Measuring Mental Health Service Disparities in Child Welfare

Background:  A number of studies conclude that youth of color placed in foster care are less likely than their Caucasian counterparts to utilize mental health services. However, few studies offer sound justification for the methods they used to calculate disparities in mental health service receipt, despite the fact that the magnitude and direction of disparity may vary, depending on the measurement of disparity chosen. This study offers a comparative and critical analysis of different approaches to the measurement of mental health service disparities among a nationally representative sample of racially/ethnically diverse youth referred to the child welfare system (CWS).

Methods: This study uses the National Survey of Child and Adolescent Well-being, a nationally representative, longitudinal sample of children who had contact with the CWS between October 1999 and December 2000 (Cohort 1) and between February 2008 and April 2009 (Cohort II), respectively. The current investigation focuses on 1,732 older children between the ages of 12 and 17 (1,005 NSCAW-I and 727 NSCAW-II children). Analyses focus on data collected at Wave 3 for NSCAW-I and Wave 2 for NSCAW-II. By then, all children in both cohorts were reported to the CWS within the past 18 months, allotting sufficient time for children to access services after the initial report of abuse at baseline if warranted. We analyze trends in four summary measures of relative disparity (Rate Ratio, Index of Disparity, Theil index, and Mean Log Deviation) and two absolute disparity measures (Rate Difference and Between-Group Variance).  In addition, a subgroup analysis was conducted to determine whether mental health service disparities are more salient among children in need as detected by the Child Behavior Checklist.

Results: All measures of disparity indicated that racial/ethnic disparities in mental health service use increased between Cohort I and Cohort II. However, the magnitude of the increase varied widely across disparity measures. Among the measures of relative disparity, for example, the Rate Ratio increased by 4.1 percent, while the Index of Disparity, Theil index, and Mean Log Deviation showed a 55.5, 63.1, and 56.5 percent increase, respectively. The magnitude of the increase also differed across absolute disparity measures, ranging from a 23.5 percent increase in the Rate Difference to a 44.5 percent increase in Between-Group Variance.  In addition, the estimated increases in disparity were even greater when the sample was limited to children in need. For instance, Theil index and Mean Log Deviation increased by 252.9 and 260.1 percent.          

Conclusions: This study showed that the same data may produce different magnitudes of disparity, depending on which metric is implemented and whether mental health need is accounted for. Thus, a greater understanding of and justification for selection of methods to examine mental health disparities among child welfare researchers and policy makers is warranted. Which metric is used may influence whether and how interventions are targeted to reduce disparities for this vulnerable population.