M A J O R A R T I C L E

Vancomycin Exposure in Patients With

Methicillin-Resistant Staphylococcus aureus
Bloodstream Infections: How Much Is Enough?
Thomas P. Lodise,
1
George L. Drusano,
2
Evan Zasowski,
1
Amanda Dihmess,
1
Victoria Lazariu,
3
Leon Cosler,
1
and
Louise-Anne McNutt
3
1
Albany College of Pharmacy and Health Sciences, New York;
2
Institute for Therapeutic Innovation, College of Medicine, University of Florida, Lake Nona;
and
3
University at Albany, State University of New York
Background. Contemporary vancomycin dosing schemes are designed to achieve an area under the curve
(AUC) to minimum inhibitory concentration (MIC) ratio of 400. However, scant clinical data exist to support
this target and available data relied on pharmacokinetic formulas based on daily vancomycin dose and estimated
renal function (demographic pharmacokinetic model) to estimate AUCs.
Methods. A cohort study of hospitalized, adult, nondialysis patients with methicillin-resistant Staphylococcus
aureus bloodstream infections treated with vancomycin was performed to quantitatively evaluate the relationship
between vancomycin exposure and outcomes. Bayesian techniques were used to estimate vancomycin exposure pro-
le for day 1 and 2 of therapy for each patient based on their dosing schedule and collected concentrations. Clas-
sication and Regression Tree (CART) analysis was used to identify day 1 and 2 exposure thresholds associated with
an increased risk of failure. Failure was dened as 30-day mortality, bacteremia was 7 days, or recurrence.
Results. During the study period, 123 cases met criteria. Failure was uniformly less pronounced (approximately
20% less in absolute value) in patients who achieved the CART-derived day 1 and 2 thresholds for AUC/MIC by
broth microdilution and AUC/MIC by Etest. In the multivariate analyses, all risk ratios were approximately 0.5 for all
CART-derived AUC/MIC exposure thresholds, indicating that achievement of CART-derived AUC/MIC exposure
thresholds was associated with a 2-fold decrease in failure.
Conclusions. These ndings establish the critical importance of daily AUC/MIC ratios during the rst 2 days of
therapy. As with all observational studies, these ndings should be interpreted cautiously and validated in a multi-
center randomized trial before adoption into practice.
Keywords. MRSA; outcomes; pharmacodynamics; pharmacokinetics; vancomycin.
Despite its introduction more than a half-century ago,
the optimal dosing strategy for vancomycin remains
undened. Contemporary vancomycin dosing schemes
are designed to achieve an area under the curve (AUC)
to minimum inhibitory concentration (MIC) ratio
400 for serious methicillin-resistant Staphylococcus
aureus (MRSA) infections [1, 2]. Although this target
is based on the best available evidence [16], it is largely
derived from neutropenic mouse thigh infection model
data [3]. The best clinical evidence supporting AUC/
MIC ratio 400 is drawn from a retrospective evalua-
tionof patients withS. aureus pneumonia [5]. Tworecent
studies of patients with MRSA bloodstream infections
(BSIs) have also identied similar vancomycin AUC/
MIC ratio targets [7, 8]. Although these evaluations
provide further evidence that the vancomycin pharma-
codynamic target is an AUC/MIC ratio of at least 400,
all evaluations used a simple formula based on daily van-
comycin dose and estimated renal function to estimate
AUC values [5, 7, 8]. In most cases, they used the Cock-
croft-Gault creatinine clearance (CrCl) formula. There
Received 4 March 2014; accepted 19 May 2014; electronically published 27 May
2014.
Correspondence: Thomas Lodise, PharmD, PhD, Pharmacy Practice, Albany Col-
lege of Pharmacy and Health Sciences, Albany, NY 12208-3492 (thomas.lodise@
acphs.edu).
Clinical Infectious Diseases 2014;59(5):66675
The Author 2014. Published by Oxford University Press on behalf of the Infectious
Diseases Society of America. All rights reserved. For Permissions, please e-mail:
journals.permissions@oup.com.
DOI: 10.1093/cid/ciu398
666 CID 2014:59 (1 September) Lodise et al

a
t

U
n
i
v
e
r
s
i
d
a
d

d
e

A
n
t
i
o
q
u
i
a

o
n

A
u
g
u
s
t

2
2
,

2
0
1
4
h
t
t
p
:
/
/
c
i
d
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
/
D
o
w
n
l
o
a
d
e
d

f
r
o
m

is considerable interpatient variability in vancomycin exposure
proles in clinical practice and it is difcult to generate valid es-
timates of exposure variables in a given individual based on glo-
merular ltration estimation formulas alone [911]. To date, we
are only aware of 2 small-scale vancomycin exposureresponse
clinical evaluations that considered individualized estimates of
the vancomycin AUC based on collected levels and doses received
[12, 13]. Thus, there is a critical need for additional, larger-scale
clinical studies that utilize individualized estimates of exposure
proles based on measured concentrations.
Although AUC/MIC ratio is the prevailing vancomycin expo-
sure target, AUCs are not determined in clinical practice due to
the perceived difculty in calculating the AUC [2]. Expert guide-
lines recommend maintaining a minimum (trough) concentra-
tion (C
min
) between 15 and 20 mg/L as a surrogate marker for an
AUC/MIC ratio 400 [1, 2]. However, the clinical benets of
maintaining higher vancomycin trough values have not been
well described [1419]. The intent of this study was to quantita-
tively evaluate the relationship between vancomycin exposure
variables (ie, C
min
/MIC, AUC/MIC) and outcomes among pa-
tients with MRSA BSIs. Bayesian techniques [2022] were used
to estimate the vancomycin concentrationtime prole for each
patient. The Bayesian approach used in this study to estimate
exposure proles has recently been validated as a method to esti-
mate vancomycin exposure values with low bias and high preci-
sion in situations where trough-only pharmacokinetic (PK) data
are available [22]. As a secondary objective, this study compared
the predictive performance of the Bayesian relative to the formula-
based approach for estimating exposure proles.
METHODS
Experimental Design and Study Population
A retrospective cohort study was performed among hospitalized
patients with MRSA BSIs treated with vancomycin at Albany
Medical Center Hospital between January 2005 and June
2009. Patients meeting the following criteria were included:
(1) age 18 years; (2) absolute neutrophil count 1000 cells/
L; (3) MRSA culture met the Centers for Disease Control
and Prevention criteria for BSI [23]; (4) index MRSA isolate
available for phenotypic characterization; (5) not receiving dial-
ysis; (6) received vancomycin within 48 hours of index culture;
(7) received vancomycin for at least 2 days; and (8) had 1 van-
comycin level collected within the rst 5 days of therapy. If a
patient had >1 MRSA BSI during the study period, additional
episodes were included if they occurred >60 days after comple-
tion of antibiotic therapy for the previous BSI. The study was
limited to patients who received vancomycin within 48 hours
of index culture collection as this has been identied as the
critical time window for delivery of appropriate antibiotics
for patients with MRSA BSIs [24]. The study was approved
by expedited review by the institutional review board of
Albany Medical Center Hospital, and a HIPAA waiver was
obtained.
Patient Data
Data elements included demographics, medical history, and co-
morbidities [18], recent healthcare institution exposure in the
past 6 months, receipt of antibiotics in the 30 days prior to
the index culture collection, hospitalization history, and CrCl
estimated by the Cockcroft-Gault formula [25] at index culture
collection. Illness severity was dened by the Acute Physiology
and Chronic Health Evaluation II score (based on the worst
physiological score in the 48 hours prior to index culture collec-
tion) [26] and the Chronic Disease ScoreInfectious Diseases
score (determined at admission) [27]. Additional data elements
included source of MRSA BSI, mortality risk associated with in-
fection source [24, 28, 29], presence of infective endocarditis
[30], infection source control intervention, microbiologic data,
treatment data, occurrence of nephrotoxicity (dened as either a
50% or 0.5 mg/dL increase in serum creatinine, whichever was
greater, from initiation of vancomycin to 48 hours postcomple-
tion among patients with a baseline serum creatinine level
<2 mg/dL [2]), and outcomes.
Microbiologic Data and Phenotypic and Genotypic
Characterization
All isolates were identied as S. aureus according to standard
methods. Initial susceptibility testing for oxacillin resistance
was performed according to Clinical and Laboratory Standards
Institute guidelines [11]. Isolates were stored in trypticase soy
broth with 20% glycerol at 70C. All available MRSA BSIs
were shipped to JMI Laboratories (North Liberty, Iowa) for de-
termination of broth microdilution (BMD) MIC [31], Etest
MIC (according to the manufacturers instructions; bioMrieux,
Marcy lEtoile, France), minimum bactericidal concentration
(MBC) [32], heterogeneous vancomycin-intermediate S. aureus
(hVISA) [33], and -lysin activity [34]. Details on these testing
methodologies are provided in the Supplementary Methods.
For patients with multiple MRSA blood cultures, only the
index isolate was considered in the analysis.
Treatment Data
All antibiotic treatment and vancomycin concentration data
were collected. Vancomycin exposure variables were estimated
using ADAPT 5, a computational modeling platform devel-
oped for PK/PD [ pharmacodynamic] applications [20].
Given the time-critical nature of the rst 48 hours of treatment
for MRSA BSIs [24], vancomycin exposure variables were esti-
mated using the maximal a posteriori probability (MAP) proce-
dure in ADAPT 5 for hours 024 (day 1) and 2448 (day 2) [20,
21]. This approach has recently been validated as a way to
Vancomycin ExposureOutcome Analyses CID 2014:59 (1 September) 667

a
t

U
n
i
v
e
r
s
i
d
a
d

d
e

A
n
t
i
o
q
u
i
a

o
n

A
u
g
u
s
t

2
2
,

2
0
1
4
h
t
t
p
:
/
/
c
i
d
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
/
D
o
w
n
l
o
a
d
e
d

f
r
o
m

estimate AUC values with low bias and high precision with
trough-only PK sampling [22].
In brief, the mean parameter vector and covariance matrix
from a previously described 2-compartment vancomycin
model [3537] was embedded in the PRIOR subroutine of
ADAPT 5 (Bayesian prior) [20]. The MAP procedure in estima-
tion module (ID) of ADAPT 5 was then used to estimate the re-
vised probability distribution of a given patients PK parameter
values after dosing and drug concentration data were taken
into account (Bayesian conditional posterior). With the Bayesian
posterior PK information for a given individual, ADAPT 5 was
used to estimate the following vancomycin exposure variables
based on the dosing schedule received: (1) C
min24h
, (2) C
min48h
,
(3) AUC
024h
, and (4) AUC
2448h
. The predictive performance
of the Bayesian approach was assessed by comparing the predict-
ed to the observed concentrations. The predictive performance of
the Cockcroft-Gault formulabased approach using the mean pa-
rameter vector of the PK model [3537] used in this study was
also determined.
Outcomes
Due to the retrospective nature of the study, an objective assess-
ment of treatment failure that included only readily measurable
study endpoints was used [18, 38]. Failure was dened as any of
the following: (1) death within 30 days of index MRSA blood
culture (30-day mortality), (2) microbiological failure (blood
culture growing MRSA obtained 7 days after the initiation of
therapy and before therapy completion), or (3) recurrence of
MRSA bacteremia within 60 days of discontinuation of therapy
[18, 38]. The Social Security Death Index was accessed to deter-
mine the vital status of patients discharged within 30 days of
MRSA bacteremia onset.
Data Analysis Plan
The primary vancomycin exposure variables considered in the
analyses included: (1) C
min24h
/MIC
BMD
, (2) C
min48h
/MIC
BMD
,
(3) AUC
024h
/MIC
BMD
, (4) AUC
2448h
/MIC
BMD
, (5) C
min24h
/
MIC
ETEST
, (6) C
min48h
/MIC
ETEST
, (7) AUC
024h
/MIC
ETEST
,
and (8) AUC
2448h
/MIC
ETEST
. Vancomycin exposure variables
were modeled as both continuous and dichotomous variables.
Bivariate associations between vancomycin exposure variables
and outcomes and baseline covariates and outcomes were as-
sessed using Fisher exact test (categorical variables) and Student
t test and MannWhitney U test (continuous variables). Break-
points in the distribution of continuous vancomycin exposure
variables were sought by Classication and Regression Tree
(CART) analysis [39]. We also examined the relationship be-
tween C
min
15 mg/L and outcomes, given the recent expert
guidelines recommendations [1, 2].
Poisson regression analyses with robust variance estimates
[40, 41] were performed to quantify the association between
each vancomycin exposure variable and failure and 30-day
mortality after adjustment for potential confounding variables.
Each vancomycin exposure variable was evaluated separately.
All potential confounding variables (baseline covariates asso-
ciated with outcomes at P < .2) were included at model entry
and were retained as confounders if the relative risk (RR) of
the vancomycin exposure variable changed by >10%. All calcu-
lations were computed using SAS software, version 9.3 (SAS In-
stitute, Cary, North Carolina) and CART software (Salford
Systems, San Diego, California).
Table 1. Distribution of Microbiologic Phenotypes, Exposure
Variables, and Outcomes in Study Cohort
Characteristic Value
Microbiologic phenotypes
MIC
ETEST
values
Range 0.383.0 mg/L
MIC
50/90
1.5/1.5 mg/L
MIC
BMD
Range 0.383.0 mg/L
MIC
50/90
0.75/1 mg/L
MBC/MIC ratio
Range 121.33
MBC/MIC
50/90
1.3/2.4
hVISA phenotype 4 (3.3%)
Agr dysfunctional 62 (50.4%)
Mean (SD) vancomycin exposure variables
C
min24h
8.6 (4.7)
C
min48h
11.2 (5.9)
C
min24h
/MIC
BMD
11.2 (6.6)
C
min48h
/MIC
BMD
14.8 (8.5)
C
min24h
/MIC
ETEST
7.4 (5.3)
C
min48h
/MIC
ETEST
9.7 (6.8)
AUC
024h
436.4 (162.5)
AUC
2448h
517.3 (197.3)
AUC
024h
/MIC
BMD
571.6 (245.8)
AUC
2448h
/MIC
BMD
680.0 (302.2)
AUC
024h
/MIC
ETEST
380.1 (215.4)
AUC
2448h
/MIC
ETEST
453.9 (273.6)
Outcomes
Failure
a
40 (32.5%)
30-d mortality 25 (20.3%)
Microbiologic failure 15 (12.2%)
Recurrence 10 (8.1%)
Nephrotoxicity
b
19 (17.9%)
Abbreviations: AUC, area under the curve; BMD, broth microdilution; C
min
,
minimum concentration; hVISA, heterogeneous vancomycin-intermediate
S. aureus; MBC, minimum bactericidal concentration; MIC, minimum
inhibitory concentration; MIC
50/90
, minimum concentration that inhibits 50%
and 90% of bacterial isolates; SD, standard deviation.
a
Of the 40 failures, 30 met 1 failure criterion and 10 met 2 criteria.
b
Nephrotoxicity occurrences among the 106 patients with a serum creatinine
level <2 mg/dL.
668 CID 2014:59 (1 September) Lodise et al

a
t

U
n
i
v
e
r
s
i
d
a
d

d
e

A
n
t
i
o
q
u
i
a

o
n

A
u
g
u
s
t

2
2
,

2
0
1
4
h
t
t
p
:
/
/
c
i
d
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
/
D
o
w
n
l
o
a
d
e
d

f
r
o
m

RESULTS
There were 238 unique episodes of MRSA BSIs during the ob-
servation period, and 123 met the inclusion criteria. Reasons for
exclusion were (1) no MRSA isolate (n = 18); (2) receipt of dial-
ysis (n = 58); (3) treatment with alternative anti-MRSA antibi-
otic (n = 20); (4) lack of anti-MRSA antibiotics (n = 10); and
(5) no vancomycin measurements (n = 10). Among cases treat-
ed with alternative antibiotics, most received either linezolid
(n = 11) or daptomycin (n = 5). Distribution of microbiologic
phenotypes, exposure variables, and outcomes among the 123
included cases are provided in Table 1. In situations where
source control was warranted, there was an attempt at source
control in most cases. Among patients who experienced fail-
ure, there was a documented intervention in the medical
record to control the source in >95% of cases.
Results of the observed vs predicted plots for the Bayesian
and formula-based estimation approaches are shown in Figure 1.
There were 282 available concentrations among the 123 cases.
The regression line from the observedpredicted plot for the
Bayesian approach was 0.994 predicted + 0.08, and the R
2
was 0.99 (Figure 1A). The regression line for the formula-
based approach was 0.54 predicted + 8.2 and the R
2
was
0.32 (Figure 1B).
Bivariate comparisons between vancomycin exposure vari-
ables and failure are displayed in Figures 2A and 2B. Break-
points were identied for all continuous vancomycin exposure
variables with CART. Cases with C
min
/MIC
BMD
values in excess
of the CART-derived thresholds had higher incidences of fail-
ure, whereas the inverse was observed for the CART-derived
C
min
/MIC
ETEST
variables (Figure 2A). Similarly, failure was
higher among patients with trough levels 15 mg/L relative to
those with trough levels <15 mg/L on both day 1 (30.6% vs
50.0%, respectively; P = .2) and day 2 (30.1% vs 40.0%, respec-
tively; P = .3). In contrast, failure was uniformly less pronounced
in patients who achieved the AUC/MIC
BMD
and AUC/MIC
ETEST
CART-derived thresholds relative to those who did not (Fig-
ure 2B). For all 4 AUC/MICexposure variables, failure was nearly
doubled in patients with exposures below the CART-derived
thresholds vs those in excess of the breakpoints. Bivariate com-
parisons between the CART-derived vancomycin exposure vari-
ables and 30-day mortality, microbiologic failure, and recurrence
are shown in Table 2. Overall, the ndings between the CART-
derived vancomycin exposure variables and 30-day mortality, mi-
crobiologic failure, and recurrence were generally consistent with
the exposurefailure analyses.
Results of the Poisson regression analyses for failure and
30-day mortality are displayed in Table 3. Baseline covariates
associated with failure or 30-day mortality at a P value <.2
were considered as potential confounders at model entry in
each set of multivariate analyses (Table 4). Overall, no notable
relationships were observed between the CART-derived C
min
/
MIC exposure variables and failure. In contrast, all adjusted rel-
ative risks were <1 for the CART-derived AUC/MIC exposure
variables. Similar to the bivariate analyses, the risk of failure
was 50% lower in those with exposures in excess of the
CART-derived AUC/MIC thresholds relative to those below
the AUC/MIC ratio breakpoints. The results from the 30-day
mortality Poisson regression analyses were largely similar to
the failure Poisson regression analyses. All CART-derived
AUC/MIC exposures and C
min024h
/MIC
ETEST
4.4 were asso-
ciated with a lower risk of 30-day mortality.
DISCUSSION
Using a validated method to determine exposure proles among
patients with trough-only PK sampling [22], the analyses pre-
sented here suggest that AUC/MIC is the pharmacodynamic
index most closely linked to outcomes for vancomycin.
Achievement of exposure values in excess of the CART-derived
AUC/MIC breakpoints were associated with failure rates of
Figure 1. Observed vs predicted concentrations for Bayesian estimation
approach (A) and formula-based approach (B).
Vancomycin ExposureOutcome Analyses CID 2014:59 (1 September) 669

a
t

U
n
i
v
e
r
s
i
d
a
d

d
e

A
n
t
i
o
q
u
i
a

o
n

A
u
g
u
s
t

2
2
,

2
0
1
4
h
t
t
p
:
/
/
c
i
d
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
/
D
o
w
n
l
o
a
d
e
d

f
r
o
m

20%25%, consistent with the near-maximal effect expected for
patients with MRSA BSIs [18, 24, 42]. In contrast, failure to
achieve the CART-derived AUC/MIC exposure thresholds
was associated with failure rates >40%. In the multivariate anal-
yses (Table 3), all RRs were approximately 0.5, indicating that
achievement of CART-derived AUC/MIC exposure thresholds
was associated with a 2-fold decrease in failure. Similarly,
achievement of the CART-derived AUC/MIC exposures was as-
sociated with a 22.5 fold reduction in 30-day mortality. Collec-
tively, these ndings establish the critical importance of the
daily AUC/MIC ratios during the rst 2 days of vancomycin
therapy.
Figure 2. Bivariate relationship between Classication and Regression Tree (CART) analysisderived day 1 and day 2 minimum concentration/minimum
inhibitory concentration (MIC) exposure variables and failure (A) and CART-derived day 1 and day 2 area under the curve/MIC exposure variables and failure
(B). Abbreviations: AUC, area under the curve; BMD, broth microdilution; CART, Classication and Regression Tree; CI, condence interval; MIC, minimum
inhibitory concentration; RR, relative risk.
670 CID 2014:59 (1 September) Lodise et al

a
t

U
n
i
v
e
r
s
i
d
a
d

d
e

A
n
t
i
o
q
u
i
a

o
n

A
u
g
u
s
t

2
2
,

2
0
1
4
h
t
t
p
:
/
/
c
i
d
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
/
D
o
w
n
l
o
a
d
e
d

f
r
o
m

Unfortunately, this study lacked power to discriminate be-
tween AUC/MIC
BMD
and AUC/MIC
ETEST
and day 1 and 2 ex-
posures. All CART-derived AUC/MIC exposure variables
across both time intervals (day 1 and day 2) were found to be
similarly associated with outcomes. Not surprisingly, the
CART-derived exposure AUC/MIC
BMD
and AUC/MIC
ETEST
Table 2. Bivariate Comparisons Between Classication and Regression Tree AnalysisDerived Vancomycin Exposure Variables and
30-Day Mortality, Microbiologic Failure, and Recurrence
Variable 30-d Mortality RR (95% CI) Microbiologic Failure RR (95% CI) Recurrence RR (95% CI)
C
min
/MIC (BMD)
C
min024h
/MIC14.9
a
(n = 28) 11 (39.3) 2.67 (1.375.20) 2 (7.1) 0.52 (.132.18) 1 (3.4) 0.38 (.052.85)
C
min024h
/MIC< 14.9
a
(n = 95) 14 (14.7) 13 (13.7) 9 (9.5)
C
min2448h
/MIC20.4
a
(n =30) 10 (33.3) 2.07 (1.044.10) 5 (16.7) 1.55 (.584.18) 2 (6.7) 0.78 (.173.45)
C
min2448h
/MIC<20.4
a
(n =93) 15 (16.1) 10 (10.8) 8 (8.6)
C
min
/MIC (ETEST)
C
min024h
/MIC4.4
a
(n =91) 15 (16.5) 0.53 (.261.05) 11 (12.1) 0.97 (.332.82) 8 (9.8) 1.41 (.326.28)
C
min024h
/MIC< 4.4
a
(n =32) 10 (31.3) 4 (12.5) 2 (6.3)
C
min2448h
/MIC11.2
a
(n =41) 8 (19.5) 0.94 (.442.00) 3 (7.3) 0.50 (.151.67) 2 (4.9) 0.50 (.112.25)
C
min2448h
/MIC<11.2
a
(n =82) 17 (20.7) 12 (14.6) 8 (9.8)
AUC/MIC (BMD)
AUC
024h
/MIC521
a
(n =67) 11 (16.4) 0.66 (.321.33) 4 (6.0) 0.30 (.10.90) 5 (7.5) 0.84 (.252.74)
AUC
024h
/MIC<521
a
(n =56) 14 (25) 11 (19.6) 5 (8.9)
AUC
2448h
/MIC650
a
(n = 65) 10 (15.4) 0.59 (.291.22) 5 (7.7) 0.45 (.161.23) 5 (7.7) 0.89 (.272.93)
AUC
2448h
/MIC<650
a
(n = 58) 15 (25.9) 10 (17.2) 5 (8.6)
AUC/MIC (ETEST)
AUC
024h
/MIC303
a
(n =73) 9 (12.3) 0.39 (.19.80) 7 (9.6) 0.60 (.231.54) 5 (6.8) 0.68 (.212.24)
AUC
024h
/MIC<303
a
(n =50) 16 (32) 8 (16.0) 5 (10)
AUC
2448h
/MIC320
a
(n = 85) 12 (14.1) 0.41 (.21.82) 9 (10.6) 0.67 (.261.75) 7 (8.2) 1.04 (.293.82)
AUC
2448h
/MIC<320
a
(n = 38) 13 (38.2) 6 (15.8) 3 (7.9)
All data presented as No. (%) unless otherwise noted.
Abbreviations: AUC, area under the curve; BMD, broth microdilution; CI, confidence interval; C
min
, minimum concentration; MIC, minimum inhibitory concentration;
RR, relative risk.
a
Classification and Regression Tree Analysisderived breakpoints.
Table 3. Association Between the Classication and Regression Tree AnalysisDerived Vancomycin Minimum Concentration and Area
Under the Curve Exposure Variables and Overall Failure and 30-Day Mortality in the Poisson Regression Analyses
Exposure
Overall Failure
a
30-d Mortality
b
RR 95% CI P Value RR 95% CI P Value
Day 1 C
min024 h
/MIC
BMD
14.9 1.24 .672.29 .50 1.65 .773.56 .19
C
min024 h
/MIC
ETEST
4.4 0.63 .371.08 .09 0.43 .22.87 .02
AUC
024 h
/MIC
BMD
521 0.54 .32.91 .02 0.43 .20.90 .03
AUC
024 h
/MIC
ETEST
303 0.48 .29.78 .003 0.32 .16.64 .001
Day 2 C
min2448 h
/MIC
BMD
20.4 1.47 .792.54 .24 1.38 .692.75 .36
C
min2448 h
/MIC
ETEST
11.2 0.80 .441.44 .46 0.97 .462.03 .93
AUC
2448 h
/MIC
BMD
650 0.58 .34.99 .05 0.50 .251.02 .06
AUC
2448 h
/MIC
ETEST
> 320 0.53 .32.88 .01 0.49 .24.98 .04
Abbreviations: AUC, area under the curve; BMD, broth microdilution; CI, confidence interval; C
min
, minimum concentration; MBC, minimum bactericidal
concentration; MIC, minimum inhibitory concentration; RR, relative risk.
a
All variables associated with failure at P .2 and considered at model entry included: Acute Physiology and Chronic Health Evaluation II (APACHE-II) score, chronic
obstructive pulmonary disease, diabetes mellitus, malignancy, recent prior surgery, MIC
ETEST
1.5 mg/L, and cumulative number of reduced vancomycin
susceptibility phenotypes.
b
Baseline covariates associated with 30-day mortality at P .2 and considered at model entry included APACHE-II score, malignancy, MIC
ETEST
1.5 mg/L, MIC
BMD
1 mg/L, and MBC/MIC ratio >4.
Vancomycin ExposureOutcome Analyses CID 2014:59 (1 September) 671

a
t

U
n
i
v
e
r
s
i
d
a
d

d
e

A
n
t
i
o
q
u
i
a

o
n

A
u
g
u
s
t

2
2
,

2
0
1
4
h
t
t
p
:
/
/
c
i
d
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
/
D
o
w
n
l
o
a
d
e
d

f
r
o
m

Table 4. Bivariate Comparisons of Baseline Features Between Failures and 30-Day Mortality
Variable Failure (n =40) Nonfailure (n =83) RR (95% CI) Died (n =25) Survived (n =98) RR (95% CI)
Demographic and clinical characteristics
Age, y, mean (SD) 63.5 (13.1) 60.3 (16.1) 64.1 (14.4) 60. (15.4
Male sex 26 (65.0) 59 (71.1) 0.83 (.491.40) 15 (60.0) 70 (71.4) 0.67 (.331.35)
Weight, kg, mean (SD) 82.7 (30.1) 85.8 (29.8) 81.7 (36.4) 85.5 (28.1)
Recent residence in healthcare
institution
26 (65.0) 50 (60.2) 1.15 (.671.97) 15 (60.0) 61 (62.2) 0.93 (.451.89)
Diabetes mellitus 13 (32.5) 41 (49.4) 0.62 (.351.07) 9 (36.0) 45 (45.9) 0.72 (.341.50)
Heart failure 9 (22.5) 29 (34.9) 0.65 (.341.23) 6 (24.0) 32 (32.7) 0.71 (.311.63)
COPD 17 (42.5) 22 (26.5) 1.59 (.972.62) 10 (40.0) 29 (29.6) 1.44 (.712.90)
Hepatic dysfunction 2 (5.0) 3 (3.6) 1.24 (.413.75) 2 (8.0) 3 (3.1) 2.1 (.666.38)
Malignancy 16 (40.0) 18 (21.7) 1.75 (1.062.86) 10 (40.0) 24 (24.5) 1.75 (.873.50)
Decubitus ulcers 8 (20.0) 19 (22.9) 0.89 (.471.7) 5 (20.0) 22 (22.5) 0.89 (.372.15)
Immunosuppressants 10 (25.0) 13 (15.7) 1.45 (.832.52) 6 (24.0) 17 (17.4) 1.37 (.623.05)
HIV 1 (2.5) 3 (3.6) 0.76 (.144.25) 1 (4.0) 3 (3.1) 1.24 (.227.02)
History of cerebrovascular event 2 (5.0) 11 (13.3) 0.45 (.121.64) 1 (4.0) 12 (12.2) 0.35 (.052.40)
Recent surgery in previous 30 d 18 (45.0) 25 (30.1) 1.52 (.922.51) 11 (44.0) 32 (32.7) 1.46 (.732.94)
Recent antibiotics in previous
30 d
21 (52.5) 47 (56.6) 0.89 (.541.49) 11 (44.0) 57 (58.2) 0.64 (.311.29)
Prior vancomycin in past 30 d 6 (15.0) 16 (19.3) 0.81 (.391.69) 3 (12.0) 19 (19.4) 0.63 (.211.91)
Length of stay in days prior to
index culture collection,
median (IQR)
4 (016.75) 0 (012) 4 (014.5) 0.5 (013.2)
Residence in ICU prior to onset 15 (37.5) 27 (32.5) 1.16 (.691.94) 10 (40.0) 32 (32.7) 1.29 (.632.61)
Intensive care unit at onset 10 (25.0) 20 (24.1) 1.03 (.581.86) 7 (28.0) 23 (23.5) 1.21 (.562.60)
Baseline CrCl, mL/min, mean
(SD)
74.7 (52.9) 73.5 (49.9) 84.2 (58.8) 71.3 (48.4)
APACHE-II score, mean (SD) 16.7 (6.4) 12.4 (5.5) 17.7 (6.7) 12.8 (5.6)
APACHE-II score 20 13 (32.5) 10 (12.0) 2.09 (1.293.39) 9 (36.0) 14 (14.3) 2.45 (1.244.82)
CDS-ID score at admission,
mean (SD)
2.54 (1.86) 2.89 (2.31) 2.40 (1.68) 2.87 (2.28)
Source of bacteremia
Intravenous catheter 17 (42.5) 32 (38.6) 1.12 (.671.86) 11 (44.0) 38 (38.8) 1.19 (.592.40)
Skin and soft tissue 8 (20.0) 22 (26.5) 0.78 (.401.49) 4 (16.0) 26 (26.5) 0.59 (.221.58)
Bone and joint 1 (2.5) 7 (8.4) 0.37 (.062.35) 1 (4.0) 7 (7.1) 0.60 (.093.88)
Respiratory tract 2 (5.0) 7 (8.4) 0.67 (.192.33) 2 (8.0) 7 (7.1) 1.10 (.313.94)
Intra-abdominal 3 (7.5) 5 (6.0) 1.17 (.462.96) 2 (8.0) 6 (6.1) 1.25 (.364.38)
Urinary tract 0 (0) 2 (2.4) NA 0 (0) 2 (2.0) NA
Infected graft/device 3 (7.5) 4 (4.8) 1.34 (.553.29) 1 (4.0) 6 (6.1) 0.69 (.114.39)
Unknown 6 (15.0) 4 (4.8) 1.99 (1.123.56) 4 (16.0) 6 (6.1) 2.15 (.925.04)
Infective endocarditis 8 (20.0) 9 (10.8) 1.56 (.872.79) 3 (12.0) 14 (14.3) 0.85 (.292.53)
High-risk source 26 (65.0) 52 (62.7) 1.07 (.631.83) 15 (60.0) 63 (64.3) 0.87 (.431.76)
Microbiologic phenotypes
MIC
BMD
1 mg/L 10 (25.0) 15 (18.1) 1.31 (.742.30) 2 (8.0) 23 (23.5) 0.34 (.091.35)
MIC
ETEST
1.5 mg/L 28 (70.0) 46 (55.4) 1.55 (.872.74) 18 (72.0) 56 (57.1) 1.70 (.773.77)
MBC/MIC ratio >4 3 (7.5) 4 (4.8) 1.34 (.553.29) 3 (12.0) 4 (4.1) 2.26 (.895.75)
hVISA 2 (5.0) 2 (2.4) 1.57 (.574.32) 0 (0) 4 (4.1) NA
Agr dysfunction 18 (45.0) 44 (53.3) 0.81 (.481.34) 10 (40.0) 52 (53.1) 0.66 (.321.34)
Concomitant antibiotics administered >48 h
Aminoglycosides 10 (25.0) 15 (18.1) 1.31 (.742.30) 7 (28.0) 18 (18.4) 1.52 (.723.24)
Clindamycin 0 (0) 1 (1.2) NA 0 (0) 1 (1.0) NA
Daptomycin 1 (2.5) 1 (1.2) 1.55 (.386.35) 0 (0) 2 (2.0) NA
Linezolid 3 (7.5) 11 (13.3) 0.63 (.221.78) 2 (8.0) 12 (12.2) 0.68 (.182.57)
672 CID 2014:59 (1 September) Lodise et al

a
t

U
n
i
v
e
r
s
i
d
a
d

d
e

A
n
t
i
o
q
u
i
a

o
n

A
u
g
u
s
t

2
2
,

2
0
1
4
h
t
t
p
:
/
/
c
i
d
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
/
D
o
w
n
l
o
a
d
e
d

f
r
o
m

variables were concordant in >80% of cases. Given that MICs
are largely unknown until day 3 of therapy, these ndings sug-
gest that clinicians should target daily AUCs that provide ade-
quate coverage against the common MIC values observed in
their practices. For most institutions, this will be an MIC
BMD
of 1 mg/L or MIC
ETEST
of 1.52 mg/L, which translates into
daily targeted AUC values of 550650 mg h/L. Software pro-
grams are readily available to estimate AUC based on collected
vancomycin trough levels [22, 43, 44]. Alternatively, vancomy-
cin AUCs can be calculated using formulas based on the collec-
tion of a few timed measurements [10]. As with all observational
studies, these ndings should be interpreted cautiously and
need to be validated in a multicenter randomized trial before
adoption into practice. However, it is important to note that
the current AUC/MIC ratio >400 PK/PD target for vancomy-
cin is based on studies [5, 7, 8] with designs similar to this one.
Two additional ndings bear mention here. First, no note-
worthy relationships between C
min
/MIC and outcomes were
observed in this study. Although expert guidelines recommend
trough monitoring [1, 2], this was not an unexpected nding
from an exposureresponse perspective. Troughs are a single
point estimate of the concentrationtime prole at the end of the
dosing interval. Whereas a trough ensures the achievement of a
minimum cumulative exposure, a wide range of concentration
time proles can result in a given C
min
[10]. Because cumulative
exposure proles associated with a given trough value are highly
variable [10], it is not surprising that C
min
was found to be un-
informative and nondiscriminatory. These ndings further sup-
port the notion that AUC monitoring should be incorporated
into clinical practice, as it has been identied as the key phar-
macodynamic index for vancomycin across a growing number
of in vitro, animal model, and clinical studies [18, 45].
Second, whereas the observed vs predicted plots for the Baye-
sian approach showed slopes and intercepts very close to the
ideal values of 1.0 and 0.0, respectively, the formula-based ap-
proach did not t the data well and only explained 35% of the
variance. The poor t associated with the Cockcroft-Gault
formula best approach does not come as a surprise. This nding
was consistent with older gentamicin PK data, which demon-
strated that CrCL-based estimation formulas did not accurately
predict the observed concentrationtime proles of gentamicin,
an antibiotic that is predominately cleared by the kidneys, sim-
ilar to vancomycin [9]. This nding further supports the notion
that individualized estimates of exposure proles based on mea-
sured vancomycin concentrations are required when evaluating
vancomycin exposureresponse relationships in clinical prac-
tice. Interestingly, the formula-based approach underestimated
exposure proles by approximately 40%50% (overestimated
clearance by 40%50%). If one considers that AUC = dose/
clearance, our daily AUC target of 550600 aligns closely
with previous evaluations [5, 7, 8] that noted the critical
AUC/MIC target to be approximately 400.
Several caveats should be noted when interpreting these nd-
ings. First, this was a study of adult, non-neutropenic, non-
dialysis patients. It is unknown if the observed ndings are
applicable to other populations. Second, only a limited number
of isolates had MIC
BMD
>1 mg/L. The current paradigm in
PK/PD is that a doubling of exposure is required with each
log
2
increase in MIC values. Before we denitively recommend
a daily AUC of 1300 for an MIC
BMD
of 2 mg/L, additional re-
search is needed. If future studies indicate the pharmacodynamic
target is consistent with the AUC/MIC
BMD
targets observed in this
study when the MIC
BMD
is 2 mg/L, it will be difcult to use van-
comycin in these instances as the vancomycin AUC needed for
effect will be associated with nephrotoxicity rates >30% [21].
We did not attempt to determine if 30-day mortality was at-
tributable to the MRSA BSI. Rather than basing microbiological
failure on persistent signs and symptoms of infection, treatment
was considered a microbiological failure only if the duration of
bacteremia was 7 days, as proposed by a number of authors
[18, 27, 38]. We believe these aforementioned denitions allow
for an objective assessment of the endpoints and minimize any
Table 4 continued.
Variable Failure (n =40) Nonfailure (n =83) RR (95% CI) Died (n =25) Survived (n =98) RR (95% CI)
Macrolide 3 (7.5) 2 (2.4) 1.91 (.894.12) 3 (12.0) 2 (2.0) 3.22 (1.437.23)
Rifampin 3 (7.5) 13 (15.7) 0.54 (.191.55) 1 (4.0) 15 (15.3) 0.28 (.041.92)
Tigecycline 2 (5.0) 1 (1.2) 0.83 (.491.40) 1 (4.0) 2 (2.0) 1.67 (.328.59)
TMP/SMX 1 (2.5) 8 (9.6) 0.32 (.052.10) 0 (0) 9 (9.2) NA
All data are presented as No. (%) unless otherwise noted. Baseline covariates associated with failure at P .2 included APACHE-II score, COPD, diabetes mellitus,
malignancy, recent prior surgery, MIC
ETEST
1.5 mg/L, and cumulative number of reduced vancomycin susceptibility phenotypes. Baseline covariates associated
with 30-day mortality at P .2 included APACHE-II score, malignancy, MIC
ETEST
1.5 mg/L, MIC
BMD
1 mg/L, and MBC/MIC ratio >4.
Abbreviations: APACHE, Acute Physiology and Chronic Health Evaluation; BMD, broth microdilution; CDS-ID, Chronic Disease ScoreInfectious Diseases; CI,
confidence interval; C
min
, minimum concentration; COPD, chronic obstructive pulmonary disease; CrCl, creatinine clearance; HIV, human immunodeficiency
virus; hVISA, heterogeneous vancomycin-intermediate S. aureus; ICU, intensive care unit; IQR, interquartile range; MBC, minimum bactericidal concentration;
MIC, minimum inhibitory concentration; NA, not available; RR, relative risk; SD, standard deviation; TMP-SMX, trimethoprim-sulfamethoxazole.
Vancomycin ExposureOutcome Analyses CID 2014:59 (1 September) 673

a
t

U
n
i
v
e
r
s
i
d
a
d

d
e

A
n
t
i
o
q
u
i
a

o
n

A
u
g
u
s
t

2
2
,

2
0
1
4
h
t
t
p
:
/
/
c
i
d
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
/
D
o
w
n
l
o
a
d
e
d

f
r
o
m

subjective biases that may result from assessing and interpreting
retrospective clinical data. Another consideration in the evalu-
ation of MRSA bloodstream infection studies is the adequacy of
source control. In situations when source control was warrant-
ed, there was an attempt to remove the catheter/debride the
wound in almost all cases. Among patients who experienced a
failure, there was a documented intervention in the medical re-
cord to control the source in >95% of cases. The frequency of
source control attempts were not different between CART-
derived exposure groups.
In conclusion, our ndings suggest that that AUC/MIC, not
C
min
/MIC, is the pharmacodynamic index most closely linked
to outcomes for patients with MRSA BSIs. Clinicians should
conservatively target AUC values needed to provide adequate
exposure against the common MIC values observed in their in-
stitution, given that MICs are largely unknown until therapy
day 3. Further research is still needed among patients with
MRSA BSIs with vancomycin MIC
BMD
>1 mg/L. As this was
a retrospective observational study, these ndings need to be
validated in a multicenter vancomycin AUC dose-optimized
randomized outcomes trial before they can be incorporated
into clinical practice.
Supplementary Data
Supplementary materials are available at Clinical Infectious Diseases online
(http://cid.oxfordjournals.org). Supplementary materials consist of data
provided by the author that are published to benet the reader. The posted
materials are not copyedited. The contents of all supplementary data are the
sole responsibility of the authors. Questions or messages regarding errors
should be addressed to the author.
Notes
Acknowledgments. We extend gratitude to Nadia El-Fawal, Jill Butter-
eld, Benjamin Woo, and Rasha Masoud for data collection and database
entry, and to Ron Jones, MD, and Rodrigo E. Mendes, PhD, at JMI Labora-
tories (North Liberty, Iowa) for characterizing the phenotypic and genotypic
proles of the MRSA isolates. This article has greatly beneted from the
thoughtful editing (grammar and spelling) of Allison Krug, who was paid
via grant funding.
Disclaimer. Cubist provided support to complete the project and was
not involved in the design and conduct of the study; collection, manage-
ment, analysis, and interpretation of the data; or preparation and review
of the manuscript.
Financial support. This work was supported by an investigator-initiated
research grant from Cubist Pharmaceuticals ( principal investigator, T. P. L.).
Potential conicts of interest. T. P. L. is a consultant for Cubist. G. L. D.
is a consultant for Cubist Pharmaceuticals. All other authors report no
potential conicts.
All authors have submitted the ICMJE Form for Disclosure of Potential
Conicts of Interest. Conicts that the editors consider relevant to the con-
tent of the manuscript have been disclosed.
References
1. Liu C, Bayer A, Cosgrove SE, et al. Clinical practice guidelines by
the Infectious Diseases Society of America for the treatment of
methicillin-resistant Staphylococcus aureus infections in adults and
children: executive summary. Clin Infect Dis 2011; 52:28592.
2. Rybak M, Lomaestro B, Rotschafer JC, et al. Therapeutic monitoring of
vancomycin in adult patients: a consensus review of the American So-
ciety of Health-System Pharmacists, the Infectious Diseases Society of
America, and the Society of Infectious Diseases Pharmacists. Am J
Health Syst Pharm 2009; 66:8298.
3. Craig WA. Basic pharmacodynamics of antibacterials with clinical ap-
plications to the use of beta-lactams, glycopeptides, and linezolid. Infect
Dis Clin North Am 2003; 17:479501.
4. Drusano GL. Antimicrobial pharmacodynamics: critical interactions of
bug and drug. Nat Rev Microbiol 2004; 2:289300.
5. Moise-Broder PA, Forrest A, Birmingham MC, Schentag JJ. Pharmaco-
dynamics of vancomycin and other antimicrobials in patients with
Staphylococcus aureus lower respiratory tract infections. Clin Pharma-
cokinet 2004; 43:92542.
6. Rybak MJ. The pharmacokinetic and pharmacodynamic properties of
vancomycin. Clin Infect Dis 2006; 42(suppl 1):S359.
7. Kullar R, Leonard SN, Davis SL, et al. Validation of the effectiveness of a
vancomycin nomogram in achieving target trough concentrations of
15-20 mg/L suggested by the vancomycin consensus guidelines. Phar-
macotherapy 2011; 31:4418.
8. Holmes NE, Turnidge JD, Munckhof WJ, et al. Vancomycin AUC/MIC
ratio and 30-day mortality in patients with Staphylococcus aureus bac-
teremia. Antimicrob Agents Chemother 2013; 57:165463.
9. Lesar TS, Rotschafer JC, Strand LM, Solem LD, Zaske DE. Gentamicin
dosing errors with four commonly used nomograms. JAMA 1982;
248:11903.
10. Patel N, Pai MP, Rodvold KA, Lomaestro B, Drusano GL, Lodise TP.
Vancomycin: we cant get there from here. Clin Infect Dis 2011;
52:96974.
11. Clinical and Laboratory Standards Institute. Performance standards for
antimicrobial disk susceptibility tests; approved standards9th ed.
CLSI document M2-M9. Wayne, PA: CLSI, 2006.
12. Jung Y, Song KH, Cho J, et al. Area under the concentration-time curve
to minimum inhibitory concentration ratio as a predictor of vancomy-
cin treatment outcome in methicillin-resistant Staphylococcus aureus
bacteraemia. Int J Antimicrob Agents 2014; 43:17983.
13. Brown J, Brown K, Forrest A. Vancomycin AUC24/MIC ratio in pa-
tients with complicated bacteremia and infective endocarditis due to
methicillin-resistant Staphylococcus aureus and its association with at-
tributable mortality during hospitalization. Antimicrob Agents Chemo-
ther 2012; 56:6348.
14. Chung J, Oh JM, Cho EM, et al. Optimal dose of vancomycin for treat-
ing methicillin-resistant Staphylococcus aureus pneumonia in critically
ill patients. Anaesth Intensive Care 2011; 39:10307.
15. Hermsen ED, Hanson M, Sankaranarayanan J, Stoner JA, Florescu MC,
Rupp ME. Clinical outcomes and nephrotoxicity associated with vanco-
mycin trough concentrations during treatment of deep-seated infec-
tions. Expert Opin Drug Saf 2010; 9:914.
16. Hidayat LK, Hsu DI, Quist R, Shriner KA, Wong-Beringer A. High-dose
vancomycin therapy for methicillin-resistant Staphylococcus aureus in-
fections: efcacy and toxicity. Arch Intern Med 2006; 166:213844.
17. Kralovicova K, Spanik S, Halko J, et al. Do vancomycin serum levels pre-
dict failures of vancomycin therapy or nephrotoxicity in cancer pa-
tients? J Chemother 1997; 9:4206.
18. Lodise TP, Graves J, Evans A, et al. Relationship between vancomycin
MIC and failure among patients with methicillin-resistant Staphylococ-
cus aureus bacteremia treated with vancomycin. Antimicrob Agents
Chemother 2008; 52:331520.
19. Zimmermann AE, Katona BG, Plaisance KI. Association of vancomycin
serum concentrations with outcomes in patients with gram-positive
bacteremia. Pharmacotherapy 1995; 15:8591.
20. DArgenio DZ, Schumitzky A, Wang X. ADAPT 5 Users guide: phar-
macokinetic/pharmacodynamic systems analysis software. Los Angeles:
Biomedical Simulations Resource, 2009.
674 CID 2014:59 (1 September) Lodise et al

a
t

U
n
i
v
e
r
s
i
d
a
d

d
e

A
n
t
i
o
q
u
i
a

o
n

A
u
g
u
s
t

2
2
,

2
0
1
4
h
t
t
p
:
/
/
c
i
d
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
/
D
o
w
n
l
o
a
d
e
d

f
r
o
m

21. Lodise TP, Patel N, Lomaestro BM, Rodvold KA, Drusano GL. Relation-
ship between initial vancomycin concentration-time prole and
nephrotoxicity among hospitalized patients. Clin Infect Dis 2009;
49:50714.
22. Neely MN, Youn G, Jones B, et al. Are vancomycin troughs adequate for
optimal dosing? Antimicrob Agents Chemother 2014; 58:30916.
23. Garner JS, Jarvis WR, Emori TG, Horan TC, Hughes JM. CDCdenitions
for nosocomial infections, 1988. Am J Infect Control 1988; 16:12840.
24. Lodise TP, McKinnon PS, Swiderski L, Rybak MJ. Outcomes analysis of
delayed antibiotic treatment for hospital-acquired Staphylococcus aure-
us bacteremia. Clin Infect Dis 2003; 36:141823.
25. Cockcroft DW, Gault MH. Prediction of creatinine clearance from
serum creatinine. Nephron 1976; 16:3141.
26. Knaus WA, Draper EA, Wagner DP, Zimmerman JE. APACHE II: a se-
verity of disease classication system. Crit Care Med 1985; 13:81829.
27. McGregor JC, Rich SE, Harris AD, et al. A systematic review of the meth-
ods used to assess the association between appropriate antibiotic therapy
and mortality in bacteremic patients. Clin Infect Dis 2007; 45:32937.
28. Soriano A, Martinez JA, Mensa J, et al. Pathogenic signicance of meth-
icillin resistance for patients with Staphylococcus aureus bacteremia.
Clin Infect Dis 2000; 30:36873.
29. Blot SI, Vandewoude KH, Hoste EA, Colardyn FA. Outcome and attrib-
utable mortality in critically ill patients with bacteremia involving meth-
icillin-susceptible and methicillin-resistant Staphylococcus aureus. Arch
Intern Med 2002; 162:222935.
30. Li JS, Sexton DJ, Mick N, et al. Proposed modications to the Duke cri-
teria for the diagnosis of infective endocarditis. Clin Infect Dis 2000;
30:6338.
31. Clinical and Laboratory Standards Institute. Methods for dilution anti-
microbial susceptibility tests for bacteria that growaerobically; approved
standard9th ed. CLSI document M07-A9. Wayne, PA: CLSI, 2012.
32. Moody J, Knapp C. Tests to assess bactericidal activity. In: Insenberg
HD, ed. Clinical microbiology procedures handbook. Washington,
DC: ASM Press: 5.10.11.1115.10.13.16, 2004.
33. Wootton M, MacGowan AP, Walsh TR, Howe RA. A multicenter study
evaluating the current strategies for isolating Staphylococcus aureus
strains with reduced susceptibility to glycopeptides. J Clin Microbiol
2007; 45:32932.
34. Sakoulas G, Eliopoulos GM, Moellering RC Jr, et al. Accessory gene reg-
ulator (agr) locus in geographically diverse Staphylococcus aureus
isolates with reduced susceptibility to vancomycin. Antimicrob Agents
Chemother 2002; 46:1492502.
35. Pryka RD, Rodvold KA, Garrison M, Rotschafer JC. Individualizing
vancomycin dosage regimens: one- versus two-compartment Bayesian
models. Ther Drug Monit 1989; 11:4504.
36. Rodvold KA, Pryka RD, Garrison M, Rotschafer JC. Evaluation of a
two-compartment Bayesian forecasting program for predicting vanco-
mycin concentrations. Ther Drug Monit 1989; 11:26975.
37. Drusano GL, Ambrose PG, Bhavnani SM, et al. Vancomycin dose
recommendations for hospital-, ventilator- or health care-associated
pneumonia and the attainment of vancomycin trough concentrations
of 15-20 mg/L: cognitive dissonance. In: 45th Annual Meeting of the
Infectious Diseases Society of America, 47 October, 2007, San
Diego, CA.
38. Jenkins TC, Price CS, Sabel AL, Mehler PS, Burman WJ. Impact of rou-
tine infectious diseases service consultation on the evaluation, manage-
ment, and outcomes of Staphylococcus aureus bacteremia. Clin Infect
Dis 2008; 46:10008.
39. Zhang H, Burthon S. Recursive partitioning in the health sciences.
New York: Springer, 1999.
40. McNutt LA, Wu C, Xue X, Hafner JP. Estimating the relative risk in co-
hort studies and clinical trials of common outcomes. Am J Epidemiol
2003; 157:9403.
41. Spiegelman D, Hertzmark E. Easy SAS calculations for risk or
prevalence ratios and differences. Am J Epidemiol 2005; 162:
199200.
42. Fowler VG Jr, Olsen MK, Corey GR, et al. Clinical identiers of com-
plicated Staphylococcus aureus bacteremia. Arch Intern Med 2003;
163:206672.
43. Fuchs A, Csajka C, Thoma Y, Buclin T, Widmer N. Benchmarking ther-
apeutic drug monitoring software: a review of available computer tools.
Clin Pharmacokinet 2013; 52:922.
44. Suzuki Y, Kawasaki K, Sato Y, et al. Is peak concentration needed
in therapeutic drug monitoring of vancomycin? A pharmacokinetic-
pharmacodynamic analysis in patients with methicillin-resistant Staph-
ylococcus aureus pneumonia. Chemotherapy 2012; 58:30812.
45. Nicasio AM, Bulitta JB, Lodise TP, et al. Evaluation of once-daily van-
comycin against methicillin-resistant Staphylococcus aureus in a hol-
low-ber infection model. Antimicrob Agents Chemother 2012;
56:6826.
Vancomycin ExposureOutcome Analyses CID 2014:59 (1 September) 675

a
t

U
n
i
v
e
r
s
i
d
a
d

d
e

A
n
t
i
o
q
u
i
a

o
n

A
u
g
u
s
t

2
2
,

2
0
1
4
h
t
t
p
:
/
/
c
i
d
.
o
x
f
o
r
d
j
o
u
r
n
a
l
s
.
o
r
g
/
D
o
w
n
l
o
a
d
e
d

f
r
o
m