Authors: Tianqi Shen, Xingxing Yin, Xiaomeng Yu, Haibo Song, Menglu Li, Xiangjie Li, Shan Qin
Categories: Systematic Review, Electroconvulsive therapy, Hyperventilation, Hypocapnia, Seizures, Meta-analysis
Source: BMC Psychiatry
Authors: Tianqi Shen, Xingxing Yin, Xiaomeng Yu, Haibo Song, Menglu Li, Xiangjie Li, Shan Qin
Electroconvulsive therapy (ECT) is a treatment used for severe psychiatric conditions that involves the induction of seizures under general anesthesia. It has been proposed that hyperventilation, which is the rapid breathing either voluntary or induced to lower carbon dioxide levels, might have an impact on the effectiveness of ECT. This systematic review and meta-analysis aimed to consolidate the current evidence on how hyperventilation may alter the course and outcomes of ECT by potentially affecting seizure duration and the resultant therapeutic outcomes, and assess its implications for both safety and potential therapeutic outcomes for patients.
We conducted a systematic literature search of articles published up to April 18, 2024, in databases including PubMed, the Cochrane Library, Embase, Web of Science Core Collection, Ovid, Wanfang Data, the China National Knowledge Infrastructure, and the SINOMED database. Revised Cochrane Risk of Bias Tool for Randomized Trials (RoB 2.0) was used to evaluate the quality of these studies. Statistical analysis was conducted using RStudio version 4.3.2, with the Egger’s test for publication bias and sensitivity analysis for result robustness.
Seven studies with 620 ECT sessions were analyzed. Hyperventilation significantly extended muscle (MD 2.880 s, 95% CI 0.917–4.842 s, GRADE moderate) and EEG (MD 5.426 s, 95% CI 0.656–10.195 s, GRADE moderate) seizure durations. No significant differences in safety indices were observed between groups. Additionally, the overall bias across the included studies was significant, and the majority of studies relied solely on seizure duration as an indicator of ECT efficacy.
Hyperventilation during ECT induction phase may improve seizure outcomes and is well-tolerated, though further research is needed to confirm these preliminary findings.
The online version contains supplementary material available at 10.1186/s12888-025-07105-7.
Electroconvulsive Therapy (ECT) is a medical procedure performed under general anesthesia to treat various psychiatric conditions [1]. For over eight decades, ECT has remained one of the most effective treatments for severe depression, mania, and clozapine-resistant schizophrenia [2, 3]. It has also demonstrated efficacy in reducing suicidal ideation and may enhance the therapeutic effects of clozapine in certain psychiatric disorders [4]. During the procedure, electrical currents are delivered through the scalp to stimulate the brain, intentionally inducing a controlled seizure. This seizure activity is believed to modulate brain chemistry, neural activity, and synaptic connectivity, thereby producing therapeutic benefits.
Patient response to ECT may be associated with the duration of induced seizures, although this relationship is debated [5]. Despite this uncertainty, clinicians have conducted numerous studies aiming to explore methods that might prolong seizure duration without increasing the intensity of electrical stimulation [6]. These investigations have predominantly focused on optimizing anesthetic protocols, encompassing modifications to sedative agents (notably etomidate and ketamine) [7, 8], adjustments in the anesthetic-ECT time [9], and alterations in ventilation modalities [10], with the duration of seizure being used as a metric for efficacy, although the validity of this metric is debated. Among the strategies explored, hyperventilation during the induction of anesthesia has been proposed to potentially prolong seizure duration [11]. The role of hyperventilation has been largely described in several previous works. However, variation in the methodology of studies which makes it difficult to draw firm conclusions. The differences are mainly due to the different procedures of this technique and evaluated outcomes as explained in previous reviews. The objective of this review is to explore the impact of seizure duration and safety of concurrent hyperventilation with ECT.
The protocol of this systematic review followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 statement [12] and was registered at PROSPERO (ID: CRD42024533151; Date: April 18, 2024).
The research team independently searched multiple academic databases, such as PubMed, the Cochrane Library, Embase, Web of Science Core Collection, Ovid, Wanfang Data, China National Knowledge Infrastructure (CNKI), and SINOMED, up to April 18, 2024. We used search terms including “hyperventilation” and “electroconvulsive therapy,” along with their synonyms, including Medical Subject Headings (MeSH) and keywords. To ensure thoroughness, we also manually checked the reference lists of the studies we found for any additional relevant documents that might have been missed in the initial search. The complete search strategy is outlined in Supplement Table 1.
We have included all randomized controlled trials that compare hyperventilation with normal ventilation in our review. Hyperventilation is defined as a breathing rate or tidal volume that is higher than what is used during standard anesthesia induction.
The inclusion criteria for this study were as (1) the study population comprised psychiatric patients who were undergoing ECT, without limitations based on the type of psychiatric disorder; (2) participants were aged 18 years or older, without restrictions regarding nationality or ethnicity; (3) the experimental group was subjected to hyperventilation during the induction of anesthesia; (4) the study design was that of a randomized controlled trial; (5) no restrictions on the language or geographical area of the studies included in the review.
Exclusion criteria for the review (1) studies for which the full text was not accessible; (2) studies lacking outcome measures that included both efficacy and safety metrics; (3) studies with an incomplete presentation of data.
The process of filtering the literature was carried out using Endnote 20 software. Each author independently reviewed the titles and abstracts of the identified studies, eliminating those that were clearly irrelevant based on the predetermined inclusion and exclusion criteria. Following this, the authors conducted a detailed review of the remaining studies to determine the final selection for inclusion in our analysis. The literature screening and selection process was conducted independently by Tianqi Shen and Xingxing Yin, with any disputes resolved by Shan Qin.
During the data extraction phase, authors independently extracted information from the eligible studies. The details systematically extracted the lead researcher’s name, the year of publication, demographic and clinical characteristics of the patient group (such as age, gender, and psychiatric diagnosis), the specific anesthetic agent used during the induction phase, and the goals of the hyperventilation intervention. For studies with multiple experimental groups, data from each arm were pooled for analysis. WebPlotDigitizer v4.6 [13] was utilized to extract data from graphical representations present in the literature.
The primary outcome measure was set as the duration of seizure occurrence following ECT, with the observation method differentiated by muscle contractions (detected by oculomyography of the upper limbs or other body parts) or by EEG identified seizure duration.
Secondary outcome measures include other indicators used to assess the therapeutic effect, efficacy, and safety of ECT.
Each author was tasked with independently assessing the methodological rigor of the included studies. The Revised Cochrane Risk of Bias Tool for Randomized Trials (RoB 2.0) [14] was deployed to detect potential biases, encompassing those associated with the randomization process, deviations from the prescribed intervention, incomplete outcome data, biases in outcome measurement, and overall outcome bias. For studies featuring crossover designs, the assessment also considered the potential bias introduced by period and carryover effects during the randomization process [15]. In the event of discrepancies in the individual evaluations, the principal investigator, or corresponding author, would deliver the definitive verdict.
The statistical analyses were performed utilizing R Studio version 4.3.2. Continuous outcomes are presented as mean differences (MD) alongside their respective 95% confidence intervals (CI). For those studies that provided means alongside their 95% CIs, we engaged in data adjustment calculations employing methodologies outlined in the Cochrane Handbook [16]. The defects pertaining to merging were quantified using an inverse variance model. Heterogeneity among the included studies was appraised. An I^2^ statistic exceeding 50% signified considerable heterogeneity, which was managed by employing a random-effects model. We conducted subgroup analyses to examine the study characteristics that might contribute to heterogeneity in the research. Additionally, sensitivity analyses were conducted to ascertain the robustness of the aggregate findings. Publication bias was assessed using funnel plots and Egger’s test.
A comprehensive evaluation of the evidence was conducted in strict adherence to the Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) guidelines [17, 18]. The GRADE approach dictates that the evidence be classified according to the risk of bias, inconsistency, indirectness, and imprecision. The GRADEpro GDT online tool was utilized for assessing the certainty of the evidence and for generating summary tables of the findings.
A thorough search across nine databases yielded a total of 1,983 citations. After deduplication was performed using EndNote, the number of unique citations was condensed to 1,713. Subsequently, 20 studies were identified based on their titles and abstracts and were retrieved in full text for further analysis. The selection process culminated in the inclusion of seven studies [19–25], which encompassed data from 185 patients across 620 ECT sessions. Among these, four studies concentrated on patients diagnosed with depression, while five studies utilized a specific expiratory carbon dioxide (ETCO2) target value as a metric for hyperventilation. A detailed depiction of the literature screening process is presented in Fig. 1, and general information of the studies included in the review was presented in Table 1.
Fig. 1Literature screening process
Table 1General information of included studiesStudyCountryDesignSex(M/F)Age(years)No. of patientsNo. of sessionsDignosesInductionType of muscarinicOpioid agonistHyperventilationRegular ventilationStimulus methodsChater1988UKCrossover8/2231–913090major depresionmethohexitonesuxamethoniumwithout20 breaths/10 breathsno breathbilateralPande1990USAParallel3/1261.03 ± 14.261575depressionmethohexitalsuccinylcholinewithout20/25mmHg30mmHgright unilateralSawayama2008JapanParallel9/1052.84 ± 15.6419114schizophrenia/depressionpropofolsuxamethoniumwithout30mmHg40mmHgbilateralMayur2010AustraliaParallel-40.00 ± 8.212575major depressionthiopentonesuxamethoniumwithout20 maximum capacity breathsnot more than 3 positive pressure ventilationright unilateralChoukalas2010USACrossover--1836major depressionpropofolsuccinylcholineremifentanil18mmHg25mmHg-Nishikawa2017JapanParallel34/2459.15 ± 9.1238190schizophrenia/depressionpropofolsuxamethoniumremifentanil30-35mmHg40-45mmHgbilateralGundogdu2020TurkeyParallel-38.00 ± 9.764040-propofolrocuroniumwithout25-30mmHg35-40mmHgbilateral
We employed Rob 2.0 to appraise potential biases stemming from the randomization process, deviations from the intended interventions, issues related to missing outcome data, inaccuracies in the measurement of outcomes, and selective reporting of results. The findings are articulated in Fig. 2, which reveals that one study [20] was characterized as having a high risk of bias attributable to an imbalanced randomization process. Additionally, concerns of bias were raised regarding the potential deviation from the prescribed intervention in one study [22]. Furthermore, two studies [19, 23] were flagged for high risk of bias due to the influence of period effects and carryover effects inherent in their crossover design [26]. Fig. 2.
Fig. 2Summary of bias assessment
A total of six studies [19–21, 23–25] were identified, which reported on the duration of motor seizure subsequent to hyperventilation. These studies encompassed 545 sessions across 160 patients. The aggregated results indicated that hyperventilation significantly elongated the duration of motor seizure (MD 2.880 s; 95% CI: 0.917 s to 4.842 s; p = 0.004), with the analysis exhibiting a low degree of heterogeneity (I^2^ = 0%; p = 0.43). Fig. 3-A.
Fig. 3Forest plot of seizure duration. (A) Muscle seizure duration. (B) Electroencephalogram seizure duration
Three studies [21, 22, 24] reported on the duration of electroencephalogram (EEG) seizures following hyperventilation, included a total of 379 electroconvulsive stimulus across 82 patients. The findings indicated that hyperventilation significantly extended the duration of EEG seizures (MD 5.426 s; 95% CI: 0.656 s to 10.195 s; p = 0.026). However, the results also revealed a moderate degree of heterogeneity among the studies (I^2^ = 56%; p = 0.18). Fig. 3-B.
Some of studies in our analysis assessing ECT dosages indicated a protocol-driven increase in electrical stimulation requirements over multiple sessions for all participants [22, 24]. This protocol-driven dosage escalation aimed to ensure the therapeutic benefits for all patients in both groups during ECT. Nishikawa et al.‘s study compared the required increments in stimulation to maintain efficacy post-ECT between the two groups and found that the hyperventilation group demonstrated a significantly smaller increase compared to the conventional ventilation group, with this difference achieving statistical significance (p = 0.010) [24].
Three studies have examined the safety of administering ECT under conditions of hyperventilation. The findings indicate that both the hyperventilated and normoventilated groups exhibited similar respiratory function parameters, including respiratory rate (respiration per minute): 31 versus 32, and inspiratory/expiratory 0.6 versus 1 [23]. Additionally, haemodynamic indices were comparable between the two groups, with blood pressures recorded as 96.25 ± 15.11 mmHg versus 95.40 ± 9.70 mmHg (p = 0.83), and heart rates as 88.40 ± 13.92 bpm versus 88.75 ± 12.15 bpm (p = 0.93). Tissue oxygenation indices also showed no significant difference, with peripheral oxygen saturation at 99.45 ± 0.99% versus 99.55 ± 0.75% (p = 0.72), and regional cerebral oxygenation saturation at 80.90 ± 9.26% versus 76.45 ± 8.61% (p = 0.13) following the seizure [25]. Furthermore, the hyperventilated group displayed a more rapid recovery rate, with a mean recovery time of 9.7 ± 2.02 min versus 12.3 ± 2.27 min in the regular ventilated group (p = 0.001). Mayur’s study [22] revealed a clinically meaningful 34% prolongation in orientation time among non-hyperventilated patients compared to the hyperventilation group (time ratio 1.34, 95% CI 0.94–1.92; P = 0.103), but the difference did not reach statistical significance at conventional thresholds.
A subgroup analysis was conducted to assess the impact of anesthetic-inducing drugs, the type of muscle relaxant agents, the inclusion or exclusion of opioid medications, study design, and bias assessment outcomes on muscle seizure duration. As depicted in Fig. 4, the analysis revealed that none of these factors had a significant mediating effect on muscle seizure duration. However, it is noteworthy that some factors exhibited a considerable degree of within-subgroup heterogeneity.
Fig. 4Forest plot of subgroup analysis
A sensitivity analysis for muscle seizure duration was performed using a one-by-one exclusion approach. The results consistently indicated a statistically significant advantage for the hyperventilation group, accompanied by low heterogeneity (Supplemental Fig. 1). Nevertheless, due to the scarcity of studies available for EEG seizure duration, it was not viable to conduct a sensitivity analysis for this parameter.
Due to the limited number of included RCTs (n < 10), analysis of publication bias, as per standard guidelines, was not feasible [27].
The GRADEpro online tool was utilized for the application of the GRADE system. Both outcomes, muscle and EEG seizure durations, were downgraded to account for the bias ratings associated with the included studies. In the final evaluation, the quality of evidence for muscle seizure duration and EEG seizure duration were rated as moderate. (Table 2)
Table 2Summary of findings. CI: confidence interval; SMD: standardised mean difference; a. Some of the studies were rated as medium or high risk of biasCertainty assessmentNo. of patientsEffectCertainty№ of studiesStudy designRisk of biasInconsistencyIndirectnessImprecisionOther considerationsHyperventilationRegularRelative(95% CI)Absolute(95% CI) Muscle seizure duration 6randomised trialsserious^a^not seriousnot seriousnot seriousnone297248-MD 2.880 longer(0.917 longer to 4.842 longer)⨁⨁⨁〇Moderate EEG seizure duration 3randomised trialsserious^a^not seriousnot seriousnot seriousnone191188-MD 5.426 longer(0.656 longer to 10.195 longer)⨁⨁⨁〇Moderate
Although ECT is recognized as an excellent treatment for some disorders [28, 29], there is an observed escalation in the electrical dose required for individual sessions as the total number of ECT stimulus increases [30]. This increment in electrical dose is accompanied by a heightened risk of side effects, notably including cognitive impairment [31]. The pursuit of strategies to augment seizure duration by modulating anesthetic agents or ventilation techniques, thereby aiming to lower the seizure threshold, has been a focal point for anaesthesiologists [32, 33]. Since initial case report in 1980 [34], the role of hyperventilation—an intervention noted for its simplicity and ease of implementation—has garnered significant interest within the context of ECT. In response to the burgeoning evidence that underscores the efficacy of the laryngeal mask in facilitating precise EtCO2 measurements and enhancing hyperventilation [35], there has been a notable increase in the volume of research conducted in recent years [25, 36].
The present study has demonstrated that hyperventilation can significantly extend the duration of EEG seizures by a mean of 5.426 s (95% CI: 0.656 s to 10.195 s; p = 0.026), and muscle seizures by a mean of 2.880 s (95% CI: 0.917 s to 4.842 s; p = 0.004). There is an emerging consensus that hyperventilation may be associated with increased neuronal supersynchronous peak activity [37]. It is hypothesized that the decrease in the partial pressure of carbon dioxide within the bloodstream, consequent to hyperventilation, may result in a diminished cerebral blood flow velocity and induce a transient state of respiratory alkalosis. Such physiological changes are posited to potentially destabilize the neuronal tissue membranes, rendering neurons more amenable to excitation and more likely to fire [38, 39]. This mechanistic insight may underpin the observed phenomenon that ECT, when preceded by hyperventilation, can elicit more prolonged seizure episodes. Consonant with these findings, analogous results have been documented in other observational studies [10, 40].
The control of EtCO2 varied across the studies included, with the hyperventilation group’s EtCO2 ranging from 18 mmHg to 35 mmHg. EtCO2 is not only related to the quality of seizures but also closely associated with adverse events during and after ECT. The duration and severity of intraoperative hypocapnia are intimately linked to the severity of delirium [41], and sustained EtCO2 below 25 mmHg for more than 5 min are correlated with the occurrence of delirium within 7 days [42]. Additionally, studies have indicated that excessively low EtCO2 levels (EtCO2 < 28 mmHg) during surgery are associated with postoperative pulmonary complications within 30 days [43]. This review includes a limited number of studies on the safety indicators following hyperventilation and lacks research on complications associated with stimulus. The safety of ECT under conditions of hyperventilation is also a direction for future research.
During our screening process, the study by Taylor and colleagues [9] was not included in this meta-analysis due to the absence of specific numerical data on seizure duration, a critical parameter for our analysis. However, their findings underscore the significance of the Anesthetic-ECT time interval, which was found to markedly influence the quality of seizures induced by ECT. This interval may be a key determinant of anaesthetic concentration at the time of stimulation, thereby affecting the therapeutic efficacy of ECT. The study highlights the clinical importance of monitoring and potentially optimizing the Anesthetic-ECT time interval to enhance seizure quality and, by extension, the overall effectiveness of ECT stimulus.
In the studies included in this research, there is a protocol-driven increase in the absolute electrical dose during ECT treatment in some cases, yet such an increase does not necessarily correlate with individual patient needs or therapeutic outcomes. The relationship between the dose and the seizure threshold is a more significant determinant of cognitive side effects than the absolute dose itself [44]. We must acknowledge the variability in patient responses and the importance of personalized treatment protocols [45, 46]. The seizure threshold plays a crucial role in determining the appropriate dosage, seeking to balance efficacy and the minimization of side effects [31, 47].
The results of this review suggest that hyperventilation may modulate the electrical dosage necessary for ECT. However, it is important to note that this observation does not necessarily imply a direct correlation with therapeutic efficacy. Additionally, no differences in seizure induction thresholds or postictal suppression were observed between the two ventilation groups [22], suggesting that dosage adjustments, while protocol-mandated, may not always align with individual patient responses. The relationship between side effects and dosage is nuanced, being more dependent on the dose in relation to the seizure threshold rather than the absolute dose. The primary objective of dosage adjustments is to maintain this relationship, thereby balancing treatment efficacy and minimizing side effects.
In addition to the muscle tremors caused by depolarizing neuromuscular blocking agents used during anesthetic induction, which increase tissue oxygen consumption [36], the seizure induced by ECT is associated with an increase in cerebral oxygen consumption [48]. Despite the administration of pure oxygen prior to anesthesia, there remains a potential for hypoxia to occur following the ECT procedure [49, 50]. It is commonly believed that changes in PaCO2 can alter the diameter of cerebral small arteries, thereby determining the nutritional microcirculatory blood flow through the capillary bed, which reduces the oxygen supply to the brain. However, the large cerebral arteries are generally considered to be non-compliant [51]. A study [52] observed that regional cerebral oxygen saturation (rSO2) increased during hyperventilation and decreased sharply immediately after stimulation. Gundogdu et al. have shown that, compared to conventional ventilation, hyperventilation can lead to an increase in rSO2 [25]. However, Veraar et al. observed patients undergoing elective cardiac surgery at different EtCO2 levels and found that rSO2 decreased during periods of hypocapnia, while it correspondingly increased during periods of hypercapnia [53]. In contrast, Brandi et al.‘s study indicated that short-term moderate hyperventilation had no effect on cerebral oxygen supply [54]. Furthermore, in addition to rSO2, the Oxygen Reserve Index (ORi) has also been identified as a sensitive indicator for the detection of hypoxic states following ECT [55, 56].
During ECT, patients are typically preoxygenated in the supine position using a face mask, and manual mask ventilation is maintained post-ECT. However, this approach has significant limitations in obese patients or those with potential difficult airways [40, 57]. With a better understanding of the factors that affect efficacy and cognitive side effects, numerous studies have focused on improving ventilation methods or techniques during anesthesia induction [32, 58, 59]., including the use of laryngeal masks or the hook technique in ECT [58]. Some scholars strongly recommend the use of supraglottic airway devices in these patients to ensure safety [36]. Although supraglottic devices can better monitor EtCO2, the studies included in this review used anesthetic masks for ventilation, which may affect the measurement of EtCO2 and increase experimental error.
In this meta-analysis, the GRADE method rated the quality of evidence for both muscle seizure duration and EEG seizure duration as moderate. The high overall risk of bias across the included studies may have compromised the reliability of the results. Although we validated the robustness of the findings through sensitivity analysis, the presence of bias still limited the certainty of the evidence. Additionally, the presence of moderate heterogeneity in some of the results (I² = 56%) further impacted the precision of the evidence.
This study has its limitations. The validity of seizure duration as a metric for assessing the efficacy of ECT is currently a subject of scrutiny [11]. Nonetheless, the majority of studies encompassed in this analysis have utilized seizure duration as the principal indicator for gauging the effectiveness of ECT. A significant correlation has been established between postictal suppression index and the therapeutic efficacy of ECT [60]. In this review, only one study compared postictal suppression between the hyperventilation and regular ventilation groups, and no significant difference was observed between them. Furthermore, there is a lack of evidence regarding whether hyperventilation actually improves clinical outcomes. The studies included in this review do not provide sufficient data to determine if the observed changes in seizure duration under hyperventilation translate into better patient outcomes. Additionally, there is insufficient evidence to support any sustained effect of hyperventilation on seizure duration or morphology over a course of treatment. The majority of the studies focused on the immediate effects of hyperventilation during a single ECT session, and long-term follow-up data are scarce.
To the best of our knowledge, this review represents the first meta-analysis to examine the impact of hyperventilation on ECT administered during the induction phase of ECT. It offers a moderate level of evidence, albeit with several limitations. Firstly, the overall bias across the included studies was significant, attributable to the extensive time span covered by the studies. Secondly, the majority of studies relied solely on seizure duration as an indicator of ECT efficacy, utilizing a single outcome measure. This approach limited the ability to conduct a comprehensive analysis of the effects of hyperventilation. Further research with larger sample sizes and more rigorously designed randomized controlled trials is necessary to confirm the safety and efficacy of hyperventilation in ECT.
This systematic review and meta-analysis suggests that hyperventilation during the induction phase of ECT can not only improve seizure duration but also appears to be well-tolerated by patients. However, further research is required to validate its impact on clinical outcomes and to explore its long-term clinical benefits, in order to enhance treatment efficacy and patient safety.
Below is the link to the electronic supplementary material.
Supplementary Material 1
Supplementary Material 2
Supplementary Material 3
Supplementary Material 4