Time of day is an inaccurate surrogate marker for sleep-wake state in an epilepsy monitoring unit.

OBJECTIVE

To assess the correlation between time of day and sleep-wake state prior to seizure onset for seizures recorded in an inpatient epilepsy monitoring unit.

METHODS

We prospectively enrolled a consecutive series of patients undergoing inpatient epilepsy monitoring. For each epileptic seizure recorded, continuous EEG data preceding seizure onset was reviewed and scored as W, N1-3, or REM in ten 30-second epochs. Time of day was divided into four 6-h phases (0600-1159, 1200-1759, 1800-2359, 0000-0559). The preictal sleep-wake state was then correlated to nocturnal (0000-0559) versus diurnal (0600-2359) times of day.

RESULTS

A total of 102 seizures from 42 patients met enrollment criteria over a period of 19 months. Eighty-five seizures occurred during the diurnal phase, and 17 seizures occurred during the nocturnal phase. Thirty-six percent of all seizures (n = 37) were preceded by at least 1 epoch of sleep. The proportion of patients sleeping prior to a seizure within each 6-h phase varied significantly from the overall distribution only during nocturnal phase. Seventy-six percent of nocturnal seizures and 28 % of diurnal seizures were preceded by sleep. Therefore, the nocturnal time window from 0000-0559 had a sensitivity of 0.65 (95 % confidence interval 0.48-0.78), a specificity of 0.06 (0.02-0.15), a positive predictive value of 0.28 (0.20-0.39), and a negative predictive value of 0.24 (0.10-0.39) in association with sleep-onset seizures.

SIGNIFICANCE

The time of day is an inaccurate surrogate for preictal sleep-wake state in the epilepsy monitoring unit despite a correlation between nocturnal period and sleep. Diurnal sleep is a common phenomenon in the inpatient unit.