Achieving agreement on service needs in child protection. Comparing children’s, mothers’ and practitioners’ views over time and between approaches

Abstract

Purpose

While studies on service users’ participation and their perceptions on the quality of services exist, agreement between family members’ and practitioners’ assessments of the family’s situation has received less interest. The purpose of this paper is to investigate agreement and its effect on outcomes by comparing the viewpoints of three groups of informants (children, mothers and practitioners) in the context of statutory child protection in two study groups – one applying a systemic approach (SPM) and a service-as-usual control group (SAU).

Design/methodology/approach

A quasi-experimental repeated-measures study design was applied. Outcome data comprised 112 cases (SPM cases n = 56 and SAU cases n = 56) at three sites. Data was collected from all participants at baseline and six months later.

Findings

First, practitioners’ analyses of a child’s need for protection did not meet family members’ expressed need for help. Second, child–mother agreement on the need for service intervention at T1 predicted a decrease in practitioner-assessed abuse or neglect from T1 to T2. In this sample, no differences were found between the two groups.

Originality/value

This study highlights the importance of making explicit the viewpoints of children, parents and practitioners in casework and research to improve understanding of how their perspectives differ over the course of the process and how possible initial disagreements affect outcomes.

Citation

Aaltio, E. and Kannasoja, S. (2023), "Achieving agreement on service needs in child protection. Comparing children’s, mothers’ and practitioners’ views over time and between approaches", Journal of Children's Services, Vol. 18 No. 1, pp. 1-17. https://doi.org/10.1108/JCS-12-2021-0052

Introduction

Service user’s viewpoints should be considered in children’s social care when assessing the need for services and service outcomes. First, according to the United Nations Convention on the Rights of the Child (CRC), children have the right to express their views and participate in decision-making process affecting their lives. Any organisation claiming to operate rights-based practice should make efforts to enable children to freely express their opinions and feel that their voices are heard (Falch-Eriksen et al., 2021). National legislation in many countries, such as in England (Carr, 2004) or Finland (Social Welfare Act 1301/2014), requires service user participation in social care services. Second, the families’ participation can help to improve the quality and possibly also outcomes of children’s services (Gladstone et al., 2012; Heimer et al., 2018). However, in several studies, service users have reported negative experiences when participating in service design and delivery (Bekaert et al., 2021; Falch-Eriksen et al., 2021).

Service user engagement and participation can refer to collective involvement in service development or to users’ individual involvement in cases that concern them personally (Diaz, 2020; Kiili et al., 2021). In this study, we use the terms engagement, participation and involvement interchangeably to refer to the active contribution of children and parents in their own social care cases.

New models or frameworks for practice have been developed to improve the quality and outcomes of child protection, such as Signs of Safety (Turnell and Murphy, 2017) and Reclaiming Social Work (RSW) (Goodman and Trowler, 2012). While the theoretical basis and key skills of these models differ, they both promote the importance of service users’ participation in improving service quality and outcomes. This study focused on the Systemic Practice Model (SPM), a Finnish adaptation of RSW, that has been widely adopted in Finland since 2017. RSW and the SPM apply the ideas and methods of systemic family therapy in statutory child and family social work. In the SPM (Lahtinen et al., 2017; Author’s own), service user participation, relationship-based practice and power sharing are among the key principles.

While the idea of acknowledging the limits of practitioners’ knowledge underpins systemic practice, child protection authorities have a legal responsibility to assess a child’s need for protection and act accordingly. In Finland (Child Welfare Act 417/2007), this responsibility is assigned to a case-holding social worker (henceforth practitioner). While a child’s parents (or custodians) are primarily responsible for the child’s well-being and development, children also have a right, as laid down by the CRC, to be heard on these issues. Hence, three viewpoints should be considered when assessing a child’s need for protection and services: those of the child, the practitioner and the parent(s). Research on service users’ participation and their perceptions on the quality of services (Bekaert et al., 2021; Falch-Eriksen et al., 2021) has rather neglected the issue of agreement between these three parties and the effect of such agreement on service outcomes. Forrester et al. (2013) explored agreement between the family and practitioner by comparing correlations between practitioners’ and parents’ ratings of a group applying RSW and a service-as-usual group. They found a higher level of agreement on key issues in the family in the RSW group. However, the sample did not include children as informants. In addition, the association with outcomes was not analysed.

This study investigated agreement and its effect on outcomes in the context of statutory child protection by comparing the viewpoints of three parties (child, parent and practitioner) in two study groups – a group applying a systemic approach (SPM) and a service-as-usual control group (SAU). Specifically, we explored informant discrepancy, that is, the mean level differences between the participants’ scores on families’ service needs and family dynamics, and the extent of agreement, that is, inter-rater correspondence. By applying a quasi-experimental repeated-measures study design, it is possible to investigate change in agreement longitudinally, here over a six-month period and compare changes between study groups. Finally, we explored the relationship between informant discrepancy at baseline and child protection outcomes at follow-up. The main hypothesis was that agreement between practitioners and two family members would increase over time as the three groups of informants explored the family’s problems and service needs together. Achieving some form of agreement or understanding with service users on their problems and possible solutions is one of the goals of direct social work practice (Trevithick, 2012). A second hypothesis was that agreement would be higher in the SPM group, as practitioners in this group have received additional training and support for involving families and co-reflecting with them. A third hypothesis was that the higher the level of agreement, the better the child protection outcome.

The research questions were:

RQ1.

Are there discrepancies between the informants’ views on the family’s situation and service needs? If so, then what differences occur over time and between the study groups?

RQ2.

To what extent do the informants agree on the family’s situation and the need for child protection? What differences occur over time between the informants and between the two study groups?

RQ3.

Is agreement at T1 related to the outcome at T2?

Service user participation

Various reasons have been given for taking service users’ personal views on their well-being and service needs into account in the context of child protection services. First, engaging families in decision-making may be a prerequisite for a successful intervention. Meta-analyses on a range of therapeutic interventions for children and adolescents (Shirk and Karver, 2003) and adults (Horvath and Symonds, 1991) have found consistent, although modest, associations between engagement and positive treatment outcomes. Similar associations are expected in child protection interventions; for example, Horwitz and Marshall (2015) state that families only benefit from interventions they are committed to. This statement has been empirically supported. Gladstone et al. (2012) found that parents’ perceived engagement was associated with positive outcomes for parents, such as satisfaction with the service, subjective perceptions of improved parenting and feeling that their children were safer. In turn, based on coded observations of practice and outcome data collected from parents, Forrester et al. (2019) found that practitioners’ engagement skills seemed less important in producing positive outcomes than the skills termed by the authors as “good authority,” that is, purposefulness, clarity about concerns and a focus on the child. Nonetheless, the authors state that service quality and respecting service users are important even if child protection could not resolve the family’s problems.

Second, according to the CRC, authorities are obliged to hear children and make decisions that are in the child’s best interest. To this end, practitioners need to canvass children’s opinions and preferences (Falch-Eriksen et al., 2021). Moreover, parents and children may frame the situation differently, especially when children are at risk because of lack of parental care. Based on their analysis of child protection referrals and care plans, Heimer et al. (2018) found that while children’s perceptions were heard at the beginning of the child protection process, parents’ views became increasingly dominant in the process from assessment to investigation and the provision of care. When an intervention was based solely on parents’ perceptions, it was likely to poorly match the problem. The authors concluded that involving children improves the quality of decision-making and intervention design.

Third, collecting information both from children and parents enables practitioners to compare these viewpoints and address any disagreement. Parent–child discrepancy provides valuable information for use when working with the family and designing an intervention. Goolsby et al. (2018), who evaluated a psychotherapy intervention for children with psychosocial deficits, found that parent–child discrepancy was associated with poorer treatment outcomes. To improve the effectiveness of treatment, the authors suggest that practitioners should address such discrepancies and examine the factors causing them, such as family functioning, awareness of problems, motivation to be involved in treatment and conflicting perspectives.

While many argue for service user participation and family engagement, these ideals are not necessarily met in child protection practice. In the UK, for example, Diaz (2020) found that while practitioners regard children’s participation as important, they either lack understanding and knowledge of how to implement meaningful participation or are simply too busy include it. Hence, contrary to the national guidelines on good social work practice, children were not informed about how decisions regarding them were made and felt their views were ignored. A systematic review (Falch-Eriksen et al., 2021) on children’s participation in child protection found that in several countries, their perceptions are consistently overlooked. Similarly, in a metasynthesis on family members’ perceptions, Bekaert et al. (2021) found many studies where service users reported negative experiences of practitioners’ engagement with them. For instance, parents felt practitioners held narrow pre-conceived ideas about their problems and did not take time to hear parents’ interpretations of the problem or service plan.

Systemic approaches and child protection systems in England and Finland

The Finnish SPM for child protection is an adaptation of the RSW model originally developed in England (Goodman and Trowler, 2012). The English child protective system has been typologised as a hybrid system that provides support for families but is also child protection oriented, whereas the Finnish system is primarily family-service oriented (Gilbert et al., 2011). The Finnish system is less regulated than the English system (Berrick et al., 2015), but caseloads are higher in Finland than in England (Berrick et al., 2016).

In England, the majority of local authorities have adapted some framework or model to guide their practice (Baginsky et al., 2020). While Signs of Safety is the most popular practice model in England, the government-funded reform programme for child and family services in Finland 2016–2019 was inspired by RSW. RSW combines the methods and principles of systemic family therapy with statutory child protection practice. RSW incorporates reforms from leadership to case management and frontline casework. The aim of these changes and activities is to enable relationship- and strength-based practice, where the family is seen as a system interacting with other systems, such as the extended family and professionals. Systemic practice seeks to engage the family, reflect multiple perspectives and act differently in the light of these insights (Bostock et al., 2019). In previous research, RSW has been associated with improved quality of child protection practice. In a mixed-methods evaluation study conducted in the original site pioneering RSW, Forrester et al. (2013) observed that practitioners in RSW units were more empathic and received more positive feedback from parents than practitioners in service-as-usual teams. Moreover, agreement between practitioners and parents about the nature and seriousness of the family’s issues was higher in the RSW units. Bostock et al. (2019) explored new sites implementing RSW and found an association between systemic case supervision and practitioner skills, that is, better relationship-building skills and “good authority” (Forrester et al., 2019).

In the SPM (Lahtinen et al., 2017), child protection organisations form systemic teams comprising a consultant social worker, a family therapist, a unit coordinator, one to three social workers and occasionally also family practitioners. These teams hold weekly meetings, where the consultant social worker and family therapist provide systemic case supervision. The purpose of these sessions is to formulate hypotheses from multiple perspectives and plan the next steps to take with the family. Compared to service as usual, practitioners should engage more in mutual reflection, share responsibility and knowledge of cases and receive case supervision more frequently (Aaltio and Isokuortti, 2021). Regarding direct practice with families, practitioners are expected to spend more time with families and apply systemic tools and methods, such as genograms and hypothesis. Using these methods should provide both practitioners and service users with insights and new perspectives on families’ dynamics, histories, patterns and narratives (Aaltio and Isokuortti, 2021). While systemic practice is inspired by ideas central to systemic family therapy (Lorås et al., 2017), the intention is not to turn social work practice into therapy. Instead, systemic practice involves collaborating and sharing expertise with families and other professionals and formulating goals and creating meanings together with families while acknowledging the authority position of the practitioner. The SPM is outlined as a child- and family-based model where service users’ needs are the starting point for casework (Lahtinen et al., 2017).

In previous research, practitioners implementing the SPM during the first stage of national dissemination reported difficulty in applying the model, but considered it had potential and especially valued the contribution of a family therapist in case discussions (Isokuortti and Aaltio, 2020). A quasi-experimental study comparing child- and family-level outcomes found no differences between SPM and SAU groups in children’s safety or well-being or in family dynamics (Aaltio, 2022). However, the SPM practitioners met families more often than their service-as-usual counterparts. While these findings suggest that SPM did not outperform conventional practice in service outcomes, it has yet been analysed whether it may improve agreement between the parties involved in child protection case work.

Method

Study design and conditions

This study used a quasi-experimental, repeated measures design. Outcome data were collected from three groups of informants: children, parents and practitioners.

The SPM was developed and disseminated by the Finnish Institute for Health and Welfare, which also conducted a national evaluation of SPM implementation including the data collection for this study.

During 2017–2018, The Finnish Institute for Health and Welfare trained 58 local trainers in use of the SPM. By summer 2018, these had trained 52 local teams at 31 sites. Local training comprised six days of team training and subsequent group supervision sessions led by the trainers. The training included an introduction to systemic thinking and family therapy, the functioning of a systemic team, team roles and systemic weekly meetings, documentation, principles of systemic practice and an introduction to systemic methods and techniques. After training, teams began to hold systemic weekly meetings and apply systemic ideas and methods in their casework. All teams participating in this study included a consultant social worker, a unit coordinator and a systemic family therapist participating in their weekly team meetings. However, all the teams but one contained more than three social workers, that is, the teams were larger than intended.

The Institute recommended that the municipalities participating in the project reduce caseloads in the SPM teams to 20 children per practitioner. As most of the municipalities did not have the resources for this, caseloads in the SPM teams remained high. In the SPM as well as SAU teams participating in this study, caseloads averaged 37 children per practitioner.

Sample

The sample comprised a group of families supported by practitioners in SPM teams (n = 9) and a group of families supported by practitioners in SAU teams (n = 9). Data were collected from three municipal children’s service sites. One was a large city in the metropolitan area of Finland and two were medium-sized regional centres, one in Central and one in Eastern Finland. All teams worked with child protection cases, that is, with children who had been assessed as needing protection but could nevertheless remain in their parents’ custody.

The SAU teams did not receive any additional training and their practice was not guided by any shared model or framework. The baseline characteristics of the study groups are shown in Table 1. The only significant difference between the study groups was that the SAU group children were on average older (U = 2,023 and p = 0.004).

Procedure

Purposive sampling was used. The first data collection (T1) was conducted at each site over a three-month period in Spring 2018. Practitioners were asked to complete assessment questionnaires pertaining to the families they were currently actively working with. The idea was to deliver and complete these questionnaires as part of everyday practice so that the information collected from and with the families would simultaneously support casework. Adolescents and parents could choose to complete the questionnaire alone or during a meeting. Children’s questionnaires were to be completed during a meeting with a practitioner. The follow-up data (T2) were collected approximately six months later. The practitioner data were used with the permission of the municipalities participating in the study. Participation in the research was voluntary for the children and their parents. Informed consent was requested from guardians and children over age 12 years to use the children’s forms and from parents to use the parents’ forms. All the parents and children assessed at T1 were asked to complete the self-assessment questionnaires at T2 irrespective of their participation at T1. Ethical approval was granted by the Finnish Institute for Health and Welfare (2017–09). All personal information collected in this study has been treated as highly sensitive.

Instruments

Questionnaire data on well-being and the need for service intervention were collected from three groups of informants (practitioners, children and parents). The practitioner’s questionnaire included items on child and parent demographics and outcome measures. The self-assessment questionnaires included subjective outcome measures. Two versions of the children’s questionnaire were used (7- to 12- and 13- to 17-year-olds). Both versions contained the same item content but with simpler wording for the younger participants.

Measures

Key outcome measures

Need for service intervention was measured with a single-item question developed for this study. For practitioners, the item was “Based on your overall judgement, how much in need of child protection is the child at present?” Answers were given on a 11-point response scale (from 0 = no need to 10 = the child’s safety is seriously endangered). For children and parents, the question was “Think about your family’s overall situation at present. How much outsider help do you think you need?” Answers were given on a scale from 0 = none to 10 = very much.

Child’s subjective well-being was measured by KINDL, a generic instrument for assessing health-related quality of life in children and adolescents (Ravens-Sieberer and Bullinger, 1998). Both versions of the questionnaire contain 24 items with a five-point response scale: never, seldom, sometimes, often and all the time. The respondent is asked how they have felt during the past week with respect to five dimensions: physical well-being (e.g. “I felt ill”), emotional well-being (e.g. “I felt alone”), self-esteem (e.g. “I was proud of myself”), family (e.g. “I got on with my parents”), friends (e.g. “I played with friends”) and everyday functioning in school (e.g. “I enjoyed my lessons”). For items asking about poor quality of life, the values are reversed. The sub-scales were combined to produce a total score ranging from 0 to 100. High scores indicate higher well-being. [T1: α = 0.84 and T2: α = 0.84].

Signs of abuse or neglect were measured in the practitioner questionnaire with a set of 21 items on different types of abuse or neglect. The wording was positive (e.g. “The child’s daily activities are done in safe surroundings” and “The parent treats the child age-appropriately”). Responses were given on four-point scale (0 = no signs of abuse or neglect, 1 = some signs, 2 = serious signs and 3 = cannot say). Excluding value 3, responses were combined to form a sum variable ranging from 0 to 42. The measure was originally developed in a previous research project (Aaltio, 2015). For the present study, the wording and number of items was modified based on consultation with experts in child protection and child psychiatry. [T1: α = 0.88 and T2: α = 0.89].

Other measures.

Family dynamics according to the child and to the parent were measured using SCORE-15, an instrument developed to monitor progress and outcome in systemic family therapy (Stratton et al, 2010). SCORE-15 contains 15 items describing aspects of family functioning (e.g. “It feels risky to disagree in our family” and “We trust each other”). Responses were given on five-point Likert scales ranging from 1 = describes my family very well to 5 = does not at all describe my family. For items asking about poor family dynamics, the values are reversed. The total score is the mean score for all items and ranges from one to five. Low scores indicate better adjustment. [Mothers T1: α = 0.90 and T2: α = 0.89; children T1: α = 0.84 and T2: α = 0.86]

Child-related case characteristics were measured in the practitioner’s questionnaire with a set of ten items on the child’s physical, mental, cognitive and social well-being and behaviour (e.g. “Does the child have mental health-related problems, such as fatigue, exhaustion or tiredness?” and “Does the child have any friends?”). Responses were given on a four-point scale where 1 indicated no problems, 2 some problems, 3 serious problems and 4, cannot say. The exact wording depended on the question. Responses 2 and 3 were counted as indicating problems.

Sample size and attrition

A total of 65 cases were assessed by practitioners at both T1 and T2 (attrition 42%). Missing data analyses showed that missing data were not related to study group or the child’s age, sex, language or living arrangements. The children’s questionnaire was completed at both measurement points by 50 children, that is, 50% of all the children aged 7–17 (n = 100) participating in the study. Missing data were not related to study group or child demographics. 83 mothers participated in the study at T1 and 43 completed the parent self-assessment questionnaire at both measurement points (attrition 48%). Non-Finnish speaking mothers were less likely to participate at both T1 and T2 (Χ² (1) = 4,09 and p = 0.043). Missing data were not related to study group, child demographics or mother’s age, education, employment status, relationship with the father (i.e. marriage, cohabitation or divorced) or relationship with the child (i.e. birth parent or stepmother). In the study, 25 fathers participated at T1, and 9 completed the parent self-assessment questionnaire at both measurement points. In all these cases, the mother had also replied. To keep the sample more homogenous, we decided to use mothers’ responses only.

Statistical analyses

The analyses were carried out in the following steps. First, baseline differences between the two groups were tested using t-test or the Mann–Whitney U-test for continuous variables and Chi-square test for nominal variables. Second, to assess informant discrepancy (mean level differences) on the need for service intervention and the family dynamics, paired-sample t-tests were used to compare each pair of informants at T1 and T2. Third, to assess the change in this discrepancy between T1 and T2 in the whole sample and between the study groups, the difference between the group means on the need for service intervention and family dynamics were calculated. These new variables were then used to test the effects of group and time and group-by-time interaction by using repeated measures analysis of variance. Fourth, to answer the research question on the extent of agreement between the child, the child’s mother and the practitioner at T1 and T2, separately, Pearson’s Product-Moment correlation was used. Fifth, to examine if the extent of agreement had increased more in the SPM than SAU group, Fisher’s Z-test was used. Sixth, to examine the effect of agreement on outcomes, three change scores (T2–T1) were calculated for the selected key outcome variables: the need for child protection according to the practitioner, signs of abuse or neglect as assessed by the practitioner and the child’s subjective well-being. Thereafter, multiple regression analysis was conducted. Child and mother reports separately and the interaction terms were used as predictors for change scores of all three outcome variables. Practitioner–child and practitioner–mother reports separately, and the interaction terms were used as predictors for change in signs of abuse or neglect and the child’s well-being. SPSS Statistics 27 was used for all analyses.

Results

Preliminary analysis

The baseline characteristics of the study groups with respect to the child-related measures are shown in Table 2. No significant between-group differences were found.

Discrepancies

The need for service intervention according to the child, the child’s mother and the practitioner

The results across the sample indicate a significant mother-child discrepancy on the need for service intervention at T1, t(58) = –4.48, p < 0.001 and at T2, t(39) = –2.69, p = 0.01. Mothers reported significantly higher scores than children at both measurement points (Figure 1) A significant child-practitioner discrepancy was found at T1, t(65) = –5.44 and p < 0.001 but not at T2, t(42) = –1.72 and p = 0.093. No significant differences between mothers’ and practitioners’ assessments were found at T1, t(80) = –0.70 and p = 0.487 or at T2, t(41) = 0.34 and p = 0.73).

The difference in means between the practitioners’ and children’s groups decreased across the sample from 3.1 to 1.3 over the follow-up period. This change was significant (Wilks’ Λ = 0.80, F(1, 34) = 8.57, p = 0.006 and η_p² = 0.201). In the SPM group, the mean difference decreased from 4.3 to 1.8 and in the service-as-usual group from 2.3 to 1.0. No statistically significant group-by-time interaction effect was observed (Wilks’ Λ = 0.98, F(1, 34) = 0.74, p = 0.395 and η_p² = 0.021), indicating that the systemic approach and service-as-usual were equally effective in increasing the level of agreement between a practitioners and a children.

The difference in means between the practitioners’ and mothers’ groups decreased from 0.8 to –0.1, indicating that at T1, the family’s need for child protection was rated higher by practitioners than mothers, whereas at T2, mothers were slightly more concerned about the situation than practitioners. However, no significant effect of time was observed (Wilks’ Λ = 0.91, F(1,34) = 3.20, p = 0.082 and η_p² = 0.086). In the SPM group, the mean difference decreased from 1.3 to –0.1. In the SAU, the absolute difference remained the same (0.1), although practitioners were more concerned than mothers at T1 and mothers more concerned than practitioners at T2. No significant group-by-time interaction effect was observed (Wilks’ Λ = 0.96, F(1,34) = 1.38, p = 0.248 and η_p² = 0.039).

The difference in means between the mothers and children decreased from 2.2 to 1.6, but no significant effect of time was found (Wilks’ Λ = 0.94, F(1, 30) = 1.87, p = 0.181 and η_p² = 0.059). In the SPM group, the difference between the means decreased from 2.5 to 1.2 and in the SAU from 2.1 to 1.8. No statistically significant group-by-time interaction effect was observed (Wilks’ Λ = 0.97, F(1,30) = 0.79, p = 0.382 and η_p² = 0.026).

The family dynamics according to children and mothers.

Paired-sample t-tests found no significant differences between children’s and mothers’ assessment of their family’s dynamics at either T1 (children: M = 2.0, mothers: M = 2.1, t(61) = –0.64 and p = 0.524), or T2 (children: M = 2.0 and mothers: M = 2.0, t(39) = 1.47 and p = 0.149). The finding indicates that on average, children and mothers gave similar assessments of the quality of their family’s dynamics. The difference in means between children and mothers was 0.0 at T1 and 0.1 at T2 both in the whole sample and in each study group. These differences showed no significant effect of time (Wilks’ Λ = 0.95, F(1,31) = 1.76, p = 0.195 and η_p² = 0.054) or effect of group-by-time interaction (Wilks’ Λ = 1.00, F(1,31) = 0.14, p = 0.712 and η_p² = 0.004).

The extent of agreement

The need for service intervention according to children, mothers and practitioners.

Practitioners’ assessments of the need for service intervention did not correlate with children’s assessments across the sample at either T1 or T2 (Table 3). Similarly, no association was found between practitioners’ and mothers’ assessments at T1 or T2. Children’s scores showed a medium-level (r = 0.33) correlation (Cohen, 1992) with mothers’ scores at T1 but not at T2. These findings indicate that the extent of agreement between practitioners and family members was poor. However, child–mother agreement was moderate at baseline but not at follow-up.

To explore the difference in correlations between the SPM and SAU groups at T2, correlations between the same variables were calculated by study group. None of the correlations were statistically significant.

Family dynamics according to children and mothers.

A significant association across the sample was found between children’s and mothers’ assessments of family dynamics at both T1 and T2 (Table 3). The strength of the correlations (r > 0.50) (Cohen, 1992) indicates a high extent of child–mother agreement on their family’s dynamics.

At T2, the correlation was r(21) = 0.58 and p = 0.003 in the SAU group and r(15) = 0.52 and p = 0.034 in the SPM group. The difference in the correlations was not significant, z = −0.28 and p = 0.39.

Other study variables.

We also explored the associations between the need for service intervention, family dynamics and two variables used in the outcome evaluation, that is, the child’s subjective well-being and practitioner-assessed signs of abuse or neglect. At both measurement points, children’s subjective well-being correlated with children’s assessment of family dynamics (Table 3). This correlation was negative owing to the scoring protocol. Hence, these findings indicate that poor family dynamics was associated with children’s poor subjective well-being. At T1, mothers' assessments of their family's dynamics correlated with their children’s subjective well-being, but this association was no longer found at T2. A significant association was found between practitioner-assessed signs of abuse or neglect and the need for service intervention at both T1 and T2. Interestingly, practitioners’ assessments of abuse or neglect did not correlate with children’s subjective well-being. A medium correlation (r = 0.33) (Cohen, 1992) was found between practitioner-assessed abuse or neglect and child-assessed family dynamics at T2.

When the study groups were examined separately, significant correlations were found between mothers’ assessments of the need for help and their family’s dynamics: r(21) = 0.61 and p = 0.002 in the SPM group and r(22) = 0.51 and p = 0.011 in the SAU group. Significant correlations were also found between mothers’ and children’s assessments of their family’s dynamics: r(15) = 0.52 and p = 0.034 in the SPM group and r(21) = 0.58 and p = 0.003 in the SAU group. However, these between-group differences in the correlations were not significant.

Effect of agreement on outcomes

Multiple regression was conducted to explore the relationship between the informants’ views on the need for service intervention at T1 and change from T1 to T2 in three outcomes: need for service intervention according to practitioners, signs of abuse or neglect according to practitioners and children’s subjective well-being.

Children’s and mothers’ assessments of service need across the sample did not, separately, predict change in outcomes. However, their interaction did predict change in abuse or neglect (Table 4). Based on the analysis, greater agreement between child and mother on the need for help at T1 predicted a decrease in the amount of abuse or neglect. Child–mother agreement on no need for help predicted a greater decrease in abuse or neglect than child–mother agreement on the need for help (Figure 2). Disagreement between children and their mothers predicted an increase in the amount of abuse or neglect reported by the practitioner.

A similar regression was conducted for each study group. In the SPM group, no significant relationships were found. In the SAU group, a significant relationship between child–mother discrepancy and signs of abuse or neglect was found (B = –2.283, SE B = 0.84, β = –0.52, t = –2.73 and p = 0.013).

Children’s and practitioners’ assessments on the need for service intervention at T1, both separately and by their interaction, did not predict change in the outcome variables across the sample. No significant differences within the study groups were detected. Similarly, mother’s and practitioner’s views and their interaction did not predict changes in outcome variables in the whole sample or in the study groups.

Discussion

This study explored whether the viewpoints of children, mothers and practitioners converge over time, and if so, then whether this change differs between a group applying the systemic approach to child protection and a SAU. The effect of agreement on child protection outcomes was also explored.

The first hypothesis was that agreement would increase across the sample over time. This was expected, given that one task of a practitioner is to achieve at least some form of agreement with the service user before proceeding to shared decision-making and planning how to overcome problems (Trevithick, 2012). In this sample, the level of agreement between practitioners and children on the need for service intervention increased over time. At baseline, practitioners gave graver assessments than children of the child’s situation. However, six months later, the difference between these groups was no longer significant. No differences were found between mothers’ and practitioners’ assessments.

Interestingly, while children and mothers tended to agree on the quality of their family’s dynamics, mothers saw more need for outsider help at both measurement points. This finding can be interpreted in several ways. First, mothers differ from children in their awareness of the service system and support that might be available. Second, as they are responsible for taking care of the child, mothers need to assess the balance between their personal resources and the difficulties within the family that they are trying to tackle. The differences between the role of a child and that of a carer in the family mean that children’s and carers’ perspectives and experiences will also differ (Miller and Bentovim, 2003).

The correlations found between the study variables revealed that practitioners’ analyses of a family’s need for service intervention was not associated with either children’s or mothers’ expressed need for help. At T2, an association was found between children’s assessments of their family’s dynamics and practitioner-reported abuse or neglect. Otherwise, practitioners’ and family members’ views did not converge during the process. These findings could indicate less successful progress than expected in this sample. The casework with families had not yet resulted in shared understandings of problems. Several explanations can be offered for the disparity between practitioners and service users’ views. First, families might seek to protect their private sphere. Children’s loyalty to parents may prevent them from discussing their family life with social workers (Pölkki et al., 2012; Wilson et al., 2020). As Trevithick (2011) points out, “some service users do not want us to know what they know – and this may be a rational or aware stance that they are adopting.” Second, children’s poor relationship with their social worker or lack of the support or information needed to meaningfully participate in child protection processes may impair their ability or willingness to be involved (Diaz, 2020). Furthermore, psychological and emotional stress caused by the process of investigating abuse, and mistrust of social workers may hinder children from disclosing abuse and neglect (Wilson et al., 2020). Third, in some cases, violence experienced at home or by peers in the school environment may become “normalised” (Allnock and Atkinson, 2019; Paterson-Young, 2021). Similarly, some parents with a personal history of childhood trauma may have difficulties in distinguishing between acceptable boundary setting and abusive behaviour in parenting (Siverns and Morgan, 2019). Thus, a situation that may seem normal to a child or a parent may be of concern to a social worker.

The second hypothesis was that the agreement would be higher in the SPM group, as the practitioners had received special training and support, including mutual reflection, to increase families’ involvement in case work. However, no between-group differences in discrepancies were found. Unfortunately, exploring between-group differences in correlations was not possible. In this sample, the systemic approach did not promote child–mother or family member–practitioner agreement any better than service-as-usual. In their meta-analyses, Bekaert et al. (2021) and Falch-Eriksen et al. (2021) conclude that to engage and support families, practitioners need more time, reasonable caseloads and adequate budgets. Factors facilitating participation were active listening, access to the practitioner, private discussions with the practitioner and practitioner effort to build a trusting relationship with the child (Falch-Eriksen et al., 2021). In both groups, high caseloads may have affected practitioners’ possibilities to facilitate families’ engagement.

Finally, we explored the effect of agreement on outcomes. The results showed that child-mother disagreement on the need for service intervention predicted negative and child–mother agreement positive outcomes in terms of abuse or neglect. Interestingly, this decrease was larger when both child and mother reported that the family was not in need of outsider help. Despite this, the child protection practitioners seem to have carried out a successful intervention. The present findings suggest that child protection practice is less successful when one family member sees the need for service intervention, while another does not.

Strengths and limitations

This study is among the few in the context of children’s social services that has collected quantitative data from three groups of informants and compared their viewpoints. Outcomes were assessed by two informants, the child and the practitioner. This decision was based on the legal responsibility of social workers to assess children’s need for protection and on children’s legal right to express their views and participate in the decision-making process.

The study has its limitations. First, the small sample size limits the representativeness of the study, and the attrition between the two measurement points hindered between-group comparisons. However, at T1, apart from the mean higher age of the SAU group children, no between-group differences in background variables were found. Attrition was not associated with study group. Thus, the study groups were similar enough to allow reasonable comparisons between them. Second, the low participation of fathers resulted in their exclusion. In future research, more attention should be paid to recruiting fathers.

Third, in the absence of any single indicator of what constitutes success in a child protection intervention (Forrester, 2017), we used a combination of measures designed and validated in other settings (i.e. SCORE-15, KINDL) and non-validated measures designed for this study (need for service intervention, abuse or neglect). Validating these measures for this population was beyond the scope of this study. Thus, to develop and validate suitable outcome measures for use in the child protection context requires more research.

Finally, according to a previous report (Isokuortti and Aaltio, 2020), the systemic approach had not yet been fully implemented during the study period. This means that had differences between the study groups been found; this would have limited the conclusions. More collaboration between implementers and researchers is clearly needed to improve implementation quality and success in data collection.

Conclusions and implications

The findings from this sample do not suggest that the SPM, as implemented during the first stage of national dissemination, is likely to outperform service as usual in improving agreement between service users and practitioners. Further research is needed to explore the effectiveness of systemic practice on agreement.

The present findings for the whole sample have more general implications for future practice. First, practitioner assessments of the family’s situation might not reflect child or parent assessments. Hence, to identify alternative hypotheses and next-step strategies, the situation should be explored from multiple perspectives using multiple sources of information.

Second, possible disagreements between family members should be addressed in practice, as child–parent disagreement might have a negative effect on child protection outcomes. To this end, practitioners should first form a good relationship with both child and parent. Based on previous research on shared decision-making, practitioners need better understanding, additional training and more time to engage service users (Diaz, 2020). As further described by Stabler et al. (2019), practitioners need to work collaboratively with the family to create open dialogue, motivate the family to engage with the process, build trust and reduce the shame associated with discussing difficult issues. This has implications for organisations, who need to provide supervision and develop an organisational culture that supports good practice. Moreover, service users suffer from continual changes in, and time constraints on, their social workers. Hence, organisations need to find strategies to decrease practitioners’ high caseloads and reduce staff turnover rates (Diaz, 2020).

Third, findings indicate that child–parent agreement predicts a successful intervention whether they agree or disagree on the need for professional help. A possible hypothesis is that in general child–parent agreement provides a basis for practitioner-family collaboration. However, if related to the child’s loyalty towards the parent or the “normalisation” of problems, then agreement may also conceal possible risk factors within the family. Thus, in both research and practice, the mechanisms underlying agreement should be further explored and critically appraised.

To conclude, information should be collected more often from multiple sources, allowing comparison between practitioners’ and family members’ viewpoints. Research, both quantitative and qualitative, such as individual interviews, is needed to identify the mechanisms underlying agreement and disagreement and monitor changes in practitioners’, children’s and parents’ viewpoints over time.

Figures

Figure 1

Need for service intervention according to a child, a mother, and a practitioner at T1 and T2

Figure 2

Interaction between child-reported and mother-reported need at T1 predicted change in practitioner-reported abuse or neglect

Table 1

Baseline (T1) socio-demographic characteristics of the children in the systemic practice model and service-as-usual control group groups

Socio-demographic charasteristic	Total N = 112	SPM N = 56	SAU N = 56	Significant difference SPM vs SAU p
Age
Median (Range)	13.0 (1–17)	11.0 (1–17)	13.5 (5–17)	0.004
Sex
Female	56 (50%)	33 (59%)	23 (41%)	ns.
Male	56 (50%)	23 (41%)	33 (59%)
First language Finnish	103 (93%)	54 (96%)	49 (89%)	ns.

Table 2

Child-related characteristics and outcome measures in the systemic practice model and service-as-usual control group groups at T1

Child-related characteristic	Total n (%)	SPM n (%)	SAU n (%)	Significant difference SPM vs. SAU p
Some or many problems (according to the practitioner) with
Physical health	23 (35.4)	9 (32.1)	14 (37.8)	0.634
Mental health	55 (84.6)	21 (75.0)	34 (91.9)	0.062
Cognitive functions	36 (55.4)	13 (46.4)	23 (62.2)	0.206
Contacting other people	33 (50.8)	13 (46.4)	20 (54.1)	0.543
Being bullied	34 (52.3)	14 (50.0)	20 (54.1)	0.746
Controlling emotions	46 (70.8)	17 (60.7)	29 (78.4)	0.121
Neglecting one’s own basic physical needs	38 (58.5)	17 (60.7)	21 (56.8)	0.749
Outcome measure	Total M (SD)	SPM M (SD)	SAU M (SD)	p
Signs of abuse or neglect according to the practitioner	8.2 (5.9)	8.0 (5.4)	8.4 (6.5)	0.752
Need for service intervention according to the practitioner	6.2 (2.4)	6.6 (2.1)	5.9 (2.6)	0.122
Subjective well-being according to the child	69.8 (12.9)	72.3 (12.6)	68.2 (13.0)	0.200

Table 3

Correlations for outcome variables at T1 and at T2

Variable		1	2	3	4	5	6
1. Need for service intervention, practitioner	T1 T2	– –
2. Need for service intervention, child	T1 T2	0.12 −0.14	– –
3. Need for service intervention, mother	T1 T2	0.02 0.12	0.33* 0.22	– –
4. Family dynamics, child	T1 T2	0.10 0.20	0.30* 0.07	0.11 0.13	– –
5. Family dynamics, mother	T1 T2	0.07 0.20	0.28* 0.12	0.50** 0.54**	0.50** 0.57**	– –
6. Subjective well-being, child	T1 T2	−0.07 −0.05	−0.23 −0.09	−0.09 −0.27	−0.58** −0.43**	−0.43** −0.27	– –
7. Abuse or neglect, practitioner	T1 T2	0.53** 0.40**	0.02 −0.14	−0.08 −0.13	0.20 0.33*	0.07 0.09	−0.08 0.12

Notes:

*p < 0.05; **p < 0.01

Table 4

Multiple regression analysis on the relationship between child–mother discrepancy (mean-centred variables) on need for service intervention at T1 and change in scores (T2–T1) of three outcome variables

Variable	B	SE B	β	p
Need for service intervention according to practitioner
Model 1
Child-reported need	0.22	0.57	0.07	0.706
Mother-reported need	−0.10	0.49	−0.04	0.846
R2 = 0.004, Δ F(2, 34) = 0.074 and p = 0.928
Model 2
Child-reported need	0.25	0.61	0.08	0.681
Mother-reported need	−0.11	0.50	−0.04	0.830
Child × mother interaction	−0.11	0.58	−0.03	0.852
R2 = 0.005, Δ F(1, 33) = 0.035 and p = 0.852
Signs of abuse or neglect according to practitioner
Model 1
Child-reported need	−0.99	1.16	−0.15	0.400
Mother-reported need	1.59	1.00	0.28	0.121
R2 = 0.072, Δ F(2, 34) = 1.31 and p = 0.283
Model 2
Child-reported need	−0.07	1.13	−0.01	0.954
Mother-reported need	1.26	0.94	0.22	0.188
Child × mother interaction	−2.77	1.08	−0.41	0.015
R2 = 0.226, Δ F(1, 33) = 6.57 and p = 0.015
Child’s subjective well-being
Model 1
Child-reported need	2.94	2.10	0.25	0.172
Mother-reported need	1.40	1.82	0.14	0.45
R2 = 0.095, Δ F(2, 30) = 1.58 and p = 0.223
Model 2
Child-reported need	2.62	2.32	0.22	0.268
Mother-reported need	1.42	1.85	0.14	0.448
Child × mother interaction	0.81	2.31	0.07	0.727
R2 = 0.10, Δ F(1, 29) = 0.12 and p = 0.727