Detection of children at risk of maltreatment: longitudinal understanding of teachers’ concern for children’s home situation in Sweden

Teachers in preschools and schools have a central role in detecting children at risk of maltreatment. The aim of this study was to provide a longitudinal understanding of teachers ’ temporary and prolonged concerns for children ’ s home situations (CCHS), from preschool (age 3 – 5 years) to year 1 – 3 of elementary school (age 8 – 10 years). Teacher-and parent reports from an ongoing prospective longitudinal study was used (baseline n = 2,113; 5-year follow-up n = 1,829). The study applied the bio-ecological systems framework by assessing how teachers ’ CCHS were associated with person, process, and context-related factors. Both temporary and prolonged CCHS were primarily associated with teachers ’ concern for di ﬀ erent aspects of the children ’ s development, poor contact with parents and low parental SES, while school-related factors were less prominent. The longitudinal pattern indicates three risk levels and highlights the importance of the contact between teachers and parents in understanding when CCHS arises, ceases, or persists over time.


Introduction
Child maltreatment is the abuse and neglect experienced by children under 18 years of age (World Health Organization 2022).The distressing and disruptive effects caused by maltreatment during childhood significantly influence a child's health, development, learning capacity, and have long-term consequences for a range of outcomes, as educational and employment-related outcomes (Bunting et al. 2018;Fry et al. 2018), health and health behaviours (Gilbert et al. 2009;Grummitt et al. 2021;Hughes et al. 2021), and criminal behaviour (Gilbert et al. 2009).Early detection of children at risk of maltreatment and the provision of adequate support is therefore crucial for the development of the child (Felitti et al. 1998).Teachers in preschools and schools have a central role in the early detection process, as they have almost daily interactions with children and their parents.
Research within this area has primarily focused on the obligation of professionals to refer matters to the Child Welfare Services (CWS) in cases of suspect child maltreatment.The proportion of referrals is generally low and affected by factors at several levels: Characteristics of the child (Björkenstam et al. 2018;Hjern, Weitoft, and Lindblad 2010;Karlsson and Lundström 2021); the professional's experience and the relationship between the professional and the family (Gilbert et al. 2009;Kuruppu et al. 2020;Kvist et al. 2014;Piltz and Wachtel 2009;Svärd 2017;Svensson and Janson 2008;Talsma, Boström, and Östberg 2015), and sociodemographic conditions (Bywaters et al. 2015;Sidebotham and Heron 2006;Webb et al. 2020).Noumana et al. (2020) have also identified important case related factors such as professionals' assessed severity of harm to the child, perceived lack of cooperation by parents and indications that provided support to the family has not benefitted the child.Research shows that apart from maltreatment severity, children living in single-parent families, teenage parent families, poor families, or immigrant families are more likely to be referred to the CWS than other children (Kalin et al. 2022).Clearly, professionals' concern of children's home situation and possible referrals to the CWS are related to more factors than suspicion of maltreatment as such.Previous findings raise questions about the grounds on which professionals become concerned about children's home situation and highlight the importance of understanding the combination of individual and contextual factors.More knowledge is also needed about stability and change of concern as indicator of risk of child maltreatment.Research shows that teachers' concerns for children's home situations (CCHS) can be understood as a puzzle of different indicators that become clearer over time for each individual child (Svensson and Janson 2008).However, disentangling risk factors may also entails a risk of bias on group level due to preconceived notions and norms (Kalin et al. 2022;Krase 2015).
The present study focuses on CCHS as a possible early indicator of child maltreatment.Three aspects are particularly relevant in relation to central risk factors for child maltreatment and recurring reasons for referrals to Child Welfare Services (CWS): (1) concern of parental capability such as lack of care, lack of guidance and/or lack of emotional support, (2) concern of the child's safety due to conflicts or violence in any form, and (3) concern of the living conditions, such as overcrowding, poverty or lack of security.Although teachers' reports of CCHS according to these aspects do not necessarily reflect actual child maltreatment, CCHS is still highly relevant in relation to early detection of children at risk of maltreatment, especially in Sweden where there is a low threshold for referrals to CWS.Here, suspicion of maltreatment is enough to refer to CWS (National Board of Health and Welfare 2019).
The Swedish longitudinal study Social and Physical Development, Interventions and Adaptation (SOFIA) is unique in examining CCHS across time.The present study is a follow up on a previous SOFIA-based study focused on the pre-school years (Svensson, Andershed, and Janson 2015) which showed that preschool teachers' CCHS was related to elevated concerns for several aspects of the child's health and development, need for special support in preschool, poor teacher-parent contact, and parental socioeconomic status.However, Svensson, Andershed, and Janson (2015) did not estimate the strength of independent associations between included variables and CCHS.Also, this study expands the previous one by examine CCHS not only in preschool but also into the first years of school.A better understanding of which factors that are independently associated with teachers temporary and prolonged concern for the children's home situations, from preschool to children's first years of school, can provide important knowledge to any efforts made to improve teacher-referrals to CWS.
The aim of the current study was to provide a longitudinal understanding of teachers' temporary and prolonged concern for children's home situation, from preschool to year 3 of elementary school.In order to achieve this goal, we draw on the bio-ecological system approach.
Research questions: . What characterizes the following categories of teacher-rated concern for children's home situation: no concern, temporary concern (in preschool or school), and prolonged concern (from preschool to school)? .How are factors related to person, process, and context of the child associated with teachers' temporary concern (in preschool or school) and prolonged concern (from preschool to school) for children's home situation?

Theoretical perspective
According to the bio-ecological model (Bronfenbrenner and Morris 2006), children develop over time through an interplay of the child's personal characteristics, processes in the child's immediate environment, and the context including factors at micro, meso, exo, and macro levels.Accordingly, it is reasonable to assume that variation in CCHS across time is related to the child's personal characteristics, processes in the child's immediate environment, and the context around the child.In this study, we therefore investigate the associations over time between teachers' CCHS and factors related to person, process, and context (measures described in more detail below).

Preschool and elementary school in Sweden
In Sweden, more than 85% of children aged 1-5 years participate in preschool (Swedish National Agency for Education 2020).According to the curriculum, preschool should stimulate children's development and learning in a safe environment (Swedish National Agency for Education 2019).Mandatory schooling is free of charge and runs from age 6 (preschool class) to year 9 when students are 16 years of age (Swedish National Agency for Education 2018).According to the Swedish Education Act, preschools and schools have a responsibility to ensure that children receive special support if it is deemed necessary, both for their educational learning and for their social and psychological development (SFS 2010:800).In addition, in Sweden, preschool and school teachers are legally responsible to report to social services any suspicion of a child being at risk according to chapter 14 § 1 of the Swedish Social Services Act.

Population and setting
This study used data from the ongoing prospective longitudinal school-based study SOFIA that aims to provide a better understanding of children's behavior, social adjustment, and psychological and physical health.The SOFIA-study was initiated in collaboration with toplevel decision-makers in the municipal Child and Adolescent Department and all preschool principals.For more details on the recruitment and procedure of the SOFIA-study, see Colins et al. (2014).Included children were born in 2005, 2006, and 2007 and attended preschool in a medium-sized Swedish municipality in the spring of 2010 (n = 2,542).In 2010, the municipality had approximately 85,000 citizens and was demographically similar to the rest of Sweden in terms of age, education level, level of employment, and the mixture of urban and rural areas.At baseline, 2,121 of the children's caregivers (from here onwards referred to as parents) gave active consent to participation.In total, the SOFIA-study includes six waves of data collection with multiple informants.The present study is based on data from the baseline data collection (wave 1, 2010), the 1-year follow-up (wave 2, 2011), and the 5-year follow-up (wave 4, 2015) and includes both teacher and parent ratings.

Participants
At baseline, children were between three and five years of age (m = 3.86, sd = 0.86).
Responses were received from 2,113 (99.6%) of the teachers, and from 2,026 (95.5%) of the parents.At the 1-year follow-up, when children were between the ages of four and six, 2,041 teachers (96.2%) and 1,964 parents (92.6%) responded to the questionnaire.At the 5-year follow-up, when children were between the ages of 8 and 10 and in their first, second, or third year of primary school, 1,829 teachers (86.2%) and 1,654 parents (78%) participated.
Concerning drop-outs (16% of population in 2010), previous analyses conclude that the nonparticipating group did not differ significantly from participants regarding the parents' socioeconomic status and country of birth, or the children's levels of conduct problems and internalizing problems (Colins et al. 2014).

Outcome
Concern for children´s home situation (CCHS).At all three time points, preschool and schoolteachers were asked to assess concern for parental capability, the child's safety, and the living conditions in the home, using a scale ranging from 'No concern' (1) to 'Great concern' (5).The variable was dichotomized into 'No concern' (response option 1) and 'Concern' (response options 2-5).To identify stability and change in teachers' CCHS, children were categorized into four mutually exclusive groups: (0) No CCHS at any time (1) Temporary CCHS in preschool included children for whom teachers expressed CCHS at either baseline or 1-year follow-up, or both but not at the 5-year follow-up; (2) Temporary CCHS in school included those children for whom teachers expressed CCHS at the 5-year follow-up but not at baseline or 1-year follow-up; (3) Prolonged CCHS included children who teachers expressed CCHS for across all data collections, i.e. both in preschool and school.

Independent variables
In keeping with the bio-ecological system approach, factors, potentially associated with CCHS, were categorized into Person, Process and Context.

Person
Gender: girl or boy as indicated by parents at baseline.
Teachers' concern for the child's health and development included the following four aspects: o Learning capability (ability to learn new things and concentrate).o Psychological well-being (anxiety, sadness, or depression).o Social development (difficulty with friendships, functioning in a group, or taking responsibility).o Behavior (deviation, aggression, lying, difficulty following rules).
Degree of concern was estimated by teachers using a scale ranging from 'No concern' (1) to 'Great concern' (5).These questions have been used in previous studies on parental and professional concerns (McCue Horwitz et al. 2003).We used data from the 1-and 5year follow-up.Each item was dichotomized into 'No concern' (response options 1 and 2) and 'Concern' (response options 3-5).Apart from the individual items, sum-score composite measures based on the continuous variables were constructed for each time point (range 4-20).These were used in the regression analyses.
The child's disability status was based on preschool teacher information from the 1and 5-year follow-up on whether the child had any diagnosed disability or chronic illness, for example diabetes, asthma, autism, ADHD, or physical disability.The response options were 'No' and 'Yes'.
School performance.Since children do not receive grades in primary school, teachers were asked to assess the child's academic performance in mathematics and Swedish in relation to learning objectives for their age at the 5-year follow-up (see e.g.Frogner et al. 2020).The response scale was 'Inadequate knowledge' (1); 'Acceptable knowledge' (2); and 'More than acceptable knowledge' (3).Each item was dichotomized into 'Yes' (response options 2 and 3) and 'No' (response option 1), representing children who either met or did not meet the set knowledge requirements in mathematics and Swedish.In the regression analyses, we used a composite variable combining these two variables (range 0-2).

Process
Relationships with peers.At the 1-and 5-year follow-up, teachers assessed children's peer relationship through rating the statement 'He/she is a child that others want to play/be with' using the response scale 'Agree completely' (1) to 'Disagree' (4).The variable was dichotomized into 'Yes' (response options 1 and 2) and 'No' (response options 3 and 4).The latter is referred to as 'Poor peer relationships'.
School attachment.Teachers at the 1-and 5-year follow-up assessed the child's attachment to (pre) school: 'He/she seems to like preschool very much'/ 'The child likes school'.The item was rated from 'Agree completely' (1) to 'Disagree' (4).For descriptive purposes, the variable was dichotomized into 'Yes' (response options 1-3) and 'No' (response option 4).The latter is referred to as 'Poor school attachment'.In order to allow for more variation, the original scale (range 1-4) was used in the regression analyses.

Context
Parents' country of birth was dichotomized as: Both parents born in Sweden (0), one or both parents born outside of Sweden (1).
Parent's level of educational was indicated by the following categories: elementary school (1), vocational or theoretical secondary education (2) and college or university (3).In the analyses, this variable was used as a continuous measure (range 1-3).
The variable contact between teacher and parents was specifically constructed for the SOFIA-study.Teachers were asked to rate their satisfaction with their contact with the child's caregiver 'Describe your contact with the caregivers with whom you have the most contact', on a scale from 'Very good' (1) to 'Not good at all' (4).We used data from the 1-and 5-year follow-up and the variable was dichotomized into 'Yes' (response options 1 and 2) and 'No' (response options 3 and 4).The latter is referred to as 'Poor contact teacherparent contact'.In the regression analyses, the original scale was used (range 1-4).

Statistical analysis
First, One way ANOVA and Pearson Chi Square tests were used for descriptive statistics and between-group analyses.Second, adjusted multinomial logistic regressions were performed to estimate the strength of independent associations between included independent variables and the CCHS categories, i.e.Temporary CCHS in preschool, Temporary CCHS in school, Prolonged CCHS in preschool and school.No CCHS at any time was set as reference category.Level of significance was set at 0.05.All data analyses were performed in IBM SPSS 27.

Ethics statement
All procedures and methods are in agreement with the ethical guide-lines outlined by the Swedish Research Council, which are in line with Ethical Code for Early Childhood Researchers (EECERA).Prior to data collection, the questionnaires were developed and tested in collaboration with a reference group representing teachers in preschools and school.We provided written information about the study to parents of all participating children and all parents gave their written consent prior to data collection.We also informed all preschools and schools about the study and they also agreed to participate prior to data collection.The Swedish Ethical Review Authority has approved the procedure used in the SOFIA study (Waves 1-3, Dnr #2009/429; Wave 4, Dnr #2015/024).

Results
What characterizes the following categories of teacher-rated concern for children's home situation: no concern, temporary concern (in preschool or school), and prolonged concern (from preschool to school)?
As presented in Table 1, 9.7% of preschool teachers reported temporary CCHS at the 1year follow-up (i.e.temporary CCHS in preschool).At the 5-year follow-up, 4.6% of teachers reported temporary CCHS (i.e.temporary CCHS in school).Prolonged CCHS was reported for 3.3% of the children, indicating that teachers expressed concern for these children in both preschool and in school.The degree of CCHS was consistently higher in the prolonged CCHS group and significantly higher during school, compared to temporary CCHS.
Below we present results for person-process-and context-related factors as found in Table 2.With regard to person-related factors, teachers reported significantly more temporary and prolonged CCHS for boys compared to girls and the gender differences increased over time.Almost eight of ten children in the group 'Prolonged CCHS' were boys.The level of CCHS was higher for children with disabilities than their peers and this difference increased across time.
Teachers' concerns for the children's learning, social development, mental health, and behavior co-varied with CCHS over time.For example, in the group 'Temporary CCHS (preschool)', the proportion of 'concern for the child's social development' at the 1-year follow-up was 30.3% while the corresponding proportion at the 5-year follow-up was 11.4%.When CCHS arose (5-year follow-up), concerns for all aspects of the children's development had gradually increased since preschool.A sharp increase was identified during the school years.The greatest proportions as well as highest level of teacherrated concerns for children's health and development (at both the 1-year and 5-year follow up) were found among those whom teachers have reported CCHS in both preschool and school (prolonged CCHS).
Children failing to reach sufficient mathematics and/or Swedish skills were overrepresented in both the Temporary CCHS in school and Prolonged CCHS group.The greatest proportion of children who did not meet the set knowledge requirements in one or both subjects were found in the prolonged CCHS category: 20.7% compared to only 2.4% in the no CCHS category.
Concerning process factors, the proportion of children with poor relationships with peers doubled over time, both among those for whom CCHS arose in school and in the prolonged CCHS group.Correspondingly, the proportion of children with poor relationships with peers decreased over time when there was only temporary CCHS in preschool.The same pattern was identified regarding school attachment: if CCHS was present, school attachment was likely to be poorer than if no CCHS was reported.More than 70% of the children in the prolonged CCHS group had poor school attachment at the 5-year follow up.The context factor of poor contact between teachers and parents was strongly related to prolonged CCHS.When CCHS arose (from preschool to school), the contact between teachers and parents deteriorated, and when CCHS ceased, the contact between teachers and parents clearly improved.
The parents of children whom teachers reported CCHS had significantly lower income and level of education than parents of children in the non-CCHS group, particularly the prolonged CCHS group.Preschool teachers, but not schoolteachers, were more likely to report CCHS if one or both parents were born outside Sweden than if both parents were born in Sweden (Table 2).
How are factors related to person, process, and context of the child uniquely associated with teachers' temporary concern (during preschool or school) and prolonged concern (from preschool to school) for children's home situation?
Table 3 shows results from the adjusted regression analyses.Of the person-related factors, concern for children's health and development at both 1-year and 5-year follow-ups uniquely increased the odds of belonging to any of the temporary (in preschool and school) and prolonged CCHS groups.None of the process factors was significantly associated with CCHS in the adjusted models.Reduced quality of teacherparent contact at 1-year follow-up (context-related factor) increased the odds of temporary CCHS in preschool and prolonged CCHS, but not temporary CCHS in school.At 5year follow-up, associations were found for temporary CCHS in school and prolonged CCHS.The context-related factor of increasing parental income was uniquely associated with decreased odds of temporary CCHS in preschool and school but not prolonged CCHS at 5-year follow-up.The higher parental level of education, the lower the probability of prolonged teacher CCHS but not temporary CCHS.Having one or both parents born outside Sweden increased the odds of temporary CCHS in preschool.

Discussion
This study aimed to provide a longitudinal understanding of teachers' temporary and prolonged CCHS, from preschool to year 1-3 of elementary school.Drawing on the bio-ecological model and the key factors related to person, we note that teachers were more likely to report prolonged concern for boys and for children for whom concerns for different aspects of development were present.The process-related factors (poor relationships with peers and low degree of school attachment) were significantly associated with concern for home environment in bivariate analyses but not when jointly analyzed with person-and context-related factors.In line with Svensson, Andershed, and Janson (2015), teachers' CCHS correlated with concern for different aspects of the Note. 1 Means with different subscripts (a = no concern; b = temporary concern preschool; c = temporary concern school; d = prolonged concern) were significantly different in post hoc comparisons.Significant values after applying the Bonferroni correction (p < 0.001).
Table 3. Multinomial logistic regressions with temporary CHS in preschool, temporary CHS in school, and prolonged CHS and person, process, and context related predictors at 1-and 5-year follow-up.
Fully adjusted models using 1-yr follow up (4-6 yr) Fully adjusted models using 5-yr follow up (8-10 yr) children's health and development.When CCHS arose (5-year follow-up), concerns for all aspects of the children's development had gradually increased since preschool.The greatest proportions as well as highest level of teacher-rated concerns for children's health and development were found among those whom teachers have reported prolonged CCHS.The study also shows that concern for the children's health and development persist and increase over time, regardless of whether the child has a disability or not.The connection between CCHS for children with disabilities appears rather temporary (only during preschool).
The result highlights the importance of recognizing how risk of maltreatment is related to multiple risks factors regarding the child's development.As identified in previous research, child maltreatment has negative impact on a child's development (Bunting et al. 2018;Fry et al. 2018;Gilbert et al. 2009;Grummitt et al. 2021;Hughes et al. 2021).Referrals to CWS should therefore include information about the child's development from several aspects as well as change over time.Such detailed information, which today is usually missing in referrals (Kalin et al. 2022), could contribute to identifying the need for early interventions and constitute an important basis for CWS' continued assessment.
In line with previous studies (Kuruppu et al. 2020;Noumana et al. 2020;Svensson, Andershed, and Janson 2015), our findings also particularly highlight the importance of context (teacher-parent contact, parental level of income and education, and country of birth).Poor contact between teachers and parents also showed a distinct longitudinal pattern: Teachers' contact with parents improved when CCHS ceased, deteriorated when CCHS arose and was stably poor when CCHS persisted.Poor contact between teachers and parents also appears to be the single strongest factor in understanding when concerns for children's home situation arise, cease, or persist over time.The study thus provides a deeper understanding of the previously identified importance of relationships between professionals and families (Gilbert et al. 2009;Kuruppu et al. 2020;Kvist et al. 2014;Piltz and Wachtel 2009;Svärd 2017;Svensson and Janson 2008;Talsma, Boström, and Östberg 2015).Although these findings contribute to a deeper understanding of the complexity of vulnerability, it also implies a risk for bias.On one hand there is a risk that teachers fail to detect children at risk of maltreatment, as long as their parents maintain expected contact with preschool and school, have relatively high socioeconomic status, and are of Swedish origin.On the other hand, there is a risk that parenting practices and ways of communication that do not fit the Swedish norm might lead to CCHS irrespective of the presence or absence of other risk factors related to child maltreatment.As shown in previous research, cultural differences between educational personal and families can result in racial disproportionality and disparity in reports of maltreatment by educational personnel (Krase 2015).Thus, our findings, showing that teachers are more likely to rate higher CCHS for children born outside Sweden, could at least partially be explained, or caused by racial bias.In line with this, our results should also be interpreted in relation to recent findings showing that teachers largely refrain from referring non-immigrant, mid-high SES-family children to the CWS despite high levels of severe child self-reported vulnerability (Kalin et al. 2022).Overall, Kalin et al. (2022) identified distinct associations between referrals and socioeconomic factors and country of origin.Our findings thus highlight the importance of recognizing signs of maltreatment, regardless of the parent's education level, income level and origin.To achieve this, teachers may require a better understanding of how these factors can impact their ability to identify children at risk for maltreatment.Without such knowledge, there is a risk that they rely on their own (sometime narrowed) culture lens, during this process.This is particularly relevant if they perceive the definition of what qualifies as children at risk for maltreatment to be unclear (Berger, McDaniel, and Paxson 2005).
In conclusion, this study has provided a longitudinal understanding of teachers' temporary and prolonged CCHS, from preschool to year 1-3 of elementary school.Teachers in both preschools and schools identify three different groups of children; children with a low risk of maltreatment (no CCHS), children with a medium risk of maltreatment (temporary CCHS), and children with a high risk of maltreatment (prolonged CCHS).With a focus on factors related to person-, process-and context, the study has increased the knowledge of how professionals detect children at risk of maltreatment within a preschool and school context.In particular, the findings highlight the importance of the contact between teachers and parents as well as the awareness of risk for bias, in understanding when CCHS arise, cease, or persist over time.

Strengths and limitations
SOFIA is the first known study that examines preschool and schoolteachers' concerns for children's home situations over time.Another strength is the high response rate and relatively large sample.Through the application of the bio-ecological framework and the inclusion of more specific school related factors, this study expands previous findings (Svensson, Andershed, and Janson 2015) and contributes to a deeper understanding of the complexity surrounding identification of children at risk of maltreatment.
Limitations include the fact that the survey do not cover any direct questions regarding suspicions of child maltreatment and the obligation of teachers to refer matters to the CWS.Another limitation is that the question of CCHS include several relevant aspects in one question; parental capability, child's safety and living condition.This limitation may be of extra importance for the association found between non-Swedish born children and increased reports of CCHS, since research show that ambiguous definitions of what constitutions maltreatment may influence bias (Berger, McDaniel, and Paxson 2005).These limitations are largely because the aim of the SOFIA-study is to investigate children's health and development in a wider perspective, not only children at risk of maltreatment.The underrepresentation of parents born outside of Sweden represents another limitation.

Practical implications
To reduce the risk of bias, such as failing to detect children due to well-functioning contact between guardians and teachers or due to preconceived notions about families from other cultures, the present study highlights the importance of communication; between teachers and children as well as parents from all backgrounds about different aspects of the child's development, school-related factors, and home conditions.Also, the present results point towards the importance of raising awareness of the significance of origin and culture in referral processes, and to offer teachers training in how to detect children at risk of maltreatment.Such training would enhance the quality of teachers' referrals rather than just focus on policies to increase the frequency of reporting.The results may also contribute to more accurate decision-making of professionals when planning for early interventions.For example, an awareness of the distinction between different levels of risk over time, opens up for more nuanced decisions together with children and parents, within preschools and schools and in collaboration with health care and the CWS.The results support the development of early coordinated initiatives as a complement to the CWS' more comprehensive measures and intervention (National Board of Health and Welfare 2021).Future research should acknowledge the importance of factors related to person, process, and context in detecting children at risk for maltreatment.

Table 1 .
Distribution of teacher-rated grade of CHS from 1-year follow-up (preschool) to 5-year followup (3rd grade of elementary school).

Table 2 .
Person-, process-and context-related factors at 1-and 5-year follow up in relation to CHS.