Relationship of Major Bleeding to Pharmacologic Venous Thromboembolism Prophylaxis in Hospitalized Medical Patients

Background: Guidelines recommend pharmacologic venous thromboembolism (VTE) prophylaxis for high risk hospitalized medical patients that do not have an elevated bleeding risk.  A systematic review of randomized trials estimates the overall bleeding risk with pharmacologic prophylaxis to be low, at 1/1000 (0.1%).  However, it is unclear if the bleeding events were related to pharmacologic prophylaxis versus patients’ underlying medical conditions. We sought to determine the relationship between pharmacologic prophylaxis and major bleeding.

Methods: The Michigan Hospital Medicine Safety Consortium is a payer-funded quality collaborative of 47 hospitals with a goal of reducing adverse events in hospitalized medical patients. Trained abstractors collected data regarding VTE risk factors, type of VTE prophylaxis used, and VTE and major bleeding (International Society of Thrombosis and Hemostasis defined) events at 90 days follow up. In 2015, a panel of eight hospitalists adjudicated major bleeding events from the last quarter of 2014 and the first quarter of 2015 and classified them as unrelated, possibly, probably, and definitely related to pharmacologic prophylaxis using a modified Delphi process. The panel reviewed standardized abstracted patient-level demographic, clinical and laboratory information and indicated their initial decision, followed by discussion and then final decision by consensus. Access to the full database, which included abstractors’ notes, was available during review.

Results: There were 38 (0.5%) major bleeding events in 7834 cases of patients receiving pharmacologic prophylaxis during the 2 quarters and 35 cases had sufficient information for review.  Consensus was achieved in all 35 cases and there was unanimous agreement in 83% (95% CI, 70-95%). The number (proportion [95% CI]) of cases of major bleeding deemed to be unrelated, possibly related, probably related and definitely related to pharmacologic prophylaxis were 12 (34% [18-50%]) , 15 (43% [27-59%]), 8 (23% [9-37%]) and 0 cases respectively.

Conclusions: Major bleeding events were low and consistent with clinical trials in a large real world cohort of hospitalized medical patients and a significant proportion of these events may not be related to the receipt of prophylactic anticoagulants.