15581-m-nanninga

DETERMINANTS - SYSTEMATIC REVIEW 69 Study selection The search method resulted in 3,594 records, of which title and abstract were independently screened for eligibility by three authors (MN all, JT and DEMCJ half). A study was eligible if data were presented on determinants of passing one of the filters (decision to consult a professional/seek help, recognition of problems by a professional, referral to specialized care, or admission to inpatient care) or use of any care aimed at reducing or making manageable psychosocial problems (emotional, behavioural, social, or mental health problems) of children aged 0-18 in high-income countries (countries with a gross national income per capita over 12,276 United States Dollar in 2010 according to the World Bank’s list of high-income economies). Studies that focused on the general child population or on children with a specific type of psychosocial problem were included. Studies that selected a specific group based on a certain demographic characteristic − for example, studies including only boys, low-income families, children living in a rural area, or only ethnic minorities − were also included as long as it was compared with a group which was representative of the community, or when it regarded children in care, compared with other children in care. Qualitative studies, evaluation studies, single-case studies, reviews, editorial letters, and books were also excluded. This resulted in the exclusion of 3,203 records based on their title and abstract (Figure 2). Interrater Cohen’s kappas for inclusion/exclusion were 0.75 (MN/JT) with p pos =0.79 and p neg =0.97, and 0.70 (MN/DEMCJ) with p pos =0.73 and p neg =0.97 [21, 22]. It is important to note that we just indicated one reason of exclusion per paper in Figure 2, i.e. the first observed issue leading to its exclusion, but a paper could actually meet more than one exclusion reason in the selection process. Full texts of the remaining 391 studies were independently reviewed by the three authors, leading to the exclusion of 245 studies for various reasons (Figure 2). Interrater Cohen’s kappas for inclusion/exclusion were 0.72 (MN/JT) with p pos =0.82 and p neg =0.90, and 0.74 (MN/DEMCJ) with p pos =0.84 and p neg =0.90 [21, 22]. Discrepancies between reviewers, both in screening title/abstract and full-text, were resolved by discussion or by consultation of the other reviewer. Quality assessment The remaining 146 studies underwent quality assessment and content extraction by using modified versions of the tools of Gyorkos and colleagues [23, 24]. Study quality was assessed by scoring the following dimensions: (1) sample/cohort selection including response rate and generalizability, (2) inclusion of sufficient confounding factors in multivariable analyses among which the child’s psychosocial problems, (3) clarity of measurement of determinants and (4) outcome variables, and (5) completeness of