Quasi-experimental study of sodium citrate locks and the risk of acute hemodialysis catheter infection among critically ill patients

Abstract

Critically ill patients who require renal replacement therapy (RRT) are vulnerable to catheter-related bloodstream infections (CRBSI). This study compared the risks of dialysis catheter infection according to the choice of locking solution in the intensive care unit (ICU). A prospective quasi-experimental study with marginal structural models (MSM) and 2:1 greedy propensity-score matching (PSM) was conducted at nine university-affiliated hospitals and three general hospitals. A total of 596 critically ill patients received either saline solution or heparin lock solution (the standard of care [SOC]) from 2004 to 2007 in the Cathedia cohort (n = 464 for MSM; n = 124 for PSM) or 46.7% citrate lock from 2011 to 2012 in the citrate (CLock) cohort (n = 132 for MSM; n = 62 for PSM) to perform RRT using intermittent hemodialysis. Catheter-tip colonization and CRBSI were analyzed. The mean duration (standard deviation [SD]) of catheterization was 7.1 days (6.1) in the SOC group and 7.0 days (5.9) in the CLock group (P = 0.84). The risk of dialysis catheter-tip colonization was lower in the CLock group (20.5 versus 38.7 per 1,000 catheter-days in the SOC group; hazard ratio [HR] from MSM, 0.73; 95% confidence interval [CI], 0.57 to 0.93; P < 0.02). Consistent findings were found from PSM (HR, 0.46; 95% CI, 0.22 to 0.95; P < 0.04). The risk of CRBSI was nonsignificantly different in the CLock group (1.1 versus 1.8 per 1,000 catheter-days in the SOC group; HR from MSM, 0.48; 95% CI, 0.12 to 1.87; P = 0.29). By reducing the risk of catheter-tip colonization, citrate lock has the potential to improve hemodialysis safety in the ICU. Additional studies are warranted before the routine use of citrate locks can be recommended in the ICU.

Copyright © 2014, American Society for Microbiology. All Rights Reserved.

Figures

FIG 1

Kaplan-Meier curves of time to catheter colonization at the time of catheter removal (A) and at the time of dysfunction (B) in the overall cohort (n = 596). IPWT, inverse probability weighting.

FIG 2

Kaplan-Meier curves of time to catheter colonization at the time of catheter removal (A) and at the time of dysfunction (B) after propensity-score matching (n = 186).