Procedure
All empirical studies outlined in this thesis were part of the longitudinal research project “Early Detection – Early Intervention in Preschools” (TUTI) [31].
Preschools from a stratified sample of various-sized Swedish municipalities were invited to participate. The preschool managements in the various municipalities were contacted and consent was requested for participation of their preschool units. The preschool teachers then asked all parents for their written informed consent. Answers were based on all the preschool teachers with professional knowledge of the child. Teachers were required to have known the child for at least six months, and were asked to base their ratings on at least the two preceding weeks (22).
Participants
In the first data collection set, 1615 children were invited to participate. The parents of 663 of these children (41%) subsequently gave their consent. During the second year of the study, the parents of a further 91 children participated, and in the third year a further 73 children were incorporated into the study. In total, 845 children participated: 438 boys and 407 girls with a mean age of three and a half years, or 42 months (SD=16, range 13–71). There was no significant mean age difference between genders.
At the time of investigation, a total of 91% of the children lived with both their biological parents, 3.1% lived alternately with each parent and 5.1% lived with only their mother. Altogether, 28% of the children had a mother tongue other than Swedish and, according to preschool teachers’ estimation, 5.8% received special support in preschool. These figures correspond quite well with Statistics Sweden (SCB) (Statistics Sweden, 2013; Swedish National Agency for Education, 2017) estimates for this age group.
The preschool group mean size was 22 (range 8–50) children, with a mean of 4 preschool teachers (range 1–7) in the group. The child to teacher ratio averaged 5.6 children/teacher. Over the three years that the study was conducted, 49 preschools participated, with 110 preschool classes. They were drawn from a stratified sample of six differently sized Swedish municipalities, representing large, medium and smaller municipalities. Approximately 47% of the children lived in small municipalities (<50,000), 45% in medium-sized municipalities (50,000–200,000) and 8% in large municipalities (>200,000). In Sweden as a whole, 43% live in small municipalities, 16% in medium ones and 41% in large; thus, large municipalities are under-represented and medium-sized ones are over- represented in our sample.
Instruments
Strengths and Difficulties Questionnaire (SDQ)
The SDQ is a well-known questionnaire in Europe, consisting of 25 items that measure child behaviours, and can be rated by parents, teachers or self-rated (from 11 years). The SDQ for ages 4–18 years displays good construct and concurrent validity, as well as some evidence of predictive validity [32-34]. In this study, the SDQ teacher version for children aged 2–4 was used [35]. The SDQ has been translated into Swedish and validated for parental and teacher use for children between 5 and 17 years of age, and the self-rated version has been validated for children aged 11–16 years; it has demonstrated good psychometric properties [31, 36-38]. Recently, the SDQ for parents has been used in a normative sample of children aged 2–5 years in Sweden, and here risk groups are reported as percentiles to identify them [39]. The SDQ has been confirmed as having satisfactory psychometric properties in identifying 3- and 4-year-olds with emotional and behavioural difficulties [40].
The items are divided into five subscales of five items each, generating scores for four problem subscales: conduct problems, hyperactivity, emotional symptoms and peer relationship problems, and one strength subscale: prosocial behaviours. Responses are given on a 3-point Likert scale: (0) “not true” (1) “somewhat true” (2) “certainly true”. The total scores for the behavioural difficulties scale can be divided into three subgroups (normal, abnormal and borderline) by the use of cut-off scores [41].
The impairment supplement questions, which describe how behavioural problems affect the child’s level of functioning, are subsequently included in the impact score. Ratings for these impairment supplement questions are: (0) “not at all” and (0) “only a little” (1) “quite a lot” and (2) “a great deal” [35]. For the question “Do the difficulties interfere with the child’s everyday life in the following areas?” the specification “Peer relationship and Learning” is substituted in this study with the situations: “Free play, Organised situations and Routine situations”. This is an adaption to the structure of the Swedish preschool environment, which is not as “classroom-like” as in many other countries.
The Children’s Engagement Questionnaire (CEQ)
Preschool teachers used the Child Engagement Questionnaire [42] in order to measure engagement. Designed in the USA, the original 32-item instrument was used to rate how children typically spend their time when in preschool, this being identified as a measurement of global engagement. The questionnaire consists of 32 items on a four-point Likert scale. The response alternatives for the Swedish rating scale is (1) almost never happens (2) sometimes happens (3) happens quite often or (4) happens very often; this version was the one used in the present study. Minor adaptations were employed in the translation of CEQ into Swedish, which resulted in the use of 29 of the original 32 items [43]. Earlier studies have reported good content and construct validity, as well as intra-rate reliability, for CEQ [43-45].
The International Classification of Functioning, Disability and Health : Children and Youth Version (ICF-CY)
The International Classification of Functioning, Disability and Health is the WHO-framework for measuring health and disability at both individual and population levels [20].
In this study, 7 items were used to assess developmental delay, regarding bodily function and cognition (3 questions each) and language (1 question) using the ICF-CY Code Sets [46]. These were responded to on a scale of: (1) “not true at all” (2) “partly true” and (3) “completely true”.
Statistics
The data were analysed in SPSS version 27. Demographic data are presented with mean, standard deviation (SD), median and cut-offs for the 90th and 10th percentile.
Risk groups
Risk groups for participants were constructed based on emotional subscale of the SDQ as well as the engagement scale derived from the CEQ-questionaire at baseline. For the emotional subscale of SDQ, a value of 1 and above were considered at risk regarding emotional problems. In the case of missing values in the emotional subscale where the total value for the subscale was still 1 or above, the participants were considered at risk regarding emotional problems. The cut-off of 1 was chosen based on the mean value of this dataset as well as similar corresponding datasets from Sweden (23, 24). Using a higher cut-off of 3 for the emotional subscale when constructing the risk groups yielded similar results.
In this dataset, the values obtained from the CEQ-questionnaire were not normally distributed. Hence, the cut off was based on the cohort median rather than the mean. Hence, below median values for engagement derived from the CEQ-questionnaire were considered at risk regarding engagement.
At baseline, children with CEQ-values above median as well as values below 1 in the emotional subscale of SDQ were placed in the “No risk”-group while children with CEQ-values above median but who scored 1 or higher in the emotional subscale of SDQ were placed in the “Emotional risk”-group. Furthermore, Children with CEQ-values below median but with values below 1 in the emotional subscale of SDQ were placed in the “Engagement risk”-group. Children with CEQ-values below the median who also scored 1 or higher in the emotional subscale of SDQ were placed in the “Combined risk”-group.
General linear models
For graphic longitudinal presentation, we used general linear models (GLM). Longitudinal development for SDQ was calculated for each subgroup at baseline compared to year two and three. For each time point in this analysis, a SDQ total for that time point was calculated where the emotional subscale as well as the prosocial subscale was excluded. Considering the GLM for risk groups according to ICF, a ICF total was calculated at each time point. When assessing the GLM for risk groups according to emotional subscale as well as CEQ, totals for both subscales was calculated at each time point.
A test for trend was conducted for the GLMs. As the assumption of sphericity was not met for the analyses, Huynh–Feldt correction was used to estimate the trend. To investigate the association between risk group status and outcomes at year three, a one-way Anova was conducted. A post-hoc Tukey test was used to further analyze the difference between the risk groups at year three.
For the multivariable analysis, continuous variables were added as covariates in the GLMs, while categorical values were added as between subject factors. The models were then assessed regarding main effects.
Risk group status at year three
A risk group variable based on emotional subscale of the SDQ as well as the engagement scale derived from the CEQ-questionaire at year three was constructed based on the same cut-offs as for the risk groups at baseline. Chi-square and linear-by-linear analyses were used to investigate the relationship between riskgroup at baseline and riskgroup at year three.
Ethical considerations
This study was approved by the Regional Ethical Review Board in Linköping (Dnr 2012/199–31). Preschool management, preschool teachers and both parents of each child provided written informed consent. All questionnaires were coded and the coding key was kept separate from the questionnaires after the data was collected. If preschool teachers identified children with previously unknown mental health problems in the course of the study, they were instructed to refer them to child healthcare for support.