Interobserver Variance in Clinical Scoring for Cystic Fibrosis

Multicenter clinical research would benefit from a simple, reliable scoring system for comparison of the clinical status of patients at different centers. In this study, five physicians performed simultaneous, independent scoring of 41 individuals with cystic fibrosis using the Doershuk modification of the Shwachman-Kulczycld scoring system for history, physical examination, and nutrition, and the Birmingham scoring system for chest roentgenograms. These were added together to obtain a clinical score. Interobserver variance of the scores was calculated. Mean individual observer variance from the consensus mean was 1.6-2.9 score points of a possible 25 for each category, 4.5-6.0 of a possible 100 for the total score. Coefficient of variance about the mean was approximately 10 percent for the individual categories, 6.7 percent for the total score. We concluded that the interobserver variance of this scoring system is within acceptable limit for most clinical studies. The total consensus score correlated with the NIH clinical score, chest roentgenogram score alone, and predicted values for forced vital capacity and FEV1 with a high degree of confidence.