Background
We investigated the performance of the Perfusion Index (PI) derived from pulse oximetry waveform as a tool for assessment of anesthetic depth in comparison with A-line Autoregression Index (AAI) derived from analysis of Middle-Latency Auditory Evoked Potentials (MLAEP) waveform integrated by aepEXplus monitor in children receiving sevoflurane anesthesia for tonsillectomy.

Methods
Forty-one patients (4–12 years old) were included in this study. The PI and AAI were recorded simultaneously every minute during different stages of anesthesia delivery. The statistical tests included descriptive analysis, significance tests, correlation tests, and Receiver Operating Characteristic (ROC) curve. The AAI served as a reference.

Results
The PI significantly decreased during light anesthesia and recovery, and significantly increased during deeper planes of anesthesia, with an inverse mirror-image relationship with the AAI. A negative correlation of low to moderate degree was detected between PI and AAI during the study (p >  0.05), that reached a statistical significance at the 5th minute during sevoflurane mask induction (r = -0.457, p =  0.008). ROC analysis at an AAI < 25 extracted the best cut-off value for PI before intubation as 1.48 (AUC = 0.698 [0.537–0.859], 94.4% sensitivity, 44.5% specificity) and at 10-minute intraoperatively as 2.4 (AUC = 0.537 [0.354- 0.721], 91.7% sensitivity, 31% specificity). During recovery, at an AAI ≥ 50, the best cutoff was 1.82, (AUC = 0.661 [0.46–0.863], 100% sensitivity and 50% specificity) 2-minutes before spontaneous eye opening.

Conclusions
Compared with the AAI, the PI can track changes in depth of anesthesia in pediatric patients undergoing tonsillectomy under sevoflurane anesthesia.