SLEEP CHARACTERISTICS OF DUTCH ADOLESCENTS ARE RELATED TO SCHOOL PERFORMANCE W.F. Hofman, L. Steenhof Department of Psychology, University of Amsterdam, Amsterdam INTRODUCTION During adolescence the changes in sleep pattern do not seem to coincide with the demands of the relatively early starting times of schools (Wolfson and Carskadon, 1996; Kowalski and Allen, 1995). Adolescents develop a sleep debt during the school week, as their bedtimes are increasingly delayed with age. The children seem to make up for this debt by sleeping in on the weekends. This sleep phase delay with shorter sleep times during the school week, sleepiness during the school day and longer, later sleep during the weekend may have a negative effect on school performance and, eventually, the health of the school children. In these studies no distinction was made between morning types and evening types. It is hypothesized that evening types, because of their natural tendency to delay their bedtimes while the school start times are the same as for morning types, may experience a more detrimental effect of this sleep phase delay on school performance. Also most studies were performed in places with very early school starting times (between 7.00 am and 8.00 am). Starting times of Dutch schools are generally between 8.00 am and 9.00 am. This study describes the changes in sleep characteristics and the relation with performance in Dutch adolescents from 6 different high schools. Possible differences between morning and evening types were also investigated. METHODS A sleep habit questionnaire consisting of 51 questions was sent to 6 different high schools in an area around Utrecht, in the center of the Netherlands. The response rate was 65%, resulting in a group of 1500 Dutch adolescents completing the questionnaire. Inclusion criteria were: good health, enough motivation for school and no use of drugs other than alcohol, caffeine, nicotine and/or weed. 604 boys and girls were finally selected for this study. The age range was 12 - 18 years and in the schools 3 different levels of education were represented: the lowest level VBO/MAVO (130 subjects), the intermediate level HAVO (229 subjects) and the highest level VWO (245 subjects).

51

ANALYSIS Nine variables were derived from the questionnaire to describe the changes in sleep characteristics with age: • Week bedtime • Weekend bedtime • Week rise time • Weekend rise time • Sleep lag between week/weekend bed and rise times • Total time in bed • Peak alertness • Through alertness • Sleep quality The relation between these variables and age was estimated with linear regression. Morningness/eveningness (M/E) was measured with the Dutch questionnaire of Kerkhof (1984) and some extra questions. As it became apparent that the answers on the M/E questionnaire were based on the starting times of the schools only the subjective judgement of their morningness/eveningness was used. A product moment correlation was calculated to measure the relation of morningness/ eveningness with age. The relation between sleep characteristics and school performance was measured by means of multiple regression analysis. In order to determine the variables that might also have an effect on school performance besides sleep characteristics a correlation matrix was calculated between all variables. Based also on this correlation matrix the independent variables of the multiple regression (with school performance as dependent variable) were: gender, level of education, total drug use, sleep lag, total time in bed, sleep need, sleep quality, peak alertness, trough alertness, morningness/ eveningness. When any main effect was found further regression analyses were done to identify variables that might have affected these main effects. RESULTS Changes in sleep characteristics with age. The mean values and standard deviations of the sleep characteristics are shown for each age group in Table 1, together with the explained variance of the linear regression analysis. Bedtime (during the week as well as during the weekend) was significantly delayed with age and total time in bed was significantly reduced. Although rise time was significantly delayed with age and sleep lag was significantly increased the explained variance was small. No significant change was found between 12 and 18 years of age for the variables sleep quality, peak alertness and trough alertness.

52

Table 1: Mean values (SD between brackets) of the sleep characteristics per age group. Last column: explained variance of the linear regression of each variable with age. * = p