Inter-observer variation in the assessment of skin ulceration

Six observers (three physicians and three nurses) assessed 27 wounds of different types in 20 elderly patients, using five different classification systems. Using classification by colours, inter-observer agreement was moderate when six colours were used (70% agreement, group kappa 53%), and good if this was reduced to three (80% agreement, group kappa 64%). Results were only moderate using the Shea scoring system (68% agreement, group kappa 42%). Assessment for signs of infection was fair to moderate (75-85% agreement, group kappa 29-55%), but there was high inter-observer agreement for size and area (analysis of variance, p > or = 0.88).