External validation of the DASH prediction rule: a retrospective cohort study

Abstract

Background The DASH prediction model has been proposed as a guide to identify patients at low risk of recurrence of venous thromboembolism (VTE), but has never been validated in an independent cohort. Aims To validate the calibration and discrimination of the DASH prediction model, and to evaluate the DASH score in a predefined patient subgroup aged > 65 years. Methods Patients with a proximal unprovoked deep vein thrombosis (DVT) or pulmonary embolism (PE) who received a full course of vitamin K antagonist or direct oral anticoagulant (> 3 months) and had D-dimer measured after treatment withdrawal were eligible. The DASH score was computed on the basis of the D-dimer level after therapy withdrawal and personal characteristics at the time of the event. Recurrent VTE events were symptomatic proximal or distal DVT/PE, and were analyzed with a time-dependent analysis. Observed 12-month and 24-month recurrence rates were compared with recurrence rates predicted by the DASH model. Results We analyzed a total of 827 patients, of whom 100 (12.1%) had an objectively documented recurrence. As compared with the original DASH cohort, there was a greater proportion of subjects with a 'low-risk' (≤ 1) DASH score (66.3% versus 51.6%, P < 0.001). The slope of the observed versus expected cumulative incidence at 2 years was 0.71 (95% confidence interval 0.51-1.45). The c-statistic was lower for subjects aged > 65 years (0.54) than for younger subjects (0.72). Conclusions These results confirm the validity of DASH prediction model, particularly in young subjects. The recurrence risk in elderly patients (> 65 years) was, however, > 5% even in those with the lowest DASH scores.