MAJOR ARTICLE

Antimicrobial Lock Solutions for the Prevention of Infections Associated with Intravascular Catheters in Patients Undergoing Hemodialysis: Systematic Review and Meta-analysis of Randomized, Controlled Trials
Dafna Yahav,1,5,a Benaya Rozen-Zvi,2,5,a Anat Gafter-Gvili,3,5 Leonard Leibovici,1,5 Uzi Gafter,2 and Mical Paul4,5
Departments of 1Medicine E, 2Nephrology, and 3Hematology and 4Infectious Disease Unit, Rabin Medical Center, Beilinson Campus, Petah-Tiqva, and 5Sackler School of Medicine, Tel Aviv University, Ramat Aviv, Israel

Downloaded from http://cid.oxfordjournals.org/ by guest on September 8, 2013

Background. Prevention of catheter-related bloodstream infections in patients undergoing hemodialysis by use of antimicrobial catheter lock solutions has been examined in several trials, but no consensus is available for clinical practice. Methods. A systematic review and meta-analysis were performed of randomized controlled trials that compared single or combination antimicrobial catheter lock solutions with heparin or another antimicrobial for the prevention of infections in patients undergoing hemodialysis. The primary outcomes assessed were bloodstream infections, catheter-related bloodstream infections, and the need for catheter removal. Relative risks with 95% condence intervals (CIs) for individual trials were pooled. Results. Eleven trials (924 patients) that assessed antibiotic catheter lock solutions and 5 trials (661 patients) that assessed nonantibiotic antimicrobial catheter lock solutions met inclusion criteria. None of the trials assessed all bloodstream infections. Antibiotic catheter lock solutions signicantly reduced catheter-related bloodstream infections (relative risk, 0.44; 95% CI, 0.380.50). Signicant heterogeneity for this outcome could be explained by smaller effect estimates in larger trials that reported adequate randomization methods (relative risk, 0.60; 95% CI, 0.540.67). Efcacy was higher when additional preventive measures were used and to prevent the rst episode of catheter-related bloodstream infection. Catheter removal rates were signicantly reduced (relative risk, 0.35; 95% CI, 0.230.55). Resistance development was documented in a single patient. Data concerning nonantibiotic antimicrobial lock solutions were limited and heterogenous. High-quality trials that used additional preventive measures showed a signicant reduction in catheter-related bloodstream infections (relative risk, 0.25; 95% CI, 0.130.50). Conclusions. Antibiotic catheter lock solutions reduce catheter-related bloodstream infections, with a number needed to treat of 4 patients (95% CI, 45), and catheter removal rates in patients undergoing hemodialysis. The use of antibiotic catheter lock solutions should be considered in routine clinical practice in conjunction with other prevention modalities. Central venous catheterrelated bloodstream infection is an important cause of morbidity and mortality among patients undergoing hemodialysis. The incidence of dialysis-related, catheter-related bloodstream infections is reported to be 2.55.5 cases per 1000 catheter-days or 0.92 episodes per patient-year in patients using catheters as their vascular access [1]. According to the US Renal Data System, infection is the second leading cause of death in patients with end-stage renal disease [2]. Several modalities have been assessed as a means to prevent catheter-related infections. Tunneling of central venous catheters [3, 4] and cufng of the catheter hub have been shown to reduce the rate of catheter-related bloodstream infections [5]. Topical agents applied to the catheter exit site, such as povidone iodine, mupirocin, and bacitracin zinc and polymyxin B sulfate ointCatheter Lock Solutions for Hemodialysis CID 2008:47 (1 July) 83

Received 3 December 2007; accepted 22 February 2008; electronically published 22 May 2008. a D.Y. and B.R.-Z. contributed equally to this article. Reprints and correspondence: Dr. Dafna Yahav, Dept. of Medicine E, Rabin Medical Center, Beilinson Campus, 49100 Petah-Tiqva, Israel (dafna.yahav @gmail.com). Clinical Infectious Diseases 2008; 47:8393 2008 by the Infectious Diseases Society of America. All rights reserved. 1058-4838/2008/4701-0014$15.00 DOI: 10.1086/588667

ments, have been proven effective [610]. Oral rifampin or nasal mupirocin ointment reduced the incidence of Staphylococcus aureus bacteremia [1113]. Silver coating of tunneled catheters was not shown to reduce the incidence of catheterrelated bloodstream infections in patients undergoing hemodialysis, because biolm formation was not prevented [14, 15], but antibiotic coating of uncuffed catheters was benecial in patients with cancer experiencing acute renal failure [16]. Instilling an antimicrobial solution into the catheter lumen limits biolm formation [1719]. The use of antimicrobial catheter lock solutions for the prevention of catheter-related bloodstream infections in hemodialysis has been examined in several randomized controlled trials, but to date no consensus is available for clinical practice [20, 21]. We therefore performed a systematic review of all randomized controlled trials that assessed antimicrobial catheter lock solutions for the prevention of catheter-related infections in patients undergoing hemodialysis. METHODS Inclusion criteria and outcomes. We included randomized controlled trials that assessed adults or children with end-stage renal disease undergoing hemodialysis through central catheters. We allowed up to 30% of patients with acute renal failure to be included in the study. We included trials that compared any single or combination antimicrobial (antibiotic, citrate, taurolidine, or alcohol) catheter lock solution with heparin or another antimicrobial catheter lock solution. All antimicrobial solutions could be given with heparin or EDTA. Analyses were separated by antibiotic and nonantibiotic antimicrobial locks. The primary outcomes assessed were clinically signicant bacteremia of any cause, catheter-related bloodstream infections, and removal of the catheter for any cause. Secondary outcomes included all-cause mortality, exit site infections, tunnel infections, catheter thrombosis (dened as thrombosis or need for thrombolytic therapy or removal of the catheter because of ow problems), need for hospitalization and hospitalization duration, emergence of resistance (dened as development of catheter-related bloodstream infections or colonization by bacteria resistant to the antibiotic given), fungal infections, and adverse events. Search strategy and selection criteria. We searched the Cochrane Register of Controlled Trials (Central) and PubMed databases. Unpublished trials were sought in references of all selected studies, relevant conference proceedings, trial registries, ongoing trial databases, new drug application documents, and personal contacts with the investigators of included trials. No language or date restrictions were imposed. The last search was performed in November 2007. The terms lock*, ush, solution*, antibiotic*, and specic lock solutions were combined with the medical subject heading terms dialysis or kidney disease and
84 CID 2008:47 (1 July) Yahav et al.

intravascular catheters. All were combined with the Cochrane lter for randomized controlled trials in PubMed [22]. Study selection and data extraction. Two reviewers (D.Y. and B.R.-Z.) independently performed the search, applied inclusion criteria, and extracted the data. Outcomes were extracted preferentially by intention to treat. For the primary outcomes, we extracted the number of episodes both per patients and per catheter-days. Study follow-up was documented, and outcomes were extracted at the end of follow-up. We extracted data on baseline patient characteristics, catheter characteristics, and additional prophylactic measures used that may have affected outcomes. Missing data were requested from the authors. Methodologic quality assessment was performed using the individual component approach, and its effect was assessed through sensitivity analyses. Allocation concealment and generation were graded as adequate, unclear, or inadequate, by use of criteria suggested in the Cochrane handbook [22]. Statistical analysis. Relative risks with 95% CIs were calculated for individual trials. Meta-analysis was performed using the Mantel-Haenszel xed-effects model (Review Manager 4.2). Heterogeneity in the results of the trials was assessed using the x2 test for heterogeneity and the I 2 measure of inconsistency [23]. Substantial heterogeneity (I 2 1 50% ) was investigated through subgroup analyses and metaregression (Comprehensive Meta-analysis 2.2) for catheter tunneling and incident (!90 days) versus prevalent (longer-term) hemodialysis. A funnel plot was visually inspected, and the small-studies effect was formally assessed using Eggers test of the intercept. RESULTS The search yielded 162 potentially relevant studies, of which 127 were irrelevant. Thirty-ve studies were further evaluated, of which 19 were excluded [2442] (gure 1). Sixteen randomized controlled trials, conducted from 1998 through 2006, were included in the review [4358] (table 1). Eleven trials assessed antibiotic catheter lock solutions. The antibiotics assessed were gentamicin in 3 trials [46, 52, 58], gentamicin and citrate in 3 trials [47, 53, 54], gentamicin and vancomycin in 1 trial [43], gentamicin and cefazolin in 1 trial [50], cefotaxime in 2 trials [55, 56], and minocycline with EDTA in 2 trials [45, 53]. These trials included 924 patients with 176,332 catheter-days. Eight trials included tunneled catheters, 2 included nontunneled catheters, and 1 included both types. Five trials assessed nonantibiotic antimicrobial catheter lock solutions, including citrate in 4 trials [48, 49, 51, 57] and citrate with taurolidine in 1 trial [44]. We did not identify trials that assessed alcohol lock therapy for prevention. The trials included 661 patients with 63,345 catheter-days. Three trials included tunneled catheters, and 2 included both tunneled and nontunneled catheters.

Downloaded from http://cid.oxfordjournals.org/ by guest on September 8, 2013

Figure 1. Flowchart for selection of trials

All trials compared the intervention with heparin. The follow-up duration ranged from 6 to 16 months. None of the trials assessed bloodstream infections of any cause. The need for hospitalization was reported in only 2 trials, whereas none of the trials reported the length of hospitalization. All-cause mortality at the end of follow-up was reported in 7 trials. Adequate allocation concealment and allocation-sequence generation were described in 7 trials. Three additional trials described only adequate allocation generation, and another trial described only adequate allocation concealment. Six trials were double-blind, whereas all others were open-label trials (table 1). Any antibiotic lock versus heparin. The rates of catheterrelated bloodstream infections were signicantly lower with antibiotic catheter lock solutions compared with heparin lock alone both per patient (relative risk, 0.44; 95% CI, 0.380.50; all 11 trials included) and per catheter-day (relative risk, 0.37; 95% CI, 0.300.47) (gure 2). Statistically signicant heterogeneity was present in both analyses. Heterogeneity could not be explained by catheter tunneling, incident hemodialysis (reported in 6 trials), diabetes, or denitions used for catheterrelated bloodstream infections. Effect estimates for catheter lock solutions were statistically signicantly higher when additional prophylaxis measures were used, with longer duration of catheter lock solution treatment, when only the rst catheter-related bloodstream infection was assessed and with advancing year of the study (table 2). In addition, the funnel plot revealed a statistically signicant small-studies effect (gure 3), and sensitivity analyses for the trials methodologic quality showed that

adequate randomization methods and double-blinding were consistently associated with effect estimates of lower magnitude. Catheter removal rates were signicantly lower in the intervention group per patient (relative risk, 0.35; 95% CI, 0.23 0.55; 5 trials; 552 patients) and per catheter-day (relative risk, 0.34; 95% CI, 0.210.55; 135,769 catheter-days), without signicant heterogeneity. Catheter survival was reported variously as means or medians or by use of a Kaplan-Meier analysis. Results could not be compiled; however, all trials showed a benet of antibiotic catheter lock solutions, with a median difference among trial arms in the range 425 days. Catheter thrombosis rates were signicantly lower with the intervention (relative risk, 0.48; 95% CI, 0.320.72; 142,271 catheter-days), without signicant heterogeneity. Exit site infections were reduced without statistical signicance per patient (relative risk, 0.69; 95% CI, 0.481.01; 6 trials; 574 patients) and per catheter-day (relative risk, 0.73; 95% CI, 0.481.12; 152,056 catheter-days), without signicant heterogeneity. Tunnel infections were not reported. The emergence of clinically signicant resistant strains was reported in 5 trials, including 316 patients receiving intervention and 211 control patients. Only 1 case of gentamicin-resistant S. aureus was reported in a patient receiving gentamicin and citrate during 16 months of follow-up. Colonization rates were not assessed. No fungal infections in the intervention group were documented in 7 trials. Adverse events were addressed by 5 trials that included 425 patients. No adverse events occurred in 3 trials, whereas 2 trials that assessed gentamicin and citrate reported 1 case of rash that required treatment discontinuation [53] and 4 cases of dizziness without vertigo [47] among patients receiving intervention. Overall mortality was reported in 5 comparisons. No deaths occurred in 3 trials, and in 1 trial, including 2 comparisons, no statistically signicant difference was found among the trial arms. Because we combined antibiotic regimens with a heterogeneous spectrum of coverage in the main analysis, we separated the analysis for trials that assessed gentamicin alone or with citrate and other trials that assessed gentamicin combined with vancomycin or cefazolin, cefotaxime alone, or minocycline (table 3). Outcomes with gentamicin as a single antibiotic were not inferior to those obtained from trials that used additional or other antibiotics. Seven trials reported the specic isolates that caused catheter-related bloodstream infections. Both gentamicin and other antibiotics statistically signicantly reduced the rates of catheter-related bloodstream infections caused by gram-negative and gram-positive bacteria. Both gentamicin and other antibiotics reduced S. aureus catheter-related bloodstream infections, with only the latter achieving statistical signicance. Six trials specically reported methicillin-resistant S. aureus infections. Three methicillin-resistant S. aureus catheter-related
Catheter Lock Solutions for Hemodialysis CID 2008:47 (1 July) 85

Downloaded from http://cid.oxfordjournals.org/ by guest on September 8, 2013

Table 1.

Summary of clinical trials on catheter lock solutions, 19982006.

Study

Interventions assessed

Location

Incident hemodialysis, No. of No. of No. of Mean age, Diabetes, Tunneled % Study years patients catheters catheter-days years % catheters, %

Additional preventive measures

Antibiotic catheter lock solutions Vancomycin (25 mg/mL) plus Gm (40 mg/mL) Saudi Arabia plus heparin (5000 U/mL) vs. heparin (5000 U/mL) Minocycline (3 mg/mL) plus EDTA (30 mg/ mL) vs. heparin Gm vs. heparin Gm (40 mg/mL) plus citrate (3.13%) vs. heparin (5000 U/mL) Gm (5 mg/mL) plus cefazolin (10 mg/mL) plus heparin (1000 U/mL) vs. heparin (1000 U/mL) Gm (5 mg/mL) plus heparin (5000 U/mL) vs. heparin (5000 U/mL) Gm (4 mg/mL) plus citrate (3.13%) vs. minocycline (3 mg/mL) plus EDTA (30 mg/mL) vs. heparin (5000 U/mL) Gm (18 mg/mL) plus citrate (4.6%) vs. heparin (5000 U/mL) Ctax (10 mg/mL) plus heparin (5000 U/mL) vs. heparin (5000 U/mL) Ctax (10 mg/mL) plus heparin (5000 U/mL) vs. heparin (5000 U/mL) Gm (4 mg/mL) plus heparin (5500 U/mL) vs. heparin (5500 U/mL) China 20042006 101 Saudi Arabia 20022003 118 124 Saudi Arabia 20022003 208 208 93,805 43,435 9300 USA 19992000 36 36 2924 50 USA 20032004 62 6189 58 UK 20022003 50 5722 61 26 55 Korea 20012003 120 4503 55 52.5 0 Australia 19992001 83 112 5923 58 34.7 100 USA 36 3095 100 USA 19981999 60 4454 54 38.9 86 0 40 20052006 63 81 14,867 47 20.4 100 Exit site topical iodine/ chlorhexidine Nasal mupirocin, exit site topical iodine/ chlorhexidine 100 B, B, OL

Al-Hwiesh et al., 2007 [43]

Bleyer et al., 2005 [45]

A, A DB B, B, OL A, A, DB

Cooper et al., 1999 [46]

Dogra et al., 2002 [47]

Kim et al., 2006 [50]

A, A, DB

McIntyre et al., 2004 [52]

100 100

0 15

B, A, OL A,B OL

Nori et al., 2006 [53]

Downloaded from http://cid.oxfordjournals.org/ by guest on September 8, 2013

86
Citrate (4%) plus taurolidine (1.35%) vs. heparin (5000 U/mL) Citrate (46.7%) vs. heparin (5000 U/mL) Citrate (5%) vs. heparin (5000 U/mL) Citrate (4%) vs. heparin (5000 U/mL) Citrate (30%) vs. heparin (5000 U/mL) Canada Netherlands Belgium UK 20002000 20012002 Netherlands 20022003 58 67 3404 232 19 61 291 291 36,108 3196 4090 16,547

Pervez et al., 2002 [54]

41.7

100

Exit site topical iodine, plastic bag over catheter hub 48 77 29.3 35.6 0 100 100 42

A, B, OL

Saxena et al., 2005 [55]

A, A, DB A, A, DB A, B, OL

Saxena et al., 2006 [56]

Zhang et al., 2006 [58]

Nonantibiotic antimicrobial catheter lock solutions 54 27.7 23.7 Nasal mupirocin, exit site topical iodine/ chlorhexidine 70 66 61 52.5 29.5 100 100 100 33.7 0 0 Nasal mupirocin, exit site topical iodine/ chlorhexidine A, A, OL

Betjes et al., 2004 [44]

Duncan et al., 2005 [48]

B, B, OL B, B, OL B, B, OL A, A, DB

Hendricks et al., 2001 [49]

MacRae et al., [51]

Weijmer et al., 2005 [57]

NOTE. Ctax, cefotaxime; Gm, gentamicin.

The rst letter is a grade for the quality of allocation generation, and the second letter is a grade for the quality of allocation concealment; both are graded as A for adequate or B for unclear or unknown methods. DB, double-blind; OL, open-label. b The original trial had 3 arms; a plastic bug over the catheter hub was added to both arms described here. A third arm evaluated heparin treatment alone.

Downloaded from http://cid.oxfordjournals.org/ by guest on September 8, 2013

Figure 2. Catheter-related bloodstream infections per patient (A) and per catheter-day (B) for trials that compared any antibiotic catheter lock solution with heparin. The analysis is subcategorized by the concomitant use of citrate with the antibiotic. A relative risk (RR) of !1.0 favors the antimicrobial lock solution examined. Studies are given as the rst authors surname followed by the last 2 digits of the year published and an abbreviation for the substance studied: CefaGH, cefazolin-gentamicin-heparin; CefoH, cefotaxime-heparin; G, gentamicin; GC, gentamicin-citrate; GCbag, gentamicincitrate plastic bag over catheter hub; GH, gentamicin-heparin; ME, minocycline-EDTA; VGH, vancomycin-gentamicin-heparin.
87

Table 2. Effect of study characteristics on the relative risk (RR) for catheter-related bloodstream infections per patient in trials comparing an antibiotic catheter lock with heparin alone.
Studied patient characteristic Catheter tunneling Tunneled Nontunneled Topical exit-site antibiotics Used Not used Mupirocin nasal decontamination Used Not used Lock duration Repeat catheter-related bloodstream infections included in outcome assessment No Yes or not dened Diabetes mellitus, % Study size Above median Below median Allocation concealment Adequate Not reported Allocation generation Adequate Not reported Double-blind Yes No Year end Denition of catheter-related bloodstream infections CDC criteria Other criteria or not dened RR, 0.59; 95% CI, 0.530.66; 7 trials; P p .001; I2 p 73% RR, 0.20; 95% CI, 0.060.65; 4 trials; P p .89; I2 p 0% RR, 0.61; 95% CI, 0.550.68; 5 trials; P p .31; I2 p 16% RR, 0.14; 95% CI, 0.070.26; 6 trials; P p .91; I2 p 0% 0.769 per 1-year increment .023 NS RR, 0.60; 95% CI, 0.540.67; 8 trials; P p .18; I2 p 31% RR, 0.13; 95% CI, 0.060.27; 3 trials; P p .99; I2 p 0% .026 RR, 0.60; 95% CI, 0.540.67; 6 trials; P p .16; I2 p 37% RR, 0.14; 95% CI, 0.070.28; 5 trials; P p .96; I2 p 0% .016 RR, 0.61; 95% CI, 0.540.68; 5 trials; P p .25; I2 p 26% RR, 0.14; 95% CI, 0.070.27; 6 trials; P p .88; I2 p 0%
!.001
a a

Statistical ndings of study

P NS

RR, 0.54; 95% CI, 0.450.65; 9 trials,; P p .003; I2 p 65% RR, 0.61; 95% CI, 0.530.69; 2 trials; P p .17; I2 p 47%
!.001

RR, 0.15; 95% CI, 0.070.31; 3 trials; P p .52; I2 p 0% RR, 0.60; 95% CI, 0.540.67; 8 trials; P p .11; I2 p 40% NS RR, 0.06; 95% CI, 00.99; 1 trial RR, 0.59; 95% CI, 0.530.65; 10 trials; P p .01; I2 p 61% 1.00014 per 1-month increment RR, 0.12; 95% CI, 0.040.41; 4 trials; P p .88; I2 p 0% RR, 0.59; 95% CI, 0.530.66; 7 trials; P p .005; I2 p 68% 1.003 per 1% increment 1.013 per 1% increment NS NS .033 .01

Incidence hemodialysis, %

Downloaded from http://cid.oxfordjournals.org/ by guest on September 8, 2013

!.001

NOTE. CDC, Centers for Disease Control and Prevention; NS, not statistically signicant. On the basis of 9 trials reporting the percentage of patients with diabetes mellitus and 6 trials reporting the percentage of patients who underwent incident hemodialysis.
a

bloodstream infections occurred in trials that assessed gentamicin (2 trials), 2 with heparin treatment and 1 with gentamicin treatment, whereas no cases occurred in trials that assessed other antibiotics. Nonantibiotic antimicrobial locks versus heparin. Rates of catheter-related bloodstream infections were signicantly lower with antimicrobial catheter lock solutions than with heparin alone per patient (relative risk, 0.46; 95% CI, 0.290.71; 4 trials; 642 patients) and per catheter-day (relative risk, 0.48; 95% CI, 0.300.76; 60,149 catheter-days) (gure 4). Statistically significant heterogeneity was present overall and among trials that assessed citrate, explained by the differential use of additional prophylactic measures. Two trials used mupirocin nasal decontamination and topical iodine at the exit site, and their relative risk was 0.25 (95% CI, 0.130.50), whereas 2 trials that did not use additional prophylactic measures did not show a signicant advantage to catheter lock solutions (relative risk,
88 CID 2008:47 (1 July) Yahav et al.

0.90; 95% CI, 0.481.69; P p .01 for the difference). The same trials that used mupirocin and iodine reported adequate allocation concealment and generation, and 1 trial was double blinded. Relative risks were similar for tunneled and nontunneled catheters. Catheter removal rates were signicantly lower in the study group per patient (relative risk, 0.61; 95% CI, 0.450.81; 3 trials; 581 patients) and per catheter-day (relative risk, 0.62; 95% CI, 0.440.88; 56,089 catheter-days), without signicant heterogeneity. No difference was found between study groups with regard to catheter thrombosis (relative risk, 0.97; 95% CI, 0.72 1.30; 4 trials; 27,237 catheter-days), without signicant heterogeneity. Catheter survival at 3 months was higher with catheter lock solutions in 2 trials that assessed citrate (relative risk, 1.28; 95% CI, 1.041.57). Exit site infections were signicantly reduced by catheter lock solutions per patient (relative risk, 0.58; 95% CI, 0.400.84; 5 trials; 661 patients) and per catheter-day

Figure 3. Funnel plot of the SE of the log relative risk (RR) versus RRs for catheter-related bloodstream infections per patient, showing a statistically signicant small-studies effect (Eggers intercept, 1.51312; 95% CI, 2.29929 to 0.72694; t p 4.35388; 9 df; P ! .001).

Downloaded from http://cid.oxfordjournals.org/ by guest on September 8, 2013

(relative risk, 0.60; 95% CI, 0.400.90; 63,345 catheter-days). Tunnel infections were not reported. Adverse events were addressed by 4 trials. No events were reported in the trial that assessed citrate and taurolidine [44]. Two trials reported a higher frequency of adverse events, mainly bleeding, in the control group [51, 57], and 1 trial reported a higher rate of treatment discontinuations with citrate (20 of 132 patients) compared with heparin (0 of 100 patients), without specifying the causes [48]. Overall mortality was reported in the 2 high-quality trials, without a signicant difference (relative risk, 0.65; 95% CI, 0.341.24). When all trials that assessed citrate were pooled (alone, with gentamicin, or with taurolidine; 7 trials; 818 patients; 75,185 catheter-days), signicant reductions were found in catheterrelated bloodstream infections (relative risk, 0.36; 95% CI, 0.240.55), the need for catheter removal (relative risk, 0.56; 95% CI, 0.420.74), and exit site infections (relative risk, 0.34; 95% CI, 0.190.61) per patient (similar results per catheter day; data not shown). No difference was found in the incidence of catheter thrombosis. DISCUSSION The objective of our systematic review was to assess the efcacy of antimicrobial lock solutions for the prevention of catheterassociated infections in patients undergoing hemodialysis. We assessed separately antibiotic locks and other nonantibiotic antimicrobial locks. Antibiotic catheter lock solutions signicantly reduced catheter-related bloodstream infection rates (relative risk, 0.44; 95% CI, 0.380.50) and catheter removal rates (relative risk, 0.35; 95% CI, 0.230.55) compared with heparin lock alone. All studies showed a benet for lock solutions with regard to catheter-related bloodstream infections. However, the size of the

effect varied signicantly, with some studies demonstrating large effect estimates (risk reduction of up to 93%) and others demonstrating more modest protective effects. Catheter lock solutions were more effective when additional preventive measures, such as nasal decontamination with mupirocin and iodine dressing at the exit site, were used. A statistically signicant effect of trial methodologic quality and size on effect estimates was found, with smaller studies and those not reporting adequate randomization methods exaggerating the benecial effect of antibiotic lock solutions. However, the protective effect remained statistically signicant, although of lower magnitude, among the other trials as well. Only 5 trials compared nonantibiotic catheter lock solutions, mainly citrate, with heparin. When the catheter lock solution was used in conjunction with nasal mupirocin and exit site iodine dressing in good-quality trials, the intervention was highly effective (relative risk, 0.25; 95% CI, 0.130.50). No statistically signicant effect could be shown for nonantibiotic catheter lock solution alone. Citrate catheter lock solution, with or without antibiotics, was assessed in 7 trials and statistically signicantly reduced infection and catheter-removal rates, with no effect on catheter thrombosis. Taurolidine was assessed in a single small randomized trial (58 patients), which showed nonsignicant reductions in infection-related outcomes. Our results are in accordance with a previous systematic review that assessed vancomycin catheter lock solutions in patients not undergoing hemodialysis [59]. Safdar and Maki demonstrated that catheter lock solutions retained in the catheter, as for patients undergoing hemodialysis, provide signicantly higher protection than ush catheter lock solutions. Various antibiotic regimens were used in the trials included in our review. Gentamicin alone was as effective as combination therapy or other broader-spectrum antibiotic catheter lock soluCatheter Lock Solutions for Hemodialysis CID 2008:47 (1 July) 89

Table 3. Effects of catheter lock solutions using gentamicin as a single antibiotic compared with other single or combination antibiotics.
Comparison No. of trials [references] No of patients No. of catheter-days No. of catheter-related bloodstream infections (95% CI) Per patient Per catheter-day c No. of catheter-related bloodstream infections caused (95% CI) By gram-positive bacteria By gram-negative bacteria By Staphylococcus aureus No. of catheters removed (95% CI) Per patient Per catheter-day No. of catheter thrombosis events per catheter-day (95% CI) No. of exit site infections (95% CI) Per patient Per catheter-day Overall mortality (95% CI)
a b

Gentamicin alone 6 [46, 47, 5254, 58] 332 29,833 0.12 (0.050.33) 0.09 (0.030.26) 0.11 (0.020.54) 0.10 (0.020.52) 0.22 (0.041.26) 0.23 (0.050.98) 0.19 (0.040.87) 0.66 (0.301.47) 0.21 (0.041.21) 0.18 (0.031.12) 1.00 (0.109.75)

Other antibiotics

6 [43, 45, 50, 53, 55, 56] 592 146,499 0.49 (0.430.57) 0.42 (0.330.54) 0.38 (0.260.56) 0.36 (0.230.56) 0.41 (0.230.73) 0.37 (0.240.59) 0.37 (0.220.61) 0.42 (0.260.69) 0.76 (0.521.12) 0.82 (0.521.27) 1.50 (0.0733.89)

Downloaded from http://cid.oxfordjournals.org/ by guest on September 8, 2013

Including gentamicin combined with vancomycin or cefazolin, cefotaxime alone, and minocycline alone. Nori et al. [53] randomized patients into 3 arms: gentamicin-citrate treatment, minocycline-EDTA treatment, and heparin treatment; these patients are included in both analyses. c Data on specic pathogens causing catheter-related bloodstream infections per study group were available from 3 trials that assessed gentamicin alone [47, 52, 58] and 4 trials that assessed other antibiotics [45, 50, 53, 56].

tions and prevented equally gram-positive and gram-negative infections. Minocycline-EDTA was also effective, but the number of evaluated patients was small (88 patients in 2 trials). Gentamicin and minocycline have a broad spectrum of activity against gram-negative and gram-positive bacteria [60, 61], especially with drug levels achieved in the catheter lumen. An advantage of both is that they are not used empirically for the treatment of sepsis in patients undergoing hemodialysis [62]. In vitro and animal studies show good efcacy of minocycline plus EDTA in prevention of colonization and eradication of the relevant bacteria in biolms, whereas vancomycin was found to be less effective [1719]. Concerns regarding the use of gentamicin locks include potential ototoxicity. In our review, no cases of ototoxicity were reported. So far, only 1 case of ototoxocity with use of an amikacin lock has been reported [63]. Development of resistance is the major factor against the use of antibiotic catheter lock solutions and the reason that these locks are not routinely recommended [21]. Only 1 case of development of resistance was reported in 5 trials included in our review that addressed this issue, after a follow-up of 12 months in these trials [54]. No surveillance cultures were performed in all trials. In the previous meta-analysis [59], resistance to vancomycin was not encountered in all trials, including 2 trials that performed surveillance cultures. Lack of resistance was attributed to undetectable blood levels of vancomycin in
90 CID 2008:47 (1 July) Yahav et al.

all studies. A follow-up of 1 year in a dialysis unit using gentamicin locks found no increase in resistance rates compared with historical rates [29]. However, all these results do not preclude development of resistance with longer and more extensive use of antibiotic catheter lock solutions. Several limitations of our conclusions should be stated. Bloodstream infections of any cause were not assessed in these trials. We regard this as a limitation because the outcome of catheter-related bloodstream infections might not capture the overall effect of the intervention and may permit a detection bias. Similarly, length of hospital stay was not reported, and all-cause mortality was reported in only 7 of 16 trials. We observed a small-studies effect and potential bias that favored the intervention in open trials that did not report randomization methods. We thus believe that the truer effect estimate for the reduction of catheter-related bloodstream infections is that of the higher-quality trials (relative risk, 0.60; 95% CI, 0.540.67). Finally, despite the inclusion of unpublished trials in our review, we cannot rule out publication bias. Catheter lock solutions for the prevention of infection in patients with central venous catheters are not generally recommended [21]. The Kidney Disease Outcomes Quality Initiative guidelines do not advocate the use of catheter lock solutions for the prevention of catheter-related bacteremia in hemodialysis [20]. The Centers for Disease Control and Prevention guidelines state that catheter lock solutions can be con-

Downloaded from http://cid.oxfordjournals.org/ by guest on September 8, 2013

Figure 4. Catheter-related bloodstream infections per patient (A) and per catheter-day (B) for trials comparing nonantibiotic antimicrobial catheter lock solution with heparin. The analysis is subcategorized by the type of antimicrobial. Studies are given as the rst authors surname followed by the last 2 digits of the year published and an abbreviation for the substance studied: C, citrate; CT, citrate-taurolidine. RR, relative risk.

sidered only under special circumstances, such as multiple catheter-related bloodstream infections, despite maximal adherence to aseptic technique [21]. Our results demonstrate a statistically signicant reduction in catheter-related bloodstream infections, catheter removal, and catheter thrombosis rates with the use of antibiotic locks, without the emergence of resistant organisms. The number of patients needed to treat to prevent 1 catheter-related bloodstream infection on the basis of the relative risk of the higher-quality trials (0.60) is 4 patients (95%

CI, 45 patients), for a median treatment duration of 3 weeks (range, 115 weeks) in these trials. Our results do not support the role of catheter lock solutions for patients with recurrent catheter-related bloodstream infections, because the effect of antibiotic catheter lock solutions was seen mainly for the rst episode of catheter-related bloodstream infection. We therefore conclude that antibiotic catheter lock solutions should be considered in routine clinical practice, with the caveat that development of resistance might not have been adequately assessed.
Catheter Lock Solutions for Hemodialysis CID 2008:47 (1 July) 91

Catheters constitute the least desired method of vascular access for long-term hemodialysis. The US National Kidney Foundation has dened the reduction of catheter use as a goal for hemodialysis centers [20]. For those patients with no alternative vascular access, use of catheter lock solutions should be considered in conjunction with other proven infection prevention and control modalities. On the basis of our subgroup analysis and considerations of antibiotic resistance, antibiotic catheter lock solutions might be most benecial to prevent the rst episode of catheter-related bloodstream infection in patients receiving short-term catheter hemodialysis, such as patients undergoing incident hemodialysis. Locks that contain gentamicin or minocycline plus EDTA are as efcacious as broader-spectrum regimens. Currently, evidence does not show the efcacy of citrate or citrate-taurolidine catheter lock solutions as single agents, when applied without other preventive measures. Further studies are needed to determine their efcacy compared with heparin and antibiotics, stratifying for preventive measures with proven efcacy in infection prevention. Induction of resistance to antibiotic catheter lock solutions should be assessed in prospective long-term studies.
Acknowledgments
We thank the authors who responded to our letters and supplied additional data on their trials. Potential conicts of interests. All authors: no conicts.

References
1. Katneni R, Hedayati SS. Central venous catheter-related bacteremia in chronic hemodialysis patients: epidemiology and evidence-based management. Nat Clin Pract Nephrol 2007; 3:25666. 2. US Renal Data System. USRDS 2003 annual data report: atlas of endstage renal disease in the United States. Bethesda, MD: National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases, 2003. 3. Randolph AG, Cook DJ, Gonzales CA, Brun-Buisson C. Tunneling short-term central venous catheters to prevent catheter-related infection: a meta-analysis of randomized, controlled trials. Crit Care Med 1998; 26:14527. 4. Timsit JF, Bruneel F, Cheval C, et al. Use of tunneled femoral catheters to prevent catheter-related infection: a randomized, controlled trial. Ann Intern Med 1999; 130:72935. 5. Flowers RH III, Schwenzer KJ, Kopel RF, Fisch MJ, Tucker SI, Farr BM. Efcacy of an attachable subcutaneous cuff for the prevention of intravascular catheter-related infection: a randomized, controlled trial. JAMA 1989; 261:87883. 6. Fong IW. Prevention of haemodialysis and peritoneal dialysis catheter related infection by topical povidone-iodine. Postgrad Med J 1993; 69(Suppl 3):S157. 7. Johnson DW, MacGinley R, Kay TD, et al. A randomized controlled trial of topical exit site mupirocin application in patients with tunnelled, cuffed haemodialysis catheters. Nephrol Dial Transplant 2002; 17:18027. 8. Levin A, Mason AJ, Jindal KK, Fong IW, Goldstein MB. Prevention of hemodialysis subclavian vein catheter infections by topical povidoneiodine. Kidney Int 1991; 40:9348. 9. Lok CE, Stanley KE, Hux JE, Richardson R, Tobe SW, Conly J. Hemodialysis infection prevention with polysporin ointment. J Am Soc Nephrol 2003; 14:16979.

10. Sesso R, Barbosa D, Leme IL, et al. Staphylococcus aureus prophylaxis in hemodialysis patients using central venous catheter: effect of mupirocin ointment. J Am Soc Nephrol 1998; 9:108592. 11. Boelaert JR, De Smedt RA, De Baere YA, et al. The inuence of calcium mupirocin nasal ointment on the incidence of Staphylococcus aureus infections in haemodialysis patients. Nephrol Dial Transplant 1989; 4: 27881. 12. Tacconelli E, Carmeli Y, Aizer A, Ferreira G, Foreman MG, DAgata EM. Mupirocin prophylaxis to prevent Staphylococcus aureus infection in patients undergoing dialysis: a meta-analysis. Clin Infect Dis 2003;37: 162938. 13. Yu VL, Goetz A, Wagener M, et al. Staphylococcus aureus nasal carriage and infection in patients on hemodialysis: efcacy of antibiotic prophylaxis. N Engl J Med 1986; 315:916. 14. Dahlberg PJ, Agger WA, Singer JR, et al. Subclavian hemodialysis catheter infections: a prospective, randomized trial of an attachable silverimpregnated cuff for prevention of catheter-related infections. Infect Control Hosp Epidemiol 1995; 16:50611. 15. Trerotola SO, Johnson MS, Shah H, et al. Tunneled hemodialysis catheters: use of a silver-coated catheter for prevention of infection: a randomized study. Radiology 1998; 207:4916. 16. Chatzinikolaou I, Finkel K, Hanna H, et al. Antibiotic-coated hemodialysis catheters for the prevention of vascular catheter-related infections: a prospective, randomized study. Am J Med 2003; 115:3527. 17. Raad I, Hachem R, Tcholakian RK, Sherertz R. Efcacy of minocycline and EDTA lock solution in preventing catheter-related bacteremia, septic phlebitis, and endocarditis in rabbits. Antimicrob Agents Chemother 2002; 46:32732. 18. Raad I, Hanna H, Jiang Y, et al. Comparative activities of daptomycin, linezolid, and tigecycline against catheter-related methicillin-resistant Staphylococcus bacteremic isolates embedded in biolm. Antimicrob Agents Chemother 2007; 51:165660. 19. Sherertz RJ, Boger MS, Collins CA, Mason L, Raad II. Comparative in vitro efcacies of various catheter lock solutions. Antimicrob Agents Chemother 2006; 50:18658. 20. National Kidney Foundation. National kidney foundation clinical practice guidelines and clinical practice recommendations for 2006 updates: HD adequacy, peritoneal dialysis adequacy and vascular access. Available at: http://www.kidney.org/professional/Kdoqi/guideline_upHD_ PD_VA/index.htm:S1-S322. Accessed 17 May 2008. 21. OGrady NP, Alexander M, Dellinger EP, et al. Guidelines for the prevention of intravascular catheter-related infections. MMWR Recomm Rep 2002; 51:129. 22. Higgins JP, Green S. Cochrane handbook for systematic reviews of interventions 4.2.5: updated May 2005. Chichester, UK: John Wiley & Sons, 2007. 23. Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses. BMJ 2003; 327:55760. 24. Allon M. Prophylaxis against dialysis catheterrelated bacteremia with a novel antimicrobial lock solution. Clin Infect Dis 2003; 36:153944. 25. Ash S, Mannkus R, Sutton J, et al. Concentrated sodium citrate as a catheter lock solution [abstract]. J Am Soc Nephrol 1999; 10:272A. 26. Bleyer A, Mason L, Raad I, Sherertz RJ. A randomized, double-blind trial comparing heparin vs minocycline/EDTA as ush solutions for hemodialysis catheters [abstract P-S1-31]. In: Program and abstracts of the Fourth Decennial International Conference of Nosocomial and Healthcare-Associated Infections (Atlanta). 2000:91. 27. Buturovic J, Ponikvar R, Kandus A, Boh M, Klinkmann J, Ivanovich P. Filling hemodialysis catheters in the interdialytic period: heparin versus citrate versus polygeline: a prospective randomized study. Artif Organs 1998; 22:9457. 28. Chiou PF, Chang CC, Wen YK, Yang Y. Antibiotic lock technique reduces the incidence of temporary catheter-related infections. Clin Nephrol 2006; 65:41922. 29. Fluck R, Taal M, McIntyre C. A single centre, one-year experience with antibiotic locking for tunnelled venous catheters. Nephrol Dial Transplant 2006; 21(Suppl 4):iv45960.

Downloaded from http://cid.oxfordjournals.org/ by guest on September 8, 2013

92 CID 2008:47 (1 July) Yahav et al.

30. Geron R, Tanchilevski O, Kristal B. Catheter lock solution: taurolock for prevention of catheter-related bacteremia in hemodialysis patients [in Hebrew]. Harefuah 2006; 145:881943. 31. Grudzinski L, Quinan P, Kwok S, Pierratos A. Sodium citrate 4% locking solution for central venous dialysis catheters: an effective, more cost-efcient alternative to heparin. Nephrol Dial Transplant 2007; 22: 4716. 32. Jurewitsch B, Lee T, Park J, Jeejeebhoy K. Taurolidine 2% as an antimicrobial lock solution for prevention of recurrent catheter-related bloodstream infections. JPEN J Parenter Enteral Nutr 1998; 22:2424. 33. Lambie SH, Hulme LJ, Taal M, Fluck RJ, McIntyre CW. Prospective study of gentamicin locking of tunnelled dialysis catheters: the effect on infection rates and CRP. Kidney Int 2005; 67:378. 34. Lok CE, Appleton D, Bhola C, Khoo B, Richardson RM. Trisodium citrate 4%: an alternative to heparin capping of haemodialysis catheters. Nephrol Dial Transplant 2007; 22:47783. 35. Lynn RI. Reduced incidence of hemodialysis (HD) catheters related bacteremia (CRB) using a weekly citrate-gentamycin lock [abstract 1006]. J Am Soc Nephrol 2000; 11:189A. 36. Meeus G, Kuypers DR, Claes K, Evenepoel P, Maes B, Vanrenterghem Y. A prospective, randomized, double-blind crossover study on the use of 5% citrate lock versus 10% citrate lock in permanent hemodialysis catheters. Blood Purif 2005; 23:1015. 37. Nori US, Fisher KA, Osman Y, Manoharan A, Besarab A. Efcacy of antibiotic lock solutions in preventing hemodialysis catheter related bacteremia [abstract SU-FC076]. J Am Soc Nephrol 2004; 15:60A. 38. Saxena AK, Panhotra BR, Sundaram DS, et al. Tunneled catheters outcome optimization among diabetics on dialysis through antibioticlock placement. Kidney Int 2006; 70:162935. 39. Schwab SJ, Weiss MA, Rushton F, et al. Multicenter clinical trial results with the LifeSite hemodialysis access system. Kidney Int 2002; 62: 102633. 40. Sodemann K, Lubrich-Birkner I, Berger O, Baumert J, Feldmer B, Von Hodenberg E. Gentamicin/sodium-citrate mixture as antibiotic-lock technique for salvage and prevention of catheter-related infections: a four year trial. J Am Soc Nephrol 1997; 8:173A. 41. Stas KJ, Vanwalleghem J, De Moor B, Keuleers H. Trisodium citrate 30% vs. heparin 5% as catheter lock in the interdialytic period in twinor double-lumen dialysis catheters for intermittent haemodialysis. Nephrol Dial Transplant 2001; 16:15212. 42. Weijmer M, van den Dorpel R, van Geelen J, et al. Substantial reduction of infectious complications in hemodialysis patients with trisodium citrate 30% as catheter locking solution: a prospective multicenter double-blind randomised controlled trial [abstract]. J Am Soc Nephrol 2003; 14:54A. 43. Al-Hwiesh AK, Abdul-Rahman IS. Successful prevention of tunneled, central catheter infection by antibiotic lock therapy using vancomycin and gentamycin. Saudi J Kidney Dis Transpl 2007; 18:23947. 44. Betjes MG, van Agteren M. Prevention of dialysis catheter-related sepsis with a citrate-taurolidine-containing lock solution. Nephrol Dial Transplant 2004; 19:154651. 45. Bleyer AJ, Mason L, Russell G, Raad II, Sherertz RJ. A randomized, controlled trial of a new vascular catheter ush solution (minocyclineEDTA) in temporary hemodialysis access. Infect Control Hosp Epidemiol 2005; 26:5204. 46. Cooper R, Saad T. Prevention of bacteraemia in patients with tunneled cuffed permanent hemodialysis catheters (PCs) using gentamicin catheter packing [abstract]. J Am Soc Nephrol 1999; 10:203A. 47. Dogra GK, Herson H, Hutchison B, et al. Prevention of tunneled hemodialysis catheter-related infections using catheter-restricted lling

48.

49.

50.

51.

52.

53.

54.

55.

56.

57.

58.

59.

60.

61.

62. 63.

with gentamicin and citrate: a randomized controlled study. J Am Soc Nephrol 2002; 13:21339. Duncan N, Singh S, Amao M. A single centre randomized control trial of sodium citrate versus heparin line locks for cuffed central venous catheters. J Am Soc Nephrol 2005; 16:451A. Hendrickx L, Kuypers D, Evenepoel P, Maes B, Messiaen T, Vanrenterghem Y. A comparative prospective study on the use of low concentrate citrate lock versus heparin lock in permanent dialysis catheters. Int J Artif Organs 2001; 24:20811. Kim SH, Song KI, Chang JW, et al. Prevention of uncuffed hemodialysis catheter-related bacteremia using an antibiotic lock technique: a prospective, randomized clinical trial. Kidney Int 2006; 69:1614. MacRae JM, Dojcinovic I, Jung J, Shalansky S, Levin A, Kiaii M. 4% Citrate vs Heparin and the Reduction of Thrombosis Study (CHARTS) [abstract SA-PO057]. In: Programs and abstracts of Renal Week (San Diego). 2006. McIntyre CW, Hulme LJ, Taal M, Fluck RJ. Locking of tunneled hemodialysis catheters with gentamicin and heparin. Kidney Int 2004; 66:8015. Nori US, Manoharan A, Yee J, Besarab A. Comparison of low-dose gentamicin with minocycline as catheter lock solutions in the prevention of catheter-related bacteremia. Am J Kidney Dis 2006; 48:596605. Pervez A, Ahmed M, Ram S, et al. Antibiotic lock technique for prevention of cuffed tunnel catheter associated bacteremia. J Vasc Access 2002; 3:10813. Saxena AK, Panhotra BR. The impact of catheter-restricted lling with cefotaxime and heparin on the lifespan of temporary hemodialysis catheters: a case controlled study. J Nephrol 2005; 18:75563. Saxena AK, Panhotra BR, Sundaram DS, Morsy MN, Al-Ghamdi AM. Enhancing the survival of tunneled haemodialysis catheters using an antibiotic lock in the elderly: a randomised, double-blind clinical trial. Nephrology (Carlton) 2006; 11:299305. Weijmer MC, van den Dorpel MA, Van de Ven PJ, et al. Randomized, clinical trial comparison of trisodium citrate 30% and heparin as catheter-locking solution in hemodialysis patients. J Am Soc Nephrol 2005; 16:276977. Zhang P, Zhang W, He Q, et al. A randomized controlled study on prevention of cuff-tunneled catheter related bacteremia with gentamicin-heparin lock solution: the metaphase result [abstract SA-PO073]. In: Programs and abstracts of Renal Week (San Diego). 2006. Safdar N, Maki DG. Use of vancomycin-containing lock or ush solutions for prevention of bloodstream infection associated with central venous access devices: a meta-analysis of prospective, randomized trials. Clin Infect Dis 2006; 43:47484. Pfaller MA, Jones RN, Doern GV, Sader HS, Kugler KC, Beach ML. Survey of blood stream infections attributable to gram-positive cocci: frequency of occurrence and antimicrobial susceptibility of isolates collected in 1997 in the United States, Canada, and Latin America from the SENTRY Antimicrobial Surveillance Program. SENTRY Participants Group. Diagn Microbiol Infect Dis 1999; 33:28397. Rhomberg PR, Jones RN. Contemporary activity of meropenem and comparator broad-spectrum agents: MYSTIC program report from the United States component (2005). Diagn Microbiol Infect Dis 2007; 57: 20715. Bleyer AJ. Use of antimicrobial catheter lock solutions to prevent catheter-related bacteremia. Clin J Am Soc Nephrol 2007; 2:10738. Saxena AK, Panhotra BR, Naguib M. Sudden irreversible sensory-neural hearing loss in a patient with diabetes receiving amikacin as an antibiotic-heparin lock. Pharmacotherapy 2002; 22:1058.

Downloaded from http://cid.oxfordjournals.org/ by guest on September 8, 2013

Catheter Lock Solutions for Hemodialysis CID 2008:47 (1 July) 93