Child outpatient mental health service use: why doesn't insurance matter?

BACKGROUND: Several recent studies of child outpatient mental health service use in the US have shown that having private insurance has no effect on the propensity to use services. Some studies also find that public coverage has no beneficial effect relative to no insurance. AIMS: This study explores several potential explanations, including inadequate measurement of mental health status, bandwagon effects, unobservable heterogeneity and public sector substitution for private services, for the lack of an effect of private insurance on service use. METHODS: We use secondary analysis of data from the three mainland US sites of NIMH's 1992 field trial of the Cooperative Agreement for Methodological Research for Multi-Site Surveys of Mental Disorders in Child and Adolescent Populations (MECA) Study. We examine whether or not a subject used any mental health service, school-based mental health services or outpatient mental health services, and the number of outpatient visits among users. We also examine use of general medical services as a check on our results. We conduct regression analysis; instrumental variables analysis, using instruments based on employment and parental history of mental health problems to identify insurance choice, and bivariate probit analysis to examine multiservice use. RESULTS: We find evidence that children with private health insurance have fewer observable (measured) mental health problems. They also appear to have a lower unobservable (latent) propensity to use mental health services than do children without coverage and those with Medicaid coverage. Unobserved differences in mental health status that relate to insurance choice are found to contribute to the absence of a positive effect for private insurance relative to no coverage in service use regressions. We find no evidence to suggest that differences in attitudes or differences in service availability in children's census tracts of residence explain the non-effect of insurance. Finally, we find that the lack of a difference is not a consequence of substitution of school-based for office-based services. School-based and office-based specialty mental health services are complements rather than substitutes. School-based services are used by the same children who use office-based services, even after controlling for mental health status. DISCUSSION: Our results are consistent with at least two explanations. First, limits on coverage under private insurance may discourage families who anticipate a need for child mental health services from purchasing such insurance. Second, publicly funded services may be readily available substitutes for private services, so that lack of insurance is not a barrier to adequate care. Despite the richness of data in the MECA dataset, cross-sectional data based on epidemiological surveys do not appear to be sufficient to fully understand the surprising result that insurance does not enable access to care. IMPLICATIONS FOR POLICY AND RESEARCH: Limits on coverage under private mental health insurance combined with a relatively extensive system of public mental health coverage have apparently generated a situation where there is no observed advantage to the marginal family of obtaining private mental health insurance coverage. Further research using longitudinal data is needed to better understand the nature of selection in the child mental health insurance market. Further research using better measures of the nature of treatment provided in different settings is needed to better understand how the private and public mental health systems operate.