The CORECT Project stands for Competing Risks and Endpoints in Clinical Trials.

Yes, it is ironic that CORECT is spelled incorrectly.

The goal of this project is to develop better methods to risk-stratify populations with competing risks. We have shown that standard models (for example, Cox regression models of survival and disease-free survival) are suboptimal for this purpose in competing risks settings. This is because they do not discriminate effects of covariates on competing events, in effect averaging heterogeneous effects over a set of disparate events. This reduces efficiency and leads, among other problems, to more expensive clinical trials. 

Recently, we developed a novel method to better risk stratify patients with high competing event risks, using a generalized competing event model. See our publication first showing this in endometrial cancer (PMID: 24969798). The figure above illustrates this phenomenon. Using our model, it is possible to better stratify patients according to the ratio of the hazard for cancer mortality to the hazard for overall mortality, represented by lower case omega (on the y-axis) for any given level of risk for overall mortality (on the x-axis). Optimizing this ratio leads to more efficient clinical trial designs. The figure below shows why. Standard sample size estimation schemes do not vary with this factor, whereas in reality the sample size asymptotically increases with increasing values of omega, whenever the true effect of treatment on non-cancer mortality is null or adverse.