HERC Health Economics Seminar
Choosing Models for Cost Analyses: Issues of Nonlinearity and Endogeneity
Melissa Garrido, Ph.D.
Seminar date: 7/20/2011
Description: Following the American Recovery and Reinvestment Act’s support for comparative effectiveness research, concerns regarding conclusions drawn from healthcare cost analyses have grown. Skewness, non-negative outcomes, and censoring are commonly encountered in cost analyses. A growing literature focuses on the development and evaluation of statistical models designed to yield consistent, unbiased and reliable estimates in the presence of these issues. However, these developments rarely address endogeneity of key regressors, another common characteristic of cost analyses. Using VA cost and utilization data, we compare several methods for estimating costs with endogeneity (propensity scores, two-stage least squares, control function [CF], and full information maximum simulated likelihood [FIMSL]) and illustrate the impact of model specification on estimates obtained. Additionally, we address important features of cost analyses for treatments that are often overlooked, including marginal effect (ME) distributions and specification of the functional form of residuals included in control function (CF) models. To illustrate differences among models, we examine the effect of an inpatient palliative care consultation (PCCT) on direct costs of care per day for 3,321 inpatients hospitalized in five VA acute care facilities in 2004-2007 with one or more life-limiting diseases (advanced cancer, advanced HIV/AIDS, or severe congestive heart failure or chronic obstructive pulmonary disease).
We're sorry - this video format is no longer supported. You may still download the resources below.
Request PDF Handout |
Audio only (mp3) |
transcript not yet available