Estimating need for mental health services across the U.S

In the absence of small-area needs assessments, estimating the need for mental health services in small geographic areas of the U.S. is challenging because (1) no national health survey is fully comprehensive in its coverage of mental disorders, (2) national surveys are not designed to generate small area estimates, and (3) no national utilization survey contains adequate samples of both individuals with and without SMI. To address these limitations, estimates from different surveys were combined. First, the National Comorbidity Survey Replication (NCSR) was used to estimate a logistic model (n=9,257) that predicts SMI status from demographic characteristics. This was used to estimate the prevalence of SMI by county (n=3,141 counties) yielding a mean prevalence of 4.6%. Utilization estimates from individuals with SMI in the NCSR (n=350 with SMI) and from those without SMI in the Medical Expenditure Panel Survey (n=16,728) were imputed to the populations with and without SMI by county, resulting in small-area estimates of need for different types of mental health providers. Both of these surveys provide detailed information on use of multiple types of mental health providers. Utilization was recorded in terms of visit minutes which were translated into provider FTEs. Findings indicate that about 40% of individuals with SMI saw a mental health provider in the past year compared to 6.5% of those without SMI. Service users spent about 2 hours in the past year with a prescriber and 8 hours with a nonprescriber provider. Strengths, limitations, and implications for mental health service planning are discussed.