Objectives: Emergency room (ER) evaluation may differ when physicians see patients in the context of clinical trials compared to routine care. We aimed to determine whether patterns in the use of a systemic pretest probability (PTP) stratification tool and D-dimer testing change over time and whether this influences their clinical utility.Methods: Charts were reviewed of 974 cases that had D-dimer testing for VTE exclusion in two time periods; 474 consecutive patients evaluated between 07/02 and 10/02 and 500 between 07/04 and 10/04. The former cohort was managed by ER physicians that had just participated in a clinical trial on VTE diagnosis including PTP assessment and D-dimer testing, whereas the latter group had received no formal training after 2002. In both cohorts, pretest scoring as low, intermediate or high risk according to Wells criteria for DVT and PE was performed in every case since this was requested from the laboratory. D-dimer testing (Vidas® D-Dimer, Biomerieux) was performed only in the low and moderate risk group after reception of the PTP assessment form. Physicians were also asked to check out a form detailing individual Wells criteria leading to the overall assessment but were not mandated to do so in order to obtain the D-dimer result from the laboratory.Results: Pretest probability was evaluated as low, moderate or high in 66,7%, 31,9% and 1,3% vs. 78,4%, 21,2% and 0% (all comparisons are 2002 vs. 2004). There was a significant increase in proportion of undetailed evaluations of PTP (32,7% vs. 61%). Detailed forms in both cohorts had similar risk distribution (low/moderate risk 51,7/46,4% vs. 59,2/40,8%) whereas undetailed forms were usually quoted as low risk (97,4% vs. 90.5%). Number of D-dimer tests performed per month was stable over the three year period of observation and in the two evaluated cohorts (119/mth vs. 125/mth). D-dimer results were negative in 299/474 cases (63,1%) in 2002 and 359/500 (71,8%) in 2004 (p=0,003), although this difference was less apparent when analysed according to the individual PTP risk groups (low/moderate risk 77,3/51,9% vs. 71,2/48,3%). Incidence of VTE events decreased over time from 5,3 to 1,6% (p=0,002). Incidence in the low/moderate risk groups was 1,6%/10,6% vs. 0,3%/6,6%. Only one false negative result (popliteal vein DVT) was observed in the two cohorts (NPP = 99,9%).Conclusion: Our results show a decreasing incidence in VTE over time in an ER population screened by D-dimer testing and PTP even though the number of tests performed were stable over time. This was accompanied by a decreasing number of cases considered to have an intermediate PTP. These findings suggest a change of practice over time resulting in an increasing use of D-dimer testing for very low risk patients and a decrease in their use for intermediate risk patients. A decrease in the proper use of the PTP tool over time might result in overestimation of the physician perceived risk and therefore lead to an increase in imaging resource utilisation. Broader studies including imaging prescription trends over time will be needed to confirm this hypothesis.