Spline-based estimation of cure rates: An application to the analysis of breast cancer data

This paper considers the utility of spline approximations of the hazard function for estimation of cure rates from survival data. With this approach, it is possible to assess dissimilar effects of various covariates on the probability of tumor cure and on the timing of the endpoint (death, tumor recurrence, etc.) under study. The usefulness of the approach is illustrated with an application to data on survival of patients treated for breast cancer.