Content area
Full Text
Introduction
The evidence for the use of real-time ultrasound guidance in the placement of central venous catheters (CVC) has been so convincing that the Agency for Healthcare Research and Quality has listed it as one of 11 patient safety-related practices to reduce morbidity and mortality. 1 Three systematic reviews of ultrasound-guided CVC placement by Randolph et al , 2 Hind et al 3 and Keenan 4 were limited to intensive care unit patients. None of these reviews 2-4 included studies of the efficacy of ultrasound-guided CVC placement in an Emergency Department (ED) setting. We systematically reviewed the literature comparing the success and complication rates between ultrasound- and landmark-guided CVC placement by ED physicians.
Methods
Search strategy
We utilised a (PICO) format to ask our clinical question: Is real time ultrasound-guided CVC placement more successful than a traditional landmark approach in the ED setting?
P atients: Our study included ED patients over the age of 18 years requiring CVC placement for any reason deemed necessary by the ED physician, with the exception of placement for the purposes of cardiopulmonary resuscitation. I ntervention: Central line placement using real-time ultrasound guidance. C omparator: Central line placement by traditional anatomical landmarks. O utcome: Comparison of success rates of CVC placement between ultrasound- versus landmark-guided techniques.
Outcome measures and data abstraction
Our primary outcome was the successful CVC cannulation. Secondary outcome included complications of CVC placement. Our target study design limited our review to randomised trials where participants underwent either ultrasound- or landmark-guided CVC placement.
Primary data analysis
CVC cannulation success rates were compared using a Forest Plot of RR (95% CI) between ultrasound- and landmark-guided techniques using Review Manager (RevMan) (computer program), Version 5.0 (Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration, 2008). Pooled treatment effects were estimated using RR for CVC success rates after intervention/comparator, with Mantel-Haenszel RR, using a random-effects model. Heterogeneity was assessed by χ2 and the I2 statistic. Manuscripts were investigated for potential bias by the following criteria: randomisation, concealment, blinding, intention-to-treat and completeness of follow-up.
Data collection and processing
MEDLINE with the PubMed interface was searched for articles from 1965 through November 2010 and EMBASE from 1980 to November 2010 with the Ovid Technologies interface (See online appendix A for complete MEDLINE...