Assessing medical costs from a longitudinal model
by
Joseph C. Gardiner
Michigan State University
Coauthors: Corina M. Sirbu

In an era of constrained health care budgets, economic evaluations of health care interventions are increasingly important as policymakers seek to prioritize health care expenditures. An assessment of the elements of resource use, patient and intervention characteristics that influence costs and health benefits is important in informing resource allocation decisions in health care. We use a dynamic regression model in which costs are incurred in random amounts during sojourn in health states and at transition times between states. A Markov model describes the unfolding over time of patients' event histories, with transition intensities depending upon patient specific demographic and clinical characteristics through a multiplicative intensity model. A random effects model is used for transition and sojourn costs. We then estimate the net present value of expenditures incurred over a finite time horizon, quality-adjusted survival time and net health cost. While incorporating explanatory variables, the joint model can accommodate heteroscedasticity, skewness and censoring in cost and health outcome data and provides a flexible approach to analyses of health care costs and outcomes.

Date received: August 25, 2003