State Child Welfare Data Linkages Descriptive Study Technical Report: Study Findings

We highlight key findings below by section of the report. 

Section II: What are the characteristics of the state child welfare agencies that participated in the SCW Descriptive Study? 

• In 70 percent of states, the child welfare agency was part of a larger agency. 

• States varied in terms of who provided different child welfare services (state agency, counties, and/or contracted providers). Most states reported that the state child welfare agency provided child abuse and neglect investigations (84 percent), foster care placements (84 percent), child welfare case management after investigations (82 percent), prevention services (75 percent), and services defined in the state’s Prevention Plan for the Family First Prevention Services Act (75 percent). In at least half of states, contracted providers provided domestic violence or intimate partner violence services (68 percent), services defined in the state’s Prevention Plan for the Family First Prevention Services Act (66 percent), prevention services (61 percent), substance misuse or substance use services (61 percent), mental health services (57 percent), and foster care placements (50 percent). 

Section III: What are the characteristics of states’ systems of record? 

• States varied in terms of the status of their systems of record for child maltreatment reporting to the National Child Abuse and Neglect Data System (NCANDS) “system of record”. Twenty-six states (59 percent) reported that they were using a legacy system—for example, their Statewide Automated Child Welfare Information System (SACWIS). Twenty-one states (48 percent) reported that they were actively working on developing or replacing their system of record. 

• All states reported that frontline child protection staff who provided direct services entered data into systems of record. More than 80 percent of states indicated that state child welfare staff who were not frontline child protection staff also entered data directly into systems of record for child welfare. 

• Medicaid eligibility records were the most commonly integrated service data that were not child welfare data. Only 15 states (34 percent) did not systematically integrate Medicaid eligibility records in their systems of record. 

• Intellectual or developmental disability data were rarely integrated; 38 states (86 percent) did not include such data in their system of record. Mental health data were also rarely integrated; 37 states (84 percent) did not include such data in their system of record. 

• Twenty-two states (50 percent) required administrative data on child maltreatment to be purged or expunged; twenty-four states (55 percent) indefinitely retained data on screened-out or not investigated reports; and nineteen states (43 percent) indefinitely retained data on unsubstantiated child maltreatment reports. 

Section IV: Findings about Integrated Data 

• The most common use reported for integrated data was practice, such as casework, case management, or service delivery, selected by 76 percent of states, followed by continuous quality improvement, selected by 66 percent of states. Other responses selected by 45 to 48 percent of states were legislative report or mandate, program evaluation research, state policy analysis or research, and performance monitoring. Fewer than one-third of states selected predictive modeling and other research as a use of their integrated data. 

• Eighty-nine percent of states said that their integrated data covered statewide populations. 

• Eighty-six percent of states connected data with the system of record at the person or individual level, and 43 percent connected data with the system of record at the family level. 

• Eighty-six percent of states used deterministic or exact linking to connect their integrated data; one state used proprietary probabilistic linkage software; and no state used open-source probabilistic linkage software. 

Section V: Findings for linked data sets at the state level 

• Forty-one states (93 percent) reported the existence of data-sharing agreements intended to support caseworker practice, external research, or a linked analytic data set about families reported for child maltreatment or families receiving child welfare services. Three states (7 percent) indicated that they had no data-sharing agreements in place to support caseworker practice, external research, or a linked analytic data set. 

• Twenty-seven states (61 percent) were sharing data to support caseworker practice. Thirty-three states (75 percent) reporting having data-sharing agreements to support external research. Twenty-eight states (64 percent) shared data to support the creation of an analytic file about families either reported to them or receiving services from them. Thirty-seven states (84 percent) indicated that they created an analytic data set other than the NCANDS child file to create dashboards, conduct analyses, produce analytic files, or develop other reports related to the incidence of child maltreatment. 

• Twenty-one states (47.7 percent) reported having state requirements related to privacy or data transparency that were specific to child welfare data and 11 states (25 percent) reported state requirements regarding the submission of child welfare data to other agencies or divisions. 

Section VI: Findings for linked data at the data set level 

• The most common uses for the 37 linked data sets reported on in the Connected Data Survey were program evaluation research (49 percent), state policy analysis research (46 percent), continuous quality improvement (41 percent), and practice (38 percent) such as case management or service delivery. 

• The most common users of linked data sets other than staff of the managing organization were researchers (35 percent) and state government staff (32 percent). Thirty-five percent of linked data sets had no additional active users outside of the managing organization. 

• 43 percent of linked data sets could be accessed by a direct request to a designated contact, and 19 percent of linked data sets were not available to outside users. 

• Almost three-quarters (70 percent) of linked data sets had documented data governance procedures. 

Section VII: What are states’ plans for connected data, and how are they building capacity? 

• Eighteen states (41 percent) reported that existing state and federal funding to support data infrastructure was sufficient for their needs. 

• Twenty-three states (52 percent) reported that they had immediate plans to establish new data-sharing agreements or partnerships.