[Usefulness of clinical prediction rules for ruling out deep vein thrombosis in a hospital emergency department].

Objectives: To compare the efficacy of the most familiar clinical prediction rules in combination with D-dimer testing to rule out a diagnosis of deep vein thrombosis (DVT) in a hospital emergency department.

Material And Methods: Retrospective cross-sectional analysis of the case records of all patients attending a hospital emergency department with suspected lower-limb DVT between 1998 and 2002. Ten clinical prediction scores were calculated and D-dimer levels were available for all patients. The gold standard was ultrasound diagnosis of DVT by an independent radiologist who was blinded to clinical records. For each prediction rule, we analyzed the effectiveness of the prediction strategy defined by "low clinical probability and negative D-dimer level" against the ultrasound diagnosis.

Results: A total of 861 case records were reviewed and 577 cases were selected; the mean (SD) age was 66.7 (14.2) years. DVT was diagnosed in 145 patients (25.1%). Only the Wells clinical prediction rule and 4 other models had a false negative rate under 2%. The Wells criteria and the score published by Johanning and colleagues identified higher percentages of cases (15.6% and 11.6%, respectively).

Conclusion: This study shows that several clinical prediction rules can be safely used in the emergency department, although none of them have proven more effective than the Wells criteria.