontal placement had similar outcomes to those treated with bitemporal placement in our TRP sample. For positive symptoms, the benefit of combining ECT and clozapine was isolated to patients receiving bifrontal electrode placement, suggesting that bifrontal placement may be of particular benefit for positive symptoms in patients with TRP receiving clozapine. Conversely, bitemporal placement was associated with a greater reduction in negative symptoms among depressed patients in our sample, suggesting that negative symptoms may be better targeted by bitemporal ECT in this group.
For seizure outcomes, bifrontal electrode placement was associated with lower electrical doses required for ECT versus bitemporal when an anticonvulsant was present, suggesting a possible strategy for optimizing ECT delivery when concomitant anticonvulsant use is felt to be clinically warranted. Bifrontal ECT has evidence for less severe cognitive effects [
], further underlining the importance of measures that decrease the required stimulus intensity. Taken together, these findings suggest that bifrontal deserves further consideration for use and study in TRP, where combination approaches to pharmacotherapy are common and there is a need to optimize delivery of ECT, especially given baseline cognitive impairment in this population [
]. It is also unclear how bifrontal versus bitemporal placement affects the electrical dose required to elicit a seizure. We found no significant difference between placements overall in this regard, but our finding that bifrontal ECT was associated with lower electrical doses in the presence of an anticonvulsant suggests that this placement may more effectively bypass the inhibitory effect of these medications on seizure response than bitemporal, which is an intriguing finding warranting further study.
4.5 Interactions between ECT, clozapine, and anticonvulsants
Our exploratory analysis revealed large effects of anticonvulsants on decreasing seizure duration (though these remained well above adequate, i.e., ≥15 s), and on increasing mean electrical dose, as expected. Clozapine was associated with a moderate effect on increasing seizure duration, equivalent to an increase in duration of 2.6 s per dose difference of 100 mg between patients. However, clozapine also appeared to be associated with a small increase in mean electrical dose. These conflicting findings suggest that clozapine's posited effect on decreasing seizure threshold [
] may be of limited practical impact in the setting of a clinical ECT practice. We did not see a significant effect of benzodiazepines on either measure, possibly reflecting the regular practice of holding these the night before ECT.
The interactions observed between anticonvulsants, seizure expression, and outcomes in our study merit further discussion. While anticonvulsants with mood-stabilizing properties have evidence for improving symptoms of psychosis, these same medications can impact ECT seizure adequacy by raising seizure threshold and therefore threaten effectiveness of the procedure. Our findings reflect this paradox: the use of ECT in the setting of concomitant anticonvulsant treatment was associated with a larger reduction in positive symptoms than ECT alone, while a large reduction in seizure duration and increase in apparent seizure threshold was associated with anticonvulsant use. The effect on positive symptoms appeared to be independent from, and possibly additive to, the effect of clozapine.
A possible implication is that continuation of anticonvulsant medication during a course of ECT may not adversely impact positive symptom improvement, provided seizures remain adequate. The further possibility that anticonvulsant medication can exert an additive benefit on this outcome in patients treated with ECT should be treated with caution, as this was an exploratory analysis. Nonetheless, from a practical clinical standpoint, our findings suggest that TRP patients who have benefited or are expected to benefit from continuation of an anticonvulsant need not be excluded from consideration of ECT, and vice-versa. Clinicians must balance the benefit of anticonvulsant continuation against the possibility of increased electrical dose-related side effects, mitigate these where possible (e.g., holding such medications the night before ECT), and carefully monitor seizure adequacy.
4.6 Limitations
As this was a retrospective study, any causative inferences from our models should be treated with caution. In particular, as allocation of treatments was not randomized, there may be potential confounders that could not be accounted for. For instance, comparisons between ECT patients taking clozapine and those not should be interpreted alongside the consideration that ECT patients not taking clozapine were primarily those who did not tolerate the medication. Our analyses of electrode placement and the impact of anticonvulsants should be treated as hypothesis-generating, as some subgroups within these analyses were relatively small. Finally, there were some limitations related to the pragmatic nature of this study, including that PANSS raters were not blinded to treatment condition, medication (including clozapine) dosages were not repeatedly measured during admission for non-ECT patients, and that clozapine serum levels and ECT adverse effects were not systematically recorded. However, we believe that our analytic approach, combined with our relatively large sample size, has mitigated these limitations to the extent that is feasible.
5. Conclusion
This retrospective study provides evidence for clinical effectiveness of the ECT-clozapine combination in patients with highly refractory psychosis. Our data suggest that bifrontal electrode placement is non-inferior to bitemporal for improvement of positive symptoms and may require lower electrical doses to achieve adequate seizures in the presence of an anticonvulsant, while bitemporal is potentially superior to bifrontal for negative symptoms. The continuation of anticonvulsants during ECT did not result in unacceptable decline in seizure adequacy, while the effect of clozapine on increasing seizure duration appears to be modest. Future prospective research is needed to further investigate the comparative efficacy of bifrontal versus bitemporal placement in this population, as well as the interactions between clozapine and anticonvulsant medication with ECT.
CRediT authorship contribution statement
Nicholas J. Ainsworth: Conceptualization, Methodology, Investigation, Formal analysis, Writing – original draft, preparation. A. Michelle Avina-Galindo: Investigation, Resources, Data curation, Project administration, Writing – review & editing. Randall F. White: Investigation, Resources, Data curation, Project administration, Writing – review & editing, Supervision. Denghuang Zhan: Methodology, Validation, Formal analysis. Elizabeth C. Gregory: Methodology, Validation, Writing – review & editing. William G. Honer: Conceptualization, Resources, Writing – review & editing, Supervision. Fidel Vila-Rodriguez: Conceptualization, Methodology, Resources, Writing – review & editing, Supervision.
Declaration of competing interest
The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Dr. Ainsworth receives research support from the University of British Columbia Clinician Investigator Program. Dr. White receives honoraria from HLS Therapeutics, is on an advisory panel with HLS Therapeutics, and receives consulting fees from the Canadian Agency for Drugs and Technologies in Health. Dr. Honer has received consulting fees or sat on paid advisory boards for In Silico Biosciences, Translational Life Sciences, and AbbVie. Dr. Vila-Rodriguez receives research support from CIHR, Brain Canada, Michael Smith Foundation for Health Research, Vancouver Coastal Health Research Institute, and Weston Brain Institute for investigator-initiated research; philanthropic support from Seedlings Foundation; and in-kind equipment support from MagVenture. He has received honoraria for participation in an advisory board for Allergan. Ms. Avina-Galindo, Mr. Zhan, and Ms. Gregory have no disclosures.
Acknowledgements
The authors wish to thank Dr. Geoffrey Smith, Prescilla Carrion, Ashley DeGraaf, and Dr. Olga Leonova for their contributions to maintaining the BC Psychosis Program clinical database, from which data were drawn for this analysis.
Appendix A. Supplementary data
The following are the Supplementary data to this article:
Treatment-resistant schizophrenia: treatment response and resistance in psychosis (TRRIP) working group consensus guidelines on diagnosis and terminology.
Double-blind, placebo-controlled, multicenter trial of selegiline augmentation of antipsychotic medication to treat negative symptoms in outpatients with schizophrenia.
Combined use of electroconvulsive therapy and antipsychotics (both clozapine and non-clozapine) in treatment resistant schizophrenia: a comparative meta-analysis.
Comparative study of effectiveness of augmentation with ECT in clozapine resistant schizophrenia (CRS) and non-clozapine resistant schizophrenia (Non-CRS).
Double-blind randomized controlled study showing symptomatic and cognitive superiority of bifrontal over bitemporal electrode placement during electroconvulsive therapy for schizophrenia.
00190-5&id=gr1.jpg){kind=link}
00190-5&id=gr2.jpg){kind=link}