The 131st Annual Meeting (November 15-19, 2003) of APHA

Abstract #62906

Sunny Kim, PhD, Health Psychology, Ohio University, 236 Porter Hall, Athens, OH 45701, (740)597-1963, kims2@ohio.edu and Mark Boye, PhD, Health Economics, Pharmacia Corporation, Mailstop4-420, 100 Route 206 North, Peapack, NJ 07977.

Objectives: For the comparison of cost-effectiveness among treatment groups, researchers routinely form the ratio of the cost to the effect (C/E) of treatment. However, we often face some limitations in applying C/E ratios in practice including conceptual and statistical difficulties of ratio variables as well as issues concerning the control of confounding biases. Treatment and comparison groups, in observational study, are rarely comparable. Propensity score methods (PSMs) have been increasingly used in medical treatment and program evaluation settings to overcome confounding bias. However, to our knowledge, no published study has applied the PSM to the analysis of C/E ratios. The objectives of this study, therefore, are first to illustrate the difficulties of using C/E ratio in cost-effectiveness studies and second to report on the C/E ratio test, the PSM, and a statistical model. Methods: Problems resulting from the use of C/E ratios were illustrated through the use of data. Thereafter and with the same data we compared results derived from the C/E ratio test, the PSM, and a statistical model. Results: Statistical model revealed that comparison of the C/E ratio is confounded by patients’ characteristics. Although average treatment-control group differences estimated through the use of PSM can serve to reduce confounding, issues of data dependency may limit generalizability. Conclusions: Application of statistical modeling can effectively control for confounding bias; estimates of treatment effects can be compared for a given amount of cost by incorporating a cost variable into the model.

Learning Objectives:

At the conclusion of this session, the participant in this session will be able to:
• Recognize the problems of using cost-effect ratio.
• Understand the propensity score approach and statistical modeling in a way to overcome confounding effect.

Keywords: Decision-Making, Statistics

Presenting author's disclosure statement:
I do not have any significant financial interest/arrangement or affiliation with any organization/institution whose products or services are being discussed in this session.