Grading and scoring in histopathology

In many areas of histopathology a nominal category, such as a diagnosis of breast carcinoma, does not give enough information for the referring clinician to make decisions about patient prognosis and treatment. Therefore scoring and grading systems have been developed which provide additional information. This article reviews the principles behind these systems with particular reference to the relationships between the natural clustering (or nonclustering) of cases and the imposition of arbitrary class boundaries on such distributions. The difference between real numbers and the ordinal categorical numeric labels, which are often produced by histopathology scoring systems, is discussed. The reproducibility of scoring and grading systems is reviewed and generic suggestions are given for developing new systems and for their validation.