The use of age-dependent D-dimer cut-off values to exclude deep vein thrombosis. Reply to "Using an age-dependent D-dimer cut-off value increases the number of older patients in whom deep vein thrombosis can be safely excluded". Haematologica 2012;97(10):1507-13.

Following on from recent studies which suggest in patients over 50 years of age an age-dependent D-dimer cut off of ‘age multiplied by 10’, in conjunction with a non-high clinical probability, can be used to safely exclude pulmonary embolism (PE), we read with interest the application of the same age-dependent D-dimer cut off to the exclusion of deep vein thrombosis (DVT) by Douma et al. Their retrospective analysis of a number of different patient cohorts found incorporation of this higher D-dimer cut off was associated with only a minimal increase in diagnostic failure rate compared to the conventional cut off of 500 μg/L (0.8% vs. 0.7%), a finding replicated in the recent data of Schouten et al. We analyzed our data covering 6,599 consecutive patient episodes (individual patients may have had more than one attendance) in the out-patient DVT clinic (i.e. patients with a suspected lower limb proximal DVT not requiring hospital admission) from 2007 to 2011 to evaluate what proportion of patient management would have differed had an age-dependent D-dimer cut off been used and what the diagnostic failure rate would have been. In this cohort, 826 (12.5%) had DVT proven on a proximal compression ultrasound. As shown in Figure 1, our diagnostic algorithm for the investigation of suspected lower limb proximal DVT includes two points at which the D-dimer (STA-Liatest) may be measured. Point 1: patients with an ‘unlikely’ clinical probability score (Wells score ≤1); Point 2: patients with a ‘likely’ clinical probability score (Wells score ≥2) and a negative first lower limb proximal compression ultrasound. Patients who have already received low molecular weight heparin are ineligible for D-dimer testing. Using an age-dependent D-dimer cut off at Point 1, 89 additional patients (1.3% of total patient cohort) would have been discharged without radiological investigation, 2 of whom did have DVT demonstrated on proximal compression ultrasound giving a diagnostic failure rate of 0.03% (for total patient cohort). Using the same agedependent D-dimer cut off at Point 2, 342 patients would not have undergone a second proximal compression ultrasound (5.2% of total patient cohort), one of whom had DVT demonstrated on the second scan giving a diagnostic failure rate of 0.015% (for total patient cohort). Therefore, if an age-dependent D-dimer cut off had been implemented in our cohort it would have resulted in 431 fewer scans being performed at the expense of failing to diagnose 3 DVTs in 6,599 patients, increasing the diagnostic failure rate by 0.045%, a rate similar to that observed by Douma et al. Sixty-five percent of the scans ‘avoided’ using an age-dependent cut off would have involved patients over 70 years of age. This both correlates with the findings of Douma et al. and reflects the observation that the D-dimer increases in a continuum with age, supporting the use of a coefficient to correct for age rather than a fixed higher cut off. Although compression ultrasound is a non-invasive test with few complications for the patient, it requires hospital attendance, is only available in normal working hours, and incurs a cost of £60 per procedure (UK National Health Service reference cost). This retrospective analysis of our data suggests an age-dependent Ddimer cut off could be safely implemented, which would have meant an additional 1.5% of patients would not