Objective: To discuss the estimation on gene-environment interaction in partial case-control studies when gene information of the controls was partly missing.
Methods: The results of hot deck multiple imputation and listwise deletion analysis were compared when missing data was generated using Monte Carlo method in Stata 9.0.
Results: Coefficients of environment effect, gene effect and gene-environment interaction were respectively estimated by means of hot deck multiple imputation and listwise deletion when approaching to those complete data with missing part less than 50 percent. Both estimated variances of the two methods were increasing with the increased proportion of missing data, but the estimated variance of hot deck multiple imputation was smaller than the one with listwise deletion in each proportion.
Conclusion: Hot deck imputation could be adopted to make full use of existing information to estimate gene-environment interaction in the partial case-control study when missing proportion of gene data of controls was less than 50 percent so as to increase the precision of the estimation.