The Danish Youth Cohort: characteristics of participants and non-participants and determinants of attrition.

AIMS: The aim of this paper is to describe the design and methods used in the Danish Youth Cohort and to give a description of the study participants with special attention to a comparison between participants and non-participants regarding sociodemographic characteristics. METHODS: A total of 1,945 schools were invited, out of which 506 participated. The participating 7th grades comprised a total of 12,498 responding adolescents. The response rate for the Danish Youth Cohort established in 2005 was 63%. The sample of 12,498 adolescents represents 18.2% of all pupils (n = 68,764) in the 7th grade (mean age: 13.4 years) in Danish schools in 2005. The cohort was followed up in spring 2006 and spring 2007, where the adolescents were in the 8th (mean age: 14.4 years) and 9th (mean age: 15.3 years) grades, respectively. RESULTS: We found that compared with non-participants the participants were significantly more likely to be girls, to be of Danish ethnicity, and to live in one-family houses. Furthermore, participants more often came from families with two or three children, were more likely to have parents with a high occupational status, parents who were married and parents with a higher total income. Loss to follow-up was only associated with adolescents' higher probability of drinking and use of tobacco, and none of the other factors were associated with attrition. CONCLUSIONS: The participants in the Danish Youth Cohort represent a great variety of different groups of socio-demographic factors, although they differ from non-participants as regards a range of socio-demographic factors. This should be taken into account in future analyses.

Psychometric properties of brief indexes designed to measure social-cognitive predictors of smoking initiation.

BACKGROUND: Constructing indexes which measure factors that may predict smoking initiation is essential for planning prevention programs. Our aim was to examine the criterion-related construct validity of brief psychological indexes of attitude, social influence and self-efficacy to be used in future studies for predicting smoking initiation in adolescents. METHODS: Five indexes were evaluated using cross-sectional data on 4819 adolescents age 13 in a random sample of schools in six Danish counties. Item analyses were performed with the graphical log-linear Rasch model, a modification of the Rasch model that allows for differential item functioning and positive local dependence. RESULTS: The three indexes social influence-norms, social influence-pressure and attitude showed acceptable deviations from the Rasch model, while the two indexes self-efficacy and social influence-behavior were invalidated by negative local dependence. CONCLUSION: Of the five tested indexes, three (social influence-norms, social influence-pressure and attitude) showed acceptable criterion-related construct validity and may be considered unbiased representations of the theory-based factors if statistical analyses are appropriately adjusted. The results of this population-based study show that the graphical log-linear Rasch model is useful for evaluating health-related behavior indexes and identifying problems, which can be dealt with when using the indexes in future studies.

Does educational level influence the effects of smoking, alcohol, physical activity, and obesity on mortality? A prospective population study.

OBJECTIVES: This study aims at examining whether the relation between established risk factors and mortality differs with socioeconomic status as measured by level of education. METHODS: A population-based sample of 14,399 women and 16,236 men aged 20-93 years from Copenhagen was stratified into three educational levels measured as basic schooling, and the effect of smoking habits, alcohol consumption, physical activity, and body mass index, respectively, on mortality was assessed. RESULTS: Those with the lowest level of education were most frequently heavy smokers, heavy drinkers, physically inactive, and obese. During a mean follow up of 16 years 10,952 subjects died. Compared with subjects with the lowest educational level, women with the highest educational level had a relative risk of 0.80 (95% CI; 0.70-0.91), and men of 0.71 (0.65-0.78). Heavy smoking compared with never smoking implied a more than twofold increased risk at all three educational levels among both men and women. The relation between alcohol intake and mortality was J-shaped on all three educational levels. There were decreasing risk functions describing the relations between physical activity and mortality on all three strata. Further, subjects who were either very lean or obese had increased risks of death compared with those of normal weight at all educational levels in both genders. CONCLUSIONS: The difference in distribution of the main known risk factors may be part of the explanation for the differences in mortality risk. However, these risk factors seem to influence mortality equally at different educational levels. Therefore, social inequalities in mortality do not seem to be explained only by differences in effect of lifestyle risk factors, but are also related to the social rank or unexamined factors within.