Which central venous catheters have the highest rate of catheter-associated deep venous thrombosis: a prospective analysis of 2,128 catheter days in the surgical intensive care unit.

Malinoski, Darren; Ewing, Tyler; Bhakta, Akash; Schutz, Randi; Imayanagita, Bryan; Casas, Tamara; Woo, Noah; Margulies, Daniel; Barrios, Cristobal; Lekawa, Michael; Chung, Rex; Bukur, Marko; Kong, Allen

BACKGROUND Catheter-associated deep venous thromboses (CADVTs) are a common occurrence in the surgical intensive care unit (SICU), necessitating central venous catheter (CVC) removal and replacement. Previous studies evaluating risk factors for CADVT in SICU patients are limited, and most lack a true denominator of all CVC days. We sought to determine the true incidence of and risk factors for CADVT based on patient characteristics as well as CVC site, type, and duration of insertion. METHODS The following data from all SICU patients in two urban Level I trauma centers were prospectively collected from 2009 to 2012: demographics, risk factors for DVT, CVC site/type/duration, and duplex results. Sites included the subclavian (SC), internal jugular (IJ), arm (for peripherally inserted central catheter [PICC] lines), and femoral. Types included multilumen (ML), introducer/hemodialysis (I/HD), and PICC. High-risk patients received weekly screening duplex examinations and a CADVT was defined as a DVT being detected on duplex with a CVC in place or within 7 days of removal. Rates of CADVT were normalized per 1,000 CVC days, and independent predictors of CADVT were determined using logistic regression. RESULTS Data were complete for 184 patients, 354 CVCs, and 2,128 CVC days. Fifty-nine CADVTs were diagnosed in 28% of patients. Rates of CADVT were 9 per 1,000 catheter days for SC, 61 for IJ (p < 0.01 vs. SC), 27 for arm (p < 0.01), 36 for femoral (p < 0.01), 22 for ML, 57 for I/HD (p < 0.01 vs. ML), and 27 for PICC (p = 0.24). After adjusting for patient risk factors, predictors of CADVT included the IJ and arm sites (odds ratio, 6.0 and 3.0 compared with SC) and the I/HD type (odds ratio, 2.6 compared with ML, all p < 0.05). CONCLUSION The IJ and arm sites and I/HD type are associated with increased CADVT. These data may be used to determine the optimal site and type of CVC for insertion. LEVEL OF EVIDENCE Prognostic/epidemiologic study, level III.