On the use of survival analysis techniques to estimate medical care costs

Measurement of treatment costs is important in the evaluation of medical in terventions. Accurate cost estimation is problematic, when cost records are incomplete. Methods from the survival analysis literature have been propos ed for estimating costs using available data. In this article, we clarify a ssumptions necessary for validity of these techniques. We demonstrate how a ssumptions needed for valid survival analysis may be violated when these me thods are applied to cost estimation. Our observations are confirmed throug h simulations and empirical data analysis. We conclude that survival analys is approaches are not generally appropriate for the analysis of medical cos ts and review several valid alternatives. (C) 1999 Elsevier Science B.V. Al l rights reserved. JEL classification: C000 Mathematical and quantitative m ethods: general; C100 Econometric and statistical methods: general; C400 Ec onometric and statistical methods: special topics: general.