A0438
Title: A multiple imputation-based sensitivity analysis approach for regression analysis with a MNAR covariate
Authors: Chiu-Hsieh Hsu - University of Arizona (United States) [presenting]
Yulei He - CDC (United States)
Chengcheng Hu - University of Arizona (United States)
Wei Zhou - University of Arizona (United States)
Abstract: Missing covariate problems are common in biomedical studies. If there is a suspicion of missing not at random, researchers often perform sensitivity analysis to evaluate the impact of various missingness mechanisms. Under the selection modeling framework, we propose a sensitivity analysis approach with a standardized sensitivity parameter using a nonparametric multiple imputation strategy. The proposed approach requires fitting two working models for deriving two predictive scores and specifying the correlation coefficient between missing covariate values and selection probabilities. For each missing covariate observation, the two predictive scores are used to select the nearest neighborhood and the correlation coefficient is used to define an imputing set based on the selected neighborhood. The proposed approach is expected to be more robust against mis-specifications of the selection model and the sensitivity parameter since the selection model is only used to induce missing not at random and the sensitivity parameter is only used to define imputing sets. For illustration, the proposed approach is applied to evaluate the relationship between post-operative outcomes and incomplete pre-operative Hemoglobin A1c levels for surgical high-grade carotid artery stenosisa patients. A simulation study is conducted to evaluate the performance of the proposed approach.
