Performance of Selected Nonparametric Tests for Discrete Longitudinal Data Under Different Patterns of Missing Data

Abstract

Comparison of changes over time of a continuous response variable between treatment groups is often of main interest in clinical trials. When the distributional properties of the continuous response variable are not regular enough, or when the response is discrete, nonparametric techniques have been used. The relative performances of selected repeated measures nonparametric two-sample tests proposed by Wei and Lachin, Koziol, Wei and Johnson, and the adapted Wilcoxon Rank-Sum test are compared through simulations based on quality of life data. The Wilcoxon Rank-Sum test is the most powerful and is not significantly affected by the different patterns of missing data.

Key Words:

• Informative
• MAR
• MCAR
• Nonparametric tests
• Simulations
• Wilcoxon

ACKNOWLEDGMENTS

We would like to thank Prof. Geert Molenberghs for his valuable comments on the work, which has led to this paper. We also like to acknowledge Bristol-Myers Squibb Pharmaceuticals, Belgium, for allowing the authors to use one of their oncology QOL study data as a guide in realistic simulation of incomplete, discrete longitudinal data.

Notes

Note: *Code 999 indicates a missing score and this is of interest.