Heart rate variability and stroke volume variability to ...

38 downloads 155 Views 714KB Size Report
Mar 27, 2015 - measures of HRV were reduced during hypovolemia compared to ... We conclude that HRV% and SVIntegral detect hypovolemia during.
Home

Search

Collections

Journals

About

Contact us

My IOPscience

Heart rate variability and stroke volume variability to detect central hypovolemia during spontaneous breathing and supported ventilation in young, healthy volunteers

This content has been downloaded from IOPscience. Please scroll down to see the full text. 2015 Physiol. Meas. 36 671 (http://iopscience.iop.org/0967-3334/36/4/671) View the table of contents for this issue, or go to the journal homepage for more

Download details: IP Address: 129.240.194.207 This content was downloaded on 27/03/2015 at 14:14

Please note that terms and conditions apply.

Institute of Physics and Engineering in Medicine Physiol. Meas. 36 (2015) 671–681

Physiological Measurement doi:10.1088/0967-3334/36/4/671

Heart rate variability and stroke volume variability to detect central hypovolemia during spontaneous breathing and supported ventilation in young, healthy volunteers Maja Elstad and Lars Walløe Department of Physiology, Institute of Basic Medical Sciences, University of Oslo, Blindern, 0317 Oslo, Norway E-mail: [email protected] Received 6 October 2014, revised 13 November 2014 Accepted for publication 14 November 2014 Published 23 March 2015 Abstract

Cardiovascular oscillations exist in many different variables and may give important diagnostic and prognostic information in patients. Variability in cardiac stroke volume (SVV) is used in clinical practice for diagnosis of hypovolemia, but currently is limited to patients on mechanical ventilation. We investigated if SVV and heart rate variability (HRV) could detect central hypovolemia in spontaneously breathing humans: We also compared cardiovascular variability during spontaneous breathing with supported mechanical ventilation. Ten subjects underwent simulated central hypovolemia by lower body negative pressure (LBNP) with >10% reduction of cardiac stroke volume. The subjects breathed spontaneously and with supported mechanical ventilation. Heart rate, respiratory frequency and mean arterial blood pressure were measured. Stroke volume (SV) was estimated by ModelFlow (Finometer). Respiratory SVV was calculated by: 1) SVV% = (SVmax − SVmin)/SVmean during one respiratory cycle, 2) SVIntegral from the power spectra (Fourier transform) at 0.15–0.4 Hz and 3) SVV_norm = (√SVIntegral)/SVmean. HRV was calculated by the same methods. During spontaneous breathing two measures of SVV and all three measures of HRV were reduced during hypovolemia compared to baseline. During spontaneous breathing SVIntegral and HRV% were best to detect hypovolemia (area under receiver operating curve 0.81). HRV% ≤ 11% and SVIntegral ≤ 12 ml2 differentiated between hypovolemia and baseline during spontaneous breathing.

0967-3334/15/040671+11$33.00  © 2015 Institute of Physics and Engineering in Medicine  Printed in the UK

671

M Elstad and L Walløe

Physiol. Meas. 36 (2015) 671

During supported mechanical ventilation, none of the three measures of SVV changed and two of the HRV measures were reduced during hypovolemia. Neither measures of SVV nor HRV were classified as a good detector of hypovolemia. We conclude that HRV% and SVIntegral detect hypovolemia during spontaneous breathing and both are candidates for further clinical testing. Keywords: cardiac stroke volume variability, heart rate variability, respiration, hypovolemia (Some figures may appear in colour only in the online journal) 1. Introduction Cardiovascular oscillations exist in many forms in all cardiovascular variables. In the present paper we focus on the respiration induced cardiovascular oscillations, more specifically on the variability in cardiac stroke volume and heart rate at respiratory frequency. Interactions between respiration and circulation are both mechanically and neurally regulated, and both mechanical and neural factors can be modified in a clinical setting. An additional advantage of respiratory variability is that even short recordings can provide a good estimation of the cardiovascular variability (Task Force of the European Society of Cardiology and The North American Society of Pacing and Electrophysiology 1996) which is of great value in critically ill patients. There are many proposed methods of measuring cardiovascular variability, in this study we test three different measures. Cardiovascular oscillations are of increasing interest as clinical tools to detect hypovolemia due to their potential non-invasive nature and the fact that cardiovascular variability may change prior to observable changes in their absolute levels as shown in end-organ damage due to hypertension (Tatasciore et al 2007). Trauma is a leading cause of death in the young population. A substantial part of these deaths could be prevented if serious hemorrhages were detected at an earlier stage (Cales and Trunkey 1985, Anderson et al 1988, Sauaia et al 1995, Hanson et al 1998). A major obstacle when assessing an ongoing internal bleeding is the compensatory physiological mechanisms which camouflage the level of hemorrhage until the edge of collapse. In the clinical setting, vital signs such as heart rate, blood pressure and respiratory rate, are monitored to distinguish the compensatory phase of a bleeding patient from the non-bleeding patient. Classically, heart rate and respiratory rate increase, while blood pressure decreases during a hemorrhage. However, these vital signs are not good enough to detect bleeding at an early stage (Pacagnella et al 2013) as the reference values show a wide range, and the time between a compensated hemorrhage with next-to-normal cardiovascular variables and a cardiovascular collapse could be seconds to minutes. A non-invasive technique to detect ongoing bleeding is warranted, especially in a pre-clinical setting. Cardiac stroke volume is known to change immediately with hypovolemia (caused by for instance hemorrhage), as preload is reduced. There is however no current validated method to distinguish hypovolemia from normovolemia on the basis of cardiac stroke volume measurements. Cardiac stroke volume shows large variability at respiratory frequency, due to mechanical difference in preload during spontaneous expiration and inspiration (Toska and Eriksen 1993, Elstad et al 2001, Elstad 2012). On the other hand, cardiac stroke volume variability (SVV) is an established technique for diagnosing fluid-responsive hypovolemia during mechanical ventilation (Marik et al 2009). SVV is in other words already in clinical use, but 672

M Elstad and L Walløe

Physiol. Meas. 36 (2015) 671

currently limited to patients on mechanical ventilation, and most of the patients with ongoing, undetected bleeding are breathing spontaneously. We therefore wanted to investigate a technique to detect mild hypovolemia in spontaneously breathing volunteers and compare this with mechanical ventilation in the same subjects. Heart rate also varies at different frequencies, and this is thought to reflect the tone of autonomic nervous system. Since the compensatory mechanisms during bleeding mediated by the parasympathetic and sympathetic nerves, heart rate variability (HRV) has been under extensive investigation for its potential as detector of hypovolemia (Ryan et al 2011). The aim of the current study was to investigate whether SVV and HRV could be used to detect central hypovolemia in spontaneously breathing, young, healthy humans. Central hypovolemia induced by lower body negative pressure (LBNP) simulates the cardiovascular changes obtained during a hemorrhage (Johnson et al 2014). In the current study we therefore compared the changes in SVV and HRV during LBNP during both spontaneous breathing and supported mechanical ventilation. Our primary hypothesis was that SVV could detect hypovolemia during spontaneous breathing. 2. Methods 2.1. Subjects

Sixteen young healthy volunteers (nine females), aged between 19 and 25 years, were recruited. None of the subjects had any symptoms of cardiovascular disorder, none smoked and none used any medication (except contraceptive pills). Their physical fitness was normal, and all took weekly exercise (median 5 h, range 2–10 h). Written informed consent was obtained from all subjects. 2.2.  Experimental protocol

The subjects visited the laboratory twice before the two experimental days. During these visits they were acclimatized to the laboratory and the non-invasive intermittent positive pressure ventilation (IPPV). Each subject had individually set ventilation settings, which were adjusted on the experimental days. The subjects were instructed to abstain from coffee, tea and exercise on the experimental day and to have a light meal two hours prior to each experiment. They abstained from alcohol for at least 24 h prior to the experiment. On the experimental day, the subject rested supine on a special bench and was fitted in the lower body negative pressure chamber to prevent any air leakage that could evacuate the negative pressure within the chamber. The subjects took breaths of normal depth when breathing spontaneously and the subjects were provided individually set breaths by non-invasive IPPV (VIVO50, Diacor a/s, Oslo, Norway, matched to the spontaneous breathing). The frequency of the IPPV was decided by the subject’s spontaneous respiratory frequency. The non-invasive IPPV was provided through a facial mask individually fitted (Respireo Primo F Non Vented, Air Liquide Medical Systems, Italy). We set the inspiration time (median 1.5 s (range 1.2–1.8 s), inspiratory pressure individually (minimum: median 6 cm H2O, maximum: median 15 cm H2O) and we set a low expiratory pressure (range 2–3 cm H2O). One experimental protocol consisted of a baseline recording of ten minutes. Immediately (within 0.3 s, (Hisdal et al 2003)) after baseline, followed a simulated central hypovolemia of ten minutes produced by lower body negative pressure (LBNP) at −30 mmHg. After LBNP, 673

M Elstad and L Walløe

Physiol. Meas. 36 (2015) 671

10 min

10 min

10 min

0

LBNP -10 (mm Hg) -20 -30 A: SB B: IPPV

IPPV SB

SB IPPV

IPPV SB

Figure 1.  The protocol. Baseline, hypovolemia and recovery lasted ten minutes each.

Ventilation mode either started with spontaneous breathing and continued as illustrated in A or started with IPPV and continued as illustrated in B. IPPV, intermittent positive pressure ventilation; LBNP, lower body negative pressure; SB, spontaneous breathing.

the chamber pressure returned to atmospheric pressure within 0.3 s and a ten-minute-recording was made during recovery. All three experimental situations (baseline, LBNP and recovery) included five minutes of spontaneous breathing and five minutes of IPPV. The starting order of the ventilation mode was randomized, and to prevent too many shifts between ventilation modes the subsequent ventilation modes were fixed (figure 1). The analysis of cardiovascular variability was performed on one experimental run from the second day. The starting ventilation mode was randomized, so six subjects started with spontaneous breathing and four with IPPV. The LBNP was terminated if the subject experienced any symptoms (dizziness, nausea, vision loss) or predefined cardiovascular signs of presyncope such as reduction of blood pressure >15 mmHg or increase in heart rate > 120 beats per minute. All 16 subjects completed the protocol without any subjective discomfort or other termination criteria. 2.3.  Instrumentation and recordings

Respiratory chest movement was obtained using a belt around the upper abdomen (Respiration and Body position Amplifier, Scan-Med a/s, Drammen, Norway). HR was obtained from the duration of each R-R interval of the three-lead ECG signal (SD-100). Finger arterial pressure was recorded continuously from the middle left finger positioned at heart level (Finometer, Finapres Medical System, Amsterdam, The Netherlands). The pressure output was transferred to the recording computer, and beat-by-beat mean arterial blood pressure (MAP) was calculated by numerical integration. The Finometer also provided SV calculated by ModelFlow (Bogert and van Lieshout 2005). SV measured by ModelFlow is in good accordance with SV measured by ultrasound Doppler during supine rest (Van Lieshout et al 2003) and reflects the progression of SV during LBNP (Reisner et al 2011). The signals were sampled at 100 Hz and transferred on-line to a recording computer running a dedicated data collection and analysis program (program for real time data acquisition: Morten Eriksen, Oslo, Norway). A capnograph (inbuilt in VIVO50) registered the expiratory CO2 level and indicated if a subject hypoor hyperventilated. 674

M Elstad and L Walløe

Physiol. Meas. 36 (2015) 671

2.4.  Signal processing and analysis

Every recorded signal from each experimental run was visually inspected, and only time intervals with successful recordings were included in the subsequent analysis. In addition, each selected continuous sequence with acceptable measurements had to last for at least ten respiratory cycles (Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology 1996). The original recording was sampled at 300 Hz for ECG, 100 Hz for lung volume and SV and beat-by-beat for HR and MAP. Out of the 16 recruited subjects, six had less than 10% decrease in SV during LBNP. We had chosen a reduction of SV ≥ 10% to be included in the analysis as the clinical criteria for fluid-responsive hypovolemia is an increase in SV by >10% when fluid is supplied. For the ten subjects with a significant SV reduction, we calculated respiratory heart rate variability and stroke volume variability by three different methods. Respiratory SVV was calculated by three methods (equations (1)–(3)) and respiratory HRV was calculated by the same three methods. SVV% = (SVmax − SVmin) / (SVmean) 

(1)

SVIntegral = Integral of power spectra at high frequency 

(2)

(

)

SVV norm = √ SVIntegral / SVmean 

(3)

SVV% was calculated during one respiratory cycle and was set to the median of five breaths. Within the five breaths the median ratio between maximum and minimum of SVV% was 1.9 (range 1.2–4.2). Equation  (1) reflects current clinical practice for SVV estimation (Guinot et al 2014). Many studies calculate average SVV% over 20–30 s, and the five breaths median is a corresponding method. SVIntegral was calculated from 240 s epochs. Within 240 s the individual numbers of breaths ranged from 24 and 76. Power density spectra were calculated for each of the variables in the separate time intervals to obtain variability at 0.15–0.4 Hz, also called high frequency power. The power spectra were calculated by the fast Fourier transform algorithm. Prior to analysis, the beat-to-beat signals were converted into equidistant time samples by interpolation. The distance between samples after interpolation ensured that the resulting number of samples was an integer power of two, which was a requirement for the subsequent analysis. SV, sampled at 100 Hz, did not need conversion to equidistant time samples. The spectra were smoothed by a sliding Gaussian function with standard deviation of 0.01 Hz. Equation  (2) has been previously published for SVV (Elstad 2012) and is standard for HRV (Task Force of the European Society of Cardiology and The North American Society of Pacing and Electrophysiology 1996). Equation (3) was suggested as a normalization for HRV to correct for HR differences between groups, termed as the coefficient of component variance (Hayano et al 1990). This is also relevant to SV, which decreases during hypovolemia. We performed this non-standardized normalization of SVV and HRV to obtain a measure for the magnitude of fluctuations relative to the mean value. In addition to this, SVV is also shown to change when HR changes (Roeth et al 2014), which we have not accounted for in our analysis. 2.5. Statistics

The non-parametric median and 95% confidence intervals (CI) were calculated by Hodges– Lehmann’s estimates (StatExact, Cytel Studio 7; Cytel Inc., Cambridge, MA, USA). Predictive 675

M Elstad and L Walløe

Physiol. Meas. 36 (2015) 671

120

Stroke 100 Volume (ml) 80 60 70

Heart rate (bpm)

60

50

40

0

-10

LBNP (mm Hg) -20 -30 150

165

1801005 1020 1335 1350

Time (s)

Figure 2.  Heart rate and stroke volume during baseline and  −30 mmHg lower body negative pressure in one subject during spontaneous breathing. Heart rate was increased and stroke volume is reduced during central hypovolemia. The respiratory variations in heart rate and stroke volume were reduced during central hypovolemia in this spontaneously breathing subject. LBNP, lower body negative pressure.

values of the variations were calculated by the area under receiver operating curves (AUROC, SPSS 20, Chicago, IL, USA). We considered AUROC ≥ 0.80 to be a good detector of hypovolemia. The Wilcoxon signed rank sum test against a two-sided alternative was used to test for differences between situations. The Wilcoxon median and upper and lower limits of the 95% confidence interval (CI) are reported, which corresponds to the non-parametric one-sample test (Hollander and Wolfe 1999). 2  ×  2 tables were tested with n − 1 X2 test (Campbell 2007). P ≤ 0.05 was considered significant. 3. Results Figure 2 shows the recordings from one subject during four respiratory cycles at baseline and during −30 mmHg LBNP. Heart rate increased (+10 bpm), while stroke volume (−20 ml) and mean arterial blood pressure (−2.5 mmHg) decreased during −30 mmHg LBNP compared to baseline. Respiratory frequency, respiratory depth and CO2 were unchanged throughout the protocol. 676

M Elstad and L Walløe

Physiol. Meas. 36 (2015) 671

Table 1.

  Hemodynamic variability during spontaneous breathing. Median and 95% confidence interval

Equations (1)

Hemodynamic variability SVV% (%) SVIntegral (ml2) SVV_norm

(2) (3)

SV (ml) (1)

HRV% (%) HRIntegral (bpm2) HRV_norm

(2) (3)

HR (bpm)

Baseline

−30 mmHg LBNP

Recovery

16 (9, 21) 17.5 (9.8, 25.8) 0.047 (0.028, 0.055) 87.2 (73.8, 97.5) 17 (13, 25) 11.0 (4.9, 16.8) 0.056 (0.034, 0.067) 57.5 (47.5, 61.8)

10* (7, 15) 6.1* (2.8, 11.0) 0.034 (0.020, 0.042) 70.6 (60.3, 78.1) 9* (6, 14) 4.1* (2.0, 11.1) 0.032* (0.021, 0.039) 63.3 (54.3, 67.7)

16 (11, 21) 19.1 (10.6, 26.8) 0.047 (0.029, 0.056) 91.9 (79.1, 100.1) 14 (11, 21) 6.7 (4.2, 27.5) 0.049 (0.032, 0.056) 55.9 (46.7, 59.6)

AUROC (Baseline and LBNP) 0.64 0.81 0.68

0.81 0.74 0.79

AUROC, area under receiver operating curve; bpm, beats per minute; HRV, heart rate variability; LBNP, lower body negative pressure; SVV, stroke volume variability. *indicates p