Two decades of an indigenously developed brief-pulse electroconvulsive therapy device: A review of research work from National Institute of Mental Health and Neurosciences.

In 1993, a device to administer brief-pulse electroconvulsive therapy was indigenously developed through collaboration between the National Institution for Quality and Reliability and the National Institute of Mental Health and Neurosciences (NIMHANS), Bengaluru, Karnataka, India. The additional feature of computerized recording of the electroencephalograph and electrocardiograph for both online and offline use had substantial clinical and research implications. Over the past two decades, this device has been used extensively in different academic and nonacademic settings. A considerable body of research with clinical and heuristic interest has also emanated using this device. In this paper, we present the development of this device and follow it up with a review of research conducted at NIMHANS that validate the features and potentials of this device.

INTRODUCTION

Indian researchers have contributed substantially to the field of electroconvulsive therapy (ECT). Their contribution has gone beyond academic research – the team of researchers in the National Institute of Mental Health and Neurosciences (NIMHANS), Bengaluru, Karnataka, India, has developed indigenous ECT devices that have stood the test of time in terms of their use in clinical practice as well as in their contribution to research. Initially, sine-wave ECT device was developed at NIMHANS. It delivered bidirectional stimuli, and facilities for measuring various parameters of electrical dose had been incorporated in the device.[1] With the demonstration of cognitive superiority and comparable efficacy of brief-pulse ECT,[23] brief-pulse ECT equipment soon became the standard of treatment. The national workshop on ECT (1988) recommended the development of state-of-the-art ECT devices with brief-pulse waveform to be available indigenously. No ECT machine sold in India during that period had this feature.

A committee including the member of the Bengaluru chapter of the National Institution for Quality and Reliability, a psychiatrist (BNG), and a biomedical engineer (VSC) took up the task of developing a brief-pulse ECT device indigenously. After several laboratory tests, the machine was finally found suitable for safe use in 1992. It incorporated all the features then considered necessary and featured in international models. The focus on the simplest of the models was the ease of use by psychiatrists. This was possible by a simple microcontroller unit that transmits a brief-pulse (1.5 ms as width) of bipolar nature (at a frequency of 125 pulses/s) and constant current (800 mA). The new device ensured constant current through an electronically controlled feedback mechanism. The stimulus was calculated in milli-Coulombs (mC). Ten steps of stimulus dose was made available by adjusting the total duration of stimulus from 0.2 s to 3.6 s and keeping current, pulse width, and pulse frequency constant. The first two stimulus steps were 30 mC and 60 mC; the next eight steps were in multiples of 60 mC up to 540 mC. The stimulus was delivered by pressing the switch provided on the hand-held electrode. The duration of stimulus was preprogrammed based on the total charge chosen and would be shorter if the switch was released earlier. This was a safety provided to assort stimulus manually. The same switch also started a seconds counter to help time the observed motor seizures. The treating physician should keep the switch pressed to let that seconds counter on; release of switch at the termination of seizure stops the counter and indicates the duration (in seconds) of the motor seizure. Release of the switch earlier than 20 s (failing to see the convulsion of duration longer than 20 s) automatically would increase the stimulus dose setting to next level for re-stimulation if the doctor chooses to do so. Features such as internal trigger to disconnect the electroencephalograph (EEG) electrodes from external amplifiers during stimulus helped using standby EEG/electrocardiograph (ECG) monitoring free of artifacts.

ECT device could be connected to a conventional computer. This allowed changing the settings of other electrical parameters such as pulse width and frequency if required while keeping the default setting same as the earlier microcontroller-programmed steps. However, it gave liberty to set either 10 steps of increment as in the basic model or 20 steps increase of 15 mC as 15, 30, 45, 60,… up to 540 mC. Automatic upgradation of the stimulus dose, if the observed seizure was <20 s, was preserved. By 1994, the computer allowed paperless recording of EEG and ECG.[4] The computer monitor would display the EEG in two, four, or eight channels along with ECG. After EEG recording is terminated, a 10-s calibration signal of 100 μv was recorded. There was also option of choosing different EEG settings, including number of channels, and display size (zoom).

This refined EEG and computerized model is in use in NIMHANS since 1993. In 1998–1999, the option of automatic setting of threshold stimulus dose in mC, by feeding data on age, sex, and inion-nasion distance (IND), was developed. The research related to measurement of impedance and stimulus threshold of ECT in NIMHANS contributed to it.[5] Around the same time, feature to store ECT seizure and ECG for offline analysis was made available. With popularity of bifrontal ECT (BFECT), special concave stimulus electrodes were added. Most recent development includes facility of administering ultra-brief-pulse (<0.5 ms) ECT that is expected to cause lesser cognitive dysfunction.[6] The latest integrated ECT machine, with four-channel EEG/ECG, has intelligent features to independently operate without interfacing with a computer system. EEG/ECG is displayed on 5.7" color graphic LCD screen, and it also has SD card memory and printer interface. A separate portable EEG amplifier independent of ECT can remain attached to the patient and the signals can be seen on any hand-held device that has Bluetooth.

All ECT machines at NIMHANS are periodically calibrated for pulse width, frequency, amplitude, and stimulus duration stages. The frequency and duration are crystal-driven measures and highly stable. Current has also remained stable at 0.8 mA. Over the past two decades, this device has been used to administer ECT to about 700–800 patients annually. With each patient receiving about 6–7 ECTs on an average through his/her course, about 5000–7000 ECTs are delivered every year. The precision and flexibility of electrical parameters provided by this ECT machine have fostered quality research on ECT at NIMHANS. We have selected for review those studies from NIMHANS which have utilized specific features of this device. Some of the important findings from these studies are presented below.

SEIZURE THRESHOLD

Stimulus relative to seizure threshold is more critical than the absolute stimulus.[7] A few techniques of threshold estimation included the half-age method and dose titration method[89] but have limitations. The former was dependent on the model of the device while the latter might expose the patient to inadequate stimuli. The study at NIMHANS in this regard on 100 consecutive patients receiving bilateral ECT (BLECT) revealed that age was the most important predictor of threshold stimulus accounting for 23% variance in multivariate analysis.[10] The other significant but less robust factors were IND and severity of illness that accounted for 5% and 4% of variance in threshold, respectively. The psychotropic medications and head circumference were not noted to influence the threshold. The major positive association with age was in conformity with the existing literature.[1112] A formula as mentioned below for determining stimulus threshold was devised using forward stepwise linear regression model based on age alone and was prospectively tested at the 1st and 6th ECT sessions.

T1 = 1.67 (age) + 48.7

T6 = 1.8 (age) + 71.9

T1 = Threshold at 1st session, T6 = Threshold at 6th session.

This could produce the adequate seizure in 82% and 84% of patients, respectively.[1314] The computerized version of stimulus setting allows use of this regression equation with ease. With the computerized version, ECT sessions of each patient can be archived in a spreadsheet. A similar effort was made to find the seizure threshold for unilateral ECT (ULECT). Here, age was the only significant predictor of seizure threshold (15% of variance). A formula based on regression analysis was developed which is as follows:

Threshold (mC) = Exp (Age [0.015412] + 3.667387), where unit of age is years.

Then, in an independent sample (n = 30) of patients receiving ULECT, the threshold was estimated (but not administered) using the above formula. The value obtained was rounded off to the nearest higher level on the ECT device. This estimated threshold was then compared with the actual threshold stimulus which was determined by using the titration method. The results showed that 22 (73%) patients would have developed adequate seizure with the first stimulus estimated by the formula method, but it overestimated threshold stimulus by 30–60 mC in majority of them. Moreover, it underestimated the threshold in 8 (27%) patients. Hence, the formula method may not be suitable for threshold estimation in ULECT. However, if the ECT administrator wishes to use it, he or she should do so with the stimulus set at one or two levels lower than the formula-estimated threshold estimation, and further titration may be done as needed.[5]

Although stimulus at 2–6 times the threshold intensity is associated with greater clinical efficacy in ULECT, the guidelines were less clear about this in BLECT. From 2006, there was change in practice in NIMAHNS and unless otherwise requested from the referring psychiatrist, all BLECTs are administered at 1.5 times the threshold stimulus charge. Before 2006, the threshold stimulus was administered for BLECT. Hence, a retrospective data analysis was done to see whether new practice has better outcome. The data of 100 consecutive inpatients who received BLECT at threshold level before the change in practice implemented were compared with that of 101 who received BLECT at 1.5 times threshold level after the change implemented. The outcome measures were the number of ECTs administered and the number of inpatient days after the start of ECT. There was no change in the outcome of patients with schizophrenia and depression. However, the change to suprathreshold BLECT conferred advantage in the treatment of mania: Those receiving suprathreshold ECT achieved clinical improvement with, on an average, 2 ECTs less and their inpatient days reduced by 10 days.[15]

ELECTROENCEPHALOGRAM MONITORING

ECT guidelines mandate the need for seizure monitoring during ECT.[1617] Initially, feasibility, reliability, and validity of EEG seizure measurement were studied at NIMHANS using a separate paper EEG machine.[18] Subsequently, a paperless EEG amplifier, which was attached to a computer monitor displaying the EEG, was designed at NIMHANS.[4] Display of both EEG and the digital counter showing time duration was triggered by the ECT stimulus switch (currently, EEG monitoring is available in a display in the ECT machine obviating the need for a computer attached to the ECT machine). This paperless EEG was validated against a paper EEG measurement which showed high correlation in terms of seizure duration (intraclass correlation, r = 0.99, P < 0.001). In later studies using this paperless EEG machine, unequivocal absence of epileptiform transients for five or more seconds on both channels was taken as the end of EEG seizure.[192021] The use of this paperless EEG simplified EEG monitoring by reducing mechanical problems associated with paper EEG and simplified the storing process. Further, this allowed us to apply different algorithms to study EEG during ECT.

Jayaprakash et al. (1998) monitored 158 consecutive patients referred for ECT on their first ECT for both motor and EEG seizures.[19] All patients received bilateral threshold ECT. About 20.3% of the sample had prolonged EEG seizures (>120 s). Around 40% of patients with prolonged EEG seizures had a seizure duration >180 s, necessitating termination. Motor seizure of 90 s failed to predict prolonged EEG seizures (120 s or more) in 60% of occasions. A sizable minority (8%) failed to obtain adequate motor seizure on the cuffed forearm although their EEG seizures were considered adequate by the study criteria (≥25 s). A similar study was conducted by Mayur et al. with 232 consecutive patients posted for either BLECT (54%) or ULECT (46%). In this study, 16% had prolonged EEG seizures, 39% of which was missed by motor monitoring, and 7% of adequate EEG seizures were classified as inadequate by motor seizure monitoring.[20] Around 1% of sessions with ULECT and 11% with BLECT were considered inadequate by motor seizure monitoring despite adequate EEG seizures. Motor seizure correlated well with EEG seizure when the latter was adequate but not when prolonged. These findings were again replicated in a further larger study with 485 consecutive patients receiving either BLECT or ULECT.[21] In the later study, two patients classified as having inadequate motor seizures (<15 s) had prolonged EEG seizures (>120 s). This study identified risk factors for prolonged seizures. Patients with prolonged seizures were younger, had lower seizure threshold, were more likely to receive a diagnosis of mania, and were treated with lithium.

The above studies emphasize the need for EEG monitoring during ECT to prevent unnecessary re-stimulation when motor seizures monitoring incorrectly label the treatment as inadequate and also to detect and abort the prolonged seizures, which appears to be common in Indian setting. A study was conducted on 287 consecutive consenting patients receiving either BLECT or ULECT to explore the factors influencing ratio of motor and EEG seizure duration (ME ratio).[22] Patients were classified into low (≤0.8; n = 146) and high (>0.8, n = 141) ME-ratio groups. Significantly, more patients in the low ME-ratio group received BLECT, had prolonged EEG seizure, and were on concurrent lithium compared with the high ME-ratio group. This study underscores the importance for EEG monitoring at least inpatients receiving lithium and BLECT.

Recent international guidelines have recognized the limitations of seizure duration as a marker of adequacy of treatment and emphasize the pattern of EEG seizure.[16] A study was conducted to compare the old definition of seizure adequacy (based on seizure duration) with the new definition (a seizure with polyspikes and a 3-Hz activity) in 102 computerized EEG recordings.[23] Only 2% of the 58 “inadequate” seizures by the old definition were found to be adequate by the new definition. This suggested that seizures meeting newer criteria almost invariably last for at least 25 s; in the Indian context, the change in the criteria would make little change in the day-to-day practice of ECT.

Numerous ictal EEG markers have been elucidated to predict treatment response of ECT including postictal suppression, postictal coherence and amplitude, amount of slowing and time to onset of slowing, global EEG power, largest Lyapunov exponent, and strength symmetry index (SSI).[24]

Gangadhar et al. evaluated SSI in 12 patients receiving BLECT and equal number of patients receiving ULECT. SSI was computed from the early- and mid-seizure EEG epochs using the fractal dimension (FD).[25] The seizures of ULECT were characterized by significantly smaller SSI in both phases, despite having seizure duration comparable to BLECT. It was suggested that SSI may be a potential measure of seizure adequacy. In another study, SSI was computed in 40 patients receiving either high-dose or low-dose bitemporal ECT (BTECT).[26] SSI of the early seizure was higher in the high dose than in the low dose ECT group.

Another method of analyzing EEG is to measure the components of the EEG such as its frequencies across several nonoverlapping frequency bands: Delta, theta, alpha, and beta through fast Fourier transform.[24] The delta band of EEG is said to be relevant to the therapeutic properties of ECT. EEG of 25 patients receiving either BLECT or ULECT was subjected fast Fourier transform, and the spectral power of the delta (1–4 Hz) band was computed.[27] BLECT resulted in symmetrical seizure spectral power; ULECT seizures had significantly lower spectral power in the delta band on the unstimulated side.

Postictal suppression of EEG has been evaluated as predictor of treatment response.[24] Ictal EEG of 40 patients with melancholic depression posted for BLECT was evaluated. FD was computed for early-, mid-, and post-seizure phases of the first ECT. FD is a geometrical method of analysis of EEG, which was calculated based on the principles described by Katz.[28] It is a measure of complexity of the EEG wave in the terms of amplitude and frequency. Smaller postseizure FD after the 1st ECT was associated with remission status at 2 weeks. Smaller postictal FD was considered a marker for postictal suppression suggesting that postictal suppression may be a marker for early antidepressant response for ECT.[29] A similar study was conducted in 51 right-handed, drug-free patients with major depressive disorder receiving right ULECTs at 2.5 times their seizure threshold.[30] Spectral power (dB) for 2–6 Hz band and FD of postseizure EEG were estimated. Smaller FD and spectral power (i.e. greater postseizure EEG suppression) were seen in early responders compared to the late responders.

FD has also been studied as a measure of seizure intensity. Ictal EEG FD was calculated from 26 EEG records of patients receiving either BFECT or BTECT for schizophrenia. Maximum FD at the 2nd or 3rd ECT had an inverse correlation with psychosis severity after 6 ECTs as well as with the number of ECTs provided. Similarly, maximum FD had direct correlation with maximum heart rate. Hence, maximum FD during the early part of the ECT course in schizophrenia patients is predictive of greater benefit in short-term psychopathology.[31] In another analysis of data from the sample, it was seen that maximum FD does not change over six ECTs despite decline in seizure duration over ECT.[32] Thus, EEG morphology of BLECT induced seizures show little variation over the course of treatment which is in keeping with other studies.[33]

EEG has also been used to measure the depth of anesthesia. Bispectral index (BIS) measures the level of hypnosis (sedation) during anesthesia. BIS values of awaked individuals with schizophrenia were studied through a course of bitemporal brief-pulse ECT. It was recorded using a BIS monitor (A-2000, Aspect Medical Systems Inc., Newton, Massachusetts, USA), with sensor applied to the forehead on the right side. It was found that after six ECTs, 60% of patients had their resting BIS values in the anesthetized/sedated range even when they are awake. Thus, BIS may not provide accurate estimation of the depth of anesthesia during ECT after the initial ECT sessions.[34]

ELECTRODE PLACEMENT

BFECT is found to be as efficacious as BTECT for depression but associated with fewer cognitive adverse effects.[35] However, its efficacy in other indications had not been well studied and hence, it was planned in NIMHANS. Assessing BFECT in acute mania in double-blind randomized controlled trial showed faster decline in YMRS scores and earlier attainment of response compared to BTECT without differences in adverse effects.[36] Both groups obtained comparable seizure variables. A double-blind, randomized controlled trial demonstrated the superior clinical as well as cognitive effects of BFECT over BTECT in-patients with schizophrenia.[37] Seizure duration significantly reduced only with BFECT and not with BTECT over the course of ECT.[38] This provided the indirect support to possible better anticonvulsant mechanism of action of BFECT in schizophrenia. These studies used electrodes with inner concave surface specifically for BFECT and later, we have been using these electrodes in clinical practice for BFECT in NIMHANS.

ELECTRICAL STIMULUS PARAMETERS

The effect of electrical stimulus parameters of ECT on prolactin response was studied. Stimulus dose, frequency of ECT, and seizure variables such as seizure duration, seizure strength, pattern, and symmetry did not predict chance in prolactin level. Similarly, none of the clinical variables including baseline severity of illness, presence of psychotic symptoms, drug-naive status, and degree of clinical improvement predicted the prolactin response. The prolactin response however, was significantly more in female participants which may be due to their lower seizure threshold and inherent biological predisposition.[39]

A study comparing low- and high-stimulus doses in a double-blind, randomized controlled design revealed that EEG seizure was of comparable duration in the two stimulus groups.[26] Another study with a cross-over design looking into the effect of different stimulus frequency found that lower pulse frequency (50 PPS) stimulus had lower seizure threshold and produced fewer subconvulsive stimulations compared with high-frequency stimulus (200 PPS) without altering seizure durations or the cardiovascular responses.[40] The findings confirmed the available literature.[4142] The low-frequency stimulus produced similar clinical benefits as that of high frequency in subsequent study.[43] Thus, lowering down the frequency of pulses may be a viable option when seizure threshold is high due to various reasons. The successful use of this strategy was subsequently reported.[44] The lower stimulus frequency leads to longer stimulus train duration and hence, had lower rate (mC/s) of stimulus dissipation.

CARDIOVASCULAR SYSTEM MONITORING

The cardiovascular system (CVS) effects of ECT were also studied with the features of automated cardiac monitor and ambulatory ECG system in EEG machine. To study the effect of atropine on CVS response to ECT, 30 patients were randomly allocated before the third ECT to receive atropine or no premedication and found that atropine contributed to 32% of the variance in rate pressure product (RPP) with lower values in those who did not receive atropine.[45] However, withholding atropine did not affect the seizure duration and did not produce bradyarrhythmias. This study hence indicated that avoiding atropine as premedication can control RPP and thus can improve cardiac stability during ECT without affecting other aspects.

In a large sample of 124 patients, it was found that postictal RPP was significantly higher after BLECT than ULECT even after eliminating the effect of possible confounding variables such as subconvulsions and concurrent medications. The other important predictors were age and pre-ECT RPP. The difference of 10% postictal RPP increments between the right ULECT and BLECT groups found in this study is of clinical relevance in elderly or patients at risk of cardiac complications.[46] The postictal RPP may be considered as a peripheral marker of the degree of seizure generalization. In addition, if the hypothesis of seizure generalization to hypothalamus as a prerequisite for clinical efficacy[47] is given consideration, then lower RPP of threshold ULECT compared to BLECT can explain lower clinical efficacy of ULECT and may reflect the less than optimal hypothalamic stimulation. The correlation of RPP with clinical efficacy was further supported by the lower postictal RPP with unilateral threshold ECT compared to that with unilateral suprathreshold ECT.[48] However, atropine could mask the differential cardiovascular responses of the effective and ineffective seizures.[49]

In another study, the effect of anesthetic modification of motor seizure on RPP was also explored. 1 mg/kg dose of succinylcholine improved the number of patients with successful modification compared to 0.5 mg/kg but did not influence ictal or postictal RPP.[50] It thus ruled out the significant CVS alteration due to depolarizing effect of succinylcholine and 1 mg/kg dose may be used routinely without fear of exaggerated CVS response.

Besides these, ictal RPP was directly compared with EEG seizure and motor seizure. It confirmed the above findings that RPP response correlates with the cerebral seizure instead of motor component of seizure. Ictal RPP did not differ between groups with or without adequate motor seizure if EEG seizure was adequate. However, the group with inadequate EEG seizure had the least ictal RPP.[51] Hemodynamic response with BTECT and BFECT were examined at five-time points starting from before anesthesia induction to 2 min after the seizure.[52] The maximum values of systolic blood pressure, heart rate, and RPP were similar in both groups, suggesting that cardiovascular load is similar in BFECT and BTECT. It also indicates that differential frontal lobe stimulation rather than seizure generalization may be playing a role in better efficacy of BFECT over BTECT in schizophrenia.

Lithium given concurrently with ECT can have serious interactions with various aspects of the procedure. At the same time, not combining ECT and lithium in certain patients may have adverse clinical consequences. In the absence of systematically collected data on adverse effects of lithium co-prescribed with ECT, patients on lithium (n=27) and those not on lithium (n=28) were compared while they received ECT. Seizure variables and time to recovery from anesthesia were not affected by concurrent lithium. The sympathetic cardiovascular response was lower in those who were on lithium and the time to post-ECT recovery directly correlated with serum lithium level. Safety of lithium during ECT was satisfactory particularly when serum lithium was at a lower therapeutic range in young patients without significant cardiovascular risk factors.[53]

To conclude, the indigenously developed brief-pulse ECT-EEG device has been used to treat thousands of patients in the past two decades. Its extensive use in research has yielded results with important heuristic and clinical implications. These have been summarized in Box 1.

Box 1

Summary of the findings of research conducted utilizing the specific features of the indigenously developed brief-pulse ECT device*

Many more issues related to ECT need exploration in future; these include altering amplitude of electrical stimulus, ultra-brief ECT in indications other than depression, novel techniques of different electrode's sizes, and placements to reduce the cognitive adverse effects without compromising the clinical efficacy.

Financial support and sponsorship

Nil.

Conflicts of interest

The last author, Mr. V.S. Candade is the Director of the company, Niviqure Meditech Pvt. Ltd., Bengaluru, Karnataka, India, which manufactures the brief-pulse ECT device mentioned in this review article.