September 2005Vol. 6, No. 7Sampling Bias in Using Medicaid Data to Identify Children in Foster Care

A recent study suggests that Medicaid data alone may not be reliable for determining the true needs and costs of providing health care services to children in foster care. Researchers found a sampling bias when Medicaid data were used to identify children in foster care and to estimate the children's use of health care services. The sampling bias resulted in a significant overrepresentation of those children in foster care who used more services and a modest underrepresentation of the total Medicaid costs of the group. Findings came from a study that explored the reliability of Medicaid data by examining foster care administrative files and Medicaid data for 5,683 children in their first year of foster care.

According to the study's authors, the misrepresentation stems from cases in which children were incorrectly coded as not being in foster care when their information was entered into the Medicaid system. Analysis showed that nearly one-third of children were never coded as being in foster care in their Medicaid eligibility files during the first year after placement; in addition, these incorrectly classified children disproportionately accounted for those who never used any health services after entering foster care. In contrast, children who were correctly coded as being in foster care were more likely to have used health care services and to have reimbursed claims. This correctly classified group tended to be older, non-Hispanic White, with more placements, and more time in foster care, or with residence in a group home or residential facility. In fact, researchers suggest that the increased contact with the child welfare system or with health care providers may have brought these children to the attention of administrators who were able to correctly switch their coverage to the Medicaid foster care program.

Based on these findings, the authors suggest that linking child welfare records to regional Medicaid data may provide a truer picture than Medicaid data alone of the needs and costs of health care services for children in foster care.

The study, "A Sampling Bias in Identifying Children in Foster Care Using Medicaid Data," by D. M. Rubin, S. Pati, X. Luan, and E. A. Alessandrini, was published in Ambulatory Pediatrics, Volume 5. The article can be purchased online at www.ambulatorypediatrics.org/article/S1530-1567(05)60199-2/abstract