United States General Accounting Office GAO Report to Congressional Requesters July 2003 CHILD WELFARE Most States Are Developing Statewide Information Systems, but the Reliability of Child Welfare Data Could Be Improved GAO-03-809 July 2003 CHILD WELFARE Most States Are Developing Statewide Highlights of GAO-03-809, a report to Information Systems, but the Reliability Congressional Requesters of Child Welfare Data Could Be Improved To better monitor children and HHS reported that 47 states are developing or operating a SACWIS, but many families served by state child continue to face challenges developing their systems. Most state officials welfare agencies, Congress said they recognize the benefit their state will achieve by developing authorized matching funds for the SACWIS, such as contributing to the timeliness of child abuse and neglect development of statewide investigations; however, despite the availability of federal funds since 1994, automated child welfare information systems (SACWIS) and states reported a median delay of 2-½ years beyond the timeframes they set required that the Department of for completion. States reported that they encountered some difficulties Health and Human Services (HHS) during SACWIS development, such as challenges receiving state funding and compile information on the creating a system that reflected their work processes. In response to some of children served by state agencies. these challenges, HHS has provided technical assistance to help states This report reviews (1) states’ develop their systems and conducted on-site reviews of SACWIS to verify experiences in developing child that the systems meet federal requirements. welfare information systems and HHS’s role in assisting in their Despite efforts to implement comprehensive information systems, several development, (2) factors that affect factors affect the states’ ability to collect and report reliable adoption, foster the reliability of data that states care, and child abuse and neglect data. States responding to GAO’s survey collect and report on children served by their child welfare and officials in the 5 states GAO visited reported that insufficient agencies and HHS’s role in caseworker training and inaccurate and incomplete data entry affect the ensuring the reliability of those quality of the data reported to HHS. In addition, states reported technical data, and (3) practices that child challenges reporting data. Despite HHS’s assistance, many states report welfare agencies use to overcome ongoing challenges, such as the lack of clear and documented guidance on challenges associated with SACWIS how to report child welfare data. In addition, although states were mandated development and data reliability. to begin reporting data to the Adoption and Foster Care Analysis and Reporting System (AFCARS) in 1995, few reviews of states’ AFCARS reporting capabilities have been conducted to assist states in resolving some In order to improve the reliability of their reporting challenges. of state-reported child welfare data, GAO recommends that the Some states are using a variety of practices to address the challenges Secretary of HHS consider ways to associated with developing SACWIS and improving data reliability. For enhance the guidance and example, 44 states included caseworkers and other system users in the assistance offered to states to help design and testing of SACWIS, and 28 states reported using approaches to them overcome the key challenges help caseworkers identify and better understand the data elements that are in collecting and reporting child required for federal reporting. welfare data. In commenting on this report, HHS generally agreed Factors That Affect the Reliability of Data Reported to HHS with GAO’s findings and commented that the report Factors affecting data quality: Which may lead to provides a useful perspective of the inaccurate measures of: i on problems states face in collecting Inaccurate and incomplete data- at rm entry by caseworkers In fo State performance data and of HHS’s effort to provide on federal outcomes Insufficient caseworker training ongoing technical assistance to Differences between state and Children's improve child welfare data. federal data definitions Resulting in experiences, potentially unreliable such as time Lack of clear, documented www.gao.gov/cgi-bin/getrpt?GAO-03-809. information on abused spent in guidance from HHS and neglected foster Difficulty accessing technical To view the full product, including the scope children available in care assistance from HHS and methodology, click on the link above. federal data systems For more information, contact Cornelia M. Source: GAO analysis. Ashby at (202) 512-8403 or AshbyC@gao.gov. Contents Letter 1 Results in Brief 2 Background 5 Most States Are Developing SACWIS, but Challenges Remain Despite HHS’s Oversight and Technical Assistance 10 Several Factors Affect the States’ Ability to Ensure Reliable Data on Children’s Experiences, and Some of HHS’s Oversight and Assistance Is Problematic 24 States Are Using Various Practices to Overcome System Development Challenges and Improve Data on Children’s Experiences 38 Conclusion 42 Recommendation to the Secretary of Health and Human Services 43 Agency Comments 43 Appendix I Scope and Methodology 47 Appendix II State SACWIS Stages of Development 49 Appendix III Comments from the Department of Health and Human Services 50 Appendix IV GAO Contacts and Acknowledgments 62 GAO Contacts 62 Staff Acknowledgments 62 Bibliography 63 Related GAO Products 64 Tables Table 1: Significant Child Welfare Information and Data Systems 6 Page i GAO-03-809 Child Welfare Table 2: Number of States in Various Stages of SACWIS Development 12 Table 3: Selected SACWIS Child Welfare and Administrative Services 14 Table 4: Commonly Used SACWIS Development Participants and Their Level of Helpfulness 15 Table 5: State-Reported Benefits of SACWIS Development 17 Table 6: Number of Months States Delayed in SACWIS Development 18 Table 7: Common SACWIS Development Challenges 19 Table 8: Selected AFCARS Elements and Six States’ Levels of Compliance in Meeting Reporting Requirements 32 Figures Figure 1: Most Common Caseworker Issues That Affect Data Quality 26 Figure 2: Federal Practices That Affect Data Quality 35 Abbreviations ACF Administration for Children and Families ADP advance planning document AFCARS Adoption and Foster Care Analysis and Reporting System CAPTA Child Abuse Prevention and Treatment Act CRSR Child and Family Services Reviews FFP federal financial participation HHS Department of Health and Human Services IT information technology NCANDS National Child Abuse and Neglect Data System NYTD National Youth in Transition Database SACWIS Statewide Automated Child Welfare Information System VCIS Voluntary Cooperative Information System This is a work of the U.S. Government and is not subject to copyright protection in the United States. It may be reproduced and distributed in its entirety without further permission from GAO. It may contain copyrighted graphics, images or other materials. Permission from the copyright holder may be necessary should you wish to reproduce copyrighted materials separately from GAO’s product. Page ii GAO-03-809 Child Welfare United States General Accounting Office Washington, DC 20548 July 31, 2003 The Honorable Charles E. Grassley Chairman, Committee on Finance United States Senate The Honorable Tom DeLay Majority Leader House of Representatives Recent news reports of tragedies involving child welfare agencies across the country highlight the long-standing problems states have had monitoring children in their care. Given that hundreds of thousands of children are found to be victims of abuse and neglect and are estimated to spend some time in foster care each year, the Congress required that the Department of Health and Human Services (HHS) compile information on the children served by state agencies and authorized federal funds to match those of states for use in the development of state child welfare information systems. Since 1994, designated federal matching funds have been available to states to develop and implement comprehensive case management systems—Statewide Automated Child Welfare Information Systems (SACWIS)—to manage their child welfare cases as well as to report child abuse and neglect, foster care, and adoption information to the federal government. States have the option to implement a SACWIS or develop different information systems without using SACWIS funds to support their child welfare agencies and collect information on their child welfare cases. Regardless of the type of system a state develops, child welfare caseworkers at the county or local level are the key personnel who collect and document information on children and families served by child welfare agencies, in addition to performing a wide range of services to protect children—such as investigating child abuse or neglect reports or providing support services to maintain the children in their homes. Currently, HHS compiles state-reported child welfare data in two databases: the Adoption and Foster Care Analysis and Reporting System (AFCARS) and the National Child Abuse and Neglect Data System (NCANDS). HHS relies on the information available in its databases to analyze and track children’s experiences in the child welfare system, to determine states’ performance on federal child welfare outcome measures, and to report to Congress on children’s well being and child welfare experiences. Page 1 GAO-03-809 Child Welfare Because of your interest in how states have developed systems to collect and report information on the children they serve and the reliability of that information, you asked us to determine (1) states’ experiences in developing child welfare information systems and HHS’s role in assisting in their development; (2) factors that affect the reliability of data that states collect and report on children served by their child welfare agencies, and HHS’s role in ensuring the reliability of those data; and (3) practices that child welfare agencies use to overcome challenges associated with SACWIS development and data reliability. To conduct our work, we surveyed all 50 states and the District of Columbia regarding their experiences in developing and using information systems and their ability to report data to HHS. We received responses from 49 states and the District of Columbia,1 although all states did not respond to every question. We also reviewed a variety of HHS documents, including the protocol and reports for its reviews of SACWIS systems and states’ AFCARS reporting capabilities. In addition, we visited 5 states— Colorado, Iowa, New York, North Carolina, and Oklahoma—to obtain first- hand information on their experiences developing SACWIS and reporting data to HHS. We selected these states to represent geographic diversity and different stages of SACWIS implementation. Finally, we interviewed HHS officials and child welfare and data experts and reviewed relevant literature. We conducted our work between June 2002 and June 2003 in accordance with generally accepted government auditing standards. A more detailed discussion of our scope and methodology appears in appendix I. HHS reported that 47 states are developing or operating a SACWIS, but Results in Brief many continue to face challenges despite HHS’s oversight and technical assistance. Most states are using federal SACWIS funds and are in various stages of development. States reported in our survey that they have spent approximately $2.4 billion in federal, state, and local funding on SACWIS. Most state officials said they recognize the benefit their state will achieve by developing SACWIS, such as contributing to the timeliness of child abuse and neglect investigations; however, despite the availability of federal funds since 1994, many child welfare agencies lag behind the 1 Throughout this report, references to state survey responses include the District of Columbia. Forty-six of these states reported that they are developing or operating a SACWIS. Nevada, which HHS reported has an operational SACWIS, did not respond to our survey. Page 2 GAO-03-809 Child Welfare timeframes they set for completion, with delays ranging from 2 months to 8 years. Forty-two of the 46 states responding to our survey that they are developing SACWIS reported at least some challenge obtaining state funding. In Iowa, for example, state officials reported that insufficient state funds delayed them in making the necessary modifications to meet federal requirements for system completion. Some states had difficulties developing a system that met the state child welfare agency’s needs statewide. For example, state officials in New York—a state where the counties are responsible for administering child welfare services—said that building a uniform system was stalled when significant frustration with the system’s design led some county officials to request that the state stop SACWIS development. In addition, 32 states reported at least some challenge securing information technology contractors with knowledge of child welfare practice to develop their SACWIS. In response to some of these challenges, HHS has provided technical assistance to help states develop their systems and conducted on-site reviews of SACWIS to verify that the systems meet all federal requirements. For example, at the time of our review, HHS had conducted on-site reviews in 26 states with operational SACWIS to ensure that the systems met all federal requirements and to offer assistance to states that faced challenges completing the development of their SACWIS. Several factors affect states’ ability to collect and report reliable data on children served by state child welfare agencies, and some problems exist, such as a lack of clear and documented guidance, with HHS’s oversight and technical assistance. Almost all of the states responding to our survey reported that insufficient caseworker training and inaccurate and incomplete data entry into their information system affect the quality of the data reported to HHS. Although most states reported these as separate factors, HHS and the states we visited found that insufficient training and inaccurate and incomplete data entry are often linked. In addition, 36 of the 50 states that responded to our survey reported that technical challenges, such as matching their state data element definitions to HHS’s data categories, affected the quality of the data that they report to the federal government. Similarly, during assessments of 6 states’ compliance with AFCARS reporting standards, HHS found that these issues affect data reliability. Despite HHS’s assistance in helping states improve their data, such as testing state data quality and providing the results to the states to aid them in resubmitting data, states report ongoing challenges. For example, 41 of the 50 states responding to our survey reported that a lack of clear and documented guidance from HHS affects the quality of the data they report to AFCARS, and 25 states said the lack of clear, documented guidance also affected the data reported to NCANDS. In addition, although Page 3 GAO-03-809 Child Welfare states were mandated to begin reporting AFCARS in 1995, few reviews of states AFCARS reporting capabilities have been conducted to assist states in resolving some of their challenges. Some states are using a variety of practices to address the challenges associated with developing SACWIS and improving data reliability, although no formal evaluations are available on their effectiveness. To overcome the challenge of developing a system to meet statewide needs, many states relied on caseworkers and other system users for input on design and testing of SACWIS. Few states reported that they devised strategies to overcome the other challenges, such as limited funding and finding contractors with knowledge of child welfare. However, Oklahoma child welfare officials—in order to maximize the limited state funding for maintaining their SACWIS—reported saving $1 million each year by hiring some of the contractors who developed their SACWIS as permanent staff. To improve data reliability, the 5 states we visited routinely review their data to identify data entry errors so that managers can ensure that the missing data are entered appropriately. In addition, some states reported that frequent use of the data, such as publishing periodic management reports detailing local offices’ performance on outcome measures, helps caseworkers understand the importance of entering timely information. In order to improve the reliability of state-reported child welfare data, we are recommending that the Secretary of HHS consider ways to enhance the guidance and assistance offered to states to help them overcome the key challenges in collecting and reporting child welfare data. In commenting on this report, HHS’s Administration for Children and Families (ACF) generally agreed with our findings and commented that the report provides a useful perspective of the problems states face in collecting data and of ACF’s effort to provide ongoing technical assistance to improve the quality of child welfare data. In response to our recommendation, ACF said that we did not recognize the long-term efforts it has taken to provide AFCARS and NCANDS guidance. ACF also noted that the data definitions need to be updated and revised and said it is currently in the process of revising the AFCARS regulations to further standardize the information states are to report—which we acknowledge in our report. Further, ACF added that although staff turnover in state child welfare agencies is a significant contributor to data quality issues, we did not focus on this as a significant factor. However, because we recently issued a detailed report on a variety of caseworker issues, we primarily focused in this report on the key data entry challenges caseworkers face and refer readers to our previous work for additional information on challenges related to caseworker recruitment and retention and their Page 4 GAO-03-809 Child Welfare affect on child welfare agencies. ACF commented that it is firmly committed to continue to support the states and to provide technical assistance and other guidance as its resources will permit. We believe that the recent activities to formally obtain, document, and incorporate feedback from the states with regard to collection and reporting adoption and foster care data represent are a step towards improving states’ data. Our recommendation encourages HHS to consider ways to enhance the guidance and assistance already offered to states as a step to helping them better comply with the reporting requirements. ACF’s Children’s Bureau is responsible for the administration and Background oversight of federal funding to states for child welfare services under Titles IV-B and IV-E of the Social Security Act. However, the monitoring of children served by state child welfare agencies is the responsibility of the state agencies that provide the services to these children and their families. Child welfare caseworkers at the county or local level are the key personnel responsible for documenting the wide range of services offered to children and families, such as investigations of abuse and neglect; treatment services offered to families to keep them intact and prevent the need for foster care; and arrangements made for permanent or adoptive placements when children must be removed from their homes. Caseworkers are supported by supervisors who typically assign new cases to workers and monitor caseworkers’ progress in achieving desired outcomes, analyzing and addressing problems, and making decisions about cases. A number of efforts at the national level have been taken to implement comprehensive data systems that capture, report, and analyze the child welfare information collected by the states (see table 1 for information on national data systems as well as information on state systems). Page 5 GAO-03-809 Child Welfare Table 1: Significant Child Welfare Information and Data Systems System History Use State information systems Statewide Automated Child Welfare Information The Omnibus Budget Reconciliation Act A SACWIS is designed and System (SACWIS) (OBRA) of 1993 authorized the use of an developed for use by states’ enhanced federal financial participation caseworkers and other (FFP) rate of 75 percent to assist states personnel for the purposes of develop uniform automated information establishing an electronic case systems—SACWIS—that support the file for children and families administration of services offered under their served by the state child welfare child welfare programs.a The enhanced FFP agency. Some of the data was available initially from federal fiscal captured in SACWIS are years 1994 through 1996 and subsequently reported to HHS. extended through federal fiscal year 1997. After 1997, states receive a 50 percent match for SACWIS-related activities. Funding approval for SACWIS is based on states’ estimated costs for development and operation, and no time limits are placed on the receipt of federal funding. National data systems Voluntary Cooperative Information System (VCIS) With funding from HHS, the American Public The data available in VCIS was Welfare Association—now known as the used to produce reports at the American Public Human Services national level on the Association—established VCIS in an effort characteristics of children in to compile state-specific child welfare data. foster care and those adopted States voluntarily reported aggregate-level from state child welfare data on the characteristics of children in agencies. However, the data foster care and those adopted from state reported to VCIS were child welfare agencies. VCIS compiled data inconsistent.b Following the between 1982 and 1994. implementation of AFCARS, states no longer reported to VCIS. Page 6 GAO-03-809 Child Welfare System History Use Adoption and Foster Care Analysis and Reporting In 1986, Congress added Section 479 to Some of the key uses of System (AFCARS) Title IV-E of the Social Security Act, which AFCARS data include: required HHS to establish and implement a mandatory foster care and adoption data collection system. Despite the legislative Calculating the national standard requirement that HHS complete final for five federal outcome regulations for the system by the end of measures—incidence of child calendar year 1988, regulations were not abuse and/or neglect in foster issued until December 1993. States began care, foster care re-entries, reporting to AFCARS in 1995 on the stability of foster care demographic characteristics of adoptive and placements, length of time to foster children and their parents as well as achieve reunification, and length foster children’s type of placement and of time to achieve adoption. permanency goals. States are required to report 66 foster care and 37 adoption data elements for each child in care during The publication of the annual 6-month periods. Between 1998 and Child Welfare Outcomes Report. 2001, states not meeting certain reporting standards were subject to penalties. The Determining the allocation of penalties were rescinded in January 2002 funds in the adoption incentive following an appeal filed by 12 states. program and Chafee Foster Care Independence Program.c National Child Abuse and Neglect Data System In 1988, Congress amended the Child Some of the key uses of (NCANDS) Abuse Prevention and Treatment Act NCANDS data include: (CAPTA) by directing HHS to establish a national data collection and analysis system consisting of state child abuse and neglect Calculating the national standard information. Walter R. McDonald & for two federal outcome Associates was awarded the contract to measures on child safety— compile and analyze the state-reported data. recurrence of maltreatment and States began voluntarily reporting annually incidence of child abuse and/or in 1990; however, the 1996 CAPTA neglect in foster care. amendments directed states to report certain data to receive CAPTA grant funding. States submit either child-specific The publication of the annual records or aggregate-level data. In 2001— Child Maltreatment Report. the most recently available data from HHS— 39 states and the District of Columbia The publication of the annual submitted child-specific data, which includes Child Welfare Outcomes Report. the demographics of the children and their perpetrators, the types of maltreatment, investigation or assessment findings, risk factors, and services provided as a result of the investigation or assessment. The remaining 11 states submitted aggregate data. Page 7 GAO-03-809 Child Welfare System History Use National Youth in Transition Database (NYTD) In response to requirements included in the In response to the legislation, Foster Care Independence Act of 1999, data available in the NYTD will HHS has proposed to develop the NYTD to be used, in part, to track (1) the capture more detailed data than is currently number and characteristics of reported to AFCARS on older foster children older youth receiving who receive independent living services. independent living services, Such services include life-skills training in (2) the type and quantity of financial management or career planning. services states provided, and HHS reported to Congress in September (3) state performance on the 2001 that nationwide implementation of outcome measures required by NYTD was planned for October 2003, but the legislation. recent HHS information indicates that the actual implementation date will likely be 2005 or 2006. Once implemented, states will be required to report to NYTD. Source: GAO analysis based on federal legislation, regulations, and other documents. a The allowable costs under the 75 percent FFP included planning, design, development, and installation of a SACWIS. Other costs, such as operation of the SACWIS, were covered at 50 percent. b See www.acf.hhs.gov/programs/cb/dis/vcis/maintoc.htm for more complete details on the shortcomings of the data available in VCIS. c The total number of finalized adoptions reported to AFCARS are used to determine the amount, if any, of the adoption incentive payments awarded to states. In addition, AFCARS data are used to determine the ratio of the number of children in foster care in each state to the total number of children in foster care in all states to calculate state allotments under the Chafee Foster Care Independence Program. To qualify for federal funding for SACWIS, states must prepare and submit an advance planning document (APD) to ACF’s Children’s Bureau, in which they describe the state’s plan for managing the design, development, implementation, and operation of a SACWIS that meets federal requirements and state needs in an efficient, comprehensive, and cost- effective manner.2 In addition, the state must establish SACWIS and program performance goals in terms of projected costs and benefits in the APD. States are required to submit separate APDs for the planning and development phases, in addition to periodic updates. 2 When states choose to develop information systems that include other human services, such as food stamps, child support enforcement, or Medicaid, states must submit APDs to each cognizant federal agency. In a hearing held last year before the Subcommittee on Technology and Procurement Policy, House Committee on Government Reform, we testified that the federal agencies do not have systems to monitor states requests for federal approval and funding through the life cycle of a state request. (See U.S. General Accounting Office, Human Services: Federal Approval and Funding Processes for States’ Information System, GAO-02-347T (Washington, D.C: July 9, 2002.)) Page 8 GAO-03-809 Child Welfare Since the administration and structure of state child welfare agencies vary across the nation, states can design their SACWIS to meet their state needs, as long as states meet certain federal requirements. Federal funding is available to states for SACWIS that • meet the requirements for reporting AFCARS data to HHS; • to the extent practicable, are capable of linking with the state data collection system that collects information on child abuse and neglect; • to the extent practicable, are capable of linking with, and retrieving information from, the state data collection system that collects information on the eligibility of individuals under Title IV-A—Temporary Assistance for Needy Families; and • provides for more efficient, economical, and effective administration of the programs carried out under a state’s plans approved under Titles IV-B and IV-E of the Social Security Act. A SACWIS must operate uniformly as a single system in each state and must encompass all entities that administer programs provided under Titles IV-B and IV-E. In some cases, HHS will allow the statewide system to link to another state system to perform required functions, such as linking to financial systems to issue and reconcile payments to child welfare service providers. The state’s APD must describe how its SACWIS will link to other systems to meet the requirements in the SACWIS regulations. In addition to monitoring the APDs of the states that are developing SACWIS, HHS reviews state information systems through formal SACWIS assessment reviews and the Child and Family Services Reviews (CFSR)—a federal review process to monitor states’ compliance with child welfare laws and federal outcome measures. The formal SACWIS reviews are conducted by ACF’s Children’s Bureau to determine if a state has developed and implemented all components detailed in the state’s APD and if the system adheres to federal requirements. The CFSR assesses statewide information systems, along with other systemic factors, to determine if the state is operating a system that can readily identify the status, demographic characteristics, location, and goals for placement of every child who is in foster care. This systemic factor is reviewed in all states, regardless of whether the state is developing a SACWIS or the stage of system development. According to results from the fiscal years 2001 and 2002 CFSRs, 4 of the 32 states in which HHS reviewed were not in substantial conformity on the statewide information system indicator. These 4 states must address how they will come into conformity with this Page 9 GAO-03-809 Child Welfare factor in a program improvement plan. HHS has also conducted SACWIS reviews in 2 of these states. While 47 states are developing or operating a SACWIS, many challenges Most States Are remain despite HHS’s oversight and technical assistance. Since 1994, Developing SACWIS, states reported that they have spent approximately $2.4 billion in federal, state, and local funding on SACWIS. While most state officials we but Challenges interviewed and those responding to our survey said that they recognize Remain Despite HHS’s the benefits their state will achieve by developing a statewide system, many states reported that the development of their SACWIS is delayed Oversight and between 2 months and 8 years beyond the time frames the states set for Technical Assistance completion, with a median delay of 2-½ years. Most states responding to our survey faced challenges, such as obtaining state funding and developing a system that met the child welfare agency’s needs statewide. In response to some of these challenges, HHS has provided technical assistance to help states develop their systems and conducted on-site SACWIS reviews to verify that the systems meet all federal requirements. States Are Using Federal Currently, 47 states are developing or operating a SACWIS and are in and State Funds and various stages of development—ranging from planning to complete. The Various Participants to states responding to our survey reported using approximately $1.3 billion in federal funds3 and approximately $1.1 billion in state and local funds4 for Develop MultiComponent their SACWIS. However, HHS estimated that it allocated approximately SACWIS $821 million between fiscal years 1994 and 2001 in SACWIS developmental funds5 and $173 million between fiscal years 1999 and 2001 in SACWIS 3 Forty-four states provided information on the total amount of federal funds they received to develop and operate SACWIS. Alaska, Hawaii, Missouri, North Carolina, Texas, and Vermont did not report federal funding information. Nevada did not respond to our survey. State-reported figures may include some funding allocated in fiscal year 2003 since the survey was issued in October 2002 and completed as late as December 2002. 4 Forty-four states provided information on the total amount of state funds used to develop and operate SACWIS. Arkansas, Hawaii, Missouri, North Carolina, Texas, and Vermont did not report state funding information. Nevada did not respond to our survey. State-reported figures may include some funding allocated in fiscal year 2003 since the survey was issued in October 2002 and completed as late as December 2002. 5 This figure includes developmental funds allocated by HHS to 49 states and the District of Columbia. Hawaii did not take any federal money for SACWIS development. Page 10 GAO-03-809 Child Welfare operational funds.6 The total amount of federal funding provided to states for SACWIS is unknown because states claimed operational costs as a part of their Title IV-E administrative expenses prior to 1999.7 Although the federal government matched state funding at an enhanced rate of 75 percent beginning in 1994, many states did not apply for federal funding or begin SACWIS development until 1996 and 1997 when more than $467 million—the bulk of federal funds—were allocated. Most states were still developing their SACWIS by the time enhanced funding expired in 1997, after which states could receive a 50 percent FFP for SACWIS development and operation. Although 47 states are currently developing or operating a SACWIS, all states except Hawaii received some federal SACWIS funds. For example, according to figures provided by HHS, North Carolina received approximately $9.6 million in developmental funds and North Dakota received approximately $2.4 million in developmental funds and $245,000 in operational funds for SACWIS, but both states encountered difficulties that prevented them from completing their systems. In these situations, HHS entered into negotiations with the states about the amount of money that the states must return to the federal government. In order to track states’ SACWIS development, HHS places them in six categories that identify their stage of development (see table 2). States are required to submit APD updates periodically, which inform HHS of their progress in developing SACWIS. See appendix II for a complete list of states’ phases of development. Although most states continue to advance in the development of their systems, some encounter problems that cause HHS to recategorize them into a lower stage of development. In Pennsylvania, for example, the child welfare agency encountered difficulties, such as inadequate computer software to support a comprehensive SACWIS, after attempting to implement its SACWIS in 6 This figure includes operational funds allocated to 35 states. States begin claiming operational costs when some or all components of their SACWIS are operating in local offices. Operational activities include routine maintenance, minor enhancements, and other changes that do not significantly increase or modify the functionality of the system. 7 According to HHS officials, prior to fiscal year 2000, states reported SACWIS operational expenses as part of their Title IV-E administrative expenses because the claims sheet states used for reporting did not have a separate column for SACWIS operational expenditures. In fiscal year 2000, states were required to use a claims sheet that was reformatted to provide space for SACWIS operational expenditures. In addition, an HHS official explained that the difference between the state-reported figures and the federal figures may be due to states claiming some SACWIS expenses under different programs, such as Title IV-E administrative funds, rather than separately as SACWIS expenses. Page 11 GAO-03-809 Child Welfare 2000. Due to these problems, the state is in the process of shutting down the system and has resubmitted an APD for a new system to HHS for review and approval for further federal funding. According to figures provided by HHS, Pennsylvania has received approximately $9.7 million in federal funding thus far. In addition, while HHS may classify a state system as complete following an assessment of their SACWIS, a state may make additional changes to the system since SACWIS, like other computer systems, continually evolve as technology and child welfare practices change. States can claim federal funding for these changes as operational expenses. For example, Oklahoma’s SACWIS was the first system to be determined complete, but it has made enhancements to its system since HHS found the system in compliance with federal requirements in 1998. In addition, Oklahoma is currently considering moving to a Web-based system. An HHS official reported that such changes do not need prior approval unless they are in excess of $5 million. Table 2: Number of States in Various Stages of SACWIS Development Stage Number of states Completea 5 b Operational 24 c Partially operational 9 d Implementation 2 Planninge 7 f No SACWIS 4 Source: HHS. Note: Status is as of May 22, 2003. a SACWIS assessment process is completed, and all functional requirements and specifications set forth in the APD are either included in the system or in an accepted corrective action plan. b All functional requirements and specifications in APD are included in system, and system is functional statewide, but state has not completed SACWIS assessment or is working on other issues. c State is still rolling out system to field sites or still adding functions to systems that are operational statewide. d In active design and development, even if delayed while waiting to resolve problems such as funding. e Working through options for a SACWIS. f Have never pursued SACWIS funding or have abandoned plans to develop a system. In developing a system, states have considerable flexibility in the design of their SACWIS. According to HHS officials, a state should be using its SACWIS as a case management tool that uses automation to support the various aspects of state child welfare programs, such as recording child Page 12 GAO-03-809 Child Welfare protection, out of home care, and foster care and adoption services. To further assist child welfare practice, states have designed their systems to follow the natural flow of child welfare practice in their state and have added design features to help track key events during a case. For example, in Iowa child welfare work is divided between child abuse and neglect investigations and ongoing case management for children brought into the care of the child welfare agency. As a result, Iowa designed a SACWIS to reflect this work process by linking two databases—one to record child abuse and neglect information and one to record ongoing case records— that share information with one another.8 In Rhode Island, the SACWIS was designed to alert caseworkers if an alleged perpetrator has been the subject of three reports of abuse or neglect within 1 year. Regardless of the findings of each report, this alert notifies the caseworker to initiate an investigation when a third report is received. Since many states are in different phases of SACWIS development, their systems currently support to varying degrees a variety of child welfare and administrative components (see table 3). According to HHS, while the components represented in table 3 are required for a state’s SACWIS to be considered compliant with federal guidance—either through an interface or built within the system—some of the subcomponents, such as a function that helps caseworkers manage their caseloads, are optional. HHS has encouraged states to automate as many functions as possible in the SACWIS in an effort to cut down on the additional paperwork or duplicative steps inherent in manual data collection. One of these services, tracking independent living, is becoming more important for states as HHS decides how to implement the Foster Care Independence Act of 1999 and considers the development of the NYTD.9 Some states have already started collecting data on older youth and the services they receive. Currently, 27 states reported in our survey that they are at some stage of using their SACWIS to track independent living services, and an additional 14 states plan to include this component in their system in preparation for the 8 Although the Iowa state officials described their SACWIS as including the child abuse and neglect system, HHS commented on a draft of this report that it does not view the child abuse and neglect system as part of the state’s SACWIS. However, HHS said that the state has met the SACWIS requirement in this area by building an interface between the two systems. 9 The Foster Care Independence Act of 1999 increased federal support to states for independent living programs—programs designed to assist youth who are identified as likely to remain in foster care until age 18. Independent living services can include education or training necessary for the youth to obtain employment. Page 13 GAO-03-809 Child Welfare requirements. However, 21 of the 46 states reporting to our survey that they are developing or operating a SACWIS reported that they would have to make substantial changes to their SACWIS in order to capture this information. Table 3: Selected SACWIS Child Welfare and Administrative Services Fully or partially Planned Service operational in SACWIS for SACWIS Child Welfare Services a Child protection 38 5 b Out of home care 35 8 Adoption 34 9 Independent living 27 14 c Intensive home-based services 27 13 Administrative Services Workload management 32 8 IV-E eligibilityd 29 14 Foster care maintenance payments 28 14 Adoption assistance payments 25 17 Contract provider payment 24 15 Source: GAO survey. Note: Based on responses from 46 states developing or operating a SACWIS. The rows for the columns “fully or partially operational” and “planned” do not add to 46 because the respondents may have answered “not supported,” “don’t know,” or “no answer.” a Child protection includes services such as intake and screening, investigation, and disposition. b Out of home care includes things such as foster care, group homes, and residential placement. c Intensive home-based services include efforts to avoid placing a child in foster care. d IV-E funding is available for foster care, adoption, and independent living services. To assist with the design of their SACWIS, states relied on a number of different participants including internal users, such as caseworkers and managers, information technology (IT) staff, and contractors. Most states found these participants to be extremely or very helpful in the process (see table 4). In Oklahoma, for example, 150 child welfare staff from the field worked closely with the contractor in intensive work group sessions to design and test the system. To complement the caseworkers’ knowledge of child welfare practice, 43 states relied on IT staff. In Colorado, for example, IT staff said that during SACWIS design and development, they shared office space with program staff that had been assigned to help with Page 14 GAO-03-809 Child Welfare SACWIS development. This co-location of staff aided in the exchange of information pertaining to the development of the system. Finally, 42 states reported that they hired private contractors to conduct a large part of SACWIS design and development. The contractors helped states meet federal requirements, designed the system with state specific options, wrote the necessary software, tested and implemented the system, and trained users. Table 4: Commonly Used SACWIS Development Participants and Their Level of Helpfulness SACWIS development Number of states Number of states rating participant participant using participant as extremely or very helpful Internal users 44 41 IT staff 43 31 Private contractors 42 37 Source: GAO survey. Note: Based on responses from 46 states developing or operating a SACWIS. At the time of our review, HHS reported that 4 states were not pursuing SACWIS development and most of these states reported various reasons in our survey for not developing a system. In Hawaii, the child welfare agency chose not to pursue SACWIS because it already had a statewide system in place that it believed was adequately meeting its needs and which was collecting and reporting federal child welfare data. After an attempt to develop a system, North Carolina cancelled its efforts because it could not build consensus across its 100 counties on the design of a uniform system. On our site visit to North Carolina, child welfare officials reported that they are currently working on a statewide information system that will encompass a number of social services, such as food stamps and mental health services, but an HHS official reported that North Carolina is not seeking federal SACWIS funding to support the development of this system. Vermont officials reported that they did not pursue SACWIS because the legislature declined to provide the matching state funds. In retrospect, they believe that the choice not to develop SACWIS was best for the state because they found the SACWIS requirements too restrictive to enable the state to design a system to meet its needs. Officials said that the state would not use a number of the required SACWIS components, such as developing all the required electronic links to other agencies’ systems, especially since the state has a small child welfare population. Another state—North Dakota—did not report in our survey the reason for stopping SACWIS development; Page 15 GAO-03-809 Child Welfare however, HHS officials reported that the state had attempted to develop a SACWIS, but faced a variety of problems, such as receiving state funding. States Accrue Benefits While most state child welfare agency officials said they recognize the from Using SACWIS, but benefits the state will achieve by developing SACWIS, such as enhancing their ability to track the whereabouts of foster children, 31 state agencies Several Issues Create lag behind the time frames they set for completion, with 26 states Delays in Completing reporting delays ranging from 2 months to 8 years. State officials reported States’ Systems in our survey and during site visits that SACWIS has contributed to more efficient and effective agency functioning, which can improve states’ capabilities to manage their child welfare cases, including keeping track of where the children are living and the services they are receiving. Child welfare officials in Colorado reported that automation has improved agency functioning by making child welfare case information available statewide, which is especially helpful when families move from one county to another. In Oklahoma, caseworkers and state officials noted that they believe their children are safer since the implementation of SACWIS simply because the information on the children is easily accessible to the caseworkers and their supervisors. According to survey results, automated systems provided easier access to data and allowed caseworkers to better monitor children in their care, which may contribute to additional child welfare and administrative benefits, such as decreased incidences of child abuse and neglect, shortened length of time to achieve adoption, timeliness of payments to foster families, and timeliness of payments to foster facilities (see table 5). New Jersey reported in our survey that its goal in developing a SACWIS is to integrate the more than 40 stand-alone systems that currently capture information on the children served by their child welfare agency. By pulling all of these systems together into a uniform SACWIS, the state hopes to improve the recording of casework activities in a timely manner and to develop a tool to better target resources and services. Effectively integrating these systems will require the state to use a disciplined IT management approach that includes (1) detailed analyses of users’ needs and requirements, (2) a clearly defined strategy for addressing information needs, and (3) sufficient technical expertise and resources to support the effort. Page 16 GAO-03-809 Child Welfare Table 5: State-Reported Benefits of SACWIS Development Number of states identifying SACWIS as extremely or very effective in Child welfare benefits achieving measure Decreased incidence of child abuse and neglect 25 Shortened length of time to achieve adoption 24 Decreased recurrence of child maltreatment 23 Shortened length of time to achieve reunification 22 Management benefits Timeliness of payment to foster families 36 Timeliness of payment to foster facilities 34 Overall case management for children and families 33 Timeliness of child abuse and neglect investigations 33 Source: GAO survey. Note: Based on responses from 46 states developing or operating SACWIS. States not included answered “moderately effective,” “somewhat effective,” “not at all effective,” “not a state goal,” “system doesn’t support,” “don’t know,” or “no answer.” Despite the benefits that many states have accrued with SACWIS, 31 states reported in our survey that they have been delayed in system completion beyond their initial deadline and identified a number of challenges that have led to the delay (see table 6).10 Some of the common difficulties states reported in developing SACWIS included receiving state funding approval, reaching internal agreement on system development, and creating a system that reflects child welfare work processes and is user friendly (see table 7). 10 Twelve of the 46 states reporting that they are developing or operating a SACWIS reported that they have not experienced delays in developing their systems. In response to the length of the delays reported by 26 states in our survey, ACF commented on a draft of this report that these states may be using different definitions in defining their delays. However, ACF did not provide further information on how the delays represented in this report differ from its perception of states’ experiences. In our survey, we asked states to report on the delays that exceeded the time line outlined in their initial APD. Page 17 GAO-03-809 Child Welfare Table 6: Number of Months States Delayed in SACWIS Development State Length of delay in monthsa Alabama 36 Arkansas 6 California 36 Colorado 26 Connecticut 96 District of Columbia 36 Georgia 25 Idaho 21 Illinois 79 Indiana 6 Kansas 72 Louisiana 12 Maryland 12 Michigan 26 Minnesota 12 Mississippi 12 New Jersey 42 New Mexico 3 Ohio 36 Oregon 70 Rhode Island 14 South Carolina 47 Tennessee 36 Utah 48 Virginia 2 Washington 36 Source: GAO survey. Note: While 31 states reported in the survey that they have experienced a delay in SACWIS development, only 26 states reported the length of their delay. The survey was issued in October 2002 and completed by states as late as December 2002. a States were asked to report the number of months the delays exceeded the time line outlined in their APD. Page 18 GAO-03-809 Child Welfare Table 7: Common SACWIS Development Challenges Number of states experiencing at a SACWIS development challenges least some challenge Receiving state funding approval 42 Reaching internal agreement on system development 41 Creating a system that reflects work processes 40 Creating a user friendly system 39 Insufficient state funding allocation 32 Securing contractors with knowledge of child welfare 32 Source: GAO survey. Note: Based on responses from 46 states developing or operating a SACWIS. a States were asked the extent to which certain factors were a challenge in developing SACWIS using the following scale: very great, great, moderate, some, or no challenge. The number of states reported in our analysis of challenges represents the total number of states reporting any extent the factor posed a challenge. States not included answered “no challenge,” “have not yet reached this stage,” “don’t know,” or “no answer.” Forty-two states reported challenges receiving funding approval, and 32 states reported that insufficient state funding allocations for SACWIS development were a challenge in developing a comprehensive system. For example, Vermont officials reported that the state legislature declined to provide the matching state funds needed to secure federal funding for SACWIS. As a result, the state could not pursue development. In addition to the development challenges reported in our survey, 2 of the 5 states we visited reported that insufficient funding affected ongoing SACWIS maintenance. In Colorado, state agencies have received a series of budget cuts, which child welfare officials report have impacted their ability to respond to child welfare caseworkers’ needs for system improvements. In Iowa, child welfare officials reported that the state legislature appropriated $17,000 for state fiscal year 2002 for all child welfare automated systems activity, which they noted was an insufficient amount to maintain and upgrade systems as well as to pay staff. They reported that, as a result, the child welfare agency provided the information systems department with approximately $1 million from other parts of the agency’s budget. Despite user involvement in system design, some states still faced challenges trying to reach internal agreement among agency officials and caseworkers on the design of a system, resulting in a delay in development. In New York—a state where the counties are responsible for Page 19 GAO-03-809 Child Welfare administering child welfare services—the development of SACWIS was stalled when significant frustration with the system’s design led Commissioners from five large counties and New York City to request that the state stop SACWIS development until a reassessment of the design and plans for the implementation of the system was completed. After a thorough evaluation of the project, the state made changes to the project plan and developed statewide work groups to ensure all counties were in agreement with the system design. In addition, they hired a contractor to monitor system development and ensure that all users’ requirements are seriously considered. Similarly, despite states’ heavy reliance on contractors, many reported that securing contractors with knowledge of child welfare practice was a challenge for timely SACWIS development. Contractors are hired by the state for their system development knowledge but often are unfamiliar with child welfare policies and practices, especially since they vary from state to state. Officials in Colorado, for example, said they encountered difficulties with their contractors because of high turnover among the contractor staff and their lack of knowledge of child welfare policies. A contractor who has worked with 7 states to develop their SACWIS reported that contractors are asked to learn the child welfare business practices of a state in a short amount of time and that states cannot devote many resources, such as caseworkers, to help in the design process because caseworkers need to devote their time to providing services to children and families. Therefore, contractors often have to acquire knowledge on their own. Many states reported that creating a system that reflects child welfare work processes and is user friendly was a challenge in developing SACWIS. These issues were also identified in the federal reviews of states’ SACWIS. For example, one state explained in the SACWIS review that it had designed a system to meet the caseworkers’ needs and reflect the nature of the child welfare work processes by developing a system that required events to be documented as they occurred. However, this design limited the SACWIS’s functionality because it did not allow the caseworkers to go back and enter information after an event happened. The state explained that caseworkers do not use the system in real time, but provide services to the children and families and then record the information in the system. The state had to redesign the system to correct for this design flaw. In addition, the 14 states reporting that they have adapted a system from another state have experienced some challenge modifying the systems to reflect their work processes. While HHS advises states to consider adapting another state’s system if it requires few Page 20 GAO-03-809 Child Welfare changes, states report that they and their contractors were not always aware of the changes that would need to be made to adapt a system. Although Colorado and New York decided to modify another state’s SACWIS instead of designing a new system, child welfare officials in these states reported that adapting a SACWIS from another state has created more problems than anticipated. Colorado and New York adapted systems from state-administered child welfare agencies, which required extensive modifications to meet their needs as county-administered states. For example, Colorado needed a system that supported its administrative structure and could handle a larger number of cases. HHS Provides Some HHS has assisted states in a variety of ways in developing and completing Assistance to Help States their SACWIS.11 As a part of its regulatory responsibilities, HHS must Meet SACWIS review, assess, and inspect the planning, design, development, installation, and operation of SACWIS. In addition to reviewing and monitoring states’ Requirements APDs, HHS conducts on-site SACWIS reviews to comply with these responsibilities. HHS officials told us that these reviews are a detailed and thorough assessment of state systems to ensure the systems’ compliance with SACWIS requirements. In addition, officials reported that they provide verbal technical assistance during the on-site review to help states that do not fully conform with the applicable regulations and policies. At the time of our review, HHS had conducted 26 SACWIS reviews—5 of which were determined as meeting all the requirements and classified as complete. HHS officials told us that since states have the flexibility to build a SACWIS that meets their needs, a large portion of the formal reviews concentrate on ensuring that the systems conform to state business practices. For example, while SACWIS regulations require that a state report all AFCARS data from their SACWIS, one state HHS reviewed relied on a separate state system to report data on the children served by the juvenile justice agency who are eligible for IV-E foster care funds. The state proved it had developed an automated process to merge data from both systems to compile a single AFCARS report that included children captured in both their SACWIS and juvenile justice systems. Therefore, 11 With regard to the budget difficulties that states reported facing, since 1994 the federal government has made a commitment to help states develop and maintain their SACWIS by matching 75 percent of states’ development funds through 1997 and providing an ongoing match of 50 percent of state funding for the development and maintenance of their systems. However, since the states’ legislatures must make the initial commitment to fund SACWIS, the federal government cannot assist state child welfare agencies with this challenge. Page 21 GAO-03-809 Child Welfare HHS recognized that this process best met the state’s needs and determined the SACWIS to be complete and meeting all requirements. Few systems have been determined complete after an on-site review because of unresolved issues, such as not being able to build links to other state information systems or not implementing certain eligibility determination functions. To help states address some of these development challenges, the SACWIS review team provides the state with recommendations for complying with SACWIS requirements. For example, HHS observed during a review in one state that the SACWIS was available statewide, but information collected in one county was not available to caseworkers in other counties. The federal officials offered recommendations to the state to meet the SACWIS requirement that all information be available statewide. In addition, HHS officials reported that once the draft report with the results of the SACWIS review are completed, federal staff schedule a conference call with the state officials to walk through the system’s deficiencies and offer guidance on how the state can move forward. HHS facilitates the sharing of information between states developing SACWIS through an automated system users group that allows state and federal officials to exchange information, ideas, and concerns. According to some state and HHS officials, the trust level at these meetings is very high, which promotes open discussions and also creates an atmosphere for informal dialogue with HHS. The systems users group developed out of another active group—the child welfare users group—when HHS solicited state representatives to help HHS define a model child welfare information system, which was later used as the basis for the SACWIS functional requirements after the passage of the 1993 legislation authorizing enhanced federal funding. State officials in Iowa and New York reported that the systems users group continues to play an important role in providing a forum for the honest exchange of information on SACWIS development. For example, child welfare and technical officials in New York said that the systems users group has been very beneficial because they have learned from other states’ positive and negative experiences in developing SACWIS, as well as the experiences unique to states with county-administered agencies. In addition to the users group, HHS officials also sponsor a listserv—an electronic mailing list—that allows state officials to exchange information, and a monthly conference call with state information technology directors. Iowa child welfare information Page 22 GAO-03-809 Child Welfare technology officials said that they find the monthly SACWIS telephone conference call helpful because project managers discuss issues such as promising practices and new regulations.12 Technical assistance for SACWIS development is also available to states through the National Resource Center for Information Technology in Child Welfare (Resource Center). According to survey results, 9 states said they used the Resource Center for assistance in developing SACWIS and 14 states reported using it for help with SACWIS maintenance and improvements. According to Resource Center officials, they assist states with SACWIS development by helping states understand the technology that is available for use, providing information on the automation of child welfare work and converting data, and reviewing the APD documentation. For example, the Resource Center offered technical assistance to Pennsylvania to help the state decide if it should continue development of its current SACWIS, abandon the SACWIS project and allow the counties to operate individual systems, or design a different SACWIS. The Resource Center evaluated the current SACWIS to determine if it could capture information based on the SACWIS regulations and if it was user friendly for the caseworker. Following the Resource Center’s analysis, Pennsylvania decided to discontinue the existing SACWIS and develop a new SACWIS. When the Resource Center opened in 1999—5 years after many states started developing SACWIS—staff were not very familiar with many of the efforts states made during development. In an attempt to remedy this lack of knowledge on states’ issues developing SACWIS, Resource Center staff participated in some of the on-site SACWIS reviews conducted by HHS. Both HHS and Resource Center officials believe this exposure to the SACWIS systems enhanced the availability of technical assistance resources and knowledge available to the states. 12 In commenting on a draft of this report, HHS indicated that a Web resource is available to states interested in learning about other states’ efforts to develop human services—child welfare, food stamps, Temporary Assistance to Needy Families, child care, and child support enforcement—information systems at http://www.acf.hhs.gov/nhsitrc. Page 23 GAO-03-809 Child Welfare Several factors affect states’ ability to collect and report reliable13 data on Several Factors Affect children served by state child welfare agencies, and some problems exist, the States’ Ability to such as a lack of clear and documented guidance, with HHS’s oversight and technical assistance. Almost all of the states responding to our survey Ensure Reliable Data reported that insufficient caseworker training and inaccurate and on Children’s incomplete data entry affect the quality of the data reported to HHS.14 In addition, 36 of the 5015 states that responded to our survey reported that Experiences, and technical challenges, such as matching their state data element definitions Some of HHS’s to HHS’s data categories, affected the quality of the data that they report to Oversight and the federal government. For example, North Carolina officials told us that while state policy mandates that they count every location in which a child Assistance Is resides, including hospital stays, AFCARS regulations say that hospital Problematic stays and other short-term placements should not be included in the count of foster care placements. In cases where state policy differs from federal policy, state officials must carefully re-format their data in order to meet federal reporting requirements. Similarly, during assessments of 6 states’ compliance with AFCARS reporting standards, HHS found that these issues affect data reliability. Despite the assistance that HHS offers to states, such as testing state data quality and providing the results to states to aid them in resubmitting data, states report ongoing challenges receiving clear and documented guidance and accessing technical assistance. 13 Data are reliable when they are complete and accurate. A subcategory of accuracy is consistency. Consistency refers to the need to obtain and use data that are clear and well- defined enough to yield similar results in similar analysis. See U.S. General Accounting Office, Assessing the Reliability of Computer-Processed Data, GAO-02-15G (Washington, D.C.: Sept. 2002). 14 States were asked the extent to which certain problems may decrease the quality of the data submitted to AFCARS and NCANDS using the following scale: very great, great, moderate, some, and no affect. 15 The analysis of survey responses about reporting data to HHS is based on responses from 49 states and the District of Columbia. All states, regardless of SACWIS development, were asked to complete these questions. Page 24 GAO-03-809 Child Welfare Insufficient Caseworker Almost every state responding to our survey and all the states we visited Training and Inaccurate reported that insufficient training for caseworkers and inaccurate and and Incomplete Data Entry incomplete data entry affect the quality of the data reported to AFCARS and NCANDS (see fig. 1). Although most states reported these as separate Are the Most Common factors, HHS and the states we visited found that insufficient training and Factors That Affect Data inaccurate and incomplete data entry are often linked. For example, in Reliability official reviews of states’ information systems capability to capture data and report them to AFCARS, HHS advised states to offer additional training to caseworkers on several AFCARS data elements, such as recording the reasons for a child leaving foster care, to improve the accuracy of the data submitted. Similarly, Oklahoma reported that the state found that caseworkers were misinterpreting reports of policy violations by foster parents and inaccurately recording them as abuse or neglect allegations. However, state officials told us that training is typically one of the first programs cut when states face tight budget restrictions. For example, Iowa officials told us that training has been significantly reduced in recent years because of budget cuts and new workers may wait 2 to 3 months before being trained how to enter data appropriately into their SACWIS. Page 25 GAO-03-809 Child Welfare Figure 1: Most Common Caseworker Issues That Affect Data Quality States reporting at least some affect in data quality 50 49 49 46 45 43 42 15 40 18 12 11 17 12 30 20 34 33 31 31 29 31 10 0 Inaccurate data Insufficient Incomplete data entry by training for entry by caseworkers caseworkers caseworkers AFCARS NCANDS Source: GAO survey. Notes: Based on responses from 50 states. The results reported in the figure are a sum of the states that reported the issue had a very great affect, great affect, moderate affect, or some affect on the quality of state data submitted to HHS. Very great and great affect responses are represented in the top section of each bar. Moderate and some affect responses are represented in the bottom section of each bar. States not included answered “no affect,” “don’t know,” or “no answer.” Inaccurate and incomplete data entry can also result from a number of other factors, such as caseworkers’ hesitation to ask families for sensitive information. For example, caseworkers in Oklahoma reported that they did not feel comfortable asking if a child’s mother was married at the time of birth or if a child is of Hispanic origin—both of which are required AFCARS data elements. In commenting on a draft of this report, Oklahoma added that caseworkers did not understand why the data elements were required and how the federal government used the information. In addition, Iowa state officials said that caseworkers may guess the racial backgrounds of children in their care or record them as unknown, especially when children come from mixed racial backgrounds, rather than asking the family for the information. HHS noted similar issues in Page 26 GAO-03-809 Child Welfare 5 states that have had an AFCARS review.16 Caseworkers were inaccurately recording a child’s race as “unable to determine” even though this option should be selected only if the child’s parents or relatives cannot provide the information, such as when a child is abandoned.17 Caseworkers, supervisors, and managers in the 5 states we visited reported that additional factors, such as difficulties balancing data entry with the time that they spend with the families and children, contributed to inaccurate or incomplete data entry. In addition, our recent work on caseworker recruitment and retention found that caseworkers struggle to balance the time they spend with children and data entry, and reportedly spend at least 50 percent of their time documenting case records.18 Supervisors in Iowa explained that since caseworkers are responsible for ensuring that children and their families receive the services they need, the caseworkers tend to initially limit data entry to the information that is necessary to ensure timely payment to foster care providers, and complete all other data elements when the caseworkers have time. In addition, caseworkers in Colorado said that they are between 30 and 60 days behind in their data entry, so the information in the automated system may not accurately reflect the current circumstances of children in care. The caseworkers reported that they tend to concentrate only on entering data that will allow them to open a case in their SACWIS. HHS’s Inspector General recently issued a report in which more than two-thirds of the states reported that caseworkers’ workloads, turnover, a lack of training, and untimely and incomplete data entry affected the reporting of AFCARS data.19 16 We reviewed AFCARS reports from 6 of the 8 states assessed by HHS—Arkansas, Connecticut, New Mexico, Texas, Vermont, and Wyoming. HHS conducted reviews in Delaware and West Virginia after we completed our analysis. 17 In commenting on a draft of this report, ACF said that the finding from the AFCARS reviews indicates that information is often defaulted to the response “unable to determine” in order for the element not to fail the missing data standard, not that workers are recording “unknown”; however, the report findings we used in this analysis instruct states to fix the defaults and address caseworker practice by enhancing training on the correct use of “unable to determine” when noting a child’s race. 18 See U.S. General Accounting Office, Child Welfare: HHS Could Play a Greater Role in Helping Child Welfare Agencies Recruit and Retain Staff, GAO-03-357 (Washington, D.C.: Mar. 31, 2003). 19 Department of Health and Human Services, Office of Inspector General, Adoption and Foster Care Analysis and Reporting System (AFCARS): Challenges and Limitations, OEI-07-01-00660 (Washington, D.C.: Mar. 2003). Page 27 GAO-03-809 Child Welfare Technical Challenges, such In addition to data quality being affected by caseworker issues, many as Matching State states experienced technical challenges reporting their data to HHS. The Definitions to Federal problems reported by states are typically a result of challenges associated with data “mapping”—matching state data elements to the federal data Definitions, Affect Data elements. For example, 36 states reported in our survey that matching Reliability their state-defined data to HHS’s definitions affected the quality of the data reported to NCANDS and AFCARS. Similarly, 24 states reported that matching the more detailed data options available in their states’ information systems to the federal data elements affected the quality of the data reported to NCANDS. Twenty-nine states reported that this issue created challenges in reporting data to AFCARS. For example, following an AFCARS assessment, HHS instructed a state that collects detailed information on children’s disabilities, such as Downs Syndrome, Attention Deficit Disorder, and eating disorders, to map the information to the more limited options in AFCARS, such as mental retardation and emotionally disturbed. The Inspector General’s report found that states faced similar challenges mapping their data to meet the AFCARS reporting requirements. In many cases, states have to balance state policy with federal requirements to ensure that they are reporting accurate data to AFCARS and NCANDS, but are not contradicting their state policies. For example, Texas officials reported that although the findings of their AFCARS review instructed them to modify their SACWIS to collect, map, and extract data on guardianship placements, the state does not support guardianship arrangements.20 In addition, a recent report from the Child Welfare League of America (CWLA) found that when reporting the number of times children move from one foster care placement to another, states varied in the type of placements included in that count.21 For example, 29 percent of the states responding to CWLA’s survey included respite,22 25 percent included runaways, and 16 percent included trial home visits when reporting the number of placements a child had during the AFCARS report period. According to federal guidance, the “number of placements” element is meant to gather information on the number of times the child 20 Guardianship arrangements occur when permanent legal custody of a child is awarded to an individual, such as a relative, but the child is not legally adopted. 21 Child Welfare League of America. National Working Group Highlights, “Placement Stability Measure and Diverse Out-of-Home Care Populations” (Washington, D.C., Apr. 2002). 22 Respite care provides temporary childcare for children away from their caretakers. Page 28 GAO-03-809 Child Welfare welfare agency found it necessary to move a child while in foster care and that by including runaways or trial home visits, a state is inflating the number of moves a child experienced. However, North Carolina officials told us that although the federal definition for placements instructs states not to include such stays when counting the number of children’s foster care placements, the state instructs them to count each time a child is sleeping in a different place as a new placement. The Inspector General reported that the placement definitions were the most commonly cited source of confusion among the states surveyed.23 In addition to the challenges reported in our survey, HHS reported that transferring data from older data systems into SACWIS affects the quality of the data reported to AFCARS and NCANDS. HHS officials reported that they have observed that states experience the biggest change in data quality when they begin reporting from their SACWIS. In general, the first data submissions are of low quality because of the time it takes states to transfer data or the system re-sets the information for data elements. For example, in 1 state, 65 percent of the records reviewed by HHS during an AFCARS assessment recorded the date the children were removed from their homes as July 28, 1997—the date the SACWIS came on-line; however, the actual dates of removal for these children ranged from 1988 to 1997. Although HHS Has Taken HHS provides technical assistance for AFCARS and NCANDS reporting Steps to Help States through a number of resources. HHS officials in the central office and Improve Their Data, Some NCANDS contractor staff serve as the points of contact for states to ask questions and seek guidance on reporting child welfare data. HHS officials Problems with Its Efforts reported that assistance is offered in a number of ways, including Exist telephone and e-mail communication. The officials in 3 of the 5 states that we visited said that the one-on-one focused technical assistance was useful when provided in a timely fashion. Most state officials found the NCANDS data easier to report, in part because more people were available for consultation and they were more accessible and responsive. For example, states have access to four NCANDS specialists and staff in the contractor’s central office when they need assistance reporting child abuse and neglect information. However, some of the states we visited 23 Although the findings from the Inspector General’s report and our study are more recent, ACF commented on a draft of this report that it issued policy clarifications regarding placement information on July 5, 2002. Page 29 GAO-03-809 Child Welfare reported that only one or two staff in HHS’s central office are available to assist with AFCARS reporting. In addition, the Resource Center offers states assistance with improving data quality; however, Resource Center staff reported that the assistance is geared more towards improving the limited data used in the federal review process to monitor states’ compliance with child welfare laws and federal outcome measures—CFSR—rather than all the data reported to HHS. The Resource Center also sponsors an annual information technology conference during which sessions covering all data-related issues are held, including practices for ensuring data quality and outcome evaluation in child welfare. In conjunction with the national data conference, the HHS officials and the contractors that operate NCANDS hold an annual technical assistance meeting for states to share ideas with one another, discuss data elements that pose difficulties, and explore ways to address these problems. For example, at a recent technical assistance meeting, approximately 43 state representatives attended sessions on preparing the calendar year 2002 NCANDS data submissions and received a detailed explanation of how the NCANDS staff test states’ data submissions for quality. In addition, an NCANDS state advisory group meets annually to talk with HHS officials about NCANDS data and their experiences reporting data. From these meetings, the state advisory group proposes changes or improvements to NCANDS. HHS and state officials reported that this partnership has helped ease some of the challenges in reporting child abuse and neglect data. In addition to the direct assistance through consultation with HHS officials and the Resource Center, HHS has made available to states the software it uses to examine states’ AFCARS and NCANDS submissions for inconsistencies and invalid data. Officials in all the states we visited said that they regularly use this software, and an HHS official said that nearly every state has used the software at least once. When the data are submitted to HHS, they are run through the same software, and HHS notifies the states of areas where data are missing or inconsistent and allows the states to resubmit the data after errors are corrected. For example, HHS officials said that they worked with one state that was trying to determine the source of data errors in reporting to AFCARS the race or ethnicity of children in their care. The state was not able to determine the source of the problem, so an HHS official examined the state’s submissions and helped correct the data errors. The officials reported that these tests help them to identify some data quality errors, such as missing data, and said that they believe that, in general, data have Page 30 GAO-03-809 Child Welfare improved in recent years. However, they indicated that the tests cannot pinpoint the underlying problems contributing to these errors. Furthermore, one official reported that no specific efforts have been conducted to track the individual data elements and, therefore, HHS cannot report on how data quality has changed over time. The results of these quality tests had been the basis for penalties levied against states that submitted low quality AFCARS data before the penalties were rescinded. HHS officials reported that the penalties served as an effective motivation to states to correct their data. Although HHS was not able to report how the lack of penalties might be affecting recent data quality, an official reported that the agency plans to conduct this analysis in the future. In an attempt to help states comply with the reporting standards and address some of the factors that contribute to data quality problems, HHS performs comprehensive reviews of state information systems’ ability to capture AFCARS data to identify problems associated with data collection and reporting, and to ensure that the information in the automated system correctly reflects children’s experiences in care. The assessments include a technical review of the states’ computer code, a comparison of the data from selected cases available in the information system to the case files, and an improvement plan to resolve any errors. In addition, HHS officials offer guidance to the states on improvements that can be made to the information system and changes to program code used to report the AFCARS data. HHS conducted pilot reviews in eight states between 1996 and 2000. By March 2003, HHS had conducted eight official reviews—even though states began reporting to AFCARS in 1995. According to results from six of the eight official AFCARS assessments we reviewed, no state met the reporting requirements for all AFCARS data elements. Table 8 shows a selection of the data elements and the states’ ratings. The problems noted in the reviews are similar to those we heard from states responding to our survey and those we visited. For example, most states Page 31 GAO-03-809 Child Welfare received ratings of 2 or 3, indicating technical and/or data entry errors that affect the AFCARS data quality.24 Table 8: Selected AFCARS Elements and Six States’ Levels of Compliance in Meeting Reporting Requirements Rating factora Data Elements (# of states) Foster Care Child race 2 (4 states) (American Indian or Alaska native, Asian, Black or African American, Native Hawaiian or 3 (1 state) other Pacific Islander, White, unable to determine) 4 (1 state) Has the child been diagnosed with a disability? 2 (6 states) (yes, no, not yet determined) Mental retardation, visually/hearing impaired, physically disabled, emotionally disturbed, 2 (5 states) other diagnosed conditionb 3 (1 state) (does not apply, applies) Has child ever been adopted? 1 (1 state) (yes, no, unable to determine) 2 (3 states) 3 (2 states) c Total number of removals from home 2 (3 states) 3 (2 states) 4 (1 state) Date of discharge from previous episodec 2 (3 states) 3 (2 states) 4 (1 state) Date of latest removalc 2 (3 states) 3 (3 states) Date of placement in current setting 2 (5 states) 4 (1 state) c Number of previous placement settings in this episode 2 (6 states) Current placement setting 2 (4 states) (pre-adoptive home, foster family home-relative, foster family home-nonrelative, group 3 (1 state) home, institution, supervised independent living, runaway, trial home visit) 4 (1 state) Out of state placement (yes, no) 2 (4 states) 3 (1 state) 4 (1 state) 24 A rating of 2 indicates that the state’s information system does not fully meet the requirements for AFCARS reporting, whereas if a state receives a 3, the information system requirements for AFCARS reporting are in place, but there are data entry problems affecting the quality of data. According to an HHS official, data elements that have a combination of technical and data entry problems are rated as 2 until the technical issues are resolved. HHS will then rate the element as a 3 until the data entry practices are changed. A state receives a rating of 4 if in compliance with the AFCARS requirements or a 1 if the requirement is not addressed in the state’s information system. Page 32 GAO-03-809 Child Welfare a Rating factor Data Elements (# of states) Most recent case plan goal 1 (1 state) (reunify with parents or principal care takers, live with other relative(s), adoption, long term 2 (5 states) foster care, emancipation, guardianship, case plan goal not yet established) c Date of discharge 2 (2 states) 3 (2 states) 4 (2 states) Reason for dischargec 2 (4 states) (not applicable, reunification with parent(s) or primary caretaker(s), living with other 3 (1 state) relative(s), adoption, emancipation, guardianship, transfer to another agency, runaway, 4 (1 state) death of child) Adoption Child race 2 (4 states) (American Indian or Alaska native, Asian, Black or African American, Native Hawaiian or 3 (1 state) other Pacific Islander, White, unable to determine) 4 (1 state) Primary basis for determining special needs 2 (6 states) (not applicable; racial/original background; age; membership in a sibling group; medical conditions or mental, physical, or emotional disabilities; other) Mental retardation, physically disabled, emotionally disturbedb 2 (5 states) (does not apply, applies) 3 (1 state) Visually/hearing impaired 2 (4 states) (does not apply, applies) 3 (2 states) Other diagnosed condition 2 (6 states) (does not apply, applies) Source: HHS. Note: Analysis based on results of AFCARS reviews in 6 states. States are rated on 66 foster care elements and 37 adoption elements. a HHS rates each data element using a four-point scale: (1) the AFCARS requirement(s) has not been implemented in the information system; (2) the technical system requirements for AFCARS reporting do not fully meet the standards; (3) the technical system requirements for AFCARS reporting are in place, but there are data entry problems affecting the quality of the data; (4) all of the AFCARS requirements have been met. b States are rated on each category. The data elements were combined for purposes of this analysis. c AFCARS data element used in the CFSR. State officials in these 6 states reported that they found the reviews useful for improving their AFCARS data submissions. In particular, they valued the thorough review by HHS officials of the computer code states use to report the data. Some of these officials reported that if all states were reviewed, the quality of data available in AFCARS would improve tremendously. However, HHS officials reported that they are not mandated to conduct the AFCARS reviews and that priority is placed on other reviews, such as the CFSR and SACWIS reviews. In addition, officials explained that the AFCARS reviews are not conducted in states developing SACWIS until the systems are operational. HHS expects to Page 33 GAO-03-809 Child Welfare complete approximately four reviews each year depending on available resources and has scheduled states through 2006. Similar to the SACWIS reviews, HHS officials offer recommendations and technical assistance to states during the review on how they can improve the quality of the data reported to AFCARS. Although the states we visited appreciated some of HHS’s efforts to assist with improving state data quality, they and most states responding to our survey agreed that the assistance is not always consistent or easily accessible (see fig. 2). States reported similar information to the Inspector General—AFCARS data elements were not clearly and consistently defined and technical assistance is effective but difficult to access. Page 34 GAO-03-809 Child Welfare Figure 2: Federal Practices That Affect Data Quality States reporting at least some affect in data quality 50 45 41 40 17 35 35 11 6 30 9 25 5 20 17 30 29 28 4 26 20 10 13 0 Imprecise Lack of clear, Difficulty definitions for documented accessing required fields guidance technical assistance and guidance AFCARS NCANDS Source: GAO survey. Notes: Based on responses from 50 states. The results reported in the figure are a sum of the states that reported the issue had a very great affect, great affect, moderate affect, or some affect on the quality of state data submitted to HHS. Very great and great affect responses are represented in the top section of each bar. Moderate and some affect responses are represented in the bottom section of each bar. States not included answered “no affect,” “don’t know,” or “no answer.” The primary concerns reported by the states we visited were delays in receiving clear, written guidance on defining and reporting certain data elements and the lack of state input in suggesting changes to AFCARS. Despite the written guidance available to states in the form of regulations and an on-line policy manual, states reported that the variation in state policies and practices make it difficult to interpret how to apply the general guidance. As a result, states consult with HHS to ensure they are applying the regulations appropriately. However, in commenting on a draft of this report, officials in Oklahoma told us that a common concern among the states is the lack of timely response from HHS when seeking guidance on how to report data. In addition, officials in New York explained they Page 35 GAO-03-809 Child Welfare have made it a practice to check the HHS Web site on a regular basis for current guidance, but have not found it a useful tool, and may turn to other states for guidance on AFCARS reporting. In commenting on a draft of this report, HHS explained that it first refers states to its Web site for information and believes that the available guidance addresses states’ concerns in most instances. In addition, the states that have had an AFCARS review experienced delays in obtaining guidance on how to proceed following the on-site review. Although they found the review to be very helpful, some states reported that HHS officials are delayed in responding to their questions. For example, Texas officials reported that the state sought clarification on its improvement plan and submitted additional questions to HHS following the review; however, when we spoke with the state officials, they said that they had been waiting 3 months for a response on how to proceed. An HHS official told us that since the review process is relatively new, the agency is still developing a process to respond to the states and recognizes that it has not been responsive to the states already reviewed. In addition, HHS is taking steps to gather feedback from states and other users of AFCARS data to determine how to improve the system to make the data more accurate and useable. As a part of these efforts, HHS has published a Federal Register notice soliciting comments and held focus group meetings at national conferences. The difficulties states face in receiving federal guidance and assistance, as well as the other challenges they face in reporting data, may negatively impact the reliability of the data available in AFCARS and NCANDS. As a result, states are concerned that the national standards used in the CFSR are based on unreliable data and should not be used as a basis of comparison and potential financial penalty. The variation in states’ reporting practices may affect the validity of the measures and may place some states at a disadvantage. For example, the CWLA and Inspector General studies found that approximately half the states include the juvenile justice population in their AFCARS reports, while the other states Page 36 GAO-03-809 Child Welfare do not.25 Child welfare experts and some state officials believe that the states that include children served by the juvenile justice agency in their AFCARS report may report a higher number of re-entries into the child welfare system or a higher number of moves within the system when compared to states that do not have IV-E agreements26 with their juvenile justice systems.27 As a result, a state that includes such children in their AFCARS report are likely to fare less favorably when compared to the national standard than other states on two outcome measures—foster care re-entries and stability of foster care placements—and may face financial penalties associated with the CFSR. 25 Federal guidance states that children who are (1) removed from their home and initially placed in a juvenile justice facility are not to be included in AFCARS reporting; (2) in a foster care setting who are moved to a juvenile justice facility and who are expected to be returned to a foster care setting should continue to be included in the AFCARS reporting population; and (3) in a foster care setting and are moved to a juvenile justice facility and who become the responsibility of another agency should not be included in the AFCARS reporting population. However, these studies show that the arrangements between child welfare and juvenile justice agencies vary, which affect the population of children reported to AFCARS. For example, some state child welfare agencies have responsibility for all children in the juvenile justice system and include these children in their AFCARS reports, while other states only report children who are in the custody of the juvenile justice system, but receiving Title IV-E funding. 26 According to HHS, state child welfare agencies can claim Title IV-E foster care maintenance funds for eligible children for costs incurred by juvenile justice agencies that have entered into an agreement with the child welfare agencies. Delinquent children served by these agencies who meet the Title IV-E eligibility criteria are eligible for Title IV-E foster care maintenance funds. States must meet all Titles IV-B and IV-E program and/or eligibility requirements with respect to the children who are adjudicated delinquent and are receiving Title IV-E funding or for children in foster care under the placement and care responsibility of the child welfare agency. 27 Children in the juvenile justice system may enter care because of behavior problems and return home when the behavior is controlled. However, the problem may arise again and the child could re-enter state custody. Also, the children may experience a greater number of placements while in foster care because of their behavior problems or a “step-down” approach that some states use to gradually decrease the level of security required to care for the children. Page 37 GAO-03-809 Child Welfare Some states are using a variety of practices to address the challenges States Are Using associated with developing SACWIS and improving data reliability, Various Practices to although no formal evaluations are available on their effectiveness. To address the challenge of developing a system to meet statewide needs, Overcome System states relied on caseworkers and supervisors from local offices to assist in Development the design and testing of the system. Few states reported in our survey strategies to overcome the other key challenges, such as limited funding Challenges and and securing knowledgeable contractors, but some states we visited have Improve Data on devised some useful approaches. For example, Oklahoma child welfare Children’s officials—in order to maximize the limited state funding for maintaining their SACWIS—reported saving $1 million each year by hiring some of the Experiences contractors who developed their SACWIS as permanent staff. To improve data reliability, the 5 states we visited routinely review their data to identify data entry errors so that managers can ensure that the missing data are entered appropriately. In addition, some states reported that frequent use of the data, such as publishing periodic management reports detailing local offices’ performance on outcome measures, helps caseworkers understand the importance of entering timely information. States Are Primarily To overcome development challenges, survey respondents emphasized the Relying on SACWIS Users importance of including system users in the various phases of completing to Overcome Some of the SACWIS—planning, design, development, testing, and implementation. Past GAO work and other research efforts have determined similar Challenges to Completing approaches as best practices in building information systems.28 Forty-four Their Systems of the 46 states responding to our survey that they are developing or operating a SACWIS indicated that they relied on internal users, such as caseworkers and supervisors, in the development of their systems and 34 of these states said that they were extremely helpful participants. The extent to which the users were involved in development differed across the states. For example, in Texas, caseworkers from all of their child welfare regions were recruited to provide input on design and development, as well as during initial testing, pilot testing, and implementation of the system. Arkansas reported establishing a committee made up of users to review the work plan and sign off on recommended changes. In addition, states reported that their system users served a 28 See U.S. General Accounting Office, Executive Guide: Improving Mission Performance Through Strategic Information Management and Technology, GAO/AIMD-94-115 (Washington, D.C.: May 1, 1994); Center for Technology in Government, University of Albany, SUNY. Tying a Sensible Knot: A Practical Guide to State-Local Information Systems. Albany, N.Y., June 1997. Page 38 GAO-03-809 Child Welfare number of purposes, including serving as experts on the different specialties within child welfare, such as child abuse, foster care, or adoption, and as representatives from local or county offices to assist in identifying the diverse approaches to capturing information across the state. For example, Indiana reported that caseworkers involved in SACWIS development represented the unique needs of the different geographical areas of the state and helped design a uniform statewide system to meet the diverse needs of large, intermediate, and small local offices. Ten states noted that user input should not be limited to frontline workers, such as caseworkers, but should include representatives from other areas of the agency, such as the financial staff, and other agencies that serve children, such as child support enforcement.29 Since many SACWIS link with other state information systems, states advised that developing a collaborative relationship with other state agencies will help the development of the system. While not one of the most common challenges reported in our survey, New Hampshire reported that one of its challenges with meeting its SACWIS timeframe was not working collaboratively with other agencies, such as Temporary Assistance for Needy Families (TANF)30 and child support enforcement, to develop the payment component of SACWIS. Similarly, we previously reported that the difficulty developing linkages between social services agencies limits the effectiveness of all the programs to serve families.31 To attempt to overcome this challenge, 26 of the 46 states responding to our survey that they are developing or operating a SACWIS indicated that they included external public agency users and 23 reported using representatives from other state agencies that serve children in developing their SACWIS. Indiana said that a task force made up of representatives from the TANF and child support enforcement agencies was developed to design the 29 The Child Support Enforcement Program is a joint federal, state, and local partnership that was established in 1975 under Title IV-D of the Social Security Act. Each state runs a child support program, which provides four major services: locating non-custodial parents, establishing paternity, establishing child support obligations, and collecting child support for families. 30 In 1996, the Congress created the block grant Temporary Assistance for Needy Families program replacing the Aid to Families with Dependent Children (AFDC) and related welfare programs. States were given increased flexibility in designing the eligibility criteria and benefit rules, which require work in exchange for time-limited benefits. 31 See U.S. General Accounting Office, Welfare Reform: Improving State Automated Systems Requires Coordinated Federal Effort, GAO/HEHS-00-48 (Washington, D.C.: Apr. 27, 2000). Page 39 GAO-03-809 Child Welfare linkages between the systems. In addition, Colorado officials reported that they are working with the Department of Youth Corrections—an agency that shares the SACWIS with child welfare—to ensure that the shared screens use the same definitions. In addition to seeking input from caseworkers and other system users while developing SACWIS, many states continue to include users as a part of the implementation teams, to serve as contacts in the field and provide ongoing assistance, and to provide input on system enhancements. Alabama responded in our survey that the state had “mentors” in each county to help caseworkers adjust to the new system. These mentors continue to provide ongoing support now that the system is implemented. Similarly, Oklahoma developed Field Implementation Teams consisting of one contractor and one child welfare staff person. During system implementation, the teams went to field offices to provide on-site assistance with using SACWIS and becoming accustomed with the new method of recording child welfare information. Furthermore, Oklahoma recruits experienced child welfare field staff for its SACWIS help desk because of their knowledge of the system and child welfare policy and practice. Although states faced other challenges in completing their SACWIS, few reported implementing approaches to overcome the barriers. According to survey results, a common problem states faced in developing SACWIS was receiving insufficient state funding for development. However, in our previous work on managing information technology, we found that the IT products can become obsolete in a matter of months rather than years, calling for more frequent investments in upgrades and enhancements.32 In addition, officials in Iowa told us that maintaining systems takes just as much money as building them. States did not report in our survey approaches for obtaining more funding for developing SACWIS, and few states reported developing strategies in an attempt to overcome the challenges associated with tight budgets for maintaining their systems. For example, Iowa officials engaged in careful planning with system users to ensure that they addressed the highest priorities when enhancing the system. In particular, the officials reported that maintaining tight control over the development and maintenance processes helps them avoid investing inordinate amounts of resources to make corrections to the 32 See U.S. General Accounting Office, Managing Technology: Best Practices Can Improve Performance and Produce Results, GAO/T-AIMD-97-38 (Washington, D.C.: Jan. 31, 1997). Page 40 GAO-03-809 Child Welfare system. In Oklahoma, child welfare officials reported that they relied on the contractors who developed their SACWIS to conduct ongoing maintenance activities until the contract expired in 2001. At that time, the agency hired some of the contract staff as full-time state employees to continue with the maintenance activities. State officials explained that this approach ensured continuity of service, in addition to saving the agency approximately $1 million each year. Similarly, few states reported on approaches to overcome the challenge of finding contractors with knowledge of child welfare practice. However, Iowa officials explained that once the contract staff are hired, they are required to attend the same training as new caseworkers to ensure that they are familiar with the state’s child welfare policies and to familiarize themselves with casework practices. States Use Strategies, such Twenty-eight states reported using approaches to help caseworkers as Producing Reports That identify the data elements that are required for federal reporting and to Identify Missing Data, in an help them better understand the importance of entering timely and accurate data. Ten states responding to our survey reported reviewing the Attempt to Improve the federal reporting requirements in training sessions as a promising Reliability of the Data approach they use to improve data quality or as a lesson learned. For Reported to HHS example, Tennessee reported that the state added a component about AFCARS to the initial and ongoing training workers receive about using SACWIS. The curriculum addresses the AFCARS report in general and the individual data elements to help the caseworkers better understand the purpose of collecting the information. In Nebraska, a “desk aid” that explains the data elements and where and why to enter them in the system is available on the caseworkers’ computer desktops. In addition, New York has developed a step-by-step guide explaining to workers how NCANDS data should be entered, with references to the policy or statute requiring the information. To improve data reliability, some states have designed their information systems with special features to encourage caseworkers to enter the information. Four states responding to our survey and 3 states we visited designed their SACWIS with color-coded fields to draw attention to the data elements that caseworker are required to enter. For example, the AFCARS data fields in Oklahoma’s system are coded red until the data are entered, after which the fields change to blue. In addition, workers can look at a single screen in the Oklahoma system to see what AFCARS data elements need to be completed without having to scroll through the entire case record. Colorado, Iowa, New York, and Oklahoma have built into their systems alerts—also known as “ticklers”—to remind caseworkers Page 41 GAO-03-809 Child Welfare and supervisors of tasks that they need to complete. For example, in Iowa, alerts are sent to supervisors if a caseworker fails to enter the data necessary to complete a payment to a foster care provider. Whereas, in Oklahoma, a stoplight icon on the caseworker’s computer desktop reminds the worker when tasks are due. A green light indicates that nothing is due within 5 days; a yellow light means that something is due within 5 days; and a red light means that something is overdue. Caseworkers and supervisors in the states we visited had mixed responses about the usefulness and effectiveness of the alerts. Some caseworkers found them to be a nuisance, while other caseworkers and supervisors found them to be useful tools in managing workloads and prioritizing daily tasks. Six states reported that the best way to improve data quality was to use the data in published reports and hold the caseworkers and supervisors accountable for the outcomes of the children in their care. In addition, 6 states responding to our survey reported using the data available in their information systems to measure state outcomes similar to the CFSR. State officials reported that this approach is an effective way to get local offices invested in the quality of the data. For example, North Carolina publishes monthly reports for each county comparing their performance on state data indicators, such as the length of time children spend in care, to counties of similar size and the state as a whole. County officials reported that these reports encourage workers to improve the quality of the data collected and entered into the state system since their performance is being widely published and compared to other counties. In addition, all the states we visited reported that frequent review of their data, such as using software from HHS to test their AFCARS and NCANDS data to pin-point data entry errors prior to submitting them to HHS, has helped improve data quality. When the states identify poor data, they alert the caseworkers and supervisors of needed corrections and data entry improvements. For example, Colorado runs these reports about 4 to 5 times a year, with one run occurring approximately 6 weeks before each AFCARS submission. When the data specialists find errors, they notify the caseworker to clean up the data. New York officials told us that they incorporate the results from these tests in training if a consistent pattern of errors is identified. While most states are developing statewide information systems, Conclusion challenges with data reliability remain. Although SACWIS development is delayed in many states, state officials recognize the benefits of having a Page 42 GAO-03-809 Child Welfare uniform system that enhances the states’ ability to monitor the services provided and the outcomes for children in their care. Although states began reporting to NCANDS in1990 and were mandated to begin reporting to AFCARS in 1995, most states continue to face challenges providing complete, accurate, and consistent data to HHS. In addition, the results of more recent HHS efforts, such as conducting AFCARS-related focus groups, are unknown. Reliable data are essential to the federal government’s development of policies that address the needs of the children served by state child welfare agencies and its ability to assist states in improving child welfare system deficiencies. Without well- documented, clearer guidance and the completion of more comprehensive reviews of states’ AFCARS reporting capabilities, states are limited in overcoming challenges that affect data reliability. Because these challenges still remain, HHS may be using some questionable data as the foundation for national reports and national standards for the CFSR and may not have a clear picture of how states meet the needs of children in their care. To improve the reliability of state-reported child welfare data, we are Recommendation to recommending that the Secretary of HHS consider, in addition to HHS’s the Secretary of recent efforts to improve AFCARS data, ways to enhance the guidance and assistance offered to states to help them overcome the key challenges in Health and Human collecting and reporting child welfare data. These efforts could include a Services stronger emphasis placed on conducting AFCARS reviews and more timely follow-up to help states implement their improvement plans or identifying a useful method to provide clear and consistent guidance on AFCARS and NCANDS reporting. We obtained comments on a draft of this report from the Department of Agency Comments Health and Human Services’ Administration for Children and Families (ACF). These comments are reproduced in appendix III. ACF also provided technical clarifications, which we incorporated when appropriate. ACF generally agreed with our findings and commented that the report provides a useful perspective of the problems states face in collecting data and of ACF’s effort to provide ongoing technical assistance to improve the quality of child welfare data. In response to our recommendation, ACF said that we categorized its efforts as “recent” and did not recognize the long-term efforts to provide AFCARS and NCANDS related guidance to the states. Although we did not discuss each effort in depth, we do mention Page 43 GAO-03-809 Child Welfare the agency’s ongoing efforts in our report. However, we refer to the recent efforts in the recommendation in recognition of the agency’s current activities to formally obtain, document, and incorporate feedback from the states with regard to collecting and reporting adoption and foster care data. ACF also noted in its comments that the data definitions need to be updated and revised and said it is currently in the process of revising the AFCARS regulations to further standardize the information states are to report—which we acknowledge in our report. In addition to the steps HHS is taking to further improve the AFCARS data, our recommendation encourages HHS to consider ways to enhance the ongoing guidance and assistance offered to states to help them overcome the key challenges in collecting and reporting child welfare data. ACF requested specific recommendations on approaches to overcome the difficulty of collecting and merging information from multiple state and county programs into a single national database. While there may be additional methodologies that the agency could use to overcome such challenges, our recommendation focuses on improving the guidance already offered to the states as a step to helping them better comply with the reporting requirements. In addition, ACF added that although staff turnover in state child welfare agencies is a significant contributor to data quality issues, we did not focus on this as a significant factor. ACF also commented that it is firmly committed to continue to support the states and to provide technical assistance and other guidance as its resources will permit. However, because we recently issued a detailed report on a variety of caseworker issues,33 we primarily focused in this report on the key data entry challenges caseworkers face and refer readers to our previous work for additional information on challenges related to caseworker recruitment and retention and their affect on child welfare agencies. In commenting on our previous work prior to its release, HHS indicated that it does not have the authority to require states to address factors that contribute to staff turnover, such as high caseloads and said that it has limited resources to assist the states in the area of staff recruitment and retention. ACF commented that it provided increased funding to the National Resource Centers in fiscal year 2003, which they believe will improve 33 See U.S. General Accounting Office, Child Welfare: HHS Could Play a Greater Role in Helping Child Welfare Agencies Recruit and Retain Staff. GAO-03-357. (Washington, D.C.: Mar. 31, 2003). Page 44 GAO-03-809 Child Welfare ACF’s ability to provide assistance to the states. After receiving the draft report for comment, HHS separately provided information on an additional service the National Resource Center for Information Technology in Child Welfare provides to states. In an effort to assist states with improving the quality of their AFCARS data, the Resource Center will review states’ programming code used for AFCARS data. As of June 2003, HHS reported that the Resource Center provided this assistance to Arkansas, Louisiana, Mississippi, North Carolina, Nevada, New Jersey, and Rhode Island, and 3 states—Maryland, Michigan, and Wisconsin—and the District of Columbia have requested the assistance. In response to our survey methodology, ACF requested that we explain why the territory of Puerto Rico was not included in the state survey. Although Puerto Rico receives federal child welfare funds, we traditionally focus on the states and therefore do not include the U.S. territories, including American Samoa, the Commonwealth of the Northern Mariana Islands, Guam, Puerto Rico, and the Virgin Islands, in the scope of our reviews. Finally, in response to our discussion of the AFCARS review process, ACF provided a few clarifications. During the course of our review, an HHS official characterized the AFCARS review process as relatively new and explained that the agency is still developing a process to respond to the states following the completion of the on-site review. When responding to a draft of this report, ACF disagreed with this characterization. ACF commented that the review process has been in place since 1996, pointing to the pilot reviews as evidence that the agency has a defined process. However, when we requested AFCARS reports for review, HHS explained that the states undergoing pilot reviews would be re-reviewed and that the official process was formalized in 2001 with the release of an AFCARS review guide and the start of the official reviews. In addition, ACF commented that SACWIS reviews do not take priority over AFCARS reviews. However, officials had previously explained that although SACWIS and AFCARS reviews can happen at the same time, in practice, the AFCARS reviews are scheduled to occur in the states that are developing SACWIS after they have participated in a SACWIS review. Furthermore, ACF explained that states do not develop their improvement plan following the conclusion of the AFCARS review. Instead, ACF officials draft the plan for the state. Although state representatives had described a challenge in receiving timely feedback on their improvement plan, we have changed the language in the report to reflect ACF’s comment. Page 45 GAO-03-809 Child Welfare We also provided a copy of our draft to child welfare officials in the 5 states we visited—Colorado, Iowa, North Carolina, New York, and Oklahoma. Iowa and New York had technical clarifications, which we in incorporated when appropriate. Oklahoma provided additional information, which was incorporated. Colorado had no suggested corrections or edits. North Carolina did not provide any comments. As agreed with your offices, unless you publicly announce its contents earlier, we plan no further distribution of this report until 30 days after its issue date. At that time, we will send copies of this report to the Secretary of Health and Human Services, state child welfare directors, and other interested parties. We will make copies available to others on request. In addition, the report will be available at no charge on GAO’s Web site at http://www.gao.gov. If you or your staff have any questions or wish to discuss this material further, please call me at (202) 512-8403 or Diana Pietrowiak at (202) 512-6239. Key contributors to this report are listed in appendix IV. Cornelia M. Ashby Director, Education, Workforce, and Income Security Issues Page 46 GAO-03-809 Child Welfare Appendix I: Scope and Methodology Appendix I: Scope and Methodology To determine the progress states have made in developing Statewide Automated Child Welfare Information Systems (SACWIS), we surveyed all 50 states and the District of Columbia through a Web-based survey. We pretested the survey instrument in Maryland and the District of Columbia. We received responses from 49 states and the District of Columbia. The state of Nevada did not respond to the survey. We discarded a question that asked states to report the date their advance planning document (APD) was approved by the Department of Health and Human Services (HHS). Due to a technical error, the date was truncated and a valid answer was not stored in the responses. Of the 50 survey responses, 46 were from states that are pursuing SACWIS development. The 4 states not developing SACWIS were asked to skip sections of the survey that asked about SACWIS development, system modifications, and supported services and links. We did not independently verify the survey responses. In addition, we visited 5 states to obtain more detailed and qualitative information regarding states’ experiences developing SACWIS. We conducted site visits in Colorado, Iowa, New York, North Carolina, and Oklahoma. We selected these states to represent a range of SACWIS development stages, sizes of foster care populations, and geographic locations. During our site visits, we interviewed state and local child welfare staff, state and local staff that regularly exchange information with the child welfare agency, and private contractors. We also spoke with HHS staff in the central and regional offices, National Resource Center officials, contractors involved in SACWIS development, and child welfare experts from the Child Welfare League of America and the American Public Human Services Association. To determine how states and HHS ensure reliable data exist on children served by child welfare agencies we surveyed states using the above- mentioned survey instrument. In addition, we interviewed state and HHS officials on their efforts to analyze and compile data and HHS’s role in providing technical assistance to states. We spoke with state officials during our site visits and HHS officials in the central and regional offices and attended the 6th National Child Welfare Data Conference. We obtained and reviewed available SACWIS and Adoption and Foster Care Analysis and Reporting System (AFCARS) reports. At the time of our review, HHS had conducted 26 SACWIS reviews. We obtained and reviewed 23 reports. The remaining reports were not available for review because HHS has not yet completed the report or shared the results with the state. Most of the SACWIS reports were considered drafts since many states are in the process of resolving issues with completing their systems. We reviewed AFCARS assessment reports from 6 of the 8 states assessed by HHS—Arkansas, Connecticut, New Mexico, Texas, Vermont, and Wyoming. HHS conducted reviews in Delaware and West Virginia after we Page 47 GAO-03-809 Child Welfare Appendix I: Scope and Methodology completed our analysis. We did not review any of the eight pilot review reports since these were not final reports and HHS plans to conduct official reviews in these states. These AFCARS assessment reports were analyzed to understand the breadth of on-site assistance HHS provides to states during the review and to identify common data collection and reporting difficulties among states. Finally, we talked with officials in 6 of the 8 states that had an AFCARS review about their experiences during the review and child welfare experts. To identify practices state and local child welfare agencies are using to help ensure the accuracy, timeliness, and completeness of child welfare data we interviewed state and local child welfare officials on our site visits and inquired about the practices they are employing. We also included questions on practices and lessons learned in our survey. In addition, we spoke with numerous child welfare experts, including individuals from the National Resource Center for Information Technology in Child Welfare, the Child Welfare League of America, and the American Public Human Services Association.1 1 The Child Welfare League of America is an association of almost 1,200 public and private nonprofit agencies that assist abused and neglected children and their families with a wide range of services. The American Public Human Services Association is an organization of individuals and agencies concerned with human services, including state and local human service agencies and individuals who work in or otherwise have an interest in human service programs. Page 48 GAO-03-809 Child Welfare Appendix II: State SACWIS Stages of Appendix II: State SACWIS Stages of Development Development SACWIS stages of development Completed Partially operational Arizona Alabama Massachusetts Florida Oklahoma Illinois Rhode Island Mississippi West Virginia New York Oregon Operational South Carolina Arkansas Tennessee California Wisconsin Colorado Connecticut Implementation Delaware Alaska District of Columbia Maryland Idaho Indiana Planning Iowa Georgia Kentucky Kansas Maine Louisiana Michigan Missouri Minnesota New Jersey Montana Ohio Nebraska Pennsylvania Nevada New Hampshire No SACWIS New Mexico Hawaii South Dakota North Carolina Texas North Dakota Utah Vermont Virginia Washington Wyoming Source: HHS. Note: Status is as of May 22, 2003. Page 49 GAO-03-809 Child Welfare Appendix III: Comments from the Department of Health and Human Services Appendix III: Comments from the Department of Health and Human Services Page 50 GAO-03-809 Child Welfare Appendix III: Comments from the Department of Health and Human Services Page 51 GAO-03-809 Child Welfare Appendix III: Comments from the Department of Health and Human Services Page 52 GAO-03-809 Child Welfare Appendix III: Comments from the Department of Health and Human Services Page 53 GAO-03-809 Child Welfare Appendix III: Comments from the Department of Health and Human Services Page 54 GAO-03-809 Child Welfare Appendix III: Comments from the Department of Health and Human Services Page 55 GAO-03-809 Child Welfare Appendix III: Comments from the Department of Health and Human Services Page 56 GAO-03-809 Child Welfare Appendix III: Comments from the Department of Health and Human Services Page 57 GAO-03-809 Child Welfare Appendix III: Comments from the Department of Health and Human Services Page 58 GAO-03-809 Child Welfare Appendix III: Comments from the Department of Health and Human Services Page 59 GAO-03-809 Child Welfare Appendix III: Comments from the Department of Health and Human Services Page 60 GAO-03-809 Child Welfare Appendix III: Comments from the Department of Health and Human Services Page 61 GAO-03-809 Child Welfare Appendix IV: GAO Contacts and Appendix IV: GAO Contacts and Acknowledgments Acknowledgments Diana Pietrowiak, (202) 512-6239 GAO Contacts Sara L. Schibanoff, (202) 512-4176 In addition to those named above, Leah DeWolf and Rachel Seid made key Staff contributions to this report. Avrum Ashery, Patrick DiBattista, Acknowledgments Barbara Johnson, Valerie Melvin, and Rebecca Shea also provided key technical assistance. Page 62 GAO-03-809 Child Welfare Bibliography Bibliography The American Public Welfare Association. Statewide Automated Child Welfare Information Systems: Survey of State Progress. Washington, D.C., July 1997. The American Public Welfare Association. Child Welfare Information Systems: Some Concepts and Their Implications. Washington, D.C., July 1994. The American Public Welfare Association. Survey of State Child Welfare Information Systems: Status of AFCARS and SACWIS. Washington, D.C., April 1995. Caliber Associates, Analysis of State Child Welfare Data: VCIS Survey Data from 1990 through 1994, May 1998, Department of Health and Human Services. Center for Technology in Government, University of Albany, SUNY. Tying a Sensible Knot: A Practical Guide to State-Local Information Systems. Albany, N.Y., June 1997. Child Welfare League of America. National Working Group Highlights, “Child Maltreatment in Foster Care: Understanding the Data.” Washington, D.C., October 2002. Child Welfare League of America. National Working Group Highlights, “Placement Stability Measure and Diverse Out-of-Home Care Populations.” Washington, D.C., April 2002. U.S. Department of Health and Human Services, Administration for Children and Families, Administration on Children, Youth and Families, Children’s Bureau. Child Maltreatment 2001. Washington, D.C., 2003. U.S. Department of Health and Human Services, Administration for Children and Families, Administration on Children, Youth and Families, Children’s Bureau. Child Welfare Outcomes 1999: Annual Report. Washington, D.C., n.d.. U.S. Department of Health and Human Services, Office of Inspector General. Adoption and Foster Care Analysis and Reporting System (AFCARS): Challenges and Limitations. Washington, D.C., March 2003. Page 63 GAO-03-809 Child Welfare Related GAO Products Related GAO Products Child Welfare and Juvenile Justice: Federal Agencies Could Play a Stronger Role in Helping States Reduce the Number of Children Placed Solely to Obtain Mental Health Services. GAO-03-397. Washington, D.C.: April 21, 2003. Child Welfare: HHS Could Play a Greater Role in Helping Child Welfare Agencies Recruit and Retain Staff. GAO-03-357. Washington, D.C.: March 31, 2003. Human Services: Federal Approval and Funding Processes for States’ Information Systems. GAO-02-347T. Washington, D.C.: July 9, 2002. Foster Care: Recent Legislation Helps States Focus on Finding Permanent Homes for Children, but Long-Standing Barriers Remain. GAO-02-585. Washington, D.C.: June 28, 2002. Human Services Integration: Results of a GAO Cosponsored Conference on Modernizing Information Systems. GAO-02-121. Washington, D.C.: January 31, 2002. District of Columbia Child Welfare: Long-Term Challenges to Ensuring Children’s Well-Being. GAO-01-191. Washington, D.C.: December 29, 2000. Child Welfare: New Financing and Service Strategies Hold Promise, but Effects Unknown. GAO/T-HEHS-00-158. Washington, D.C.: July 20, 2000. Welfare Reform: Improving State Automated Systems Requires Coordinated Federal Effort. GAO/HEHS-00-48. Washington, D.C.: April 27, 2000. Foster Care: States’ Early Experiences Implementing the Adoption and Safe Families Act. GAO/HEHS-00-1. Washington, D.C.: December 22, 1999. Foster Care: HHS Could Better Facilitate the Interjurisdictional Adoption Process. GAO/HEHS-00-12. Washington, D.C.: November 19, 1999. Foster Care: Effectiveness of Independent Living Services Unknown. GAO/HEHS-00-13. Washington, D.C.: November 5, 1999. Foster Care: Kinship Care Quality and Permanency Issues. GAO/HEHS- 99-32. Washington, D.C.: May 6, 1999. Page 64 GAO-03-809 Child Welfare Related GAO Products Juvenile Courts: Reforms Aim to Better Serve Maltreated Children. GAO/HEHS-99-13. Washington, D.C.: January 11, 1999. Child Welfare: Early Experiences Implementing a Managed Care Approach. GAO/HEHS-99-8. Washington, D.C.: October 21, 1998. Foster Care: Agencies Face Challenges Securing Stable Homes for Children of Substance Abusers. GAO/ HEHS-98-182. Washington, D.C.: September 30, 1998. Managing Technology: Best Practices Can Improve Performance and Produce Results. GAO/T-AIMD-97-38, January 31, 1997. Child Welfare: HHS Begins to Assume Leadership to Implement National and State Systems. GAO/AIMD-94-37. Washington, D.C.: June 8, 1994. Executive Guide: Improving Mission Performance Through Strategic Information Management and Technology. GAO/AIMD-94-115. May 1, 1994. (130163) Page 65 GAO-03-809 Child Welfare The General Accounting Office, the audit, evaluation and investigative arm of GAO’s Mission Congress, exists to support Congress in meeting its constitutional responsibilities and to help improve the performance and accountability of the federal government for the American people. GAO examines the use of public funds; evaluates federal programs and policies; and provides analyses, recommendations, and other assistance to help Congress make informed oversight, policy, and funding decisions. GAO’s commitment to good government is reflected in its core values of accountability, integrity, and reliability. The fastest and easiest way to obtain copies of GAO documents at no cost is Obtaining Copies of through the Internet. GAO’s Web site (www.gao.gov) contains abstracts and full- GAO Reports and text files of current reports and testimony and an expanding archive of older products. The Web site features a search engine to help you locate documents Testimony using key words and phrases. You can print these documents in their entirety, including charts and other graphics. Each day, GAO issues a list of newly released reports, testimony, and correspondence. GAO posts this list, known as “Today’s Reports,” on its Web site daily. The list contains links to the full-text document files. To have GAO e-mail this list to you every afternoon, go to www.gao.gov and select “Subscribe to daily E-mail alert for newly released products” under the GAO Reports heading. Order by Mail or Phone The first copy of each printed report is free. Additional copies are $2 each. A check or money order should be made out to the Superintendent of Documents. GAO also accepts VISA and Mastercard. Orders for 100 or more copies mailed to a single address are discounted 25 percent. Orders should be sent to: U.S. General Accounting Office 441 G Street NW, Room LM Washington, D.C. 20548 To order by Phone: Voice: (202) 512-6000 TDD: (202) 512-2537 Fax: (202) 512-6061 Contact: To Report Fraud, Web site: www.gao.gov/fraudnet/fraudnet.htm Waste, and Abuse in E-mail: firstname.lastname@example.org Federal Programs Automated answering system: (800) 424-5454 or (202) 512-7470 Jeff Nelligan, Managing Director, NelliganJ@gao.gov (202) 512-4800 Public Affairs U.S. General Accounting Office, 441 G Street NW, Room 7149 Washington, D.C. 20548
Child Welfare: Most States Are Developing Statewide Information Systems, but the Reliability of Child Welfare Data Could Be Improved
Published by the Government Accountability Office on 2003-07-31.
Below is a raw (and likely hideous) rendition of the original report. (PDF)