Incidence and Risk Factors for In-Hospital Bleeding in Acutely Ill Medical Patients: Findings from IMPROVE.

Background We examined data from acutely ill medical patients enrolled in The International Medical Prevention Registry on Venous Thromboembolism (IMPROVE) to determine factors that are independently associated with an increased risk of bleeding during hospitalization.

Methods Patients ≥18 years old, hospitalized for ≥3 days with an acute medical illness have been enrolled consecutively since July 2002. Exclusion criteria are: therapeutic antithrombotics/thrombolytics at admission; major surgery or trauma during 3 months prior to admission; and venous thromboembolism (VTE) treatment within 24 hours of admission. Patients with bleeding immediately prior to, or at admission were excluded from this analysis. Factors considered were: age, ICU stay, reduced creatinine clearance, severe infection, cerebrovascular stroke, cancer, diabetes, severe renal failure, hypertension, lower limb paralysis, bleeding disorders, hemorrhagic stroke, thrombocytopenia, gastro-duodenal ulcer, hepatic failure, central venous catheter at admission, hormonal therapy for cancer, platelet count, BMI, immobility, and length of hospital stay. Factors increasing the risk of bleeding were identified by univariate analysis (P≤0.20) and included in a multiple logistic regression model. Factors with significance of P≤0.05 were retained in the model. Bleeding events were defined as major or clinically significant non-major according to published criteria (

Results. Data were from 2816 patients enrolled up to 30 June 2004 in 34 hospitals in 10 countries. Patients were: 48% female, mean age 64 years, mean weight 72 kg, mean length of hospital stay 12 days, and 37% were immobile for ≥3 days (median duration of immobility 6 days, including immobility immediately prior to admission). Only 89 (3.2%) patients had in-hospital bleeding: 1.2% major, 1.9% clinically significant non-major, 0.1% unspecified. Factors that were independently associated with an increased risk of bleeding (major or non-major) in acutely ill medical patients are shown in Table 1.

Conclusion Only 3.2% of acutely ill medical patients in IMPROVE had in-hospital bleeding. Major bleeding (1.2%) was similar to that observed in a major clinical trial on VTE prophylaxis, MEDENOX (1.0%;

Table 1. Factors independently associated with an increased risk of bleeding in acutely ill medical patients