Ventilation-perfusion lung scans for pulmonary emboli. Accuracy of reporting.

Reports were made on combined ventilation-perfusion lung scans by three observers on three occasions and by another observer once. Reproducibility for each observer varied between 80 and 88%. There was complete agreement about the areas of scans reported as abnormal. Agreement between observes on whether or not the abnormality represented a pulmonary embolus averaged 77%. There was 86% agreement with the final clinical diagnosis. Our results show that reporting of ventilation perfusion lung scans by eye is reproducible. They support the claims that, under routine clinical conditions, the technique is 91% to 95% accurate for pulmonary emboli.