Rater agreement in lung scintigraphy.

Abstract

The PIOPED criteria in their original and revised forms are today's standards in the interpretation of ventilation-perfusion scintigraphy. When the PIOPED criteria are used by experienced raters with training in consensus interpretation, the agreement rates have been demonstrated to be excellent. Our purpose was to investigate the rates of agreement between 2 experienced raters from different hospital who had no training in consensus interpretation. The 2 raters investigated a population of 195 patients. This group included 72 patients from a previous study who had an intermediate probability of pulmonary embolism and who had also been examined by pulmonary angiography. The results demonstrated moderate agreement rates with a kappa value of 0.54 (0.45-0.63 in a 95% confidence interval), which is similar to the kappa value of the PIOPED study but significantly lower than the kappa values of agreement rates among consensus-trained raters. There was a low consistency in the intermediate probability category, with a proportional agreement rate of 0.39 between the experienced raters. The moderate agreement rates between raters from different hospitals make it difficult to compare study populations of a certain scintigraphic category in different hospitals. Further investigations are mandatory for accurate diagnosis when the scintigrams are in the category of intermediate probability of pulmonary embolism.