Multiple Imputation for Non-Random Missing Data in Longitudinal Studies of Health-Related Quality of Life
Missing health-related quality of life (QOL) data in longitudinal studies rarely occur for a single reason. Some individuals drop out of the study as a result of mortality or morbidity of the disease or its treatment, others because their disease symptoms have disappeared. Individuals may have intermittent missing assessments for reasons both unrelated and related to treatment. Most methods for analysis focus on a single mechanism, such as informative dropout, and do not address multiple causes and patterns of missing data. These methods are generally based on strong assumptions that cannot be tested because of the missing data. Further, the estimates are often sensitive to the assumptions underlying the methods. The concepts of multiple imputation can be extended to longitudinal studies of QOL allowing 1) the incorporation of a mixture of models reflecting the variety of causes of missing data that occurs in these studies and 2) the uncertainty associated with the underlying assumptions of the models. However, the assumptions underlying the methods for imputing the missing data should be carefully examined in settings where the missing assessments are nonignorable.
KeywordsPropensity Score Multiple Imputation Imputation Model Missing Observation Miss Data Mechanism
Unable to display preview. Download preview PDF.
- 7.Troxel, A.B., Harrington, D.R and Lipsitz, S.R. (1998). Analysis of Longitudinal Data with Nonignorable non-monotone missing values. Applied Statistics 47, 425–438.Google Scholar
- 8.SAS Institute (1996). The MIXED procedure. SAS/STAT Software. Changes and Enhancements (through Release 6.11). Cary, NC: SAS Institute.Google Scholar
- 9.Schluchter, M.D. (1990). 5V: unbalanced repeated measures models with structured covariance matrices. In: BMDP Statistical Software Manual Part 2. Berkley, CA: University of California Press.Google Scholar
- 10.Bonomi, P., Kim, K.M., Fairclough, D., Cella, D., Kugler, J., Rowinsky, E., Jiroutek, M. and Johnson, D. (2000). Comparison of survival and quality of life in advanced non-small cell lung cancer patients treated with two dose levels of paclitaxel combined with cisplatin versus etoposide-cisplatin: results of an Eastern Cooperative Oncology Group trial. Journal of Clinical Oncology 18, 623–631.PubMedGoogle Scholar
- 12.Cella, D.F., Tulsky, D.S., Gray, G., Sarafian, B., Linn, E.., Bonomi, A., Silberman, M., Yellen, S.B., Winicour, P. and Brannon J. (1993). The Functional Assessment of Cancer Therapy (FACT) scales: development and validation of the general measure. Journal of Clinical Oncology 11, 570–579.PubMedGoogle Scholar
- 21.Little, R.J. and Rubin, D.B. (1987). Statistical Analysis with Missing Data. New York: John Wiley and Sons.Google Scholar
- 22.Statistical Solutions (1997). Solas for Missing Data Analysis 1.0 User Reference.Google Scholar
- 23.Yuan, Y.C. (2000). Multiple Imputation for Missing Data: Concepts and New Development. Rockville, MD: SAS Institue Inc. Cork, Ireland: Statistical Solutions Ltd.Google Scholar
- 26.Neter, J. and Wasserman, W. (1974). Applied Linear Statistical Models. Home-wood, IL: Richard Irwin, Inc.Google Scholar