Integrating Rapid Pathogen Identification

and Antimicrobial Stewardship

Significantly Decreases Hospital Costs
Katherine K. Perez, PharmD; Randall J. Olsen, MD, PhD; William L. Musick, PharmD; Patricia L. Cernoch, BS; James R. Davis, PhD;
Geoffrey A. Land, PhD; Leif E. Peterson, PhD; James M. Musser, MD, PhD

 Context.Early diagnosis of gram-negative bloodstream

infections, prompt identification of the infecting organism,
and appropriate antibiotic therapy improve patient care
outcomes and decrease health care expenditures. In an era
of increasing antimicrobial resistance, methods to acquire
and rapidly translate critical results into timely therapies
for gram-negative bloodstream infections are needed.
Objective.To determine whether mass spectrometry
technology coupled with antimicrobial stewardship provides a substantially improved alternative to conventional
laboratory methods.
Design.An evidence-based intervention that integrated matrix-assisted laser desorption and ionization time-offlight mass spectrometry, rapid antimicrobial susceptibility
testing, and nearreal-time antimicrobial stewardship
practices was implemented. Outcomes in patients hospitalized prior to initiation of the study intervention were
compared to those in patients treated after implementation. Differences in length of hospitalization and hospital
costs were assessed in survivors.

Results.The mean hospital length of stay in the

preintervention group survivors (n 100) was 11.9 versus
9.3 days in the intervention group (n 101; P .01). After
multivariate analysis, factors independently associated
with decreased length of hospitalization included the
intervention (hazard ratio, 1.38; 95% confidence interval,
1.011.88) and active therapy at 48 hours (hazard ratio,
2.9; confidence interval, 1.157.33). Mean hospital costs
per patient were $45 709 in the preintervention group and
$26 162 in the intervention group (P .009).
Conclusions.Integration of rapid identification and
susceptibility techniques with antimicrobial stewardship
significantly improved time to optimal therapy, and it
decreased hospital length of stay and total costs. This
innovative strategy has ramifications for other areas of
patient care.
(Arch Pathol Lab Med. 2013;137:12471254; doi:
10.5858/arpa.2012-0651-OA)

identification of bacteria and yeast grown on agar media, and

more recently has been used to identify gram-negative
organisms directly from positive blood culture bottles within
minutes.912 Further, testing of specimens taken directly from
positive blood cultures has been reported to decrease the
time required for antimicrobial susceptibility testing results.12
With the enhanced speed and technical advances made
possible by MS, in principle, opportunities exist for
antimicrobial stewardship practices to enhance antimicrobial management.8,13,14 However, the clinical and health care
cost implications of integrating new molecular diagnostic
microbiology techniques with antimicrobial stewardship
practices have not been well established. Here, we tested
the hypothesis that patient care would be enhanced by
combining new rapid methods to obtain identification and
susceptibility results with real-time clinical interpretation
and action by an infectious diseases pharmacist.

rompt and aggressive initiation of antimicrobial therapy

is the mainstay of treatment for patients with bloodstream infections (BSIs).13 Moreover, increasing rates of
drug resistance have forced clinicians to expose patients
with presumed bacterial infection to treatment with empiric
combination broad-spectrum antibiotics.48 It follows that
the time required for bacterial identification and drug
susceptibility testing has a critical impact on guiding
therapy.
Matrix-assisted laser desorption ionization time-of-flight
mass spectrometry (MALDI-TOF MS) analysis of microbial
proteins is a new method for rapid and reliable species
Accepted for publication October 30, 2012.
Published as an Early Online Release December 6, 2012.
From the Departments of Pharmacy (Drs Perez and Musick) and
Pathology and Genomic Medicine (Drs Olsen, Davis, Land, and
Musser, and Ms Cernoch), The Methodist Hospital, Houston, Texas;
and the Center for Biostatistics (Dr Peterson), The Methodist Hospital
Research Institute, Houston, Texas.
The authors have no relevant financial interest in the products or
companies described in this article.
Reprints: James M. Musser, MD, PhD, Department of Pathology
and Genomic Medicine, The Methodist Hospital, 6565 Fannin, B490, Houston, TX 77030 (e-mail: jmmusser@tmhs.org).
Arch Pathol Lab MedVol 137, September 2013

MATERIALS AND METHODS

Study Location and Patient Population
We implemented our study intervention at The Methodist
Hospital in Houston, Texas, a 1000-bed quaternary care academic
hospital on February 1, 2012. Eligibility screening was conducted
Rapid Diagnostics and Stewardship Reduce CostPerez et al 1247

among consecutive patients ages 18 years and older with 1 or more

blood cultures with a gram-negative organism who were hospitalized between August 15, 2011, and November 30, 2011 (preintervention period), and February 1, 2012, through May 25, 2012
(intervention period). For simplicity, only the first BSI episode was
evaluated for each patient; subsequent episodes that occurred in
the study period were excluded from the analysis. Patients were
excluded if the pathogen ultimately identified was not a facultative
anaerobic or aerobic gram-negative bacillus; if the patient had a
polymicrobial BSI, experienced prior or subsequent BSIs with an
organism(s) other than the study pathogen, or died prior to the
index blood culture becoming positive; and/or if criteria for
discharge were determined by institutional policies rather than at
the treating physicians discretion (ie, medical circumstances other
than the BSI required extended hospitalization). This study was
approved by the Institutional Review Board of The Methodist
Hospital Research Institute (IRB1011-0200). No private sector
funds were used in support of this study. The authors vouch for the
accuracy and completeness of the reported data and the fidelity of
this report to the study protocol.

Data Collection
Patients were identified through the clinical microbiology
laboratory records for the designated study periods. We collected
data on demographic characteristics, manifestations of BSI,
microbiology results, antibiotic therapy, and severity of illness.15,16
Cost data were obtained from the institutions financial management department. The Acute Physiology and Chronic Health
Evaluation II (APACHE II) score was calculated based on clinical
data present during the 24 hours preceding the index blood culture.
Underlying illnesses, recent hospitalizations (previous 90 days),
and reason for admission were noted. Immunosuppressive therapy
was defined as receipt of cytotoxic agents within 6 weeks,
corticosteroids at a dosage of 15 mg or more of prednisolone daily
for longer than 1 week within 4 weeks, or other immunosuppressive agents within 2 weeks before bacteremia onset. Bloodstream
infection onset was defined as the time the first blood sample
yielding the study isolate (index blood culture) was collected.
Infection-related characteristics examined were infection source;
pathogen species and susceptibility data; and time, dose, and route
of therapy with individual antimicrobial agents relative to time of
index culture collection. The source of bacteremia was determined
according to the definitions published by the Centers for Disease
Control and Prevention.17 In each case, an effort was made to
establish a primary focus of infection.
Appropriateness of antibiotic therapy was assessed on a case-bycase basis by an infectious diseasestrained pharmacist and was
evaluated at BSI onset, at the index culture time-to-positivity (TTP),
and at 24 and 48 hours after BSI onset in the 2 groups. Active
therapy was defined as when the regimen included 1 or more
antimicrobial agents to which the causative pathogen was
susceptible in vitro. Therapy was appropriate when the administered regimen was active and, when available, was in accordance
with current clinical guidelines regarding dosing and route of
administration.16 Therapy was defined as inactive if the blood
isolate was resistant to the agent(s) used or in the absence of any
antibacterials.
Empiric therapy refers to antimicrobial agents administered during
the period before identification of the blood culture isolate and
susceptibility results were available. De-escalation was defined as
switching to a narrower-spectrum agent or decreasing the number
of antibiotics from 2 or more agents to a single agent when clinically
appropriate. Unnecessary gram-positive coverage was defined as
empiric use of an antigram-positive agent (eg, glycopeptides)
without subsequent microbiologic or clinical indication. Therapy was
considered optimized when antibiotics were de-escalated based on
available identification and susceptibility testing results, when dosing
or administration route modifications were made based on organ
function, or if antimicrobial spectrum was expanded based on
patient-specific history or local antibiogram in the absence of final
pathogen susceptibility testing results. Patients who initially received
1248 Arch Pathol Lab MedVol 137, September 2013

broad-spectrum or combined empiric therapy were considered

candidates for treatment de-escalation.
Duration of hospital and intensive care unit stays were
determined as the difference in days between admission and
discharge. Inasmuch as death can artificially decrease length of stay
(LOS), LOS analysis was done solely with patients who survived to
hospital discharge.

Study Design
All blood specimens obtained in each of the 2 study periods were
processed identically using the BACTEC FX automated blood
culture system (BD Diagnostics, Sparks, Maryland) and standard
aerobic and anaerobic blood culture media. Specimens were Gram
stained when the blood culture bottle became positive. In the
preintervention phase, microbiology laboratory personnel directly
notified the nursing staff with the Gram stain result. Additionally, if
an infectious diseases physician was recorded on the provider list,
the individual was called with the result. Positive blood culture
specimens were inoculated on appropriate solid agar media and
subsequently identified by conventional clinical microbiology
procedures. The final organism identification and antimicrobial
susceptibilities were performed using the BD Phoenix system (BD
Diagnostics). Once obtained, the results were reported to the
electronic medical record without additional active notification of
the patient care team. Multidrug resistant organisms (MDROs)
and/or extended-spectrum beta-lactamase producing (ESBL) organisms were telephoned to the appropriate nursing unit during
both study periods. Susceptibility testing was performed according
to guidelines and breakpoints established by the Clinical Laboratory and Standards Institute.18

The Intervention
After a validation process we published recently,12 the clinical
microbiology laboratory implemented MALDI-TOF MS (Bruker
Daltonics, Fremont, California) for routine species identification of
gram-negative bacteria directly from early-positive blood cultures.
Immediately after the Gram stain results were obtained, the
microbiology laboratory staff telephoned the appropriate member
of the patient care team, a procedure identical to that used before
the intervention. If a gram-negative isolate was identified, the
specimen was then analyzed by the MALDI-TOF MS and
simultaneously set up for antimicrobial susceptibility testing by
the BD Phoenix system. Positive blood culture specimens also were
inoculated on appropriate agar media for identification by
conventional bacteriologic methods, if necessary. Species identification based on the BD Phoenix result was considered to be the
reference standard. Microbiology laboratory personnel called the
infectious diseases pharmacist with each result obtained for every
hospitalized patient, 24 hours a day and 7 days a week. The on-call
infectious diseases pharmacist had remote access privileges to
patients electronic medical records. After review, and when
necessary, the infectious diseases pharmacist would contact the
treating physician to discuss the results and formulate the most
effective, targeted antimicrobial therapy. The pharmacist recommended de-escalation to targeted therapy when the final bacterial
identification and/or susceptibility test results were available.
Recommendations related to dosing/route modifications or to
escalate therapy were made when clinically indicated after review
of the medical record. The MALDI-TOF MS analysis was initially
performed 3 times daily (at ~1000, 1300, and 1900 hours), but on
March 20, 2012, an additional run was instituted on the night shift
(0500 hours; 4 times daily) every day.

Outcomes
We modeled exposure to the study intervention according to
clinically relevant temporal variables, comparing values collected
prior to and during the invention period. Clinical outcomes
evaluated included differences in time to final identification and
susceptibilities results, de-escalation rates, time to active therapy
Rapid Diagnostics and Stewardship Reduce CostPerez et al

Figure 1. Eligibility and inclusion of the study participants. The most common reasons for ineligibility among patients were medical circumstances
requiring prolonged hospitalization unrelated to the patients bloodstream infection (BSI; 24.4%), including patients receiving extracorporeal
membrane oxygenation (ECMO) for cardiorespiratory failure; advanced heart failure requiring ventricular assist devices (VADs) or an artificial heart;
and elective admissions for bone marrow transplantation (BMT). Length of stay (LOS) and hospital cost analyses were conducted in those patients
surviving to hospital discharge. Abbreviation: TTP, time-to-positivity of index blood culture.

when initial therapy was inactive, hospital LOS, total hospital costs,
and 30-day mortality rates between the 2 study periods.

Cost Analysis
Total hospital costs were calculated by adding up the costs
incurred across all cost centers, including room and board,
pharmacy, radiology, and laboratory. Cost data were obtained
from an individual in The Methodist Hospital accounting department who was independent of the team. All reported costs
represent actual costs for the administration of patient care as
determined by the individual departmental finance sections. No
changes were made in how these costs were calculated during the
period of the preintervention and intervention protocols.
Arch Pathol Lab MedVol 137, September 2013

Statistical Analysis
Summary statistics for continuous variables were reported as
mean 6 SD, and results for categoric variables were presented as
frequencies. The Mann-Whitney test was employed to identify
significantly different central locations between groups for continuously scaled variables, whereas the v2 test was used to determine
significantly different configurations across groups of categoric
data. All tests were 2-tailed, and a P value .05 represented
statistical significance. P values for v2 test were based on Fisher
exact test. To evaluate the independent impact of the study
intervention on LOS, we performed univariate and multivariate
Cox proportional hazards regression using LOS as the time-toRapid Diagnostics and Stewardship Reduce CostPerez et al 1249

Table 1.

Demographics and Baseline Characteristicsa

Characteristic

Preintervention Cohort (n 112)

Intervention Cohort (n 107)

Age, y
Male sex, No. (%)
Body mass index, kg/m2
APACHE II scoreb
Hospitalization in previous 90 d, No. (%)
Preexisting conditions, No. (%)
Immunosuppressed
HIV/AIDSc
Organ transplant
Malignancy
Solid tumor
Hematologic malignancy
Chronic lung disease
Cardiovascular disease
Cerebrovascular disease
Diabetes
Chronic kidney disease
End-stage renal disease
Liver disease
Infection source
Urinary
Vascular catheter associated
Gastrointestinal
Respiratory
Surgical site or skin structure
Unidentified
Organism, No. (%)
Escherichia coli
Klebsiella spp
Pseudomonas aeruginosa
Enterobacter spp
Proteus spp
Serratia spp
Other
MDRO/ESBL, No. (%)
Health care associated, No. (%)d
Nosocomial, No. (%)e
Preinfection LOS, d, median (IQR)f
ICU admissiong
ICU admission 48 h after BSI onset

66.9 6 14.3
56 (50)
27.4 6 7.4
14.3 6 6
60 (53.6)

65.3 6 16.3
56 (52.3)
27.1 6 6.8
15.4 6 5.3
52 (48.6)

47
0
16
27
19
8
17
33
27
38
27
19
11

(42)
(0)
(14.3)
(24.1)
(16.7)
(7.1)
(15.2)
(29.5)
(24.1)
(33.9)
(24.1)
(17)
(9.8)

42
4
16
22
12
11
11
37
17
42
35
19
16

(39.3)
(3.7)
(15)
(20.6)
(11.2)
(10.3)
(10.3)
(34.6)
(15.6)
(39.3)
(32.7)
(17.8)
(15)

48
22
19
9
7
7

(42.9)
(20)
(17)
(8)
(6.3)
(6.3)

37
24
14
10
15
7

(34.6)
(22.4)
(13.1)
(9.4)
(14)
(6.5)

56
26
9
8
5
2
4
11
78
22
9.5
55
11

(50)
(23.2)
(8)
(7.1)
(4.4)
(1.8)
(3.6)
(9.8)
(69.6)
(19.6)
(5.613)
(49.1)
(9.8)

46
21
15
7
8
7
3
10
72
15
8.2
39
5

(43)
(19.7)
(14)
(6.5)
(7.5)
(6.5)
(2.8)
(9.3)
(67.3)
(14)
(6.29.2)
(36.5)
(4.7)

Abbreviations: APACHE II, Acute Physiology and Chronic Health Evaluation II; BSI, bloodstream infection; HIV/AIDS, human immunodeficiency
virus/acquired immunodeficiency syndrome; ICU, intensive care unit; IQR, interquartile range; LOS, length of stay; MDRO/ESBL, multidrug-resistant
organism/extended-spectrum beta-lactamaseproducing organism.
a
Plus-minus values are means 6 SDs.
b
The APACHE II score was calculated based on clinical data present during the 24 hours preceding the index blood culture. Missing variables were
assumed to be normal. Glasgow Coma Score values were not available for inclusion.
c
P .04 for the comparison between preintervention cohort and intervention cohort.
d
Includes patients with recent contact with the health care system (recipients of recent intravenous therapy, dialysis, home wound care, residence at
long-term care facilities, and recent hospitalizations).
e
Limited to patients hospitalized for 48 hours prior to collection of the index culture (BSI onset).
f
Preinfection LOS limited to patients with nosocomial acquisition of BSI and who survived to hospital discharge (preintervention group [n 18] and
intervention group [n 12]).
g
Included patients with an ICU admission at any time during the index hospitalization.

event and discharge as the failure, and we included predictor

covariates likely to affect LOS, regardless of diagnosis timing.
These included age, sex, comorbidities, severity of illness, in vitro
antibiotic activity at 24 and 48 hours from BSI onset, preinfection
LOS (for nosocomial infections), and hospital mortality. The
expectation for Cox proportional hazards modeling is that if the
intervention shortens LOS, the resulting hazard ratio will be
greater than 1, reflecting a beneficial outcome. Kaplan-Meier
survival analysis was performed by partitioning LOS into groups
representing preintervention and intervention study periods, and
the equality of LOS was tested using the log-rank test.

RESULTS
Patients (N 317) with gram-negative BSIs were
evaluated for inclusion between August 2011 and November
1250 Arch Pathol Lab MedVol 137, September 2013

2011 (preintervention) and February 2012 and May 2012

(intervention). After the defined criteria were applied, 219
patients with gram-negative BSIs were included in the final
analysis. Of these, 112 patients were included in the
preintervention study group and 107 patients were in the
intervention study group (Figure 1).
Among the 219 evaluable patients, the mean age of
participants was 66.1 6 15.3 years (range, 25100 years),
with 112 men (51.1% Table 1). The overall mean APACHE II
score was 14.8 6 5.7 (range, 339). The most common
infection source was the urinary tract, accounting for 42.9%
of BSIs in the preintervention group and 34.6% in the
intervention group, followed by intravascular catheter
associated (20% versus 22.4%, respectively). Escherichia coli
was isolated most frequently (in 50% of preintervention and
Rapid Diagnostics and Stewardship Reduce CostPerez et al

Figure 2. Timeline comparison of preintervention and intervention study periods depicting the differences in laboratory procedure and their
respective impact on adjusted therapy. Adjusted therapy included, when clinically indicated, de-escalation/escalation of antibiotic therapy, dosing/
route modifications, and/or discontinuation of unnecessary gram-positive coverage. White boxes denote the average times (hours) until the
corresponding information was obtained or action implemented in the preintervention (PI) and intervention (Int) groups. The bottom horizontal line
represents the global study/patient timeline (hours) and includes point measurements (below) for patients on inactive therapy at 0, 24, and 48 hours
in both groups. Abbreviations: EMR, electronic medical record; MALDI-TOF MS, matrix-assisted laser desorption and ionization time-of-flight mass
spectrometry.

43% of intervention study groups), followed by Klebsiella

spp (in 23.3% and 19.7%, respectively). Nosocomial
infections accounted for 20.5% of BSIs in the preintervention group and 14% in the intervention group.
Outcomes
The average time from the blood culture TTP to final
species identification and antimicrobial susceptibility results
was 47.1 6 13.7 hours for the preintervention study group
versus 24.4 6 11.4 hours for the intervention study group (P
, .001). The mean TTP was 15.6 6 12 hours (range, 4109
hours) for both groups and was not significantly different.
The mean time to gram-negative organism identification
was significantly longer in the preintervention group versus
the intervention arm (36.6 6 15.3 versus 11.1 6 10.2 hours,
respectively; P , .001; Figure 2).
During the intervention study period, a total of 246
interventions were recommended to the prescribing physician by the infectious diseases pharmacist, of which 225
Table 2.
Outcome
Hospital length of stay
Hospital length of stay after BSI onset
ICU length of stay
ICU length of stay after BSI onset
Total hospital costs
MS DRG weight

were accepted. Overall, efforts to adjust and optimize

antimicrobial management occurred, on average, at 75
hours from TTP in 80% of eligible patients during the
preintervention period compared with 29 hours, on average,
in 94% of eligible patients in the intervention study group
(79 of 99 and 86 of 92 patients, respectively; P .004; Figure
2).
At TTP, similar proportions of patients on inactive therapy
were identified in the 2 groups (22 of 112 preintervention
patients and 16 of 107 intervention patients; Figure 2). At 24
hours from BSI onset, 22 patients (19.6%) in the preintervention period were on inactive therapy compared with 5
patients (4.7%) in the intervention group. The average time
to initiation of an active agent was 73.2 hours in the
preintervention group (n 22) compared with 36.5 hours in
the intervention group (n 5; P , .001). At 48 hours,
inactive therapy was only identified in 15 of 112 patients
from the preintervention study period (P , .001).

Length of Stay and Cost Outcomes in Survivorsa

Preintervention Cohort (n 100)
11.9
9.9
7.3
6.1
$45 709
2.7

6
6
6
6
6
6

9.3
7.1
8.5
6
$61 806
2.4

Intervention Cohort (n 101)

9.3
8.1
6.3
4.9
$26 162

6 7.6
6 6.4
6 8.7
6 6.7
6 $28 996
61.9

P
.01
.01
.05
.09
.009
54

Abbreviations: BSI, bloodstream infection; ICU, intensive care unit; MS DRG, Medicare Diagnosis-Related Group.
a
Values for length of stay outcomes are given as days, mean 6 SD. Costs are reported as cost per hospitalization, mean 6 SD.
Arch Pathol Lab MedVol 137, September 2013

Rapid Diagnostics and Stewardship Reduce CostPerez et al 1251

tion period ($26 162 versus $45 709, respectively; P .009;

Table 2). Diagnosis-related group coding weight means
were similar in both groups (2.3 for the intervention period
and 2.7 for the preintervention period), and the most
frequently identified diagnosis-related group codes were for
septicemia or severe sepsis (33 of 101 and 34 of 100,
respectively; P 54).

Figure 3. Impact of combining matrix-assisted laser desorption and

ionization time-of-flight mass spectrometry with antimicrobial stewardship on hospital length of stay after bloodstream infection onset.
Kaplan-Meier curves compare time to hospital discharge between
preintervention and intervention study periods for hospital survivors (P
.02, log-rank test).

The mean hospital LOS for survivors (n 100) in the

preintervention group was 11.9 days, and this was
significantly decreased to 9.3 days for the intervention
group (n 101; P .01; Table 2). The Kaplan-Meier analysis
(Figure 3) showed a significantly decreased probability of
remaining hospitalized after BSI onset as a function of the
study intervention (P .02). Multivariate Cox proportional
hazards regression analysis revealed that our study intervention remained independently associated with a decreased length of hospitalization in patients with gramnegative BSIs. Patients in the intervention study group had a
38% greater rate of discharge compared with the preintervention group (hazard ratio for intervention group, 1.38;
95% confidence interval, 1.011.88). Additionally, receipt of
active antimicrobial therapy at 48 hours was independently
associated with decreased LOS (hazard ratio, 2.9; 95%
confidence interval, 1.157.33). Preexisting pulmonary
disease, preinfection LOS, and APACHE II score were
independent predictors of prolonged hospitalization (Table
3). All-cause 30-day mortality rates were lower in the
intervention period compared with the preintervention
period (5.6% versus 10.7%, respectively; P .19).
Mean hospital costs per survivor were $19 547 less in the
intervention study period compared with the preintervenTable 3.

COMMENT
The goal of integrating rapid diagnostic microbiology
laboratory techniques (ie, pathogen identification and
susceptibility testing) with antimicrobial stewardship practices was to improve outcomes among hospitalized patients
with gram-negative BSIs. During the intervention period,
the average turnaround time for final culture identification
and antimicrobial susceptibility results was a full day quicker
compared with the preintervention study group (24.4 versus
47.1 hours, respectively; P , .001; Figure 2). Earlier initiation
of active, targeted antimicrobial therapy, informed by more
rapid identification and susceptibility testing results, demonstrated improved patient care outcomes (decreased LOS,
decreased mortality) and reduced health care expenditures.
The substantial benefit realized by our study intervention
protocol demonstrates the paramount relationship between
prompt, appropriate initiation of antibiotic therapy and
improved patient care, and it further highlights the key
advantage of fast and reliable information flow. Our findings
demonstrated a significant decrease in hospital LOS for
patients in the intervention group compared with the
previous institutional standard of care (9 versus 12 days,
respectively, P .01). This striking outcome was independent of multiple confounders, including comorbidities and
severity of illness (Table 2). Additionally, active therapy at 48
hours from BSI onset was also an independent predictor of
decreased LOS (Table 3); importantly, active therapy was
initiated in all intervention patients at 48 hours, eliminating
this risk factor (P , .001). Our results identified a decreased
rate in 30-day mortality that clearly favored the intervention
study group compared with the preintervention study group
(6 of 107 versus 12 of 112 patients, respectively; P .19).
Larger sample sizes are required to determine whether these
differences reach statistical significance.
Use of MALDI-TOF MS technology for routine bacterial
identification in clinical practice is at its infancy; hence, few
studies have the ability to measure and evaluate the
potential impact of this rapid intervention. In contrast, our
findings in the preintervention period are consistent with
the negative outcomes associated with inactive antimicrobial therapy reported in previous studies.13,1923 Increased

Independent Factors Associated with Length of Staya

Multivariateb

Univariate
Factor
Active antibiotic therapy at 48 h
MALDI-TOF MS antimicrobial stewardship intervention
APACHE II
Preinfection LOS
Preexisting lung disease

HR
2.24
1.40
0.96
0.87
0.62

95% CI

HR

95% CI

1.234.08
1.061.85
0.930.99
0.830.91
0.400.94

.009
.02
.003
,.001
.02

2.90
1.38
0.97
0.86
0.54

1.157.33
1.011.88
0.930.999
0.830.91
0.350.84

.02
.04
.05
,.001
.006

Abbreviations: APACHE II, Acute Physiology and Chronic Health Evaluation II; CI, confidence interval; HR, hazard ratio; LOS, length of stay;
MALDI-TOF MS, matrix-assisted laser desorption and ionization time-of-flight mass spectrometry.
a
Independent factors based on univariate and multivariate Cox proportional hazards regression models using data for patients surviving to discharge
(n 201).
b
Multivariate model additionally included sex, age, in vitro antibiotic activity at 24 hours, and underlying conditions, including
immunosuppression, cerebrovascular disease, cardiovascular disease, diabetes, chronic kidney disease, and liver disease (results not shown).
1252 Arch Pathol Lab MedVol 137, September 2013

Rapid Diagnostics and Stewardship Reduce CostPerez et al

morbidity and excess hospital LOS have been documented

in patients given initially inadequate antibiotic therapy at 24
hours in Staphylococcus aureus bacteremia, communityacquired pneumonia, and, more recently, gram-negative
septicemia.21,2426
A total of 11 MDRO/ESBL organisms (9.8%) were isolated
from patients in the preintervention study period, and 10
(9.3%) were isolated from patients in the intervention
period. These percentages are on the lower end of reported
MDRO/ESBL BSI incidence.23,27 Our study population was
limited to monomicrobial BSIs and excluded patients with
any other BSI during the index hospitalization in an effort to
decrease extraneous influences on LOS, costs, or mortality.5,7,16 There was a trend toward decreased intensive care
unit admissions in the intervention group compared with
the preintervention group (Table 1). Although not significant, this trend suggests a potential benefit (preventing
clinical deterioration) of rapid results and timely initiation of
appropriate therapy, even in the face of life-threatening
infection. The persistent benefit favoring the intervention
group despite similar rates of MDRO/ESBL isolates and
nosocomial acquisition, as well as an absence of difference
in baseline characteristics and severity of illness, increases
the generalizability of our data.5,20,21,28
The Infectious Diseases Society of America and the
Society for Healthcare Epidemiology of America recommend Prospective Audit with Intervention and Feedback
(PAIF) as a key mechanism to (1) provide actionable
diagnostic microbiology results, (2) decrease inappropriate
use of antimicrobial agents, and (3) improve patient
outcomes.8,13,29 PAIF involves direct interaction between
the prescriber and an infectious disease physician or clinical
pharmacist to target appropriate antimicrobial therapy.
Through PAIF, the infectious diseases pharmacist significantly enhanced the utility of the rapid results provided by
MALDI-TOF MS and rapid antimicrobial susceptibility
testing by tailoring therapy. Other studies have evaluated
the utility of molecular diagnostics with and without
antimicrobial stewardship programs, and they have found
that when used alone, the decreased turnaround times were
not likely to affect the time to appropriate antibiotic therapy
or reduce hospital LOS.13,3033 In addition, the benefits
associated with de-escalation have indicated decreased
costs, low risk of adverse drug events, and less selective
pressure for resistance and superinfections.34
Cost Discussion
Relatively few publications have demonstrated the costeffectiveness of molecular testing.11,13,30 A significant difference in hospital costs favoring the intervention study period
was observed in our study and was attributed primarily to
the decreased LOS. Total hospitalization costs per patient
decreased, on average, by $19 547 during the intervention
period compared with the preintervention group. Survivors
of septicemia often require prolonged hospitalizations with
direct hospital costs exceeding $40 000 per episode.4,23,28 In
our 1000-bed quaternary care hospital, we project a cost
savings of ~$18 million annually with the implementation
of this strategy for the management of gram-negative BSIs.
Limitations
Our study has several limitations. First, we analyzed a
limited sample size at a single academic quaternary care
hospital, which means that the results may not be applicable
to other settings. However, as noted above, the findings
Arch Pathol Lab MedVol 137, September 2013

from other investigations corroborate the role of timely,

appropriate antibiotics as a determinant of improved
outcomes in patients with gram-negative BSIs. Second,
patients in the preintervention study period were analyzed
retrospectively, creating the possibility of information bias.
Third, we did not examine all aspects of care that may have
influenced the clinical outcomes (eg, training and experience
of the treating physicians). Fourth, when the study started,
we had not yet validated the MALDI-TOF MS for
identification of yeasts, gram-positive anaerobic organisms,
or gram-negative anaerobic organisms directly from blood
culture bottles, thereby preventing inclusion of patients with
infections caused by these organisms. Future analyses
warrant their evaluation for potential impact.3,6,13,25,35
In an era of increasing resistance to bacterial antimicrobial
agents, optimal and timely management of patients with
gram-negative BSIs is essential. Many strategies have been
proposed and tried to further improve the consequences of
these detrimental infections. However, our study is the first
to demonstrate that integrating rapid molecular analysis by
novel application of MS with antimicrobial stewardship in
near real time significantly enhanced clinical care and
financial outcomes. Our interdisciplinary collaborative study
provides an important framework for productively addressing many other clinical care problems.
We thank the Clinical Microbiology Laboratory staff and the
Clinical Pharmacy Services for assistance with the study; D. Meuth,
MBA (The Methodist Hospital [TMH], Houston, Texas), for
providing institutional financial data; J. O. Ikwuagwu, PharmD
(TMH), for advice and assistance; and P. Cagle, MD (TMH), A.
Drews, MD (TMH), V. Fainstein, MD (TMH), R. Harris, MD
(TMH), M. Liebl, PharmD (TMH), D. Low, MD (Mt Sinai Hospital,
Toronto, Ontario, Canada), A. McGeer, MD (Mt. Sinai Hospital,
Toronto), I. Nachamkin, DrPH, MPH (University of Pennsylvania,
Philadelphia, Pennsylvania), R. Rapp, PharmD (University of
Kentucky, Lexington, Kentucky), E. Septimus, MD (Hospital
Corporation of America Inc, Houston, Texas), S. Shelburne, MD,
PhD (MD Anderson Cancer Center, Houston, Texas), and K.
Traugott, PharmD (Ochsner Health System, New Orleans,
Louisiana), for critical review of the manuscript. We gratefully
acknowledge K. E. Stockbauer, PhD, Office of Academic Development, Department of Pathology and Genomic Medicine, TMH, for
assistance with manuscript preparation.
References
1. Ibrahim EH, Sherman G, Ward S, Fraser VJ, Kollef MH. The influence of
inadequate antimicrobial treatment of bloodstream infections on patient
outcomes in the ICU setting. Chest. 2000;118(1):146155.
2. Harbarth S, Garbino J, Pugin J, Romand JA, Lew D, Pittet D. Inappropriate
initial antimicrobial therapy and its effect on survival in a clinical trial of
immunomodulating therapy for severe sepsis. Am J Med. 2003;115(7):529535.
3. Garey KW, Rege M, Pai MP, et al. Time to initiation of fluconazole therapy
impacts mortality in patients with candidemia: a multi-institutional study. Clin
Infect Dis. 2006;43(1):2531.
4. Micek ST, Lloyd AE, Ritchie DJ, Reichley RM, Fraser VJ, Kollef MH.
Pseudomonas aeruginosa bloodstream infection: importance of appropriate
initial antimicrobial treatment. Antimicrob Agents Chemother. 2005;49(4):
13061311.
5. Hidron AI, Edwards JR, Patel J, et al. NHSN annual update: antimicrobialresistant pathogens associated with healthcare-associated infections: annual
summary of data reported to the national healthcare safety network at the centers
for disease control and prevention, 20062007. Infect Control Hosp Epidemiol.
2008;29(11):9961011.
6. Gaynes R, Edwards JR; National Nosocomial Infections Surveillance
System. Overview of nosocomial infections caused by gram-negative bacilli.
Clin Infect Dis. 2005;41(6):848854.
7. Peleg AY, Hooper DC. Hospital-acquired infections due to gram-negative
bacteria. N Engl J Med. 2010;362(19):18041813.
8. Dellit TH, Owens RC, McGowan JE Jr, et al. Infectious Diseases Society of
America and the Society for Healthcare Epidemiology of America guidelines for
developing an institutional program to enhance antimicrobial stewardship. Clin
Infect Dis. 2007;44(2):159177.

Rapid Diagnostics and Stewardship Reduce CostPerez et al 1253

9. Seng P, Drancourt M, Gouriet F, et al. Ongoing revolution in bacteriology:

routine identification of bacteria by matrix-assisted laser desorption ionization
time-of-flight mass spectrometry. Clin Infect Dis. 2009;49(4):543551.
10. Prodhom G, Bizzini A, Durussel C, Bille J, Greub G. Matrix-assisted laser
desorption ionization-time of flight mass spectrometry for direct bacterial
identification from positive blood culture pellets. J Clin Microbiol. 2010;48(4):
14811483.
11. Schubert S, Weinert K, Wagner C, et al. Novel, improved sample
preparation for rapid, direct identification from positive blood cultures using
matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass
spectrometry. J Mol Diagn. 2011;13(6):701706.
12. Wimmer JL, Long SW, Cernoch P, et al. Strategy for rapid identification and
antibiotic susceptibility testing of gram-negative bacteria directly recovered from
positive blood cultures using the Bruker MALDI Biotyper and the BD Phoenix
system. J Clin Microbiol. 2012;50(7):24522454.
13. Bauer KA, West JE, Balada-Llasat JM, Pancholi P, Stevenson KB, Goff DA.
An antimicrobial stewardship programs impact with rapid polymerase chain
reaction methicillin-resistant Staphylococcus aureus/S. aureus blood culture test
in patients with S. aureus bacteremia. Clin Infect Dis. 2010;51(9):10741080.
14. Boucher HW, Talbot GH, Bradley JS, et al. Bad bugs, no drugs: no
ESKAPE!: an update from the Infectious Diseases Society of America. Clin Infect
Dis. 2009;48(1):112.
15. Knaus WA, Draper EA, Wagner DP, Zimmerman JE. APACHEII: a severity
of disease classification system. Crit Care Med. 1985;13(10):818829.
16. McGregor JC, Rich SE, Harris AD, et al. A systematic review of the methods
used to assess the association between appropriate antibiotic therapy and
mortality in bacteremic patients. Clin Infect Dis. 2007;45(3):329337.
17. Garner JS, Jarvis WR, Emori TG, Horan TC, Hughes JM. CDC definitions for
nosocomial infections, 1988. Am J Infect Control. 1988;16(3):128140.
18. Clinical and Laboratory Standards Institute. Performance Standards for
Antimicrobial Susceptibility Testing: Twenty-First Informational Supplement.
Wayne, PA: Clinical and Laboratory Standards Institute; 2011. CLSI document
M100-S21.
19. Kang CI, Kim SH, Kim HB, et al. Pseudomonas aeruginosa bacteremia: risk
factors for mortality and influence of delayed receipt of effective antimicrobial
therapy on clinical outcome. Clin Infect Dis. 2003;37(6):745751.
20. Lee YT, Kuo SC, Yang SP, et al. Impact of appropriate antimicrobial therapy
on mortality associated with Acinetobacter baumannii bacteremia: relation to
severity of infection. Clin Infect Dis. 2012;55(2):209215.
21. Shorr AF, Micek ST, Welch EC, Doherty JA, Reichley RM, Kollef MH.
Inappropriate antibiotic therapy in gram-negative sepsis increases hospital length
of stay. Crit Care Med. 2011;39(1):4651.
22. Lodise TP Jr, Patel N, Kwa A, et al. Predictors of 30-day mortality among
patients with Pseudomonas aeruginosa bloodstream infections: impact of delayed
appropriate antibiotic selection. Antimicrob Agents Chemother. 2007;51(10):
35103515.

1254 Arch Pathol Lab MedVol 137, September 2013

23. Tumbarello M, Spanu T, Di Bidino R, et al. Costs of bloodstream infections

caused by Escherichia coli and influence of extended-spectrum-beta-lactamase
production and inadequate initial antibiotic therapy. Antimicrob Agents Chemother. 2010;54(10):40854091.
24. Shorr AF, Micek ST, Kollef MH. Inappropriate therapy for methicillinresistant Staphylococcus aureus: resource utilization and cost implications. Crit
Care Med. 2008;36(8):23352340.
25. Lodise TP, McKinnon PS, Swiderski L, Rybak MJ. Outcomes analysis of
delayed antibiotic treatment for hospital-acquired Staphylococcus aureus
bacteremia. Clin Infect Dis. 2003;36(11):14181423.
26. Kollef KE, Schramm GE, Wills AR, Reichley RM, Micek ST, Kollef MH.
Predictors of 30-day mortality and hospital costs in patients with ventilatorassociated pneumonia attributed to potentially antibiotic-resistant gram-negative
bacteria. Chest. 2008;134(2):281287.
27. Schwaber MJ, Carmeli Y. Mortality and delay in effective therapy
associated with extended-spectrum beta-lactamase production in Enterobacteriaceae bacteraemia: a systematic review and meta-analysis. J Antimicrob
Chemother. 2007;60(5):913920.
28. Martin GS, Mannino DM, Eaton S, Moss M. The epidemiology of sepsis in
the United States from 1979 through 2000. N Engl J Med. 2003;348(16):1546
1554.
29. Camins BC, King MD, Wells JB, et al. Impact of an antimicrobial utilization
program on antimicrobial use at a large teaching hospital: a randomized
controlled trial. Infect Control Hosp Epidemiol. 2009;30(10):931938.
30. Forrest GN, Mehta S, Weekes E, Lincalis DP, Johnson JK, Venezia RA.
Impact of rapid in situ hybridization testing on coagulase-negative staphylococci
positive blood cultures. J Antimicrob Chemother. 2006;58(1):154158.
31. Holtzman C, Whitney D, Barlam T, Miller NS. Assessment of impact of
peptide nucleic acid fluorescence in situ hybridization for rapid identification of
coagulase-negative staphylococci in the absence of antimicrobial stewardship
intervention. J Clin Microbiol. 2011;49(4):15811582.
32. Roberts RR, Hota B, Ahmad I, et al. Hospital and societal costs of
antimicrobial-resistant infections in a Chicago teaching hospital: implications for
antibiotic stewardship. Clin Infect Dis. 2009;49(8):11751184.
33. Charani E, Edwards R, Sevdalis N, et al. Behavior change strategies to
influence antimicrobial prescribing in acute care: a systematic review. Clin Infect
Dis. 2011;53(7):651662.
34. Singh N, Rogers P, Atwood CW, Wagener MM, Yu VL. Short-course
empiric antibiotic therapy for patients with pulmonary infiltrates in the intensive
care unit: a proposed solution for indiscriminate antibiotic prescription. Am J
Respir Crit Care Med. 2000;162(2, pt 1):505511.
35. Neidell MJ, Cohen B, Furuya Y, et al. Costs of healthcare- and communityassociated infections with antimicrobial-resistant versus susceptible organisms.
Clin Infect Dis. 2012;55(6):807815.

Rapid Diagnostics and Stewardship Reduce CostPerez et al