Concordance of the modified wells score with CTPA-confirmed pulmonary embolism in an imaging-selected cohort: A retrospective single-center study

Abstract

Background Pulmonary embolism (PE) is a major diagnostic challenge in emergency medicine. Computed tomography pulmonary angiography (CTPA) is the gold standard but is frequently overused. The modified Wells Score (mWS) estimates pre-test probability and may reduce inappropriate imaging at triage. We evaluated the concordance between mWS and CTPA-confirmed PE in a tertiary emergency department. Methods We retrospectively analyzed all patients undergoing CTPA for suspected PE over 24 months. As all patients had already undergone CTPA, this design reflects concordance within an imaging-selected cohort, not triage performance. mWS values were assigned by consensus between two readers, blinded to CTPA results; D-dimer was not incorporated. AUC was computed from the full ordinal mWS range; sensitivity, specificity, PPV and NPV were calculated at the conventional mWS ≥ 4 threshold. Spearman's correlation assessed the mWS–PE severity association. Results Among 691 patients, 103 (14.9%) had CTPA-confirmed PE. At mWS ≥ 4, sensitivity was 86.4%, specificity 98.6%, AUC 0.943 (95% CI: 0.91–0.97). The Youden index identified mWS ≥ 3 as optimal (sensitivity 89.3%, specificity 95.7%); ≥ 4 was retained for analysis. Under this threshold, 594 patients (86.0%) fell below the cut-off, suggesting avoidable imaging, though unconfirmed in non-imaged patients. Fourteen false-negative (3 central PE) and eight false-positive cases were identified. Higher mWS correlated moderately with PE burden (ρ = 0.578, p < 0.01). Conclusion mWS showed strong concordance with CTPA-confirmed PE; diagnostic accuracy exceeding pooled meta-analytic estimates likely reflects consensus-scoring methodology rather than genuine superiority. Prospective, multicentre validation incorporating D-dimer testing and independent scoring is warranted before routine implementation.