Portal de Periódicos da CAPES

Comparison of Low-Dose Gentamicin With Minocycline as Catheter Lock Solutions in the Prevention of Catheter-Related Bacteremia

2006; Elsevier BV; Volume: 48; Issue: 4 Linguagem: Inglês

10.1053/j.ajkd.2006.06.012

1523-6838

Uday Nori, Anup Manoharan, Jerry Yee, Anatole Besarab,

Vascular Procedures and Complications

Background: Catheter-restricted antibiotic lock solutions were found to be effective in the prevention of catheter-related bacteremia (CRB), but insufficient data are available about the ideal agent and dose. We hypothesized that a low concentration of gentamicin would be as effective as the high doses studied in the past. Methods: In this prospective, open-labeled, randomized, clinical trial of patients on long-term hemodialysis therapy, patients were randomly assigned to administration of an antibiotic lock solution of gentamicin/citrate (4 mg/mL), minocycline/EDTA, or the control solution of heparin. Patients were followed up until the study end point of CRB was reached or a censoring event occurred. Interim data analysis was performed after 6 months to assess data safety; efficacy was noted and the study was terminated early. Results: Sixty-two patients were enrolled into the study, evenly distributed in 3 arms, with data from 1 patient excluded from analysis. Seven of 20 patients in the heparin group (4.0 events/1,000 catheter days), 1 of 21 patients in the minocycline group (0.4 events/1,000 catheter days), and none of 20 patients in the gentamicin group developed bacteremia. Results were statistically significant by using 2-tailed Fisher exact test; heparin versus gentamicin, P = 0.008, and heparin versus minocycline, P = 0.020. Conclusion: Antibiotic lock solutions are superior to the standard heparin lock alone in the prevention of CRBs, and low-dose gentamicin solution has efficacy similar to that of greater concentrations used in previous studies. Background: Catheter-restricted antibiotic lock solutions were found to be effective in the prevention of catheter-related bacteremia (CRB), but insufficient data are available about the ideal agent and dose. We hypothesized that a low concentration of gentamicin would be as effective as the high doses studied in the past. Methods: In this prospective, open-labeled, randomized, clinical trial of patients on long-term hemodialysis therapy, patients were randomly assigned to administration of an antibiotic lock solution of gentamicin/citrate (4 mg/mL), minocycline/EDTA, or the control solution of heparin. Patients were followed up until the study end point of CRB was reached or a censoring event occurred. Interim data analysis was performed after 6 months to assess data safety; efficacy was noted and the study was terminated early. Results: Sixty-two patients were enrolled into the study, evenly distributed in 3 arms, with data from 1 patient excluded from analysis. Seven of 20 patients in the heparin group (4.0 events/1,000 catheter days), 1 of 21 patients in the minocycline group (0.4 events/1,000 catheter days), and none of 20 patients in the gentamicin group developed bacteremia. Results were statistically significant by using 2-tailed Fisher exact test; heparin versus gentamicin, P = 0.008, and heparin versus minocycline, P = 0.020. Conclusion: Antibiotic lock solutions are superior to the standard heparin lock alone in the prevention of CRBs, and low-dose gentamicin solution has efficacy similar to that of greater concentrations used in previous studies. INFECTION IS A COMMON cause of morbidity and mortality in hemodialysis (HD) patients.1Allon M. Radeva M. Bailey J. et al.HEMO Study GroupThe spectrum of infection-related morbidity in hospitalized haemodialysis patients.Nephrol Dial Transplant. 2005; 20 (Epub March 15, 2005): 1180-1186Crossref PubMed Scopus (52) Google Scholar, 2US Renal Data SystemUSRDS 2003 Annual Data Report. The National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Disease, Bethesda, MD2003Google Scholar Despite clear recommendations for a native arteriovenous fistula as the preferred access for HD patients, international data from the Dialysis Outcomes and Practice Patterns Study3Pisoni R.L. Young E.W. Dykstra D.M. et al.Vascular access use in Europe and the United States: Results from the DOPPS.Kidney Int. 2002; 61: 305-316Crossref PubMed Scopus (722) Google Scholar and US data from the End-Stage Renal Disease Clinical Performance Measures Project4Centers for Medicare & Medicaid Services2005 Annual Report, End-Stage Renal Disease Clinical Performance Measures Project. Department of Health and Human Services, Centers for Medicare & Medicaid Services, Center for Beneficiary Choices, Baltimore, MD2005Google Scholar and regional Network data5Besarab A. Adams M. Amatucci S. et al.Unraveling the realities of vascular access: The Network 11 experience.Adv Ren Replace Ther. 2000; 7: S65-S70PubMed Google Scholar showed consistent and persistent use of catheters in more than 60% of incident and up to 30% of prevalent patients. Perhaps the use of some catheters as vascular access in the United States is unavoidable in a proportion of patients because of failed arteriovenous fistulae and grafts.6Neumann M.E. "Fistula first" initiative pushes for new standards in access care.Nephrol News Issues. 2004; 18 (43): 47-48Google Scholar Twenty-eight percent of long-term dialysis patients use a catheter as permanent HD access.7Rayner H.C. Besarab A. Brown W.W. Disney A. Saito A. Pisoni R.L. Vascular access results from the Dialysis Outcomes and Practice Patterns Study (DOPPS): Performance against Kidney Disease Outcomes Quality Initiative (K/DOQI) Clinical Practice Guidelines.Am J Kidney Dis. 2004; 44: S22-S26Abstract Full Text Full Text PDF PubMed Scopus (186) Google Scholar HD catheters contribute significantly to bacteremia.8Blankestijn P.J. Treatment and prevention of catheter-related infections in haemodialysis patients.Nephrol Dial Transplant. 2001; 16: 1975-1978Crossref PubMed Scopus (33) Google Scholar Septicemia rates in HD patients continue to increase, and hospital admissions for vascular access infection have doubled in the last decade.9US Renal Data SystemUSRDS 2004 Annual Data Report.Am J Kidney Dis. 2005; 45: 8-280Abstract Full Text Full Text PDF Google Scholar The use of cuffed tunneled HD catheters instead of uncuffed catheters has not translated to a significant decrease in the incidence of catheter-related bacteremia (CRB) and resultant infective endocarditis in this population.10Tokars J.I. Miller E.R. Stein G. New national surveillance system for hemodialysis-associated infections: Initial results.Am J Infect Control. 2002; 30: 288-295Abstract Full Text Full Text PDF PubMed Scopus (136) Google Scholar The most devastating complication is the development of endocarditis, to which patients with chronic kidney disease are particularly prone, with a 1-year survival rate of only 62%.11Shroff G.R. Herzog C.A. Ma J.Z. Collins A.J. Long-term survival of dialysis patients with bacterial endocarditis in the United States.Am J Kidney Dis. 2004; 44: 1077-1082Abstract Full Text Full Text PDF PubMed Scopus (42) Google Scholar Increasingly, Staphylococcus aureus is the responsible microbe, typically acquired outside the hospital and increasingly resistant to methicillin.12Fowler Jr, V.G. Miro J.M. Hoen B. et al.the ICE InvestigatorsStaphylococcus aureus endocarditis: A consequence of medical progress.JAMA. 2005; 293 ([Erratum in JAMA 294:900, 2005]): 3012-3021Crossref PubMed Scopus (889) Google Scholar All indwelling vascular catheters are colonized by microorganisms within 24 hours after insertion.13Raad I. Costerton W. Sabharwal U. Sacilowski M. Anaissie E. Bodey G.P. Ultrastructural analysis of indwelling vascular catheters: A quantitative relationship between luminal colonization and duration of placement.J Infect Dis. 1993; 168: 400-407Crossref PubMed Scopus (472) Google Scholar The formation of "biofilm" on external and internal surfaces of vascular catheters is thought to have an important role in the colonization process. Although both external and internal surfaces develop biofilm, the absence of evidence of external site or tunnel infection in the majority of episodes of CRB suggests that the more important biofilm is endoluminal. Internal-surface biofilm also may be modifiable with endoluminal antibiotic locks. The biofilm produced by a combination of host factors and microbial products (eg, glycocalyx, or "slime") has a critical role in bacterial antimicrobial resistance and recalcitrant infections.14Lewis K. Riddle of the biofilm resistance.Antimicrob Agents Chemother. 2001; 45: 999-1007Crossref PubMed Scopus (1468) Google Scholar Such bacteria as S aureus are able to communicate in groups to alter virulence and create biofilm.15Greenberg E.P. Bacterial communication: Tiny teamwork.Nature. 2003; 424: 134Crossref PubMed Scopus (79) Google Scholar Systemic antibiotics used to treat bacteremia do not penetrate the catheter lumen and therefore do not eradicate the biofilm,16Bastani B. Minton J. Islam S. Insufficient penetration of systemic vancomycin into the PermCath lumen.Nephrol Dial Transplant. 2000; 15: 1035-1037Crossref PubMed Scopus (26) Google Scholar leading to potential treatment failures and eventual sacrifice of the catheter. A variety of techniques have been used to prevent CRB, summarized by the Centers for Disease Control and Prevention (CDC).17Centers for Disease Control and PreventionGuidelines for the prevention of intravascular catheter-related infections.MMWR Morb Mortal Wkly Rep. 2002; 51: 1-31Google Scholar A protocol based on Kidney Disease Outcomes Quality Initiative (KDOQI) guidelines focusing on "hub care"18Beathard G.A. Catheter management protocol for catheter-related bacteremia prophylaxis.Semin Dial. 2003; 16: 403-405Crossref PubMed Scopus (56) Google Scholar was able to decrease the incidence of CRB from 6.8 to 1.6 episodes/1,000 catheter-days. Antibiotic lock solutions also were studied in vascular catheters in many clinical settings and successfully prevented CRB, including in HD catheters.19Saxena A.K. Panhotra B.R. Prevention of catheter-related bloodstream infections: An appraisal of developments in designing an infection-resistant 'dream dialysis-catheter.'.Nephrology (Carlton). 2005; 10: 240-248Crossref PubMed Scopus (12) Google Scholar This effectiveness is commensurate with the elimination of biofilm from the catheter lumen.20Andris D.A. Krzywda E.A. Edmiston C.E. Krepel C.J. Gohr C.M. Elimination of intraluminal colonization by antibiotic lock in silicone vascular catheters.Nutrition. 1998; 14: 427-432Abstract Full Text Full Text PDF PubMed Scopus (58) Google Scholar However, debate continues on the appropriate antibacterial agent, as well as its optimal concentration, to eradicate a wide variety of organisms,21Krishnasami Z. Carlton D. Bimbo L. et al.Management of hemodialysis catheter-related bacteremia with an adjunctive antibiotic lock solution.Kidney Int. 2002; 61: 1136-1142Crossref PubMed Scopus (216) Google Scholar, 22Poole C.V. Carlton D. Bimbo L. Allon M. Treatment of catheter-related bacteraemia with an antibiotic lock protocol: Effect of bacterial pathogen.Nephrol Dial Transplant. 2004; 19: 1237-1244Crossref PubMed Scopus (165) Google Scholar, 23Allon M. Saving infected catheters: Why and how.Blood Purif. 2005; 23: 23-28Crossref PubMed Scopus (16) Google Scholar, 24Allon M. Dialysis catheter-related bacteremia: Treatment and prophylaxis.Am J Kidney Dis. 2004; 44: 779-791Abstract Full Text Full Text PDF PubMed Scopus (267) Google Scholar as well as concerns for ultimate bacterial resistance. When used after CRB has occurred, antibiotic lock solutions in conjunction with systemic antibiotics salvage catheters, but the success rate22Poole C.V. Carlton D. Bimbo L. Allon M. Treatment of catheter-related bacteraemia with an antibiotic lock protocol: Effect of bacterial pathogen.Nephrol Dial Transplant. 2004; 19: 1237-1244Crossref PubMed Scopus (165) Google Scholar is still less than that with catheter exchange. Several studies examined primary prevention of CRB by means of intraluminal instillation of antibiotics: minocycline/EDTA (M/EDTA),25Bleyer A.J. Mason L. Russell G. Raad I.I. Sherertz R.J. A randomized, controlled trial of a new vascular catheter flush solution (minocycline-EDTA) in temporary hemodialysis access.Infect Control Hosp Epidemiol. 2005; 26: 520-524Crossref PubMed Scopus (89) Google Scholar vancomycin/heparin,26Se Sio L. Jenker A. Milano G.M. et al.Antibiotic lock with vancomycin and urokinase can successfully treat colonized central venous catheters in pediatric cancer patients.Pediatr Infect Dis J. 2004; 23: 963-965Crossref PubMed Scopus (32) Google Scholar and gentamicin.27Dogra G.K. Herson H. Hutchison B. et al.Prevention of tunneled hemodialysis catheter-related infections using catheter-restricted filling with gentamicin and citrate: A randomized controlled study.J Am Soc Nephrol. 2002; 13: 2133-2139Crossref PubMed Scopus (239) Google Scholar Dogra et al27Dogra G.K. Herson H. Hutchison B. et al.Prevention of tunneled hemodialysis catheter-related infections using catheter-restricted filling with gentamicin and citrate: A randomized controlled study.J Am Soc Nephrol. 2002; 13: 2133-2139Crossref PubMed Scopus (239) Google Scholar studied gentamicin at an intraluminal concentration of 40 mg/mL mixed with sodium citrate and successfully decreased the frequency of CRB. However, detectable blood levels of gentamicin evident before delivery of the following dose raised concerns for ototoxicity. Because the intraluminal concentrations used have concentrations orders of magnitude greater than those achieved systematically, we postulated that gentamicin at a lower concentration also might be effective. We also sought to determine whether duration of catheter implantation (vintage) before antibiotic lock solution was begun affected potential benefit. The study design is a prospective, open-label, randomized, controlled trial conducted at a large tertiary-care urban medical center. The primary aim is to evaluate the efficacy of low-dose gentamicin compared with heparin as a locking solution at the end of treatments. The secondary goal is to determine whether prior duration of use affected the efficacy of the antibiotic lock. We used a 10-fold lower gentamicin concentration of 4 mg/mL mixed in a low concentration of citrate as the experimental antibiotic lock solution. Citrate was chosen as the anticoagulant because the stability of gentamicin in heparin for the required periods up to 1 week, either at room temperature or 4°C, could not be attained consistently. Physical stability for 1 week, along with biological efficacy against a spectrum of pathogens found with CRBs (shown in vitro in the microbiology laboratory), was crucial to the study to make it practical. To ensure efficacy, we also compared outcomes of the gentamicin lock with an M/EDTA antibiotic lock solution previously found effective in preventing bacteremia in HD patients.28Raad I. Buzaid A. Rhyne J. et al.Minocycline and ethylenediaminetetraacetate for the prevention of recurrent vascular catheter infections.Clin Infect Dis. 1997; 25: 149-151Crossref PubMed Scopus (95) Google Scholar The orange color of the minocycline solution precluded blinding of either patients or medical staff without masking the solutions. All antibiotic lock solutions were prepared at the central research pharmacy of Henry Ford Hospital (Detroit, MI) and shipped weekly, then stored at 4°C at regional dialysis centers. Solutions older than 1 week were discarded. Because it was observed that the orange color of the reconstituted M/EDTA solution discolored rapidly within 24 hours at room temperature, but not if refrigerated, we were concerned that inadvertent change during transfer from pharmacy to dialysis center or within dialysis units could lead to potential anxiety for patients and nursing staff, as well as alter bactericidal properties. We therefore evaluated in vitro bactericidal activity of the solution microbiologically. The M/EDTA mixture as used in the clinical trial was divided into 2 aliquots: 1 refrigerated and another stored at room temperature. Immediately after preparation and at 1, 2, and 7 days of storage, 100 μL from each of the refrigerated and room-temperature bottles was inoculated into separate 13-mm diameter blank filter paper disks (BBL Taxo; Becton Dickenson Microbiology Systems, Cockeysville, MD) and allowed to dry. Bactericidal activity was tested against a methicillin-resistant S aureus isolate from a patient. Culture plates were inoculated on blood agar in 5% carbon dioxide; after 18 to 24 hours, a direct colony suspension was prepared by transferring isolated colonies to a tube of sterile saline. Inoculum density was measured by using a spectrophotometer (Vitek DensiChek, St Louis, MO) to obtain turbidity optically similar to a 0.5 McFarland turbidity standard. The resulting suspension was inoculated onto the surface of a Mueller Hinton agar plate (BBL; Remel, Lenexa, KS) in the conventional manner and preprepared discs were placed onto the surface of inoculated plates. Plates were examined after a 16-hour incubation at 35°C, and the zone of inhibition around each disc was measured. The zone of inhibition was 43 mm initially, as well as after 1, 2, and 7 days of storage for both the refrigerated and room-temperature samples. Similar experiments were conducted with gentamicin/trisodium citrate (G/TC) with freshly prepared solutions stored for 7 days. Patients were enrolled from 3 HD centers within the Greenfield Health System (GHS) between October 4, 2003, and April 30, 2004. To adjust for individual center effect on the incidence of bacteremia (baseline CRB rates varied 2-fold among the 3 dialysis centers), permutated block randomization was performed at each center. All prevalent patients with either tunneled or nontunneled (only if placed in the internal jugular vein) catheters as their primary vascular access were eligible for the study. However, only tunneled catheters were studied because of a GHS policy prohibiting dialysis in their centers of patients with nontunneled catheters. Catheters of all vintages were included. Patients were excluded if they were younger than 18 years, required a surrogate decision maker, had antibiotic treatment within 2 weeks before the date of enrollment, had catheters with blood flow rates less than 300 mL/min, or required frequent thrombolytic solution dwells in the catheter lumen for malfunction. Patients also were excluded if they were admitted to an outside hospital for any illness or required thrombolytics for catheter thromboses on more than 3 occasions. Enrolled patients were randomly assigned by using a block randomization protocol into 1 of 3 arms containing the following solutions: G/TC (4 mg/mL and 3.13%, respectively), M/EDTA (3 mg/mL and 30 mg/mL, respectively), and standard heparin solution alone (HS; 5,000 U/mL). Antibiotic solutions were prepared in the research pharmacy, transported to the individual centers on a weekly basis, and stored in the refrigerator. Nurses were trained to instill the solution into each of the 2 ports of the HD catheter at the end of each treatment, using the exact fill volume of each port. Similarly, this lock solution was withdrawn and discarded before the beginning of each subsequent treatment. In the event of hospitalization, the appropriate instillation of the 2 antibiotic lock solutions or HS was continued at the end of each dialysis treatment by the staff at Henry Ford Hospital. Tunneled HD catheters were inserted by experienced vascular access surgeons. Enrolled patients were routinely monitored clinically for symptoms and signs of bacteremia. Blood cultures were drawn if patients had fever, chills, rigors, sweats, change in mental status, or exit-site infection. The study protocol required peripheral and catheter blood cultures and an exit-site swab to be collected, if indicated. Empiric vancomycin and/or gentamicin also was administered based on clinical judgment. Catheters were removed based on CDC recommendations.29O'Grady N.P. Alexander M. Dellinger E.P. et al.Centers for Disease Control and PreventionGuidelines for the prevention of intravascular catheter-related infections.MMWR Recomm Rep. 2002; 51: 1-29PubMed Google Scholar If the catheter was removed, the catheter tip was cultured. Patients with positive blood culture results were treated with systemic antibiotics driven by type of organism and antibiotic susceptibility. Blood stream infections were defined by CDC criteria.29O'Grady N.P. Alexander M. Dellinger E.P. et al.Centers for Disease Control and PreventionGuidelines for the prevention of intravascular catheter-related infections.MMWR Recomm Rep. 2002; 51: 1-29PubMed Google Scholar This is defined as the same organism from a semiquantitative culture of the catheter tip (>15 colony-forming units/catheter segment) and a peripheral or catheter blood sample in a symptomatic patient with no other apparent source of infection. This is defined as defervescence of symptoms after antibiotic therapy with or without removal of the catheter in the setting in which blood cultures confirm infection, but catheter tip does not, or if catheter tip does, blood cultures do not, in a symptomatic patient with no other apparent source of infection. We chose not to include "possible" blood stream infection, defined as the absence of laboratory confirmation of blood stream infection. Primary end points of the study are the occurrence of bacteremia or a censoring event, such as removal of the catheter for any reason, need for thrombolytic instillation on more than 2 occasions, loss of follow-up of patients, or death or withdrawal from dialysis therapy of the patient. Infection-free survival of the catheter is defined as the number of days from instillation of the first catheter lock solution after randomization to the diagnosis of CRB, censure point, or end of study. Patient data were to be excluded from analysis if a patient in any group developed an interval acute medical illness within 2 weeks after enrollment because this would have precluded an effective period of therapy, particularly in a catheter of older vintage. An audit of all HD patients at GHS HD centers with tunneled and nontunneled catheters showed a baseline risk for infection of 3.2 episodes/1,000 catheter-days for 2002, with the rate unchanged through the first 6 months of 2003. Using this rate of catheter infection, sample size for α of 0.05 and 80% power was calculated to be 280 catheters (3 groups) followed up for a total of 1,000 days, assuming an effect size of 50% and no difference between the 2 antibiotic lock solution groups, G/TC and M/EDTA. Conversely, if much larger effect sizes of 80% occurred, as reported for other interventions,8Blankestijn P.J. Treatment and prevention of catheter-related infections in haemodialysis patients.Nephrol Dial Transplant. 2001; 16: 1975-1978Crossref PubMed Scopus (33) Google Scholar, 18Beathard G.A. Catheter management protocol for catheter-related bacteremia prophylaxis.Semin Dial. 2003; 16: 403-405Crossref PubMed Scopus (56) Google Scholar, 27Dogra G.K. Herson H. Hutchison B. et al.Prevention of tunneled hemodialysis catheter-related infections using catheter-restricted filling with gentamicin and citrate: A randomized controlled study.J Am Soc Nephrol. 2002; 13: 2133-2139Crossref PubMed Scopus (239) Google Scholar sample size could be decreased to 79 catheters followed up for 1,000 days. To avoid exposure of patients to unnecessary risk if antibiotic lock solution had large effects, interim analyses were planned after 5,000 and 10,000 total days at risk had accrued in enrolled patients. Fisher exact test to determine differences in absolute events and Kaplan-Meier method and log-rank test were used to determine whether differences in cumulative infection-free catheter survival occurred. Infection-free catheter survival, adjusted for baseline covariates, was analyzed by using Cox proportional hazards test. All results are presented as mean ± SEM or 95% confidence intervals. Differences among groups in demographic characteristics were assessed by using chi-square or Fisher exact test. Differences in continuous variables were assessed by using analysis of variance followed by a protected t-test, if appropriate. Some variables were not normally distributed; therefore, Wilcoxon/Kruskal-Wallis rank-sum test also was used to compare the groups. P less than 0.05 is accepted as significant. P of 0.05 to 0.10 are given as absolute values in tables. All values greater than 0.10 are listed as not significant. At the start of the study, 116 of 523 patients (22%) from the 3 HD centers were being dialyzed with tunneled catheters. Patient flow is shown in Fig 1. Of these, 53 patients were eligible and agreed to participate in the study, and an additional 9 incident patients were added during the following months. One prevalent patient (assigned to the control group) was excluded from analysis because of the development of bacteremia within 3 days of study enrollment, and the individual's demographic data also were removed from analyses. All data reported therefore are for the per-protocol population. All patients were seen on a weekly basis in the HD unit, and clinical events were examined closely. No patient was lost to follow-up. All catheters in use were tunneled because our institutional policy precluded use of acute catheters in GHS dialysis units. Baseline demographic and clinical data were similar for all groups (Table 1). Average age was 58 to 60 years, and patients were predominantly Afro-American. Male-female ratio was nearly 1:1 in the HS and G/TC groups; the M/EDTA group had a slight excess of males (13:8), but there was no statistically significant difference between groups (P > 0.5 for all comparisons). Diabetes was underrepresented in the M/EDTA heparin group, 8 of 21 patients compared with the other 2 groups, in which diabetes was present in 60% or greater of subjects; however, the maximum difference did not reach significance (P = 0.061). The prerandomization vintage of catheters in days in use was not normally distributed, but was similar in all 3 groups: G/TC, 145 ± 33 days; M/EDTA, 198 ± 45 days; and HS, 111 ± 25 days. Median durations were 122, 98, and 63 days, respectively (P = 0.45 for differences among distributions). At the time of the first interim analysis, catheter-days at risk also did not differ: G/TC, 95 ± 11 days; M/EDTA, 97 ± 18 days; and HS, 85 ± 18 days. A total of 6,189 days at risk had accrued, distributed as 1,734 days in the HS group, 2,002 days in the G/TC group, and 2,453 days in the M/EDTA group. The higher total days at risk in the M/EDTA group accrued from a smaller number of censure events caused by death, withdrawal, or tissue-type plasminogen activator (tPA) use (Table 2). During the course of the study, 4 patients died: 1 each in the HS and M/EDTA groups and 2 in the G/TC group. Overall rate was 26% per patient-year at risk. No death was caused by CRB.Table 1Demographic Characteristics of Study GroupsHS (control)G/TCM/EDTAPNo. of subjects202021Patient age (y)59 ± 458 ± 358 ± 3>0.5Catheter vintage (d)111 ± 25145 ± 33198 ± 450.28Median catheter vintage (d)63122980.27Catheter-days at risk85 ± 1895 ± 1197 ± 18>0.5Median catheter-days at risk7196800.45RaceNS Asian010 Afro-American121015 Caucasian786 Hispanic110Diabetes (yes/no)12:814:68:130.061⁎Fisher exact P for largest difference, G/TC versus M/EDTA.Male-female ratio10:1011:913:8NSNOTE. Data expressed as mean ± SEM unless noted otherwise. Analysis of variance used for comparison among means and Wilcoxon/Kruskal-Wallis rank sum used for comparison of medians.Abbreviation: NS, not significant. Fisher exact P for largest difference, G/TC versus M/EDTA. Open table in a new tab Table 2Mortality and Use of Thrombolytics for Access PatencyHSG/TCM/EDTAP⁎For all P > 0.1, NS is used.Mortality Deaths†No death was associated with CRB.121NS Withdrawn from dialysis110NSUse of tPA No. of subjects321NS No. of instillations1041NS‡Calculated as number of tPA instillations/all dialysis treatments. No. withdrawn (tPA > 2/wk)110NSAbbreviation: NS, not significant. For all P > 0.1, NS is used.† No death was associated with CRB.‡ Calculated as number of tPA instillations/all dialysis treatments. Open table in a new tab NOTE. Data expressed as mean ± SEM unless noted otherwise. Analysis of variance used for comparison among means and Wilcoxon/Kruskal-Wallis rank sum used for comparison of medians. Abbreviation: NS, not significant. Abbreviation: NS, not significant. Seven patients developed CRB in the HS (control) group at the time of the first interim analysis, whereas only 1 patient had CRB in the M/EDTA arm and no patient had CRB in the G/TC arm (2-tailed Fisher exact test for HS versus G/TC, P = 0.008, and for HS versus M/EDTA, P = 0.02). There was no difference between the M/EDTA and G/TC groups. Clinical examination and follow-up of patients with bacteremia did not show sources of infection in these patients other than the catheter. Some infected catheters were salvaged with use of intravenous antibiotic therapy; therefore, bacteriological identification of the catheter tip was not possible in all catheters. CRB rates were 4 episodes/1,000 catheter-days in the HS group and 0.4 episodes/1,000 catheter-days in the M/EDTA group. Because of these findings, the study was terminated. Kaplan-Meier analysis is shown in Fig 2. Log-rank test for difference was 0.0067. Infection-free survival rates at 120 days were 100% for the G/TC group, 95% for M/EDTA group, and 56% for the HS group. There was no difference in survival between the G/TC and M/EDTA groups. Cox proportional hazard analysis of factors associated with risk for developing CRB showed that the CRB event rate occurred independent of patient age, catheter vintage, presence of diabetes, sex, center at which the patients received HD, and prerandomization blood