Ruling out low- and moderate-risk probability pulmonary emboli without radiological imaging: appraisal of a clinical prediction algorithm after implementation and revision with higher D-dimer thresholds.

Abstract

A study conducted at an Australian hospital between 2007 and 2009 led to the implementation of a clinical prediction algorithm for pulmonary embolism (PE) consisting of the Wells criteria and D-dimer testing. Since then, studies conducted in other countries have recommended new D-dimer cut-offs to the algorithm. To appraise the implemented algorithm and see if its ability to stratify patient risk has been maintained and to test retrospectively new D-dimer thresholds. Data from 300 patients with computed tomography pulmonary angiograms (CTPA) scans between February 2010 and March 2012 were included. Results were reviewed for algorithm compliance, patient risk stratification and PE prevalence. A higher D-dimer and an age-adjusted D-dimer cut-off were then retrospectively applied to the algorithm and results were re-evaluated. Pre-test probabilities between the current and original study remain within 1%, with slightly higher PE prevalence in the original study (12.7% vs 13.8%). Of the imaged cases, 14% (42/300) were not compliant with the algorithm. Of the low- and moderate-risk patients that underwent D-dimer testing, the retrospective application of 0.5 mg/L D-dimer cut-off resulted in a further 12.1% (95% confidence intervals 8.0-17.1%) of patients excluded from undergoing a CTPA. When combined with an age-adjusted D-dimer cut-off, 27.9% (60/215 95% confidence intervals 22.0-34.3%) were excluded from undergoing a CTPA. The implemented algorithm has maintained risk stratification and PE prevalence results. Along with evidence in other studies, we have shown that raising the D-dimer and applying age-adjusted D-dimer cut-offs might improve the efficiency of the clinical prediction algorithm in patients aged over 50.