Evaluation of classification systems for nonspecific idiopathic orbital inflammation

To systematically analyze existing classification systems for idiopathic orbital inflammation (IOI) and propose and test a new best practice classification system.

A systematic literature search was conducted to find all studies that described and applied a classification system for IOI. Classification categories used in more than two studies were extracted, and criteria for these categories were defined using common descriptors. Using patient data, these newly defined classification systems were evaluated. Reliability was tested by inter- and intrarater agreement of two raters and distinction tested by evaluating clinical differences among classification categories. Feasibility, face validity, and content validity were qualitatively tested.

The most frequently encountered IOI classification systems were based on onset (acute, chronic), histopathology (classic, granulomatous, sclerosing), or localization (diffuse, extraocular muscle, lacrimal gland, sclera, optic nerve). Systems based on histopathology and localization showed good reliability (κ values range 0.74-0.89), were easy to apply (feasibility), and described the biologic process (face validity). Because of static sampling, histopathology-based systems had moderate content validity and moderate distinction between classification categories. Being a static measure, localization had moderate content validity, but good distinction. It was found that content validity was improved by combining histopathology and localization into a two-dimensional classification system.

This combined histopathology and localization-based classification system provides a repeatable, easy to use, plausible, and complete classification system that can be used to further advance the research of IOI.