The Inuence of Anaesthetic Choice on Electroconvulsive Therapy Parameters; Etomidate Versus Methohexital

Many of the anaesthetic drugs used for electroconvulsive therapy have anticonvulsant properties and may inuence ecacy of electroconvulsive therapy. With this study we aim to provide more information on the effect of etomidate and methohexital on seizure duration. We explore the relationship between induction drug, motor and electroencephalography seizure duration. Moreover, we study the relationship of seizure duration and number of therapies. In this retrospective study we collected data from patient records from 2005 until 2016. Inclusion criteria were the use of etomidate and/or methohexital and documentation of dosage, electroconvulsive therapy dosage and seizure duration. Exclusion criteria were missing data on either induction drug, dosage or seizure duration. Hemodynamic data, induction drug and dosage, duration of seizure (both motor and EEG seizures) and demographic data were all recorded in patient records. They studied a small group = 10) of with a total of 90 ECTs in a randomized cross-over study comparing etomidate and methohexital. Their results showed longer durations of EEG and motor seizures after etomidate with no clear dose-effect relationship.


Introduction
Electroconvulsive therapy (ECT) is a safe and effective treatment of severe and medication resistant psychiatric disorders such as depression and mania 1 . An electrical current is applied to the brain via transcutaneous electrodes while the patient is under general anaesthesia. This electrical current results in a generalized (motor) seizure accompanied by an acute cardiovascular response due to activation of the autonomic nervous system.
Despite uncertainties in the literature regarding good indicators for the e cacy of ECT, there is evidence that there is a relationship between the seizure duration and the effectiveness of ECT [2][3][4][5] . Many of the anaesthetic drugs used for ECT have anticonvulsant properties and may alter seizure parameters, e.g. the decrease of the duration of the induced seizure activity. They might therefore decrease the effectiveness of ECT.
Methohexital, a short-acting barbiturate, remains the most widely used anaesthetic for ECT due to its potential for long seizure duration and is recommended as drug of rst choice for ECT by the American Psychiatry Association [6][7][8] . Despite this fact, etomidate gained popularity as an anaesthetic agent for ECT, due to its positive effects on electroencephalography (EEG) and motor seizure duration. Furthermore, it has a fast onset of action, short duration of effect and offers haemodynamic stability 8,9 . The use of etomidate is associated with longer EEG and motor seizure durations. This might be because the absence of anticonvulsant effect of etomidate. Thus, it could be the induction drug of choice in patients experiencing inadequate seizure activity 6,7,10,11 . Furthermore, it is described as the only induction agent that may reduce ECT dosage 6 .
There are a limited number of studies comparing methohexital and etomidate in the use for ECT and these data are inconsistent 10,[12][13][14][15] . Some studies show no difference between etomidate and methohexital. Others show longer seizure duration and reduction of ECT dosage in favour of etomidate. Nonetheless, superiority or inferiority over methohexital could not be demonstrated with the presently available literature 8, 10, 12-14 . Study aim In our hospital either methohexital or etomidate is used in patients undergoing an ECT. The choice of induction drug depends on the preference of the anaesthesiologist. In the same patient, etomidate and methohexital might both have been used. This minimizes potential confounders and facilitates a 'crossover' design.
With the present study we aim to provide more information on the effectiveness of etomidate compared to methohexital on ECT parameters. We will explore the relationship between induction drug and seizure duration measured by both motor seizure activity and EEG. Moreover, we will study the relationship of seizure duration and number of treatments.

Study design
The present study is a retrospective analysis of electroconvulsive therapy performed in our clinical institution from 2005 until 2016. The study was approved by the medical ethical committee (Toetsingscommissie Wetenschappelijk Onderzoek Rotterdam en omstreken; TWOR, 2016-61) and the need for written informed consent was waived. This study was performed in accordance with the Declaration of Helsinki. Hemodynamic data, induction drug and dosage, duration of seizure (both motor and EEG seizures) and demographic data were all recorded in patient records.
We included patients receiving electroconvulsive therapy between 2005 and 2016. Patients were included if they received etomidate and/or methohexital and if dosage, ECT dosage and duration of seizure were recorded. Patients with missing data on either induction drug, dosage or seizure duration were excluded.

Outcome
We explored the relationship between induction drug and seizure duration measured by both motor and EEG seizure activity. Motor seizure duration was de ned as time from electrical stimulation to resolution of clonic activity in an isolated limb. EEG seizure duration was the time from delivery of electrical stimulus to postictal EEG suppression.
Furthermore, we studied the relationship between seizure duration and induction drug dosage and number of past treatments.

Data collection
Data was collected from the hospital records (both paper and electronic). Baseline data included age, sex, number of ECT sessions, psychiatric diagnoses and diagnosis of kidney failure, cerebrovascular accident or epilepsy. In addition, we collected the subsequent data per ECT session; anesthetic agent and dosage used per therapy, opioid use, antihypertensive drug use, benzodiazepine use, antipsychotic drug use, tricyclic antidepressant use, uni-or bilateral stimulation, impedance, pulse width of the electrical stimulus, frequency used, ECT dosage, motor and EEG seizure duration and hemodynamics (systolic and diastolic pressure, heart rate). Because of differences in schedule time of the ECT between different days, we also documented the day at which the ECT took place. We performed a strati ed analysis of these data.
The data was handled in accordance with the Dutch Data Protection Act and the privacy regulation of the Maasstad Hospital. Each patient was allocated an unique identi cation number. Data was directly entered into a secured online database management system. Permission for this study was granted by the Institutional Review Board.

Statistical analysis
Descriptive statistics Descriptive statistics were presented as counts and percentages (dichotomous variables) or means and standard deviations (normally distributed continuous variables). Non-normally distributed, continuous variables will be described as medians with interquartile range. Continuous data were compared using the paired t-test. Categorical data were compared using the McNemar's test. Effects were described with their 95% con dence intervals (95% CI).
Relationship between induction drug and seizure duration.
The relationship between seizure duration and induction agent was modeled by repeated measures analysis using a linear mixed model with random intercept for patient and random slopes for treatment number and induction drug. Measurements up to 365 days after start of convulsion therapy were included. To allow for easier interpretation of parameters, covariate ECT dosage was centered at 100%, while covariates drug dosage for etomidate and methohexital were centered at their most frequently prescribed dosages of 20 mg and 100 mg respectively. Covariates include a dummy variable for induction drug (etomidate, reference methohexital), number of treatment, ECT dosage (range 0-200%, centred at 100%) and induction drug dosage, centred at 20 mg (etomidate) and 100 mg (methohexital).
For both outcomes (motor seizure activity and seizure activity on EEG), the following model building strategy was undertaken. First, relationships between outcome and covariates time, ECT dosage, induction drug and dosage were graphically assessed by scatter plots. Secondly, saturated models were t for each drug separately, including all interactions for time, dose and induction threshold. Starting with the three-way interaction, interaction terms were sequentially removed in case the likelihood ratio-test Chi2-statistic was not signi cant for that term (which proved to be the case for all the interaction terms). Thirdly, models for both drugs were combined, starting with all two-way interactions between induction drug and covariates time, ECT dosage and dosage (the drug-dosage interaction term being forced into the model for interpretation). Again, interactions terms were sequentially removed based on the likelihood ratio-test Chi2-statistic. All models were t with a random intercept for patient and random slopes for treatment occasion and an unstructured covariance for the random effects under full maximum likelihood. For the nal models, the addition of a random slope for induction drug and allowing for heterogeneity in residual errors by drug were tested by the likelihood test under restricted maximum likelihood. These resulted in a small but signi cant improvement in model t and were hence incorporated in the nal models.

Demographics
From 2005 until 2016, 102 patients were treated with ECT. Yet, we only found complete hospital documentation of 37 patients (see gure 1).
A number of 37 patients were analysed. The mean age of the patients was 52.3 years (median 50, sd 13.7, interquartile range (IQR) 44-63 (number (N) = 37). Seventy six percent were female. Most patients were suffering from affective disorders (81%) and three patients (8%) were diagnosed with schizophrenia. Psychiatric diagnosis was unclear in four patients (11%). See Table 1 for the demographic data.
Medication use and comorbidity are described in Table 1. Benzodiazepines and antipsychotic drugs were the most frequently used drugs. We also documented the number of patients with missing information on their medication use.

Discussion
In our study, we have shown that etomidate is associated with a long seizure duration (both EEG and motor) compared to methohexital. Furthermore, there is a negative dose-response relationship with the use of both induction agents. We also demonstrated a negative association between seizure duration and number of treatment and between seizure duration and ECT dosage.
Literature regarding the optimal induction drug for ECT is inconsistent. Our ndings differ from a study by Eser et al 12 . They performed a retrospective analysis of 5482 ECTs in 455 patients treated between 1995 and 2003. There was no difference between induction with etomidate or methohexital in seizure duration 12,13 . The mean dosage of administered methohexital and etomidate was higher compared to our patient population (130 mg and 32 mg compared to 100 mg and 20 mg, respectively) 16 . And to achieve adequate seizures, patients with higher ECT dosage received methohexital more frequently than other induction drugs. In our patient group, the choice for using methohexital or etomidate was not based on ECT dosage but rather on the preference of the attending anaesthesiologist. Switches in induction drug in one patient were thus at random. The higher drug dosage and the use of methohexital in patients with higher ECT dosage could be an explanation for the different results.
A more recent systemic review and meta-analysis of Singh and colleagues 8 included seventeen trials involving the use of etomidate, methohexital, thiopental and propofol as induction agents for ECT. Four trials compared etomidate with methohexital [13][14][15]17 .
A total of 84 ECT settings were included in this meta-analysis. The etomidate group showed a longer EEG and motor seizure duration of etomidate compared to methohexital (2.23 and 1.45 seconds, respectively). This difference was, however, not statistically signi cant. Thus, the study was underpowered to demonstrate superiority of etomidate over methohexital. Another study in line with our nding is the research of Avramov's et al 10,14,15 . They studied a small group (N = 10) of patients with chronic depression with a total of 90 ECTs in a randomized cross-over study comparing etomidate and methohexital. Their results showed longer durations of EEG and motor seizures after etomidate with no clear dose-effect relationship.
Limitations of this study include the retrospective design and the limited number of patients (N = 37). Though there were 102 patients treated from 2005 until 2016, we only found hospital documentation of 37 patients. The large number of missing patient records might affect the internal validity of our study.
Comparing baseline characteristics of patients excluded because of missing data, no differences were found. This decreases the risk of non-randomly distributed missing data. We therefore think that our result are valid, both internally as externally. Despite the large amount of missing data, we were able to show a signi cant result, thus our study was not underpowered. Another limitation is the retrospective design and the large study period. During this study period, treatment remained largely the same. Any treatment effect found in this study is less likely due to chances in treatment but more likely due to a true difference in seizure duration between etomidate and methohexital. Regarding our outcome measures, we were unable to study effectiveness of ECT. We used seizure duration as a proxy. Although there are uncertainties in the literature regarding good indicators for the e cacy of ECT, seizure duration is universally accepted as marker for effectiveness of ECT 2-5 . Antiepileptic medication was not corrected for in the nal model. Although mentioned in the guidelines as medication which affects the seizure duration, there is no reason to presume medication used differed between the two groups. 11 The strength of this study was the fact that we were able to include a good number of patients who were anesthetized with both etomidate and methohexital. This made the patients their own control, thus elimination several confounders. Furthermore, the choice of induction drug was based on the preference of the attending anaesthesiologist. This made confounding by indication less likely. We used this in our analysis by using linear mixed model and at the same time we could correct for confounders as ECT dosage, induction drug dosage and sensitization effect (number of treatments). We believe that, due to lack of confounding by indication, non-random loss-to-follow-up and an advanced statistical model, our results are valid. Although a difference of 6,69 seconds in motor seizure duration and 13.4 seconds in EEG seizure duration might not seem clinically relevant, it's a decrease of 23% and 39% of the average motor and EEG seizure duration respectively.
In conclusion, this study demonstrated a longer EEG and motor seizure duration with the use of etomidate in comparison to methohexital. Moreover, we observed a negative association between seizure duration and number of treatments and a negative association between seizure duration and ECT dosage. With this study we contribute to the available literature comparing methohexital and etomidate as induction agents for ECT.
Declarations Figure 1 Flowchart Measurements up to 365 days after start of electroconvulsive therapy were included. We excluded other anaesthetic induction agents than etomidate or methohexital. We excluded patients with missing data of motor or EEG seizure duration, ECT dosage and induction dose as well. A total number of 679 and 551