Dosing methods in electroconvulsive therapy: should the Scandinavian time-titration method be resumed?

Abstract

Aim

To describe and evaluate the different dosing methods in ECT, and bringing back into focus the Scandinavian time-titration method.

Method

A narrative, unsystematic, and selective review and discussion.

Results

There are five dosing methods: 1) The Scandinavian time-titration to tonic convulsion, using low-frequency pulses and a long maximal pulse train duration, was highly efficacious with right unilateral (RUL) ECT, comparable to bitemporal (BT) ECT. However, the device used went out of production in the 1990s. Because US devices until 1990 had a short maximal pulse train duration, time-titration went out of use. 2) Fixed high dosing at 50–100% of the device’s maximal output, and 3) Formula-based dosing, initially the Age method, long prevailed in USA. Later, the Half-Age method was introduced for BT ECT. 4) Charge-dosing as a multiple of titrated seizure threshold (ST) demonstrated RUL and BT ECT to have comparable outcomes when dosed about six and two times the ST, respectively. 5) Dosing from benchmark is based on a high dose at the first session, to ensure a high peak heart rate and tonic-clonic convulsions. In later sessions the lowest dose producing similar outcomes is chosen.

Conclusions

No dosing method is documented superior to the others. Seizure threshold-based and benchmark dosing seems to be more accurate than fixed high and formula-based dosing. However, time-titration dosing makes it possible to adjust the dose at every session, and may be the most efficient method.

Keywords:

• Electroconvulsive therapy
• dosing
• titration
• pulse train
• pulse frequency

Introduction

Electroconvulsive therapy (ECT) was first given in April 1938 by Cerletti and Bini in Rome. Bitemporal (BT) electrode placement and sinus wave current were used. Later, several ECT devices were manufactured and different stimulus types and electrode placements tested [1,2]. Using pulse current instead of sinus wave current, the electrical dose could be significantly reduced and cognitive side-effects were less, with no change in clinical response. The seizure-eliciting and antidepressant potentials of the pulse current depend on pulse width, amperage, charge, pulse frequency, and number of pulses, i.e. pulse train duration. Pulse frequency and train duration are relevant to our presentation, because decreasing pulse frequency is associated with lower seizure threshold (ST) and better treatment outcome [3–11]. Accordingly, Roepke et al. [8] and Gangadhar et al. [7,10,11] found higher response rates with 80 vs. 200 pulses per second (pps), and with 50 vs. 125 pps, respectively. Some data suggest that the optimal pulse frequency may lie in the 25–60 pps range [6]. Additionally, a sufficiently high maximal pulse train duration is a prerequisite for a low frequency pulse stimulus to deliver a high enough number of pulses.

In the first decades after the introduction of ECT, different traditions regarding the choices of stimulation technique and ECT devices developed in the US and Europe [1,12]. In the US, two devices have dominated for the last 40 years, i.e. MECTA, first produced by Blachly in 1973, later tested at Colombia University, and Thymatron, introduced by Abrams and Swartz in 1985 [13–15]. The maximal pulse train duration with MECTA was 2 s from 1973, 2.8 s from 1985, 6 s from 1997, and 8 s from 2003. With Thymatron it was 4 s from 1985 and 8 s from 1990. In Scandinavia, leading ECT researchers used the Siemens Konvulsator, first described in 1951 by von Braunmühl [1,16]. It delivered a maximal pulse train duration of 10 s. The US devices required delivery of a preset amount of energy, whereas the Konvulsator allowed the stimulus duration to be determined by the therapist while watching the motor signs of the seizure and ending the current when a seizure presumed to be adequate had been elicited [12,17]. This is the Scandinavian time-titration dosing. The Konvulsator went out of production in the 1990s and was replaced with MECTA and Thymatron. Both have the advantages of being constant current devices with built-in EEG-recorders, whereas the Konvulsator was a constant voltage device without an EEG-recorder. With the 8 s maximal stimulus duration of the latest versions of MECTA andThymatron, the Scandinavian technique could have been resumed. This can be done by using the maximal pulse train duration, a low frequency, and releasing the stimulus button when a presumed adequate seizure has been elicited. However, this has not been done.

The optimal method for individualizing the stimulus dose in ECT is not known [18–21]. Consequently, several techniques are in use. In the 1980s, one of us (P. B.), supervised by Giacomo d’Elia, practiced the Scandinavian technique, which we think may have advantages. Therefore, we present a narrative, unsystematic and selective review and discussion of dosing strategies, that has the explicit aim of bringing back into focus the Scandinavian time-titration dosing.

Dosing methods

Over the years, five dosing methods have been developed: 1) Time-titrated stimulation until a generalized tonic convulsion appears, 2) stimulation with a fixed high dose, 3) formula/age-based dosing, 4) dosing as a multiple of ST, and 5) dosing from benchmark. Optionally, the dose is adjusted during the treatment course according to clinical response, EEG-recorded seizure quality and peak heart rate.

The Scandinavian time-titration dosing

In the 1950s and 60 s, Cronholm and Ottosson [22,23] compared two ECT devices in BT ECT, the Elther ES and the Siemens Konvulsator III. The Elther ES delivered ultrabrief rectangular pulses of 0.1 ms with a frequency of 15 pps and a pulse train duration of 2 s for each age decade [1]. The Konvulsator III delivered quarter sine-wave pulses of 50 pps from 50 Hertz sine wave current. For every fourth pulse, the next four could be left out, giving intermittent volleys of four pulses with 95 ms intervals. Maximum pulse train duration was 10 s [1]. The later version Konvulsator 622 delivered modified quarter sine-wave pulses which started abruptly 0.7 ms before the apex, resembling brief rectangular pulses of 1.4 ms with a steeply declining tail of 4.3 ms [24]. Cronholm and Ottosson found that clinical improvement was greater with the Konvulsator, probably because ultrabrief pulses with the Elter ES sometimes led to insufficient seizures. The influence on memory did not differ. Therefore, they chose the Konvulsator for further use.

With the Siemens Konvulsator 622, d’Elia, Ottosson, and Sand Strömgren developed the Scandinavian practice of ECT, summarized in 1983 [17]: (1) unilateral nondominant/right parietotemporal electrode placement with at least 12 cm between the midpoints of the electrodes, i.e. d’Elia position, (2) stimulus with few pps and a long and not prefixed pulse train duration, and (3) interruption of the stimulus when the initial muscle contractions shift to the generalized tonic phase, ultimately observed by maximal tonic extension of the big toes. This leads to a dose slightly above the limit necessary to elicit a self-sustaining generalized tonic-clonic seizure.

The convulsions from maximum seizure activity usually end gradually, starting distally, and are succeeded by a comatose stage from which consciousness is regained progressively. The optimal time of disorientation, i.e. the postictal reorientation time (PRT) [25,26], is not known. However, it may be about half an hour. Accordingly, in three randomized controlled trials (RCT) by Sackeim, McCall et al. [27–29] PRT was in the range of 30–40 versus 10–20 min in the groups with best and poorest treatment outcomes, respectively. Moreover, in the elderly, Bjølseth et al. [26] found that longer PRTs (from the resumption of spontaneous respiration and eye opening) at the first and third ECT session predicted more rapid and better response. None of the patients with PRTs less than five minutes recovered, whereas all with PRTs of 35 min or more remitted after 12 sessions or less.

Right unilateral (RUL) ECT applying this time-titration method using the Siemens Konvulsator 622 was documented highly efficacious and equal to bitemporal (BT) ECT in endogenous depression, but with markedly less cognitive side-effects, in a double-blind RCT by d’Elia [30] in 1970. These good results were replicated in a double-blind RCT by Sand Strömgren, Fromholt and Christensen [31,32] in 1973, although the less appropriate Lancaster position was used. A second replication of the comparable efficacy of the time-titration method with the d’Elia position was published in 1994 [24]. The pulse train duration with intermittent volleys was 1.8–8.0 s in the first two trials. In the last trial, it was 4–10 s probably due to a relatively high dose of 3.5–4.5 mg/kg thiopental used in that study.

Sand Strömgren [33] also reviewed 57 ECT-series in 51 patients with diagnoses other than endogenous depression treated at Aarhus Psychiatric Hospital in 1984. The time-titration method with the d’Elia position and intermittent volleys were used. Only in three series two to three supplementary BT sessions were given. The diagnoses were mania (15), mixed state (19), reactive psychosis (8), schizophrenia (6), delirium (1), and other (2). The effect was satisfactory in 53% and moderate in 30% of the series, although the relapse rate was 30% within three months.

Furthermore, in 1984, Sand Strömgren [34] published a retrospective study in patients who did not respond to 6–10 RUL ECT sessions, comparing switching to BT ECT with continuation of RUL ECT. No significant difference was found, i.e. 41% (11/27) improved considerably after being switched to BT ECT, 59% (19/34) after continuation of RUL ECT. This is the only comparative study of this important clinical question. The technique was once again time-titration with intermittent volleys of low-frequency pulses.

Fixed high dose and formula/age-based dosing

In the US two dosing methods long prevailed: fixed high dosing at 50–100% of the maximal output (576 mC for MECTA and 504 mC for Thymatron), and formula/age-based dosing [35–43]. The fixed high dosing is an old tradition. We have found no reference to when it was introduced. The Full-Age dosing of five times age mC was introduced by Abrams and Swartz in 1985 [35]. However, according to Petrides and Fink [12], these methods were criticized for causing overstimulation when using BT ECT, exposing patients to unnecessary cognitive side-effects. Therefore, in 1996 they tested and introduced the Half-Age method, i.e. 2.5 times age mC, in BT ECT. Only one out of 55 patients in their study had to be re-stimulated. In case seizure quality diminished, the dose was increased with 10 energy percent (50 mC). No one needed more than 350 mC, and the clinical response was very good. Later, Abrams [35] summarized studies showing that a mean doses of ≥378 mC with brief-pulse RUL ECT and ≥187 mC with BT ECT had improvement/response rates from 65 to 89%.

With the formula-based Full-Age and Half-Age methods other factors such as gender may be included [41]. The ST is higher in men than in women, but the difference varies greatly, from minimal to nearly 60% in different studies [44–48]. In two Norwegian RCTs, the age-based dose was adjusted upwards with 5–10 energy percent for men and downwards with 5–10 energy percent for women [49,50].

In 2001, Kellner [39] recommended the start dose in RUL ECT to approximate 75% of the maximal output, in BT ECT to lie in the middle third, and in bifrontal (BF) ECT to approximate 50% of the maximal output. In 2002, Abrams [42] recommended starting with a fixed maximal dose in RUL ECT, using the Full-Age method in BF ECT, and the Half-Age method in BT ECT. Swartz and Nelson [51] were in favor of applying the Half-Age method for the three bilateral electrode placements, i.e. BT, BF and left anterior right temporal (LART), given an amperage of 900 mA. Notably, in an RCT comparing RUL vs. BF ECT applying the Full-Age and Half-Age methods, respectively, Bjølseth et al. [49] found significantly higher increments in doses during the courses and a higher proportion of sessions in which re-stimulation was necessary, with the BF placement. The findings indicate that the Full-Age method, and not the Half-Age method, is more appropriate in BF ECT, in line with Abram’s recommendation. Bennett et al. [52] concluded retrospectively that with BT ECT a fixed dose of 200 mC for those <65 years and 250 mC for those >65 years would simplify the ECT process and result in more patients receiving effective treatment at the first session than with the Half-Age and ST-based dosing.

Seizure threshold-based dosing

In 1987, Sackeim et al. [53,54] reported a titration procedure to define the ST in mC, and demonstrated that BT ECT has superior efficacy compared to RUL ECT when the dose approximates the ST. They assumed that the treatment outcome of RUL ECT is dependent on the extent to which the dose exceeds the ST, i.e. that a generalized seizure is necessary, but not sufficient, for a good clinical response. In the years to follow, they developed the ST-based dosing. In 1993 and 2000, they published RCTs that lend support to their hypothesis [28,29]. BT ECT retained its superior efficacy over RUL ECT with a dose of 2.5 times ST. With a ST dose BT ECT was slightly less efficacious, whereas RUL ECT did not work. However, RUL ECT dosed 6 times ST compared to BT ECT dosed 2.5 times ST. RUL ECT dosed 1.5 times and 2.5 times ST did not. Additionally, in 2000, McCall et al. [27] demonstrated with RUL ECT that a fixed high dose of 403 mC was more efficacious than a dose of 2.25 times ST. ST titration was performed in all their patients. Therefore, it could also be shown that doses of 8–12 times ST were more efficacious than 3–5 times ST. However, higher doses were associated with more cognitive side-effects, assessed 1–2 days after the last session.

A meta-analysis of eight trials from 1993 to 2010 found no difference in efficacy between BT and BF ECT at 1.0–1.5 times ST and RUL ECT at various doses up to six times ST [55]. Mini-Mental State Examination decline was less with BF than BT ECT. RUL ECT impaired Complex figure recall more than BF ECT, but BF ECT impaired word recall more than RUL ECT. A meta-analysis of seven trials from 2000 to 2016 found no difference in efficacy between BT ECT at 1.0 to 2.5 times ST and RUL ECT at 5–8 times ST [56]. Retrograde autobiographical memory was better with RUL ECT, however, not in the study using a dose of eight times ST [57].

Benchmark dosing

Benchmark dosing means measuring the quality of each ECT session against an optimum in that patient, and adjusting the dose accordingly along the treatment course. The method was developed by Swartz et al. [58–66] and first described in 2002 [63]. At the first session, a relatively high stimulus dose is selected in order to produce an intense seizure. In the latest description by Swartz [38] this dose is 2.5 times age mC with LART or BT ECT and five times age mC with RUL ECT, at 900 mA and 0.5 ms pulse width. Anaesthesia includes glycopyrrolate or atropine, low-dose methohexitone, succinylcholine, and maximal hyperventilation. Peak heart rate (PHR) and tonic motor seizure are monitored during the course. If no tonic motor activity occurs, the stimulus dose is increased. The dose is also increased if the PHR is less than 140 bpm, unless the patient is older than 80 years or has a specific medical reason for a low heart rate, such as cardiovascular disease, the use of propofol or a betablocker. Swartz et al. [60–62,64,66] have found the PHR, which is controlled by the locus of cardio-acceleration in the right medulla, to be a more reliable physiological indicator of dose and clinical response than EEG ictal parameters. The benchmark dosing has not been used with RUL ECT [38]. However, it has been very efficacious in small studies with bilateral electrode placements, especially LART [38,62,67–70], which was recently shown to be a favorable placement in a computational modelling study [71].

Discussion

Recommendations from clinical guidelines

Following the ST-studies in the 1990s, the American Psychiatric Association in 2001 recommended formula/age- and ST-based dosing, whereas the use of a fixed high dose should be reserved for special situations [41]. ST-titration was considered most accurate. The recommended dose was 2.5–6 times ST in RUL ECT and 1.5–2.5 times ST in BT ECT. In the 1990s, the Sackeim group applied 1.5 ms pulses and a maximal pulse train duration of 2 and 4 s [28,29]. Later on, pulses in the 0.25–1 ms range and a maximal pulse train duration of 8 s became the norm, thus lowering the ST. Accordingly, it is now recommended to dose RUL ECT 5–6 times ST, whereas BT and BF ECT should be dosed 1.5–2.5 times ST, as before [37,72–74]. The British guidelines of 2019 claim that ‘Direct measurement of the ST has real advantages to maximise the clinical effectiveness of treatment and minimise the risk of adverse effects’ [75]. The Australian and New Zealand guidelines of 2019 recommend the dose titration method over age-based methods [73], and the Canadian guidelines of 2016 mention the ST-titration method only [74].

Critique of seizure threshold-based dosing

In 2001, Swartz [76], in a guest editorial in the Journal of ECT, summed up a thorough critique of the ST-based method in 13 points. He pointed out that a fixed multiple of the ST does not have similar effects in different patients, and that there is no rationale for selecting a specific multiple for the individual patient. In 2001, Kellner [39] wrote that titrated dosing is unnecessary for most patients. In 2002, Abrams [42] came with a comprehensive critique of titrated dosing, and in 2014, Fink [77] concluded with ST determination neither being necessary nor useful.

In cases when formula/age-based or fixed dosing is applied, a significant number of patients will receive higher or lower doses than with ST-titration [19,78]. Titrated dosing makes it possible to detect patients who have extra high or low STs [79]. However, titration is done in steps with untested intervals, and the threshold EEG is not always distinct, leaving the ST at a rough estimate [80–82]. This is reflected in a three-fold variation in mean STs among different studies with equal electrode placements and pulse lengths, in spite of small age differences [48,83]. Furthermore, a significant proportion of patients may have a ST lower than the first step of many existing dose titration protocols [83].

The critique of the ST-based dosing was resumed in 2018 by Rosenman [84]. He presents strong arguments for his conclusion that ‘Seizure threshold titration … is not a proven technique of dose optimization. … It is a prematurely settled answer to an unsettled question that discourages further enquiry. It is an example of how practices, assumed scientific, enter medicine by obscure paths.’ Kellner and Borys [18] commented that they ‘completely agree and believe this represents the field having rushed from small research datasets to premature implications in clinical practice.’ They suggest the possibility of a good response to low-dose RUL ECT in a substantial minority of patients, and question the widely accepted idea of the extreme range and variability of STs.

McLoughlin [20] commented that ‘the best way to administer ECT with regard to optimizing therapeutic benefit and simultaneously minimizing cognitive side-effects is not known.’ This was recently supported by Landry et al. [19], who published a systematic review of the clinical relevance of the ST-based method compared to the age- and fixed dose methods, based on a literature search in March 2020. They concluded that no clear recommendations could be drawn regarding the clinical superiority of one method.

Advantages and limitations of the Scandinavian time-titration dosing

Sienaert [21] comments on Rosenman [84] that ‘An alternative adequate and evidence-based dosing-strategy would be welcomed by the field,’ and points to the lack of replication studies. The Scandinavian time-titration dosing may be such an alternative, because it is simple, efficacious and replicated in several studies. It was once an established technique, dependent on a device delivering pulses with a low frequency and a long maximal pulse-train duration of 10 s. It fell out of use with the disappearance of the Siemens Konvulsator. However, at about the same time the maximal pulse-train duration of the US devices was increased to 8 s. This was the longest time needed in all but one of the Scandinavian studies [24,30–34]. Consequently, time-titration is possible with these devices. The Thymatron even have a program for intermittent pulse-volley stimulus mimicking the Konvulsator [85].

The ST often increases during an ECT course [40,41,44,47,80,86–91], implying that ST-based and formula/age-based doses apply to the first session only. The dose has to be adjusted upwards at subsequent sessions in the absence of clinical response and optionally, if EEG-recorded seizure activity abates [37,72,92]. Alternatively, the ST may be re-titrated. In maintenance-ECT, the dose may have to be reduced [93,94]. With time-titration to tonic convulsion this problem is avoided, because the dose is titrated beyond the ST and directly to a presumed therapeutic level at every session, i.e. to slightly above the limit for a self-sustained, generalized, tonic-clonic seizure. Such a titration is performed by using a pulse current with low frequency and variable pulse train duration. During the stimulation, the contractions should be continuously observed. When these turn into a generalized tonic phase, ultimately observed by maximal extension of the feet and big toes, the stimulus is interrupted [17,24,84]. Because stimulation is continued beyond the initial muscle contractions corresponding to the ST, it fits with the importance of dosing far above the ST with RUL ECT. The tonic phase soon turns into a clonic phase. The seizure is succeeded by a comatose stage from which consciousness should be regained gradually [17], indicating activation of deep cerebral structures. The optimal duration of the PRT is not known, However, it may be about 30 min [26–29], and should probably be no less than five minutes from the resumption of spontaneous respiration and eye opening [26].

Direct observation of tonic convulsion as the dosing criterion may seem crude, however, so is the dosing criteria of the other methods, including seizure threshold determination. Formal reliability studies are lacking for all the methods. However, the validity of the time-titration dosing is supported by good clinical results over several decades. It requires experience by the observer, pressing and releasing the stimulus button, and the feet must not be covered with socks or clothes. As soon as the feet and big toes are extended maximally in a tonic position the current is interrupted. A common error is releasing the button too early. This may result in one of two submaximal convulsions [17]. The one is ‘dissociate convulsion,’ which superficially resembles an optimal seizure, but is followed by early awakening. The other is ‘clonic convulsion’, which has no tonic phase, ends abruptly and simultaneously in the whole body, and usually lasts only 10–20 s. The dose of succinylcholine chloride should be 0.50–0.75 mg/kg [17]. If a larger dose is required the Hamilton cuff technique used on the right ankle may be applied [37,75].

In conclusion, the optimal method for individualizing the stimulus dose in ECT is not known. The previously established Scandinavian time-titration dosing may be the simplest and most efficacious stimulation method for individualized dosing [3,6]. Time-titration secures a sufficient and individualized number of pulses with an efficacious low pulse frequency, given a device delivering a long maximal pulse train duration. A sufficiently long PRT may be the ultimate sign of a therapeutic seizure. The Scandinavian time-titration dosing has not been tested against other methods. We think it was prematurely abandoned, should be resumed, studied more closely, and compared to other methods.

Disclosure statement

There are no conflicts of interest associated with this publication, and there has been no financial support for this work that could have influenced its outcome.

Data availability statement

Data sharing is not applicable to this article as no datasets were generated or analyzed during the current study.

Correction Statement

This article has been republished with minor changes. These changes do not impact the academic content of the article.

Additional information

Notes on contributors

Per Bergsholm

Per Bergsholm (b. 1945) is a Norwegian specialist in neurology, clinical neurophysiology, and psychiatry. He has a dissertation on ECT guided by Giacomo d'Elia, and has published about 50 papers, mostly on mood disorders and ECT. He participates in providing national courses on ECT for Norwegian clinicians.

Tor Magne Bjølseth

Tor Magne Bjølseth (b. 1965) has since 2002 worked as a senior consultant in geriatric psychiatry at Diakonhjemmet Hospital, Oslo. He is in charge of the ECT clinic at the hospital. In 2016 he defended his thesis on predicting the treatment outcome of ECT, emphasizing the post-ictal reorientation time.