Child abuse and neglect re-reports: Combining and comparing data from two national sources

This study compared child-level estimates of child maltreatment re-report and recurrence in two national sets of data on child maltreatment: state administrative data submitted to the National Child Abuse and Neglect Data System (NCANDS) and caseworker interviews from the National Survey of Child and Adolescent Well-Being (NSCAW). Maltreatment data from NCANDS and NSCAW were merged for 2230 children that had intersecting information from both sets. The percentage of child cases with at least one re-report of abuse or neglect over the study period differed depending on the data set. The NCANDS re-report estimate was 32.3% (95% CI 26.5%, 38.6%) whereas the estimate based on NSCAW caseworker interviews was 22.9% (95% CI 17.6%, 29.2%). More than a quarter of the children with observations from the union of the two data sets had a re-report identified by one source but not the other (set difference). Most often, the set difference in re-reports appeared in NCANDS, but was not reported by an NSCAW caseworker. When the set difference from NCANDS was added to the re-reports by NSCAW caseworkers, the resulting union of re-reports increased the point estimate in the NSCAW-NCANDS intersection to 40.9% (95% CI 34.3%, 47.8%). Restricting the comparisons to only substantiated re-reports (recurrence) narrowed the differences in absolute terms but the set difference in recurrence rates was proportionally similar. Potential explanations for non-intersecting re-reports and recurrence between the two data sets were examined. FINDINGS illuminate methodological challenges that may arise when child maltreatment re-report and recurrence data from administrative and survey sources are merged, and the value of the union of state-level administrative data with national survey data for studies of safety and well-being of children reported for maltreatment.