The majority of students (70%) were mildly to severely sleep deprived based on Carskadon's  finding that adolescents require at least 8.5 hours of weeknight sleep. Excessive daytime sleepiness was common, with 40.9% of H/W students and 41.9% of G/E students reporting ESS score greater than 10. Five factors, gender, hours of weeknight sleep, consistent bedtime pattern, up late to study and drinking caffeineated beverages after 6 P.M. were statistically significantly associated with ESS >10.
The impact on performance was important, with a statistically significant association between ESS > 10 and five performance measures, grades dropping, lates, being extremely sleepy at school, having decreased extracurricular activities and missing social or sport events or work. Further, there was a linear relationship between ESS scores and performance measures.
Morning sleepiness was common as the largest proportion of students reported being very sleepy between 8 and 10 A.M. (Figure 1). In addition, a significant proportion of students, even those with low Epworth scores, reported they were late for school in the previous month. Many students reported difficulty waking in the morning and being tired on waking These findings likely reflect the effect of "sleep phase delay" and corresponding change in circadian rhythms which begins in adolescence [2, 3], confirming that adolescents are often not ready to function early in the morning. The students also went to sleep later on weekends, thus reinforcing the sleep phase delay. With sleep phase delay, adolescents are really in a "sleep" state early in the morning, but our data show that many of these students manage to perform reasonably well. For those who were very sleepy only between 8 and 10 A.M. performance measures were just 3–5% less satisfactory than their peers, but those who were also very sleepy between 10 A.M. and 12 P.M. exhibited increases in proportion of performance problems of 46–220%. Sleep phase delay is a physiological phenomenon with onset at puberty . Parents and educators need to know that adolescents' tendency to go to bed late and wake late is normal, and this must be considered in addressing sleep habits and in academic scheduling and transportation.
Early morning is a difficult time for many students to absorb and remember complicated material. This conforms to the findings of Hansen ; in a study of 60 high school seniors, students in early morning classes reported being wearier, being less alert and having to expend greater effort. All students performed better in the afternoon.
The 24.4% of Grand Erie and 22.7% of Hamilton Wentworth students who felt their grades had dropped because of sleepiness reported 26 and 27 minutes less weeknight sleep respectively than their peers. These times are comparable to the 25 minutes found in Wolfson's study . The association of these modest sleep losses with lowered academic performance has now been demonstrated in three student samples. The potential significance of an alteration in performance associated with about 25 minutes extra sleep warrants further exploration.
Memory is critical to learning and adequate sleep is critical for memory. Memory is consolidated, and may be enhanced, during sleep after a learning exposure. The relationships between memory, learning and sleep have been outlined by Smith  who noted that declarative material (rote memorizing) is usually explicitly (consciously) learned, while procedural material (learning a novel cognitive or motor task) is usually implicitly (unconsciously) acquired. While declarative material does not seem very sensitive to sleep loss, procedural is very sensitive to sleep loss.
It is worrisome that educators continue to fail to recognize students who may have disrupted sleep patterns and that physicians fail to ask about sleep. Owen  found 44% of paediatricians did not routinely screen for sleep disorders in adolescents and underdiagnosis of sleep problems has been noted by a number of authors [24–29].
There are potential concerns with the study. Although the consistency of results from seven schools in different geographic and socioeconomic environments with a time interval of three years lends confidence to the findings, the data were based entirely on anonymous self-completed questionnaires and thus subject to responder bias. The potential problem of clustering (sampling seven schools) was addressed in the analysis. The ESS performed well in this survey; however the addition of a more recent validated sleep questionnaire for comparison would be desirable. Sleep times can be accurately measured using actigraphy; although this would be impractical for such a large sample, the use of a nested random sample of students would be valuable. There is evidence that sleep quality is a better measure for adequate sleep than sleep times; this needs to be explored. Student grades are regarded as highly confidential by the education system; the number of releases necessary to obtain these could be impractical. Again, the use of a random sample for comparison could be utilized. School "lates" records are accurately maintained in these jurisdictions and would be accessible. Measures of bedtime sleep habits would be enhanced by the use of parent questionnaires that would have a secondary benefit of focusing the attention of parents on appropriate ways to improve "sleep hygiene". A major deficit was the lack of a well validated questionnaire to identify sleep disorders in this population. It was the intent in the G/E survey to identify these students, but an unfortunate and unforeseen circumstance made that impossible. In addition, there was no clear differentiation between the environmental, physiological and behavioural causes of sleepiness as opposed to medical causes such as depression or Attention Deficit Disorders. This is an area of research that needs to be explored more thoroughly. Participation rates were different for the two Boards, 73% and 28%. Despite this the results were remarkably similar. The 28% rate in G/E reflects the difficulties of obtaining signed consent in the school system; however strategies were not in place for follow-up or reminders. This will need to be addressed in subsequent research.
The students were drawn from one of the most populous areas in Canada, so caution is needed in generalizing the results or applying them to other populations.