Research to Consider While Effectively Re-Designing Child Welfare Services

Abstract

An intense appetite for reforming and transforming child welfare services in the United States is yielding many new initiatives. Vulnerable children and families who become involved with child welfare clearly deserve higher quality and more effective services. New policies, programs, and practices should be built on sound evidence. Reforms based on misunderstandings about what the current data show may ultimately harm families. This review highlights 10 commonly held misconceptions which we assert are inconsistent with the best available contemporary evidence. Implications for better alignment of evidence and reform are discussed.

The United States is poised for major child welfare reform. Nationally recognized child welfare administrator Bryan Samuels (2020) recently published a treatise entitled “The Moment is Now: Family and Child Well-Being: An Urgent Call to Action.” The U.S. Department of Health and Human Services’ (DHHS) Administration for Children and Families, in concert with several national foundations and children’s services organizations, has developed a first-of-its-kind national partnership “Thriving families, safer children: A national commitment to well-being”—to “prove that it is possible to fundamentally rethink how child welfare systems function” (DHHS, 2020b, para. 1). More extreme voices are calling for abolition. The Center for the Study for Social Policy and the University of Houston School of Social Work have developed the abolitionist #upEND movement, to reconsider a system absent of mandated child maltreatment reporting, parent-child separations, termination of parental rights, and a number of other changes (Meltzer, 2020).

In concert with these voices, the authors of this paper are in accord that major reforms are warranted. Due to a number of factors, most prominently the high rates of family poverty and income inequality in the U.S. and the few universal family support services available, child welfare systems touch the lives of many American families. Fully one-third of U.S. children, and one-half of Black children will be investigated for maltreatment at some point during childhood (Kim, Wildeman, Jonson-Reid, & Drake, 2017), one in ten Black children will be separated from their families and experience foster care (Wildeman & Emanuel, 2014). The quality of services children and families receive is highly variable. Alternative preventive approaches that support families and improved responses when prevention fails are clearly needed.

Reforms to child welfare should be informed by a wide array of stakeholders together with reliable data and the best available research evidence. Yet, some have voiced skepticism about the value of empirical research as justification for implementing their ideas (Dettlaff et al., 2020). This is unfortunate, as research has contributed significantly to ending negative practices and building new supports. For example, data from studies conducted by the Association of American Indian Affairs on egregious removal and adoption practices laid the foundation for the Indian Child Welfare Act in 1978 which established the rights of federal tribes to be engaged in all aspects of placement decision making for eligible native children (Mannes, 1996). We learned from early research that time-unlimited foster care led to very long stays marked by multiple moves (Maas & Engler, 1959) and that spells over 18 months were less likely to result in reunification (Fanshel & Shinn, 1978). These data, in part, resulted in greater emphasis on permanence and efforts to reunify families. Decades of overestimating the risks of kinship foster care led to delays in affording that placement option until research emerged to indicate that kinship care, generally, has outcomes commensurate with non-kinship care (Berrick, Barth, & Needell, 1994; Koh & Testa, 2008). Evidence about the benefits of guardianship (often with kin), provided a valuable positive permanency option to adoption for many children (Rolock & White, 2016; Testa, 2004). Research on the serious shortcomings of supports for youth emancipating from foster care (Barth, 1990; Festinger, 1983) resulted in a number of policies and financial investments to provide greater transitional supports for youth leaving care. The greatest improvements in outcomes for former foster youth appear to be through extended federal and state foster care provisions which were supported by more recent child welfare research (Courtney et al., 2020).

Clearly, much more work is required, but the modern child welfare system is comparatively young and our ability to collect and apply data has greatly accelerated. We believe this provides the optimum conditions for drawing on research to inform targets of reform. Misunderstandings about how the child welfare system works can bring about misguided perspectives on child welfare reform. Some current calls for reform are based in scientific evidence. This is very pragmatic as we have evidence that agencies that use research to guide their work have higher rates of permanency (Wulczyn et al., 2016).

Other approaches to reform—especially the abolitionist movement—are based on false assumptions about how the child welfare system operates or misrepresentations of historical practice. Reactive policy-making based on flawed assumptions risks repeating past mistakes and ill-conceived practices by discounting the current context and evidence that could support better outcomes (e.g., Jagannathan & Camasso, 2017; Testa & Poertner, 2010).

Reform can be especially effective when empirical data are coupled with the views of a variety of stakeholders. This includes acknowledging the pain of parents who were, or who have felt mistreated, and especially those who were unfairly separated from their children. It also honors the tragic stories of children who suffered harm or even death due to insufficient protection. These are reminders of the fundamental issues at stake in child welfare. Child welfare policy and practice must be guided both by attentive listening to those we serve and also to data which show us what clients are experiencing across the entire served population. Effective large-scale reform requires information describing what is happening at the population level; planning for specific services, implementation, and quality improvement need stakeholder voices. The qualitative experiences of selected parents or youth can sensitize us to issues that need to be addressed that may be related to localized practices, training, policy implementation, or other system failings. Likewise, stakeholder planning can be informed by understanding whether experiences are reflective of the whole so that positive practices can be retained or replicated. To this end, all stakeholders in reform—including families and communities—need access to the most current, accurate, relevant, and understandable evidence upon which to base decisions. A summary of current empirical research related to 10 areas of child welfare is provided so that as reform proposals advance, they can benefit from a clear understanding of what, on average, is occurring. Perhaps this will encourage stakeholders to learn from and apply data to guide child welfare services development and evaluation so we can realize a new era in supporting families and protecting children.

Misconceptions to Reconsider

An evidence-informed understanding of the current system’s functioning can help us avoid over- or under-correcting, or repeating past failures. We believe there are several areas in which underlying evidence is misunderstood or misrepresented. This is a serious problem, in that it may lead to ineffective or harmful policies. As child welfare researchers, we accept some responsibility for this as we have not done enough to generate and effectively disseminate timely research on the experience and outcomes of child welfare. Evaluating empirical data, however, can be difficult. Findings may differ because of issues with samples, variables, methods, geography, biases, and interpretation of the data. These problems can be addressed by drawing on information from multiple sources. In this way, consistent empirical findings can be identified and outcomes for various sub-groups of children—particularly outcomes for different ages and different ethno-racial groups, can be explored. We have written about our understanding of a set of child welfare outcomes, broadly and with special attention to outcomes for Black children, in a separate paper (Barth et al., 2020). The current paper focuses on issues impacting child welfare contact and subsequent service trajectories.

We are aware that the findings we present may not be universally applicable across populations, places, or agencies. Place matters, and although we benefit from some national general principles, funding schemes, and outcome measures, there is substantial geographic variation in population characteristics, policies, administrative structures, and community resources. In addition to such variation in communities and practice, we acknowledge limitations in the body of child welfare research. In spite of increasingly sophisticated research designs and analyses, we need more replication to allow detailed assessment of trends and effects across place and populations.

Our review considers the preponderance of current policy and services research on 10 topics related to current reform discussions arranged in rough time order. We refer to the system that responds to maltreatment reporting and that assesses, investigates, and determines next steps as the Child Protective Services (CPS) system; services and supports to families, including foster care and adoption, are thereafter referred to as the Child Welfare System (CWS). Topic areas are labeled as a query about misconceptions commonly expressed in the field.

Are Low-Income Children Inappropriately Referred to Child Protective Services due to Implicit Bias?

Low-income children are reported to CPS at higher rates than other children (Kim, Drake, & Jonson-Reid, 2018; Putnam-Hornstein & Needell, 2011; Sedlak et al., 2010). Why? It could be because low-income children are unfairly over-identified as a result of a biased reporting system, or it could be because low-income families face higher levels of stressors and have fewer resources with which to work. This question has been discussed for over four decades now, and the consensus is that low-income children have contact with CPS at higher rates than other children because they are at higher risk (Pelton, 2015).

Perhaps the simplest way of addressing this issue is to ask “is the low-income/non-low-income difference in CPS reports consistent with other known low-income/non-low-income differences in child well-being?” If CPS reports were driven by anti-poor bias, then CPS would serve a much higher proportion of low-income people than other systems that work to support child well-being. They do not. Low-income children experience outcomes such as death (Reno & Hyder, 2018) or other negative child outcome and well-being measures (Chaudry & Wimer, 2016) at differentials consistent with or higher than their CPS reporting differences (Drake et al., 2011).

Another way to look at these issues is to compare self-reports to official reports. Low-income persons are much more likely than higher income counterparts to self-report maltreatment (Slopen et al., 2016; Steele et al., 2016). A recent meta-analysis found no significant difference in agreement between retrospective and prospective reports regardless of whether maltreatment was measured by (1) CPS report and then later self-report, or (2) instances when both retrospective and prospective reports were self-reported (Baldwin et al., 2019). Put simply, accounts of maltreatment by victims themselves confirm that low-income children are at higher risk; they are not merely over-reported due to bias.

Yet, another way to determine if low-income children are unfairly over-reported is to look closely at which low-income children are reported. Low-income children who are reported to CPS are far more likely to have a range of medical, behavioral, delinquency, and other negative outcomes compared to equally low-income children not reported to CPS in studies using a variety of observational and administrative data methods (Bunting et al., 2018; Jonson-Reid, Drake, & Kohl, 2009; Widom, Czaja, Bentley, & Johnson, 2012). If poor families were being capriciously and randomly reported simply based on class bias, then child outcomes should be similar between poor children who are not reported and poor children who are reported. The data do not support that assumption.

Dismissal of high rates of reporting among low-income populations as merely bias has the potential to distract the nation from the urgent need to address poverty in general, and the poverty-related needs of CPS client families in particular (Pelton, 1978, 2015). This is especially unfortunate at a time when recent research shows that material need and income supports can reduce maltreatment and child welfare involvement (e.g., Beimers & Coulton, 2011; Berger et al., 2017; Brown & De Cao, 2020; Courtin et al., 2019; Rostad et al., 2017).

Are Families who Receive Public Social Services and Have Contact With Mandated Reporters Disproportionately Likely to be Referred to Child Protective Services?

It certainly seems reasonable to assume that those who encounter more mandated reporters are much more likely to be reported. It also seems reasonable that a low-income individual served in a clinic may be much more likely to be reported than a middle-class person seeing a family doctor with whom they have a longstanding relationship. These assumptions, however prevalent, are very likely inaccurate (Pelton, 1978, 2015).

There are three studies which directly assess this proposed dynamic. The first (Chaffin & Bard, 2006) followed children in a single state who were served by family support programs, parenting programs or in-home services, and then tracked later reports and report sources. The second study was national in scope and tracked re-reports among CPS referred families who had or had not received services (Drake, Jonson-Reid, & Kim, 2017). Both studies found that possible surveillance bias effects increased CPS reporting by less than 2%. Another study comparing children in families receiving income maintenance with and without CPS reports found those reported to CPS also had higher rates of official delinquency, mental health problems, and hospital care for injury—again suggesting that the reports were likely indicators of concern and not just a result of surveilling low-income families receiving aid (Jonson-Reid et al., 2009).

Further, national and state level reports show that as individual or community poverty increases, the proportion of mandated reporters among all reports decreases. Low-income people are proportionately less likely to be reported by mandated reporters. This is exactly the opposite of what would be required for the surveillance bias model to be true (Drake, Lee, & Jonson-Reid, 2009; Kim et al., 2018). In summary, while surveillance bias may exist to a very small degree, no meaningfully large impact on report rates has ever been found.

We acknowledge that there are important critiques relating to mandated reporting in general (Wald, 2014), and that reforms aimed at improving consistency are warranted (Rebbe, 2018), but constraining mandated reporters from referring children who may be in danger is unlikely to improve children’s safety. Unless new empirical evidence can convincingly demonstrate meaningfully high levels of surveillance bias, it seems it is time to turn away from this persistent but unsupported misconception.

Is the Racial Disproportionality of Black Children in CPS Substantially Driven by Bias?

Black children are almost twice as likely as White children to have a CPS referral (DHHS, 2020a). Given extreme income inequality by race and the strong correlation between poverty and maltreatment, how should we view this?

Compared to White children, Black children in the U.S. are about three times more likely to live in poverty (Kids Count Data Center, 2020), and poverty is powerfully associated with child maltreatment (Pelton, 2015). The real question is “Once poverty is considered, are Black children disproportionately reported?” This question can be answered through the use of multivariate models which control for poverty and other risk factors. Once poverty is controlled, several studies show that Black children are at similar, or perhaps slightly lower risk of being reported than White children (Kim & Drake, 2018; Maloney, Jiang, Putnam-Hornstein, Dalton, & Vaithianathan, 2017; Putnam-Hornstein & Needell, 2011). When we compare (generally poorer) Black children to (generally wealthier) White children, we are making an invalid “apples to oranges” comparison. When we make an economically fair “apples to apples” comparison, the overrepresentation of Black children disappears. By no means do we mean to minimize the profound wrongness of economic disadvantage facing Black children—we instead wish to shine a light directly upon it. Racially based economic stratification is a profound and crippling societal flaw which must be addressed.

Myths about large-scale over-reporting of Black children persist nonetheless. These views may be shaped by small, poorly conducted, and outdated hospital studies showing high rates of reporting for Black mothers with positive toxicology screens (e.g., Chasnoff, Landress, & Barrett, 1990). A careful look at more recent studies using population-based samples suggests that Black substance-exposed infants were less likely to be referred for maltreatment than White or Hispanic substance-exposed infants (Putnam-Hornstein, Prindle, & Leventhal, 2016). While this study did not address possible inequalities in hospital decisions to screen children, it did show that once identified, Black substance-exposed children were not more likely to be referred to CPS.

Another way to look at this question is by comparison of child maltreatment disparity to disparity in objective, unbiased measures of child well-being, or risk. This approach also suggests that it is unlikely Black children are over-reported, with Black/White disparities in maltreatment reporting being consistently lower than disparities in other objective indicators of child well-being, such as mortality (Drake et al., 2011). That is, the same socio-economic factors that disproportionately place Black children at risk of a range of adverse health and social indicators also increase the risk of maltreatment among low-income Black (and White) children (Lanier, Maguire-Jack, Walsh, Drake, & Hubel, 2014). If Black children were being over-reported due to bias, we would expect a persistence of over-reporting once income is controlled for and we would also expect that reports to CPS would be more racially disparate than objective measures of child well-being. The data do not bear out these assumptions. Data suggest that efforts to address the notable racial disproportionality of Black children referred to Child Protective Services should focus on the socio-economic antecedents that place Black children at unequal risk of poverty, and that directly address family poverty in the U.S.

Are Decisions to Substantiate or Place in Foster Care Largely Driven by Racial Bias?

Child maltreatment reports are made both by public professionals and community members. Once a case is investigated, we can examine if child welfare professionals’ recommendations relating to substantiation or child placement vary by race. The large majority of studies using recent data (e.g., Drake et al., 2021; Fix & Nair, 2020; Maloney et al., 2017; Putnam-Hornstein, Needell, King, & Johnson-Motoyama, 2013) show that as they move through the system, socio-economically disadvantaged Black children are generally less likely to be substantiated or removed into foster care compared to White children. Two exceptions to this trend in the literature using older data can be found (Dettlaff et al., 2011; Maguire-Jack, Font, & Dillard, 2020), although the Dettlaff study only found higher rates of substantiation for Black children once a measure of caseworker risk perception was included.

A persistent racial difference which clearly does exist is that Black children stay in foster care about 25% longer than White children, perhaps due to the lower likelihood of reunification and adoption as exit opportunities (Wulczyn, 2020). Due to the nature of the national data used in studying length of stay, however, adequate poverty controls have not been employed, suggesting that a current assessment of race as the sole predictor of reunification or adoption is unpersuasive. Efforts to address the thin evidence base on the effectiveness of reunification services could offer more favorable opportunities for stable family reunions among Black families. Some of these differences may reflect a longer but still successful process for reunification for Black children. This would be in keeping with federal encouragement to de-emphasize timelines in favor of permanency (DHHS, 2021b). For children who are unable to reunify, efforts to promote alternative exit opportunities—including kinship guardianship and adoption—are likely warranted.

In summary, with the exception of longer foster care stays for Black children, the evidence suggests that Black children are being reported, substantiated, and placed at rates commensurate with or even lower than their increased poverty and exposure to risk factors would suggest. Tackling the underlying race-based social and economic inequalities in our society requires urgent attention and is likely to have very positive impacts on families’ child welfare involvement.

Is Child Neglect Synonymous With Family Poverty?

The majority of substantiated maltreatment reports and fatalities (over 70%) include child neglect (DHHS, 2021a). There is clear evidence establishing the relationship between poverty and child neglect (Cancian, Yang, & Slack, 2013; Drake & Jonson-Reid, 2014; Pelton, 2015), which may, in part, lead to misunderstandings and assertions in the popular press that neglect is just another word for poverty or an excuse to punish poor parents (Riley, 2020). The reference to material needs in some state definitions of neglect may also lead to misconceptions. It is important to remember that the association of poverty with heightened risk of neglect does not mean they are one and the same.

Differentiating between child neglect and poverty can be made more difficult because poverty alone can have significant negative impacts on child development (e.g., Luby, 2015) and poverty is also associated with a number of adverse childhood experiences (Slopen et al., 2016; Walsh, McCartney, Smith, & Armour, 2019) in addition to maltreatment. Despite these possible confounds, for decades studies have suggested that both poverty and child neglect exert separate negative impacts on children’s development (Brooks-Gunn, Klebanov, Fong-Ruey, & Duncan, 1995; Hildyard & Wolfe, 2002). These effects occur across the lifespan. A study of young low-income children found that neglected children fared worse in measures of kindergarten behavior and first grade academic performance (Manly, Lynch, Oshri, Herzog, & Wortel, 2013). In a birth cohort study (Strathearn et al., 2020), neglect was related to a number of poor outcomes across domains while controlling for additional maltreatment types and socio-economic status. In a longitudinal observational study, Nikulina, Widom, & Czaja (2011) found that child neglect had effects distinct from poverty in areas such as mental health, criminal justice involvement, and academic outcomes in adulthood. Another study of low-income young adults found that those previously referred to CPS for child neglect had higher risk of criminal justice involvement, poor economic outcomes, low educational attainment, and early parenthood (Font & Maguire-Jack, 2020a).

Another way to assess how neglect may differ from poverty is to look at the constellation of risk factors associated with the occurrence of neglect as compared to poverty. A recent meta-analysis found that while poverty was consistently associated with neglect, the mean effect size for other risk factors like parental antisocial behavior, mental, or physical problems was higher (Mulder, Kuiper, van der Put, Stams, & Assink, 2018). Among low-income parents, those reported for neglect have been found to experience a range of hardships and suffer work and residential instability, show less parental warmth and are more likely to use corporal punishment (Slack, Holl, McDaniel, Yoo, & Bolger, 2004). Relying on a combination of youth self-report and CPS records, a study examining low-income children found that children who were physically neglected were more likely to have parents who struggled with depression, parents who reported maltreatment in childhood, and the children reported living in lower quality neighborhoods (Shanahan, Runyan, Martin, & Kotch, 2017). Similar findings exist for studies using self-report measures of neglect. Using the Multidimensional Neglectful Behavior Scale (Straus, 2006), a population level study found that risk factors for child neglect include maternal mental health and substance abuse, and paternal work family balance and economic stress (Clément et al., Bérubé, & Chamberland, 2016). Another study measured frequency of parenting behaviors related to neglect (e.g., lack of caring about rule breaking, can’t be bothered to ensure child has enough to eat, ignore child’s feelings, not caring if child does something dangerous, etc.). This study found that parental burnout predicted neglectful and violent behaviors, controlling for economic status (Mikolajczak, Brianda, Avalosse, & Roskam, 2018). Studies using officially reported neglect and those using scales measuring self-reported neglecting behaviors find unique constellations of risks and/or parenting behaviors associated with child neglect.

Finally, most studies indicate that compared to other forms of maltreatment, neglect is more likely to recur and frequently co-occurs with other forms of maltreatment (Jonson-Reid, Drake, & Kohl, 2017; Kim & Drake, 2018; Mennen, Kim, Sang, & Trickett, 2010; Proctor & Dubowitz, 2014; Strathearn et al., 2020). The presence of factors unique to neglect driving recurrence is important as the risk of poor outcomes has been found to increase with chronic maltreatment or multiple forms of maltreatment (Jaffee & Maikovich-Fong, 2011; Warmingham, Handley, Rogosch, Manly, & Cicchetti, 2019). While a large-scale study of economic intervention showed promise in reducing recurrence in samples of predominantly neglect cases, the effect was modest (about 11%) suggesting the presence of other factors beyond poverty that influence recurrence among such cases (Rostad et al., 2017). Studies of outcomes following recurrent maltreatment suggest neglect has an equal role compared to other forms of maltreatment and greater when compared to poverty alone (Jonson-Reid, Kohl, & Drake, 2012; Warmingham et al., 2019).

Additional confusion may arise due to how neglect is defined in some state policies guiding reporting. The large majority of states include in their definition lack of supervision and abandonment, lack of medical care, and some form of lack of basic needs; they also range from requiring risk of harm to actual harm in order to respond (Rebbe, 2018). Many states try to differentiate between what may be termed, involuntary neglect—conditions associated with poverty alone—from cases that should be determined as maltreatment because parental behaviors or inaction stem from more than lack of resources (Rebbe, 2018). Indeed, only a small subset of neglect referrals (perhaps one in four) are due to material needs, and those cases are only about a quarter as likely to be substantiated as other neglect cases (Jonson-Reid, Drake, & Zhou, 2013; Kang, Bae, & Fuller, 2015). There are well-founded critiques regarding the variability of child neglect laws across the country (Milner & Kelly, 2020; Wald, 2014); efforts to clarify and narrow these definitional frames for certain forms of intervention are warranted (Day et al., 2021). On the other hand, it is not clear how such clarifications should be implemented. As Font and Mcguire-Jack (2020b) have stated: “… whereas the distinction between poverty and neglect can and should drive decisions about how to intervene—in-home services versus foster care, court-mandated services versus voluntary community supports—it is not clear such distinctions should determine whether to intervene” (p. 29). For example, while some may prefer a narrower requirement of actual harm to drive intervention, using “threat of harm” in state definitions allows states greater leeway to offer preventive services to families (Rebbe, 2018).

Taken together, the evidence overwhelmingly suggests that neglect is a marker for conditions that may be associated with, but are distinct from, poverty. Narratives that conflate poverty and child neglect unfairly characterize low-income families, the majority of whom provide appropriate care for their children (Carter & Myers, 2007). As discussed in earlier sections, families of color are more likely to suffer from poverty related to structural racism. The greater likelihood of poverty among families of color should motivate primary prevention related to socio-economic circumstance (which would address at least one major risk factor for neglect) rather than averting our gaze from the many circumstances in which neglectful parenting diverts from and must be treated differently than poverty alone. Policies and practices that address family poverty are urgently needed and would likely reduce but not eliminate the incidence of child neglect (Berger et al., 2017; Rostad et al., 2013). Moreover, efforts to characterize child neglect as simply a problem of poverty have the potential to obscure the real harms associated with neglect. These harms and questions of response are reviewed in the next section.

Is Child Neglect Harmful to Children?

Whether we define child neglect as maltreatment is, in part, related to our understanding of neglect as harmful to children. The “neglect of neglect” in child welfare research (Dubowitz, 1994; McSherry, 2007) may contribute to misconceptions about the serious nature of the phenomenon. In spite of the variability in definition, child neglect, whether measured by official reports, self-report, or parent report, is associated with a range of serious, negative outcomes. The effects of neglect may become evident when a lack of the basic nurturing, care, and supervision needs of a child reach a threshold that results in neurobiological and/or socioemotional harm (Proctor & Dubowitz, 2014). Neglect, particularly during sensitive developmental periods or chronically, can lead to significant harm such as death (Jonson-Reid, Chance, & Drake, 2007; DHHS, 2021a, 2021b) or hospitalization (Rebbe et al., 2021), as well as longer term alterations in children’s developmental trajectory (e.g., Pereira, Li, & Power, 2017).

Compared to children who are abused, children suffering from neglect exhibit equal or more serious developmental deficits across the lifespan (Strathearn et al., 2020). These include an increased likelihood of poor cognitive or academic outcomes (Font & Maguire-Jack, 2020a; Manly et al., 2013; Nikulina et al., 2011; Strathearn et al., 2020), trauma symptoms and mental health problems (Cohen, Menon, Shorey, Le, & Temple, 2017; Nikulina et al., 2011; Strathearn et al., 2020; Turner, Vanderminden, Finkelhor, & Hamby, 2019), second generation perpetration (Ben-David, Jonson-Reid, Drake, & Kohl, 2015), increased likelihood of juvenile justice or criminal justice involvement and substance abuse (Cohen et al., 2017; Font & Maguire-Jack, 2020a; Strathearn et al., 2020; Vidal et al., 2017), sexual risk taking or early parenthood (Font & Maguire-Jack, 2020a; Strathearn et al., 2020; Wilson & Widom, 2010), and reduced economic well-being in adulthood (Currie & Spatz Widom, 2010). These findings hold true across studies whether maltreatment is based on self/parent report or child welfare contact.

Building off the prior section, poor outcomes associated with neglect are also found not only when compared to abuse but also when controlling for poverty. Such studies find both behavioral and academic problems across childhood and into young adulthood (e.g., Bunting et al., 2018; Font & Maguire-Jack, 2020b; Manly et al., 2013; Mills et al., 2011; Nikulina et al., 2011; Vidal et al., 2017). Some research also indicates that child neglect is linked to additional risk of other forms of victimization as well as trauma symptoms (Turner et al., 2019).

For too long, the public’s understanding about the phenomenon of child neglect has been misguided. Child neglect brings its own host of risks to healthy development across the lifespan. Perhaps due to this misconception regarding the nature of and harm related to the phenomena of child neglect, less attention has been paid to developing effective interventions. One of the few exceptions is SafeCare which is building the evidence base in regard to intervention with child welfare involved families (Whitaker et al., 2020) and has also been shown to be culturally acceptable and effective with underrepresented minority families (Chaffin, Bard, Bigfoot, & Maher, 2012; Rogers-Brown et al., 2020). Current efforts to test approaches to address poverty and material needs show promise of reducing the incidence of maltreatment, but more work is needed to understand the best means of delivery and full impact of these approaches (Drake, Jonson-Reid, & Dvalishvili, in press). We can engage in healthy debate about the best way to address both prevention and intervention to promote healthy outcomes, but any effort to support child well-being must address child neglect as a real and serious public health concern.

Are Research-Supported Practices Effective for Families of Color?

Contrary to much popular belief, decades of well-designed and executed research findings are informed by the experiences of culturally diverse youth and adults (Huey et al., 2014). Indeed, Huey and Jones (2013) identified 30 interventions classified as “probably or possibly efficacious” for ethnic minority children and adolescents in the general population. Huey and Polo’s (2008) review also concluded that cultural adaptations did not add very much to the general efficacy of the original model.

While noteworthy, these early reviews did not grapple with efficacy for families of color in the child welfare system. Since the early 2000s, findings from randomized controlled trials and quasi-experimental studies (e.g., Chaffin et al., 2004; Painter, 2009) challenge the assertion that research-supported practices or research-supported treatments (RSTs) are ineffective for families of color who are involved in child welfare. Garcia, DeNard, Morones, & Eldeeb (2019) identified four RSTs from the California Evidence-Based Clearinghouse (CEBC) that are deemed “well-supported” by at least one of these rigorous designs, and achieved a diversity threshold in which at least 40% of the study samples included children and families of color. They include Parent-Child Interaction Therapy, Trauma-Focused Cognitive Behavioral Therapy, Level 4 Triple P (Pathways Positive Parenting Program), and Multi-Systemic Therapy. Collectively, these RSTs address a range of issues of concern, including child externalizing and disruptive behaviors; attention deficit hyperactivity disorder, post-traumatic stress disorder, and depression symptoms; juvenile justice system involvement; parenting stress; and school absence and poor school functioning. More broadly, Miranda et al. (2005) have also shown that cognitive behavioral therapy (CBT), which is the basis of many interventions, is a robust intervention across populations. Although not listed by the CEBC, the evidence of CBT’s general benefit is promising. Scholars and agency leaders who have implemented and/or evaluated RSTs in child welfare reported during semi-structured interviews that while RSTs are indeed effective for diverse populations, these groups are more likely to experience inequities in access to RSTs, and barriers to actively engaging in them (Garcia et al., 2019).

Simply put, RSTs are effective as long as child welfare workers are trained and motivated to refer clients to them (Myers et al., 2020), and when clinicians are skilled at balancing fidelity and flexibility (Garcia, DeNard, Ohene, Morones, Connaughton, 2018). In child welfare contexts, “balancing” may involve implementing RSTs as intended, albeit diversifying engagement strategies to increase access to and active participation among children and families of color. This might entail introducing case examples that can relate to diverse populations, and capitalizing on familial and cultural bonds between peers, relatives, and fictive kin during intervention planning (Garcia et al., 2018; McCabe & Yeh, 2009). To that end, fidelity is not compromised, but strategies to prevent premature discontinuation are considered.

Mischaracterizations of RSTs as ineffective for families of color have the potential to dissuade professionals from offering these services to diverse families. The result could be that families of color are denied access to services that work. Studies over the past two decades reveal that some child and family focused-RSTs are robust and helpful across diverse populations, if implemented to fidelity. Despite these advances, more evidence is needed for a range of practices in the field of child welfare across diverse populations, settings, and contexts. To that end, practices should continue to be assessed for their efficacy and applicability with the diverse population that child welfare serves.

Do Foster Children “Grow Up” in Foster Care?

Child welfare researchers have been analyzing children’s trajectories in and through care with increasing sophistication. The prodigious quarter-century effort to develop child and family service review metrics has paid off with a clearer national picture of foster care entrances, exits, re-entrances, types of placements, and lengths of stay. Efforts like Chapin Hall’s Center for State Child Welfare Data (2021) and many state-university partnerships have also provided a more precise look at our services by age, race, and location.

The evidence is overwhelming that children, today—unlike those who may have experienced foster care in the 20th century—do not grow up in foster care. Although anecdotes from multiple sources and some limited data from previous decades suggest long-term foster care was a pattern for many children (Fanshel, Finch, & Grundy, 1990), efforts to infuse permanency into the terminology, governing policies and practices of child welfare professionals, attorneys, and judicial decision makers appear to have borne fruit. In two recent studies examining the likelihood that an infant placed in care would experience a spell of continuous foster care to age 18, both found that far fewer than one percent of infants experienced a childhood in care (Magruder & Berrick, 2021; Wulczyn, 2020). For children aged 13 or younger, the risk of remaining in foster care to age 18 is less than 10%. Examining entries to care over time, Wulczyn (2020) finds that in more recent years, the likelihood of experiencing long-term foster care declines further, suggesting that the trend toward shorter stays in care continues.

If the majority of children do not grow up in foster care, do they spend a preponderance of their childhood in care? Wulczyn (2020) examines these issues showing that children who enter care as infants spend the longest duration in care. Including entries and re-entries to care, infants spend approximately 10% of their entire childhood in care (number of days in care divided by the number of days in childhood [365 × 18 years], though their longer duration in care is largely driven by their greater likelihood of adoption as a foster care outcome. Less than 4% of 10-year-old children admitted to foster care in 2010 reached their 18th birthday in care. Those children will spend less than 8% of their childhoods in care.

Another significant sub-population of children who enter foster care do so at age 14 or older experiencing behavioral health problems (Barth, Wildfire, & Green, 2006). Even these youth have only a one-in-eight chance (13%) of remaining in foster care until age 18. Most return home sooner. In all, about half of the youth who remain in foster care at age 18 were admitted after age 15 (Wulczyn, 2020), another indicator that the difficulties they may experience in young adulthood are typically not caused by long stays in foster care.

Placement dynamics vary by age, race, geography, and other factors. Black children experience more lengthy stays in out-of-home care compared to White and Latinx children. This is largely due to the fact that their exit patterns differ. Black children have a lower likelihood of leaving care to reunification, though the differences are not pronounced during the first 6 months of care when reunification rates are about 20% for all groups. Black children are also less likely to leave care via adoption (Wulczyn, 2020).

Children’s placement dynamics vary by a range of factors; most children experience temporary stays in care and the large majority exit to reunification, adoption, or guardianship. The population of children who “age out” are primarily youth who enter care as teenagers. Reform efforts based on the misconception that children typically grow up in foster care first do a disservice to the child welfare and allied professionals who have embraced concepts of permanency and who have worked diligently to shorten lengths of stay. Second, they overemphasize the role that out-of-home care likely plays in children’s young adult and adult outcomes. And third, they undermine the importance of out-of-home care as one option available in the continuum of services, and used only when absolutely necessary and when other services along the continuum have been exhausted.

Does Foster Care Cause Poor Outcomes for Children and Youth?

Studies of young adults who have experienced foster care generally conclude that outcomes across a range of domains are worse than outcomes for young adults in the general population (Courtney & Dworsky, 2006; Courtney et al., 2011, 2020; Pecora et al., 2005). Based on this information alone, some erroneously conclude that foster care causes poor outcomes. The appropriate reference group for comparison, however, is essential to our understanding of foster care outcomes. We know, first, that child maltreatment is associated with a range of serious, long-term, negative outcomes (Widom, 2014). A question to be addressed is whether time in foster care increases the likelihood of negative young adult outcomes over and above the effects of maltreatment.

Independent of a foster care experience, children entering care typically exhibit a range of characteristics associated with increased risk for negative adult outcomes (Berger, Bruch, Johnson, James, & Rubin, 2009; Lindquist & Santavirta, 2014). These children often struggle greatly as adults. Studies of former foster youth without an appropriate comparison group often shape the narrative, however, supporting the idea that time in foster care causes poor outcomes after foster care. Berzin (2008; 2010) shows that the techniques used to match foster youth with an appropriate comparison group may impact results in transition outcomes. Barth and colleagues (2020) review dozens of methodologically rigorous studies examining outcomes following foster care in multiple life course domains such as safety, criminal justice involvement, education, health and behavioral health, and permanence and stability. The review by Barth and colleagues, which does not limit studies to children who age out of care, shows that it is unlikely foster care worsens children’s outcomes. In some areas such as child safety, children’s outcomes appear improved: Foster care may prevent early mortality among Black children, and decrease the likelihood of Black girls experiencing teen pregnancy and juvenile justice involvement.

Why, then, are there widespread misunderstandings about the causal role of foster care in children’s outcomes? Foster care is an umbrella term that captures considerable experiential variability. One source of misunderstanding is the vivid impact of powerful anecdotal narratives about negative foster care experiences which are both compelling and deeply disturbing. Children should never be subject to poor-quality care. Case studies may highlight situations which are concerning, but cannot tell us about experiences across an entire population. Some studies combine the voices of children (through interviews) with rigorous sampling and statistical analyses of survey data to produce a more generalizable picture of foster care experiences. Numerous mixed-methods studies, both early and recent, (e.g., Barth, 1990; Courtney et al., 2020; Festinger, 1983) have elicited the perspectives of current and former foster youth via in-depth interviews and/or detailed surveys. Courtney and colleagues (2007, 2010; 2011; 2020) found that most young adults reflected favorably on their foster care experiences after exiting care. Studies show that most younger children in care also report their care experience neutrally or favorably (e.g., Chapman, Wall, & Barth, 2004; Dunn, Culhane, & Taussig, 2010; Fox & Berrick, 2007). Not discounting the perspectives of young people who report dissatisfaction with foster care services, studies show that most youth are satisfied with their out-of-home living arrangements, have positive relationships with substitute caregivers, and receive quality care from resource families (Chapman et al., 2004; Fox & Berrick, 2007). Jones (2015) found that eight in 10 former foster youth believed that their home removals were justified by necessity, and that their foster care placements in some way benefitted their life circumstances. Other qualitative studies by Fox, Berrick, & Frasch. (2008) and Wilson and Conroy (1999, as cited in Fox & Berrick, 2007) found that most children felt safer in out-of-home care than in their homes of origin. Themes of safety and satisfaction emerging from youth voices in the literature run counter to pervasive negative public perceptions of foster care.

Another source of misunderstanding may be rooted in an over-reliance on studies of children aging out of care, compared to an inappropriate reference group. Compared to their peers in the general population, foster youth who age-out face elevated risks of homelessness, unemployment, poverty, low educational attainment, and health problems (Fernandes-Alcantara, 2019). Most studies on adult outcomes target young people who entered care as adolescents and later aged out of care (e.g., Courtney et al., 2007). For adolescents entering care, the accumulation of significant pre-placement risk likely contributes to negative post-care outcomes. Further, in 2008, the upper age limit of foster care eligibility was extended—at state discretion—from 18 to 21 years. Most studies in the relevant literature base were conducted prior to this key policy change. Subsequent research suggests that extended care functions as a protective factor for older youth. Courtney and colleagues (2020), for example, found extended foster care to be associated with improved educational attainment and increased financial assets. Longer time in extended care is also associated with reduced odds of arrest, criminal conviction, early pregnancy/parenthood, and homelessness/housing instability for 17 to 21 year olds (Courtney et al., 2018).

Research to date underscores the diversity of children’s pre-placement risks and care experiences, challenging the widely held misconception that foster care, in and of itself, causes negative outcomes. An evidence-informed understanding of the role of foster care in the lives of maltreated children indicates that the average experience of care is more favorable than conditions in the birth home at the time of removal. Although a major part of any reform agenda should focus on the pre-placement conditions that place children at risk of entering foster care, misinterpreted narratives about the negative consequences of care have the potential to dissuade the public from stepping forward to fulfill the role of foster parent. This diminishes the capacity of foster care systems to provide needed services, and inhibits the participation of stakeholders in efforts to improve the quality of foster care.

Is Adoption Breakdown Common for Former Foster Children?

Efforts to achieve permanency for U.S. children in out-of-home care are taken seriously by child welfare professionals. As a result, almost half of the children placed in care are reunified, another one-quarter are adopted, and 10% exit care to guardianship (DHHS, 2020c). For young children, adoption can be a particularly likely outcome (Wulczyn, 2020). Some have raised concerns that the push for permanency may result in hasty adoption determinations, decreasing the likelihood of success (e.g., Festinger & Pratt, 2002).

Misunderstandings about the likelihood of adoption dissolution are perpetuated, in part, by misleading terminology. Discontinuity might arise prior to adoption finalization or post-finalization. Several authors refer to adoption disruption as occurring prior to finalization and adoption dissolution occurring post-finalization (Coakley & Berrick, 2008; Palacios, Rolock, Selwyn, & Barbosa-Ducharne, 2019). Here, we attend to the phenomenon of adoption dissolution only; adoption disruption, though concerning and distressing for all involved, can be conceptualized as another form of placement instability.

The U.S. emphasis on adoption as, generally, the best outcome for children who cannot return home emerged in the 1980s as we understood that children placed in out-of-home care often experience considerable instability (Goldstein, Freud, Solnit, & Burlingham, 1984). Reunification continues to be the child welfare priority, although recent estimates indicate that almost one in three children who are reunified eventually return to care (Wulczyn, Parolini, Schmits, Magruder, & Webster, 2020).

By comparison, adoption affords children considerable stability. Across a range of studies, adoption dissolution rates typically fall well under five percent. A GAO report from about 20 years ago, using data collected from almost half of the states, showed the incidence of dissolution at about one percent. A more recent study by Smith (2014, as cited in Palacios et al., 2019) reported a dissolution rate of 2.2%. Some studies have reported dissolution rates that are higher. For example, Rolock and colleagues (2019) examined rates of returning from adoption to foster care in Illinois and New Jersey. About five percent had returned to foster care before reaching the age of majority. Sattler and Font (2020) found that the overall rate of dissolution among Texas adoptions after 8 years, was two percent (compared to seven percent for guardianships).

As in all data about the foster care experience, averages can mask considerable variability by age, race, initial placement type, and other factors. Adoption dissolution for very young children is unlikely. In a retrospective study of children placed in care during infancy and followed for 18 years, Magruder and Berrick (2021) found that almost half (46%) left their first episode in care to adoption and 54% were eventually adopted. They estimate that approximately three percent of these children’s adoptions dissolved before age 18.

The misconception that adoption poses considerable risk of impermanence for children can be re-conceptualized as adoption being a stable permanency alternative for children who otherwise cannot be reunified. Reform efforts that seek to curtail opportunities for adoption among children who cannot be reunified would deny most children the lifetime of permanency that our laws seek to promote.

Conclusion

Child welfare professionals, advocates, and researchers have long worked toward improving child welfare services. The core principles of attending to safety, the importance of family, the need for permanency, and the central role of tribal child welfare have been in place for more than 40 years, but are still evolving. A little over 20 years ago, child well-being was included in law as a child welfare system aspiration and greater emphasis was placed on transition services for older youth in care. More recently, guidance was provided to strengthen the implementation of the Indian Child Welfare Act. Much has changed during the decades as researchers have built stronger evidence about maltreatment and child welfare involvement, but it is also distressing to see how much improvement is still needed. We are encouraged by the current attention to growing and improving preventive resources and in improving transitions to successful adult living. We want to see similar innovation and investments in supporting birth families before placements and in improving the experiences of children, birth families, and families who provide care when children cannot safely remain at home.

A host of issues not covered here await more evidence to guide the way forward. For example, there are undoubtedly promising advances in tribal child welfare (e.g., Chaffin et al., 2012; Haight et al., 2019; Scannapieco & Iannone, 2012) or regional practices (Lothridge, McCroskey, Pecora, Chambers, & Fatemi, 2012) emerging in the research literature that will strengthen capacity to implement effective reform. Reform should arise from a healthy mixture of ideas from children, parents, kin, judges, child welfare professionals, taxpayers, and child welfare and prevention science scholars, to name just a few key stakeholders. Children, families, communities, and policy makers wrestling with how best to prevent maltreatment and intervene after it occurs deserve access to the best available data whether that be on the nature and scope of maltreatment, the current system response, or on existing practices that are available to replicate or adapt. Here we offered a summary of some of this knowledge. Why is this important? Misunderstandings can arise when what is being written in commentaries and social media does not reflect actual practice.

Sometimes, these views seem familiar and accurate—but are from an earlier era or rely on small case studies that do not reflect the common or current experience. Child welfare is ever changing and research literature is not as readily accessible and frequently revisited as it should be. Basing future innovations on a dated view of child welfare or on assumptions that are not well anchored in research are not likely to generate optimal outcomes for children, families, or society. We are optimistic that the current move to reform or transform child welfare can benefit from and move forward with the best available evidence. This includes bringing the most rigorous methods to bear to fill key gaps as new ideas develop, supporting the effective dissemination of the newly accumulated knowledge, and using accurate baseline data to assess progress.