Objectives: The current literature on diagnosis of periprosthetic joint infection (PJI) provides controversial evidence on the diagnostic accuracy of D-dimer. Therefore, this critical literature search and meta-analysis was aimed to summarize the diagnostic accuracy of D-dimer for diagnosing PJI.

Content: We searched MEDLINE, Scopus, and Web of Science, for studies on D-dimer for diagnosing PJI, according to the PRISMA flowchart. QUADAS was used for assessing study quality. Sensitivity, specificity, positive (PLR) and negative likelihood ratio (NLR), and diagnostic odds ratio (DOR) were analyzed using bivariate diagnostic random-effects model. The area under the receiver-operating curve (AUC-ROC) was calculated. Subgroup analysis and univariate meta-regression were carried out for detecting potential sources of heterogeneity.

Summary: We included 12 articles, totaling 1,818 patients (539 with PJI). The pooled sensitivity and specificity of D-dimer for diagnosing PJI were 0.739 (95% CI: 0.616-0.833) and 0.785 (95% CI: 0.679-0.863). The pooled PLR, NLR, DOR were 3.359 (95% CI, 2.340-4.821), 0.295 (95% CI, 0.180-0.484), and 11.787 (95% CI, 5.785-24.018). The cumulative ROC plot displayed an AUC of 0.688 (95% CI, 0.663-0.713; p<0.001). No threshold effects could be observed. The type of blood sample was identified as possible source of heterogeneity for DOR (p=0.01).

Outlook: Evidence emerged from this meta-analysis suggests that D-dimer displays sufficient diagnostic accuracy to rule out PJI. The type of blood sample (plasma vs. serum) and the study design could influence the results in terms of DOR and sensitivity. However, further perspective studies would be needed to validate its potential diagnostic usefulness.