Comparison of eeg power spectra between patients with primary insomnia and good sleeper controls, accounting for the effects of age, gender, and sleep period, using a large compilation of PSG recordings from three clinical trials

Abstract

Introduction Spectral analysis has been used to quantify differences in the sleep EEG between primary insomniacs (PI) and good sleeper controls (GSC). Previous studies have been based on relatively small samples with varying inclusion criteria, and have not typically examined the effects of important factors such as age, gender, and sleep period of the night. These limitations may be responsible for the conflicting results reported not only with respect to the magnitude but also in the direction of differences in EEG power spectra between PI and GSC. By utilizing a large database of sleep laboratory EEG recordings from clinical trials from the same development program, this study provides a comprehensive comparison of EEG power spectra in NREM and REM within sleep period between patients with PI and GSC, within gender and age subgroups. Materials and methods Spectral analysis was performed on PSG recordings from 847 PI patients and 818 GSC collected in sleep laboratory setting in three clinical trials. EEG power spectra of PI and GSC were computed during NREM and REM sleep over the entire night and by sleep period (defined by thirds of the night or by sleep cycle) as a function of age and gender. Results Differences in the EEG spectral profile between PI and GSC in NREM strongly depend on age, gender, and sleep period (defined by either thirds of the night or sleep cycle). Specifically, PI aged 35 years but not in males of any age. Results of analyses by parts of the night were in general agreement with those by sleep cycle. EEG spectral power was higher for females than males of all ages and in all sleep periods across almost all frequencies for both PI and GSC. Within each group (PI and GSC), gender differences in spectral power appear to be larger than the differences observed between PI and GSC. Differences in REM sleep were similar but less consistent. Conclusion Analysis of a large database of sleep EEG recordings from several clinical trials with consistent recording and scoring procedures reveals differences in the EEG spectral profile of PI and GSC that are dependent on age, gender, and sleep period, thus providing new information on characteristics of the sleep EEG in patients with primary insomnia. Acknowledgement This work was supported by funding from Merck&Co., Inc.