Objective: To determine to what extent the 'non-diagnostic' lung scans made because of a clinical suspicion of pulmonary embolism enable further determination of the risk of pulmonary embolism.
Design: Retrospective.
Method: All non-diagnostic lung perfusion ventilation scans made in the Academic Medical Centre of Amsterdam in 1997 of 114 patients in succession (55 males and 59 females aged 27-85 years) were subjected to blind and independent re-evaluation by three observers (an experienced nuclear medicine expert, an experienced and an inexperienced intern) who, using a lung segment chart, estimated the risk of embolism as < 25%, 25-50%, 50-75% and > 75%. They did this first without and then with the chest X-ray. The findings were grouped on the basis of accordance or non-accordance with the pulmonary angiogram. The interobserver agreement was calculated by means of kappa statistics.
Results: Of 58 patients the lung scan could be compared with a chest X-ray and a pulmonary angiogram. In 43 patients with a normal angiogram the observers in an average of 50% of the scans estimated the risk of pulmonary embolism as < 25%, as against 25-50% in 27%, 50-75% in 9% and > 75% in 5%. In 15 patients with a deviant pulmonary angiogram, these figures were 22%, 38%, 20%, and 12%, respectively. The interobserver kappa for evaluation without chest X-ray was < or = 0.16, as against < or = 0.41% with the chest X-ray.
Conclusions: A reliable classification of the risk of pulmonary embolism was not possible on the basis of non-diagnostic lung scans, regardless of whether the patient did or did not have pulmonary embolism. The interobserver variability was less when the lung scan was evaluated together with the chest X-ray, but even so it was unacceptably high.