Crit Care Med 2012 Vol. 40, No. 11 2933 N early 250,000 healthcare- associated infections occur annually in patients with central lines placed to deliver life-saving medical care (1). Furthermore, 25% of patients contracting a central line-associated bloodstream infection (CLABSI) in the intensive care unit (ICU) die, totaling 31,000 deaths annually in the United States (2). A recent review esti- mated an added annual cost of $9 billion to the U.S. healthcare system (3). Previous quality improvement studies suggest that these infections are largely preventable (4–8). However, these studies were based on nonrandomized trials with historical or contemporaneous controls. These designs might overestimate the effect of the intervention and may not be sufficient to establish a causal relationship between the interventions and the reduced infections (9, 10), especially when a consistent national decline in CLABSIs was found in all types of ICUs in the United States over the same period (11). The Keystone ICU collaborative in Michigan (12) used a bundle of evidence-based bloodstream infection prevention practices coupled with a program to improve patient safety, communication, and teamwork, known as the Comprehensive Unit–based Safety Program (CUSP) (13). Together these interventions reduced the overall CLABSI rate by 66% in a cohort of ICUs (7). Nevertheless, this cohort study, with no concurrent control group, was not able to establish a causal relationship between the intervention and the reduced CLABSI rate. The rational next scientific step was to test a causal relationship between this multifaceted intervention and reduced CLABSI rates in a randomized controlled trial (RCT) to evaluate the magnitude of the effectiveness of the intervention. This article reports our findings. METHODS Design and Setting. We used a multicenter, phased, cluster RCT to implement and test the multifaceted intervention designed to improve safety, safety climate, and the use of evidence- based practices to prevent bloodstream infec- tions. Two faith-based, affiliated health systems with hospitals in the West (Adventist Health) and in the Midwest and Southeast regions (Adventist Health System) of the country were Copyright © 2012 by the Society of Critical Care Medicine and Lippincott Williams and Wilkins DOI: 10.1097/CCM.0b013e31825fd4d8 Objectives: To determine the causal effects of an intervention proven effective in pre-post studies in reducing central line-asso- ciated bloodstream infections in the intensive care unit. Design: We conducted a multicenter, phased, cluster-random- ized controlled trial in which hospitals were randomized into two groups. The intervention group started in March 2007 and the con- trol group started in October 2007; the study period ended Septem- ber 2008. Baseline data for both groups are from 2006. Setting: Forty-five intensive care units from 35 hospitals in two Adventist healthcare systems. Interventions: A multifaceted intervention involving evidence- based practices to prevent central line-associated bloodstream infections and the Comprehensive Unit–based Safety Program to improve safety, teamwork, and communication. Measurements and Results: We measured central line-associ- ated bloodstream infections per 1,000 central line days and reported quarterly rates. Baseline average central line-associated blood- stream infections per 1,000 central line days was 4.48 and 2.71, for the intervention and control groups (p = .28), respectively. By October to December 2007, the infection rate declined to 1.33 in the interven- tion group compared to 2.16 in the control group (adjusted incidence rate ratio 0.19; p = .003; 95% confidence interval 0.06–0.57). The intervention group sustained rates <1/1,000 central line days at 19 months (an 81% reduction). The control group also reduced infection rates to <1/1,000 central line days (a 69% reduction) at 12 months. Conclusions: This study demonstrated a causal relationship between the multifaceted intervention and the reduced central line-associated bloodstream infections. Both groups decreased infection rates after implementation and sustained these results over time, replicating the results found in previous, pre-post stud- ies of this multifaceted intervention and providing further evidence that most central line-associated bloodstream infections are pre- ventable. (Crit Care Med 2012; 40:2933–2939) KEY WORDS: catheter-related infections; evidence-based prac- tice; intensive care units; prevention and control; quality improve- ment; randomized controlled trial A multicenter, phased, cluster-randomized controlled trial to reduce central line-associated bloodstream infections in intensive care units* Jill A. Marsteller, PhD, MPP; J. Bryan Sexton, PhD; Yea-Jen Hsu, PhD, MHA; Chun-Ju Hsiao, PhD, MHS; Christine G. Holzmueller, BLA; Peter J. Pronovost, MD, PhD, FCCM; David A. Thompson, DNSc, MS, RN Feature Articles *See also p. 3083. From the Department of Health Policy and Management (JAM, JBS, Y-JH, C-JH, PJP), Johns Hopkins Bloomberg School of Public Health, Baltimore, MD; and Department of Anesthesiology and Critical Care Medicine (JAM, JBS, CGH, PJP, DAT), Johns Hopkins School of Medicine, Baltimore, MD. The work was performed at Adventist Health and the Adventist Health System. Supported by a grant from the Robert Wood Johnson Foundation’s Interdisciplinary Nursing Quality Research Initiative. Dr. Sexton received honoraria from Michigan Hospital Association & Trinity. Dr. Pronovost is a consul- tant for Leigh Bureau Association for Professionals in Infection Control and Epidemiology. He receives grant support from AHRQ, NIH, RAND, Common Wealth Fund, and receives royalties from his book, Safe Patients, Smart Hospitals. The remaining authors have not dis- closed any potential conflicts of interest. For information regarding this article, E-mail: jmarstel@jhsph.edu