Pain assessment is important when providing analgesia, but is still currently a challenge, particularly in noncommunicating patients. Self-rating pain assessment by a numerical rating scale (NRS) or visual analogue scale is considered good clinical practice in alert patients. For sedated and anaesthetised patients, several methods of pain assessment, including heartbeat intervals, plethysmographic pulse wave amplitude, surgical stress index, skin conductance, pupillary response and electroencephalographic indices, have recently been investigated.1–9 However, mean arterial blood pressure (BP), heart rate (HR) and heart rate variability (HRV) do not correlate with NRS scores, and are neither sensitive nor specific for pain.10 Therefore, children and patients with limited ability to communicate such as those in the ICU, under sedation, suffering from dementia, cerebral ischaemia, coma and cognitive impairment may not be able to cooperate and, as a result, may receive inadequate analgesia.
A promising noninvasive technique is the continuous monitoring of HRV transformed into an analgesia nociception index (ANI, 0–100), which assesses parasympathetic activity as a possible measure of nociception.11–13 Logier described this novel technique of HRV measurement, whereas Jeanne and others defined a derived ANI.12–16 The ANI provides greater stability than raw indices of HRV. It seems to be ideal for assessing pain and its clinical use has been demonstrated in patients under general anaesthesia.12,13,17,18 ANI monitoring seems to be sensitive when detecting moderate nociceptive stimuli during propofol anaesthesia and may be helpful in terms of optimising remifentanil administration.19 However, protocol-driven intraoperative analgesia guided by ANI monitoring did not reduce postoperative pain after elective laparoscopic cholecystectomy.20 Recent evidence concerning the efficacy of ANI in conscious patients for the assessment of acute postoperative pain and pain during labour is also inconclusive.21–24 Heterogeneous study populations, different kinds of acute pain, individual differences and several confounding factors in the environment (e.g. anaesthetic recovery room) cloud objective results.
To our knowledge, the robustness of ANI as a tool for pain assessment has never been investigated in a standardised experimental pain model. We designed a prospective study with randomised expected and unexpected electrically induced pain, nonpainful and sham stimuli. Our primary objective was to assess pain intensity in conscious volunteers using ANI. We focussed on magnitudes, descending slopes and time courses of ANI and expected a negative correlation between ANI and NRS.
This single-blinded, randomised crossover study was conducted with volunteers who gave written, informed consent before data collection. The study protocol was approved by the Ethics Committee of the Faculty of Medicine, Ruhr University, Bochum, Germany; No. 4501-12; date of approval: 23 October 2012. The experimental procedure was in accordance with the Declaration of Helsinki.
To avoid sex differences, 20 male students were consecutively recruited to the study through local advertising. All of the participants were right handed and healthy. Volunteers with disorders, medical consultation (<6 months), regular medication or recent exposure (<3 months) to medications for analgesia or affecting HR, cardiac rhythm or presenting with anything other than sinus rhythm were excluded.
Heart rate variability
HRV is related to autonomic nervous system activity.14 Real-time analysis of HRV has been performed using wavelet transformation to evaluate pain.25 Growing evidence demonstrates that pain results in a decrease of HRV, in particular of high-frequency (0.15–0.4 Hz) power, indicating a drop of parasympathetic tone during unpleasant stimuli or emotions.26–28 To prevent artefact-induced, inaccurate measurement of R-R intervals (time between two QRS complexes in an ECG), an effective filtering algorithm has been developed.15 Recording R-R intervals via ECG allows observation of changing patterns of respiratory sinus arrhythmia in relation to painful and nonpainful stimuli. To stabilise and simplify HRV data, these can be transformed into an ANI. This is based on measurement of the respiratory influence on the R-R interval in the ECG and allows qualitative and quantitative measurement of HRV using the ANI Monitor (MDoloris Medical Systems, Lille, France). This displays a graph using an index of 0–100 to provide an actual measure of ANI.11 An index of 100 reflects low HRV and high parasympathetic activity. HRV increases during sympathetic stimulation, and the effect of respiration on the R-R interval can be used to assess sympathetic tone, and therefore the analgesia-nociception balance.12,13
All volunteers participated in a single session. They lay alone supine in a darkened and silent room and were asked to keep their eyes closed. Two single-use ANI sensors were placed in the middle of the sternum and mid-clavicle on the left chest, and BP and HR were determined (pre-BP, pre-HR). A self-made electrode consisting of 12 punctate pins was fixed on the right ventral forearm 5 cm distal to the cubital fossa so that a painful electrical stimulus of 2 mA in intensity could be applied for 5 s at 100 Hz with a constant current stimulator (DS7A, Digitimer, Welwyn Garden City, UK).
The session started with a baseline measurement for 2.5 min without any disturbance (−150 to 0 s, prebase 1). Then, four different stimuli were applied (Fig. 1). All participants were blinded to our hypothesis, study order and stimuli. They were told that all four identical, painful stimuli would be applied after an announcement, and that they would be asked to rate their pain level after each stimulus on a NRS between 0 and 10 (0 = no pain, 10 = most intense pain imaginable). For each participant, however, the first stimulus was an electrical, unexpected painful stimulus (UPS). The next was an electrical, expected painful stimulus (EPS), a neutral, nonpainful stimulus (NPS) or a placebo stimulus. All subsequent stimuli were given in a previously generated computer-based random order (six different order options) at ‘0 s’, each following a repeated announcement, within a latency time of 30 s. EPS was set to have identical intensity to UPS, but this time it was expected by the participant. NPS consisted of a spray of disinfectant on the ventral forearm, and after announcement of placebo stimulus, no stimulus was given at all. Each stimulus was followed by a recovery time of 5 min (0–150 s observation, followed by a new baseline for 150 s for the next stimulus (postbase = prebase 2). After the last stimulus and a last recovery period, BP was checked again (post-BP). Observational periods and durations of the sessions were determined after five pilot measurements under comparable study conditions.
Data and statistical analysis
The ANI Monitor recorded HRV continuously and returned ANI values. An averaged ANI was recorded every second. To simplify, we reported ANI means of predetermined different time intervals of at least 30 to a maximum of 150 s. Each stimulus (UPS, EPS, NPS or placebo) was set at ‘0 s’ and the time count started again. Observation periods for ANI monitoring were determined before and after applying a stimulus. There were two different baselines prior to the stimuli. Prebase 1 was overall base at the beginning of all measurements prior to UPS at −150 to 0 s. Prebase 2 was prior to EPS, NPS or placebo stimulus at −180 to −30 s. Prebase 1 was used as a reference baseline for all calculations. The announcement was given 30 s before the stimuli (announcement: −30 to 0 s). After each stimulus, the observational period was divided into different time intervals of about 30 s at 30, 60, 90, 120 and 150 s. A baseline was recorded again at postbase 150–300 s. Postbase and prebase 2 could be identical before and after a different stimulus. End base was defined as the last baseline after all measurements (Fig. 1).
Quantitative data (ANI means, amplitudes, maxima, minima and slopes) are presented as mean ± standard deviation, median with interquartile range or mean difference with 95% confidence interval (CI). The amplitudes were measured between the maxima and minima of the ANI within the particular intervals. ANI slope for comparing ANI time courses was set between two ANI values at 0, 30, 60, 90, 120 and 150 s. As no comparable studies are available, the number of participants was based on pragmatism. Statistical Package for Social Sciences software (Version 22, IBM, Chicago, Illinois, USA) was used for analysis. Data were checked for normal distribution before performing analyses by using the Shapiro–Wilk test. As the ANI permits the comparison of differences of values and has a fixed zero value, it is defined as a ratio scale. For the comparison of NRS values (UPS versus EPS) and time in seconds until reaching ANI minima following a painful or nonpainful stimulus, Wilcoxon signed rank or paired Student t tests were used. Quantitative ANI data were first analysed using a two-way analysis of variance (ANOVA) for repeated measurements to test only for main effects [‘stimulus’ (UPS, EPS, NPS, placebo stimulus) and ‘time’ (different time intervals)]. Post-hoc one-way ANOVA for repeated measurements was performed to compare ANI values across different stimuli and for comparison across ‘time intervals for each stimulus’. Tests for sequence and sequence–stimulus interactions as part of the crossover analysis were performed. Greenhouse–Geisser and Bonferroni-adjusted P values are reported as appropriate. The Spearman rank correlation coefficient was used to assess the associations between NRS values and time in seconds to reach ANI minima versus ANI minima. UPS and EPS were grouped as painful, whereas NPS and placebo stimulus were grouped as nonpainful stimuli. Planned subgroup analyses were performed for volunteers rating UPS or EPS at NRS less than 3 and NRS of at least 3 (clinically relevant pain). A two-sided P value less than 0.05 was considered statistically significant.
All 20 participants aged (mean ± standard deviation) 24.2 ± 1.9 years were included in the data analysis. No adverse effects from the ANI electrodes or the painful electrical stimuli were reported. There were no significant differences between baseline and postbase data in HR or BP (Table 1).
NRS scores were significantly (P = 0.008) higher when the participants expected the pain stimulus (EPS: 5.25; 3.5–6.75, UPS: 4.5; 3–5) despite equal stimulus intensity. Only five participants evaluated the NPS with a rating of more than zero (0.00; 0.00–0.75). All participants assessed the placebo stimulus as zero (Table 1).
Time courses of analgesia nociception index
Resting alert volunteers exhibited ANI values of 82.1 ± 10.7. Prebase 1, prebase 2 (81.6 ± 10.1), postbase (81.9 ± 9.5
) and end-base values (82.8 ± 7.9) were similar (P = 0.94). There were no significant effects of stimulus or time (P = 0.20) and crossover tests of sequence with sequence × time were also not significant (P = 0.57). One-way ANOVA showed that differences in ANI values between UPS, EPS, NPS and placebo stimulus were not significant (P = 0.18), but there was a significant effect for different time intervals (P < 0.001).
In general, ANI scores were lower after each stimulus. However, ANI dropped significantly within the first 2 min after each stimulus (P < 0.001). ANI minima following painful (UPS and EPS) and nonpainful (NPS and placebo) stimuli were reached similarly (P = 0.95) after 78.9 ± 21.3 and 79.3 ± 31.6 s, respectively. After the announcement of all stimuli, the ANI mean showed no change at −30 to 0 s (P = 0.24). A significant drop of ANI directly after UPS, EPS, NPS and placebo stimuli occurred with a latency of at least 30 s (Table 2). After UPS, EPS, NPS and placebo stimuli, the ANI mean demonstrated a maximal decrease of 25.0 ± 7.3% at the time interval of 60–90 s compared with prebase 1 (UPS = 32.6%, EPS = 29.2%, NPS = 21.5%, placebo stimulus = 16.6%). All ANI values returned to baseline after a resting phase of 120–300 s following a stimulus (Fig. 2). In addition, regarding ANI minima, a significant decline was observed during each stimulus course (P < 0.01, Table 3). Maxima and amplitudes showed comparable results.
ANI values at the time intervals of 0–30 (P = 0.70), 30–60 (P = 0.13), 90–120 (P = 0.20) and 120–150 s (P = 0.57) did not differ between the stimulus settings. During 60–90 s, only UPS versus placebo stimulus [−9.7 (95% CI: −18.4; 1.0), P = 0.023] significantly differed but not UPS versus NPS, EPS versus placebo stimulus, EPS versus NPS or NPS versus placebo stimulus. The ANI minima during the various measurements only showed significant differences between UPS and placebo stimulus at the interval of 60–90 s [−8.8 (95% CI: 17.7; 8.6), P = 0.025].
Within the time courses of all stimuli settings, a significant change of slope was detected (P < 0.001). The sharpest decline was observed at 30–60 s for UPS, EPS and NPS (−0.34 ± 0.252). The sharpest rise was seen at 120–150 s for UPS, NPS and placebo stimuli (0.33 ± 0.30). There was no significant difference of ANI slope between stimuli conditions (P = 0.79).
There was no correlation between ANI minima and NRS of UPS (rs = −0.096, P = 0.69) or EPS (rs = −0.05, P = 0.84) in the interval 60–90 s. A scatterplot summarises the results combining UPS and EPS (Fig. 3). Furthermore, there was no correlation between ANI slope in the time interval of sharpest decline and NRS values of painful stimuli (rs = −0.09, P = 0.60). There was no correlation between the time to reach ANI minima and NRS of painful stimuli (rs = −0.01, P = 0.97).
In this study, HRV calculated by ANI was used to assess pain intensity in a standardised pain experiment conducted with alert volunteers. ANI values decreased after the application of a random stimulus, but did not allow a specific differentiation of painful, nonpainful or sham stimuli. There was no negative linear relationship between ANI and self-rated pain, as described previously.21,22,24
Recently, objective pain assessments using ANI have been investigated in alert patients, but conflicting results have been shown to date.9 In an observational study, ANI measurements seemed to be in agreement with pain and analgesia evaluations during and after general anaesthesia for total knee replacement.11,29 Meanwhile, a negative linear relationship between ANI and NRS has been observed during uterine contractions in labouring women before initiation of epidural analgesia.24 In addition, Boselli et al.22 showed that the assessment of ANI during the immediate postoperative period after general anaesthesia was significantly negatively correlated with pain intensity in the postanaesthesia care unit. Furthermore, the measurement of ANI immediately before extubation was significantly and negatively related to pain intensity on arrival in the postanaesthesia care unit after general anaesthesia using an inhalational agent and remifentanil.30 In contrast, an Australian investigation after surgery did not show any differences in ANI scores between different levels of pain, although lower ANI scores were found in ‘severe’ versus ‘no’ pain.23 In addition, ANI was significantly higher with deep sedation compared with full consciousness, whereas after a bolus of fentanyl, ANI scores did not significantly differ.23 Such contradictory results might have been caused by the interactions and effects of opioids and inhalation anaesthesia.21,30 Heterogeneous study populations, different kinds of acute pain, individual aspects and several confounding factors in the environment (e.g. anaesthetic recovery room) cloud objective results.
In the present study, NRS values were higher after the expected pain stimulus. In particular, the influence of pain expectation is likely to be very important in this kind of experimental setup.31 From an ethical point of view, we had to inform the participants about triggering electrical pain stimuli. That alone may cause a certain expectation. Nevertheless, to be able to study the issue of ANI after unexpected and expected pain, the first stimulus was administered without warning. This may have created impaired attention and thus cortical involvement resulting in avoidance of pain. The latter is potentially affected by not warning, and EPS might be higher in reality.
Low ANI values should demonstrate relatively low parasympathetic tone as a surrogate for nociception. Thus, in our opinion, the minimum of the ANI is the ideal measure for correlation with pain intensity. To evaluate the relationship between pain sensation and decreased ANI, we compared the ANI minima during the progress of all intervals but did not observe the expected negative correlation. Mean ANI minima after painful stimuli were 55.6 (UPS) and 56.4 (EPS) versus NRS means of 4.3 and 5.0, which are moderate pain intensities. The minimum value of ANI obtained in this study was similar to that observed after surgery. Boselli et al. concluded that an ANI value of less than 50 at arousal from general anaesthesia was predictive of moderate-to-severe pain (NRS > 3)30 and that a value higher than 48 predicted that 99% of patients did not have severe pain in the immediate postoperative period.22 However, as there are no convincing differences in our ANI values for nonpainful or sham stimuli, those predictions are not relevant for detecting nociception in our study group. To assess the progress of ANI scores more precisely, we analysed the slopes and amplitudes of all intervals within all four stimulations. However, again, no conclusive differences were found between the different stimuli. During general anaesthesia, hypnotics (especially halogenated) and opiates are likely to modify the gain and slope of the baroreflex. This may explain differences in patients under general anaesthesia and awake patients on the importance of changes in the R-R interval and its relationship with the NRS scores.
An overall ANI baseline in alert participants seems to be around 82; ANI values of 100 were not reached. The robustness and informative value of ANI appears to be limited in conscious patients. Even in healthy volunteers, ANI is not adequate to assess pain intensity. ANI responds equally to pain and other environmental stimuli. Limitations presented by the manufacturer (e.g. arrhythmia, apnoea, medications) and factors that are known to increase sympathetic activity (e.g. noises, BP measurement, anxiety, agitation) can influence ANI in alert patients. We chose a uniform participant group with no sedatives or analgesics and used a standardised study protocol, so further confounding factors such as age, level of consciousness, changing autonomic or haemodynamic conditions, preexisting cardiac conditions, affecting medications or narcotics could be disregarded. In our study, confounding factors like abrupt movement, variability of respiratory pattern caused by agitation or application of painful stimulation cannot be neglected. A further limitation of our small sample size should be mentioned. A post-hoc power analysis calculated using G*Power (Version 3.1) for the nonsignificant repeated measures ANOVA with ‘stimulus’ factor showed a power of 83% for an estimated F = 0.25 as the effect size. Assessment of more intense pain intensities and prolonged duration of pain would give additional information about the usefulness of ANI. Therefore, another limitation of this study is a limited consideration of the whole range of pain.
The announcement of a painful stimulus had no impact on the time courses of ANI. Regardless of stimulus quality, there was a latency of ANI decrease of at least 30 s. Nevertheless, the pain intensity caused by electrical pain could not be objectively predicted from ANI values, nor could a convincing differentiation be detected for real, painful stimuli (UPS and EPS), nonpainful (NPS) and placebo stimuli. Finally, when there was a small differentiation between the painful and the placebo stimuli, no differentiation could be demonstrated between NPS and placebo stimulus. Although the curves of placebo stimulus and NPS were flattened compared with EPS and UPS, ultimately, there was no conclusive difference between the two. These results are consistent with investigations by De Jonckheere et al.32,33 who proved that ANI could be a good indicator of parasympathetic changes in emotional situations.
In conclusion, we have shown for the first time that the measurement of HRV derived by ANI did not allow a differentiation of painful, nonpainful or sham stimuli in alert volunteers. Under standardised experimental conditions without the influence of sedatives and analgesics or confounding factors, ANI was not specific when it came to assessing pain intensity in alert volunteers. For the future, ANI should not be considered as an objective tool to differentiate or quantify pain intensity without NRS and should only be cautiously interpreted when optimising pain management in patients with a limited ability to communicate. Pain assessment and optimisation of analgesic therapy remain a challenge.
Acknowledgements relating to this article
Assistance with the study: none.
Financial support and sponsorship: the ANI-Monitor (MetroDoloris, Lille, France) was provided by the manufacturer for the duration of the investigation. Financial support was not supplied. Expendable materials and proband fees were obtained from institutional sources.
Conflicts of interest: none.
Presentation: preliminary data for this study were presented as a poster presentation at the German Society of Anaesthesiology and Intensive Care Medicine ‘DAC 2013’, 20–23 April 2013, Nürnberg, Germany, and presented at ‘DAC 2015’, 7–9 May 2015, Düsseldorf, Germany.
1. Bonhomme V, Uutela K, Hans G, et al. Comparison of the surgical Pleth IndexTM
with haemodynamic variables to assess nociception-antinociception balance during general anaesthesia. Br J Anaesth
2. Gjerstad AC, Storm H, Hagen R, et al. Skin conductance or entropy for detection of nonnoxious stimulation during different clinical levels of sedation. Acta Anaesthesiol Scand
3. Gjerstad AC, Storm H, Hagen R, et al. Comparison of skin conductance with entropy during intubation, tetanic stimulation and emergence from general anaesthesia. Acta Anaesthesiol Scand
4. Barvais L, Engelman E, Eba JM, et al. Effect site concentrations of remifentanil and pupil response to noxious stimulation. Br J Anaesth
5. Bonhomme V, Llabres V, Dewandre PY, et al. Combined use of Bispectral Index and A-Line Autoregressive Index to assess antinociceptive component of balanced anaesthesia during lumbar arthrodesis. Br J Anaesth
6. Takamatsu I, Ozaki M, Kazama T. Entropy indices vs the bispectral index for estimating nociception during sevoflurane anaesthesia. Br J Anaesth
7. Ledowski T, Bromilow J, Paech MJ, et al. Monitoring of skin conductance to assess postoperative pain intensity. Br J Anaesth
8. Ledowski T, Ang B, Schmarbeck T, Rhodes J. Monitoring of sympathetic tone to assess postoperative pain: skin conductance vs surgical stress index. Anaesthesia
9. Cowen R, Stasiowska MK, Laycock H, Bantel C. Assessing pain objectively: the use of physiological markers. Anaesthesia
2015; doi: 10. 1111/anae.13018.
10. Ledowski T, Reimer M, Chavez V, et al. Effects of acute postoperative pain on catecholamine plasma levels, hemodynamic parameters, and cardiac autonomic control. Pain
11. Logier R, Jeanne M, De Jonckheere J, et al. PhysioDoloris: a monitoring device for analgesia /nociception balance evaluation using heart rate variability analysis. Conf Proc IEEE Eng Med Biol Soc
12. Jeanne M, Logier R, De Jonckheere J, Tavernier B. Heart rate variability during total intravenous anesthesia: effects of nociception and analgesia. Auton Neurosci
13. Jeanne M, Logier R, De Jonckheere J, Tavernier B. Validation of a graphic measurement of heart rate variability to assess analgesia/nociception balance during general anesthesia. Conf Proc IEEE Eng Med Biol Soc
14. 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. Eur Heart J
15. Logier R, De Jonckheere J, Dassonneville A. An efficient algorithm for R-R intervals series filtering. Conf Proc IEEE Eng Med Biol Soc
16. Logier R, Jeanne M, Tavernier B, De Jonckheere J. Pain/analgesia evaluation using heart rate variability analysis. Conf Proc IEEE Eng Med Biol Soc
17. Huhle R, Burghardt M, Zaunseder S, et al. Effects of awareness and nociception on heart rate variability during general anaesthesia. Physiol Meas
18. Gruenewald M, Ilies C, Herz J, et al. Influence of nociceptive stimulation on analgesia nociception index (ANI) during propofol-remifentanil anaesthesia. Br J Anaesth
19. Jeanne M, Clément C, De Jonckheere J, et al. Variations of the analgesia nociception index during general anaesthesia for laparoscopic abdominal surgery. J Clin Monit Comput
20. Szental JA, Webb A, Weeraratne C, et al. Postoperative pain after laparoscopic cholecystectomy is not reduced by intraoperative analgesia guided by analgesia nociception index (ANI) monitoring: a randomized clinical trial. Br J Anaesth
21. Boselli E, Jeanne M. Analgesia/nociception index for the assessment of acute postoperative pain. Br J Anaesth
22. Boselli E, Daniela-Ionescu M, Bégou G, et al. Prospective observational study of the noninvasive assessment of immediate postoperative pain using the analgesia/nociception index (ANI). Br J Anaesth
23. Ledowski T, Tiong WS, Lee C, et al. Analgesia nociception index: evaluation as a new parameter for acute postoperative pain. Br J Anaesth
24. Le Guen M, Jeanne M, Sievert K, et al. The Analgesia Nociception Index: a pilot study to evaluation of a new pain parameter during labor. Int J Obstet Anesth
25. Pichot V, Gaspoz JM, Molliex S, et al. Wavelet transform to quantify heart rate variability and to assess its instantaneous changes. J Appl Physiol
1999; 1985: 86:1081–1091.
26. Miu AC, Heilman RM, Miclea M. Reduced heart rate variability and vagal tone in anxiety: trait versus state, and the effects of autogenic training. Auton Neurosci
27. Demaree HA, Robinson JL, Everhart DE, Schmeichel BJ. Resting RSA is associated with natural and self-regulated responses to negative emotional stimuli. Brain Cogn
28. Appelhans BM, Luecken LJ. Heart rate variability and pain: associations of two interrelated homeostatic processes. Biol Psychol
29. Jeanne M, Delecroix M, De Jonckheere J, et al. Variations of the analgesia nociception index during propofol anesthesia for total knee replacement. Clin J Pain
30. Boselli E, Bouvet L, Bégou G, et al. Prediction of immediate postoperative pain using the analgesia/nociception index: a prospective observational study. Br J Anaesth
31. Bingel U, Wanigasekera V, Wiech K, et al. The effect of treatment expectation on drug efficacy: imaging the analgesic benefit of the opioid remifentanil. Sci Transl Med
32. De Jonckheere J, Rommel D, Nandrino JL, et al. Heart rate variability analysis as an index of emotion regulation processes: interest of the Analgesia Nociception Index (ANI). Conf Proc IEEE Eng Med Biol Soc
33. De Jonckheere J, Logier R, Jounwaz R, et al. From pain to stress evaluation using heart rate variability analysis: development of an evaluation platform. Conf Proc IEEE Eng Med Biol Soc