More Nomograms or Better Evidence of Efficacy: What Do We Need in Urologic Oncology?

Abstract

The publication of the nomogram to predict radical prostatectomy recurrence using preoperative parameters by Kattan et al in 1998 was an important event in urologic oncology [1]. The authors used statistical regression to measure the impact of various clinicopathologic parameters on the likelihood of postoperative tumour recurrence. The final nomogram represented an amalgamation of these scores and accurately identified outcomes in 74–79% of patients. This report revealed the importance of preoperative measures for determining subsequent outcome and brought nomograms to the attention of the urologic community. Because nomograms provide individualised risk predictions for patients in an illustrated understandable manner and combine multiple statistical analyses, they have become very popular among clinicians and statisticians. Nomograms have now been applied to nearly every aspect of urology. Because the nomogram is only as good as the analysis and input data, with time the training data sets have become larger whilst the pool of suitably expert statisticians remains small. An example of this paradox is demonstrated by the authors whose timely review of the subject is published in this month’s journal [2]. This group has developed and published numerous nomograms that can be used to guide treatment choices in urologic oncology. These are freely available on the Internet [3] and both patients and clinicians are encouraged to use them. By collaboration with large units, the authors have created