CLINICAL MICROBIOLOGY REVIEWS, July 1992, p. 302-327 Vol. 5, No.

3
0893-8512/92/030302-26$02.00/0
Copyright X 1992, American Society for Microbiology

Automated Systems for Identification of Microorganisms

CHARLES E. STAGER' AND JAMES R. DAVIS2*
Department of Pathology, Ben Taub General Hospital, 1 and Department of Pathology,
The Methodist Hospital,2 Baylor College of Medicine, Houston, Texas 77030

INTRODUCTION ........................................................ 302

VITEK ........................................................ 305
SENSITITRE ........................................................ 308
WALKAWAY-96, WALKAWAY-40, AND AUTOSCAN-4 ........................................................ 310
ALADIN AND AUTOREADER ........................................................ 315
BIOLOG ......................................................... 316
MIDI MICROBLIL IDENTIFICATION SYSTEM ......................................................... 318
AUTOSCEPTOR ......................................................... 320
STUDIES COMPARING AUTOMATED IDENTIFICATION SYSTEMS ..........................................321
DISCUSSION ........................................................ 321
ACKNOWLEDGMENTS ........................................................ 324
REFERENCES .........................................................324

INTRODUCTION primary goal was to enhance data acquisition and process-

ing, particularly with regard to decreasing turnaround time.
Clinical microbiology has been an especially dynamic Although the instruments available today are improve-
discipline during the past 10 to 15 years. The exciting ments over the original formulations, they still represent the
developments include the recognition of several new etio- first generation of instruments used to identify microorgan-
logic agents, the reemergence of some classic pathogens, isms. These instruments are widely accepted and very
development of molecular diagnostic tools, and automation helpful; however, like the instruments used in clinical chem-
of antimicrobial susceptibility testing and microbial identifi- istry and hematology laboratories, they will continue to
cation. This article explores the development of automated evolve to better meet the needs of the clinical microbiology
identification systems and reviews their performance. laboratory. If we compare the modem clinical chemistry
To limit confusion, we will avoid terms often used in the analyzer, with its discrete multianalytes requiring no sample
literature such as semiautomated or partially automated. preparation, with instruments available in clinical chemistry
Webster's Dictionary defines automation as the "automati- during the 1960s, we believe we get a glimpse of what the
cally controlled operation of an apparatus, process, or future can be in microbiology. At the very least, we should
system by mechanical or electronic devices that take the target that level of automation for clinical microbiology and
place of human organs of observation, effort, and decision" expect future generations of equipment to be highly auto-
(77a). None of the systems described is totally automated. mated, cost-effective, accurate, reliable, and flexible and to
We use the term "automated" to describe the instruments provide rapid turnaround time.
discussed here and trust the reader to understand that some Among the first automated microbial identification sys-
instruments are more automated than others. tems were the Autobac Series II (formerly called the Auto-
The criteria used for inclusion of an instrument in this bac; Organon-Teknika, Durham, N.C.) and the Avantage
review are as follows. (i) The minimum requirement is Microbiology Center (formerly called the Abbott MS-2) and
automated result entry and identification of microorganisms. Quantum II Microbiology System (Abbott Laboratories Di-
Systems requiring manual result entry are not discussed. (ii) agnostic Division, Irving, Tex.). These systems are no
The instrument must have a data base for the identification
of a large variety of different microorganisms. Instruments longer manufactured but are still in service in some labora-
such as automated enzyme immunoassay systems that iden- tories. As an introduction to the systems used for automated
tify a relatively small number of microorganisms are not identification, we believe that it is appropriate to provide a
described. (iii) The instrument must be available in the brief review of these systems.
United States. The Autobac Series II uses a 19-chambered plastic cuvette
For studies that have compared the identification accuracy and automatically interprets results of biochemical tests.
of two or more automated identification systems, the per- One chamber is a growth control, and the other 18 contain
centile (P value) of the chi-square distribution as determined substrates composed of antibiotics, dyes, or other chemi-
by the chi-square test has been calculated. cals. Common members of the family Enterobacteriaceae
The development of the first generation of automated and six species or groups of nonfermentative and oxidase-
equipment for clinical microbiology involved essentially two positive gram-negative bacilli can be identified by differential
approaches. One can be described as the mechanization of growth inhibition. After off-line incubation of the cuvette for
existing techniques. The second combined mechanization 3 to 6 h, a photometer automatically determines growth
with other changes, such as miniaturization and/or incorpo- inhibition by analyzing the light-scattering index of each
ration of innovative substrates, inhibitors, or indicators. The chamber. A two-stage quadratic discriminant analysis pro-
gram is then used to identify the isolate.
Early studies of the Autobac Series II with currently used
*
Corresponding author. substrates demonstrated that 87.7 to 94.8% of the organisms
302
VOL. 5, 1992 AUTOMATED IDENTIFICATION SYSTEMS 303

tested were correctly identified (10, 11, 18, 27, 60, 106, 111). test results for bacteria and the germ-tube test result for
Commonly isolated organisms such as Escherichia coli, yeasts are manually entered) with a probability matrix.
Kiebsiella pneumoniae, Proteus mirabilis, and Pseudomo- There have been two reports of the Avantage BIC with
nas aeruginosa were correctly identified. However, rela- currently used substrates. Jorgensen et al. (55) reported the
tively common organisms such as Enterobacter cloacae, results of a collaborative evaluation of the Avantage BIC
Citrobacter freundii, Proteus vulgans, Providencia spp., used to identify commonly isolated nonfermentative or oxi-
Salmonella spp., Shigella spp., and Pseudomonas cepacia, dase-positive gram-negative bacilli in 5 h. Conventional
as well as uncommon organisms, were frequently incorrectly biochemicals were used as the reference system. The organ-
identified. In addition, when organisms not represented in isms included in the data base were Acinetobacter anitratus,
the data base were tested, they were frequently given an Acinetobacter iwoffi, Aeromonas hydrophila, Flavobacte-
identification. Only one of these early studies involved a rium meningosepticum IIb group, P. aeruginosa, P. cepa-
comparison of the Autobac Series II with another automated cia, Pseudomonas fluorescens-Pseudomonas putida group,
identification system, the Vitek system (bioMerieux Vitek, X. maltophilia, and Plesiomonas shigelloides. In phase I of
Inc., Hazelwood, Mo.). The Vitek system will be described the study, 200 challenge strains were tested by each of three
later in this paper. In this study, Barry et al. (10) tested 1,510 laboratories. The overall accuracy was 96%, with 95%
members of the family Enterobacteriaceae and nonfermen- correct results for isolates in the data base and 98% for
tative gram-negative bacilli. The Autobac Series II correctly recognition of biotypes not in the data base. Of 200 isolates,
identified 1,443 (95.6%) of the isolates tested, and 1,457 11 either produced a correct identification with low likeli-
(96.5%) were identified by the Vitek system (P > 0.10). hood (probability, <80%) or required tests of oxidative
Truant et al. (120) recently compared the Autobac Series II fermentation of glucose or 10% lactose to separate A.
with the Vitek system and a visually read microtiter system. anitratus from A. iwoffi. In phase II, 100 to 200 routine
They tested 434 clinical isolates whose distribution was clinical isolates or selected stock strains were tested by each
similar to that expected in the clinical setting. If the three laboratory. In phase II, 95% of the isolates were correctly
identified. Only 11 (2.5%) of 437 isolates yielded a low
systems disagreed on identification, the API 20E (Analytab likelihood of identification, and only 4 (4%) of 103 isolates
Products, Plainview, N.Y.) or conventional biochemicals not in the Avantage data base produced misidentifications.
were used as the reference method. The Autobac Series II P. fluorescens-P. putida presented a significant problem,
correctly identified 357 (82.3%) of the isolates. Organisms with only 17 of 21 isolates correctly identified.
with a low relative-probability identification value or uniden- The other report on the Avantage BIC was an abstract by
tified organisms accounted for 32% of the errors. Organisms Snyder et al. (109). Of 342 members of the family Entero-
that posed problems in this group included Citrobacter bacteriaceae and 63 gram-negative other bacteria tested,
diversus, Enterobacter cloacae, E. coli, Proteus mirabilis, 94% were correctly identified by the BIC. The API 20E was
P. aenrginosa, and Serratia marcescens. Misidentification the reference method. Serratia spp. and Enterobacter spp.
of the organism, particularly with Citrobacter spp., Entero- were most frequently responsible for incorrect or inconclu-
bacter spp., and E. coli, accounted for 68% of the errors. sive results with the BIC. The authors did not provide the
The Autobac Series II also misidentified organisms from definition of an inconclusive result.
many different genera as either C. freundii or Enterobacter Early reports on the Quantum II BIC (85, 93, 116) found
agglomerans. There was no discernible pattern to explain that 90.1 to 98.2% of the members of the family Enterobac-
these misidentifications. The earlier studies of the Autobac teriaceae and 83.3 to 93.2% of the nonfermenters and
also reported frequent problems in identification of Citrobac- oxidase-positive fermenters tested were correctly identified.
ter and Enterobacter spp. but not E. coli, as reported by Only small numbers of relatively uncommon clinical organ-
Truant et al. (120). In the study of Truant et al. (120), the isms were tested, and generally these organisms posed the
Autobac Series II correctly identified 82.3% of the tested greatest challenge to the Quantum II BIC. The reactions
isolates, versus 95.6% for the Vitek system (P < 0.001). most frequently responsible for incorrect identifications
The Avantage Microbiology Center and Quantum II Mi- were the lysine, ornithine, esculin, adonitol, inositol, and
crobiology System use a 20-chamber clear-plastic cartridge rhamnose reactions. Only one of these reports was a com-
for identification of aerobic gram-negative bacilli or yeasts. parative study involving another automated system, the
The Abbott Bacterial Identification Cartridge (BIC) and Vitek system (93). In this study, 382 members of the family
Abbott Yeast Identification Cartridge (YIC) each contain 20 Enterobacteriaceae and 119 nonenteric gram-negative bacilli
lyophilized substrates. The Avantage and Quantum II data were tested. The Quantum II BIC correctly identified 375
base includes information for identification of 29 genera or (98.2%) of the organisms belonging to the family Enterobac-
species of the Enterobacteriaceae, 2 Acinetobacter spp., teriaceae and 374 (97.9%) were identified by the Vitek
Xanthomonas maltophilia, 3 Pseudomonas species or system (P > 0.05). The Quantum II BIC correctly identified
groups, 5 other oxidase-positive gram-negative bacilli, 15 111 (93.2%) of the nonenteric organisms tested, and the
Candida spp., 8 Cryptococcus spp., S Rhodotorula spp., and Vitek system identified 115 (96.6%) (P > 0.05). Rhoden and
4 additional yeast genera. After inoculation of the cartridges, O'Hara (98) evaluated an updated Quantum II system, with
baseline turbidimetric or colorimetric readings are obtained changes in the software, an expanded data base, and the
for the biochemical chambers. With the Avantage, these change of some biochemical formulations. The authors
optical readings are simultaneously determined by light- tested 335 isolates, which consisted of 258 members of the
emitting diodes and matched photodetectors, whereas the family Enterobacteriaceae, 55 nonfermenters, and 22 oxi-
Quantum II uses a dual-wavelength spectrophotometer to dase-positive fermenters. No more than 10 strains in each of
read biochemical chambers in sequence. After off-line incu- 65 species were tested. The isolates consisted of both typical
bation at 35 to 37°C for 4 to 6 h (BIC) or at 30°C for 22 to 24 and atypical strains obtained from collections of the Centers
h (YIC), readings are again obtained to determine turbidi- for Disease Control and were not representative of those
metric or colorimetric changes. Isolates are identified by commonly found in the clinical laboratory. The isolates were
comparison of biochemical test results (oxidase and indole identified by conventional biochemical and serologic meth-
304 STAGER AND DAVIS CLIN. MICROBIOL. REV.

ods. The Quantum II BIC correctly identified 92.6% of stocked clinical yeast isolates. The API 20C and conven-
members of the family Enterobacteniaceae, 92.7% of the tional methods served as the reference. The YIC correctly
nonfermenters, and 91% of the oxidase-positive fermenters. identified 86% of the isolates. C. albicans, Candida guillier-
To obtain these levels of correct identification, additional mondii, Candida krusei, C. lusitaniae, C. pseudotropicalis,
serologic (42 isolates), biochemical (34 isolates), or serologic C. glabrata, Cryptococcus terreus, Cryptococcus laurentii,
and biochemical (17 isolates) tests were necessary as indi- Hansenula anomala, Rhodotorula glutinis, Rhodotorula ru-
cated by the Quantum II data base. The distribution of bra, and Saccharomyces cerevisiae were identified with
results was as follows: 184 (54.9%) were identified with a 100% accuracy. Organisms identified with less accuracy
probability of 95% or greater; 51 (15.2%) were identified included Cryptococcus neoformans (95%), C. parapsilosis
within the range of 80 to 94.9%; and 57 (17%) were identified (84%), C. tropicalis (80%), and T. beigelii (86%).
with a probability of less than 80%. There were 25 misiden- Pfaller et al. (94) compared the Quantum II YIC with the
tified organisms. Four nonfermenters (two P. fluorescens Yeast Ident (Analytab Products) and the Vitek systems for
and two P. putida) and two oxidase-positive fermenters
(Plesiomonas shigelloides) grew slowly, and this was be- the identification of 120 common and 101 relatively uncom-
lieved to contribute to their misidentification. Three strains mon clinical yeast isolates. The API 20C was the standard
(one indole-negative E. coli, one Kluyvera cryocrescens, and for comparison. Discrepancies were arbitrated by conven-
one Yersinia enterocolitica) had biochemical profiles not tional methods and morphologic testing. The YIC correctly
recognized by the data base. The remaining 16 misidentified identified 94% of the common isolates and 67% of the
organisms were generally newly recognized genera, and the uncommon isolates. Overall, the YIC correctly identified
misidentifications were due to false-positive or false-nega- 82% of the isolates. Of the 40 misidentifications, 15 occurred
tive test reactions. at the genus level, 21 occurred at the species level, and 4
Cooper et al. (26) published a collaborative evaluation of isolates were unidentified even though they were in the
the MS-2 YIC. In phase I of the study, 91% of 179 stock Quantum II data base. C. parapsilosis, which was not in the
cultures of yeasts were correctly identified. Conventional data base, was most frequently misidentified, followed by C.
methods were used as the reference system. Isolates difficult tropicalis and Geotrichum spp. Multiple false-positive and
to identify included Trichosporon beigelii, Geotrichum spp., false-negative reactions were responsible for most of the
Candidafamata, and Candida humicola. In phase II, 96% of misidentifications. The authors speculated that standardiza-
378 clinical isolates were correctly identified. When there tion of the inoculum with a spectrophotometer or hemacy-
was a discrepancy between the routine laboratory test and tometer might enhance the performance of the system.
the YIC, conventional methods were used for identification. Compared with the Quantum II YIC, the Vitek correctly
Only 2% of those correctly identified had a low likelihood of identified 83% of the isolates (P > 0.05).
correct identification (probability, <80%). Candida lusita- Recent abstracts regarding the Quantum II YIC have
niae, Candida rugosa, and Cryptococcus laurentii were not reported that 91 and 99% of the tested yeasts were correctly
properly identified, although in each case the total number of identified (5, 48).
isolates tested was small. In summary, early studies with the Autobac Series II
The only evaluation of the Avantage YIC was reported by demonstrated that most of the gram-negative bacilli com-
Connelly and Jerris (25). One hundred eighteen yeast iso- monly isolated in the clinical laboratory were correctly
lates grown on Sabouraud dextrose agar with and without identified (10, 11, 18, 27, 60, 106, 111). However, problems
penicillin (20 U/ml) and streptomycin (40 U/ml) were identi- were reported in the identification of such organisms as
fied by the YIC and conventional methods. The identity of Enterobacter cloacae, C. freundii, Proteus vulgaris, Provi-
isolates grown on the nonselective and selective media and dencia spp., Salmonella spp., Shigella spp., P. cepacia, and
identified by the YIC agreed with results obtained from many uncommon organisms. Truant et al. (120) demon-
conventional methods 80 and 83% of the time, respectively. strated additional unexplained problems in identification of
Of 18 isolates grown on Mycosel agar, 17 (94%) were some common organisms such as E. coli, Proteus mirabilis,
correctly identified. All isolates of Candida albicans (N = 7), and P. aeruginosa. The most recent versions of the Avan-
Candida tropicalis (N = 13), Candida glabrata (N = 7), and
Cryptococcus neoformans (N = 15) grown on selective tage BIC correctly identified approximately 95% of the
media were correctly identified by the YIC. gram-negative bacillus strains tested (55, 116), whereas the
Kiehn et al. (61) compared the Quantum II YIC with the most recent version of the Quantum II BIC correctly iden-
API 20C (Analytab Products) and the BBL Minitek (Becton tified approximately 92% of typical and atypical gram-
Dickinson, Towson, Md.) systems for the identification of negative bacillus strains tested (98). Organisms that particu-
245 yeasts. Discrepancies were resolved by conventional larly posed identification problems for the Abbott
tests and morphologic observations on commeal-Tween 80 identification systems included P. fluorescens, P. putida,
agar. The YIC correctly identified 80.4%, misidentified Serratia spp., and uncommon organisms. The MS-2 YIC
7.8%, and did not identify 11.8% of the isolates. Most of the (26), Avantage YIC (25), and Quantum II YIC (61, 94, 102,
misidentifications occurred because of false-positive assim- 104) performed well in the identification of common yeast
ilation reactions. Of the 29 yeasts not identified, 25 were isolates, but less commonly encountered yeasts were fre-
Candida spp., of which 50% were either Candida parapsi- quently misidentified. The number of genera, species, or
losis or Candida pseudotropicalis. groups of gram-negative bacilli or yeasts in the data base of
The Quantum II YIC was evaluated by Salkin et al. (102) the Autobac Series II or Abbott identification systems is
for identification of 239 yeasts. The API 20C was the limited. Consequently, these early identification systems
reference system. The YIC correctly identified 86% of the have had significant problems in identification of uncommon
isolates, but common yeasts (e.g., C. albicans and C. organisms and newly recognized genera.
glabrata) were more readily identified (92% correct) than These first automated identification systems are of histor-
were less frequently encountered yeasts (73% correct). ical importance and can serve as a reference of performance
Sekhon et al. (104) evaluated the Quantum II YIC with 115 level for the systems to be described in this review.
VOL. 5, 1992 AUTOMATED IDENTIFICATION SYSTEMS 305

VITEK 0.5% saline and adjusting the suspension to the equivalent of

a no. 1 (GNI and GPI) or a no. 2 (YBC) McFarland standard.
The Vitek system had its origins in the 1960s, when The inoculum is automatically transferred to the card via a
McDonnell Douglas was contracted by the National Aero- transfer tube during the vacuum cycle of the filling module.
nautics and Space Administration to develop an automated The GNI and GPI are placed in plastic trays, each tray
system to detect and identify pathogens directly from urine holding up to 30 cards. The tray is placed in the reader-
specimens of astronauts in space. This system was subse- incubator at 35°C, and at hourly intervals a digitized analog
quently modified and introduced to the clinical microbiology optical reading, proportional to the light attenuation for each
laboratory as the AutoMicrobic System (AMS) in 1976. The test well, is obtained for each card. The first reading usually
Vitek system, based on bacterial growth in microwells of establishes a baseline value, and the amount of light reduc-
thin plastic cards, could identify nine common urinary tract tion caused by growth or a biochemical reaction in the
pathogens directly from urine specimens and used the most- microwell is determined on subsequent readings. A prede-
probable-number concept to determine the presence of more termined minimum change is required to differentiate be-
than 4.5 x 104 CFU per ml of microbes of urine. Vitek later tween positive and negative reactions. Final identification by
developed additional cards that required pure cultures for the GNI is reported between 4 and 18 h. Most of the
inocula. These 30 microwell cards contained antibiotics or non-glucose-fermenting gram-negative bacilli are reported at
biochemical substrates. Susceptibility test cards are avail- 18 h. Organisms are identified by the GPI between 4 and 15
able for both gram-negative bacilli and gram-positive bacte- h. The YBC is incubated off-line at 30°C for 24 h and then
ria with 11 antimicrobial agents per card. Results are avail- placed in the reader-incubator for a single reading. A mes-
able in 4 to 8 h. Vitek has a variety of standard test kits, and sage "reincubate for 24 h at 30°C" indicates that a definitive
custom-defined test kits can be purchased. Over 40 antimi- identification requires more incubation time. At 48 h, one
crobial agents are currently available on the cards, and must fill in the 48H mark on the card and obtain a second
results from each test include an interpolated MIC, as well as reading. The biochemical test results for all cards are com-
the National Committee for Clinical Laboratory Standards pared with the data base, and the first and second choices, as
categories of susceptible, moderately susceptible, interme- well as their absolute likelihoods and normalized percent
diate, and resistant. Identification cards automatically inter- probabilities, are reported. The biochemical test results, as
preted by the Vitek system are the Gram-Negative Identifi- well as supplemental tests if required, are printed.
cation Test Kit (GNI), the Gram-Positive Identification Test The Vitek system original gram-negative identification
Kit (GPI), and the Yeast Biochemical Test Kit (YBC). card was called the Enterobacteriaceae Biochemical Card
Identification cards that require off-line incubation and man- (EBC) and was designed for the identification of members of
ual entry of the results into the Vitek computer are the the family Enterobacteriaceae within 8 h. The EBC con-
Anaerobe Identification Test Kit, Neisseria/Haemophilus tained 23 conventional biochemical substrates, 2 inhibitors,
Identification Test Kit, and Enteric Pathogen Screen Test and 1 nonconventional substrate. Even though the original
Kit. Vitek system had a limited data base, numerous studies
The GNI and GPI each contain 29 substrates and a growth found the EBC to correctly identify 92 to 99% of the tested
control medium. The GNI substrates include 25 conven- organisms (7, 12, 17, 29, 39, 41, 50, 53, 58). Organisms not
tional biochemical substrates, 3 proprietary substrates, and 1 represented in the data base, uncommon organisms, and
antibiotic. The GNI must be marked if the organism is incorrect reactions for certain biochemicals such as adonitol,
oxidase positive. The GNI data base includes information arginine, citrate, H2S, and malonate were responsible for
for identification of 46 species of members of the family most misidentifications. The lack of an indole reaction was
Enterobactenaceae and 39 species of other gram-negative also considered a weakness of the system. Ferraro et al. (39)
organisms. The GPI substrates include 26 based on conven- found that commonly isolated members of the family Entero-
tional biochemical tests, two antibiotics, and one dye. The bacteriaceae could be presumptively identified by the EBC
GPI must be marked for catalase-negative, beta-hemolytic in 4 h. They found that 97% of the members of the Entero-
organisms or for coagulase-positive organisms that are cat- bacteriaceae isolated in their laboratory belonged to 11
alase positive. The GPI data base includes information for species of six genera. When the EBC was limited to those 11
identification of 23 Streptococcus species, 4 Enterococcus species, 83% of the isolates were correctly identified to
species, 16 Staphylococcus species, and 4 Corynebacterium, genus or species at 4 h. Two percent of the isolates were
Aerococcus, Listenia monocytogenes, and Erysipelothrix misidentified, and 15% were not reported until 8 h.
rhusiopathiae species or groups. The YBC contains 26 Vitek then introduced the EBC+, which, with the addition
substrates which are based on conventional methods. The of acetamide, cetrimide, glucose oxidation, and a supple-
YBC data base includes information for identification of 16 mental oxidase test, was capable of identifying members of
Candida species, 6 Cryptococcus species, 3 Rhodotorula the family Enterobacteriaceae within 8 h and nonfermenta-
species, 2 Tnichosporon species, 3 Geotnichum species, 2 tive and oxidase-positive gram-negative bacilli within 18 h.
Prototheca species, and single species of four additional Studies of the EBC+ demonstrated that 90 to 96% of the
genera. members of the family Enterobacteriaceae and 86 to 97% of
The Vitek system is an integrated modular system consist- the gram-negative organisms that were not members of the
ing of a filling-sealer unit, reader-incubator, computer con- Enterobacteriaceae tested were correctly identified (7-10,
trol module, data terminal, and multicopy printer. The Vitek 44, 56, 125). However, uncommon organisms were fre-
system can be purchased with a capacity of 30, 60, 120, or quently misidentified or had a low percent likelihood of
240 cards and can be interfaced with other computers. A identification. The new substrates added to the EBC+ were
data management center can be added. of limited value in the identification of gram-negative organ-
Inocula for the identification cards are prepared from isms that were not members of the Enterobacteriaceae, and
selective (GNI or GPI) or nonselective (GNI, GPI, and only Aeromonas iwoffi, Aeromonas hydrophila, and P.
YBC) agar media. Inocula for the GNI, GPI, and YBC are aeruginosa were consistently correctly identified (7, 8, 54,
prepared by suspending several colonies in 1.8 ml of 0.45 to 108, 125). Barry and Badal (8) determined the reliability of
306 STAGER AND DAVIS CLIN. MICROBIOL. REV.

the EBC+ for early preliminary identification of isolates. and 11 species). The API 20E and conventional biochemical
They reported that Vitek system identifications were 96% tests were used as the reference system. The GNI correctly
accurate at 4 h if all results with a probability of <80% were identified 96.2% of the members of the Enterobactenaceae
excluded and if Morganella morganii, Acinetobacter spp., and 90.1% of the nonenteric bacilli. Of the 20 misidentifica-
Yersinia spp., Salmonella spp. (other than Salmonella tions, 4 were at the genus level, 4 were at the species level,
typhi), Shigella spp. (other than Shigella sonnei), Enterobac- and 12 organisms were not identified even though they were
ter agglomerans, P. cepacia, the P. putida-P. fluorescens included in the data base. Acinetobacter spp., P. aerugi-
group, or other uncommon genera such as Klebsiella spp. nosa, and Enterobacter spp. were most frequently misiden-
(other than K pneumoniae or Klebsiella oxytoca), Citrobac- tified. The GNI identified 58% of the members of the
ter amalonaticus, Serratia liquefaciens, or Vibno spp. were Enterobacteriaceae at 4 h, 31% at 5 to 8 h, and 11% at 9 to
not reported. If these criteria were adhered to, about half of 13 h. Of the nonenteric isolates, 15% were identified at 4 h,
the isolates in their series could be identified accurately at 4 an additional 45% were identified at 5 to 8 h, and an
h. additional 40% were identified at 9 to 18 h. To determine
The GNI has now replaced the EBC+, and there have interlaboratory agreement, a common set of 134 isolates was
been changes and modifications in the substrates and up- tested at the two participating laboratories. The overall
dates in the data base. Pfaller et al. (93) reported on the agreement between the two laboratories was 92%.
accuracy of the GNI for the identification of 382 members of Recent abstracts involving the GNI have been published.
the family Enterobacteriaceae and 119 nonenteric gram- Colonna et al. (24) tested 358 members of the family Entero-
negative bacilli (representing five genera and nine species). bacteriaceae and 142 nonfermenters with the GNI. The API
The API 20E served as the reference method. The GNI 20E and API NFT (Analytab Products) were used as the
correctly identified 97.9% of the members of the Enterobac- reference systems. They found that 94.7% of the members of
teniaceae and 96.6% of the nonenteric organisms. Only two the Enterobacteriaceae were correctly identified, while 3.6%
members of the Enterobacteriaceae and one of the nonen- were inconclusive, 1.1% were incorrectly identified, and
teric bacteria required additional testing. Of the 12 organ- 0.6% yielded no identification. However, only 79.6% of the
isms misidentified, 2 were misidentified at the genus level nonfermenters were properly identified, whereas 13.4%
and 2 were misidentified at the species level. The remaining were inconclusive, 2.8% were incorrectly identified, and
eight isolates were not identified even though they were in 4.2% yielded no identification.
the data base. Several aberrant biochemical reactions were In 1989, the GNI data base was expanded to include Vibrio
responsible for most misidentifications. Enterobacter spp. alginolyticus, Vibrio damsela, Vbrio fluvialis, and Vibrio
were most frequently misidentified. The GNI identified 35% vulnificus. Farnham et al. (38) evaluated the GNI and this
of the organisms within S h (44% of the members of the expanded data base with 212 members of the family iri-
Enterobacteriaceae and 7% of the nonenteric organisms) onaceae, including 24 A. hydrophila, 27 Plesiomonas
and 63% within 6 h (75% of the members of the Enterobac- shigelloides, 35 V. alginolyticus, 37 Vibrio cholerae, 12 V.
tenaceae and 25% of the nonenteric organisms). damsela, 18 V. fluvialis, and 24 Vibio parahaemolyticus
Truant et al. (120) reported on the performance of the isolates and 35 Vibrio spp. Reference systems were the API
GNI, testing 434 clinical isolates whose distribution was 20E and conventional biochemicals. The overall correlation
similar to that expected in the clinical setting. The GNI of the GNI with the reference identification was 95%. Of the
correctly identified 95.6% of the isolates with a likelihood of new species in the data base, 90% of V. alginolyticus, 97% of
.90%. However, many uncommon members of the family V. vulnificus, 100% of V. damsela, and 100% of V. fluvialis
Enterobacteriaceae were not evaluated, and only three isolates were correctly identified.
genera and seven species of nonenteric bacilli were in the There have been limited studies of the GPI for identifica-
series. Enterobacter cloacae was the only problem organ- tion of Staphylococcus spp. Three studies (2, 45, 101) were
ism, with 8 of 41 isolates yielding either a low probability or conducted soon after the GPI was introduced, at which time
being unidentified. This was thought to be due to the there were 27 substrates and only seven Staphylococcus
premature reading of the arginine, lysine, and omithine species in the data base. Staphylococcus aureus, Staphylo-
reactions. coccus epidermidis, and Staphylococcus saprophyticus were
Plorde et al. (96) evaluated the GNI for identification of correctly identified in the range of 88 to 100%. However, the
419 non-glucose-fermenting gram-negative bacilli represent- accuracy for the remaining four species in the data base
ing 14 genera and 35 species. The isolates were identified by ranged from 0 to 71%. In addition, 47 to 65% of the
conventional biochemicals. Of the 419 tested organisms, 63 misidentified species were incorrectly identified as other
belonged to species not included in the data base. Fifty-eight Staphylococcus species. A recent study of the GPI involved
(92%) of these isolates were appropriately reported as uni- 130 strains representing 14 species of staphylococci of bo-
dentified. Of the 356 test organisms included in the data vine origin (72). The isolates were identified by conventional
base, 307 (86.2%) were correctly identified, 36 (10.1%) were methods. Eighty-five of the strains (representing 12 species)
not identified, and 13 (3.7%) were misidentified. If the were in the Vitek system data base, and 58 (68%) were
first-choice identification had been accepted as correct, 18 correctly identified. Forty-seven of these were identified at
(4.3%) isolates would have been erroneously reported. the .90% level of confidence. The GPI correctly identified
When first-choice identifications were rejected if the report 100% of the S. aureus and S. epidermidis, 91% of the
indicated "questionable biopattern" and when all isolates Staphylococcus xylosus, and 80% of the Staphylococcus
were screened for characteristic odor and an appropriate simulans isolates. Staphylococcus chromogenes (N = 43)
antimicrobial profile before acceptance of the report, misi- was not in the data base; 39 strains were misidentified (27 as
dentifications were reduced to 1.2%. The average time to S. simulans), and 4 strains were unidentified. A recent
identification was 15 h. abstract (46) compared species identification of the coagu-
The most recent study of the GNI was reported by Pfaller lase-negative staphylococci by the GPI with species identi-
et al. (95). They tested 292 members of the family Entero- fication by the methods of Kloos and Schleifer (64). The GPI
bactenaceae and 91 nonenteric bacilli (including 8 genera correctly identified 148 of 149 S. epidennidis, 37 of 39
VOL. 5, 1992 AUTOMATED IDENTIFICATION SYSTEMS 307

Staphylococcus haemolyticus, and 20 of 27 other coagulase- ical characteristics were used. The manufacturer recom-
negative staphylococcal strains. mended incubation of the YBC for 24 h. Oblack et al. (88)
Streptococcus pyogenes and species of group D strepto- recorded reactions at 24, 48, and 72 h for 236 isolates. Eleven
cocci and enterococci were accurately identified with a high isolates were misidentified at 24 h, but 7 of these were
level of confidence (3, 36, 101) by the GPI. The accuracy of correctly identified at 48 h. The other four isolates were not
identification of Streptococcus agalactiae varied from 85.7 correctly identified even at 72 h. However, seven isolates
to 100%. However, only 61.9 to 85.7% of these strains were correctly identified at 24 h were misidentified at 48 or 72 h.
identified with a high level of confidence. Ruoff et al. (101) Pfaller et al. (94) reported on the YBC for the identification
and Appelbaum et al. (3) found that the GPI correctly of 120 common clinical yeast isolates and 101 less common
identified 36.3 to 76.2% of the beta-hemolytic group C, F, clinical yeast isolates, with the API 20C as the reference.
and G streptococci. Ruoff et al. (101) reported excellent The YBC correctly identified 103 (86%) of the common
results for Streptococcus pneumoniae with the GPI. How- isolates and 81 (80%) of the uncommon isolates, for an 83%
ever, two subsequent studies identified 80.9 to 84.4% of overall accuracy. The YBC misidentified 37 isolates (13 at
Streptococcus pneumoniae isolates, and only 76.6 to 76.9% the genus level and 18 at the species level), and 6 isolates
of the results were at a high level of confidence (3, 36). Four were not identified even though they were in the Vitek data
studies of the GPI for identification of the viridans group base. Several isolates, including Candida humicola, Can-
showed an accuracy of 57.2 to 79%, and many of the strains dida lambica, C. lusitaniae, Candida paratropicalis, and
were identified at a low level of confidence (3, 36, 92, 101). Prototheca wickerhamii, were not in the Vitek data base and
There have been single studies of the GPI for identification were incorrectly identified. Five yeasts (C. tropicalis, C.
of Listeria and Aerococcus spp. Ruoff et al. (101) reported parapsilosis, C. lambica, C. paratropicalis, and C. glabrata)
that 30 of 30 L. monocytogenes isolates were correctly accounted for 62% of the YBC misidentifications. Multiple
identified, all with a likelihood of between 90 and 100%. false-positive and false-negative biochemical tests were re-
Facklam et al. (36) found that 24 (92.3%) of 26 Aerococcus sponsible for the misidentifications.
spp. were correctly identified, 84.6% with a high level of Vitek expanded the YBC data base in 1989 and allowed for
confidence. There are no reported studies of the GPI for a 48-h incubation when necessary. El-Zaatari et al. (34)
identification of Corynebacterium spp. or Erysipelothrix evaluated this data base with 398 clinical isolates, including
rhusiopathiae. 9 genera and 26 species of yeasts and yeastlike fungi, the
Recent abstracts describing the GPI used for viridans hyphomycete Geotrichum candidum, and the achlorophyl-
streptococci include one by Schiminsky and Ferrieri (103), lous alga P. wickerhamii. The API 20C was the reference
who tested 84 viridans streptococci, 56 enterococci, and 3 method. The YBC identified 243 (99.2%) of 245 of the
group D streptococci. Isolates were identified by conven- common isolates (C. albicans, C. parapsilosis, C. tropicalis,
tional biochemical tests. Only one enterococcus was improp- C. glabrata, and Cryptococcus neoformans) and 144 (94.1%)
erly identified to species level. Streptococcus sanguis II, of 153 of the uncommon isolates. The overall accuracy was
Streptococcus sanguis I, Streptococcus mitis, and Strepto- 97.2%. Of the 387 correctly identified isolates, 282 (70.8%)
coccus salivarius were most frequently isolated, and 79% of were identified at 24 h, 105 (26.4%) were identified at 48 h, 67
these isolates were correctly identified. Overall, 63 (75%) of (17.8%) required additional morphological evaluation, and
the 84 viridans streptococci were properly identified. 15 (3.8%) required both additional biochemical tests and
Adridge et al. (1) identified 222 clinical isolates of viridans morphological studies. They concluded that the YBC with
streptococci from blood and odontogenic infections by using an updated and expanded data base was markedly improved
the GPI without supplemental biochemical tests. Conven- and provided a rapid and reliable identification of yeasts.
tional methods served as the reference system. The GPI Recent abstracts confirm that the new YBC data base
correctly identified 46.1% of the isolates, with 13% yielding provides accurate identification of clinical yeast isolates.
no identification. Hinnebusch et al. (51) tested 186 viridans Forsythe et al. (40) evaluated the YBC with 34 stock strains
streptococcal isolates (150 blood culture isolates and 36 and 144 clinical yeast isolates. The standard for comparison
stock strains) by using the GPI and conventional biochemi- was the API 20C. The overall agreement between systems
cals. The GPI properly identified 60.2% of the isolates was 97%. The agreement was 99.5% when only clinical
without supplemental tests. With supplemental tests, 80.6% isolates were examined. The YBC was evaluated by Yamane
were correctly identified. If three rarely encountered species et al. (127), who used 292 strains (26 species) of yeasts. The
(Streptococcus morbillorum, Streptococcus uberis, and API 20C, and morphological examination when necessary,
Streptococcus acidominimus) were excluded, the GPI cor- was the standard for comparison. Excellent agreement was
rectly identified 79.7% of the remaining isolates without obtained for C. albicans, C. tropicalis, C. parapsilosis, C.
supplemental tests. glabrata, S. cerevisiae, and Cryptococcus neoformans.
Early studies of the YBC demonstrated 84.9 to 96% Overall, 95.2% of the isolates, including those that required
accuracy of identification of yeasts (49, 66, 88). Generally, supplemental tests, were correctly identified. Of the less
common clinical isolates such as C. albicans, C. tropicalis, common species, 91.8% were correctly identified. Quinn and
C. pseudotropicalis, C. parapsilosis, C. glabrata, Crypto- Horstmeier (97) tested the YBC with 368 commonly encoun-
coccus neoformans, Cryptococcus albidus, R. rubra, and S. tered clinical yeast isolates. The standard method of identi-
cerevisiae were readily identified with a high level of confi- fication was not mentioned. The YBC correctly identified
dence. However, uncommon clinical isolates posed more of 93.5% of the isolates. Supplemental testing or morphological
a problem for the system. Hasyn and Buckley (49) identified examination was required for 13.8% of the isolates.
84.9% of 352 freshly isolated, germ tube-negative yeasts with McWhirter et al. (77) tested the YBC with 105 germ tube-
the YBC, but 10.7% of these required microscopic examina- negative clinical yeast isolates. Conventional methods were
tion of inoculated cornmeal-agar medium for morphological used as the standard. The YBC correctly identified 86% of
characteristics. Oblack et al. (88) reported an accuracy of the isolates in 24 h. After supplemental tests were per-
84% for 253 clinical yeast isolates when YBC biochemical formed, 98% of the isolates were correctly identified. Sim-
data were used, but this improved to 96% when morpholog- monds et al. (107) reported on the YBC for identification of
308 STAGER AND DAVIS CLIN. MICROBIOL. REV.

TABLE 1. Performance of automated identification systems for identification of members of the family Enterobactenaceae
No. of % of identifications that were:
System organisms Reference
tested Correct Inconclusive Not made Incorrect
Vitek GNI 382 97.4 0.5a 1.6 0.5 93
Vitek GNI 339 96.2 _b _C -C 120
Vitek GNI 292 94.5 1.7d 1.7 2.1 95
Vitek GNI 358 94.7 3.6e 0.6 1.1 24
Sensititre AP80 358 93.3 0.3e 0 6.4 24
Sensititre AP80 180 78.9 f _c _c 70
autoSCAN-4 GNB 307 84.0 11.49 2.0 2.6 99
autoSCAN-4 GNB 372 95.4 2.79 0 1.9 42
WalkAway-96 R-GNB 292 86.3 8.99 0.3 4.5 95
WalkAway-96 R-GNB 358 83.0 7.8e 0 9.2 24
WalkAway-96 R-GNB 180 86.1 f _c _- 70
Biolog GN MicroPlate 212 52.0" 2.0 38.0 8.0 79
60.0' 10.2 21.3 8.5
Percent correct identification with first-choice likelihood of <80%. Supplemental tests were required to confirm correct identification.
bIdentifications with first-choice likelihood of <90% were considered inconclusive; the percent inconclusive identifications not stated.
c Performance could not be determined from data presented.
d Percent correct identification with first-choice likelihood of <90%.
e Identification likelihood not stated; percent inconclusive identifications that were correct identifications not stated.
f Identification likelihood not stated; percent inconclusive identifications not stated.
g Percent correct identification with first-choice likelihood of <85%.
h Correct identification to species level at 4 h.
Correct identification to species level at 24 h.

232 clinical yeast isolates from 10 genera and 29 species. The The Vitek system is a comprehensive microbial identifi-
API 20C was used as the standard method, and discrepan- cation system that needs further improvement in its identi-
cies were resolved with conventional assimilation and fer- fication accuracy for some Staphylococcus and Streptococ-
mentation tests. The overall accuracy of the YBC was cus species, as well as for unusual and uncommon bacteria
85.3%. A 48-h reading of the YBC was required for 47.8% of and yeasts. In addition, a shortened time for identification of
the isolates. When biotypes not found in the data base were non-glucose-fermenting gram-negative bacilli, gram-positive
excluded, the accuracy was 94%. Morphological observa- bacteria, and yeasts would be desirable. On-line incubation
tions were needed to identify >50% of the isolates, and of the YBC, Anaerobe Identification Test Kit, Neisseria/
confirmatory biochemical tests were required for identifica- Haemophilus Identification Test Kit, and Enteric Pathogen
tion of 8% of the isolates by the YBC. Screen Test Kit cards would further enhance the utility of
As will be true for all of the automated identification the Vitek system.
systems considered, Vitek Systems has continued to im- Tables 1, 2, 3, and 4 demonstrate the identification accu-
prove the capability of its systems to accurately identify racy for members of the family Enterobacteriaceae, for
microorganisms. Consequently, when attempting to deter- strains of other gram-negative organisms, for Staphylococ-
mine the performance of the Vitek system, one should cus strains, and for strains of yeasts and yeastlike organisms,
consider the publications that have evaluated or compared respectively, for each of the automated identification sys-
the latest software, data base, biochemical configuration, or tems that can currently be purchased.
other performance characteristics of their system. In consid-
ering recent studies of the GNI, members of the family SENSMTTRE
Enterobactenaceae were correctly identified in the range of
94.7 to 96.2%, with Enterobacter spp. the most frequently The Sensititre fluorogenic system (Radiometer America,
misidentified (24, 95, 120). Gram-negative organisms that Inc., Westlake, Ohio) is a modular system composed of a
were not members of the Enterobactenaceae were correctly computer and an automated reader. This system identifies
identified in the range of 79.6 to 95%, with P. aeruginosa and gram-negative bacilli and performs susceptibility tests on
Acinetobacter spp. the most commonly misidentified (24, 38, both gram-positive and gram-negative bacteria in either 5 or
95, 120). S. aureus, S. epidermidis, S. saprophyticus, S. 18 h. Breakpoint or MIC capability for gram-negative and
xylosus, and S. haemolyticus have been correctly identified gram-positive bacteria is available for 54 antimicrobial
by the GPI in the range of 88 to 100%, but less commonly agents. The Sensititre AP80 panel data base includes infor-
encountered species have frequently been misidentified or mation for the identification of 84 members of the family
identified with a low likelihood (2, 45, 46, 72, 101). Strepto- Enterobacteriaceae, 24 oxidase-positive fermenters, 16
coccus pyogenes and species of group D streptococci and pseudomonads, and 16 other nonfermenters.
enterococci have been accurately identified with a high level The automated Sensititre system is composed of a Sensi-
of confidence by the GPI, but other Streptococcus species titre Fluorescence AutoReader, a Digital PRO 380 computer
have posed more of a problem for the system (1, 3, 36, 51, with 768K of main memory, 20 or 30 MB hard-disk and dual
92, 101, 103). In recent publications, the YBC correctly floppy-disk drives, data terminal, and multicopy printer. A
identified 93.5 to 98% of tested yeasts and yeastlike isolates, Sensititre Automatic Inoculator is also available. The Sen-
when organisms not represented in the data base were sititre software has menus for demographics, interpretation
excluded and when required supplemental tests or morpho- of culture results, data analysis, report generation, and
logical examinations were performed (34, 40, 77, 97, 107, epidemiological evaluations.
127). The Sensititre AP80 gram-negative identification test plate
VOL. 5, 1992 AUTOMATED IDENTIFICATION SYSTEMS 309

TABLE 2. Performance of automated identification systems for identification of strains of gram-negative organisms that were not
members of the Enterobacteriaceae
No. of % of identifications that were:
System organisms Reference
tested Correct Inconclusive Not made Incorrect
Vitek GNI 119 95.8 0.8a 1.7 1.7 93
Vitek GNI 95 93.7 _b -C -C 120
Vitek GNI 356 74.7 11.5d 10.1 3.7 96
Vitek GNI 91 86.8 3.3d 7.7 2.2 95
Vitek GNI 142 79.6 13.4e 4.2 2.8 24
Vitek GNI 212 95.0 -f c _c 38
Sensititre AP80 142 71.1 5.6e 7.0 16.2 24
Sensititre AP80 60 35.0 f -c -c 70
autoSCAN-4 GNB 98 58.2 36.79 4.1 1.0 99
WalkAway-96 R-GNB 91 92.3 3.39 0 4.4 95
WalkAway-96 GNB 310 41.3 21.39 15.0 22.4 117
WalkAway-96 R-GNB 310 50.2 40.5 0 9.3 117
WalkAway-96 R-GNB 142 74.6 14.1w 0 11.3 24
WalkAway-96 R-GNB 60 81.7 -f c -c 70
Biolog GN MicroPlate 140 38.6h 12.8 33.6 15.0 79
55.8' 12.6 20.8 10.8
a
Percent correct identification with first-choice likelihood of <80%. Supplemental tests were required to confirm the correct identification.
b
Identifications with first-choice likelihood of <90% were considered inconclusive; percent inconclusive identifications not stated.
c Performance could not be determined from data presented.
dPercent correct identification with first-choice likelihood of <90%.
eIdentification likelihood not stated; percent inconclusive identifications that were correct identifications not stated.
f Identification likelihood not stated; percent inconclusive identifications not stated.
g Percent correct identification with first-choice likelihood of <85%.
h Correct identification to species level at 4 h.
Correct identification to species level at 24 h.

will test three separate organisms, with each section con- cence. Therefore, absence of fluorescence indicates a posi-
taining 32 dried, fluorescently labeled substrates. Substrates tive test. In other cases (carbon assimilation reactions,
are linked to a fluorophore (4-methylumbelliferone or 7-ami- decarboxylase reactions, and urea hydrolysis), the pH is
no-4-methylcoumarin). Some of the substrates (peptidase, increased, thus producing fluorescence.
pyranosidase, phosphatase, and glucuronidase) do not di- Inocula are prepared, using a built-in nephelometer in the
rectly fluoresce. However, when enzymes release the fluo- Automatic Inoculator, by transferring one or more colonies
rophore, it will fluoresce. The fluorogenic reaction occurs from either nonselective or selective agar media to sterile
when UV light is absorbed and electrons are raised to higher distilled water to equal a no. 0.5 McFarland standard. The
energy levels. When these electrons return to their original nephelometer is calibrated to be linear over the range of no.
states, photons with lower energies and hence longer wave- 0.1 to 0.9 McFarland standard. The Automatic Inoculator
lengths are emitted. Fluorometric reactions, which detect dispenses 50 ,ul of inoculum into each test well and then
pH changes, are also used. In one case, the pH of the dispenses 150 ,ul of oil into the urea and a proprietary
substrate is originally alkaline and the fluorophore is fluo- substrate microwell. A transparent adhesive seal with per-
rescent. With oxidation or fermentation of carbohydrates, forations over the wells containing malonate, pyruvate,
acid production lowers the pH and quenches the fluores- citrate, and agmatine is applied to the plate. The test plate is

TABLE 3. Performance of automated identification systems for identification of Staphylococcus species

No. of % of identifications that were:
System organisms Reference
tested Correct Inconclusive Not made Incorrect
Vitek GPI 150 67.3a 0 0 32.7 2
Vitek GPI 190 83.2b 6.8c 3.7 6.3 45
Vitek GPI 148 89.9 4.1c 0.7 5.3 101
Vitek GPI 85d 68.2e 0 11.8 20.0 72
130d"f 44.6 0 11.6 43.8
Vitek GPI 215 95.3 0 1.9 2.8 46
autoSCAN-4 GPB 175 83.4 -9 -g 52
WalkAway-96 R-GPB 239 91.6 3.8 2.1 2.5 114
a
Percent correct identification regardless of first-choice likelihood percentage.
b Percent correct identification with required supplemental physiologic tests.
c Identified as Staphylococcus spp.
d Staphylococcal strains of bovine origin.
e
Percent correct identification for staphylococcal species present in Vitek data base.
f Forty-five staphylococcal strains were from species not contained in the Vitek data base.
g Performance could not be determined from available data.
310 STAGER AND DAVIS CLIN. MICROBIOL. REV.

TABLE 4. Performance of automated identification systems for ported on Sensititre AP80, using the MicroScan Negative
identification of yeasts and yeastlike organisms Combo type 7 panel for identification of gram-negative rods
as the reference system. Discrepant results were resolved
~No. of % Correctly
with either the API 20E or the API NFT. For enteric
System organisms identified Reference
tested gram-negative rods, 79% (142 of 180 strains) were correctly
identified to species, whereas for nonfermentative gram-
Vitek YBC 352 84.9a 49 negative rods, only 35% (21 of 60 strains) were correctly
Vitek YBC 1,106 93.4a 66 identified to species. Incorrect identifications occurred with
Vitek YBC 253 95.7a 88
Vitek YBC 120b 85.8c,d 94 7.9% (19 of 240) of the strains tested. Weckbach et al. (122)
Vitek YBC iole 80 2C,d 94 reported a clinical trial of the Sensititre AP80 identification
Vitek YBC 245b 99.2a 34 system at three study centers. Additional data were given at
Vitek YBC 153e 94.la 34 the time of presentation of the paper. The expanded data
Vitek YBC 178 97.0f 40 (110) included 879 members of the family Enterobac-
Vitek YBC 292 95.2a 127 teriaceae, 100 Pseudomonas spp., 33 Acinetobacter spp.,
Vitek YBC 368 93.5a 97 and 11 other gram-negative bacilli (including Aeromonas,
Vitek YBC 105 98.la 77 Plesiomonas, and Bordetella spp.). Results for the AP80
Vitek YBC 232 94.0a 107
autoSCAN-4 YIP 232 96.6a 107 were compared with results for the API 20E at S and 18 h
autoSCAN-4 YIP 357 62e 112 independently. Discrepant results were arbitrated by the
autoSCAN-4 YIP 189 79d 14 Centers for Disease Control with conventional biochemicals.
At 5 h, 95.9% of the isolates were identified to genus, while
a
Correct identification with required microscopic morphological examina- 92.4% had an acceptable species match. A total genus match
tions or supplemental physiologic tests.
b Common
yeast isolates. for isolates tested at 18 h was 98.4%, and an acceptable
c Percent correct identification with first-choice likelihood of 280%. species match was 96.3%. For the AP80, the percentage of
d No supplemental microscopic morphological examinations or physiologic isolates that required additional rapid tests at 5 h was 3.5
tests performed. versus 2.0% at 18 h. The incidence of additional tests
e Uncommon yeast isolates.
f Identification likelihood not stated; requirements for microscopic morpho- required by the API 20E was 5.6%. Even though the glucose
logical examinations or supplemental physiologic tests not stated. nonfermenters represented only 13% of the isolates tested,
33% required reincubation after the 5-h reading.
Limitations of the Sensititre system include the require-
ment for off-line incubation of AP80 panels, the necessity of
incubated off-line at 35 to 37°C and may be read on the performing additional rapid tests for some isolates, and the
Sensititre Fluorescence AutoReader after 5 or 18 h of necessity for reincubating some AP80 panels. In addition,
incubation. the development of systems for the rapid identification of
Reagent addition is not required with the AP80 identifica- gram-positive bacteria, anaerobes, and yeasts would expand
tion panel. The test plate is placed into the AutoReader, the utility of the Sensititre fluorogenic system. Radiometer
where it is automatically transported to the reader station. America will introduce the ARIS (Automated Reading and
The light source is a broad-band xenon flash lamp (360 nm) Incubation System) to the United States in 1992. ARIS will
that generates microsecond pulses of high-peak-power light. provide "walk-away" automation and plate reading technol-
The light goes through interference filters and a beam- ogy based on the Sensititre AutoReader.
splitting cube with wavelength-selective coatings to lenses The limited results available with the Sensititre AP80
that focus the light onto the test well and the detector. The identification system are very encouraging. However, fur-
detector is a photomultiplier tube that transmits the raw ther evaluation of this identification system, particularly in
fluorescence data to the DEC PRO 380 computer. It requires defining its accuracy for both common and uncommon
approximately 30 s to read a plate. The biocode generated is organisms at both 5 and 18 h, is needed.
matched to the Sensititre data base. There may be additional
test prompts for indole, oxidase, motility, or pigment pro- WALKAWAY-96, WALKAWAY-40, AND AUTOSCAN-4
duction. A reincubate prompt also occurs at 5 h if there is no
identification or a low probability of identification. The result The WalkAway-96 (formerly called the autoSCAN-W/A)
of each test is printed, and for each organism listed, the and WalkAway-40 (Baxter Diagnostics, Inc., MicroScan
quality of the identification (excellent, good, group identifi- Division, West Sacramento, Calif.) are computer-controlled
cation, etc.) is determined on the basis of calculated proba- systems that will incubate microtiter panels and automati-
bility values. Test results against any identified organism(s) cally interpret biochemical or susceptibility results with
are also listed. either a photometric or a fluorogenic reader. MicroScan also
There have been three recent studies of the Sensititre manufactures the autoSCAN-4, which requires off-line incu-
AP80 identification panel. Colonna et al. (24) tested 358 bation and, with the exception of fluorogenic panels, will test
members of the family Enterobacteriaceae and 142 nonfer- the same panels as the other two automated systems. All
menters isolated from clinical specimens. The API 20E and three systems perform susceptibility tests on aerobic gram-
API NFT were used as the reference systems. Results for negative bacilli, gram-positive bacteria, and anaerobes. Flu-
the members of the Enterobacteriaceae were 93.3% correct, orogenic panels provides a 3.5- to 7-h susceptibility result for
6.4% incorrect, and 0.3% inconclusive, and results for the aerobic gram-negative bacilli and a 3.5- to 15-h susceptibility
nonfermenters were 71.1% correct, 16.2% incorrect, 5.6% result for aerobic gram-positive bacteria. Conventional pan-
inconclusive, and 7.0% unidentified. Because of the limita- els provides a 15- to 24-h susceptibility result for aerobic
tion inherent in an abstract, it was not stated whether bacteria and a 24- to 48-h susceptibility result for anaerobes.
isolates were correctly identified to genus or to species. MicroScan has numerous panel types, including both MIC
Also, it was not stated whether these data applied to the 5- or and breakpoint susceptibility panels. Custom panels are also
18-h identification with the AP80. Lyznicki et al. (70) re- available. The three systems automatically identify gram-
VOL. 5, 1992 AUTOMATED IDENTIFICATION SYSTEMS 311

negative bacilli, gram-positive bacteria, fastidious aerobic The inoculum for the AIP must be prepared from nonse-
bacteria, anaerobes, and yeasts. lective media, whereas the inoculum for all other MicroScan
The rapid fluorogenic panels for identification of gram- identification panels can be prepared from selective media.
negative bacilli (R-GNB) and gram-positive bacteria (R- Media containing blood, yeast extract-malt extract agar, and
GPB) use fluorogenic substrates (4-methylumbelliferone or inhibitory mold agar should not be used for yeasts. Several
7-amino-4-methylcoumarin attached to a phosphate, sugar morphologically similar colonies are picked and suspended
moiety, or amino acid) or fluorometric indicators. Identifi- in either sterile distilled water (GNB, GPB, AIP, and YIP),
cation is based on hydrolysis of fluorogenic substrates, pH sterile 0.4% saline (R-GNB and R-GPB), or MicroScan
changes following substrate utilization, production of spe- HNID Inoculum Broth (HNID). The suspension is adjusted
cific metabolic by-products, or the rate of production of to the equivalent of a no. 0.5 McFarland turbidity standard
specific metabolic by-products after 2 h of incubation. The (GNB, R-GNB, GPB, and R-GPB), a no. 3 McFarland
modified conventional panel for identification of gram-nega- standard (HNID and AIP), or a no. 5 McFarland standard
tive bacilli (GNB) has 29 modified conventional or chro- (YIP). The inoculum is further diluted in sterile water for the
mogenic tests and six antibiotics, and results are available in GNB and GPB and in MicroScan Inoculum Broth for the
15 to 42 h. Both the R-GNB and GNB data bases include R-GNB and R-GPB. The inoculum is poured into a seed
information for identification of 59 groups, genera, or species trough and covered with a transfer lid. The RENOK rehy-
of members of the Enterobacteriaceae and 57 groups, gen- drator/inoculator (Baxter Diagnostics, Inc., MicroScan Di-
era, or species of nonfermentative and oxidase-positive vision), a manual pipettor that simultaneously rehydrates
gram-negative bacilli. The modified conventional panel for and inoculates MicroScan conventional panels or designated
identification of gram-positive bacteria (GPB) has 25 modi- rapid panels, is attached to the transfer lid. A lever is lifted
fied conventional or chromogenic substrates, one dye, and on the RENOK, creating a vacuum; as a result, inoculum is
three antibiotics and yields an identification in 15 to 42 h. drawn into the inoculator transfer lid. The inoculum-filled
Both the R-GPB and GPB data bases include information for transfer lid is then positioned on the GNB, R-GNB, GPR, or
identification of 24 genera or species of the family Micrococ- R-GPR, and with the depression of a release button on the
caceae, 18 members of the family Streptococcaceae, 4 RENOK, 115 + 10 ,ul of inoculum is dispensed into each
Enterococcus spp., Aerococcus viridans, and L. monocyto- microwell. For the HNID, AIP, and YIP, each microwell is
genes. The MicroScan Rapid Haemophilus and Neisseria manually filled with 50 ,ul of inoculum with a manual
Identification Panel (HNID) has 17 modified conventional or pipetting device.
chromogenic tests and one antibiotic, and the data base Before incubation of the panels, a mineral oil overlay must
includes information for identification of Haemophilus influ- be added to the lysine, omithine, and decarboxylase base
enzae (seven biotypes), Haemophilus parainfluenzae (four wells of the R-GNB; to the glucose, urea, H2S, lysine,
biotypes), Haemophilus aphrophilus-Haemophilus paraph- arginine, ornithine, and decarboxylase base wells of the
rophilus, Haemophilus haemolyticus, four Neisseria spe- GNB; to the arginine and urea wells of the GPB; to the urea
cies, Moraxella catarrhalis, and Gardnerella vaginalis in 4 and indole wells of the AIP; and to the urea well of the
h. The MicroScan Rapid Anaerobe Identification Panel HNID. A sealing strip is placed over the citrate, malonate,
(AIP) has 24 modified conventional or chromogenic sub- o-nitrophenyl-3-D-galactopyranoside, tartrate, acetamide,
strates, and the data base includes information for identifi- cetrimide, oxidation-fermentation glucose, oxidation-fer-
cation of 21 anaerobic gram-negative bacilli, 13 anaerobic mentation base, and decarboxylase base wells when the
non-spore-forming gram-positive bacilli, 8 anaerobic gram- GNB is used with oxidase-positive organisms.
positive cocci, and 16 clostridia in 4 h without anaerobic A bar code label for patient identification is printed by the
incubation. The MicroScan Rapid Yeast Identification Panel DMS and attached to the side of the panel by the operator.
(YIP) has 27 modified conventional or chromogenic sub- The WalkAway-96 will process up to 96 panels, and the
strates, and the data base includes information for identifi- WalkAway-40 will process up to 40 panels. Any combination
cation of 16 Candida spp. or biotypes, 8 Cryptococcus spp., of fluorogenic or conventional panels can be randomly
and 11 genera (representing 16 species) of yeasts or yeastlike loaded into the tower shelves of the carousel. After the
organisms in 4 h. incubator door is closed, the bar code reader scans each
The WalkAway-96 and WalkAway-40 are composed of an panel in the tower and the computer then schedules baseline
incubator, ultrasonic humidifier, carousel holding towers, readings, incubation, reagent addition when required, and
bar code reader, spectrophotometer, fluorometric reader, readings for each panel. Currently, up to 12 reagents can be
reagent-dispensing subsystem, panel-accessing mechanism, dispensed. For the fluorogenic reader, the light source is a
and computer for panel scheduling and tracking. The auto- quartz-halogen incandescent bulb filtered so that UV light of
SCAN-4 is a colorimetric optical system that performs 370 nm is focused into each of the 96 wells. Any free
similarly to the spectrophotometer in the WalkAway-96 and fluorophore present in the wells will then emit light at 450
WalkAway-40. Panels are processed one at a time after nm. Nonfluorophore wavelengths are filtered from this light,
off-line incubation. The technologist communicates with and the light is focused onto a detector. Normal instrument
these systems from the keyboard and monitor of an IBM variations in panel readings are compensated for by a
Personal Computer AT or IBM Personal System/2 com- fluorescing reference disk. When conventional panels are
puter, model 60 or 80, via MicroScan Data Management read on the WalkAway systems, a tungsten-halogen light
System (DMS) software. The DMS produces chartable re- source passes through interference filters on a color wheel
ports, epidemiologic reports, work logs, and antibiograms. and is focused on 97 optical fibers, 96 of which will indepen-
The results can be displayed on a monitor or printed. A dently illuminate a separate well in the 96-well panel. The
Pharm-Link program, which allows the pharmacy to interact 97th photometer yields a baseline reading with which all
with the MicroScan system, can be added. The DMS will other photometer readings are compared. A color wheel
support two MicroScan instruments and multiple worksta- rotates during each read cycle and, depending on the optimal
tions. One- or two-way interfaces with other computer light source for each substrate, provides a reading at one of
systems are available. six different wavelengths. The data are transferred to the
312 STAGER AND DAVIS CLIN. MICROBIOL. REV.

IBM computer, where the results are converted to a biotype cation. The R-GNB identified 86.3% of the members of the
number and compared with probabilities in the DMS data Enterobacteriaceae and 92.3% of the nonenteric bacilli.
base. All possible identifications are printed in descending Overall, 87.7% of all isolates were correctly identified, with
order from the highest probability to a cumulative total of a likelihood of 285%. An additional 29 isolates (7.6%)
99.9%. yielded a correct identification but with a likelihood of <85%
The original automated reader for MicroScan, the auto (overall identification, 95.3%). They found that 8.9% of the
SCAN-3, was studied by Woolfrey et al. (126), who used 678 members of the Enterobacteriaceae and 7.6% of all isolates
clinical isolates of members of the family Enterobac- would require supplemental tests for correct identification.
teriaceae that had been identified by the API 20E and Of 17 misidentifications, 12 were at the genus level and 5
conventional biochemical tests. The autoSCAN-3 correctly were at the species level. In one instance no identification
identified 95.1% at the genus level and 94.9% at the species was given even though that organism was in the data base.
level versus 97.9% at the genus level by visual interpretation Enterobacter spp. and members of the tribe Proteeae re-
(P < 0.01). Organisms most frequently misidentified at the quired supplemental testing most frequently. A common set
genus level were E. coli (5 of 141 isolates), Enterobacter of 134 isolates was tested in the two participating laborato-
cloacae (6 of 53 isolates), and Shigella spp. (15 of 42 ries to determine interlaboratory agreement. The overall
isolates). Of 42 Shigella strains, 97.9% were correctly iden- agreement was 86%.
tified by visual interpretation of the panels compared with Tenover et al. (117) tested 310 well-characterized non-
64% by the autoSCAN-3. Overall, approximately one-third glucose-fermenting gram-negative bacilli on the WalkAway-
of the panels had to be visually inspected for at least one 96, using both the GNB and R-GNB. The GNB reported
reaction as a result of equivocation by the autoSCAN-3 41.3% correct at 285% probability, 21.3% correct with low
reader. The biochemicals that required visual interpretation probability, 22.4% incorrect, and 15.0% unidentified. The
were randomly distributed, thus indicating the need for R-GNB reported 50.1% correct at .85% probability, 40.5%
improved discrimination by the autoSCAN-3 reader. In correct with low probability, and 9.3% incorrect. Alcali-
addition, 37 isolates were designated as very rare biotypes genes xylosoxidans subsp. xylosoxidans, P. putida, P. fluo-
by the autoSCAN-3 and required visual interpretation of the rescens, and X. maltophilia accounted for 57.8% of the
plate. This resulted in correct identification of 27 of the misidentifications with the colorimetric panel, and P. fluo-
isolates as Serratia marcescens and 1 as Proteus mirabilis. rescens accounted for 28% of the misidentifications with the
Of the remaining nine isolates, four were correctly identified rapid fluorometric panel. Supplemental tests were required
(three Enterobacter cloacae and one Serratia marcescens) for 36.3% of the isolates tested with the GNB and 40.5%
when the telephone computer service was contacted. tested with the R-GNB. With the R-GNB, supplemental
Rhoden et al. (99) evaluated a production model of the tests could be set up after the interpretation at 2 h. Only
autoSCAN-4 with typical and atypical strains of members of 13.5% of the GNB tests could be interpreted at 18 h, with the
the family Enterobacteriaceae, nonfermenters, and nonen- remainder requiring 42 h of incubation. When the identifica-
teric fermenters, that had been identified at the Centers for tions were recalculated with an updated R-GNB data base
Disease Control by conventional methods. These MicroScan and revised software, 77.1% were correctly identified at
test trays required freezer storage. The autoSCAN-4 identi- >85% probability and 8.1% were misidentified at this same
fied 95.4% of the members of the Enterobacteriaceae (N = probability.
307), 96.6% of the nonenteric fermenters (N = 29), and Published abstracts regarding the WalkAway-96 R-GNB
94.2% of the nonfermenters (N = 69). Visual interpretation have reported an identification accuracy that has varied from
of the trays yielded essentially the same results. Additional 66.4 to 97.8% (24, 30, 32, 47, 57, 59, 70, 75, 76, 87, 118).
tests were required for as few as 5.2% of the members of the Organisms reported in some studies to be frequently misi-
Enterobacteriaceae and as many as 51.8% of the nonenteric dentified or to yield inconclusive results included K pneu-
fermenters. "Very rare biotype" was printed by the auto moniae, K oxytoca, Enterobacter cloacae, C. freundii,
SCAN-4 for only four strains of the Enterobacteriaceae and Serratia marcescens, Proteus mirabilis, M. morganii, Shi-
three nonfermenters. Three of six Edwardsiella tarda iso- gella spp., and P. fluorescens (30, 32, 59, 75, 118). One study
lates were misidentified because weakly positive H2S tests reported that 21% of the P. aeruginosa isolates tested in the
were read as negative. In several instances, the autoSCAN-4 WalkAway-96 failed to grow and could not be identified (59).
reader did not detect a weakly positive arginine test. The The original MicroScan GPB panel for identification of
authors concluded that the autoSCAN-4 was a highly effi- gram-positive bacteria was supplied and stored frozen and
cient, accurate, and reliable system for identification of contained 27 tests. Tritz et al. (119) evaluated the GPB
gram-negative bacilli. frozen panel for identification of 100 Enterococcus strains.
Gavini et al. (42) evaluated the autoSCAN-4 for identifi- Conventional methods were used for identification of entero-
cation of 372 clinical isolates of members of the family cocci and were based on reactions listed by Facklam and
Enterobacteriaceae with a dry panel and a new formulation. Collins (37). Sixty isolates of streptococci, identified by
Conventional tests were used for reference identification. conventional methods, were also evaluated. The GPB tests
The autoSCAN-4 correctly identified 95.4% of the isolates at were read on the autoSCAN-4, and 94 enterococci (77
the species level and 98.4% at the genus level. Of the 17 Enterococcus faecalis, 14 Enterococcus faecium, and 3
isolates not identified at the species level, 16 required Enterococcus durans strains) were correctly identified. Of
additional tests. A correct identification at the genus level the remaining six strains, four were identified as Enterococ-
was obtained in 13 of these cases. Only one isolate was cus faecalis by GPB and as Enterococcus solitarius by the
misidentified, an Enterobacter agglomerans isolate that was conventional method and two were identified as Enterococ-
identified as enteric group 41. cus durans by GPB and as Enterococcus hirae by the
Pfaller et al. (95) reported on the WalkAway-96 R-GNB conventional method. The GPB did not identify any of the 60
for identification of 292 members of the family Enterobac- Streptococcus spp. as Enterococcus spp. Hussain et al. (52)
teriaceae and 91 nonenteric bacilli. The API 20E and con- determined the accuracy of the GPB for identification of 175
ventional biochemical tests were used for reference identifi- clinical isolates of coagulase-negative staphylococci. The
VOL. 5, 1992 AUTOMATED IDENTIFICATION SYSTEMS 313

GPB results were compared with those of the API Staph- commonly isolated Staphylococcus strains, none were incor-
Ident system, and discrepancies were resolved by the con- rectly identified and only 1 (Staphylococcus lugdunensis)
ventional method of Kloos and Schleifer (64). The GPB uses required additional tests for correct identification.
18 of the 27 substrates for identification of Staphylococcus Godsey et al. (43) evaluated the WalkAway-96 R-GPS for
spp. The GPB was interpreted both visually and by the identification of 607 staphylococci, 704 streptococci, 42
autoSCAN-4. After overnight incubation, only 36% of the Micrococcus isolates, 28 Aerococcus isolates, and 30 L.
identifications were complete; the remaining isolates re- monocytogenes isolates. The accuracy of identification
quired an additional 24 h of incubation. The GPB correctly ranged from 79% (S. warnen) to 100% (multiple species).
identified 83.4% of the coagulase-negative staphylococci, The overall identification accuracy was 95.7%.
with 97% of S. epidermidis and S. saprophyticus isolates Stoakes et al. (113) evaluated the AIP both by visual
yielding an excellent identification. An acceptable identifi- interpretation and with the autoSCAN-4. The results were
cation (>80%) was obtained for most S. haemolyticus, S. compared with those obtained by the Virginia Polytechnic
simulans, Staphylococcus capitis, Staphylococcus cohnii, Institute conventional method. A total of 237 anaerobes
and S. xylosus isolates, but only 35.2% of Staphylococcus were tested, of which 166 (70%) were correctly identified to
hominis, 69.5% of Staphylococcus warnei, and 60% of species level (probability .85%) by visual readings versus
Staphylococcus sciun isolates were properly identified. Sup- 157 (66.2%) correctly identified by the autoSCAN-4 (P >
plemental tests were not required with the GPB. The only 0.30). When supplemental tests were performed, 80.1 and
reading errors by the autoSCAN-4, compared with visual 76.7%, respectively, were correctly identified (P > 0.30).
readings, were observed with bacitracin, crystal violet, or The visual and automated interpretations produced 94 dis-
novobiocin susceptibility and were due to air bubbles or crepant reactions, including 37 with gram-negative bacilli, 48
scratches on the plastic well. The GPB biochemical tests that with clostridia, and 9 with other organisms. The auto
differed from conventional tests included the pyrrolidonyl- SCAN-4 could not interpret a reaction in 46 instances,
3-naphthylamide, ,-D-galactopyranosidase, lactose, and whereas 93 reactions were difficult to interpret visually.
urease tests (false-negative reactions) and the arginine, man- Difficulties in interpretation of reactions were randomly
nose, novobiocin, and lactose tests (false-positive reac- distributed with the autoSCAN-4 but were frequent for
tions). L-lysine-,3-naphthylamide reactions with Bacteroidesfragilis
Kloos and George (62) compared the GPB and R-GPB and for bis-p-nitrophenyl phosphate reactions with gram-
tests for the identification of 25 different Staphylococcus positive organisms when read visually. To determine the
spp. The 920 strains tested were obtained from reference and reproducibility of the system, 50 strains (30 gram-negative
type collections, clinical specimens, and the skin of humans bacilli and 20 clostridia) representing 1,200 reactions were
and animals. The nonreference strains were identified by the tested in duplicate. Discrepant results were 69 (5.8%) with
characteristics and methods described by Kloos and Lambe the autoSCAN-4 and 57 (4.8%) with visual readings (P >
(63), and selected strains from each species were verified via 0.20). As a result, 8% of the strains were changed from a
DNA-DNA hybridization with reference or type strains. The correct to an incorrect identification with the autoSCAN-4
GPB was read visually at 15 to 48 h, and the R-GPB was read and 6% with visual readings.
by the WalkAway-96 at 2 h. Both systems correctly identi- Land et al. (67) evaluated the YIP with 437 recent clinical
fied 290% of strains of Staphylococcus arlettae, S. aureus, yeast isolates, using the API 20C as the reference method.
Staphylococcus auricularis, S. capitis subsp. capitis, Staph- However, the MicroScan automated instruments were not
ylococcus camosus, S. cohnii subsp. cohnii, Staphylococcus used and only visual readings were obtained. The YIP
lentus, S. saprophyticus, and S. sciuri. In addition, the identified 85% of all isolates and 92% of the taxa included in
R-GPB correctly identified >90% of strains of Staphylococ- the data base. The YIP identified 94% of rapidly growing
cus caprae, Staphylococcus caseolyticus, S. epidermidis, yeasts in the genera Candida, Hansenula, Pichia, Rhodo-
Staphylococcus kloosii, S. warneri subsp. 2, and S. xylosus. torula, Saccharomyces, and Torulopsis (98% when organ-
Both systems correctly identified between 80 and 90% of isms and biocodes not in the data base were excluded).
strains of Staphylococcus equorum, S. haemolyticus subsp. However, only 65% of identifications of slower-growing
1, Staphylococcus hyicus, and Staphylococcus intermedius. yeasts (Blastoschizomyces, Cryptococcus, Geotnichum, Hy-
The R-GPB also correctly identified between 80 and 89% of phopichia, Phaeococcomyces, Prototheca, and Trichos-
strains of S. capitis subsp. ureolyticus, S. cohnii subsp. poron species) correlated with results of the API 20C. When
urealyticum, S. hominis, S. simulans, and S. wameri subsp. unrecognized organisms and biocodes were excluded, the
1. Both systems identified 50 to 75% of strains of S. correlation improved only to 68%. The yeasts that posed the
chromogenes, Staphylococcus gallinarum, Staphylococcus greatest problem for the YIP were Torulopsis beigelii, Blas-
lugdunensis, and Staphylococcus schleiferi. toschizomyces capitatus, Geotrichum spp., and Cryptococ-
Stoakes et al. (114) evaluated the WalkAway-96 R-GPB cus neoformans. The YIP correctly identified 40 (67%) of 60
with 239 strains of staphylococci belonging to 17 species. known serotypes of Cryptococcus neofornans. One strain of
Ten or more strains of commonly occurring clinical strains serotype A was incorrectly identified, and 19 isolates (3 of 14
were tested. Conventional methods were used to establish serotype A, 3 of 15 serotype B, 2 of 15 serotype C, and 11 of
the true identity of the strains. Of the tested strains, 219 15 serotype D isolates) yielded biocodes not found in the
(91.6%) were correctly identified without additional tests, 9 data base. The authors concluded that the expansion of the
(3.8%) were correctly identified after additional tests, 6 YIP data base and adjustment of the indoxyl phosphatases
(2.5%) were incorrectly identified, and 5 (2.1%) were classi- and pH indicators for sucroses 1 and 2 and trehalose should
fied as rare biotypes and not identified. The misidentified make the YIP an excellent system for identification of
stains (all considered common isolates), were S. capitis (2 of medically important yeasts.
21 strains), S. hominis (2 of 20 strains), S. wameri (1 of 25 St.-Germain and Beauchesne (112) reported on a modified
strains), and S. xylosus (1 of 12 strains). One discrepant version of the MicroScan yeast identification system. Four
reaction (positive in each instance) was responsible for each substrates (isoleucine, urea, N-acetylgalactosamine, and
misidentification. Of the 37 (representing 7 species) less trehalose) had been reformulated and the data base regener-
314 STAGER AND DAVIS CLIN. MICROBIOL. REV.

ated for Version 17 software. The authors evaluated the YIP identified 86.3% of members of the family Enterobac-
for identification of 357 yeastlike clinical isolates of 11 teriaceae and 92.3% of gram-negative nonenteric bacilli
genera and 30 species. Of these, 217 were common isolates tested. Supplemental tests were required for 7.6% of the
and 140 were relatively uncommon. The YIP results were isolates for correct identification. Tenover et al. (117) found
read both visually, by two observers, and with the auto that the WalkAway-96 GNB and R-GNB correctly identified
SCAN-4. The API 20C and morphological characterization 62.6 and 90.6%, respectively, of the gram-negative nonen-
on cornmeal-Tween 80 agar were the reference systems. teric bacilli tested. However, approximately 35 and 45%,
Conventional tests were used to resolve any discrepancies respectively, of the correct identifications with the GNB and
between the API 20C and the YIP. Both the YIP, as R-GNB had a probability of less than 85%. In addition,
interpreted by the autoSCAN-4, and the API 20C identified approximately 40% of the isolates required supplemental
278 (78%) of 357 strains without supplemental tests. With tests for correct identification by both the GNB and R-GNB.
supplemental tests, the YIP correctly identified 345 strains, Only 13.5% of the GNB results could be interpreted at 18 h,
incorrectly identified 10 strains, and did not identify 2 with the remainder requiring 42 h of incubation. A. xylosox-
strains, for an overall accuracy of 96.6%. It accurately
identified 99.5% of the common strains and 92.1% of the less idans subsp. xylosoxidans, P. putida, P. fluorescens, and X.
common strains. Of the common strains, the autoSCAN-4 maltophilia were most frequently misidentified by the GNB,
incorrectly identified only one strain of C. glabrata (99.5% whereas P. fluorescens accounted for 28% of the misidenti-
accuracy), whereas none were incorrectly identified when fications by the R-GNB. Published abstracts concerning the
the panels were visually interpreted. Of the less common WalkAway-96 R-GNB have reported an identification accu-
strains, the autoSCAN-4 incorrectly identified nine strains racy that has ranged from 66.4 to 97.8% (24, 30, 32, 47, 57,
and failed to identify two strains (92.1% accuracy). On visual 59, 70, 75, 76, 87, 118). There have been no studies of the
interpretation 12 strains were incorrectly identified and 2 or autoSCAN-4 involving the GPB panel with dried substrates.
3 strains were not identified (89.3% accuracy). There was an Kloos and George (62) found that the WalkAway-96 R-GPB
overall accuracy of 96.6% for the autoSCAN-4 versus 95.8% correctly identified common Staphylococcus spp. and many
for visually interpreted panels (P > 0.50). Identical profile uncommon Staphylococcus spp. at a probability of >85%.
numbers were obtained by both observers and the auto Stoakes et al. (114) found that the WalkAway-96 R-GPB
SCAN-4 for 44% of the 357 strains. Difficulties with the correctly identified 91.6% of Staphylococcus spp. tested
interpretation of chromogenic substrates, particularly without supplemental tests and 95.4% with required supple-
,-naphthylamide substrates and nitrophenyl-linked sub- mental tests. None of the less commonly isolated Staphylo-
strates, resulted in discrepancies in profile numbers between coccus spp. (37 of 239 strains) were incorrectly identified.
the two observers and the autoSCAN-4. Godsey et al. (43) tested 1,411 gram-positive bacteria, in-
Abstracts regarding the YIP have recently been published. cluding staphylococci, streptococci, Micrococcus isolates.
Simmonds et al. (107) reported on the YIP for identification Aerococcus isolates, and L. monocytogenes, with the Walk-
of 232 clinical yeast isolates from 10 genera and 29 species. Away-96 R-GPB and found an overall identification accu-
The API 20C was used as the standard method, and discrep-
ancies were resolved with conventional assimilation and racy of 95.7%. The only evaluation of the AIP was with the
fermentation tests. The overall accuracy of the YIP was 59% autoSCAN-4 (113). When required supplemental tests were
when interpreted by the autoSCAN-4 and 72% when inter- performed, the autoSCAN-4 AIP correctly identified 76.7%
preted visually (P < 0.01). When biotypes not found in the of the anaerobes tested (N = 237). Results for correct
data base were excluded, the accuracy was 62% by auto identification with the autoSCAN-4 compared favorably
SCAN-4 and 76% by visual interpretation (P < 0.01). Mor- with visual interpretation of the same panels (P > 0.30).
phological observations were needed to identify more than St.-Germain and Beauchesne (112) reported on a recent
50% of the isolates, and confirmatory biochemical tests were modified version of the autoSCAN-4 YIP and found that
required for identification of 29% of the isolates by the 99.5% of the common yeast isolates and 92.1% of the less
autoSCAN-4 versus 16% by visual interpretation. Belcher et common yeast isolates tested were correctly identified when
al. (14) tested the YIP with 189 clinical yeast isolates from 7 the required supplemental tests were performed. There was
genera and 20 species. The Minitek Yeast Carbon Assimila- an overall accuracy of 96.6% for the autoSCAN-4 versus
tion procedure (Becton-Dickinson) was the standard for 95.8% when the same panels were visually interpreted (P >
comparison. The autoSCAN-4 correctly identified 79% of 0.50). Recent abstracts on the autoSCAN-4 YIP have re-
the isolates, whereas visual interpretation correctly identi- ported that 62 and 79% of the yeasts tested were correctly
fied 86% (P > 0.05). No supplemental morphological or identified (14, 107). There have been no published studies of
physiologic tests were performed. Stockman and Roberts the WalkAway-96 YIP.
(115) tested the YIP with 260 yeast isolates from 24 species. The MicroScan automated systems will identify a wide
The API 20C was the reference method. It was not stated
whether readings were obtained by an automated system or variety of organisms. However, further improvements in the
by visual inspection. The YIP correctly identified 75.3% of data base and software for all panel types are needed. An
the isolates with a probability of >85%. Supplemental tests automated method for the addition of mineral oil to the
were required for 21% of the isolates. Two C. famata, two various panels and for inoculation of the HNID, AIP, and
Candida zeylanoides, and one each of C. glabrata, C. YIP would enhance work flow.
lipolytica, C. parapsilosis, C. tropicalis, Kluveromyces lac- There have been no evaluations of the HNID panel and
tis, S. cerevisiae, and Sporobolomyces isolates were misi- GPB dried-substrate panels by using MicroScans automated
dentified by the YIP. systems. Evaluation of the WalkAway-40 and further eval-
In summary, the most recent study of the autoSCAN-4 uation of the WalkAway-96 and autoSCAN-4 with appro-
GNB correctly identified 95.4% of tested members of the priate MicroScan panel types will be required as Micro
family Enterobacteriaceae to the species level (42). Pfaller et Scan continues to update the data bases and software
al. (95) reported that the WalkAway-96 R-GNB correctly programs.
VOL. 5, 1992 AUTOMATED IDENTIFICATION SYSTEMS 315

ALADIN AND AUTOREADER (approximately 85 i±l) of the suspension. When the UniScept
20E is to be incubated off-line and then read on the UniScept
The Automated Laboratory Diagnostic Instrument (ALA- AutoReader, the cupule section of the arginine, lysine,
DIN; Analytab Products) is a computer-assisted system with ornithine, urea, and H2S tubes must be manually overlaid
on-line incubation that interprets biochemical and suscepti- with mineral oil.
bility tests by a video image process. Analytab Products also The ALADIN has a 60-specimen-capacity incubator, with
manufactures the UniScept system AutoReader, which tests up to two UniScept panels (identification and/or susceptibil-
the same panels as the ALADIN but requires off-line incu- ity) being processed per specimen. This allows the process-
bation and which performs automatic interpretation of re- ing of up to 120 individual panels. A virtually unlimited
sults through photometric evaluation of test wells at multiple number of panels may be incubated off-line for reading on
wavelengths. The ALADIN and UniScept systems perform the ALADIN. One or two UniScept panels per specimen are
both MIC and qualitative susceptibility tests on aerobic placed in a universal carrier, which is a molded plastic frame
gram-negative bacilli and gram-positive bacteria. Both sys- approximately 5 by 85/8 in. (ca. 13 by 22 cm) in size. The
tems identify gram-negative bacteria (UniScept 20E), gram- panels are inoculated, and the universal carrier is placed into
positive bacteria (UniScept 20GP), and anaerobes (AN- 1 of 60 test slots (two rows of 30) in the incubator. The
Ident). Other Analytab Products identification panels can be elevator and grabber arms automatically transfer the panels
used with either system, but manual entry of the reactions is to the read station. The specimen number and panel type are
required. These panels identify gram-negative bacteria (Rap- interpreted by video image processing. This activates the
id-E and Rapid NFT), Staphylococcus spp. (Staph-Ident), appropriate incubation cycle and reagent addition for each
and yeasts (Yeast-Ident). panel. After appropriate incubation, panels are transferred
The UniScept 20E has 20 modified conventional sub- to the reagent-dispensing station for addition of reagents;
strates, and the data base includes information for identifi- after further appropriate incubation, they are returned to the
cation of 55 species of the family Enterobacteriaceae and 50 reader for a final examination. Currently, up to nine reagents
groups, genera, or species of nonfermentative, oxidase- can be dispensed. Each microtube is examined by a black-
positive, gram-negative bacilli in 18 to 24 h. The UniScept and-white video image processor through one to four colored
20GP has four chromogenic, one ,B-naphthol-labeled, one filters selected by the computer, depending on which reac-
0-naphthylamide-labeled, and 14 modified conventional sub- tion is being analyzed. The video imaging camera isolates an
strates, and the data base includes information for identifi- area of interest for each microtube and uses 200 to 300
cation of 13 Staphylococcus spp., 3 Enterococcus spp., and picture elements, or pixels, to view this area. For example,
2 Streptococcus spp. (group D, nonenterococcus) in 18 to 24 fermentation reactions are read at the middle or bottom of
h. The AN-Ident has nine chromogenic substrates whose the microtube. The density of each pixel perceived by the
reactions are detected by the liberation of indoxyl, o-nitro- camera is a specific voltage that is represented digitally for
phenol, or p-nitrophenol from the substrates, nine 3-naph- computer processing. These specific results are converted
thylamide-labeled substrates, one modified conventional into plus or minus reactions, and the computer generates a
substrate, and a test for catalase. The AN-Ident data base seven-digit profile number. This profile number, or "bio-
includes information for identification of 83 groups, genera, type," identifies the most probable organisms of each bio-
or species of anaerobic bacteria in 4 h. type. The panels are then transferred to the disposal station,
The ALADIN system is composed of an incubator, eleva- where they are discarded into a waste bag by miniature
tor, reagent-dispensing station, grabber arms, image proces- forklifts.
sor, and disposal station. A separate workstation interacts Panels incubated off-line are placed in a tray and read on
with the ALADIN and is composed of a keyboard, computer the UniScept AutoReader or the ALADIN. For the Uni
(Compaq 386 Sx model 40), UniScept deziner-er Software, Scept AutoReader, a photometer examines each microwell
and Okidata 320 printer. The software contains the data base at multiple positions and wavelengths. A photodiode detects
for all UniScept products, susceptibility programs, patient the transmitted light, and the resulting voltage is converted
demographics, storage of data, and antibiograms. A bidirec- to an optical density value, which is equated to a predeter-
tional mainframe computer interface that allows up-load of mined positive or negative reaction. The resulting biochem-
daily test results and down-load of demographics may be ical profile is compared with the data base for identification.
purchased. Shulman et al. (105) reported on the ability of the ALA-
Inocula for the UniScept 20E and UniScept 20GP can be DIN to read UniScept 20E panels. A total of 300 isolates (144
grown on either selective or nonselective agar media, E. coli, 45 Proteus/Providencia spp., 27 KiebsiellalEntero-
whereas inocula for the AN-Ident must be grown on nonse- bacter spp., 18 Serratia spp., 12 Pseudomonas spp., and 54
lective agar media (excluding 5% sheep blood Trypticase soy other organisms) were tested. On a test-for-test basis, the
agar [TSA] or Schaedler's blood agar). To prepare inocula, ALADIN and visual readings of 300 organisms showed a
several morphologically similar colonies are selected and 99.0% level of correlation. These differences in readings did
suspended in sterile 0.85% saline (UniScept 20E and Uni not affect the final identification of any of the tested organ-
Scept 20GP) or sterile distilled water (UniScept 20E, Uni isms. The reference system to establish the true identity of
Scept 20GP, and AN-Ident). The suspension is adjusted to the organisms and the percentage of organisms correctly
the equivalent of a no. 0.5 McFarland turbidity standard identified was not mentioned.
(UniScept 20E and UniScept 20GP) or a no. 5 McFarland Navarro et al. (86) reported on the ALADIN for its ability
standard (AN-Ident). The inocula for the UniScept 20E and to read AN-Ident panels. A total of 125 anaerobes (48
UniScept 20GP are diluted 1:100 in sterile 0.85% saline. With Clostridium spp., 37 Bacteroides spp., 15 Fusobacterium
either the UniScept Autoinoculator or the manual pipetting spp., 9Actinomyces spp., 5 Propionibacterium spp., and 11
device, 100 ,ul of inoculum is dispensed into each micro- other organisms) were tested. With respect to visual read-
tube of the UniScept 20E (300 RI into the citrate, Voges- ings of the 125 organisms, the ALADIN showed a 96.1%
Proskauer, and gelatin microtubes) and UniScept 20GP. For level of correlation on a test-for-test basis. The abstract did
the AN-Ident, each microtube is manually filled with 2 drops not mention whether these differences in readings affected
316 STAGER AND DAVIS CLIN. MICROBIOL. REV.

the final identification of any of the tested organisms by the organisms that can be identified by the UniScept 20GP
ALADIN. Also, the reference system to establish the true represent most of the gram-positive bacteria routinely en-
identity of the organisms and the percentage of organisms countered in the clinical laboratory, it is desirable to have an
correctly identified were not mentioned. automated system that could identify a broader range of
The only extensive report on the ALADIN identification significant gram-positive bacteria. The only rapid (4-h) panel
system was a collaborative study at three medical centers that can be automatically read by the ALADIN and Uni
(28). In this study, 318 aerobic and facultatively anaerobic Scept system AutoReader is the AN-Ident. Automated read-
gram-negative bacteria and 148 obligately anaerobic isolates ing of other rapid Analytab Products identification panels
were tested. All bacteria were identified by established would enhance the usefulness of these systems.
Lonventional methods. The UniScept 20E and AN-Ident It is commonly accepted that the UniScept 20E, UniScept
biochemical reactions were interpreted on the ALADIN by 20GP, and AN-Ident have excellent data bases for identifi-
video imaging. Then the panels were visually interpreted. cation of bacteria. However, there are no reported studies
The results were compared on a test-for-test basis, with the with either the ALADIN or UniScept system AutoReader
visual interpretation serving as the reference. When two or and the UniScept 20GP. The studies by D'Amato et al. (28)
more discrepant biochemical test results per test system and O'Hara et al. (90) indirectly suggest that the ALADIN
were obtained, the isolate was retested. The results of the and UniScept system AutoReader, respectively, will reliably
retest were considered final. Altogether, 6,360 individual identify isolates with either the UniScept 20E or the AN-
tests were performed on the 318 gram-negative bacteria, Ident. However, studies to prove the accuracy of identifica-
with an overall agreement of 96.5%. False-negative readings tion of the UniScept 20E, UniScept 20GP, and AN-Ident
for indole production, mannitol fermentation, and tryp- when tested on these systems have not been reported.
tophan deaminase produced =90% agreement. The authors
suggested that the computer algorithms for these substrates, BIOLOG
all carbohydrates, and the o-nitrophenyl-3-D-galactoside re-
action be adjusted. The Biolog system (Biolog, Inc., Hayward, Calif.) was
With anaerobic test results, the overall agreement be- introduced in 1989 for identification of aerobic gram-negative
tween the ALADIN and visual interpretation was 95.2%. bacteria (enteric bacilli, nonfermenters, and fastidious spe-
False-positive results with indoxylacetate yielded an 81.1% cies) by determination of carbon source utilization profiles.
agreement for this substrate. False-negative results, partic- Recently, Biolog has added the capability to identify a broad
ularly with carbohydrate and some ,B-naphthylamide deriv- range (cocci, bacilli, and spore-forming bacilli) of aerobic
ative tests, were noted. It was suggested that the indole test gram-positive bacteria. The Biolog GN MicroPlate (for iden-
might be improved by the substitution of dimethylaminocin- tification of gram-negative bacteria) and the Biolog GP
namaldehyde for p-dimethylaminobenzaldehyde. This has MicroPlate (for identification of gram-positive bacteria) are
now been incorporated into the ALADIN. 96-well dehydrated panels containing tetrazolium violet, a
Intralaboratory and interlaboratory reproducibility with buffered nutrient medium, and a different carbon source for
the ALADIN biochemical profiles averaged 96.0 and 91.5%, each well except the control, which does not contain a
respectively, in the three laboratories. Similar values were carbon source. The microwells are rehydrated with a cell
obtained for visually determined results. suspension and read at either 4 h or overnight (16 to 24 h) for
The report of D'Amato et al. (28) verified that there was the ability of the bacteria to utilize the carbon source.
good agreement between video image and visual interpreta- Tetrazolium violet is a redox dye used to detect electrons
tion of biochemical reactions for the UniScept 20E and donated by NADH to the electron transport system. Re-
AN-Ident, but it did not report the effect of false-positive or duced tetrazolium violet is a purple formazan. When a
false-negative results by video image on identification of the carbon source is not used, the microwell remains colorless,
organisms tested. as does the control well. The resulting pattern of purple wells
There are some limitations to the use of the UniScept 20E yields a "metabolic fingerprint" of the bacterium tested (15,
and the ALADIN. With the UniScept 20E, some slow- 16).
growing members of the family Enterobacteriaceae and For metabolic capability studies, the MT MicroPlate con-
gram-negative organisms that are not members of the En- tains only tetrazolium and a buffered nutrient medium with-
terobacteriaceae require supplemental tests (oxidase, oxida- out a carbon source in any of the 96 wells. The user can add
tion-fermentation glucose medium, or motility medium) and various carbon sources to the microplate. The ES Mi-
incubation for 36 to 48 h. The ALADIN will not identify croPlate contains 95 different carbon sources and is designed
these organisms, since reagents are added and panels are for characterizing and/or identifying different E. coli and
read at 24 h. These panels are automatically returned to the Salmonella strains, for mutant strain characterization or for
incubator for removal and off-line incubation. quality control testing of E. coli K-12 or Salmonella typhi-
The only study of the UniScept system AutoReader was murium LT2 strains carrying recombinant plasmids, and for
reported by O'Hara et al. (90). They compared the agree- epidemiological studies. Although there is no Biolog data
ment of automated and visual readings of biochemical tests base for the MT MicroPlate and ES MicroPlate, a custom
with the results of the UniScept 20E. They tested 291 data base can be created by using Biolog software. Biolog is
oxidase-negative and 49 oxidase-positive glucose-fermenting developing the capacity for identification of yeasts, anaer-
isolates. Since the AutoReader is not designed to read a obes, and additional environmental and oligotrophic bacte-
UniScept 20E at 48 h, nonfermentative bacteria were not ria.
evaluated. Of the 6,800 tests compared, only 45 readings The 95 substrates contained in the Biolog GN MicroPlate
(0.7%) did not match. Discrepant results involved 16 bio- and the Biolog GP MicroPlate include carbohydrates, car-
chemicals, with indole and citrate disagreeing most often (13 boxylic acids, amides, esters, amino acids, peptides, amines,
of 45). alcohols, aromatic chemicals, halogenated chemicals, phos-
The ALADIN and UniScept system AutoReader identify phorus- and sulfur-containing chemicals, and polymeric
a limited number of gram-positive bacteria. Although the chemicals. The Biolog system data base includes informa-
VOL. 5, 1992 AUTOMATED IDENTIFICATION SYSTEMS 317

tion for identification of 569 species or groups of aerobic strains were correctly classified to species. However, Aero-
gram-negative bacteria and 225 species or groups of gram- monas species were not included in the Biolog data base at
positive bacteria and encompasses almost all known human that time, and this study demonstrated only that Aeromonas
pathogens and most important environmental species. isolates could be identified to species with substrates present
The Biolog automated system consists of a manual eight- in the Biolog GN panel.
channel repeating pipettor, a turbidimeter, a MicroPlate Armon et al. (4) tested the Biolog GN panel with Legion-
Reader, a MicroLog Program Disk, and any DOS-based ella spp., also before this organism was included in the
IBM-compatible PC, including XT and AT 286, 386, and 486 Biolog data base. They tested one strain each of Legionella
models. A manual system without the MicroPlate Reader is pneumophila (Philadelphia 1), Legionella pneumophila (en-
also available. The computer must have a hard drive of at vironmental isolate), Legionella micdadei, Legionella
least 20 MB, at least one floppy drive (3.5 or 5.25 in. [ca. 8.9 oakridgensis, Legionella longbeachae, and Legionella gor-
or 13.3 cm]), and, when the automated reader is used, an manii. The strains were identified by reference biochemicals
enhanced 101 keyboard. The software runs on systems with but were not serogrouped. The strains were grown for 3 days
VGA or EGA color graphics or with monochrome displays. at 35.5°C on buffered charcoal-yeast extract agar supple-
Different software versions can be purchased, either for mented with L-cysteine. The organisms were harvested, and
manual entry or for automated reading of the test results. the inoculum for each strain was prepared in two different
The latter software version allows data to be printed and buffers [0.1 mol/liter of phosphate buffer (pH 7.0) and 0.05
saved in computer files, which can be utilized by user- mol/liter of N-(2-acetoamido)-2-aminoethane sulfonic acid
provided software routines. Alternatively, the data can be (ACES) buffer supplemented with 4 ml of 10% L-cysteine
filed in a "user data base" in which the reaction patterns are (pH 7.0) per liter]. The inoculum prepared in each buffer was
permanently saved. An unknown biochemical profile can be inoculated into a separate Biolog GN panel and, after
compared with the Biolog GN or Biolog GP data base, the incubation, was read manually. The Legionella strains tested
user data base, or a combination of the two. Other features yielded biochemical profile variations that allowed distinc-
of the software include on-line information about any species tion between the tested isolates. With the existing data base,
in the library, cluster analysis programs in the form of the authors found some overlap of biochemical profiles with
dendrograms and two-dimensional and three-dimensional Moraxella bovis. The biochemical reactions showed an
plots to demonstrate the relatedness of strains or species, enhanced color reaction when ACES buffer with L-cysteine
and the separation of the GN data base into clinical and was used as diluent for the inoculum.
environmental versions. Mauchline and Keevil (73) used the Biolog system to
The inoculum for gram-negative organisms is prepared establish a new data base and identify asaccharolytic
from TSA or TSA-blood agar medium. The inoculum for Legionella spp. They tested single type strains of Legionella
gram-positive clinical and food isolates, with few excep- pneumophila serogroups 1 through 14 (excluding serogroups
tions, must be grown on Biolog Universal Growth Medium 4 and 9) and a single type strain of Legionella bozemanii,
with 5% sheep's blood (Biolog, Inc.). Environmental isolates Legionella dumoffii, Legionella feeleii, Legionella hacke-
typically do not require the addition of blood. Generally, liae, Legionella israelensis, Legionella rubrilucens, Legion-
bacteria are grown for 4 to 18 h. A swab is gently rolled over ella longbeachae, and Legionella micdadei. The isolates
the colonies to prevent carryover of nutrients from the agar were grown on buffered charcoal-yeast extract agar at 37°C
medium into the saline (0.85%) suspension of bacteria. The for 72 h. Inocula were prepared in Page's amoebal saline,
colorimeter or spectrophotometer (optical density at 590 nm and Biolog plates were inoculated and then incubated at 37°C
of 0) is blanked with a tube containing uninoculated saline. in either air or a low-oxygen (=4%) atmosphere. The plates
The bacterial suspension is adjusted to within a low-stan- were read at 24-h intervals up to 72 h, both visually and with
dard-to-high-standard range. The inoculum should be used a Merertech Microplate Reader (Atlas Bioscan, Bognor
within 10 min. The plates are inoculated with 150 ,ul per well Regis, United Kingdom). When the tested legionellae were
and incubated at 30 to 35°C either with or without CO2. The incubated in air, some of the reactions were not visible at 24
plates are read at 4 h either manually or on a computer- h and the substrates required longer incubations to turn
controlled microplate reader. The MicroLog software sub- positive. However, when the same strains were incubated in
tracts the background cell density from the negative control a reduced-oxygen atmosphere, definite reactions were ap-
well and interprets all tests above a threshold as positive. parent in 24 h. When tested under these conditions, none of
The metabolic profile of the organism is matched to a data the legionellae had a metabolic profile that closely matched
base of patterns by using the MicroLog Software Program. any other bacteria in the Biolog data base, indicating that the
Data from the MicroPlates can be permanently saved in profiles obtained were specific for Legionella species. In
computer files to facilitate subsequent analyses. The auto- addition, the authors tested Biolog plates with environmen-
mated reader processes a plate in 5 s. tal isolates that had been provisionally identified by serologic
There have been four published studies on the use of the testing as various Legionella species. They compared the
Biolog GN, one of which used the MicroPlate Reader. results with a combination of the Biolog data base and their
Carnahan et al. (19) tested 20 clinical strains each of Aero- own data base and found that the results agreed with those
monas hydrophila, Aeromonas caviae, and Aeromonas so- obtained by serologic testing. In further experiments, it was
bria with the Biolog GN. The reference methods for identi- determined by using distilled water as the diluent and a
fication of the isolates were not stated. Isolates were grown normal aerobic atmosphere during incubation that adequate
overnight on TSA, and the Biolog GN was inoculated as numbers of positive reactions occurred at 24 h to allow
specified by the manufacturer. Inoculated plates were incu- accurate identification of the tested strains. The authors
bated at 35°C for 18 to 20 h and then read manually. For the concluded that the Biolog system had the ability to identify
three species tested, nine substrates were found to yield the tested Legionella strains at least to the species level but
good, discriminatory values. Seven of these substrates had that multiple strains from all known species would have to be
not been previously identified as being useful for identifying characterized to establish a comprehensive and stable data
Aeromonas isolates to the species level. All 60 Aeromonas base.
318 STAGER AND DAVIS CLIN. MICROBIOL. REV.

Miller and Rhoden (79) evaluated an early version of the features in its metabolic profile and that strains that were
Biolog GN data base with a diverse group of clinically epidemiologically related had similar metabolic profiles.
relevant members of the family Enterobactenaceae (212 However, the results suggested that the Biolog system
strains) and other gram-negative organisms (105 nonferment- would not replace standard methods for species identifica-
ers and 35 oxidase-positive fermenters) consisting of 91 tion of Brucella isolates. Ewalt et al. (35) evaluated the
species. The isolates consisted of usual and unusual organ- manual Biolog system for the ability to differentiate Brucella
isms in proportions likely never to be found in the clinical biovars. The reference system for identification was not
laboratory and were identified by conventional biochemical mentioned. They tested isolates of B. abortus bv. 19; B.
and serologic techniques. The Biolog GN was read on the abortus bv. 1, 2, and 4; B. suis bv. 1, 2, and 3; Brucella ovis;
MicroPlate Reader, and only 266 (75.6%) of 352 organisms Brucella canis; B. melitensis bv. 1 and 2; and Brucella
tested were identified with an acceptable similarity index neotomae. There were no visible reactions at 4 h, but there
(SI). Of the 266 strains identified, 87.3% were correct at the were definite reaction patterns for each Brucella sp. and for
genus level and 75.6% were correct at the species level after the majority of biovars at 24 h. Isolates of the same species
24 h. The error rate was 12.8%. In this study of 352 strains, and biovar demonstrated some variation in their metabolic
46.6% of the species were correctly identified at 4 h and profile. Methylpyruvate and DL-lactic acid were utilized by
57.1% were correct at 24 h. The error rate was 10.4% at 4 h most of the isolates.
and 9.6% at 24 h. When the isolates used in this study were There is an obvious need for further evaluation of the
reevaluated with an upgrade of the data base (Release 2.00), Biolog GN and MicroPlate Reader for identification of
86.4% of the members of the family Enterobacteniaceae and gram-negative bacilli. There have been no published studies
80.4% of the other gram-negative organisms were correctly of the Biolog GP. With the large Biolog data base for
identified. The authors pointed out that the SIs of the gram-negative and gram-positive bacteria, the Biolog system
common clinical isolates may be weakened by slow-growing could enhance the ability of laboratories to identify unusual
environmental strains. In late 1991, an expanded and revised bacteria. As the capabilities of the Biolog system continue to
version of the data base (Release 3.00) was introduced. This expand and the accuracies of the new products are verified,
version has not yet been evaluated. the value of the Biolog system in the clinical laboratory
Recent abstracts have reported on the Biolog GN, but should increase.
only one of these studies used the MicroPlate Reader.
McLaughlin et al. (76) reported on the Biolog GN and MIDI MICROBIAL IDENTIFICATION SYSTEM
MicroPlate Reader for identification of infrequently isolated
gram-negative human pathogens. All strains tested were Gas chromatography of cellular fatty acids is a rapid and
obtained from the American Type Culture Collection. Biolog reliable means of identifying organisms encountered in the
GN correctly identified 68.5% (89 of 130 strains) to species clinical laboratory (80, 81). Because of the large number of
level and 79.2% (103 of 130 strains) to genus level. Ten fatty acids found in the cell wall and cell membranes of
percent (13 of 130 strains) were incorrectly identified. Barth bacteria and because the composition of cellular fatty acids
et al. (12) evaluated the Biolog GN but did not use the is a very stable genetic trait that is highly conserved within a
MicroPlate Reader. They tested the Biolog GN with 46 taxonomic group, fatty acid composition can be successfully
miscellaneous gram-negative bacteria recently isolated from used for identification of bacteria. The MIDI Microbial
humans. The isolates had been identified by conventional Identification System (MIS; Microbial ID Inc., Newark,
biochemical tests. The Biolog GN results were read at 4 h Del.) is a fully automated, computerized, high-resolution gas
and at 16 to 24 h. In most cases, the 4-h reading was the same chromatography system that can analyze more than 300 fatty
as the later reading. Fifty-nine percent of the organisms were acid methyl esters ranging in length from 9 to 20 carbons.
correctly identified to species level, and 83% were identified The MIS computer then searches data bases of known
to genus level. Four of the organisms were listed as the compositions to automatically identify yeasts, anaerobic
second choice for the Biolog GN, and there was no correla- bacteria, and aerobic bacteria, including mycobacteria.
tion between conventional and Biolog GN identifications for Welch (123) has recently reviewed the applications of cellu-
four other organisms. The colorimetric reactions were diffi- lar fatty acid analysis in clinical microbiology and has
cult to interpret. Roman et al. (100) reported on the Biolog described the fatty acid profiles found in various microor-
GN but did not use the MicroPlate Reader. They tested 75 ganisms.
strains of P. cepacia isolated from cystic fibrosis patients in The MIS data base includes information for identification
Ohio, in Utah, and in Toronto, Canada. The strains had been of 15 genera and 65 species or subspecies of the family
identified by using conventional biochemicals. At 24 h, 53 Enterobacteriaceae; 45 Pseudomonas species, subspecies,
(86%) of 62 strains were correctly identified by the Biolog or biovars; 18 Staphylococcus species or subspecies; 19
GN. To determine genus similarities and species differences, Bacillus species; and 53 additional genera of aerobic bacteria
Wong (124) tested 12 strains of Brucella melitensis, 9 strains containing 197 species, subspecies, or biovars. In addition,
of Brucella abortus, and 6 strains of Brucella suis with the MIS data base includes information for identification of
Biolog plates. Strains that were epidemiologically related 32 species, subgroups, or complexes of mycobacteria; 31
(strains isolated from common-source infections and labora- genera of anaerobes containing 254 species, types, or
tory accidents) were also tested. After incubation, the plates groups; and 23 genera of yeasts including 195 species or
were read visually. Methylpyruvate, monomethyl succinate, subspecies. Periodically, an expanded and updated data base
and DL-lactic acid were utilized by all Brucella spp. tested. is provided at no cost to users of the system.
All B. suis isolates tested utilized L-arabinose, a-ketobutyric The MIS is composed of a gas chromatograph (model
acid, uronic acid, ,B-hydroxybutyric acid, and uridine. The 5890A; Hewlett-Packard Co., Avondale, Pa.) equipped with
last two substrates were positive only with B. suis. D-Fruc- a fused-silica capillary column (25 m by 0.2 mm [inner
tose, alaninamide, L-alanine, L-asparagine, and L-glutamic diameter]) containing cross-linked methyl-phenyl silicone as
acid were utilized by all B. abortus isolates tested. It was the stationary phase, an automatic injector, a sample con-
concluded that each Brucella strain had some distinguishable troller, a sample tray, a flame ionization detector, an elec-
VOL. 5, 1992 AUTOMATED IDENTIFICATION SYSTEMS 319

tronic integrator controlled by a DOS-based 386 computer plots of principal-component analyses. The cluster analysis
such as the Hewlett-Packard Vectra, and a ThinkJet Printer/ programs are valuable for tracking nosocomial infections.
Plotter. The MIS software includes programs for operation The MIS has proven valuable in some cases for the
of the gas chromatograph, automatic peak naming, data differentiation of phenotypically similar organisms and for
storage, and comparison of the unknown profile with one or subgrouping or subspecies characterization of organisms.
more data bases by using pattern recognition algorithms. Wallace et al. (121) used the MIS to compare the fatty acid
The data base contains more than 100,000 profiles of strains profiles of Kingella denitrificans, Kingella kingae, and Kin-
collected worldwide and grown under standardized condi- gella indologenes with the phenotypically similar Cardio-
tions. Generally, 20 or more strains of a species or subspe- bacterium hominis and Eikenella corrodens. Kingella in-
cies are included in the data base. dologenes, Cardiobacterium hominis, and Eikenella
Environmental aerobic bacteria are cultured on Trypticase corrodens demonstrated large amounts of cis-vaccenic and
soy broth agar (TSBA) at 28°C. Bacteria that grow poorly on palmitic acids, whereas myristic and palmitic acids were the
this medium are grown on a more enriched medium (e.g., major acids in Kingella denitnificans and Kingella kingae.
Neissenia spp. are grown on chocolate agar). Clinical aerobic Only Cardiobacterium hominis lacked 3-hydroxypalmitic
bacteria are incubated at 35°C on TSBA. Mycobacteria are acid and 3-hydroxymyristic acid and could be differentiated
grown at 35°C on Middlebrook 7H10 or 7H11 agar with from Kingella indologenes and Eikenella corrodens by the
oleate-albumin-dextrose-catalase enrichment (Mycobacteri- presence of 3-hydroxypalmitic acid and 3-hydroxymyristic
um marinum is grown at 30°C). Yeasts are cultured on acid, respectively, in these bacteria.
Christenson et al. (22) isolated Pseudomonas gladioli from
Sabouraud dextrose agar at 28°C. Anaerobic bacteria are
grown overnight in peptone-yeast extract-glucose broth at 11 patients with cystic fibrosis and found that it was not
35°C and harvested by centrifugation. Approximately 40 mg associated with any infectious complications. P. gladioli is
of cells is saponified with 1.0 ml of 1.2 M NaOH in 50% primarily a plant pathogen and is rarely isolated from hu-
aqueous methanol by heating the cells in a boiling-water bath mans. Since it resembles P. cepacia, a known infectious
for 30 min. The saponified cellular lipids are methylated with agent in cystic fibrosis patients, the authors confirmed the
2 ml of methylation reagent (325 ml of 6 N HCI and 275 ml of identity of the isolates by using conventional biochemicals,
DNA hybridization studies, and analysis of cellular fatty
methanol) for 10 min at 80°C, and the fatty acids are acid profiles with the MIS. Most of the P. gladioli isolates
extracted with 1.25 ml of extraction reagent (200 ml of contained 3-OH C10:0 fatty acids, whereas all 58 strains of P.
hexane and 200 ml of methyl tert-butyl ether) by rotating the cepacia lacked this fatty acid.
mixture for 10 min at room temperature. The lower, aqueous Mukwaya and Welch (84) determined the cellular fatty
phase is removed; 3 ml of sample cleanup reagent (10.8 g of acid profiles for 42 strains of P. cepacia isolated from
NaOH dissolved in 900 ml of distilled water) is added; and patients at five cystic fibrosis centers. Hexadecanoic (C16:0)
the tube is rotated for 5 min. About two-thirds of the organic acid, cis-9 hexadecenoic (C16:1 ci,9) acid, and an isomer of
phase is then transferred to a septum-capped sample vial and octadecenoic (C18:1) acid were present in significant amounts
placed in the sample tray. Samples are then logged into the in all strains, and none of the fatty acids had fewer than 14
computer. The automatic injector injects 2,ul of the extract carbon atoms. Through numerical analysis of the fatty acid
through a heated, self-sealing rubber septum in the heated data, the authors identified a different subgroup present at
injection port (250°C). The autosampler allows the system to each of the cystic fibrosis centers.
be operated unattended for up to 2 days at a time. The Lambert and Moss (65) used the MIS to determine the
injected sample is volatilized and swept through the column cellular fatty acid profiles of 182 Legionella strains repre-
by a stream of carrier gas (hydrogen). The column is encased senting 23 species. The 23 species differed in the relative
in a thermoregulated oven, and the MIS computer raises the amounts of 14-methylpentadecanoic (i-C16:0), hexadecanoic
temperature from 170 to 270°C at 5°C per min. At the end of (C16:1), and 12-methyltetradecanoic (a-C15:0) acids and could
the analysis, the column is cleaned by heating (310°C for 2 be placed into three major fatty acid groups. When the fatty
min). The flame ionization detector (300°C) sends the elec- acid profiles of these Legionella strains were linked with
tronic signals produced by the analytes to integrators that their ubiquinone contents, the strains could be distinguished
amplify and process the signals. These data are passed to the from other gram-negative bacteria.
computer, where they are stored and can be compared with deBoer and Sasser (31) examined the cellular fatty acid
the data base. MIS identifications are listed with a confi- compositions of Erwinia carotovora strains and found that
dence measurement (SI) on a scale of 0 to 1.0. The total test Erwinia carotovora subsp. carotovora and Erwinia caroto-
time for each specimen is 30 min. vora subsp. atroseptica had six common fatty acids but that
The calibration standard used with the MIS is a mixture of the subspecies could be differentiated because three of these
straight-chain saturated fatty acids from 9 to 20 carbons in fatty acids had different ratios.
length and five hydroxyl acids. With the calibration mixture, Moss et al. (83) found that Moraxella spp. could be
the retention time of the various peaks can be converted to differentiated from Acinetobacter spp., Psychrobacter im-
equivalent chain length data for fatty acid identification. The mobilis, Oligella urethralis, and CDC groups EO-2, EO-3,
equivalent-chain-length value for each unknown compound M-5, and M-6 on the basis of differences in cellular fatty
is compared with the external standard for peak naming. acids. The MIS also determined that Moraxella bovis, Mor-
Changes in sample injection volume and variables such as axella nonliquefaciens, and some strains of Moraxella lacu-
carrier gas flow rates and column and detector temperatures nata have a common fatty acid group, while Moraxella
will affect the sample retention time. Therefore, the calibra- osloensis, Moraxella phenylpyruvica, Moraxella atlantae,
tion mixture is analyzed after each 10 sample analyses to and strains of Moraxella lacunata have species-specific fatty
correct for any possible drift of retention time. acid profiles. They also used pigment production, cellular
The Library Generation Software allows the user to gen- morphology, and cellular fatty acid profiles of strains orig-
erate data bases. The software contains cluster analysis inally classified as CDC group EO-2 to identify them as
programs that will generate dendrograms or two-dimensional Psychrobacter immobilis, EO-3, or EO-2.
320 STAGER AND DAVIS CLIN. MICROBIOL. REV.

Osterhout et al. (91) evaluated the MIS with 573 isolates of when full-bodied colonies were obtained from 5% sheep
gram-negative nonfermentative bacteria including Pseudo- blood agar that had been incubated at 35°C for 24 h.
monas, Shewanella, Comamonas, Flavimonas, Xanthomo- Euclidian distance cuts were used to define groups, and
nas, Acinetobacter, Agrobacterium, Alcaligenes, Borde- reference strains or phenotypes were used to name the
tella, Flavobacterium, Methylobacterium, Moraxella, defined groups. The authors found that all 29 named species
Ochrobactrum, CDC group EO-2, CDC group VB-3, CDC were identified by the generated library and 20 unnamed
group M-5, CDC group IVC-2, Chryseomonas, Sphingobac- staphylococcal strains (clusters) were recognized. Moss et
tenum, Oligella, and Weeksella species. Of these, 536 were al. (82) used the MIS to create cellular fatty acid library
fresh clinical isolates and 37 were reference strains. All entries for 358 bacterial species or unnamed groups com-
isolates were identified by conventional tests. Isolates were monly recovered from clinical specimens. There were 91
cultured at 28°C for 22 to 26 h on TSBA with 5% sheep cellular fatty acid groups, of which some were specific at the
blood. MIS identifications with an SI of 20.5 were consid- genus or species level while others contained species from
ered a good match. The MIS correctly identified 478 (90%) of different genera. Conventional culture and biochemical
532 strains contained in the data base. However, only 314 methods served as the reference identification for the iso-
(59%) had an SI of .0.5. Of the 54 strains incorrectly lates. Of 2,018 isolates, the MIS correctly identified 78% to
identified, 26 were Acinetobacter, Moraxella, orAlcaligenes the proper cellular fatty acid group, 13% were equivocal
strains and 12 were Pseudomonas pickettii strains. Of 41 because of one or more unusual biochemical reactions, 6%
isolates (representing 12 species) not in the data base, 33 were unidentified by both cellular fatty acid and conven-
were not identified or were named with very low SIs. The tional methods, and 3% were misidentified by cellular fatty
authors attributed discrepancies to incorrect or poorly de- acids. Master et al. (71) identified 13 isolates of pink-
fined data base profiles or an inability to differentiate species pigmented, oxidase-positive bacteria with the MIS, API
that are genetically and chemically closely related. Refer- 20E, Rapid NFT, BBL Sceptor (Becton Dickinson), Corning
ence strains of P. aeruginosa and X. maltophilia demon- Uni-NF Tek (Flow Laboratories, Inc., Rosalyn, N.Y.),
strated a significant variation in SIs when they were tested MicroScan, Pasco (Difco, Inc., Wheat Ridge, Co.), and
on 100 different occasions. There was significant improve- Vitek GNI. Identifications of the same isolates by conven-
ment in SIs when organisms were incubated at 35°C and tional biochemical methods served as reference identifica-
analyzed by a data base generated at this temperature with tions. The unknown fatty acid profiles were analyzed by
the Library Generation Software. It was concluded that the both the routine and an experimental MIS data base. Ten
development of a data base for the culture conditions rou- Methylobactenum spp. and one Pseudomonas vesicularis
tinely used in clinical microbiology laboratories might im- isolate were correctly identified, and one isolate was cor-
prove the accuracy of the MIS. rectly classified as a gram-positive bacillus by the MIS. The
Abstracts concerning the identification of common and remaining isolate was not identified. The commercial bio-
uncommon clinical isolates by the MIS have recently been chemical systems were unable to identify any of the 13
published. deTurck et al. (33) evaluated the MIS with isolates.
well-characterized bacterial strains. The MIS correctly iden- Extensive evaluations of the MIS for the accurate identi-
tified 102 (93.4%) of 109 strains of Pseudomonas spp. fication of common and uncommon clinical isolates (aer-
(including 13 species) and 58 (81.2%) of 71 strains of other obes, mycobacteria, yeasts, and anaerobes) and for isolates
gram-negative nonfermenters (including 12 species). None of which are misidentified, unidentified, or identified with a low
the strains were incorrectly named; rather, they produced no likelihood by commonly used commercial systems will be
identification. Of 30 strains of Bacteroidesffragilis tested, 29 required to determine the utility of the MIS in the routine
(97%) were correctly identified. The authors reported that clinical setting.
118 additional anaerobic bacteria of various genera were all
correctly identified to the genus level. The MIS, using the AUTOSCEPTOR
Virginia Polytechnic Institute anaerobe library, was com-
pared with the RapID ANA II System (Analytab Products) The autoSceptor (Becton Dickinson) will automatically
for identification of anaerobic clinical isolates (74). Discrep- read and report up to 15 Sceptor panels that have been
ancies between the two systems were resolved at Virginia incubated off-line. The autoSceptor is capable of identifying
Polytechnic Institute by conventional testing. The isolates gram-negative bacilli and interpreting breakpoint or MIC
tested were 25 Bacteroides spp., 8 other anaerobic gram- tests on gram-positive and gram-negative bacteria after 18 to
negative bacilli, 20 Clostndium spp., 10 nonsporeforming 24 h of incubation. Various panel formats are available for
gram-positive bacilli, and 14 anaerobic gram-positive cocci. susceptibility testing.
The MIS and the RapID ANA System identified 97 and 94%, The identification panel contains 24 dried, modified, con-
respectively, of the anaerobic isolates tested (P > 0.20). ventional substrates. The data base includes information for
Ayers and Solomon (6) cultured Staphylococcus strains on identification of 42 species of the family Enterobactenaceae
TSBA at 28°C (595 cultures) or 5% sheep blood agar at 350C and 36 groups, genera, or species of nonfermentative and
(1,318 culture). The standard method for identification of oxidase-positive gram-negative bacilli in 18 to 24 h.
isolates was not mentioned. The strains were chromato- The autoSceptor is composed of a modified InterMed
graphed for cell wall fatty acids in the MIS, and the data ImmunoReader NJ-2000 microELISA reader, Digital Profes-
were stored by Library Generation Software. Cluster anal- sional 350/380 computer, and a Data Management Center
ysis with two-dimensional plots and dendrograms were used (DMC). The DMC has menus for demographic entry, data
to demonstrate isolate relationships. Factors that adversely analysis, editing, report generation, and epidemiological
affected relatedness included growth on TSBA, small-colony evaluations. A bidirectional mainframe interface is available.
variants, the initial culture from frozen or lyophilized refer- Inocula in 0.85% saline are prepared by suspending colo-
ence strains, cell wall fatty acid heterogeneity with some nies grown overnight on selective or nonselective media to
species, and the similar taxonomic position of some species. the equivalent of a 0.5 McFarland turbidity standard. Panels
The MIS software recognized tight clusters with high SIs are inoculated (100 ,ul per well) with a computer-controlled,
VOL. 5, 1992 AUTOMATED IDENTIFICATION SYSTEMS 321

TABLE 5. Studies comparing automated identification systems

No. of System (% identification accuracy)
organisms P value Reference
tested Vitek autoSCAN-4 WalkAway-96 Sensititre
246 GNI (99.2) R-GNB (94.3) <0.01 30
292a GNI (96.2) R-GNB (95.2) >0.50 95
91b GNI (90.1) R-GNB (95.6) >0.10 95
180a R-GNB (86.0) AP80 (79.0) >0.05 70
60b R-GNB (82.0) AP80 (35.0) <0.001 70
358a GNI (94.7) R-GNB (83.0) AP80 (93.3) <0.001 (GNI Vs R-GNB) 24
>0.20 (GNI vs AP80)
<0.001 (AP80 vs R-GNB)
142b GNI (79.6) R-GNB (74.6) AP80 (71.1) >0.30 (GNI vs R-GNB) 24
>0.05 (GNI vs AP80)
>0.50 (R-GNB vs AP80)
232 YBC (85.0) YIP (59.0) <0.001 107
a Members of the family Enterobacteriaceae.
b Gram-negative organisms that were not members of the Enterobacteriaceae.

seven-tip automatic inoculator or the totally automated for each study, and the cited references. The details of how
Sceptor preparation station. Panels are incubated off-line each automated system performed versus the reference
aerobically at 35°C for 18 to 24 h. Before panels are loaded system are found earlier in this paper where that particular
into the autoSceptor, specimens are logged into the DMC, automated system is reviewed.
bar code labels are attached to the panels, and the bar codes Table 6 compares various features of the automated iden-
are scanned with a bar code wand. This allows linking of the tification systems reviewed in this paper.
unique sequence number on the labeled panel to the patient
specimen information. Kovac's reagent is added to the DISCUSSION
indole well, and ferric chloride is added to the tryptophan
deaminase well. Immediately before the autoSceptor reads The systems reviewed here vary considerably in their
the panel, an instrument-mounted scanner reads the bar approach to the identification of microorganisms. The MIS,
code and links that panel with the patient specimen informa- for example, analyzes cellular material, whereas others use
tion in the DMC. To obtain test results, light absorbances of more conventional end points such as increase in cell density
the substrates in Sceptor panel wells are measured. Light or color changes due to shifts in pH. There is also some
from a halogen-tungsten lamp is projected onto a row of variability in the degree of automation, spectrum of organ-
seven wells. The light from each well is reflected, con- isms identified, and turnaround time. The constant is that
densed, filtered, and directed to a multibranch glass fiber, most of these instruments have proven their applicability in
which branches the light from each well to a separate clinical microbiology laboratories throughout the country.
photodiode detector. Absorbance values from selected fil- Specific capabilities, as well as a review of their operation
ters are used to determine color changes in the colorimetric and the details regarding accuracy of identification, are
tests. After all biochemical reactions are read, the DMC described in the individual sections of this review and will
determines the identification of the isolate and links that not be summarized in the discussion. Although the systems
result to any other patient data. available today have proven capabilities, there is ample
The only report in the literature on the autoSceptor is an room for improvement and we can expect an expanded range
abstract (21). The authors evaluated the accuracy of the of identification capabilities, shorter turnaround time, and
autoSceptor with 570 gram-negative bacilli. Panels read on greater automation, particularly in the area of data manage-
the autoSceptor were also examined visually, and the bio- ment, to be available in the future. The devices described
chemical test results were compared. There was more than here are, with the exception of the Vitek urine card, based
97% agreement for biochemical tests, and identification to on pure culture techniques. In other words, the organism
species on the autoSceptor was 95%. The reference system must be isolated before the identification process is per-
used to establish the true identity of the isolates was not formed. There are, however, procedures, both manual and
mentioned. automated, that can identify organisms directly in specimens
The disadvantages for the autoSceptor are off-line incuba- by using antibodies or nucleic acid probes. We believe that
tion of panels, manual addition of reagents, and identifica- these direct approaches have a very definite but limited
tion of only gram-negative bacilli. The initial report on the utility in the immediate future. For conditions such as
autoSceptor is favorable, but this system must be further meningitis, the consequences of the infection and the ex-
evaluated to determine its utility in the clinical laboratory. pected range of pathogens are limited enough to make such
an approach desirable and feasible, but in other situations,
STUDIES COMPARING AUTOMATED such as diarrhea, it would not be economically or often
IDENTIFICATION SYSTEMS medically justified. A variety of other approaches for iden-
tification of microorganisms have been described, such as
Five studies have compared the identification accuracy of flow cytometry, image analysis, and mass spectrometry (13,
either two or three different automated identification sys- 20, 69). These all have the potential to provide rapid identi-
tems. Table 5 shows the identification accuracy for each of fication of microorganisms but at present are beyond the
the automated systems, the percentile (P value) of the scope of the routine clinical microbiology laboratory.
chi-square distribution as determined by the chi-square test As discussed by Miller (78), because there is no conven-
322 STAGER AND DAVIS CLIN. MICROBIOL. REV.

0
la
_- 4_
0 ut ce c u
4) '-4
C0I - ;zzzzzz0 >4 >4 04 4 Z° >. >.
t--

Cu

*9 *0
zzzzzzz*Y E E E E E E
0 00

0 H- 0 0
c.)

ix z z >k
(14 en cn c C)e '413
0~~~~~~
C)
t' 1
n-C u nn
o o o o o o
0m zzo
t zZ d- d-
00
d1

: o~~~~ Cu
0 ~ ~ ~

C)

0 0 .0U
CZ
0
t zzo zo azoZ,
*-- 00 en Z ZZ
ZO ZZ o >Z >.ZR
0

CD

z 0U V° Q L

z Cu

0
E..
Cu)

0
'000
(0
C)
C.)
.o
0
4-o c0 C)

0 C)
00 '4

Cu
r.
E o o
0 0
0=zzzzzz0 zo 0o
HE
._

u C)
(A
0

0
10 co2 " rA o rCuCW
0
la
C) "D oo 00
Cu a-'i
Cu3
C.
0
Cu
co
t) t ; *00
0 0

00
0 ct z

0- 3 0 0

03 C._
oy~~~~~~ IC,"-q zt C.)

o o~~~~~~o
0 'IC00000 0-
14 ._
C.)

I1 Cu 4
I- 0 z z .0 0
C

3 00
009 0

i - - - - ~~~~~~- 000
Cu
D
80m .D
d
,It
~ ~ o° pe
''
0
Z
ooo
'
0
It
0 IR
"It

Cu-
.0 .
C14~~~~~~~~~~C
00
t
0-4 (=10
*-
aC;
tr,o 0 zz Z
4-
0 o
00

d-i
t
z
tt
>4 > >. >, > tn
D'4
w
0
00
Z ( U w 0 > 0
I14 _
u 0 m~-
;t.

Cu~~~=

4-0
~~ ~ ~ ~ ~ ~ ~
VOL. 5, 1992 AUTOMATED IDENTIFICATION SYSTEMS 323

tion for what constitutes an accurate or inaccurate test result country. As the data bases are expanded to include these
for automated identification systems and under what circum- organisms, the accuracy of the systems should improve.
stances they should be tested, the overall accuracy of the Would more substrates in a panel improve the accuracy of
systems is difficult to determine. Miller (78) presents a very identification? Lapage et al. (68) designed a computer pro-
good description of steps which could lead to a consensus gram to address this question. The results indicated that
approach to resolve many of the issues that have plagued us aberrant strains of members of the family Enterobac-
during our review of the literature on automated identifica- teriaceae and nonfermentative gram-negative bacilli re-
tion systems. We are unable to objectively answer the quired an average of 29 to 32 tests for reliable identification
question of which instrument is best for which application. and that additional tests did not improve the accuracy. Most
Among the many reasons is that the accuracy of a system is of the systems reviewed in this paper contain 29 or more
highly dependent on the organisms tested and that these substrates in their identification system. Certainly, there are
populations are never the same in the various studies. Also, circumstances under which the availability of fewer than 29
by the time a study is reported, the manufacturer has often substrates is limiting. For example, the Avantage BIC, with
changed the data base, altered thresholds on the reader, or 20 substrates, will identify only 10 members of the gram-
modified, added, or replaced some of the substrates. In negative organisms that are not members of the Enterobac-
addition, some studies have used photometrically standard- teriaceae.
ized inocula, whereas others have used generally less reli- How effective are automated identification systems in
able visually prepared inocula. Also, the reference system in determining the relatedness of isolates for epidemiologic
some studies may not always yield the correct identification, purposes? The Biolog system and the MIS have cluster
and hence the system under study can be unfairly penalized. analysis programs in the form of dendrograms and two-
In a recent reevaluation of the API 20E, O'Hara et al'. (89) dimensional or three-dimensional plots to demonstrate the
showed that a system did not even necessarily provide the relatedness of strains or species. There are, however, only
same level of accuracy over time. They were not able to limited reports on the effectiveness of these systems for
explain why this was so, but an obvious difference was an epidemiologic purposes. Wong (124) used the Biolog system
to test Brucella strains which were isolated from common-
expanded data base. The list of variables goes on! However, source infections or laboratory accidents and found that
it is obvious from the literature that automated systems strains that were epidemiologically related did have similar
accurately identify common clinical isolates. Rare biotypes metabolic profiles. However, it was not indicated whether
of common organisms and unusual organisms are often cluster analysis or dendrograms were used in that study. The
misidentified, identified at low likelihood, or not identified. study suggested that the Biolog system should prove useful
The accuracy of systems has been observed to vary even in epidemiologic studies. Mukwaya and Welch (84) used the
with the same kind of organism when it is evaluated in MIS to determine the cellular fatty acid profiles for 42 strains
different laboratories. Kelly et al. (60), Stevens et al. (111), of P. cepacia isolated from patients at five cystic fibrosis
and Truant et al. (120) suggested that variation in the centers and found, through numerical analysis of the data,
biotypes of individual species encountered in different areas that a different subgroup was present at each of the centers.
of the country may contribute to this performance variation. Clarridge and Harrison (23) evaluated the MIS for strain
In support of this hypothesis, Kelly et al. (60) reported a differentiation of X. maltophilia from the surgical intensive
collaborative evaluation of the Autobac by four laboratories care unit (7 strains) and medical intensive care unit (9
in various parts of the country. Each laboratory tested strains) in a hospital. When it was assumed that a group of
organisms collected in its geographic area. Correct identifi- strains from different patients were clustered with the same
cation of Citrobacter spp. ranged from 69 to 100% among the degree of relatedness as defined by all strains from one
laboratories, and that of Enterobacter spp. ranged from 62 to patient during different extractions and runs on the MIS, the
100%. The reason for this variation was not evident. The medical intensive care strains made up a single group
authors also noted a variation in accuracy with Acinetobac- whereas the surgical intensive care strains were grouped into
ter spp. at two of the laboratories. One of these laboratories three clusters. Only two studies have evaluated the repro-
correctly identified 65% of their strains, whereas the other ducibility of biochemical tests with automated instruments.
correctly identified 100% of their strains. When the strains Stoakes et al. (113) retested 50 strains of anaerobes (30
from the first laboratory were retested at the second labora- gram-negative bacilli and 20 clostridia) with the autoSCAN-4
tory, the accuracy forAcinetobacter spp. was similar to that AIP. Of the 1,200 reactions, 69 (5.6%) were recorded differ-
of the first laboratory. Land et al. (67) reported that only 27% ently between the two trials. As a result, 8% of the strains
of serogroup D isolates of Cryptococcus neoformans, which changed classification from correct identification to incorrect
are common to Europe and other temperate regions, were identification. Murray et al. (85) evaluated the reproducibil-
correctly identified by the MicroScan YIP. On the other ity of the biochemical reactions obtained with the Quantum
hand, 83% of serogroups B and C, serogroups geographically II BIC by testing, on three consecutive days, 40 gram-
found nearest to the manufacturer, were correctly identified. negative organisms belonging to members of the family
Serogroup A isolates, which are commonly found in the Enterobacteriaceae. Identical biocodes for all three tests
remainder of the United States, were identified 73% of the were obtained for only 10 (25%) of the 40 organisms. After
time. The results of Kelly et al. (60) and Land et al. (67) modification of the photometer, an additional 25 isolates
suggest that the data bases of these automated identification were evaluated. For the triplicate tests, identical biocodes
systems include limited biotypes from certain geographic were obtained for 13 organisms (52%). The results of these
areas. That geographic or regional variation exists is not last two studies suggest that the automated systems evalu-
surprising, but apparently some manufacturers have not ated are not highly reproducible in generating identical
adequately regionalized the strains in their data base, and biocodes and that the error rate would be too great for the
this has caused problems. We know that several manufac- systems to be of value in epidemiological studies. How
turers have programs for acquisition of biotypes of common reliable other automated systems would be in generating
and unusual organisms from various geographic areas of the identical biocodes has not been reported in the literature.
324 STAGER AND DAVIS CLIN. MICROBIOL. REV.

We believe that more standardized evaluations of auto- teriaceae biochemical cards versus Enterobacteriaceae-plus
mated identification systems should be performed but that biochemical cards. J. Clin. Microbiol. 15:575-581.
anyone contemplating such studies should examine very 8. Barry, A. L., and R. E. Badal. 1982. Reliability of early
carefully the observations of Miller (78). When trying to identifications obtained with Enterobacteriaceae-plus bio-
compare or determine the accuracy of the available instru- chemical cards in the AutoMicrobic system. J. Clin. Microbiol.
16:257-265.
ments, we were surprised to find that comprehensive evalu- 9. Barry, A. L., T. L. Gavan, R. E. Badal, and M. J. Telenson.
ations have not been reported, particularly since many of the 1982. Sensitivity, specificity, and reproducibility of the Auto
changes in data bases and other improvements have been Microbic system (with the Enterobacteriaceae-plus biochemi-
made. Some automated systems or parts of systems have not cal card) for identifying clinical isolates of gram-negative
been independently evaluated. bacilli. J. Clin. Microbiol. 15:582-588.
Manufacturers should consider providing test cards or 10. Barry, A. L., T. L. Gavan, R. E. Badal, and M. J. Telenson.
panels that contain only the supplemental substrates re- 1982. Direct comparison of two mechanized systems for iden-
quired for their identification systems. When supplemental tification of gram-negative bacilli. Am. J. Clin. Pathol. 78:462-
tests are necessary, they are frequently not readily available, 470.
and significant delay and expense are required for identifi- 11. Barry, A. L., T. L. Gavan, P. B. Smith, J. M. Matsen, J. A.
cation of what should not be a difficult isolate. There is a Morello, and B. H. Sielaff. 1982. Accuracy and precision of the
Autobac system for rapid identification of gram-negative ba-
need to develop techniques to identify and determine sus- cilli: a collaborative evaluation. J. Clin. Microbiol. 15:1111-
ceptibility patterns for life-threatening infections more 1119.
quickly. We need methods for direct testing of positive blood 12. Barth, S. S., K. B. Williams, S. J. Gibson, and L. B. Elliott.
cultures, for example. 1991. Rapid identification of gram-negative bacteria by carbon
We should make better use of the computer capabilities of source oxidation, abstr. C-211, p. 377. Abstr. 91st Gen. Meet.
the systems. Epidemiology, determination of the signifi- Am. Soc. Microbiol. 1991. American Society for Microbiol-
cance of organisms identified, and continuing education ogy, Washington, D.C.
could easily be enhanced. For example, when a relatively 13. Beckmann, E., and P. Connolly. 1990. Flow cytometry: intro-
duction and microbiological applications. Clin. Microbiol.
rare isolate is identified, the operator should be able to query Newsl. 12:105-112.
the computer data base for information regarding its likeli- 14. Belcher, K., N. Warren, and H. P. Dalton. 1990. Evaluation of
hood of causing infection, its probable susceptibility pattern, the MicroScan rapid yeast identification panel (YIP) and API
and other relevant data. 20C yeast system for the identification of yeasts, abstr. C-152,
Although it is difficult to imagine a more exciting and p. 369. Abstr. 90th Annu. Meet. Am. Soc. Microbiol. 1990.
stimulating period for clinical microbiologists than the recent American Society for Microbiology, Washington, D.C.
15. Bochner, B. 1989. "Breathprints" at the microbial level. ASM
past, the immediate future appears to be at least as stimu- News 55:536-539.
lating. 16. Bochner, B. R. 1989. Sleuthing out bacterial identities. Nature
ACKNOWLEDGMENTS (London) 339:157-158.
17. Burdash, N. M., G. Teti, M. E. West, E. R. Bannister, and J. P.
We thank Janice Edwards-Bryant and Susan Fogg for typing the Manos. 1981. Evaluation of an automated computerized sys-
manuscript. tem (AutoMicrobic system) for Enterobacteriaceae identifica-
tion. J. Clin. Microbiol. 13:331-334.
REFERENCES 18. Burdash, N. M., A. L. Welborn, G. Teti, E. R. Bannister, and
1. Aldridge, K. E., J. H. Wall, R. Facklam, L. B. Reller, A. J. P. Manos. 1985. Identification of gram-negative bacilli using
Janney, and C. V. Sanders. 1988. Identification of viridans the Autobac IDX. Diagn. Microbiol. Infect. Dis. 3:59-64.
streptococci by three commercial systems, abstr. C-346, p. 19. Carnahan, A. M., S. W. Joseph, and J. M. Janda. 1989. Species
389. Abstr. 88th Annu. Meet. Am. Soc. Microbiol. 1988. identification of Aeromonas strains based on carbon substrate
American Society for Microbiology, Washington, D.C. oxidation profiles. J. Clin. Microbiol. 27:2128-2129.
2. Almeida, R. J., J. H. Jorgenson, and J. E. Johnson. 1983. 20. Castleman, K. R. 1979. Images and digital processing, p. 3-13.
Evaluation of the AutoMicrobic system gram-positive identifi- In K. R. Castleman (ed.), Digital image processing, 1st ed.
cation card for species identification of coagulase-negative Prentice-Hall, Inc., Englewood Cliffs, N.J.
staphylococci. J. Clin. Microbiol. 18:438-439. 21. Chambers, C., J. Henja, B. Bradley, C. Patterson, C. Kennedy,
3. Appelbaum, P. C., M. R. Jacobs, J. I. Heald, W. M. Palko, A. L. Green, and G. Controni. 1990. Comparison of an automated
Duffett, R. Crist, and P. A. Naugle. 1984. Comparative evalu- system (AutoSceptor) to the manual Sceptor system, abstr.
ation of the API 20S system and the AutoMicrobic system C-166, p. 371. Abstr. 90th Annu. Meet. Am. Soc. Microbiol.
gram-positive identification card for species identification of 1990. American Society for Microbiology, Washington, D.C.
streptococci. J. Clin. Microbiol. 19:164-168. 22. Christenson, J. C., D. F. Welch, G. Mukwaya, M. J. Muszyn-
4. Armon, R., I. Potasman, and M. Green. 1990. Biochemical ski, R. E. Weaver, and D. J. Brenner. 1989. Recovery of
fingerprints of Legionella spp. by the BIOLOG system: pre- Pseudomonas gladioli from respiratory tract specimens of
sumptive identification of clinical and environmental isolates. patients with cystic fibrosis. J. Clin. Microbiol. 27:270-273.
Lett. Appl. Microbiol. 11:290-292. 23. Clarridge, J. E., and E. V. Harrison. 1991. The use of fatty acid
5. Aure, M., J. Brisker, M. Hirschmann, R. Almazan, and J. analysis for the tracking of Xanthomonas maltophilia (Xm)
Keiser. 1988. Comparative evaluation of the Quantum II and strains in a hospital, abstr. L-36, p. 428. Abstr. 91st Gen. Meet.
the API 20C systems for identification of germ tube negative Am. Soc. Microbiol. 1991. American Society for Microbiol-
yeasts, abstr. C-305, p. 382. Abstr. 88th Annu. Meet. Am. Soc. ogy, Washington, D.C.
Microbiol. 1988. American Society for Microbiology, Wash- 24. Colonna, P., D. Nikolai, and D. Bruckner. 1990. Comparison of
ington, D.C. MicroScan autoSCAN-W/A, Radiometer Sensititre and Vitek
6. Ayers, L. W., and M. C. Solomon. 1991. Performance of the systems for rapid identification of gram-negative bacilli, abstr.
Library Generation Software of the MIS system in the man- C-157, p. 370. Abstr. 90th Annu. Meet. Am. Soc. Microbiol.
agement of Staphylococcus speciation, abstr. C-156, p. 368. 1990. American Society for Microbiology, Washington, D.C.
Abstr. 91st Gen. Meet. Am. Soc. Microbiol. 1991. American 25. Connelly, M. R., and R. C. Jerris. 1987. Identification of yeasts
Society for Microbiology, Washington, D.C. isolated on antimicrobic-containing media using the Abbott
7. Barry, A. L., and R. E. Badal. 1982. Identification of Entero- Avantage System, abstr. C-348, p. 381. Abstr. 87th Annu.
bacteriaceae by the AutoMicrobic system: Enterobac- Meet. Am. Soc. Microbiol. 1987. American Society for Micro-
VOL. 5, 1992 AUTOMATED IDENTIFICATION SYSTEMS 325

biology, Washington, D.C. Microbiol. 1988. American Society for Microbiology, Wash-
26. Cooper, B. H., S. Prowant, B. Alexander, and D. H. Brunson. ington, D.C.
1984. Collaborative evaluation of the Abbott yeast identifica- 44. Goldstein, J., J. J. Guarneri, P. Della-Latta, and J. Scherer.
tion system. J. Clin. Microbiol. 19:853-856. 1982. Use of the AutoMicrobic and Enteric-Tek systems for
27. Costigan, W. J., and G. E. Hollick. 1984. Use of the Autobac identification of Enterobacteriaceae. J. Clin. Microbiol. 15:
IDX system for rapid identification of Enterobacteriaceae and 654-659.
nonfermentative gram-negative bacilli. J. Clin. Microbiol. 19: 45. Grasmiek, A. E., N. Naito, and D. A. Bruckner. 1983. Clinical
301-302. comparison of the AutoMicrobic system gram-positive identi-
28. D'Amato, R. F., H. D. Isenberg, G. A. McKinley, E. J. Baron, fication card, API Staph-Ident, and conventional methods in
R. Tepper, and M. Shulman. 1988. Novel application of video the identification of coagulase-negative Staphylococcus spp. J.
image processing to biochemical and antimicrobial susceptibil- Clin. Microbiol. 18:1323-1328.
ity testing. J. Clin. Microbiol. 26:1492-1495. 46. Gregson, D., D. Gopaul, P. Hostetler, H. Bialkowska-Hobrzan-
29. Davis, J. R., C. E. Stager, R. D. Wende, and S. M. H. Qadri. ska, and 0. Hammerberg. 1991. Evaluation of the Automicro-
1981. Clinical laboratory evaluation of the AutoMicrobic sys- bic System's ability to speciate and detect oxacillin-resistance
tem Enterobacteriaceae biochemical card. J. Clin. Microbiol. in coagulase-negative staphylococci, abstr. C-233, p. 381.
14:370-375. Abstr. 91st Gen. Meet. Am. Soc. Microbiol. 1991. American
30. Debates, M., R. Van Enk, and K. Thompson. 1989. Comparison Society for Microbiology, Washington, D.C.
of MicroScan autoSCAN W/A (rapid panels) with Vitek AMS 47. Hadfield, T. L., P. Lauderdale, B. Fisher, and R. Tapia. 1989.
for identification and susceptibility testing, abstr. C-265, p. Comparison of the Vitek AMS and MicroScan Walkaway
437. Abstr. 89th Annu. Meet. Am. Soc. Microbiol. 1989. automated microbial identification systems, abstr. C-264, p.
American Society for Microbiology, Washington, D.C. 437. Abstr. 89th Annu. Meet. Am. Soc. Microbiol. 1989.
31. DeBoer, S. H., and M. Sasser. 1986. Differentiation of Erwinia American Society for Microbiology, Washington, D.C.
carotovora spp. atroseptica on the basis of cellular fatty acid 48. Hare, D., G. Hall, B. Stinson, and M. Szakal. 1989. Compara-
composition. Can. J. Microbiol. 32:796-800. tive evaluation of the Quantum II, API Yeast-IDENT, and
32. Denys, G., M. Pfaller, L. Priller, M. Yu, and N. Yamane. 1990. Vitek AMS-YBC systems for rapid identification of yeasts not
Collaborative evaluation of the autoSCAN-W/A rapid gram Candida albicans, abstr. C-300, p. 443. Abstr. 89th Annu.
neg ID system, abstr. C-159, p. 370. Abstr. 90th Annu. Meet. Meet. Am. Soc. Microbiol. 1989. American Society for Micro-
Am. Soc. Microbiol. 1990. American Society for Microbiol- biology, Washington, D.C.
ogy, Washington, D.C. 49. Hasyn, J. J., and H. R. Buckley. 1982. Evaluation of the
33. deTurck, J., D. Brooks, M. Pease, and D. Bobey. 1987. Critical AutoMicrobic system for identification of yeasts. J. Clin.
evaluation of computerized gas liquid chromatography (GLC) Microbiol. 16:901-904.
system for bacterial identification, abstr. C-62, p. 333. Abstr. 50. Hasyn, J. J., K. R. Cundy, C. C. Dietz, and W. Wong. 1981.
87th Annu. Meet. Am. Soc. Microbiol. 1987. American Society Clinical laboratory evaluation of the AutoMicrobic system for
for Microbiology, Washington, D.C. rapid identification of Enterobacteriaceae. J. Clin. Microbiol.
34. El-Zaatari, M., L. Pasarell, M. R. McGinnis, J. Buckner, G. A. 13:491497.
Land, and I. F. Salkin. 1990. Evaluation of the updated Vitek 51. Hinnebusch, C., D. Nikolai, and D. Bruckner. 1989. Compari-
yeast identification data base. J. Clin. Microbiol. 28:1938- son of API Rapid Strep, IDS Rapid STR System, BBL Minitek
1941. Differential Identification System, and Vitek GPI with conven-
35. Ewalt, D. R., J. B. Payeur, and B. R. Bochner. 1990. Differen- tional biochemicals for the identification of viridans group
tiation of Brucella biovars using the Biolog system, p. 73. streptococci, abstr. C-259, p. 436. Abstr. 89th Annu. Meet.
Abstr. Meet. Am. Assoc. Vet. Lab. Diagnosticians 1990. Am. Soc. Microbiol. 1989. American Society for Microbiol-
36. Facklam, R., G. S. Bosley, D. Rhoden, A. R. Franklin, N. ogy, Washington, D.C.
Weaver, and R. Schulman. 1985. Comparative evaluation of the 52. Hussain, Z., L. Stoakes, D. L. Stevens, B. C. Schieven, R.
API 20S and AutoMicrobic gram-positive identification sys- Lannigan, and C. Jones. 1986. Comparison of MicroScan
tems for non-beta-hemolytic streptococci and aerococci. J. system with the API Staph-Ident system for species identifica-
Clin. Microbiol. 21:535-541. tion of coagulase-negative staphylococci. J. Clin. Microbiol.
37. Facklam, R., and M. D. Collins. 1989. Identification of Entero- 23:126-128.
coccus species isolated from human infections by a conven- 53. Isenberg, H. D., T. L. Gavan, P. B. Smith, A. Sonnenwirth, W.
tional test scheme. J. Clin. Microbiol. 27:731-734. Taylor, W. J. Martin, D. Rhoden, and A. Balows. 1980.
38. Farnham, S., N. Moss, and J. Scott. 1989. The identification of Collaborative investigation of the AutoMicrobic system En-
Vibrionaceae using the Vitek Gram-Negative Identification terobacteriaceae biochemical card. J. Clin. Microbiol. 11:694-
Card (GNI) with expanded data base, abstr. C-263, p. 437. 702.
Abstr. 89th Annu. Meet. Am. Soc. Microbiol. 1989. American 54. Johnson, J. E., and A. W. Brinkley. 1982. Comparison of the
Society for Microbiology, Washington, D.C. AutoMicrobic system and a conventional tube system for
39. Ferraro, M. J. B., M. A. Edelblut, and L. J. Kunz. 1981. identification of nonfermentative and oxidase-positive gram-
Accurate automated identification of selected Enterobac- negative bacilli. J. Clin. Microbiol. 15:25-27.
teriaceae at four hours. J. Clin. Microbiol. 13:151-157. 55. Jorgensen, J. H., J. W. Dyke, N. G. P. Helgeson, B. H. Cooper,
40. Forsythe, W. F., J. A. Brandt, and J. E. Lewis. 1990. Compar- J. S. Redding, S. A. Crawford, M. T. Andruszewski, and S. A.
ison study of Vitek YBC system and API 20C for identification Prowant. 1984. Collaborative evaluation of the Abbott Avan-
of clinical yeast isolates, abstr. C-155, p. 369. Abstr. 90th tage System for identification of frequently isolated nonfer-
Annu. Meet. Am. Soc. Microbiol. 1990. American Society for mentative or oxidase-positive gram-negative bacteria. J. Clin.
Microbiology, Washington, D.C. Microbiol. 20:899-904.
41. Freeman, J. W., R. W. Rowland, S. B. Overman, and N. L. 56. Jorgensen, J. H., J. E. Johnson, G. A. Alexander, R. Paxson,
Goodman. 1981. Laboratory evaluation of the AutoMicrobic and G. L. Alderson. 1983. Comparison of automated and rapid
system for identification of Enterobacteriaceae. J. Clin. Micro- manual methods for the same-day identification of Enterobac-
biol. 13:895-898. teriaceae. Am. J. Clin. Pathol. 79:683-687.
42. Gavini, F., M. 0. Husson, D. Izard, A. Berniguad, and B. 57. Kelley, R. W., S. Bascomb, M. Newland, D. Nothaft, R. Fields,
Quiviger. 1988. Evaluation of AutoSCAN-4 for identification of and M. V. Lancaster. 1987. A fluorescent system for two hour
members of the family Enterobacteriaceae. J. Clin. Microbiol. identification of gram-negative bacteria, abstr. C-57, p. 332.
26:1586-1588. Abstr. 87th Annu. Meet. Am. Soc. Microbiol. 1987. American
43. Godsey, J., R. Kelley, D. Nothaft, J. Bobolis, G. Enscoe, and K. Society for Microbiology, Washington, D.C.
Tomfohrde. 1988. Evaluation of the MicroScan Rapid Pos ID 58. Kelly, M. T., and J. M. Latimer. 1980. Comparison of the
panel, abstr. C-296, p. 381. Abstr. 88th Annu. Meet. Am. Soc. AutoMicrobic system with API, Enterotube, Micro-ID, Micro-
326 STAGER AND DAVIS CLIN. MICROBIOL. REV.

Media systems, and conventional methods for identification of 76. McLaughlin, J. C., W. G. Barron, T. L. Merlin, J. Fierro, and
Enterobactenaceae. J. Clin. Microbiol. 12:659-662. W. C. Thompson. 1991. A comparison of Biolog, Microscan,
59. Kelly, M. T., and C. Leicester. 1990. Evaluation of the Auto and Vitek AMS for the identification of infrequently isolated
scan-W/A rapid gram-negative identification-susceptibility human gram-negative bacterial pathogens, abstr. C-210, p. 377.
testing system, abstr. C-167, p. 371. Abstr. 90th Annu. Meet. Abstr. 91st Gen. Meet. Am. Soc. Microbiol. 1991. American
Am. Soc. Microbiol. 1990. American Society for Microbiol- Society for Microbiology, Washington, D.C.
ogy, Washington, D.C. 77. McWhirter, R. H., C. Sand, and P. C. Kibsey. 1990. Compar-
60. Kelly, M. T., J. M. Matsen, J. A. Morello, P. B. Smith, and ison of the Vitek YBC System and the Microscan Rapid Yeast
R. C. Tilton. 1984. Collaborative clinical evaluation of the ID System for the identification of germ tube negative yeast,
Autobac IDX system for identification of gram-negative bacilli. abstr. F-12, p. 410. Abstr. 90th Annu. Meet. Am. Soc. Micro-
J. Clin. Microbiol. 19:529-533. biol. 1990. American Society for Microbiology, Washington,
61. Kiehn, T. E., F. F. Edwards, D. Tom, G. Lieberman, E. M. D.C.
Bernard, and D. Armstrong. 1985. Evaluation of the Quantum 77a.Merriam-Webster Inc. 1981. Webster's new collegiate dictio-
II yeast identification system. J. Clin. Microbiol. 22:216-219. nary, 150th anniversary ed., p. 76. The G. & C. Merriam Co.,
62. Kloos, W. E., and C. G. George. 1991. Identification of Springfield, Mass.
Staphylococcus species and subspecies with the MicroScan 78. Miller, J. M. 1991. Evaluating biochemical identification sys-
Pos ID and Rapid Pos ID panel system. J. Clin. Microbiol. tems. J. Clin. Microbiol. 29:1559-1561.
29:738-744. 79. Miller, J. M., and D. L. Rhoden. 1991. Preliminary evaluation
63. Kloos, W. E., and D. W. Lambe, Jr. 1991. Staphylococcus, p. of Biolog, a carbon source utilization method for bacterial
222-237. In A. Balows, W. J. Hausler, Jr., K. Herrmann, H. identification. J. Clin. Microbiol. 29:1143-1147.
Isenberg, and H. J. Shadomy (ed.), Manual of clinical micro- 80. Morgan, M. A. 1984. Gas-liquid chromatography in the clinical
biology, 5th ed. American Society for Microbiology, Washing- microbiology laboratory. Lab. Med. 15:544-550.
ton, D.C. 81. Moss, C. W. 1981. Gas-liquid chromatography as an analytical
64. Kloos, W. E., and K. H. Schleifer. 1975. Simplified scheme for tool in microbiology. J. Chromatogr. 203:337-347.
routine identification of human Staphylococcus species. J. 82. Moss, C. W., M. I. Daneshrar, and D. G. Hollis. 1991.
Clin. Microbiol. 1:82-88. Automated GLC computer analysis of cellular fatty acids for
65. Lambert, M. A., and C. W. Moss. 1989. Cellular fatty acid identification of bacteria, abstr. C-121, p. 362. Abstr. 91st Gen.
compositions and isoprenoid quinone contents of 23 Legionella Meet. Am. Soc. Microbiol. 1991. American Society for Micro-
species. J. Clin. Microbiol. 27:465-473. biology, Washington, D.C.
66. Land, G., R. Stotler, K. Land, and J. Staneck. 1984. Update 83. Moss, C. W., P. L. Wallace, D. G. Hollis, and R. E. Weaver.
and evaluation of the AutoMicrobic yeast identification sys- 1988. Cultural and chemical characterization of CDC groups
tem. J. Clin. Microbiol. 20:649-652. EO-2, M-5, and M-6, Moraxella (Moraxella) species, Oligella
67. Land, G. A., I. F. Salkin, M. ElIZaatari, M. R. McGinnis, and urethralis, Acinetobacter species, and Psychrobacter immobi-
G. Hashem. 1991. Evaluation of the Baxter-MicroScan 4-hour lis. J. Clin. Microbiol. 26:484-492.
enzyme-based yeast identification system. J. Clin. Microbiol. 84. Mukwaya, G. M., and D. F. Welch. 1989. Subgrouping of
29:718-722. Pseudomonas cepacia by cellular fatty acid composition. J.
68. Lapage, S. P., S. Bascomb, W. R. Willcox, and M. A. Curtis. Clin. Microbiol. 27:2640-2646.
1973. Identification of bacteria by computer: general aspects 85. Murray, P. R., A. Gauthier, and A. Niles. 1984. Evaluation of
and perspectives. J. Gen. Microbiol. 77:273-290. the Quantum II and Rapid E identification systems. J. Clin.
69. Larsson, L. 1991. Gas chromatography and mass spectrome- Microbiol. 20:509-514.
try, p. 153-166. In J. H. Jorgensen (ed.), Automation in clinical 86. Navarro, M. C., M. A. Shulman, J. A. Bahrenburg, L. Klints,
microbiology, 1st ed. CRC Press, Inc., Boca Raton, Fla. and D. Schreier. 1987. Comparison of ALADIN video image
70. Lyznicki, J., S. Lester, D. Springer, L. Gvazdinskas, and W. processing to manual interpretation of An-IDENT, abstr. C-59,
Landau. 1991. Comparison of Microscan and Sensititre panels p. 333. Abstr. 87th Annu. Meet. Am. Soc. Microbiol. 1987.
for the rapid identification and susceptibility testing of gram- American Society for Microbiology, Washington, D.C.
negative rods, abstr. C-208, p. 376. Abstr. 91st Gen. Meet. 87. Nea, L., L. Van Pelt, S. Bascomb, K. Tomfohrde, D. W. Rogers,
Am. Soc. Microbiol. 1991. American Society for Microbiol- W. A. Sybers, and K. Irmen. 1989. Evaluation of the Micro
ogy, Washington, D.C. Scan rapid ID/MIC system, abstr. C-267, p. 438. Abstr. 89th
71. Master, R. N., R. L. Sautter, W. J. Brown, J. W. Hong, and Annu. Meet. Am. Soc. Microbiol. 1989. American Society for
A. E. Crist, Jr. 1991. Comparison of the Microbial Identifica- Microbiology, Washington, D.C.
tion System with conventional and commercially available 88. Oblack, D. L., J. C. Rhodes, and W. J. Martin. 1981. Clinical
systems for the identification of pink-pigmented, oxidase- evaluation of the AutoMicrobic system yeast biochemical card
positive bacteria, abstr. C-218, p. 378. Abstr. 91st Gen. Meet. for rapid identification of medically important yeasts. J. Clin.
Am. Soc. Microbiol. 1991. American Society for Microbiol- Microbiol. 13:351-355.
ogy, Washington, D.C. 89. O'Hara, C. M., D. L. Rhoden, and J. M. Miller. 1992.
72. Matthews, K. R., S. P. Oliver, and S. H. King. 1990. Compar- Reevaluation of the API 20E identification system versus
ison of Vitek gram-positive identification system with API conventional biochemicals for identification of members of the
Staph-Trac system for species identification of staphylococci family Enterobacteriaceae: a new look at an old product. J.
of bovine origin. J. Clin. Microbiol. 28:1649-1651. Clin. Microbiol. 30:123-125.
73. Mauchline, W. S., and C. W. Keevil. 1991. Development of the 90. O'Hara, C. M., D. L. Rhoden, and P. B. Smith. 1990. Agree-
BIOLOG substrate utilization system for identification of ment between visual and automated UniScept API readings. J.
Legionella spp. Appl. Environ. Microbiol. 57:3345-3349. Clin. Microbiol. 28:452-454.
74. McAllister, J. M., R. Master, and J. A. Poupard. 1991. Com- 91. Osterhout, G. J., V. H. Shull, and J. D. Dick. 1991. Identifica-
parison of the Microbial Identification System and the Rapid tion of clinical isolates of gram-negative nonfermentative bac-
ANA II system for the identification of anaerobic bacteria, teria by an automated cellular fatty acid identification system.
abstr. C-95, p. 358. Abstr. 91st Gen. Meet. Am. Soc. Micro- J. Clin. Microbiol. 29:1822-1830.
biol. 1991. American Society for Microbiology, Washington, 92. Peterson, E. M., J. T. Shigei, A. Woolard, and L. M. de la
D.C. Maza. 1988. Identification of viridans streptococci by three
75. McHugh, L. A., M. T. Clifford, B. A. Basille, and J. A. commercial systems. Am. J. Clin. Pathol. 90:87-91.
Washington. 1989. Evaluation of the MicroScan Walk-Away 93. Pfaller, M. A., M. J. Bale, K. R. Schulte, and F. P. Koontz.
rapid identification and susceptibility testing system, abstr. 1986. Comparison of the Quantum II bacterial identification
C-266, p. 437. Abstr. 89th Annu. Meet. Am. Soc. Microbiol. system and the AutoMicrobic system for the identification of
1989. American Society for Microbiology, Washington, D.C. gram-negative bacilli. J. Clin. Microbiol. 23:1-5.
VOL. 5, 1992 AUTOMATED IDENTIFICATION SYSTEMS 327

94. Pfaller, M. A., T. Preston, M. Bale, F. P. Koontz, and B. A. Abstr. 90th Annu. Meet. Am. Soc. Microbiol. 1990. American
Body. 1988. Comparison of the Quantum II, API Yeast Ident, Society for Microbiology, Washington, D.C.
and AutoMicrobic systems for identification of clinical yeast 110. Staneck, J. L. Personal communication.
isolates. J. Clin. Microbiol. 26:2054-2058. 111. Stevens, M., R. K. A. Feltham, F. Schneider, C. Grasmick, F.
95. Pfaller, M. A., D. Sahm, C. O'Hara, C. Ciaglia, M. Yu, N. Schaak, and P. Roos. 1984. A collaborative evaluation of a
Yamane, G. Scharnweber, and D. Rhoden. 1991. Comparison rapid automated bacterial identification system: the Autobac
of the AutoSCAN-W/A rapid bacterial identification system IDX. Eur. J. Clin. Microbiol. 3:419-423.
and the Vitek AutoMicrobic system for identification of gram- 112. St.-Germain, G., and D. Beauchesne. 1991. Evaluation of the
negative bacilli. J. Clin. Microbiol. 29:1422-1428. MicroScan rapid yeast identification panel. J. Clin. Microbiol.
96. Plorde, J. J., J. A. Gates, L. G. Carlson, and F. C. Tenover. 29:2296-2299.
1986. Critical evaluation of the AutoMicrobic system gram- 113. Stoakes, L., T. Kelly, K. Manarin, B. Schieven, R. Lannigan, D.
negative identification of glucose-nonfermenting gram-negative Groves, and Z. Hussain. 1990. Accuracy and reproducibility of
rods. J. Clin. Microbiol. 23:251-257. the MicroScan rapid anaerobe identification system with an
97. Quinn, P. R., and C. D. Horstmeier. 1990. Evaluation of Vitek automated reader. J. Clin. Microbiol. 28:1135-1138.
Yeast Biochemical Card (YBC) with an expanded data base for 114. Stoakes, L., B. C. Schieven, E. Ofori, P. Ewan, R. Lannigan,
the identification of clinically important yeasts and yeastlike and Z. Hussain. 1992. Evaluation of MicroScan Rapid Pos
organisms, abstr. F-11, p. 410. Abstr. 90th Annu. Meet. Am. Combo panels for identification of staphylococci. J. Clin.
Soc. Microbiol. 1990. American Society for Microbiology, Microbiol. 30:93-95.
Washington, D.C. 115. Stockman, L., and G. D. Roberts. 1991. Evaluation of Micro-
98. Rhoden, D. L., and C. M. O'Hara. 1989. Evaluation of the Scan Rapid Yeast Identification Panel (version 17), abstr. F-23,
updated QUANTUM II system for the identification of gram- p. 412. Abstr. 91st Gen. Meet. Am. Soc. Microbiol. 1991.
negative bacilli. J. Clin. Microbiol. 27:2420-2422. American Society for Microbiology, Washington, D.C.
99. Rhoden, D. L., P. B. Smith, C. N. Baker, B. Schable, and S. 116. Sylvester, M. K., and J. A. Washington II. 1984. Evaluation of
Stocker. 1985. AutoSCAN-4 system for identification of gram- the Quantum II microbiology system for bacterial identifica-
negative bacilli. J. Clin. Microbiol. 22:915-918. tion. J. Clin. Microbiol. 20:1196-1197.
100. Roman, S. B., L. A. Carson, C. M. O'Hara, D. A. Pegues, and 117. Tenover, F. C., T. S. Mizuki, and L. G. Carlson. 1990.
M. Miller. 1991. Comparison of four identification methods Evaluation of autoSCAN-W/A automated system for the iden-
used to identify Pseudomonas cepacia isolated from sputum of tification of non-glucose-fermenting gram-negative bacilli. J.
cystic fibrosis patients, abstr. C-222, p. 379. Abstr. 91st Gen. Clin. Microbiol. 28:1628-1634.
Meet. Am. Soc. Microbiol. 1991. American Society for Micro- 118. Tomfohrde, K., R. Kelley, D. Nothaft, J. Bobolis, G. Enscoe,
biology, Washington, D.C. and J. Godsey. 1988. Evaluation of the MicroScan Rapid Neg
101. Ruoff, K. L., M. J. Ferraro, M. E. Jerz, and J. Kissling. 1982. ID Panel, abstr. C-283, p. 379. Abstr. 88th Annu. Meet. Am.
Automated identification of gram-positive bacteria. J. Clin. Soc. Microbiol. 1988. American Society for Microbiology,
Microbiol. 16:1091-1095. Washington, D.C.
102. Salkin, I. F., K. H. Schadow, L. A. Bankaitis, M. R. McGinnis, 119. Tritz, D. M., P. C. Iwen, and G. L. Woods. 1990. Evaluation of
and M. E. Kemna. 1985. Evaluation of Abbott Quantum II MicroScan for identification of Enterococcus species. J. Clin.
yeast identification system. J. Clin. Microbiol. 22:442-444. Microbiol. 28:1477-1478.
103. Schiminsky, N., and P. Ferrieri. 1987. The use of the Vitek 120. Truant, A. L., E. Starr, C. A. Nevel, M. Tsolakis, and E. F.
AMS for speciation of the viridans and group D streptococci, Fiss. 1989. Comparison of AMS-Vitek, MicroScan, and Auto-
abstr. C-94, p. 339. Abstr. 87th Annu. Meet. Am. Soc. Micro- bac Series II for the identification of gram-negative bacilli.
biol. 1987. American Society for Microbiology, Washington,
D.C. Diagn. Microbiol. Infect. Dis. 12:211-215.
104. Sekhon, A. S., A. A. Padhye, A. K. Garg, and W. R. Pruitt. 121. Wallace, P. L., D. G. Hollis, R. E. Weaver, and C. W. Moss.
1987. Evaluation of the Abbott Quantum II yeast identification 1988. Cellular fatty acid composition of Kingella species,
system. Mykosen 30:408-411. Cardiobacterium hominis, and Eikenella corrodens. J. Clin.
105. Shulman, M. A., M. C. Navarro, J. A. Bahrenburg, S. Jahnke, Microbiol. 26:1592-1594.
and D. Schreier. 1987. Comparison of the ALADIN video 122. Weckbach, L. S., J. L. Staneck, R. C. Tilton, D. Douglas, R.
image processing to manual interpretation of UniScept 20E, Zabransky, L. Bayola-Mueller, C. M. O'Hara, and D. L.
abstr. C-58, p. 333. Abstr. 87th Annu. Meet. Am. Soc. Micro- Rhoden. 1990. Abstr. 90th Annu. Meet Am. Soc. Microbiol.
biol. 1987. American Society for Microbiology, Washington, 1990, C-234, p. 383.
D.C. 123. Welch, D. F. 1991. Applications of cellular fatty acid analysis.
106. Sielaff, B. H., J. M. Matsen, and J. E. McKie. 1982. Novel Clin. Microbiol. Rev. 4:422-438.
approach to bacterial identification that uses the Autobac 124. Wong, J. 1989. Carbon utilization patterns of Brucella sp.,
system. J. Clin. Microbiol. 15:1103-1110. abstr. C-210, p. 428. Abstr. 89th Annu. Meet. Am. Soc.
107. Simmonds, J. E., M. R. Motyl, and J. C. McKitrick. 1990. Microbiol. 1989. American Society for Microbiology, Wash-
Evaluation of Vitek Yeast Biochemical Card (YBC) with ington, D.C.
expanded data base and two rapid enzymatic panels: Micro 125. Woolfrey, B. F., R. T. Lally, M. N. Ederer, and C. 0. Quall.
scan Yeast Identification Panel (YIP) and API Yeast-Ident 1984. Evaluation of the AutoMicrobic system for identification
(Y-I) systems for identification of medically important yeasts, and susceptibility testing of gram-negative bacilli. J. Clin.
abstr. 1068, p. 264. Program Abstr. 30th Intersci. Conf. Anti- Microbiol. 20:1053-1059.
microb. Agents Chemother. American Society for Microbiol- 126. Woolfrey, B. F., R. T. Lally, and C. 0. Quall. 1983. Evaluation
ogy, Washington, D.C. of the AutoSCAN-3 and Sceptor systems for Enterobac-
108. Smith, S. M., K. R. Cundy, G. L. Gilardi, and W. Wong. 1982. teriaceae identification. J. Clin. Microbiol. 17:807-813.
Evaluation of the AutoMicrobic system for identification of 127. Yamane, N., F. Koontz, M. Lindsey, and V. Hirst. 1990.
glucose-nonfermenting gram-negative rods. J. Clin. Microbiol. Comparison of the Vitek YeastAPIBiochemical Card (YBC) with
15:302-307. the updated data base to the 20C for the identification of
109. Snyder, J., G. Buck, and B. Nohinek. 1990. Evaluation of the Soc.
yeast, abstr. C-233, p. 382. Abstr. 90th Annu. Meet. Am.Wash-
Abbott Avantage for bacterial identification, susceptibility Microbiol. 1990. American Society for Microbiology,
testing, and impact on antibiotic selection, abstr. C-172, p. 372. ington, D.C.