Perioperative pain carries significant morbidity, including cardiovascular complications, immunosuppression, and development of chronic pain [1]. Intraoperative nociceptive stimuli are still processed centrally even in presence of deep general anesthesia [2], therefore, optimal nociception control is pivotal in anesthesia practice, as advocated by the Safetots initiative [3]. Monitoring of intraoperative nociception (neural encoding and the processing of noxious stimuli without conscious perception) has gained popularity in the recent years.

An ideal nociception monitoring would optimize the administration of intraoperative analgesics and reduce overdose side effects, infer the quality of intraoperative regional blocks, and avoid unnecessary oversedation [4]. The collection of intraoperative data of the nociceptive response would also promote research on the interaction between intraoperative nociception and postoperative outcomes [4].

The monitoring and modulation of intraoperative nociception represents a big challenge for researchers and clinicians. First, the nociceptive ascending pathways or subcortical centers can be evaluated only through complex investigations, such as functional magnetic resonance imaging [4], which is clearly unfeasible in the daily practice. Second, there are various classes of nociceptors, which respond to temperature, pressure, chemical and tissue damage stimuli [5]. Nociceptors are characterized by an all-or-none response, which is transmitted on different speeds by axons of different size and myelinization [5]. As a results, nociception pathways may not all be blocked by a single analgesic agent, such as opioids [6]. Third, there is no gold standard for intraoperative nociception monitoring, which makes the validation of any new modality to measure nociception difficult. Finally, data in pediatrics are limited, and any extrapolation from the adult literature remains artificial.

Noxious stimulus during surgery leads to a peripheral autonomic response that results in lacrimation, patient movements, tachypnea, tachycardia, and hypertension [4]. Patient’s movements, tachycardia, and hypertension are still the most common parameters used to guesstimate the level of intraoperative analgesia, and are often used as reference to validate emerging analgesia monitors [7]. However, these responses can be affected by factors not related to nociception, such as medications (i.e., muscle relaxants, beta blockers) and medical conditions (i.e., heart transplant, pacemaker, hypovolemia).

Heart rate and blood pressure poorly correlate with brain and spinal cord nociception [2]. In children under general anesthesia, maximal tetanic stimulation may lead to an increase in heart rate of only 5% [8]. Therefore, hemodynamic changes are not always and uniquely related to intraoperative nociception. When variations in heart rate and blood pressure occur, the administration of analgesic (opioids) may be appropriate (the dose is proportional to a nociceptive stimulus), inappropriate (the dose is insufficient or excessive in comparison to a nociceptive stimulus), or unnecessary (the hemodynamic changes are not secondary to a nociceptive stimulus). Consequently, the use of heart rate and blood pressure as markers of adequate analgesia has led to intraoperative opioids overuse and possible related side effects such as postoperative hyperalgesia [6].

Electroencephalogram (EEG) signal has been integrated in various nociception monitors such as the Brain Anaesthesia Response monitor (BAR, the Cortical Dynamics Ltd., North Perth, Australia), the qNOX (Quantium Medical S.L., Barcelona, Spain), and the Spectral Entropy (GE Healthcare, Helsinki, Finland) ^[9][10][9,10]. Common EEG intraoperative monitors group wave signals in four frequency bands: <4 Hz (delta), 4–8 Hz (theta), 8–13 Hz (alpha),13–25 Hz (beta) and 25–40 Hz (gamma) [11]. These clusters have been used to develop specific indexes and Density Spectral Array (DSA) outputs [12].

During general anesthesia, a noxious stimulus can trigger the beta arousal (increased power in the beta-frequency band), delta arousal (increased power in the delta band) and alpha dropout (decreased alpha power) ^[10][13][10,13]. Beta arousals typically occur during light anesthesia, contrary to delta arousals and alpha dropouts that occur during a deep anesthesia. In the EEG spectrogram these three events are visualized as: (1) an increase of warm colors (power) in the beta range, (2) an intensification of warm colors in the delta range and (3) a sudden, temporary turn from warm to cooler colors in the alpha range [10].

The main limitation of EEG is represented by its interpretation in the context of general anesthesia: while a beta arousal can be seen as a trend toward patient awakening, delta arousal and alpha dropout may be misclassified as an excessive level of hypnosis [10]. Moreover, patient’s conditions such as neurodegeneration, stroke, age, cognitive impairments alter the baseline EEG waves [10]. Intraoperative EEG has shown limitations in monitoring anesthesia depth of small children and particularly infants [12].

2.4. Functional Near-Infrared Spectroscopy (fNIRS)

Near infrared spectroscopy (NIRS) continously measures regional tissue oxigenated and de-oxyganated hemoglobin. Initially developed to monitor oxygen uptake/consumtpion of the brain, NIRS has been applied to other tissues such as kidney and has been widely used in pediatric open-heart surgery and intensive care [14]. Functional NIRS (fNIRS) measures changes in the hemoglobin oxygenation (oxygenated and de-oxygenated) as a function of cerebral activity [15]. The technology has recently been employed to investigate nociception-related brain activity in a number of diseases and conditions [16]. Changes of 0.3 mM of oxygenated hemoglobin in specific regions of the brain (i.e., somatosensory and frontopolar cortexes) have been associated to intraoperative nociceptive events [17]. Data on fNIRS and nociception are still preliminary and limited to experimental data in adults.

2.5. Spectral Entropy

Spectral entropy monitor analyzes the EEG entropy (i.e., the degree of perturbation or randomness) and the electromyography (EMG) signal to calculate the response entropy (RE) and the state entropy (SE) as a measure of intraoperative analgesia ^[12][18][19][12,18,19]. The RE is computed from a frequency range of 0.8–47 Hz and integrates both EEG and EMG signals, whereas the SE derives from an EEG frequency range of 0.8–32 Hz and represents the depth of hypnosis [18]. A difference between the two (ΔRE-SE) less than 10 was associated with a decrease of intraoperative opioid administration ^[9][18][9,18].

2.6. qNOX Index

The qNOX index is one component of the CONOX^® monitor (Fresenius Kabi, Brézins, France), which is based on the integration of an artificial neural network with a fuzzy logic system ^[20][21][20,21]. The qNOX index was developed from a Ramsay scale 5 and 6 as reference and integrated with the qCON component (developed from EEG data) [20]. The qNOX uses raw EEG and EMG signals to predict the likelihood of a response to nociception. The score is displayed on a 100-point scale (0–99), being values >60 indicative of high likelihood of nociception in adults ^[21][22][21,22]. A recent investigation, however, showed that qNOX correlated poorly (r = 0.3) with the intraoperative remifentanil infusion rate and the Analgesia Nociception Index (ANI) values [21]. No data exists in children.

2.7. Nociceptive Flexor Reflex

The electrical intensity needed to elicit a spinal polysynaptic withdrawal reflex can be used as surrogate of the level of nociception [18]. A clinically adequate, opioid-based general anesthesia (absence of somatic and hemodynamic responses to high intensity tetanic stimulations) abolishes only 59% of the spinal cord and brain nociception, which is still detectable with functional magnetic resonance imaging [2]. The Nociceptive flexor reflex (NFR or RIII reflex, Dolosys GmbH, Berlin, Germany) measures the EMG activity secondary to the electrical stimulus (on the biceps femoris muscle) and has been studied during general anesthesia ^[18][23][18,23].

2.8. Newborn Infant Parasympathetic Evaluation (NIPE) and Analgesia Nociception Index (ANI)

The newborn infant parasympathetic evaluation (NIPE, MDoloris Medical Systems, Loos, France) is a non-invasive, standardized continuous measurement of HRV. The cardiac signal is extrapolated from the electrocardiogram, and a wavelet based high pass filter over 0.15 Hz is applied in order to keep parasympathetic related variations ^[24][36]. The NIPE monitor displays two averaged measurements: the average NIPE (NIPEa) results from the average of NIPE measured over the previous 20 min, and the current NIPE (NIPEc) is calculated on a 64-s sliding window. An algorithm ^[25][37] derived from the high frequency changes of the HRV calculates a score between 0 and 100, where a score of 0 indicates minimal parasympathetic tone and maximal nociception or discomfort. Non-anesthetized infants undergoing procedural pain (heel pick) showed a median NIPEc max values of 52.5 (43.0–59.0) and 50.0 (44.5–59.0) for no/mild/moderate and severe pain, respectively ^[26][38].

The NIPE index is a modification of the ANI (Mdoloris Medical Systems, Loos, France) and was developed for infants and young children who have baseline heart rate higher than adults, resulting in a possible lower variability. Similar to NIPE, ANI expresses the relative amount of parasympathetic tone present as compared to sum of sympathetic and parasympathetic activities. The ANI Monitor displays two averaged ANI measurements: the ANIi results from the average of ANI measured over the previous 120 s, and the ANIm results from the average of ANI measured over the previous 240 s. The ANI algorithm is set for a heart rate range of 30–180 beats/min, whereas the NIPE algorithm range is between 80 and 250 beats/min. The ANI has shown a good performance in predicting intraoperative nociceptive stimuli in animals ^[27][41], adults ^[27][28][31,41] and older children ^{[8][29][30][31][32]}[8,42,43,44,45]. In adults, ANI showed 88% sensitivity and 83% specificity in predicting hemodynamic changes [22]. In children, sensitivity and specificity of ANI values ≤50 to predict intraoperative nociception (increased heart rate by 10% during skin incision) were 79% and 62%, respectively ^[29][42]. However, it has been reported that 30–50% of patients may lie in the inconclusive zone ^[30][43], suggesting that further studies are warranted.

2.9. Pupillometry

The pupillary radial muscle has a sympathetic innervation and causes pupillary dilatation, whereas the circular muscle has a parasympathetic innervation and causes pupillary constriction ^[33][34][47,48]. After a nociceptive stimulus, the sympathetic-mediated pupillary reflex dilation begins in 3 s and peaks within a minute ^[33][47]. A PRD amplitude between 13% and 25% from baseline is considered indicative of nociception in absence of hemodynamic response ^[34][48]. In children aged 2–15 years, PRD variations were more sensitive to surgical skin incision than hemodynamic changes, increasing by 160–200% contrary to 7–10% of heart rate and 5–8% of systolic blood pressure ^[35][49]. In children with burn injuries, aged 1–13 years and anesthetized with ketamine ^[36][50], the pupillary diameter increased linearly with the incrementation of the tetanic stimulations to a maximal mean dilation of 39% (±19%). It must be noted that in above studies, the baseline pupillary diameter varied from 2.3 to 3.4 mm, which may explain the variability in the PRD response (i.e., higher basal pupillary diameter, lower maximal possible variation). A PRD amplitude above 32% showed a 65% sensitivity and 77% specificity for movement response to nociception. In children aged 3–12 years, PRD-guided analgesia (PRD between 5–30%) was associated with a 25% decrease in remifentanil consumption compared to blood pressure-guided analgesia (defined as changes of ±20% from the baseline) ^[37][51]. The PPI can be measured through an infrared videopupillometer applied over the orbit with the aid of an opaque silicon cylinder ^[33][47] while a tetanic stimulation (200-microseconds, 100 Hz) is delivered on the patient ulnar nerve. From a starting intensity of 10 mA, each step consists in stimulations increased by 10 mA up to 60 mA, after which the intensity remains constant and the duration is prolonged by 1 s for a maximum of 3 s ^[38][39][52,53]. An PRD increase of 13% determined the PPI ^[38][39][52,53]. The PPI ranges from 1 to 10, being 10 maximal pupillary reactivity ^[33][47], and a PPI > 7 suggests insufficient analgesia [18]. 

2.10. Skin Conductance

The Skin Conductance Algesimeter (SCA, MedStorm innovations, AS, Oslo, Norway) aims to measure the skin conductance caused by rapid micro-fluctuations of water permeability at the level of the palms (or soles), as the sweat glands in these regions are exclusively innervate by the sympathetic system ^[22][33][22,47]. Nociception, by increasing the sympathetic tone (i.e., increasing sweat), leads to an increase of frequency and amplitude of the skin conductance ^[33][47]. Skin conductance is measured as number of skin conductance fluctuations per second (NFSC) [22], which have the advantage of fast response (<2 s) and short duration (<0.7 s) ^[40][54]. In adults, it correlates with intraoperative blood pressure and plasma catecholamines concentrations, but not with opioid administration ^[33][47].

2.11. Surgical Pleth Index (SPI)

Similar to palmar sweat glands, distal arterioles exclusively contain sympathetic alpha₁ receptors ^[33][47]. The nociception-induced vasoconstriction results in a change of blood flow wave amplitude that can be measured by photoplethysmography ^[33][47]. The Surgical Pleth Index (SPI, GE healthcare, Helsinki, Finland) derives from the former Surgical Stress Index (SSI). The SPI combines the normalized photoplethysmography wave amplitude (PPGA_norm) and the normalized heartbeat interval (HBI_norm) into an algorithm that displays SPI values on a 100-point scale, being 100 maximum nociception ^[33][41][46,47].

2.12. Nociception Level (NoL) Index

Photoplethysmographic pulse wave, galvanic skin conductance, accelerometer, and peripheral temperature have been integrated in the Nociception Level Index (NoL, Medasense, Ramat Gan, Israel) ^[33][47]. From these four parameters, a number of derivates are extrapolated, namely the photoplethysmographic waveform amplitude (PPGA), high frequency band HRV (HRV-HF) power, number of skin conductance fluctuations (NSCF) and others ^[42][63]. Currently, NoL is the only 4-parameter monitor on the market. So far, NoL has been not investigated in pediatrics, mainly because the apparatus resides in finger probe, which has not been designed for children. In comparison with adults, children have a different vascular elastance which affects the photoplethysmographic wave pattern and its derivates ^[43][64]. Without a proper validation, the use of current NoL algorithm may remain inaccurate in pediatrics.