Social-emotional problems among Swedish three-year-olds: an Item Response Theory analysis of the Ages and Stages Questionnaires: Social-Emotional

This study took place in the Region Västerbotten in the northern part of Sweden with 40 Child Health Care (CHC) centres for children up to the age of six years. Almost every child (about 3000 children annually) turning three years of age visits the CHC. The ASQ:SE 36-month interval was used for data collection within CHC at the 3-year-olds’ ordinary visit. Parents filled out the questionnaire at home prior to the visit and had a dialogue with the CHC nurse about their child on the basis of their responses during the visit. This data collection was done in collaboration with the Salut Child Health Promotion Programme [20].

Parental-reported data on 8214 3-year-olds were collected during January 2014–September 2017, which corresponds to a response rate of 80% of those attending CHC (that has an almost 100% coverage). Out of these, 7179 of the children remained for the study after considering the exclusion criteria. The fact that 1035 children were excluded was due to any of these three reasons: the parents did not consent to the research (n = 447); the age of the child could not be determined or was outside the recommended age range of the 36-month interval of the ASQ:SE (n = 513); and the number of unanswered ASQ:SE questions were more than three (n = 75).

We used the first edition of ASQ:SE (36-month interval) [13], which has been translated to Swedish according to established recommendations [21]. The allowed age ranged from 33 months and 0 days to 41 months and 29 days according to the ASQ:SE User’s Guide [22]. Child age was calculated by using the date for the check-up visit for the child at the CHC centre and the birth date of the child. The 36-month interval of the ASQ:SE consisted of 31 items, using a 3-point Likert scale (0, 5 or 10 points), and in addition, for each item the parents could indicate if it was a ‘concern’ for them (5 points). Thus, the range of possible total score was 0–465 (zero means no behaviour problem in any item and 465 means having problem in all items plus that parents expressed concern). In addition, the ASQ:SE includes three open-ended questions that should not be used for calculating the total score and were therefore not included.

The Regional Ethical Review Board in Umeå has approved the study (2013–268-31Ö). Only children whose parents have given written informed consent were included in the study. The study was carried out according to the ethical principles available in the Helsinki Declaration of 1975 (as revised in 1983).

We have included ASQ:SE responses for 7179 3-year old children with total scores ranging 0–215. Internal consistency measured with Cronbach’s alpha was 0.78. Generally, boys had higher scores (mean = 31.2, SD = 24.9) than girls (mean = 23.9, SD = 20.6). Further, ASQ:SE scores were higher for boys compared to girls across all the quartiles (boys with quartile 1 = 15, median = 25, quartile 3 = 45, and girls with quartile 1 = 10, median = 20, quartile 3 = 35) (Table 1). These results show that normative values for boys were nearer to the cut-off (59) for social-emotional problems than for girls.

Previous study on the distribution of the ASQ:SE items [10] has shown that parents were mainly concerned about the following items: children’s eating situation, problems when asked whether the child checks that you are nearby, does what you ask him/her, complies with your requests, cries or screams for a long time, or about a child’s hyperactivity. Although many children may have difficulties in these areas, it does not necessarily mean that they should be considered having social-emotional problems. According to our present study all these items occurred at the level of very low social-emotional trait/problems. In contrast, very few children demonstrated – based on parental-report – problems on items regarding interest in sexual words and activities, too little sleep, disinterest in things around, unhappiness, and self-injury. We have shown that these behaviours are more commonly exhibited by children with high levels of social-emotional problems. Thus, we suggest that children with these latter behaviours are given attention at times when the complete ASQ:SE is inaccessible or has not been carried out, as is still the situation in most other CHC settings in Sweden and elsewhere.

Gender differences in emotional and behavioural problems have previously been observed among pre-schoolers in Sweden [7] and outside Scandinavia [25, 26]. We showed that a higher percentage of boys (12.3%) had a parental reported score above the cut-off compared to girls (5.6%). Other authors have reported that, among children of school age the gender imbalance is reversed [15]. One reason for such discrepancy may be that, at an early age children’s emotional and behavioural problems are mainly assessed through teachers and parental observations and questionnaires, while older children usually self-report their problems. Furthermore, the gender difference in ASQ:SE scores has been acknowledged in the ASQ:SE-2 User’s Guide when it comes to referral [14]. It is suggested that girls should be referred for specialized assessment and earlier therapeutic interventions already with scores lower than the cut-off point (i.e. in the monitoring zone). The gender differences shown through our IRT analysis suggest that at the same level of social-emotional problems both boys and girls exhibited difficulties in some areas of self-regulation. However, boys were more likely to have interaction and communication problems, while girls were more likely to show vulnerabilities in autonomy and compliance. Our results are in line with a previous study on ASQ:SE conducted on a population sample from Brazil, China, South Korea and USA [27]. Their results indicated that with equal levels of social-emotional problems, parents of girls were more likely to report internalizing problematic behaviours than parents of boys, such as ‘Does your child cling to you more than you expect?’. On the other hand, parents of boys were more likely to report externalizing problematic behaviours, such as ‘Does your child try to hurt other children, adults, or animals?’. It has been shown that, boys’ expressions of mental health problems may be easier to observe as these are more externalized and therefore largely reported by parents, while internalized psychological symptoms are more common among girls [15]. The latter demand more developed communication skills to be verbalized, and thus could easily be missed in reports by parents of younger children. However, whether the gender differences detected in our analysis are because of the children’s own tangible performance or their parents’ perceptions and expectations remains unclear and requires further investigations.

One of the important strengths of this study is that the results are derived from a comprehensive population-based study with high participation rates (about 80%). The use of ASQ:SE is another asset of our study, as it has shown to be a promising tool in detecting social-emotional problems in pre-school children [13]. Applying advanced statistical methods such as IRT is another advantage of our study, as it has made it possible to examine the quality of each item and understand how each item functions across the spectrum of the latent trait of the social-emotional problems.

There are also some weaknesses in our study that need to be acknowledged, however, these does not challenge our conclusions. The cut-off point of 59 was based on an American study [22] and may not entirely reflect the social-emotional problems in our population. However, using this cut-off made it possible to compare our results with many other international studies, which has been valuable. We have used only the Swedish and English version of the ASQ:SE. This likely contributes to lower response rate among immigrant families who might have different cultural beliefs and perceptions in regard to children’s socio-emotional behaviours. Although our aim was not to investigate the difference across cultures or ethnicities, the gender differences detected in our analysis were in line with other cross-cultural studies [27].