Article Text
PDF
Abstract
Background The Valsalva manoeuvre (VM) is used to treat supraventricular tachycardia (SVT) by inducing a vagal response (drop in HR). There is debate as to the best position in which to carry out the VM and how the strain should be delivered in practice. We aimed to compare vagal responses induced with supine and modified VMs using strains delivered with a standardised manometer or novel Valsalva Assist Device (VAD), a simple device to provide resistance to exhalation.
Methods We conducted a repeated measures randomised trial of four VMs (two supine VM and two modified VMs), in healthy adult volunteers, with strains delivered using an adapted sphygmomanometer (manometer) or a VAD. Changes in HR, pressure and duration of strain and adverse events were monitored and compared between the techniques and devices. The trial was approved by the University of Exeter Medical School Research ethics committee.
Results 75 healthy participants aged 19–55 years were recruited over a 4-month period. A mixed-effects linear regression showed the modified VM resulted in a 3.8 beats per min (bpm) greater drop in HR compared with the supine VM (p=0.002, 95% CI 2.2 to 5.4). VM strains produced by the VAD were of a similar pressure but of slightly shorter duration and resulted in a 1.9 bpm smaller drop in HR compared with the manometer (p=0.01, 95% CI 0.4 to 3.4). There were no differences in adverse events.
Conclusions Modified VM was associated with a greater drop in HR than a supine VM with no increase in adverse events in healthy volunteers. The VAD can be used to safely generate the recommended VM strain pressure, but produced a smaller drop in HR compared with a manometer and requires modification to enable the recommended strain duration to be achieved consistently.
Trial registration number NCT03298880.
- acute care
- cardiac care
- arrhythmia
Statistics from Altmetric.com
Key messages
What is already known on this subject
The Valsalva manoeuvre (VM) is an internationally recommended first-line treatment for SVT.
Recent evidence suggests a modified manoeuvre is much more effective than the semi-recumbent manoeuvre, however, it has not previously been compared to a fully supine VM.
Current practical methods of generating the VM strain, such as blowing on a syringe, are unreliable in generating the recommended strain pressures. A purpose designed Valsalva Assist Device (VAD) has been developed but not yet tested.
What this study adds
The modified manoeuvre was shown to be physiologically superior to the purely supine manoeuvre.
The VAD can be used to generate recommended VM strain pressures , similar to strains delivered with a manometer but modifications to the VAD are required to allow adequate duration of strain and testing in patients with SVT .
Introduction
The Valsalva manoeuvre (VM) is an internationally recommended first-line treatment for haemodynamically stable supraventricular tachycardia (SVT).1 The VM involves an exhalation strain against resistance which causes a reflex slowing of HR mediated by the vagus nerve. The highest VM cardioversion rates in emergency medicine practice to date have been achieved in a study using a postural modification and a manometer delivered strain compared with semi-recumbent controls. In this study, intervention patients with SVT undertook Valsalva strains using an adapted sphygmomanometer, blowing for 15 s at 40 mm Hg in a semi-recumbent position before immediate supine repositioning with passive leg raise at the end of the strain.2
However, there is debate whether this modification has any advantage over a purely supine VM which is associated with a greater vagal response in healthy volunteers compared with sitting and Trendelenberg position VMs3 and may reduce the risk of adverse events.4 Adapted manual sphygmomanometers are also not routinely available for VMs carried out in normal practice and while surrogates such as blowing on empty syringes are commonly used, they have been shown to be unreliable.5
A simple hand held Valsalva Assist Device (VAD) has been developed to provide 40 mm Hg resistance to exhalation and is available as a CE marked device (‘Valse-Valve’ Valsalva Assist Device, Meditech Systems, Shrublands Estate, Sherstock, Shaftesbury, Dorset, England; see figure 1). This device has a mouth piece, a pressure indicator and an air leak to prevent mouth pressure only generated strains (pressure not transmitted to the thorax). However, it has not been tested against a sphygmomanometer in healthy volunteers or in the postures that might be used in clinical practice. Such a device used in clinical practice would be advantageous to control and standardise strains and could also carry clear instructions for the recommended VM posture and strain duration.
The Valsalva Assist Device (shown without connections for manometer and 40mmHg target line).
We conducted a repeated measures trial to compare the vagal responses (drop in HR) and strain characteristics achieved by healthy volunteers performing VMs in both supine and modified positions using the VAD and adapted sphygmomanometer (‘manometer’). Our objectives were to compare the vagal response with modified and supine VMs and to assess the performance of the VAD.
Methods
Participants
The study was approved by the University of Exeter ethic committee and registered with ClinicalTrials.gov (NCT03298880) prior to commencement of recruitment. Healthy adult volunteers (aged 18–60 years) from University of Exeter or Royal Devon & Exeter Hospital NHS Foundation Trust staff were invited to take part through posters and social media. All participants were screened for eligibility (figure 2) and provided informed written consent prior to participation.
Trial inclusion and exclusion criteria.
Screening involved physical examination, including observations of pulse, oxygen saturations, RR and BP and the recording of a 12-lead ECG. Testing was conducted in the clinical research facility of the Royal Devon & Exeter Hospital, according to a strict trial protocol.
Sample size calculations
Previous volunteer studies have suggested a difference of at least 3 beats per min (bpm) as an important difference between techniques.3 We therefore powered our study to detect a difference of 4 bpm, also taking into account informal pilot work suggesting at least this level of difference might be expected between techniques while ensuring a reasonable size study population and appropriately narrow CIs. Calculations were based on a simple paired t-test using the SD reported by Smith et al3 in a similar study (about 12 for both individual measures and for differences between measures). With these parameters, a sample size of at least 73 participants would provide 80% power at the 5% level of significance. We planned to recruit a total of 75 participants in case of unexpected dropouts or device failure.
Procedures
We conducted a randomised repeated measures trial between 1 November 2017 and 5 February 2018 with participants undergoing a total of four VMs of the following variations in random order, stratified by all possible orders.
Study VM interventions
Supine VM using manometer. Supine Valsalva strain using a manometer visible to the participant with a target of 40 mm Hg for 15 s.
Modified VM using manometer. Semi-recumbent (at 45 degrees) Valsalva strain using a manometer visible to the participant with a target of 40 mm Hg for 15 s followed by supine positioning and passive 45 degree leg lift immediately at the end of the strain for a further 15 s.
Supine VM using device. Supine Valsalva strain using the device connected to manometer invisible to the participant but visible to a researcher for 15 s.
Modified VM using device. Semi-recumbent (at 45 degrees) Valsalva strain using the device connected to manometer invisible to the participant but visible to a researcher for 15 s followed by supine positioning and passive 45 degree leg lift immediately at the end of the strain for a further 15 s.
All testing was performed on a standard hospital trolley with a manually adjustable back rest. A 45 degree angle template guide was used to ensure consistent back rest and leg elevation angles. Participants lay at rest for 5 min prior to testing to ensure baseline resting HR was achieved. Participants were read clear, standardised instructions before each manoeuvre and target pressures were marked on the manometer and device gauges. No practices were allowed. A new device and a new 92 cm length of green oxygen bubble tubing, for the manometer, were used for each participant.
A stop watch was used to time all procedures and was visible to participants and researchers. Participants were instructed to stop blowing after the 15 s strain but no other encouragement or instruction was allowed. For safety, participants were not allowed to blow >50 mm Hg, as measured on the manometer, whether using the manometer or device to generate the strain. It was planned that in event of device malfunction (ie, it provides no resistance or resistance is >50 mm Hg), the VM would be abandoned and the malfunction recorded as an adverse incident. The particular manoeuvre would then be restarted using a new device, if the participant was happy to continue.
There was a 3 min washout period between strains including 2 min rest after any change in posture. Continuous 3-lead ECG monitoring on the same, previously calibrated, print enabled defibrillator (Smart Biphasic AED, Philips Heartstart XL) was used to assess HR during the manoeuvre. Standard ECG rhythm strip traces (25 mm/s) were printed for 45 s (15 s before, during and 15 s after each VM). They were marked at the onset of each Valsalva strain, labelled with a code and subsequently analysed in batches, blind to technique according to the method described by Smith et al.3 These traces were second read by another researcher, who was not part of the study team nor present at testing and also blind to allocation.
Premanoeuvre HRs were determined by calculating the mean R-R interval of the 10 beats preceding each manoeuvre before converting it to HR in bpm ((25/mean R-R interval)×60). The lowest postmanoeuvre HR was determined by measuring and recording the longest R-R interval during and up to 15 s postmanoeuvre, converted to an HR in bpm ((25/longest R-R interval)×60). The difference between the premanoeuvre and postmanoeuvre HR indicated the degree of vagal tone or slowing of HR induced by each manoeuvre. Where there was disagreement in ECG measurements between the two readers, the mean of the two figures was taken. The peak sustained strain pressures achieved, as observed on the manometer and duration of longest strain attempt during all VMs were also recorded on a standard report card.
Participants were monitored for any adverse events. Participants who felt unwell or who developed any significant or persistent ECG abnormalities were immediately withdrawn from further testing and appropriate further clinical assessment arranged. All adverse events were recorded, graded and reported according to Good Clinical Practice guidelines.
Statistical analysis
The HRs measured under each of the four testing scenarios were summarised appropriately, for example, mean and SD, with the expectation that HR would be approximately normally distributed. The following comparisons of drop in HR were planned:
Supine VM versus modified VM (recognising that this comparison may or may not be different according to how the strain was generated—manometer or device).
Manometer versus device (recognising that this comparison may or may not be different according to the posture used—supine or modified).
Analysis was based on mixed-effects linear regression (with appropriate prior assessment of assumptions, eg, normality), assessing post-VM HR with individual as a random effect, and posture (supine/modified) and strain method (manometer/device) as fixed effects. An interaction term (posture×strain method) was examined to consider whether there was any evidence of a differential effect (of strain method according to posture, or equivalently posture according to strain method), but was planned to be dropped from the model if p>0.1 (no evidence of interaction). If the interaction term was to be retained then comparison 1 between postures would be presented separately by strain method, and comparison 2 between strain methods would be presented separately by posture type; otherwise the two comparisons would be presented overall.
Two versions of the model were used, one adjusting for pre-VM HR as a covariate, the other not doing so. The pre-VM HR was measured having assumed the relevant starting position for the next manoeuvre, and was (unsurprisingly) lower for the supine manoeuvres than the modified versions, which start in a semi-recumbent position. Hence, these pre-VM HRs are not true baselines comparable across the four scenarios. Moreover, since all four manoeuvres were undertaken by each person (hence essentially comparing results ‘within’ person), the starting HR adds little to statistical efficiency.
Results
Eighty volunteers were screened and five were excluded (heart murmurs (2), ECG abnormalities (2) and participant on medication (1)). Seventy-five healthy participants aged 19–55 years (mean 26) underwent trial VMs. Forty-five (60%) were female. All participants completed all four VMs and there was no missing data. There was one device failure where the pressure observed by manometer was 60 mm Hg while the device was showing a pressure 40 mm Hg. This attempt was abandoned and redone with a new device and reported as an adverse event as described in the ’Methods' section.
Agreement of readers
For the traces analysed for 10 beats prior to the manoeuvre, the two readers had very good agreement for measuring the relevant interval: they both reported overall means of 180.8 mm, with 71% of the 300 readings being identical; 99% of readings were within 1 mm and the (single) worst discepancy was 2.5 mm. For the postmanoeuvre readings of the longest R-R interval, the two readers both recorded overall means of 27.8 mm, with 80% of readings identical and 100% within 1 mm. As agreement of the trace readings was very good (intraclass correlation 0.999 or higher), a simple average of the two readings was subsequently used in the analyses.
Effects on HR
All manoeuvres were associated with a substantial drop in HR. Mean HRs observed for the four VMs are summarised in table 1 together with Valsalva ratios (highest HR/lowest HR recorded during the VM).
Mean HR (SD) with study Valsalva manoeuvres
The table also shows the ‘marginal’ results for all supine manoeuvres (combining manometer and device results); all modified manoeuvres (similarly), all manometer use (combining supine and modified manoeuvres) and all VAD use (similarly).
These unadjusted means in table 1 suggest that the modified manoeuvre was associated with a greater reduction in HR than the supine manoeuvre, a difference of about 8 bpm. Moreover, this difference is consistent across strain methods, with a similar difference whether the manometer or device is used. The table also suggests a small difference between the two strain methods, with a slightly larger reduction in HR for the manometer compared with the VAD, again with consistency across the two postures used.
The mixed-effects regression (including pre-VM HR as a covariate) resulted in a small interaction (strain method×posture of VM) with a p value of 0.70, hence was dropped as planned. The resulting model then showed a significant effect for VM posture used, with the modified version reducing HR by 3.8 bpm more than the supine version (95% CI 2.2 to 5.4; p<0.001). The strain method also showed a significant effect (although weaker), with the manometer reducing HR by 1.9 bpm more than the VAD (95% CI 0.4 to 3.4; p=0.01) (table 2).
Adjusted differences in fall of HR between postures and strain methods using mixed-effects regression to adjusting for baseline (pre-VM) HR
Repeating the regression without the pre-VM HR as a covariate produced similar results: a small non-significant interaction (p=0.69), and significant benefits for the modified posture (2.3 bpm, 95% CI 0.8 to 3.8; p=0.003) and manometer (2.0 bpm, 95% CI 0.5 to 3.5; p=0.01).
Strain characteristics
The mean strain pressures delivered by manometer and VAD were similar (39.96 vs 42.46 mm Hg), but manometer was more precise with 97% of participants straining between 35 and 45 mm Hg compared with 75% when using the VAD. The use of VAD was also associated with significantly shorter total duration of strain and anecdotally was due to subjects running out of breath. This was particularly evident in females with only about a third of female participants achieving a full 15 s strain with the device compared with 95% of males (data not shown) (table 3).
Mean pressure and strain duration characteristics
Adverse events
Thirty-four (11%) of the 300 VM undertaken were associated with symptoms reported during the attempts. All of these were transient and did not prevent completion of VMs and were similar between the four different VM under investigation. The most common side effect was a headache and lightheadedness. No serious adverse events were recorded (table 4).
Recorded adverse events by allocation
Discussion
Our results demonstrate the physiological advantage of the described modified Valsalva over a purely supine VM. This postural modification was associated with a greater vagal response, shown by the absolute and relative drop in HR compared with the supine VM. Despite supine positioning being associated with a lower initial HR (likely because of initial increased vagal tone), the exaggerated effects on venous return resulting from semi-recumbent position in the strain phase and leg elevation in the supine relaxation phase (Valsalva stage 3) resulted in more intense vagal stimulation overall with the modified VM.
Previous work has suggested that vagal techniques associated with a greater drop in HR and so larger Valsalva ratio, are correlated with a greater chance of SVT being terminated when that vagal manoeuvre is used to treat this arrhythmia.6 So, although caution must be employed in extrapolating volunteer studies, demonstration of a significantly greater drop in HR with the modified VM in this study, is consistent with the efficacy of this postural modification in clinical trials2 7 and benefits of a modified VM over a purely supine technique.
Although straining in upright postures is in theory associated with a greater drop in BP and an increased risk of syncope during straining,7 this complication was not seen in our study which used a controlled and defined strain and is consistent with our experience of the modified VM in clinical practice and trials. There were no serious adverse events and the number of non-serious adverse events was similar between the two different posture groups. We believe this study therefore further supports the case for routine use of a modified VM as described, over a supine VM.
Overall, VMs with VAD generated strains resulted in a meaningful fall in HR of 29 bpm. This is greater than that seen in similar volunteer studies using a manometer.3 8 The VAD, however, did not perform quite as well as the manometer in terms of vagal response in our study. Although the difference in drop of mean HR between these strain methods just reached statistical significance, it was less than our previously stated clinically meaningful difference of 4 bpm (which was also excluded from the 95% CI) and it is debatable whether this small difference would affect cardioversion rates if it were replicated in clinical practice.
The device produced a similar mean pressure of strain. Although the variation of pressures was greater than with manometer, this was within a clinically appropriate range and considerably better than that seen with use of a syringe.4 The VAD, however, was associated with significantly shorter strain durations than strains using the manometer. This was almost certainly due to the predesigned air leak which was probably a little too large for most volunteers compared with the manometer which is a sealed system with no leak. Shorter durations of strain were seen most with VAD delivered strains by female participants. Female sex is associated with a physiologically lower functional residual lung capacity and supports the theory that the device leak was the cause for the reduced strain duration observed with the VAD. This degree of leak and shorter duration of VM strain might account for the devices marginally reduced effect on HR. This has been fed back to the manufacturer to consider refinements to their design. There was no significant difference but a trend towards a lower number of adverse events in the VAD group. Further study could reassess performance of such device modifications and compare it again with the manometer or with the syringe, as the most commonly used alternative in practice.
We conducted our study on mainly young, healthy volunteers with a slight preponderance of females. This is the demographic of many patients with SVT; however, there is a second peak of incidence in older age, often in patients with associated comorbidities and so consideration should also be given to repeating this study and assessing VAD performance in the older population.
Finally, findings from this volunteer study using the VAD should not be used alone in changing the care of patients with SVT. Other work will be needed to assess the performance of such devices in clinical practice. A feasibility trial is currently underway to look at use of the VAD (after modifications) by paramedics in the South West of England to treat patients with SVT (EVADE Study, NCT03514628).
Conclusions
This study indicates that a modified VM results in a greater vagal response than a standard supine VM with no increase in adverse effects when used with a controlled strain in healthy volunteers. Our findings give support for the physiological advantage of this specific manoeuvre over a purely supine VM.
We have also shown a simple hand held device can be used to generate recommended VM strain pressures but its use in volunteers resulted in shorter strain durations and a slightly smaller drop in HR compared with a modified sphygmomanometer. Refinements to this device may improve its function further and allow it to be assessed in clinical practice to treat SVT.
Acknowledgments
The authors would like to thank Anna Steele and the staff at the CRF for their support in conducting the study. We would also like to acknowledge and thank Gerens Curnow for their independent reading of the ECG traces.
Footnotes
Contributors IF recruited the participants, ran the study and wrote the first draft of the manuscript. PE advised on study design, conduced the statistical analysis and reviewed the manuscript drafts. IL co-supervised IF, advised on the project and reviewed the manuscript. AA conceived the idea for the study, co-supervised IF, reviewed and revised the manuscript and acts as guarantor. The project was part of IF’s Masters by Research (MbyRes) project at the University of Exeter and was also submitted to the RCEM undergraduate research competition.
Funding The project was supported by the University of Exeter Medical school as part of a Masters by Research project (IF) and the Academic Department of Emergency Medicine. IL’s contribution to this research was supported by the National Institute for Health Research (NIHR) Collaboration for Leadership in Applied Health Research and Care South West Peninsula. Valsalva Assist Devices were supplied by Meditech Systems and purchased jointly by the Academic Department of Emergency Medicine and the Academic Health Science Network (SW).
Disclaimer This paper presents independent research funded by the Academic Department of Emergency Medicine, Royal Devon & Exeter Hospital NHS Foundation Trust supported by the National Institute for Health Research (NIHR) Exeter Clinical Research Facility. The views expressed are those of the authors and not necessarily those of the Royal Devon & Exeter Hospital NHS Foundation Trust, the NIHR or the Department of Health.
Competing interests None declared.
Patient consent Not required.
Ethics approval This study was approved by the University of Exeter Medical School (The Research, Innovation, Learning and Development Building, Royal Devon and Exeter Hospital, Barrack Road, Exeter, UK) research committee.
Provenance and peer review Not commissioned; externally peer reviewed.