Effectiveness of Minocycline and Rifampin vs
Chlorhexidine and Silver Sulfadiazine-Impregnated
Central Venous Catheters in Preventing Central
Line-Associated Bloodstream Infection in a High-Volume
Academic Intensive Care Unit: A Before and after Trial
Stephanie Bonne, MD, FACS, John E Mazuski, MD, PhD, FACS, Carie Sona, RN, Marilyn Schallom, RN,
Walter Boyle, MD, Timothy G Buchman, PhD, MD, FACS, Grant V Bochicchio, MD, MPH, FACS,
Craig M Coopersmith, MD, FACS, Douglas JE Schuerer, MD, FACS
BACKGROUND: Use of chlorhexidine and silver sulfadiazine-impregnated (CSS) central venous catheters
(CVCs) has not been shown to decrease the catheter-related bloodstream infection rate
in an ICU. The purpose of this study was to determine if use of minocycline and
rifampin-impregnated (MR) CVCs would decrease central line-associated bloodstream
infection (CLABSI) rates compared with those observed with use of CSS-impregnated
CVCs.
STUDY DESIGN: A total of 7,181 patients were admitted to a 24-bed university hospital surgical ICU: 2,551
between March 2004 and August 2005 (period 1) and 4,630 between April 2006 and July
2008 (period 2). All patients requiring CVC placement in period 1 had a CSS catheter
inserted, and in period 2 all patients had MR CVCs placed.
RESULTS: Twenty-two CLABSIs occurred during 7,732 catheter days (2.7 per 1,000 catheter days) in
the 18-month period when CSS lines were used. After the introduction of MR CVCs, 21
catheter-related bloodstream infections occurred during 15,722 catheter days (1.4 per 1,000
catheter days). This represents a significant (p < 0.05) decrease in the CLABSI rate after
introduction of MR CVCs. Mean length of time to infection developing after catheterization
(8.6 days for CSS vs 6.1 days for MR) was also different (p ¼ 0.04). The presence of MR did
not alter the microbiologic profile of catheter-related infections, and it did not increase the
incidence of resistant organisms.
CONCLUSIONS: The CLABSI rate decreased more with the use of MR CVCs compared with CSS CVCs in an
ICU where the CLABSI rate was already low. The types of organisms causing infection were
similar. With continued use of MR-impregnated CVCs in our ICU in the subsequent 5 years,
we have seen sustained low rates of CLABSIs. (J Am Coll Surg 2015;221:739e747. � 2015
by the American College of Surgeons)
The CDC estimated that a total of 18,000 central line-
associated blood stream infections (CLABSIs) occurred
in ICUs in the United States in 2009.1 An additional
estimated 28,000 infections occurred in inpatient wards.
Disclosure Information: Nothing to disclose.
Disclosures outside the scope of this work: Dr Bochicchio’s institution
received a grant from Cook Medical.
Abstract presented at the 38th Critical Care Congress, Society of Critical
Care Medicine, Nashville, TN, February 2009.
Received January 19, 2015; Revised April 6, 2015; Accepted May 14, 2015.
739
ª 2015 by the American College of Surgeons
Published by Elsevier Inc.
Although this represents a 58% reduction from 2001
levels,2 the CLABSI rate is still well above zero. Current
(2014) CDC data estimate that 41,000 CLABSIs occur
annually in US hospitals.3 With the National Healthcare
From the Departments of Surgery (Bonne, Mazuski, Bochicchio, Schuerer) and
Anesthesiology (Boyle), Washington University School of Medicine, Barnes-
Jewish Hospital (Sona, Schallom), St Louis, MO, and Department of Surgery,
Emory University School of Medicine, Atlanta, GA (Buchman, Coopersmith).
Correspondence address: Stephanie Bonne, MD, FACS, Department of
Surgery, Washington University School of Medicine, 660 S Euclid, Box
8109, St Louis, MO 63122. email: bonnes@wudosis.wustl.edu
http://dx.doi.org/10.1016/j.jamcollsurg.2015.05.013
ISSN 1072-7515/15
Abbreviations and Acronyms
CLABSI ¼ central line-associated blood stream infection
CSS ¼ chlorhexidine and silver sulfadiazine
CVC ¼ central venous catheter
MR ¼ minocycline and rifampin
NHSN ¼ National Healthcare Safety Network
SICU ¼ surgical ICU
740 Bonne et al Preventing Central Line-Associated Infection J Am Coll Surg
Safety Network (NHSN) declaration of CLABSI as a
“never” event, hospitals and practitioners are under
increasing pressure to decrease their CLABSI rates to
zero. Additionally, although the mortality of CLABSI
is variable in the literature, with estimates ranging
from 0 to 35%,4-9 the development of CLABSI places
patients in danger of adverse events. These infections
also increase length of stay in both ICUs and hospitals,
placing an additional financial burden on the health
care system.5,7-13 Estimates of cost per line range from
$12,000 to $56,000 per infection.10-13
Exposure to central venous catheters (CVC) is high in
ICU patients, with use rates ranging from 0.36 CVC days
per total number of patient days in coronary care units to
0.83 CVC days per total number of patient days in
cardiothoracic ICUs. Surgical and medical ICUs have
rates of 0.63 and 0.52, respectively.2 Multiple studies
have shown that most, if not all, CLABSIs are prevent-
able, and extensive guidelines have been developed to
guide best practice for CVC placement.14 Successful stra-
tegies to reduce CLABSIs include implementation of
educational programs directed toward the entire multidis-
ciplinary ICU team.15-18 These programs typically stress
adherence to best practice behaviors known to prevent
CLABSIs, including the use of full barrier precautions,19
skin preparation at the insertion site,20 appropriate hand
hygiene,21,22 and specifying the anatomic site of CVC
placement (subclavian preferred to internal jugular
preferred to femoral).23,24 The introduction of checklists
and central line “bundles,” along with educational pro-
grams, have additionally been shown to improve out-
comes and increase adherence to best practices,
including decreasing CLABSIs.25,26 Although the decrease
in CLABSIs varies depending on which intervention is be-
ing studied, comprehensive prevention programs can
result in almost complete elimination of CLABSIs from
a surgical ICU (SICU).27,28
Another tool to prevent CLABSIs is the use of anti-
septic or antimicrobial-impregnated catheters. To date,
there have been >20 randomized trials, as well as several
meta-analyses and cost to benefit analyses on the efficacy
of either chlorhexidine and silver sulfadiazine (CSS)
CVCs or minocycline and rifampin (MR) CVCs.29-33
The majority of these demonstrate substantial decreases
in CLABSI rates without the development of resistant or-
ganisms, regardless of which catheter was used.31,32 Of
note, placement of antiseptic-impregnated catheters has
been shown to be effective in decreasing CLABSIs if
evidence-based practices are adhered to, but infection
rates are still higher than desired.33 The accumulated
weight of these studies has resulted in recommendations
from the CDC,1 the Agency for Healthcare Research
and Quality,34 and from thought leaders in the field35-37
to use antiseptic or antimicrobial-impregnated catheters,
although their efficacy has not been universally accepted.
Although antiseptic or antimicrobial-impregnated cath-
eters are approximately $15 to $20 more expensive (pric-
ing varies by institution) than standard CVCs, cost to
benefit analyses favor their use due to the attributable
cost of an infection being much greater than the cost of
a catheter. There has been substantial interest in deter-
mining at-risk populations that would most benefit
from the use of these catheters. The CDC recommends
that an antiseptic or antimicrobial-impregnated CVCs
be placed “in adults whose catheter is expected to remain
in place >5 days if, after implementing a comprehensive
strategy to reduce rates of CLABSI, the CLABSI rate re-
mains above the goal set by the individual institution
based on benchmark rates and local factors.”14 The
comprehensive strategy is defined as an educational pro-
gram directed toward those who insert and maintain
CVCs, as well as use of both maximal barrier precautions
and 2% chlorhexidine as skin preparation solution for
CVC insertion.
Previously, we instituted educational and behavioral in-
terventions designed to prevent CLABSIs.15,17 These
resulted in infection rates that were one-third the national
average, as determined by the NHSN at the time of their
publication. To determine if there was additional efficacy
in using CSS CVCs despite our low CLABSI rates, we
studied the effect of inserting these catheters in all patients
in our SICU during an 18-month period. These data
showed that CSS CVCs were only effective in specific pa-
tient populations, but did not statistically decrease the rate
of CLABSI in the SICU.38 In an attempt to study the
potential benefit of antibiotic-impregnated catheters, we
changed our practice pattern to use MR CVCs on all
patients in the SICU and studied the efficacy of these
catheters compared with CSS CVCs.
These results could be consistent with the CDC recom-
mendation to use these catheters in settings where the
CLABSI rates remain elevated after other interventions
were made. However, there are actually few other
data currently available on the efficacy of antiseptic or
Vol. 221, No. 3, September 2015 Bonne et al Preventing Central Line-Associated Infection 741
antimicrobial-impregnated catheters in ICUs that have
successfully instituted the comprehensive strategy out-
lined by the CDC, and whether or not different types
of catheters might have different efficacy in this setting.
The purpose of this study was to address the use of anti-
septic- or antibiotic-impregnated catheters (specifically
looking at MR CVCs) in an ICU with low infection rates
after implementation of a successful educational program
to prevent CLABSI.
METHODS
Study location and patient population
Barnes-Jewish Hospital is a 1,305-bed tertiary care,
university-affiliated teaching hospital. All patients
admitted to the SICU between March 1, 2004 and
August 31, 2005, and subsequently between April 1,
2006 and July 31, 2008, were included in the study.
The SICU is a 24-bed unit that admits all noncardio-
thoracic and non-neurosurgical critically ill surgical and
trauma patients in the hospital, as well as selected medical
patients. Mean length of stay was constant throughout the
study at 4.0 (�0.8) days, as were nursing and physician
staffing ratios. Throughout the study, all patients
admitted to the SICU were followed prospectively by
an infection-control team and surveyed for bloodstream
infections. Data in this study were collected after comple-
tion of our published educational and behavioral inter-
ventions to prevent CLABSI in the SICU,15,17 and after
the completion of our study showing no decrease in rates
of CLABSI with second-generation CSS CVCs over tradi-
tional, nonimpregnated catheters.37
Demographic data for patients in the study was
retrieved from the Project IMPACT database (Cerner).
All positive blood cultures were classified as primary or
secondary based on CDC definitions. A CLABSI was
identified when a pathogen was isolated from blood cul-
ture not considered to be the manifestation of an infection
at another site or in a patient with fever >38.5�C, chills,
or hypotension. Common skin contaminant, such as
coagulase-negative staphylococci, were considered patho-
gens if they were isolated from two blood cultures drawn
on separate occasions and were unrelated to infection at
another site, or if the attending physician chose to provide
a therapeutic course of antimicrobial therapy based on a
single positive blood culture. Secondary bacteremias
were not included in this analysis. Central line-
associated related bloodstream infections were reported
when they were identified in first-time central line or
subsequent central line placements.
Updates on CLABSI rates were reported monthly at the
SICU’s quality-improvement conference and were
compared with earlier rates within the SICU before the
universal implementation of MR CVCs. Data collections
and surveillance protocols remained the same across the
two study periods. However, aggregate results using the
MR CVCs were neither reported nor compared with rates
obtained when the CSS CVCs were being used until
completion of the trial.
The study was approved by the IRB of the Human
Research Protection Office of Washington University
School of Medicine. A requirement for informed consent
from the patients was waived.
Study design
In the preintervention group, or period 1, all patients
requiring a central line in the ICU had a second-
generation CCS CVC placed (Arrowgard Blue Plus;
Arrow International). All patients admitted to the
SICU in the postintervention group, or period 2, who
required CVC insertion had MR CVCs (Cook Medical,
Cook Group) placed. These catheters are polyurethane
catheters that are manufactured with both external
coating and internal impregnation of the antibiotic
agents. The decision to use catheters exclusively was
based on a decision that any CVC placed in the SICU
was likely to stay in place for 5 days and therefore
met the CDC criteria for placing antimicrobial or
antiseptic-impregnated catheters. Mandatory educa-
tional modules on how to prevent CLABSI continued
on a regular basis throughout the study for both physi-
cians and nurses in the SICU.
A pre-hoc decision was made that the primary end
point would be CLABSI rates for CVCs placed in the
SICU, with a secondary end point being CLABSI rates
for all CVCs identified in the surgical ICU regardless of
where they were placed. The rationale behind this deci-
sion was to eliminate the confounding effect of nonim-
pregnated CVCs placed by individuals in the operating
room, emergency department, hospital wards, interven-
tional radiology suite, or at another hospital for the pri-
mary analysis. The same primary end point was used in
a previous study from this ICU.38 All CVCs placed in
the SICU throughout the study were inserted by super-
vised residents, fellows and nurse practitioners, or
attending physicians.
Statistical analysis
Data were analyzed using the statistical software program
GraphPad Prism software, version 4.0 (GraphPad Software).
Infection rates and contingency tables (ie, when comparing
insertion site between CSS and MR groups) were analyzed
using chi-square test. Demographic and microbiology
comparisons between pre- and postintervention groups
Figure 1. Central line-associated bloodstream infection (CLABSI)
rate per 1,000 catheter days. C/SS, chlorhexidine and silver
sulfadiazine-impregnated catheter; M/R, minocycline and rifampin-
impregnated catheter.
Table 1. Demographics of All Patients with a Central
Venous Catheter in Place
Variable
March 2004
to August
2005
April 2006
to July
2008
Patients admitted, n 2,551 4,630
Patients with CVC, n 1,949 3,727
Where CVC placed, %
In SICU 50.1 43.8
Outside SICU (OR, ED,
radiology, ward, other) 49.9 56.2
Age, y, mean 59.3 57.5
CVC duration, d, mean 3.4 3.3
APACHE II score 18.2 17.9
ICU length of stay, d, mean 8.9 8.0
Hospital length of stay of
patients, d, mean 23.3 21.4
Duration of CVC, %
<7 days 82.7 84.4
7 to 10 days 11.8 11.1
>10 days 5.4 4.6
Insertion site, %
Subclavian 40.0 43.0
Internal jugular 52.0 48.3
Femoral 8.0 8.7
CVC, central venous catheter; ED, emergency department; OR, operating
room; SICU, surgical ICU.
742 Bonne et al Preventing Central Line-Associated Infection J Am Coll Surg
were performed using the Mann-Whitney test. A p value
<0.05 was considered statistically significant.
RESULTS
Patients
There were 7,181 patients admitted to the SICU during
the 46 months of the study. This includes 2,551 patients,
of which 1,949 had CVCs in 18 months from March
2004 through August 2005, when CSS CVCs were stan-
dard in the ICU. There were 4,630, patients, of which
3,727 had CVCs in 28 months from April 2006 through
July 2008, during which time all patients who had a CVC
placed in the SICU had an MR CVC inserted. Demo-
graphics for all patients with a CVC in place were similar
in the pre- and postintervention groups (Table 1). The
majority of all CVCs were placed in the internal jugular
vein, but this was predominantly due to CVCs placed
in the operating room (nearly exclusively placed in the in-
ternal jugular) because the majority of CVCs inserted in
the SICU were placed into the subclavian vein (data not
shown). Patients who had a CVC inserted had higher
severity of illness throughout the study than those who
did not, because patients who had a CVC in place had
higher mean APACHE II scores (18.2 vs 15.6) and had
longer lengths of stay than those without CVCs (6.9 vs
4.3 days, data not shown).
Central line-associated blood stream infection in
chlorhexidine and silver sulfadiazine central venous
catheters placed in surgical intensive care units
In the first time period of the study, when CSS CVCs
were being placed, there was a total of 7,732 catheter
days. In this time, there were 16 CLABSIs identified, or
2.1 per 1,000 catheter days. During the second time
period, when MR CVCs were placed, there were a total
of 21 CLABSIs in 15,722 catheter days, or 1.4 per
1,000 catheter days (p < 0.05) (Fig. 1). The demo-
graphics of patients with CLABSI in CVCs placed in
the SICU were similar regardless of whether the infection
was in a CSS CVC or MR CVC (Table 2). Length of stay
and anatomic site of catheter insertion were also similar
between groups.
Microbiology of central line-associated blood
stream infection in intensive care unit
No statistically significant differences in microbiology
were noted in infection in CVCs placed in the SICU,
regardless of whether the patient had a CSS CVC inserted
or MR CVC, Gram-positive organisms predominated in
both the pre- and postintervention groups (Fig. 2). In
the CCS-impregnated catheters, 50% of the infections
were due to Gram-positive microorganisms, with a pre-
dominance of Staphylococcus sp. and Enterococcus sp.
Twenty-five percent were due to Gram-negative microor-
ganisms and 25% were due to fungal microorganisms. In
the MR catheters, 57% were due to Gram-positive, 19%
were due to Gram-negative and 24% were due to fungal
microorganisms (p ¼ NS).
Figure 2. Microorganisms isolated from patients with central line-
associated bloodstream infection from central venous catheters
placed in the surgical ICU. Blue bar, chlorhexidine and silver
sulfadiazine-impregnated; red bar, minocycline/rifampin-impregnated.
Table 2. Demographics of Patients Who Developed Cen-
tral Line-Associated Bloodstream Infection from Central
Venous Catheters Placed in the Surgical Intensive Care Unit
Variable CSS MR
Age, y, mean � SD 55.2 � 20.0 57.5 � 16.4
Male, n (%) 9 (56) 15 (62.5)
CVC duration before infection, d 8.6 6.1
Insertion site, n (%)
Subclavian 8 (50.0) 10 (41.6)
Internal jugular 8 (50.0) 8 (33.3)
Femoral 0 3 (12.5)
CSS, chlorhexidine and silver sulfadiazine-impregnated catheter; CVC,
central venous catheter; MR, minocycline and rifampin-impregnated
catheter.
Vol. 221, No. 3, September 2015 Bonne et al Preventing Central Line-Associated Infection 743
Time to development of infection in central venous
catheters
During the initial study period, when CSS CVCs were
used, the mean length of time a SICU-placed CVC was
in before development of CLABSI was >8 days. This is
substantially greater than the mean length of time the
typical CVC remained in place, because <15% of all
CVCs stayed in >7 days. During the subsequent study
period, when MR CVCs were placed, catheters were in
place for a shorter period of time before development of
an infection (6.1 days; p ¼ 0.04). However, the mean
duration that CVC was in place (3.3 days; p ¼ NS)
was not significantly different, suggesting that earlier
discontinuation of CVCs during the latter period did
not account for the differences observed in the infection
rates or their timing. For patients in whom an infection
did eventually develop, the time to CLABSI development
was 8.6 days in the CSS CVC group and 6.1 days in the
MR CVC group (p ¼ 0.04) (Fig. 3).
Figure 3. Days to line infection by catheter type. C/SS, chlorhexi-
dine/silver sulfadiazine-impregnated; M/R, minocycline and
rifampin-impregnated catheter.
Central line-associated blood stream infection in all
central venous catheters by placement location
In addition to CLABSIs in CVCs placed in the SICU,
there were 6 CLABSIs in CVCs placed before patient
arrival in the SICU in period 1 when CSS CVCs were be-
ing placed in the SICU, and 3 in period 2, when MR
catheters were placed. There were 22 infections in the
CSS group (2.7 per 1,000 catheter days). This includes
16 infections in CSS CVCs placed in the SICU and 6
standard catheters placed outside the SICU, which were
not impregnated. In the MR group, there were a total
of 21 central line-associated bloodstream infection in
15,722 catheter days, or 1.4 per 1,000 catheter days,
including 3 lines placed outside the SICU that were not
impregnated (Fig. 4). By comparison, CDC data for
SICUs nationwide in 2009 show a rate of 1.9 per 1,000
catheter days in 2009. Throughout the study, CLABSI
rates were lower in the SICU than CDC rates published
for each year (data not shown).2,39
DISCUSSION
This study demonstrates that placing MR-coated cathe-
ters in all SICU patients requiring CVC insertion signif-
icantly decreased CLABSI rates below the already low
rates obtained in an ICU using second-generation CSS-
impregnated catheters. This was done in a SICU in which
a comprehensive educational program and adherence to
best practice behaviors program was already in place. To
our knowledge, this is the only direct comparison of a
second-generation antiseptic catheter vs antimicrobial-
impregnated catheter.
Existing literature supports the use of impregnated
antiseptic or antimicrobial catheters to decrease the rate
of CLABSIs in the ICU. The CDC currently has a 1A
Figure 4. Infected line placement location. Blue bar, inside surgical
ICU; Red bar, outside ICU. C/SS, chlorhexidine and silver
sulfadiazine-impregnated catheter; M/R, minocycline and rifampin-
impregnated catheter.
744 Bonne et al Preventing Central Line-Associated Infection J Am Coll Surg
recommendation that either an antiseptic or antimicrobial
CVC be placed when the catheter is likely to stay in place
for >5 days and the ICU in which the CVC is to be
placed has a record of higher than national average
CLABSI rates after implementation of a comprehensive
strategy to decrease CLABSI. However, the decision as
to which catheter should be placed is left to the practi-
tioner in the guidelines, with little available data
comparing one catheter with another.40
In light of the fact that multiple ICUs have eliminated
CLABSI, the goal CLABSI rate for the SICU at Barnes-
Jewish Hospital should be zero. Despite the fact that our
CLABSI rates have been less than mean NHSN and
CDC levels since 1999, our CLABSI rates were, in fact,
higher than the institutional goal. As such, strict implemen-
tation of the CDC guidelines would result in our using
antiseptic or antibiotic-impregnated catheters (either CSS
or MR) on all of our patients expected to have a catheter
in place for >5 days. As NHSN rates of CLABSI and sub-
sequently the CDC target rates of CLABSI drop year by
year, an important question was whether antiseptic-
impregnated or antimicrobial-impregnated catheters could
decrease infection rates in an ICU if education and compli-
ance with best practice led to CLABSI rates that were low,
but not zero. Previously published results from our group
comparing CSS CVCs with standard nonimpregnated
catheters did not demonstrate a substantially lower rate
of infection in these antiseptic-impregnated catheters.39
Based on these results here, however, the use of the MR
CVCs does appear to be justified from the standpoint of
efficacy alone. Previously, we reported that the CLABSI
rate of CSS CVCs and traditional catheters were
equivalent. Based on the current data, CLABSI rates
should also be lower when MR CVCs are used rather
than traditional catheters in our ICU setting. This conclu-
sion is also supported by other studies.41-43 In addition,
when the cost of a CLABSI is considered, the use of
MR CVCs becomes worth the small additional cost of
the catheters. Actual cost savings would need to be consid-
ered on an institutional basis, as agreements about cost of
materials vary by institution. However, regardless of
specific cost at a given institution, the additional cost of
catheters should be quite low compared with the over-
whelming cost of treating a CLABSI.
The demographics of all patients with CVCs were strik-
ingly similar throughout the 46 months of the study, as
were the characteristics of patients whose catheters became
infected, regardless of whether they were CSS CVCs or
MR CVCs. There was a significant difference in the
time to infection between the CSS CVCs and the MR
CVCs. Of note, central lines were removed from the pa-
tient once a CLABSI was identified. At first look, the data
on time to infection appear to favor the CSS CVCs
because the time to infection was longer for these cathe-
ters. However, the overall duration of catheter use was
not different between the two groups. In addition, even
if the time to infection of MR CVCs is shorter, overall,
there were still fewer patients with a CLABSI among pa-
tients having MR CVCs placed compared with those hav-
ing CSS CVCs placed. Although the reason bloodstream
infections became manifest slightly earlier with the MR
CVCs remains elusive, it does not obviate the overall
benefit observed with MR CVCs. Additionally, we do
not advocate for routine line change at 7 days, based on
these data. If overall line infection is short and the time
to a line infection developing, if one is to develop, is
<7 days, there should be no evidence-based reason for
routine catheter changes.
One other finding with the use of both the CSS and
MR CVCs was a nonsignificant trend toward increased
fungemia. In the overall experience, there were 2 funge-
mias during 56 months before use of any antiseptic- or
antibiotic-coated CVCs, 4 in the subsequent 18 months
when the CSS catheters were used, and then 5 in the 28
months when the MR CVCs were used. Longer-term
follow-up would be needed to determine if this consti-
tuted a clinically significant finding.
Although we believe this study answers an important,
previously unanswered question about the direct compar-
ison of antiseptic and antimicrobial catheters, it has a
number of limitations. Because our previous educational
program decreased CLABSI rates >2-fold, we already
have a low CLABSI rate and, therefore, few patients
with infections to study. Although our CLABSI rate was
Figure 5. Five-year follow-up data. Purple line, rate of infection per 1,000 catheter days; C/SS CVCs,
chlorhexidine and silver sulfadiazine-impregnated central venous catheters; M/R CVCs, minocycline and
rifampin-impregnated central venous catheters.
Vol. 221, No. 3, September 2015 Bonne et al Preventing Central Line-Associated Infection 745
downtrending after implementation of education pro-
grams, we still sought to bring our CLABSI rate closer
to zero, as is the gold standard for a “never” event. Despite
having a low number of total infections, we were still able
to detect a 49% decrease in CLABSI with use of MR
CVCs. Of note, we have had sustained low rates of
CLABSI in our ICU in the subsequent period of time
since the study period ended, which is tracked as part of
our SICU quality-improvement process (Fig. 5). We
conclude from these observations that education
programs decrease CLABSI rates to a degree, but a multi-
disciplinary approach, including the use of antibiotic-
impregnated catheters, brings rates closer to zero than
education programming alone.
Other limitations exist with the pre- and poststudy
design used. Compared with a prospective randomized
trial, this type of study is susceptible to temporal bias
due to unrecognized changes in patient populations or in
ICU practices. In addition, it makes a blinded comparison
impossible. However, the similarity of the patients in the 2
cohorts and the similar duration of CVC use suggest that a
systematic bias due to changes in patient characteristics was
not present. In addition, the stability of the CLABSI rate
after study completion also argues against other interven-
tions in the SICU contributing to the decreased rate of
infection. Ultimately, with the overall low rate of CLABSI,
a single-center randomized controlled trial would not be
feasible, and even a multicenter trial would likely be
impossible due to the need to enroll several thousand pa-
tients to detect a statistically significant difference without
an appreciable type II error.
The study design called for all patients having CVCs
placed in the SICU to have the MR CVC used. It was
believed that there was a considerable chance that those
catheters would be left in place for 5 days, to meet the
CDC recommendation for antiseptic or antimicrobial-
catheter placement. However, >80% of CVCs in the
SICU remained in place for <7 days, and the mean dura-
tion of catheter use was 3.4 and 3.3 days in the 2 cohorts.
Although this suggests the a priori assumption that cath-
eters would be used for at least 5 days was not valid, this
discrepancy might be more apparent than real. However,
the data presented in Table 1 demonstrate that >80% of
CVCs in the SICU were left in <7 days. This discrepancy
is not as great as it might seem on the surface because the
majority of CVCs in SICU patients were placed outside
the SICU, and these were likely to be in a shorter period
of time than catheters placed in the SICU. The short
duration of use of CVCs placed outside of the SICU
was also likely because of standard SICU protocols.
Most CVCs placed in the emergency department, for
instance, were removed within 24 to 48 hours. Central
venous catheters placed in the operating room were
frequently used for intraoperative monitoring or drug
administration, and were many times removed within a
day or 2 after the operation, if there was no longer an
746 Bonne et al Preventing Central Line-Associated Infection J Am Coll Surg
indication for use, following the CDC guideline for daily
assessment of need. In contrast, CVCs were usually placed
in the SICU because of clinical necessity, such as a need
for longer-term access for administration of medications,
such as parenteral nutrition or vasoactive medications, or
for ongoing monitoring of central venous pressures of ox-
ygen saturations. This increases the likelihood that a CVC
placed in the SICU would be left in place for >5 days. An
unknown number of CVCs placed in the SICU during
this study were likely removed within 5 days. However,
from the standpoint of facilitating a standard approach
by multiple different providers, using a single catheter
was believed to be far more feasible than relying on a
more complicated protocol asking individuals to speculate
on the duration of catheter use to choose the desired one.
CONCLUSIONS
This study has important implications for ICUs, which
have successfully decreased CLABSI rates to less than the
NHSN and CDC means, but have been unable to
