BackgroundEvidence supporting use of an inferior vena cava filter (VCF) to prevent death or recurrent venous thromboembolism (rVTE) in cancer patients who are hospitalized for acute VTE is limited. AimsTo determine the effectiveness of VCF placement on the 15-day and 30-day incidence of death and the 180-day incidence of rVTE manifested as pulmonary embolism (PE) or recurrent deep-vein thrombosis (DVT) alone among cancer patients hospitalized for acute-VTE. MethodsUsing a large retrospective observational study of discharge records in California, we analyzed outcomes after VCF placement in cancer patients hospitalized 2005-2009 for acute VTE using propensity-score methodology. We excluded all patients who had a history of a prior VCF placement (1991-2009). Outcomes were death <15-days and <30-days and rVTE (as PE or DVT alone) at 6 months. We used 3 analytic methods: 1) standard risk-adjusted multivariable analysis, 2) adjustment using propensity-score and inverse probability weighing (IPW) and 3) comparison based on matching (2:1) based on propensity score (caliper method). The multivariate model used to generate the propensity score included age, race/ethnicity, insurance coverage, expected bleeding risk, metastatic disease, bleeding present-on-admission, location of bleeding, recent or impending major surgery, use of thrombolytic agents, number of chronic co-morbidities, severity-of-illness (ascertained by 3M, APR-DRG grouper), index PE vs. DVT, and hospital characteristics. IPW of propensity score was applied to a risk-adjusted logistic model to predict death; IPW was applied to risk-adjusted Cox models predicting rVTE (as PE or DVT alone). In the model predicting death, risk-of-mortality on admission was used instead of severity-of-illness. ResultsAmong 14,000 cancer-associated acute-VTE cases, the overall crude 15-day mortality rate was 1396 (10%) and the 30-day mortality was 2247 (16.1%). For 11,253 no-VCF patients, the crude 15-day mortality was 1089 (9.7%) and at 30 days it was 1727 (15.3%). A VCF was placed in 2747 patients (19.6%). The crude mortality in VCF patients was 307 (11.2%) at 15-days, and 520 (18.9%) at 30-days. After accounting for propensity to insert a VCF (using IPW) in a risk-adjusted model, there was no significant reduction in the risk of death associated with VCF use at 15-days (OR=0.90, CI:0.8-1.1, p=0.26) or 30 days (OR=1.04, 95%CI:0.9-1.2, p=0.57); findings were the same using standard multivariable analysis and matching based on propensity score. The crude 180-day incidence of recurrent PE was 3.3%: 2.6% in VCF patients and 3.4% in the no-VCF patients. In the adjusted model using IPW the risk of rVTE manifested as PE, the risk was lower in VCF patients (HR=0.81 95%CI:0.6-1.1, p=0.14) but this did not reach statistical significance. The crude 180-day incidence of rVTE manifested as DVT alone was 4.2% overall: 5.4% in VCF patients and 3.9% in no-VCF patients. In the IPW propensity score model, the risk of rVTE manifested as DVT at 180 days was significantly higher in VCF patients (HR=1.55, 95%CI:1.3-1.9, p<0.001). Models for recurrent VTE manifested by PE or DVT gave similar results whether based on propensity-score matching or multivariable analysis. ConclusionsUse of a VCF in cancer patients hospitalized specifically for acute VTE was not associated with a significant reduction in the risk of death at 15 or 30 days, and the overall 30-day mortality was high. There was a 20% reduction in the risk of rVTE manifested as PE at 180-days but this did not reach the level of statistical significance (p>0.05). VCF use was associated with a 55% higher risk of rVTE manifested as DVT at 180 days. Further refinements in modeling incorporating competing outcomes (e.g., death) are underway. Disclosures:Ho:American Society of Hematology: ASH HONORS trainee research award Other.