Identifying Patients at High Risk for Deep Vein Thrombosis (DVT) After Acute Stroke Using Clinical and Laboratory Predictors.

Abstract 3825

The national Venous Thromboembolism prevention programme in England emphasises reducing hospital-associated thrombosis by ensuring adequate prophylaxis, but this is problematic in stroke patients due to poor uptake of pharmacological prophylaxis and lack of evidence for mechanical measures. Our prospective study aimed to develop clinical and laboratory predictors of thrombosis in patients after acute stroke. Participants were recruited within 48 hours of admission to the Acute Stroke Centre at King's College Hospital London UK, between June 2009 and June 2010. 142 participants were recruited with complete data available from 86 participants. Racial distribution was as follows: 75% Caucasian, 19% African/Caribbean, 6% Asian. Mean age was 69 years (SD15) and 45/86 (52%) were female. 73 strokes were ischaemic and 13 haemorrhagic. In the ischaemic stroke group, 33 of 73 (45%) were thrombolysed as per National stroke strategy. Baseline demographic data and clinical evaluation using the NIH Stroke Scale (NIHSS) and Barthel Index (BI) were recorded. Analysis of prothrombotic markers was performed including D-dimer (DD), Clauss fibrinogen (CFIB) and thrombin generation (TG) measured by calibrated automated thrombography (CAT). The TG assay measured kinetics in platelet poor plasma including lag time (LT) - time taken to generate 10nM of thrombin; time to peak (TTP) - time in reach maximum TG; peak thrombin generation (PTG)- maximum thrombin concentration; and endogenous thrombin potential (ETP) - area under the curve (expressed in nmol/L of thrombin). In the second week all laboratory tests were repeated and lower limb compression ultrasound scan (CUSS) performed. 18/86 (20.9%) had objectively confirmed DVT (5 proximal and 13 distal). 4 DVTs (2 proximal and 2 distal) occurred in patients with haemorrhagic stroke, and 14 DVTs (3 proximal and 11 distal) occurred in patients with ischaemic stroke. Both BI (p=0.02) and body mass index (p=0.05) but not NIHSS (p=0.13) were associated with DVT. TG assay parameters and CFIB were not associated with the risk of DVT. Elevated DD was significantly associated with DVT development both at baseline (DD 2425 ng/ml vs 1010 ng/ml, p=0.0010) and at week 2 (DD 2770 ng/ml vs 830 ng/ml, P<0.001). A subsequent analysis examining the effects of NIHSS, BI and DD using multiple logistic regression, indicated that only DD at baseline was significantly associated with DVT. After adjusting for the effect of this variable, no other variables were found to be significant. ROC curves were used to determine the optimum cut-off for each of the three key variables (NIHSS, BI and DD). The results indicated that the area under the ROC curve (AUC) was greatest for DD (0.77), followed by BI (0.70), and NIHSS (0.62). The best predictive performance for DVT was for DD, with the cut-off of 1230 ng/ml providing a sensitivity of 88% and a specificity of 64%. Limitations of the analysis include the relatively small number of DVT cases and further patients will be accrued before the study is complete. In conclusion, the incidence of DVT after stroke remains high (20.9%) despite the trend towards more aggressive treatment with thrombolysis. BI and DD (but not NIHSS, TG and CFIB) are important predictors of DVT in this population, and are of potential use in identifying high-risk patients for thromboprophylaxis following acute stroke.