Competing risks occur frequently in follow-up clinical studies. To assess treatment or covariate effects, measures of clinical impact based on crude cumulative incidence should be considered, such as relative risks or the absolute risk reduction. In this work, transformation models through suitable link functions provide a straightforward approach to obtain point and interval estimates of such measures. An extension of the Klein and Andersen proposal, based on pseudo-values, is considered. Non-additive effects were tested by interactions between baseline (spline function on time) and covariates. The methods are applied to the evaluation of the impact of axillary lymph node nanometastases on metastatic relapse of breast cancer patients. Further, a literature data set on prostate cancer was used for illustration.
- Clinically useful measures
- Competing risks
- Generalized estimating equations
ASJC Scopus subject areas
- Statistics and Probability