Validity and reproducibility of the Cooperative Cataract Research Group (CCRG) cataract classification system

The validity and reproducibility with which six classifiers [one experienced (L.T.C.), and five novices (W.G., F.G., W.W., J.W. and O.W.)] used the CCRG cataract classification system was assessed. The validity of index classifications was assessed by computing sensitivities and pairwise interclass correlations between experienced and novice classifiers using the former's classification as the standard. The number of unordered combinations of terms in the CCRG's classification was reduced by combining cortical terms according to the CCRG's accepted system of staged simplification. The number of combinations of terms at each stage is as follows: Stage I (greater than 1000); II (127); III (63); IV (15); V (7); VI and VII (3) and VIII (2). Excellent agreement was obtained between the experienced and novice classifiers for Stages VII and VIII of the classification, good agreement for Stages V and VI and poor agreement for Stages IV, III and II (sensitivities of 97, 96, 72, 59, 40, 24 and 20% respectively). Good agreement was also achieved for the classifications of single lenticular regions, except for subcapsular regions. The intra- and interobserver reproducibility was assessed by computing the Kappa statistic to (1) compare classifications between novice observers and (2) compare repeat classifications made by the same observer by viewing the same cataract once on each of three different days. The novice classifiers had excellent intraobserver reproducibility for Stages VII and VIII (Kappas of 0.87 and 0.97 respectively), good reproducibility for Stages IV, V and VI (Kappas of 0.53, 0.62 and 0.62, respectively) and marginal reproducibility for stages II and III (Kappas of 0.39 and 0.40, respectively). The intraobserver reproducibility of the experienced classifier was superior to the others for virtually all characteristics with excellent reproducibility for Stages IV, V, VI, VII and VIII with Kappas of 0.79, 0.90, 1.0, 1.0 and 1.0, respectively and good reproducibility for Stages II and III (Kappas of 0.55 and 0.64, respectively). These results indicate that the simplified CCRG cataract classification system (Stages IV-VIII) passes the minimum standards for reproducibility. The performance of the experienced classifier far exceeds the minimum standards and indicates the feasibility of improving classifier performance with training and practice.