Statistical Analysis of Case-Cohort Designs in Semi-Competing Risks

Abstract

Semi-competing risks refers to the setting where interest lies in some non-terminal event, the occurrence of which is subject to some terminal event (typically, but not always, death). While existing analysis methods generally assume complete data on all relevant covariates, it is often the case, particularly with electronic health records databases and disease registries, that some information is not readily-available. To mitigate this, outcome-dependent sampling is a commonly used design tool to collect otherwise unavailable information on a subset of participants rather than all participants. This is particularly useful in research settings where one or more covariates of interest may not be readily available, whether cost-prohibitive, time-consuming, or difficult to obtain in a resource-limited setting. Currently, researchers have only limited options for outcome-dependent sampling in the semi-competing risks setting.

In this dissertation, we present a novel class of case-cohort designs for semi-competing risks within which researchers have the flexibility to tailor allocation of resources in a variety of ways that best suit the disease context and study goals at-hand. For estimation and inference, we propose to use inverse-probability weighting for a hazard regression-based frailty illness-death model. We present asymptotic results along with a practical estimator of the asymptotic variance. Simulation results verify performance of the proposed analysis methods in finite settings and illustrate potential efficiency gains associated with the design. Additionally, we develop a framework to evaluate design options via simulations for both the case-cohort and nested case-control designs and compare various design choices. We compare the bias and efficiency of design choices through rigorous simulations and discuss practical design considerations. Finally, we illustrate to a general audience the flexibility of these two designs to tailor resource allocation for cohort-based studies with resource limitations and provide guidance for study design with a simulation-based algorithm for power and sample size analysis. The two designs are motivated by and illustrated with real-world data from the Center for International Blood \& Marrow Transplant Research, where we demonstrate the utility of the two classes of outcome-dependent sampling designs.