Missing Confounding Data in Marginal Structural Models: A Comparison of Inverse Probability Weighting and Multiple Imputation

• Published in: The International Journal of Biostatistics Vol. 4; no. 1; pp. 13 - 37
• Main Authors: Moodie, Erica E. M, Delaney, Joseph A.C, Lefebvre, Geneviève, Platt, Robert W
• Format: Journal Article
• Language: English
• Published: Germany bepress 2008; De Gruyter
• Subjects: Adrenergic beta-Antagonists - therapeutic use; Bias; Biostatistics - methods; causal inference; confounding; Data Interpretation, Statistical; Databases, Factual; double robustness; Humans; inverse probability weighting; marginal structural models; missing data; Models, Statistical; Monte Carlo Method; multiple imputation; Myocardial Infarction - drug therapy; Myocardial Infarction - mortality; Probability; simulations

Abstract Standard statistical analyses of observational data often exclude valuable information from individuals with incomplete measurements. This may lead to biased estimates of the treatment effect and loss of precision. The issue of missing data for inverse probability of treatment weighted esti...