Common needs but divergent interventions for U.S. homeless and foster care children: results from a systematic review


Cheryl Zlotnick, Children’s Hospital & Research Center Oakland, CHORI/Center for the Vulnerable Child, 747 52nd Street, Oakland, CA 94609-1809, USA


Many children living in homeless situations in the U.S. have temporary stays in foster care, and both populations suffer disproportionately higher rates of physical, psychological and social difficulties compared with other children. However, very little is known about which specific interventions achieve the best outcomes for children in these overlapping transitional living situations. To address this gap, we review existing literature to identify the most promising practices for children living in transition. A standardised vocabulary specific to each of three electronic databases (i.e. Medline, PsychINFO and CINAHL) was employed to identify studies that described an intervention specifically targeting foster care or homeless children and families. Separate systematic searches were conducted for homeless and foster children, and only studies published in English between January 1993 and February 2009 were selected. The final sample (n = 43) of articles described interventions that fell into two categories: mental health (n = 17) and case management (n = 26). No article included a sample containing both homeless and foster care children, and most studies on homeless children used case management interventions while most studies on foster care children focused on mental health interventions. Few articles employed rigorous study designs. Although repeatedly studies have demonstrated the overlap between populations of homeless and foster care children, studies focused on one population or the other. Virtually all studies on both homeless and foster children devised interventions to reduce trauma and family instability; yet, no evidence-based practice addresses the overlapping needs and potentially relevant evidence-based practice for these two populations. An important and vital next step is to establish an effective evidence-based intervention that reduces the impact of trauma on both U.S. populations of children living in transition.

What is known about the topic

  •  U.S. Homeless and foster children both suffer from trauma and health manifestations of that trauma.
  •  Mental health and case management services have been identified as vital services for both U.S. homeless and foster care children.
  •  A subset of children may be in both homeless and foster populations during their childhood.

What this paper adds

  •  A systematic literature view of both (separate) bodies of literature (i.e. homeless children and foster care children in the U.S.) involving promising or evidence-based practices applicable to both populations.
  •  The acknowledgement and background demonstrating the two overlapping U.S. populations of homeless and foster children.
  •  Identifying and describing through the reviews of the overlapping needs and potentially relevant EBPs for these populations.

The populations – connection of homeless to foster care children in the U.S.

Trauma, substance abuse and mental health problems provide the bridge that connects the two ostensibly distinct U.S. populations of foster care and homeless populations together from childhood to young adulthood. The connection begins with young children whose birth parents have problems with substance abuse and mental health problems that can lead to child neglect and eventual placement into foster care (U.S. General Accounting Office, 1994, 1997, Hoffman & Rosenheck 2001, Zlotnick et al. 2003) (Figure 1). In fact, a study with a nationally representative sample of U.S. homeless adults found that more than 70% of homeless mothers were not living with their children (Burt et al. 1999), and approximately 25% of them were in foster care (Zlotnick et al. 2007). Another study demonstrated that nearly half of the young children entering foster care had been removed from homeless birth parents (Zlotnick et al. 1998). This connection continues in youth. Many runaways or homeless youth, particularly those with histories of abuse, have lived in foster care (Greene et al. 1999, Lindsay et al. 2000, Thompson & Pollio 2006). Later, in young adulthood, approximately 20% of youth ageing out of the foster care system became homeless within 2 years (Pecora et al. 2006, Kushel et al. 2007, Dworsky & Courtney 2009). The studies describing the connection between homelessness and foster care in the different stages of life recognise the important role that substance abuse, mental health problems and trauma play in this connection (Gerstel et al. 1996, Medrano et al. 1999, Hoffman & Rosenheck 2001, Grella et al. 2006). Despite the mounting evidence demonstrating this overlap, most studies investigate interventions designed exclusively for children/youth who are homeless or those in foster care.

Figure 1.

 Cycle of family homelessness and link to foster care.

Service needs of U.S. homeless and foster children

The two populations of homeless and foster care children have been studied separately, yet similar conclusions have been made on the types of services needed for each child population. Homeless children have disproportionately high rates of unmet health needs including acute health problems, trauma-related injuries and chronic diseases as a result of living in unstable and often unsafe environments (Committee on Community Health Services 1996, Weinreb et al. 1998). They also have higher rates of developmental delays and behavioural concerns that may have been aggravated by school absenteeism (Committee on Community Health Services 1996, Vostanis et al. 1997, Zima et al. 1997, Johnston et al. 2006). Risky behaviours, including early sexual activity that increases the risk for HIV infection, sexually transmitted disease and incidence of pregnancy, also are problematic among homeless youth (Zima et al. 1997, Kidd & Scrimenti 2004). Similarly, children entering foster care also have disproportionately high rates of physical and mental health concerns (Chernoff et al. 1994, Simms et al. 2000). Despite the strong evidence demonstrating the high need for services in this population, only half of child welfare agencies routinely assess foster care children for mental health problems (Leslie et al. 2003).

Service needs vary widely among families experiencing homelessness or involved in the foster care system. Many are concrete, such as help finding and applying for food, housing or services, or transportation to attend appointments. Parental mental health and/or substance abuse treatment services are also needed by many of these families. Case management, a service designed to obtain and coordinate a variety of resources, is commonly used in health and social service agencies working with homeless families. Case managers help parents navigate a society that assumes families live in stable homes and have the financial and organisational abilities to advocate for their children. They assist families to identify needs and create and implement plans to address them (Zlotnick & Marks 2002). While homeless families frequently rely on case managers for assistance, parents in the foster care system must co-operate with their court-assigned child protective services (CPS) worker. CPS workers work with both birth parents and foster parents to ensure the children’s needs are met. One role of CPS caseworkers is to review reunifying parents’ case plans and provide information on services and items such as bus tickets so parents can obtain court-identified services that are prerequisites for reunification (Ellis et al. 2003, pp. 2–4).

Although studies are finding a consistent connection between the homelessness and foster care populations, and studies have demonstrated that this overlapping population suffers similar and disproportionately high rates of physical, psychological and social needs, no study has examined the existing literature to identify the most promising practices among the physical, mental health and case management services used for children living in transition – whether in homeless and foster care living situations.


Study design/approach

The authors examined articles on interventions for homeless and foster care children and families to identify promising practices specifically designed for these populations. Inclusionary criteria were (i) a sample that contained either homeless or foster care children and their families; and (ii) focus on a specific intervention. Exclusions were case studies based on a diagnosis, meta-analyses, observational studies solely designed to describe a population, policy discussions and systematic reviews. To ensure an objective, consistent and systematic approach, no attempts were made to identify ongoing interventions that had not been submitted and published in professional journals.

The sample of articles was obtained using a systematic search strategy that incorporated the standardised vocabulary, subject headings or thesaurus specific to each database (e.g. Medical Subject Headings or MESH for Medline) on the three major electronic databases cataloguing articles on clinical interventions: medical (PubMed), psychological and social services (PsychINFO), and nursing and allied health services (e.g. physical therapy, occupational therapy and speech therapy) literature (CINAHL). Only articles located in searches of these databases were used. Because this review was designed to locate interventions or promising practices specifically designed for homeless and foster care children and families, public policy, social studies, sociological and databases that target non-clinical disciplines were not a part of the search.

Two sets of searches were conducted on each of the three databases. The first set targeted homeless children and families, and the second targeted foster children and their families. Search words included the following: foster, homeless, services, family and child. Limiting criteria included publication date between January 1993 and February 2009, and English language.

Many articles met the search criteria for the two populations of children (Figure 2). Homeless persons yielded many articles in three large bibliographic databases: PubMed (n = 2290), PsychINFO (n = 3636) and CINAHL (n = 1214). However, when the search was narrowed to include services, and either youth, children or family, only a sparse 1.8% (n = 126) of articles were found with most identified in PsychINFO (n = 52) followed by PubMed (n = 43) and CINAHL (n = 31). Literature on foster care populations fared better: PubMed (n = 1594), PsychINFO (n = 648) and CINAHL (n = 754). When the search was narrowed by services, 8.7% (n = 262) remained, including articles in PsychINFO (n = 164), CINAHL (n = 80) and PubMed (n = 18).

Figure 2.

 Search findings on children living in transitional families.

The articles’ abstracts were reviewed by two investigators to ensure each article presented a study that focused on services for either foster care children or homeless children and their families (Figure 1). After the exclusionary criteria were imposed, the final sample of articles (n = 43) consisted of 70% (n = 30) on foster children and 30% (n = 13) on homeless children and their families.

Instrumentation and inter-rater reliability

A standardised, semi-structured instrument was devised to review each article. Development of this instrument was informed by other study abstraction instruments. Abstraction instruments used to review studies were biased towards randomised controlled clinical trials (RCCTs) as historically double-blinded RCCTs have been considered the ‘gold standard’ yielding the most reliable and reproducible findings. Recent comparisons of RCCT and rigorous observational studies have found that investigations using observational epidemiological designs including cohort, cross-sectional and retrospective studies have produced similar results (Benson & Hartz 2000, Evans 2003). These findings indicate that in addition to the study design, the elements contained within the abstraction tools need to include the study population and sample, interventions, outcomes measures, statistical analysis and results (Agency for Healthcare Research and Quality, 2002, Lohr 2004).

Incorporating these perspectives, the instrument designed to abstract the articles elicited basic information that was recommended for both RCCT and observational studies, including the following: the theory or model that was being examined, target population, demographic characteristics and numbers included in the sample, the study design, the instruments, fidelity or validity measures used to test the instruments, major findings and conclusions. Three reviewers each tested the instrument on six articles. The reviewers compared their independent results to identify discrepancies and to finalise the instrument. To ensure the most systematic and uniform article assessment, a primary and a secondary reviewer examined every article. Discrepancies were discussed in terms of the content of the article as well as the interpretation and use of the instrument to inform modification to the instrument. To ensure the most systematic and uniform article assessment, a primary and a secondary reviewer examined every article. Dissimilar findings were resolved using a consensus process. In this process, reviewers who identified a discrepancy would proceed by locating the text within the article that supported the reviewer’s position. Both primary and secondary reviewers would explain the reasoning for the classification and findings. If consensus was not achieved, the third reviewer was asked to conduct an independent review of the article to assist the primary and secondary reviewers achieve consensus.

The articles were organised based on the primary target populations, classification of interventions (i.e. case management, mental health, substance abuse or a combination of services), and rigour of study design. Similarities and dissimilarities among the types of interventions were assessed between populations of children. The rigour of the study design and evidence presented by each article was ranked with a commonly used five-level scale that considers the study’s power, as assessed by sample size, significance level and Type II errors (Sackett 1989). Several methods have been used to rank evidence; however, the Sackett model has been used in a variety of journals employing many different study designs (Sackett 1989, Guyatt et al. 1998, Wright et al. 2003, Wupperman et al. 2007).


The 43 studies described two major types of services (Table 1). Approximately 40% (n = 17) described mental and psychosocial health services, and 60% (n = 26) case management services. No article included both homeless and foster care children. Approximately 70% (n = 30) of the articles targeted children or their parents who had contact with the foster care system (including group homes). The remainder targeted homeless children, youth or families.

Table 1.   Quality of evidence by type of service
Level of quality of evidenceService type
Total (n = 43)Mental health (n = 17)Case management (n = 26)
IRandomised controlled trials with low false positive (α– 95% confidence intervals) and low false negatives (β)   
IIRandomised controlled trials with high false positive (α– trends but not significant) and high false negatives (β)7% (3)12% (2)4% (1)
IIINon-randomised, prospective, cohort or case–control with concurrent comparison group26% (11)29% (5)23% (6)
IVNon-randomised, cohort studies or retrospective studies with historical comparison groups   
VStudies without control groups including cross-sectional, descriptive, case series or reports67% (29)59% (10)73% (19)

Mental health

Of the 17 mental health and/or psychosocial services studies, there were few similarities in types of intervention, study design and rigour, use of instrumentation, child age range or population, or outcomes (Table 2). Most interventions were provided in shelters, homes, or places outside of health or public institutions. Others needed specialised equipment (e.g. observation rooms) or training, or were tailored to children of specific age ranges. Most mental health interventions solely targeted children in foster care.

Table 2.   Mental health
AuthorsAge of children(years)Specific modelPopulation and sampleInterventionMethodologyOutcomeConclusionCategory of evidence
Christenson & McMurtry (2007) Not givenFoster pride/adoptive pride programmeFoster care/group homes
N = 288
This article describes 9, 3-hour sessions to promote knowledge on topics such as nurturing children, addressing developmental delays and making connections with children with the goal of reunification with birth parents. A comparison was made between responses from kinship versus non-kinship foster parents. The instrument used to assess knowledge was pilot tested for internal consistencyCross-sectional
Two groups of foster parents: kinship (n = 69)
Non-kinship (n = 159)
Comparison of pretest and posttest responses indicated foster parents understood the goal of reunification and were willing to work with birth parents. Kinship caregivers demonstrated less improvement than non-kinship caregiversFamilies demonstrated increased knowledge after participating in the curriculumF
Collado & Levine (2007) 4–17MH consultation/therapyFoster care/group homes
N = 548
The article describes the implementation of a mental health consultation approach used in two sites (Bronx and Staten Island) among two large NY city agencies working with foster children who were at high risk for placement changes because of their behaviours. The intervention involved individual and family therapy, and psychoeducational groups. Case examples were presented to indicate the ability of the intervention to decrease placement changesQuasi-experimental
Two sites, consultations conducted at different agencies: GSS-Bronx (n = 325)
Seaman-Staten Island (n = 233)
Multiple placement changes were found prior to the mental health consultation. Only two children experienced placement changes among 108 who received mental health consultationTrauma-based therapy and MH consultation contribute to decreased placement changes. Mental health credentials as well as the clinicians’ need for supervision also were discussedE
Farmer et al. (2003) 3–17, avg 13Treatment foster care (TFC)Foster care/Residential N = 184This study interviewed foster care parents of children in residential treatment for youth, ages 3–17, who have psychiatric disorders and exhibit aggressive behaviour, at 6, 12 and 18 monthsProspective/Longitudinal
TFCs are used as a step-down placement. However, children who move out of the TFC were more likely to have behavioural concerns. Increased age was significantly correlated with an increase risk of placement. Any improvements from remaining in the TFC were not sustained a year after TFC dischargeTFCs were not always used as step-down placementsD
Glisson & Green (2006) 4–18, avg 14.43 SD 2.39 Foster care/group homes
N = 1241
Using a sample obtained from the juvenile courts of 22 counties in Tennessee, this study examined who obtained specialty mental health services among those who were assessed with mental health problems, and among those who were more likely to be in out of home placements. Baseline interview and follow-up at 6 months and for 1.5 years. Three validated tools, Short form Assessment for Children (SAC), Columbia Impairment Scale (CIS) and the Brief Symptom Inventory (BSI), were used to assess the child’s mental healthProspective-longitudinal
Children, who needed mental health services, were less likely to receive the services after child welfare or juvenile justice contact compared to other times. This finding indicates that when most needed, mental health services were not provided. Also, child’s mental health problems and substance abuse were associated with higher rates of out-of-home placement. Specialty mental health care (from a mental health professional) reduced odds of an out of home placementThis indicates the importance of screening and treatment of mental health concerns by a mental health professionalD
Holody & Maher (1996) 9–11Life book processFoster care/group homes
N = 4
The authors describe the approach and utility of the Life Book and use the experiences of four foster children aged 9–11 years to exemplify the approachDescriptiveNo outcomes or data were providedThe authors suggest that the Life Book Process may be a good intervention for latency age children that can be implemented by all educational levels of case workerG
Kliman (1996) 3–12Personal life bookFoster care/group homes
N = 52
The author described how creating a personal life history book was used as a treatment to express and deal with emotions and trauma of foster care. A case–control study using foster care children was conducted, matching comparison group by age. Intervention was use of Lifebook, but not specifically describedCase–control
24 cases (treatment) group
28 matched comparison group
Study found that children treated with the Lifebook were 11 times more likely not to have a placement movePersonal Lifebook may sublimate behaviours and thereby decrease difficult behavioursE
Lewis et al. (1995) 1–17Family-based intensive treatment (FIT) used for family reunification projectReunifying children who were in foster care
N = 110
This study compared the usual child welfare service to the family reunification projects application of the family-based intensive treatment. Children were aged 1–17 years. The model is based on Rogerian Theory. The treatment group received the FIT caseworker services, which averaged 37.6 hours of services for each family over the 90 day period. The control group received the routine child welfare services. Assessment of services and outcomes reported by caseworker. Some validity measures conducted on outcome measuresRandomised control trial
57 experimental group
53 control group
The authors found improvement among the following was linked to reunification within 12 months: parenting skills, anger management, school performance services. Concrete services such as transportation, tangible help and coaching clients was also linked to successful reunificationWorking in partnership with clients promoted client cooperation and this relationship was at the foundation of the success in the FIT programme.B
McWey (2008) Not givenIn-home therapyBirth parents at risk of having children removed
N = 20
The study included 20 families who received a 6-week series of in-home family therapy sessions to decrease the likelihood of having children placed into foster care. This study obtained client quotations on their perceptions of in-home therapyQualitativeFamilies reported utility of services such as support, skill building (e.g., modalities of providing child discipline) and availability of services. Therapist’s personality and approach were important factors in family’s use of servicesObtaining client perspective on therapy and its utility is important when devising in-home therapyG
Ornelas et al. (2007) <6 (n = 387); 6–12 (n = 212)Mental health for permanencyFoster care, guardian and adoptive homes
N = 859
The kinship centre employed the mental health permanency model with wraparound services for children who were in temporary foster care and moving into permanency, kinship, adoption or guardianship. All children were Medicaid-insured. Validated measures were used for different ages; they included: Ages and Stages Questionnaire, Child Behaviour Checklist, Kinship Center Attachment Questionnaire, Child and Adolescent Functional Assessment Scale were used. Baseline and follow-up administration of measures to assess change in behaviour and mental health functionProspective-longitudinal
Two cohorts of children
No significant results reported although some changes in attachment and externalizing behaviours noted; and
among children with attachment disorder, scores suggest some improvement
The article highlights the need of mental health services for children exiting the child welfare systemD
Orfirer & Kronstadt (2002) Birth to 3 yearsServices to enhance early development (SEED)Foster care/group homesThis study describes a collaboration between a county Child Welfare department and a hospital-based department to improve placements of young children under three by assessing child’s cues and relationships, and providing mental health servicesQualitativeCase studies demonstrate the benefits of assessment, relationship-based therapy, and collaboration with child-welfare workers and public health nursesThe authors stress the need for multidisciplinary collaborations to improve the placements of children in foster careG
Swenson et al. (2000) 1 day–16 years, avg 8Charleston collaborative project (CCP)Foster care families
N = 45 families with 72 children
This study described a collaboration between a mental health agency and state agencies serving foster children. A randomised controlled clinical trial was used to examine family therapy services and their ability to promote family reunification, and increased child–caregiver function. Validated measures included the Addiction Severity Index, Parenting Stress Index and Child Behaviour Checklist were used. Fidelity of assessment and treatment were measured. Costs of programme also were measured.Experimental/Randomised control trial
Random assignment to CCP versus current services
Two follow-ups at 90 days and 3 months
Some decreases in cost were demonstrated in medical care utilization. No significant differences in family reunification were found; however, the authors suggest power, number and intensity of interventions may be the reason for lack of significanceImplementation of a multi-faceted programme in ‘real world’ setting was problematic and resulted in a non-optimal implementation of the modelB
Timmer et al. (2006a) 3–5Parent child interaction therapy (PCIT)Foster care/group homes
N = 1
A case study of a foster-adoptive child who had experienced five placements is used to describe and demonstrate the utility of Parent Child interaction therapy (PCIT). PCIT involves relationship-based therapy with two phases: child-directed and parent-directed. The latter phase includes coaching sessions in which a one-way mirror with adjoining observation room is used so clinicians can observe and give prompts to parents as they interact with their child. The following standardised measures were used: Eyberg Child Behaviour Inventory, Child Behaviour Checklist, Child Abuse Potential Inventory and Parenting Stress IndexQualitative
Cohort given treatment and child’s behaviour was assessed at pre-, mid- and posttreatment
The study acknowledged learning through PCIT if the observers noted a certain number of reflective statements, and praises. Changes in child behavioural measures from pretreatment to mid. Graph shows improved behaviour was correlated with the number of praises at treatment sessionsPCIT may improve child’s behaviour, but it also demands the dedication and involvement of parent/caregiverG
Timmer et al. (2006b) Bio sample avg = 4.61 (SD 1.62); Foster sample avg = 4.31 (SD 1.66Parent child interaction therapy (PCIT)Foster care/group homes
N = 385
This study conducted pre- and postmeasurements to assess impact of Parent Child Interaction Therapy (PCIT) on children under 5 years. Although 691 foster parent–child dyads enrolled, only 385 completed treatment (two groups – birth parents and foster care parents). The following standardized measures were used: Child Abuse Potential Inventory, Child Behaviour Checklist, Parenting Stress Index., Eyberg Child Behaviour Inventory, and Symptom Checklist 90RQuasi-experimental
Two groups were compared at pretest–posttest
birth parent–child (n = 222); and foster parent–child (n = 163)
Both birth and foster care parents were equally likely to complete treatment. Increased drop-out rate for African Americans and children with high symptom severity. Parent–child interaction therapy (PCIT) was found effective for reducing child behaviour problems– more for birth parents than for foster care parentsPCIT may instill child behaviour improvements because of parents’ ability to coach and make a change or because parents’ expectations changedE
Tischler et al. (2000) Not givenBrief supportive therapyHomeless families
N = 40 families (122 children)
This UK study evaluated the utility of a brief mental health intervention provided to families entering the hostels/shelters. The model provides time-limited behavioural therapy to help parents and staff deal with problematic child behaviours. Brief supportive psychotherapy is provided to children who have experienced traumaQualitative description with cross-sectional dataInterventions reported but no resultsA relationship between parent and child mental health concerns was reported to tease apart mental health issues of parents from those of the children. A discussion of the importance of mental health treatment is providedG
Westermark et al. (2007) Not givenMultidimensional treatment foster careFoster care/group homes
N = 28
This study conducted on foster parents in Sweden rated the Multidimensional Treatment Foster Care (MTFC) designed to help teens with behavioural problems. MFTC is a labour-intensive structured treatment including family therapy, home visiting, didactic instruction and behavioural interventions over an 8- to 12-month period. Using mixed methods, the ‘hands-on’ approach was found to be positive by foster parentsQualitative description with cross-sectional dataFoster family satisfaction with the model, although more positive than negative, indicated that there were two types of foster parents – those who want to be professional and distant, versus those who wanted to integrate child into family. Satisfaction differed based on type of foster parentFoster families appreciated use of treatment tools delineating interventions with youthG
Williams et al. (2001) 9–11Pynoos therapy modelFoster care/group Homes
N = 5
This study examined the experiences of five children to assess the school-based group therapy model. There were seven goals: promoting peer member understanding; increasing capacity to regulate emotional responses with peers; fostering empathy; enhancing social skills; encouraging help-seeking behaviour; linking current feelings and behaviours with prior trauma; and facilitating an environment that allows expression of the consequences and meanings of traumatic exposureQualitativeYouth made progress towards all of the seven goals. Qualitative examples of this progress are providedThe Pynoos Group model has been examined with other children; it was not specifically designed for foster children. Still, it proved effective. One difficulty was recruitmentG
Wotherspoon et al. (2008) <3 yearsCollaborative mental health careFoster care vignettesThis article describes the Collaborative Mental Health Care in Canada, a combination of early childhood development and infant mental health that works with the child welfare workers. Developmental services were provided by a speech therapist. Clinical vignettes illustrate the utility of this modelQualitativeImprovements in the infant’s development and behaviour were noted in the vignettes. Among the issues addressed were parent–child interaction, mother’s health and child’s healthThis article discussed the importance of integrating needed services, considering child development, maternal child health and the ethics of confidentialityG

Interventions targeting age ranges

Few of these mental health interventions were assessed in multiple articles. There were two exceptions. The first included four studies that examined relationship-based, child– or infant–parent psychotherapy (Orfirer & Kronstadt 2002, Timmer et al. 2006a,b, Wotherspoon et al. 2008) using different models. Certain structures associated with relationship-based psychotherapy were similar across the four studies including clinicians with highly specialised education and skills, and equipment such as one-sided mirrors and observation rooms. These studies targeted young children, between the ages of birth and five, in foster care; none targeted children in homeless families.

Two studies described ‘Lifebooks’ intervention, geared for older children (Holody & Maher 1996, Kliman 1996). While considered ideal for latency age children, it was used in children as young as 3 years old. In both of these studies, the goal was to sublimate difficult behaviours by engaging children to tell their stories in art and words and thereby reduce the number of foster care placements. Again, Lifebooks was used only for children in foster care, not for those living in homeless families.


Four (24%) of the studies on mental health and/or psychosocial services described multisession parent-focused treatment designed to promote caregiver function (Lewis et al. 1995, Swenson et al. 2000, Christenson & McMurtry 2007, McWey 2008). These studies focused on children and caregivers who had contact with the Child Welfare System. The goal was to promote understanding and communication to stabilise the relationship between foster care parents and children in foster care (Christenson & McMurtry 2007), or between birth parents and their children who had been or were at risk of entering foster care (Lewis et al. 1995, Swenson et al. 2000, McWey 2008).

About half (53% or n = 9) of the mental health and/or psychosocial services studies were designed to decrease child’s problematic behaviours (Holody & Maher 1996, Kliman 1996, Tischler et al. 2000, Williams et al. 2001, Farmer et al. 2003, Glisson & Green 2006, Collado & Levine 2007, Ornelas et al. 2007, Westermark et al. 2007). The interventions included a variety of combinations of groups and therapies such as Brief Mental Health intervention, Multidimensional Treatment Foster Care and Pynoos Groups.

Most studies targeted foster children (Holody & Maher 1996, Kliman 1996, Williams et al. 2001, Farmer et al. 2003, Glisson & Green 2006, Collado & Levine 2007, Ornelas et al. 2007, Westermark et al. 2007). The goal of the interventions targeting foster care children was to interrupt the cycle of changing foster care placements attributed to problematic child behaviours. Several articles indicated that the problematic behaviours were the result of trauma. The foster care child’s trauma may have begun prior to foster care placement (e.g. abuse or neglect); however, entry into foster care and cycling through foster homes often compounded the trauma. Only one UK study described interventions designed to reduce problematic behaviours that were associated with the trauma of homelessness as a means of facilitating family stability (Tischler et al. 2000).

Study designs

The rigour of the study designs testing these mental health modalities varied markedly (Table 1). Among the 17 mental health and/or psychosocial services studies reviewed, only 12% (n = 2) used unblinded, randomised controlled groups with experimental study designs (Level II) (Lewis et al. 1995, Swenson et al. 2000). Approximately 29% (n = 5) compared their intervention with another regular service by employing non-randomised comparison groups (Kliman 1996, Collado & Levine 2007) with cohort or pretest–posttest study designs and sample sizes of >10 (Level III) (Timmer et al. 2006b, Christenson & McMurtry 2007, Ornelas et al. 2007). More than half (59% or n = 10) used cohort designs without comparison groups (Farmer et al. 2003, Glisson & Green 2006) or qualitative designs with descriptions of interventions using varying size samples (Level V) (Holody & Maher 1996, Tischler et al. 2000, Williams et al. 2001, Orfirer & Kronstadt 2002, Timmer et al. 2006a, Westermark et al. 2007, McWey 2008, Wotherspoon et al. 2008).

Case management

These articles focused on case management, which is frequently a composite of several services (Table 3). Virtually all 26 articles described case management services as having the ultimate goal of helping clients achieve stability; however, the focus of the intervention varied among studies. More than a third (n = 10) of these studies targeted homeless populations and with one exception described the importance of shelter- or school-based services (Grant 1991, Schram & Giovengo 1991, Rog et al. 1995, Davey 2004, Nabors et al. 2004, Anderson et al. 2006, Ferguson 2007, Slesnick et al. 2007, 2008, Taylor et al. 2007, Weinreb et al. 2007).

Table 3.   Case management
AuthorsAge of children(years)Specific modelPopulation and sampleInterventionMethodologyOutcomeConclusionCategory of evidence
Anderson et al. (2006) Not givenFamily support team/workerHomeless families
N = 21
This study describes the utility of the UK-based Family Support Team (FST) programme and parents satisfaction with the programme. The FST and the Family Support Worker (FSW) provide a range of wrap-around and mental health services. Using in-depth interviews and participant observation, the study assessed the utility of the FSTQualitativeClients voiced many positive comments on the utility of having a FSW and FST responding to the families’ needs for referrals and other types of supportFST was seen as positive but results were mixed with views on UK’s hostelsG
Blatt et al. (1997) <5 years (57%) 5–12 years (33%) 12+ years (10%ENHANCE programmeFoster care children N = 548The ENHANCE Programme incorporates a multidisciplinary approach to providing care to children in foster care. The team includes clinic staff, case worker, clinical nurse specialist, paediatrician, nurse practitioner, psychologist, and child welfare worker liaison, and is designed to provide health and mental health needs assessment on admission to out-of-home placementDescriptiveThe team described the high prevalence rates of physical and psychological concerns that they identified in newly placed foster care childrenChildren in out-of-home care benefit from a multidisciplinary team that employs frequent communication among the members of the teamG
Carten (1996) 1–15 yearsFamily rehabilitation programmeBirth mothers who had been reported to the Child Welfare System (CPS)
N = 20
The Family Rehabilitation Programme, designed for mothers with substance abuse problems that have an impact on their abilities to parent, was created to be an alternative to placing their child in foster care. Priority is given to newborns with positive toxicologies for drugs. Caseloads of the child welfare workers (CWW) are much lower than usualQualitativeFollow-up survey six months after discharge of mothers who had no new CPS reports and had successfully completed the Programme indicated that most mothers had initially considered the programme intrusive, but appreciated the joint planning process and developed good relationships with their cww, who they felt were respectful and supportive. Staff also reported high job satisfactionA new approach to mothers with substance abuse may have more benefits than the current programme of placing a child into foster care. This family preservation approach may be particularly important to address the current inequities that exist in CPS for low-income people of colourG
Davey (2004) Avg 7.95 yearsFamily retreatHomeless families
N = 39
An 8-week series was condensed into a weekend to prevent attrition common in shelters. The weekend retreat was held for multiple families who were living in the homeless shelter and was held over a weekend. The sessions were designed to provide education on family strengths, communication, stress management and decision-makingQualitativeFamilies evaluated the retreat and reported learning how to deal with stress and appreciated the family activitiesFamily retreats are a helpful way of encouraging family participation and reducing feelings of isolationG
Ferguson (2007) 18–24 yearsSocial enterprise interventionHomeless youth
N = 28
This study describes a pilot study of the Social Enterprise Intervention comprised of conducting outreach to hard-to-reach homeless youth, engaging them to enter a programme that teaches vocational skills and simultaneously serves as an entry point for mental health services. Entry point was a youth drop-in centre. Didactic programme lasted 7 months, and another 6 months was needed to engage in the vocational cooperativeDescriptiveLogic model indicates several desired outcomes including attaining job skills and gaining employment, increased housing stability and decreased high-risk behaviours. No specific outcomes were providedProgrammes designed to work with street youth must recognise that their economic survival arises from high-risk activities and therefore must incorporate into their programmes other means of employmentG
Gillespie et al. (1995) <5 (64%), 6–11 (19%), 12–17 (17%)Intensive Family preservation services (IFPS)-specialised foster parent trainingFoster Care/Group Homes
N = 42
This study enrolled 42 children to examine the utility of Intensive Family Preservation Services (IFPS) to promote reunification. The IFPS programme included mental health intervention, groups, parenting and home visits. The goal was reunificationQuasi-Experimental
One group pretest–posttest
Of the 42 children, 79% were successfully reunified with parents or relatives. One year after follow-up, 91% of reunified children were still living with their family. High intensity of services with parent visitation increased likelihood of reunificationA positive maternal attitude was key to programme participationE
Grant (1991) 2 years 9 months–5 years Homeless families
N = 78
The study describes developmental needs of children residing with their families in a NY City-based vouchered motel. A day care programme was initiated for 15 months including teachers with anecdotal records and periodic progress reports. The study was a retrospective description of the findingsDescriptiveMany health and behavioural problems were noted on admission. Many were addressed and reduced in severity after 2–3 months of stability. No specific indicators on how behaviours were measuredInitial trauma of homelessness or movement into the shelter elicited temporary behaviour problems in children that decreased in intensity after stabilityG
Hobbie et al. (2000) <1 (9%) 1–4 (31%) 5–9 (41%) 10–14 (18%) >14 (1%)Medical assessment at entry into foster careNew foster care children
n = 4053
This evaluation assesses the health status of new foster children who obtain medical examinations at a children’s hospital-based programme. The programme is designed for all children entering foster care in the county. A substantial number of children had one or more medical diagnoses upon admissionRetrospectiveThe programme was valued for its ability to identify health risks and to connect children with a primary care providerChildren entering foster care are at high risk of medical problems and need an admission assessmentD
Kessler & Greene (1999) Not givenHome visitCase worker
N = 4 and Supervisor
N = 1
Case workers were taught to plan, conduct and end home visits using several educational modalities: individualised training, role-play and group training. A pre–post comparison was made on implementation of visitsQualitativeIndividualised and group training yielded similar benefits. Group training was more efficient; however, even after training, not all elements of the training were incorporated into visitsTraining is important; however, quality control to determine incorporation of training is essentialG
Kirk & Griffith (2004) % in two samples:
0–2 (30/33%)
3–5 (17/20%)
6–10 (28/26%)
11–12 (10/7%)
13+ (15/14%)
Intensive family preservation services (IFPS)Foster care children
N = 26 364
This study enrolled children from 51 of 100 North Carolina counties to examine the effectiveness of Intensive Family Preservation Services (IFPS) at reducing number of placements for the most high risk children. Study had two groups. They enrolled IFPS treated children from counties that used this modality and enrolled non-IFPS from the counties that did not. IFPS involved approximately 73.1 hours (SD = 34.5) of services. Report of fidelity to the model based on time with client. The goal was preventive – not to have an out-of-home placementProspective/Longitudinal
Two Groups – IFPS (n = 542) versus regular services (n = 25 822)
Significant differences between IFPS and non-IFPS in proportion of out-of-home placement. IFPS outperformed traditional child welfare services in identifying the most high risk children and preventing placements. The outcome is influenced by additional high-risk factors identified in the child, and the impact of the treatment appears to waneIFPS is effective when used targeted to high-risk children and families who are willing to accept the interventionC
Landy & Munro (1998) Not givenShared parentingFoster care/group homes
N = 13
This study assessed a programme in which foster parents help an entire family, the parents and children, with reunification. Authors indicated difficulties with recruitment of families. Several standardised tests were used: Center for Epidemiologic Studies Depression Scale, Child Behaviour Checklist, Procidano Perceived Social Support Questionnaire for Family and Friends. The goal was to integrate child back into birth familyQuasi-ExperimentalOnly 30% of families were reintegrated into birth family. Findings suggest that the programme was more effective with families with fewer risk factors than those with more. Also, socioeconomic status was associated with outcomeShared Parenting model may work with some but not all families. Birth families need to be invested in the programme to make it a successE
Lewandowski & Pierce (2004) Pilot: avg 9.9 years; comparison avg 7.5 yearsFamily-centred out-of-home care (FCOHC)Foster care children
N = 472
This study compared days in out-of-home care between a pilot group that received Family-centred out-of-home care (FCOHC) and a group that did not. FCOHC, unlike other preservation models, is designed to reunify children with families after out-of-home placement has occurred. FCOHC caseloads are higher than other family preservation child welfare workers, but smaller than the regular out-of-home workersProspective
Two Groups –FCOHC (n = 294) versus regular services (n = 178)
Children who received FCOHC services were reunified more quickly than the comparison group. Parental involvement and number of services were significantly related to reunificationAlthough FCOHC demonstrated success, there were a group of very high-risk families who needed the most services and were still less likely to reunify. This group may need many more services to achieve successful reunificationC
Lindsay et al. (1993) 0–1.9 (31%) 2–4.9 (28%) 5–11.9 (33%) 12+ (8%)Health passport – The San Diego modelFoster care children
N = 431
This study described the findings of children who received health passports when they were in family reunificationDescriptiveChildren exhibited multiple physical, developmental and psychological and/or behavioural concernsHealth passports are necessary for this high-risk populationG
Martin et al. (2002) KAOP Sample avg. 2.25 (urban), 5.67 (rural). Foster sample avg. 11.1 (urban), 10.4 (rural)Kentucky adoptions opportunity project (KAOP)Foster care/group homes
N = 1442
The KAOP model’s goal is for expedited permanency planning for high-risk children (most under 8 years) in both urban and rural areas of KY. The model incorporated risk assessment, maintaining same attorney for the child throughout court process, and the identification/implementation of early placement. Complexity of case plans was scored. Higher scores resulted among families with high amounts of parental morbidity or risk factors (e.g. mental health, substance abuse and developmental disorders). Comparisons between urban and rural counties were also made because of differences in resources. The goal of KAOP is to promote a single permanent placement of a foster childQuasi-experimental
KAOP N = 84
Other N = 1239
KAOP N = 30
Other N = 89
Lengths of stay were shorter for KAOP families than others. More KAOP families achieved permanency compared to others. Parental morbidity or risk factors and urban/rural were confounders that increased the length of stay and case score complexityCompliance with the case plan was key. The KAOP model may be a beneficial programme given new legislative requirements for permanency placementsE
McBeath & Meezan (2008) Pilot avg. 6.74 (SD 4.78); Nonpilot Avg 5.22 (SD 3.78)Service provision by child welfare and other agenciesFoster children
N = 243
This longitudinal study examines the services provided to foster care children under two types of service modalities fee-for-service and managed care. County system assigned child to agency (some pilot and others not) based on alphabet and capacity. Authors assumed randomisation based on this procedure. Outcome was service provisionProspective
Two groups –Pilot group (n = 175) versus regular services (n = 68)
Even after accounting for client characteristics and other confounders, service modality is associated with service provision. Managed care appears to curtail the number of services and increase disparitiesManaged-care is not an ideal model for service delivery of care for foster care children. Intensity of service was unrelated to client characteristics. Case workers differed in service delivery, based on experienceC
McDonald et al. (2003) Not givenInterdisciplinary primary care and support projectFoster care families
N = 10
This study conducted a 30- to 90-minute interview to assess foster care families perceptions of the multidisciplinary approach of primary and social servicesCross-sectionalBased on interviews, authors found foster parents had positive perceptions of services and felt the services improved their abilities to care for childrenPartnerships between the services provider and family, as well as a multidisciplinary team approach are more useful than hierarchical approaches that consist of a single type of serviceF
Nabors et al. (2004) 6–11 years (avg 7 years 2 months)Empowerment zone projectHomeless and at-risk youth
N = 141
This study describes a ‘summer camp’ programme for two groups of children: homeless and at-risk elementary youth children. The Empowerment Zone Project provided 10 classroom sessions on different topics such as health promotion, self-esteem, setting and achieving goals, conflict resolution. It also included small group sessions for mental health and health promotion. The outcome was to examine the perceptions of children and of teachersQuasi-Experimental
Two Groups –Homeless (n = 55) versus low-income (n = 86)
Demographic characteristics indicated that low-income children were more likely to have a primary care provider than homeless children. More homeless than low-income youth had histories of mental health intervention. Overall, youth reported satisfaction with the programme. No differences between homeless and at-risk youthSchools are a natural setting for students. More programmes need to use school settings to help introduce programmes designed that promote function and well-beingE
Risley-Curtiss & Stites (2007) Average Ages: Tx Urban 6.14 years; Tx Rural 7.16; Ctrl Urban 7.85; Ctrl Rural 7.32Instituting routine health examinationsFoster care children
N = 2507
The study compared rural and urban sites to assess barriers to routine health examinations for children in foster care via the Health Exam Pilot Project (HEP). Examinations also included mental health and developmental screenings. Comparisons were made between rural and urban sitesQuasi-experimental
Two groups
n = 954 HEP
n = 1385 control
n = 106 HEP
n = 62 control
More children in experimental group had completed physical exams compared to control group. The majority of child welfare workers did not report an improved level of health care services and also were not including findings into their reportsData collection system via computerised system was used, and case workers data input was considered problematic. There were many barriers and logistical problems that interfered with foster care children from obtaining physical examinationsE
Rog et al. (1995) Not givenRobert wood johnson foundation-funded homeless families programmeHomeless families
N = 1259
This study reported characteristics and need for participation of a foundation-funded programme for homeless families implemented in nine sites nationwide. The programme offered housing vouchers, case management and other services. A variety of standardised measures were used including portions of the National Health and Nutrition Examination Survey (NHANES), the Center for Epidemiological studies on Depression (CES-D), Conflicts Tactics Scale (CTS) and others. The goal was to examine service needsCross-sectionalAmong the nine sites, there were many demographic differences in the mothers (e.g. childhood risk factors) and need. Based on the findings, authors delineated different typologies of homeless families that incorporated human capital needs, health, mental health/substance abuse, and needs in multiple areasHomeless families are not a homogenous population. Family assessment is vital to determine service needs.F
Sanchirico & Jablonka (2000) Not givenFoster parent trainingFoster parents
N = 650
This study conducted an interview of foster parents in the NY State registry of foster parents to assess the impact of training on the role and responsibilities to facilitate child contact with birth parents. Two-stage sampling was used to ensure New York city and upstate representation, and the second stage for random selection of 3000 surveys (half to city and half to upstate). A total of 1160 surveys were returned; however, only 650 were used. These included respondents who were currently caring for at least one foster child and who were required to participate in child–parent contact. The goal was to assess the activities in which the foster parents engaged that promoted child contact with birth parents and to examine the role of training and child welfare system supportCross-sectionalThe results indicated that training and support received by foster parents increased the reunification promoting activities Most foster parents engaged in at least one activity that promoted contact with birth parents such as taking a child for a visit, helping with phone calls, providing supervised visitationIt is a vital responsibility of foster care parents to help their foster care children maintain contact with birth parents. Agencies should be actively promoting these activitiesF
Schram & Giovengo (1991) 16–18 years (median 17.8 years)Threshold model like boston bridgeHomeless youth
N = 24
This study examined the impact of a residential programme designed to help homeless female youth obtain independence and stability. Most have had trauma such as sexual or physical abuse, or neglect. The goal was to help the female youth continue education and/or gain employment within 19-month follow-up. The programme provided counselling (individual, group and family); medical/dental treatment; psychosocial educational groups for anger management, substance abuse, and parenting. Follow-up 19 months later indicated positive results for 42% of the sampleQuasi-Experimental
Cohort of female youth
Approximately 42% of female youth were able to complete programme requirements for school/employment, lived in a stable situation and were free from criminal/substance abuse. No explanation presented on why some youth succeeded and others did notFew programmes target this group of homeless female youth. Success rate is positive considering the amount of risk with which the female youth in this group presentedE
Slesnick et al. (2008) 14–24 yearsCommunity reinforcement approach (CRA)Runaway youth
N = 172
Youth aged 14–24 years in Albuquerque NM received the Community Reinforcement Approach (CRA) that addressed substance abuse, homelessness and mental health problems. In addition, wrap-around services were provided by a case manager. Pre- and posttest at baseline, 6 and 12 months using a series of measures assessed change. The following standardised instruments were used: Brief Symptom Inventory; Global Severity Index and Alcohol and Drug Use Form 90. The goals were to decrease substance abuse, promote stability with housing and employment, and increase psychological functionQuasi-experimental
Findings suggest that substance abuse among participants was decreased, and mental health was improved. Mental health was related to better housingAuthors concluded that costs for programmes such as this, which have a positive impact on substance abuse and mental health, are less than for prisonE
Slesnick et al. (2007) 14–22 years (mean 19.21, SD 2.15)Community reinforcement approach (CRA)Homeless youth
N = 180
The community reinforcement approach (CRA) was compared to treatment as usual to 180 homeless youth aged 14–22 years who made contact with a drop-in centre. CRA is a mental health approach with 12 sessions for on areas of life that clients wanted to change (e.g. social relationships, housing, employment, anxiety, depression) and four added sessions on AIDS education. Several standardised measures were used to make comparisons: National Youth Survey Delinquency Scale, Coping Inventory for Stressful Situations, and Beck Depression Inventory. Treatment fidelity was measured by assessing audiotapes of the treatment. The goal was to decrease substance abuseExperimental/RCC
Two groups –CRA (n = 96) versus treatment as usual (n = 84)
Not all youth completed the sessions. Among those who completed the sessions, there was higher prevalence of Marijuana and lower prevalence of alcohol dependence. HIV risk also decreased among homeless youth in CRA group. Other findings were that the CRA group demonstrated more stability, less depression and better social outcomes than the treatment as usual group. Some differences in impact of CRA based on age (higher impact for older youth compared to younger youth)CRA shows some promise, but the treatment as usual group also progressed in some areasB
Taylor et al. (2007) 16–23 years, avg 19Strong minded programme at foyer sitesRunaway youth
N = 26
The study described the use of ‘Strong Minded’ mental health programme for 18 youth who stayed in one of five Foyer sites in the UK. Several participants, in addition to the 1:1 mental health programme, were provided with additional mental health classes for anger management, etc. The goal was to assess satisfaction with services by youthQualitativeStrong minded programme reported by youth as being useful. Having a single person to contact was noted as particularly helpfulThere is a benefit of integrating mental health components with other educational programmes for homeless youthG
Weinreb et al. (2007) Not givenIntegrative behavioural healthBirth families
N = 999
This study described the model of a clinic using a model that combines health care with other services including mental health, support services, substance abuse and case management. The goal was to describe utilisation of the programmeCross-sectionalMothers were provided treatment at most encounters. The majority of behavioural services were provided to entire families, very few targeted individual children. The authors concluded that because of the multiple needs of families; the optimal programme not only will contain physical health services, but also the full spectrum of mental health and substance abuse services. No outcomes were providedAn integrative service model with multiple services best fits the complicated needs of homeless familiesF
Zlotnick et al. (1997) <1 (1.5%), 1–1.9 (37.4%), 2–2.9 (30.3%), 3–3.9 (30.8%)Case managementFoster care children
N = 130
This study used a cross-sectional sample of a longitudinal study that obtained a random sample of children newly entering foster care. The goal of the study was to examine the case management needs of newly placed foster care children and their familiesCross-sectionalChildren newly placed in foster care required many services; yet many of recommended services (at intake) were not provided at follow-up. Families with most need for case management services were also in the most need for mental health services, particularly parentingChildren newly entering foster care have multiple needsF

Interventions targeting age ranges

Many of these case management services studies were not targeting a specific age range of children; in fact, several did not identify the children’s age groups at all (Rog et al. 1995, McDonald et al. 2003, Anderson et al. 2006). Others served a wide age range of children (Lindsay et al. 1993, Gillespie et al. 1995, Carten 1996, Blatt et al. 1997, Martin et al. 2002, Kirk & Griffith 2004, Lewandowski & Pierce 2004, Risley-Curtiss & Stites 2007). Others, particularly programmes designed for homeless and runaway youth served clients, specified that the ages of the youth served ranged from 14–24 years old (Schram & Giovengo 1991, Ferguson 2007, Slesnick et al. 2007, 2008, Taylor et al. 2007).


Half of these articles (n = 13 or 50%) focused on family stability as a key outcome, describing programmes that supported family relationships including mental health interventions (Gillespie et al. 1995, Rog et al. 1995, Carten 1996, Blatt et al. 1997, Klee et al. 1997, Landy & Munro 1998, Martin et al. 2002, McDonald et al. 2003, Davey 2004, Kirk & Griffith 2004, Lewandowski & Pierce 2004, Anderson et al. 2006, Weinreb et al. 2007). Of these studies, the majority (n = 9) described programmes that worked in collaboration with the child welfare system, in some cases focusing on birth parents and in others foster care parents (Gillespie et al. 1995, Carten 1996, Blatt et al. 1997, Klee et al. 1997, Landy & Munro 1998, Martin et al. 2002, McDonald et al. 2003, Kirk & Griffith 2004, Lewandowski & Pierce 2004). Two of these studies focused on stability of homeless parents and families. None of the studies targeted both homeless and foster care populations.

Most other case management studies (n = 12) described programmes that focused on child outcomes. Four of these were designed to ensure that children obtained health-care through an infrastructure change or a new model that promoted health service access and utilisation (Lindsay et al. 1993, Hobbie et al. 2000, Risley-Curtiss & Stites 2007, McBeath & Meezan 2008). Six assessed the child’s function and stability after implementing a combination of wrap-around and psychosocial services (Schram & Giovengo 1991, Nabors et al. 2004, Ferguson 2007, Slesnick et al. 2007, 2008, Taylor et al. 2007). Two were school-based: one was situated in a shelter-based preschool programme (Grant 1991) and another was located in an elementary school-based summer camp (Nabors et al. 2004). The other studies targeting homeless and runaway youth employed programmes that combined mental health, vocational, educational and housing services (Schram & Giovengo 1991, Ferguson 2007, Slesnick et al. 2007, 2008, Taylor et al. 2007).

Two studies described programmes designed to improve child welfare services. One described a programme to promote child stability and function by focusing on the relationships between birth and foster care parents, ensuring that this connection facilitated the child’s visitation with birth parents (Sanchirico & Jablonka 2000). Another evaluated the home visit skills of their case workers (Kessler & Greene 1999).

Study designs

Similar to the mental health studies, the study design rigour differed among studies (Table 1). Of the 26 studies on case management, just one used an unblinded, randomised controlled, clinical trial study design (Level II) (Slesnick et al. 2007). Approximately 23% (n = 6) examined interventions using non-randomised, comparison groups with prospective cohort or quasi-experimental study designs (Martin et al. 2002, Kirk & Griffith 2004, Lewandowski & Pierce 2004, Nabors et al. 2004, Risley-Curtiss & Stites 2007, McBeath & Meezan 2008). The remaining 73% (n = 19) of the case management studies used a variety of quasi-experimental designs without comparison groups, descriptive or qualitative studies (Level V) (Grant 1991, Schram & Giovengo 1991, Lindsay et al. 1993, Gillespie et al. 1995, Rog et al. 1995, Carten 1996, Blatt et al. 1997, Klee et al. 1997, Landy & Munro 1998, Kessler & Greene 1999, Hobbie et al. 2000, Sanchirico & Jablonka 2000, McDonald et al. 2003, Davey 2004, Anderson et al. 2006, Ferguson 2007, Taylor et al. 2007, Weinreb et al. 2007, Slesnick et al. 2008).


This study reviews and synthesises existing literature to highlight best practice interventions and identify the overlap in services for U.S. children living in transition – including in foster care and literal homelessness. Surprisingly, this systematic literature search uncovered very few promising practices at all, and the studies all targeted children or youth in either foster care or homeless situations, not both. This review provides a valuable service to administrators, policy-makers and providers, and a starting place for the development of more effective interventions.

Research has clearly demonstrated that it is not uncommon for children and youth to experience both foster care and homeless situations through various phases of their developing years, and that the population has a multitude of physical and mental health needs in common. Some legislation has similarly acknowledged these overlaps. Yet, the articles reviewed employed samples of either homeless or foster children/youth, never both. The split in the intervention literature is not entirely surprising given the distinctive service systems that have emerged to address the specific populations. For example, the homeless service system has historically focused on responding to the diverse, wide-ranging issues and needs that the experience of homelessness creates for individuals and families. While some of these needs can only effectively be met with customised care, they are primarily caused by social and structural breakdowns; homelessness puts up barriers to services more readily available to housed persons. ‘Case management’ is therefore a critical service for homeless programmes, as a case manager’s primary role is to locate and coordinate services and resources for individuals. Because available services and resources differ widely, depending on location, politics, budgets, timing and so on, it is not possible or even desirable to narrowly define the case manager role; he/she must be a creative problem-solver addressing immediate needs. It is not surprising, given the history and reality of homelessness, that interventions and services arising in response are diffuse and include, but are not as narrowly limited to, mental health services we see in the foster care system. The focus among children in CPS was on mental health services, rather than case management. This was anticipated as the CPS worker’s role, like a case manager, is designed to help families with resources (although, unlike a case manager, the resources usually facilitate the implementation of their court-ordered case plan). Thus, case management is incorporated into the child welfare system and the CPS worker’s role. Most studies describing interventions for children in foster care refer to the trauma that preceded foster care entry, such as neglect or abuse, the trauma associated with child removal and the child’s health status upon entry to foster care. It is important to note that often the mental health services provided to children in foster care were screening, observation and assessment because of the increasing awareness of trauma’s impact on child behavioural development and mental health (Van der Kolk 2003), and the heightened attention to the long-term effects of childhood foster care on physical and mental health morbidity (Felitti et al. 1998, Zlotnick et al. 2004, Afifi et al. 2008), efforts to improve the mental health of foster care children is understandable.

This literature review reveals the potential for these two parallel universes – of U.S. foster care and homeless service systems – to learn from each other. Acknowledgement of the overlap, in the U.S. populations and in the children and families’ needs, is an essential first step to understanding which promising and evidence-based practice interventions may benefit both groups. Trauma-informed services, for example, will aid children and families served within homeless shelters as well as those within the foster care system. Targeted interventions like Lifebooks, shown to benefit children of various ages in foster care, may be equally effective for children living in homeless shelters or transitional housing programmes.

Although scales that rank evidence such as Sackett’s ordinal scale are used to assess quantitative and qualitative articles in a wide array of clinical disciplines (Sackett 1989, Guyatt et al. 1998, Wright et al. 2003, Wupperman et al. 2007), the criteria place all studies without control groups including observational studies that are cross-sectional, descriptive studies, case studies or series, and other qualitative studies in the same level. Consequently, it is not ideal for examining qualitative studies, which are known for their richness of information, rather than their precision of numbers.

On the whole, there is a dearth of research focusing on promising or evidence-based practice interventions for children and families experiencing either the foster care system and/or homelessness, and even within this body of research, the rigour of the study designs and methodologies is quite weak. Although this literature review focuses on U.S. populations of transitional families, it provides an important opportunity to assess the current status of existing promising practices for these populations, and by revealing the overlap in characteristics and needs between them affords new insight into how they can benefit individuals within both systems and lead to creative thinking about new best practices. This also opens the door to considering other systems these individuals may encounter and transition through – such as health clinics, schools and juvenile justice systems – when thinking about what prevention measures and targeted interventions might benefit these children and families.