Views 164
Citations 0
Comment & Response
April 2016

Competing Risk of Death in Kaplan-Meier Curves When Analyzing Subsequent Keratinocyte Cancer

Author Affiliations
  • 1Department of Dermatology, Erasmus University Medical Center, Rotterdam, the Netherlands
  • 2Department of Research, Netherlands Comprehensive Cancer Organization, Utrecht, the Netherlands

Copyright 2016 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.

JAMA Dermatol. 2016;152(4):493-494. doi:10.1001/jamadermatol.2015.5152

To the Editor We have read with great interest the article by Wehner et al1 about the timing of subsequent new keratinocyte carcinomas in patients who present with basal cell carcinoma (BCC) or cutaneous squamous cell carcinoma (SCC).1 The authors estimated the probability of developing a subsequent KC by calculating 1 minus the Kaplan-Meier (KM) survival probability. The use of the KM method for other end points than overall mortality can lead to a violation of a key assumption, which is the independent censoring assumption. In a KM curve with subsequent KC as the event of interest, patients who die are censored. The independent censoring assumption means that we assume that patients who are censored at time t have the same risk of developing the event of interest as those patients who are still in follow-up at time t. It is impossible to develop a KC after death, and not adjusting for this will lead to an overestimation of the probability of developing a new KC.

First Page Preview View Large
First page PDF preview
First page PDF preview