Objectives: A binary system is reputed to be superior to the World Health Organization (WHO) system in grading oral dysplasia. We aimed to validate its reproducibility and prognostic ability and examine whether it could be improved.
Study design: Three pathologists graded 141 oral epithelial dysplasia biopsies with the use of both systems. Observer variability and prognostic ability were assessed with the use of kappa and logistic regression models.
Results: The binary system showed superior agreement to the WHO system (multirater kappa 0.59 vs. 0.49, respectively) but similar prognostic ability (odds ratio [OR] 4.59 [P = .014] vs. OR 2.25 [P = .02], respectively). Adding smoking and alcohol slightly improved the prognostic ability of both systems (OR 5.10 vs. OR 2.42, respectively). Our new binary system with a refined diagnostic threshold demonstrated a slightly greater prognostic ability and improved ability to differentiate between high- and low-risk moderate dysplasia cases.
Conclusions: The binary system has similar prognostic ability but superior reproducibility compared with the WHO system. Prognostication is improved still further by using a new threshold.
Copyright © 2013 Elsevier Inc. All rights reserved.