Usefulness of predictive models of pulmonary embolism: Experience in real-world clinical practice in a regional hospital

Abstract Background Faced with the suspicion of pulmonary embolism (PE), the guidelines recommend the use of clinical probability scales, measurement of D-dimer and, in certain cases, confirmation by pulmonary angiography by computed tomography (CTPA) or scintigraphy. Recently, it has been proposed to adjust the D-dimer according to age or use simpler scales (YEARS algorithm) for a better selection of patients. Objective To define the degree of application of the guidelines in our population and compare sensitivity, specificity and positive and negative predictive values of the different diagnostic models: standard model (Wells 2 categories + D-dimer), model adjusted for age (Wells 2 categories + D-dimer adjusted for age), YEARS algorithm. Material and methods A retrospective study of all patients who underwent APTC at our centre for the diagnosis of PE over one year. Results Of 618 cases (85.4% of initial APTC), 544 patients were included. A total of 113 EPs were diagnosed (20.8%). The degree of application of the standard model was very high (90.1%) and proved to have the best sensitivity and negative predictive value ratio (sensitivity = 1.0, negative predictive value = 1.0). The new models could reduce the number of scans (17, 3.2% model adjusted for age and 48, 8.8% model YEARS) with a risk of false negatives (2 PE [1.8%] undiagnosed respectively). Conclusions The current diagnostic models for PE lead to a large number of unnecessary explorations. The new models could reduce the number of APTC although with a minimum risk of false negatives.