This prospective, observational cohort study (NCT02481999) was approved by our local ethics committee on 12th March 2015 (Ethics committee Charité, University Medicine of Berlin / Chairperson Prof. Dr. R. Seeland / EA2/027/15). Written informed consent of either parents or legal proxies was obtained according to the Declaration of Helsinki. We included children aged 0.5 – 8 years, undergoing elective surgery with a planned duration of > 60min at our University hospital – Charité / Berlin. Between September 8, 2015 and February 22, 2017 parents or legal proxies of the child were approached by study staff members during stay in the preoperative evaluation center on the day prior to surgery. Exclusion criteria comprised any history of neurological or psychiatric diseases, any signs of delayed development in the child, isolation required because of multiresistant bacteria, inability of the parents to speak, read or understand the German language, as well as concurrent enrollment in another study.
The a priori primary outcome was ED assessed during stay in the recovery room with the Pediatric Assessment of Emergence Delirium (PAED) Score (15). Values of at least 10 were considered as ED.
Secondary outcomes related to frontal EEG recordings were depth of anaesthesia and burst suppression duration.
Oral premedication with midazolam was administered in all children. Children were anaesthetised with either propofol or sevoflurane as decided by the anaesthesiologist in charge. Standard monitoring included non-invasive blood pressure, electrocardiogram and pulse oximetry. Patients received i.v. propofol (assessed as mg kg-1 body weight) or mask induction / maintenance with sevoflurane (assessed as endtidal concentrations et Vol %). Dosage of sevoflurane and propofol was given according to clinical needs and chosen by the anaesthesiologist in charge. Remifentanil was administered as analgesic agent during the induction period in all children, according to clinical needs. If muscle relaxation was needed, cis-atracurium was administered adapted to body weight. Before the end of surgery children received metamizol or paracetamol and / or piritramid for analgesia. Some children also received regional anaesthesia. The complete anaesthetic procedure and medication were outside of the scope of this study.
ED assessment
ED was assessed according to the PAED Score. From admission to the post-anaesthesiological care unit until discharge the PAED score was determined 1min after extubation, at arrival in the recovery room, 5min, 10min, 15min, 30min, 45min, and 60min after arrival and at discharge from the recovery room by a member of the study staff sitting next to the child. Values > 10 were considered as an emergence delirium (15). The “Faces Legs Activity Cry Consolability Pain Scale” was used to determine pain events (16). Richmond Agitation Sedation Scale score was used to assess the level of consciousness (17). PAED score was only included in the analysis if the Richmond Agitation Sedation Scale score was above -2 and if it was unlikely that pain was triggering agitated behavior. If inadequate behavior during the stay in the recovery room improved after pain-medication these periods of agitated behavior were not classified as ED. If a member of the study team was unable to take the PAED score in the recovery room the child was excluded from further evaluation.
EEG recording and analysis
Bi-frontal EEGs were obtained with the Narcotrend Monitor (MT Monitor Technik, Bad Bramstedt, Germany). The EEG was recorded continuously from baseline before start of anaesthesia until the end of stay in the recovery room. After skin preparation with alcohol four electrodes (Ambu BlueSensor, Bad Nauheim Germany) were placed on the patients’ forehead at positions Fz, F7 and F8, with a reference electrode at Fp2. The impedances were kept below 8kOhm, differences between electrodes were less than 2kOhm. During the EEG recording event markers comprising “start of anaesthesia”, “intubation”, “operation” and “extubation” and “recovery room” were placed. “Start of anaesthesia” was defined as the time-point when the anaesthesiologist began to administer the anaesthetic agent, i.e., either sevoflurane, propofol or a mixture of both. “Intubation” was defined as the time point, when the anaesthesiologist in charge began to intubate the child. “Operation” indicated a time point within 15 min to 30 min after intubation of stable surgery and anaesthesia, without severe pain events or intraoperative bolus application of propofol. “Extubation” was defined as the time point, at which the anaesthesiologist in charge extubated the child.
Raw EEG data were recorded with a high pass filter of 0.5 Hz and a low pass filter of 45 Hz, sampling rate was 128 / sec. Visual EEG analysis (EEG viewer software: 50µV-100µV and 1s/div) was performed from the time point of “start of anaesthesia” until “extubation”. The raw EEG was analysed by an expert (S. K., neurologist with specialization in clinical neurophysiology and EEG) blinded to the ED outcome, the medication patients received and further clinical data. The presence of Burst Suppression periods was validated by a second expert (C.P. pediatric neurologist with specialization in electroencephalography in children). Short periods (< 5 min within the total anaesthesia procedure) with artifacts (muscle, eyelid, and electricity) were excluded from any further analysis. If persistent artifacts or repeating artifacts were seen, these EEGs were excluded from further analysis.
Burst Suppression periods were assessed by visual inspection of the raw EEG (EEG viewer software: 50µV and 1s/div). Burst Suppression segments were included if duration of isoelectric line exceeded 0.5 seconds. Duration of intraoperative Burst Suppression was calculated from start of the first isoelectric line segment until the end of the last isoelectric line. Isoelectric line is classified as an EEG activity below 5µV (18). We calculated the duration of the isoelectric line activity by distracting the time duration of the intermittent burst activity. Duration of the isoelectric line is the sum of all periods of isoelectric line present in the raw EEG. Additionally, we calculated the Burst Suppression strength by dividing total isoelectric line duration by Burst Suppression duration. The Burst Suppression strength indicates the fraction of the isoelectric line within the Burst Suppression pattern, a prolonged duration of an isoelectric line will show a Burst Suppression strength value tending towards 1, which indicates a deeper stage of coma or a deeper level of anaesthesia.
Statistical Analysis
The present study was designed as a prospective, observational study. Statistical analysis was performed using SPSS, version 25, copyright SPSS, Inc., Chicago, IL 60606, USA. Data are presented as means + SD, in case of unbalanced data distribution as medians (IQR 25/75) or as frequencies (%). For nominal data statistical analysis was performed with the Chi-square test from Pearson. Numerical data were analysed using the Mann-Whitney-U test for non-parametric data. To determine the impact of age, anaesthetic medication, Burst Suppression and depth of anaesthesia on the incidence of ED we performed a univariate logistic regression. Odds Ratio with 95% confidence intervals and corresponding p-values were calculated for each risk factor. The Spearman Rho correlation test was used to analyse the correlation between age (in months) and Burst Suppression duration, isoelectric line duration and Burst Suppression strength. Statistical significance was assumed at p<0.05.
The statistical analysis plan was made prior to data assessment. To calculate the sample size needed we postulated an ED incidence of 10.5% (from a pilot study in our department 2013 (NCT02358278), resulting in a 10.5% incidence of ED in children aged 0-14 years) and an increased risk to develop ED with an increase in depth of anaesthesia and the occurrence of Burst Suppression periods. 470 patients were initially planned to be included in this study. An interim analysis was planned (after approximately 1/3 – 1/2 of the total planned sample size) with recalculation of the initial sample size calculation to adopt the study procedure, if the initial assumptions differed strongly. At the interim analysis the ED incidence rate was distinctly higher, which is most likely related to the fact that we mainly included younger patients in our actual study compared to the pilot study 2013 in our department. We re-ran the sample size calculation taking into account the new incidence rate of 41%. Initially, the study was planned for an odds ratio of 1.6 with an ED incidence rate of 10.5% and a R2 of the other covariates of 0.2 resulting into a sample size of 470 children with a power of 80%. Considering the new incidence rate of 41% and without changes of the other parameters we achieve a power of 80% with 97 children and an odds ratio of 2.1.