Screening of toxic-metabolic encephalopathy with and without epileptic seizure with density spectral array

Although toxic metabolic encephalopathy (TME) is clinically common, distinguishing between TME with/without epileptic findings remains challenging. We examined the efficacy of density spectral array (DSA), a form of power-spectrum electroencephalography (EEG) screening tool, for clinicians not specializing in EEG interpretation to make this distinction.

Among 346 patients with suspected TME who underwent EEG for acute impaired consciousness (2012-2023), 149 were ultimately diagnosed with TME (mean age 68.9 ± 13.3 years) were enrolled. Using EEG data, we operationally classified DSAs based on frequency changes, as follows: 1) flame or cyclic (presence of temporal frequency change), 2) band (presence of continuous alpha-range activity), 3) gradation (alpha to delta-range activity), and 4) other patterns. The inter-rater agreement rate for DSA pattern assignment was evaluated in a double-blind manner to confirm the reasonableness of the classification. Additionally, we evaluated the sensitivity and specificity of each DSA pattern at determining the ultimate diagnostic outcomes (TME alone or TME with epileptic findings).

TME alone and TME with epileptic findings were 136 and 13, respectively. The inter-rater agreement for DSA classification was high among clinicians (κ = 0.72-0.92). The flame or cyclic pattern exhibited high specificity (97.1 %), but low sensitivity (23.1 %) for the diagnosis of TME with epileptic findings. Conversely, the band and gradation patterns showed a high specificity (76.9-84.6 %) for the diagnosis of TME alone.

Overall, our DSA classification demonstrated a high inter-rater agreement rate, indicating utility as a simple yet specific tool for distinguishing TME with and without epileptic findings.