Stagewise pseudo‐value regression for time‐varying effects on the cumulative incidence

In a competing risks setting, the cumulative incidence of an event of interest describes the absolute risk for this event as a function of time. For regression analysis, one can either choose to model all competing events by separate cause‐specific hazard models or directly model the association between covariates and the cumulative incidence of one of the events. With a suitable link function, direct regression models allow for a straightforward interpretation of covariate effects on the cumulative incidence. In practice, where data can be right‐censored, these regression models are implemented using a pseudo‐value approach. For a grid of time points, the possibly unobserved binary event status is replaced by a jackknife pseudo‐value based on the Aalen–Johansen method. We combine a stagewise regression technique with the pseudo‐value approach to provide variable selection while allowing for time‐varying effects. This is implemented by coupling variable selection between the grid times, but determining estimates separately. The effect estimates are regularized to also allow for model fitting with a low to moderate number of observations. This technique is illustrated in an application using clinical cancer registry data from hepatocellular carcinoma patients. The results are contrasted with traditional hazard‐based modeling. In addition to a more straightforward interpretation, when using the proposed technique, the identification of time‐varying effect patterns on the cumulative incidence is seen to be feasible with a moderate number of observations. Copyright © 2015 John Wiley & Sons, Ltd.