Design and participants
The materials used in this study were collected from a cross-sectional health survey among pupils in 10th grade (15–16 year old) in both urban and rural regions of Norway, including Oslo, and the counties of Hedmark and Oppland in the southeastern part of Norway. The Norwegian Institute of Public Health was in charge of conducting the data collection.
All 10th grade school classes in the selected counties were invited to participate in the study. This is the last year of obligatory schooling in Norway.
Participation in this health survey was voluntary and pupils gave their written informed consent for using the data for research purposes before filling out the questionnaires. Parents also had the option of withdrawing their child from the study.
Questionnaires were filled out during school hours in the classrooms. Information about the survey and instructions on how to complete the questionnaire was given in the classrooms by specially trained field workers who were also responsible for gathering the completed questionnaires.
Those students not present were given the questionnaire to be answered later. Students still not responding were mailed the questionnaire at home to be answered and returned in an already stamped and addressed envelope. The final response rate was 88% and the total number of respondents was 11154.
The study protocol was approved by the Regional Comity for Medical Research Ethics and by the Norwegian Data Inspectorate.
Mental health status
Mental health status was measured using the Hopkins Symptom Check List, 10 item version (HSCL-10), which is a condensed version of the HSCL-25 [17, 18]. The HSCL-10 asks for presence of symptoms during the last week, and includes 3 questions of symptoms of anxiety, 6 questions of depression, and one question relevant for both anxiety and depression. Responses are encoded on a four-point Likert scale from "not troubled" to "heavily troubled". For the purpose of the first aim (dimensionality in symptom load in relation to help-seeking), we apply the mean-score of the scale as a continuous variable (each sum-score value labelled with percentiles in the total population). For the last aim, we apply the conventional dichotomy for case-level symptom load with cut-off at and above 1.85 on the mean-score (range 1–4).
Help-seeking for mental health problems
Participants were asked whether they during the last 12 months had "had a mental health problem for which help had been sought". Response categories were "yes" or "no".
Use of health services
Separate from the question on help-seeking for mental health problems cited above, the pupils were asked about their use of various health services during the last 12 months. They were given a list including school health services, youth health clinic, adolescent pedagogical psychological service, psychologist or psychiatrist, family counselling, other medical specialist, emergency ward or outpatient clinic, hospital inpatient, municipal social service agency, physiotherapist, and alternative therapy. For each of these the pupils were asked whether they had used this health service; none, 1–3 times, or 4 or more times during the last 12 months. The content and quality of the contact with health service providers could not be determined from the questionnaire. The response categories 1–3 times and 4 or more times were therefore collapsed into one category for analyses.
Hierarchical arrangement of services specializing in mental health problems
Adolescents commonly reported contact with multiple service providers. In order to examine degree of mental health specialization in the contact adolescents achieved, we defined a hierarchical ranking of level of specialization. At the lowest level, we defined services mainly provided by nurses or other health-providers without academic degree or special training in mental health problems. On the next level, we defined all MDs regardless of specialization, psychiatrists excluded. The GP was not addressed as a separate level because referral to specialist services is done by the GP. Consequently, most adolescents attending other medical specialists would have been examined by the GP. The third level comprised pedagogical-psychological services and family counselling, where most providers have some specialisation in mental health, as they commonly handle behavioural problems. The fourth level included clinical psychologist and psychiatrist, regarded as the health care providers with the most advanced training in mental health problems. Adolescents with multiple contacts were in this hierarchal arrangement encoded as to their most specialized health-care provider contacted.
Somatic and psychosomatic symptoms
Somatic and psychosomatic conditions were defined as report of having had asthma, allergy, eczema, or within the last 12 months having had inflammation of the ear, soar throat, bronchitis, lung inflammation head ache, or pains in shoulder, neck, legs, knees, stomach, or back. Response categories were "yes" or "no".
The dimensional analysis of associations between symptom load (HSCL-10 score) and having a mental health problem for which help had been sought during the last 12 months, was illustrated graphically with proportions (with 95% confidence intervals [CI]) having sought help for each symptom load. The second analysis concerned the degree of specialization in health services contacted by adolescents who report help-seeking for mental health problems, and the comparison of this with those who not report such help-seeking. For this purpose we applied cross-tabulations. In our third analysis, we attempted to disentangle help-seeking for mental health problems from somatic and psychosomatic conditions in adolescents' help-seeking towards each of the different health services listed. For this purpose, we applied logistic regression analyses with help-seeking toward each of the help-providers listed as dependent variable and case-level mental health problems as independent variable. Associations between mental health problems and help-seeking were reported as odds ratios (OR) with 95% CI with and without adjustment for somatic psychosomatic conditions. SPSS version 11.5 was used for all analyses.