[Skip to Navigation]
Views 199
Citations 0
Comment & Response
December 2018

Incomplete Data in Cutaneous Squamous Cell Carcinoma Staging System Analysis—Reply

Author Affiliations
  • 1Department of Dermatology, Oslo University Hospital, Oslo, Norway
  • 2Oslo Centre for Biostatistics and Epidemiology, Norway
JAMA Dermatol. 2018;154(12):1489. doi:10.1001/jamadermatol.2018.4018

In Reply On behalf of our coauthors, we thank Jambusaria-Pahlajani and colleagues for their interest in and comments to our study1 and appreciate the opportunity to respond to their concerns.

In retrospect, we recognize that the phrase, “the Breuninger system gave the best results,” following our assessment of the validity of 4 staging systems for identification of patients with cutaneous squamous cell carcinoma (cSCC) at high risk for metastasis, was not accurate, particularly after performing a reanalysis of the data. In our study, we categorized patients based on available information on risk factors, but realize that some patients potentially could have been categorized in a higher risk category if more data on variables necessary for accurate staging had been available. After performing multiple imputation of missing data, the categorization of some patients changed, leading to changes in the odds ratios for metastasis in all 4 staging systems. Also, the sensitivity, specificity, correctly classified, concordance index (C index), and R2 value changed, including significant improvements for the Brigham & Women’s Hospital (BWH) system, coming on par with and in some respects even above the Breuninger system. Therefore, the reanalysis does not support our original comment that the Breuninger system gave the best results.

Add or change institution