Abstract:
|
Time-to-event data in the presence of competing risks has been well studied in recent years. One popular approach to this problem is to model the subdistribution of competing risks with a proportional hazards assumption, first proposed by Fine and Gray (1999). This approach is appealing because it utilizes the cumulative incidence function, which has a direct interpretation in clinical practice. However, the estimator resulting from this model does not perform as expected when the covariates are measured with error. We ran simulations to determine under which conditions the Fine & Gray estimator accrues bias in the presence of covariate measurement error. We plan to apply our results to Alzheimer's disease data, which examine the association of measurement-error prone cerebrospinal fluid (CBF) biomarkers and risk of conversion to Alzheimer's disease.
|