Relationship between subjective sleep rating and objective sleep parameters: a case study.

The subjective sleep of a male subject was related to 22 objective polygraphic sleep parameters on 39 experimental nights. Subjective sleep was rated by the sleeper himself as "good," "moderate," or "poor" upon arising, for 10, 24, and 5 nights, respectively. Differences in the 22 sleep parameters for these three ratings were examined, and poor sleep showed a significant increase in stage W (waking), an elongation of sleep latency, a decrease in mean sleep depth during the night and 2 hr after going to bed, a decrease in integrated value of sleep depth during the night, an increase in the gradient and decrease in the intercept of regression line of sleep depth against time, and a shortening of total sleep time. The average delta wave percentage per epoch was greater for a subjective rating of "moderate" than for the other groups. Canonical discriminant analysis was conducted using 16 sleep parameters, and the overall correct identification rate of three subjective sleep ratings by the objective sleep parameters was 89.7%. Variables of the first or second axis with large standard coefficients were stages 1, 2, REM (rapid eye movement), integrated voltage of electromyogram, and sleep latency. From the above two sets of analyses, sleep latency was the most useful parameter for predicting subjective sleep.