Analgesia Nociception Index (ANI) as a monitor of peri-operative nociception-antinociception balance in paediatric craniotomies: a prospective observational study

Analgesia Nociception Index (ANI) as a monitor of peri-operative nociception-anti-nociception balance has not been studied in paediatric neurosurgery. The objectives were to study the correlation between ANI (Mdoloris Education system) and revised FLACC (r-FLACC) score for the prediction of acute postoperative pain in paediatric population undergoing elective craniotomies and to compare the changes in ANI values with heart rate (HR), mean arterial pressure (MAP) and surgical plethysmographic index (SPI) during various time points of intraoperative noxious stimulation and before and after opioid administration. This prospective observational pilot study included 14 patients between 2 and 12 years of age undergoing elective craniotomies. HR, MAP, SPI, ANI instantaneous (ANIi) and ANI mean (ANIm) values were recorded intraoperatively and before and after opioid administration. Postoperatively HR, MAP, ANIi and ANIm, and pain scores (r-FLACC scale) were recorded. There was a statistically significant negative correlation between ANIi and ANIm with r-FLACC during the time course of PACU stay (r =  − 0.89, p < 0.001 and r =  − 0.88 and p < 0.001 respectively). Intraoperatively, in patients with ANIi values < 50, with additional fentanyl administration, there was an increasing trend in values beyond 50, which was statistically significant (p < 0.05) at 3, 4, 5 and 10 min. The trend in changes of SPI after opioid administration was not found to be significant for patients irrespective of the baseline SPI values. The ANI is a reliable tool for objective assessment of acute postoperative pain as assessed by r-FLACC in children undergoing craniotomies for intracranial lesions. It may be used as a guide to nociception-antinociception balance during the peri-operative period in this population.


Introduction
Evidence suggests the presence of moderate to severe intensity of pain following major craniotomy procedures [1,2]. Studies in paediatric population have highlighted that up to 42 percent of the children suffer at least one episode of moderate to severe pain following intracranial surgeries [3], with younger age group, longer surgery duration and occipital craniotomies being relevant risk factors [4,5]. This intensity of pain has relevance due to its propensity to trigger hypertensive episodes, anxiety and postoperative vomiting, often contributing to increased postoperative complications of raised intracranial pressure and bleeding, eventually affecting outcome [6].
Most relevant issues in the paediatric neurosurgical population are the limitations in self-reporting of pain both due to the young age or altered sensorium and under-treatment of pain due to the fear of sedative-analgesics masking neurological examination. Thus, postoperative analgesia after neurosurgical procedures remains frequently overlooked.

3
These factors highlight the need for valid and reliable objective measures of nociception-antinociception balance in the paediatric neurosurgical population, to discern the need for analgesic interventions or to gauge the efficacy of such interventions.
The current practice includes the administration of analgesics based on the hemodynamic changes. However, many confounders (hypovolemia from peri-operative use of diuretics and blood loss, wearing off of muscle relaxants, inadequate depth of anaesthesia) exist in the peri-operative period which make the use of hemodynamic parameters as markers of nociception-antinociception balance unreliable. While a gold standard for monitoring nociception in peri-operative period does not exist, reactions to nociceptive stimuli can be assessed using various commercially available devices. The Analgesia Nociception Index (ANI) has been shown to reflect the nociception-antinociception balance effectively in adult neurosurgeries [7,8]. But, its reliability in paediatric neurosurgery is unknown. With the advancing age from infancy, there is a decrease in heart rate and an increase in the heart rate variability (HRV) [9] which is related to progressive decline in the sympathetic modulation of the SA Node. Bobkowski et al. reported that the only indices which do not alter significantly with age between 3 and 18 years are those reflecting the fast oscillations (high-frequency component of HRV) [10]. These high frequencies express the parasympathetic tone variations, mainly in relation with respiratory sinus arrhythmia. Since the ANI algorithm is entirely based on high-frequency oscillations [11], it is unlikely that the patient age will significantly contribute to the derived values. Another monitor, the surgical plethysmographic index (SPI) based on the non-invasive acquisition of a plethysmographic pulse wave has been studied as a monitor of surgical stress in anaesthetised patients. Ledowski et al. have reported that the published SPI cutoff of 50 had neither any clinically relevant sensitivity nor specificity to predict acute pain in PACU in 2-16 year old patients [12]. Thus, we hypothesised that ANI can be used as a predictor of nociception and analgesia in the intraoperative and the immediate postoperative period in children aged 2-12 years undergoing elective craniotomies for intracranial lesions.
The primary objective was to study the correlation between ANI and revised FLACC (r-FLACC) for the prediction of acute postoperative pain in paediatric population undergoing elective craniotomies for intracranial lesions. The r-FLACC pain score has been proven to be valid and reliable in assessing postoperative pain in children with cerebral palsy [13], and varying degrees of motor deficits and cognitive impairment who are unable to self-report pain [14]. Since no other pain scale has been validated for use in the paediatric neurosurgical population specifically, we used the r-FLACC scale. The secondary objectives were to compare the changes in ANI values with heart rate (HR), mean arterial pressure (MAP) and SPI during various time points of noxious stimulation and before and after opioid administration.

Materials and methods
This study was designed as a prospective observational pilot study conducted over 6 months at the National Institute of Mental Health and Neurosciences, Bengaluru, India. The study was approved by the Institute Ethics Committee NIMH/ DO/IEC (BS & NS DIV)/2018-2019/30th Nov 2019).
The study population included children between 2 and 12 years of age undergoing elective craniotomy for intracranial lesions. After performing a thorough preoperative assessment of the child, informed written consent was obtained from the parent/guardian and assent for participatory decision was obtained from children aged 7-12 years, based on their comprehension.
Children planned for elective ventilation postoperatively or shifted with endotracheal tube (ETT) in situ on T-piece, children receiving α2 agonists and ASA physical status 3 and above were excluded from the study.

Study protocol
All children received premedication with intravenous midazolam 0.1 mg/kg to alleviate parental separation anxiety. In the operation theatre, standard monitoring with pulse oximetry (SPO 2 ), non-invasive blood pressure (NIBP), electrocardiography (ECG) for recording heart rate, respiratory rate (RR) and end-tidal carbon dioxide (EtCO 2 ) levels was done. Analgesic Nociceptive Index (ANI: Metro Doloris Education system) values were derived from the ECG data obtained by the sensors placed on the patient's chest wall. Surgical plethysmographic index (SPI: GE health care) was recorded from the photoplethysmogram obtained by the sensor placed on the index finger not used for the BP measurement.
The anaesthetic protocol for induction and maintenance was standardised for all patients with fentanyl 2 mcg/kg iv, thiopentone 3-5 mg/kg iv and atracurium 0.5 mg/kg iv. Anaesthesia was maintained with sevoflurane (1.8-2.2%) in air and oxygen mixture titrated to target entropy (GE Entropy™ Module) values between 40 and 60. Mechanical ventilation was titrated to achieve an EtCO2 of 35 ± 5 mmHg. Scalp block or pin site infiltration with 0.25% bupivacaine was administered prior to the Mayfield clamp application.
Fentanyl boluses were administered at the dose of 1 mcg/ kg every hourly or as deemed appropriate by the in charge anaesthesiologist as per the changes in clinical parameters (HR, MAP) and/or SPI and were not guided by the changes in ANI values. Weight appropriate dose of paracetamol was administered towards the end of surgery. All children received neostigmine 0.05 mg/kg iv and glycopyrrolate 0.01 mg/kg iv for reversal after which patient's trachea was extubated.
Postoperatively children were observed in post anaesthesia care unit (PACU) wherein monitoring of the hemodynamic variables and ANI was continued. Recovery from anaesthesia was assessed using modified observer's assessment of alertness/sedation scale (M-OAAS). Once M-OASS score reached 4 points, pain was assessed using revised FLACC (r-FLACC) scale (score 0-10). A r-FLACC score of < 4 was considered as the absence of pain or mild pain, and ≥ 4 was taken as moderate to severe pain. Children were discharged from PACU once the modified Aldrete score was ≥ 9 [15], at which point data recording was concluded.

Data collection
The demographic profile, hemodynamic variables (HR, MAP, SBP, DBP), SPI, ANIi (ANI instantaneous) and ANIm (ANI mean: obtained by a two minute averaging of ANIi) were recorded peri-operatively. The time points of measurements included baseline (before administration of anaesthetic agents), at various noxious stimuli (during intubation, pin insertion, skin incision, during craniotomy, at durotomy, at dural closure and skin closure) and every 15 min throughout the surgery.
For surgeries in prone position, new baseline values were noted at ten minutes after turning the child prone, following which the skull pins were applied. During opioid administration, the variables were recorded 5 min prior to fentanyl administration and thereafter every minute for the next 5 min and at 10th minute post-fentanyl administration. Postoperatively, haemodynamic variables, GCS, r-FLACC scores, ANIi and ANIm were recorded.

Statistical analysis
As our study was a pilot design, no formal sample size was calculated. On analysing the data from the 14 included patients, the power achieved was found to be > 90%. Data was collated on a Microsoft Excel 2007 spreadsheet and analysed using R software [16] version 3.5.2. R package "lmerTest" [17] and "rmcorr" was used for analysis and "ggplot2" [18] & "gridExtra" [19] used for graphical visualization. Linear mixed-effect models with random intercept by subject were used for observing influence of 3 sets of time points (intraoperative (IO); after opioid administration and PACU) on the variables of interest. Repeated measures correlation was used for observing correlations between variables of interest. A p value < 0.05 was used as level for statistical significance.

Results
Data were collected over 6 months from December 2019 to May 2020. Out of the total 18 patients initially screened and recruited in the study, four had to be excluded from data analysis either due to deviation from the study protocol or technical difficulties in data collection. There were 10 male and 4 female children, with average age 7 ± 3 years. The ratio of supratentorial to infratentorial pathologies was 1:1 (4 posterior fossa medulloblastomas, 3 cerebellar pilocytic astrocytomas, 2 craniopharyngiomas, 2 supratentorial gliomas, 2 choroid plexus papillomas and 1 orbito-frontal spindle cell tumour).
Supplementary Table 1 (Table S1) shows the descriptive statistics (mean ± standard deviation) of all the study variables across intraoperative study time points. Table 1 shows the correlation coefficient between intraoperative values of ANIi and ANIm with hemodynamic variables and SPI. The correlation was found to be moderately strong and negative (p < 0.001) for all variables ( Table 1). The correlation was stronger for ANIi than ANIm. Any increase in HR, MAP and SPI was associated with a reduction in the values of ANIi and ANIm.
The trend in changes of study variables (ANIi, ANIm and SPI) after administration of opioids is represented in Fig. 1. Table 2 represents the results of linear mixed-effect modelling of study variables with time and group as main effects and time*group as interaction effect. Further grouping was done based on the baseline values of respective modalities that were noted 5 min prior to opioid administration. In the nociception group, baseline value was less than 50 for ANIi and ANIm, while in the anti-nociception group, value was ≥ 50. For SPI, the nociception group was defined by baseline value > 50, while anti-nociception was defined by values ≤ 50. The main effect of time represents the average change in study variable from baseline, after accounting for differences due to group. The changes were found to be statistically significant and positive for ANIi, but not found to be statistically significant for ANIm (p < 0.05 at T10) or SPI.
The main effect of group represents the difference in study variable between the groups after accounting for change across time points. The negative estimate for main effect of group in ANIi shows that, on average, the nociception group had a lower value (− 26.623) compared to the antinociception group (p < 0.001). The difference was found to not be statistically significant for ANIm (− 12.962, p = 0.088), while the same was found to be positive and significant for SPI (estimate = 32.533, p < 0.001).
The interaction effect shows the difference in trend of change of values across time points, between the two groups. The positive estimate in the interaction effect of ANIi shows that, on average, the nociception group had larger change of ANIi from baseline compared to the antinociception group at that time point. The same was not statistically significant for ANIm and SPI at any time point (minutes after administration of opioids).
There was a statistically significant negative strong correlation between ANIi and ANIm with r-FLACC during the time course of PACU stay (r = − 0.89, p < 0.001 and r = − 0.88 and p < 0.001 respectively). A similar association was observed between HR and r-FLACC (r = 0.74, p < 0.001); however, the correlation between blood pressure variables and r-FLACC was moderate and statistically significant (Fig. 2).

Discussion
The ANI in postoperative period was strongly negatively correlated to the r-FLACC score in paediatric patients undergoing craniotomies. Thus, ANI can distinguish between mild and moderate to severe intensity pain in PACU in paediatric neurosurgical population. To the best of our knowledge, there is no previous literature commenting on the utility of ANI in identifying peri-operative nociception-antinociception balance either in the postoperative or intraoperative periods in the paediatric neurosurgical population.
Julien Marsollier et al. studied the diagnostic value of monitoring ANI in 49 non-neurosurgical children aged 2-12 years undergoing ENT, urological, abdominal and orthopaedic procedures to detect surgical simulation [20]. Thus, ANI has been reported to provide a more sensitive assessment of the nociception/antinociception balance than hemodynamic variables in paediatric surgical patients.
However, the ability of ANI to predict postoperative pain remains a controversial issue. One concern in the postoperative period is the uncertainty of whether the manufacturers' recommended cutoff values for adequate analgesia in unconscious patients are applicable in conscious patients. Ledowski et al. found only a small negative correlation between ANI and the NRS scores in adult patients undergoing non-emergency surgery. They concluded that arousal, agitation, noise, anxiety and use of inhalational over intravenous maintenance agents may be possible confounders [21]. The cutoff values for SPI and ANI in predicting postoperative pain have been quoted as 44 (sensitivity: 84%, specificity: 53%) and 63 (sensitivity: 52%, specificity: 82%), respectively for adult non-neurosurgical population [22]. O Gall et al. investigated the relationship between ANI and postoperative pain intensity in < 7-year-old and cognitively impaired children after surgical or imaging procedures (taken as control) under general anaesthesia. Their results showed that instantaneous values of ANI were lower in the surgical group than in the control group, and these values coincided with a FLACC (Faces, Legs, Activity, Cry and Consolability) score of > 4 (moderate to severe pain) in the surgical group [23].
In our institute, in the PACU, we mainly utilise the FACES pain scale and haemodynamic variables for pain assessment. Our primary objective thus was to study the performance of ANI in the PACU for postoperative pain identification. Contrasting some other studies in elective surgeries, where a single postoperative value of ANI taken at 15th minute of PACU arrival was utilised for assessment of postoperative pain [24] or ANI measured immediately before tracheal extubation was utilised to predict immediate postoperative pain [25]. We measured ANI values after complete recovery of children at every 10-min interval along with r-FLACC score and hemodynamics to minimise the effect of confounding variables. Our study protocol was similar to the one utilised by O Gall [23] in this aspect. But unlike them, we utilised the r-FLACC scale instead of the FLACC scale for postoperative pain assessment.
The relationship between ANI and SPI or haemodynamic variables in the paediatric neurosurgical population has not been studied as most previous studies either excluded neurosurgical population or included very few patients undergoing craniotomies. We found that intraoperative noxious stimuli were associated with a drop in the ANI values, with The correlation values between intraoperative MAP, HR and SPI with ANIi and ANIm were statistically significant (p < 0.001) but only moderately strong. This is similar to the findings of a study conducted at our institute in the adult neurosurgical population where the authors explain that low values of R (correlation coefficient) could be due to differing baseline BP and HR based on patient's individual sympatho-vagal balance [7]. It could also reflect that many other factors in play such as blood loss and use of vasopressor medications during the intraoperative period may confound the relationship between ANI and hemodynamic variables. Moreover, before craniotomy and removal of bone flap, raised ICP affecting hemodynamics and in turn ANI maybe a concern.
The variation of study variables around the administration of opioids yielded interesting results. We routinely utilise SPI to guide intraoperative analgesia administration at our institute. In the present study, the trend of change in SPI was not found to be significant irrespective of baseline nociception and anti-nociception (cutoff 50 for defining nociception vs. antinociception). ANIi on the other hand appeared to be a better indicator of insufficient anti-nociception than SPI in our paediatric cases. But for ANIm, it seems as if the averaging period maybe too long to reflect the almost immediate changes in the nociception/antinociception balance during a surgical procedure, as changes in ANIm require some time to reach a steady state after a noxious stimulus and treatment interventions. Weber et al. had similar results in paediatric elective surgeries where ANIi values returned to baseline within 2 min after opioid administration [26]. A study by Park et al. comparing SPI-guided analgesia with conventional analgesia in children aged 3-10 years concluded that SPI does not appear to be valid in children. They proposed that this may be due to differences in blood vessel distensibility and baseline higher heart rates in children versus adults [27]; hence, the cutoff of SPI for determination of nociception-antinociception balance in the paediatric age group may be different from adults.

Limitations
The positive findings of this study need to be confirmed in clinical studies utilising larger sample sizes and randomisation. We did not take into consideration the effect of other possible confounding variables in the intraoperative period which may affect HRV and thus ANI. We used the cutoff of 50 in our data analysis for both SPI and ANI. The validity of these values needs to be proven, as some studies have shown that a lower cutoff of SPI maybe more appropriate for paediatric population. Moreover, the cutoff value of ANI is recommended by the manufacturer in the intraoperative state, and the effect of a transitional state from controlled ventilations to spontaneous respirations may be a confounder in the postoperative period. Collective assessment of different types of surgeries can be a major limitation as the intensity of pain varies with the type of craniotomy (supratentorial or infratentorial) [28]. Although these are promising results, their extrapolation into clinical practice should be evaluated within the context of these limitations.

Conclusion
The ANI is a reliable tool for objective assessment of acute postoperative pain as assessed by r-FLACC in children undergoing craniotomies for intracranial lesions. It may be used as a guide to nociception-antinociception balance during the peri-operative period in this population. Further studies utilising ANIi for guiding administration of opioids intraoperatively and comparing the side-effect profile or total dose of opioids administered, management of pain in paediatric craniotomies in PACU and utility in children with spontaneous respirations vs. controlled ventilation may be formulated based on our results.