States’ Use of Medicaid Waivers for Children with Behavioral Health Needs is Associated with Lower Rates of Voluntary Child Welfare Placements

Abstract

Background:

For many children with significant behavioral health needs, the inability to obtain funding for adequate behavioral health care through private means results in parents “voluntarily” placing their child in the state’s child welfare system. This study investigated the association between states’ use of 1915(c) Medicaid waiver policies targeting children and youth with the most complex behavioral health needs and state rates of voluntary foster care placements.

Method:

This study used data from the Children’s Bureau Adoption and Foster Care Analysis and Reporting System (AFCARS), from 2010 to 2019. Multilevel, linear regression models with state-year fixed-effects assessed the relationship between states’ use of a 1915(c) waiver and their total new entry voluntary foster care placement rates for children with behavioral or developmental health care needs, controlling for a variety of state-level policy and demographic characteristics.

Results:

States’ use of 1915(c) waivers is significantly associated with lower total voluntary placement rates with behavioral health needs, both in total (β= −1.51 [−2.82, −0.21]) and proportionally (β=−1.03 [−2.07, 0.02]).

Conclusion:

HCBS Medicaid waivers targeting children with complex behavioral health needs may reduce state rates of voluntary foster care placements.

Approximately 22% of children ages 6 to 17 years old experience mental, emotional or behavioral problems, including ADHD, conduct disorders, anxiety disorders, and mood disorders (Bethell et al., 2022). Within this population, a smaller group of children and youth—5 to 6% of the general child and youth population—demonstrate particularly significant or complex behavioral health needs (Williams et al., 2017). This can encompass a wide range of mental, emotional or behavioral disorders—including developmental disorders and other neurodevelopmental disabilities—that result in significant role functioning impairment in at least two of the following domains: family, peers, educational settings, or the community (Costello et al., 1998).

Caregivers and families of these youth are impacted by their children’s struggles. Parents of children with emotional, developmental, or behavioral problems report poor mental health, aggravation with their children, and poor coping at significantly higher rates than parents of children with no emotional, developmental or behavioral problems (Bethell et al., 2022; Graaf, Hughes, et al., 2024). Caregivers of children with more complex behavioral health care needs—including developmental disabilities—experience significant stress from managing a wide range of behaviors in their children. These include running away, chronic suspensions in school, violence towards others, substance use, self-harming (Herbell & Banks, 2020) and concerns for child and family safety (Brown et al., 2018).

In this article, we explore the association of public funding for intensive, community-based behavioral health care for children with the systemic barriers families face in accessing adequate mental health treatment for their children with complex behavioral health needs. For some families, systemic and financial barriers result in voluntarily placing their children into out-of-home (OOH) foster care settings under the auspices of the child welfare system as means of accessing needed care (Herbell & Graaf, 2023). One strategy for thwarting this practice is to facilitate access to affordable intensive community-based mental health care through 1915(c) Home and Community-Based Services (HCBS) Medicaid waivers, specifically tailored to the needs of children with significant behavioral health needs. In this paper we aim to measure the association between states’ use of this policy and state rates of voluntary placements of children with behavioral health care needs in the foster care system.

Barriers to Child Mental Health Care

Families of children with complex behavioral health care needs often experience substantial barriers to accessing adequate mental health care for their child (Bullock, 2005; Graaf, Kitchens, et al., 2024; Herbell & Banks, 2020; LaBrenz et al., 2020). While outpatient services may be sufficient for children and youth with mild to moderate mental health needs, which can be addressed with therapy and psychotropic medication (Baker-Ericzén et al., 2010), youth with significant needs may need more specialized or “specialty” mental health care. This can include intensive community-based, in-home, inpatient, or residential mental health care (LaBrenz et al., 2020; Olson et al., 2021; Rose et al., 2021). Though children with more severe conditions are almost twice as likely to access mental health services than those with mild or moderate conditions (Merikangas et al., 2011), getting the appropriate intensity and level of treatment or intervention can often prove uniquely challenging for these youth (Herbell & Banks, 2020; LaBrenz et al., 2020).

The most commonly cited barriers to behavioral health care and other family support services include lack of knowledge about where to obtain services, transportation or service location problems, insufficiently/inadequately trained professionals, the cost of services, and insurance-related problems (Bullock, 2005; Cadigan et al., 2018; Graaf et al., 2022; Graaf, Hooley, et al., 2023; Herbell & Banks, 2020; Martin et al., 2018). These barriers compound one another and often compel families to make drastic decisions to obtain support for their children, such as voluntary involvement with the child welfare system (Bullock, 2005; Herbell & Graaf, 2023; Hill, 2017; Semanchin Jones et al., 2018). This process often results in caregivers surrendering their children to child welfare OOH placements settings, commonly referred to as foster care.

Voluntary Placement Agreements

According to current federal statutes, a voluntary foster care placement is an OOH placement of a minor for which the guardians of the minor has requested the assistance of the agency and signed a voluntary placement agreement with that state agency (Social Security Act, 1935). A voluntary placement agreement is a time-limited, written, binding document between the state agency and the minor’s guardians which specifies the rights and obligations of the parents, the child, and the agency—while the child is in the state agency’s care. In theory, such agreements can be revoked by caregivers at any time (Social Security Act, 1935).

This practice has been referred to in a number of ways in research and policy briefs, including voluntary placement (Semanchin Jones et al., 2018), custody relinquishment (Cohen et al., 1993; Stroul, 2019), or custody transfers (Cohen et al., 1991). When this transfer of custody occurs solely for the purpose of accessing mental health services not otherwise available to the family, it is informally referred to a “custody for treatment trade” (Graaf & Snowden, 2020; Herbell & Graaf, 2023; McCarthy, 2023; Stroul, 2019). However, though custody relinquishment can be a type of voluntary placement and may result from a voluntary placement agreement, it is distinct in that it involves parental or caregiver relinquishment of legal and physical custody (Bullock, 2005). In contrast, voluntary placements only require relinquishment of physical custody of a child in most states (Semanchin Jones et al., 2018). Though often used interchangeably, due to nuanced conceptual differences among states’ child welfare policies, we use the term voluntary placement in this paper to refer to theoretically temporary OOH placements structured under voluntary placement agreements with state child welfare agencies, in which physical custody—not legal custody—is transferred to the state. This does not refer to termination of parental rights or custody relinquishment that is planned to be permanent. These agreements all involve OOH placements, and—in some states and jurisdictions—may also require the relinquishment of parents’ legal custody.

Most often, when families turn to the child welfare system for support for their child’s complex behavioral health needs, parents are seeking residential care or intensive Home and Community Based Services (HCBS) due to safety concerns resulting from the child’s or youths’ attempts or threats to harm themselves or others (Graaf & Snowden, 2019; Brown et al., 2018). One reason for pursuing assistance through the child welfare system is that foster care placement, under Title IV-E of the Social Security Act, automatically confers Medicaid coverage (Medicaid and CHIP Payment and Access Commission, 2015) and, thus, financing for specialized mental health care (Graaf & Snowden, 2019). A result of federal mandates for states to fund Early and Periodic Screening, Diagnosis, and Treatment (EPSDT) through Medicaid, many states provide Medicaid reimbursement for Home and Community-based Services (HCBS) services such as youth or parent peer support, respite care, independent living skill development, or care coordination for children with complex health care needs–including mental health needs (Musumeci & Chidambaram, 2019; Pires et al., 2018). As such, public health insurance provides more comprehensive coverage for children’s mental health services than most private insurance plans (Graaf & Palmer, 2022; Musumeci & Chidambaram, 2019). Further, for children needing access to psychiatric residential care, Medicaid coverage pays for the treatment costs of such care in full or in part. However, in some states, Title IV-E funds—only available to children in foster care—are also required to finance the room and board portion of residential care (The Center for Health Care Strategies, 2020).

Prevalence of Voluntary Foster Care Placements

In 2003, the General Accounting Office (GAO) reported that in 2001 over 12,000 children in 19 states were voluntarily placed into foster care when families were desperate to obtain mental health services for their child (United States General Accounting Office, 2003). Though this phenomenon has not been systematically tracked—or estimated nationally since 2003—examination of national foster care data may suggest the rates at which this occurs. Drawing on the Adoption and Foster Care Analysis and Reporting System (AFCARS) data from 2013 from all fifty states and Washington D.C., differences between children voluntarily placed into the foster care system and those entering the system through court orders were examined (Hill, 2017). The analysis found that rates of disability (intellectual and physical) and emotional disturbance were significantly higher among youth voluntarily placed in foster care. Further, rates of placement in group homes, residential institutions, and supervised independent living were significantly higher for voluntarily placed children than for children entering the foster care system through court orders (Hill, 2017).

Another recent study used latent class analysis to identify subgroups of children voluntarily placed into OOH foster care, using AFCARS data from 2019 (Simmel, Zhang, & Graaf, Manuscript Under Review). Results pointed to a latent class of children whose reasons for removal were primarily related to children’s behavioral health problems, developmental disabilities, and caregivers’ inability to cope with the child. Given that caregivers of children and youth with complex behavioral health or developmental needs report poorer coping that caregivers of other children (Graaf, Hughes, et al., 2024), and because developmental disabilities commonly co-occur in child populations with complex behavioral health needs (Graaf, Whitfield, et al., 2023), this group may represent those voluntarily placed in order to access behavioral health care. Though it is impossible to distinguish between children with developmental, behavioral or dual disabilities in this cluster, study results also show that while in OOH care, this subgroup was more likely to have frequent moves within the system and to be placed in more restrictive types of settings, compared to other subgroups of voluntarily placed children. Almost 30% of children identified as voluntarily placed in child welfare in 2019 fell into this classification (Simmel, Zhang, & Graaf, Manuscript Under Review), representing over four thousand children and youth. Other subgroups included children experiencing 1) maltreatment, 2) caregiver death, incarceration, abandonment, relinquishment, or 3) caregiver risk factors, including caregiver substance use, neglect, or housing instability.

State Policies to Prevent Voluntary Placements

The use of voluntary placement agreements to facilitate access to children’s mental health care can have damaging consequences for families. For instance, many parents may experience grief about the removal of their children from their care as well as stigma about undergoing this process (Herbell & Graaf, 2023). Further, in states where legal custody must be relinquished, after a certain period of time, parents give up the right to participate in decision making regarding their child’s well-being, care, and treatment (Herbell & Graaf, 2023). In some cases, parents who voluntarily place their children in state custody may face challenges in regaining parental custody (Semanchin Jones et al., 2018; Stroul, 2019) and may be forced to pay child support to the state (Herbell & Graaf, 2023). Further, this process can cause trauma for the child through family separation and feelings of abandonment, punishment, and rejection (Stroul, 2019).

To reduce this practice and the negative impacts it can have on families, states report using policy manuals and guidelines, along with statewide trainings for child welfare, mental health providers, and parent support workers about alternative paths to needed mental health care (Stroul, 2019). Many states have developed specific diversion protocols in juvenile justice and child welfare settings aimed at delivering alternative, publicly funded mental health focused HCBS to the child and their family (Stroul, 2019). Forty states have voluntary placement statutes and many offer other types of formal guidelines that structure voluntary placements for children in the foster care system. Nine states’ voluntary placement statutes expressly address the needs of children with behavioral health problems, suggesting that some state statutes are specifically crafted for these children (Semanchin Jones et al., 2018). These policies vary significantly in the types of placements they authorized, the timelines surrounding these placements, and the extent to which the justice system become formally involved (Semanchin Jones et al., 2018).

Nearly all states report that the development and expansion of HCBS for children with mental health needs is a key strategy for preventing the use of voluntary placements to access mental health care (Stroul, 2019). Some states achieve this by adding specialty mental health HCBS to their state Medicaid plan under the EPSDT mandate–allowing access to these services for all Medicaid beneficiaries in the state (Graaf & Snowden, 2019). Other states use alternative Medicaid-based mechanisms to finance specialty mental health care for their Medicaid eligible population, including 1115 waivers, 1915(i) and (k) State Plan Amendments (SPAs), Money Follows the Person (MFP), or the Balancing Incentive Program (BIP) (Graaf & Snowden, 2017, 2019, 2021).

However, states offering specialty mental health care through their Medicaid plan keep these services unavailable to families whose incomes disqualify them for Medicaid. Some states address this concern by expanding their rich Medicaid-funded service array to families with higher incomes by offering their full Medicaid plan to higher income families eligible for coverage under the Children’s Health Insurance Plan (CHIP) (Graaf & Snowden, 2019). Several states use the Katie Beckett option—which omits the family income means test and extends standard Medicaid coverage to children with complex health care needs—to extend their broader selection of mental health benefits to children and youth with significant mental health needs in middle income families (Graaf & Snowden, 2017). Other states organize and finance intensive community mental health care for children through state allocations and initiatives, making them available to families regardless of income (Graaf & Snowden, 2019). These efforts are often funded by federal System of Care grants, the allocation of state general funds for HCBS behavioral health care, and cross system strategies that braid funding for behavioral health care from multiple state child-serving systems (Graaf & Snowden, 2019).

1915(c) Medicaid Waivers

Over the past two decades, in an effort to reduce voluntary placements and the excessive health care costs of serving children in potentially avoidable OOH settings, several states have adopted 1915(c) HCBS Medicaid waivers for children with significant mental health needs (Friesen et al., 2003; Graaf & Snowden, 2017). 1915(c) Medicaid waivers are federally authorized state Medicaid policies aimed at supporting individuals—including children—with complex medical and behavioral health needs in their homes and communities. Participation in such waiver programs grants participants access to the customary Medicaid benefits, as well as the additional specialized home and community-based services authorized under the waiver (Harrington et al., 2012). By determining Medicaid eligibility based on youths’ demonstration of significant behavioral health need (rather than family income) and expanding the array of Medicaid-financed HCBS offered to waiver participants, such policies facilitate access to behavioral health HCBS for children with the most significant mental health needs, theoretically diverting them from OOH care.

1915(c) HCBS waivers have been used extensively by states to increase home and community-based services for children with developmental disabilities, but far less often for children with serious behavioral health treatment needs (Keim-Malpass et al., 2023; Musumeci et al., 2020). However, since the dissemination of the 2003 GAO Report in which greater attention was paid to the voluntary placement process, states’ use of these policies to serve children with complex behavioral health needs has accelerated. A 5-year federal demonstration project, conducted from 2007 to 2012, evaluated the use of 1915(c) HCBS waivers to reduce OOH in Psychiatric Residential Treatment Facilities (PRTFs). It concluded that waiver-funded HCBS could be key in reducing risk for OOH care and associated public costs—particularly for children and youth with the most significant developmental and behavioral health care needs (Urdapilleta et al., 2013). In response to these findings, the Centers for Medicare and Medicaid Services (CMS) issued an informational bulletin formally encouraging states to adopt 1915(c) waivers (Mann & Hyde, 2013).

As of 2024, eleven states use 1915(c) HCBS Medicaid waivers to facilitate access to a broader array of HCBS for youth with complex behavioral health needs, and several additional states participated in the federal waiver demonstration program between 2007 and 2012 (Graaf & Snowden, 2017). In studies comparing other state strategies for funding HCBS for children and youth with the use of these waivers, researchers found states’ waiver use to be associated with higher odds of children and youth having both public and private health coverage (suggesting they are reaching the non-Medicaid eligible population), reduced rates of unmet mental health need in publicly insured children with significant mental health needs, and higher penetration rates in their public mental health systems (Graaf et al., 2021; Graaf & Snowden, 2018). Though this and other limited research suggests that these waivers may be helping to expand access to mental health care in community-based settings and divert children from OOH care generally (inclusive of justice, psychiatric and foster care settings) (Frimpong et al., 2018; Graaf et al., 2021; Graaf & Snowden, 2018; Urdapilleta et al., 2013), the impact of these policies on rates of voluntary placements into OOH care is unknown. To address this gap, this exploratory study draws on ten years of data from the Adoption and Foster Care Analysis and Reporting System (AFCARS) to determine if states’ adoption of 1915(c) HCBS Medicaid waivers for this population is associated with reductions in state voluntary placements among children with behavioral health needs. We hypothesize that states’ use of 1915(c) HCBS Medicaid waivers for children and youth with mental health needs is associated with lower rates of voluntary placements of children with behavioral health needs into state foster care systems.

Method

This project drew on ten years of data from the Children’s Bureau Adoption and Foster Care Analysis and Reporting System (AFCARS) for youth from foster care, from 2010 to 2019. These data were combined with state level demographic characteristics drawn from the U.S. Census American Community Survey (ACS), state health and mental health policy data drawn from Medicaid.gov’s State Waiver List, Kaiser Family Foundation Medicaid reports, and the Substance Abuse and Mental Health Services Administration’s Unified Reporting System (URS) and the National Mental Health Services Survey. Due to significant administrative changes in the reporting and collection of AFCARS data prior to 2010 (Green et al., 2015), the research team excluded AFCARS data prior to 2010.

Data Sources

Adoption and Foster Care Analysis and Reporting System (AFCARS)

The outcome variables for this study—annual state rates of voluntary placements in OOH care—were derived from the AFCARS dataset. AFCARS includes individual-level information on all foster youth in state custody through child protective services for all fifty states and Washington D.C. States and tribal Title IV-E agencies collect the data and report it semi-annually to the Children’s Bureau. The data are used to construct the AFCARS database, which is used to assist in policy development and program management at the federal, state, and Tribal level. Use of the AFCARS by researchers is managed through the National Data Archive on Child Abuse and Neglect (NDACAN), and is funded by the Children’s Bureau, Administration on Children, Youth, and Families, and the U.S. Department of Health Human Services. The dataset includes demographic information on the foster child, the foster and adoptive parents, as well as information about removal and placement experiences (National Data Archive on Child Abuse and Neglect, 2013).

State Waiver Lists

The study predictor variable—states’ use of 1915(c) Medicaid waivers for children with mental health needs—was drawn from the State Waiver List at Medicaid.gov (Medicaid.gov, 2023). This website lists and describes every Medicaid waiver for all fifty states and provides links to the most recent waiver applications filed by the state with the Centers for Medicaid and Medicare Services (CMS). Which states used 1915(c) waivers, whether through the federal PRTF demonstration or apart from it, and in which years the policy was adopted or de-adopted, is listed in Tables 1 and 2.

Table 1.

Count of State 1915(c) HCBS Medicaid Waivers for Children with Behavioral Health Needs, by Type and Year

Year	Regular Waivers
2010–19	Texas, Louisiana, Michigan, Wyoming, Kansas, New York, Wisconsin
2011–2019	Iowa
	PRTF Waivers*
2010–2012	Alaska, Kansas,
2010–2014	Georgia, Maryland, Mississippi
2010–2015	Montana
2010–2017	Indiana, South Carolina, Virginia

^*PRTF Waivers were policies offered only during the five-year federal PRTF demonstration program from 2007–2012. Some participating states “phased out” these demonstration programs by continuing the program beyond the federal demonstration for three to five years. Regular Waivers are programs offered by states outside of the federal demonstration. Kansas participated in the PRTF Demonstration while simultaneously offering a regular waiver program.

Table 2.

State Waiver Count, by Type and Year

YEAR	Regular Waivers	PRTF Waivers	Any 1915(c) Waivers
2010	7	9	15
2011	8	9	16
2012	8	9	16
2013	8	7	15
2014	8	7	15
2015	8	4	12
2016	8	3	11
2017	8	3	11
2018	8	0	8
2019	8	0	8
Total	79	51	127

Kaiser Family Foundation Kaiser Commission on Medicaid and the Uninsured

State-level control variables, annual state Medicaid expenditures for children and Medicaid eligibility income limits for children 6 to 18 years of age, was drawn from annual reports from the Kaiser Commission on Medicaid and the Uninsured’ s Annual 50-State Survey of Eligibility Rules, Enrollment and Renewal Procedures, And Cost Sharing Practices In Medicaid And CHIP (Annual Updates on Eligibility Rules, Enrollment and Renewal Procedures, and Cost-Sharing Practices in Medicaid and CHIP, 2022). The survey is conducted by the Kaiser Commission on Medicaid and the Uninsured in partnership with the Georgetown University Center for Children and Families and the Center on Budget and Policy Priorities. Data are collected through in-depth telephone interviews with state Medicaid and CHIP officials and verified through follow-up communications via email and phone (Brooks et al., 2019).

Substance Abuse and Mental Health Services Administration (SAMHSA)

Additional state level controls, total state mental health authority (SMHA) expenditures and total child mental health providers per state, were drawn from SAMHSA’s Unified Reporting System (URS) and the National Mental Health Services Survey (N-MHSS). The URS provides uniform, comparable, high-quality statistics on mental health services for each state. The URS was developed to enable the use of data in decision support and planning for public mental health systems and to promote program accountability and has been used by State Mental Health Agencies (SMHAs) to aggregate and report annual data. The dataset reports the sociodemographic characteristics of people served, outcomes from services, state’s use of specific evidence-based practices, consumer assessments of care, and state expenditures and revenues.

The Substance Abuse and Mental Health Services Administration also administers the National Mental Health Services Survey (N-MHSS, 2010–2019) and makes its data available through the Substance Abuse and Mental Health Data Archive. The N-MHSS is a point-prevalence survey, providing a “snapshot” of information about the U.S. mental health treatment system and the consumers it serves on a pre-selected reference date of April 30 of each year. Data is collected annually from all known public and private mental health facilities in the US. The dataset includes information about the location, characteristics, and use of formal mental health service providers throughout all fifty states, the District of Columbia, and the U.S. territories, but excludes the following facilities: military treatment facilities, Indian Health Service-administered or tribally operated facilities, private practitioners or small group practices not licensed as a mental health clinic or center, and jails or prisons.

U.S. Census American Community Survey (ACS)

State level demographic characteristics were drawn from the ACS. The ACS is an ongoing survey conducted by the U.S. Census Bureau that collects annual information at the household level on wide range of demographic and economic information that affects individuals and families, including employment, education, housing status, and health care coverage. This data is collected for use in public budgeting and policy making at federal, state, and local levels.

Study Variables

Predictor Variable: 1915c Medicaid Waivers

Drawing on waiver program descriptions listed on Medicaid.gov’s State Waiver List, and the state waiver applications available through this website, a binary variable representing a state’s use of a 1915(c) waiver for children with mental health needs was constructed for each state for each year. States using 1915(c) waivers for this population were coded with a “1” for each year they served children under this policy. This included states participating in the federal PRTF Demonstration as well as states customarily offering these programs outside of the demonstration. Because the first three years of this study included the three last years of the PRTF Demonstration, states’ participation in the demonstration was verified through the final Report to the President and Congress for the Medicaid Home and Community-Based Alternatives to Psychiatric Residential Treatment Facilities Demonstration (Urdapilleta et al., 2013). Continuation of 1915(c) waiver programs after the demonstration ended was verified through examination of waiver applications.

Response Variable: State Rates of New Entry Voluntary Placements for Children or Youth with Significant Behavioral Health Need

State voluntary placement rates for children with significant behavioral health needs each year were drawn from the AFCARS data, which reports each child’s Manner of Removal and Reason for Removal variable. Each case in the AFCARS is coded for both variables. The Manner of Removal is a mutually exclusive variable that specifies how the child was removed into foster care, with the three following categories: voluntary, court ordered, or not yet determined. The Reason for Removal variable is a non-mutually exclusive variable that specifies the reason that a child entered foster care and can include: physical abuse, sexual abuse, neglect, alcohol abuse by a parent, drug abuse by a parent, alcohol abuse by the child, drug abuse by the child, a child’s disability, a child’s behavioral problem, parental death, parental incarceration, parental inability to cope, parental abandonment, parental relinquishment, or inadequate housing.

Drawing on the results of a latent class analysis (Simmel, Zhang, & Graaf, Manuscript Under Review), which identified Reasons for Removal likely to be indicative of voluntary placement due to need for behavioral health care, the following Reasons for Removal were utilized to identify voluntarily placed children with significant behavioral health needs: alcohol abuse by the child, drug abuse by the child, a child’s disability, a child’s behavioral problem, and parental inability to cope. The Voluntary Placements with Significant Behavioral Health Need variable was constructed from counts of children with voluntary manner of removal codes and one or more of the Reasons for Removal listed above. Totals for new entry children meeting these criteria in each state were calculated for each year (Annual Total New Entry Voluntary Placements with Significant Behavioral Health Need). These totals were used to generate a proportional measure (Annual Rate of New Entry Voluntary Placements with Significant Behavioral Health Need (Proportional Measure))—representing the proportion of children entering each state system for each year who were voluntarily placed with one or more of the behavioral health-related Reasons for Removal. Proportions were calculated by dividing the total new entry Voluntary Placements with Significant Behavioral Health Need divided by the total number of new entry children served in the child welfare system in each state (court-ordered, voluntarily placed, or other) for each year. Rate of missingness for the variables used to construct the outcome variable (manner of removal and reason for removal) was approximately 4 to 5% for each year. Observations missing values for these variables were excluded from state totals.

Control Variables: State-level demographic and policy characteristics

Study estimates of the association between 1915(c) Medicaid waivers and voluntary placement rates for children with behavioral health needs controlled for a variety of time-varying state characteristics that might confound voluntary placement rates – due to family, environmental, or mental health system risk factors. Family or environmental risk factors, drawn for the years 2010 to 2019 from the ACS single year estimates, included median income (thousands), total state child population (million), state proportions of African American (both Hispanic and non-Hispanic), White (both Hispanic and non-Hispanic), and Hispanic (any race) child populations, proportions of families using public assistance, child poverty rates, family education rates, unemployment rates, and rates of single-parent households. These factors have been demonstrated to contribute to child maltreatment rates (Kim & Drake, 2017; Kovski et al., 2021) as well as rates of child behavioral health needs and family stress (Currie, 2024; Yoshikawa et al., 2012).

State policy variables that may impact access to behavioral health care, aside from states’ use of HCBS Medicaid waivers, include public mental health spending (Le Cook et al., 2013), Medicaid generosity (Snowden et al., 2016, 2017; Thomas et al., 2012), and the size and location of behavioral health provider networks (Cook et al., 2013). To control for the potentially confounding effects of these state-level factors, models included total state child Medicaid expenditures and Medicaid eligibility limits for children ages 6 to 18, expressed as a percentage of the Federal Poverty Line (FPL). These data were drawn from Kaiser Commission on Medicaid and the Uninsured Reports, for the years 2010–2019.

The total number of mental health providers in a state were estimated from the N-MHSS, using data from 2010, 2012 and from 2014–2019. Estimates of total child mental health providers from 2010 were applied for the years 2010 and 2011, and data for 2012 was applied for 2012 and 2013 due to gaps in survey administration and data collection. Total state investment in mental health services was controlled for by including total SMHA expenditures for each state in each year. Due to gaps in URS data collection and reporting, 2015 expenditure values are applied to both 2015 and 2016.

Study Population Preparation

To create the dataset for this analysis, all 10 years of AFCARS voluntary placement totals and proportions for each state were combined into a single dataset, with one observation for each state for each year. The AFCARS estimates were then merged with state-level control variables for 2010 to 2019. The total sample size, representing fifty states plus Washington D.C. observed from 2010 to 2019, was 510 state-year observations. To prepare the data for multilevel modeling, the “xtset state year” command was used to organize the data by state and over year. Sample preparation and analysis was completed in StataMP 16.0.

Analysis

Univariate descriptive analysis identified average state rates of voluntary placements, overall and for children with mental health needs, proportions of voluntary placements overall and for children with mental health needs, and control variable means and standard deviations. Because some states reported no voluntary placements for any reason for removal and some states reported hundreds of voluntary placements with a wide range of reasons for removal, the natural log of all outcome variables was calculated for each outcome variable, adding a very small constant (0.0001) to manage variable heteroskedasticity (Hammouri et al., 2020; Mullahy & Norton, 2022). Log transformed variables were used as outcomes in further bivariate and multivariable analysis.

Bivariate analysis used cross tabulation with chi-square tests to assess for significance of differences between means for each variable by state 1915c waiver policy use. To test the hypothesis that states’ use of 1915(c) HCBS Medicaid waivers for children and youth with mental health needs is associated with lower rates of voluntary placements of children with behavioral health needs into state foster care system, two multivariable, multilevel linear regression models were used, nesting annual observations within states. Regression models for the proportional variable were adjusted for all the state control variables listed above. Regression models for the total new entries (per child in each state) variable were adjusted for all the state control variables listed above. The inclusion of the total child population in each state in each model controls for the number of children within a given state, which directly impacts the total number of children in foster care as well as in voluntary placements; this control variable also accounts for the ratio of total state spending (Medicaid and SMHA) and mental health providers per child. Models included state and time fixed effects and robust standards errors to further manage outcome variable heteroskedasticity.

Results

Changes in all voluntary placements and voluntary placements with behavioral health needs from 2010 to 2019 are reported in Figure 1. Voluntary placements, both with and without behavioral health needs, generally decreased over the years of this study in all states. A slight upward trend starting in 2018 is observed in all rates for non-waiver states. However, in states with waivers, voluntary placements with behavioral health needs did not increase in 2018. State mean voluntary placement rates for all ten years, expressed as totals per child living in the state and as proportions of all foster care placements, are reported in Table 3. We report rates for all voluntary placements and voluntary placements with behavioral health needs. Rhode Island has the highest rate of voluntary placements (1177.39 per 1 child in the state; 20.03 % of total foster care placements), while Minnesota has the highest rate of voluntary placements with behavioral health needs (603.22 per child in the state; 12.23% of total foster care placements). Arkansas, Delaware, and Missouri report no voluntary placements at all, and Illinois and North Dakota report no voluntary placements with behavioral health needs.

Figure 1.

Voluntary Placements Over Time: Waiver vs. Non-Waiver States

Table 3.

Annual State Rate of Voluntary Placements, Totals and Proportions, by State, 2010–2019 (Raw Values)

	Annual State Rate of Voluntary Placements - All Types (per child in the state)				Annual State Rate of Voluntary Placements - All Types: Proportional Measure				Annual State Rate of Voluntary Placements with Behavioral Health Needs (per child in the state)				Annual State Rate of Voluntary Placements with Behavioral Health Needs: Proportional Measure
	Mean	SE	95% CI		Mean (%)	SE	95% CI		Mean	SE	95% CI		Mean (%)	SE	95% CI
AL	73.99	10.07	54.20	93.78	2.51	0.42	1.69	3.33	29.74	4.08	21.72	37.76	1.00	0.16	0.67	1.32
AK	127.26	15.03	97.74	156.78	2.07	0.28	1.52	2.62	34.44	5.49	23.67	45.22	0.56	0.09	0.39	0.73
AZ	195.94	8.50	179.25	212.64	3.16	0.24	2.70	3.62	46.54	13.00	20.99	72.08	0.81	0.26	0.30	1.32
AR	(omitted)				(omitted)				(omitted)				(omitted)
CA	78.96	10.86	57.63	100.28	2.27	0.28	1.71	2.83	30.23	5.44	19.54	40.92	0.86	0.15	0.58	1.15
CO	175.29	13.82	148.13	202.44	4.32	0.27	3.78	4.85	135.11	13.30	108.98	161.24	3.32	0.28	2.77	3.87
CT	88.40	8.85	71.02	105.79	3.49	0.43	2.64	4.33	67.59	10.08	47.79	87.40	2.67	0.44	1.81	3.53
DE	(omitted)				(omitted)				(omitted)				(omitted)
DC	0.92	0.92	−0.89	2.72	0.02	0.02	−0.02	0.06	0.92	0.92	−0.89	2.72	0.02	0.02	−0.02	0.06
FL	30.66	16.89	−2.52	63.85	0.81	0.46	−0.09	1.71	3.49	0.72	2.08	4.89	0.09	0.02	0.05	0.13
GA	31.29	3.29	24.82	37.76	1.15	0.17	0.82	1.48	21.90	1.98	18.00	25.80	0.79	0.09	0.62	0.97
HI	674.59	92.61	492.64	856.55	18.89	2.94	13.11	24.67	127.56	23.74	80.91	174.21	3.60	0.73	2.16	5.03
ID	33.98	16.68	1.22	66.74	1.15	0.55	0.07	2.23	29.51	14.73	0.57	58.45	0.99	0.48	0.05	1.93
IL	0.19	0.14	−0.08	0.46	0.01	0.01	0.00	0.03	(omitted)				(omitted)
IN	23.36	4.81	13.90	32.82	0.36	0.07	0.22	0.49	4.08	1.04	2.03	6.13	0.06	0.02	0.03	0.09
IA	643.53	261.74	129.30	1157.76	10.49	4.06	2.52	18.47	306.63	112.45	85.71	527.55	5.02	1.74	1.60	8.44
KS	0.28	0.19	−0.09	0.65	0.01	0.00	0.00	0.01	0.14	0.14	−0.13	0.41	0.00	0.00	0.00	0.01
KY	33.56	9.88	14.15	52.98	0.54	0.13	0.28	0.79	19.05	5.21	8.81	29.30	0.31	0.07	0.17	0.45
LA	7.21	0.86	5.52	8.90	0.22	0.02	0.18	0.27	2.88	0.64	1.62	4.13	0.09	0.02	0.05	0.12
ME	46.36	8.48	29.70	63.02	1.50	0.38	0.75	2.25	20.92	5.26	10.58	31.26	0.70	0.22	0.27	1.12
MD	97.07	10.55	76.33	117.80	5.44	0.50	4.47	6.42	91.33	10.14	71.41	111.25	5.12	0.48	4.18	6.06
MA	275.13	32.25	211.77	338.49	6.62	0.89	4.88	8.37	183.72	23.56	137.43	230.01	4.43	0.64	3.17	5.69
MI	16.57	7.39	2.05	31.10	0.53	0.24	0.05	1.01	2.58	1.48	−0.32	5.49	0.08	0.04	−0.01	0.17
MN	787.17	30.32	727.60	846.74	15.94	0.96	14.06	17.81	603.22	25.52	553.07	653.37	12.23	0.78	10.69	13.76
MS	10.21	1.54	7.18	13.23	0.27	0.04	0.18	0.35	5.22	1.59	2.09	8.34	0.15	0.05	0.05	0.24
MO	(omitted)				(omitted)				(omitted)				(omitted)
MT	679.31	103.14	476.68	881.94	8.75	0.82	7.13	10.37	47.33	15.54	16.80	77.86	0.87	0.36	0.16	1.59
NE	242.69	56.04	132.60	352.79	4.05	0.72	2.65	5.46	85.34	23.16	39.83	130.85	1.40	0.31	0.80	2.01
NV	51.00	3.58	43.97	58.03	1.08	0.09	0.91	1.25	14.19	1.41	11.42	16.95	0.30	0.04	0.23	0.37
NH	7.72	2.47	2.87	12.57	0.35	0.13	0.09	0.61	4.51	1.48	1.60	7.42	0.17	0.05	0.07	0.28
NJ	28.68	3.64	21.52	35.84	1.28	0.12	1.05	1.51	19.59	3.06	13.58	25.61	0.86	0.11	0.65	1.08
NM	5.01	0.75	3.53	6.49	0.14	0.02	0.10	0.18	1.21	0.44	0.33	2.08	0.03	0.01	0.01	0.06
NY	81.82	8.44	65.24	98.41	3.67	0.44	2.79	4.54	67.49	6.89	53.96	81.03	3.02	0.36	2.32	3.73
NC	83.43	14.64	54.66	112.19	3.57	0.66	2.28	4.86	43.95	8.26	27.72	60.18	1.89	0.37	1.15	2.62
ND	3.00	1.59	−0.13	6.13	0.05	0.03	0.00	0.10	(omitted)				(omitted)
OH	267.71	24.63	219.32	316.11	6.68	0.61	5.48	7.87	194.93	23.36	149.03	240.82	4.81	0.50	3.84	5.79
OK	15.32	3.85	7.75	22.88	0.29	0.08	0.13	0.45	8.51	2.19	4.21	12.81	0.16	0.04	0.07	0.25
OR	156.93	23.28	111.19	202.67	3.36	0.41	2.56	4.16	139.96	22.02	96.69	183.23	2.99	0.38	2.23	3.74
PA	395.16	65.77	265.95	524.37	9.53	1.48	6.63	12.43	248.91	41.64	167.09	330.72	5.97	0.88	4.24	7.70
RI	1177.39	101.04	978.88	1375.91	20.03	1.91	16.28	23.77	511.15	45.87	421.03	601.26	8.67	0.80	7.09	10.25
SC	33.01	4.07	25.01	41.01	1.07	0.14	0.79	1.35	15.69	2.72	10.35	21.04	0.51	0.09	0.33	0.69
SD	82.87	12.01	59.27	106.47	1.43	0.15	1.13	1.74	25.74	4.85	16.22	35.26	0.44	0.07	0.30	0.58
TN	219.71	33.22	154.43	284.98	5.21	0.81	3.63	6.80	200.36	31.94	137.61	263.11	4.75	0.77	3.24	6.27
TX	0.07	0.03	0.01	0.13	0.00	0.00	0.00	0.01	0.03	0.02	−0.01	0.07	0.00	0.00	0.00	0.00
UT	5.78	0.68	4.45	7.11	0.24	0.03	0.18	0.29	4.56	0.88	2.83	6.30	0.19	0.04	0.12	0.26
VT	124.47	17.57	89.95	159.00	2.18	0.41	1.38	2.99	71.00	12.44	46.55	95.45	1.28	0.29	0.71	1.85
VA	69.20	6.85	55.73	82.67	4.65	0.51	3.65	5.66	43.39	5.34	32.91	53.87	2.93	0.39	2.16	3.69
WA	325.80	46.54	234.37	417.24	9.09	1.18	6.78	11.40	90.39	17.23	56.54	124.25	2.50	0.42	1.68	3.32
WV	126.94	12.11	103.15	150.73	1.25	0.17	0.93	1.58	118.21	10.88	96.83	139.58	1.17	0.16	0.86	1.48
WI	100.92	10.10	81.09	120.76	2.82	0.32	2.19	3.44	59.28	8.81	41.98	76.58	1.66	0.26	1.14	2.18
WY	146.19	16.41	113.95	178.43	1.97	0.24	1.50	2.44	71.66	12.52	47.07	96.25	0.97	0.17	0.63	1.31

We hypothesized that states’ use of 1915(c) HCBS Medicaid waivers for children and youth with mental health needs is associated with lower rates of voluntary placements of children with behavioral health needs into state foster care systems. The hypothesis was initially supported by the stratified bivariate analysis, comparing outcomes and controls for waiver versus non-waiver states. Table 4 reports means for each variable across all states and compares values across states with and without 1915(c) waivers. Approximately 25% state-years had 1915(c) waivers. The total number of new entry voluntary placements with significant behavioral health needs varied widely across states, with a mean of 93.06 (SD=181.67). State-years with no waiver program had a significantly (p<.03) higher mean number of voluntary placements with behavioral health needs (m=103.01, SD=197.70) than state-years with waiver programs (m=63.06, SD=116.71). However, when the proportion of the total foster care population was considered, though voluntary placements with behavioral health needs were a smaller proportion of new entries in state-years with waiver programs (m=1.45, SD=2.37) than non-waiver state-years (m=1.88, SD=2.84), the difference was not significant (p<0.12).

Table 4.

State Characteristics by State 1915c Waiver Status, 2010–2019

	All States and District of Columbia		States with No 1915(c) Waiver for Children with Behavioral Health Needs(n=383 state years)		States with a 1915(c) Waiver for Children with Behavioral Health Needs (n=127 state years)
	Mean	SD	Mean	SD	Mean	SD
Annual State Voluntary Placements with Behavioral Health Needs
Total New Entries	93.06	181.67	103.01	197.70	63.06*	116.71
Proportion of Total New Entries	1.77	2.73	1.88	2.84	1.45	2.37
State Context Variables
Total Child Mental Health Providers	137.90	107.88	128.78	101.39	165.40**	121.79
Total Child Medicaid Spending (millions)	$4109.56	$4678.782	$ 3,644.45	$ 5,512.21	$ 6,489.17**	$ 695.40
Total SMHA Expenditures (millions)	$224.26	$129.83	$ 219.73	$ 133.67	$ 237.93	$ 116.90
Medicaid Child Eligibility, Ages 6–18 (% FPL)	146.60	50.54	151.04	53.36	133.23**	37.99
Median Income (thousands)	$ 66.55	$ 13.60	$ 67.46	$ 14.38	$ 63.81*	$ 10.50
Total child population (millions)	1.44	1.67	1.30	1.57	1.85**	1.86
% Black/African American children	12.91	12.38	11.81	12.12	16.23**	12.63
% White children	70.15	14.82	70.12	15.59	70.23**	12.24
% Hispanic/Latino children	16.58	12.98	17.27	13.42	14.50*	11.34
% Children in families using public assistance	26.09	6.10	26.14	6.07	25.94	6.21
% Child poverty	19.25	5.27	18.98	5.19	20.05*	5.44
% High school graduate or higher	88.67	3.09	88.80	2.92	88.28	3.55
% Bachelor’s degree or higher	30.21	6.31	30.74	6.74	28.63**	4.44
% Unemployment	6.81	2.48	6.73	2.47	7.02	2.49
% Children in single parent households	33.87	5.66	33.73	5.86	34.31	5.00

FPL: Federal Poverty Line;

^*difference between state groups is significant at p<0.05;

^**difference is significant at p<0.01.

Our hypothesis was further supported in multivariate, multilevel analysis results. Coefficients and confidence intervals representing the adjusted association between states’ use of 1915(c) waivers and state voluntary placement rates are reported in Table 5. Though not significant in bivariate analysis, when adjusting for covariates, states’ use of a 1915(c) waiver is significantly (p<0.02) associated with reductions in total new annual new entry voluntary placements with behavioral health needs (β= −1.51 [−2.82, −0.21]), and with reductions in the proportion of all new entries (β=−1.03 [−2.07, .02]) that are voluntarily placed with behavioral health needs (p<0.05).

Table 5.

Associations between States’ Use of 1915(c) Medicaid Waivers for Children with Mental Health Needs and Annual State Rates of Voluntary Placements with Significant Mental Health Needs

	Total Annual Voluntary Placements with Significant Behavioral Health Need				Annual Voluntary Placements with Significant Behavioral Health Need: Proportion of all Placements
	β	p	95% CI		β	p	95% CI
1915c HCBS Waiver for Children with SED	−1.51	0.02	−2.82	−0.21	−1.03	0.05	−2.07	0.02
Total Child Mental Health Providers	0.01	0.06	0.00	0.02	0.01	0.07	0.00	0.01
Total Child Medicaid Spending (millions)	0.00	0.56	0.00	0.00	0.00	0.27	0.00	0.00
Total SMHA Expenditures	0.00	0.09	0.00	0.00	0.00	0.06	0.00	0.00
Medicaid Child Eligibility FPL 6–18	0.00	0.80	−0.01	0.01	0.00	0.75	−0.01	0.00
Median Income (thousands)	−0.03	0.56	−0.12	0.07	−0.02	0.63	−0.09	0.05
Total child population (millions)	−7.11	0.01	−12.57	−1.64	−5.32	0.00	−8.89	−1.74
% Black/African American children	0.14	0.31	−0.13	0.41	0.11	0.28	−0.10	0.33
% White children	0.08	0.54	−0.17	0.33	0.06	0.54	−0.14	0.26
% Hispanic/Latino children	0.20	0.30	−0.18	0.59	0.21	0.18	−0.10	0.52
% Children in families using public assistance	0.07	0.50	−0.13	0.26	0.05	0.48	−0.10	0.20
% Child poverty	−0.10	0.51	−0.40	0.20	−0.10	0.40	−0.35	0.14
% High school graduate or higher	0.44	0.13	−0.13	1.01	0.42	0.06	−0.03	0.86
% Bachelor’s degree or higher	0.00	0.99	−0.80	0.81	−0.02	0.96	−0.68	0.65
% Unemployment	−0.09	0.54	−0.40	0.22	−0.04	0.76	−0.27	0.20
% Children in single parent households	0.14	0.27	−0.11	0.39	0.13	0.21	−0.07	0.32
Year	−0.39	0.01	−0.69	−0.08	−0.35	0.01	−0.60	−0.11

^*All models regress on the natural log of each outcome; FPL: Federal Poverty Line;

Discussion

This exploratory study sought to examine the extent to which states’ adoption of 1915(c) HCBS Medicaid waivers is linked with a reduction of families’ use of voluntary placement agreements to place for children with complex behavioral health needs in states’ child welfare systems. The results of this study demonstrate that states’ adoption and use of 1915(c) HCBS Medicaid waivers is associated with significantly lower state total annual voluntary placements for children with complex behavioral health challenges as well as the proportion of children entering OOH care settings through such placements. Though more rigorous assessments are needed, this suggests that these programs may be an effective means of reducing families’ use of voluntary placement agreements to obtain services, treatment, and interventions for their children with complex behavioral health needs. This finding may expand the small body of findings that identifies specific policy mechanisms that expand access to behavioral health care for children and youth (So et al., 2019). This study also contributes new findings to an even smaller body of knowledge about factors that can expand access specifically for families whose children have the most significant mental health needs.

Existing research has demonstrated that waivers are associated with reductions in addressing unmet mental health need for children with emotional and behavioral problems more broadly (Graaf et al., 2021) and that waiver enrollment reduces total Medicaid spending on mental health care for children enrolled in the program (Frimpong et al., 2018). This suggests that waivers’ association with reductions in behavioral health-related voluntary placements may be due to expanded access to publicly funded, less costly HCBS for behavioral health needs in waiver states, which provides routes for families to obtain treatment and intervention for their children outside of the foster care system. Alternatively, because states adopt waivers to reduce preventable OOH care and reduce costs for residential care (Graaf & Snowden, 2020), these states may also mandate inter-organizational and gatekeeping processes that work to divert children more effectively from voluntary placements. These processes may include enhanced communication across multi-sector provider organizations, the use of mobile crisis response teams, or swift and comprehensive post-discharge care upon ED or inpatient psychiatric visits. However, more research is needed to understand how waiver programs are implemented, how families are impacted by them, the processes for diversion from OOH child welfare settings, and the systemic differences between waiver-funded and non-waiver funded behavioral health programs.

Implications for Policy and Research

Though these findings are promising, they are exploratory and more must be learned about these families’ experiences with the behavioral health and child welfare systems to enable more robust assessments of policies that can reduce this practice. Much remains to be understood about how families enter the decision to surrender their children to OOH care through voluntary placement agreements. It is important to understand these processes and experiences—from the caregivers,’ mental health providers,’ and child welfare workers’ perspectives—to identify the practice and policy levers with which to prevent such arrangements. Of particular importance is understanding how the option is first presented to families and by whom, as well as what is explained to them prior to their agreement. Limited research suggests that families were not fully aware of their rights or the legal repercussions of such actions (Herbell & Graaf, 2023). Additionally, understanding the legal and systemic processes that take place when a child enters and exists such a voluntary placement, how these vary across states, and what the impact of these variations are for youth and families are critical for understanding best practices in such arrangements.

Understanding sources of state differences may also enable researchers to identify children in the foster care system who are placed exclusively for behavioral health treatment more precisely. For example, the number of voluntarily placed children reported varies widely across states: some states report none and other states report hundreds per year. One explanation for this may be that five states (Iowa, Louisana, New Jersey, Virginia, and West Virginia) require court involvement in the execution of voluntary placement agreements (Semanchin Jones et al., 2018). These states may have alternative procedures for categorizing children that are in foster care via voluntary placement agreements—noting them as court-ordered placements rather than voluntary placements. Further analysis is required to understand if states requiring court involvement for voluntary placements are the same states that also report no voluntary placement in AFCARS.

State Policy Choices

It appears that state adoption of 1915(c) HCBS Medicaid waivers for children with behavioral health needs are associated with reductions in the rate at which children with behavioral health care needs are experiencing voluntary placements in state child welfare systems. However, given the potential success of these policies in diverting children from foster care, it is surprising that less than a quarter of states offer them. States’ motivation to adopt these policies rest on priorities to reduce costs for OOH care, but administrators are often hindered by ideological misgivings about Medicaid expansions in general or extending safety net programs to non-indigent families (Graaf & Snowden, 2020). Other states hesitate to adopt these programs due to concerns about extending public health coverage to high needs and high costs populations—that the costs incurred through these programs would be unsustainable for state budgets (Graaf & Snowden, 2020).

For states choosing not to use 1915(c) HCBS waivers, which are likely due to the reasons cited above, other formal policies were described above that may be helping to reduce voluntary placements expressly for the purpose of accessing mental health care. These include voluntary placement statutes, state funded HCBS programs, other types of Medicaid HCBS programs (1115 waivers, MFP, BIPs or SPAs), 1915(c) Medicaid programs for other populations, or states expanding Medicaid coverage through the Katie Beckett option or CHIP. Further, states use 1915(c) HCBS Medicaid waivers to finance and organize HCBS for a wide range of other child populations (Keim-Malpass et al., 2019) and coverage under waivers for children with mental health needs is relatively limited compared to many of these other programs (Keim-Malpass et al., 2023). States’ use of these types of programs may also help to reduce the voluntary placement of children with disabilities or complex medical conditions into OOH, many of whom also have co-occurring behavioral health conditions. However, these waivers often have extremely long waitlists compared to waivers for children with SED (Musumeci et al., 2019), partly because many of them serve both adults and children (Keim-Malpass et al., 2019; Velott et al., 2016). In contrast, all SED waivers only serve youth under 21 years of age or less, and frequent churn in SED waiver enrollment has been documented (Graaf, Whitfield, et al., 2023)—suggesting that SED waivers may have more ongoing capacity to serve more children than waivers serving other youth populations.

The structures of these other Medicaid and state-funded programs, guidelines, and statutes—and the ways in which behavioral health and child welfare providers implement them—may differentially impact the rates at which families agree to voluntary placements to access behavioral health care. Assessments of the impacts of these policies and how they compare to states’ use of 1915(c) waivers, have yet to be conducted and are warranted. Further, the costs incurred by states under each of these solutions, as well as the effectiveness of each solution in adequately supporting the mental health needs of children, should be assessed. Finally, in adequately and accurately evaluating the comparative effectiveness of these policies’ impacts of the use of voluntary placement agreements to facilitate access to mental health care, it will be essential understand how each of these policies is implemented in practice and how the implementation varies—both across states and across regions within states. Differences in program implementation may play a key role in understanding the population-level effects of these policies.

Limitations

A key strength of this study is that use of a longitudinal approach, which allows for a larger sample of states and enables models to capture the effects of de-implementation in some states. However, because of data quality concerns in the AFCARS prior to 2010, observations were limited to 2010–2019. As a result, the majority of waiver programs were in effect at the start of the study period. Though nine states de-adopted the policy between 2010 and 2019, there is not sufficient data prior to policy adoption in most states to constitute a true difference-in-difference design and no causal inference can be made. It should also be noted that structure and generosity of these programs typically varies across states (Keim-Malpass et al., 2019; Velott et al., 2016). Further, variation in implementation of these policy likely affects state level outcomes (McGinty et al., 2021). This variation is not captured or accounted for in this study, limiting precision or generalizability of findings (Schuler et al., 2021).

Study findings are limited in a few additional ways. First, five states reported no voluntary placements with behavioral health need for all ten years of the study. Many (14) states also reported, on average, very few voluntary placements with behavioral health needs (less than ten per year). Further, the overall rate of voluntary placements, including those with behavioral health needs, decreased throughout the time of our study, often even in non-waiver states. Some of these low and declining rates may be due to the use of many other state policies and practices that are effectively preventing these types of placements and models did not specifically control for the existence of these policies in non-waiver states. However, the majority of these potentially confounding policies did not vary over the time of our study and thus their effects would have been managed through the use of state fixed effects(Clarke et al., 2010).

Second, the use of administrative data comes with inherent limitations, including the availability of variables and related limited insight into the behavioral health needs of children in the study. Reasons for removal, upon which our outcome variable is structured, is based on reports from caseworkers and thus subject to variation in application and use across agencies, caseworkers, and states (Green et al., 2015). For example, many states have voluntary placement statutes that require court involvement before the voluntary placement can begin (Semanchin Jones et al., 2018). As such, some states may code voluntary placements as court ordered. Furthermore, the AFCARS administration data align with the federal fiscal year (begins October 1 and ends September 30), while the 1915(c) waivers dataset corresponds to the calendar year. While the impact of this slight timetable variation on the study results is minimal, observing states’ waiver policy use and associated voluntary placement rates on a monthly or quarterly basis would help improve the robustness of the study outcomes. Additionally, key unobserved variations in state behavioral health and child welfare systems may be confounding findings. Finally, due to the limit on the number of states in the U.S., the sample size in this study is relatively small and power is thus limited. However, because this study includes the full population of states in the U.S., findings are generalizable at the national level.

Conclusion

This study documents that states’ adoption or de-adoption of 1915(c) waivers targeting children with complex behavioral health needs is associated with reductions in state rates of voluntary placements of children with behavioral health needs in the foster care system. Findings provide preliminary evidence to support wider adoption of these policies to address the mental health needs of children while preserving families and maintaining children and youth in their homes, schools, and communities. However, additional research is needed to understand youth and families’ experiences of “custody for treatment trades” and the state and provider processes for facilitating these placements. Understanding the factors that contribute to the use of voluntarily placements agreements to facilitate treatment access for children with significant behavioral health needs will enable more rigorous examinations, with greater causal inference, of state level polices and provider processes that can reduce this practice (Schuler et al., 2021).