Application of Markov Chain Monte-Carlo Multiple Imputation Method to Deal with Missing Data from the Mechanism of MNAR in Sensitivity Analysis for a Longitudinal Clinical Trial

Part of the ICSA Book Series in Statistics book series (ICSABSS)


Missing data in clinical trials could potentially arise from the mechanism of Missing Not At Random (MNAR) . In order to understand the impact on a longitudinal clinical trial findings from missing data under the MNAR assumption, the sensitivity analyses could be carried out by multiple imputations. By progressively decreasing the treatment differences in those treated subjects who fell into an assumed MNAR pattern, the departure from Missing At Random (MAR) assumption could be investigated. This chapter aims to apply Markov Chain Monte-Carlo (MCMC) Multiple Imputation method to investigate that, under an MNAR assumption, that the missing data pattern of subjects who receive clinical trial treatment are similar to, worse than, or better than those of subjects who receive placebo with similar observed outcomes for two scenarios of early discontinuation: (1) discontinuation due to lack of efficacy and non-disease progression related Adverse Events (AEs); (2) discontinuation due to any reason. We also demonstrate how to apply MCMC multiple imputation method without assuming the data to have normal distribution.


Missing data Multiple imputation Markov Chain Monte-Carlo Longitudinal clinical trial Missing not at random Early discontinuation Sensitivity analysis ANCOVA Wilcoxon rank-sum test 


  1. Barnard, J., & Rubin, D. B. (1999). Small-sample degrees of freedom with multiple imputation. Biometrika, 86, 948–955.MathSciNetCrossRefzbMATHGoogle Scholar
  2. Galbraith, S. (2012). Applied missing data analysis by Craig K Enders. Australian & New Zealand Journal of Statistics, 54, 251. doi: 10.1111/j.1467-842X.2012.00656.x.CrossRefGoogle Scholar
  3. Mogg, R., & Mehrotra, D. V. (2007). Analysis of antiretroviral immunotherapy trials with potentially non-normal and incomplete longitudinal data. Statistics in Medicine, 26(3), 484–497.Google Scholar
  4. National Research Council. (2010). The prevention and treatment of missing data in clinical trials. Washington, DC: National Academies Press.
  5. Rubin, D. B. (1987). Multiple imputation for nonresponse in surveys. New York: Wiley.CrossRefzbMATHGoogle Scholar
  6. Schafer, J. L. (1997). Analysis of incomplete multivariate data. New York: Chapman and Hall.CrossRefzbMATHGoogle Scholar
  7. Verbeke, G., & Molenberghs, G. (2000). Linear mixed models for longitudinal data. New York: Springer.zbMATHGoogle Scholar
  8. Wang, N., & Robins, J. M. (1998). Large sample theory for parametric multiple imputation procedures. Biometrika, 85, 935–948.MathSciNetCrossRefzbMATHGoogle Scholar

Copyright information

© Springer Nature Singapore Pte Ltd. 2017

Authors and Affiliations

  1. 1.Manager Biostatistician at Otsuka AmericaNewyorkUSA

Personalised recommendations