[Skip to Navigation]
Comment & Response
April 2015

For Deep Vein Thrombosis, Follow the Randomized Trials

Author Affiliations
  • 1Society of Interventional Radiology, North, Fairfax, Virginia
  • 2American Venous Forum, Milwaukee, Wisconsin
  • 3American College of Phlebology, Fairfield, Connecticut
JAMA Intern Med. 2015;175(4):653. doi:10.1001/jamainternmed.2014.8171

To the Editor We appreciate the efforts on the recent article by Bashir et al.1 However, we are concerned that the potential for bias in this nonrandomized study has been underestimated. Some associations drawn between catheter-directed thrombolysis (CDT) and adverse outcomes may not be attributable to CDT.

First, propensity analysis cannot eliminate bias from retrospective observational comparison of a first-line treatment (anticoagulation) against a treatment reserved for selected patients with the most severe clinical presentations (CDT). Catheter-directed thrombolysis is often used as salvage therapy after physician-perceived failure of first-line anticoagulant therapy. Many factors that prompt a clinician to request CDT connote a higher risk of pulmonary embolism (PE), filter placement, and longer hospital stay and were not accounted for (eg, increase in pain and/or swelling despite anticoagulation, anatomic thrombus progression, or acute limb-threatening circulatory compromise). The administrative data, which rely entirely on correct coding for its accuracy, does not distinguish whether PE events occurred prior to, during, or after CDT, or whether “CDT” cohort patients truly received CDT for deep vein thrombosis (DVT) rather than systemic thrombolysis or CDT for other conditions. Hence, the observed increases in PE, filter placement, and hospital stay may simply represent native differences between the 2 cohorts rather than consequences of CDT.

Add or change institution