The high-risk bleeding category of different scores in patients with venous thromboembolism: Systematic review and meta-analysis.

BACKGROUND

In patients with venous thromboembolism (VTE), bleeding risk should be carefully assessed but none of the available risk scores is currently recommended. The aim of this study was to systematically evaluate the performance of bleeding scores in patients with VTE focusing on high-risk patients.

METHODS

Longitudinal studies were searched in Medline and Cochrane Library, as well as reviews and references of retrieved articles. Studies were identified, data were extracted, and reporting quality was evaluated. We determined the sensitivity, specificity, positive likelihood ratio (LR+), and diagnostic odds ratio (DOR) of the 'high risk' category of each bleeding score. Random effects meta-analysis was performed in order to derive the central estimates and 95% confidence intervals (95% CI).

RESULTS

Twenty-one studies and ten bleeding scores fulfilled the inclusion criteria. VTE-BLEED showed the highest sensitivity but the second-lowest specificity (Se 76%; Sp 61%), followed by ACCP (Se 59%; Sp 57%). The remaining scores had high specificity (> 80%) but a low sensitivity (< 20%). HEMORRHAGES and Niewenhuis score showed the best performance regarding LR+ that was 2.67 and 5.91, respectively. Regarding DOR, the Niewenhuis score and VTE-BLEED were the best performers with 9.04; 95% CI 3.87-21.09 and 4.94 95% CI 2.66-9.09, respectively. In a cohort with patients predominantly treated with direct oral anticoagulants (DOACs), VTE-BLEED had the highest sensitivity (Se 77%; Sp 60%).

CONCLUSIONS

Overall, the majority of the risk scores showed a moderate ability to forecast major bleeding events, with the VTE-BLEED as the most sensitive in patients treated with DOACs.