There has been increasing emphasis on cost comparison and cost-effectiveness analysis in health care delivery systems. It is important that health care policies are based on reliable inferences about medical care costs and reliable predictions about future costs for medical care. However, reliable inferences and prediction may be impeded by the following four characteristics of medical cost data: (1) non-zero cost observations are highly skewed to the right; (2) a certain proportion of the population is expected to incur no medical care costs during the whole study period; (3) cost observations are hierarchically structured; and (4) cost observations typically exhibit heteroscedasticity. In this application, investigators will develop a general statistical methodology for regression analysis of health care cost data, adjusting for these four characteristics. Furthermore, contributions will be demonstrated in the area of health services research by applying the proposed methods to analyze of total health care costs in four existing datasets.
Song, Xiao; Zhou, Xiao-Hua (2011) Evaluating markers for treatment selection based on survival time. Stat Med 30:2251-64 |
Zhou, Xiao-Hua; Cheng, Hao (2008) A computer program for estimating the re-transformed mean in heteroscedastic two-part models. Comput Methods Programs Biomed 90:210-6 |
Harel, Ofer; Zhou, Xiao-Hua (2007) Multiple imputation: review of theory, implementation and software. Stat Med 26:3057-77 |
Song, Xiao; Zhou, Xiao-Hua (2005) A marginal model approach for analysis of multi-reader multi-test receiver operating characteristic (ROC) data. Biostatistics 6:303-12 |
Jiang, Hongyu; Zhou, Xiao-Hua (2004) Bootstrap confidence intervals for medical costs with censored observations. Stat Med 23:3365-76 |