Variation in the diagnostic performance of D-dimer for suspected deep vein thrombosis.

BACKGROUND: Numerous studies have evaluated the accuracy of D-dimer in diagnosing suspected deep vein thrombosis (DVT), but results are conflicting.

AIM: To overview estimates of the diagnostic accuracy of D-dimer and identify causes of variation.

DESIGN: Systematic review, meta-analysis and meta-regression.

METHODS: We searched Medline, EMBASE, CINAHL, Web of Science, Cochrane Database of Systematic Reviews, Cochrane Controlled Trials Register, Database of Reviews of Effectiveness, the ACP Journal Club, citation lists, and contacted manufacturers. We selected studies that compared D-dimer to a reference standard in patients with suspected DVT. Data were analysed by random effects meta-analysis and meta-regression.

RESULTS: We included 97 studies reporting 198 assays in 99 different patient groups. Overall estimated sensitivity and specificity of D-dimer were 90.5% and 54.7%, but both estimates were subject to significant heterogeneity (p < 0.001). Meta-regression identified that some heterogeneity was explained by study setting, exclusion criteria, whether recruitment was consecutive or the study prospective, whether D-dimer and the reference standard were measured blind, and whether the D-dimer threshold was determined a priori. Sensitivity and specificity also varied between ELISA (94% and 45% respectively), latex (89% and 55%) and whole blood agglutination assays (87% and 68%). Sensitivity was higher for proximal than distal DVT. Specificity was dependent upon whether clinical probability of DVT was high (specificity 51%), intermediate (67%) or low (78%).

DISCUSSION: D-dimer has good sensitivity, but poor specificity, for DVT. Estimates are subject to substantial heterogeneity from various sources. D-dimer specificity appears to be strongly dependent upon the pre-test clinical probability of DVT.