Use of Prediction Intervals to Characterize Electroencephalographic Changes Caused by Diphenhydramine

The authors report on progress in the use of a two-channel, microprocessor-based electroencephalographic (EEG) device and the statistical strategy of prediction intervals for characterizing the normal limits of variability in cerebral state in older subjects and recognition of excessive change caused by mildly toxic medications. Prediction intervals for relative delta, theta, alpha, and beta power were calculated from repeated EEG measures on 54 medically stable elderly subjects. The authors subsequently evaluated the effects of diphenhydramine (25-75 mg) administered to 10 additional subjects. Significant drug-induced changes at the group level were apparent only in relative theta power. Significant change in relative theta power at the individual level was observed in five of nine subjects for whom data at 50 mg were available.