Translating behavioral health services research into benefits policy.

Abstract

This article uses a 4-pronged statistical approach to examine the impact of a mental health carve-out at a major employer. To examine net financial impact of the carve-out, the authors perform a pre-post, multivariate regression analysis of changes in costs. Using a random-effects model, the authors explore the ultimate financial impact of the carve-out for patients and for the firm. Using a multinomial logistic regression, they examine differing program effects by intensity of use. A fixed-effects negative binomial regression models the episodic nature of outpatient care, controlling for patient-specific unobserved characteristics that influence health care utilization. The carve-out slightly reduced overall mental health costs and utilization while expanding entry-level access to routine services. At the same time, the specific carve-out shifted financial burdens from the firm onto high-utilization patients. Therefore, this carve-out appears poorly suited to the care of individuals experiencing severe and debilitating psychiatric disorders.