Practice patterns and complication rates of thrombolysis for pulmonary embolism.

Practice patterns, bleeding complication rates, and outcomes of thrombolytic use for pulmonary embolism (PE) outside of clinical trials are not well characterized. Retrospective cohort study of patients with PE, using an enhanced administrative database of approximately 20 % of hospitalizations in the United States, 2008-2011. We used hierarchical logistic regression to identify patient- and hospital-level factors associated with thrombolytic use and quantified between-hospital variation in thrombolysis. We calculated rates of hemorrhagic complications and mortality in patients receiving thrombolytics stratified by vasopressor use. We identified 91,860 hospitalizations for PE; thrombolytics were utilized in 13.2 % (578/4363) who received vasopressors and 1.6 % (1320/82,997) who did not receive vasopressors. Patients who received thrombolytics were younger, more likely white, with private insurance, fewer comorbidities and more acute organ failures. Hospitals varied widely in thrombolysis rates for PE (0-100 % of patients on vasopressors, 0-12.5 % not on vasopressors) with 17 % of variation unexplained by patient or hospital characteristics. Bleeding complication rates for patients receiving thrombolytics versus no thrombolytics, respectively, were: ICH (0.8 vs. 0.08 %, p < 0.01), major bleeding (13 vs. 3 %, p < 0.01), and major bleeding with transfusion (2.5 vs. 0.05 %, p < 0.01). Among patients given vasopressors, thrombolytics were not associated with hospital mortality (thrombolysis 41 % vs. no thrombolysis 35 % mortality; adjusted OR 0.97, 95 % CI 0.93-1.01, p = 0.15). There is wide variation in thrombolytic use for PE in the US. Patient characteristics and complications associated with real-world thrombolytic use were similar to published randomized trials.