Autonomic dysfunction in the surgical population is a frequently overlooked complication of cardiovascular diseases, such as heart failure, coronary artery disease and diabetes mellitus.1,2 The presence of autonomic dysfunction may have important clinical and prognostic implications due to its association with peri-operative haemodynamic instability, postoperative outcome, increased mortality after myocardial ischaemia and sudden cardiac death.3–5
Evaluation of autonomic function is commonly performed by quantitative assessment of beat-to-beat heart rate variability (HRV) reflecting parasympathetic and sympathetic control of the sinoatrial node.6–8 Particularly, in case of impaired autonomic innervation, the characteristic oscillations in heart rate are reduced. International standards recommend the use of R–R intervals derived from an electrocardiogram (ECG) to calculate HRV.9 In the intra-operative setting, movement artefacts and electrical interference due to diathermy may, however, complicate R-wave detection. Pulse rate variability (PRV) derived from continuous blood pressure measurements may provide a feasible alternative for HRV with less sensitivity to peri-operative environmental disturbances. However, several potential sources of error exist that may interfere with accurate determination of cardiac cycle length. In contrast to the ECG R-wave, a pressure pulse wave lacks a sharp peak. Furthermore, the shape and velocity of the pressure pulse wave strongly depends on stroke volume, ventricular pressure, vascular tone and pulse transit time.10 These potential limitations for accurate pulse peak detection raised the question whether PRV is comparable to HRV and can be used to evaluate autonomic function.
Recently, the Nexfin HemoDynamics monitor has become available which uses an updated implementation of the Finapres method for non-invasive, beat-to-beat blood pressure measurements.11,12 Using Nexfin HemoDynamics, we investigated the level of agreement between PRV derived from blood pressure waveforms and traditional HRV derived from an ECG.
This prospective observational study (Ethical Committee number 2009/249) was approved by the Human Subjects Ethics Committee of the VU University Medical Center, Amsterdam, the Netherlands (President Professor P.E. Postmus) on 17 September 2009. All participants gave written informed consent.
Healthy male individuals were recruited among institute staff, residents and medical students. Individuals with a history of cardiovascular disease, antihypertensive treatment, diabetes mellitus or a BMI less than 15 kg/m2 or more than 35 kg/m2 were excluded.
Individuals were studied on two occasions. Both measurements were performed by the same investigator between 8:00 and 10:00 a.m. after overnight fasting. Individuals were asked to refrain from smoking and caffeine-containing beverages 12 h before testing. On each occasion, individuals were studied during 5 min of spontaneous breathing in the supine position. After stabilisation of heart rate and blood pressure, R–R intervals were obtained using standard bipolar ECG leads connected to a recorder. Simultaneously, interbeat intervals (IBIs) derived from blood pressure waveforms were recorded using a continuous finger arterial blood pressure measurement device (Nexfin HemoDynamics, BMEYE, the Netherlands). The finger cuff was applied to the middle finger of the right hand according to the manufacturer's instructions. The Nexfin uses updated technology of the Finapres device which was developed in the 1980s based on the volume clamp methodology by Peňãz.13 Briefly, volume clamping keeps the finger artery at its unloaded volume, the point at which finger cuff pressure and intra-arterial pressure are equal and transmural pressure is zero. The arteries are kept at this unloaded diameter by changing finger cuff pressure in parallel with intra-arterial pressure. Cuff pressure thus provides an indirect measure of intra-arterial pressure.14,15
Heart rate variability analysis
R–R intervals were recorded using a QRS detector with a sample rate of 1000 Hz and the IBI data obtained with the Nexfin were recorded with a sample rate of 200 Hz. Heart rate and blood pressure signals were visually inspected for premature or abnormal beats and movement artefacts. Both signals were digitised and stored on a personal computer for further analysis using free available software (Kubios HRV version 2.0, University of Kuopio, Finland).16 The software extracts R–R or interbeat intervals, constructs time series and subsequently performs spectral analysis using fast Fourier transformation. This technique divides the overall variability of a signal into its composing frequencies and provides insight into what extent a frequency contributes to the overall variability of the signal. The power spectrum of HRV and PRV has been shown to consist of three peaks: the very low frequency (VLF) band (<0.04 Hz), the low-frequency band (0.04–0.12 Hz) and the high-frequency band (0.12–0.40 Hz).8,17 The VLF fluctuations are thought to be mediated primarily by the sympathetic system, the low-frequency fluctuations by both the sympathetic and parasympathetic system and the high-frequency fluctuations are under parasympathetic control.
All statistical analyses were carried out using SPSS statistical software version 15.0 (SPSS Inc., Chicago, Illinois, USA). Sample size was calculated on the basis of the correlation between ECG-derived HRV and Nexfin-derived PRV. We chose a correlation of at least 0.9 as a potentially meaningful correlation. For an α = 0.01 and a β = 0.10, a sample size of 10 patients was calculated. To compensate for dropouts, 20 patients were enrolled.
The association between HRV and PRV was analysed using Spearman's rank correlation coefficient. Level of agreement was analysed by constructing Bland–Altman plots. HRV and PRV parameters were checked for heteroscedasticity and, if necessary, log-transformed prior to calculation of the limits of agreement (LoA).18
Finally, to assess test–retest reliability of HRV and PRV, the intra-class correlation coefficients (ICCs) were calculated for pairs of repeated measurements (two-way random model, absolute agreement, average measures). A P value of less than 0.05 was considered statistically significant.
Twenty healthy male individuals aged between 19 and 34 years were studied. Data were obtained on two occasions with 57 ± 14 days apart. In one individual, the retest data were excluded because of significant heart rhythm disturbances affecting the results of spectral analysis.
Correlation and agreement between heart rate variability and pulse rate variability
The correlation between the variability in heart rate and pulse rate is shown in Table 1, which was 0.99 for all frequency bands. Bland–Altman analysis (Fig. 1) revealed a mean difference between PRV and HRV of 1% in the VLF (panel a) and low-frequency band (panel b), whereas PRV was 14% higher than HRV in the high-frequency band (panel c). Antilogs of the LoA revealed that for 95% of cases in the VLF band, PRV will be between 0.97 and 1.04 times HRV. This indicates that PRV measurements may differ from HRV measurements by 3% below to 4% above. The difference between PRV and HRV ranges from 5% below to 8% above in the low-frequency band and from 4% below to 36% above in the high-frequency band (Table 2).
Test–retest reliability of heart rate variability and pulse rate variability
ICCs calculated for the test–retest reliability of HRV and PRV are shown in Table 3. The ICC values range from moderate (r = 0.38) to good (r = 0.93) and were comparable for both HRV and PRV.
The present study provides merits to use PRV as alternative for traditional HRV in the detection of disturbances in autonomic function. PRV derived from blood pressure waveforms using non-invasive, beat-to-beat blood pressure monitoring showed a high correlation and level of agreement with traditional HRV obtained with an ECG. Furthermore, both PRV and HRV showed a moderate to high test–retest reproducibility. These results imply that PRV, which is less influenced by environmental disturbances than HRV, may be used for autonomic function assessment in the peri-operative period.
Our findings are consistent with previous reports investigating alternative methods for HRV assessment.19–21 McKinley et al.19 studied the reliability of interbeat intervals by finger plethysmography (Finapres) and ECG for calculation of HRV parameters under different conditions. Although correspondence between the two methods was excellent, the Finapres overestimated the power in the high-frequency (0.15–0.40 Hz) band during supine rest. This overestimation was more pronounced during exercise, but the level of agreement remained acceptable. Giardino et al.21 also compared finger plethysmography with ECG in providing accurate cardiac cycle duration for calculation of HRV. Both methods correlated highly for low-frequency (0.07–0.15 Hz) and high-frequency (0.15–0.40 Hz) HRV. However, during rest and exercise, plethysmography-derived HRV overestimated the power in the high-frequency band, and to a lesser extent the power in the low-frequency band. Finally, Carrasco et al.20 studied ECG-derived and Finapres-derived HRV and reported a high correlation between the two acquisition methods, but with significant overestimation of the high-frequency band using Finapres data during standing and exercise. Our data also demonstrate excellent correspondence for the VLF and low-frequency band. Although the high-frequency band derived from PRV was overestimated, these data are clinically acceptable.
The explanation for the overestimation of high-frequency variability under various conditions remains unclear, but several factors may contribute to differences between PRV and HRV. Giardino et al.21 investigated the influence of sampling rate on the correlation between ECG-derived and plethysmograph-derived intervals. The correlation between the two methods decreased with a decrease in sampling rate. In our study, Nexfin-derived intervals were obtained with a sample rate of 200 per second and the R–R intervals with a rate of 1000 per second, which may have contributed partly to the difference in ECG-derived and Nexfin-derived high-frequency band. Another possible explanation is suggested by the observation that the differences between the two methods occurred periodically within the frequency range of respiratory activity.21 Whether the periodical influence of respiration is indeed capable of influencing high-frequency PRV is a subject of further research.
Cardiac cycle duration detected with finger plethysmography may also differ from ECG-derived intervals because the shape of the pressure waveform is dependent on different characteristics of the peripheral vasculature. For example, various studies showed that arterial stiffness decreases in healthy individuals during exercise, thereby reducing pulse transit time and pulse wave velocity.22,23 Other factors that may influence arterial stiffness are increasing age, arteriosclerosis and hypertension. Whether changes in arterial stiffness and pulse transit time alter the shape of the pressure waveform in a way that affects determination of cardiac cycle duration remains to be elucidated. Therefore, the possibility of extrapolating our results to conditions that affect shape and velocity of the pulse wave, such as peripheral or cardiovascular disease and critically ill patients requiring vasoactive medication, needs to be explored in future studies.
The test–retest reliability of HRV and PRV measurements are comparable. Previous studies focusing on reliability of HRV using short-term recordings have shown similar results.24,25 The relatively low ICC for the VLF band is consistent with the report from the Task Force of the European Society of Cardiology that recommend to assess the VLF band from long-term (24 h) recording to ensure maximal reliability.9
The following limitations should be taken into account in the interpretation of the present results. As previously stated, in this study, we included only healthy individuals and therefore no conclusions can be drawn about the use of PRV under pathological conditions, such as increased sympathetic stress or autonomic dysfunction. Furthermore, the agreement and reliability of PRV were obtained during supine rest and not during exercise, which limits comparison to other studies. However, current guidelines recommend to perform spectral analysis under resting conditions for optimal systematic evaluation.2 In addition, age-specific and sex-specific reference values for the different frequency bands are studied only for resting HRV, making this the most important clinical parameter.24,26 Finally, in this study, we used the Nexfin, which derives arterial blood pressure waveforms with a specific finger probe by the volume clamp methodology. Although this methodology has been established in autonomic function testing, its intra-operative use under different clinical conditions largely remains to be validated.11,24 We, therefore, emphasise that further studies investigating the value of PRV to evaluate peri-operative autonomic function under various clinical conditions are warranted.
In conclusion, our data show that under resting conditions, PRV derived from non-invasive blood pressure waveforms obtained with the Nexfin corresponds well with traditional HRV derived from ECG. These results indicate that in healthy individuals under standard conditions blood pressure waveforms may replace HRV in the evaluation of autonomic function. Therefore, PRV can be used in research settings in which an ECG signal is absent or unwanted. These results further expand the use of PRV in autonomic function testing and future investigations should explore whether its use can be extrapolated to different clinical conditions and patient subgroups.
This work was supported by the Department of Anesthesiology, VU University Medical Center, Amsterdam, the Netherlands.
R.A.B. is supported by a Dr E. Dekker research career grant of the Netherlands Heart foundation (grant 2008T003).
There are no conflicts of interest.
1 Maser RE, Mitchell BD, Vinik AI, Freeman R. The association between cardiovascular autonomic neuropathy and mortality in individuals with diabetes: a meta-analysis. Diabetes Care 2003; 26:1895–1901.
2 Vinik AI, Ziegler D. Diabetic cardiovascular autonomic neuropathy. Circulation 2007; 115:387–397.
3 Kahn JK, Sisson JC, Vinik AI. Prediction of sudden cardiac death in diabetic autonomic neuropathy. J Nucl Med 1988; 29:1605–1606.
4 Suarez GA, Clark VM, Norell JE, et al
. Sudden cardiac death in diabetes mellitus: risk factors in the Rochester diabetic neuropathy study. J Neurol Neurosurg Psychiatry 2005; 76:240–245.
5 Burgos LG, Ebert TJ, Asiddao C, et al
. Increased intraoperative cardiovascular morbidity in diabetics with autonomic neuropathy. Anesthesiology 1989; 70:591–597.
6 Hyndman BW, Kitney RI, Sayers BM. Spontaneous rhythms in physiological control systems. Nature 1971; 233:339–341.
7 Hyndman BW. The role of rhythms in homeostasis. Kybernetik 1974; 15:227–236.
8 Akselrod S, Gordon D, Ubel FA, et al
. Power spectrum analysis of heart rate fluctuation: a quantitative probe of beat-to-beat cardiovascular control. Science 1981; 213:220–222.
9 Heart rate variability: standards of measurement, physiological interpretation and clinical use. Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology. Circulation
10 Safar ME. Pulse pressure, arterial stiffness and wave reflections (augmentation index) as cardiovascular risk factors in hypertension. Ther Adv Cardiovasc Dis 2008; 2:13–24.
11 Eeftinck Schattenkerk DW, van Lieshout JJ, van den Meiracker AH, et al
. Nexfin noninvasive continuous blood pressure validated against Riva-Rocci/Korotkoff. Am J Hypertens 2009; 22:378–383.
12 Molhoek GP, Wesseling KH, Settels JJ, et al
. Evaluation of the Penaz servo-plethysmo-manometer for the continuous, noninvasive measurement of finger blood pressure. Basic Res Cardiol 1984; 79:598–609.
13 Peňãz J. Current photoelectric recording of blood flow through the finger. Cesk Fysiol 1975; 24:349–352.
14 Wesseling KH, de Wit B, Beneken JE. Arterial haemodynamic parameters derived from noninvasively recorded pulsewaves, using parameter estimation. Med Biol Eng 1973; 11:724–731.
15 Imholz BP, Wieling W, van Montfrans GA, Wesseling KH. Fifteen years experience with finger arterial pressure monitoring: assessment of the technology. Cardiovasc Res 1998; 38:605–616.
16 Niskanen JP, Tarvainen MP, Ranta-Aho PO, Karjalainen PA. Software for advanced HRV analysis. Comput Methods Programs Biomed 2004; 76:73–81.
17 Parati G, Saul JP, Di RM, Mancia G. Spectral analysis of blood pressure and heart rate variability in evaluating cardiovascular regulation. A critical appraisal. Hypertension 1995; 25:1276–1286.
18 Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986; 1:307–310.
19 McKinley PS, Shapiro PA, Bagiella E, et al
. Deriving heart period variability from blood pressure waveforms. J Appl Physiol 2003; 95:1431–1438.
20 Carrasco S, Gonzalez R, Jimenez J, et al
. Comparison of the heart rate variability parameters obtained from the electrocardiogram and the blood pressure wave. J Med Eng Technol 1998; 22:195–205.
21 Giardino ND, Lehrer PM, Edelberg R. Comparison of finger plethysmograph to ECG in the measurement of heart rate variability. Psychophysiology 2002; 39:246–253.
22 Tordi N, Mourot L, Colin E, Regnard J. Intermittent versus constant aerobic exercise: effects on arterial stiffness. Eur J Appl Physiol 2010; 108:801–809.
23 Currie KD, Thomas SG, Goodman JM. Effects of short-term endurance exercise training on vascular function in young males. Eur J Appl Physiol 2009; 107:211–218.
24 Gerritsen J, TenVoorde BJ, Dekker JM, et al
. Measures of cardiovascular autonomic nervous function: agreement, reproducibility, and reference values in middle age and elderly subjects. Diabetologia 2003; 46:330–338.
25 Carrasco S, Gonzalez R, Gaitan MJ, Yanez O. Reproducibility of heart rate variability from short-term recordings during five manoeuvres in normal subjects. J Med Eng Technol 2003; 27:241–248.
26 Ziegler D, Laux G, Dannehl K, et al
. Assessment of cardiovascular autonomic function: age-related normal ranges and reproducibility of spectral analysis, vector analysis, and standard tests of heart rate variation and blood pressure responses. Diabet Med 1992; 9:166–175.