JAN

REVIEW PAPER

Catheter-related bloodstream infections in intensive care units: a systematic review with meta-analysis
Prabha Ramritu, Kate Halton, David Cook, Michael Whitby & Nicholas Graves
Accepted for publication 28 August 2007

Correspondence to N. Graves: e-mail: n.graves@qut.edu.au Prabha Ramritu MN RN Project Ofcer The Centre for Healthcare Related Infection Surveillance & Prevention, Princess Alexandra Hospital, Brisbane, Qld, Australia Kate Halton MSc Project Ofcer The Centre for Healthcare Related Infection Surveillance & Prevention, Princess Alexandra Hospital, Brisbane, Qld, Australia Institute of Biomedical and Health Innovation, Brisbane, Qld, Australia David Cook MBBS PhD Medical Ofcer Intensive Care Unit, Princess Alexandra Hospital, Brisbane, Qld, Australia Michael Whitby FRACP MPH Director of Infectious Diseases The Centre for Healthcare Related Infection Surveillance & Prevention, Princess Alexandra Hospital, Brisbane, Qld, Australia Nicholas Graves PhD Senior Research Fellow in Health Economics The Centre for Healthcare Related Infection Surveillance & Prevention, Princess Alexandra Hospital, Brisbane, Qld, Australia Institute of Biomedical and Health Innovation, Brisbane, Qld, Australia

R A M R I T U P . , H A L T O N K . , C O O K D . , W H I T B Y M . & G R A V E S N . ( 2 0 0 8 ) Catheterrelated bloodstream infections in intensive care units: a systematic review with meta-analysis. Journal of Advanced Nursing 62(1), 321 doi: 10.1111/j.1365-2648.2007.04564.x

Abstract
Title. Catheter-related bloodstream infections in intensive care units: a systematic review with meta-analysis. Aim. This paper is a report of a systematic review and meta-analysis of strategies, other than antimicrobial coated catheters, hypothesized to reduce risk of catheterrelated bloodstream infections and catheter colonization in the intensive care unit setting. Background. Catheter-related bloodstream infections occur at a rate of 5 per 1000 catheter days in the intensive care unit setting and cause substantial mortality and excess cost. Reducing risk of catheter-related bloodstream infections among intensive care unit patients will save costs, reduce length of stay, and improve outcomes. Methods. A systematic review of studies published between January 1985 and February 2007 was carried out using the keywords catheterization central venous with combinations of infection*, prevention* and bloodstream*. All included studies were screened by two reviewers, a validated data extraction instrument was used and data collection was completed by two blinded independent reviewers. Risk ratios for catheter-related bloodstream infections and catheter colonization were estimated with 95% condence intervals for each study. Results from studies of similar interventions were pooled using meta-analyses. Results. Twenty-three studies were included in the review. The strategies that reduced catheter colonization included insertion of central venous catheters in the subclavian vein rather than other sites, use of alternate skin disinfection solutions before catheter insertion and use of Vitacuff in combination with polymyxin, neomycin and bacitracin ointment. Strategies to reduce catheter-related bloodstream infection included staff education multifaceted infection control programmes and performance feedback. Conclusion. A range of interventions may reduce risks of catheter-related bloodstream infection, in addition to antimicrobial catheters. Keywords: bloodstream infection, central venous catheters, intensive care unit, meta-analysis, nursing, prevention, systematic review

2008 The Authors. Journal compilation 2008 Blackwell Publishing Ltd

P. Ramritu et al.

Introduction
Catheter-related bloodstream infections (CRBSI) occur at a rate of 50 per 1000 catheter days in the intensive care unit (ICU) setting (McLaws & Taylor 2003, Anonymous, 2004). The mortality attributable to these infections may be as high as 246% and length of stay is extended by between 7 and 191 days; the cost per case has been estimated to be $US4888 (c. 2448) and 16,356 (c. 11,139) (Orsi et al. 2002, Rosenthal et al. 2003a). Reducing risk of CRBSI among ICU patients will save costs, reduce length of stay and improve mortality and morbidity. The risk of CRBSI in the ICU setting depends on many factors. Although many relate to patient diagnosis and underlying health state, others are associated with discretionary decisions made by healthcare professionals. Many studies have focused on the use of anti-microbial catheters to reduce rates in this clinical context and a number of meta-analyses have been published that describe the effectiveness of different types of anti-microbial catheters for the ICU setting (Veenstra et al. 1999, Marin et al. 2000, Mermel 2000, Walder et al. 2002, Geffers et al. 2003). However, a multitude of other interventions used at various points in the catheter insertion and management pathway are available. There are currently no systematic reviews summarizing the evidence for these other interventions specically in relation to their effectiveness in the adult ICU setting, or attempting to bring this information together in one coherent summary (Halton & Graves 2007).

February 2007: MEDLINE; Cumulative Index of Nursing and Allied Health Literature (CINAHL); Current Contents; Current Contents Connect; Australian Medical Index; Biological Abstracts; EMBASE; Science Citation; National Library of Medicine; PubMed; Dissertation Abstracts; Database of Review of Abstracts of Effectiveness; Cochrane Library; Health Services Technology (National Health Service, United Kingdom and United States of America); National Clearing House (Agency for Healthcare Research and Quality); Center for Disease Control guideline and reports; Bandolier and Clinical Evidence (Wales BMJ Publishing group). The broad MeSH search term, catheterization central venous with combinations of keywords infection*, prevention* and bloodstream* was used. The reference lists of all relevant guidelines, systematic reviews, and full articles of all studies included were hand-checked for additional studies. Details of the inclusion and exclusion criteria are presented in Table 1. Observational and randomized controlled trial (RCT) designs were included. Studies were included if a denition of CRBSI was provided that required the isolation of the same organism from the culture of a catheter segment and peripheral blood cultures. Catheter colonization results were reported if the study used a recognized denition of a positive catheter colonization of either 15 colony forming units (CFU) by semi-quantitative culture (Maki et al. 1977) or 103 CFU/ mL by quantitative technique from culture of the distal end of the catheter (Brun-Buisson et al. 1987).

The review
Aim
The aim of the review was to evaluate strategies, other than antimicrobial coated catheters, hypothesized to reduce risk of CRBSI and catheter colonization in the ICU setting.

Table 1 Inclusion and exclusion criteria Inclusion criteria Full report of observational study or randomized controlled trial Investigate short-term (<21 day) non-tunnelled catheters Report incidence of CRBSI as an outcome State denition of catheter colonization and CRBSI used Conducted in adult (18 years) patients within an ICU Written in English Published/completed between January 1985 and February 2007. Exclusion criteria Non-research papers Studies that investigated tunnelled, totally implantable or peripherally inserted central venous catheters or haemodialysis catheters Studies conducted in diverse settings and complete data not presented separately for ICU Studies included arterial and central venous catheters and complete data not presented separately for CVCs Studies included children and adults and complete data not presented separately for adults. CRBSI, catheter-related bloodstream infection; ICU, intensive care unit; CVC, central venous catheter.

Design
A systematic review was conducted according to the methods described by the Centre for Reviews and Dissemination, University of York (Centre for Reviews and Dissemination, 2001).

Search methods
Databases and search terms A search was conducted of the following electronic databases for research published between January 1985 and
4

2008 The Authors. Journal compilation 2008 Blackwell Publishing Ltd

JAN: REVIEW PAPER

Catheter-related bloodstream infections in intensive care units

Search outcome Five hundred and sixty-nine abstracts were reviewed and 513 excluded. The remaining 56 articles were read in full and judged explicitly against the inclusion criteria. Twenty-seven were excluded and 29 met the inclusion criteria. Complete data sets were obtained from twenty-three studies, published between 1985 and February 2007: 13 RCTs, two trials and nine observational cohort studies, see Table 2. Six were excluded as part of the quality assessment exercise because we could not obtain all the necessary data despite attempts to contact the authors, see Table 3 for all exclusion criteria. The process is illustrated by Figure 1. Quality appraisal Validated data extraction instruments for experimental and observational studies chosen for clarity, comprehensiveness and use in previous central venous catheter (CVC)-related reviews [specically the Evidence-based Practice Infection Control (EPIC) project] were pilot tested and used (Pratt et al. 2001). This instrument and the Scottish Intercollegiate Guidelines Network (SIGN) checklist were used to assess study quality (SIGN 2001). As in the EPIC project, studies were grouped into three quality categories (Pratt et al. 2001): category 1, well-designed study, ndings can be generalized to most hospitals; category 2, acceptable experimental study, conclusions may be generalized and in absence of category 1 evidence, study accepted into review; category 3, study methodology fatally awed and rejected from review. Details of the data extracted are provided in Appendix 1. Data abstraction All studies were screened by two reviewers using the abstract or full paper. Data were extracted independently by two reviewers (PR and either NG or DC or KH), blinded to the author, date of publication and author afliations, with inconsistencies resolved by consultation. If required, corresponding authors were contacted by e-mail, twice in 4 weeks, to obtain missing information or clarify ambiguity. Synthesis For each study, the risk ratios (RR) with 95% condence intervals (CI) were calculated for the outcomes of CRBSI and catheter colonization. A RR is a measure of relative risk derived from two proportions. Relative risk is the risk of developing a disease relative to exposure. Data from studies investigating the same intervention were pooled. The DerSimonian & Laird method was used to calculate summary RR with 95% CI for both outcomes via a random effects model (Petitti 2000), normally distributed errors were assumed. This method incorporates an assumption that the different studies

are estimating different, yet related, treatment effects. Statistical heterogeneity was assessed using the MantelHaenzel test statistic. Studies brought together in any systematic review will differ and these differences are known as statistical heterogeneity. STATA 9TM software package (StataCorp, College Station, TX, USA) was used for all the statistical analysis. No sensitivity or subgroup analysis was undertaken due to the small number of studies anticipated to be available for any one type of intervention.

Results
Study characteristics and a summary of results are presented in Tables 3 and 4. Studies were grouped into eight categories: site of CVC insertion; choice of skin disinfectant; catheter replacement at a new site vs. exchange over a guidewire; connectors and hubs; attachable cuffs; number of lumens on the CVC; educational programme to increase healthcare professional awareness of strategies to prevent infection; and combined interventions. All 23 reported rates of CRBSI and 16 studies reported catheter colonization. The range of denitions used in the studies are summarized in Table 5.

Site of insertion (two studies)

Short-term non-tunnelled CVC are inserted into the internal jugular, subclavian, femoral or axillary veins and the risk of CRBSI is believed to vary by site. One RCT (Merrer et al., 2001) found a higher proportion of catheters inserted in the femoral vein were colonized compared to the subclavian (RR 64; 95% CI: 19212; P < 0001) but there was no difference in CRBSI rates (RR 202; 95% CI: 019221; P = 06). A prospective comparative open study (Martin et al. 1998) found a similar colonization rate for internal jugular vs. axillary vein (RR 106; 95% CI: 034332; P = 05) and no signicant difference in the CRBSI rate (RR 042; 95% CI: 004398; P = 041).

Skin disinfectant solutions (three studies)

The skin can be a source of CRBSI, with skin ora believed to migrate along the subcutaneous insertion tract. The use of skin disinfectants prior to insertion and ongoing management of the catheter may be an important risk reducing strategy. Three studies compared different types of skin disinfectants. One RCT (Maki et al. 1991) compared 2% aqueous chlorhexidine (Chl) gluconate to 10% povidone iodine (PI) and 70% isopropyl alcohol. The 2% aqueous Chl solution compared to 10% PI reduced colonization by 69% (RR 031; 95% CI: 017088; P = 001) but there was no statistically
5

2008 The Authors. Journal compilation 2008 Blackwell Publishing Ltd

6 Participants Mean age (years) % Males Study design Setting (no. ICUs) Interventions Sample size RCT OBS Mixed (1) Axillary vs. internal jugular M, S (8) Subclavian vs. femoral 932 24 136 134 62 79 597 619 54 59 62 70 79 75 M/S; M, NS (4) S (1) 10%PI vs. 70% alcohol vs. 2% Chl 10% PI vs. 05% Chl 12 7 RCT CORT M (2) 5% PI/70% ethanol vs. 10% aqueous PI 117 125 77 32 67 106* 117 80 79 68 64 622 583 53 53 51 544 615 615 625 NR NR NR NR NR 66 61 12 RCT CS (1) New site vs. guidewire exchange NR RCT RCT RCT RCT Mixed (1) M, S (3) M, S (7) Standard hub vs. Segur Lock ICU (1) & surgical units Standard hub vs. Segur Lock 18 15 3 NR Hub protection box and standard hub vs. needle-less closed connector with multio hub Standard 3-way stopcock vs. disinfectable needle free connector No cuff & PNB vs. vitacuff & PNB No cuff vs. vitacuff 27* 24* 114 116 37 40 139* 139* NR NR 596 581 NR NR 587 553 NR NR 649 586 NR NR 755 705 RCT NRT S (2) S (1) 9 13 29* 26* 64 90 Single vs. double Single vs. triple Single vs. triple Single vs. triple 491 425 47 57 812 790 NR NR RCT OBS OBS RCT Mixed (4) M/S (1) M (1) NR (1) 12 15 6 22 25 25 63* 157* 99* 309* 51 54 NR NR NR NR NR NR 635 655 NR NR NR NR NR NR NR NR

Table 2 Study characteristics

P. Ramritu et al.

Strategy & Author

Year

Duration of study (months)

Site of insertion Merrer

2001

Martin

1998

Insertion site skin disinfecting solution Humar 2000 RCT

Maki

1991

Parienti

2004

Catheter replacement Bach 1992

Devices: connectors Luna

2000

Leon

2003

Lucet

2000

Yebenes

2004

Devices: cuff Flowers

1989

Hasaniya

1996

Lumens Gupta

1995

Gil

1989

Hilton

1988

2008 The Authors. Journal compilation 2008 Blackwell Publishing Ltd

Farkas

1992

Table 2 (Continued) Participants Mean age (years) % Males

JAN: REVIEW PAPER

Strategy & Author

Year

Study design

Setting (no. ICUs)

Interventions

Sample size

Duration of study (months)

Educational programme Warren 2003 OBS M,S (2) Preintervention (no details) vs. intervention education, posters, fact sheets posted in ICU; 10-page self-study module completed 674 541 71 71 52 52

29

Multiple interventions Bonawitz 1991 RCT S (1)

14

Cobb

1992

RCT

M, S (2)

40* 35* 45* 39* 35 40 41 44 147 34 156 59 NR NR NR

NR NR NR NR 57 59 55

NR NR NR NR 68 50 59 52 NR NR NR

16

2008 The Authors. Journal compilation 2008 Blackwell Publishing Ltd OBS Trauma (1) 15 OBS M (1) 316 190 266 132 338 NR NR NR 443 459 NR NR NR 455 482 11 23 OBS M/S NS (2) OBS M/S C (4) Gp 1 cuff & CVC removed at 3 days Gp 2 cuff & CVC removed at 7 days Gp 3 no cuff & CVC removed at 3 days Gp 4 no cuff & CVC removed at 7 days Gp 1 CVC replaced every 3 days + insertion at new site Gp 2 CVC replaced every 3 days + exchange over guidewire Gp 3 CVC replaced when clinically indicated + insertion at new site Gp 4 CVC replaced when clinically indicated + exchange over a guidewire Phase 1: PI for skin preparation Ph 2: triple lumen Chg/ssd CVCs Phase 3 chlorhexidine skin cleanser, change in criteria for guidewire exchange; extension of safe period for catheter extension from 2 to 4 days Phase 1 pretest, observation of CVC Phase 2 effect of education program, standardized catheter care practices Phase 3 monthly CRBSI rate feedback, provision of CRBSI prevention guide to all medical residents Phase 1 active surveillance without process control Phase 2 infection control education program; process control of CVC care; compliance with infection control practices and CVC care Phase 1 surveillance of CRBSI rates Phase 2 training and education Phase 3 performance feedback on compliance with infection control programme and CVC care NR NR NR 71 71 71 488 536 536 28

Civetta

1996

Lobo

2005

Higuera

2005

Rosenthal

2003

CRBSI, catheter-related bloodstream infection; ICU, intensive care unit; CVC, central venous catheter; NR, not reported; RCT, randomized controlled trial; OBS: cohort observational study; CORT, cross-over unit-randomized trial; NRT, non-randomized trial; M, medical; M/S, medical/surgical; C, cardiac; S, surgical; CS, cardiac surgical; NS, neuro-surgical. *Number of CVCs reported only.

Catheter-related bloodstream infections in intensive care units

P. Ramritu et al. Table 3 Reasons for excluding studies following full review Excluded papers Roberts & Cheung (1998) Berenholtz et al. (2004) Carrer et al. (2005) Coopersmith et al. (2002) Warren et al. (2004) Badley et al. (1996) Reasons for exclusion Denition of CRBSI was not clear. Insufcient information on inclusion criteria Insufcient information on inclusion criteria Insufcient information on inclusion criteria Insufcient information on inclusion criteria Insufcient data on proven CRBSI to calculate rate of CRBSI in the three comparison groups

Catheter replacement on new site vs. exchange over guidewire (one study)
Catheters can be replaced by inserting a replacement catheter in a new site, or by replacing the catheter over a guide wire in the same insertion site. The one RCT included in this review that looked at this intervention found no signicant difference in colonization rates with guidewire exchange relative to new site replacement (RR 037; 95% CI: 01135; P = 010) and no cases of CRBSI were observed (Bach et al. 1992).

Use of devices: connectors (four studies)

Central venous catheter hubs may be a source of intraluminal catheter colonization and CRBSI due to frequent manipulation and use of a PI connection shield might reduce risk. Two RCTs (Lucet et al. 2000, Leon et al. 2003) compared the rate of CRBSI observed with use of a Segur-Lock (Inibsa Laboratories, Barcelona, Spain) connector relative to that observed under use of a standard hub. Luna (Luna et al. 2000) included patients from ICU and surgical units and data from ICU participants only are included in the analysis. The pooled ndings of these two studies indicated a non-signicant reduction in risk of CRBSI with Segur-Lock, see Figure 2 (RR 027; 95% CI: 007105). Lucet et al. (2000) compared antiseptic impregnated hub protection boxes to needle-less closed connectors in a RCT, and reported a non-signicant difference in risk of catheter colonization and CRBSI (RR 125; 95% CI: 066237; P = 03 and RR 096; 95% CI: 0061499; P = 074, respectively). The fourth RCT (Yebenes et al. 2004) compared a disinfectable needle-free connector to the standard three-way stopcock. The rates of CRBSI and catheter colonization were lower in the needle-free connector group, with this difference approaching statistical signicance for CRBSI (RR 014; 95% CI: 002114; P = 003), but not for catheter colonization (RR 069; 95% CI: 0316; P = 037).

CRBSI, catheter-related bloodstream infection.

Search strategy

569 abstracts for review

513 rejected

56 articles for further review

27 excluded (did not meet inclusion criteria)

29 included for full review

6 excluded (complete data sets not available)

Figure 1 Search strategy and study selection.

23 included

signicant reduction for CRBSI (RR 023; 95% CI: 003 192; P = 014). Compared to the 70% alcohol solution the 2% Chl showed no statistically signicant difference in colonization (RR 038; 95% CI: 011133; P = 011) or CRBSI rates (RR 024; 95% CI: 002254; P = 024). Another RCT (Humar et al. 2000) compared 10% PI solution with 05% tincture of Chl solution and found no difference in rates of colonization or CRBSI (RR 08; 95% CI: 052126; P = 022; RR 107; 95% CI: 027417; P = 06 respectively). The third study (Parienti et al. 2004) used a cross-over unit-randomized trial to compare 10% aqueous PI solution with 5% PI in 70% ethanol based aqueous solution. Colonization rates were signicantly lower for alcoholic PI (RR 038; 95% CI: 022 to 065; P < 0001) but there was no signicant difference in CRBSI rate (RR 03; 95% CI: 00324; P = 021).

Use of devices: cuffs (two studies)

A biodegradable collagen cuff impregnated with bactericidal silver ions (Vitacuff; Vitaphore Corp, Menlo Park, CA, USA) has been suggested to prevent CVC colonization from skin organisms migrating along the subcutaneous tract. One RCT (Flowers et al. 1989) and one non-randomized trial (Hasaniya et al. 1996) investigated triple-lumen catheters with an attachable cuff (Vitacuff) relative to catheters with no cuff. In the RCT, colonization rates with

2008 The Authors. Journal compilation 2008 Blackwell Publishing Ltd

Table 4 Summary results of included studies (grouped together according to preventative strategy examined) Year Intervention Proportion of colonization RR (95% CI) Proportion of CRBSI RR (95% CI)

Preventative strategy/ Author

JAN: REVIEW PAPER

Site of insertion Merrer 2001 Subclavian vs. femoral

Martin

1998

Axillary vs. internal Jugular

3/136 19/134 RR (using femoral site) 64 (19212, P < 0001) 5/62 6/79 RR (using axillary vein) 106 (034332, P = 05)

1/136 2/134 RR (using femoral site) 202 (0192212, P = 06) 1/62 3/79 RR (using axillary vein) 042 (004398, P = 041) 41/1000 or 4/177 46/1000 or 4/125 RR (using 10% PI) 107 (027417, P = 06)

Insertion site skin disinfecting solution Humar 2000 10% PI vs. 05% Chl

Maki

1991

10% PI vs. 70% alcohol vs. 2% Chl gluconate

2008 The Authors. Journal compilation 2008 Blackwell Publishing Ltd 2004 5% PI/70% ethanol vs. 10% aqueous PI 46/1000 or 24/88 34/1000 or 31/92 RR (using 10% PI) 081 (052126, P = 022) 15/77 5/32 4/67 RR (using 2% Chl vs. 10%PI) 031 (017088, P = 001) RR (using 2% Chl vs. 70% alcohol) 038 (011133, P = 011) RR (using 10% PI vs. 70% alcohol) 125 (049314, P = 043) 14/106 41/117 RR (using 5% PI) 038 (022065, P < 0001) 3/80 8/79 RR (using new site) 037 (01135, P = 010) 5/77 2/32 1/67 RR (using 2% Chl vs. 10%PI) 023 (003192, P = 014) RR (using 2% Chl vs. 70% alcohol) 024 (002254, P = 024) RR (using 10% PI vs. 70% alcohol) 104 (021508, P = 067) 1/106 4/117 RR (using 5% PI) 03 (00324, P = 021) 1992 New site vs. guidewire 0/80 0/79 2000 Standard hub vs. Segur Lock 1/27 0/24 2003 Standard hub vs. Segur Lock 2/27 1/24 RR (using Segur Lock) 056 (005582, P = 054) N/R NR 2000 Hub protection box and standard hub vs. needle-less closed connector with multio hub Standard 3-way stopcock vs. disinfectable needle free connector 2004 13/67 17/70 RR (using needleless connector) 125 (066237, P = 031) 13/139 9/139 RR (using connector) 069 (0316, P = 037) 8/114 2/116 RR using Segur-Lock 024 (005113, P = 005) 1/67 1/70 RR (using needleless connector) 096 (0061499, P = 074) 7/139 1/139 RR (using connector) 014 (002114, P = 003)

Parienti

Catheter replacement Bach

Devices: connectors Luna

Leon

Lucet

Yebenes

Catheter-related bloodstream infections in intensive care units

10 Year Intervention Proportion of colonization RR (95% CI) Proportion of CRBSI RR (95% CI) 1989 No cuff & PNB vs. vitacuff & PNB 4/29 0/26 1996 No cuff vs. vitacuff 10/29 2/26 RR (using Vitacuff) 022 (005092, P = 002) 23/135 12/100 RR (using Vitacuff) 070 (037135, P = 019) 4/135 3/100 RR (using Vitacuff) 101 (023442, P = 064) 1/25 0/25 1995 Single vs. double 1989 Single vs. triple 5/63 6/ 157 RR (using triple lumen) 048 (015152, P = 017) 1988 1992 Single vs. triple 17/68 12/61 RR (using triple lumen) 078 (041151, P = 04) N/R N/R Single vs. triple 6/25 4/25 RR (using double lumen) 067 (021: 208, P = 036) 10/63 32/ 157 RR (using triple lumen) 128 (067245, P = 028) NR NR 3/99 17/309 RR (using triple lumen) 182 (054606, P = 024) 5/68 5/61 RR (using triple lumen) 111 (034366, P = 085) 2003 Preintervention Intervention education posters, fact sheets posted in ICU; 10-page self-study module completed Gp Gp Gp Gp 1 2 3 4 cuff & CVC removed at 3 days cuff & CVC removed at 7 days no cuff & CVC removed at 3 days no cuff & CVC removed at 7 days 49/1000 31/1000 RR (using post intervention) 043 (022084) 1991 3/40 8/35 9/45 6/39 RR (using Gp 1 vs. Gp 3 038 (011129, P = 009) RR (using Gp 1 vs. Gp 4) 049 (013181, P = 031) RR (using Gp 2 vs. Gp 4) 149 (057386, P = 03) RR (using Gp 4 vs. Gp 3) 077 (030197, P = 04) 2/40 2/35 0/45 0/45 RR (using Gp 1 vs. Gp 2) 087 (013589, P = 064)

Table 4 (Continued)

P. Ramritu et al.

Preventative strategy/ Author

Devices: cuff Flowers

Hasaniya

Lumens Gupta

Gil

Hilton

Farkas

Educational programme Warren

2008 The Authors. Journal compilation 2008 Blackwell Publishing Ltd

Multiple interventions Bonawitz

Table 4 (Continued) Year 1992 10/220 5/94 5/100 2/100 1/94 4/220 Gp 1 CVC replaced every 3 days + insertion at new site Gp 2 CVC replaced every 3 days + exchange over guidewire Gp 3 CVC replaced when clinically indicated + insertion at new site Gp 4 CVC replaced when clinically indicated + exchange over a guidewire 6/109 1/109 Intervention Proportion of colonization RR (95% CI) Proportion of CRBSI RR (95% CI)

Preventative strategy/ Author

JAN: REVIEW PAPER

Cobb

2008 The Authors. Journal compilation 2008 Blackwell Publishing Ltd 1996 Phase 1: PI for skin preparation; Phase 3 chlorhexidine skin cleanser, change in criteria for guidewire exchange; extension of safe period for catheter extension from 2 to 4 days RR (using Gp 1 vs. 2) 050 (005446, P = 046) RR (using Gp 3 vs. 4) 053 (005577, P = 052) RR (using Gp 1 vs. 3) 086 (005136, P = 071) RR (using Gp 2 vs. 4) 091 (017488, P = 061) 8/387 6/298 RR (using phase 3) 097 (034278, P = 059) 2005 RR (using Gp 1 vs. 2) 121 (045325, P = 044) RR (using Gp 3 vs. 4) 106 (032356, P = 059) RR (using Gp 1 vs. 3) 103 (033328, P = 060) RR (using Gp 2 vs. 4) 091 (032260, P = 053) Phase 1 59/387 26/298 RR (using phase 3) 057 (037088, P = 001) NR NR NR Phase 1 pretest, observation of CVC care practices Phase 2 effect of education programme, standardized catheter care practices Phase 3 monthly CRBSI rate feedback, provision of CRBSI prevention guide to all medical residents 2005 NR NR 20/1000 or 48/2450 12/1000 or 16/1381 12/1000 or 22/1701 RR (using phase 2 vs. phase 1 059 (033103, P = 006) RR (using phase 3 vs. phase 1) 066 (04108, P = 010) 46/1000 or 28/132 195/ 1000 or 55/338 RR 042 (027066 P = 00001) NR Phase 1 vs. Phases 2 &3 4594/1000 1110/1000 RR 025 (017036) 2003 Phase 1 active surveillance without process control Phase 2 infection control education programme; process control of CVC care; compliance with infection control practices and CVC care Phase 1 surveillance of CRBSI rates Phase 2 training and education Phase 3 performance feedback on compliance with infection control programme and CVC care

Civetta

Lobo

Higuera

Rosenthal

CVC, central venous catheter; PNB, polymyxin, neomycin & bacitracin; PI, povidone iodine; Chl, chlorhexidine; bzlk, benzylkonium chloride; bchl, benzyl alcohol.

Catheter-related bloodstream infections in intensive care units

11

P. Ramritu et al. Table 5 Denitions of catheter-related bloodstream infection used in studies included in the review Author Bach Bonawitz Civetta Cobb Farkas Flowers Gil Gupta Hasaniya Higuera Hilton Humar Leon Lobo Lucet Luna Maki Martin Merrer Parienti Rosenthal Warren Yebenes Year 1992 1991 1996 1992 1992 1989 1989 1995 1996 2005 1988 2000 2000 2005 2000 2000 1991 1998 2001 2004 2003 2003 2004 SO SO/NS SO/CS SO/NS/CS SO/NS/CS/AB SO/NS/RS SO/CS/RS NA rSO/CS/NS r r r r r r d r r r r r e r r rSO/CS r r d r

SO, same organism isolated from catheter and peripheral blood; NS, no other source of infection; CS, clinical symptoms of sepsis/infection; AB, physician instituted antimicrobial therapy; RS, resolution of symptoms on catheter removal. r, only intravascular catheter segment cultured. h, intravascular & subcutaneous catheter segments cultured. , intravascular and negative infusate culture. d, intravascular and skin swab at insertion site. s, intravascular & subcutaneous & other (insertion site, hub, infusate) catheter segment cultured. e, not stated if catheter segment was cultured. NA, denition not provided or not clear or deduced from article without a denition being provided. +, CRBSI was dened according to the CDC denitions, no reference provided.

Risk ratio Study Luna et al (2000) 037 (002, 875) Leon et al (2003) 025 (005, 113) Overall 810 190 (95% CI) % Weight

027 (007, 105) 1000

015920

1 Risk ratio

628107

Figure 2 Catheter-related bloodstream infection: Segur-Lock vs. standard hub.

Vitacuff were reduced by 78% (RR 022; 95% CI: 005 092; P = 002), and no cases of CRBSI were observed in the Vitacuff group compared to four cases amongst the no cuff catheters. Conversely, the non-randomized trial found no signicant difference in colonization (RR 070; 95% CI: 037135; P = 019) and CRBSI rates (RR 101; 95% CI: 023442; P = 064) between groups. Polymyxin, neomycin and bacitracin ointment (PNB) was applied to the insertion site of all catheters in the RCT and it is possible that this may contribute to differences in the outcome of the studies. The pooling of these studies is illustrated in Figure 3 and shows no statistically signicant reduction in colonization (RR 047; 95% CI: 016139) or CRBSI (RR 050; 95% CI: 007376).

12

2008 The Authors. Journal compilation 2008 Blackwell Publishing Ltd

JAN: REVIEW PAPER

Catheter-related bloodstream infections in intensive care units

Risk ratio Study (95% CI) 048 ( 015, 152) 182 ( 054, 607) 111 ( 034, 367) % Weight 348 323 330

Risk ratio Study (95% CI) % Weight

Gil et al (1989)

Flowers et al (1989) Hasaniya et al (1996)

012 ( 001, 219)

333

Hilton et al (1988) Farkas et al (1992) Overall

101 ( 023, 442)

667

097 ( 045, 210) 1000

Overall

050 ( 007, 376) 1000

152430 1 Risk ratio 656036

006964

1 Risk ratio

143591

Figure 3 Catheter-related bloodstream infection: vitacuff vs. no cuff.

Figure 4 Catheter-related bloodstream infection: triple vs. single lumens.

Number of lumens (four studies)

The risk of colonization and CRBSI may be related to the number of lumens attached to the CVC. One RCT (Gupta et al. 1995) compared double lumen CVC relative to single lumen catheters. One patient in the single lumen catheter group had a CRBSI with no cases in the double-lumen group but neither this difference nor the reduction in risk of catheter tip colonization observed with double-lumen catheters were signicant (RR 067; 95% CI: 021208; P = 036). Three studies (Hilton et al. 1988, Gil et al. 1989, Farkas et al. 1992) compared CRBSI rates for triple versus single-lumen catheters, with two studies (Gil et al. 1989, Farkas et al. 1992) also reporting rates of catheter colonization. Pooling results from the studies did not show a statistically signicant reduction in risk of either catheter colonization (RR 096; 95% CI: 053 172) or CRBSI (RR 097; 95% CI: 045210; Figure 4).

Education of healthcare workers (one study)

Staff education and the promotion of preventative strategies in clinical practice may reduce risks of catheter colonization and CRBSI in the ICU setting. In one prepost intervention study, posters and fact sheets were posted in the ICU and nurses and physicians completed a 10-page self-study module (Warren et al. 2003). A reduction in CRBSI from 49 cases per 1000 catheter-days to 21 cases per 1000 catheter days was reported for the pre and post postintervention periods (RR 043; 95% CI: 022084). All CVC used in this study were bonded with Chl and silver sulfadiazine.

Combinations of interventions (six studies)

Prevention of CRBSI requires the implementation of multiple strategies at insertion and management over the duration of the catheter. Six studies investigated the effectiveness of using

strategies in combination to prevent CRBSI. One RCT (Bonawitz et al. 1991) investigated the effect of using silver-impregnated implantable collagen cuffs in combination with a specic catheter removal policy (removal at either 3 or 7 days). There was no statistically signicant difference in colonization or CRBSI rates between any of the four groups. Another RCT investigated the combined effect of mode of catheter replacement (via a new site or over guidewire) and replacement schedule (every 3 days vs. when clinically indicated) (Cobb et al. 1992) Again, there was no statistically signicant difference in catheter colonization or CRBSI between the 4 groups. Civetta et al. (1996) reported on a prospective three-phase continuous quality management intervention, which replaced the use of PI solution prior to catheter insertion with Chl skin cleanser, changed the criteria for guidewire exchange and extended the safe period considered for exchange of catheter from 2 to 4 days. They observed a signicant reduction in the catheter colonization rate between the rst and the third phase (RR 057; 95% CI: 037088; P = 001) but no reduction in the rate of CRBSI (RR 097; 95% CI: 034278; P = 059). Three prepost intervention studies investigated combined interventions of education and training, a standardized catheter care protocol as well as performance feedback. Lobo et al. (2005) conducted a pre- and postintervention study with two phases. Following the rst phase of baseline observation, phase 2 investigated the effect of an education programme and a standardized catheter care policy on CRBSI rates. The third phase studied the additional impact of providing monthly feedback on CRBSI rates to the unit and provision of a CRBSI prevention guide to all medical residents. The reduction in CRBSI rates following the rst intervention (phase 2) approached statistical signicance (RR 059; 95% CI: 033103; P = 006), whilst the additional interventions provided in phase 3 appeared to provide no additional reduction in risk. Higuera et al. (2005) studied the
13

2008 The Authors. Journal compilation 2008 Blackwell Publishing Ltd

P. Ramritu et al. Table 6 Quality assessment of randomized controlled trials Blinding of analysis personnel NR NR Y NR Y NR NR NR Y NR NR NR Y Power/effect size calculation included in study design Y Y NR NR Y Y NR NR Y NR NR NR Y

References Merrer et al. (2001) Humar et al. (2000) Maki et al. (1991) Bach et al. (1992) Luna et al. (2000) Leon et al. (2003) Lucet et al. (2000) Yebenes et al. (2004) Flowers et al. (1989) Gupta et al. (1995) Farkas et al. (1992) Bonawitz et al. (1991) Cobb et al. (1992)

Reporting of randomization 1 1 1 2 2 1 2 2 1 2 2 1 1

Allocation concealment Y NR NR NR NR NR NR NR NR NR NR NR NR

Blinding of care givers NP U NP NP NP NP NP NP NP NP NP NP NP

Intention to treat analysis performed N Y NR NR NR Y NR Y N NR N N N

Yes, investigators and researchers not aware of allocation of intervention; NP, blinding of intervention not possible; U, unclear, probably not possible; Y, yes; NR, not reported and, not clear from analysis. 1, detailed randomization process provided; 2, randomly allocated, details of randomization not provided.

impact of introducing a combined intervention on top of their baseline surveillance activities for catheter associated infections and compliance with site care and hygiene. The intervention comprised education in catheter care, implementation of process control measures based on the Centres for Disease Control and Prevention (CDC) guidelines, observation of practices and performance feedback on compliance with hand hygiene and catheter care. The authors found a statistically signicant reduction in CRBSI rates from baseline following the intervention (RR 042; 95% CI: 027066; P = 00001). Following an initial baseline observation phase, Rosenthal (Rosenthal et al. 2003b) investigated the effectiveness of incremental implementation of education of healthcare workers on care of CVC. This formed the second phase, which was followed by a third phase in which performance feedback was given to workers and administrators in each ICU on rates of compliance with specic site care activities including hand hygiene, placement of gauze dressing on insertion site and documentation of replacement date on administration set. From baseline phase 1 there was a statistically signicant reduction in CRBSI rates following the incremental implementation of education of healthcare workers in phase 2 (RR 037; 95% CI: 019073; P < 0001). The rate continued to fall with the introduction of performance feedback in the third phase but the difference between phase 2 and 3 did not reach signicance (RR 058; 95% CI: 029118; P = 011). The overall reduction in rate of CRBSI over phases 2 and 3 combined, relative to phase 1 was approximately 75% (RR 025; 95% CI: 017036; P < 0001).
14

Quality of included studies

All studies included in the review were judged to meet category 2 criteria: acceptable experimental study, conclusions may be generalized and, in absence of category 1 evidence, study accepted into review. In addition the RCTs included in the review were assessed according to the reporting of randomization, allocation concealment, blinding of caregivers and research staff, intention to treat analysis and whether the power/effect size calculation was included in study design, see Table 6. Seven of the 13 RCTs provided details of randomization process; with a small number reporting on allocation concealment and intention to treat analysis. Four studies reported on blinding of microbiology and/or research staff.

Discussion
This systematic review summarizes the evidence for a broad range of interventions for the prevention of CRBSI and focuses specically on their effectiveness in the ICU setting. It complements earlier reviews set in this clinical context that have focused exclusively on the use of antimicrobial catheters by providing important information on the effectiveness of many of the other interventions and practices which are used in the insertion and management of CVCs. We found good evidence for the effectiveness of appropriately designed and delivered education programmes introduced alone (Warren et al. 2003) or in combination with focused infection control policies and performance feedback on compliance with these policies and/or CRBSI rates

2008 The Authors. Journal compilation 2008 Blackwell Publishing Ltd

JAN: REVIEW PAPER

Catheter-related bloodstream infections in intensive care units

(Rosenthal et al. 2003b, Higuera et al. 2005) in reducing rates of CRBSI in a variety of critical care settings. The potential for education programmes to reduce rates of CRBSI across a broader range of clinical contexts has been commented upon in other reviews (Gastmeier & Geffers 2006), and, along with surveillance and feedback of infection rates, have long been recognized as an important tool in infection control. A number of points should be made about the strength of conclusions that can be drawn from this evidence. First, it may be questionable whether these educational interventions are actually effective or as effective as they appear. The evidence in this review for education programmes comes mainly from before and after intervention studies which provide less convincing evidence of effectiveness than RCTs, and the outcomes of these interventions can be vulnerable to the Hawthorne effect. Moreover, studies which report dramatic falls in infection rates following an educational intervention often arise from settings where baseline rates of CRBSI are relatively high in comparison to the national averages (Rosenthal et al. 2003b, Lobo et al. 2005) and the same results may not be achievable in hospitals with lower rates. Second, the difculty in using this evidence to inform practice in other critical care settings, even if the evidence is credible, arises from whether the results can be replicated in a different context. The success of educational programmes has been shown to be dependent to varying degrees upon key individuals involved in delivering and promoting the intervention in each setting. Indeed, the importance of strong leadership and recruitment of key individuals is recognized as crucial to the success of the United States Institute for Healthcare Improvement Five Million Lives Campaigns. Also, strategies to reduce CRBSI that include educational components are often highly complex and delivered in conjunction with other changes, such as modied catheter insertion and management protocols or surveillance and feedback strategies. This makes interpretation of results more complex as it cannot be identied which element or elements of the combined programme is most useful. Amongst the other interventions included in this review, a number of strategies were shown to reduce rates of catheter colonization. Catheter colonization is often used as a surrogate outcome for CRBSI (Rijnders & Van Wijngaerden 2002), and these interventions may represent a good focus for further research. Interventions which may merit particular focus include preferential insertion of the CVC in the subclavian vein (Merrer et al., 2001); use of one of two types of skin disinfectant solutions: 2% aqueous Chl (Maki et al. 1991) or 5% alcoholic PI (Parienti et al. 2004); use of Vitacuff in combination with PNB ointment on the insertion

site (Flowers et al. 1989); and a combined intervention of Chl skin disinfectant solution, catheters coated with Chl and silver sulfadiazine and the extension of safe periods for catheter removal of suspected catheter-related sepsis from 2 to 4 days (Civetta et al. 1996). Evidence from this review supports the effectiveness in the ICU context of the education of healthcare workers, and suggests that preferential use of the subclavian site for insertion and skin antisepsis with 2% Chl may also be benecial in preventing CRBSI. These interventions are already recommended in the CVC management guidelines published by the CDC (Centres for Disease Control and Prevention, 2002) that are aimed at a more general context. This suggests that, in the absence of ICU-specic evidence, information derived from studies in non-ICU settings, studies using surrogate endpoints only, such as colonization, expert opinion and in vitro studies as incorporated in the CDC guidelines, provide relevant information with which to inform clinical practice in the ICU. It is also worth noting that these relatively simple interventions that relate to good catheter management may have the potential to benet a wider range of critical care settings, including resource-poor hospitals, than high-cost technological interventions such as antimicrobial catheters.

Review strengths and limitations

The strengths of this review are that we summarize the evidence for a wide range of interventions employed to reduce CRBSI, and the application of systematic review methods eliminated bias in selection, appraisal and data extraction. Its weaknesses include the possibility of publication bias from only including studies published with an English language abstract and exclusion of studies where authors did not respond to questions to clarify data and heterogeneity between studies, which makes comparison difcult or irrelevant. The conclusions are limited by underlying issues in the evidence included. Only 13 of the 23 studies (57%) used a RCT study design, with the others using a mixture of quasiexperimental or observational study designs. The predominance of these types of studies in the infection control literature has already been commented upon (Harris & Lautenbach 2005), and many questions relevant to infection control and critical care practice are not amenable to study by clinical trial. It should also be noted that the results of our quality assessment indicate that the RCT evidence may also be vulnerable to sources of bias that relate to the way these studies were conducted and analysed. Careful thought should be given in future trials of any study design to how sources of
15

2008 The Authors. Journal compilation 2008 Blackwell Publishing Ltd

Table 7 Clinical heterogeneity in included studies

16 No. lumens Dressing CVC insertion procedure Insertion site skin disinfecting solution Dressing change regimen, days Mean duration of CVC, days LSD,SG, SGl, M,C SG; SGl; M, C PI SSD daily PI or Chl or I Opsite IV3000 PRN, SP- 16 No No 1,2,3 NC 1, SP Subclavian 111 Femoral 94 Axillary 15 Jugular 15 MBP, SGl; SG, M, LD SGl 10%PI; 70%Alc; 2% Chlg SG 2 10% PI or 05%chl SG PRN, 3 No No NR 1,2,3,4, SP MBP, SD 4%PI; followed by 10% Aq PI or 5% AlcPI. PI 10% PI; AO SG AO SG TSD 3 10%PI 83 05% Chl 69 10%PI 53 70%Alc 52 2% Chlg 53 Aq PI 9 Alch PI 87 NR 3 7 St. hub 13 S. Lock 14 2 week NR Gauze TPU NR 3 daily PRN, 2 2 Yes 2 2,3 SP SG, SGl, C,M No Yes 2,3 NR M, NC 3, SP 3 M,G,SD PI M,SG,SGl, SD 10% PI, AO SG, LSD,M,C,SGl Chl soap, 10% PI As per CDC NR SG,Gl,C, M; LSD; vs. SG only in experimental group Standard aseptic, SD 10%PI Gauze, AO No Yes Yes Yes Yes St hub 112 S. Lock 108 St hub & box 73 Multi-o hub 70 99 No cuff 44 Vitacuff 55 No cuff 77 Cuff 62 I & spirit NR NR PI SG Sterile dressing Sterile dry dressing Opsite daily daily NS 2 1,2 1,3, SP 1,2,3 1,3, SP C, M, SGl, SG C,M,G,SGl NS C, M, SG, SGl, SD No Yes Yes Yes 85 NR Single 6 Triple 9 Single 116 Triple 91 All CVCs ChgSS coated NS NS NS NS Pre 91 Post 96 NS

P. Ramritu et al.

Preventative strategy/Author

Year

Insertion site

>1 CVC allowed

Site of insertion Merrer

2001

F, SC

Martin

1998

A, IJ

Insertion site skin disinfecting solution Humar 2000 IJ, SC 1, M, SP

Maki

1991

SC, IJ; F

Parienti

2004

IJ, SC, F

Catheter replacement Bach 1992 Devices: connectors Luna 2000

IJ

SC, J, F

Leon

2003

SC, IJ

Lucet

2000

NR

Yebenes Devices: cuff Flowers

2004

SC/ J

1989

SC

Hasaniya

1996

NR

Lumens Gupta Gil Hilton

1995 1989 1988

SC IJ,SC,F NR

Farkas

1992

SG, IJ

2008 The Authors. Journal compilation 2008 Blackwell Publishing Ltd

Educational programme Warren

2003

NS

JAN: REVIEW PAPER

Table 7 (Continued)

Preventative strategy/Author Dressing

Year

Insertion site

No. lumens

CVC insertion procedure

Insertion site skin disinfecting solution

Dressing change regimen, days

Mean duration of CVC, days

>1 CVC allowed

Multiple interventions Bonawitz 1991 SC; IJ SC, IJ 3 M, SG, SGl, SD 10%PI 2 M S, SG, SGl, C, M NS 2-3 Gauze pad & Opsite G, AO

Yes Yes

Cobb

1992

2008 The Authors. Journal compilation 2008 Blackwell Publishing Ltd NR SG/AO G G SC mostly NS NS S 3 M,C,SG,SGl Phase1: PI Phase 2: 4% chl PI NS PRN, 3 PRN, daily NS Gps1&3 3 Gps2&4 7 Gp1 3 Gp2 3 Gp3 6 Gp4 7 Phase1 45 Phase3 54 NS 73* Yes NS NS NS S SG,SGl, LD, C,M Full barrier precaution sometimes NS NS G NS 707* NS

Civetta

1996

Lobo Higuera

2005 2005

Rosenthal

2003

NS, not specied; NR, not reported; CVC, central venous catheter; A, axillary; F. femoral; IJ, internal jugular; J, jugular; NR, not reported; S, supraclavicular; SC, subclavian; NC, noncoated CVCs; Chg/ssd, chlorhexidine gluconate/silver sulfadiazine coated CVCs; M, multiple; S, standard; SP, standard polyurethane CVC; D, drapes; LD, large drape; MBP, maximal barrier precaution; SD, sterile drape; LSD, large sterile drape; G. gown; SG. sterile gown; SGl, sterile gloves; Gl, gloves; M, mask; C, cap; Alc, alcohol; AlcPI. alcoholic povidone iodine; Aq PI, aqueous povidone iodine; AO, antibiotic/antimicrobial ointment; Chlg, chlorhexidine gluconate; Chl, chlorhexidine; I, iodine tincture; PI, povidone iodine; SSD, standard sterile dressing; TSD, transparent sterile dressing; TPU, transparent polyurethane dressing; SG, sterile gauze; G, gauze; THD, transparent hydrocolloid dressing; SPU, standard polyurethane dressing; AO, antiseptic/antimicrobial ointment applied on insertion site; PRN, changed as required; SP, study site preference. *Information obtained from authors via e-mail.

Catheter-related bloodstream infections in intensive care units

17

P. Ramritu et al.

What is already known about this topic

Catheter-related bloodstream infection arising in the intensive care unit setting is a serious clinical and economic problem. Antimicrobial catheters are effective at reducing risk of catheter-related bloodstream infection in the intensive care unit setting. Less is known about the effectiveness of other risk reducing strategies.

such as site of insertion, use of barrier precautions and type of disinfecting solution. Other problems arise from the ndings being inuenced by the absence of a consistent denition for CRBSI, see Table 5, and inconsistent catheter management protocols across studies. In short, we still do not have a full understanding of the most effective ways to prevent CRBSI in the ICU context although this review and others can help to indicate where some of the largest gaps in our knowledge lie.

Conclusion
What this paper adds
There is some evidence for strategies other than antimicrobial catheters, such as staff education, multifaceted infection control programmes and performance feedback. This quality of this evidence is mixed and studies are characterized by small samples. These interventions may be less costly and easier to implement than widespread use of antimicrobial catheters, but further work to model the cost-effectiveness of all strategies is required. Interventions other than antimicrobial catheters may be useful for reducing risks of CRBSI in the ICU setting. Infection control is typically nurse-led within a wider multidisciplinary team and nurses play a key role in the care of ICU patients, including assisting with the insertion and daily management of CVCs. Nurse awareness and implementation of appropriate risk reducing strategies is important. This review is relevant to evidence-based nursing practice in many international settings. It highlights the strength of evidence underlying each intervention and the recent trend to use a combination of most effective interventions to manage the care of CVC. The strategies we recommend are relatively low-cost, pose minimal risks for patients and constitute good nursing practice. Finally, good decision-making requires that the cost-effectiveness of all interventions be assessed and compared, but understanding their effectiveness is an important rst step.

error and bias can be minimized at all stages of the research. Other limitations of the available evidence include the inadequate power of many of the studies to detect a real difference in CRBSI and colonization rates. Where possible we combined evidence in meta-analyses, but the small number of studies available for which we could pool data still precluded the provision of statistical support for the effectiveness of interventions. For example, colonization rates were lower but did not reach signicance at the 5% level for a number of interventions such as 10% PI vs. 05% Chl (Humar et al. 2000) and new site vs. guidewire exchange for CVC replacement (Bach et al. 1992). We suggest future studies be designed with a sample size large enough to detect a real difference or to inform a subsequent meta-analysis. The appraisal of evidence for reducing catheter colonization and CRBSI is further complicated by the fact that a number of different interventions are used concurrently in the insertion and ongoing management of CVC. Unless all studies investigating a particular intervention control for the effect of these co-interventions it will be difcult to determine the interventions that are most effective in reducing colonization and CRBSI, and if we can expect to achieve this effect in other settings. As shown in Table 7, there was much clinical heterogeneity in the studies and important confounding data was often not provided for some co-interventions
18

Acknowledgements
Financial support for the study was provided by the National Health & Medical Research Council of Australia.

Author contributions
PR and NG were responsible for the study conception and design and PR, DC and NG were responsible for the drafting of the manuscript. PR, KH, DC and NG performed the data collection and PR performed the searches and data analysis. MW and NG obtained funding. PR, KH, DC and NG made critical revisions to the paper. KH provided statistical advice. DC, MW and NG supervised the study.

References
Anonymous (2004) National Nosocomial Infections Surveillance System. National Nosocomial Infections Surveillance (NNIS) System report, data summary from January 1992 through June 2004,

2008 The Authors. Journal compilation 2008 Blackwell Publishing Ltd

JAN: REVIEW PAPER issued October 2004. American Journal of Infection Control 32, 470485. Bach A., Stubbig K. & Geiss H. (1992) Infectious risk of replacing venous catheters by the guide-wire technique. Zentralblatt fur Hygiene und Umweltmedizin 193, 150159. Badley A., Steckelberg J., Wollan P. & Thompson R. (1996) Infectious rates of central venous pressure catheters: comparison between newly placed catheters and those that have been changed. Mayo Clinic Proceedings 71, 838846. Berenholtz S., Pronovost P., Lipsett P., Hobson D., Earsing K., Farley J., Milanovich S., Garrett-Mayer E., Winters B., Rubin H., Dorman T. & Perl T. (2004) Eliminating catheter-related bloodstream infections in the intensive care unit. Critical Care Medicine 32, 20142020. Bonawitz S., Hammell E. & Kirkpatrick J. (1991) Prevention of central venous catheter sepsis: a prospective randomized trial. American Surgeon 57, 618623. Brun-Buisson C., Abrouk F., Legrand P., Huet Y., Larabi S. & Rapin M. (1987) Diagnosis of central venous catheter-related sepsis. Archives of Internal Medicine 147, 873877. Carrer S., Bocchi A., Bortolotti M., Braga N., Gilli G., Candini M. & Tartari S. (2005) Effect of different sterile barrier precautions and central venous catheter dressing on the skin colonization around the insertion site. Minerva Anestesiologica 71, 197206. Centre for Reviews and Dissemination (2001) Undertaking Systematic Reviews of Research on Effectiveness: CRDs Guidance for Those Carrying out or Commissioning Reviews. CRD Report Number 4, 2nd edn. University of York, York. Centres for Disease Control and Prevention (2002) Guidelines for the prevention of intravascular catheter-related infections. MMWR 51(RR10), 126. Civetta J., Hudson-Civetta J. & Ball S. (1996) Decreasing catheterrelated infection and hospital costs by continuous quality improvement. Critical Care Medicine 24, 16601665. Cobb D., High K., Sawyer R., Sable C., Adams R., Lindley D., Prvett T., Schwenzer K. & Farr B. (1992) A controlled trial of scheduled replacement of central venous and pulmonary-artery catheters. [comment]. New England Journal of Medicine 327, 10621068. Coopersmith C., Rebmann T., Zack J., Ward M., Corcoran R., Schallom M., Sona C., Buchman T., Boyle W., Polish L. & Fraser V. (2002) Effect of an education program on decreasing catheterrelated bloodstream infections in the surgical intensive care unit. Critical Care Medicine 30, 5964. Farkas J., Liu N., Bleriot J., Chevret S., Goldstein F. & Carlet J. (1992) Single-versus triple-lumen central catheter-related sepsis: a prospective randomized study in a critically ill population. American Journal of Medicine 93, 277282. Flowers R., Schwenzer K., Kopel R., Fisch M., Tucker S. & Farr B. (1989) Efficacy of an attachable subcutaneous cuff for the prevention of intravascular catheter-related infection. A randomized, controlled trial. JAMA 261, 878883. Gastmeier P. & Geffers C. (2006) Prevention of catheter-related bloodstream infections: analysis of studies published between 2002 and 2005. Journal of Hospital Infection 64, 326335. Geffers C., Zuschneid I., Eckmanns T., Ruden H. & Gastmeier P. (2003) The relationship between methodological trial quality and the effects of impregnated central venous catheters. Intensive Care Medicine 29, 403409.

Catheter-related bloodstream infections in intensive care units Gil R., Kruse J., Thill-Baharozian M. & Carlson R. (1989) Triple- vs. single-lumen central venous catheters: a prospective study in a critically ill population. Archives of Internal Medicine 149, 1139 1143. Gupta S., Batra Y., Puri G., Panigrahi D. & Roy S. (1995) Infection rates in single- and double-lumen central venous catheters in critically ill patients. National Medical Journal of India 8, 114117. Halton K. & Graves N. (2007) Economic evaluation and catheterrelated bloodstream infections. Emergency Infectious Disease 13, 815823. Harris A., Lautenbach E. & Prebencevich E. (2005) A systematic review of quasi-experimental study designs in the fields of infection control and antibiotic resistance. Clinical Infectious Diseases 41, 7782. Hasaniya N., Angelis M., Brown M. & Yu M. (1996) Efficacy of subcutaneous silver-impregnated cuffs in preventing central venous catheter infections. Chest 109, 10301032. Higuera F., Rosenthal V., Duarte P., Ruiz J., Franco G. & Safdar N. (2005) The effect of process control on the incidence of central venous catheter-associated bloodstream infections and mortality in intensive care units in Mexico. Critical Care Medicine 33, 2022 2027. Hilton E., Haslet T., Borenstein M., Tucci V. & Isenberg H. (1988) Central venous infections: single- versus triple-lumen catheters: influence of guidewires on infection rates when used for replacement of catheters. American Journal of Medicine 84, 667672. Humar A., Ostromecki A., Direnfeld J., Marshall J., Lazar N., Houston P., Boiteau P. & Conly J. (2000) Prospective randomized trial of 10% povidone-iodine versus 0.5% tincture of chlorhexidine as cutaneous antisepsis for prevention of central venous catheter infection. Clinical Infectious Diseases 31, 10011007. Leon C., Alvarez Lerma F., Ruiz Santana S., Gonzalez V., De La Torre M., Sierra R., Leon M. & Rodrigo J. (2003) Antiseptic chambercontaining hub reduces central venous catheter-related infection: a prospective, randomized study. Critical Care Medicine 31, 1318 1324. Lobo R., Levin A., Gomes L., Cursino R., Park M., Figueiredo V., Taniguchi L., Polido C. & Costa S. (2005) Impact of an educational program and policy changes on decreasing catheter-associated bloodstream infections in a medical intensive care unit in Brazil. American Journal of Infection Control 33, 8387. Lucet J., Hayon J., Bruneel F., Dumoulin J. & Joly-Guillou M. (2000) Microbiological evaluation of central venous catheter administration hubs. Infection Control and Hospital Epidemiology 21, 40 42. Luna J., Masdeu G., Perez M., Claramonte R., Forcadell J., Barrachina F. & Panisello M. (2000) Clinical trial evaluating a new hub device designed to prevent catheter-related sepsis. European Journal of Clinical Microbiology and Infectious Diseases 19, 655662. Maki D., Weise C. & Sarofin H. (1977) A semiquantitative culture method for identifying intravenous catheter-related infection. New England Journal of Medicine 296, 13051309. Maki D., Ringer M. & Alvarado C. (1991) Prospective randomised trial of povidone-iodine, alcohol, and chlorhexidine for prevention of infection associated with central venous and arterial catheters. Lancet 338, 339343. Marin M., Lee J. & Skurnick J. (2000) Prevention of nosocomial bloodstream infections: effectiveness of antimicrobial-impregnated

2008 The Authors. Journal compilation 2008 Blackwell Publishing Ltd

19

P. Ramritu et al. and heparin-bonded central venous catheters. Critical Care Medicine 28, 33323338. Martin C., Bruder N., Papazian L., Saux P. & Gouin F. (1998) Catheter-related infections following axillary vein catheterization. Acta Anaesthesiologica Scandinavica 42, 5256. McLaws M. & Taylor P. (2003) The hospital infection standardised surveillance (HISS) programme: analysis of a two-year pilot. Journal of Hospital Infection 53, 259267. Mermel L. (2000) Prevention of intravacular catheter-related infections. Annals Internal Medicine 132, 391402. Merrer J., De Jonghe B., Golliot F., Lefrant J., Raffy B., Barre E., Rigaud J., Casciani D., Misset B., Outin Bosquet C., Brun-Buisson C., Nitenberg G. & French Catheter Study Group (2001) Complications of femoral and subclavian venous catheterization in critically ill patients - A randomized controlled trial. JAMA 286, 700707. Orsi G., Stefano L. & Noah N. (2002) Hospital-acquired, laboratory-confirmed bloodstream infection: increased hospital stay and direct costs. Infection Control and Hospital Epidemiology 23, 190197. Parienti J., Du Cheyron D., Ramakers M., Malbruny B., Leclercq R., Le Coutour X. & Charbonneau P. (2004) Alcoholic povidone-iodine to prevent central venous catheter colonization: a randomized unit-crossover study. Critical Care Medicine 32, 708713. Petitti D. (2000) Meta-analysis, Decision Analysis, and Cost-Effectiveness Analysis: Methods for Quantitative Synthesis in Medicine. Oxford Univeristy Press, Oxford. Pratt R., Pellowe C., Loveday H., Robinson N., Smith G., Barrett S., Davey P., Harper P., Loveday C., Mcdougall C., Mulhall A., Privett S., Smales C., Taylor L., Weller B. & Wilcox M. (2001) The Epic project. Developing national evidence based guidelines for preventing hospital acquired infections associated with the use of central venous catheters. Technical report Part A & B. Journal of Hospital Infections 47(Suppl), S3S82. Rijnders B. & Van Wijngaerden E. (2002) Catheter-tip colonization as a surrogate end point in clinical studies on catheter-related bloodstream infection: how strong is the evidence? Clinical Infectious Diseases 35, 10531058. Roberts B. & Cheung D. (1998) Biopatcha new concept in antimicrobial dressings for invasive devices. Australian Critical Care 11, 1619. Rosenthal V., Guzman S. & Migone O. (2003a) The attributable cost, length of stay, and mortality of central line-associated bloodstream infection in intensive care departments in Argentina: a prospective, matched analysis. American Journal of Infection Control 31, 475480. Rosenthal V., Guzman S., Pezzotto S. & Crnich C. (2003b) Effect of an infection control program using education and performance feedback on rates of intravascular device-associated bloodstream infections in intensive care units in Argentina. American Journal of Infection Control 31, 405409. Scottish Intercollegiate Guidelines Network (2001) SIGN 50: A Guideline Developers Handbook. Royal College of Physicians of Edinburgh, Edinburgh. Available at http://www.sign.ac.uk. accessed on 5 February 2007. Veenstra D., Saint S., Saha S., Lumley T. & Sullivan S. (1999) Efficacy of antiseptic-impregnated central venous catheters in preventing catheter-related bloodstream infection: a meta-analysis. JAMA 281, 261267. Walder B., Pittet D. & Tramer M. (2002) Prevention of bloodstream ` infections with central venous catheters treated with anti-infective agents depends on catheter type and insertion time: evidence from a meta-analysis. Infection Control and Hospital Epidemiology 23, 748756. Warren D., Zack J., Cox M., Cohen M. & Fraser V. (2003) An educational intervention to prevent catheter-associated bloodstream infections in a nonteaching, community medical center. Critical Care Medicine 31, 19591963. Warren D., Zack J., Mayfield J., Chen A., Prentice D., Fraser V. & Kollef M. (2004) The effect of an education program on the inciderce of central venous catheter-associated bloodstream infection in a medical ICU. Chest 126, 16121618. Yebenes J., Vidaur L., Serra-Prat M., Sirvent J., Batlle J., Motje M., Bonet A. & Palomar M. (2004) Prevention of catheter-related bloodstream infection in critically ill patients using a disinfectable, needle-free connector: a randomized controlled trial. American Journal of Infection Control 32, 291295.

20

2008 The Authors. Journal compilation 2008 Blackwell Publishing Ltd

JAN: REVIEW PAPER

Catheter-related bloodstream infections in intensive care units

Appendix 1
Details of data extracted.
Clinical data: experimental and observational studies Study aim/objective Denitions of study outcomes Study design: type, appropriateness to aim Study setting/characteristics of sample/inclusion criteria/sample size; study dates Details of intervention type, regimen, co-interventions Outcomes baseline measures, details of measurement, summary results Duration of follow-up Investigators conclusions, reviewer agreements and any reservations Hand search of reference list Data analysis Preintervention comparability table Similarity of subjects at start of study Quality: experimental studies Method of allocation, randomization; blinding Sample size power calculation Equal treatment of groups other than intervention Bias due to drop out, characteristics and reasons for withdrawals Appropriateness of statistical analysis, intention to treat or as treated Dealing with confounding variables Quality Observational studies Method of sample selection Comparability of groups at baseline Drop-out rate Details of withdrawal Appropriateness of statistical analysis Summary of reviewers judgement for experimental and observational studies Soundness of methodology design Generalizability of ndings Clinical importance of outcomes Harm/benet analysis Risk of bias Need for author contact for further information/ statistical assessment required Acceptance of paper into review Study grading for experimental studies Category 1: study well designed experimental study; ndings generalizable Category 2: acceptable design, in absence of category 1, accept study into review Category 3: methodologically fatally awed, reject study. Study grading for observational studies Category 1: well designed observational study, ndings generalizable Category 2: acceptable design, in absence of category 1, accept study into review Category 3: methodologically fatally awed, reject study. Sources: (Scottish Intercollegiate Guidelines Network 2001) & (Pratt et al. 2001).

2008 The Authors. Journal compilation 2008 Blackwell Publishing Ltd

21