Surveillance of Antimicrobial Use and Antimicrobial Resistance in United States Hospitals:

Project ICARE Phase 2

Author(s): Scott K. Fridkin, Christine D. Steward, Jonathan R. Edwards, Erica R. Pryor, John
E. McGowan, Jr., Lennox K. Archibald, Robert P. Gaynes, Fred C. Tenover, Project Intensive
Care Antimicrobial Resistance Epidemiology (ICARE) Hospitals
Source: Clinical Infectious Diseases, Vol. 29, No. 2 (Aug., 1999), pp. 245-252
Published by: Oxford University Press
Stable URL: http://www.jstor.org/stable/4460867 .
Accessed: 28/02/2011 16:20

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at .
http://www.jstor.org/action/showPublisher?publisherCode=oup. .

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

Oxford University Press is collaborating with JSTOR to digitize, preserve and extend access to Clinical
Infectious Diseases.

http://www.jstor.org
 245

Surveillance of Antimicrobial Use and Antimicrobial Resistance in United States

Hospitals: Project ICARE Phase 2
Scott K. Fridkin, Christine D. Steward, From the Hospital Infections Program, National Centerfor Infectious
Jonathan R. Edwards, Erica R. Pryor, Diseases, Centersfor Disease Control and Prevention, and the
John E. McGowan, Jr., Lennox K. Archibald, Departmentof Epidemiology,Rollins School of Public Health, Emory
Robert P. Gaynes, Fred C. Tenover, and Project University,Atlanta, Georgia
Intensive Care Antimicrobial Resistance
Epidemiology (ICARE) Hospitals

The search for the means to understand and control the emergence and spread of antimicrobial
resistance has become a public health priority. Project ICARE (Intensive Care Antimicrobial
Resistance Epidemiology)has establishedlaboratory-basedsurveillancefor antimicrobialresistance
and antimicrobialuse at a subset of hospitals participatingin the National Nosocomial Infection
Surveillancesystem. These data illustrate that for most antimicrobial-resistantorganisms studied,
rates of resistance were highest in the intensive care unit (ICU) areas and lowest in the outpatient
areas. A notable exception was ciprofloxacin-or ofloxacin-resistantPseudomonasaeruginosa,for
which resistance rates were highest in the outpatient areas. For most of the antimicrobialagents
associatedwith this resistance,the rate of use was highest in the ICU areas, in parallelto the pattern
seen for resistance.These comparativedata on use and resistanceamong similar areas (i.e., ICU or
other inpatient areas) can be used as a benchmarkby participatinghospitals to focus their efforts at
addressingantimicrobialresistance.

Antimicrobial-resistant pathogenspose an enormouschal- Disease ControlandPrevention(CDC)andthe NationalFoun-

lenge for clinicians, infection-controlpersonnel,and hospital
administrators.These pathogens have become a prevalent
cause of hospital-acquired infections,particularlyin patientsin
intensive care units (ICUs) [1, 2]. However, reportsdemon-
stratethat these antimicrobial-resistantpathogenshave spread
beyond the walls of the traditionalacute-carehospital [3-6].
These reports heighten the need to extend surveillancefor
antimicrobialresistanceto includeareasoutsideof traditional
settings.
As antimicrobialuse generallyprecedesthe emergenceof
antimicrobialresistance,preventingthe emergenceand spread
of antimicrobial-resistantpathogensclearlyrequiresoptimizing
antimicrobialuse. A workshopsponsoredby the Centersfor
This articleis part of a series of paperspresentedat a symposiumentitled
"SurveillanceandControlof Pathogensof EpidemiologicImportance" thatwas
held on 3-4 April 1998 in Orlando,Florida.The symposiumwas fundedby
Rh6ne-PoulencRorerPharmaceuticals.
Financialsupport:this work was supportedin partby grantsto the Rollins
School of Public Health of Emory University by Zeneca Pharmaceuticals
(Wilmington,DE) as a full sponsor,and Rh6ne-PoulencRorer(Collegeville,
PA), the NationalFoundationfor InfectiousDiseases (Bethesda,MD), Roche
Laboratories(Nutley,NJ), the AmericanSociety for HealthSystemsPharma-
cists Researchand EducationFoundation(Bethesda,MD), Kimberly-Clark
(Roswell,GA), andBayerCorporation, PharmaceuticalsDivision(WestHaven,
CT), as partialsponsors.
Reprintsor correspondence:Dr. Scott K. Fridkin,HospitalInfectionsPro-
gram,MailstopE-55, Centersfor DiseaseControlandPrevention,1600 Clifton
Road,Atlanta,Georgia30333 (skf0@cdc.gov).
ClinicalInfectiousDiseases 1999;29:245-52
? 1999by the InfectiousDiseasesSocietyof America.All rightsreserved.
1058-4838/99/2902-0004$03.00
dation for Infectious Diseases made recommendationsthat
hospitals should monitor antimicrobialuse in an attemptto
reduce the emergence and spread of antimicrobial-resistant
pathogens [7]. Such monitoringalso can aid the infection-
controlcommunityin determininghow to focus theireffortsin
reducingthe emergenceand spreadof antimicrobial-resistant
pathogens.To createa monitoringtool to use in such efforts,
CDC's HospitalInfectionsProgram,in cooperationwith the
Rollins School of Public Health at EmoryUniversity,began
ProjectICARE(IntensiveCareAntimicrobialResistanceEpi-
demiology)at a subsetof hospitalsparticipatingin the National
NosocomialInfectionsSurveillance(NNIS) system at CDC.
ProjectICAREprovidesdata on the prevalenceof antimi-
crobialresistanceand antimicrobialuse in U.S. hospital set-
tings. Data fromthe eight hospitalsparticipatingin phase 1 of
this project suggestedthat most antimicrobialresistancewas
focused in the ICUs of the hospitals [1, 8]. Here we present
data from 41 hospitals participatingin phase 2 of Project
ICARE.In additionto the datacollected,clinicalisolateswere
collected from each hospitalto verify reportedresistance.
Methods
Participating hospitals. Hospitals that participated in the
ICU componentof NNIS were eligible to participatein phase
2 of Project ICARE. The surveillancemethodology of the
NNIS system has been describedpreviously [9]. Hospitals
were asked to reportat least 12 monthsof data,beginningin
January1996. All participatinghospitalswere requiredto sub-
246 Fridkinet al.
mit data from one or more ICUs to the ICU componentof
NNIS. For ICARE,hospitalsreportedadditionalmicrobiology
and pharmacydata to CDC. These data were aggregatedfor
each month and stratifiedby hospital area. Hospital areas
includedeach ICU separately(i.e., unitsthatprovideintensive
observation,diagnosis,andtherapeuticproceduresfor critically
ill patients), non-ICU inpatientareas (i.e., areas where the
patientstays at least one nightin the hospitalexcludingICUs),
and outpatientareas(i.e., unitsthatperformsame-daysurgery,
simple diagnosticprocedures,or therapy,such as chemother-
apy, hemodialysis, cardiac catheterization,urgent care, or
emergencyunits).
Microbiologydata. Microbiologydataincludedantimicro-
bial susceptibilitytest resultsfor all nonduplicateclinical iso-
lates processedby the laboratoryduringeach study month.A
duplicateisolate was definedas an isolate of the same species
of bacteriawith the same antimicrobialsusceptibilitypatternin
the same patient,whateverthe site of isolation, duringeach
month. To accommodatedifferencesin susceptibilitytesting
panels amonghospitals,reportingof resistancewas requested
for one or more of a groupof relatedantimicrobials.
Susceptible,intermediate, andresistantisolatesof selectedor-
ganismswere stratifiedby locationin the hospitalwhere the
organismwas acquired(eachICU ward,inpatientnon-ICUunits
combined,and outpatientareas)for each month,for 12 sentinel
antimicrobial-resistant organisms.These included methicillin-
resistantcoagulase-negativestaphylococci;methicillin-resistant
Staphylococcus aureus; vancomycin-resistant enterococci;
penicillin-resistant Streptococcus pneumoniae; ciprofloxacin-
or ofloxacin-resistant
Escherichiacoli; E. coli resistantto cefta-
zidime,cefotaxime,or ceftriaxone;
Klebsiellapneumoniaeresis-
tant to these third-generationcephalosporins;Enterobacter
species resistant to these third-generationcephalosporins;
piperacillin-resistant Pseudomonas aeruginosa; ceftazidime-
resistantP. aeruginosa; ciprofloxacin-or ofloxacin-resistant
P. aeruginosa; and imipenem-resistantP. aeruginosa. All isolates
fittingthe selectedorganism-antimicrobialcombinations, whether
for or
responsible hospital- community-acquired infectionor for
colonization,were reportedto ProjectICAREby participating
hospitals.Hospitalsused interpretivestandardsfromthe National
Committeefor ClinicalLaboratory Standards(NCCLS)for MIC
or zone diametertestingstandards to reportnumbersof suscepti-
ble, intermediate,or resistantorganisms[10-12].
Microbiology isolates. Each ICARE hospital's microbiol-
ogy laboratorycollectedup to 20 nonduplicateclinicalisolates
of select organismsfor validationtestingat the ProjectICARE
laboratoryat CDC. Only isolates having MICs or zone sizes
classifiedas intermediateor resistantto a specificantimicrobial
agentwere sent for validation.Eachorganismsentwas accom-
paniedby a completedreportformand antibiogramindicating
susceptibilitytest methodand hospital-derivedresistancepat-
tern. The resultsof this validationtestingprovideda sense of
CID 1999;29 (August)
the accuracyof each laboratory'smethods of susceptibility
testing in identifyingorganismsas resistant.
The followingorganismswere collectedfor validitytesting:
methicillin-resistantS. aureus; vancomycin-resistantentero-
cocci; ceftazidime-, cefotaxime-, or ceftriaxone-resistant
K.pneumoniae;ciprofloxacin- or ofloxacin-resistant
Enterobacte-
riaceae;imipenem-resistant Enterobacteriaceae; ciprofloxacin-or
ofloxacin-resistantP. aeruginosa; and imipenem-resistant
P. aeruginosa.Brothmicrodilution testingwas done
susceptibility
for these isolatesaccordingto NCCLSmethodsby meansof in
house-prepared plates[10].
For the isolates tested for validation,each hospital-derived
MIC or disk diffusion zone size was comparedwith results
frombrothmicrodilutiontestingby use of NCCLSbreakpoint
interpretations.Organismsdefined as causing major errors
were those categorizedby the hospitalas resistantandfoundto
be susceptibleby validitytesting. Organismsdefinedas caus-
ing minor errorswere classified as intermediateby validity
testing and as susceptibleor resistantby the hospitalor clas-
sified as intermediateby the hospital and as susceptibleor
resistantby validitytesting.
Pharmacydata. The pharmacydata included in-hospital
use of selected oral and parenteralantimicrobialagents in
grams.For purposeof analysis,gramsof antimicrobialswere
convertedinto numberof defineddaily doses used each month
in each hospital area. A defined daily dose is the typical
numberof gramsof an antimicrobialused per day for a typical
adult(table 1) [13]. Antimicrobialuse was stratifiedby routeof
administrationand hospitalarea (i.e., each ICU and inpatient
non-ICU areas combined).Because outpatientantimicrobial
use could not be estimatedreliably from hospital pharmacy
records,we did not collect dataaboutoutpatientantimicrobial
use. Finally,antimicrobialswith similarspectrumsor clinical
indicationswere grouped(table 1).
Statisticalanalyses. Theresistancerateof eachantimicrobial-
resistantorganismwas evaluatedin two ways. First,to provide
feedbackof datato the individualhospitals,the individualresis-
tance rateswere calculatedat each institution,by hospitalarea
(i.e., individuallyfor each ICU,combinedfor all non-ICUinpa-
tient areasand for all outpatientareas).Data from each of the
hospitalsareas were pooled over the entire study period. To
calculatea single rate for each hospitalarea, we dividedthe
numberof resistantisolatesby the total numberof isolatesfor
whichsusceptibility testinghadbeendone.Resistancerateswere
calculatedfor a particular hospitalareaonly if > 10 isolateswere
testedfor susceptibility. Second,to compareratesbetweenhospi-
tal areas(i.e., all ICUs combined,non-ICUinpatientareas,out-
patientareas),a pooledmeanresistancerateforeachcombination
of antimicrobial-resistant organismswas calculatedby poolingthe
data from all ICAREhospitals.This rate was the sum of all
resistantorganisms,dividedby the sumof all organismstestedfor
susceptibility,by hospitalarea,amongall ICAREhospitals.
CID 1999;29 (August) AntimicrobialUse and Resistance 247

Table 1. Class, grouping, and defined daily dose of antimicrobial agents-Project ICARE, phase 2.

Class Group Antimicrobialagent Defineddaily dose

/3-Lactams Penicillin PenicillinG 12 X 106U

ProcainepenicillinG 2.4 x 106 U
PenicillinG benzathine 1.2 X 106U
PenicillinV 1g
Ampicillingroup Ampicillin(parenteral) 4g
Ampicillin(oral) 2g
Ampicillin/sulbactam 6g
Amoxicillin(oral) 1.5 g
Amoxicillin/clavulanicacid (oral) 1.5 g
Antistaphylococcalpenicillins(methicillingroup) Methicillin 4g
Nafcillin 4g
Oxacillin 4g
Cloxacillin(oral) 2g
Dicloxacillin(oral) 2g
Antipseudomonalpenicillins Piperacillin 18 g
Piperacillin/tazobactam 13.5 g
Mezlocillin 18 g
Ticarcillin 18 g
Ticarcillin/clavulanic
acid 12.4 g
First-generation
cephalosporins Cefazolin 3g
Cephalothin 4g
Cefadroxil(oral) 2g
Cephalexin(oral) 2g
Second-generation
cephalosporins Cefotetan 2g
Cefmetazole 4g
Cefoxitin 4g
Cefuroxime 3g
Cefuroximeaxetil (oral) 1g
Cefaclor(oral) 1g
Third-generation
cephalosporins Cefotaxime 3g
Ceftazidime 3g
Ceftizoxime 3g
Ceftriaxone 1g
Cefixime(oral) 0.4 g
Cefoperazone 4g
Other 3-lactams Imipenemcilastatin 2g
Aztreonam 4g
Glycopepetides Vancomycin(parenteral) 2g
Vancomycin(oral) 1g
Fluoroquinolones Ciprofloxacin(parenteral) 0.8 g
Ciprofloxacin(oral) 1.5 g
Ofloxacin(parenteral) 0.8 g
Ofloxacin(oral) 0.8 g
Trimethoprim-sulfamethoxazole Trimethoprimcomponent(oral) 0.32 g
Trimethoprimcomponent(iv) 0.84 g

To controlfor the populationat riskfor receivingthesedrugs,
we expresseduse as a usagedensityrate:numberof defineddaily
doses per 1,000 patient-days.Use of antimicrobial agentswas
determined at eachhospital,by hospitalarea.Thepoolednumber
of gramsof each antimicrobial agentused,by hospitalarea,was
dividedby the numberof gramsper defineddaily dose for the
specificantimicrobialagent,thendividedby thepoolednumberof
patient-daysin the correspondinghospitalarea,andmultipliedby
1,000 to derive the numberof defineddaily doses per 1,000
patient-days.PediatricICUs were excludedfromthe analysisof
antimicrobialuse, sincethe defineddailydose of mostantimicro-
bials studiedis less well definedin the pediatricpopulationand
commonlydependson weightor age.
Comparisons of proportions(i.e., pooledmeanresistancerates
of all ICAREhospitals)betweenhospitalareasweredoneby use
of the X2test. Comparisons of the medianratesof antimicrobial
usebetweenhospitalareas(i.e.,ICUsvs. non-ICUinpatientareas)
were doneby use of the Wilcoxonranksumtest.
Results
Participating hospitals. Phase 2 hospitals were located in
18 statesandthe Districtof Columbia:15 (37%)of these were
248 Fridkinet al.
locatedin the southernAtlanticstates.The demographicsof the
phase 2 hospitals were similar to those of other hospitals
participatingin NNIS (table 2). Forty-onehospitalsreported
data from 108 adult ICUs for a medianof 12 months(inter-
quartilerange,9-18 months),40 reporteddatafromnon-ICU
inpatientareasfor a medianof 12 months(interquartilerange,
12-20 months),and 38 reporteddatafromthe outpatientareas
for a medianof 12 months(interquartile range,12-18 months).
Ten of these hospitalsreporteddataconcerningpediatricICUs
(median,12 months;range,6-24 months).Susceptibilitydata
were reportedfor a total of 290,045 isolates.
Validation of resistant isolates. Of isolates tested for val-
idationtesting,the frequenciesof minoror majorerrorswere
within an acceptablerange: 11.1% minor errors and 9.5%
majorerrors,with one exception(not includedin these calcu-
lations). Validity testing by broth microdilution of 84
imipenem-resistant isolates of Enterobacteriaceae sent by hos-
pitals revealed 13 minor errors (15.5%) and 67 majorerrors
(79.8%), and testing of 223 imipenem-resistantisolates of
P. aeruginosarevealed64 minorerrors(28.7%)and 47 major
errors(21.9%).The rate of errorsvaried slightly betweenthe
two most common hospital original test methods: Vitek
(bioMerieuxVitek, Hazelwood, MO) and MicroScan(Dade
MicroScan,West Sacramento,CA). Amongisolatesof Entero-
bacteriaceae,the proportionof majorerrorswas 67%for Vitek
users and 81% for MicroScanusers, whereasthe proportion
amongisolates of P. aeruginosawas 22% for Vitek users and
17%for MicroScanusers [14].
Resistancerates. Formostof the organismsstudied,the per-
centageof resistantisolatesreportedfrommonthlydatacollection
decreasedin a stepwisemannerby areaof the hospital(figure1).
The highestresistanceratesoccurredamongisolatesfrom ICU
patients,followed in decreasingorderby rates among isolates
from non-ICUinpatientsand rates among isolates from out-
patients. These organisms included methicillin-resistant
coagulase-negative S. aureus,
staphylococci,methicillin-resistant
vancomycin-resistant P.
enterococci,piperacillin-resistantaerugi-
nosa, ceftazidime-resistant P. aeruginosa, and ceftazidime-,
cefotaxime-, or ceftriaxone-resistantEnterobacter species.All of
Table 2. Characteristicsof Phase 2 ICARE hospitals and all Na-
tional Nosocomial InfectionsSurveillance(NNIS) hospitals, 1996-
1997.
ICARE-NNIS OtherNNIS
Hospitals hospitals
Characteristic (n = 41) (n = 221)
Bed size category
<200 3 (8) 9 (4)
200-500 25 (61) 146 (66)
>500 13 (32) 66 (30)
Departmentof VeteransAffairsHospital 4 (10) 11 (5)
Academicaffiliation 28 (70) 163 (74)
NOTE. Data are no. of hospitals(%).
CID 1999;29 (August)
80
70
60
50
40
30O
20
10
MR-CNS MRSA VE Pr- PA CF-PA CF3-ENB
Antimicrobial-resistant
pathogen
Figure 1. Comparisonof pooled mean rates of resistanceamong
isolatesfromadultintensivecareunit (ICU)patients(solid bars),non-
ICU inpatients(shadedbars),andoutpatients(openbars)from41 U.S.
hospitals.CF-PA = ceftazidime-resistant Pseudomonasaeruginosa;
CF3-ENB= Enterobacter speciesresistantto anythird-generation
ceph-
alosporin;MR-CNS= methicillin-resistant staphy-
coagulase-negative
lococci; MRSA = methicillin-resistant Staphylococcusaureus; PIP-
P. aeruginosa;VRE = vancomycin-resistant
PA = piperacillin-resistant
enterococci.
these stepwisedecreaseswere statisticallysignificantexceptfor
vancomycin-resistant enterococci,for which the differencebor-
dered on statisticalsignificance(table 3). However,a second
patternwas seen for K pneumoniaeand E. coli resistantto
ceftazidime,cefotaxime,or ceftriaxoneand for ofloxacin-or
ciprofloxacin-resistantE. coli. The pooledmeanresistancerates
were similarbetweenthe adult ICU areas combinedand the
non-ICUinpatientareascombined(figure2), even thoughthe
resistancerateswerestillsignificantlyhigherforall inpatientareas
combinedthanfor outpatientareas(table3).
S. pneumoniae,the rateof
In contrast,for penicillin-resistant
resistancewas similarin all hospitalareas(i.e., ICU, non-ICU
inpatient, outpatient), while the rate of ciprofloxacin- or
ofloxacin-resistantP. aeruginosa was significantlyhigher in
outpatientareasthan inpatientareas(table 3).
The rates of antimicrobial-resistant organismsamong pa-
tients frompediatricICUs was not significantlydifferentfrom
the rates in adult ICUs (data not shown), with some notable
exceptions.Penicillin-resistant pneumococciwere more likely
isolated from patientsin pediatricICUs than adult ICUs (pe-
diatric, 15 [20.5%] of 73, vs. adult, 27 [9.5%] of 284;
P = .009). In addition,pediatricICUs reportedsignificantly
lower rates of ciprofloxacin-or ofloxacin-resistantP. aerugi-
nosa (pediatric,7 [3.7%] of 191, vs. adult, 593 [16.4%] of
3,622; P < .001), methicillin-resistant S. aureus(pediatric,13
[5.7%] of 227, vs. adult, 1,673 [35.2%] of 4,750; P < .001),
and vancomycin-resistant enterococci(pediatric,2 [3.6%] of
56, vs. all others,403 [13%]of 3,097; P = .036).
Pharmacydata. The medianrateof antimicrobialuse was
significantlyhigherin adultICU areasthan in non-ICUareas
combinedfor third-generation cephalosporinscombined,cefta-
zidime alone, intravenousvancomycin,anti-pseudomonal pen-
icillins, intravenousfluoroquinolones,or imipenem(figure3).
CID 1999;29 (August) AntimicrobialUse and Resistance 249

Table 3. Resistance to specific antimicrobials in isolates from different hospital areas, phase 2 of Project ICARE, January 1996 to December
1997.

Inpatientareas

Antimicrobial-resistant
organism ICUs Non-ICUinpatient P value* Outpatientareas P valuet

Methicillin-resistant
coagulase-negativestaphylococci 3,307/4,411 (75.0) 7,240/11,988(60.4) <.01 3,611/8,120(44.5) <.01
[n = 82] [n = 39] [n = 35]
Staphylococcusaureus
Methicillin-resistant 1,673/4,750(35.2) 1,597/5,097(31.9) <.01 2,231/12,580 (17.7) <.01
[n = 90] [n = 40] [n = 36]
enterococci
Vancomycin-resistant 403/3,097 (13.0) 1,763/14,917(11.8) .06 206/8,285 (2.5) <.01
[n = 75] [n = 39] [n = 35]
Pseudomonasaeruginosa
Piperacillin-resistant 381/3,117 (12.2) 695/8,423 (8.3) <.01 300/5,028 (6.0) <.01
[n = 70] [n = 37] [n = 33]
P. aeruginosa
Ceftazidime-resistant 369/3,601 (10.2) 751/10,461(7.2) <.01 293/5,784 (5.1) <.01
[n = 78] [n = 39] [n = 36]
Third-generation Enterobacter
cephalosporin-resistant 491/1,966 (25.0) 971/4,351 (22.3) .02 228/2,265 (10.1) <.01
species [n = 61] [n = 40] [n = 34]
Third-generation Klebsiella
cephalosporin-resistant 69/1,887 (3.7) 262/7,030 (3.7) .98 76/5,434 (1.4) <.01
pneumoniae [n = 62] [n = 39] [n = 35]
Third-generationcephalosporin-resistant 27/3,012 (0.9) 148/19,188(0.8) .59 67/32,618 (0.2) <.01
Escherichiacoli [n = 80] [n = 39] [n = 36]
E. coli
Ofloxacin-or ciprofloxacin-resistant 37/2,794 (1.3) 249/17,836 (1.4) .63 185/28,314(0.7) <.01
[n = 77] [n = 39] [n = 35]
P. aeruginosa
Ofloxacin-or ciprofloxacin-resistant 593/3,622 (16.4) 1,817/10,312(17.6) .10 1,177/5,871(20.0) <.01
[n = 80] [n = 40] [n = 36]
Penicillin-resistant
Streptococcuspneumoniae 27/284 (9.5) 190/1,830(10.4) .65 179/1,818(9.8) .28
[n = 17] [n = 40] [n = 31]

NOTE. ICU = intensivecare unit. Data are no. of resistantisolates/totalno. of isolates tested (%) [no. of hospitalunits reportingresultsfor > 10 isolates].
* Comparisonof ICU patientareasvs. non-ICUinpatientareascombined,X2.
t Comparisonof inpatientareasvs. outpatientareas,X2.

However, there was no significant difference in use between 140- GroupI Group2 Group3
ICU and non-ICU areas for the antistaphylococcal penicillins 120- ICU> non-ICU ICU= non-ICU ICU <non-ICU
(i.e., methicillin group), first-generation cephalosporins,
second-generation cephalosporins, or aztreonam (figure 3). In WIO 100
80l
. 60

20- n
141 I I 60 A

12-

Antimicrobial agent or group

6-
M4- Figure 3. Comparisonof median rates of antimicrobialuse (i.e.,
2- defineddaily doses per 1,000patient-days)reportedin adultintensive
care units (ICUs;solid bars; n = 108) and non-ICUareascombined
CF-Kc 03-EC OIL-EC 0PS OIL-PA (open bars; n = 40). Categorizationof antimicrobialsis shown in
Antimicrobial-resistant
pathogen table 1. All use is iv and orally (po), except as noted. AP Pcn =
antipseudomonalpenicillins;Atm = aztreonam;CF1, CF2, CF3 =
Figure 2. Comparisonof pooled mean rates of resistanceamong first-, second-, and third-generationcephalosporins,respectively;
isolatesfromadultintensivecareunit (ICU)patients(solid bars),non- Czid = ceftazidime; FQ = fluoroquinolones; Imi = imipenem;
ICU inpatients(shadedbars),andoutpatients(openbars)from41 U.S. Meth = methicillin; Tmp-Smz = trimethoprim-sulfamethoxazole;
hospitals.CF3-EC = Escherichiacoli resistantto third-generation Vm = vancomycin.Group 1: antimicrobialagents used at signifi-
cephalosporins; CF3-KP = Klebsiellapneumoniaeresistantto third- cantlyhigherratesamongICUpatientsthanamongnon-ICUpatients;
generationcephalosporins;OFL-EC = ofloxacin- or ciprofloxacin- group2: agentsfor whichuse is similarregardlessof hospitallocation
resistantE. coli; OFL-PA= ofloxacin-orciprofloxacin-resistant
Pseudo- (P < .05); group3: agentsfor which use is significantlyloweramong
monasaeruginosa;PRSP = penicillin-resistant pneumococci. ICU patientsthan among non-ICUpatients(P < .05).
250 Fridkinet al. CID 1999;29 (August)

which were includedin the outpatientareasof ICAREhospi-

Higher resistance in ICUs than in Higher use in ICUs than in non-
non-ICU areas ICU areas tals. Community-acquired infection due to P. aeruginosare-
E. coli, P. aeruginosa:ceftazidime 3rd generation cephalosporins sistantto multipleantimicrobialshas previouslybeen associ-
P. aeruginosa:piperacillin Ureido/carboxypenicillins atedwith cystic fibrosis[15] andotherunderlyingillness, such
Enterococci:vancomycin Vancomycin
as infectionwith HIV [16]. The relationshipof our findingsto
Similar or less resistance in ICUs Similar or less use in ICUs than infectionsamongpatientsin these types of outpatientsettings
than in non-ICU areas in non-ICU areas is not clear.
Pneumococcus:penicillin Aminopenicillins We evaluatedin detail how these fluoroquinoloneswere
E. coli, P. aeruginosa:quinolones Quinolones
used in each hospitalarea.Althoughintravenousfluoroquino-
lone use was similarin the ICU and non-ICUinpatientareas,
Figure 4. Summary of ratesof resistance andantimicrobial use in
intensivecare units (ICUs) comparedwith non-ICUareas,for select the oral fluoroquinoloneswere used at a significantlyhigher
antimicrobial-resistant organismsfor which use andresistancefollow rate in the non-ICUinpatientareasthanin the ICU areas.We
a similarpattern-Project ICARE,phase 2, 1996-1997. E. coli = suspect that exposureto oral fluoroquinolonesmay be high
Escherichiacoli; P. aeruginosa = Pseudomonasaeruginosa. outside the hospital as well, as they are commonlyused for
many community-acquired infections.
Thesedataalso illustratethreedifficultiesin reportingaccu-
contrast, significantly lower rates of use were reported in the rate data from multicenterlaboratory-based antimicrobialre-
adultICU areasfor trimethoprim-sulfamethoxazole, oral van- sistancesurveillancesystems:uncertainclinicalrelevance,as-
comycin, or oral fluoroquinolones.However,if oral and par- suringvalidityof results,andgeneralizabilityof the data.First,
enteral fluoroquinoloneuses were combined,rates of usage to minimize the burdenof data collection, we requestedall
were similar between ICU and non-ICU areas. Except for
laboratorysusceptibilitydata,not just susceptibilitydatafrom
parenteralfluoroquinolones,for each of the antimicrobial
patientswith nosocomialinfections.The lattertype of data is
groupswith higheruse in ICU areas,therewas a correspond- collected
by hospitalpersonnelperformingactive surveillance
ingly higherrate of the respectiveresistantorganismsamong for nosocomialinfections,such as is reportedto the Surveil-
isolates from ICU patientscomparedwith non-ICUinpatients
lance and Controlof Pathogensof EpidemiologicImportance
(figure 4).
(SCOPE)Project[17] or to the non-ICAREcomponentsof the
NNIS [9] system. Furthermore, ProjectICAREreceived data
Discussion for all isolates, not just sterile site or respiratoryisolates.
Our data illustrate that for many common nosocomial or- Consequently,the numberof isolatesreportedfromthe ICUsin
ProjectICARE is approximatelythree-foldgreaterthan that
ganisms, resistance tends to be focused in the intensive care
unit setting.Similarly,for manyantimicrobials,the rateof use reported fromthe same ICUs over the same time periodto the
is higherin the ICU areasthannon-ICUareas.This higherrate ICU component of NNIS (i.e., an infection-basedsurveillance
of use parallelsthe higherrates of resistanceobservedin the system) [18]. This laboratory-based reportingof all isolates
ICU areas for third-generationcephalosporinsand Enter- parallels the type of reports many hospital microbiologylabo-
obacterspecies, vancomycinand enterococci,and antipseudo- ratories generate to track cumulative antimicrobial susceptibil-
monal penicillins or third-generationcephalosporins and ity data. Although this type of reporting minimizes the burden
P. aeruginosa(figure4). In otherinstances,rates of use were of data collection, the clinical relevance of the isolatesreported
similar throughout the hospital, but resistance remained high in to a laboratory-based system is unclear (i.e., the exact natureof
ICUs; for example, use of the methicillin group of agents was any infection associated with the isolate is uncertain). How-
similarin all hospitalareas,but rates for methicillin-resistant ever, despite the ambiguityof the "infection"statusof these
S. aureus or methicillin-resistant coagulase-negativestaphylo- isolates,they definethe burdenof bacteriain the ICU,whichin
cocci were higher in ICUs. This suggests that other factors, turnrelatesto the risk of cross-transmission.
such as use of other drugs or cross-transmission, play an Second,thereshouldbe a mechanismto validatethe results
important role in propagation of these organisms. Further stud- of susceptibilitytestingto some degree.We requestedup to 20
ies to correlateresistanceandantimicrobialuse amongICUsor isolates with specific resistanceprofiles and retestedthem as
hospitalsare underway. described to providea measureof the validityof the data.We
A notableexceptionto the patterndescribedabove was the observedthatmisclassificationof imipenem-susceptible strains
higher prevalence of ofloxacin- or ciprofloxacin-resistant of Enterobacteriaceae and P. aeruginosaas resistantoccurred
P. aeruginosain outpatientareas (20%) than in the non-ICU among ProjectICAREhospitals[14]. We suspectthat this is
areasor ICU areas(- 17%).These datasuggestthatofloxacin- occurringin many U.S. hospitals [19]. Imipenemdegrades
or ciprofloxacin-resistant strains of P. aeruginosa now may be easily. Hospitallaboratoriesshouldmonitorthe storagecondi-
present more frequentlyamong patients in settings such as tions of susceptibilitypanels, cards, and disks carefullyand
emergencywardsand urgentcare and outpatientclinics, all of check qualitycontrolresultsfrequently.This experienceillus-
CID 1999;29 (August) AntimicrobialUse and Resistance
tratesthatattemptsto aggregatecumulativesusceptibilitydata
fromseveralinstitutionsshouldincludevalidationof the accu-
racy of laboratorytesting techniques.In phase 3 of Project
ICARE,we have addeda proficiencytesting programfor the
participatinglaboratoriesat the time of enrollmentas an addi-
tional step to improvedataquality.
Althoughthis projectassessed antimicrobialuse for a wide
range of antimicrobialagents in a large numberof hospitals,
these dataare not from a representativesampleof U.S. hospi-
tals. Thehospitalsparticipatingin ProjectICAREaresimilarto
the other hospitals reportingdata to the NNIS system but
underrepresentsmaller hospitals and certain geographicre-
gions [20]. However,because use varied so significantlybe-
tween the hospitalareas,we believe that reportingof antimi-
crobialuse must be stratifiedby hospital area to make valid
comparisonsbetweenhospitals.Furtherstudiesare ongoingto
determinethe importanceof specific ICU type as well as
regionalvariationsin patternsof antimicrobialuse. Further-
more, from these data we cannotassess the relativecontribu-
tion of specializedpatientpopulations,such as hemodialysis,
hematology,or organtransplantpatients,to patternsof inter-
pretingantimicrobialuse and resistance.In these data, these
specialized patient populationsare included in the non-ICU
inpatientareas.These types of patientsmay have very atypical
dosing patternsof antimicrobials,and assessmentof the de-
fined daily dose may not be the best methodto characterize
antimicrobialuse in such specializedpatientpopulations.To
addressthis limitation,data for these specializedpatientsare
currentlybeing collected in phase 3 of ProjectICARE.
ProjectICAREhospitalshave used these data in a quality
improvementprocessby focusing effortsin those areaswhere
reporteduse is in excess of that in other ICARE hospitals.
Routine surveillanceof antimicrobialuse, such as that de-
scribed here, can aid hospitals in targetinginfection-control
efforts. However, furtherstudies are needed to determinethe
most accurateway of comparingusage dataamonginstitutions
for qualityimprovementpurposesand to study the impactof
antimicrobialcontrolpracticeson rates of antimicrobialresis-
tance in the healthcare setting.
Surveillanceof antimicrobialresistanceamong organisms
infectionsis difficultbecauseof the
causinghospital-acquired
paucityof appropriatedataconcerningseveralparametersthat
are relevant for the study of antimicrobialresistance.These
include the case-mix of the patientpopulation,validationof
microbiologyresults, severityof illness, and accurateantimi-
crobialusage data.In addition,emergenceand spreadof these
organismsdo not occurjust withinthe walls of the acute-care
hospital.In the eraof managedcare,extended-caresettingsand
the outpatientsettinghave become associatedwith such resis-
tance.Thus, a surveillancesystem for antimicrobialresistance
amongorganismscausinghospital-acquired infectionneeds to
addressthese concerns at both the local and nationallevels
(i.e., a multicentersurveillancesystemis needed).Phase2 data
251
from Project ICARE underscore the importance of obtaining
accurate data about antimicrobial usage and of validating mi-
crobiology results. In addition, they suggest that monitoring
antimicrobial use in hospitals should help hospitals distinguish
problem areas where antimicrobial use appears to be a primary
factor contributing to antimicrobial resistance from other areas
where factors such as cross-transmission may need increased
attention. With information on use and resistance, the infec-
tious disease community will be one step closer to winning the
battle against the emergence and spread of antimicrobial-
resistant organisms in the health care setting.
Acknowledgments
The authorsthank the infection-control,pharmacy,and micro-
biology personnel from the participating ICARE hospitals for
making this study possible. Also, they thank Sarah C. Pichette,
M.P.H., for assistance with data management, and Jasmine Mo-
hammed, M.P.H., Elise Felicione, M.P.H., and SusannahHubert
for assistance with confirmationof isolate identity and suscepti-
bilities.
References
1. ArchibaldL, PhillipsL, McGowanJE Jr,MonnetD, TenoverFC, Gaynes
RP. Antimicrobialresistancein isolates frominpatientsand outpatients
in the UnitedStates:the increasingimportanceof the intensivecareunit.
Clin InfectDis 1997;24:211-5.
2. Centersfor Disease Controland Prevention.Nosocomial enterococcire-
sistantto vancomycin-United States,1989-1993. MMWRMorbMor-
tal Wkly Rep 1993;42:597-9.
3. Boyce JM.Are the epidemiologyandmicrobiologyof methicillin-resistant
Staphylococcusaureuschanging?JAMA 1998;279:623-4.
4. Herold BC, ImmergluckLC, MarananMC, et al. Community-acquired
Staphylococcusaureus in childrenwith no identi-
methicillin-resistant
fied predisposingrisk. JAMA 1998;279:593-8.
5. Moreno F, Crisp C, JorgensenJH, PattersonJE. Methicillin-resistant
Staphylococcusaureusas a communityorganism.Clin InfectDis 1995;
21:1308-12.
6. StrausbaughLJ, Crossley KB, Nurse BA, ThruppLD. Antimicrobial
resistancein long-term-carefacilities. Infect ControlHosp Epidemiol
1996;17:129-40.
7. GoldmannDA, WeinsteinRA, WenzelRP, et al. Strategiesto preventand
controlthe emergenceandspreadof antimicrobial-resistant microorgan-
isms in hospitals.JAMA 1996;275:234-40.
8. Monnet DL, ArchibaldLK, Phillips L, et al. Antimicrobialuse and
resistancein eight U.S. hospitals:complexitiesof analysisand model-
ing. Infect ControlHosp Epidemiol1998;19:388-94.
9. Emori G, CulverDH, HoranTC, et al. NationalNosocomial Infections
Surveillance(NNIS): descriptionof surveillancemethods.Am J Infect
Control1991;19:19-35.
10. NationalCommitteefor ClinicalLaboratoryStandards.Methodsfor dilu-
tion antimicrobialsusceptibilitytests for bacteriathatgrow aerobically:
approvedstandard.NCCLSdocumentM7-A3. Villanova,PA: National
Committeefor ClinicalLaboratoryStandards,1993.
11. NationalCommitteefor ClinicalLaboratoryStandards.Performancestan-
dards for antimicrobialdisk susceptibilitytests: approvedstandard.
NCCLS documentM2-A5. Villanova, PA: National Committeefor
ClinicalLaboratoryStandards,1993.
12. NationalCommitteefor ClinicalLaboratoryStandards.Performancestan-
dardsfor antimicrobialsusceptibilitytesting: approvedstandard.NC-
252Fridkin
CLS documentM100-S6.Wayne,PA: NationalCommitteefor Clinical
LaboratoryStandards,1995.
13. Amsden GW, SchentagJJ. Tables of antimicrobialagent pharmacology.
In: MandellGL, BennettJE, Dolin R, eds. Principlesand practiceof
infectiousdiseases.New York:ChurchillLivingstone,1995:492-528.
14. Steward CD, McGowan JE Jr, Tenover FC, Fridkin SK, Gaynes RP.
Multicenterstudyof susceptibilityof imipenemdemonstratesthattest-
ing problemspersist [abstractno 769]. Clin InfectDis 1998;27:1064.
15. Hoiby N. Microbiologyof cystic fibrosis.In: HudsenME, Geddes DM,
eds. Cystic fibrosis.London:Chapmanand Hall Medical, 1995:84.
16. KielhofnerM, AtmarRL. Life-threatening Pseudomonasaeruginosain-
fections in patientswith humanimmunodeficiencyvirus infection.Clin
Infect Dis 1992;14:403-11.
17. PfallerMA, Jones RN, MarshallSA, EdmondMB, Wenzel RP, SCOPE
Hospital Study Group.Nosocomial streptococcalblood streaminfec-
et al.CID 1999;29 (August)
tions in the SCOPE Program:species occurrenceand antimicrobial
resistance.Diagn MicrobiolInfectDis 1997;29:259-63.
18. FridkinSK, PryorER, EdwardsJR,McGowanJE Jr,TenoverFC, Gaynes
RP. Do hospital antibiogramsreflect resistancein nosocomial infec-
tions? [abstractno 90]. In: Programand abstractsof the 8th annual
meetingof the SocietyforHealthcareEpidemiologyof America(Wood-
bury,NJ). 1998:40.
19. White RL, Kays MB, FriedrichLV, Brown EW, Koonce JR. Pseudore-
sistance of Pseudomonasaeruginosa resulting from degradationof
imipenemin an automatedsusceptibilitytesting system with predried
panels. J Clin Microbiol1991;29:398-400.
20. SartorC, EdwardsJR, Gaynes RP, Culver DH. Evolution of hospital
participationin the NationalNosocomial InfectionsSurveillanceSys-
tem, 1986 to 1993. Am J Infect Control1995;23:364-8.