A Bayesian Two-Part Latent Class Model for Longitudinal Medical Expenditure Data

• Activity Number: 268
• Type: Invited
• Date/Time: Tuesday, July 31, 2012 : 8:30 AM to 10:20 AM
• Sponsor: ENAR
• Author(s): Brian Neelon*+ and James O'Malley and Sharon-Lise T. Normand
• Companies: Duke University Medical Center and Harvard Medical School and Harvard Medical School
• Keywords: Bayesian analysis ; growth mixture model ; latent class model ; mental health parity ; semi-continuous data ; two-part model

Abstract

In 2001, the U.S. Office of Personnel Management required health plans participating in the Federal Employees Health Benefits Program to offer mental health benefits on par with general medical benefits. The initial evaluation found that, on average, parity did not increase spending or service use over time. However, some groups may have benefited from parity more than others. To address this question, we propose a Bayesian two-part latent class model to characterize the effect of parity on mental health use and expenditures. Within each class, we fit a two-part random effects model to separately model the probability of using mental health services and the mean spending trajectories among users. The regression coefficients and random effect covariances are allowed to vary across classes, thus permitting class-specific correlation structures between the two model components. Our analysis identified three classes: a group of low spenders who rarely used services; a group of moderate spenders who had an increase in both use and mean spending after the introduction of parity; and a group of high spenders with chronic service use and constant spending patterns.