Reporting of ventilation perfusion images for pulmonary embolism: accuracy and precision.

Thirty-four hospital patients with suspected pulmonary embolism (PE) had ventilation perfusion (VP) imaging after pulmonary angiography to assess the performance of highly specific criteria for diagnosis of PE and to evaluate the precision of image reporting. A diagnostic result was obtained from VP imaging in 21 of 34 patients giving a sensitivity for PE of 80% and a specificity of 100% in this subgroup. Six of 11 patients with PE had an indeterminate study which reduced the overall sensitivity of the technique for PE in the whole group. A critical analysis of image reporting by two readers gave an 85% inter-observer agreement on ranking as diagnostic or indeterminate for PE (k 0.69) and 91%-94% agreement for consensus reproducibility (k 0.82). We conclude that the degree of accuracy and precision of reporting that can be obtained using specific criteria will provide a reliable diagnosis in a significant number of patients suspected of having PE. Consensus reporting by two readers is shown to be more reliable than individual reporting.