Review of criteria appropriate for a very low probability of pulmonary embolism on ventilation-perfusion lung scans: a position paper.

The "low-probability" interpretation of ventilation-perfusion lung scans has been characterized as misleading or even dangerous because of the high prevalence of pulmonary embolism associated with such an interpretation. Since the completion of the Prospective Investigation of Pulmonary Embolism Diagnosis (PIOPED) study, analyses of the PIOPED database have allowed identification of several abnormalities seen on ventilation-perfusion scans that have a positive predictive value (PPV) for pulmonary embolism of less than 10%. These include nonsegmental perfusion abnormalities (PPV = 8%), perfusion defects smaller than the corresponding areas of increased opacity at chest radiography (PPV = 8%), matched ventilation-perfusion abnormalities in two or three zones of a single lung (PPV = 3%), one to three small segmental perfusion defects (PPV = 1%), triple matched defects in the upper or middle lung zone (PPV = 4%), and the stripe sign (PPV = 7%). Use of these abnormalities as interpretative criteria constitutes "very low probability" interpretation and will reduce the number of low-probability interpretations of ventilation-perfusion lung scans, which may be considered nondiagnostic because of the unacceptably high rate of false-negative results. This will enhance the utility of the ventilation-perfusion lung scan for screening patients with suspected pulmonary embolism.