JOURNALOF CLINICAL MICROBIOLOGY,Dec.1978,p.695-699 Vol.8,
0095-1137/78/0008-0695$02.00/0
Copyright © 1978 AmericanSociety forMicrobiology PrintedinU
Evaluation of the Repliscan System for Identification of
Enterobacteriaceae
STEVEN D. BROWN AND JOHN A. WASHINGTON II Mayo Clinic andMayoFounrdation, Rochester, Minnesota55901
Receivedforpublication28August 1978
TheRepliscan system,asemiautomated method foridentifying gram-negative
bacilli,wasevaluatedfor itspotential usefulness in clinical microbiology
labora-tories. A total of1,877 isolates, including 1,712 fermentative and 165
nonfermen-tative organisms, were tested in parallel with the Repliscan and Enterotube
methodsof enteric identification. Discrepancieswereretested in eachsystemas
well as with conventional methods. The Repliscan method correctly identified
91%, misidentified 2%, and failed to identify 7%of the fermentative organisms
tested. Thesystem consistently failedtorecognize satisfactorily nonfermentative
organisms. Ofthegeneraunderstudy,Enterobacterposedthegreatest problem
tothesystemin terns ofoverallidentificationrates.TheRepliscan appears tobe
anefficient, economic,and effectivelaboratorytool for identification of
Entero-bacteriaceae.
,No.
J.S.A.
The accurate identification of
Enterobacteri-aceaefamily members represents aconsiderable
expense for a microbiology laboratory interms
ofatechnologist'stime and material costs (6).
During thepastdecade, a considerable amount ofefforthasbeen directedtoward the develop-mentof avariety ofkits,systems,methods,and
procedures for identifying these organisms.
Commercial kitsnowavailable have proven to
behighly reliable (10) butarerelatively costly.
Inexpensive replicate-plating techniques have
beenused formany yearsforagardilution
an-tibioticsusceptibilitytesting (3, 8) and are now
being used inalimitedwayfortheidentification
ofmicroorganisms (1,5, 9).The recent
comput-erization of these methods (7; B. Filburn, F.
Houston, V. Shull, W. L. Krause, and P.
Char-ache, Abstr. Annu. Meet. Am. Soc. Microbiol.
1978, C165, p. 304) has led to more accurate
identifications and a decrease in overall cost.
The Repliscan system (Cathra International,
Ontario, Canada) is a commercially available
replica-plating method for the identification of
Enterobacteriaceae. This report presents the
results of an evaluation of the computerized
systemfor itsaccuracyin identifying
fermenta-tive, gram-negative bacilli.
MATERIALS AND METHODS Cultures tested. A total of 1,786 gram-negative
bacilli, including 1,621 fermentative and165 nonfer-mentative strains, were selected from clinical
speci-menssubmittedto amicrobiologylaboratory.An ad-ditional91 stock organismswereutilized toprovide
species that are normallyencountered infrequently.
Procedures for isolation and identification ofclinically
important bacteria have been described (2, 4). All
organismswereassignedanumerical code and identi-fied blindly by the respective test methods.
Nonfer-mentativeorgnism wereincluded in thisevaluation
inan effort to exceed the stated capabilities of the systemandto testforpossiblefalse identificationas a fermentative species.
Identificationmethods.All organisms were iden-tified inparallelbyusingthe Enterotubeidentification system(RocheDiagnostic Div.,Nutley,N.J.) and the
Repliscan method. Discrepancies were retested in
each system. TheEnterotubetestsystemwas inocu-lated, incubated, and read in accordance with the
manufacturer's instructions. Since the performance
characteristics of the Enterotube have been firmly established (10), furtherevaluationof thisproduct is beyond the scope of this publication. Classical tests (2) wereused to identify all stock cultureorganisma, aswellastoresolvediscrepancies between the Enter-otubeandRepliscansystems.
Repliscansystem.TheRepliscan system consists ofavariety ofbiochemical and antibiotic plated media
(Table 1), amultiple-inoculum replicating device, a
viewing table allowing visual inspection and electronic recording of individualreactions,andacomputer
ter-minai complete withhard-copy printout. The plated
biochemical mediawerepurchased fromthe manufac-turerandstoredasdirected.Antibiotic-containing me-dium wasprepared asrecommendedbythe Interna-tionalCollaborativeStudy (ICS) (3) andasdescribed indetail elsewherebyWashingtonandBarry(11) for agar dilution susceptibility testing. All plates were inoculated according to the manufacturer's instruc-tions. One to three colonies of thebacterium to be identifiedwereinoculated into2mlof Mueller-Hinton broth and grown to aturbidity equaltothatof a 0.5 McFarland barium sulfate turbidity standard. One-695
on February 7, 2020 by guest
http://jcm.asm.org/
TABLE 1.
Basic
testcomponents of Repliscan systemSusceptibility testing Identification
Antibiotic ,ug/ml
Mueller-Hinton Tetracycline 4
(growthcontrol)
Citrate Chloramphenicol 16
Lysine decarboxylase Kanamycin 8 Ornithine decarboxyl- Cephalothin 8
ase
Urea Gentamicin 4
Deoxyribonuclease Ampicillin 8
Colistin Cephalothin H2S Bile esculin Arginine dihydrolase Glucose
Lactose Sucrose Mannitol Inositol Arabinose
half milliliter of this suspensionwasthen placedinan
individual well of the replicating device and inoculated
ontoeach of the23 biochemicaland antibioticplates. Thereplicating deviceallows thesimultaneous inoc-ulation ofupto36organismsonasingleagarplate. AUl plateswereincubatedaerobicallyat35°C for18to
20 h. The presence of 3% agar in the biochemical mediumeffectivelyinhibited theswarming of Proteus
spp., as well as limiting the diffusion of metabolic
products. After incubation,theplateswerearranged on aloadingtrayand inserted into theviewing table. Thegrowthofasingleisolateoneach of the 23 types
of mediawasobserved bytransmitted light through
individualwindows intheviewingtable.Positiveand negativereactionswererecordedbyusingapenlight
andphotosensitivereceptorslocateddirectly adjacent totheviewingwindow for eachtestplate.When the reactions on all test plates had been recorded, the computeranalyzedthe information entered and pro-vided theidentificationof theorganism. The limited antibiogramwasusedbythecomputertoconfirm the biochemical identification of the organism. When a
given isolate had been identified, the viewing table automaticallyadvanced theplatestothenextreading position,and theidentificationcyclewasrepeatedfor
each of the 36isolatesperplate.
Because the formulation ofsomeof the mediaused in theRepliscansystemdiffersconsiderablyfrom that ofconventional media, certaintests produce results that differed from those reported by Edwards and Ewing (2)andEwingand Martin(3). Since, however, themanufacturer'scomputerizeddatabasewas
devel-oped independentlyof the classical reactions of Ed-wards and Ewing (2), a test-by-test comparison of
individualreactionswasdeterminedto be of limited value and thuswasnotperformed.The evaluationwas
designedand conductedtodetermine theoverall
ac-curacyofidentification bythe experimental system.
Antibioticsusceptibilitytests. Themultiple in-oculator used in the Repliscan system consists of con-ical stainless-steel rods which taper to a point1mm in diameter. Since the volume delivered by these rods differs from that ofaSteersreplicator (8) and dilution of the inoculum is not made as for the ICS agar dilution method (3, 11), a study was undertaken to compare the minimalinhibitory concentrations (MIC)
of organisms tested in each system. The ICS agar
dilution technique (3)served as thereference method. Atotal of 242consecutivegram-negative isolates was tested by each method. The results of the two systems wereexpressed as an MIC ratio(ICS method/ Replis-can method). If the MIC value for each of the two methods wereidentical, theresulting ratio would be 1; MICs which were within ±1 doubling dilution pro-duced ratios of0.5and 2;MICs within ±2 doubling dilutionsproduced ratios of0.25and 4.
RESULTS
Biochemical identification. The Repliscan
systemcorrectlyidentified 91% of the 1,712
fer-mentative organisms tested in this evaluation (Table 2). Thirty-five organiams (2%) were
in-correctly identified by thesystem,andan
addi-tional120organisms (7%) couldnotbeidentified andproducedinconclusiveresults.
Escherichia coli and Klebsiellapneumoniae represented the majority of strains tested,
pro-ducing correct identification rates of97.5 and
88.8%, respectively (Table 3). Proteus rettgeri, Providencia stuartii, Enterobactercloacae, En-terobacteragglomerans, Enterobacter hafniae, Edwardsiella tarda, and Yersinia enterocolit-icaprovidedthegreatestchallengetothesystem
with fewer than 85% of the strains being
cor-rectly identified. The remaining10species tested produced identification rates of greater than 85%.K.pneumoniae presentedthelargestsingle species identification problemto thesystem in
terms of identification errors. Thirteen ofthe
341isolates ofK.pneumoniae (3.8%)were misi-dentifiedasKlebsiellaozaenae(Table4).These
isolatesrepresented38% of thetotal
identifica-tionerrorsandweretheresultoffalsely negative
citrate reactions.All of the strains testedproved
tobecitratepositive bythe Enterotube method
andonconventionalmedia, although many were
delayedpositivereactionsthatrequired48hof incubation. The genus Enterobacter also
repre-sented 38% of the identificationerrorswith no
consistentexplanationfor themisidentifications.
There were no trendsestablished withinagiven
TABLE 2. Organismidentification summary
Organism No. % % %
characteristic tested Cor- Error Inconf
rect clusive
Fermentative 1,712 91.0 2.0 7.0
Nonfermentative 165 73.9 1.2 24.8
on February 7, 2020 by guest
http://jcm.asm.org/
REPLISCAN SYSTEM 697 TABLE 3. Organisms tested
Organism Escherichiacoli Klebsiella pneumo-niae Kozaenae Proteus mirabilis P.morganii P.vulgaris P.rettgeri Providencia stuar-tii Enterobacter cloa-cae E.aerogenes E.agglomerans E.hafniae Serratia marces-cens Citrobacter diver-sus C.freundii Salmonella Shigella Arizona hinshawii Edwardsiella tarda Yersinia enteroco-litica Pseudomonas aeruginosa P.maltophilia Acinetobacter cal-coaceticus Miscellaneous No. %
tested Cor.rect
640 97.5 341 88.8 2 100 137 97.8 55 94.5 39 92.3 21 76.2 15 60 125 70.4 72 93.1 16 18.7 6 17 71 95.7 45 97.8 52 85.6 32 100 30 93 7 86 2 50 4 0 97 97.9 10 80 34 11.7 24 62.5 Error 0.3 3.8 Incon-clusive 2.2 7.3 o o 0 2.2 0 5.5 2.6 5.1 0 23.8 7 33 6.4 23.2 0 6.9 25.0 56.3 33 50 1.4 2.8 0 2.2 3.9 10.5 o o 0 7 0 14 0 50 25 75 0 2.1 0 20 2.9 85.3 4.2 33.3
species andnoobviouscausefor the inconclusive
or wrong identifications; however, the number
of strains testedwasinsufficient foranin-depth
analysis of the identification errors.
Seven percent of the organisms studied
pro-duced inconclusiveresults andwerenot
identi-fied by the Repliscan's computerized datum
base. The highest percentage of inconclusive
biochemical patterns was from species within
thegenusEnterobacter. Noexplanationcanbe
offered for the relatively poor performance in
thisareaother thanpossible inadequacies within
the datum base itself. The small number of
strains studiedprevented the recognition of
spe-cific trendsor"biotypes."
Although oxidase-psoitive, nonfermentative organism arespecificallybeyond the stated
ca-pabilities of the system, a limited number of
these strains were included in thestudy.
Rec-ognition of the biochemicalpatternasbelonging
toanonfermentative specieswasrecordedas a
correctresponsesincegenus andspecies
desig-nations were beyond the scope of the
instru-ment. Of the 165 nonfermentative organisms
studied, 122 (73.9%) wererecognized asnot
be-longingtothe familyEnterobacteriaceae(Table
2). Twoorganisms,Aeromonashydrophila and
Acinetobacter calcoaceticus,weremisidentified
as E. coli (Table 4) and represented an error
rateof 1.2%. Theremainingisolates(24.8%)were
not recognized by the system's computerized
datum base.
Antibiotic susceptibilities. Comparison of
theICSandRepliscan methods of susceptibility
testing revealed that 99.4% of the end points
were withina range of+1 10g2dilution (Table
5).All but six end points (0.2%)fellwitharange
of 2 dilutions (data not shown). The greatest
variationwasnoted forampicillin,cephalothin,
TABLE 4. Errors inidentification byRepliscan
system Organism (no.)
Klebsiella pneumoniae (13) Citrobacterfreundii(2). Enterobacter cloacae (3). E.cloacae (2). E.cloacae (3). E.agglomerans (1). E.agglomerans (1). E.agglomerans (2). E.hafniae (1). E.hafniae (1)
Serratiamarcescens (1).
Proteus vulgaris (1) ...
Escherichia coli (1).
E. coli (1).
Providencia stuartii(1). Yersinia enterocolitica (1) Aeromonashydrophila (1) A. calcoaceticus (1).
Incorrectly identifiedas:
K.ozaenae
E.coli E.agglomerans C.freundii
C. diversusorE.
ag-glomerans E.coli C.freundii K.rhinoschleromatis K. ozaenae E.coli S.liquefaciens P.mirabilis Afermentative K.ozaenae P.alcalifaciens E.coli E.coli E.coli
TABLE 5. Comparison ofICS andRepliscan
methodsforthequantitative susceptibilitytests MIC' ratio(%ofstrains) Antibiotic
<0.25 0.5 1.0 2.0 24.0
Amikacin 0.0 0.0 99.6 0.4 0.0
Ampicillin 1.6 10.7 83.1 4.5 0.0
Carbenicillin 0.4 5.8 92.1 1.6 0.0
Cephalothin 1.6 14.9 77.3 5.8 0.4
Chloramphenicol 0.8 5.0 79.3 13.2 1.6
Gentamicin 0.0 4.1 94.6 1.2 0.0
Kanamycin 0.0 2.0 97.5 0.4 0.0
Nalidixicacid 0.0 1.6 98.3 0.0 0.0
Nitrofurantoin 0.0 3.7 94.6 1.6 0.0 Tetracycline 0.4 0.8 71.9 26.5 0.4
Tobramycin 0.0 1.2 96.3 2.5 0.0
Trimethoprim- 0.0 0.4 98.8 0.8 0.0
sulfamethoxa-zole
% ofall tests 0.4 4.2 90.3 4.9 0.2 aMICratio,ICS/Repliscan,see text.
VOL. 8,1978
on February 7, 2020 by guest
http://jcm.asm.org/
chloramphenicol, and tetracycline. Repliscan MICvaluesweregenerally lower than ICS
val-ues for chloramphenicol and tetracycline but higherforampicillin andcephalothin. End point disagreementwas evenlydistributed amongall species tested. All remaining antibiotics
pro-duced endpointagreement in greater than92%
ofthe strainstested.
DISCUSSION
The resultsof thisevaluation showeda good
correlationbetween the Repliscanand conven-tionalidentifications.Many of the identification
errors could be eliminated by adding one or
moreconfirmatory tests.Whenforexample, an organism is identified by Repliscan as K.
ozaenae,aSimmons citrate testcould be inoc-ulated and held for48h. The final identification shouldreflect the results of the additional test
result.
The positive and negative biochemical
reac-tionswere easilyidentifiable inmostinstances. Ornithine decarboxylase and arginine dihydro-lase occasionally produced ambiguous results.In
manycases,these questionablereactionsdid not
affect the finalidentification of the organismin
question. In otherinstances, the ambiguous
re-actions led to inconclusive identifications, and
theorganisms required additional testing forus toachieveproperidentification.
The turbidity of the inoculumin theICSagar
dilution method (3, 11) is usually adjusted to
match that of one-half of a McFarland no. 1
barium sulfate standard andisdiluted 1:20
be-foreapplicationontheagarsurface, usually by
meansofaSteers replicator (8), each inoculating
prong of which is 3 mmin diameter. The MIC
is determined by exaning the agar surface withreflected light. With Repliscan the adjusted butundiluted inoculum isappliedwithan inoc-ulatingprong 1mmindiameter,andtheMIC is determined by
examining
the agar withtrans-mittedlight.Despite these technicaldifferences, MIC values obtained with each method didnot
differsignificantly. Moreover, fine, barely visible hazes, ignoredin theICSmethod(3,11),are not
visiblewithtransmittedlight,sothat endpoints
intheRepliscanweremoredefinite.
TheRepliscansystemwas asimple, economic,
and efficient method for the identification of
fermentative,gram-negative bacilli.Inoculation ofplates, reading, and recording of results can be accomplished in approximately 1.5 min per
organism.Mixedcultureswereeasilyrecognized
onthe solid media.
The economy realized in using the system
takestwoforms.The primarysavingsis in the
form ofdecreased mediacosts. Since up to 36
J. CLIN. MICROBIOL. organisms can beinoculated on a single plate,
thecostofperformingagiventestwasnominal ($0.02). Assuming100%utilization (36organisms perplate),the material cost for theidentification
of Enterobacteriaceae was determined to be $0.46per isolate.Thisfigure represents less than
20% of the costs of the API 20E orconventional
seven-tube sets (6). Identification of onlyseven
organisms per platewould cost$2.37 perisolate,
afigurewhich isroughly comparabletothecost
of many commercial systems. No allowances
weremade foroverhead expenses,
instrumenta-tion costs,orthe technologist'stime. A further savingswasnotedwith themergingof the
bio-chemical testingwith anagar dilution suscepti-bility method. Combining these two methods
has allowed efficient work organization and a
decreased requirement for technologists' time.
Whereastheday-to-dayoperatingexpense of thesystemisquite low,theinitial investmentin
instrumentation is rather substantial. For this reason,thefeasibility studiesof the system must bedetermined byindividual laboratoriesonthe
basis of need and the total number of isolates
tested on adaily basis.Laboratories
identifying
more than 18 organisms per day could expect
the system to become cost-effective within 4
years basedonmaterialsavings alone. Asmaller, lessexpensivemodel, knownastheReplireader (not evaluated in thisstudy) could be expected
to become cost-effective in approximately 2
yearsorless.
Replicate-plating methods provide a distinct advantagein terms ofquality control ofmedia
used in the system. Four quality-control
orga-nismscanbeinoculatedoneachtypeof medium andidentifiedasusualalong with32other iso-lates. E. coli, Pseudomonas aeruginosa,
Pro-teus mirabilis, and Serratia marcescens were
selectedasquality-control organismsbecause of theirabilitytoprovide apositive and negative
responsefor each reaction tested.Tabulation of
the controlreactionswasfacilitated bytheuse
of hexadecimal code numbers provided bythe
computer which identified the reactions of all
the biochemicaltests.
The major limitation to the system wasthe fact that 7% of the fermentativeorganisms tested produced inconclusiveidentifications; however, this percentage doesnotexceedthat which we
have noted with other systems for identifying
the Enterobacteriaceae (10). These organisms required identificationby conventionalmethods, resultingin another 24to48hdelayinreporting.
As moredata are obtained, additional
identifi-cationprofilescanbe entered into thecomputer
datum baseofthesystem andhopefullyreduce
thenumber ofstrainsrequiringadditional
test-ing. Until thiscan beaccomplished, a suitable
on February 7, 2020 by guest
http://jcm.asm.org/
699
back-up identification system must be main-tained to identify accurately those organisms producing inconclusive resultsontheRepliscan
system. Thisrequirement poses noproblem to
larger clinicalcentersbut mayadverselyaffect smallhospital laboratories.
Thenecessity foraback-upsystem,together withthe initial instrumentation cost,placesthe Repliscansystembeyondthelimits of practical-ity for manysmaller laboratories. However,the
Replireadersystem (amethodusingthe
Replis-candatum base butless instrumentation) may
bewell within the graspoflaboratories identi-fyingasfewas 10organisms perday.
Inconclusion, the results of this study indicate that the Repliscan system foridentification of Enterobacteriaceae compared favorably with conventional methods. The advantages of de-creased mediacostsandtechnologist's time
out-weigh the present limitations and the initial
expense of the system. This method of enteric
identificationpromisestobeavaluableadjunct
to a clinical laboratory which identifies large numbers of isolatesdaily.
LITERATURE CITED
1. Chadwick, P.,G. J.Delisle,and M.Byer. 1974. Bio-chemical identification of hospital enterobacteria by replica agarplating.Can. J. Microbiol. 20:1653-1663. 2. Edwards,P.R.,and W. H.Ewing.1972.Identification
ofEnterobacteriaceae, 3rd ed. Burgess Publishing Co., Minneapolis.
3. Ericsson, H. M., and J.C. Sherris. 1971. Antibiotic
sensitivitytesting report of an international collabora-tive study. Acta. Pathol. Microbiol. Scand. Sect. B 217(Suppl): 1-90.
4. Ewing, W.H.,and W. J. Martin. 1974. Enterobacteri-aceae, p. 189-221. In E. H.Lennette, E. H. Spaulding and J. P. Truant(ed.),Manual of clinical microbiology, 2nded. American Society for Microbiology, Washing-ton, D.C.
5. Fuchs, P.C. 1975. The replicator method for identifica-tion andbiotyping of common bacterial isolates. Lab. Med. 6:6-11.
6. Robertson, E. A., G. C. Macks, and J. C. MacLowry. 1976. Analysis of cost andaccuracy of alternative strat-egies forEnterobacteriaceae identification. J. Clin. Mi-crobiol. 3:421-424.
7. Sheaff, E. T., and N. A. Hinton. 1973. Development of anon-linecomputer system forevaluation of biochem-ical and sensitivity tests in a diagnostic bacteriology laboratory. Can. J. Pub. Health 64:82.
8. Steers, E., E. L.Foltz, and B. S. Graves. 1959. An inocula replicating apparatus for routine testing of bac-terialsusceptibility to antibiotics. Antibiot. Chemother. 9:307-311.
9. Stolzky,G.1964.Replica plating techniques for studying microbial interactions in soil. Can. J. Microbiol. 11: 629-636.
10.Washington, J. A., II. 1976. Laboratory approachesto
theidentification of Enterobacteriaceae. Human
Pa-thol. 7:151-159.
11.Washington,J.A., II, and A. L Barry. 1974. Dilution testprocedures, p. 410-417. In E. H. Lennette, E. H.
Spaulding and J. P.Truant (ed.), Manual of clinical microbiology, 2nd ed.American Society for Microbiol-ogy,Washington,D.C.
VOL. 8,1978
on February 7, 2020 by guest
http://jcm.asm.org/
