Utilizing Administrative Data Linkages to Estimate Child Maltreatment in Anchorage, Alaska

BACKGROUND:  Multiple studies have documented the difficulty in estimating child maltreatment (CM) rates due to differences in operational definitions.  Most CM estimates are derived from reports investigated or substantiated by state child protective services (CPS) agencies, which have jurisdiction only over CM by a primary caretaker and are subject to variations based on institutional capacity. Because CPS data are known to underestimate the actual burden of CM experienced, cross-jurisdictional data collection methods are needed. This study demonstrates the utility of linking CPS data with additional administrative data sources in an effort to more accurately estimate the burden of CM experienced within the city of Anchorage.

METHODS:  We linked 2008-14 administrative data from CPS with child advocacy center records and Anchorage police reports. Data were limited to reports of victimization made in Anchorage against children <10. Deduplication and probabilistic linkages were conducted using three identifiers: date of birth, first name, and last name. The contributions of each data source were quantified using a specific definition of CM, which included only substantiated CPS reports; and a sensitive definition, which included all CPS reports. Rates were calculated using population estimates from the Alaska Department of Labor.

RESULTS:  Using the specific and sensitive definitions of CM, we respectively identified 8,265 or 18,620 individual children with CM reports during 2008-14. By the specific and sensitive definitions, CPS records respectively contained 3,723 (45.0%) or 17,199 (92.4%) of the children identified. The child advocacy center contributed 2,900 (15.6% or 35.1%), and the police department 3,598 (19.3% or 43.5%). No children who were not identified in CPS data were reported to both the child advocacy center and the police department. The average number of individual children identified each year was 1,373 or 4,151; which corresponds to an average annual range of 13.0-39.4 victims of CM per 1,000 children in Anchorage during 2008-14.

CONCLUSIONS: Reports to CPS capture the majority of identified CM cases. However, these data can be difficult to interpret because of differences in case identification, availability of services, and database quality. For jurisdictions interested in improving their estimates or without access to complete CPS data, there is utility in linking alternate data sources. Using cross-jurisdictional data also encourages communication across disciplines and makes estimates less susceptible to changes in resource availability or reporting methodology.