The Practical Use of Different Strategies to Handle Dropout in Longitudinal Studies
- PDF / 1,010,492 Bytes
- 16 Pages / 504 x 719.759 pts Page_size
- 82 Downloads / 134 Views
0092-86 I 51200 I
Copyright 0 1-001 Drug Information Association Inc.
THE PRACTICAL USE OF DIFFERENT STRATEGIES TO HANDLE DROPOUT IN LONGITUDINAL STUDIES GEERTVERBEKE, EMMANUEL LESAFFRE, AND BARTSPIESSENS Biostatistical Centre, Catholic University of Leuven, Leuven, Belgium
In the presence of dropout, valid statistical inferences based on longitudinal data can, in general, only be obtained from modeling the measurement process and the dropout process simultaneously. Many models have been proposed in the statistical literature, most of which have been formulated within the framework of selection models or patternmixture models. In this paper; we will use continuous data from a longitudinal clinical trial with a 24% dropout rate to illustrate some ofthe models frequeritly used in practice. We emphasize the underlying implicit assumptions made by the different approaches, and the sensitivity of the results with respect to these assumptions. The merits and drawbacks of the procedures are extensively discussed and compared from a practical point of view. Key Words: Linear mixed model; Selection model; Pattern-mixture model; Sensitivity
analysis
INTRODUCTION IN MEDICAL RESEARCH, studies are often designed in which specific parameters are measured repeatedly over time in the participating subjects. This allows the process of change to be modeled within each subject separately, based on both subject-specific factors (such as gender) and experiment-specific factors (such as treatment). The analysis of such longitudinal data requires statistical models that take into account the association between the measurements within subjects. During the last decade, a lot of effort has been put into the search for flexible longitudinal models. An overview can be found in Diggle, Liang, and Zeger (1). In practice, longitudinal studies often suffer from attrition, that is, subjects dropping out earlier than scheduled, for reasons outside the control of the investigator. The resulting data are unbalanced with unequal numbers of measures for each participant. Nowadays, several statistical packages (eg, SAS procedure MIXED [ 2 ] ,LME function in S-plus [3], or BMDP procedure 5V [4]) are available for the analysis of unbalanced longitudinal data. However, they only yield valid inferences under specific assumptions for the dropout process. In general, valid inferences can only be obtained by modeling the response measurements and the dropout process simultaneously. Making various assumptions about the dropout Presented at the 9th Annual European DIA Workshop “Statistical Methodology in Clinical Research & Development,” April 20-22, 1998, Amsterdam, The Netherlands. Reprint address: Geert Verbeke, Biostatistical Centre, Catholic University of Leuven, U.Z. St.-Rafael, Kapucijnenvoer 35, B-3000 Leuven, Belgium.
419
Downloaded from dij.sagepub.com at NANYANG TECH UNIV LIBRARY on April 28, 2015
420
Geert Verbeke, Emmanuel Lesaffre, and Bart Spiessens
mechanism, a large variety of models for continuous as well as categorical outcomes hav
Data Loading...
