Observer variation in the interpretation of ventilation-perfusion lung scintigraphy

The diagnosis of pulmonary embolism (PE) remains one of the most difficult in clinical medicine, and the diagnostic value of lung scintigraphy has been questioned. To evaluate the observer variation in the interpretation of ventilation-perfusion lung scanning in the diagnosis of PE, 87 lung scintigrams from consecutive patients with phlebography-proven deep venous thrombosis and without clinical signs of PE were randomly mixed with 50 reference lung scintigrams from patients with PE symptoms. The scintigrams were reevaluated blind by two experienced clinical physiologists. Each observer evaluated each lung scintigram twice and recorded whether the lung scan was normal or abnormal. If it was abnormal, the location and number of segment defects were registered. The intraobserver agreement, including number and location of segments, ranged from 0.77 to 0.85 and for the diagnosis of PE from 0.88 to 0.92 with a kappa of 0.80-0.84. The values for the interobserver agreement for the diagnosis of PE were 0.73-0.80 with a kappa of 0.56-0.67. It is concluded that in the interpretation of ventilation-perfusion lung scintigraphy the use of a simple scheme-deciding whether there is segmental ventilation-perfusion mismatch or not-has a good reproducibility with a high kappa for inter- and intraobserver variation and can serve as a simple routine method for diagnosing PE.