Background: To determine the optimal diagnostic cut-off point using a simplified criterion for the detection of pulmonary embolus (PE) and to evaluate the criterion's utility and reporter reproducibility.
Methods: Lung scintigraphy was carried out in 924 patients for the diagnosis of PE. This group consisted of 316 men and 608 women with median age of 63 years (range 18-94 years). Ventilation imaging was carried out with Tc-99m Technegas followed by perfusion imaging using 190 MBq Tc-99m macroaggregated albumin. Studies were classified using a 6-category probability criterion of incremental ventilation/perfusion (V/Q) mismatch: A, normal; B, low (minor matched V/Q defects or segmental matched V/Q defects without opacity on chest X-ray); C, low-moderate (a partial segment of V/Q mismatch); D, moderate (1 segment of mismatch); E, moderate-high (1-2 segments of V/Q mismatch) and F, high probability (=2 segments of V/Q mismatch). Clinical end-points at 3 and 6 months were death by PE or PE treated with anticoagulation therapy. Three-reporter reproducibility was determined by kappa statistic on a subgroup of patients (53/924).
Results: A total of 122 patients (13%) had a confirmed diagnosis of PE at 3 months and no additional cases were registered at 6 months. The lung scintigraphy probability classification showed: normal 152 (16%), low 620 (67%), low-moderate 20 (2%), moderate 28 (3%), moderate-high 24 (3%) and high 80 (9%). The respective sensitivities and specificities, where the diagnostic cut-offs were established at F, high; E, moderate-high; D, moderate and C, low-moderate probability, were F, 64 and 100%; E, 82 and 99%; D, 95 and 98% and C, 98 and 96%. The respective false-negative cases for F, E, D and C cut-offs were 44, 22, 7 and 3. Using the revised Prospective Investigation of Pulmonary Embolism Diagnosis reporting classification reporter agreement showed kappa values of 0.31-0.48. Using a simplified 2-category (>0.5 segment of V/Q mismatch positive, all others negative) criterion resulted in a higher reporting agreement (kappa 0.74-0.83). There were only 3% of indeterminate cases if this was defined by the D category and a maximum of 8% if categories C, D and E were included.
Conclusions: Using a simplified diagnostic criterion where all studies showing >0.5 segments of V/Q mismatch are regarded as positive and all others as negative, lung scintigraphy, incorporating Tc-99m Technegas ventilation imaging or its equivalent, can achieve a very high diagnostic accuracy for the detection of PE. Using this technique, less than 5% of scans are indeterminate. A simplified, unambiguous approach to reporting is recommended.