Bispectral Index–Derived EEG Metrics for Objective Perioperative Pain Assessment in Non‑Communicative Patients

Pain assessment in older adults with cognitive impairment is challenging because self‑report is often unreliable and observational tools have psychometric limitations. In this study, we conducted a prospective observation of 30 older patients (mean age 83.5 ± 7.9 years) undergoing surgery for femoral fracture. Pain was assessed with the Numerical Rating Scale (NRS) in cognitively intact patients and with the Pain Assessment in Advanced Dementia (PAINAD) scale in cognitively impaired patients. Frontal EEG was recorded with Bispectral Index (BIS) monitor at three perioperative phases: pre-anaesthesia, anaesthesia, and post-anaesthesia. Spectral edge frequency (SEF08), median frequency (MEDFRQ), and total power (TOTPOW08) were extracted. Across 86 valid phase recordings, SEF08 and MEDFRQ were significantly higher when pain was present (p < 0.01), whereas TOTPOW08 and patient demographicswere not.Multivariable logistic regression, confirmed SEF08 (OR 1.34–1.70) and MEDFRQ (OR 1.41–1.86) as independent predictors of pain across both outcomes. Pain peaked during anaesthesia (OR 10–26× baseline). Cognitive impairment reduced self-reported pain but not EEG markers, indicating a dissociation between reporting and neurophysiological activity. These findings suggest that EEG-derived spectral features may provide reliable, objective markers of pain, particularly useful in non-communicative or cognitively impaired patients. Larger studies are warranted to validate the integration of such EEG metrics into clinical pain monitoring.