Longitudinal Modeling of Health Care Expenditures

University Of Wisconsin-Madison, Madison WI

Investigators

Abstract

[unreadable] DESCRIPTION (PROVIDED BY APPLICANT): The long-term aims of this project are to provide tools to policymakers and healthcare professionals to help assess the impact of system changes, such as the success of disease management programs and cost-effectiveness analyses. These tools will help increase the understanding of the magnitude and timing of health care expenditures. In an atmosphere of searching for ways to constrain resources spent on healthcare, a decision of whether a healthcare program is successful requires an estimate of costs. Usually the evaluation of these programs involves costs from multiple years and could involve costs incurred on some people across years. The proposed work will emphasize the analysis of longitudinal data, with the goal of better aligning analytic models with empirical data patterns. Specific areas of focus will be in the areas of chronic diseases and on costs impacting the elderly. [unreadable] [unreadable] This research will introduce statistical models and methods to handle data that (1) are likely to be large (long-tailed data) or (2) may be a combination of zeros and positive values (two-part data). Regardless of these tendencies to be non-normal, longitudinal data still appear as repeated observations over time and hence tend to be clustered. This project will introduce copula-based methods for handling this clustering. A copula is a tool for understanding relationships among multivariate outcomes; it is a function that links univariate marginal distributions to multivariate distributions. This proposed research will test whether these tools can provide a better fit to data, and will be more flexible and more computationally convenient when compared to traditional longitudinal methods that use latent variables for modeling dependence. [unreadable] [unreadable] We will demonstrate the applicability of these copula-based models using Medical Expenditure Panel Survey data and will examine expenditures related to those with chronic diseases. In addition, we will also apply these models to data related to Wisconsin nursing homes. [unreadable] [unreadable] [unreadable]