Introduction

Bowlby (1969/1982) emphasized the paramount importance of attachment relationships: whether secure or insecure, these relationships serve multiple key functions in child development. As highlighted by Bowlby (e.g. 1973) there are, consequently, substantial risks associated with prolonged and repeated separations from caregivers, and this insight was pivotal for his formulation of attachment theory (e.g. Duschinsky, 2020). At the same time, low caregiver sensitivity, frightening/frightened behavior, and maltreatment predict insecure and disorganized attachment and suboptimal child development (Ainsworth et al., 2015; Groh et al., 2017; Madigan et al., 2023). Child protection assessments hence actualize key attachment-related questions, and reconciling children’s need for continuity in attachment relationships with sensitive care can be difficult.

Not surprisingly then, attachment theory has become heavily drawn upon in child protection contexts (Melinder et al., 2021; Søberg Bjerre et al., 2021). Concerns have, however, been voiced about misconceptions that may result in erroneous decision-making concerning out-of-home placements (Granqvist, 2016; van IJzendoorn et al., 2018b). Attachment scholars have therefore published international consensus statements to counter misinformation and provide recommendations for evidence-based use of the theory (Forslund et al., 2022; Granqvist et al., 2017). Yet, there is scarce empirical knowledge about the actual understanding and use of attachment in child protection assessments. Recent studies of such assessments have corroborated the concerns expressed in the consensus statements (Beckwith et al., 2022; Hammarlund et al., 2022), but it is unclear whether or not the misconceptions translate to documentation and decision-making in court. Therefore, the present study examined the perception and application of attachment concepts in child-removal decision-making in Swedish court cases.

Attachment, separation, and child protection

Attachment theory comprises two analytical levels. The first, stemming from Bowlby’s (1969–1980) foundational theoretical and empirical work (Bowlby, 1944; Bowlby et al., 1952), concerns species-typical (or “normative”) aspects. As canalized (Waddington, 1957) by evolution, infants are strongly predisposed to develop attachment relationships to their caregivers (or “attachment figures”). For protection, and ultimately survival, children show attachment behavior (e.g. clinging, following) toward their attachment figures, maintain proximity, and protest involuntary separations. Children who have developed attachment also tend to approach their attachment figures as their primary sources of comfort when vulnerable (i.e. as safe havens), and use them as bases from which to explore (i.e. as secure bases). These key functions distinguish attachment relationships from other relationships and are crucial for child development. Reflecting the importance of these relationships, children react with intense anxiety upon involuntary separation, and mourn the loss of their attachment figures (Bowlby, 1973, 1980).

Almost all children develop attachment relationships, because almost all children have caregivers who are sufficiently available and responsive for this species-wide phenomenon to develop. Most children also experience continuity in their attachment-relationships, because their caregivers are only replaced or lost under exceptional circumstances. The centrality of the species-typical aspects of attachment is thus mainly seen following severe perturbations to the development and maintenance of attachment relationships. Such events, fairly uncommon in many countries, include institutional care (Ismayilova et al., 2023; van IJzendoorn et al., 2020), and prolonged separations from and losses of caregivers due to parental work migration (Hewage et al., 2011) and war and forced migration (Jones-Mason et al., 2021). Notably, such disruptions are linked to grave developmental risks, including for drug abuse and suicidal behavior (van IJzendoorn et al., 2020).

In many countries, child protection assessments constitute a particularly salient situation that may result in disrupted attachments through out-of-home placements. In Sweden, 26 500 children were placed in out-of-home care during 2022 (32% involuntarily), and a majority of these children experience placement breakdown, and thus discontinuous care via multiple placements (Socialstyrelsen, 2023). Attachment theory constitutes a key perspective in child protection, and attachment scholars have recently reiterated the importance of its species-typical aspects in this context (Forslund et al., 2022; van Ijzendoorn & Bakersmans-Kranenburg, 2024). More specifically, they have emphasized that children need continuity of good-enough care from attachment figures, that extreme caution should be exercised in disrupting attachment relationships, and that the theory is primarily relevant for directing supportive interventions. The potential benefits of supportive interventions must also be contrasted with the notable risk, following child removal, of placement breakdowns. Indeed, one-fifth to two thirds of all long-term placements end prematurely, which constitutes an additional risk-factor (i.e. repeated attachment disruption; e.g. Konijn et al., 2019; Sallnäs et al., 2004). Cognizant of these issues Bowlby (1958) argued that:

It is because the removal of a young child from his home to the care of strangers is usually productive of intense anxiety, misery, and anger (often even when the home he leaves is a bad one), and also because it all too frequently leads to further moves, that I believe it should be undertaken with great circumspection. Efforts to improve the home, though requiring time, skill, and patience, are in many cases preferable in the long run.

The second level of attachment theory concerns dyadic/individual variations in attachment quality (secure, insecure, disorganized), originating in pioneering observations by Ainsworth et al. (2015), and extended by Main and Solomon (1986, 1990). This level is based on variations in children’s expectations about their caregivers’ responsivity toward comfort seeking, and is captured by distinct patterns of attention, affect, and behavior toward the caregiver following attachment activation. At the group-level, insecure attachment is reliably predicted by somewhat lower levels of caregiver sensitivity than is secure attachment (de Wolff & Van Ijzendoorn, 1997), and disorganized attachment is predicted by maltreatment and subtle forms of frightening caregiver behavior (Madigan et al., 2007; Schuengel et al., 1999). Relative to secure attachment, insecure and disorganized attachment also increase the risk for suboptimal developmental outcomes, such as slightly higher levels of internalizing and externalizing problems (Groh et al., 2017). These replicable, nomothetic associations highlight the potential appeal of using attachment quality in child protection. Through attachment quality, the theory may seem to offer solutions to the complex issues of assessing caregiving competence and future risk. However, scholars from various fields have observed and critiqued the use of attachment quality for such purposes (e.g. Hammarlund et al., 2022; Melinder et al., 2021; White et al., 2019).

The two analytical levels of attachment theory may occasionally be in conflict, such as when parents struggle to provide good-enough care. Depending on the severity of the risks, assessors and decision-makers are then tasked with the challenging responsibility of determining whether the home environment poses a risk to such an extent that out-of-home placement is in the child’s best interest. In such cases, continuity in attachment relationships (i.e. a species-typical matter), and sufficiently sensitive care (i.e. a matter of variation), can be difficult to reconcile. Given the centrality of attachment theory, it is crucial that child protection workers have a thorough understanding of the theory, its two analytical levels, and how the theory is applied in a supportive, evidence-based manner. The concerns recently expressed by attachment scholars (e.g. Forslund et al., 2022; Granqvist et al., 2017; van Ijzendoorn & Bakersmans-Kranenburg, 2024) mainly regard a fear that variation in attachment quality is given undue weight, overshadowing the foundational importance of species-typical matters. For instance, attachment quality is sometimes erroneously viewed by child protection workers as a strong determinant of child development, and as a reliable index of the child’s caregiving history (van IJzendoorn et al., 2018a).

Empirical research on these matters is scarce, and has largely corroborated the concerns expressed by attachment scholars (Beckwith et al., 2022; Søberg Bjerre et al., 2021). For instance, Beckwith et al. (2022), using Q-sort methodology, found that UK child mental health service practitioners differed from attachment researchers in the perception of certain aspects of attachment theory and its application. Some of the largest differences pertained to disorganized attachment; while attachment researchers strongly disagreed with the notions that childhood trauma is behind all cases of disorganized attachment, and that disorganized attachment can be used to identify maltreatment, practitioners were more positive. Moreover, Hammarlund et al. (2022) found that almost all Swedish child protection workers studied formed opinions about children’s attachment quality (absent validated assessment methods), and many also had overconfident beliefs about the usefulness of information about attachment quality. For example, a sizeable proportion of the respondents recommended out-of-home care for children with disorganized attachment, despite warnings that disorganized attachment by no means necessarily indicates child maltreatment (Granqvist et al., 2017). Yet, Foster (2023) found partially contradictory results in a UK sample of child protection workers. The respondents varied in their understanding and application of attachment theory, and most were ambivalent or critical about using attachment classifications. There was also no clear evidence of overconfident application, and disorganized attachment was not misconstrued as indicative of maltreatment. Yet, the links between academic attachment research and child welfare practice were weak (cf. Duschinsky et al., 2021). More research is clearly needed.

Child protection – a complex matter

According to the UN Convention on the Rights of the Child (UNCRC; UN general assembly, 1989), nation states have a responsibility to protect children from maltreatment and to implement appropriate legislative, administrative, social, and educational measures to achieve this goal (Article 19). Child out-of-home placement constitutes an important intervention to safeguard children’s interests when caregivers are unable to provide good-enough care. At the same time, states must respect the integrity of families, refrain from unnecessary interference, and complement caregivers through support. In all decision-making regarding children, the best interest of the child shall be a primary consideration (Article 3). Yet, determining what is in a child’s best interest is challenging. For instance, it requires a prognostic risk assessment based on the presumed outcomes of different alternatives. This, in turn, requires relying on developmental research regarding childhood experiences that increase palpable risks in child development. Decision-making in this context hence means walking a tightrope: underestimating risks can have devastating consequences, yet so can unwarranted intervention through child-caregiver separations (van IJzendoorn et al., 2020).

The decision-making systems in a country can assent to the removal of a child only after a certain threshold has been met. There are considerable variations among countries regarding where that threshold is set, how it is defined, and who the final decision-maker is, but there is usually a blend of social-administrative and legal decision-making. This socio-legal process entails the regulation of child welfare through legislation, ensuring adherence to the principles of the rule of law and due process, while being firmly grounded in social work practice and knowledge concerning children’s needs and development (Burns et al., 2016). Especially in matters involving state coercion, decision-makers must act according to the principle of proportionality and give priority to the least intrusive measure justified by the circumstances. The “socio” in socio-legal highlights the relationship between society and law. Law is filtered through a societal context and a changing knowledge base, which leads to international variations and shifts over time in what is deemed legally relevant and regulatable. This also forms the basis for perception of causal relationships crucial to the legal pattern of action (Westerhäll, 2005). Increased general interest in attachment theory may then also lead to this theory entering the court room.

Providing effective support to prevent child-caregiver separation, and obtaining necessary information for accurate assessment, can also be challenging. Like social work in general, child protection work is multifaceted, and conducted in diverse environments wherein practitioners employ various theoretical frameworks to identify and act upon the social needs of individuals, families, groups and communities. Not only are these needs often complex, but so are the relationships among social workers, families, and the organizational context in which child protection is carried out (Khoo et al., 2020). In Sweden, more than a third of the social workers who conduct child protection investigations has been subjected to or threatened by violence (Karsten, 2018). Organizational changes have also led to increased workloads, insufficient resources, and high staff turnover (Karsten, 2018; Bruhn et al., 2020). Consequently, the average work experience among Swedish child protection workers has decreased significantly (Tham, 2018), and high levels of stress are common (Travis et al., 2016). Complexity often increases further when there are disagreements between the family and the child protection workers regarding the child’s needs, and distrust and suspicion towards the child protection services’ involvement. According to a recent report commissioned by the Swedish Psychological Defense Agency (Ranstorp & Ahlerup, 2023), disinformation about Swedish social services has spread internationally. In some cases, threats against the Swedish nation were expressed, so the government took action to counter disinformation and protect employees (Regeringen, 2023). Nonetheless, child protection services cannot evade the responsibility of protecting potentially endangered children, and must act on the information available.

Judicial regulations on child protection in Sweden

Countries vary in their approach to child protection (Gilbert et al., 2011). Sweden is considered to have a family-service oriented system characterized by universal services and benefits. Public services primarily aim to support parents through these measures, along with targeted services tailored to specific needs (Höjer & Pösö, 2023). The main legal framework governing child protection is called the Social Services Act (2001:453). This act emphasizes collaboration between social authorities and families, and the promotion of child development through voluntary means. The social services of each municipality are responsible for child protection investigations. If an investigation reveals caregiving deficits that pose a palpable risk of detriment to the child’s health and development, and the family does not consent to voluntary intervention, the Social Services Act no longer applies. Matters are then regulated under the Care of Young Persons (Special provisions) Act (CYPA, 1990:52). Then, a social welfare committee consisting of laypersons determine whether to pursue care proceedings through the administrative court. The number of Swedish children aged 0–6 years in involuntary care has increased, although adolescents still constitute the majority (Sallnäs & Wiklund, 2017; Socialstyrelsen, 2016).

The care of young persons (special provisions) act (1990:52)

The CYPA is relevant when there is palpable risk to a child’s health or development due to the home environment (2 §) or the child’s own behavior (3 §), and necessary care cannot be provided with the consent of the custodians or the child (if older than 15; 1 §). For young children, risk is always linked to the environment rather than the child’s own behavior (e.g. criminality, substance abuse). The assessment is conducted in two steps. It must first be established that the child’s circumstances correspond to those intended by the law. This part is retrospective, based on environmental circumstances that have already occurred. The second part then concerns whether these circumstances entail palpable risk to the child’s health and development. This part is prospective and requires making a prognosis about the likelihood of a negative outcome that has not yet occurred. There is thus an environmental requirement and a prognostic requirement. In some cases, the courts may consider a sufficiently negative outcome as already established, in which case the requisite may be viewed as fulfilled without further prognostic judgments.

Preparatory work has concluded that the prognostic requirement presupposes a “complex and thorough risk assessment” (Government Bill 2002/03:53, pp. 81–82). It has also been specified that there cannot be a question of minor risk of harm: it must be established that the risk has such an impact that there is a clear need for care (Government Bill 1989/90:28). Further, the risk must be indicated by concrete (observable) circumstances. Precedents from the Supreme Administrative court further clarify that hypothetical reasoning about future events cannot form the basis for compulsory care (RÅ, 2009 ref. 64). Assessment of future risk becomes more difficult the younger the child is, due to difficulties substantiating assumptions with sufficiently concrete circumstances. Nonetheless, investigations must be based on assumptions that, as far as possible, are grounded in relevant observations of parents and children (HFD, 2011; note 101).

While preparatory works, precedents, and legal doctrine clarify applicable law, the law’s wording leaves a relatively wide scope for interpretation. The members of the court (judges, lay judges) primarily use a legal-dogmatic method based on legal source doctrine, where especially preparatory works and precedents provide guidance. However, preparatory works cannot cover all conceivable situations, and precedents do not match every single case. The members of the court, therefore, need to assess whether the allegations presented by the social services are substantiated, based on their own knowledge and experience. The members of the court therefore need to complement legal guidance with extrajudicial knowledge, for example, in the field of child development. The individual investigator and decision-maker are thus entrusted with a great deal of responsibility, with ample room for subjective judgements (Kaldal, 2010). Some understanding of children’s developmental needs and the likely consequences of a particular care deficiency are consequently necessary to make accurate predictions. The court’s appraisals of the relationships between caregiving quality and attachment security, and between attachment security and child development, are therefore relevant for the prognostic and environmental requirements. Yet, judges may have difficulties evaluating the scientific merits of psychological claims and may therefore have to trust the expert assessments (Neal et al., 2019; Scott & Emery, 2014).

The present study

The aim of the present study was to examine how attachment concepts are perceived and applied in judicial decisions concerning child removals. We addressed this matter in an interdisciplinary fashion, focusing on how extralegal considerations, based on attachment theory and research, are expressed and interpreted in the legal context, and influence legal practice. More specifically, we first conducted an inductive analysis to identify latent perceptions about attachment concepts. Secondly, we analyzed the evidential function and assessment of attachment concepts in relation to the legal (i.e. environmental and prognostic risk) requirements.

Method

As described in the following, the analysis combines summative content analysis with a legal dogmatic approach to provide a comprehensive understanding of how attachment is used in judicial decisions concerning child removals.

Data collection

The empirical data consisted of judicial decisions from the Administrative Court of Appeal, as well as related decisions from the first instance court, regarding applications under Section 2 of the Care of Young Persons (Special Provisions) Act (CYPA) during 2019. The inclusion criteria for the qualitative analysis were that 1) the decisions concerned a child aged 0–2 years, 2) emotional neglect constituted a substantial or dominant motive for care, and 3) the decision included references to attachment concepts. The age range was based on research suggesting that attachment may be referenced to a greater extent when assessments concern infants and toddlers (Hammarlund et al., 2022). The focus on decisions where emotional neglect was a prominent motive for care was motivated by a hypothesis that attachment may play a particularly prominent role in such cases, given the theory’s focus on child-caregiver interactions and relationships, and its influence on children’s socio-emotional development.

Judicial decisions are public records in Sweden and openly available according to the principle of public access. Access to decisions from the first instance courts usually requires a case number or personal identification number for the child and/or the guardians. However, decisions from the Court of Appeal are available in legal databases (e.g. JP Socialnet) through which the first author (SNL) gathered all CYPA-decisions regarding children within the age group. Related decisions from the first instance court were attached to these decisions. This phase of the data collection yielded 173 decisions from the Court of Appeal (including pertinent decisions from the first instance court). Next, the first author (SNL) coded all decisions based on the motives for considering care according to CYPA. Statements concerning emotional neglect were present in 99 decisions (57%). These decisions were then sorted into three categories based on the degree of descriptions of emotional neglect (relative to other motives for care), namely (i) small, (ii) substantial, or (iii) dominant. Of the initial 173 decisions, 56 decisions (32%) included emotional neglect to a substantial or dominant degree. The rationale for this approach was to maximize the likelihood that statements concerning attachment underwent legal assessment. According to 30 § in the Administrative Court Procedure Act (1971/291), the “court´s decision shall state the reasons that determined the case,” but if there are several cumulative grounds, not every single circumstance used in the prognosis needs to be proven in the court’s decision-making (Diesen, 2003). Of the 56 decisions involving neglect to at least a substantial degree, 28 (50%) contained references to attachment. These cases constituted the final sample that was analyzed qualitatively.

Material

Each judicial decision opened with an account of the parties involved, including the guardians, the social welfare committee and the child through his/her representatives and guardian ad litem. This was followed by an account of the parties’ position on the issue and their opinions. Finally, the court described how they assessed the statements, and whether the requisites for CYPA were regarded as substantiated. The decisions from the first instance courts were generally longer and more descriptive than those of the Administrative Court of Appeal, with the exception of cases in which the Court of Appeal made a different assessment than the Administrative Court, which then had to be more clearly justified.

The structure and language of the judicial decisions follow an evidential logic that characterizes the text material, albeit to varying extents. In some cases, the different stages of the assessment are visible in the sense that the court breaks down the issue into evidential themes that are systematically handled. In other cases, this logic is more implicit. The decisions, including the recounts of the parties’ key statements, are drafted by the court and does not necessarily reflect the full extent of the parties’ claims as they were expressed at the oral hearing, or in the written care application from the social welfare committee.

Analysis

We first conducted a Summative Content Analysis (SCA; Hsieh & Shannon, 2005) of the 28 decisions that met the inclusion criteria. SCA is used to identify, quantify and, most importantly, understand contextual use of certain words or content in units of (usually) descriptive text, such as textbooks or manuscripts. The analysis was first done individually by the first (SNL) and the second (FI) authors, who read through the material several times to understand the explicit and latent meaning of the data. Each coder began the analysis with identifying text units where manifest frequencies of attachment terms were present. The text units could be one or several sentences long and ranged between 11 to 206 words (mean = 60; median = 102; type value = 56). We also noted the court’s decision (approving or declining the application), and the source of each statement concerning attachment (the social welfare committee, the guardian/s, the child’s legal representative and guardian ad litem, witnesses or experts, or the Administrative Courts ruling). Next, we conducted an inductive analysis to identify the latent perceptions of attachment concepts and sorted them into categories and subcategories. We then compared and discussed the results, and agreed on an interpretation, in continuous discussion with the third (PG) and fourth (TF) authors.

For the analysis of the legal assessment of attachment concepts, we integrated content analysis with a legal dogmatic approach (Kleineman, 2018). This involved analyzing the use of attachment in the decisions through the lens of the legal definition of palpable risk, as formulated within legal sources (precedents and preparatory work).

Ethical considerations

The study was based on case material made public via the Swedish principle of publicity (swe: offentlighetsprincipen). As such, we did not recruit any participants but instead analyzed publicly available information. To ensure the anonymity of all individuals involved, the material was fully anonymized prior to our analysis. Since our research focus is not on individual cases of particular interest but on how attachment is perceived and applied in court decisions collectively, we have also chosen not to display the original case numbers in the analysis. Furthermore, the initial empirical work was conducted by the first author as part of her master’s thesis in social work, which was covered under the educational exemption from the Swedish legislation on research ethics.

Results

The results from the summative content analyses (SCAs) are presented in two parts, mirroring the research questions. First, we present results concerning the perception of attachment concepts. Second, we present results regarding the role of attachment concepts in the decisions and the assessment of attachment in relation to the legal requisites.

Perception and application of attachment concepts

The SCA of formulations and statements about attachment yielded three categories: direction of attachment relationships, variations in descriptions of attachment quality, and unsystematic assessments of child attachment and pertinent caregiving behavior. Within these categories, recurring approaches to applying attachment concepts were identified.

Direction of attachment relationships

Formulations indicated variations in the perception of the nature of attachment relationships. This was particularly notable concerning the perceived direction of attachment bonds, which were described in three distinct ways. In some cases, children were described as developing attachments to their caregivers, in line with attachment theory. In other cases, caregivers were depicted as developing attachments to their children, or rather failing to do so. In yet other instances, attachment was described as developing between children and caregivers, leaving its direction unclear. Statements that suggested a reversed directionality, with caregivers expected to develop attachments to their children, were most common. This view was represented by one or more of the parties in 9 of the 28 decisions, while descriptions of children attaching to their caregivers were only represented in 5 decisions. In one case, the social welfare committee stated that the deficiencies in care concerned “the [father’s] attachment to the child” (case 9) and in another, summarized the mother’s capabilities by stating that she was “lacking regarding emotional attachment to the children” (case 14). In yet another case, the parents were described as having undergone interaction therapy “to strengthen their attachment to the child” (case 7). Adding to the confusion, reversed roles were also suggested by parents, most commonly when opposing the claims from the social welfare committee. For instance, in one case, the parents stated that they were indeed attached to their child (case 1). In another case, a mother’s legal counsel gave context to her caregiving difficulties by stating that she suffered from a postpartum depression that “influenced her ability to attach to the child” (case 4). Finally, attachment was described as a mutual process in 7 decisions, through statements from the social welfare committee such as “ … there are attachment issues between the children and their mother” (case 8) or “the attachment between the child and mother has developed slowly” (case 17). It is unclear to what extent these vague descriptions imply that children and caregivers are perceived as forming mutual (horizontal) attachment relationships, but this way of describing attachment relationships was recurring.

Variations in descriptions of attachment quality

Attachment quality was described in varying ways. In a minority of the decisions (8 out of 28), terms originating from attachment theory, such as “insecure” or “disorganized,” were used. In these cases, the observed behavior of the parent and/or child was seen as an “indication” or a “clear sign” of a certain type of attachment quality. For example, in one case the court concluded that: “the family therapists’ overall assessment is that the children rarely seek support or comfort from their parents and that there are indications that all the children have developed a disorganized attachment pattern” (case 24). It is, however, unclear whether rarely seeking support or comfort is to be understood as indicative of disorganized attachment, or as a separate statement.

More commonly, other terms than those typically used in attachment theory were used to describe attachment (quality), such as “deficient” (case 17), “poor” (case 7), or “problematic” (case 7). Sometimes, concerns were conveyed through broad statements such as “there are attachment issues between [the child] and the mother” (case 28).

In the vast majority of cases, attachment-related statements concerned attachment quality. In only a small minority of the cases the focus was on attachment as a species-typical phenomenon (e.g. the presence or absence of attachment relationships). For instance, a family therapist representing a mother asserted that she “has observed an attachment between mother and child” (case 11), and a child’s council stated that “thus far, the foster family believes that the child is not attaching [to the father] during the home visits” (case 8).

Unsystematic assessments of child attachment and pertinent caregiving behavior

The assertions concerning attachment (quality) were conclusions drawn from some form of observations. Yet, no scientifically validated attachment measure seems to have been used. Instead, the assertions were supported by information from unsystematic observations, which were rarely described in much detail.

Clear descriptions of parental behavior underlying assumptions about attachment quality were scarce (10 of 28 decisions) and tended to take the form of broad representations such as “lacking in emotional availability” (case 24) or having “difficulties prioritizing the child” (case 19). In a few instances, there were clarifications through descriptions of specific parental behaviors when interacting with the child. For example, one mother’s “difficulties interpreting the child’s cues correctly” was illustrated by behaviors such as “not comforting the child when she is upset … , not making eye contact with the child, and handing over the child’s care to others” (case 17).

It was common to infer attachment (quality) through descriptions of children’s discrete behaviors (12 of 28 decisions). Sometimes, these were derived from observations of the child interacting with the parents, but relatively frequently, the child’s behavior was described without a relational context and thus portrayed as an individual characteristic of the child. In yet other instances, the observations were based on the child’s interaction with social workers or foster parents. Children were described as “having difficulties with eye contact” (case 7), being “indiscriminate in seeking contact with other adults.” (case 8), or having difficulties “giving responsive smiles, and making responsive sounds” (case 21).

Descriptions of the children’s behavior were sometimes extensive. In case 7, the child’s behavior appeared to carry considerable significance as an indicator of attachment, and was regarded as indicative of attachment disorder:

The child showed low persistence and quickly became overstimulated. Maintaining eye contact with the child and eliciting curiosity and smiles from him to a normal extent was difficult. He did not smile back when adults smiled at him, instead averting his gaze during attempts at eye contact, which is a sign of avoidance on the child’s part and a serious symptom of attachment disorder. Additionally, the fact that he slept more than what is normal for a child of his age is a protective mechanism in response to poor attachment.

It is unclear what is meant by attachment disorder in this case; whether it refers to Reactive Attachment Disorder (RAD), insecure attachment (i.e. “avoidance”), or a general concern regarding the child’s social behavior. RAD is also a very uncommon psychiatric diagnosis, comprising symptoms which overlap with much more commonly occurring, neurologically based conditions such as autism (Minnis et al., 2013). The child in this case was five months old at the time of the hearing in the Administrative Court of Appeal, and the observations had been made in an assessment home, where the family was placed after the child’s birth, and/or in the residential care home where the family received treatment thereafter. The pertinent observations must have been made from the time the child was a newborn up to a maximum of three months of age, considerably preceding the age when infants can be expected to have established clear-cut attachment relationships (i.e. the second half of the first year). The parental behavior that allegedly caused the child’s symptoms were also sparsely and vaguely described. For instance, it was stated that the parent “had difficulties focusing” and “often broke contact with the child.” Alternative explanations, such as early signs of autism in the child (see Jones & Klin, 2013), which can make attachment observations more difficult, were not explicitly considered.

In another case, the court similarly concluded that the child, who had spent a few weeks at an assessment home with his parents immediately following birth, was negatively impacted by the parents’ perceived deficiencies. The parents’ alleged difficulties were not elaborated in more detail than “extensive and severe deficiencies in their parenting” and “unable to meet the child’s needs for care, emotional responsiveness, attachment, stimulation, protection, security, and stability” (case 6). However, there were detailed descriptions of the child’s behavior, which were regarded as symptoms of insufficient care. The child was, for example, described as “showing bodily tension” when held, and “averting his gaze and looking past the parents,” behaviors deemed “serious indications that the child’s ability to form attachments with key caregivers was jeopardized.” The court concluded that the child’s behavior had been confirmed by the emergency foster home, which reported that “the child was tense, did not seek eye contact, appeared unaccustomed to feeding situations, and slept a lot.” Again, the child´s perceived symptoms were assumed to result from inadequate parental care, but the reasons for this conclusion were not clearly substantiated, and alternative explanations were not considered.

The legal function and assessment of statements concerning attachment

In this section, we explore how attachment was referenced in legal arguments supporting and opposing CYPA, and the legal evaluation of descriptions of attachment in relation to the requirements stipulated by law (i.e. environmental and risk requirements). We also present findings regarding the overall logic and reasoning in the decisions, because this has important implications for how the legal evaluation of attachment concepts is carried out.

Findings regarding the logic and reasoning in the decisions

With the exception of two instances, the Administrative Court and the Administrative Court of Appeal granted the municipality’s applications for out-of-home-care. In the two cases (15 and 19) where either the Administrative Court or the Administrative Court of Appeal rejected the social welfare committee’s care application, the rejections were not based on considerations related to attachment. One of these cases (case 15) is not cited in this article, while the other (case 19, described below) is cited to exemplify broad representations of parental behavior and statements about future risk. Thus, in a large majority of all cases studied, the circumstances presented by the social welfare committees were regarded as substantiating the environmental and prognostic risk requirements. Attachment-related information was presented and evaluated in relation to both requirements, but to varying extent and degree of explicitness.

The decisions focused on alleged caregiving deficiencies while prognostic claims about future risk, and what made it palpable, occurred to a much lesser extent. The concluding section, where the court outlined the grounds for its assessment in relation to the legislation, often consisted of a summary of what the parties had presented, followed by a statement whether legal requirements were considered as fulfilled. The decisions thus offered little guidance and transparency about the courts’ reasoning, and we were therefore reliant on the social welfare committees’ descriptions and a few clarifications from the court itself.

When future (prognostic) risks were at all mentioned, it was through broad statements that offered little concrete information about what the risk entailed, such as “The child’s health and development are at risk of being harmed both in the short term and in the long term due to the mother’s lack of caregiving ability” (case 19). In consequence, a clear distinction was rarely made between the environmental and risk requisites. The two stages of the CYPA assessment, to clarify if there are inadequacies in child care, and conclude whether these inadequacies meet the presupposed level of risk, were obscured. Since a vast majority of the applications were approved, absent explicit prognostic statements, the representations of care deficiencies must implicitly be understood as meeting the prognostic requirement. The court also engaged in few independent analyses, where the arguments made by the social welfare committee were scrutinized. Combined with the high level of approvals, this indicates that the courts displayed great confidence in the assessments presented by the social welfare committee.

Attachment considerations were primarily used to support out-of-home placement

Attachment concepts were predominantly referenced by the social welfare committee and the courts to bolster the conclusion that there were caregiving deficiencies that posed palpable risks to the child’s well-being. Hence, attachment concepts were primarily used to advocate for the necessity of child out-of-home care. In one case, the social welfare committee referenced attachment as the desirable outcome of out-of-home-care, since it would provide the child with an opportunity to “attach to stable adults who can meet all of her needs” (case 26). In no cases were the presumed benefits of removals explicitly compared with other options, neither by the social welfare committee nor by the court. Parents referenced attachment (quality) in 13 of the 28 decisions, often through statements that explicitly or implicitly opposed the social welfare committee’s assertions. For example, in one case, the parents stated that “there is also nothing indicating that the child’s difficulties with contact and attachment are due to any deficiencies on their [the parents’] part” (case 6), and in another, “the father has a good attachment and interaction with the child” (case 26). Parents only referenced normative aspects of attachment in two instances. In the first, it was stated that “the interaction and attachment to [the child] will not improve by having [the child] taken into care. Separating [the child] from [the parent] could be harmful, and [the child] might suffer lifelong consequences” (case 17). In the second, it was stated that “what’s best for the children is to live with [their father], since the father is their primary attachment figure” (case 4). Consequently, attachment was rarely referenced to emphasize the importance of maintaining attachment relationships and the risks associated with prolonged separation from and losses of caregivers. In the few instances when this was mentioned, it was exclusively by parents. Instead, attachment-related arguments were primarily focused on children’s (future) attachment quality, and used to motivate child-caregiver separations.

Attachment and the risk requisite

Attachment was mainly referenced when describing alleged care deficiencies and thus primarily used to substantiate the environmental requirement. For example, in one case the court concluded that the investigation “indicates that there are attachment problems between the children and the mother […] which suggests that the mother has not been able to meet their needs for emotional security and attention” (case 8). Since the applications were approved when prognostic statements were absent or unclear, the descriptions of attachment as care deficiencies could also be understood as substantiating the risk requirement.

Attachment was explicitly linked to the risk requirement through prognostic statements in merely 5 decisions. In these cases, the descriptions centered on statements regarding risk for insecure or disorganized attachment. In one case the social welfare committee stated that the child is “at increased risk of developing an insecure attachment pattern due to the parents’ inability to meet his needs.” To clarify the risk, the social welfare committee concluded that “This has already occurred to some extent, as the child is perceived as somewhat withdrawn and does not seek contact in an age-appropriate manner” (case 25). In another case, insecure attachment was similarly referenced when describing future risks for a child whose parents had substance abuse:

There are also risks to the attachment. A prerequisite for a secure attachment is that [the child’s] parents respond to his signals and can provide a secure base. Parents with substance abuse problems may miss important signals from the child, jeopardizing development and posing risks to attachment.

In yet another case, the court concluded that the investigation showed that attachment between the child and its mother did not “function naturally” and that “the mother requires continuous support for the child to develop a secure attachment to her” (case 17). Secure attachment thus appeared to constitute a norm and insecure attachment a palpable risk.

Attachment was most often one of multiple substantial concerns. Yet, in a few instances, attachment seemed to constitute the main reason for child removal. For instance, in one case, the social welfare committee stated that:

After six weeks at the investigation home, it was suspected that the child suffered from impaired attachment to his parents. Therefore, the child was placed in an emergency foster home and thereafter in a foster family.

In another case, the social welfare committee stated that the “problematic attachment is the caregiving deficiency that poses the most serious harm to the child’s health and development” (case 7). These cases entailed few depictions of parental behavior, but extensive descriptions of the children’s perceived symptoms, which the court appeared to rely on as valid indices of caregiving deficiencies and palpable risk.

In yet another case, the court referenced attachment as a pivotal factor in mandating involuntary care removal: “Based on what has emerged in conversations with [the older] children, the observations made by the family therapists, and the assessment made by the family therapists of the children’s attachment patterns, the administrative court considers it likely that the municipality has made it probable that the children’s needs for emotional care are not being met” (case 24). Neither in this case, nor in any other cases, did it appear that a formal attachment assessment was conducted. The family therapists had indicated that the children had developed disorganized attachment, and the court seemed to equate this with having conducted an attachment assessment procedure.

Discussion

The present study examined the perception and application of attachment concepts in Swedish judicial decisions regarding involuntary removals of young children where emotional neglect was a prominent motive. Attachment concepts were referenced in 50% of the cases, primarily by the social welfare committee, and mainly to support child out-of-home placement. The application of attachment concepts was imprecise and characterized by misunderstandings, with descriptions often deviating from established theory. Conclusions about caregiving deficiencies were often based on concerns about child behavior and insecure attachment, yet descriptions of relevant parental behavior were limited. The risks associated with non-optimal attachment also appeared exaggerated when used as evidence of palpable risk. As discussed in the following, these findings corroborate the scholarly concerns about misconceptions of attachment concepts in child-protection work expressed in recent international consensus statements (e.g. Forslund et al., 2022; Granqvist et al., 2017; see also van IJzendoorn et al., 2018a). Our findings also extend the scarce empirical research on these matters (e.g. Hammarlund et al., 2022). Yet, while these findings point to alarming issues concerning the perception and use of attachment concepts in child-protection work, this does not automatically imply that the removal decisions studied here were incorrect. In other words, the decisions may have been correct, but poorly justified.

Imprecise application of attachment concepts

Attachment concepts were used imprecisely, and there were recurrent misconceptions. For example, attachment quality was commonly described through labels such as “deficient,” “poor” and “problematic,” which deviate from terminology in established attachment theory. Moreover, attachment-bonds were often described as inverted, with caregivers expected to develop attachments to their children and failing to do so. It should be noted that the case workers investigations are reviewed at multiple levels (managers, social welfare committees, judges, lay judges), and misconceptions about attachment therefore seem to be shared by the various actors in the decision-making process. These findings are perhaps explained by the divergent discourses concerning attachment concepts that have been observed in different domains. Notably, Duschinsky et al. (2021) discuss six discourse domains, with the child welfare domain perhaps more closely aligned with the popular (“allodoxic”) discourse domain than the scientific ones (i.e. developmental, social, clinical psychology and psychiatry).

Terminology such as insecure and disorganized attachment, and attachment disorder, were present in some cases. However, these labels were not supported by clear and relevant descriptions of concrete behavior. References to attachment disorders have also been found to be arbitrary and excessive, without a proper diagnostic process (Allen & Schuengel, 2020; Woolgar & Baldock, 2015). In the present study, the label did not seem to refer to the psychiatric diagnosis, and alternative explanations (e.g. heritable neurological conditions) were not considered. Taken together, statements and conclusions about attachment were often misguided.

The multiple misconceptions and imprecise uses of attachment concepts documented in this study are highly problematic. First, these findings suggest that attachment concepts have been lost and distorted in translation, in other words that there is a notable researcher-practitioner gap (Granqvist, 2016) or a lack of shared definitions and meanings among researchers and practitioners (Beckwith et al., 2022; Duschinsky et al., 2021). This unfortunate situation likely has multiple reasons. For instance, there has not always been consensus within the attachment community, which has debated relevant topics such as the potential benefits and risks of using assessments of attachment quality in child protection (e.g. van IJzendoorn et al., 2018a vs. Spieker & Crittenden, 2018). This may have contributed to inconsistencies in curricula about attachment, and to misunderstandings among practitioners. Moreover, concepts such as sensitivity and (in)security have technical meanings in attachment theory and research, while having different connotations in everyday language (Duschinsky et al., 2021).

Second, misunderstandings and imprecise application of attachment concepts can have significant consequences for children and families, if they influence the courts’ decision-making. Yet, we reiterate that the children in the present study may indeed have been subjected to care deficiencies that posed a palpable risk of harm, and the removal decisions may thus have been correct, even if poorly justified in the decision protocols. Though speculative, accurate but nonsystematic observations and assessments may have been communicated using attachment terminology, to legitimize rightful concerns through references to a developmental theory with high scientific status. While the misconceptions and imprecise uses of attachment concepts uncovered here do not necessarily suggest erroneous decisions, they do create ambiguity in the judicial statements and exacerbate the risk for errors. The potentially grave negative consequences of such errors, including the risk for unwarranted child removals, underscore the critical need for accurate assessments.

Third, the misunderstandings and imprecise application of attachment concepts are concerning from a welfare state and legal perspective. Child removals constitute one of society’s most intrusive measures and, when erroneous, violate children’s and parents’ rights. It is hence crucial that these processes adhere to established principles of the rule of law, such as proportionality, legality and legal certainty. The pertinent legal decisions must hence be clear and predictable, and the reasons for removals and the evidence supporting the decisions must be explicitly stated and well-founded. This is of great value for the perceived legitimacy of the social authorities. If citizens perceive inconsistencies in crucial judicial decisions, or view them as arbitrary, it could weaken the delicate relationship between social authorities and families in need of support, thereby impeding cooperation and consensual support. Recent developments in Sweden also illustrate that such perceptions can be exploited to spread misinformation about child removals, potentially posing risks to national security (Ranstorp & Ahlerup, 2023).

Assessing caregiving deficiencies through attachment and child behavior

The legal requirements necessitate proving a causal link between environmental factors (i.e. caregiving deficiencies), and a palpable risk to the child. Yet, in the present study, caregiving deficiencies were often inferred from concerns about the children’s behavior and their attachment (quality), often absent clear descriptions of pertinent caregiving behavior or through broad generalizations. Moreover, the pertinent child behavior described was not empirically linked to insecure or disorganized attachment, and by no means necessarily indicate maltreatment. Difficulties with eye contact may, for example, be due to neurological problems underlying autism (Jones & Klin, 2013). A causal link between caregiving difficulties and palpable risk to the child, via insecure attachment, did therefore not appear substantiated.

Hammarlund et al. (2022) found that Swedish child protection workers often viewed a given child’s attachment quality as an indicator of the parents’ caregiving competence, as a reflection of the child’s well-being, and as a predictor of the child’s future development. Such views were discernible in this study as well, with children’s attachment quality treated as a reflection of their parents’ caregiving ability. Attachment scholars have previously emphasized the risks associated with assuming, for example, that disorganized attachment indicates maltreatment (Granqvist et al., 2017; White et al., 2019). While disorganized attachment is more frequent among maltreated children, maltreatment is by no means the singular cause of disorganized attachment – indeed, there is no such singular cause. Also, almost half of the infants with maltreating caregivers develop organized attachment relationships (Madigan et al., 2023).

Research establishing links between caregiver sensitivity and child attachment quality (Ainsworth et al., 2015/2015; de Wolff & Van Ijzendoorn, 1997), and between attachment quality and subsequent child development (Groh et al., 2017; Vaughn et al., 2019), have made important contributions to the understanding of children’s needs and how caregiving behavior can influence children’s development. However, attachment measures have been developed and validated for research on the group-level, which does not automatically translate to individual-level diagnosis or prognosis (Neal et al., 2019; van IJzendoorn et al., 2018a). While research typically establishes links between variables, and offer statistical generalizations, child protection workers investigate connections among individual children and their caregivers, to formulate personalized prognoses. It is important to note that pressure mounts for scientifically grounded (“evidence-based”) assessments, and it is thus perhaps not surprising that group-level findings may transition prematurely into child protection investigations. Indeed, a prevalent issue in court proceedings involves the potential for logically flawed conclusions in individual cases drawn from research conducted at the group level (Faigman et al., 2016).

Attachment and palpable risk

The court typically stated that the investigations had established palpable risk, without specifying what made the risk palpable, for example by clarifying when the potential harm might occur and how it could affect the child. There were also no apparent differences between the examination of the environmental and prognostic risk requisites. Consequently, the decisions provided little guidance on how palpable risk should be understood legally. We recognize that child welfare work is inherently uncertain (e.g. Fook, 2007) and that substantiating the legal prognostic requirements with clear risk assessments is challenging. Yet, since CYPA centers around ongoing or future risks, it is remarkable that these were not clarified to a greater extent.

Concerns about children’s attachment quality were recurrently referenced as a risk, with attachment insecurity a proxy for “palpable risk.” This naturally raises the question of the risks associated with insecure attachment. Roughly 50% of children in the general population develop insecure attachment to at least one caregiver (Madigan et al., 2023), and the effect sizes of the associations between insecure attachment and subsequent development are small to moderate (Groh et al., 2017). Insecure attachment can hence not be equated with palpable risk. Legislative preparatory works, and case law, emphasize that for a risk to be palpable, it must not be a matter of a negligible, unclear, or distant risk, and there must be concrete circumstances indicating risk of harm (Government Bill 1989:90). The imprecise language surrounding attachment concepts, and misconceptions about the risks associated with insecure attachment, indicate it was not clear that the legal threshold for palpable risk was met.

In sharp contrast to the recommendations provided in a recent scholarly consensus statement (Forslund et al., 2022), references to attachment concepts almost exclusively pertained to children’s perceived attachment quality (i.e. the variation/individual-differences level of attachment theory) rather than the importance of development and maintenance of attachment relationships (i.e. the species-typical, normative level of the theory). Risks associated with prolonged child-caregiver separations – and the high probability of further separations upon entering the foster care system – were never referenced by the social welfare committees or the courts. A crucial aspect of the decision-making must be whether the risks associated with remaining in the family are greater than those associated with child removal from the caregiver (cf. the prognostic risk requirement). Although speculative, the absence of such considerations may perhaps be explained by the timing of the decision-making process. More specifically, such considerations may have been prominent earlier in the process, when offering voluntary support, and when deciding whether to apply for compulsory care. In line with this hypothesis, Hammarlund et al. (2022) found that child protection workers regarded “guiding supportive interventions” as a main purpose of using attachment theory. The material we have studied, therefore, cannot clarify to what extent and how the risks associated with child-caregiver separations are weighted. However, considering the notable risks associated with losses of caregivers and prolonged separations, and in particular Bowlby’s careful descriptions of children’s strong reactions to them (i.e. the protest-despair-detachment cycle; Bowlby et al., 1952), such considerations should be an integral part of the assessment.

Challenges in child protection as a socio-legal practice

Our results must be understood in relation to the immense complexity that necessarily characterizes child protection work, and the challenging conditions under which child protection workers operate (Gibson et al., 2018; Lavee & Strier, 2018). For instance, child protection workers face increased demands for scientific rigor, while conducting an inherently difficult task in predicting the consequences (and etiology) of multifaceted problems. Moreover, organizations increasingly fail to provide necessary education, time, and resources, while individual cases become increasingly complex. This may, understandably, contribute to simplified explanatory models and misunderstandings. Out of respect for this profoundly difficult task, it is important to emphasize that the judicial decisions cannot capture the full complexity of the decision-making process. The decisions are also drafted by the court, and may not fully reflect the parties’ original statements, nor the nuances conveyed during the oral hearings and investigations. However, the decisions should be expected to present the court’s interpretation of the statements, and a sufficiently detailed account of the grounds for its decision. The Administrative Court is ultimately the final decision-maker, and should have a sufficient understanding of attachment concepts when these are drawn upon.

Attachment scholars have cautioned against assessments of attachment quality in child protection decision-making, and have emphasized that the theory should primarily be used as an overarching framework for understanding factors that contribute to children’s wellbeing and development, and for guiding supportive interventions (Forslund et al., 2022). When parental capacity is addressed, valid measures of important aspects of caregiving behavior should be prioritized, instead of taking the indirect and potentially misleading route through attachment quality. Relevant caregiving behaviors, emphasized by decades of attachment research, include caregiver sensitivity and safe haven provision, and frightening behaviors that may make children perceive their parents as a source of alarm (Madigan et al., 2007). Since well-validated (research) measures of pertinent caregiving behavior are difficult to use in child protection, due to limited time and resources, attachment scholars are currently developing and validating measures that may be more feasible to use in such settings (e.g. Forrer et al., 2024; Madigan et al., 2021).

Methodological considerations and future directions

This study is limited in that only decision protocols from one country (Sweden) was studied, and merely during one year. The number of protocols was also moderate, since we focused on protocols dealing with alleged emotional neglect and social deprivation regarding children aged 0–2 years. Moreover, the database only gave access to protocols that made it to the Appeals court. Each limitation naturally limits the generalizability of our findings, and present opportunities for future replication and extension studies. We also encourage research on how judges and other legal decision-makers perceive attachment concepts, especially vis-à-vis legal evidentiary standards.

Notwithstanding these limitations, the study makes an important contribution by extending previous research on the understanding and use of attachment concepts in child protection. More specifically, while previous studies have asked child protection workers about their understanding and use of attachment concepts, the present study sheds light on how these concepts figure in the succeeding court work and legal documents composed to justify (and oppose) child out-of-home placements. Another strength is that the material was coded by two coders, and consolidated through discussions with each other and the third and fourth authors.

Conclusion

The law mandates that a causal relationship between deficits in care, and a palpable risk to the child, be established for involuntary out-of-home care to be warranted. Assessment and decision-making must hence draw upon theory and research on caregiving and child development. Attachment theory and research uphold scientific status and focus on matters of high relevance to child protection. However, attachment researchers may not have communicated sufficiently with practitioners about the nuances – constraints, limitations, and caveats – surrounding attachment concepts, to prevent the widespread misunderstandings that we currently witness in circulation. Such misunderstandings may make causal relationships appear more well-founded than they actually are and inadvertently impact decision-making. At the same time, child protection workers face increased work-related demands, and limited resources, which makes it difficult to stay up to date with developmental theory and research. It is perhaps not surprising, then, that concerns about children’s attachment quality were referenced frequently in the legal decisions that we studied, and that the use of attachment concepts was imprecise and fraught by misconceptions. We call for more resources to child protection work, which represents a crucially important and difficult task. We also call for attachment researchers to be more active in knowledge dissemination, and to collaborate with child protection workers to address pertinent needs and counter the researcher-practitioner gap. Crucially, whether or not established via validated measures, insecure attachment does not amount to palpable risk, and cannot be used as a proxy for caregiving deficiencies.