Odds Ratios for a Continuous Outcome Variable Without Dichotomizing

• Activity Number: 124
• Type: Topic Contributed
• Date/Time: Monday, August 8, 2005 : 10:30 AM to 12:20 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Barry K. Moser*+ and Laura Coombs
• Companies: Duke University and George Washington University
• Keywords: dichotomizing ; logistic regression ; odds ratios ; linear regression ; confidence intervals

Abstract

The loss of information from dichotomizing a continuous outcome is well documented in the literature. One advantage of dichotomizing is it allows estimation of odds ratio parameters through a logistic regression analysis. The objective of this paper is to develop a new estimator of the same odds ratio parameters through regression analysis on the original continuous outcome without the inherent loss of information caused by dichotomizing. Through a mathematical, asymptotic development, the sample sizes required to attain a specified power when testing the odds ratio parameter are compared for the dichotomizing procedure and the proposed approach. The comparison highlights the substantial sample-size savings attained by the proposed approach, particularly for large values of the odds ratio parameter and small proportions of dichotomized successes or failures. In a Monte Carlo simulation, the variances and absolute biases of the two odds ratio estimators and the length of their respective confidence intervals again demonstrate the improvement attained by the proposed approach.