Abstract
Purpose::
Objective grading systems for posterior capsule opacification (PCO) are often tested against subjective grading by an ophthalmologist. In this study 4 ophthalmologists graded reflected light images of PCO to examine the inter and intra-observer repeatability both with and without the aid of standard reference images.
Methods::
Digital reflected light images were recorded from 27 eyes of 23 subjects presenting for YAG capsulotomy at the Royal Eye Unit, Kingston Hospital using standard settings on a Nikon FS-2 photo slit lamp. Images were recorded using a Kodak DCS digital camera system. For grading, the images were presented on a computer screen in random sequence to 4 ophthalmologists with the same image presented between 1 and 5 times. Graders recorded their results on a 10cm visual analogue scale (VAS) representing very mild to severe opacification of a central delineated area of the capsule. The exercise was repeated at a later date while simultaneously presenting the graders with 3 reference images representing the 1cm, 5cm and 9cm points on the VAS measured using an objective system based on total gray level in the area of interest.
Results::
With no reference images, there was a statistically significant difference (P=0.00; ANOVA) between graders with all pairs statistically significantly different except 1 (post hoc Tukey HSD). The inter-observer coefficient of repeatability was 4.16cm (total scale length 10cm) without reference images reducing to 3.45cm when reference images were available to graders. Use of reference images removed any statistically significant difference between observers (P=0.082; ANOVA). Intra-observer repeatability was assessed on 19 eyes that had between 2 and 5 repeated gradings each for all 4 graders. There was no association between the degree of severity and the standard deviation both with and without the use of reference images (P>0.5 both cases) so the standard deviation was used. There was a statistically significant difference between the standard deviation values with and without reference images (P=0.035; Mann-Whitney) with the intra-observer repeatability worse when using reference images (coefficients of repeatability 3.57cm versus 2.74cm).
Conclusions::
Comparison of objective measures against subjective grading of reflected light images of PCO may not show good agreement due to significant inter and intra-observer variability. Repeat grading of images and the use of reference images helps to reduce this variability between graders but doesn't improve intra-observer repeatability. Experience and training of the graders is likely to be a significant factor.
Keywords: posterior capsular opacification (PCO)