Abstract
Purpose :
A digital and quantitative assessment of scleritis activity can facilitate patient care and establish reproducible, masked measures for clinical research. We performed a clinical validation study of an image-based standardized grading system of scleritis using digital slit lamp photos by evaluating the agreement between clinical exam grades and digital grades. We also evaluated the intra-rater reliability of digital gradings.
Methods :
Eight scleritis patients were evaluated with clinical and digital gradings over a mean of 11.8 visits (range, 9 – 13) per patient (5 right eyes, 3 left eyes). Using this photographic database, a score of 0 to 3+ (0, 0.5+, 1+, 2+, 3+) was assigned to each quadrant of the study eye by using the photographic guidance that the NEI scleritis grading system published previously. Clinical grading based on the exam was unmasked and was performed by two investigators as part of a separate, unrelated clinical trial. Digital gradings based on slit lamp photos collected at each visit were determined by using standard screen resolution (1024 x 768). The grader was masked to the order of the images, which was randomized. Agreement between clinical and digital grading was calculated for each quadrant, and the intra-rater reliability was evaluated for each quadrant by repeated grading on a random sample of the photos.
Results :
A total of 115 photos were graded, including 95 patient visits and a random sample of 20 repeated photos. Reliability analysis demonstrated moderate agreement between clinical and digital grading (mean weighted kappa coefficient = 0.56). Bland-Altman analysis showed a mean difference of +0.48 higher grade in the digital grading (95% limits of agreement, -0.88 to 1.84). Intra-rater reliability of the digital grading was very good (mean weighted kappa coefficient = 0.89).
Conclusions :
Digital grading using a standardized grading scale for scleritis based on digital slit lamp photos is a reliable method of quantifying inflammatory activity, and it has the potential to allow use of reading centers for scleritis trials. However, the agreement between clinical and digital grading is moderate and needs to be improved with further investigations to account for systematic differences between the two gradings and to determine the clinical difference between steps of the grading scale.
This is an abstract that was submitted for the 2016 ARVO Annual Meeting, held in Seattle, Wash., May 1-5, 2016.