Background: An 'unlikely' clinical decision rule with a negative D-dimer result safely excludes pulmonary embolism (PE) in 30% of presenting patients. We aimed to simplify this diagnostic approach and to increase its efficiency.
Methods: Data for 723 consecutive patients with suspected PE were analyzed (prevalence of PE, 22%). After constructing a logistic regression model with the D-dimer test result and items from the Wells' score, we identified the most prevalent combinations of influential items and selected new D-dimer positivity thresholds. The performance was separately validated with data from 2785 consecutive patients with suspected PE.
Results: Three Wells items significantly added incremental value to the D-dimer test: hemoptysis, signs of deep vein thrombosis and 'PE most likely'. Based on the most frequent combinations of these three items, we identified two groups: (i) none of these three items positive (41%); (ii) one or more of these items positive (59%). When applying a 1000 μg/L D-dimer threshold in group 1 and 500 μg/L in group 2, PE could be excluded without CT scanning in 36%, at a false-negative rate of 1.2% (95%, 0.04-3.3%). In the validation set, these proportions were 46% and 1.9% (95% CI, 1.2-2.7%), respectively. Using the conventional Wells score with a normal D-dimer result, these rates were, respectively, 22% and 0.6% (95% CI, 0.10-2.4%).
Conclusion: Combining Wells items with the D-dimer test resulted in a simplified decision rule, which reduces the need for CT scanning in patients with suspected PE. A prospective validation is required before it can be implemented in clinical practice.
Keywords: D-dimer; diagnostic test; pulmonary embolism; tomography scanners, X-ray computed; venous thromboembolism.
© 2015 International Society on Thrombosis and Haemostasis.