Efficacy of Inferior Vena Cava Filters in Anticoagulated Patients.

Although use of inferior vena cava (IVC) filters has grown enormously over the last 10 years, data examining the benefit over standard anticoagulation (AC) regimens are limited. To assess whether the addition of IVC filters improve outcomes as compared to AC therapy alone, we retrospectively examined prospectively-collected data in the database of a large urban medical center for all patients (n=1726) who were discharged between October 1997 and September 2005 after an episode of venous thromboembolism (VTE) who had documented adequate AC (INR of 2.0) within the first 30 days. Those who received an IVC filter (n=303) in addition to AC were compared to those (n=1423) who received AC only. The last serum albumen level prior to discharge was used as a surrogate marker for morbidity. Outcome measures were new pulmonary embolism (PE) or mortality as well as recurrent deep vein thrombosis (DVT). Subsequent morbid events were assessed from the hospital record system and mortality was also assessed by hospital record or social security record search. Potential confounding variables sex, race, and cancer status at discharge were compared by chi-square; mean age and albumin were compared with independent samples t-test. Multivariable adjusted hazard ratios (HR) were estimated with Cox models. Analyses were performed with SPSS using a two-tailed alpha of .05.

Total incidence of subsequent PE, DVT and mortality were 6.5%, 36.6% and 35.3% respectively. The two groups did not significantly differ by sex or cancer status. Time in therapeutic range (TTR) and duration of AC was similar for both groups (TTR for Filter v AC-only: 52.3% v 54.3%. Patients with both filter and AC were less likely to be Hispanic, had lower mean (± sd) albumin (3.0 ± 0.9 mg/dl v 3.6 ± 0.8 mg/dl, p<.001)) and were older (65 ± 16 yrs v 60 ± 18yrs, p<.001) than those with AC only. Hazard ratios did not differ significantly for new PE (p=.8) or recurrent DVT (p=.8) when adjusted for albumin, cancer status, age, sex and race, however mortality was significantly worse for patients with a filter (HR 1.3; 95% CI 1.1,1.5, p=.009). Since patients with massive pulmonary emboli and those who are unstable may be more likely to receive an IVC filter in addition to AC, we addressed the potential indication bias by an additional analysis after exclusion of those who had a recurrent VTE (n=1) or who died (n=47) prior to discharge from the index hospitalization. This analysis of the remaining 1678 patients did not demonstrate a significant difference in difference in mortality (p=.8), new PE (p=.8) or recurrent DVT (p=.7).

These data suggest the possibility that IVC filters may not provide any substantial additional advantage for patients who can tolerate anticoagulant therapy. A randomized controlled trial will be needed to be confident that indication bias or latent confounding variables do not obscure real differences and to better define whether there is meaningful benefit to be gained from IVC filters for VTE patients already on anticoagulation.