It is important that mental health care policies are based on reliable inferences about medical care costs and reliable predictions about future costs for medical care. However, three characteristics of medical cost data can impede reliable inferences and prediction: (1) highly skewed non-zero cost observations, (2) a certain proportion of zero costs, (3) and incomplete follow-up data Without taking these three characteristics into account, statistical analyses of cost data can lead to unreliable inferences and predictions. In this proposal, we will develop a general methodology for comparisons of means of two or more groups and for regression analysis of cost data, adjusting for these three characteristics.
| Zhou, Xiao-Hua (2002) Inferences about population means of health care costs. Stat Methods Med Res 11:327-39 |
| Zhou, X H; Li, C; Gao, S et al. (2001) Methods for testing equality of means of health care costs in a paired design study. Stat Med 20:1703-20 |
| Zhou, X H; Tu, W (2000) Confidence intervals for the mean of diagnostic test charge data containing zeros. Biometrics 56:1118-25 |
| Tu, W; Zhou, X H (1999) A Wald test comparing medical costs based on log-normal distributions with zero valued costs. Stat Med 18:2749-61 |
