Reliability of the sleep cycle as a scoring unit.

Abstract

ABSTRACTA fundamental characteristic of sleep is the alternation between non‐REM sleep periods (NREMPs) and REM sleep periods (REMPs), together forming the NREM sleep cycle (measured from NREM onset to the end of the succeeding REMP). It is proposed that analysis of sleep stage data should reflect this fundamental unit of sleep by basing comparisons between nights on an equivalent number of completed NREM cycles. The night‐to‐night stability of sleep stages based on 4 completed NREM cycles was compared with an analysis in which sleep time was fixed at 6 hrs (thus including partially completed NREMPs or REMPs). Total sleep time in 4 cycles was highly reliable within subjects. Both approaches resulted in significant reliability coefficients for all sleep stages. However, reliability did not differ significantly between the two approaches.Analysis of individual NREMPs and REMPs indicated significant stability for REMPs 1 and 2 and NREMP 1 only, with intra‐individual variability increasing in later cycles. By contrast, SWS content displayed greatest reliability in cycles 2, 3, and 4.The results suggest that when subjects' sleep is compared on an equivalent‐cycles basis, intra‐individual stability of sleep patterns is as reliable as standard approaches whether stages are expressed in absolute duration or as a percent of TST. Advantages of such an approach are discussed.