Child Welfare Management and Delivery Solutions

Seventeen data-driven strategies state and local governments are using to improve results for children and families

CW Solutions Book CoverThis book highlights seventeen management and delivery solutions that public child welfare and early childhood agencies are using to overcome the operational barriers that often undercut attempts to improve the health and wellbeing of vulnerable children and their families. Throughout the country, there are remarkable cases of public agencies improving practices and outcomes when leadership and technical capacity are aligned.

For more information about the GPL’s work in child welfare and early childhood – including project features that expand upon many of the case studies referenced herein and technical guides that detail how to undertake a subset of these strategies – visit the Children & Families section of our website.


Preface: Innovative management and delivery strategies are making a difference in the lives of children and families

Imagine a future in which family strengthening programs are so effective that the number of children impacted by abuse or neglect is one-third of the 674,000 children annually victimized today; a future in which every parent experiencing addiction can access needed treatment1; a future in which dramatically fewer children are removed from their parents; and a future in which children can be cared for by a grandparent, aunt, or uncle when home is unsafe.

By 2022, the Family First Prevention Services Act will be providing more than $180 million a year to help state and local governments make investments in preventative efforts that could turn this vision a reality.2 However, this transformation will succeed only if public child welfare, early childhood, and other family support systems overcome the operational barriers that often undercut attempts to improve the health and wellbeing of children and their families.

Today upstream prevention programs inconsistently reach the children and families who would benefit most from early help. The result is that most child welfare interventions are reactive, occurring only after suspected maltreatment has occurred. When children are at risk or unsafe, governments typical fund community organizations to deliver help to these children and their families – yet we see governments treat contracting as a back-office administrative function with limited focus on working with these providers to produce better results. These services are rarely coordinated with other supportive programs, despite many families having addiction, domestic violence, housing, or other needs.

In the 2007 global economic crisis, we saw that these obstacles are amplified when budget pressures force governments to shrink agency staffing – as is likely to occur again as our country faces the COVID-19 pandemic. These operational challenges also contribute to racial inequity in the child welfare system, exacerbating the troubling overrepresentation and disparate outcomes of children and families of color.

Nevertheless, one need not “imagine” to find evidence that progress is possible. Throughout the country, there are remarkable cases of public agencies improving practices and outcomes when leadership and technical capacity are aligned. Rhode Island has decreased from 8 percent to 3 percent the share of families opening for child protection following prevention services. In Florida’s Tampa region, the state and its behavioral health providers have doubled the share of parents with substance use needs who swiftly engage in treatment. In Pennsylvania’s Allegheny County, two-thirds of children in foster care live with relatives – twice the national rate.

From these and other examples, the Harvard Kennedy School Government Performance Lab (GPL) has identified seventeen management and delivery solutions that governments are using to improve results for children and families. These strategies reflect lessons from our hands-on engagements helping jurisdictions design and implement strategies to overcome performance challenges, as well as innovations we have learned about from other communities. We have written this solutions book in the hope that these approaches will spread more rapidly.

We plan to regularly update this book as we learn of new solutions and new case studies. Please send us ideas to feature in subsequent editions. For more information about the GPL’s work in child welfare and early childhood – including project features that expand upon many of the case studies referenced herein and technical guides that detail how to undertake a subset of these strategies – visit our Children and Families page at

Jeffrey Liebman, Director, HKS Government Performance Lab
Scott Kleiman, Managing Director: Children & Families Practice, HKS Government Performance Lab

CW Solutions Book Overview

Strengthening prevention among families at risk

#1. Prioritize highest-risk families for home visiting and other prevention programs rather than filling slots ad hoc

Upstream prevention programs rarely have the scale to reach every family in a jurisdiction. The interventions supported by the strongest evidence, such as intensive family home visiting, are often too expensive to be offered universally. Many jurisdictions do not have a coordinated strategy for prioritizing families most at risk of experiencing abuse or neglect for access to these services. Developing a theory about which subpopulations to concentrate on requires customer segmentation tools that governments often lack. Few child protection hotline or investigations units consistently make referrals to prevention programs for families with screened-out or unfounded cases, despite research that tells us prior contact with the child welfare system is a meaningful risk factor for future harm.3 Prevention programs are often provided by community organizations, yet conventional fee-for-service contracting structures incentivize these providers to serve clients that are the easiest to engage regardless of their risk level.

Model approaches

Jurisdictions should pursue strategies that prioritize the highest-risk families for family home visiting services and other prevention programs. Such initiatives may require new partnerships between the child welfare agency, the public health agency, and the early childhood agency and may include:

- Determine which families should get prioritized for a community’s most intensive prevention services by analyzing historical outcomes data, such as low-birth-weight births and - child maltreatment, to uncover geographic hotspots, patterns of prior system involvement, and other family risk factors. Compare this information about who is experiencing poor outcomes with data about who is accessing services today to prioritize subpopulations for targeted outreach. This kind of segmentation analysis can also inform policy choices such as which communities to target for public health campaigns around healthy infant sleep, which teen pregnancy prevention efforts to invest resources in, and how to better reach at-risk expectant mothers early enough in pregnancy to impact birth outcomes.

- Establish centralized family home visiting triage mechanisms to route families to best-fit prevention services and ensure that slots in the most intensive evidence-based programs are reserved for families determined to need the most help. Often broadly targeted short-term home visiting programs, which offer one or two home visits to all or many families in a jurisdiction, can be useful mechanisms for figuring out which families are most likely to benefit from ongoing support.

- Conduct hospital-based screening of all births for health and developmental needs to enable prioritized service referrals for the moms and babies with elevated risk factors. For families of substance-exposed infants that receive a Plan of Safe Care, jurisdictions should monitor enrollment in referred services and offer follow-up to families that fail to engage; in too many places, these plans are simply pieces of paper shared with families rather than roadmaps that drive ongoing support. 

- Integrate performance expectations in provider contracts to reward enrollment of hard-to-serve clients and prompt the development of new outreach tactics. This can involve offering bonus payments for completing face-to-face visits with high-needs clients, requiring a minimum share of visits to be completed with clients in hotspot neighborhoods, or simply including a fixed amount of funding for outreach activities in communities where engaging clients requires substantial provider staff time.

- Use active contract management strategies to meet with family home visiting providers at high frequency to review referrals and outcomes data, reengineer systems in order to eliminate dropped handoffs, and flag families for ongoing follow-up when initial engagement fails. 

- Provide hotline screeners and child protection investigators with tools to seamlessly connect at-risk families to community-based prevention services when those families will not receive ongoing child welfare services. In particular, pregnant women who come to the attention of the child welfare agency should be considered for referral to prenatal family home visiting services. This often requires coordination with the public or behavioral health systems in jurisdictions where pregnant women fall outside of the statutory authority of the child welfare system. For example, in Michigan, a “birth match” system identifies when a parent who previously lost rights to a child has given birth to a new child and automatically assigns workers to swiftly assess the infant’s wellbeing and evaluate the risk of future harm.4

Case Study: Family home visiting in South Carolina

South Carolina’s Medicaid agency designed performance payments to incentivize community providers of Nurse-Family Partnership (NFP) to enroll hard-to-serve mothers as part of a 2016 expansion of the program that paired nurses with an additional 4,000 first-time, low-income moms. The state wanted a path for NFP to serve those mothers most at risk, who might not otherwise connect with home visiting services. To do this, project partners looked for ways to focus on high-poverty areas and ultimately agreed on an additional layer of enrollment targets for zip codes with high concentrations of poverty. The agency set a target for the percentage of enrollees living in low-income zip codes (poverty rates above 15 percent): if at least 65 percent of mothers live in these low-income zip codes at the time of enrollment, the agency pays providers a per-participant bonus payment. This payment structure has prompted providers to reorganize their outreach teams to focus engagement activities on neighborhoods with high needs and low current enrollment.5

South Carolina Case

#2. Systematically review child fatality trends to uncover earlier opportunities to intervene

“If we as a nation do nothing different to prevent child abuse and neglect fatalities, somewhere between 1,500 and 3,000 U.S. children will die from maltreatment in 2016, 2017, and beyond. We need to dramatically redesign our approach to eliminate child abuse and neglect fatalities.” - David Sanders, U.S. Commission to Eliminate Child Abuse and Neglect Fatalities6

As many as half of all child maltreatment fatality victims die without ever having previously come to the attention of the child welfare agency.7 Yet reviews of child fatalities and near fatalities in most jurisdictions focus only on children who were actively or previously involved with the child welfare system, missing opportunities to learn from the cases that never made it to protective services. And it is rare for these analyses to lead to systematic reforms: a survey conducted by the National Center for Fatality Review and Prevention Research found that only 10 percent of recommendations emerging from fatality reviews are ever implemented.8

Model approaches

A comprehensive strategy for preventing child abuse fatalities and serious injuries should focus on determining where, how, and for whom prevention interventions can be delivered more effectively. This involves examining outcomes for the entire population of at-risk children in a jurisdiction – not just those already involved with the child welfare system – and coordinating across public agencies beyond child welfare to identify earlier opportunities to link high-risk families to preventive interventions.

Jurisdictions should establish a centralized database that tracks all fatalities and near fatalities associated with maltreatment and regularly conduct analysis and improvement discussions organized around three key sets of questions:

- How effectively did the jurisdiction identify at-risk families prior to these critical events? How many victims experienced a fatality or near fatality who were not previously reported to the child welfare agency or known to be at risk by another agency? What is known about each family’s prior involvement in prevention services?

- Among families who are known to be at risk, how can those families be more consistently referred to and enrolled in appropriate prevention services?

- Among families enrolled in services, where are there opportunities to intervene earlier, reduce the variance of their persistence in care, improve service effectiveness, or match families to different service models?

These analyses should segment fatalities and near fatalities by the nature of prior interactions across the child welfare system, healthcare systems, family home visiting programs, and/or other prevention and family support resources.

For example, a tree analysis segments families who have suffered critical incidents by the nature and extent of their involvement with the child welfare system at large prior to the fatality or near fatality (see illustration below). This analysis can help agencies diagnose where opportunities for improvement exist by indicating where in the system at-risk families are falling through the cracks: Are they not being identified as at risk? Are they being identified appropriately, but not receiving the proper referrals? Have they been previously involved with the child welfare agency but are left without appropriate supports when their cases close?

Illustration of a tree analysis examining prior involvement with child welfare system

Tree Analysis

Other graphical analyses can also illustrate the nature of system challenges in ways that reveal appropriate policy responses. For example, a second approach charts fatalities and near fatalities by age and incident type to understand whether children at different ages are more vulnerable to certain kinds of maltreatment fatalities (see illustration below). This analysis can enable an agency to focus its work on a specific subset of the at-risk population and narrow possible response strategies according to the types of incidents that are most common.

In a third type of analysis, agencies can produce for each victim a graphical timeline that maps prior interactions with the child protection and prevention systems relative to the date of the critical incident. Such timeline charts can be powerful in communicating how frequently and recently families had been seen prior to the incident and by whom.

Additional tools, such as the Praed Foundation’s Safe Systems Improvement Tool used by Tennessee’s Department of Children’s Services,can help systematically catalog causes and contributing factors underlying fatalities and near fatalities to determine where systemic reforms may be needed. This approach borrows from the field of “safety science” pioneered to reduce the occurrence of catastrophic events in aviation and healthcare.

Illustration of an age and incident-type chart

Incident Chart

Regardless of the analytic approach used to generate insights about improvement opportunities, findings should be linked to prioritized recommendations about operationally-feasible opportunities to improve prevention systems. After selecting interventions to pursue, leaders from child welfare and public health systems should convene at least quarterly to actively review metrics associated with the installation of reforms and broader child safety outcomes in order to monitor progress, troubleshoot barriers, and identify opportunities for additional systems reengineering.

A detailed overview of this approach is described in the GPL’s technical guide, Uncovering earlier opportunities to keep children safe: A data-driven prevention approach to reviewing and responding to child abuse and neglect fatalities.10

Case study: Maltreatment fatality prevention in Rhode Island 

Rhode Island’s Department of Health and Department of Children, Youth, and Families have formed an innovative collaboration based on a data-driven approach to keeping kids safe. Launched in 2017, this work included matching child welfare, family visiting, and Medicaid data to identify when children interacted with the state before an incident of maltreatment, partnering with university researchers to study factors that predict child maltreatment, and selecting interventions that reflect that analysis. The state implemented four priority strategies to strengthen prevention and protect children: developing additional risk tiers for the Newborn Developmental Risk Screening to identify families at birth for outreach by the state's family home visiting services, strengthening engagement with pregnant moms reported to child protective services, introducing clear referral criteria and processes to help investigators refer to appropriate prevention services (including those administered by other public agencies), and monitoring agency progress in making those warm handoffs successful.11

#3. Improve effectiveness of screening and investigatory decision-making through regular adjustments based on outcome analysis

Within the bounds of their state policies, most child welfare agencies give hotline screening staff and safety investigators latitude in making decisions about when to screen in a call for investigation and when to remove a child for placement into foster care. Despite the expertise of these professionals, some families will unnecessarily experience maltreatment investigations, and some at-risk families who would benefit from additional services will be missed. Agencies rarely review the outcomes of families in a way that is linked back to the initial screening and investigation decision so that frontline decision-making can be adjusted or made more consistent. 

Model approaches

Agencies should provide hotline screening staff and safety investigators analytically-driven decision support tools that can help them make more consistent safety and risk determinations, as well as mitigate biases that contribute to disproportionate exposure to the child welfare system among families of color:

- Structured decision-making tools, such as those designed by the National Council on Crime & Delinquency’s Children’s Research Center, can help staff organize case information and more consistently assess safety and risk in determining when to do an investigation and when to remove a child for placement into foster care.12

- Simple heuristics – rules based on case history – can standardize screening decisions for children at particularly high risk of subsequent maltreatment. Such rules can be generated from analysis of prior outcome data that identifies profiles of families or children at particularly high risk for injury. For example, Michigan has a multiple complaint policy that automatically screens-in for investigation any report where there have been two prior reports involving the same home with a child under the age of three years.13

- Predicative analytic algorithms can combine information about similar families already in agency datasets with new information from the call. These tools can be used to provide hotline workers with real-time information that augments traditional risk scores. Such tools need to be carefully constructed to avoid engraining racial or other biases that may be reflected in the historical dataset.

- Hotline screeners and child protection investigators can often benefit from additional resources that enable them to easily connect at-risk families to community-based prevention services when those families will not receive ongoing child welfare services. Such resources can include information packets with eligibility and contact information for community-based programs, phone technology that enables hotline screeners to redirect callers to the local 2-1-1 helpline, and parent partners who can follow-up with families to provide more information on potential supports. For example, Washington State has created Child Welfare-Early Learning Liaison positions serving 5 counties to help investigators and Family Assessment Response staff connect families with early learning programs.14 

To improve performance in real time, these tools should be supported by active, data-driven reviews of worker- and unit-level outcomes data. For example, hotline staff and supervisors in New Hampshire regularly look at call processing time by worker and the daily queue of outstanding reports. To streamline the process for getting reports into the hands of investigators so families can be contacted more quickly, the team has used this data to uncover what techniques for handling calls work best and institutionalize those practices across the unit.15

In order to assess front-end decision making, an agency might also regularly compare subsequent safety outcomes (new reports of maltreatment, child deaths, child injuries) for cases just below the thresholds for doing an investigation or opening a case to those just above the thresholds. If subsequent safety outcomes are poor for those just below the thresholds, then the threshold should be lowered. If subsequent safety outcomes are very good for those above the threshold, it may be worth experimenting with raising the threshold to see if those families do similarly well with a more limited set of in-home services but without a formal case opening. To facilitate such analysis, it may be necessary to collect data on the subjective assessments of risk levels by agency staff making the decisions so that cases that are near the threshold can be identified.

“We are encouraged that the Allegheny Family Screening Tool has shown positive results by increasing accuracy, while preserving clinical judgement, and we believe that it has great potential as we continuously strive to improve our ability to keep children safe.” - Marc Cherna, Director of Allegheny County’s Department of Human Services

Case study: Hotline call screening in Allegheny Co.

Beginning in 2014, Allegheny County’s Department of Human Services in Pennsylvania collaborated with a team from Auckland University of Technology led by the Co-director of the Centre for Social Data Analytics, Rhema Vaithianathan, to develop an open-source predictive-risk modeling tool that would help improve child welfare call screening decisions. The department sought to enrich the information used by hotline staff at the entry point to child welfare, when allegations are either screened out or referred for investigation, in order to better protect children and increase consistency across staff.

The resulting Allegheny Family Screening Tool (AFST) links existing information from twenty-one sources in the county’s integrated data warehouse, including child protective services, publicly funded mental health and drug and alcohol services, and bookings in the County jail, to inform screening decisions about which calls to investigate when allegations of maltreatment are received. When a report comes in, the tool analyzes prior information about involved individuals to produce a risk score that estimates the likelihood of a future removal into foster care placement. Since the screening tool has been implemented, the county has observed encouraging results: screen-in rates have increased for higher-risk children who needed intervention supports, and disparities between Black and white children in case opening rates have decreased.16

#4. Refine maltreatment reporting campaigns by comparing child protection trends with Medicaid data

In 2017, there were 4.1 million allegations of child maltreatment reported in the U.S., yet research indicates that a large share of the children who experience maltreatment never receive official attention.17 In most states, no one is responsible for analyzing whether the child well-being system as a whole is reaching the entire population of children and families requiring protection or support. In annual data released by many state child welfare agencies, the first table often shows trends over time in reports of alleged child maltreatment. If this trend is downward, it is interpreted as progress. But reports of maltreatment can decline either because maltreatment declines or because reporting rates decline. When reporting improvement campaigns do occur, they can contribute to the disproportionate interactions experienced by families of color.

Model approaches

Agencies should regularly compare the volume of maltreatment reports received by the child welfare agency with the total number of child injuries accounted for in state Medicaid billing records. If the ratio of maltreatment reports to child injuries is declining, it may be a sign that reporting needs to be improved. Conversely, if the ratio of maltreatment reports to child injuries is increasing, reporting may be improving. It may also be worth considering if these trends are caused by unnecessary surveillance of some families. This analysis should segment trend data to identify geographic areas, demographic groups, or families with certain characteristics – such as families without children enrolled in childcare – that are more likely to be missed by the system or experience discriminatory surveillance. This information can be used to refine messaging in public education campaigns and adjust training for mandatory reporters.

Illustration of using child injury data to examine reporting effectiveness

Child injury data

Additionally, case reviews of preventable deaths and serious injuries in families that were not previously known to the state human service agencies should ask why these families were not previously identified as needing services and what services might have prevented the incidents.

As jurisdictions respond to the COVID-19 pandemic, this strategy may be helpful in assessing the extent to which reductions in the number of calls to the child welfare hotline may be leading to cases of abuse or neglect going undetected. Findings from this analysis can inform agency strategies for educating community members about how to look out for children’s and families’ safety and about family support resources that are available.

Case study: Medicaid records in Massachusetts

In 2017, Massachusetts used state Medicaid records to discover 260 serious injuries occurring in children in state care during the prior two years that were not previously known to the state child welfare agency. The state matched data from the roster of children in Department of Children and Families custody with medical records extracted from the state’s Medicaid medical information system. It identified previously unrecorded incidents including physical assaults, injuries that resulted from the use of weapons, drug overdoses or poisonings, suicide attempts, fire-related injuries, and severe burns or bone fractures. The department used this information to inform the trainings it regularly offered to mandated reporters throughout the state.18

Improving effectiveness of child welfare interventions

#5. Rebalance service mix by segmenting subpopulation needs and tracking unmet demand in addition to past utilization

Child welfare agencies often fall into patterns in which they contract every year for the same set of services with the same set of providers, resulting in funding programs that may be underperforming or are no longer meeting the needs of the community. When agencies do seek to adjust their services, two challenges are common. First, many jump immediately to selecting services without carefully assessing the different subpopulations they seek to serve and the distinct needs of each. This can result in a service array in which the right services are not available for many clients or services do not have the right components (for example, failing to require programs include transportation resources in rural communities where public transit is unavailable). Children and families with specialized needs, such as victims of trafficking, can have worse outcomes when the only resources available to them have not been tailored to their distinct circumstances.

Second, decisions about how much of each service to purchase often fail to account for underlying (and unmet) need for each service type by relying on past utilization data alone. Agencies inadvertently underestimate need for a service when they don’t adjust prior utilization rates upwards by the number of clients they would have preferred to refer but instead directed to a second-choice program due to waitlists at the best-fit option. They also overestimate the future need for a service when they don’t adjust prior utilization rates downwards by the number of clients enrolled because the first-choice option was unavailable.

Model approaches

Successfully rebalancing the mix of contracted services available to help children and families requires agencies to make investments based upon each subpopulation’s needs and volume projections that augment prior utilization data with waitlist information:

- Uncover gaps in the service mix available to clients by using data to segment the population by needs and geography, and then map those needs against the current mix of services available, including those funded outside of the child welfare agency. It is often useful to supplement this data with insights from frontline staff and individuals with lived experience as clients. Agencies can also compare outcomes for similar individuals referred to different services to inform decisions about which services are most effective (and most cost effective).

An example of a service array designed around segmented client needs is included below. In 2016, as it planned to enter new contracts for prevention and placement services, Rhode Island’s Department of Children, Youth and Families (DCYF) identified fifteen subpopulations with distinct service needs and goals. For example, population 2A in the chart below reflects the group of intact families with open child protection cases for whom the department sought services that would safely prevent removals into out-of-home care. For each of these subpopulations, the department described specific performance objectives, identified common specialized needs, and noted the potential volume of clients to be served.19

Illustration of service array designed around segmented client needs

Segmented client needs

- When it is time to estimate the volume of each type of service being purchased, combine data on past utilization trends with analysis of prior unmet needs. To do this, agencies must systematically track cases in which a child or family is referred to a less optimal service because the first-choice service did not have available slots. This can be done by revising referral forms to record first, second, and third preferences for services and logging that information in a way that can be easily aggregated (see illustration below).

Recording service preference






This data can then be analyzed to identify the services for which there was greater need than availability (to procure more of them) and the services utilized in the past only because there was not a better-fitting resource available. Additionally, identifying programs with excess capacity can uncover contracts from which freed resources can be reinvested into other programs.

Case study: Family-Based Recovery services in Connecticut

Facing a growing volume of cases with an indication of parental substance abuse driven in part by the opioid crisis, Connecticut’s Department of Youth and Families (DCF) sought in 2015 to better align its services with the needs of these families. Supporting parents and caretakers during recovery could preserve family stability and reduce the number of children requiring removal from their families. In analyzing the needs of the 18,000+ families investigated by DCF where substance use was indicated, the department discovered that family-oriented services for parents with substance use disorder were unavailable to families with children aged three to six years old; they were available only to families with children under three years old or over six years old. However, children in this range comprised a substantial share of out-of-home placements.

To respond to this gap, DCF collaborated with its provider of Family-Based Recovery services, which combine intensive parenting supports and regular home visits with substance use recovery programming, to adjust their model to be available to this segment of families and dramatically expand the number of slots to better match the increased need. The department also established a systematic process of identifying and flagging when substance use appeared in child welfare cases in order to detect if new gaps emerge in the mix of services available.20

#6. Use results-driven contracting approaches to align procurements and provider incentives with strategic goals

A very large share – sometimes half or more – of the budget of most child welfare agencies is spent on services delivered by contracted providers. Yet, agencies often treat procurements for these contracts as routine processes focused on complying with regulatory requirements rather than as strategic opportunities to produce better results.

These governments contract for services without identifying the strategic purpose they are trying to achieve and simply pay for slots in services. They usually fail to measure the outcomes achieved by contractors or to build effective outcome reporting or performance incentives into contracts. Conventional solicitations also rarely offer community stakeholders, including service providers and individuals with lived experience, meaningful opportunities to shape the programs that agencies purchase. Agencies often require providers to deliver predetermined service models, leaving little room for innovation. Complex procurement processes frequently create barriers for smaller organizations to win business, including those managed by women and people of color.

Model approaches

Major procurements with service providers should be regarded as strategic opportunities to advance the chief objectives of the department:

- Regularly prioritize the most important contracts and procurements for leadership attention, and consider prior performance when making contracting decisions. This requires strategic planning to assess service needs prior to writing a formal solicitation, identifying goals for the procurement, and selecting the procurement strategy, contract type, and compensation structure that best align with those priorities.

- Publicly engage stakeholders, such as by issuing a request for information (RFI) or holding conferences with prospective bidders, to understand the capabilities of the vendors, boost competition, and generate community input. Additional techniques for gathering stakeholder input are described in the GPL’s technical guide, Twelve strategies for gathering constructive input to improve your RFP.21

- Generate new solutions through problem-based procurements that emphasize essential functionalities and objectives over prescriptive specifications. This more open-ended approach enables vendors to propose innovative solutions that can better meet the needs of families in their community.

- Simplify procurement processes for all vendors and dedicate technical assistance resources and outreach efforts to increase vendor diversity.

Contracts also offer opportunities to increase the odds that performance improvement efforts stick:

- Establish common performance metrics and clear definitions of client success for each program.

- Integrate into contracts “micro” incentive payments – in some cases, only 1 percent of a total contract’s value – that bind the department into producing provider-specific performance information over time and enable providers to make adjustments based on long-term outcomes information about the clients they serve.

- Make future contracting, funding, and referral decisions based on past results for similar clients across like providers.

For a more detailed overview of this approach, additional case studies, and implementation resources, visit the results-driven contracting page of the GPL’s website.22

Case study: Procuring home and placement services in Rhode Island

In 2016, Rhode Island’s Department of Children, Youth, and Families (DCYF) completed a results-driven procurement that resulted in 116 new contracts collectively representing approximately $90 million of services per year. DCYF’s solicitation reflected its redesigned service array, which was organized around fifteen outcome-based service categories that were centered on subpopulation needs and tied to specific performance objectives. For eight of these subpopulations, the procurement asked providers to propose the programs that would best enable children and families to achieve the identified outcomes of interest.

The flexible nature of the solicitation leveraged the expertise of local experts and community providers to offer programs not previously considered by DCYF. As a result, DCYF made critical expansions to its family-based services and foster care resources and is innovating with new programs not previously available in Rhode Island. Since completing this procurement, DCYF has achieved a 66 percent increase in the number of contracted family-based foster homes and a 23 percent reduction in the share of foster children living in group settings, representing a 3.5x increase in the pace of reform.23

RI foster care placements

In 2018, Rhode Island’s procurement was highlighted as the nation’s single leading example of contracting for outcomes by the organization Results for America in its 2018 Invest in What Works State Standard of Excellence.24

#7. Make seamless handoffs that connect families to the best-fit services

Connecting families to the right service at the right time can often make the difference in the likelihood and speed with which they can safely and permanently exit the child welfare system.25 For this to happen, two sets of activities must go smoothly: families need to be referred to the service that best meets their needs, and families then need to swiftly enroll in those services.

However, with high caseloads and annual social worker turnover of more than 40 percent in some jurisdictions, making service decisions that best meet family needs is often challenging. Information on provider capacity, results, or cultural competency for clients is rarely available to frontline staff, so they frequently depend on word of mouth – and a single bad experience can prevent a worker from ever referring to a given provider again. Additionally, waitlists or inefficient referral and/or intake processes may mean families wait weeks or months to begin a service after needs have been identified. Even when services are mandatory, some families will never make it to the services to which they have been referred. Agency staff have limited time to follow up to see if families have enrolled in services, and providers don’t always follow up with the referral source when clients fail to show up.

Model approaches

Agencies can reengineer their systems for matching, referring, and enrolling clients in child welfare services to best meet family needs, prioritize those clients most in need of services, speed up time to service delivery, and minimize the share of referred clients who fail to receive services due to dropped handoffs.

systems matching

To increase the share of families referred to the right service, agencies should:

- Identify the characteristics of families who will benefit most from each service type by analyzing historical data for the entire population at risk of negative outcomes and formulating strategies about the set of program interventions that could improve outcomes for each segment of the population based on available evidence.

- Assess child and family risks, needs, and cultural characteristics in order to make determinations about referral decisions based upon what is most likely to enable a family to safely exit the system. Despite the large number of assessments most families in the child welfare system experience, few assessment tools result in a prioritized set of needs that services should address.

- Offer decision support resources that help caseworkers match clients to the services that can best address their needs. Many agencies employ service experts who consult with social workers on individual cases. Other agencies have created online matching tools that suggest best-fit services for a family based on a basic set of information entered by a caseworker. We have observed a handful of agencies that are also performing sophisticated analysis to figure out which programs have historically been most successful for which families.

- When program slots are limited to or designed for only the highest-need families, agencies may also want to establish mechanisms to systematically allocate these resources to the populations who will benefit from them most. For example, Annie E. Casey Foundation’s Child Welfare Strategy Group has helped many jurisdictions implement processes that require commissioner’s office signoff for any child to be placed into group care in order to prompt workers to fully explore family-based options.26

To increase the share of referred families who engage in services, agencies should:

- Design processes that make warm handoffs from agency staff to service providers rather than simply providing clients with a flyer or written information. We have observed jurisdictions invite providers to co-locate intake staff in agency field offices, require caseworkers to introduce families by phone with provider intake staff before a referral can be marked complete, and staff field offices with service navigators responsible for helping families understand and access the services to which they have been referred.

- Track service receipt in real time and collaborate with service providers to persistently reach out to those families who have not yet enrolled. This may require adding client-specific information to the data that contracted providers report to the agency or increasing the frequency of such reports. When that is not possible, agencies have used provider invoices to identify which families have and have not enrolled in services, although this process can have limited utility due to slow billing cycles.

Additionally, active contract management strategies should be used to drive ongoing adjustments to service matching and delivery mechanisms. Each month, the agency program lead should review data on the percentage of referred clients who actually received services and meet with providers to review case files of clients who failed to receive services. Identifying common barriers to service enrollment can suggest process changes to improve the fraction of the target population that is reached as well as the share that complete programming. The agency should also review cases with bad outcomes that were not referred to services and analyze what can be done to more consistently flag families who need additional help.

For a more detailed overview of this approach, additional case studies, and implementation resources, visit the service-matching and referrals page of GPL’s website.27

Case study: Enhanced service coordination in Connecticut

Connecticut’s Department of Children and Families has reengineered the way that social workers match families with services, addressing two problems: social workers were often matching families with the most familiar services rather than the services that best fit the families’ needs, and they were not recording the first-choice service for a family if no slots were available, meaning the agency did not know what services it needed to expand to meet demand. The department established a new system, enhanced service coordination, for assigning families to services that includes service navigators for a subset of family preservation services and a revamped universal referral form that focuses the referral process on determining which services are a best fit for the family's needs. By substantially reducing paperwork burden on staff, the project generated buy-in from frontline staff who might otherwise have been resistant to process changes. Initial findings have indicated that this enhanced service coordination has resulted in an improvement from thirty days to ten days in the average time it took for families to be referred to family preservation services following the identification of a need.28

Six questions to diagnose barriers in child welfare referral systems


Leaders of public child welfare agencies frequently ask us for advice about how to improve their service delivery systems so that the families they work with are consistently identified, referred, and enrolled in the right services. We have discovered six common questions that every government can ask to uncover gaps in referral systems and generate fixes that improve outcomes for their clients.

Six questions


1. Identify: Are you overlooking clients who may be a good fit for services?
2. Assess: Are you accurately assessing clients in a way that prioritizes the supports they need?
3. Match: Are you matching clients to the best-fit services for addressing identified needs?
4. Connect: Are you preparing clients and communicating with providers in a way that allows for a seamless handoff into services?
5. Enroll: Are providers engaging every client in services quickly after a referral has been made?

6. Sustain: Are providers following up and attempting to re-engage clients when they fail to show up for services?

​​​​​​​For more, visit the blogpost on our website, Six powerful questions every government ought to ask to diagnose problems in social service referral systems.

#8. Drive real-time improvements to service delivery through active contract management

Despite the many critical functions that involve work with contracted service providers, most public child welfare agencies fail to actively manage provider performance once contracts are executed. These agencies often act as though their role is complete once a contract is signed and shift to a narrow focus on making referrals, processing invoices, enforcing compliance, and occasionally assessing past results. However, some of the most important work for government occurs during the course of the contract, when ongoing adjustments to service delivery can produce better outcomes for the families being served.

Model approaches

Active contract management (ACM) transforms the relationship between a public child welfare agency and its providers into a partnership with shared responsibility for achieving common goals. Specifically, rather than only occasionally reviewing retrospective program data, agencies adopting ACM frequently review real-time performance data. Executive staff hold regular, collaborative meetings with service providers in which they use this data to identify barriers experienced by clients and design solutions for implementation.

We have observed three ways ACM can improve results:

- High-frequency troubleshooting: Real-time identification of operational challenges followed by immediate course corrections.

- Incremental improvements: Continual refinement of agency and provider practice to produce rising performance trends over time.

- Systems reengineering: Reengineering of service delivery systems to generate comprehensive remedies that dramatically improve performance.

ACM photo

On a monthly or bimonthly basis, public child welfare agencies should convene executive and program staff from groups of similar service providers to examine incremental progress against critical performance metrics indicative of longer-term success.

Emerging drops in performance can be swiftly identified and responded to, and providers can learn innovative practices from each other when brainstorming together on ways to improve results. Agencies can learn about the types of families each provider and program does best with and use that information to adjust referral patterns. Complementing these frequent discussions of key metrics with in-depth analytical attention or targeted case reviews can direct attention toward topics and practices that are critical to success but may not be regularly reported or reviewed.

For a more detailed overview of this approach, additional case studies, and implementation resources, visit the active contract management page of the GPL’s website.29 Among the resources published there is a technical guide that provides six tools for implementing ACM:30

1. A worksheet with ten planning questions for launching a new ACM practice
2. Examples to help agencies select leading and lagging performance metrics
3. Guidance for prioritizing a roadmap of performance topics for in-depth analysis
4. Three simple data techniques for revealing performance patterns
5. Strategies for fostering a collaborative, trusting ACM practice
6. Checklist of elements for maintaining an effective ACM practice

Case study: Prevention programs in Rhode Island

In early 2016, Rhode Island’s Department of Children, Youth and Families began using active contract management strategies with the state’s four regional providers of community-based prevention services for families at risk of maltreatment.31

Each month, agency leaders from programmatic and contract oversight units met with executives and program staff from all four providers to review performance data, discuss client outcomes, and share ways to improve results. Early in this process, the department and providers clarified a more explicit definition of success focused on preventing families from subsequent involvement with child protection; previously, this measure was buried in the middle of a quarterly data report.

Monthly meetings focused on developing strategies to more consistently engage all families referred for services rather than those most proactive about participating. Provider executives were expected to come prepared to each meeting with information on the enrollment status of every client referred to them and, for each unenrolled client, to know what outreach efforts had been tried by provider staff. When referral volume unexpectedly spiked, the department and providers found that waitlists were shorter at the provider where its program administrator had temporarily assigned herself cases to manage; this solution was quickly established as the norm for all providers – an example of a best practice being identified and rapidly spread across providers. In addition, the department collaborated with providers to revamp expectations for the first thirty days of support with an increased focus on stabilizing immediate risk factors before beginning longer-term service planning, as narrow adherence to the original wraparound-based practice model created barriers to meaningful engagement with families in crisis. In between monthly meetings, agency and provider staff collaborated to implement these changes.

Rhode Island has reduced the share of these clients that subsequently open to child protection from around 8 percent over the twelve months prior to these active contract management strategies to approximately 3 percent. As illustrated in the below chart, the share of families requiring state intervention fell steadily as ACM was launched. In early 2017, removals began to increase across the system; however, removal rates for families involved with these community-based prevention services increased less than removal rates for the population as a whole before returning to prior levels.

RI improvement
Rhode Island’s child welfare department has since implemented this management approach with its providers of group care and has plans to spread these strategies across all agency-contracted services. Child welfare agencies in Arizona, Connecticut, Illinois, Florida, and Riverside County, CA have also implemented active contract management strategies.

Five design features for actionable performance dashboards

There are five elements that should be part of every performance dashboard to help agency leaders spark operational changes that can immediately improve results. These features are relevant for dashboards built both for active contract management of agency providers and performance improvement activities that are internally focused.

1. Time series of the data that shows trends at monthly intervals over time (ideally, going back at least two years). This enables leaders to easily notice when trouble emerges and to monitor if operational changes produce the intended improvements.
2. A benchmark, target, or reference line that allows leaders to contextualize performance and determine the urgency of possible reforms.
3. Disaggregation by operationally meaningful subunits (such as individual field offices or service providers), aiding leaders in identifying subunits with stronger practices that can be spread as well as subunits that may need additional support.
4. An explanation of how performance on this measure influences the lives of children and families. This facilitates the design of solutions focused on producing better outcomes rather than tighter compliance.
5. Discussion questions and guidance for interpreting trends, which together enable leaders and their teams to swiftly turn their attention to considering operational changes. Developing effective questions for discussion often requires substantial pre-analysis by agency staff who have programmatic expertise.

All five of these design features are illustrated in the example dashboard in Solution #9. 

#9. Accelerate permanency by using data to uncover stalled cases and address systematic case management barriers

Many families open to the child welfare system don’t receive additional attention when case progress towards permanency stalls. A child’s time in out-of-home care can be unnecessarily extended due to delays in updating case goals, adjusting the services in which a family is enrolled, or submitting paperwork to the courts.

Model approaches

Integrating data into daily case management and supervision activities can help frontline staff proactively find new solutions for cases that aren’t making progress and shorten the duration of families’ involvement with child protective services. Such strategies include:

- Automatically flag cases for additional attention using administrative data to identify when progress of children or families stalls, such as indicating when parents of children in foster care have dropped out of reunification services or when a child has been in a foster care placement for longer than six months without a home visit.

- Set up procedures to review every case at thirty, ninety, and 180 days and record, at each point, why the child has not yet reunified. Systematically tracking barriers to permanency at pre-determined milestones can uncover new strategies for reform. For example, were an agency to discover that reunification rates stall after children are in care for more than ninety days, it might test ways to increase the frequency of family visits at this stage of cases or collaborate with providers of reunification supports on new strategies to maintain family engagement in services over time.

- Consider “nudges” based on behavioral economics principles that require periodic justification for intact family cases to remain open rather than the typical approach of requiring additional documentation only to close a case.

- Offer providers access to administrative data to help them identify children who linger in their care and prioritize these children for revisions to service plans.

We are aware of at least one agency that created an internal-facing version of active contract management to address shortfalls in its permanency outcomes.32 Other jurisdictions have adapted ChildStat, a management process pioneered in New York City that has been highlighted by the Annie E. Casey Foundation’s Child Welfare Strategy Group as a way to regularly bring together staff from multiple agency departments to jointly discuss cases and data trends for the purpose of finding opportunities to improve field practices and outcomes.33

With any of these strategies, it can be powerful to provide supervisors and administrators with dashboards and to set up a high-frequency cadence for executive and field staff meet to discuss progress, troubleshoot barriers, and conceptualize operational changes.

Below is an example dashboard for a field office administrator. This dashboard shows the frequency with which children in out-of-home care are visiting with their parents. The left chart shows the share of children statewide with 4+, 1-3, and no visits each month for the last three years. The right chart shows the share of children in each regional unit with 4+ visits each month during the same period. Such dashboards simplify the analysis of trends by caseworker, service types, and child/family characteristics to reveal systemic barriers to permanency, uncover local innovations, and identify caseworker units or workers who may need additional coaching or support.

Example of an administrator’s management dashboard for family visitation

Dashboard example

Case study: Rapid Permanency Reviews in New York City

New York City’s Administration for Children’s Services (ACS) partnered with Casey Family Programs to implement “Rapid Permanency Reviews” of the cases of children who have been in a stable family-based placement, yet linger in foster care, perhaps due to systemic barriers or administrative delays. At the case level, review team members used information from case and court records to identify and address administrative bottlenecks at critical case junctures and ensure case management activities are advancing each child’s permanency goal. Aggregated information was discussed with other systems (such as the courts and housing authorities) to resolve system-level barriers. The reviews along with other ACS initiatives have contributed to reduced length of stay in foster care and improved permanency outcomes.34

#10. Strengthen foster, kin, and adoptive resources through analytically-informed recruitment, matching, and retention

When children are temporarily removed from their parents to keep them safe, they most often do best when they can live with their relatives or foster families rather than in institutional settings. However, caseworkers often lack the time or expertise to identify a relative caregiver for every child, and many agencies struggle to maintain sufficient volumes of foster families. Nationally, 56 percent of foster families quit within one year.35 As a result, too many children in foster care are living in group settings rather than with families – a trend that is especially acute for older youth and children of color.36

Model approaches

There are many opportunities to strengthen recruitment, licensing, and retention of kinship and foster families by better using data in program design and delivery. One set of approaches can be deployed to optimize existing processes:

- Conduct market segmentation analysis to reveal common profiles of families most likely to become foster parents by examining characteristics of families in the current pool. Agencies can then target recruitment campaigns to those most likely to successfully foster, as well as craft distinct recruitment messages based on different profiles of prospective families. Such analytic techniques can be especially valuable for recruiting a pool of foster families that reflects the racial and ethnic composition of the children requiring placement.

Illustration of stage-by-stage analysis for foster family recruitment

Stage by stage analysis

- Examine the ratios of families beginning and completing each stage of the recruitment, licensing, placement, and retention process to discover where families drop out and to focus reforms at these points. For example, in the above illustration, low application completion rates may indicate an opportunity for mechanisms that periodically follow up with families to answer questions and remind them to complete their applications. Low placement acceptance rates may reflect mismatches between foster family preferences and child needs, suggesting opportunities to adjust recruitment messaging or design supports that increase family readiness to care for a child with challenging behaviors. It can also be helpful to annually survey exiting foster families to identify satisfaction trends and conceptualize retention strategies.

- Map current processes using lean process improvement tools to uncover where operational bottlenecks exist and resolve duplicative or unnecessary steps that undermine swift licensing timelines, placements, or supports. For example, a child welfare agency could collaborate with the state TANF/SNAP administrator to simplify the process for relative caregivers to access supplemental cash and food assistance programs for which they may become eligible when an additional child temporarily joins their family.

- Use data to predict kinship placements most likely to disrupt and proactively target supplemental supports to those families.

Another set of strategies can help agencies create new approaches to supporting foster children in families, including expanding kinship search programs:

- Augment busy caseworkers with kinship specialists who can conduct intensive searches for potential relative placements at the point of removal and periodically after the initial placement. Allegheny County, PA is one example of this approach. The jurisdiction contracts with provider A Second Chance to conduct all search, licensing, training, supervision, and connections to community resources for kinship caregivers. With the support of these kinship specialists, 65 percent of foster children in the county live with relatives37 – a rate double the national average of 32 percent.38 If agencies cannot offer dedicated staff to conduct kin searches in every case – and in many models, caseloads of family search specialists are kept exceptionally low to sustain in-depth attention – agencies should establish mechanisms to identify and prioritize children whose cases would benefit most from additional attention, such as those at the highest risk of entering group care.

- Other agencies are leveraging innovative procurement strategies, such as problem-based procurements, that can generate new solutions by prompting providers to offer proposals that can address a stated outcome rather than simply delivering a preset program model. For example, problem-based procurements could be used to generate new therapeutic foster care programs as an alternative to residential settings or to solicit providers to recruit more racially diverse foster families.

While these strategies may be streamlined by the existence of a technology platform that aggregates and manages foster family information, most can be achieved with the limited data infrastructure that exists in many agencies.

Case study: Foster family recruitment in Arizona

Arizona’s Department of Child Safety (DCS) developed a foster care caseload estimation tool in 2017 to support efforts to increase placement of children in the least-restrictive, most family-like settings. The estimator utilizes historical data to project the number of foster homes to be required in the future, segmenting anticipated needs by geographic and child-specific characteristics, such as age, sibling groups, and specialized care requirements. In order to achieve the target final number, the tool uses prior trends to estimate the number of prospective foster families that need to be reached at each stage of the recruitment process, such as the number of families attending information sessions, enrolling in training, and successfully licensed. The tool has enabled DCS to better collaborate with its contracted foster licensing agencies to recruit appropriate volumes and profiles of families necessary to provide foster children with stable homes.39

#11. Analyze workforce data to improve recruitment, retention, and supervision of frontline staff

The core work of child welfare agencies is delivered by its frontline staff of social workers and their supervisors. Social workers average less than two years in the job, with turnover in some jurisdictions reaching 30 to 40 percent annually.40 Many supervisors receive no formal management training, yet human resource departments in child welfare agencies often focus on administrative activities rather than strategic opportunities to help workers improve client outcomes.

Model approaches

Agencies generate large amounts of data that can be analyzed to help administrators and supervisors support frontline social workers and improve departmental effectiveness. In jurisdictions where staff are represented under collective bargaining arrangements, it may be beneficial to develop these strategies collaboratively with union leadership:

- Equip human resources leaders to predict and prepare for anticipated hiring needs by analyzing turnover trends rather than simply waiting until jobs open to begin planning how to fill them.

- Eliminate hiring inefficiencies and decrease delays that cause positions to remain unfilled for extended periods of time due to bureaucratic hurdles through lean process improvement techniques. Some agencies have experimented with streamlining the beginning stages of the hiring process by consolidating multiple steps into a single recruitment session that is offered periodically to groups of job candidates. In a single day, an applicant can speak with current staff to learn the available roles, submit application paperwork, and complete an initial job interview.41

- Develop predictive tools, based on common experiences and characteristics among existing staff, that enable agencies to proactively intervene with supports for staff at high risk of burnout or unaddressed secondary trauma.

- Disaggregate critical performance indicators by unit to help identify high-performing supervisors with management practices that should be spread to peers and those who would benefit from additional training, coaching, or supports.

- Conduct periodic, confidential surveys of frontline staff to identify managers who need additional training, to uncover and respond to concerns around team culture and psychological safety, and to spot other resource needs. Tennessee’s Department of Children’s Services annually surveys staff to measure critical indicators of organizational trust, safety, and well-being among staff using a “safety culture” survey instrument adapted from the healthcare industry (the survey asks staff, for example, about their comfort disclosing a mistake to their supervisor). Insights from this survey have led to the provision of four-wheel drive vehicles and satellite phones to staff visiting rural communities.42

Casey Family Programs’ policy brief, “How does turnover affect outcomes and what can be done to address retention?,” offers additional strategies for recruiting frontline staff.43 The National Child Welfare Workforce Institute also maintains an extensive library of resources on workforce development topics.44

Case study: Social worker retention in New Jersey

New Jersey’s Department of Children and Families (DCF) has made the recruitment, training, and retention of 4,200 child protection and permanency professionals one of its top priorities. The department collaborates with local colleges and universities to arrange for classroom visits by practicing DCF social workers to educate students on the realities of a career in social work, aiming to reduce the share of new hires that leave the workforce due to uninformed expectations about the nature of the job. The department’s dedicated recruitment unit has also focused on recruiting Spanish-speaking staff to better meet the needs of DCF clients. The office regularly shares job postings with local organizations that serve Hispanic populations and places hiring notices in news publications and public affairs TV programs that target Spanish-speaking communities.

The department’s Office of Training and Professional Development has partnered with the Rutgers School of Social Work and Stockton University to offer trainings to staff throughout the workforce. In FY17, frontline caseworkers averaged sixty-eight hours of continuing education training, and all of the department’s caseload-carrying staff and their supervisors received at least forty hours of training. Newly hired caseworkers participate in a thirteen week pre-service program, consisting of 246 hours of training that includes classroom, simulation, and field work. Over the subsequent eighteen months, new caseworkers are required to participate in online coursework focused on further developing their skills and knowledge, including modules led by New Jersey Coalition to End Domestic Violence. A specialized training curriculum was also produced for casework supervisors. It focuses on developing skills critical for successful management, such as crisis management, having difficult conversations, and interpreting and applying data to their decision-making.

Through this and other work, turnover of caseload-carrying staff in New Jersey has decreased by more than one-third – from 14.7 percent in 2005 to 8.9 percent in 2016 – placing it among the top-performing agencies in the country on this metric.45


Taking a more coordinated approach to improving outcomes

#12. Overcome service delivery silos for families involved with child welfare, behavioral health, juvenile justice, etc.

New strategies are needed for managing the overall well-being of families experiencing simultaneous involvement with child welfare, behavioral health, substance abuse treatment, workforce development, criminal justice, and other social service systems. Resources to address these complex needs are typically administered through siloed state or local agencies with distinct caseworkers, funding sources, and data systems.

Model approaches

Building out linkages between child welfare services and other programs serving families has the potential to allow for more timely interventions, provide a more streamlined experience to families, limit duplicative activities, and address underlying risk factors that contribute to families’ child welfare involvement. Agencies should:

- Strengthen referral mechanisms: Establish a systematic process for swiftly flagging when involvement in other systems occurs or is needed to prevent further involvement in child welfare, and then make seamless handoffs between systems for those services. For example, in many systems, when domestic violence puts a child at risk, the child protection staff may give the parent-victim of that violence only basic contact information for the local domestic violence advocate program – thus putting the responsibility on the individual to make contact. Instead, agencies should seek to increase take-up in these cross-system services through warm handoffs and regular follow-up with additional encouragement if a referral has not been completed. For more on improving referral practices, see the solution described elsewhere in this book.

- Coordinate care across systems: Designate care coordinators who are responsible for client success in navigating across systems when no one is otherwise accountable for whole-family outcomes. Illinois’ Department of Children and Family Services’ dually-involved youth partnership with the Conscience Community Network offers an example of contracting with a service provider to offer this navigation support (see the case study on helping service providers manage with data described elsewhere in this book). Similarly, some child welfare agencies have collaborated with local 2-1-1 call centers to establish a centralized entry point through which families needing help can access supportive resources in the community and from public agencies separate from child protection. Another approach establishes a primary caseworker who represents multiple public agencies, such as the model demonstrated with the Troubled Families Programme in the United Kingdom.46

- Set aside dedicated resources for at-risk families: Partner with other public agencies and community organizations to designate program slots for at-risk families or those already involved in child welfare. For example, New Hampshire’s child welfare agency has worked with the state’s Bureau of Housing Supports and local housing authorities to prioritize federal housing vouchers for youth and families involved with child protection. Through HUD’s Family Unification Program and Foster Youth Independence Tenant Protection Voucher Program, the state has secured dedicated housing vouchers for families for whom housing instability is a primary factor for an imminent removal of their child(ren) or a primary barrier to reunification, as well as for youth transitioning out of DCYF care.47 See the case study from Florida below for an example of reserving slots in substantial use treatment for parents referred by child welfare caseworkers.

- Offer interventions that address multiple needs: Develop new whole-family intervention models for subpopulations with complex sets of needs as alternatives to sending families to multiple distinct services at once. Common candidates for integrated programming include joint substance use recovery and parenting skill services (such as Family Based Recovery48), family shelters and permanent supportive housing that are linked to intensive case management supports, and residential substance abuse treatment that enables children to live in care settings with their parents. Some counties, such as Dakota County and Olmstead County in Minnesota, are also testing integrated service delivery models that bridge traditionally siloed programs.49 See practice 15 for more on whole-family supports.

Case study: Behavioral health referrals in Florida

Florida’s Department of Children and Families estimates caregiver substance misuse is a factor contributing to abuse or neglect in more than half of child welfare investigations, yet in mid-2018, only 7 percent of caregivers were receiving treatment within thirty days of their initial referral to addiction services. Many of these caregivers were referred by child protection workers to Family Intervention Specialists (FIS), a service designed to connect caregivers with suspected substance misuse to assessment and treatment programs.

The department’s SunCoast regional office (based in Tampa) partnered with a group of providers to use active contract management strategies to improve referral, engagement, and treatment practices. They developed a new system for tracking referrals and implemented changes to increase caregiver engagement in treatment, such as creating a telehealth system so that clients would be able to complete assessments from home and reserving priority addiction treatment slots for parents referred by child welfare.

As a result, more child welfare-involved families are engaging in substance abuse treatment and connecting with the services they need more quickly. Over the course of less than a year, the average time between referral and assessment for state-funded services decreased by five days, and the share of caregivers attending treatment services within one month of being referred has doubled.50

FIS referrals

#13. Reduce disproportionality and disparity of families’ experiences with the child welfare system

Communities across the country are seeking to address the overrepresentation of historically marginalized populations in the child welfare system and improve the equity of families’ access to resources, the care they receive, and the outcomes they experience. Children and families of color make up a greater share of the child welfare population than their representation in the general population; such racial disparities often persist at decisions points throughout the lifecycle of a case.51 Researchers estimate that 53 percent of Black children will have experienced a child protective services investigation by adulthood, compared to 37 percent among all children.52 Youth in foster care who are LGBTQ are more likely to be placed in group care and experience placement disruptions relative to their peers.53 Immigrant families involved with child welfare frequently face language, eligibility, and other barriers to accessing services.54

Model approaches

Examining disparity and disproportionality of children’s and families’ experiences with the child welfare system offers agencies a powerful lens for uncovering opportunities for service delivery improvement. Agencies should:

- Set and track progress towards equity, diversity, and inclusion goals for the agency. Integrate performance indicators for these goals into the set of metrics executives and supervisors use for daily management activities, and regularly report publicly on agency-wide progress. This work often starts by examining data disaggregated by race, geography, and other demographic variables at common case decision points and other key parts of the system. Agencies should use these analyses to uncover existing trends, establish priorities for improvement, and generate possible reforms. The callout box below suggests some specific topics agencies should examine.

- Collect high-quality data about the intersection of family identity and agency practice. This often involves making structural adjustments in data systems, such as replacing radio buttons with checkboxes that allow clients to report more than one race or changing data rules to allow for non-traditional family compositions (such as a family with two mothers). In most agency-conducted and sponsored research, racial and ethnic crosstabs should be included. Securing the raw data for reports produced by outside researchers can enable follow-on analysis on equity issues.

Analyzing child welfare systems for disproportionality and disparities

An equity lens can be incorporated into an agency’s standard problem diagnoses and continuous quality improvement processes. Below are some common areas where disparities may emerge that may be helpful starting points for such analyses:

- Outreach and engagement in family home visiting and other prevention programs
- Maltreatment reporting among mandated reporters

- Substance exposure screening for children and mothers at birth
- Risk and safety determinations as part of investigations
- Diversions to alternative response tracks (rather than investigations)
- Referral rates and engagement rates for family preservation and reunification services
Placement type, disruptions, and duration for children in out of home care
- Prescriptions of psychotropic medications for children in out of home care
- Housing, education, and employment for young adults exiting from foster care

- Include client voices and perspectives from the frontlines of service delivery as a critical component of every reform’s problem diagnosis and solution generation work. Diagnoses that reflect clients’ lived experiences with the services they receive are more likely to accurately identify opportunities for reform, and solutions that draw on client input are better positioned to yield impact and stick. Agencies should incorporate client shadowing, user experience interviews, and the co-design of solutions with those with system experience into their process for identifying what to improve and how to improve it. For example, during active contract management, service providers can shadow clients through the intake process to identify barriers to successful client engagement. In our experience, elevating client experience yields insights that are not available through analysis of administrative data (for example, prohibitively long wait times, lack of language support, or incorrect contact information), allowing the agency and service providers to work together to create solutions that are responsive to client needs.

- Provide agency staff and leaders with trainings that enhance fluency and comfort in addressing disproportionality and equity issues. Culturally responsive practices should be integrated into caseworker training, professional development, and supervision so that a parent’s ability to navigate the child welfare bureaucracy is never treated as a heuristic for his or her ability to be a safe caretaker for the child. Data on workforce diversity and language fluency should be regularly tracked to identify opportunities to adjust recruitment, training, and retention initiatives.

- Put in place strategies to align and evaluate the competencies of contracted services with the cultural composition of the populations being served. For example, an agency might use a problem-based procurement to generate new solutions for recruiting foster families of color or establish contractual requirements that providers offer clients free access to translation and childcare during services. Procurement documents can include questions that assess a providers’ cultural humility, diversity, and outcomes with clients from marginalized communities, enabling agencies to identify and contract with providers well-positioned to address disproportionality. Technical assistance can also be offered to develop the capacity of smaller, more diverse community organizations to win business from the state or to support the evaluation of program effectiveness on different subpopulations.55

As examining equity is a core way to strengthen practices systemwide, these practices do not stand alone and should be considered as part of each solution described in this book.

Other organizations have published useful resources on this topic. They include:

- From the Annie E. Casey Foundation: Race Equity and Inclusion Action Guide: 7 Steps to Advance and Embed Race Equity and Inclusion Within Your Organization56
From Casey Family Programs: How can child welfare agencies effectively support LGBTQ+ youth in care?57
From the U.S. Children’s Bureau: Issue Brief: Immigration and Child Welfare58

Case study: Prevention services procurement in New York City

In 2019, New York City’s Administration for Children’s Services released a request for proposals to restructure its $220 million per year portfolio of prevention services as it sought to provide a fuller range of supportive services for families in all five boroughs across the city and to identify providers with competencies in delivering inclusive services to diverse communities. To advance this goal, the agency included “Family Voice, Inclusivity, and Social Justice” as one of five equally-weighted evaluation criteria against which proposals were scored. In addition to responding to specific evaluation questions, applicants were also prompted to submit an example of a family-facing feedback tool that they had used in the past.

To score responses to questions in the “Family Voice, Inclusivity, and Social Justice” category, the agency set out seven expectations for programs:

1. Languages Offered: Contractors would provide culturally and linguistically competent services through staff that were representative of the community served.
2. Addressing Racial Equity: Contractors would recognize and work to redress the historical legacy of current racial inequities that results in differences in application of practices, policies, and experiences of families.
3. Providing Services in Families’ Homes and Communities: Contractors would meet with families and teens as frequently as required by each specific program model, primarily in their homes and communities.
4. Family Team Conferencing: Contractors would partner with the agency in meeting with families to develop plans to keep children safe and achieve permanency.
5. Promoting Family Voice and Choice: Contractors would ensure families were treated with respect and dignity and that families had a voice and choice in every aspect of their service experience, including their service plan.
6. Addressing Inclusivity and Social Justice: Contractors would provide a high quality of service and care that was inclusive of, but not limited to, the history, traditions, values, family systems, race and ethnicity, immigration and refugee status, religion and spirituality, sexual orientation, gender identity or expression, social class, and mental or physical abilities of client populations.
7. Listening and Customer Service: Contractors would provide a high level of customer care and satisfaction to the children and families they serve.59

Sample RFP questions to identify providers with competencies in delivering services to diverse communities

Below are sample questions governments can use in requests for proposals to identify providers with competencies in delivering services to diverse communities. Many are based on questions used by New York City in the procurement example described above:

  • Race, Ethnicity, and Language: Provide data on the race, ethnic, and language makeup of the target community. Provide data on the race, ethnic, and language makeup of all staff and board members at your organization.
  • Inclusion and Equity: Describe how your organization actively builds a culture of inclusion and equity. What is the process for identifying and addressing dynamics of racism and/or other biases within your organization?
  • Individualized and Collaborative Service Planning: Describe how your organization collaborates with youth and families to develop and implement individualized service plans. What tools and methods does the organization use to incorporate family and youth voice into service planning?
  • Shaping Program Design with Community and Families: Discuss how community members were engaged to assess their perception of need and to understand their ideas for services in the community being served. Describe how your organization will partner with families to include their feedback at each stage of service implementation (such as program planning, hiring, engagement strategies, performance evaluations, and quality improvement). 

#14. Build linkages to help young people bridge the gap from foster care into adulthood

In most jurisdictions, no single agency or entity is assigned responsibility for the outcomes of young people after they have aged out of the foster care system. These teenagers – and in particular, LGBTQ youth and youth of color – are at extraordinarily high risk of experiencing difficulties in transitioning to adulthood, such as failing to complete high school or secure quality employment, and they also have high rates of involvement in the criminal justice system. Despite federal funding for transition programs for foster youth, analysis by the Annie E. Casey Foundation indicates only 23 percent receive education support or employment assistance nationally.60 In one jurisdiction that has studied its outcomes, 42 percent of youth exiting foster care had been diagnosed with a substance use disorder and 28 percent experienced homelessness or unstable housing during the 12 months following their exit from the child welfare system.61 Agency silos, competing priorities, and misaligned incentives – where the benefits of more successful adult outcomes accrue outside of the child welfare agency’s traditional scope – make investment in new innovations rare.

Model approaches

Helping young people who are aging out of foster care successfully transition into adulthood requires a coordinated approach that uses data to support seamless handoffs into adult services and tracks outcomes for young people following their exit. Agency activities should include:

- Annually examine ninety-day, one- year, and three-year post-exit housing, employment, criminal justice, and health outcomes for young people aging out of foster care, using data matched across Medicaid, homelessness systems, the labor department, and corrections.

- Mitigate dropped handoffs by sharing referral and enrollment information across systems in a timely fashion and tracking referral completion to identify and respond to young people who fall through the cracks. For example, a child welfare agency might prioritize connections to healthcare for young people on track to age out and systematically monitor to ensure that prior to exit, each has had a visit with an internal medicine primary care physician that can provide ongoing care as these youth transition into adulthood.

- Partner with budget offices to unlock new funding opportunities by braiding financial resources across agency silos to overcome the “wrong pocket problem,” where benefits from preventive interventions accrue to a different agency than the one that made the initial investment.

- Contract with transition-specific intervention models, such as those that begin services prior to exit and continue to offer ongoing care well into adulthood, with performance incentives linked to long-term well-being outcomes. Specialized resources may be necessary for a subset of especially high-risk young people, such as new moms who were in the child welfare system as children or teens with developmental disabilities.

Case study: Youth job training and education supports in Maine

One set of promising demonstrations comes from the Learn and Earn to Achieve Potential (LEAP) initiative. This nationwide project of the Annie E. Casey Foundation aims to improve education and employment outcomes for young people who have been involved in the child welfare and justice systems or who are experiencing homelessness.

Maine's LEAP implementation is led by a collaborative partnership between Maine’s Office of Child and Family Services, the University of Southern Maine’s Cutler Institute, and other regional and statewide partners. The state’s child welfare agency created “Kids in Care” report that identifies youth in foster care eligible for supplemental programming, which is regularly provided to LEAP partners to help them more easily recruit youth for participation. The LEAP partners also collaborated with a Workforce Innovation and Opportunity Act (WIOA) provider of career programming to connect participants with additional funds for training and other assistance. In 2018, the foundation’s Fostering Youth Transitions report found that 94 percent of Maine’s foster care population earned a high school degree by age twenty-one – a rate that is two percentage points higher than for other youth in the state (92 percent) and 18 percentage points higher than the nationwide rate among foster youth (76 percent).62

#15. Provide whole-family supports to the most vulnerable children and families

Many of the families with the most complex child welfare system cases are simultaneously among the most vulnerable to poor health outcomes, criminal justice involvement, housing instability, and poverty. These families often reemerge in different parts of the government system without any single contact keeping track of the different services they are interacting with or responsible for proactively intervening early with families to link them to prevention resources that can mitigate bad outcomes.

Model approaches

Experimenting with efforts to identify the highest-need families in a jurisdiction and to provide holistic, dual-generation interventions offers an opportunity to transform service delivery for the community’s most vulnerable families. Rather than treating each facet of the family’s challenges through a separate, siloed agency, this approach allows a government to help families with multiple challenges at once. Such an approach could include:

- Identify the subset of families who have experienced the most concerning outcomes across different human services systems related to child well-being. This can be done by using administrative data to generate lists of the families who have had the most concerning outcomes over the prior eighteen months in each of three to six domains linked to child well-being (see the illustrative charts below). This might include families with the greatest number of indicated cases of maltreatment, families who have experienced the most days of homelessness, families with the highest utilization of children’s medical services for non-natural causes, families experiencing repeated domestic violence, families with the highest utilization of food assistance, or families of young people who have dropped out of school. These individual lists by domain can then be compared to identify the subset of distinct families who have experienced severe outcomes across multiple domains.

Family subsets

- Design or adapt whole-family interventions for high-risk families. Align on the critical family outcomes on which interventions should focus (e.g., parent wages, pre-K readiness, child injuries) and generate hypotheses about the characteristics of interventions that could best help target population families address these needs. Leverage empirical research and collaborate with providers and community stakeholders to refine the intervention model. Such interventions could include changes to operations internal to state agencies (e.g., establishing one lead caseworker who partners with families across benefit and service systems, creating additional flexibility for the nature of benefits offered, waiving compliance burdens such as minimum work hours requirements) or procuring new service models delivered by community providers.

- Proactively find families with the highest chance of adverse outcomes through new interagency referral protocols. For the subset of families with the highest levels of overlapping needs, identify the first time when a public human services agency became aware of a problem with the individual or family so as to uncover where the first opportunity to intervene likely occurred. Using linked administrative data, develop predictive models that appropriately balance the risks of overproviding and underproviding early intervention services. A simpler alternative may be to ask caseworkers to describe the characteristics of families that the system most struggles to serve. Use this information to design protocols to proactively flag families at high risk of becoming the most complex cases, regardless of where they initially show up in the system, and swiftly connect these families to additional services.

- Track, assess, and actively adjust program delivery. Use real-time referral, engagement, and family outcomes data with active contract management strategies to make regular adjustments to the design and delivery of interventions.

Case study: Gun violence in Delaware

Delaware’s Department of Health and Social Services has begun to develop strategies, including tools and data analytics, that could identify young people at-risk for committing or being victims of gun violence, and proactively engage them to prevent further acts of violence. In 2017, teens in Wilmington, Delaware were injured or killed by gun violence at the highest rate in the country and nearly nine times the national average. In March 2018, the department began an eighteen-month training and technical assistance grant to support technical and legal efforts that will build on a risk-stratification framework developed by the Centers for Disease Control and Prevention (CDC) by linking administrative data from multiple state agencies to identify service needs across specific population groups and geographic areas. The state, along with city and community partners and with assistance from the National Network for Safe Communities, is using this information to develop a group violence intervention that can engage a small number of the most high-risk young people and connect them to social services.63

#16. Regularly offer judges family outcomes data that can improve decision-making by the courts

Court improvement processes and other ad hoc partnerships occasionally lead judges and child welfare executives to collaboratively use data to find ways to improve family outcomes, but few jurisdictions do this on a regular basis. Court data and child welfare data are typically stored in separate data systems, which makes linking judicial activities to family outcomes difficult. As a result, judges can lack information about the long-term results of their individual decisions and about the comparative effectiveness of services that they may refer families to.

Model approaches

Matching family court data with child welfare agency data on a regular basis can help judges learn how their decisions’ outcomes compare to their peers’ outcomes and discover ways to adjust their own decision-making to improve family outcomes. For example:

- Analyze differences between judges in placement duration, court-ordered services, and family outcomes for children and families with similar characteristics to suggest opportunities for peer learning and additional training about the department’s programmatic service array.

- Compare data from judges newly appointed or rotated into the family court bench to data from their more seasoned peers to expose areas for additional orientation or training.

- Examine and discuss areas where judicial and agency decision-making is frequently misaligned, such as trends related to court-ordered service referrals that overrule caseworker recommendations or cases where social workers and service providers believe families are ready for reunification but the court does not. Scrutinizing court appearance frequency segmented by caseworker can point to staff who may need extra help preparing for hearings.

This work can be integrated into projects funded by Court Improvement Programs (CIP) grants. For example, Washington State has used CIP funds to develop an interactive dashboard using linked court and child welfare data. These dashboards are used by the courts to reveal factors that address dependency timeliness and track progress towards those goals.64 In 2019, the Supreme Court of Texas Children’s Commission used CIP grants to fund the Texas Alliance for Child and Family Services to work with judges and child welfare field offices to analyze court-specific data for use in facilitating discussions on potential systemic improvements.65

Case study: Juvenile courts engagement in Utah

Utah convened a working group that brought together juvenile court judges and executives from child welfare and juvenile justice agencies to focus on issues in the juvenile justice system, including ways to address racial disparities for youth in the dispositions they received following a determination of delinquency. Analysis indicated that, for example, young women of color were placed in the custody of the child welfare agency on delinquency charges at a substantially higher rate than their white peers. The working group developed a set of recommendations for reforming state policies to eliminate the commitment of young people to child welfare for status offenses (noncriminal conduct, such as violating curfew, prohibited only due to the youth’s status as a minor).66

#17. Help service providers manage with data and retool their business models

Many nonprofit providers of case management, child welfare interventions, and behavioral health services lack sophisticated data, finance, or operations management capabilities necessary for executing performance improvement reforms. While mechanisms like active contract management can help these providers identify opportunities to improve service delivery, they do not provide the in-depth support necessary for providers to wholly revamp internal operations.

Model approaches

Child welfare agencies can offer support to contracted and community service providers that helps improve outcomes for their mutual clients. Such assistance could include:

- Convene service providers to share best practices with each other.
- Develop trainings to improve data fluency, analytics capacity, and performance management capabilities.
- Strengthen vendor managerial accounting practices to uncover internal cost drivers and improve resource allocation, business planning, and cost efficiency.
- Partner on opportunities to create new interventions and develop rigorous evidence about the effectiveness of these interventions on local population outcomes.
- Build capacity of homegrown, community-based providers to compete on procurements with nationally-supported organizations.
- Offer providers low-cost access to the state’s internal performance improvement resources, such as Lean or Six Sigma process improvement consultants.
- Collaborate with national model developers to prioritize the technical assistance they offer to local providers implementing their programs.

Case study: Dually-involved youth service providers in Illinois

In 2015, Illinois’ Department of Children and Family Services launched a partnership with a coalition of community providers to improve service delivery for justice-involved foster youth. To help these providers uncover opportunities to adjust their practice, the department linked administrative records from the state’s case management database with providers’ internal information management systems. Newly hired provider data analysts generate weekly internal reports on status of service enrollment among referrals, frequency of client engagement, and client progress on permanency outcome goals. Providers who rank lower on these measures are connected to those who rank higher to learn how to improve their performance in real time. Frontline staff from all five providers also regularly meet to share best practices and brainstorm solutions to barriers to care. This collaboration has helped improve the quality of data entry by state caseworkers and has led to the discovery of cases in need of additional attention, such as foster children experiencing frequent placement moves.67

Notes and References

  1. Today, 8.7 million children are living with a parent with a substance use need. “Children living with parents who have a substance use disorder,” SAMHSA, August 24, 2017,

  2.   “Direct Spending and Revenue Effects of Division E of Senate Amendment 1930, the Bipartisan Budget Act of 2018,” Congressional Budget Office, February 8, 2018,

  3.   Melissa Jonson-Reid et al., “Understanding Chronically Reported Families,” Child Maltreatment 15, no. 4 (November 2010): 271–81,

  4.   “2019 Annual Progress and Services Report, Child and Family Services Plan 2015-2019,” Michigan Department of Health & Human Services, June 2018,

  5.   “South Carolina Nurse Family Partnership Pay for Success,” Harvard Kennedy School Government Performance Lab, n.d.,

  6.   “Within Our Reach: A National Strategy to Eliminate Child Abuse and Neglect Fatalities,” Commission to Eliminate Child Abuse and Neglect Fatalities, March 17, 2016,

  7.   “Within Our Reach.”

  8.   “Child Maltreatment Fatality Case Reviews: Improving Effectiveness to Improve Systems and Prevent Deaths,” The National Center for Fatality Review and Prevention, n.d.,

  9.   “Safe Systems Improvement Tool,” The John Praed Foundation, n.d.,

  10.   “Preventing Child Abuse Fatalities Guide,” Harvard Kennedy School Government Performance Lab, n.d.,

  11.   “Plans presented to protect well-being of children in R.I.,” The Providence Journal, August 20, 2018,

  12.   “Structured Decision Making (SDM) Model,” National Council on Crime and Delinquency, n.d.,

  13.   “2019 Annual Progress and Services Report, Child and Family Services Plan 2015-2019.”

  14.   “Washington Department of Children, Youth, and Families Prevention Services Integration,” Harvard Kennedy School Government Performance Lab, n.d.,

  15.   “New Hampshire Child Welfare System Transformation,” Harvard Kennedy School Government Performance Lab, n.d.,

  16. “The Allegheny Family Screening Tool,” Allegheny County, August 2016, “Impact Evaluation on Allegheny Family Screening Tool Released,” Allegheny County, May 1, 2019, 

  17. “Child Maltreatment 2017,” U.S. Department of Health & Human Services, Administration for Children and Families, Administration on Children, Youth and Families, Children’s Bureau, 2019, Ruth Gilbert et al., “Burden and Consequences of Child Maltreatment in High-Income Countries,” The Lancet 373, no. 9657 (January 3, 2009): 68–81,

  18.   “Audit of the Department of Children and Families,”, December 7, 2017,

  19.   “RFP for Home- And Placement-Based Services to Improve Outcomes for DCYF Children, Youth and Families,” Rhode Island Division of Purchases, April 26, 2016, “Rhode Island Department of Children, Youth, and Families Performance Improvement,” Harvard Kennedy School Government Performance Lab, n.d.,

  20.   “Connecticut Family Stability Pay for Success,” Harvard Kennedy School Government Performance Lab, n.d.,

  21.   “Results-Driven Contracting: Twelve Strategies for Gathering Constructive Input to Improve Your RFP,” Harvard Kennedy School Government Performance Lab, 2019,

  22.   “Results-Driven Contracting,” Harvard Kennedy School Government Performance Lab, n.d.,

  23.   “Rhode Island Department of Children, Youth, and Families Performance Improvement.”

  24.   “2018 Standard of Excellence: 14. Contracting for Outcomes,” Results for America, April 2018,

  25.   For example, see Joseph P. Ryan et al., “Timing Matters: A Randomized Control Trial of Recovery Coaches in Foster Care,” Journal of Substance Abuse Treatment 77 (2017): 178–84,

  26.   “Rightsizing Congregate Care: A Powerful First Step in Transforming Child Welfare Systems,” The Annie E. Casey Foundation, January 1, 2009,

  27.   “Service Matching & Referrals,” Harvard Kennedy School Government Performance Lab, n.d.,

  28.   “Connecticut Department of Children and Families Enhanced Service Coordination,” Harvard Kennedy School Government Performance Lab, n.d.,

  29.   “Active Contract Management,” Harvard Kennedy School Government Performance Lab, n.d.,

  30.   “Six Tools for Implementing Active Contract Management,” Harvard Kennedy School Government Performance Lab, 2019,

  31.   “Rhode Island Department of Children, Youth, and Families Performance Improvement.”

  32.   “Data-Driven Management for Agency Operations: Improving Service Planning in Rhode Island’s Child Welfare Agency,” Harvard Kennedy School Government Performance Lab, 2019,

  33.   “Implementing ChildStat,” The Annie E. Casey Foundation, February 2015,

  34.   “ACS Foster Care Strategic Blueprint FY 2019 - FY 2023 & Findings from the Rapid Permanency Reviews,” NYC Administration for Children’s Services, May 2018, “What are some effective strategies for achieving permanency?,” Casey Family Programs, September 24, 2018,

  35.   “Understanding Foster Parenting: Using Administrative Data to Explore Retention,” U.S. Department of Health & Human Services ASPE, January 1, 2005,

  36.   “Keeping Kids in Families: Trends in U.S. Foster Care Placement,” The Annie E. Casey Foundation, n.d.,

  37.   “Kin-First Culture Shift Allegheny County,” Casey Family Programs, April 9, 2019,

  38.   “Children in Foster Care by Placement Type,” Kids Count Data Center, The Annie E. Casey Foundation, n.d.,

  39.   “Arizona Child Welfare Performance Improvement,” Harvard Kennedy School Government Performance Lab, n.d.,

  40.   “Child Welfare: HHS Could Play a Greater Role in Helping Child Welfare Agencies Recruit and Retain Staff,” General Accounting Office, March 2003,

  41.   “Caseload and Workload Management,” Child Welfare Information Gateway, July 2016,

  42.   Michael J. Cull et al., “Applying Principles from Safety Science to Improve Child Protection,” Child Welfare 92, no. 2 (2013): 179–95,

  43.   “How Does Turnover Affect Outcomes and What Can Be Done to Address Retention?,” Casey Family Programs, December 29, 2017,

  44.   “Resource Library,” National Child Welfare Workforce Institute, n.d.,

  45.   Theresa McCutcheon and Christine Allegra, “Workforce Report 2016-2017 Updates,” New Jersey Department of Children and Families, 2017, “State of New Jersey 2020-2024 Child and Family Services Plan: Attachment D 2020-2024 Training Plan,” New Jersey Department of Children and Families, June 30, 2019,

  46.   Christina Economy and Gloria Gong, “UK Troubled Families Programme: Lessons from Local Authorities” (Harvard Kennedy School Government Performance Lab, 2017),

  47.   “Housing Choice Voucher Program: Special Programs,” New Hampshire Housing, n.d.,

  48.   “Connecticut Family Stability Pay for Success,” Harvard Kennedy School Government Performance Lab, n.d.,

  49.   “Pathways To Prosperity And Well-Being: A New Family-Centered Approach To Human Services Delivery,” University of Wisconsin-Madison, Institute for Research on Poverty, January 15, 2020,

  50.   “Florida Child Welfare Behavioral Health Treatment,” Harvard Kennedy School Government Performance Lab, n.d.,

  51.   “Racial Disproportionality and Disparity in Child Welfare," Child Welfare Information Gateway, November 2016,

  52.   Hyunil Kim et al., “Lifetime Prevalence of Investigating Child Maltreatment Among US Children,” American Journal of Public Health 107, no. 2 (February 1, 2017): pp. 274-280,

  53.   “LGBTQ in Child Welfare: A Systematic Review of the Literature,” The Annie E. Casey Foundation, September 22, 2016,

  54.   “Immigration and Child Welfare,” Child Welfare Information Gateway, April 2015,

  55.   While many jurisdictions have mechanisms to preference contract bids from minority-owned businesses, this mechanism is often not applicable to contracts for child welfare services, as most providers are nonprofit organizations that, by definition, do not have individual owners.

  56.   “Race Equity and Inclusion Action Guide,” The Annie E. Casey Foundation, 2014,

  57.   “How Can Child Welfare Agencies Effectively Support LGBTQ+ Youth in Care?,” Casey Family Programs, September 2019,

  58.   “Immigration and Child Welfare,” Child Welfare Information Gateway, April 2015,

  59.   “Request for Proposals: ACS Prevention Services,” New York City Administration for Children’s Services Division of Prevention Services, June 12, 2019.

  60.   “Fostering Youth Transitions: Using Data to Drive Policy and Practice Decisions,” The Annie E. Casey Foundation, November 13, 2018,

  61.   “Housing Status of Youth Exiting Foster Care, Behavioral Health and Criminal Justice Systems,” Washington State Department of Social and Health Services, June 2017,

  62.   Louisa Treskon, Kyla Wasserman, and Vicky Ho, “Connecting to Opportunity: Lessons on Adapting Interventions for Young People Experiencing Homelessness or Systems Involvement,” MDRC, September 2019, “2018 Maine Profie: Transition-Age Youth in Foster Care,” The Annie E. Casey Foundation, n.d.,

  63.   “Accelerating Youth Violence Prevention and Positive Development,” Delaware Department of Health and Social Services, January 16, 2017, “Delaware Receives UPenn Technical Assistance Grant to Develop Integrated Data System,” State of Delaware, January 3, 2018, “State brings group violence prevention strategy pioneered nationally to Wilmington,” Delaware Public Media, July 30, 2019,

  64.  “Court Improvement Program (CIP) Grant,” Washington State Administrative Office of the Courts, n.d.,

  65.   “FY2019 Court Improvement Program Sub-Grantees – Program Grant Awards,” Supreme Court of Texas Children’s Commission, n.d.,

  66.   “Juvenile Justice Policy Study,” Utah Commission on Criminal and Juvenile Justice, n.d.,

  67.   For more, visit “Dually Involved Youth Project,” Conscience Community Network, February 23, 2016,