Statistical Methods Of The Medical Cost Estimates

Estimating medical costs is an important area of research in biostatistics and phar-macoeconomics. Here proper estimation of the mean cost for a group of patients with a particular disease from possibly censored follow-up data is of great importance. Naive estimators using summary statistics on the collected data and standard survival analysis estimators may be biased. Better estimators are needed to make the estimation more accurate and robust. This paper gives an overview of two approaches and develops a new method by proposing a simple stratification estimator along with Markov assumption. The estimator is shown to be consistent. Extensive simulation studies are used to compare the estimator with others, and show that our estimator performs well in finite samples and is robust under various situations. We also provide a method to choose the best proper estimator for a variety of circumstances. Further improvement of our estimator and extensions to regression models are also discussed.