Table 1

Relevant papers

Author, date, countryPatient groupStudy typeOutcomesKey resultsStudy weaknesses
Dentali et al,1
1134 patients with suspected or confirmed CVST, diagnosed by CT venography/MRI/magnetic resonance venography (MRV)/angiography.Systematic review and meta-analysis.Diagnostic accuracy of D-dimer for CVST.10/155 CVST cases had normal (<500 ng/mL) D-dimer. 93.9% sensitivity (95% CI 87.5 to 97.1).Variable study design of included studies, for example, D-dimer assays used, exclusion of certain patients with risk factors (pregnancy and OCP). Low number of included studies. Low quality of included studies: no randomised controlled trials; only retrospective/prospective studies.
Hiltunen et al,2
71 patients from Helsinki University Central Hospital discharge register from 1987 to 2010, with radiologically confirmed CVST and D-dimer measured before initiation of anticoagulant treatment.Retrospective.D-dimer level in CVST cases.9/71 CVST cases had normal D-dimer (<0.5 mg/L). Sensitivity: 87.3%. Negative predictive value: >95%.Retrospective nature of study led to exclusion of half of patients with CVST due to inclusion criteria. Study only looked at patients presenting to Helsinki hospital; this group may not be representative of the general global population.
Alons et al,3 2015,
The Netherlands
636 patients with isolated headache and normal neurological examination. D-dimer was determined at presentation, and the presence of CVST was examined with CT venography/MRI or both.Systematic review and meta-analysis.D-dimer level in CVST cases.45/636 CVST cases, 1 had normal (<0.5 µg/mL) D-dimer. Sensitivity: 97.8%. Negative predictive value: 99.8%.Six potential articles excluded from meta-analysis due to incomplete/missing data. Assays used to determine D-dimer level varied between studies. Normal neurological examination is subjective, and assessment may vary between clinicians. The authors have not specified what they define as ‘normal’. Study only looks at ‘low risk’ patients, that is, those with isolated headache and normal neurological examination. Higher risk patients with additional neurology may have been excluded but would have still been relevant to the BET question.
Hui et al,4
139 patients with CVST presenting to their hospital between 1 January 2004 and 1 August 2014. CVST confirmed with MRI/MRV or digital subtraction angiography. D-dimer was measured within 24 hours of presentation.Prospective.D-dimer level in CVST cases.74/139 CVST cases had normal D-dimer. Sensitivity: 46.8%.The study excluded patients with CVST with a family history of thrombosis. This could skew the data and make the results less applicable to the general population.
  • BET, Best Evidence Topic; CVST, cerebral venous sinus thrombosis.