Comparison of micro ID and API 20E in rapid identification of Enterobacteriaceae

0095-1137/82/050885-06$02.00/O

Comparison of Micro-ID

and API

20E in

Rapid Identification

of Enterobacteriaceae

W.MANFORD GOOCH III*ANDGILBERT A. HILL

DepartmentsofPathology andPediatrics, Universityof UtahHealth Science Center, PrimaryChildren's Medical

Center,

and Latter

Day

Saints

Hospital,

Salt Lake

City,

Utah 84103

Received17August 1981/Accepted 21January1982

The effectiveness of Micro-ID and API 20Eassame-day identificationsystems

for Enterobacteriaceaewasevaluated in

comparison

withconventional identifica-tion by using 315 clinical isolates and 90 stock strains. The API 20E system was

heavily inoculated according to manufacturer's recommendations for same-day identification. Wefound that 83 and 81% of isolates provided adequate inocula for Micro-ID and API20E, respectively, andpurity of theheavy inocula wasnota

problem with eithersystem.Overallagreementwithconventional identificationat

genus and species levels was 93.5% with Micro-ID and 90.2% with API 20E. However, when

Klebsiella

pneumoniae and K. oxytoca were considered as a

single species and Proteus morganii was equated with Morganella

morganii,

agreement was 95.8 and 90.5%, respectively. Only 83.5% of isolates were

identifiedontheday of inoculation byAPI20E,incontrast to94.3%with Micro-ID. Theremaining isolates required supplementary overnight testing. Provisional (low selectivity) determinationswere consistentwithconventional identification with 49.3% of isolates with API 20E and 82.6% with Micro-ID. Telephone consultations with the manufacturers toresolveunprintedoctal codes requireda

maximum of 15min with Micro-ID and from2togreaterthan 48 h with API 20E.

Although they provide such benefits

as

com-puter-assisted identification, prolonged shelf

life, and efficient

use

of

storage space,

miniatur-ized "kit"

methods

of

identifying

members

of

Enterobacteriaceae

usually

require biochemical

reactions similar

to

those of conventional tubed

media and, consequently,

require incubation for

18

to

24

h. Use

of heavy inocula of

organisms

allows detection of microbial

enzymes

and

meta-bolic

products

within

hours, and this

approach

has been

successfully

incorporated into

a

com-mercial

system

(Micro-ID; General Diagnostics,

Morris

Plains,

N.J.). Recently, the manufacturer

of

API

20E

(Analytab Products, Plainview,

N.Y.) has recommended heavy inoculation of

this

widely used

identification

system to

provide

same-day

identification. The

present

communi-cation

reports a

comparison between Micro-ID

and API 20E as systems forsame-day

identifica-tion of

members

of

Enterobacteriaceae.

MATERIALSANDMETHODS

Specimens. Consecutivespecimens submittedtothe clinical microbiology laboratories of Primary Chil-dren'sMedical CenterandLatter DaySaints Hospital were considered for evaluation. Those specimens which upon primary isolation provided adequate in-oculaforprocessing of isolates of Enterobacteriaceae

by both Micro-ID and API 20E (same-day method) were entered into the study. Totals of 186 and 129

isolates from each institution, respectively, satisfied these criteria. Colonies which were morphologically consistent with Escherichia coliwereexcludedfrom thestudy after 150 strainswerecollected,so as not to create a greater preponderance of this species. In addition, 90 strains of less commonly encountered membersof Enterobacteriaceae were studied. Each of theseorganismswaspassed throughtwologphases of growth before inoculation into each of the systems.

Isolation and identification. Specimens were streakedfor isolationontoconventional isolation me-dia depending upon the source of the specimens.

Colonies of identical appearancewerepicked withan inoculating loop and added in sufficientquantityto6.5 ml of0.85%sterile saline solution to produce visible turbidity equivalent to a 1.0 McFarland standard. Samples of this suspensionwereusedtoinoculate the 20reaction chambers ofan API20Eunitaccordingto themanufacturer's instructions for same-day identifi-cation. The test chambers for arginine dihydrolase,

lysine decarboxylase, omithine decarboxylase, and urease wereoverlaid with sterile mineraloil, and the unitwasincubated for5 hat35°C. Thereafter, if the

glucosechamberwaspositive, thefollowing reagents wereaddedtotherespective chambers:1drop of40%

KOH and1drop of6%alpha-naphthol for the Voges-Proskauer test;1drop offerric chloridefortryptophan

deaminase; 1 drop of Kovacs reagent for the indole test. The resulting combination of color reactions in the 20chamberswasconverted intoaseven-digit octal code for each organism, and identification of each organismwasdeterminedby comparing the code with entries in a manual provided by the manufacturer 885

on February 7, 2020 by guest

http://jcm.asm.org/

specifically for same-day identification. Identifications listed as excellent, very good, or acceptable were utilized. Lower levels of identification with poor selec-tivity which required additional testing and overnight incubation were termed "inadequate."

After inoculation of the API 20E unit, sufficient 0.85% sterile saline was added to the remaining inocu-lum suspension to achieve visible turbidity equivalent to a0.5McFarland standard. Each of the 15 wells of a Micro-ID unit was inoculated with 0.2 ml of the resulting suspension and incubated at 35°C for 4 h. After incubation, 0.1 ml of 20% KOH was added to the well of the Voges-Proskauer test, and the unit was manipulated according to the manufacturer's instruc-tions to mix the various reactants. Identification was determined in a manner similar to that for the API 20E, using an identification manual provided by the manu-facturer (revision 07179). Identifications listed as ex-cellent, extremely good, very good, good, or accept-able wereutilized. Lower levels of identification with poorselectivity requiring additional testing and over-night incubation were termed "inadequate."

In addition, the following tubed media were inocu-lated with the suspension used to inoculate the Micro-ID: triple sugar iron agar; lysine iron agar; lysine decarboxylase; ornithine decarboxylase; methyl red-Voges-Proskauer broth; sulfide-indole-motility medi-um; Simmons citrate agar; Christensen urea agar. Reactions wereinterpreted in the conventional fashion (5, 8). When necessary for identification, additional biochemical andserological tests were performed. All isolates were tested for the presence of indophenol oxidase. Isolates identifiedas SalmonellaorShigella

in either system wereconfirmed by serological testing in the clinical laboratory (Shigella and Salmonella antisera;Difco Laboratories, Detroit, Mich.) and also were submitted to the Utah State Department of Healthfor additional biochemical and serological con-firmation.

A MacConkey agar plate was inoculated with the remaining inoculum suspension and incubated for 18 to20 hat35°Ctoevaluate thepurity of thesuspension.

Pureisolates thatyieldedcodes whichwere notlisted in the respective identification manual and which remained unidentifiable after consultation with the manufacturer by telephone were termed "unidentifi-able" with therespective system andwereconsidered tobe misidentifications at the genus level.

Unless otherwise indicated, mediawere manufac-turedby MicroBioProducts. Inc. (Phoenix, Ariz.)or were prepared in the respective hospital laboratory. Noattemptwasmadetorestrict thestudyto adefined set or setsof manfactured media. Each lot of medium and each lot of commercial kitswastested for sensitiv-ity and reliabilsensitiv-ity with negative controls and with strains of Escherichia coli, Salmonella enteritidis, Shigella sonnei, Proteus vulgaris, Klebsiella pneu-moniae, andEnterobacter cloacae.

Assessment of inoculumadequacy. Aftercompletion

of the above evaluation, the next 300 consecutive clinicalspecimenscontainingmembersof Enterobac-teriaceae wereevaluated foradequacyof inoculum in each system. Specimens were streaked for isolation ontoMacConkey agarand,for stoolspecimens,onto

Hektoen enteric agar. Colonies of identical appear-ance werepicked withaninoculating loopandadded to3.5 mlof0.85% salinetoproduce turbidity

equiva-TABLE 1. Clinicalisolates used to compare Micro-IDwithsame-day API 20E

No. of No. of

Organism clinical stock

isolates isolates

Escherichiacoli 150

Shigella sp. 2 10

Salmonella sp. 4 11

Arizonahinshawii 1 11

Citrobacterfreundii 5 9

C.diversus 1 10

Klebsiella pneumoniae 48

K. oxytoca 29

K.ozaenae 1 2

Enterobacter cloacae 9 3

E. aerogenes 5 7

E.agglomerans 4 4

Serratia marcescens 7 9

S.liquefaciens 1 4

S.rubidaea 1 2

Proteusmirabilis 35

P.vulgaris 6 2

Morganellamorganii 4 5

Providenciastuartii 2 1

lent to a 0.5 McFarland standard, as required by Micro-ID. Thereafter, if sufficient colonies remained, anadditional inoculum in 1.5 ml of saline was added to produce turbidity equivalent to a 1.0 McFarland stan-dard, as required by the same-day API 20E procedure.

RESULTS

Distribution of

the

isolates of

Enterobacteria-ceae

used in this study is listed in Table

1. A

total

of 262

(65%)

of

the

isolates

were

strains of

E.

coli,

K.pneumoniae, Klebsiella oxytoca, and

Proteus

mirabilis. The

degree

to

which

Micro-ID

and

same-day

API

20E

did

not

agree with

conventional identification is

presented in Table

2, and

specific

misidentifications

are

listed in

Table 3. Levels of

identification which would

have

postponed final determination for

an

addi-tional

day due

to

recommended

supplementary

testing

were

termed

"inadequate" and

were

considered

to

be misidentifications since

same-day

identification was not

possible. Using

this

criterion,

Micro-ID

inadequately

identified 5.7%

of

isolates

(23/405),

and in

82.6%

(19/23)

of such

instances

the

provisional identification

was

ei-ther

identical

to the

determination

made with

conventional media

or was

included

in thetwo

possible choices. The

same-day

API20E

proce-dure

provided inadequate identification of 16.5%

of isolates

(67/405),

and

only

49.3%

(33/67)

of

the

provisional identifications

were

consistent with

determinations made

with

conventional media.

Exclusive of

inadequate

determinations,

Mi-cro-ID

disagreed

with

conventional

identifica-tion

atthe

genus

levelwith

3.7%

of isolates

(14/

382) and

atthe

species

level

with

2.9%

(11/382).

Seven

of

the misidentifications at the

species

on February 7, 2020 by guest

http://jcm.asm.org/

TABLE 2. Misidentification of Enterobacteriaceaeby Micro-ID andsame-day API 20Eonday of inoculation

Levelof misidentification(no.discordant)

Organism (no.tested) Micro-ID API20E

Inadequatea Genus Species Inadequate Genus Species

Escherichia coli (150) 1 5 23 2

Klebsiella pneumoniae (48) 3 1 5 7 1

K.oxytoca(29) 1 2 4

K.ozaenae(3) 1 1 2 1

Enterobacteragglomerans (8) 1 1 7

E. aerogenes(12) 1

E.cloacae (12) 2 3 2

Serratia marcescens(16) 1 2

S.rubidaea(3) 1 1 1 1

S.liquefaciens(5) 4

Citrobacter diversus(11) 1 1 5 2 4

C.freundii(14) 2 5

Proteusvulgaris (8) 3 1 2 1

P.mirabilis (35) 3 2 2 1 2 1

Morganella morganii (9) 1 3 1

Providenciastuartii (3) 1 1 1

Arizonahinshawii (12) 1 1 1

Shigella sp.(12) 2 2 6

Salmonella sp. (15) 1

Total 23 14 11 67 17 16

aLevelsofidentification with poorselectivitywhichrequiredadditionaltestingand

overnight

incubation.

level

represented designation

of five strains of

K.

pneumoniae

as

K. oxytoca

and

two strains

of

K.

oxytoca

as K.

pneumoniae. Two

of the

misidentifications

at

the genus level

were

strains

of

P.

mirabilis

which

were

identified

as

Morgan-ella

morganii.

Overall

agreement

between

Mi-cro-ID

and

conventional media

was

93.5%

(357/

382).

API 20E

disagreed

with

conventional

me-dia

at

the

genus

level with 5.0% of isolates

(17/

338) and

at

the

species level with 4.7%

(16/338);

overall agreement

was

90.2%

(305/338).

Only

one

strain

of

K.

pneumoniae

was

misidentified

as K.

oxytoca

with this system.

Micro-ID

provided inadequate

identification

of

7.8%

(7/90)

of stock isolates and

agreed

at

the

genus

and

species

levels with conventional

iden-tification with 96.4%

(80/83).

API 20E

produced

inadequate

determinations with

18.9%

(17/90)

of

these

isolates and

agreed

with conventional

de-terminations with

74.0%

(54/73).

Identification codes which were not listed in

the

respective

manualswere

generated by

each

system. Toll-free

telephone consultation with

the

manufacturers

resolved 16 of 23 such

in-stances

with

same-day

API

20E and

8

of 10

similar instances with

Micro-ID. The isolates

which remained

unidentified

by

Micro-ID were

an E. coli and a

Citrobacter diversus, whereas

same-day

API

20E

was

unable to identify two

strains each of

P.

mirabilis and Shigella

and

one

each of

P.

vulgaris, C. diversus, and M.

mor-ganii.

The

biochemical

tests

common

to

both

sys-tems are compared

in Table

4.

Because some

authors

stress

that

lysine iron agar is

an

inappro-priate

substitute

for

Moeller's

lysine

decarbox-ylase medium (6), both media

were

inoculated

with 90

clinical isolates.

Thereafter,

since

great-er

than

95%

concordance

resulted,

lysine

iron

agar

was

used

as

the

sole

test

for

lysine

decar-boxylase

activity.

The

decarboxylase reactions

were

weak but readable with the

same-day

API

20E;

otherwise, end

points

with each

of the

commercial systems

were

easily

interpreted.

Re-actions of Micro-ID and

same-day

API

20E

were

quite

similar

to

those with

conventional

media,

with Voges-Proskauer,

H2S, and

ornithine

de-carboxylase

being

virtually

identical. The least

degree of agreement between each system and

conventional

reactions

was

with urease, but the

urease

reaction alone did

notcause a

misidentifi-cation.

Indole

was

the

only biochemical

reaction

which resulted in

substantial

misidentifications

as a result of its discordance with the

conven-tional reaction.

Four

such

misidentifications

due

to

indole occurred with

API

20E,

and 13

oc-curred

with

Micro-ID. The citrate reaction in

same-day

API-20E was

positive

withonly seven

organisms,

but

its

nonreactivity resulted in

no

misidentifications.

There

wasno

problem in

maintaining purity of

inocula when using routine laboratory

proce-dures. Of 300

consecutive clinical isolates of

Enterobacteriaceae,

248

(83%) provided

ade-15,

on February 7, 2020 by guest

http://jcm.asm.org/

TABLE 3. Disagreements between conventional system, Micro-ID, and same-day API 20E

Conventional identification API 20E(no. in Micro-ID(no. in disagreement)

(no.tested) disagreement)

Escherichia coli (150) API Group 1 (2) Shigella sp. (1)

Arizonahinshawii (3) Unidentifiable (1)

K. oxytoca (1)

APIGroup 1 (1)

K.oxytoca (5) E.coli(1)

K.pneumoniae(2) E.coli (1)

Yersiniapseudotuberculosis (1)

Yersinia enterocolitica(1) K.oxytoca(1)

E.cloacae (12)

Serratiamarcescens (16)

S.liquefaciens (5)

Enterobactersp.(1) E.aerogenes(1)

Serratiasp. (2) Serratiasp. (4)

S.rubidaea (3) Arizonahinshawii (12)

Shigellasp. (12)

Proteusvulgaris (8) P.mirabilis (35)

K.pneumoniae (1) Serratiasp. (1)

Salmonellasp. (1) Unidentifiable(2) Yersiniapestis(4) Unidentifiable(1) Unidentifiable (2) P.rettgeri (1)

Hafniaalvei (1)

Proteussp. (1) M.morganii(2) P.vulgaris (2) Morganella morganii(9)

Providencia stuartii(3) Citrobacter diversus(11)

Unidentifiable(1) Providenciasp. (1) Shigella sp.(1) Citrobactersp. (4) Unidentifiable(1)

Providencia sp.(1) Unidentifiable(1)

quateinocula forMicro-ID, and 81%were suffi-cient for thesame-dayAPI20Eprocedure. Only 73% (35/48) of

lactose-nonfermenting

isolates from stoolspecimens provided adequate inocula forMicro-ID, and 65% (31/48)wereadequatefor

thesame-day API 20E procedure. DISCUSSION

Both Micro-ID and API 20Ehave been

dem-onstrated to be reliable alternatives to

conven-tional identification of Enterobacteriaceae with-out requiring expensive and sophisticated equipment. However, Micro-ID allows

differen-tiation of theseorganismsontheday of primary

isolation whereas API 20E requires overnight

incubation. Procedural modifications allowing same-day identification would be an attractive

alternative for laboratories

already

using API

20E.

Since this study was designed to compare

Micro-ID and same-day API 20E in a clinical

laboratory setting, kits were used by bench

technologists with 315 consecutive clinical

iso-lates and 90 less common stock strains. Fresh

clinicalisolatesare morerepresentativeof

rou-tine performance of an identification system

than stock cultures. Aldridge et al. (1)

demon-strated 93.2% correlation ofan earlier

genera-tion of Micro-ID with conventional

identifica-tion by using clinical isolates and only 83.6% Klebsiella pneumoniae (48)

K.oxytoca (29) K. ozaenae(3)

Enterobacteragglomerans (8)

E.aerogenes(12)

J.CLIN.MICROBIOL.

on February 7, 2020 by guest

http://jcm.asm.org/

TABLE 4. Agreement ofbiochemical testswith conventional system, Micro-ID and same-day API

20E,using 405 cultures

%Agreement

Biochemicaltest API 20E Micro-ID API 20E

vs con- vs con- vs

Micro-ventional ventional ID

Voges-Proskauer 89 90 85

H2S 97 96 98

Indole 94 91 92

Ornithine 88 88 94 decarboxylase

Lysine 84 82 96

decarboxylase

Urease 61 71 87

ONPGa 93

Arabinose 94

Inositol 89

Sorbitol 86

aONPG,

o-Nitrophenyl-p-D-galactopyranoside.

when

using stock cultures. Similar

correlations

with API 20E

(overnight incubation)

were 95.2

and

79.5%, respectively

(1).

In the current

study, Micro-ID

performed

much better than

API20E as a

rapid identification

system

for

the

stock isolates.

API

20E

produced

18.8 and

16.2% inadequate determinations

and

36

and

4.9% incorrect determinations

with

stock

and

fresh clinical isolates,

respectively.

Incontrast,

Micro-ID

provided only

7.8 and 5.1%

inade-quate

identifications and

3.6

and

7.4% incorrect

identifications, respectively.

The better results

with Micro-ID

as

compared with

the report

of

Aldridge

etal. (1) may represent greater

experi-ence

of

the

manufacturer with unusual isolates

since

1978.

The

degree

to

which

a

same-day

system

does

not

require supplemental

overnight testing is

a

measure

of its

same-day

applicability,

and

iden-tifications which

require

additional

tests in this

context are

"inadequate." Sixty-seven (16.5%)

of isolates

in

this

study

were

inadequately

identi-fied

by same-day

API

20E in

contrastto

5.7% of

isolates with Micro-ID.

In

49.3% of

such

in-stances with API 20E

and

in

82.6% of

such

instances

with

Micro-ID,

the

provisional (low

selectivity)

interpretation included

the

conven-tional determination and therefore could be

re-ported

as a

provisional identification

to be

con-firmed upon further testing. Effectiveness of API 20E as a same-day procedure was further

compromised

because of relative unavailability

of the

telephone

consultation service. The same-day API 20E generated more unprinted system codes than

Micro-ID,

and responses to tele-phone requests for assistance required 2 to 4 h in contrast to amaximumof 15 min with Micro-ID. Slow response time was furthercomplicated by

the fact that the API 20E consultation service is

available only

from 9:00 a.m. to 5:30p.m.

(East-ern Time) and only on weekdays. Consequently, several responses from API 20E were

delayed

overnight

or over a weekend.

Previous

investigators

have

reported

from

93.2 to96.1% overall agreement between Micro-IDand conventional

identification

with

clinical

isolates similar to those

of

the present

study

(1-4). However, those authors

did

notseparate K. pneumoniae

from

K. oxytoca and did not

assign

separate

generic

status to M. morganii. These

distinctions

weremade inthis

study

because the

data bases of

Micro-ID and

API 20E

do

so. If

these distinctions had

not

been

made,

Micro-ID

would have correlatedwith conventional

identi-fication with

95.8% of isolates. After

making

the

aforementioned terminological changes

in

Kleb-sielleae and Proteeae, overall agreement be-tweenthe

same-day

API20E

procedure

and

the

conventional system was

90.5%.

The

importance of

a

single reaction

in a

bat-tery

of biochemical

tests

is primarily

a

function

of its

relationship

to

other

tests

in

the same

battery. The

most

useful index of comparison

between identification

systems

is

the

identifica-tion of

a

given isolate based

upon a pattern

of

reactions which

may be unique to the system

being used.

For

example, although

the urease

tests

of Micro-ID

and same-day API 20E agreed

with

conventional media with 71 and 61% of

isolates, respectively,

no

misidentifications

oc-curred

solely because of

a

discrepancy with

urease.

Similarly, although citrate in

the

same-day

API 20E was

positive

in only seven in-stances,

its

nonreactivity alone led to no

mis-identifications.

Adequate inocula

were provided by

83%

of

primary

isolates for Micro-ID and 81% for

same-day

API20E.

Stool

isolates

havebeen reported

to have

insufficient

inocula for Micro-ID with

35% of lactose-nonfermenting

isolates (7). In the

present

study,

73% of

lactose-nonfermenting

stool

isolates

were adequate for Micro-ID, and

65%

were

adequate for same-day

API 20E.

Although it

is well

established

as an effective

overnight identification

system for

Enterobac-teriaceae,

effectiveness

of the same-day API 20Eprocedure is compromised by the number of organisms which require additional

overnight

testing,

by inaccurate provisional

identifica-tions,

and by slow response of the telephone

consultation service for unprinted system codes.

ACKNOWLEDGMENTS

Wethankthe staffofthe Clinical Microbiology Laboratories ofPrimaryChildren's Medical Center and the LDS Hospital

fortheirinvaluable assistanceinconductingthisinvestigation. Thisstudywassupported byGeneralDiagnostics,Division

of Wamer-LambertCompany, Morris Plains,N.J.

VOL. 15, 1982

on February 7, 2020 by guest

http://jcm.asm.org/

890 AND HILL ADDENDUMINPROOF

Since thecompletionof thiswork, the identification

manual forMicro-ID has beenrevised (07179; revised January1981).

LITERATURE CITED

1. Aldridge, K.E., B. B. Gardner, S. J. Clark, and J.M. Matsen. 1978. Comparison of Micro-ID, API 20E, and

conventional mediasystemsinidentification of

Enterobac-teriaceae. J. Clin. Microbiol. 7:507-513.

2. Barry,A.L.,and R. E.Badal. 1979.Rapid identificationof

Enterobacteriaceae with the Micro-IDsystemversusAPI 20Eand conventional media. J. Clin. Microbiol. 10:293-298.

3. Edberg,S.C., B. Atkinson, C.Chambers,M. H.Moore,L.

Palumbo, C. F. Zorzon, andJ.M. Singer. 1979. Clinical evaluation of the Micro-ID, API 20E, and conventional

media systems foridentificationofEnterobacteriaceae. J. Clin. Microbiol. 10:161-167.

4. Edberg, S. C., D. Clare, M. H. Moore, andJ.M.Singer. 1979. Rapid identification of Enterobacteriaceae from blood cultures with the ID system. J. Clin. Micro-biol. 10:693-697.

5. Edwards, P. R., and W. H. Ewing. 1972.Identificationof Enterobacteriaceae, 3rd ed. Burgess Publishing Co., Min-neapolis.

6. Finegold, S. M., W. J. Martin, and E. G. Scott. 1978. Bailey andScott'sdiagnosticmicrobiology, 5th ed. C. V. Mosby Co., St. Louis.

7. Gooch, W.M., III. 1980. Evaluation ofamultitest system forrapid identificationofSalmonellaandShigella.Am.J Clin.Pathol. 73:570-573.

8. Martin, W. J., and J. A. WashingtonII.1980. Enterobac-teriaceae, p.195-219. In E. H. Lennette, A. Balows, W. J. Hauser, Jr., and J. P. Truant (ed.), Manual of clinical microbiology,3rded. AmericanSociety forMicrobiology, Washington, D.C.

on February 7, 2020 by guest

http://jcm.asm.org/