Study design
This population-based study employs previously collected data from the youth@hordaland-survey of adolescents in the county of Hordaland in Western Norway, conducted during spring in 2012. The youth@hordaland-survey is a cross sectional study with a main aim to assess mental health problems and service use in adolescents.
Sample
All adolescents in the three age cohorts in Hordaland were invited to participate in the study (n = 19 430). The adolescents received information about the study and login details via their official school e-mail, followed by an SMS reminder for the majority of the students. One school class (about 45 min) during regular school hours was allocated for the completion of the Internet based questionnaire. A teacher was present to organize the data collection and to ensure confidentiality. For those not at school during the allocated school completion, the questionnaire could be completed at other times at their convenience during the study period. Some schools arranged catch up days, and we also arranged for participation for adolescents in hospitals or institutions during the study period. Those not enrolled in school at the time of the study received log on information through postal mail. However, adolescents who had dropped out of school were not included in the current study sample, as one of the main variables was school absenteeism.
Data from the youth@hordaland-survey include information on sociodemographic variables, familial socioeconomic status, use of health care and social services, daily life functioning, as well as extensive information on mental health. Of the 19 430 adolescents who were invited to participate, 10 220 (53 %) agreed to participate and 8988 (87.9 % of the original sample) approved the linkage to administrative data on school absence.
The study and the link between youth@hordaland and data on school absence were approved by the Regional Committee for Medical and Health Research Ethics in Western Norway.
Instruments
Demographic information
Gender and year of birth are based on the personal identity number in the Norwegian national population registry. All participants were asked about their mother’s education, with the response options: ‘primary school’, ‘secondary school’, college or university: less than 4 years’ and ‘college or university: 4 years or more’.
Living situation
The participants’ living situation was based on self-report of a range of situations that were recoded as ‘living with family’, ‘living alone/with friends’, and ‘other’ for the present study. The variable ‘living with family’ includes living with biological parents, foster parents, adoptive parents, grandparents or another family. ‘Living alone/with friends’ includes living alone, living with friends or with a boyfriend/girlfriend.
School program
The educational programs reported by the participants were categorized into ‘general studies’, ‘vocational subjects in school’ (this categorization is based on the Norwegian high school system; including a program for general studies preparing for higher academic education, and a vocational education program), and a third option of ‘vocational training (work placement)’.
School absence
Administrative data on non-attendance were provided by Hordaland County Council. It included the number of days and school-hours each participant had been absent during the last semester (6 months), converted into percentage of absence relative to the total number of school days.
For the purpose of the present study high absence was defined as 15 % absence or more, based on Kearney’s criteria for problematic absence and the cut-off used in previous research on absenteeism [4, 6]. The participants were divided into three groups: Adolescents with low absence (less than 3 %), adolescents with moderate absence (between 3 and 15 %) and adolescents with high absence (15 % or more).
Self-reported absence
The participants were asked to report the number of days and school hours they had been absent during the past month. In addition, they reported location and behavior while absent, with the response alternatives: ‘I’m home’, ‘I’m out with friends’, ‘I’m at work’ or ‘I’m sick’. Other responses could be specified in an open field and multiple responses were also an option. The open responses were categorized into: ‘organizational work/politics/sport’, ‘unexcused absence’ and ‘other’.
Use of services
Service use was measured by the following question: “Have you had contact with the following services within the last school year? If yes, check how often”. The response categories used in the present study were; ‘school health services’, ‘special needs education’, ‘educational psychological service’, ‘mental health services for adolescents’, ‘mental health services for adults’, ‘general practitioner’, and ‘adolescent health clinic’. Additional services could be specified in an open field. In the present study the category ‘mental health services’ is a combination of mental health services for adolescents and adults. The participants who had been in contact with one or more services were asked to indicate the frequency of the contacts, measured by a Likert scale with the alternatives: ‘weekly’, ‘monthly’, ‘every three months’, ‘every six months’, and ‘less than every six months’, with the exception of ‘special needs education’. For the purpose of the present study, the latter two categories were combined in ‘every six months or less’ and ‘weekly’ and ‘monthly’ were combined in ‘monthly or more’.
Statistics
In this study, we investigated service use in adolescents with low absence compared to adolescents with moderate and high absence. Service use was measured by numbers and category of services visited and frequency of contact. Chi-square tests were used to examine differences between adolescents with low, moderate and high absence with regards to demographic variables, rate of contact with specific services and self-reported absence. Differences in contact with each of the services studied were examined by logistic regression, using the absence variable as the exposure variable. Age, gender, maternal education and school program were included as control variables in the regression analyses. Multinomial logistic regression was used to calculate odds ratios for the number of services visited, ranging from ‘1’ to ‘4 and more’, and the frequency of contact for the participants according to absence. Results were considered significant at the p < .05 level. IBM SPSS version 21 for Windows was used for all analyses.