BACKGROUND/ AIMS Standardised numeric grading scales are used in ophthalmic practice to improve consistency between clinicians in recording the severity of ocular conditions and to facilitate the monitoring of such changes. We investigated the intra- and inter-observer grading reliability and the agreement between subjective Cornea and Contact Lens Research Unit (CCLRU) and Efron grading scales as well as a new Advanced Ophthalmic Systems (AOS) software which uses an objective approach to grading conjunctival hyperaemia. METHODS One experienced observer graded n = 30 bulbar and n = 26 palpebral conjunctival hyperaemia images to 0.1 increments. Masked grading of randomised images was undertaken for all three methods, on two separate occasions. The agreement within and between the grading methods was assessed between sessions, and compared to the results of a novice observer. RESULTS There were no statistically significant differences (P > 0.05) between test and retest values. However, repeatability in the grading estimates of both bulbar and palpebral conjunctival hyperaemia was improved using the AOS grading method (R2 = 0.998; Coefficient of Repeatability CoR 0.10-0.13), compared to Efron (R2 = 0.926; CoR 0.62) and CCLRU (R2 = 0.885-0.911; CoR 0.50-0.78). Intraclass coefficient correlations (ICC) improved inter-observer agreement using objective (> 0.995) versus subjective methods (0.853-0.959). CONCLUSION These subjective and objective grading methods are not interchangeable. Due to the excellent repeatability and improved agreement between experienced and novice observers, the objective grading method provides a more consistent approach when grading ocular abnormalities and may achieve greater reliability in record keeping and clinical monitoring in the future.

中文翻译：

---

评估结膜充血的新客观评分软件。

背景技术/ AIMS在眼科实践中使用标准化的数字等级量表，以提高临床医生之间在记录眼部疾病的严重性方面的一致性并促进对这种变化的监测。我们调查了观察者之间和观察者之间的分级可靠性以及主观角膜和隐形眼镜研究单位（CCLRU）和Efron分级量表之间的协议，以及使用客观方法对结膜充血进行分级的新的高级眼科系统（AOS）软件。 。方法一名经验丰富的观察者将n = 30球囊和n = 26睑结膜充血图像分级为0.1增量。在两种不同的情况下，对这三种方法进行了随机图像的蒙版分级。在会议之间评估评分方法之内和之间的一致性，并与新手观察员的结果进行了比较。结果测试值和复测值之间无统计学差异（P> 0.05）。然而，与Efron（R2 = 0.926; CoR 0.62）和CCLRU（R2）相比，使用AOS分级方法（R2 = 0.998;重复性系数CoR 0.10-0.13）提高了延髓和睑结膜充血的分级估计的可重复性= 0.885-0.911; CoR 0.50-0.78）。类内系数相关性（ICC）与主观方法（0.853-0.959）相比使用客观（> 0.995）改善了观察者之间的一致性。结论这些主观和客观的评分方法是不可互换的。由于出色的可重复性以及经验丰富的和新手观察者之间的共识，