Supraventricular tachycardia (SVT) or paroxysmal SVT is a cardiogenic emergency. Annually, its incidence is ~1.8 per 10,000 visits to the emergency department (ED) in the United States of America (1). SVT may lead to palpitations, chest pain, or dyspnea. SVT may further contribute to hemodynamic instability, syncope, and even sudden cardiac death (incidence: 4%) (2, 3). In hemodynamically stable patients, pharmacological conversion is generally used most for its efficacy. However, potential severe adverse events, such as transient asystole or hypotension, limit the range for treatment (2, 4, 5). Therefore, vagal maneuvers (VagMs) are still highly recommended for acute treatment in patients with regular SVT in current guidelines and the newest edition of advanced life support (6–8).

Vagal maneuvers are techniques aimed at increasing vagal parasympathetic tone and blocking the atrioventricular (AV) node. Carotid sinus massage (CSM) is a traditional maneuver performed by giving a firm pressure to the carotid sinus in the upward and downward directions then posteriorly and medially between the examiner's fingers and the patient's cervical vertebra for 5–10 s (9, 10). However, the potential risk of thromboembolic events should be a concern when deploying CSM in older patients. The standard Valsalva maneuver (SVM) may be currently the most used VagM because of safety and easy performance. The steps to perform SVM require patients to maintain a sitting position whilst keeping an expiratory pressure of around 30–40 mmHg by blowing into a syringe for 15 s. Regretfully, the conversion rate is generally < 20% because of inadequate blow time and pressure by persons (9, 11–14). The modified Valsalva maneuver (MVM) is thereby developed by placing the patient into a supine position with the leg raised promptly for 45 s following the steps of SVM (5). The diving reflex triggered by breath holding and cold water immersion may also stimulate vagal nuclei in the brain to ameliorate AV nodal conduction (15).

Several studies attempted to evaluate the efficacy and adverse events amongst different VagMs (5, 9, 10, 12, 16, 17). These studies seemly suggested MVM as the optimal VagM for SVT. However, small number of cases were included in these reports. On the basis of the principle of evidence-based appraisal, the comparison of effectiveness and adverse events amongst all VagMs remains unclear. Thus, we have performed a systemic review and a network meta-analysis (NMA) to investigate all VagMs in adult patients with SVT, identify the effectiveness of treatment, and report the potential adverse events.

We performed a systematic review and NMA in accordance with the latest statement of the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA 2020) (18). We registered the systematic review protocol on the international website INPLASY with the registration number INPLASY2020110082 (DOI: 10.37766/inplasy2020.11.0082) and updated and recorded changes in protocol on June 20, 2021.

Clinical physicians applied MVM, SVM, and CSM for the conversion of SVT to sinus rhythm, but the most effective VagM had not been well-discussed. The preference of MVM was recommended in the newest edition of adult advanced life support in European Resuscitation Council Guidelines (8). However, rigorous appraisal was not illustrated in detail. Ceylan et al. compared the three VagMs and found that MVM was superior to SVM and CSM in terminating SVT (10). The small number of cases in each group of VagM was the limitation of the present study. The systemic review with a meta-analysis of Abdulhamid et al. demonstrated that the sinus rhythm was achieved 2.5 times more by MVM compared with that by SVM (38). However, CSM was not included in that systematic review, and CoE was not rated. The benefit of NMA was that it could compare the results related to multiple interventions and provide mixed estimates in combination with direct and indirect information. Our presentation is the first study that used NMA to compare the pooled effect of different VagMs reported by previous RCTs. High-certainty evidence of pooling direct and indirect comparison presented that MVM was far superior to other VagMs in the successful conversion of SVT to sinus rhythm regardless of initial response and the end of studies in this closed-loop NMA. Although the pooled estimates of MVM vs. SVM were contributed from only half of the low RoB data, only small portion of high RoB data were correspondingly weighted in the results. Again, most of the RCTs appraised as some-concern overall RoB were due to the insufficient information in the domain of allocation, and we considered such condition may not elicit serious threaten in the CoE of the results. The online application CINeMA, which we used to demonstrate the contribution matrix and the following RoB bar charts, can also access the confidence of NMA designed by six domains of consideration: (i) within-study bias, (ii) reporting bias, (iii) indirectness, (iv) imprecision, (v) heterogeneity, and (vi) incoherence (21, 22). We favored to use the Cochrane RoB 2.0 tool and the CoE in GRADE because both methodologies have been widely recognized and applied in the field of evidence-based medicine for many years. Both provide detailed protocols based on comprehensive and rigorous considerations, and we believe the answers are objective if following the rules.

Vagal maneuvers were mostly protocolised through the use of manometer to ensure forced expiration reaching the standard pressure. Also, the response was studied at similar timepoints. Therefore, we assumed few heterogeneities and potential confounders amongst the included studies. In our NMA, with the comparison of MVM and CSM at the end of the study, the range of direct estimates was wider than that of indirect estimates in the frequentist approach with the random-effects model (Figure 4). This finding explained the importance of indirect comparison and the benefit of NMA. The GRADE Working Group recommended that the sensitivity analysis by using variant statistical methodologies, including Bayesian and frequentist frameworks, should be performed in sparse networks (23). The reason was that the marked widening ranges of the CIs/CrIs in indirect estimates had been observed in some NMA enrolling sparse studies compared with those in direct estimates (23). Such combining direct and indirect estimates results in marked widening of CIs/CrIs than direct estimates in some NMAs (23, 39, 40), and the benefit of NMA to increased precision (narrower CIs/CrIs compared with relying on direct estimates alone) is lost. The abovementioned condition may result from insufficient data in sparse networks to reliably estimate different variances across the network. Therefore, the GRADE Working Group recommended that the authors conducting an NMA of a sparse network should plan sensitivity analyses that lead to more trustworthy estimates of effects based on the optimal choices in statistical models (23). Failing to do so might result in network estimates with low CoE, making results less useful for patients, healthcare workers, and the guideline makers.

We paid more attention to be aware of the potential of incorrect analysis, especially when the direct comparison of MVM vs. CSM was based on only one RCT in our study. The uncertainty about the value of the between-study heterogeneity parameters in the random-effects model in the Bayesian framework might result in less precise estimates of treatment effects than that from frequentist frameworks in sparse NMA (41). The pooled estimates in our NMA worked in concert with this observation. Therefore, we chose the frequentist framework with the random-effects model as the primary statistical approach. As shown in Supplementary Table 2, consistent results with similar ranges of CIs/CrIs amongst all VagM comparisons were observed in the outcomes of converting to sinus rhythm at the initial response and the end of study with various statistical methodologies. Therefore, the judgement of decision might not change when choosing the pooled estimates from different statistics. Regarding the pooled estimates of adverse events, extremely wide ranges of 95% CrIs were observed in the Bayesian approach. This finding might result from some zero event in the study arm. To handle the event of zero case in the frequentist approach, an arbitrary number of 0.5 was added to the event by the software. Such a correction resulted in narrow ranges of 95% CIs than that of 95% CrIs. Based on the CoE in GRADE, more RCTs are warranted to determine the evidence of adverse events amongst VagMs.

Our NMA demonstrated a conspicuous conversion rate of MVM compared with that of SVM. A reasonable explanation was that by adding a passive leg raise or an abrupt change from the semirecumbent position to the supine position after SVM, the modified version could increase venous return and maximize the vagal tone, leading to a decrease in heart rate by baroreflex and suppression of the AV node (17, 42, 43). This effect was instant and could sustain and maintain the successful conversion to sinus rhythm according to our result. In previous studies, the successful conversion rate from SVT to the sinus rhythm of MVM could reach ~50% (5, 16, 17). Nearly half of the patients with SVT could be successfully treated through a non-pharmacologic, free-of-charge, and simple maneuver. Although previous studies failed to show any reduction in the time of hospital stay or need of admission (5, 17), refraining from pharmacotherapy was more cost-effective and could prevent from potential severe adverse events of medications. Our result showed a potentially increased risk of adverse events in MVM compared with that in SVM. Most of the adverse events were acceptable and self-limited after the cessation of the maneuver. Thus, the adverse events might not be critical issues in performing MVM. Although no significantly increased adverse event was noted in the CSM group, some life-threatening complications, such as transient ischemic attack and stroke, had been put forward previously, especially in elderly patients (37, 44–48).

A simple handheld Valsalva assist device (VAD) had been developed to standardize the protocol of VagM reliably (49). In a recent RCT in pre-hospital settings, a higher conversion rate by VagM with VAD than usual strain method was observed (50). However, MVM was not routinely performed in both groups. Thus, the effectiveness of VAD should be further investigated. Aside from VagMs in this NMA, many ways could achieve an increasing vagal tone. These ways, which were not compared in the present study, included provoking a human diving reflex by ice packing, water immersion, or breath holding (51, 52). These procedures were seldom in clinical practice, and only some case series were reported especially for children (53). The use of a pneumatic antishock garment could also increase venous return and vagal tone (12). A new VagM, named reverse Valsalva maneuver, had been reported this year (54). RCTs are still the gold standard in validating the evidence of these VagMs. Literature regarding CSM is limited in this NMA. SVT induced by electrical stimulation involving CSM management was not included (55). Despite these limitations, MVM is the best VagM for patients with SVT due to its effectiveness with high CoE appraised by the modern methodology of GRADE.

Based on the results of NMA and CoE appraisal by GRADE, MVM is the most optimal VagM due to its high conversion rate with no significant increase in adverse events. Therefore, we recommend the use of MVM, which is a reasonable, considerable, and practicable approach clinically and can be proceeded before the pharmacologic treatment of SVT.

The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author/s.

EH and C-HC: database search, data extraction, critical analysis, interpretation of data, and drafting of the manuscript. C-YF: database search, data extraction, and drafting of the manuscript. C-WS: interpretation of data and revising the manuscript. PL and YH: double confirming of enrolled studies and data, concept of the network meta-analysis, statistical analyses, grading of risk of bias, and drafting of the manuscript. All authors contributed to the article and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. The reviewer Y-tK declared a shared affiliation, with the authors, EH, C-HC, C-YF, C-WS, to the handling editor at the time of the review.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.