A random intercept model with mental health care use as
the dependent variable was estimated to identify important
determinants. Due to the dichotomous outcome, we computed
a logistic regression model. Analyses were restricted
to the first three measurement points as some necessary
covariates (such as the SDQ for the respective age groups)
were not available in the 6-year follow-up. We included
only those participants with information on help-seeking
for at least one measurement point. All available data of an
individual were taken into account. To handle missing data,
we generated a complete cases data set.