Kinnecting Caregivers to Services, Resources, and Supports: Findings from an RCT of Colorado’s Kinship Navigator Program

Abstract

This study reflects the evidence-building journey for the Colorado Kinnected kinship navigator program. Colorado Kinnected expands the scope of services, resources, and supports offered to kinship families through an innovative approach that enhances an earlier Kinship Supports Demonstration Project. The Colorado Kinnected target population includes kinship caregivers caring for children and youth of any age with an open child welfare case. The primary purpose of the program is to reduce child welfare involvement by limiting traditional foster care and congregate care use when out-of-home placements are required. A randomized controlled trial was conducted from June to November 2020 to examine the impact of the Colorado Kinnected kinship navigator program on the kinship placement outcomes of 402 children and youth with an open child welfare case in seven Colorado counties. The initial study found that children and youth placed with kinship caregivers who received Colorado Kinnected services were significantly more likely to reunify with their parents after their kinship placements ended than were children and youth who received kinship supports as usual. The sustained effects study featured additional analyses of traditional foster and congregate care entry rates within six months of kinship placement for the same sample of children and youth. Children and youth in the intervention group were significantly less likely than children and youth in the control group to enter foster care or congregate care within six months after the end of their kinship placements. Overall, both analyses indicate that the Colorado Kinnected program promotes sustained permanency among children and youth in formal kinship placements. Data equity analyses to explore the contextual sensitivity of the intervention are detailed, and implications for practice, policy, and research are discussed.

1. Introduction

The primary purpose of the Colorado Kinnected (Kinnected) program is to reduce child welfare involvement by limiting traditional foster care and congregate care use when out-of-home placements are required. Kinnected is designed to engage families by using their existing social connections to support (1) placement/living arrangements with kinship caregivers, (2) reunification efforts, and (3) a whole-family approach to prevent the need for future child welfare involvement. If existing connections do not exist, Kinnected is also designed to assist the family in building a support network. The Kinnected target population includes kinship caregivers caring for children and youth of any age with an open child welfare case.

A randomized controlled trial (RCT) was conducted to examine the impact of Kinnected’s kinship navigator program on child welfare placement outcomes. This study includes findings from initial and follow-up data collection and analyses conducted on 402 children and youth in child welfare kinship placements whose caregivers were randomized to receive Kinnected kinship navigator program services (n = 202) or services as usual from seven Colorado county departments of human services (n = 200). The outcomes include a comparison of reunification, adoption, and guardianship rates; traditional foster care and congregate care entry rates; and placement stability at six months post-placement.

2. Literature Review

The literature review examines kinship care prevalence and outcomes in the United States, kinship caregiver needs, and kinship navigation programs and outcomes.

2.1. Kinship Care in the United States

Kinship care aims to maintain a child’s connections to their family through placement with relatives, family friends, or others already known to them. In the United States, 3% of children, or 2.4 million, were living in kinship care as of September 2023 [1]. In Colorado, 3% of children, or 34,000, were living in kinship care as of September 2023 [2]. Research points to positive outcomes for children and youth living in kinship care, such as improved placement stability, reduction in trauma, fewer disruptions to a child’s education and social lives, maintained connection with siblings and other family, and overall child well-being when compared to other forms of placement through the child welfare system [3,4,5,6]. Furthermore, kinship placement has been shown to reduce the risk of repeat child maltreatment or re-entry into out-of-home care [5,6].

2.2. Kinship Caregiver Needs

Children are placed in kinship care for a variety of reasons. including parent mental health or substance use, death, incarceration, or financial constraints [3,7]. Overwhelmingly, grandparents are the kinship caregivers that children are placed with [8,9]. Kinship caregivers, generally, have been shown to experience financial strain, with grandparent kinship caregivers specifically being likely to live below the poverty line and be out of the workforce [8,10]. Kin caregivers may still be employed, have a disability, or live in small residences [11]. As a result of these intersecting dynamics, kinship caregivers often struggle with adequate housing for kinship children, finding childcare while maintaining employment, and navigating education and healthcare systems [12]. Furthermore, older caregivers may need support for transportation, in-home services, and nutrition [11]. As a result, caring for kin can cause financial, familial, physical, and mental health stress on kinship caregivers [13,14]. To this end, kinship caregivers report needing assistance with financial support, community services and resources, health and mental health services, and legal representation [3,14,15,16].

2.3. Kinship Navigation Programs

Kinship navigator programs “provide both formal and informal kin caregivers with information, education, and referrals to a wide range of services and supports. The navigator’s goal is to maximize the caregivers’ ability to provide safety and stability and, if appropriate, permanency for the children placed in their home” [17] (p. 1). The 2018 Family First Prevention Services Act (FFPSA) demonstrates support for kinship navigation and offers funding to states to develop, implement, and evaluate these programs [18]. Kinship navigation programs are developed with the specific needs of caregivers in mind, with the goal of increasing access to services and supports to meet family needs, increase placement stability, and improve child and family well-being [3]. Programs must demonstrate effectiveness and be reviewed by the Title IV-E Prevention Services Clearinghouse (Clearinghouse) to receive federal reimbursement. To date, five programs have been rated by the Clearinghouse: 30 Days to Family is a “well-supported” program; Arizona’s Kinship Support Services and Colorado Kinnected (the focus of this article) are “supported”; and Nevada’s Foster Kinship Navigator Program and Ohio’s Kinship Supports Intervention (ProtectOHIO) are “promising” [19].

Research on rated programs shows that families receiving kinship navigation services saw fewer overall days in care, greater placement stability, and a lower likelihood of being placed in an inpatient treatment setting [5,20,21,22]. In addition, kinship navigator programs have led to an increased likelihood of caregivers being licensed (a legal document granting permission to operate a kinship foster care home), therefore increasing their access to services, such as child-only TANF [3,23]. Some caregivers receiving kinship navigation services also experience higher levels of well-being [3].

Existing kinship navigator programs have built an understanding of program components that contribute to better outcomes for kinship caregivers. For example, family-finding efforts and family supports have been shown to be effective [20,21]. Family-finding efforts involve intensive searches for families who may be interested in and able to help care for a child or otherwise support kinship caregivers. Due to social isolation and the need for social supports, family-finding interventions can be particularly impactful for kinship caregivers [3,24]. Administering needs assessments to kinship caregivers has also proven effective [5], as it allows child welfare agencies to better understand specific areas of family support based on intensity of needs or stressors, and family strengths.

3. Background

The overarching purpose of Kinnected is to maintain children and youth in a kinship placement until they can be safely reunified with their parents while strengthening the sustainable support community around children, youth, and their families. The kinship supports component provides (1) direct supports to each kinship family based on their specific needs for maintaining the placement, (2) facilitated family engagement meetings to incorporate kinship caregiver voice in child welfare case goals and keep kinship caregivers apprised of progress toward goals, and (3) family search and engagement to identify family members and friends of kinship caregivers who can provide caregiving supports and determine the level of support each person is willing to provide.

Colorado Kinnected expands the scope of services offered to kinship caregivers through the Title IV-E Waiver Demonstration Project (Waiver) Kinship Supports Intervention. Through the Waiver, the Colorado Department of Human Services (CDHS) offered county departments of human services the opportunity to trade guaranteed federal foster care maintenance and administration reimbursement for up-front funding to implement one or more of the five Waiver interventions.

The findings of the Waiver evaluation indicated that the kinship supports intervention worked in tandem with another Waiver intervention, facilitated family engagement meetings, and increased the likelihood of reunification with parents among children and youth placed in kinship care [25]. The Kinnected program builds on the synergy between the two interventions by more formally integrating kinship caregivers into child welfare case services and goals through facilitated family engagement meetings. The program also strengthens the kinship supports intervention through increased face-to-face meetings and ongoing contacts with kinship caregivers, increased assessment of caregiver needs, and the addition of family search and engagement activities to build robust support networks around kinship families.

Kinship caregivers, parents, and other stakeholders informed the development of Kinnected. In-person and virtual meetings with Colorado kinship caregivers, county departments of human services child welfare staff, and community-based partners were conducted to gather and incorporate feedback on program services, tools, and procedures.

4. Methodology

This study’s design, target population, study assignment, data collection, treatment condition, control condition, study setting, analytic sample, baseline equivalence, missing data, data analysis, and outcome measures are presented in the methodology section.

4.1. Study Design

An RCT was conducted to examine the impact of the Kinnected kinship navigator model on permanency and placement disruption outcomes for children and youth placed with kinship caregiving families in the seven Kinnected counties. The counties each placed children and youth in need of out-of-home care with kinship families; CDHS randomly assigned each kinship family to the intervention group for Kinnected services or the control group for kinship supports, as usual; and the county departments of human services administered services according to group assignment. The evaluation team collected and analyzed secondary data on case, kinship family, and child and youth characteristics and placement outcomes included in Trails, Colorado’s Comprehensive Child Welfare Information System.

The primary study hypothesis was that children and youth in the Kinnected intervention group would be more likely than children and youth in the control group to remain in their kinship placement until reunification with their parents was achieved. It was also expected that children and youth placed with kinship caregivers who received Kinnected program services would be less likely to disrupt and need another child welfare placement.

4.2. Target Population

The Kinnected target population comprised kinship families caring for children and youth of any age with an open child welfare case. This included children or youth in court-ordered kinship placements, informal living arrangements between a county department of human services and kin, and placements with kinship caregivers who were certified or uncertified as foster caregivers. To receive Kinnected services, two conditions must be met: (1) children must be in an open child welfare involvement with or without court involvement and (2) children must be entering a new certified or non-certified kinship placement/living arrangement. Child welfare kinship placements in Colorado last approximately six months on average; however, there is no expected duration. The primary objective is to keep children and youth with kin until parents can achieve case goals and reunification can safely occur. Accordingly, Kinnected is designed to meet the needs of kinship families from placement start to end, regardless of duration. Some services do gradually taper, and the bulk of services are delivered in the first 60 to 90 days of the placement. Due to COVID-19 health and safety precautions during the study, kinship caseworkers were permitted to substitute face-to-face in-person meetings for videoconference meetings as needed.

4.3. Study Assignment

The study sample includes all children and youth placed with kinship caregiving families allocated to the intervention and control groups from June 2020 to November 2020 in seven counties. The unit of assignment was kinship caregiving families. To be included in randomization, a family had to have at least one child or youth with an open child welfare case in one of the seven county departments of human services placed with them between June 2020 and November 2020. Within each of the seven counties, CDHS randomly assigned families at the start of the placement (on or just prior to the day that the placement episode began) to the treatment or control group using simple randomization procedures in SQL Server.

On the date that a family was opened as a placement provider, the caseworker assigned to the family received an automatically generated email with a link for randomizing the family to the next group in the randomization sequence. The caseworkers excluded families from randomization if they were temporary or transitional short-term caregivers, intermittent respite care-only caregivers, or caregivers residing in a different county or state. Families were also excluded if the county’s department of human services determined between the placement open date and randomization that the child(ren) and/or youth would not ultimately be placed with the family or if the family explicitly requested no kinship services from the county department of human services. After randomization, families were only excluded if the placement lasted less than seven days and the baseline needs assessment could not be completed.

4.4. Data Collection

All study data were gathered by county department of human services staff as part of routine practice and procedures. Intake and ongoing caseworkers collected demographic and risk characteristics for each child, youth, and family using existing county intake forms and the Colorado Family Risk Assessment. Kinship navigators and support workers collected kinship family characteristics during the initial assessment of kinship family needs. Kinship placement outcomes were tracked by ongoing caseworkers. County department of human services staff entered all data into Trails as they were collected.

At the end of the outcome observation period on 31 January 2022, the evaluation team collected Trails data extracts from CDHS for analysis. The data included child welfare cases, kinship family, and child and youth identification numbers; study group assignment; available family, child, and youth demographic characteristics and risk profiles; kinship family needs assessments and services; and kinship placement begin/end dates and end reasons. All data were transferred to the evaluation team in Microsoft Excel spreadsheets using Citrix ShareFile, a web-based secure file transfer service. Child, youth, and primary family demographics and risk characteristics were reported by parents prior to randomization during the abuse and/or neglect assessment and case-opening process. Kinship family characteristics were reported by caregivers during the initial Kinship Supports Needs Assessment. Kinship placement outcomes were collected through January 2022. Intervention and control group services began within 7 days of randomization and continued until the placement ended.

4.5. Treatment Condition

All kinship families in the intervention group were expected to receive, at a minimum, the following services:

(1) Colorado Kinship Navigation Guide, which includes statewide and local resources for kinship caregivers, such as food banks, public benefits, legal services, and community agencies with kinship programming;

(2) Monthly face-to-face meetings with their kinship caseworker during the first 90 days of the placement and quarterly thereafter for the remainder of the placement;

(3) Kinship Supports Needs Assessment initiated within seven business days and completed within 30 days of the placement start date, and completed again two months after the placement start date and at the end of the placement;

(4) Direct services and supports or referrals to public or community agencies for services and supports that match caregiving needs and stressors;

(5) Genogram initiated in Genopro software (2020 version) within 30 days of the placement start date and reviewed and updated during each subsequent facilitated family engagement meeting;

(6) Family Involvement Continuum for as many individuals on the genogram as is possible within seven business days of the placement start date, at 60 days after the placement start date, and at the end of the placement;

(7) Invitation and support to attend or have their voice represented at each facilitated family engagement meeting held after the placement start date;

(8) Backup placement plan identified within 30 days of the placement start date and reviewed 60 days after the placement start date and at the end of the placement.

The seven counties were expected to utilize kin-specific workers to support the kinship caregiver’s ability to care for children by (1) providing hands-on guidance and trauma-informed supports, (2) ensuring “kin voice” was heard in facilitated family engagement meetings, and (3) addressing kinship caregiver needs and social support connections. A kinship caseworker was assigned to each kinship family caring for a child or youth with an open child welfare case in each county. The kinship caseworker was responsible for ensuring that services were delivered to the family as prescribed in the program manual. Some counties had kinship caseworkers who only provided Kinnected intervention group services and kinship caseworkers who only provided control group services. Other counties had kinship caseworkers who provided both intervention group and control group services so that caseload balance could be better maintained across the caseworkers throughout randomization. All kinship caseworkers in each county were supervised by a child welfare supervisor or administrator who monitored caseloads and implementation of intervention and control group services.

Bi-weekly support calls for reinforcing initial program training and addressing implementation challenges were held throughout the study period with kinship administrators, supervisors, and caseworkers. Other implementation supports included a case-level checklist and flow chart with timeframes and processes for completing services and an implementation checklist that was completed by kinship caseworkers to gauge their understanding of and capacity to administer the program.

4.6. Control Condition

The control condition included the usual county department of human services supports for kinship families caring for children and youth with an open child welfare case in the seven counties that participated in the impact study. Each of the seven counties implemented the kinship supports intervention under the Waiver and continued to implement it as standard practice for kinship families after it ended in 2019. Bi-weekly support calls with kinship caseworkers who provided services to the control group were facilitated by CDHS to help ensure that the control group received services as usual. Kinship caseworkers who served control group kinship families, regardless of whether they provided services to families in both the treatment and control groups, were required to understand control group service expectations and attend bi-weekly support calls.

Kinship families in the control group were expected, at a minimum, to receive the following services: (1) Kinship Supports Needs Assessment completed at the start of their placement and (2) direct services and supports or referrals to public or community agencies for services and supports that match their initial caregiving needs. In addition, parents were expected to receive facilitated family engagement meetings according to county department of human service existing rules and procedures.

4.7. Study Setting

Colorado is a state-supervised, county-administered child welfare system. The Kinnected kinship navigator program was supervised by CDHS and administered in seven counties by each county’s department of human services. Kinship supports and family search and engagement services were provided by county child welfare caseworkers in county human services offices and kinship family homes. Facilitated family engagement meetings were provided by child welfare caseworkers and family engagement facilitators in county human services offices.

4.8. Analytic Sample

A total of 573 families were assessed for study eligibility during the study period. Of those, 174 were excluded from randomization for the following reasons.

• Placement did not ultimately occur (n = 121);
• Placement was known to be temporary or transitional (n = 19);
• Family lived in a different state (n = 21);
• Family requested no kinship support services (n = 7);
• Family was only serving as respite care (n = 6).

Eight families were also excluded from the treatment group and 20 were excluded from the control group after randomization because their placements lasted less than seven days and the baseline assessment could not be completed, leaving 196 families in the intervention group and 175 in the control group.

There were 273 children and youth placed with families in the intervention group and 261 placed with families in the control group. Of those, 71 children and youth were lost to follow-up in the treatment group and 61 were lost in the control group because their placements did not end by the close of the outcome observation period on 31 January 2022 (placement outcome could not be observed). The overall attrition rate across both groups was 24.7%, and the differential attrition rate between groups was 2.6%. The final analytic sample included 202 children and youth placed with caregivers assigned to the intervention group and 200 children and youth placed with caregivers assigned to the control group.

4.9. Baseline Equivalence

Seven variables were used to assess group equivalence collected during the abuse and/or neglect assessment or at the start of intervention and control group services. Two indicators of socioeconomic status were measured, including primary family financial stress and kinship family financial need. Primary family financial stress was rated on a three-point scale (1 = None, 2 = Moderate, 3 = Severe) by intake caseworkers during an investigation or by ongoing caseworkers at case open. Kinship family financial need was rated on a five-point scale (1 = None, 2 = Low, 3 = Moderate, 4 = High, 5 = Urgent) by kinship support workers at kinship placement start.

Table 1. Baseline equivalence of the intervention and control groups.

Baseline Measure	Intervention Group Children/Youth N = 202			Control Group Children/Youth N = 200			Between Groups
	N 1	Mean	SD 2	N  1	Mean	SD 2	Mean Difference	Effect Size 3
Primary family financial stress	186	2.01	0.782	200	2.00	0.750	0.010	0.013
Kinship family financial need	191	1.92	1.196	181	1.74	1.213	0.180	0.149
Child/youth gender female (%)	202	0.52	0.501	200	0.56	0.498	−0.040	0.098
Child/youth age	202	7.62	5.829	200	6.61	5.154	1.010	0.183
Child/youth racial/ethnic minority (%)	202	0.59	0.493	200	0.62	0.488	−0.030	0.076
Child/youth physical or sexual abuse (%)	202	0.19	0.392	200	0.20	0.401	−0.010	0.039
Child/youth prior child welfare case (%)	202	0.12	0.324	200	0.11	0.314	0.010	0.060
Child/youth overall risk	197	2.68	0.529	200	2.68	0.538	0.000	0.000

¹ Sample size for each measure (number of children and youth with a valid/non-missing value for the measure). ² Standard deviation. ³ Cox transformation used to convert odds ratio to standardized mean difference effect size (log odds ratio divided by 1.65).

includes the number of children and youth whose placements ended in the intervention and control groups (the analytic sample) who had a valid/non-missing value for each baseline measure, the mean and standard deviation for each group, and the mean difference and effect size between groups.

4.10. Missing Data

None of the 402 children and youth whose kinship placements ended during the outcome observation window were missing a value for the placement discharge setting variable in Trails; however, 16 (4.0%) of the children and youth were missing a value for the family financial stress baseline measure, 30 (7.5%) were missing a value for the kinship family financial need measure, and five (1.2%) were missing a value for the overall risk measure. Of the 200 children and youth in the control group, 19 were missing a kinship family financial need value in Trails. Expectation-maximization in IBM SPSS Statistics for Windows, version 25 (IBM Corp., Armonk, NY, USA) was used to estimate and impute missing data prior to including these measures as covariates in the outcome analyses. The correlation of each baseline measure and placement outcome did not differ substantially between the entire analytic sample of children and youth with imputed baseline measure values and the sample of children and youth without imputed baseline measure values, indicating a low risk for bias.

Hedges g effect size was calculated for the ordinal and scale measures and Cox’s d effect size was calculated for the nominal/dichotomous measures. Effect sizes for family financial stress, child and youth physical or sexual abuse, and child and youth overall risk were in the acceptable range (<0.05), and effect sizes for kinship family financial need and child and youth gender female, age, ethnic/racial minority, and prior child welfare case were in the adjustment range (between 0.05 and 0.25). Effect sizes falling within the adjustment range indicate an imbalance between groups that was accounted for by including them as covariates during statistical analysis.

4.11. Data Analysis

IBM SPSS Statistics for Windows was used to conduct the impact analysis. A multinomial logistic regression model was initially fitted to determine if there was an overall relationship between treatment conditions (intervention versus control group) and placement outcomes. The dependent variable for this model included six categories, one for each of the placement outcomes. Treatment conditions and the eight baseline measures were entered into the model simultaneously as predictors. Controlling for the baseline measures, the likelihood ratio test for treatment conditions was significant [χ²(5) = 11.866, p < 0.05], indicating that there was a relationship between treatment conditions and the placement outcomes.

A binary logistic regression model was then fitted for each child and youth placement outcome separately, with treatment condition as the test factor, the baseline measures as control variables, and placement outcome (experienced the outcome or did not experience the outcome) as the dependent variable. For each regression model, the treatment condition and all control variables were entered simultaneously in one block. The covariates, including family financial stress, kinship family financial need, and child or youth age and overall risk, were tested a priori before running each regression model to verify that the assumption of the linearity of the logit was not violated.

Generalized linear modeling was used to fit the binary logistic model for each permanency outcome except move to a congregate care placement. A mixed-effects binary logistic regression was used to fit the model for this outcome because the intraclass correlation coefficients (ICC) indicated that 63.2% of the likelihood of moving to a congregate care placement in the intervention group was explained by the between-county department of human services variability. The random effect included children and youth (level 1) clustered within counties (level 2), and the fixed effects included treatment conditions and the baseline variables.

4.12. Outcome Measures

The variable used for the study outcome measure was placement discharge setting. It was recorded in Trails for each child or youth by ongoing child welfare caseworkers in the county department of human services as part of standard placement end procedures. Observation of placement discharge settings for the study ended on 31 January 2022.

Table 2. Placement discharge setting values.

Placement Discharge Setting	Analytic Sample N = 402 Children and Youth
	Number	Percentage
Reunification with parents	197	49.0%
APR Allocation of Parental Rights (APR)/Guardianship	75	18.7%
Foster placement	53	13.2%
Kinship placement	50	12.4%
Adoption	9	2.2%
Residential placement	6	1.5%
Runaway	4	1.0%
ILP	3	0.7%
Group placement	2	0.5%
Death	1	0.2%
Detention/DYS	1	0.2%
Hospital	1	0.2%
Total	402	100.0%

provides descriptive statistics for the placement discharge settings of children and youth placed with kinship families enrolled in the study whose placement ended by the last day of the observation period.

Figure 1 includes the placement discharge setting or settings used for each outcome variable in the study. The 11 placement discharge setting values in Trails were collapsed into six outcome variables for the analysis of children and youth placed with kinship families allocated to the intervention or control group whose placements ended. Each outcome variable is dichotomous; for analysis, children and youth were assigned a numerical value of one if the outcome applied to them and a value of zero if it did not. Because placement discharge settings were mutually exclusive, only one placement outcome applied to each child or youth.

4.13. Initial and Sustained Effects Study Findings

The outcome analyses focused on the impact of the Kinnected kinship navigator program by examining kinship placement permanency and disruption outcomes.

4.13.1. Descriptive Statistics

The number and percentage of children and youth experiencing each kinship placement outcome in the intervention and control groups are included in

Table 3. Kinship placement outcome descriptive statistics for the analytic sample.

Kinship Placement Outcome	Intervention Group Children and Youth N = 202		Control Group Children and Youth N = 200		Percentage Difference
	Number	Percentage	Number	Percentage
Permanency
Reunification with parents	107	53.0%	90	45.0%	8.0%
Guardianship/Adoption	39	19.3%	45	22.5%	−3.2%
Disruption
Move to another kinship placement	23	11.4%	27	13.5%	−2.1%
Move to a non-relative foster care placement	19	9.4%	34	17.0%	−7.6%
Move to a congregate care placement	7	3.5%	1	0.5%	3.0%
Other 1	7	3.5%	3	1.5%	2.0%
Total	202	100%	200	100%

¹ Death, detention, hospital, independent living program, or runaway.

. The number of children and youth placed with each kinship family ranged from one to five, with 74.1% of the 147 families in the intervention group and 63.9% of the 133 families in the control group having one child or youth placed with them. The mean number of children and youth placed with each family in the intervention group was 1.43 (SD = 0.84), and the mean number of children and youth placed with each family in the control group was 1.55 (SD = 0.87). For the permanency outcomes, a greater percentage of children and youth in the intervention group than in the control group reunified with parents at placement end, and a smaller percentage experienced guardianship or were adopted. For the disruption outcomes, a smaller percentage of children and youth in the intervention group than in the control group moved to another kinship placement or moved to a non-relative foster care placement. A greater percentage moved to a congregate care placement or had an “other” placement outcome. However, only 2% (n = 8) of children and youth in the two groups combined moved to a congregate care placement, and only 2.5% (n = 10) had an “other” placement outcome.

For the potential effect of children and youth clustered within the seven county departments of human services, the ICC for all but one of the placement outcomes were very small. The unit of analysis was the individual child or youth in the analytic sample. The clustering of children and youth within kinship families did not need to be addressed in the outcome analyses.

4.13.2. Impact Study Results

The results for each binary logistic regression model are displayed in

Table 4. Binary logistic regression kinship placement outcome analysis findings.

Kinship Placement Outcome	Intervention Group Children and Youth			Control Group Children and Youth			Estimated Effect
	Sample Size	Unadjusted Mean	Adjusted Mean	Sample Size	Unadjusted Mean	Adjusted Mean	Impact	p Value	OR 1	Effect Size 2
Permanency achieved
Reunification with parents	202	0.530	0.616	200	0.450	0.506	0.110	0.032	1.565	0.271
Guardianship/Adoption	202	0.193	0.134	200	0.225	0.183	−0.049	0.150	1.312	0.165
Placement disruption
Move to another kinship placement	202	0.114	0.099	200	0.135	0.109	−0.010	0.744	1.097	0.056
Move to a non-relative foster care placement	202	0.094	0.054	200	0.170	0.105	−0.051	0.023	1.512	0.251
Move to a congregate care placement 3	202	0.035	0.049	200	0.005	0.033	0.016	0.421	1.520	0.254
Other 4	202	0.035	0.004	200	0.015	0.003	0.001	0.659	1.407	0.207

¹ Odds ratio [Exp(B)]. ² Cox transformation used to convert odds ratio to standardized mean difference effect size (log odds ratio divided by 1.65). ³ Mixed modeling used for this outcome to adjust for variance explained by clustering of children and youth within county department of human services. ⁴ Death, detention, hospital, independent living program, or runaway.

. Included are the intervention and control group sample sizes and means for each placement outcome analysis, along with the difference in adjusted outcome means between groups (impact), the significance of the treatment condition variable coefficient (p value), odds ratio [Exp(B)], and standardized mean difference effect size (log odds ratio divided by 1.65) for intervention versus control group membership.

The results indicate that Colorado Kinnected promotes reunification with parents among children and youth in child welfare kinship care placements. As expected, children and youth in the intervention group were more likely than children and youth in the control group to remain in their kinship placements until they reunified with their parents. The contribution of treatment condition to the logistic regression model for this outcome was statistically significant [B = (0.448), SE = 0.209, Wald = 4.585, p < 0.05], and the odds of reunification for children and youth in the intervention group were 56.5% greater than the odds of reunification for children and youth in the control group [Exp(B) = 1.565].

The results also indicated that Colorado Kinnected prevents moves to non-relative foster care following a kinship placement disruption. The contribution of treatment conditions to the logistic regression outcome model for this outcome was statistically significant [B = (−0.718), SE = 0.317, Wald = 5.133, p < 0.05], and the odds of moving to a non-relative foster care placement following disruption for children and youth in the intervention group were 51.2% less than the odds of moving to a non-relative foster care placement following disruption for children and youth in the control group [Exp(B) = 0.488]. Treatment conditions did not contribute significantly to the logistic regression models for the guardianship/adoption, move to another kinship placement, move to a congregate care placement, or other placement outcomes.

4.13.3. Sustained Effects Study Results

The percentage of children and youth in the intervention group who entered foster or congregate care within six months of their kinship placement end date (i.e., treatment end) was 7.9% (16 out of 202), compared to 14.5% (29 out of 200) of children and youth in the control group, which was a 6.6% difference.

The results of the regression model are shown in

Table 5. Binary logistic regression model findings for entry into traditional foster or congregate care.

Outcome	Intervention Group Children and Youth			Control Group Children and Youth			Estimated Effect
	Sample Size	Unadjusted Mean	Adjusted Mean	Sample Size	Unadjusted Mean	Adjusted Mean	Impact	p Value	OR 1	Effect Size 2
Foster or congregate care entry within six months of kinship placement end	202	0.079	0.070	200	0.160	0.138	−0.068	0.0239	0.46875	−0.4592

¹ Odds ratio [Exp(B)]. ² Cox transformation used to convert odds ratio to standardized mean difference effect size (log odds ratio divided by 1.65).

. Included are the intervention and control group sample sizes and means for each placement outcome analysis, along with the difference in adjusted outcome means between groups (impact), the significance of the treatment condition variable coefficient (p value), odds ratio [Exp(B)], and standardized mean difference effect size (log odds ratio divided by 1.65) for intervention versus control group membership.

The results indicated that the Colorado Kinnected program promoted sustained permanency among children and youth in child welfare kinship placements. Children and youth in the intervention group were less likely than children and youth in the control group to enter foster or congregate care within six months of their kinship placements ending. The contribution of treatment conditions to the logistic regression model for this outcome was statistically significant [B = (−0.7576), SE = 0.33539, Wald = 2.4, p < 0.05], and the odds of foster or congregate care entry within six months of kinship placement end (i.e., treatment end) for children and youth in the intervention group were 2.13 times less than the odds of foster or congregate entry within six months of kinship placement end for children and youth in the control group [Exp (B) = 0.46875].

5. Data Equity Study

A data equity analysis conducted on the initial study results examined the differences in experiences and outcomes of potentially vulnerable groups. Although all children and youth involved in the study were in vulnerable positions, certain sub-groups may face unique challenges. Investigation into these groups, hereby referred to as “equity subsets”, can add additional insight to the results of the kinship navigator experiment while recontextualizing discoveries made in the impact analysis.

5.1. Refitted Models

The equity subsets studied for this brief include children and youth identified as “female”, “Black”, “Hispanic”, and “White”. Additionally, a group of children and youth identified as “in financial need” was created, consisting of children and youth whose caregivers indicated that their current financial need was “Moderate”, “High”, or “Urgent”.

Subsets were chosen based on a combination of available sample sizes and field knowledge of factors affecting potential systemic vulnerability. The analysis employs three different approaches to study how the process and results of the impact study may have varied across children and youth within these groups. The data for each equity subset were analyzed separately to determine whether the experiences of these subsets diverged from those of the general group and, if so, to what extent. Since these subsets were no longer balanced according to the standards set in the RCT impact study, propensity score stratification was used to create a synthetic comparison group out of the control cases in each subset. The baseline measures are shown in

Table 6. Baseline measures for refitted models.

Standardized Mean Difference for Each Subset after Stratification
Equity Subset	Ethnic Minority	Female	Financial Stress	Financial Need	Physical/Sexual Abuse	Prior Case	Overall Risk	Age	Sample Size
Hispanic		−0.01	0.03	0.06	0.04	0.02	−0.05	0.07	182
Black		0.34	−0.06	−0.19	−0.21	−0.26	−0.22	−0.10	49
White		−0.10	−0.09	−0.04	0.00	−0.05	0.00	−0.04	160
Female	0.05		−0.06	0.00	0.00	0.01	−0.01	−0.02	216
Financial Need	0.10	−0.06			0.02	−0.11	0.05	0.05	105

.

5.2. Baseline Subset Approach

For each equity subset, baseline measures were compared to the overall data to determine whether any significant disparities existed at entry into the program. Disparities in baseline measures for equity subsets can shed light on (1) differences in outside factors relevant to their experience in the program and (2) differences in factors for the likelihood of entry into the program. Standardized mean differences (SMD) of less than 0.05 are considered acceptable; differences between 0.05 and 0.25 are in the adjustment range and should be accounted for during analysis; and differences greater than 0.25 are unacceptable and cannot be accounted for during analysis. Only two baseline measures (Black Youth Female and Black Youth Prior Case) were in the adjustment range and thus were accounted for in the analysis.

5.3. Outcomes within Treatment

Within children and youth whose kinship caregivers received kinship navigator program services (i.e., the treatment group), outcomes for equity subsets were studied. Propensity score stratification was used to improve baseline equivalence within the treatment group. Baseline measures are shown in

Table 7. Baseline measures for treatment subsets.

Standardized Mean Difference for Each Subset after Stratification
Equity Subset	Ethnic Minority	Female	Financial Stress	Financial Need	Physical/Sexual Abuse	Prior Case	Overall Risk	Age	Sample Size
Hispanic		0.05	0.00	0.00	−0.05	−0.06	0.03	−0.01	91
Black		−0.06	−0.04	−0.04	0.04	0.02	−0.07	0.02	21
White		−0.13	0.05	−0.10	0.02	−0.21	0.04	−0.12	83
Female	0.10		0.05	−0.02	0.04	−0.03	−0.07	−0.07	105
Financial Need	−0.04	0.04			0.05	0.00	−0.04	−0.03	61

. Baseline equivalence was achieved for all measures.

5.4. Data Equity Study Results

The impact study found that children and youth who received kinship navigator services were more likely to reunify with their parents with an odds ratio of 1.57 (p = 0.03). For Hispanic children and youth, the same difference was found with an odds ratio of 2.33 (p = 0.01). For Black and White children and youth, no significant differences in reunification rates were found between the treatment and comparison groups. The disparity is illustrated in Figure 2.

The impact study found that children and youth who received kinship navigator services did not have a significantly different likelihood of experiencing guardianship or adoption. However, Hispanic children and youth who received kinship navigator services had a significantly lower chance of experiencing guardianship/adoption with an odds ratio of 0.42 (p = 0.03).

Children and youth in families with financial needs who received navigator services had a significantly lower chance of experiencing guardianship/adoption, with an odds ratio of 0.29 (p = 0.007). Finally, female children and youth who received kinship navigator services had a significantly lower chance of experiencing adoption/guardianship, with an odds ratio of 0.46 (p = 0.04). These results should be interpreted in light of the results from the impact study that children and youth who received kinship navigator services were more likely to reunify with their parents, especially Hispanic children and youth. Therefore, the lower rate of guardianship/adoption in many cases may be due to the attainment of another desirable outcome.

The impact evaluation found that children and youth who received kinship navigator services did not have a significantly different likelihood of moving to another kinship placement. These results held for children and youth in all equity subsets. This is likely a favorable outcome as the most common result was children remaining with their kinship caregiver and thus not experiencing placement disruption. The impact study found that children and youth who received kinship navigator services were significantly less likely to move to non-relative foster care placement with an odds ratio of 0.49 (p = 0.02). This held for most equity subsets but was significantly greater for Hispanic children and youth with an odds ratio of 0.14 (p = 0.004). The other outcomes in the models did not have a sufficient proportion of results to reliably explore through subset analysis and were not analyzed in the data equity study.

5.5. Baseline Subset Analysis

Most disparities in baseline measures among equity subsets were minor and to be expected of smaller subsets of matched groups; however, a few stood out as potentially important. Hispanic children and youth, on average, had caregivers in higher financial need (p = 0.06) than non-Hispanic children and youth and were involved in cases with higher financial stress (p = 0.003). These disparities can be attributed largely to a significant difference within the comparison group and did not exist in the treatment group. Black and Native American children and youth were aggregated due to similarities in their baseline measures and small sample sizes, improving the ability to register statistically significant results with similar effect sizes. Black and Native American children and youth who received kinship navigator services had significantly longer placements by an average of 68 days (p = 0.02). Children and youth from families with higher financial need were significantly less likely to have reported parental physical or sexual abuse (p = 0.02).

5.6. Outcomes within Treatment Group

No significant differences in reunification rates were observed between equity subsets within the treatment group, including Hispanic children and youth. Recall that the difference between the comparison group and the treatment group in reunification rates was mostly due to a large difference among Hispanic children and youth. This difference was, in large part, due to a huge deficit of reunification rates in the comparison group among Hispanic children and youth. In other words, Hispanic children and youth saw somewhat higher reunification rates in the treatment group and substantially lower reunification rates in the comparison group.

Categorization of a group into individual identities is innately limiting in its ability to capture personal and intersectional experiences with respect to those identities. Because of this, slight variation among groups could easily be attributed to imprecision in the demographic measures themselves, and only strong disparities or large magnitude could be reasonably reported; even then, the exact estimates of differences may be less accurate. Therefore, analysis of this equity subset, while statistically sound, is only applicable specifically to the divide between “low to none” and “moderate or above”, not any other divides that exist within the responses.

6. Discussion

This study contributes to the literature showing positive outcomes for children in kinship care, more specifically, children and caregivers who received services through kinship navigator programs [5,21,22]. The impact and sustained effect studies found that children and youth placed with kinship caregivers who received Colorado Kinnected kinship navigator services were more likely to reunify with their parents at kinship placement end and less likely to enter traditional foster care or congregate care within six months of placement end than children and youth kinship caregivers who received kinship support services as usual. While kinship care can lead to greater placement stability and reduced child maltreatment, kinship caregivers face a unique set of challenges when they take relative children into their care. Kin caregivers are often older and may have a disability in addition to having to juggle employment and childcare or a housing situation that is conducive to caring for children [11,12]. Due to the common profile of these caregivers, kinship navigator programs necessitate a multi-pronged, wraparound approach that enables kin caregivers to best care for their relative children while also promoting placement stability and caregiver wellbeing. Colorado Kinnected was designed with these considerations in mind.

6.1. Implications

Given the critical role of kinship caregivers in the lives of children and the need to research the effectiveness of kinship navigator programs, this study has several implications for the broader field of kinship care. The Kinnected program received a rating of “promising” in December 2022 and “supported” in March 2023 from the Title IV-E Prevention Services Clearinghouse, making program administration and service costs eligible for up to 50 percent federal reimbursement under FFPSA. As such, the Kinnected program can now be implemented and sustained across all of Colorado’s county child welfare agencies, which means that all kinship caregivers in the state who are caring for children and youth in an out-of-home placement will be eligible for the enhanced set of services offered by the Kinnected program, regardless of geography. This will promote consistency in services and a continued positive impact on child welfare involvement outcomes statewide. The Kinnected program can also be implemented under FFPSA by other state child welfare agencies that are looking to replicate a rated kinship navigator program instead of developing and testing their own programs. A range of implementation supports are available. For example, the program includes a readiness checklist for kinship navigators that assesses the extent to which the navigator has the necessary tools, knowledge, and administrative supports to implement the program with fidelity to the model. Procedures for tracking fidelity and claiming services under FFPSA are also available to other state child welfare agencies.

The facilitated family engagement meetings and family search and engagement components of Kinnected may make it particularly effective for supporting child welfare reunification goals and preventing entry into foster or congregate care. Incorporating kinship caregiver voices into facilitated family engagement meetings allows kinship caregivers to communicate with parents about child placement progress and well-being, to voice concerns and address placement issues, and to stay apprised of progress toward case goals and reunification efforts. The family search and engagement component provides an opportunity for identifying the family and friends in each kinship caregiver’s social network and assessing the level of support each family member or friend is willing to provide, facilitating and coordinating the support, and identifying backup kinship placements should the current one be disrupted. Kinnected’s success with family search and engagement mirrors that of other research on kinship care and kinship navigator programs [3,20,21,24].

For the data equity study, Hispanic children in the treatment group reunified with their parents at higher rates than Hispanic children in the control group, while Black and White children in each group reunified at similar rates. Given the overlap between kinship care and many other issues facing marginalized communities, this finding is significant and points to an opportunity to further understand what program components or characteristics may particularly benefit caregivers based on culture, family structure, or background. Based on these findings, the following are possible practice implications for counties adopting the Kinnected kinship navigator program:

• Use of bilingual staff: If a family’s native language is not English, having a bilingual staff member who can provide information and support in the family’s native language may provide a deeper understanding of and engagement in kinship navigation services.
• Use of family engagement strategies: If a kinship family is welcomed as a member of the Kinnected team and asked to contribute to the services and supports they receive, this may provide a more positive and engaging experience. This could lead to more placement or support options that may not have become known using prior strategies.
• Use of genograms: Building a visual representation of family members and identified family support individuals may have a positive contribution in identifying and/or building families’ natural support. This could be helpful in the kinship family receiving needed supports without having to ask a system that they may not trust based on previous negative experiences.

It should be noted that there are potential reasons that willing kin are unable to provide kinship placements, which could result in children and youth being placed in traditional foster care. These may include “disqualifying factors” that emerge during the background check process, such as criminal histories, confirmed child abuse, or neglect allegations that would cause a current safety concern, or any other documented information that would cause a safety concern. Thus, not all otherwise eligible families with an open child welfare case will benefit from effective kinship navigation services.

6.2. Next Steps

Additional research is needed to understand the relative influence of each Kinnected program component on child and youth permanency outcomes, as well as differences in the experiences and outcomes of potentially vulnerable groups of kinship caregivers, children, and youth. For the data equity study, conducting interviews or surveys with kinship caregivers could help to identify whether cultural considerations are related to the differential impact of Kinnected on Hispanic children.

The next steps for practice include providing implementation planning and evaluation supports to county child welfare agencies in the roll-out and adoption of Kinnected by non-pilot counties and assisting in the expansion of Kinnected to serve non-child-welfare-involved kinship families, which includes fidelity monitoring activities. Roll-out and adoption of Kinnected by non-pilot counties will include a readiness assessment of counties, training and technical assistance, peer-to-peer learning experiences, and ongoing support for counties in transitioning their practice. Fidelity monitoring will involve collecting and analyzing county assessment data about the implementation of key program components and time frames for assessments and family contacts. Continuous quality improvement efforts also will be provided to support counties in achieving greater fidelity to the Kinnected model.