Inference about time-dependent prognostic accuracy measures in the presence of competing risks
- PDF / 1,084,025 Bytes
- 12 Pages / 595 x 791 pts Page_size
- 14 Downloads / 170 Views
(2020) 20:219
TECHNICA L A DVA NCE
Open Access
Inference about time-dependent prognostic accuracy measures in the presence of competing risks Rajib Dey1 , Giada Sebastiani2 and Paramita Saha-Chaudhuri3*
Abstract Background: Evaluating a candidate marker or developing a model for predicting risk of future conditions is one of the major goals in medicine. However, model development and assessment for a time-to-event outcome may be complicated in the presence of competing risks. In this manuscript, we propose a local and a global estimators of cause-specific AUC for right-censored survival times in the presence of competing risks. Methods: The local estimator - cause-specific weighted mean rank (cWMR) - is a local average of time-specific observed cause-specific AUCs within a neighborhood of given time t. The global estimator - cause-specific fractional polynomials (cFPL) - is based on modelling the cause-specific AUC as a function of t through fractional polynomials. Results: We investigated the performance of the proposed cWMR and cFPL estimators through simulation studies and real-life data analysis. The estimators perform well in small samples, have minimal bias and appropriate coverage. Conclusions: The local estimator cWMR and the global estimator cFPL will provide computationally efficient options for assessing the prognostic accuracy of markers for time-to-event outcome in the presence of competing risks in many practical settings. Keywords: Competing Risks, Area under the ROC curve (AUC), Cause-specific AUC, Fractional Polynomials
Background In modern evidence-based medicine, decisions on appropriate early medical intervention or the choice and timing of interventions frequently rely on prognostic markers that are informative for survival outcome. Such marker can be a single covariate or a risk score where the latter is estimated from a survival model of the association between covariates and survival time. For instance, a wellknown Framingham risk score predicts the 10-year risk of cardiovascular disease [1]. This model often guides clinicians to identify those at high/low-risk for future cardiovascular disease. Similar risk models and markers exist for other diseases, e.g. Gail score for breast cancer [2], *Correspondence: [email protected] Department of Mathematics and Statistics, University of Vermont, Burlington, USA Full list of author information is available at the end of the article 3
prostate specific antigen for prostate cancer, kidney donor risk index (KDRI) of donor for kidney transplant etc. In liver transplantation study (LT), the side-effects of medications put the organ recipients at risk for recurrent liver disease (liver fibrosis) which can lead to graft failure [3]. Liver biopsy is the gold standard for staging, managing and treating liver fibrosis, and useful in the prognosis of LT recipients. However, biopsy is often impractical due to its invasiveness and higher cost. Serum fibrosis markers such as the fibrosis score 4 (FIB-4), and the nonalcoholic fatty liver
Data Loading...
