Authors:

Lewis, G.H., ROBERTSON, I., WATSON, P. & Datta, A.K.

Reference:

Year of publication:

2003

CBU number:

6095

Abstract:

A common clinical problem is daytime hypersomnolence leading to accidents. Objective measurement of day-time vigilance would therefore be useful. To date, many measures of sustained attention look for errors of omission rather than errors of commission; these tests display a ceiling effect such that however sleepy a subject feels, they can perform well for the duration of the test. We used a new test of sustained attention that specifically examines errors of commission as well as errors of omission and reaction time (Robertson et al. 1997). Local ethical committee approval has been obtained. Analyses were performed on data from six-hourly recordings in 96-hour studies of ten healthy Cambridge undergraduates. Recordings were made in the subjectsí study-bedrooms in college and reflect a natural 24 h cycle of everyday life. Vigilance was assessed using a computerized test of digit recognition that involved both response and inhibition of response to visual presentation of digits: the sustained attention to response task (SART). Sleepiness was measured subjectively using the Stanford Sleepiness Score (Hoddes et al. 1973) and a linear visual analog sleepiness score. Sleepiness as assessed by both sleepiness scores showed a clear circadian rhythm with a large amplitude. Cosinor analysis of sleepiness revealed a maximum at 5a.m. The two scores correlated significantly with each other (rS = 0.93, p < 0.001). Vigilance performance on the SART altered with time of day: errors of commission were greatest at 1a.m.; there was an 18% reduction in these errors at 7p.m. Performance on the SART was inversely correlated with sleepiness score: errors of commission increased with increasing sleepiness. The mean reaction time for a correct response was 281ms (range 140-546ms). No circadian rhythm in mean reaction time was seen. The mean reaction time prior to making any error was 264ms; after making an error, mean reaction time increased to 285ms. This difference was significant (paired t=6.3, p < 0.001). At reaction times of 150ms there were 90% errors of commission; At reaction times of 350ms there were 35% errors. A steep and significant relationship between reaction time and errors of commission was found (rS = 0.6, p < 0.001).