Outcomes of family group conferencing in Sweden

Abstract

Objective: Between 1995 and 1997, the Swedish Association of Local Authorities implemented Family Group Conferences (FGC) in 10 local authorities throughout Sweden. This study reports on client outcomes of this implementation.

Method: 97 children involved in 66 FGCs between November 1996 and October 1997 were compared with 142 children from a random sample of 104 traditional child protection investigations by the Child Protective Services (CPS). All children were followed for exactly 3 years for future child maltreatment events reported to CPS. Effects were modeled using multiple regressions, controlling for the child’s age, gender, family background, and type and severity of problems.

Results: After controlling for initial differences, FGC-children experienced higher rates of re-referral to CPS compared to the group that had been processed in traditional investigations. They were more often re-referred due to abuse, were more often re-referred by the extended family, were longer in out-of-home placements, but tended over time to get less intrusive support from the CPS. FGCs were not related to re-referrals of neglect, of case-closure after 3 years or number of days of received services. The results suggest that the impact of the FGC was scant, accounting for 0–7% of the statistical variance of outcome variables.

Conclusions: The findings did not support the alleged effectiveness of the FGC model compared to traditional investigations in preventing future maltreatment cases. If these results are confirmed in future research, they serve as a reminder of the necessity to evaluate models based on untested theories or on extrapolations from other countries/cultures, before these models are widely spread in a national practice context.

Résumé

Objectif: Entre 1995 et 1997, l’Association Suédoise des Autorités Locales a organisé des rencontres pour le groupe familial (FCG) dans 10 autorités locales à travers la Suède. Cette étude rapporte les résultats de cette réalisation.

Méthode: 97 enfants impliqués dans 66 de ces rencontres entre novembre 1996 et octobre 1997 ont été comparés avec 142 enfants d’un échantillon choisi au hasard tiré de 104 enquêtes traditionnelles de protection de l’enfance faites par les services de Protection de l’enfance (CPS). Tous les enfants ont été suivis exactement pendant 3 ans au sujet de mauvais traitements pouvant dans le futur être rapportés au CPS. Les effets ont été modélisés en utilisant les régressions multiples, et en contrôlant chez les enfants l’âge, le sexe, le milieu familial, le type et la gravité des problèmes.

Résultats: Après contrôle pour les différences initiales, les enfants soumis aux FCG ont connu l’expérience d’être plus fréquemment référés au CPS que le groupe qui avait été soumis à des enquêtes traditionnelles. Ils ont été plus souvent référés à nouveau pour des mauvais traitements, plus souvent référés à nouveau par la famille élargie, plus souvent placés, mais à la longue ont reçu une aide moins exgérée de la part du CPS. Les enfants du groupe FGC n’étaient pas référés à nouveau pour négligence, fermeture du dossier après 3 ans ou pour le nombre de jours où ils avaient bénéficié de services. Les résultats suggèrent que l’impact des FGC était insuffisant comptant pour 0 à 7% de la variance des variables du résultat.

Conclusions: Les résultats ne prouvent pas l’efficacité attendue du modèle FGC comparé aux investigations traditionnelles dans la prévention de cas de mauvais traitements futurs. Si ces résultats étaient confirmés à l’avenir ils serviraient à rappeler qu’il est nécessaire d’évaluer les modèles basés sur des théories non prouvées ou sur des extrapolations issues d’autres cultures ou pays, avant que ces modèles ne soient répandus sous forme de pratiques à l’échelon national.

Resumen

Objetivo: Entre los años 1995 y 1997, la Asociación Sueca de Autoridades Locales implementó Conferencias de Grupos Familiares (FGC) en 10 autoridades locales en toda Suecia. Este estudio presenta los resultados de ésta implementación en los clientes.

Método: Se compararon noventa y siete niños, incluidos en 66 FGC entre noviembre 1996 y octubre 1997, con 142 niños de una muestra al azar de 104 investigaciones tradicionales sobre protección infantil realizadas por el Servicio de Protección Infantil (CPS). Se realizó el seguimiento a todos los niños sobre futuros eventos de maltrato infantil reportados al CPS durante exactamente 3 años. Los efectos fueron analizados utilizando regresiones múltiples, controlando la edad del niño, género, contexto familiar, y tipo y severidad de los problemas.

Resultados: Después de controlar las diferencias iniciales, los niños de los FGC experimentaron mayores tasas de re-referimientos al CPS que el grupo que habı́a sido procesado en las investigaciones tradicionales; con mayor frecuencia fueron re-referidos por abuso, fueron re-referidos con mayor frecuencia por la familia extendida, estuvieron por mayor tiempo en traslados fuera del hogar, pero con el tiempo, tendı́an a recibir menos apoyo del CPS. Las FGC no estuvieron relacionadas con re-referimientos por negligencia, por cierre de caso después de 3 años o número de dı́as de servicios recibidos. Los resultados sugieren que el impacto de las FGC fue escaso, siendo responsables del 0–7% de la varianza estadı́stica en los resultados de las variables.

Conclusiones: Los resultados no apoyaron la alegada efectividad del modelo de las FGC comparándolas con las investigaciones tradicionales para prevenir futuros casos de maltrato. Si estos resultados se confirman en investigaciones futuras, servirán como un recordatorio de la necesidad de evaluar los modelos basados en teorı́as no probadas o en extrapolaciones de otros paı́ses/culturas, antes de que estos modelos sean diseminados en la práctica en un contexto nacional.

Introduction

The model of Family Group Conferences (FGCs) for decision making in child welfare has, in a relatively short time, spread from its birth place New Zealand to several countries in Europe, the USA, Canada, Australia, Israel and to South Africa (e.g., Ban & Swain, 1994a, Ban & Swain, 1994b, Burford & Hudson, 2000, Immarigeon, 1996, Lupton & Nixon, 1999, Marsh & Crow, 1998). Several authors (e.g., Connolly, 1994, Lupton & Nixon, 1999; Maluccio, Ainsworth, & Thoburn, 2000) have pointed out that the wide popularity of FGCs rests more on procedural and implementation data than on outcome evidence, and have strongly stressed the need for research on long-term outcomes. This article reports on the results from a 3-year follow-up of a FGC-trial in 10 local authorities in Sweden. Outcomes were compared for 97 children involved in 66 FGCs during November 1996 and October 1997, and 142 children from a random sample of 104 traditional child protection investigations conducted by the Child Protective Services (CPS) during the same period.

FGCs have their origins in a New Zealand Ministerial inquiry (Department of Social Welfare, 1988) which addressed concern about the overrepresentation of Maori children in the care system (see also Connolly, 1994, Marsh & Allen, 1993, Ryburn, 1993). The basic premise of the FGC model is that families not only have the right to be deeply involved in decisions about their child, but that solutions found within the family are likely to be better than those imposed by professionals are. Other central assumptions are (Burford & Hudson, 2000, Hassall, 1996, Lupton & Stevens, 1998):

• families—through assuming responsibility for current problems—are better motivated to seek lasting solutions than professionals,

• the model is culturally sensitive,

• when families agree to discuss current problems in private among themselves, the likelihood of sensitive information being included into the decision making is increased,

• FGCs initiate better family functioning by way of communication, cooperation, and supervision as well as bringing together family members who have lost touch and restoring legitimate parental authority that has been undermined.

In summary, FGCs are aimed at promoting effective functioning in families by focusing on their unique strengths and by enlisting them in a problem solving process (Hudson, Galaway, Morris, & Maxwell, 1996).

Three basic practice related principles seem to prevail in all countries that have adopted the FGC model (Lupton & Stevens, 1998): (1) the term “family” should be interpreted widely to include extended family members, friends, neighbors, and significant others; (2) the family must have an opportunity to develop a protection plan in private without the professionals present; and (3) professionals should accept the plan unless it is seen to place the child at risk of significant harm. In addition, there seems to be wide agreement that FGCs should be reconvened when initial problems have been sufficiently resolved, or if there is a need for a new meeting for other reasons (Burford & Hudson, 2000; Lupton, Barnard, & Swall-Yarrington, 1995; Marsh & Crow, 1998, Pennell & Burford, 2000).

Inspiration in Sweden to start a FGC-trial was drawn from British experiences (cf, Lupton & Nixon, 1999, Marsh & Crow, 1998, Ryburn & Atherton, 1996). In 1995, the Swedish Association of Local Authorities selected 10 out of 23 interested local authorities for the trial, wanting geographic as well as socio-economic diversity, and demanding local professional and political commitment to the model. Financial support was provided for training of personnel and coordinators. The first FGCs were held in May 1996. In October 1996, all 10 local authorities had held at least one FGC. The trial operated within existing legal frameworks and policy/practice requirements. It was at the discretion of individual social workers whether families under CPS investigation should be offered a FGC or not, a reminder of the strong influence that “street level bureaucrats” have in the child welfare system (Lipsky, 1980). In practice, FGCs were but one method among several alternatives, resulting in low levels of FGC referrals, as in the UK study (Marsh & Crow, 1998; Sundell, Vinnerljung, & Ryburn, 2001). Of all families that were referred to the CPS during this period, approximately one-third (35%) were offered a FGC, and only one in four of these families accepted the offer (Sundell & Haeggman, 1999). The families offered a FGC did not differ from those not offered to do so in types of problems or severity of these problems. The main difference between the two groups is that those offered a FGC had a social worker that generally was more positive toward the FGC model. Also, social workers claimed that families who were offered FGCs were less willing to collaborate with the CPS authority during the investigation.

Research from New Zealand on outcomes of FGCs in child protection is so far sketchy. The bulk of international research on FGCs in child protection up to date has been concerned mainly with process and implementation related outcome (Lupton & Nixon, 1999, Maluccio et al., 2000). Studies of varying methodological quality have been done in several countries, as the UK (Lupton et al., 1995, Lupton & Stevens, 1998, Marsh & Crow, 1998), Sweden (Andersson & Bjerkman, 1999, Sundell, 2000, Sundell & Haeggman, 1999), Canada (Burford & Pennell, 1998, Pennell & Burford, 2000), USA (Shore, Wirth, Cahn, Yancey, & Gunderson, 2002) and Australia (Ban, 2000, Cashmore & Kiely, 2000, Crampton & Jackson, 2000; Trotter, Sheehan, Liddell, Strong, & Laragy, 1999; Vesneski & Kemp, 2000). Results on process and implementation concur so far in most respects. When used in New Zealand and in other countries, FGCs:

• Involve relatives and others from the family’s social networks in sharing responsibility for family’s problems.

• Give families who face the likelihood of statutory intervention a real chance to make their own decisions on how to solve family problems.

• Permit 9 out of 10 families to actually produce a plan for change that gains acceptance from the Child Welfare authority.

• Get high ratings for consumer satisfaction.

Follow-up data on child and family related outcome after FGCs have so far been presented in only a few studies, using relatively small samples. The results are not easily compared, due to different methodologies and legal/social context, different construction of samples and comparison groups, and varying follow-up time, even within the same studies. Taken together, the results tend to be somewhat of a mix, though more positive outcomes for FGC-clients than for comparison groups seem to prevail. Marsh and Crow (1998) presented follow-up data on 80 of the 99 children who originally participated in FGCs. Data for 16 of these children came from a 6-month follow-up, and for the remaining 64 from a 2-month follow-up. The study found considerably lower re-abuse rates compared to cited studies of CPS-populations, albeit with different follow-up times. In addition, the proportion of children still on the UK local authorities’ lists of at-risk children was significantly smaller than in comparable studies. Furthermore, the results suggested that reunification rates for children, who were placed in out-of-home care after FGCs, were the same as for children not participating in FGCs. However, reunification for FGC-children meant that more moved to members of the extended family (e.g., grandparents). In addition, a larger proportion of children reunited after FGCs tended to remain at home, not re-entering care, compared to other CPS studies. Pennell and Burford (2000) studied 28 families from Newfoundland and Labrador (including a group of indigenous families), who had participated in FGCs during 1 year. They used data from interviews with 115 family members 4–27 months after the conference, together with case file reviews. Data for a comparison group of 31 roughly matched CPS-families were accessed through similar case file reviews. The results were positive, showing declining child maltreatment rates and reduced levels of domestic violence after FGCs, compared to the development over the same time for the comparison group. Shore et al.’ s (2002) report from the Seattle area in the US is based on data from 70 FGCs, involving 137 children. Follow-up data were from at least 6 months after a convened FGC, focusing on 114 children who were in care at the time of the conference. The authors found that the number of children reunited with their parents had increased dramatically at follow-up, even though the proportion of children living with either parents or relatives was roughly the same as at the time of the FGC. Re-referral rates were lower than regional CPS-data, even for 55 children who had their FGC 2 years before follow-up. In addition, the stability of the living arrangements was high, with few re-entries to care.

The study presented here was designed to increase knowledge of long-term outcomes of FGCs. From relatively consistent claims in the literature by proponents of the model regarding FGCs assumed influence on client-level outcomes (e.g., Burford & Hudson, 2000, Hassall, 1996, Lupton & Stevens, 1998, Ryburn, 1993), six hypotheses guide the evaluation. First, it is reasonable to (1) expect positive process related results. The implementation of FGCs should furthermore (2) decrease the risk for referrals, (3) reduce the likelihood of repeated neglect and abuse, (4) when needed, FGCs should increase reports by the extended family, (5) out-of-home placement within the extended family ought to be more frequent, and (6) FGCs ought to increase the possibility of closing CPS-cases.