[Skip to Navigation]
Views 392
Citations 0
Comment & Response
April 2016

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

Author Affiliations
  • 1Department of Dermatology, University of Pennsylvania, Philadelphia
  • 2Department of Dermatology, University of California, San Francisco
  • 3Department of Epidemiology and Biostatistics, University of California, San Francisco
  • 4Dermatology Service, San Francisco Veterans Affairs Medical Center, San Francisco, California
JAMA Dermatol. 2016;152(4):494-495. doi:10.1001/jamadermatol.2015.5153

In Reply We appreciate the letter by Verkouteren et al, which challenges our specialty to think about rigorous methodological issues. We agree that death is a competing risk, as it is in every survival analysis, and that overestimation is a possibility when using standard Kaplan-Meier analyses. Competing risks analysis using a cumulative incidence curve (CIC) may give a lower estimate than Kaplan-Meier, especially when death is a frequent outcome.

The goal of our article was to examine the timing and pattern of multiple subsequent new keratinocyte cancers (KCs) to determine whether follow-up care of patients with skin cancer can be improved. Prior analyses had focused on the first subsequent KC and did not account for multiple subsequent tumors (eg, third, fourth, fifth tumors). Our study was a novel way to characterize and quantify the way individual patients develop multiple KCs over time. While we were aware of CICs and competing risks of death as a potential limitation, because our primary analysis was complex and required multiple event survival analysis, we chose to use Kaplan-Meier survival analysis rather than include competing risks in our analyses.

Add or change institution