Abstract
Purpose: :
To evaluate inter-rater agreement in applying new criteria for grading the site and activity of intraocular inflammation.
Methods: :
Cross-sectional grading of 202 eyes of 101 patients with intraocular inflammation was conducted at three uveitis subspecialty clinics by pairs of uveitis subspecialists. Agreement was calculated for gradings of the location of inflammation, anterior chamber cells, anterior chamber flare, vitreous cells (present or absent), and vitreous haze.
Results: :
Gradings of the location of intraocular inflammation using the new criteria had moderate reproducibility (kappa ranges 0.49-0.61). Excluding the newly proposed category of "anterior and intermediate uveitis," reproducibility was better (kappa ranges 0.61-0.73). Exact agreement on gradings of anterior chamber cells demonstrated low to moderate levels of agreement (kappa ranges 0.34-0.43), whereas kappa statistic ranges were in the moderate agreement range for gradings of anterior chamber flare (0.50-0.64), vitreous cells (0.48-0.51) and vitreous haze (0.53). When the standard of agreement was liberalized to include agreement within one grade, reproducibility was outstanding for anterior chamber cells (kappa ranges 0.81-1.00) and vitreous haze (kappa=0.75). For anterior chamber flare, a distribution skewed toward low grades made "within one grade" kappa statistics unstable.
Conclusions: :
Newly proposed methods for grading the site of intraocular inflammation produce moderate levels of agreement; in our hands, reproducibility was improved by excluding "both anterior and intermediate uveitis." New methods for grading inflammatory activity have moderate reproducibility for exact agreement in most instances, which improves to excellent agreement when a standard of agreement within one grade is used for anterior chamber cells and vitreous haze. Improved grading systems for anterior chamber flare and vitreous cells are needed.
Keywords: uveitis-clinical/animal model • clinical (human) or epidemiologic studies: biostatistics/epidemiology methodology • clinical (human) or epidemiologic studies: outcomes/complications