Pre- and post-test probabilities of venous thromboembolism and diagnostic accuracy of D-dimer, estimated by European clinicians working in emergency departments

In patients with suspected venous thromboembolism (VTE), it is recommended to estimate the pre-test probability of VTE, either by experience or by standardized scoring schemes (e.g. Wells or Geneva score), before performing a D-dimer test [1], [2]. Patients with a low probability or unlikely VTE should have D-dimer performed, and if negative, VTE can be excluded, without further investigations, while if positive, the patient should be referred to radiologic imaging to confirm or exclude VTE [1], [2]. Patients with a high pre-test probability or likely VTE should be referred directly to radiologic imaging without D-dimer testing. Thus, to estimate the pre-test probability of VTE before D-dimer testing is of uttermost importance since the diagnostic algorithm and the interpretation of the D-dimer result is dependent upon this.

Further, the diagnostic accuracy of D-dimer can be expressed by the likelihood ratio (LR). LR is the ratio between the probability of a test result given that VTE is present, and probability of the same test result if VTE is absent. A LR of < 1 decreases the probability of disease and a LR of > 1 increases the probability of disease. When clinicians estimate the pre-test probability for VTE, the post-test probability (after D-dimer testing) can be calculated using the LR for D-dimer either for a positive or negative test (i.e. D-dimer above or below the cut-off level, respectively). Probably, most clinicians do not know the LR for D-dimer, but it is indirectly implemented in the above mentioned algorithm for investigating a patient with suspected VTE. The advice to rule out VTE in a patient if the pre-test probability of VTE is low and the D-dimer is below the cut-off level reflects that the post-test probability of VTE in this case is sufficiently low to rule out VTE (should be < 1–2%) [1], [2], [3], [4]. However, if the pre-test probability of VTE is high or likely, a negative D-dimer cannot decrease the post-test probability enough to rule out VTE (the post-test probability will still be about 10%) [4], [5]. Therefore, only imaging should be requested when pre-test probability of VTE is high. A positive D-dimer cannot increase post-test probability of VTE enough to confirm VTE diagnosis in any case, since D-dimer is a non-specific VTE marker and levels can be increased in several other clinical situations (e.g. infection, inflammation, cancer and pregnancy).

The aims of this survey were 1) to study the relationship between pre-test probabilities estimated by clinicians using two case histories and how the same clinicians' categorized the patients into low, moderate or high probability of VTE or into unlikely or likely VTE, and 2) to examine the clinicians' perception of diagnostic accuracy of D-dimer by calculating the LRs based upon their estimated pre- and post-test probabilities.