Central line–associated bloodstream infection (CLABSI) rates per 1000 line-days per quarter (Q) from January 2001 to June 2009. Corresponding interventions are described in Table 1 for the quarters with numbers above the bars. For trend of CLABSI rates, total χ2 test, P < .001; χ2 test for linear trend, P < .001; remaining χ2 test, P = .70. Q1 indicates January through March; Q2, April through June; Q3, July through September; and Q4, October through December.
Breakdown of catheter insertion sites from 2005 through 2008. CVC indicates central venous catheter; FEM, femoral CVC; IJ, internal jugular CVC; PICC, peripherally inserted CVC; and SCL, subclavian CVC. For the proportion of PICC-days to total CVC-days: total χ2 test, P < .001; χ2 test for linear trend, P < .001; and remaining χ2 test, P < .001.
Ong A, Dysert K, Herbert C, Laux L, Granato J, Crawford J, Rodriguez A, Cortes V. Trends in Central Line–Associated Bloodstream Infections in a Trauma-Surgical Intensive Care Unit. Arch Surg. 2011;146(3):302-307. doi:10.1001/archsurg.2011.9
To report the impact of hospital-wide interventions on central line–associated bloodstream infection (CLABSI) rates in a 24-bed trauma-surgical intensive care unit.
Data were gathered retrospectively from January 1, 2001, through June 30, 2009. Interventions to reduce CLABSI rates during this period included standardization of line insertion and maintenance processes, development of a mandatory education program incorporating practical line insertion simulation sessions, frequent audits, and intensive care unit staffing modifications. We used the χ2 test and analysis of variance to analyze the data where appropriate.
Urban tertiary referral center providing level I trauma services.
Eight thousand four hundred eighty-one trauma-surgical intensive care unit admissions, of which 76% were owing to trauma.
During this period, the incidence of CLABSI declined from 6.1 to 0.3 per 1000 line-days. No CLABSIs occurred for 8 of the last 10 quarters (January 2007 to June 2009). Internal jugular sites were associated with a higher CLABSI rate than subclavian sites (P = .03). The central line utilization ratio remained high for most of the study period. When compared with the 2006-2007 Centers for Disease Control and Prevention data, the trauma-surgical intensive care unit was at the 10th percentile in CLABSIs and at the 75th to 90th percentile in central line utilization ratios.
The significant decline in the incidence of CLABSIs, which reflected the national trend, could be attributed to multiple interventions. The high central line utilization ratio compared with nationally available data represents a potential target for further improvement.
Health care–associated infections (HAIs) constitute a large economic and social burden. In 2002, an estimated 1.7 million HAIs occurred in the United States.1 Of these, an estimated 92 000 were central line–associated bloodstream infections (CLABSIs).2 In a study of patients in medical and surgical intensive care units (ICUs), an episode of CLABSI had an attributable cost of $11 971, an ICU length of stay of 2.4 days, and a hospital length of stay of 7.5 days when controlled for confounding factors.3 These findings of an excess economic burden have been duplicated in other studies.4- 6
In 2000, the Pittsburgh Regional Healthcare Initiative, an independent regional consortium of business, civic, and medical leaders, convened a committee of infection control experts to implement strategies to reduce HAIs in southwestern Pennsylvania. Allegheny General Hospital was 1 of several participating hospitals involved in a collaborative effort targeting CLABSIs. This effort resulted in multifaceted interventions and the eventual reduction of CLABSIs across participating hospitals.7 The aim of this study was to describe the impact of hospital-wide interventions on CLABSI rates in a trauma-surgical ICU (TSICU).
The TSICU is an open 24-bed ICU situated in an urban tertiary referral center providing level I trauma services. Details of all CLABSIs in the study period were obtained from prospectively collected surveillance data by the hospital Infection Control Department from January 1, 2001, through June 30, 2009. The definition of CLABSI used was the Centers for Disease Control and Prevention (CDC) definition of laboratory-confirmed bloodstream infection.8
We compared our data with nationally collected data from the National Nosocomial Infections Surveillance System (NNIS) and the National Healthcare Safety Network (NHSN). The NHSN, established in 2005, integrated and superseded the following 3 surveillance systems at the CDC: the NNIS, the Dialysis Surveillance Network, and the National Surveillance System for Healthcare Workers. Similar to the NNIS, NHSN facilities voluntarily reported their HAI surveillance data for aggregation into a single national database.9
The CLABSI rate was expressed as the number of CLABSIs per 1000 central line–days (hereinafter referred to as line-days). The central line utilization ratio was the number of line-days divided by the number of patient-days for a particular period.
Trauma patients underwent separate analysis. Injury Severity Scores and probability of survival data based on the TRISS (Trauma and Injury Severity Score) method10 were collected, allowing generation of expected mortality rates.
We used the χ2, Fisher exact, and Kruskal-Wallis tests to analyze the data where appropriate. The χ2 test for linear trend was used for analyzing trends among proportions, and the Cuzick test was used for analyzing trends among independent ordinal samples. The significance level was set at P = .05. Statistical analysis was performed using a commercially available software program (StatsDirect; StatsDirect Ltd, Cheshire, England).
Table 1 shows several of the interventions as a timeline. Initial hospital-wide efforts started in the medical ICU and coronary care unit to promote best practices in line care, insertion, and maintenance according to published CDC guidelines.11 These included full-barrier precautions during line insertion, the use of a combined 2% chlorhexidine gluconate and 70% isopropyl alcohol preparation for the insertion site, hand hygiene before line insertion, avoidance of femoral sites, and avoidance of central venous catheter (CVC) changes over guide wire. Emergently placed lines or those acquired from a referring hospital were required to be discontinued within 24 hours of arrival. The hospital ICU committee subsequently consolidated and established hospital-wide standards and disseminated these through a 4-page booklet issued in 2004 to all house staff.
A multidisciplinary central line subcommittee consisting of staff from the Infection Control Department (C.H. and J.C.), the ICUs (A.O., K.O., J.G., and V.C.), the intravenous catheter team, and the departments of Oncology, Nephrology, and Radiology and a critical care advanced practice nurse (L.L.) was formed. This group worked with vendors to create standardized line insertion kits and dressing change kits. Bundles were developed to provide all supplies to improve efficiency and to eliminate variation in practice. Policies on the care and maintenance of intravenous lines were reviewed and revised. Documentation was improved by using medical record stickers with a standardized format that could easily be filled out by house staff and by posting “line sheets” at each ICU bed, where the ICU nurse documented the type of catheter and the location and date of insertion. This facilitated discussion of each patient's need for continued bloodstream device use during daily ICU rounds.
In 2005, the coronary care unit director (J.G.) and nurse educator developed a Web-based mandatory educational module for residents and nurses. This was combined with hands-on training for house staff achieved through a 30-minute proctored mannequin session. Familiarity with standardized practices was also assessed through an online examination, with a minimum passing score of 80%. This educational program was disseminated hospital wide and was made mandatory for all residents and fellows in medicine, surgery, emergency medicine, radiology, and anesthesia annually regardless of their level of training. All house staff participating for the first time had to demonstrate competency with a 1:1 proctored mannequin session. House staff repeating the program had to participate in a mannequin “refresher” session that was proctored with a 5:1 ratio. All nurses completed an online educational module annually.
In August 2004, the practice of placing a chlorhexidine-impregnated sponge (Biopatch; Ethicon, Inc, Somerville, New Jersey) at the insertion site was started for peripherally inserted central venous catheters (PICCs) and hemodialysis catheters. From June 2005, all newly placed CVCs were required to have the sponges placed at the catheter insertion site.
Throughout the study period, the TSICU used CVCs impregnated with benzalkonium chloride on internal and external surfaces (Multi-Med CVC and Intro-Flex; Edwards Lifesciences, Irvine, California) or CVCs impregnated with a silver-platinum-carbon amalgam and benzalkonium chloride on both surfaces (Vantex antimicrobial catheter; Edwards Lifesciences).
In March 2004, a critical care advanced practice nurse position was created and filled. This nurse participated in the central line subcommittee and coordinated and performed hospital-wide audits of best practices in ICUs. This practice was continued regularly on a unit level, performed by the nursing director, nurse educator, and nurse supervisors of the TSICU.
In 2004, the Department of Nursing and the ICU committee recognized that one of the barriers to optimizing implementation and supervision of best practices was that ICU charge nurses were frequently assigned direct patient care responsibilities. To allow charge nurses to take on more supervisory responsibilities, including mentoring of less experienced nurses, the Department of Nursing increased efforts to hire and retain nurses in the TSICU.
From 2001 to 2008, there were 8481 TSICU admissions. Trauma patients accounted for 76% of the 8481 admissions and 74% of the ICU bed-days. As with total bed-days in the TSICU, there was a significant trend toward increased use of beds. Although Injury Severity Scores remained unchanged for the study period, there was a statistically significant increasing trend in patient age and a decreasing trend in ICU lengths of stay (Table 2). Mortality rates for trauma patients declined and compared favorably with the predicted rates based on the TRISS method.
The incidence of CLABSI declined from 6.1 per 1000 line-days in 2001 to 0 per 1000 line-days in 2007 and 0.3 per 1000 line days in 2008 (Figure 1). There were no CLABSIs for 8 of the last 10 quarters (January 2007 to June 2009). The central line utilization ratio remained high for most of the study period. When compared with the 2006-2007 CDC/NHSN data,9 the TSICU was at the 10th percentile mark in CLABSI rates and at the 75th to 90th percentile in central line utilization ratios.
Data regarding the line-days for each insertion site were available from 2005 onward, allowing further analysis of CLABSI rates by insertion site (Figure 2). Of 23 473 line-days, PICCs accounted for 53%, subclavian CVCs for 28%, internal jugular CVCs for 13%, and femoral CVCs for 6%. The PICCs accounted for an increasing proportion of line-days compared with the other sites from 2005 to 2008. The following CLABSI rates were found: PICCs, 11 per 12 372; subclavian CVCs, 2 per 6773; internal jugular CVCs, 5 per 2946; and femoral CVCs, 1 per 1382. There were no differences in CLABSI rates by insertion site except when we compared internal jugular CVCs with subclavian CVCs; the internal jugular site was associated with a higher CLABSI rate (Fisher exact test, P = .03).
For the CVCs associated with CLABSI, the mean (SD) duration before CVC discontinuation was 8.96 (9.30) (median, 6) days. Most of the responsible organisms were gram-positive bacteria (71.2%), with the rest divided almost equally between fungal organisms (14.9%) and gram-negative rods (13.8%). Of all organisms, those most commonly identified were Staphylococcus epidermidis (46.8%), methicillin-sensitive Staphylococcus aureus (11.7%), Candida albicans (8.5%), and methicillin-resistant S aureus (5.3%).
The following data were available during the last 3 complete study years (2006-2008) on the number of hours TSICU charge nurses were assigned to ICU patients: 1720 hours in 2006, 1485 hours in 2007, and 614 hours in 2008. With a denominator of 8760 hours (365 days × 24 hours), the number of hours charge nurses were assigned to ICU patients significantly declined during those 3 years (χ2 test for linear trend; total χ2, P = .001; χ2 for linear trend, P < .001).
Consumer demand for information about the performance of health care providers has risen during the past decade. Since 2002, 4 states (Illinois, Pennsylvania, Missouri, and Florida) have enacted legislation mandating public reporting of HAIs by health care organizations.12 Central line–associated bloodstream infections constitute a major group of HAIs, and, because they are associated with increased hospital costs and morbidity, prevention of CLABSIs should be an important priority for all health care organizations.
Several institutions had reported similar improvements in CLABSI rates with coordinated interventions. Berenholtz et al13 designed a prospective cohort study involving a study ICU and a control ICU. Interventions in the study ICU included standardization of practices, a Web-based training module, and the creation of a line insertion cart. The nurses were also empowered to stop procedures when guidelines were not followed. The control ICU saw education to increase provider awareness as the only intervention. Rates of CLABSIs declined in both ICUs, with the rate of decline significant in the study ICU.
Pronovost et al14 reported on interventions in the Keystone ICU project involving 103 Michigan ICUs and showed a decline in median CLABSI rates of 2.7 per 1000 line-days to a median of 0 per 1000 line-days; the trend was sustained during 18 months of follow-up. Interventions in their study were similar to those previously described. Similarly, in another study, an intervention consisting of a structured educational program for ICU nurses resulted in a significant reduction in CLABSI rates in a surgical ICU.15 These results and ours demonstrate that active educational interventions are a cornerstone of CLABSI eradication.
The decline in CLABSI rates in the TSICU during the past 8 years reflected a general decline in CLABSI rates in the United States since 1990.16 This was further supported by the following NNIS and NHSN data: from 2002 to 2004, of the 22 reporting trauma ICUs, the CLABSI rates were 5.2 per 1000 line-days at the 50th percentile and 1.9 per 1000 at the 10th percentile; from 2006 to 2007, the corresponding rates were 4.0 per 1000 line-days and 0.3 per 1000 for 32 reporting trauma ICUs.9,17 The national data and ours are encouraging, given that trauma ICUs traditionally have had higher incidences of CLABSI when compared with other ICUs.9 We also found that our mortality rates and lengths of stay decreased for trauma patients. However, we recognize that there were other possible contributing factors and that this study did not intend to establish an association between CLABSI rates and mortality or length of stay.
In this study, changes in ICU staffing could have influenced CLABSI rates. There was a declining trend in the number of hours that charge nurses were assigned to direct patient care, which likely resulted in more supervision and mentorship of less experienced nurses. That nurse staffing might play a role in CLABSI infections is seen in a prospective study in which there was an increased risk (odds ratio, 2.6) of CLABSI for patients cared for by float nurses for more than 60% of the time vs an assigned ICU nurse.18
The PICCs accounted for an increasing proportion of central lines placed in the TSICU. A dedicated PICC team inserted the catheters, almost always with ultrasonographic guidance at the bedside. However, it is unlikely that the increasing use of PICCs influenced CLABSI rates in the TSICU because PICC CLABSI rates were similar to those of other CVCs. Our findings were consistent with other studies reporting equivalent rates of bloodstream infection between PICCs and CVCs.19,20 The only randomized controlled trial of PICCs vs other CVCs in adult patients showed similar rates of CLABSI.21
Another factor that might have contributed to the decline in CLABSI rates was the hospital-wide use of chlorhexidine-impregnated sponges on insertion sites. Despite a meta-analysis showing no benefit by using these sponges,22 a recent multicenter randomized trial demonstrated a statistically significant reduction in CLABSI rates (1.3 vs 0.4 per 1000 catheter-days) when the group using the antimicrobial sponges was compared with the control group.23
The use of anti-infective–impregnated catheters is controversial. A systematic review24 showed that anti-infective–impregnated CVCs in general were effective in decreasing CLABSIs when compared with standard CVCs. However, there was significant heterogeneity in the reviewed studies, and when subgroups were analyzed, anti-infective–impregnated catheters with internally and externally treated surfaces and antibiotic-impregnated catheters showed efficacy in reducing CLABSI rates, whereas anti-infective–impregnated catheters with externally treated surfaces did not show a treatment effect. During the present study, the TSICU used CVCs treated with anti-infective compounds on both surfaces. Despite this, CLABSI rates did not drop until 2006. It appears, then, that measures other than use of anti-infective–treated CVCs were responsible for the decline in CLABSI rates.
It was also apparent that the TSICU had a high central line utilization ratio when we compared that ratio with the CDC/NHSN data.9 It was difficult to draw conclusions about the overuse of CVCs, since there was no accurate risk adjustment for illness severity for the total population. However, the high central line utilization ratio in our ICU represents a potential target for future efforts directed at CLABSI reduction.
Several study limitations exist. We did not have measures of illness severity or organ dysfunction scores because there was no prospective collection of these data during the study period for nontrauma patients. Hence, accurate risk adjustment could not be performed for the entire study cohort. Other factors that may have influenced CLABSI rates were not taken into account, such as improved glycemic control through wider use of insulin intravenous protocols. Whether there was more attending physician supervision of line placement by residents was also unclear. It was also difficult to measure the magnitude of impact of any given intervention in reducing CLABSI rates because several of the interventions were adopted simultaneously.
Based on our experience, we suggest several steps a similar institution can take to reduce CLABSIs:
Adopt evidence-based line insertion and maintenance practices and standardize line insertion–related materials and processes.
Establish a multidisciplinary team approach to problem solving.
Conduct annual mandatory education through a formal program for house staff and nurses with certification of proficiency through examinations and/or practical simulation sessions.
Have frequent audits of line insertion and maintenance by ICU supervisors.
Have real-time (weekly) review of infections and audit results within the ICU.
Allow ICU nurses with supervisory roles time to mentor, train, and supervise ICU personnel.
The significant decline in the incidence of CLABSI in a TSICU, which was reflective of a national trend, could be attributed to multiple interventions, including the creation of a multidisciplinary team, standardization of materials and processes relating to line care and insertion, a formal mandatory educational program for house staff, creation of an advanced critical care nurse position, and modifications of nursing ICU roles and responsibilities. The high central line utilization ratio compared with nationally available data represents a potential target for further improvement.
Correspondence: Adrian Ong, MD, Department of Surgery, Allegheny General Hospital, 320 E North Ave, Pittsburgh, PA 15212 (firstname.lastname@example.org).
Accepted for Publication: February 6, 2010.
Author Contributions:Study concept and design: Ong, Laux, Granato, and Rodriguez. Acquisition of data: Ong, Dysert, Herbert, Laux, and Crawford. Analysis and interpretation of data: Ong, Laux, and Cortes. Drafting of the manuscript: Ong, Herbert, and Crawford. Critical revision of the manuscript for important intellectual content: Ong, Dysert, Laux, Granato, Rodriguez, and Cortes. Statistical analysis: Ong. Administrative, technical, and material support: Dysert, Laux, Granato, Crawford, and Rodriguez.
Financial Disclosure: None reported.
Previous Presentation: This study was presented as a poster at the 67th Annual Meeting of the American Association for the Surgery of Trauma; September 25, 2008; Maui, Hawaii.