Agreement between mandibular and frontal patient state index values during general anaesthesia: A prospective observational study

INTRODUCTION Processed electroencephalography for depth monitoring is usually obtained from frontal electrodes; however, forehead access is often restricted in neurosurgical and head-and-neck settings. Therefore, alternative placements have been explored. For SedLine® (Masimo Corp. , Irvine, CA, USA) /Patient State Index (PSI), a validated nasal/infraorbital montage has shown high within patient agreement with the frontal montage and clinical feasibility when the forehead is unavailable. 1 In contrast, most evidence on mandibular placements arises from bispectral index (BIS) studies and suggests larger, inconsistent discrepancies relative to the frontal reference. 2–7 Direct, prospective evidence for PSI at the mandible has been lacking. We prospectively compared simultaneous mandibular (MAND) and frontal (HEAD) PSI during general anaesthesia. Our novelty was two-fold: First, to our knowledge, this is the first prospective, within-patient evaluation of a mandibular PSI montage; second, we quantified phase-specific behaviour (induction, surgery, emergence) and the influence of facial electromyography (EMG) on between-montage differences—factors known to perturb processed indices, particularly around neuromuscular block reversal and arousal. 8–12 Our primary question was the statistical agreement between MAND and HEAD across phases, that is, whether MAND PSI could be treated as a direct numerical substitute for HEAD PSI. METHODS This was a single-centre, prospective, observational study of adults undergoing a range of elective, non-neurosurgical and non–head-and-neck procedures under general anaesthesia. Patients were enroled consecutively based on the availability of the investigators. The PSI was recorded simultaneously from the HEAD and MAND montages Supplementary Figure S1. Signals (PSI, EMG, suppression ratio, and artefact indicator) were exported at 2s intervals with clock synchronisation. Artefact handling followed prior work; readings with artefact indicator (ARTF) > 50% were considered invalid. If the MAND sensor failed to display valid values, the electrode was first resecured and, if necessary, re-applied after skin preparation; failure to obtain a signal after two such attempts resulted in exclusion. Pre-specified phases were induction (first 12 min after induction), surgery (a contiguous 60 min intraoperative window anchored at surgical incision; if 50% in either montage (0/83 743, 0%). The analysis cohort was typical for adult elective surgery: median age 62 years (IQR 48–73), 28/36 (77. 8%) women, median body mass index (BMI) 21. 6 kg m-² (19. 9–24. 0), and American Society of Anesthesiologists (ASA) class I/II/III 11/23/2. The cohort included a typical distribution of elective procedures, primarily abdominal, orthopaedic, and gynaecological surgeries. The median operation and anaesthesia times were 93 min (63–180) and 132 min (104–238), respectively. At the patient level, the overall within-subject association between MAND and HEAD PSI was moderate (median r = 0. 70 IQR 0. 54–0. 85), whereas the concordance was lower (median CCC = 0. 39 0. 30–0. 68; Table 1). The overall mean difference (MAND − HEAD) was +5. 6 (SD 23. 9) PSI, with wide repeated-measures limits of agreement (95% LoA: −41. 5 to +52. 2). Across patients, the median proportion of paired readings within ±10 was 28. 7% 18. 2–59. 2 and that within ±20 was 69. 1% 36. 9–88. 1 Table 1, indicating wide dispersion and limited statistical agreement between the montages. Figure 1a and 1b illustrates the broad scatter and wide limits of agreement across the PSI range. Figure 1: Agreement between mandibular (MAND) and frontal (HEAD) Patient State Index (PSI) values and the influence of electromyographic (EMG) activity. (a) Scatterplot of simultaneous MAND and HEAD PSI values across all 2-s paired observations. The dotted identity line indicates the line of equivalence, illustrating the limited concordance between the montages. (b) Repeated-measures Bland–Altman plot showing the difference (MAND − HEAD) against the mean PSI. The positive bias and wide limits of agreement indicate wide between-montage variability and limited statistical agreement across the full range of anaesthetic states. (c) Relationship between ΔEMG (MAND − HEAD) and ΔPSI (MAND − HEAD). The fitted line represents a linear mixed-effects model (random intercept for patient), showing that a higher mandibular EMG relative to frontal EMG is associated with greater positive PSI deviations (β = 2. 32 PSI per 10% ΔEMG). The prominent vertical band at ΔEMG = 0 reflects the time points at which the mandibular and frontal EMG values were identical (ΔEMG = 0%) Table 1: Phase-specific and overall agreement between mandibular and frontal PSI valuesThe agreement varied substantially by phase Table 1. During induction, within-subject association was high (median r = 0. 85 0. 72–0. 94; median CCC = 0. 79 0. 54–0. 92) with minimal mean bias −0. 5standard deviation (SD) 21. 6 PSI but still wide limits of agreement (−43. 2 to +42. 0) ; the median proportion within ±10 was 64. 0% 46. 7–77. 0. During surgery, within-subject association was near zero (median r = 0. 06 −0. 16–0. 18) and concordance was negligible (median CCC = 0. 00 −0. 01–0. 02) with bias +5. 3 (SD 22. 9) PSI and wide limits of agreement (−39. 7 to +49. 8) ; the median proportion within ±10 was 23. 9% 3. 6–63. 8. During emergence, association again increased (median r = 0. 77 0. 44–0. 90), but concordance remained limited (median CCC = 0. 30 0. 04–0. 65), with systematic positive bias +13. 5 (SD 28. 0) PSI and the widest limits of agreement (−43. 0 to +69. 0) ; the median proportion within ±10 was 29. 0% 12. 7–50. 9 Table 1. Figure 1a and 1b illustrates phase-dependent dispersion of paired values and differences. Individual phase-specific metrics are provided in Supplementary Tables S1-3. Supplementary Table S1-1: Patient-specific agreement between mandibular and frontal PSI during inductionSupplementary Table S1-2: Patient-specific agreement between mandibular and frontal PSI during surgerySupplementary Table S1-3: Patient-specific agreement between mandibular and frontal PSI during emergenceThe EMG differences tracked the PSI divergence. In a linear mixed-effects model with a random intercept for patients, ΔPSI increased by 2. 32 PSI per 10% increase in ΔEMG β = 2. 32; 95% CI 2. 27–2. 37; p < 0. 001; Figure 1c, indicating that higher MANDEMG relative to HEADEMG was associated with greater positive PSI differences. DISCUSSION This brief report provides three key messages. First, in this prospective simultaneous comparison, the MAND PSI showed limited statistical agreement with the frontal reference and did not support its use as a direct numerical substitute, in contrast to the nasal/infraorbital alternative. 1 Second, agreement was phase-dependent, being closest during induction but substantially degraded during surgery and emergence, with wide limits of agreement even when within-subject association was moderate. Third, PSI divergence showed a consistent EMG-related component: ΔPSI increased with ΔEMG in a patient-clustered mixed-effects model, consistent with prior evidence that neuromuscular block reversal or arousal can elevate EMG and processed indices without commensurate cortical changes. 8–12 Elevated MAND EMG thus appears to be an important contributor to between-montage divergence. The limits of agreement and the proportions within ±10 and ±20 PSI were therefore interpreted as descriptive indicators of statistical agreement rather than clinical thresholds. Our conclusions are accordingly restricted to the magnitude and phase-dependence of agreement. Although elevated MAND EMG explained an important component of the divergence, residual differences likely reflect montage-dependent differences in EEG signal quality at the mandible and the limited precision of the EMG index. These findings are consistent with the BIS -literature, in which non-frontal placements show variable performance and mandibular placements often demonstrate systematic bias and false elevation. 3, 5–7 Extending this concept to PSI, our results suggest similar or worse behaviour at the mandible, with EMG-dependent inflation as a likely contributor, underscoring the need for device-specific validation rather than cross-index extrapolation. 12 Clinically, when the forehead is inaccessible, MAND PSI should be regarded only as a conditional adjunct. Interpretation should account for phase and EMG, placing limited weight on absolute values during surgery or emergence and prioritising trends under low-EMG conditions. The magnitude and variability of between-montage differences indicate that MAND PSI values should not be assumed to be numerically equivalent to HEAD in routine practice. Feasibility is an additional constraint as signal failure occurred in a meaningful minority. This study has several limitations. Its single-centre, observational design, and modest sample size, including exclusions due to mandibular signal failure, limit precision and generalisability. No a priori sample size calculation was performed; enrolment was feasibility-driven and informed by a prior nasal/infraorbital SedLine® validation study. 1 Intraoperative management, particularly neuromuscular blocking and reversal practices, was not standardised and may have influenced EMG and PSI divergence. Although dense time-series data reduce independence, this was addressed using repeated-measures Bland–Altman modelling and patient-level summaries. Future studies should examine agreement under protocolised low-EMG conditions and event-aligned analyses of stimulation and arousal. CONCLUSION The mandibular PSI showed limited statistical agreement with frontal monitoring and does not support routine direct numerical substitution for frontal PSI values. Agreement was phase- and EMG-dependent, being closest during induction under low EMG and substantially degraded during surgery and emergence, with signal failure in a meaningful minority. These findings indicate limited statistical agreement rather than a pre-defined threshold for clinical non-interchangeability. When the forehead is inaccessible, mandibular PSI should be used only as a cautious adjunct, interpreted with attention to EMG and clinical context. Presentation at conferences/CMEs and abstract publication None. Disclosure of use of artificial intelligence (AI) -assistive or generative tools The authors used ChatGPT (OpenAI, San Francisco, CA, USA) to assist with language editing and formatting. The authors take full responsibility for the content of the manuscript. Declaration of use of permitted tools No additional tools requiring declaration were used in the preparation of this manuscript. Data availability The datasets generated and analysed in the current study are not publicly available because they contain patient-level perioperative information collected at a single institution. De-identified summary data are available from the corresponding author upon reasonable request and with permission from the Institutional Ethics Committee of Hitachi General Hospital. Declaration of patient consent The authors certify that they have obtained all appropriate patient consent forms. In the form the patient (s) has/have given his/her/their consent for his/her/their images and other clinical information to be reported in the journal. The patients understand that their names and initials will not be published and due efforts will be made to conceal their identity, but anonymity cannot be guaranteed. Author contributions TS was involved in study concepts, study design, definition of intellectual content, literature search, data acquisition, data analysis, manuscript preparation, manuscript editing, and final approval of the manuscript. YY was involved in study design, data acquisition, manuscript review, and final approval of the manuscript. All the authors have participated in the review, drafting and final approval of the manuscript. Supplementary material This article has supplementary material and can be accessed at this link. Supplementary Material at https: //links. lww. com/IJOA/A73. Financial support and sponsorship Nil. Conflicts of interest There are no conflicts of interest.