Introduction Hepatic encephalopathy (HE) is the commonest complication of cirrhosis. It has a detrimental effect on quality of life and on survival. Nevertheless, there is no recognised gold standard for its diagnosis. The only objective diagnostic technique, which does not require patient co-operation, is the electroencephalogram (EEG). The current assessment of the EEG is based on measures of the mean dominant frequency (MDF) and the percentage θ (θ%) power from spectral analysis. However, this represents a fairly simplistic approach to analysis of complex shifting wave forms.
Aim To optimise the analysis of the EEG to provide more reliable criteria for the diagnosis of hepatic encephalopathy and the grading of its severity.
Method The patient population comprised 169 individuals (108 men, 61 women; mean (range) age, 55 (26–80) yr) with biopsy-proven cirrhosis, classified on the basis of clinical (Conn et al, 1977) and psychometric performance (PHES; Weissenborn et al, 2001), as neuropsychiatrically unimpaired (n=85), or as having minimal (n=21) or overt (n=63) HE. The reference population comprised 48 healthy individuals (25 men, 23 women; mean (range) age, 39 (22–58) yr). Standard 21-lead, eyes-closed EEG recordings were obtained in all subjects. EEGs were visually inspected and spectral analysis undertaken on a standard P3–P4 derivation (Amodio et al, 1999) and a SEDACA component (Montagnese et al, 2007). In addition, a measure of the stability of the posterior rhythm, F-mean, was obtained using a novel computational technique. A subset of 74 patients underwent a mean of 3 (2–9) repeat assessments over time. Performance of the variables was assessed using ROC analysis.
Results Four variables had good performance characteristics.
All four variables correlated significantly with the PHES scores (p<0.05). Significant changes in F-Mean (p<0.05), α-P3P4 and SEDACA (p<0.05) were observed over time in relation to changes in neuropsychiatric status.
Conclusion Two additional variables have been identified which have good performance characteristics for the diagnosis of any degree of HE. These new variables, in combination with the currently used EEG measures, can be used to optimise the diagnosis of HE and the grading of its severity.