Analysis of stage II pancreatic cancer data with missing covariates and time-varying covariate effects

초록

This paper aims to identify significant predictors of death from pancreatic cancer using stage II pancreatic cancer data obtained from the Surveillance, Epidemiology, and End Results (SEER) program of the National Cancer Institute (NCI). Some patients have missing covariate values in the stage II pancreatic cancer data. Excluding patients with missing covariates from the analysis and applying standard statistical methods may lead to efficiency loss and biased results. We address this issue using multiple imputation methods and then fit a cause-specific proportional hazards model for death from pancreatic cancer to examine the effects of covariates on the cause-specific hazard of death from pancreatic cancer. One covariate in the model did not satisfy the proportional hazards assumption. To address this, we apply multiple imputation methods that allow for time-varying effects and then fit a cause-specific proportional hazards model with time-varying effects to the multiply imputed datasets. The results showed that age at diagnosis, marital status, grade, tumor size, surgery, radiotherapy, and chemotherapy were significant predictors of death from pancreatic cancer in stage II pancreatic cancer.

키워드