15581-m-nanninga

BTPS-EXP 49 the scale using a confirmatory factor analysis (CFA). We examined whether the data could be captured by the four subscales, and whether the four subscales were captured by a unique higher-order factor. A Comparative Fit Index (CFI) and a Tucker Lewis Index (TLI) of >0.95, and the Root Mean Square Error of Approximation (RMSEA) of <0.08 both were considered as indicating a reasonably good fit for the model [35, 36]. Fourth, we assessed parent-adolescent agreement: Pearson correlation coefficients were calculated between parents’ and adolescents’ scores on the total scale and subscales. Fifth, we assessed the criterion validity of the BTPS-exp by examining its relatedness to basic child and family background characteristics. We anticipated that some parents and adolescents would expect barriers more frequently, because they generally had more difficulties in navigating the health care system or because these children had psychosocial problems more often. For parents who have a child with psychosocial problems, or adolescents who have psychosocial problems themselves, the situation introduced in the BTPS-exp, i.e. having (a child with) psychosocial problems, is real. It is likely that these parents and adolescents would expect more barriers as compared to participants who have to imagine the situation and speculate about it. Therefore, we expected the following characteristics to be associated with expecting multiple barriers on the total scale: non-Dutch ethnicity, low parental educational level, living in other than a two-parent family, and having (a child with) psychosocial problems [6, 37, 38, 39]. The BTPS-exp total scale was dichotomized into ‘expecting few barriers’ (parents: 1.00-1.98; adolescents: 1.00-2.63) and ‘expecting multiple barriers’ (parents: 2.00-5.00; adolescents: 2.65-5.00) based on the 25% highest scores of the community sample (in comparison, the highest 25% scores in the care sample were 2.07-5.00 for parents, and 2.51-5.00 for adolescents). We performed separate univariable and multivariable logistic regression analyses for the parent and adolescent versions, leading to odds ratios (ORs) and 95%-confidence intervals (CI). Criterion validity was also assessed by examining whether expecting multiple barriers might cause children and adolescents with psychosocial problems and their parents to avoid seeking psychosocial care. We hypothesized that the association between children’s psychosocial problems and care enrolment would become less strong, when controlled for the BTPS-exp and its interaction with psychosocial problems. We performed logistic regression analyses restricted to children with an SDQ score ≥10. A p-value <0.05 was considered statistically significant (two-sided test). For the analyses on scale consistency and validity, we used SPSS version 20 and for the CFA Mplus version 7.1.